American Diabetes Association 82nd Scientific Sessions

June 3-7, 2022; New Orleans, LA (+Virtual); Full Report – Draft

Executive Highlights

  • There was a palpable sense of optimism at this year’s ADA Scientific Sessions, which was the first in-person ADA conference since 2019. Over 10,000 virtual and in-person attendees came together from 97 countries to attend more than 190 educational sessions and award lectures. Impressively, more than 1,300 posters were accepted, and ADA featured a whopping 88 companies and other organizations inside the Exhibit Hall. Given this incredible learning, it’s so beneficial that attendees can access all the content through September 5, 2022.
  • What an enthralling year in diabetes therapy, particularly with a plethora of data for novel incretin therapies. Dual GIP/GLP agonist tirzepatide certainly stole the show (see four highlights and 10+ posters), with an array of new data showing significant improvements on kidney outcomes, consistent effects regardless of age, as well as benefits on multiple aspects of beta cell function. Of course, the tirzepatide excitement was headlined by the SURMOUNT-1 readout, which found up to 23% weight loss at the highest dose (tirzepatide 15 mg). Incretins continue to show remarkable weight loss, including among new agents in the pipeline. Given this incredibly exciting evidence for 20%+ weight loss on newer incretin therapies, we were thrilled to hear KOLs debate the topic of weight management as a primary target in type 2 diabetes care.
    • Also in therapy, on the complications front, we were pleased to continue our learning on the benefits of SGLT-2s for heart failure through both data-based and guideline-focused presentations. Several advancements in type 1 diabetes “cures” also caught our attention, including new data from SERNOVA’s cell pouch as Time in Range data and other metrics from the first patient dosed with VX-880 (99.9% TIR and still doing well as we understand it)
    • Additionally, we attended several exciting readouts for youth with diabetes, including AWARD-PEDS and PRONTO-PEDS, highlighting that the benefits of these therapies are maintained in this young population. Overall, we are thrilled to see greater recognition of the benefits of newer insulin therapies and anti-hyperglycemic agents beyond just type 2 diabetes and moving into potential adjunctive therapy for type 1s as well as more real-world evidence across the board. We are leaving ADA inspired and invigorated about a brighter future with novel therapies!
  • On the tech front, CGM continued to draw a huge focus during this year’s Sci Sessions, with many presentations focusing on the use of CGM in people with type 2 diabetes. We saw results from the IMMEDIATE RCT (n=82) showing that FreeStyle Libre use with structured DSME drove a 0.3% A1c reduction and 2.3 hour/day improvement in Time in Range at 16 weeks relative to those only receiving DSME in type 2s not on insulin, adding to a growing body of evidence in support of CGM use beyond savvy, well-controlled people with type 1 diabetes. Dr. Rich Bergenstal also gave an impressive update from the International Diabetes Center, reporting that >60% of IDC patients on CGM have FreeStyle Libre data directly integrated into Epic. We appreciated a focus on health economics, with CGM cost-effectiveness from the UK and Canada. On the next-gen CGM front, FreeStyle Libre 3 stole the show with impressive accuracy data, and data from the CGM pipeline (Medtronic’s Simplera CGM and Eversense 365 day) leaves us bullish on the future of glucose monitoring. 
  • Also in tech: real-world AID evidence was the name of the game, much like the AID data that we encountered at EASD 2021 and ATTD 2022. This year, the real-world evidence at ADA focused mainly on two AID systems, with data from a massive cohort of Control-IQ users (n>20,000) and data from early users of MiniMed 780G with the Guardian 4 CGM. We also saw additional data from Control-IQ users with the 12-month results of the CLIO study (n=1,107) and data from pediatric users (n=59) demonstrating improvements in glycemic management and psychosocial outcomes. We were also hugely excited by the data we saw from the AID pipeline – specifically, data from MiniMed 780G/Klue, Beta Bionics’ insulin-only iLet, and Omnipod 5’s type 2 feasibility study stole the show, reaffirming that the future of the AID arena is bright as can be, as long as it becomes even more possible to prompt greater awareness, particularly among HCPs and for more to gain access.

In this report, we provide our full coverage the ADA’s 82nd Scientific Sessions. Titles highlighted in blue represent new coverage that wasn’t included in our daily highlights, and sentences highlighted in yellow denote findings or commentary that we found particularly notable.

Our sections proceed as follows (you can navigate through these using our table of contents below):

  • Themes
  • GLP-1 Agonists
  • SGLT Inhibitors
  • Novel Therapies
  • Insulin Therapies
  • Type 1 “Cures” and Adjunctive Therapies
  • Treatment Algorithms, Strategies, and Guidelines
  • Precision Medicine
  • Glucose Monitoring – BGM and CGM
  • Automated Insulin Delivery, Pumps, and Pens
  • Digital Health, Telemedicine, and Decision Support
  • Time in Range and Beyond A1c
  • Big Picture of Diabetes Technology
  • Diabetes Complications
  • Obesity
  • Policy, Epidemiology, and Education
  • Prediabetes, Diabetes Remission, and Diabetes Prevention
  • Award Lectures and Additional Topics
  • Social Determinants of Health
  • Interviews with ADA Leadership
  • TCOYD and The diaTribe Foundation’s 15th Annual Diabetes Forum
  • Exhibit Hall
Table of Contents 

Themes

Diabetes Therapy

Dual GIP/GLP-1 agonist tirzepatide takes center stage with paradigm-shifting weight loss and a first look at kidney benefits, just weeks after type 2 diabetes approval

Coming in about a month after tirzepatide’s type 2 diabetes approval and topline SURMOUNT-1 results in obesity, ADA 2022 made a splash by quickly adding full SURMOUNT-1 results to its program. SURMOUNT-1 found that that tirzepatide 15 mg dose conferred an average of 23% weight loss in people with obesity. Notably, the presentation highlighted that 40% of participants had prediabetes, and astoundingly, more than 95% of those with prediabetes and treated with tirzepatide reverted to normoglycemia over the course of the trial. Unsurprisingly, these results, along with other outstanding findings, were presented to a packed room filled with palpable excitement and people were leaning over one another to take pictures of every slide. Commentators and speakers throughout ADA 2022 positioned tirzepatide as part of a new class of medications that confers >10% weight loss and is ushering in a new era of obesity pharmacotherapy – a class that Dr. Lee Kaplan (Massachusetts General Hospital) called “3rd generation anti-obesity medications” and consists of semaglutide 2.4 mg (Wegovy) and tirzepatide. The advent of these therapies are facilitating a shift in the role of obesity treatment in relation to obesity complication treatment. Discussing the implications of SURMOUNT-1, Dr. Louis Aronne (Weill Cornell Medicine) explained these results support the shift of obesity treatment upstream so that obesity treatment entails weight loss and complication prevention, as opposed to prioritizing piecemeal treatment of multiple complications.

  • While tirzepatide is not approved for obesity yet, the SURMOUNT-1 results still have major implications for people with type 2 diabetes and obesity who are eligible for tirzepatide (~90% of people with type 2 diabetes have obesity). In fact, tirzepatide’s outstanding reversion rate to normoglycemia from prediabetes has fueled discussions on normoglycemia and weight loss as type 2 diabetes treatment goals. On the last day of ADA 2022, based in part on tirzepatide’s data, Dr. Julio Rosenstock (UT Southwestern) gave a provocative talk in which he not only advocated for a new definition of diabetes remission but called for normoglycemia through weight loss to be a type 2 diabetes goal at treatment initiation. We also heard a lively debate on whether weight management should be the primary target of type 2 diabetes management between Dr. Ildiko Lingvay  (UT Southwestern) and Dr. Jeffrey Mechanick (Mount Sinai). We should note though that based on the SURPASS program, tirzepatide may have slightly dampened (but still remarkable) weight loss effects in people with diabetes. Tirzepatide’s SURPASS CVOT in people with type 2 diabetes will confirm tirzepatide’s diabetes complication reduction and the impact on cardiovascular outcomes. We are particularly curious whether any benefits are dependent on tirzepatide’s weight loss benefits – the trial is ongoing and expected to complete in October 2024. 
  • While tirzepatide has not yet completed a cardio or renal outcomes trial, ADA 2022 featured a prespecified analysis of SURPASS-4 demonstrating tirzepatide’s significant kidney benefits compared to insulin glargine. The analysis found that tirzepatide led to a 32% reduction in urine albumin-to-creatinine ratio; a reduced rate of eGFR decline by 2.21 mL/min/1.73 m2 per year; and a 42% relative risk reduction on a composite kidney outcome (macroalbuminuria, 40% eGFR decline, end stage kidney disease, or renal death). Notably, these benefits were consistent regardless of baseline SGLT-2 use, indicating the potential for additional renal benefits with combination therapy. These results were greeted with rapturous applause in part because this is the first prespecified analysis of tirzepatide, or any dual incretin agonist, to show kidney benefits. One key area for further research is the mechanism by which tirzepatide confers this benefit. Interestingly, tirzepatide’s eGFR profile reflected that of SGLT-2s: an initial acute drop in eGFR followed by eGFR stabilization. SGLT-2s are thought to lead to this eGFR profile through glomerular tubular feedback, but it is unclear how tirzepatide leads to the eGFR drop.

Success of new anti-obesity medications moves the needle on obesity management, with greater conversation about weight loss vs. glycemic control as treatment goals for type 2 diabetes

Obesity was a MAJOR theme of this year’s conference, which was headlined by the SURMOUNT-1 readout of tirzepatide in people with obesity. Beyond the landmark SURMOUNT-1 trial, we were pleased to see obesity pharmacotherapy becoming more mainstream as evidenced by several informative symposia on medical management of obesity. Of course, we hope that the greater discussion of obesity as a chronic disease among the medical community will translate into greater awareness, recognition, and interventions at the health systems and public policy levels. Between tirzepatide’s 23% weight loss in the SURMOUNT-1 trial and semaglutide’s 15-17% weight loss shown in the STEP program, there are now two agents with demonstrated double-digit weight loss, marking a new era in obesity management. With multiple next-generation anti-obesity mediations, several KOLs began to broach the exciting topic of whether or not obesity should be the primary treatment target for people with type 2 diabetes, given that major weight loss has been associated with increased diabetes remission. This paradigm shift was reflected in the ADA/EASD Consensus Report on the Management of Hyperglycemia in Type 2 Diabetes, which call for weight loss and glycemic control as “co-primary” endpoints of hyperglycemia management, alongside cardiovascular risk management and cardiorenal protection. Boy has the field made strides in recognizing the many benefits of weight loss and the many available therapies in the armamentarium to achieve clinically relevant weight loss – as well as the important interventions around therapies, including nutrition, activity/exercise, stress reduction, and other psychological and behavioral factors. We sense that multiple clinicians are very gratified to see greater conversation around best practices for management of this chronic disease, as historically not a lot has been said about obesity at ADA and indeed, it is still not necessarily yet recognized by ADA as a disease, as far as we understand it. Notably, many of the obesity sessions were packed full of eager attendees; we’re thrilled to see such strong engagement on this front as more and more physicians, trainees, nurses, educators, and others who work in diabetes and endocrinology recognize that obesity is a chronic disease and should be treated as such. We continue to assume that tirzepatide’s regulatory submission and anticipated approval in obesity will be the biggest advancement in the field of obesity medicine in history and we are eagerly awaiting updates from Lilly on the obesity regulatory timeline. 

  • In particular, we enjoyed a stirring debate by Dr. Ildiko Lingvay (UT Southwestern) and Dr. Jeffrey Mechanick (Mount Sinai) on whether weight management should be the primary target of type 2 diabetes management. On the pro side, Dr. Lingvay argued that “15 is the new 7,” referring to a 15% weight loss goal instead of a 7% A1C target as the primary focus of diabetes management. On the con side, Dr. Mechanick argued that glucose control is necessary but not sufficient to properly treat diabetes. Despite the good-hearted debate, both speakers ultimately agreed on the importance of individualizing patient care – we, too, hope to learn more about the use of precision medicine to optimally treat obesity, especially when it comes to selecting the best pharmacotherapy for each individual patient, now that multiple options are available. Dr. Lingvay and Dr. Mechanick also agreed on the point that BMI is not an accurate nor adequate measure of adiposity, reflecting a growing recognition that BMI should not be used as a diagnostic tool because it does not measure body fat directly. We heard a similar debate at ENDO 2022, where Dr. Lingvay once again took the pro-stance, while Dr. David Nathan (Massachusetts General Hospital) argued against using weight management as the primary method to treat diabetes. We imagine there will be much more discussion of treating obesity to treat type 2 diabetes going forward and we look forward to possible guideline updates in the future as obesity pharmacotherapy continues to advance.

Earlier stage incretins illuminate bright future of diabetes and obesity pharmacotherapy; new research with old favorites showcases potential for expanded indications for GLP-1 class

Incretins took center stage at the 82nd Scientific Sessions – in addition to tirzepatide, we were pleased to see the emphasis on a number of earlier stage novel incretins. In particular, dual/tri-incretin agonists, as well as combinations between incretins and other molecules, are showing impressive glycemic lowering and weight loss effects. In nearly every discussion of obesity pharmacotherapy, there is significant conversation about how these new agents will continue to fill the obesity treatment gap. Cotadutide, a dual GLP-1/glucagon agonist manufactured by AstraZeneca, is currently being investigated in phase 2 trials for NASH and CKD. CagriSema, the cagrlinitide+semaglutide combination from Novo Nordisk, is often also touted as the next big blockbuster in diabetes and obesity treatment. The randomized, multiple-ascending phase 1 trial (n=96) for cagrilintide+semaglutide combination therapy found that participants on the highest dose (2.4 mg cagrilintide+2.4mg semaglutide) lost a whopping 17.1% of their body weight in just 20 weeks, with no sign of plateauing, compared to 10% body weight reduction on semaglutide. The consensus is that the CagriSema combination could reach an unprecedented >25% reduction in body weight over a longer time period. The candidate is currently being investigated in patients with diabetes in a phase 2 study that is expected to complete in the second half 2022.

  • In addition to these later stage candidates, there are several new incretin therapies in the pipeline. The studies are predominantly smaller phase 1 trials investigating the safety and tolerability, basic clinical efficacy, and pharmacokinetics of the compounds. GI-related side effects were the most commonly reported adverse events in all of the studies, which was not unexpected given this is consistent across the rest of the incretin class. The drugs include Pemvidutide, a GLP-1/glucagon candidate made by Altimmune, Dapglutide, a GLP-1/GLP-2 receptor agonist made by Zealand, LY3502970, an oral GLP-1 molecule made by Lilly, LY3437943, a GIP/GLP-1/glucagon tri-agonist from Lilly, Mazdutide, a GLP-1/glucagon agonist from Lilly, and Danuglipron, an oral GLP-1 made by Pfizer. See this table for more details on each candidate. While it is likely not all of the compounds will make it all the way through clinical development, and while it is positive to see so much research being invested in finding additional treatments for diabetes and obesity, we also hope to see much more research in cost-effectiveness as well as areas more remote to us like implementation science, where we know there is a huge amount that we do not know that represent important factors for success.
  • While the new incretins received significant attention, we were also excited to hear the AWARDS-PEDS study readout. The phase 3 trial (n=154, ages 10 to <18 years) found that over 26 weeks, both 0.75 mg and 1.5 mg dulaglutide doses led to significant A1c reductions of -0.6% and -0.9%, respectively, from a baseline of 8.1%, compared to a 0.6% A1c increase with placebo (p<0.001 for both comparisons to placebo). In contrast to adult studies of dulaglutide, youth on dulaglutide did not see significant reductions in body weight, BMI, or waist circumference. This finding is consistent with the two previous trials of GLP-1s (liraglutide and exenatide) in youth, the Ellipse and BCB114 studies. The lack of weight reduction is also consistent with findings from SGLT-2 dapagliflozin in youth, which was published in April 2022. Thus, despite the null finding on weight, dulaglutide was proven to be an effective treatment option in children and adolescents with type 2 diabetes, for whom there are few other treatment options. We are hopeful that this could lead to an expanded indication for this population, as the rising incidence of youth-onset diabetes and obesity are of growing concern.

Insulin innovation abound: New data on once-weekly and ultrarapid-acting insulins

ADA 2022, taking place 100 years after the first insulin treatment, demonstrated that insulin therapy continues to evolve. A jam-packed symposium highlighted the latest advancements among the three innovative insulin classes in development – glucose-responsive, oral, and once-weekly insulins – which Dr. Michael Weiss (Indiana University) called “third generation” insulins. We were excited to hear Dr. Weiss share unpublished pre-clinical data from his lab’s unimolecular glucose-responsive insulin, demonstrating that movement is still being made in this exciting but early-stage area of insulin innovation. Of the third generation insulins, once-weekly insulins are by far the farthest along in development, with Novo Nordisk’s insulin icodec having released topline data for multiple phase 3 trials and Lilly’s basal insulin Fc (BIF) having initiated phase 3 trials. In fact, while not formally discussed at ADA 2022, on the first day of the conference Novo Nordisk released topline results for icodec’s ONWARDS 1 and ONWARDS 6 trials. During the conference, we heard positive phase 2 results for both icodec and BIF, with icodec demonstrating noninferiority to insulin degludec and BIF demonstrating noninferiority to insulin glargine. Specifically, icodec demonstrated similar hypoglycemia frequency, magnitude, and physiological response as glargine. As Dr. Julio Rosenstock (UT Southwestern) put it, this study and its result is “brilliant” because it addresses “the most common misconceptions about insulin icodec” – that is, misconception that icodec has more severe episodes of hypoglycemia. Reflecting this result, we saw that BIF conferred a similar A1c reduction (-1.2%) and Time in Range improvement (+2.5 hours/day) as insulin degludec in people with type 2 diabetes without increasing hypoglycemia. For patients, taking a once-weekly insulin without increased burden of hypoglycemia has the potential to improve consistency of insulin use and quality of life, along with potentially reducing clinical inertia and improving glycemic management.

  • Marking a major advancement in pediatric type 1 diabetes, the phase 3 PRONTO-Peds study showed that ultrarapid insulin lispro had similar efficacy and safety as lispro in children and adolescents. Ultrarapid insulin lispro was initially approved by the FDA in June 2020 for use in adults with either type 1 or type 2 based on the PRONTO-T1D and PRONTO-T2D studies under the brand name Lyumjev, and in August 2021 the FDA approved an expanded indication for use in insulin pumps. These were notable developments for adults since one of the major advancements of ultrarapid lispro is that it is indicated for use within 20 minutes of eating. This indication would be particularly beneficial for children and adolescents who may not be able to precisely dose their insulin prior to each meal given greater unpredictable of daily life. Among people with type 1 diabetes, adolescents and young adults have the highest A1c compared to other age groups, regardless of technology use, so while no single therapy can ameliorate that trend the availability of this ultrarapid lispro could be a step forward in improving quality of care for this patient population often not included in clinical trials.  

New research highlights progress in cures and adjunctive therapies to improve the lives of patients with T1D

New research for patients with type 1 diabetes shows tremendous progress on type 1 cures, as well as progress in adjunctive therapies to improve the lives of people with diabetes while the cures are still in development. Diabetes cures have been the talk of the town recently, and for good reason. Vertex, which has been developing stem cell-derived pancreatic islet cell replacement therapy in people with type 1 diabetes, presented new data showing that the first patient dosed in the trial achieved insulin independence at Day 270. Notably, he has been insulin independent for more than 60 days with a Time in Range of 99.9% (!) Moreover, Sernova presented preliminary data from the phase 1/2 trial (n=7) of its Cell Pouch for islet encapsulation in people with type 1 diabetes and hypoglycemia unawareness. The first three patients dosed in the trial achieved insulin independence after one supplemental intraportal islet transplantation, maintaining insulin independence at the three-month, six-month, and two-year follow-ups. Furthermore, these three patients have achieved an A1c in the normoglycemia range of ~5%. Recall that earlier this month, Sernova partnered with Evotec to combine the Cell Pouch System for beta cell encapsulation with Evotec’s work toward insulin-producing, induced pluripotent stem cell-based beta cells to create a “functional cure” for diabetes. These findings suggest that we truly are on the brink of a new era of diabetes treatment, and we will be watching these trials closely to see future results.

  • In terms of adjunctive cures, both SGLT-2s and GLP-1s were found to improve glycemic control in patients with type 1 diabetes. A double-blind, cross-over study from Canada demonstrated that low-dose empagliflozin increases Time in Range by +3.1 hours/day in patients with type 1 diabetes on AID with no diabetic ketoacidosis (DKA) events. DKA is a potential side effect for patients taking an SGLT-2 inhibitor, exacerbated in patients with type 1, and has posed a challenge to approval of adjunctive therapies in this population. Thus, the improvement in glycemic control on AID without increased DKA risk is a considerable finding. The real-world STEMT trial (n=18) demonstrated that once-weekly semaglutide 1 mg  improved weight loss and glycemic control in people with type 1 diabetes. Notably, relative weight loss of ≥5% was achieved in 60% of participants, while weight loss of ≥10% was achieved by 40% of participants, and approximately one-third (35%) of participants achieved an A1c reduction of ≥0.5% from an already low baseline A1c of 7.4%. We’re encouraged to see positive results from the first trial of blockbuster type 2 diabetes therapy Ozempic in type 1 diabetes, particularly in people with a long duration of diabetes and good baseline glycemic control. We look forward to seeing longer term results from the STEMT1 trial, and we hope this will inspire further research evaluating GLP-1s in type 1 diabetes, such as the SEMA-AP study assessing semaglutide as an adjunct to closed-loop therapy.

Diabetes Technology

The evidence base for CGM for type 2 diabetes continues to expand: RCT supporting FreeStyle Libre in non-insulin-using type 2s, cost-effectiveness of rt-CGM in insulin-using type 2s, treatment progression driven by FreeStyle Libre, and more!

As is becoming increasingly true across conferences, ADA 2022 included a bounty of data on CGM in type 2 diabetes. The data spanned an RCT evaluating FreeStyle Libre CGM in non-insulin-using type 2 diabetes (the IMMEDIATE RCT); retrospective EHR and claims data analyses looking at glycemic outcomes, treatment progression, and value when combined with a digital coaching program; and a cost-effectiveness analysis of rt-CGM in insulin-using type 2s. Of course, overall CGM uptake remains low in type 2 diabetes; two late-breaking posters showed that only 12% of type 2s are using CGM, based on Vanderbilt University Medical Center EHR analysis. Based on this analysis, CGM remains more common in type 2s who are younger, use insulin, and have a higher A1c, but we hope that the data read out at ADA 2022 (in addition to that shared at ATTD 2022 in April and future studies to be read out in the future) push the field ahead and expand the use of CGM within and beyond these subgroups.

  • One of the most exciting CGM studies coming out of ADA 2022 is the IMMEDIATE study, which found that FreeStyle Libre use (combined with diabetes education) improves A1c and Time in Range in non-insulin-using type 2s relative to diabetes education. Specifically, the study found non-insulin-using type 2s see 0.3% A1c reduction and 2.3 hour/day Time in Range improvement with FreeStyle Libre and diabetes education vs. diabetes education alone at 16 weeks and achieved a Time in Range of 76% and an A1c of 7.6% at 16 weeks. The average number of therapies that participants were on did not change from baseline to 16 weeks (2.5 therapies/participant on average in both groups), suggesting that the improvements seen in the FreeStyle Libre arm were likely due to behavior changes. These results offer two important takeaways: (i) CGM – even one without alarms like Libre “1” – improves glycemic outcomes for people with type 2 diabetes not on insulin even when therapies are not changed when combined with diabetes education; and (ii) diabetes education on its own is incredibly valuable, as those who received solely education still saw improvements in PROs and A1c. We hope that when the IMMEDIATE Study is complete and is published, its findings will encourage changes in practice guidelines, payer decisions, and prescribing behavior, particularly because of its RCT design. To us, it is clear that CGM is valuable across a wide range of people with diabetes, and we’re very moved to see researchers investing in building out the evidence base to support this belief.
  • Given the mixed data on treatment changes in CGM RCTs (e.g., MOBILE, FLASH-UK), it was fascinating to see a retrospective Canadian private payer claims analysis showing that type 2s who used FreeStyle Libre CGM were more likely to intensify their diabetes therapy treatment than those on BGM. The massive analysis included 373,871 adults with type 2 diabetes in Canada with data from May 2018-April 2021 and showed that those on FreeStyle Libre were 86%-181% more likely to progress their diabetes therapy than those on BGM (range from 86%-181% more likely depending on baseline therapy use and use of insulin during the study). These findings suggest that FreeStyle Libre use may facilitate earlier treatment intensification and improve therapeutic inertia. They make us hopeful that in the real world, CGM is being used to guide therapy decisions, and we hope that Dr. Richard Bergenstal’s new C2GM algorithm will further support CGM-guided treatment decision-making in type 2 diabetes.

Real-world evidence galore: Data from real-world AID users demonstrates sustained improvements in glycemic outcomes on par with results from pivotal trials

ADA continued the theme of a bounty of real-world evidence from AID systems that we’ve previously commented on at EASD 2021 and ATTD 2022. This year, the real-world evidence at ADA focused mainly on two AID systems with data from a massive cohort of Control-IQ users (n>20,000) and data from early users of MiniMed 780G with Guardian 4 CGM. We also saw additional data from Control-IQ users with the 12-month results of the CLIO study (n=1,107) presented in a poster and data from pediatric users (n=59) demonstrating improvements in glycemic management and psychosocial outcomes. As we’ve previously discussed, we see this growing body of real-world evidence from AID systems as clear evidence of a maturing field with more people using AID. Additionally, we continue to be incredibly impressed by the glycemic outcomes achieved by AID users across real-world cohorts that, for the most part, mirror results from the systems’ pivotal trials.

  • Data from Control-IQ users in the real-world continues to demonstrate incredibly strong glycemic outcomes. Dr. Boris Kovatchev (University of Virginia) presented data from a massive real-world population (n=20,314) including both people with type 1 (n=19,354) or type 2 (n=960) diabetes. Across users with type 1 diabetes, Time in Range improvements were seen as soon as one week following system initiation and maintained throughout the 3-month study duration. As we’ve seen across diabetes technologies, the largest improvements in glycemic outcomes came from those patients with the highest baseline A1cs. Additionally, improvements in glycemic management were seen across age cohorts and among adults with type 2 diabetes. Collectively, we think these are very important findings of this real-world investigation demonstrating that diabetes technology can be successfully adopted by people who historically may not have been considered ideal candidates for technology use (e.g., older age and poorer baseline glycemic management). As we continue to see more data in support of technology use across populations, we remain hopeful that providers will be aware of any biases they hold related to technology use to ensure patients are being offered available technologies and engaging in collaborative decision-making with providers to inform their diabetes management.
    • 12-month data from the CLIO study demonstrated similarly strong improvements in glycemic outcomes for adults using Control-IQ.  Of note, the largest improvements in glycemic management among this population came from prior MDI users once again demonstrating that prior technology use is not indicative of success on AID systems.
    • We are also starting to see real-world evidence from pediatric users of AID systems including an analysis of 59 pediatric users of Control-IQ presented at ADA. Among this cohort, AID use was associated with improved glycemic management and psychosocial outcomes, especially among parents of pediatric Control-IQ users. While pediatric Control-IQ users didn’t achieve quite the same levels of Time in Range as their adult counterparts, they still saw an impressive 2.2 hour/day increase in Time in Range from baseline, which we certainly see as a win and can only imagine the relief parents of children with diabetes must feel knowing that these systems can help safely and effectively manage glycemic excursions. While there are certainly still numerous challenges of being a child with diabetes, we see the continued label expansions of current AID systems to include pediatric and even preschool aged users is a further demonstration of a maturing field in which this technology is becoming more available to some of patients it may benefit the most.
  • ADA 2022 also featured encouraging real-world data from MiniMed 780G with Guardian 4 CGM users. Excitingly, among this population (n=7,346), the calibration-free Guardian 4 CGM enabled an average of 92% of time in Auto Mode. Overall, MiniMed 780G with Guardian 4 CGM users achieved a Time in Range of 73%, which increased even more to 79% among users with optimal system settings (active insulin time of two hours and a glucose target of 100 mg/dL) with no significant increase in time in hypoglycemia. These results build on data from ATTD 2022 demonstrating impressive population-level outcomes among MiniMed 780G users globally and again cement our view that AID can, for those interested in using the technology, drive significant improvements in glycemic outcomes in real-life settings.

What’s next in tech? Fully closed loop systems, longer CGM wear time, AID with faster insulins, DIY systems, and AID for type 2s

If you thought ATTD 2022 was where we’d see the brightest and best of next-gen technology: hold your horses. While we saw impressive data at ATTD 2022 on the CGM and AID fronts in a variety of populations, ADA brough a slew of impressive new updates from a variety of different areas in the diabetes technology field, bringing us slightly closer to answering the question: “What’s next in tech?”

  • During a highly anticipated oral presentation session, we saw feasibility study data from MiniMed 780G users in fully closed loop mode via Klue’s meal gesture technology. As a reminder, Klue was acquired by Medtronic in December 2019 and is an Apple watch-based app that detects and differentiates eating and drinking hand gestures based on motion sensors. Specifically, when the watch is worn on the dominant hand, the app use gyroscope and accelerometer data to determine the probability of an eating or drinking hand-gesture in real-time. These detected gestures are translated into carbohydrate counts using gesture-to-carb mapping, which correlates duration of meal and number of gestures with carbohydrate count. Impressively, Time in Range was consistent between Klue-enabled MiniMed 780G (without meal announcements) and MiniMed 780G with manual meal boluses (76% vs. 79%; p=0.41). While Klue didn’t do as well during high-carb meals, these findings are certainly exciting and undoubtedly bring further progress toward the effort to develop fully closed loop AID systems – other notable groups include UVA’s RocketAP CamDiab’s CamAPS HX.
  • Also on AID, clinical trial data from less prominent AID systems took center-stage at ADA. This year’s Scientific Sessions brought plenty of additional learnings on the Beta Bionics’ insulin-only iLet system. Beyond the glycemic improvements we saw at ATTD 2022, additional data reinforced that iLet can drive significant improvements in children (those with baseline A1c >9% saw +7.4 hour/day in Range), as well as widespread benefits for adults across income level, race/ethnicity, education level, and insulin delivery method subgroups. Most interestingly, an exploratory analysis of the insulin-only iLet pivotal showed minimal glycemic improvement with Fiasp compared to aspart/lispro, with no A1c differences and only a slight +29 minute/day Time in Range improvement with Fiasp. Moving to DIY AID systems, we tuned in for the landmark readout of the CREATE RCT (n=97), showing that OpenAPS demonstrated a +2.4 hour/day improvement in Time in Range relative to SAP after six months. Specifically, the AID arm achieved +2.4 hours/day Time in Range from a baseline of 61% to 71% at six months. Comparatively, those in the control arm saw a 46 minute/day decrease in Time in Range, from 58% at baseline to 55% at the conclusion of the study.
  • On CGM, we saw accuracy data from two of the most highly anticipated sensors in the diabetes technology arena. From Medtronic, we saw preliminary accuracy data from the company’s next-gen Simplera (f.k.a. Synergy) CGM in the poster hall (672-P). The interim analysis (n=241, mix of type 1s and 2s) showed topline MARDs of 10.2%, 10.7%, and 10.1% in adult arms, youth arms, and youth buttocks, respectively (±20%/20 mg/dL of 91%, 88%, and 89%, respectively). Simplera’s accuracy appears to be near or above that of FDA iCGM special controls, although we’d note that this is still below the accuracy data we’ve seen for Dexcom and Abbott CGMs (example: Dexcom G7 preschool accuracy data presented at ATTD 2022). Based on the timeline shared in Medtronic’s 1Q22 call in May, Simplera is slated for FDA submission “this summer.” From Senseonics, we caught a poster presentation (665-P) with pilot data from Senseonics’ 365-day implantable CGM, demonstrating MARD of 9.8% (n=31,482 paired glucose measurements). We are very impressed by the overall MARD of 9.8% for the full 365-day use and will be interested to see how this compares to accuracy from Senseonics’ 365-day sensor once that pivotal trial begins, which, as of the company’s 1Q22 update, is expected in 4Q22.
  • Continuing the strong momentum from ATTD 2022, we saw impressive data from Omnipod 5’s type 2 feasibility study. During the extension phase, type 2s on MDI (n=21) achieved an A1c of 8% (-1.4% from baseline) and Time in Range of 58% (+3.5 hours/day), and basal-only type 2s (n=10) achieved a whopping 7.5% A1c (-2.0% from baseline) and Time in Range of 65% (+7.9 hours/day). Elsewhere at ADA 2022, we’ve seen other encouraging data on the use of AID among adults with type 2 diabetes from both Control-IQ and MiniMed 780G users, and we remain optimistic that these systems may become available to more patients with type 2 diabetes in the future.

The early bird gets the worm: New data from CLOuD Study and 4T Study highlight the benefit of early CGM and AID initiation soon after type 1 diabetes onset

As the value of diabetes technology becomes more widely accepted (particularly in type 1 diabetes), the question has shifted from if a person with type 1 diabetes should use technology to when they should initiate it. Research presented at this year’s ADA Scientific Sessions is beginning to answer this question with the resounding answer that early initiation offers massive glycemic benefits, whether looking at CGM or AID. Together, these studies set the stage for potential guideline updates to recommend early initiation of diabetes technology (or at least the offer of early initiation).

  • This year’s Scientific Sessions offered further insight into the value of early initiation of CGM in type 1 diabetes. At ATTD 2022, we saw the readout of the pilot 4T study, which showed that new-onset pediatric type 1s who initiated CGM early and engaged in timely interventions based on CGM data saw massive glycemic benefits compared to historic comparators who did not initiate CGM. Specifically, the pilot showed that 53% of those on CGM achieved an A1c <7% at one year (from a mean baseline A1c of 11.5%) whereas only 28% of the comparator group achieved an A1c <7% at one year (baseline A1c of 10.2%). Building on the readout of Stanford’s 4T pilot trial at ATTD 2022, Dr. Priya Prahalad (Stanford) read out results of the full 4T trial at this year’s Scientific Sessions. The full trial used a lower glycemic target than the pilot (<7% vs. <7.5%) and found that two-thirds of participants who initiated CGM use within the first month of type 1 diagnosis achieved a Time in Range <7% from a mean baseline A1c of 12.3%. Together, the full study and its pilot show the clear benefit of initiating CGM – and CGM-driven education) early in a person’s type 1 diabetes progression. These results align with the seven-year real-world data published in Diabetes Care in January 2022, which showed that initiation of CGM within one year of type 1 diabetes diagnosis led to sustained improvement in A1c relative to those who did not initiate CGM or initiated CGM later than one year into their diagnosis.
  • ADA 2022 offered what we believe is the first RCT data on the value of initiating AID technology at type 1 diabetes onset. The CLOuD RCT found that initiating CGM within 21 days of type 1 diagnosis effectively improved glycemic outcomes at one year and two years in youth (ages 10-16) relative to standard care. Specifically, A1c was -0.4% lower in the CamAPS FX group than in the standard care group at one year (6.9% vs. 7.3%), and the CamAPS FX group maintained this A1c out to two years while the standard care arm saw their A1c rise to 8%. Likewise, at one year, Time in Range was +2.4 hours/day higher in the CamAPS FX group than in the standard care group (64% vs. 54%), and this gap widened further at two years when the CamAPS FX group held steady at 64% and the standard care arm saw further Time in Range declines to 49%. Although the study did not demonstrate any protective effects of AID technology in mitigating declining C-peptide levels, which was the study’s primary outcome, the improvements in glycemic outcomes were striking and hugely significant given that improvements in glycemic outcomes at this age reduces these youth’s complication risks for the rest of their life. The CLOuD RCT is continuing out to 48 months (four years), and we look forward to seeing those long-term results. Stateside, the 4T study is now testing AID initiation within 1-3 months of diagnosis, and we hope that these results confirm those of the CLOuD RCT.

Digital health marches forward: Solid outcomes, record investment, but fewer abstracts than years past

As the digital health field continues to see skyrocketing levels of financial investment, we were pleased to see some data at ADA affirming the power of two key diabetes coaching programs, Virta and Level2. As a reminder, Rock Health recorded a record $29.1 billion in digital health investment over the course of 2021, while also highlighting how the field is showing signs of maturation after its pandemic-driven balloon. While investment is surely a sign of the potential value of digital coaching and remote monitoring, we know that many remain interested in seeing clinical research that validates the efficacy of these programs by comparing users’ health from baseline to after participation, as well as the amount of user engagement necessary to see clinically relevant improvements. 

  • Given all the interest and investment around digital health over the past year, we were somewhat surprised mobile apps weren’t a bigger topic at this year’s ADA. For some context, we counted 50 abstracts this year across our digital health and telemedicine category documents. This figure is down from 78 abstracts related to digital health at ADA 2021 and ADA 2020 and continues a general decline in the number of digital health abstracts over the past few years (75 abstracts in 2019, 82 in 2018, and 95 in 2017). Given the audience of ADA (primarily HCPs in diabetes), in our view, it’s possible that digital health companies are choosing to share data on their interventions in other settings.
  • We saw a slew of data from Virta Health reinforcing the glycemic and broad metabolic benefits of the company’s nutritional intervention. Impressively, Virta presented five-year outcomes from its nutritional coaching program (n=122), showing that participants achieved an 8% weight loss (257 lbs to 237 lbs) and an 0.3% A1c reduction (7.5% to 7.2%). As a point of comparison, during Virta’s two-year analysis of VLCI, A1c improved by 0.9% (from 7.7% to 6.8%) and weight improved by 12 kg (from 254 lbs to 227 lbs). While the five-year outcomes are not quite on par with those observed at two years, certainly we can see that the strength Virta’s model has helped some populations enormously. In a separate poster (1176-P), Virta shared that 20% of the individuals completing its five-year study achieved diabetes remission (A1c<6.5%, no meds for ≥3 months). We also heard directly from Virta’s Dr. Caroline Roberts on the broad metabolic benefits observed in Virta’s five-year cohort, and also got to see two-year outcomes from a Veteran sub-study showing similar results as the five-year study.
  • In big news for the diabetes coaching arena, we saw, to our knowledge, the first data readout from Level2 since it launched in July 2020. As a reminder, Level2 is UnitedHealth Group’s type 2 diabetes remission program that provides eligible members with a Dexcom G6, a Fitbit, smartphone app-based alerts, personalized coaching, and virtual specialist consultations. With 230,000 employer-sponsored UnitedHealthcare members with type 2 diabetes as eligible to enroll, we believe Level2 is the largest type 2 diabetes program offering CGM to-date, which makes this data readout particularly significant for the broader digital coaching arena. In participants with CGM data at followup (n=378), there was an 0.7% A1c/GMI reduction from 7.7% to 7.0%, with 54% of patients seeing a >0.5% A1c reduction. In those with an A1c >9% at baseline (n=69), there was a whopping 2.6% A1c decline from 10.1% to 7.5% – whoa.

Big Picture

ADA delivers home run conference in first hybrid gathering since 2019; smooth sailing for 10,000+ virtual and in-person attendees from 97 countries!

It was phenomenal to be back together in person at ADA for the first time since 2019, when we gathered in San Francisco. Our time in New Orleans was filled with familiar faces and old friends, all gathered together to learn about the cutting-edge research in diabetes therapy, technology, and big picture. We applaud the smooth production of the hybrid event – both the in-person and virtual members of our team were able to enjoy the stellar agenda. More than 10,000 participants gathered from 97 countries to participate in this conference. The lively exhibit hall captivated our team with its elaborate booths, helpful demonstrations, and interactive displays, and we greatly enjoyed hearing representatives tout the latest and greatest innovations in cardiometabolic disease. We were blown away by the number of abstracts (1,331!) and the sheer number of researchers dedicating their work to improving the lives of people with diabetes and obesity. Outside of the 190+ conference sessions, our team greatly enjoyed exploring New Orleans, consuming beignets by the pound, and spending time with old friends. We are already looking forward to future gatherings and seeing everyone again in San Diego next June. For now, though, we hope you enjoy the curated collection of knowledge that we’ve put together in our attempts to capture the very best learning from this magnificent conference. Also, be sure to catch up on daily columns from esteemed New York Times writer Mr. Jim Hirsch, whose musings offer greater insights on the finer points of this year’s scientific sessions.

Guidelines galore on type 2 diabetes, HF, and CKD: ADA/EASD Draft Consensus report on T2D, KDIGO/ADA statement on CKD and T2D, and ADA report on HF

This year’s Scientific Sessions featured several important guidelines across type 2 diabetes management as well as key diabetes complications. Overall, the latest guidelines on the management of type 2 diabetes, as well as key diabetes complications CKD and HF, reflect growing recognition of the cardiorenal risk-reducing benefits of SGLT-2s and GLP-1s. We’re pleased to see these novel therapies moving further up the treatment algorithms from different professional societies, though we recognize that their inclusion in the guidelines is just one step in ensuring all patients who could benefit are receiving these important therapies – clearly, much remains to be done in terms of patient and provider education, financial access, and the many other factors under the broad umbrella of “clinical inertia.” With that in mind, we applaud the ADA and EASD for including a focus on implementation in the 2022 draft ADA/EASD Consensus report on type 2 diabetes management and we hope that future guidelines will follow this approach, too.

  • Ahead of the official unveiling at EASD 2022, we attended a first look session on the draft ADA/EASD Consensus Report on the management of hyperglycemia in type 2 diabetes. Broadly, the esteemed panel of KOLs from the US and across the pond advocated in favor or “holistic, person-centered” type 2 diabetes care across the four components of (i) weight management, (ii) glycemic management, (iii) cardiovascular risk management, and (iv) cardio and renal protection. Notably, the guidelines recognize glycemic management and weight management as two “co-primary” goals for patients with type 2 diabetes. We see this as a reflection of the ongoing paradigm shift toward obesity and weight management as intimately related aspects of diabetes management for people with type 2 diabetes – we’ve attended several stirring debates on this topic at both ADA and ENDO this year! Additionally, this draft consensus report, like its 2018 predecessor, includes a robust “implementation” component providing key questions and information that must be answered for clinicians to effectively implement these best practices into their clinics. Overall, we were struck by the draft consensus report’s emphasis on individualization of care and use of novel anti-hyperglycemic therapies (namely, SGLT-2s and GLP-1s), to provide cardiorenal risk reduction in people with or at risk of ASCVD and CKD.
  • We got a deep dive into the soon-to-be published ADA/KDIGO joint statement on the management of diabetes and CKD. Dr. Peter Rossing (Steno Diabetes Center Copenhagen, Denmark) explained that this joint statement reflects the high alignment between ADA and KDIGO recommendations and hoped that greater guidance alignment increases successful guidance implementation. Based on his presentation, this joint statement reflects the latest updated to the 2022 ADA Standards of Care in May 2022 and the 2022 KDIGO guidelines to be published later this year (see the draft 2022 KDIGO guidelines). The major update is that SGLT-2s are now recommended as first-line treatment with or without metformin in people with DKD and type 2 diabetes with an eGFR ≥20 mL/min/1.73 m2. Notably, the current guidelines recommend GLP-1 addition only if a patient’s individualized glycemic target is not met since there has been no kidney outcomes trial for a GLP-1 – the results of the FLOW trial of semaglutide in CKD (expected completion August 2024) may eventually change the positioning of GLP-1s.   
  • The ADA’s consensus report on heart failure was published on June 1, 2022 and presented at ADA 2022 by Dr. James Januzzi (Harvard). We were pleased to see that the report was endorsed by the American College of Cardiology, indicating the ADA’s recommendations align with the 2022 ACC/AHA/HFDSA guidelines of HF. The report emphasizes the use of SGLT-2s in people with HF, saying that first-line SGLT-2 treatment is “an expected element of care” in all people with diabetes and symptomatic HF (stages C or D) and SGLT-2s should be prioritized in people with stage B or pre-HF. The 16-page report also touches on opportunities for multidisciplinary care, providing recommendations for when to refer patients to a cardiologist, and using biomarkers to identify people with stage B (at-risk) HF.   

Keeping it real: Real-world evidence, cost-efficacy analyses, and prescribing uptake data shed light on usage of important diabetes medications outside of the four walls of the clinic

We were pleased to see a plethora of real-world evidence at this year’s ADA conference across both diabetes therapy and technology presentations. While large outcomes trials are, of course, hugely important to determine safety, efficacy, and durability of novel interventions prior to regulatory submission, real-world data provides another crucial component of the puzzle by revealing how diabetes therapies are used in practice, outside of strict clinical study protocols. As we’ve heard time and again, people with diabetes spend the vast majority of their time managing their diabetes on their own, without the direct support of a clinician and beyond the “four walls” of the clinic. Therefore, real-world evidence (RWE) is key to providing a more accurate picture of the efficacy of interventions with real, human use. Furthermore, such analyses can help reveal practical barriers to treatment initiation and continuation beyond the umbrella term of “clinical inertia” and may therefore better inform physician, nurse, certified diabetes educator (CDES), and other HCP interventions to support sustained use of these therapies. We applaud the ADA organizers for including such a broad array of real-world data and we hope these studies can better inform diabetes education protocol, particularly for more challenging and controversial therapies, such as adjunctive use of SGLT-2s and/or GLP-1s in type 1 diabetes.

  • We attended several compelling presentations on real-world trials of GLP-1s and SGLT-2s, including in people with type 1 diabetes. Particularly exciting to us were the six-month results from the STEMT1 trial, which found that once weekly semaglutide 1 mg conferred moderate weight loss and A1c reductions in 18 people with type 1 diabetes. Specifically, relative weight loss of ≥5% was achieved in 60% of participants, while weight loss of ≥10% was achieved by 40% of participants. Approximately one-third (35%) of participants achieved an A1c reduction of ≥0.5% from an already low baseline A1c of 7.4%, bringing participants almost to the ADA’s generally recommended target A1c of 7% for adults with type 1 diabetes. Also of note, a retrospective analysis of SGLT-2 treatment in type 1 diabetes conducted in Denmark, which found no cases of diabetic ketoacidosis among 134 people treated with an SGLT-2. Although the quality assurance study only included minimal data on age, sex, SGLT-2 treatment duration, and DKA frequency, we’re certainly pleased to see no incidence of DKA among a carefully selected patient population and we believe such an approach could be key to more widespread use of these important cardiorenal agents in people with type 1 diabetes.
  • On the complications front, we saw further data on the use of newly approved heart failure drugs, which revealed low utilization despite updated guidelines from the ACC/AHA/HFSA earlier this year and strong evidence of benefit across several outcomes trials (EMPEROR-Reduced, EMPEROR-Preserved, DAPA-HF, EMPULSE, CHIEF-HF). Indeed, a retrospective analysis of claims data from EMPRISE solidified empagliflozin’s ability to reduce hospitalization for heart failure compared to GLP-1s, though no significant differences on other CV outcomes was found in this analysis. Despite these clear benefits in both clinical trials and real-world populations, another retrospective claims analysis documented low use of SGLT-2s among 10,170 adults with a primary diagnosis of heart failure. Nearly one-third (30%) of patients with diabetes and heart failure and ejection fraction (HFrEF) and 13% of those with diabetes and heart failure with preserved ejection fraction (HFpEF) were on SGLT-2s, indicating that there remains significant room for optimization of guideline-directed medical therapy in HF patients. We were especially concerned to learn that SGLT-2 use was significantly higher among those with no CKD (11%) compared to those with mild CKD (7%, p<0.0001), despite the high disease burden faced by patients with comorbid HF and CKD. While this real-world analysis is highly discouraging, we know that this data combined with the right cost-effectiveness data could spur action at the provider, health system, and health policy level to ensure greater uptake of these important therapies.
  • Real-world studies also shed light on barriers to treatment initiation and continuation, such as the high cost of many newer medication classes. We were especially struck by national claims analysis data showing a correlation between out-of-pocket (OOP) costs and likelihood to initiate SGLT-2 or GLP-1 among people with type 2 diabetes with established CVD on metformin monotherapy (n=72,743). Participants in the highest OOP quartile were 21% less likely to initiate an SGLT-2 than those in lowest OOP cost quartile (HR=0.79) and were 12% less likely to initiate a GLP-1 than those in the lowest OOP cost quartile (HR=0.88) when adjusted for demographics, clinical characteristics, insurance plan, and provider and laboratory characteristics. Of particular note, the study population was fairly diverse in terms of socioeconomic status, though the majority (88%) were covered by a Medicare Advantage plan. This study also further reinforces the low uptake of SGLT-2s and GLP-1s described in JAMA studies from June 2021, April 2021, and February 2022.

GLP-1 Agonists

Phase 3 AWARD-PEDS trial: Dulaglutide demonstrates superior A1c reduction (-0.9%) compared to placebo in youth with type 2 diabetes; dulaglutide did not confer significant weight loss, in contrast to adult studies

Dr. Silva Arslanian (UPMC Children’s Hospital of Pittsburgh, PA) presented highly anticipated results from the AWARD-PEDS study investigating once-weekly GLP-1 dulaglutide vs. placebo in youth with type 2 diabetes (5-LB). The results were published on June 4, 2022 in the New England Journal of Medicine, along with a Lilly press announcement. Across 46 centers in nine countries, children and adolescents (n=154; ages 10 to <18 years) with type 2 diabetes and inadequate glycemic control were randomized 1:1:1 to receive dulaglutide 0.75 mg, dulaglutide 1.5 mg, or placebo for one year. Participants on placebo switched to dulaglutide 0.75 mg after 26 weeks. Over 26 weeks, both 0.75 mg and 1.5 mg dulaglutide doses led to significant A1c reductions of -0.6% and -0.9%, respectively, compared to a 0.6% A1c increase with placebo (p<0.001 for both comparisons to placebo). Sharing a chart of A1c change over 52 weeks (see second figure below), Dr. Arslanian pointed out the significant decline in A1c after 13 weeks in the dulaglutide arms and the drop in A1c after participants on placebo switched to dulaglutide at week 26. In a response to a question about participants’ A1c in the dulaglutide arms drifting upwards over 52 weeks, Dr. Arslanian explained that this trend, while not significant in this study, may reflect the rapid deterioration of beta cell function seen in youth with type 2 diabetes, as shown in the TODAY2 study. Pooling dulaglutide arms, about 57% of participants achieved an A1c <7%, compared to 18% on placebo. As for safety, there were no significant differences between groups in serious adverse events, adverse events leading to discontinuation, or hypoglycemia. As expected for the GLP-1 class, those on dulaglutide had an increase in transient, mild gastrointestinal events, primarily during the first two weeks after therapy initiation. Overall, Dr. Arslanian concluded that dulaglutide was superior to placebo in improving glycemic control with improvement sustained through 52 weeks and an overall safety profile similar to that seen in adult dulaglutide trials.

  • In contrast to adult studies of dulaglutide, youth on dulaglutide did not see significant reductions in body weight, BMI, or waist circumference. This finding is consistent with the two previous trials of GLP-1s (liraglutide and exenatide) in youth, the Ellipse and BCB114 studies. The lack of weight reduction is also consistent with findings from SGLT-2 dapagliflozin in youth, which was published in April 2022. The authors of AWARD-PEDS suggest two potential explanations for the lack of weight loss from dulaglutide: (i) worsening glycemic control in the placebo group may have led to weight reduction that confounded the detection of weight reduction associated with dulaglutide treatment; and (ii) previous studies suggest that as hyperglycemia worsens, GLP-1s are less effective at reducing weight, suggesting that the rapid loss of glycemic control in youth may mitigate dulaglutide’s weight loss activity. The authors rejected the idea that rapid adolescent growth could have led to this result since most participants had already reached their final height.
  • These results are especially significant considering that there are only a few approved therapies for youth with type 2 diabetes. Beyond metformin and insulin, the FDA has only approved two GLP-1s for use in youth type 2 diabetes: once-daily Victoza (liraglutide) in June 2019 and once-weekly Bydureon BCise (exenatide extended-release) in July 2021. Victoza’s approval was based on the phase 3 Ellipse study published in NEJM, and Bydureon’s approval was based on the phase 3 BCB114 study presented at ADA 2021. Both drugs are approved for people with type 2 diabetes ages 10+. While no SGLT-2 is approved in the US for youth with type 2 diabetes, in April 2022 The Lancet Diabetes & Endocrinology published phase 3 data showing that dapagliflozin conferred significant A1c reduction compared to placebo, though only when considering participants who took treatment ³80% of the study period.

Dr. Liana Billings highlights 17% reduction in stroke with GLP-1s vs. placebo in 2021 meta-analysis and calls for combination GLP-1 and SGLT-2 therapy based on complementary, interconnected cardiorenal benefits

Dr. Liana Billings (NorthShore University HealthSystem, IL) delivered a powerful call to action for GLP-1 pharmacotherapy to prevent ischemic strokes, emphasizing that “what we do matters” and urging providers to adopt a multi-risk factor approach to stroke prevention. Dr. Billings noted that stroke is the leading cause of serious long-term disability, has a high 10-year recurrence rate of 11%, and is one of the most feared diabetes complications. In particular, over half of stroke survivors aged >65 years have reduced mobility, which can lead to profound impacts on quality of life, not to mention the additional challenges of engaging in physical activity and maintaining a healthy weight, which may affect glycemic control and diabetes management. Dr. Billings noted that while individual GLP-1 CVOTs such as LEADER, PIONEER 6, and AMPLITUDE-O have not found statistically significant reductions for stroke reduction, several recent meta-analyses have helped solidify an up to 17% reduction in stroke risk with GLP-1s versus placebo. In particular, Dr. Billings cited a 2021 meta-analysis by Sattar et al. (published in The Lancet), which found an overall hazard ratio of 0.83 across eight GLP-1 CVOTs with a combined total of 60,000+ participants (p=0.0002), with very consistent lowering of the hazard ratio across all trials in favor of GLP-1s. Notably, this finding was observed in both fatal and non-fatal stroke. Likewise, meta-analyses have revealed comparable effects of GLP-1s on stroke in people with or without a history of stroke or transient ischemic attack (TIA, or “mini stroke”). Dr. Billings also highlighted a 2020 exploratory analysis by Gerstein et al. (published in The Lancet) that found a lower risk of disabling stroke with dulaglutide treatment vs. placebo (HR 0.76, p=0.0096). On potential mechanisms of action, the exploratory analysis found that about half (54%) of dulaglutide’s stroke-reducing benefit was likely derived from reductions in A1c and 14% came from effects on systolic blood pressure; the remaining one-third of benefit may be due to direct effects or other factors.

  • In contrast to GLP-1s, Dr. Billings noted that SGLT-2 meta-analyses do not show evidence of ischemic stroke risk reduction. Overall, she acknowledged the complementary and interconnected benefits of GLP-1s and SGLT-2s, calling for providers to move away from an “either/or” approach and towards an “and” (or additive) approach to pharmacotherapy for cardiovascular risk reduction.
  • Dr. Billings emphasized the importance of engaging and educating colleagues on the stroke reduction benefits of GLP-1s, since many neurologists are not yet aware of these anti-hyperglycemic agents. We would additionally call for further collaboration across specialties to maximize provider awareness and to ensure that all people who could benefit from GLP-1s, both those with and without diabetes, are receiving this valuable pharmacotherapy.

Semaglutide 2.4 mg reduces 10-year risk of type 2 diabetes by 61% in people with overweight/obesity compared to placebo, according to post hoc analysis of STEP 1 and STEP 4 trials

Dr. W. Timothy Garvey (University of Alabama at Birmingham) presented a post hoc analysis of the STEP 1 and STEP 4 trials investigating the effect of semaglutide 2.4 mg treatment on the Cardiometabolic Disease Stating (CMDS) score, which assesses type 2 diabetes risk over 10 years (2-LB). For background, STEP 1 (n=1,961) investigated weight loss with semaglutide 2.4 mg vs. placebo in people with obesity or overweight with complications, and STEP 4 (n=803) investigated the continuation of semaglutide 2.4 mg treatment for 68 weeks vs. withdrawal after 20 weeks in people with obesity or overweight with complications. Dr. Garvey explained that the CMDS score incorporates sex, age, race, BMI, blood pressure, blood glucose, triglycerides, and HDL cholesterol and serves to help clinicians identify patients with obesity who could benefit the most from treatment. Overall, this post hoc analysis found that in STEP 1 semaglutide 2.4 mg reduced 10-year type 2 diabetes risk by 61%, significantly more than the 13% reduction with placebo (p<0.001). In STEP 4, from weeks 20 to 68, those who continued semaglutide 2.4 mg had a 32% reduced 10-year type 2 diabetes risk, whereas those who switched to placebo at week 20 saw a 10-year type 2 diabetes risk increase of +41% (p<0.001). See the absolute changes in risk scores in the figure below. Notably, the relative risk reduction was similar among those with normoglycemia and prediabetes, but those with prediabetes experienced a greater absolute risk reduction since they were at higher risk at baseline. Dr. Garvey concluded that not only is semaglutide highly effective in reducing type 2 diabetes risk in people with normoglycemia and prediabetes but also sustained treatment is necessary to maintain this risk reduction and to prevent type 2 diabetes. He added that semaglutide treatment should be targeted to people with obesity and prediabetes to prevent the most cases of type 2 diabetes, which is particularly relevant considering the supply shortages for semaglutide 2.4 mg (Wegovy).

  • These results continue the narrative of weight loss therapies for type 2 diabetes prevention. Given landmark results from the SURPASS program and SURMOUNT-1 on GIP/GLP-1 tirzepatide, we’ve heard from multiple speakers on making weight loss and potentially normoglycemia the treatment goal of type 2 diabetes. In fact, SURMOUNT-1 showed that 95% of people on tirzepatide with prediabetes reverted to normoglycemia. As Dr. Louis Aronne discussed during the full SURMOUNT-1 readout at ADA 2022, the advent of therapies yielding >10% weight loss (semaglutide 2.4 mg and tirzepatide) are facilitating a new obesity paradigm that involves first treating obesity/overweight and then treating other cardiometabolic factors, such as blood glucose, if they require additional management. In other words, the impressive weight loss and reversion to normoglycemia from semaglutide and tirzepatide support the shift of obesity treatment upstream so that it entails weight loss and complication prevention, as opposed to prioritizing piecemeal treatment of multiple complications. This post hoc analysis of the STEP 1 and STEP 4 trials reinforce these reflections we heard at ADA 2022.  

Dulaglutide and liraglutide show sustained glycemic and body weight reduction at 24 months in real-world TROPHIES study

Dr. Bruno Guerci (University Hospital of Nancy, France) presented the poster on the TROPHIES study, a real-world observational study of health outcomes with dulaglutide and liraglutide in people with type 2 diabetes. The study investigated the real-world efficacy if people on once-weekly dulaglutide (n=1,1014) and once-daily liraglutide (n=991) in France, Germany, and Italy. Secondary outcomes included analysis of persistence of treatment, clinical outcomes, treatment patterns, and health care resource utilization. The study found that both dulaglutide and liraglutide conferred significant reductions in A1c and body weight, sustained over the 24-month study period. Although this was a prospective study, participants were not randomized between the two groups, meaning that causal inference cannot be made because of differences between groups at baseline. We continue to be impressed by the real-world results of GLP-1 data and are hopeful that uptake and persistence of GLP-1s will continue to grow.

  • At baseline, all patients had type 2 diabetes and were about to initiate their first injectable therapy, since target A1cs were not achieved with their current treatments. On average, patients were 59 years old, 57% male, had a BMI of 34 kg/m2, and had an A1c of 8.2% – 8.3%. In the dulaglutide group, 7.6% of patients had a macrovascular condition and 17.2% had a microvascular condition. In the liraglutide group, 11.8% of patients had a macrovascular condition and 14.4% had a microvascular condition.
  • At 24 months, both liraglutide and dulaglutide significantly reduced body weight and blood glucose from baseline. The clinically meaningful reductions were sustained over the whole treatment period. While relatively few changes in treatment were observed during the study, the most common trend was dropping a DPP-4 inhibitor from the treatment regimen. Notably, less than 10% of patients initiated any insulin during the study, suggesting that the GLP-1 is delaying progression to insulin therapy. Full results are shown in the table below.

 

Dulaglutide (n=1,014)

Liraglutide (n=991)

Mean change in A1c from baseline

-1.1%

-1.1%

Mean weight in body weight from baseline

-3.5 kg

-3.3 kg

% of patients meeting individualized A1c target

46%

40%

Patients achieving >5% weight loss

35%

35%

  • While the outcomes didn’t differ significantly between the groups, dosing patterns were slightly more varied. In the dulaglutide group, 84% of patients started on the highest available dose (1.5 mg) at baseline and 16% of patients started on a lower 0.75 mg dose. At 24 months, 92% of patients were taking the 1.5 mg. In contrast, 82% of participants in the liraglutide group started on the lowest dose (0.6 mg), while 17% of patients started on 1.2 mg. At 24 months, just 6% patients remained on the 0.6 mg dose, 57% of patients were on the 1.2 mg dose, and 37% of patients had increased to the 1.8 mg dose. We are glad to see that so many patients were escalated from the 0.6 mg dose, which is not considered to be a therapeutic dose. We were reminded at ENDO 2022 that dulaglutide is a large molecule, meaning it is not absorbed into the bloodstream as quickly and therefore doesn’t require as much titration. Thus, the once-weekly dosing and lesser need for titration may make dulaglutide an easier option for patients with diabetes. That said, the outcomes are remarkably similar between the two groups, suggesting that patients should utilize whichever GLP-1 receptor agonist they are able to get their hands on.

Education and high-touch patient care take precedence over GLP-1 titration as mechanisms for combating clinical inertia

Dr. Carol Wysham (University of Washington) presented a compelling argument for overcoming therapeutic inertia via titration to higher dose GLP-1s, though significant work must be done to reach this point. Dr. Wysham framed the issue by noting that most people with type 2 diabetes are not reaching their management targets, nor are they receiving the treatment intensification they need to help them get there. The five primary promoters of therapeutic inertia, according to Dr. Wysham, are: (i) inadequate education; (ii) inadequate time/staff; (iii) unrealistic expectations; (iv) inadequate touch points; and (v) cost. However, Dr. Wysham noted that titrating within the GLP-1 class is potentially easier than adding additional medications given that titration is expected as part of treatment, it doesn’t require glucose monitoring (unlike insulin), and there is no change in cost as the dosage of GLP-1s is increased. That said, Dr. Wysham shared estimates that only about 20% of patients on dulagtluide are taking a higher dose (3.0 mg or 4.5 mg) and around 40% of people on semaglutide are taking 1.0 mg compared to 60% on 0.5 mg. Moreover, the current usage of GLP-1s is so low in the real world, with estimates as low as 1.5% in eligible patients, that getting people on GLP-1s is the priority before the focus can turn to increasing the titration. Thus, Dr. Wysham argued for increased education for patients and higher touch interactions between the patients and the healthcare system as the priority to help patients reach their treatment goals.

Phase 3 IDegLira HIGH trial: Xultophy (insulin degludec/liraglutide) confers similar A1c reduction and significantly reduces hypoglycemia episodes compared to basal-bolus regimen in people with poorly controlled type 2 diabetes

Dr. Rodolfo Galindo (Emory University) shared results from the phase 3 IDegLira HIGH trial investigating Novo Nordisk’s Xultophy (combination insulin degludec and GLP-1 liraglutide) compared to basal-bolus insulin therapy in people with type 2 diabetes and very high A1c’s between 9%-15% (n=145). Dr. Galindo explained the basal-bolus insulin therapy is considered the most effective approach for severe hyperglycemia in type 2 diabetes, and the aim of this trial was to determine the potential of using basal insulin and GLP-1 as an alternative to basal-bolus insulin. Patients in both arms were on metformin and some patients also took an SGLT-2. Over 80% of the study population identified as Black, the mean A1c was ~11% and the mean BMI was ~32 kg/m2. Over 26 weeks, Xultophy and basal-bolus therapy led to a similar A1c reduction of approximately 2%, for a final mean A1c of ~8%. Xultophy was associated with a significantly lower rate of hypoglycemia than basal-bolus therapy (~0.8% vs. ~1.8%, respectively; p=0.03), and Xultophy was not associated with weight gain, whereas basal-bolus treatment was associated with approximately 8 kg weight gain. Furthermore, as shown in the second chart below, significantly more patients on Xultophy achieved an A1c <7% with no hypoglycemia compared to patients on basal-bolus (~35% vs. ~10%, respectively; p=0.005). In line with expected side effects of GLP-1s, Xultophy was associated with a significantly higher rate of nausea than basal-bolus treatment (17% vs 1%), but there was no difference in discontinuations between the two treatments. Dr. Galindo took these results to conclude that there is no additional benefit of using complex and burdensome basal-bolus insulin therapy instead of basal + GLP-1 therapy in patients with severe hyperglycemia and highly elevated A1cs, and he called for studies of combination once-weekly insulin and GLP-1 to include patients with very high A1c’s. Dr. Galindo noted that CGM data and patient satisfaction data were collected but remain to be analyzed. While the study sample may not be large enough for this sub-analysis, we are curious if there were any differences in treatment effects depending on background SGLT-2 use, which could shed light on the potential benefits of SGLT-2 and GLP-1 combination therapy in people with severe type 2 diabetes.  

Primary Outcome – A1c change

Secondary Outcomes

Posters – GLP-1 Agonists

Title

Authors

Details + Takeaways

Correlation of Derived Time in Range (dTIR) and Time in Range (TIR) in People with Type 2 Diabetes (T2D) Treated with IDegLira (IDL), Degludec, or Liraglutide: A Post Hoc Analysis of the DUAL I Trial

Athena Philis-Tsimikas, John M. Dcruz, Ramsathish Sivarathinasami, Cristophe de Block

  • Post hoc analysis of 52-week RCT DUAL I (n=1,663, patients with CGM data n=260) of correlation between derived Time in Range (dTIR) and Time in Range (TIR) for people with T2D uncontrolled on oral antidiabetic drugs
  • dTIR and TIR strongly correlated at baseline (correlation 0.8838) and moderately correlated at weeks 26 (correlation 0.5512) and 52 (correlation 0.5184); change in dTIR and TIR were also correlated (correlation 0.7686)
  • Correlations supported using dTIR as a surrogate endpoint when CGM data is unavailable

Insulin-Sparing Effects of Oral Semaglutide: An Analysis of PIONEER 8

Morten T. Abildlund, Rikke Agesen, Stewart B. Harris, Banafsheh Zahedi, Eiichi Araki

  • Post-hoc analysis of 52-week RCT PIONEER 8 on the effect of oral semaglutide vs placebo in reducing total insulin dose for patients with T2D without adequate control on insulin
  • Greater number of patients reduced insulin dosages by over 20%, achieved A1c below 7%, and saw no body weight gain on oral semaglutide vs placebo at week 26 and week 52

Protein Biomarkers Associated with Dulaglutide and Cardiovascular Events in REWIND

Guillaume Pare, Shun Fu Lee, Helen M. Colhoun, Hui-Rong Qian, Valentina Pirro, Anastasia Hoover, Mark Lakshmanan, Giacomo Ruotolo, Kevin L. Duffin, Hertzel C. Gerstein

  • Post-hoc analysis of REWIND trial to identify protein biomarkers associated with dulaglutide assignment and MACE outcomes
  • Dulaglutide was associated with 4 biomarkers including reduced 2-year increases in hsCRP, GDF-15, and NT-proBNP and 2-year increases in C-peptide
  • DU biomarkers had smaller increases in patients at low risk for MACE

Metabolite Biomarkers and Cardiovascular Events in REWIND: A Post Hoc Analysis

Valentina Pirro, Guillaume Pare, Shun Fu Lee, Helen M. Colhoun, Yanzhu Lin, Jonathan M. Wilson, Hui-Rong Qian, Anastasia Hoover, Kevin L. Duffin, Hertzel C. Gerstein

  • Post-hoc analysis of REWIND trial to investigate relationship between metabolite biomarkers and 3-point MACE outcome
  • Dulaglutide was associated with 2-year changes in hydroxybutyric acid, threonine, acylcarnitines, and homocitrulline, and only acylcarnitines linked to MACE

Initiation of iGlarLixi vs. Basal-Bolus Insulin (BB) in Adults with Type 2 Diabetes (T2D) Advancing from Basal Insulin (BI) Therapy: The SoliComplex Real-World Study

Kevin M. Pantalone, Caroline Heller, Rosemarie Lajara, Elisheva Lew, Xuan Li, Terry A. Dex, Rachel Kilpatrick

 

  • Comparing treatment outcomes in individuals with T2D who receive basal insulin and iGlarLixi (n=1,082) versus basal-bolus insulin (n=21,208)
  • Outcome parameters included persistence, adherence, hypoglycemia, HRU, costs, A1c change
  • Once daily iGlarLixi was associated with higher treatment persistence, adherence, and lower glycemia
  • A1c reduction was slightly larger with basal-bolus insulin than iGlarLixi

Effect of Semaglutide on Liver Steatosis/Fibrosis Indexes in Patients with Diabetes and Obesity

Juana Carretero Gomez, Francisco Javier Carrasco-Sánchez, Jose Maria Fernandez, Pedro Casado, Panish Society Of Internal Medicine Diabetes, Nutrition And Obesity Working Group Of S, Jose Carlos Arevalo Lorido, Jose Pablo Miramontes González, José Miguel Segí-Ripoll, Javier Ena

 

  • Evaluating the relationship between semaglutide treatment and liver stenosis and fibrosis outcomes (n=213)
  • Semaglutide improved liver disease outcomes, with significant reductions in the hepatic stenosis index and fibrosis-4 index, in patients with T2D and obesity, mainly due to declines in weight

Real-World Weight-Loss Effectiveness of GLP-1 Agonists among Patients with Type 2 Diabetes: A Retrospective Cohort Study

Jing Luo, Ingrid Shu, Mary T. Korytkowski, David A. Rometo, Jonathan Arnold, Gretchen E. White

 

  • A cohort study determining real-world weight-loss outcomes in participants with T2D on GLP-1s (n=2,405)
  • GLP-1s induced an average weight loss of 2% at 72 weeks
  • 1/3 of participants achieved clinically significant weight loss

Targeted Maximum Likelihood Estimation to Estimate the Effect of Liraglutide on Cardiovascular (CV) Outcomes in Race/Ethnicity Subgroups: A Post Hoc Analysis of LEADER

 

David Chen, Trine J. Abrahamsen, Lauren E.E. Dang, Jack Lawson, Richard E. Pratley

 

  • Using Targeted Maximum Likelihood Estimation (TMLE) with Super Learner to assess risk reduction of MACE in race/ethnicity subgroups on liraglutide vs placebo in LEADER trial (n=9,340)
  • Liraglutide was shown to be beneficial over the placebo on MACE in underrepresented race and ethnicity subgroups
  • Suggested TMLE may be a more precise estimator than Cox modelling, due to the ability of TMLE to narrow confidence intervals with a flexible machine learning algorithm

iGlarLixi vs. Premixed Insulin Initiation in Adults with Type 2 Diabetes (T2D) Advancing from Basal Insulin (BI) Therapy: SoliComplex Real-World Study

Rosemarie Lajara, Caroline Heller, Kevin M. Pantalone, Elisheva Lew, Xuan Li, Terry A. Dex, Rachel Kilpatrick

 

  • Comparing treatment outcomes in individuals with T2D who receive basal insulin and iGlarLixi (n=834) versus premixed insulin (n=834)
  • Outcome parameters included persistence, adherence, hypoglycemia, HRU, costs, A1c change
  • Once-daily iGlarLixi was associated with higher treatment persistence, lower hypoglycemia, and similar A1c reduction without increasing health resource utilization or cost; results in subgroup aged ≥65 are comparable to overall population — suggested potential of iGlarLixi as an alternative to premixed insulin in older populations

The Real-World Observational Prospective Study of Health Outcomes with Dulaglutide and Liraglutide in Type 2 Diabetes Patients (TROPHIES): Final 24-Month Clinical Characteristics and Treatment Pattern Analyses

Bruno Guerci, Francesco Giorgino, Kristina Boye, Martin Fuechtenbusch, Elke Heitmann, Jeremie Lebrec, Anne Dib, Marco Orsini Federici, Maria Yu, Luis-Emilio Garcia-Perez

 

  • Analyzing how long adult patients with T2D stayed on their first injectable GLP-1, as well as clinical outcomes and treatment patterns (dulaglutide n=1,014; liraglutide n=991)
  • Both dulaglutide and liraglutide were clinically effective at reducing A1c and body weight over the 24-month study period
  • Few changes in treatment patterns were observed, and most happened early in treatment; most frequent change was dropping a DPP-4

The Real-World Observational Prospective Study of Health Outcomes with Dulaglutide and Liraglutide in Type 2 Diabetes Patients (TROPHIES): Final 24-Month Patient-Reported Outcomes

Martin Fuechtenbusch, Bruno Guerci, Francesco Giorgino, Luis-Emilio Garcia-Perez, Elke Heitmann, Jeremie Lebrec, Annabel Barrett, Anne Dib, Marco Orsini Federici, Kristina Boye

  • Patient-reported outcomes data from a 24-month observational study of patients with T2Ds taking either once-weekly dulaglutide (n=1,014) or once-daily liraglutide (n=991), two of the most common GLP-1s
  • Both groups demonstrated improvement from baseline characteristics after 24 months: showing clinically meaningful decrease in both A1c and weight
  • Greater improvement in productivity, treatment satisfaction, self-perception, and quality of life scores seen in dulaglutide cohort than liraglutide cohort

Baseline Characteristics of Subjects in the Once-Weekly (OW) Semaglutide FLOW Kidney Outcomes Trial

Richard E. Pratley, Florian M.M. Baeres, George Bakris, Jack Lawson, Kenneth W. Mahaffey, Johannes F. Mann, Henriette Mersebach, Peter Rossing, Katherine R. Tuttle, Vlado Perkovic

  • An ongoing phase 3 trial that will evaluate whether OW semaglutide fares better than placebo in decreasing risk of kidney and CV mortality and delaying onset of kidney damage in patients with T2D and CKD (n=3,535)
  • Majority of FLOW participants had baseline A1c > 7% and > 10-year duration of diabetes, as well as high CV and CKD progression risk

Higher Adherence and Persistence for Dulaglutide Compared with Oral Semaglutide at Six-Months Follow-Up with U.S. Real-World Data

Rosirene Paczkowski, Meredith Hoog, Jennifer Peleshok, Maria Yu, Ahong Huang, Brendan Limone, Janna Manjelievskaia

  • Retrospective study assessing adherence to and persistence with once-weekly injectable dulaglutide (n=6,166) versus oral semaglutide (n=6,166) among patients with T2D over a 6-month period
  • Found significantly greater adherence in dulaglutide cohort (64.5% of patients) than semaglutide cohort (50.3% of patients)
  • Dulaglutide initiators showed significantly greater mean days of persistence than semaglutide initiators

Short-Acting Exenatide Added Three Times Daily to Insulin Therapy Improves Insulin Sensitivity in Type 1 Diabetes

Nicklas Johansen, Thomas F. Dejgaard, Asger Lund, Tina Vilsbøll, Henrik U. Andersen, Filip K. Knop

  • Phase 2 trial evaluated impact of adding three-times-daily exenatide, a short-acting GLP-1, to insulin therapy on insulin sensitivity in patients with T1D (n=105)
  • Exenatide addition seemingly improved insulin sensitivity, as shown by a significant 0.36 mg/kg/min increase in eGDR

Identifying Modulators of Treatment Benefits of Albiglutide (GLP-1 RA) in Preventing Major Adverse Cardiovascular Events (MACE) among Individuals with Type 2 Diabetes in the HARMONY Trial Using Machine Learning

Yilu Lin, Lizheng Shi, Vivian Fonseca, Jingchuan Guo, Hui Shao

  • Utilizing machine learning to determine patient characteristics associated with albiglutide effectiveness in preventing MACE
  • Groups with the following characteristics experienced significant MACE reduction from albiglutide: baseline A1c > 8.0%, diabetes duration < 15 years, DBP > 70 mmHg, HR < 70/hour, serum creatinine < 90 umol/L, ALP < 70 U/L

SGLT-2 Inhibitors Reduce Cardiovascular and Renal Risks Compared with GLP-1 Receptor Agonists and DPP-4 Inhibitors in Patients with Diabetic Kidney Disease: A Systematic Review and Network Meta-Analysis

Hongwei Cao, Tao Liu, Li Wang, Qiuhe Ji

  • Reviewing the MEDLINE database to evaluate effectiveness of SGLT-2s in comparison to GLP-1sand DPP-4s in CV and renal outcomes for patients with DKD
  • SGLT-2s found to confer significantly greater reductions in kidney-specific composite outcome and HHF than GLP-1s, while CV death and MACE outcomes were similar in both treatments
  • Compared with DPP-4s, GLP-1 significantly decreased MACE risk

Clinical Conversations Exchange: A Cardiologist-PCP Collaboration Discussing GLP-1 Receptor Agonists for Reducing Cardiovascular Risk in Patients with Diabetes

Russie Allen, Lauren K. Welch

  • Highlighting a series of 25 educational workshops for cardiologists and primary care physicians to better understand the clinical implications of CV outcomes data from GLP-1 studies
  • Launched a Clinical Conversations Exchange Treatment Simulator tool to assist physicians in recommending optimal CV risk management plans for patients with T2D

Success of Aligned Clinician-Patient Education Improves Clinician’s Ability to Use GLP-1 RAs in Practice and Patients’ Understanding of GLP-1RAs

Amy Larkin, Anne Le

  • Assessing whether educating both patients with T2D and clinicians would increase understanding and use of GLP-1s
  • Clinicians watched a virtual symposium, while patients were provided with visual aids and video activities; both groups were tasked with pre/post knowledge questionnaires
  • Program conferred a 22% increase in patient knowledge of GLP-1s, with 78% of patients reporting increased confidence in talking to their HCP about GLP-1s
  • After program, 80% of clinicians reported confidence in starting a patient with obesity or CVD on GLP-1s

Semaglutide 2.4 mg Improved Glucose Metabolism and Reverted Prediabetes to Normoglycemia in Adults with Overweight/Obesity vs. Liraglutide 3.0 mg (STEP 8)

Julio Rosenstock, W. Timothy Garvey, Bryan Goldman, Usman Khalid, Rasmus Sørrig, Domenica Rubino

  • Comparing semaglutide 2.4 mg with liraglutide 3.0 mg for weight management in adults with overweight or obesity
  • Semaglutide 2.4 mg was superior to liraglutide 3.0 mg in A1c and mean body weight reductions
  • 25% more participants with prediabetes in the semaglutide 2.4 mg group reverted to normoglycemia than those in the liraglutide 3.0 mg group

The Potential of Semaglutide Once-Weekly in Patients without Type 2 Diabetes with Weight Regain or Insufficient Weight Loss after Bariatric Surgery

Anne Lautenbach, Marie Wernecke, Fabian D. Stoll, Sebastian M. Meyhöfer, Svenja Meyhöfer, Jens Aberel

  • Despite common occurrence of weight regain (WR) or insufficient weight loss (IWL) after bariatric surgery, treatment options are limited
  • Patients without T2D and with WR or IWL (n=48) began taking semaglutide 64.4 months after bariatric surgery; weight loss was measured after 3 and 6 months of treatment
  • 64% of patients achieved over 5% weight loss, and 6% reached over 15% weight loss; results supported the need for a larger-scale trial

Effects of Two Weight-Loss Medications, Liraglutide and Orlistat, on Liver Enzymes in Patients with Overweight or Obesity in a Real-World Setting

Juan J. Gorgojo-Martinez, Pablo L. Lois Chicharro, Pablo J. Ferreira-Ocampo, Sandra C. Doejo-Marciales, Solange F. Barra-Malig, Francisca Almodovar-Ruiz

  • A retrospective real-world study evaluating the effect of orlistat (n=400) and liraglutide (n=100) on liver enzymes (primarily ALT and GGT) in patients with overweight, obesity, or insufficient weight loss
  • Liraglutide conferred significantly greater ALT reductions than orlistat
  • Liraglutide significantly reduced GGT levels, while orlistat had no significant effect (despite no significant difference in adjusted mean GGT changes between the two).
  • Overall improvement in liver enzyme levels was seen in both GLP-1 treatment groups.

Comparative Weight Loss with GLP-1 Receptor Agonists vs Bariatric Surgery in Obesity: A Systematic Review and Pair-Wise Meta-Analysis

Shohinee Sarma, Patricia Palcu

  • Meta-analysis utilizing random effects models to pool data compared weight loss with GLP-1s and bariatric surgery
  • Bariatric surgery was found to confer the greatest weight loss, although glycemic control outcomes in patients with diabetes and obesity were similar with surgery and GLP-1s

SGLT Inhibitors

Dr. Harriette Van Spall calls for paradigm shift to view left ventricular ejection fraction as a continuum and encourages phenotypic approach to categorize heart failure; DELIVER trial will be read out at ESC 2022

Discussing the latest in heart failure therapy, the esteemed Dr. Harriette Van Spall (McMaster University, Toronto, Canada) called for a paradigm shift away from the practical but artificial categories of heart failure based on left ventricular ejection fraction (LVEF) towards viewing heart failure as a continuum. Recall that the 2022 AHA/ACC/HFSA and 2021 ESC guidelines for heart failure categorize heart failure with reduced ejection fraction (HFrEF) as LVEF ≤40%, heart failure with mildly reduced ejection fraction as LVEF between 41% to 49%, and heart failure with preserved ejection fraction (HFpEF) as LVEF ≥50%. As just one example of how ejection fraction operates on a continuum, Dr. Van Spall discussed the EF-sex interactions with RAAS inhibitors documented in a 2020 meta-analysis by Dewan et al. The analysis found that the treatment effect of RAAS inhibitors extended to greater EF values in women relative to men, highlighting the importance of an individualized approach to HF pharmacotherapy. Although heart failure treatments effective at lower EF may still be effective at higher EF, Dr. Van Spall noted that there are lower CV and HF event rates as EF increases, meaning a lower absolute risk reduction, despite similar relative risk reduction across the spectrum of LVEF values. As EF increases, the proportion of CV death to total death decreases, and other noncardiac comorbidities become more common; Dr. Van Spall therefore encouraged providers to consider increasing attention and management of other conditions, including diabetes, renal insufficiency, hypertension, anemia, COPD, and obesity. Given the limited efficacy of most therapies beyond LVEF 55-60%, she called for a personalized approach to treatment based on myocardial deformation, an imaging technique that quantifies change of regional myocardial segments in three dimensions using strain and strain rate. Echoing the 2021 ESC guidelines and the growing push for precision medicine, Dr. Van Spall advocated for a phenotypic approach to heart failure classification. In a 2021 study by Dr. Van Spall and colleagues, a machine learning approach based on simple clinical phenotypes was more effective than LVEF in stratifying HF outcomes among 1,693 patients with heart failure. Notably, the phenotype categories were able to predict all-cause death or hospitalization at six months with greater discrimination than the guideline-recommended EF categories. Overall, we were pleased to hear Dr. Van Spall advocate a more individualized approach to heart failure treatment based on phenotypic classification, with greater attention to prevention of additional comorbidities, and we look forward to continuing to learn how precision medicine may be incorporated into the management of heart failure. 

  • Looking ahead, Dr. Van Spall stated that the DELIVER trial will read out at ESC 2022 this summer. Topline results (announced in May 2022) found that SGLT-2 Farxiga met its primary cardiovascular outcome in DELIVER: in adults with heart failure with preserved ejection fraction (HFpEF) with or without type 2 diabetes, Farxiga significantly reduced the risk of composite cardiovascular death, hospitalization for heart failure (HF), or urgent HF visit, compared to placebo. We are eagerly awaiting the full readout and expect that, as with other SGLT-2s, dapagliflozin will confer strong reduction in hospitalization for heart failure across the continuum of ejection fractions. Elsewhere in heart failure trials, the FINEARTS-HF study to evaluate the safety and efficacy of finerenone on morbidity and mortality in patients with heart failure with ejection fractions ≥40% is slated to complete in May 2024.

Real-world EMPRISE study finds that empagliflozin significantly reduces hospitalization for heart failure compared to GLP-1s; no significant difference on other cardiovascular outcomes

The real-world, observational EMPRISE study compared cardiovascular outcomes of patients on SGLT-2 inhibitor empagliflozin and GLP-1 receptor agonist liraglutide, as well as empagliflozin and the broader GLP-1 class (1079-P). The analysis utilized commercial claims databases between 2014-2019 to identify patients who had initiated either empagliflozin or a GLP-1 in that period. Using 1:1 propensity score matching, 105,955 pairs of patients taking empagliflozin or a GLP-1 and 72,498 pairs of participants taking empagliflozin or liraglutide were created. In the empagliflozin/GLP-1 class arm, participants averaged age 62 and a mean A1c of 9%; 32% had a history of CVD, 11% a history of CKD, and 23% were using insulin at baseline. In the empagliflozin/liraglutide arm, participants averaged age 61 and a mean A1c of 9%; 32% had a history of CVD, 12% a history of CKD, and 30% were using insulin at baseline. The study assessed the relative effectiveness of the two therapies within each pairing on four cardiovascular outcomes: (i) hospitalization for heart failure; (ii) myocardial infarction; (iii) stroke; and (iv) all-cause mortality. Real-world evidence is incredibly important in understanding how these drugs are being used and whether benefits are conferred outside of a clinical trial setting. We are pleased to see investment in undertaking this large retrospective cohort study.

  • Empagliflozin was only superior to GLP-1s and liraglutide in heart failure; there was no significant difference in myocardial infarction, stroke, or mortality. Given the strong evidence supporting the use of SGLT-2s in heart failure, it is not surprising that empagliflozin led to a significant reduction in hospitalization for heart failure over GLP-1s. These patterns – the difference in HF and noninferiority in other outcomes – remained true within subgroups of patients who had a history of cardiovascular disease or established chronic kidney disease. There is no significant difference in all-cause mortality, suggesting that both empagliflozin and GLP-1s/liraglutide helping keep people alive. Thus, beyond HF, these results suggest that patients should be prescribed whichever therapy makes the most sense for their specific condition to optimize outcomes.

 

Empagliflozin vs GLP-1

Hazard Ratio (95% CI)

Empagliflozin vs liraglutide

Hazard Ratio (95% CI)

Overall population

 

 

HHF

0.62 (0.53-0.71)*

0.67 (0.57-0.80)*

MI

0.95 (0.85-1.07)

0.97 (0.85-1.11)

Stroke

1.09 (0.94-1.27)

1.10 (0.92-1.31)

Mortality

0.91 (0.77-1.08)

0.97 (0.79-1.17)

Patients with CVD

 

 

HHF

0.62 (0.52-0.72)*

0.66 (0.55-0.79)*

MI

0.88 (0.75-1.02)

0.84 (0.71-1.00)

Stroke

1.14 (0.93-1.39)

1.03 (0.82-1.30)

Mortality

0.95 (0.78-1.17)

0.86 (0.69-1.09)

Patients with CKD

 

 

HHF

0.62 (0.43-0.89)*

0.73 (0.48-1.13)

MI

1.09 (0.90, 1.32)

1.21 (0.97-1.50)

Stroke

1.03 (0.81-1.31)

1.20 (0.91-1.57)

Mortality

0.84 (0.62-1.14)

1.28 (0.88-1.86)

* is statistically significant

Cost is prohibitive: National claims analysis finds type 2s with CVD enrolled in high out-of-pocket cost health plans are significantly less likely to initiate an SGLT-2 or GLP-1 than those with low out-of-pocket cost plans, although very little use across the board

In an immersive session on health economics, Dr. Jing Luo (University of Pittsburgh) presented a national claims analysis showing a correlation between out-of-pocket (OOP) costs and likelihood to initiate SGLT-2/GLP-1 in type 2s with established CVD on metformin only (130-OR). The retrospective analysis used Optum claims data from 2017-2019, which included Medicare Advantage and commercial plan beneficiaries. The cohort included 72,743 people with type 2 diabetes and CVD (age 72, 56% male) with a median follow-up of 914 days (~2.5 years). The vast majority (88%) were covered by a Medicare Advantage plan, but otherwise, the population was diverse with 30% of participants having a household income <$40,000, and 13% of participants being Black. Participants were split into four OOP cost quartiles, which ranged from an average ~$20/30-day supply in the lowest quartile for both GLP-1s and SGLT-2s and an average ~$120/30-day GLP-1 supply and ~$95/30-day SGLT-2 supply in the highest quartile; in each quartile, the out-of-pocket cost for GLP-1s were higher than those for SGLT-2s. Overall, the study found incredibly low rates of SGLT-2 and GLP-1 prescription across the board but saw an association between out-of-pocket costs and SGLT-2/GLP-1 initiation.

  • Those in the highest OOP cost quartile were less likely to initiate an SGLT-2 or GLP-1 than those in the lowest OOP cost quartile. Specifically, when adjusted for demographics, clinical characteristics, insurance plan, and provider and laboratory characteristics, those in the highest OOP cost quartile were 21% less likely to initiate an SGLT-2 than those in lowest OOP cost quartile (HR=0.79) and were 12% less likely to initiate a GLP-1 than those in the lowest OOP cost quartile (HR=0.88). Furthermore, among those who did initiate an SGLT-2 or GLP-1, those with higher OOP costs took longer to initiate the cardio-protective therapy. The median time to SGLT-2 initiation was 93 days longer in the highest OOP cost and lowest OOP cost quartiles (418 days vs. 325 days, respectively). The median time to GLP-1 initiation was similarly 87 days longer in the highest OOP cost quartile compared to the lowest OOP cost quartile (395 days vs. 308 days).
  • Regardless of out-of-pocket cost, very few participants initiated an SGLT-2 or GLP-1 during the 2.5-year follow-up period, despite having both type 2 diabetes and established CVD. Only 2.8% of those in the highest OOP cost quartile and 3.8% of those in the lowest OOP cost quartile initiated an SGLT-2. Similarly, only 2.1% of those in the highest OOP cost quartile initiated a GLP-1 compared to 2.8% in the lowest OOP cost quartile. These low rates are in line with those seen in the NEJM special report published in June 2021, a JAMA study published in April 2021, and a JAMA study published in February 2022.
  • In the lowest OOP cost quartile, the median time to initiation was equivalent for SGLT-2s and GLP-1s; however, the median time to initiate the two therapies increasingly split with each increasing OOP cost quartile. This aligns with the increasing difference in the OOP cost between SGLT-2s and GLP-1s in higher OOP cost quartiles.

Dr. Tamara Isakova highlights the research to practice gap for SGLT-2s and GLP-1s

Dr. Tamara Isakova (Northwestern University) shared data on first-time prescriber rates for SGLT-2s or GLP-1s by specialty, urging nephrologists to prescribe these kidney-protective therapies. As shown in Dr. Isakova’s chart below, SGLT-2 and GLP-1 prescription rates among nephrologists were negligible from 2013 to 2019, though we should note that the first SGLT-2 to receive a CKD indication (canagliflozin) did so in September 2019. Still, she said that this trend of low SGLT-2 and GLP-1 prescriptions holds true today – see these JAMA articles on SGLT-2 and GLP-1 prescriptions for more on this. Further highlighting the large gap between research and clinical practice, Dr. Isakova noted that it takes about 17 years to turn 14% of original research into services routinely provided in community healthcare settings. To address this gap, she emphasized that it is insufficient to just educate clinicians. Addressing the research to practice gap also requires healthcare system organization, delivery system design, clinical decision support, clinical information systems, patient activation, and patient self-management support. She specifically emphasized the importance of patient activation – getting patients to ask their clinicians for guideline-directed medical therapy. Dr. Isakova concluded her talk by advocating for research on best approaches for widespread research implementation into clinical practice. She said, “Hopefully, one day we can say we are truly protecting kidney health and not just treating kidney failure.”

Lilly symposium highlights SGLT-2s and GLP-1s as key players in treatment landscape of non-insulin therapies for type 2 diabetes, reinforcing 2022 ADA Standards of Care recommendations

In a Lilly-sponsored symposium focused on improving the lives of people with type 2 diabetes through non-insulin therapies, a rockstar KOL panel offered valuable insight and practical considerations for the use of SGLT-2s and GLP-1s. Presenters included Dr. Juan Frias (Velocity Clinical Research), Dr. John Anderson (Frist Clinic, Nashville, TN), and Dr. Davida Kruger (Henry Ford Health System). In agreement with the 2022 ADA Standards of Care, speakers recommended that SGLT-2s or GLP-1s with proven cardiovascular benefit be used for patients with type 2 diabetes where ASCVD or CKD predominates. Likewise, SGLT-2s with proven heart failure benefits should be used in patients where heart failure predominates. We were pleased to hear speakers highlight the powerful cardiorenal risk reduction with these newer agents, with Dr. Kruger stating that metformin alone is “not enough.” Indeed, while metformin is a very safe medication that has traditionally been recommended by the ADA as the first-line therapy, the 2022 SOC recommend an individualized approach to first line therapy based on an individual’s particular comorbidities, barriers, and management needs that does not necessarily call for universal metformin monotherapy as first-line treatment. We were also struck by Dr. Kruger’s emphasis on the importance of cultivating patient trust: she stressed that lack of trust often leads to “attention loss” or discontinued treatment and worse outcomes and advised physicians to take time to educate each patient about side effects and health risks in advance so that patients can anticipate – rather than be shocked by – potential side effects. Likewise, touching briefly on insulin therapy, Dr. Kruger highlighted the importance of minimizing hypoglycemia, stating that too little insulin is better than too much insulin because hypoglycemic incidents so often lead to loss of interest in pharmacotherapy and distrust of HCPs. Overall, we were pleased to hear the panelists take such a patient-centered approach, emphasizing the patient-provider relationship, quality of life, and other holistic considerations.

  • Speakers highlighted the importance of proactive treatment intensification to prevent both microvascular and macrovascular events. Indeed, data from a 2017 study published in the Journal of Diabetes found that therapeutic inertia was associated with reduced time to progression of diabetic retinopathy and increased incidence of diabetic retinopathy progression. Likewise, 2015 data published in Cardiovascular Diabetology demonstrated that a one-year delay in treatment intensification was associated with 67% higher risk of myocardial infarction, 51% higher risk of stroke, and 64% higher risk of heart failure, or a composite 62% increase in the risk of all three macrovascular outcomes (n=30,471). Recall that early intensification of therapy has also been linked with reduced risk of treatment failure, as demonstrated by the VERIFY trial. Overall, we’re glad to see the speakers draw attention to the perils of clinical inertia.
  • Dr. Anderson reminded audience members that their work in type 2 diabetes is increasingly important, with particular emphasis on geographic disparities. Most strikingly, the Southeastern U.S. experiences such high rates of type 2 diabetes that it has been termed the “diabetes belt.” According to 2018 data from the CDC’s United State Diabetes Surveillance System, nearly 61 million American adults will have diabetes by 2060. Certainly  many of these individuals will experience comorbidities such as heart and cardiovascular disease, making early, proactive treatment intensification all the more important.

Posters – SGLT Inhibitors

Title

Authors

Details + Takeaways

Prescribing Patterns of Sodium Glucose Cotransporter 2 Inhibitors in Patients with Heart Failure with or without Diabetes

Sarah R. Bermudez, Joe R. Anderson, Gretchen Ray

  • Determining SGLT-2 prescription rates among patients with HFrEF and HFpEF, with or without T2D (n=1,101)
  • SGLT-2s were prescribed in 6.5% of cases: 14.7% of patients with T2D, 0.4% without T2D; overall, SGLT-2s were underutilized in patients with HF regardless of diabetes status

Individualized Cost-Effectiveness Assessment of Sodium-Glucose Cotransporter 2 Inhibitors (SGLT2i) vs Sulfonylureas as Add-On Therapy in People with Inadequately Controlled Type 2 Diabetes (T2D) Under Metformin Monotherapy

Dawei Guan, Shu Niu, Vivian Fonseca, Lizheng Shi, Neda Laiteerapong, Jingchuan Guo, Hui Shao

 

  • Running a cost-effective analysis, using the BRAVO diabetes model and an ICER threshold of $100,000/QALY, to assess the value of SGLT-2s vs sulfonylureas (SU) for patients with T2D on metformin
  • SGLT-2 use conferred an ICER of $94,000/QALY for the overall population, and $64,000/QALY for those with a history of HF, indicating greatest cost effectiveness in individuals with CVD

The Impact of a Passive Alert on Screening for Euglycemic Diabetic Ketoacidosis in Hospitalized Patients Using a Sodium-Glucose Cotransporter 2 Inhibitor

Diana Soliman, R. Clayton Musser, Jason P. Jackson, Jashalynn German, Scott Carlson, Kathryn H. Odonnell, Tracy Setji

  • Testing a passive EHR alert system which notified providers when hospitalized patients taking SGLT-2s were at high risk for euglycemic diabetic ketoacidosis (euDKA)
  • Just under 20% (18%) of patients who triggered alert screened positive for euDKA, demonstrating success in identifying high risk patients and potential in preventing delayed diagnoses

Heart Failure Hospitalization among African American Compared with White Individuals with Type 2 Diabetes on Empagliflozin – Real World Data

Basem M. Mishriky, Doyle M. Cummings, Yuanyuan Fu, Jacquie Halladay, William S. Jones, Andrea Boan, Sara Jones Berkeley, Shivajirao P. Patil, James R. Powell, Alyssa Adams, William Irish

 

  • This retrospective study evaluated the risk of HF hospitalizations by race in individuals with T2D taking empagliflozin (n=8,994)
  • When adjusting for age, gender, and comorbidities, there was no significant difference in the risk of HF hospitalizations between African American and White individuals taking empagliflozin

PromarkerD Predicts Late-Stage Renal Function Decline in Type 2 Diabetes in the Canagliflozin Cardiovascular Assessment Study (CANVAS)

Kirsten E. Peters, Patsy Di Prinzio, Scott Bringans, Wendy A. Davis, Timothy Davis, Richard J. Lipscombe, Michael K. Hansen

 

  • PromarkerD was tested as a late-stage renal decline predictor in a controlled trial of canagliflozin vs. placebo in patients with T2D (n=3,525)
  • PromarkerD is a biomarker-based test that provided scores of low, moderate, or high risk for adverse renal outcomes; PromarkerD was a significant predictor of late-stage renal decline

Cardiovascular (CV) and Kidney Outcomes with Canagliflozin (CANA) According to Type 2 Diabetes (T2D) Treatment Targets at Baseline (BL): Data from the CANVAS Program and CREDENCE

Vincent C. Woo, Michael Tsoukas, Sheldon Tobe, April Slee, Wally Rapattoni, Fernando Ang, Jochen Seufert, David C. Wheeler

 

  • Two separate double-blind, placebo controlled, randomized trials were run to determine the efficacy of canagliflozin at reducing the risk of CV and kidney damage in patients with T2D (CANVAS trial, n=10142; CREDENCE trial, n=4401)
  • Canagliflozin consistently reduced the risk of kidney and cardiovascular death versus the placebo, regardless of other T2D treatment targets

Predictors of HbA1c and Weight Response to Sodium-Glucose Cotransporter 2 Inhibitors (SGLT2i) in the Association of British Clinical Diabetologist (ABCD) UK Nationwide Audit

Thomas S.J. Crabtree, Alison Gallagher, Siva Sivappriyan, Ketan Dhatariya, Rajeev P. Raghavan, Alex Bickerton, Jackie Elliott, Melissa L. Cull, Gerry Rayman, Ian W. Gallen, Iskandar R. Idris, Robert E. Ryder

 

  • Using a multivariate linear regression to identify predictors of A1c and weight response to SGLT-2s
  • Participant’s overall A1c fell by 0.9%, falling most in populations that were older, had greater weight loss, and shorter diabetes duration
  • Weight fell by 3.0 kg, with greatest declines in populations that were older, had lower baseline A1c, lower baseline weight, or took empagliflozin (vs. dapagliflozin)

A Territory-Wide Study of SGLT2 Inhibitor-Associated Postoperative Diabetic Ketoacidosis in Type 2 Diabetes Patients

David T.W. Lui, Tingting Wu, Xiaodong Liu, Chi Ho Au, Chi Ho Lee, Man Him Matrix Fung, Yu Cho Woo, Kathryn C. Tan, Carlos K. Wong

 

  • Determining risk of postoperative DKA in patients on SGLT-2 agonists who underwent operations between May 2015 and December 2020
  • SGLT-2 use was associated with 6x the risk of postoperative DKA
  • Emergency operation, suboptimal preoperative glycemic control, and insulin use predicted DKA risk

Differences in Heart Failure Hospitalization Risk following Initiating SGLT2 Inhibitors vs DPP4 Inhibitors across Race/Ethnicity and Rural/Urban Areas

Jingchuan Guo, Yujia Li, Jiang Bian, Daniel T. Lackland, Stephen Kimmel, Desmond Schatz, Almut G. Winterstein

 

  • Analyzing racial and rural differences in effectiveness of SGLT-2s
  • SGLT-2s were equally effective across racial lines, though showed decreased effectiveness in rural areas

Interleukin-6 and Cardiovascular Outcomes in Patients with Type 2 Diabetes: A Post Hoc Analysis of CANVAS Trial

Akihiko Koshino, Meir Schechter, Taha Sen, Brendon Neuen, Clare Arnott, Vlado Perkovic, Michael K. Hansen, Hiddo L. Heerspink

  • Investigating the association between interleukin-6 (IL6) and CV outcomes and determining the impact of canagliflozin on IL6 levels (n=3,503)
  • In patients with T2D, increased IL6 was associated with higher risk of CV outcomes
  • Canagliflozin reduced IL6 levels over one year of treatment

Long-Term Ertugliflozin Treatment and Incidence of Hypoglycemia: Analyses from VERTIS CV

Samuel Dagogo-Jack, Christopher P. Cannon, David Cherney, Francesco Cosentino, Darren K. McGuire, Jie Liu, Chih-Chin Liu, Robert Frederich, James P. Mancuso, Richard E. Pratley

  • Determining incidence of hypoglycemia in people with T2D taking SGLT-2 ertugliflozin versus a placebo (n=8,238)
  • Ertugliflozin was not found to increase rates of hypoglycemia
  • Ertugliflozin was successful as a treatment for individuals with hypoglycemia

Efficacy of Dapagliflozin by Baseline Diabetes Medications: A Prespecified Analysis from the DAPA-CKD study

Frederik Persson, Peter Rossing, Niels Jongs, Glenn M. Chertow, Fan Fan Hou, Priya Vart, John J. Mcmurray, Ricardo Correa-Rotter, Bergur Stefansson, Robert D. Toto, Anna Maria Langkilde, David C. Wheeler, Hiddo L. Heerspink, Dapa-Ckd Trial Committees And Investigators

  • A randomized, double-blind, placebo-controlled clinical trial evaluated effects of dapagliflozin on kidney, CV, and mortality outcomes in adults with and without T2D (n=4,304)
  • Dapagliflozin reduced kidney and cardiovascular events in individuals with CKD and T2D, regardless of other treatment options in play

Favorable Kidney Outcomes are Associated with Empagliflozin vs DPP4i in Patients with Diabetes and Normal Kidney Function Real-World Evidence

Meir Schechter, Cheli Melzer Cohen, Aliza Rozenberg, Ilan Yanuv, Gabriel Chodick, Ofri Mosenzon, Avraham Karasik

 

  • Assessing differences in kidney outcomes between patients with T2D using SGLT-2s vs DPP-4s (n=6,477)
  • SGLT-2s were associated with lower risk of adverse kidney outcomes than DPP-4

Reductions in Traditional Risk Factors Explain most of the Cardiovascular Benefit of SGLT2 Inhibitors and GLP-1 Receptor Agonists: An Analysis Using the BRAVO Risk Engine

Shu Niu, Dawei Guan, Naykky Singh Ospina, Kenneth Cusi, Vivian Fonseca, Lizheng Shi, Hui Shao

 

  • Analyzing eight CVOTs to determine impacts of SGLT-2s and GLP-1s on CV outcomes in patients with T2D across eight different CVOTs
  • SGLT-2 use was positively correlated with decreased hospitalization for HF (RR: 0.74), though there was also a potential stroke risk increase (RR: 1.29)
  • GLP-1s showed no additional benefit in preventing outcomes

Newer Glucose-Lowering Drugs and Risk of Dementia: A Meta-analysis of Cardiovascular Outcome Trials

Huilin Tang, Shu Niu, Joshua D. Brown, Jingkai Wei, Almut G. Winterstein, Jiang Bian, Jingchuan Guo

 

  • An analysis was run on existing observational studies to determine the association between glucose-lowering drugs and dementia risk (n=21)
  • Studied medications did not increase dementia risk
  • SGLT-2s significantly lowered risk of vascular dementia (OR: 0.11)  

Machine-Learning Models to Predict CKD and HF in Type 2 Diabetes Patients

Hiroo Tsubota, Atsushi Suzuki, Masaki Makino, Eiichiro Kanda, Toshitaka Yajima, Yoko Kidani, Naru Morita, Satomi Kanemata

 

  • Development of a novel machine learning model to predict the risk of CKD and HF among patients with T2D using real world databases
  • Model successfully identified five-year risk for patients without a history of CKD or CVD

Maximize Benefits of SGLT2 Inhibitors in People with Advanced CKD Stages with 2 Diabetes

Koji Kashima, Hiroyuki Shimizu, Masanobu Yamada

 

  • Identifying differences between patients with T2D and advanced CKD (n=75) who responded (n=60) or did not respond (n=15) to SGLT-2s
  • No major background differences were found between responders and non-responders
  • Called for further studies, especially in individuals with lower eGFR levels below 30 and more advanced CKD courses

Dapagliflozin Cost Offsets in Chronic Kidney Disease for Commercial U.S. Payers

Raymond Chang, Joanna Huang, Dan Reck, Jamie Hurst, Kat Wolf Khachatourian, Michael H. Shannon

 

  • Using a cost-offset model to quantify predicted clinical and economic benefits of dapagliflozin vs standard of care (SOC) alone in patients with CKD (n=130)
  • Dapagliflozin group saw 20 fewer HF events than SOC group
  • Study suggested commercial insurers can substantially reduce costs associated with CKD morbidity and mortality by prescribing dapagliflozin, given the 37% risk reduction of dapagliflozin compared to SOC
  • Reduced CV and renal clinical events with dapagliflozin use over a three-year time horizon, offset health resource utilization costs by $3.9 million

Patterns of Drug Utilization for Cardio- and Renal-Protective Antihyperglycemic Agents in a Commercially Insured Population

David J. Cook, Anthony Defail, Elizabeth Macfarlane, Derek Pederson

 

  • Evaluating utilization of cardioprotective and renal-protective SGLT-2s and GLP-1s in commercially insured patients with T2D (n=7,361)
  • Medications were used in approximately 25% of T2D cases with CV and renal indications
  • 21.6% of patients on GLP-1s and 19.9% of patients on SGLT-2s did not have CV or renal risk indications

Effectiveness and Safety of Empagliflozin in Routine Care: Results from the Empagliflozin Comparative Effectiveness and Safety (EMPRISE) Study

Phyo T. Htoo, Helen Tesfaye, Julie M. Paik, Deborah J. Wexler, Mehdi Najafzadeh, Robert Glynn, Anouk Deruaz-Luyet, Soulmaz F. Fazeli Farsani, Lisette Koeneman, Sebastian Schneeweiss, Elisabetta Patorno

 

  • Comparing safety and CV effectiveness of empagliflozin (n=115,116) vs DPP-4s (n=115,116) in patients with T2D and CVD using massive real world data set over five years (2014 – 2019)
  • Empagliflozin was associated with a 50% risk reduction in HHF and 12% risk reduction in myocardial infarction compared to DPP-4
  • In terms of safety, empagliflozin reduced acute kidney injury compared to DPP-4, but increased risk of hospitalization from diabetic ketoacidosis. Risk was similar between groups for lower limb amputations, non-vertebral fracture, and renal and bladder cancer

Cardiovascular Effectiveness of SGLT2 Inhibitors: Head-to-Head Comparisons

Devin Abrahami, Elvira D'andrea, Julie M. Paik, Deborah J. Wexler, Brendan M. Everett, Seoyoung C. Kim, Elisabetta Patorno

 

  •  A cohort study evaluating CV risk between SGLT-2s (canagliflozin, dapagliflozin, and empagliflozin)
  • Once weighed, there were no significant differences between risk of MI, stroke, or death across SGLT-2s
  • Empagliflozin was associated with lower risk of HHF than dapagliflozin (HR: 0.73)

Individual SGLT2 Inhibitors and the Risk of Diabetic Ketoacidosis

Devin Abrahami, Elvira D'andrea, Seoyoung C. Kim, Deborah J. Wexler, Julie M. Paik, Elisabetta Patorno

 

  • Authors conducted a cohort study to determine diabetic ketoacidosis risk between SGLT-2s (canagliflozin, dapagliflozin, and empagliflozin), GLP-1, and DPP-4 in people with T2D
  • Use of dapagliflozin was associated with reduction in risk of diabetic ketoacidosis compared to empagliflozin
  • All SGLT-2s were associated with an increased risk of diabetic ketoacidosis compared to GLP-1 and DPP-4

Prevalence of SGLT2i and GLP-1RA Use among U.S. Adults with Type 2 Diabetes

Christine Limonte, Yoshio Hall, Subbulaxmi Trikudanathan, Katherine R. Tuttle, Irl B. Hirsch, Ian De Boer, Leila Zelnick

 

  • Study analyzed NHANES population data to determine popularity of SGLT-2 and GLP-1 use among US adults with T2D 2017-2020 (n=1,375). There was no difference in drug prevalence between adults with CKD, CHF, or ASCVD versus without
  • SGLT-2 use differed by age and specialist access, while GLP-1 use differed by age, race and ethnicity, insurance status, BMI, and specialist access

Real-World Patient Perceptions of GLP-1RAs and SGLT2is with Cardiorenal Benefits

Natalie Lambert, Callahan Clark, Rozalina G. Mccoy, Alyssa Wong, David J. Cook

 

  • An analysis of posts in public T2D Facebook groups (n=250 posts, n=15 groups) from January 2019 to November 2021 to determine perceptions of GLP-1sRA and SGLT-2s
  • 44% of posts described risks (n=41) or benefits (n=6) not defined by the ADA Standards of Care
  • 63% of posts asked an information-seeking question
  • Posts mostly focused on near-term effects (quality of life, glucose control, side effects, emotion)

SGLT2 Inhibitors Reduce Cardiovascular and Renal Risks Compared with GLP-1 Receptor Agonists and DPP4 Inhibitors in Patients with Diabetic Kidney Disease: A Systematic Review and Network Meta-Analysis

Hongwei Cao, Tao Liu, Li Wang, Qiuhe Ji

 

  • Systematic review and meta-analysis comparing the effects of SGLT2-s, GLP-1s, and DPP4-s on CV and renal outcomes in patients with DKD (n=46,292)
  • SGLT-2s reduced CV and renal risk compared to the other drugs
  • DPP-4s did not reduce CV and renal risks compared to placebo
  • GLP-1s did better than placebo but worse than SGLT-2s
  • No significant difference in CV deaths between the three different treatment groups

More Pronounced Effect of Empagliflozin-Losartan Combination Therapy on Measured GFR and Blood Pressure vs Either of the Drugs: A Crossover RCT in People with Type 2 Diabetes

Daniël Van Raalte, Hiddo L. Heerspink, Rosalie Scholtes, Amsterdam, Netherlands, Groningen, Netherlands

  • One-week cross-over study comparing SGLT-2 empagliflozin, RAS inhibitor losartan, combination empagliflozin+losartan, and placebo on blood pressure (BP), heart rate (HR), and GFR outcomes (n=24 T2Ds)
  • Patients on metformin and/or sulfonylurea and had an average age of 66 ±6 years, A1c 7.4% ±0.9%, and eGFR 108 ±20 mL/min/1.73m2
  • Compared to placebo, empagliflozin reduced GFR by 7.0 mL/min, losartan by 7.3 mL/min, and empagliflozin+losartan by 10.7 mL/min
  • Empagliflozin lowered BP by 8.7 mmHg, losartan by 12.4 mmHg, and empagliflozin+losartan by 15.1 mmHg (all p<0.05)

DINAMO-Diabetes Study of Linagliptin and Empagliflozin in Children and Adolescents with Type 2 Diabetes (T2D): Innovative Study Design and Baseline Characteristics

Lori M. Laffel, Thomas Danne, William V. Tamborlane, Georgeanna J. Klingensmith, Christy Schroeder, Dietmar Neubacher, Nima Soleymanlou, Jan Marquard, Philip Zeitler, Steven M. Willi

  • RCT comparing change in A1c in youth with T2D over one year with linagliptin, empagliflozin, and placebo (n=158)
  • Youth with mean age 15, all on metformin and/or insulin, with baseline A1c 8%, BMI 36
  • 61% were female; fairly diverse cohort, with 31% Black or African American and 50% White participants
  • Study ongoing, with expected completion in Fall 2022

Cardiovascular Effectiveness of SGLT-2 Inhibitors and GLP-1 Receptor Agonists in Routine Care of Frail People with Type 2 Diabetes

Alexander Kutz, Chandrasekar Gopalakrishnan, Dae H. Kim, Elisabetta Patorno

  • Compared the effects of SGLT-2s, GLP-1s, and DPP-4s on MACE between not frail, pre-frail, and frail adults with T2D
  • The HR for MACE associated with SGLT-2 vs. DPP-4 was 0.74 in non-frail, 0.72 in pre-frail, and 0.79 in frail people (n=120,202 propensity-score matched pairs, mean age 72)
  • The HR for MACE associated with GLP-1 vs. DPP-4 was 0.75 in non-frail, 0.73 in pre-frail, and 0.83 in frail people (n=113,864 pairs, mean age 72)
  • The HR for MACE associated with SGLT-2 vs. GLP-1 was 0.96 in non-frail, 0.92 in pre-frail, and 0.89 in frail people (n=89,865 pairs, mean age 72)
  • Over mean follow-up of ~10 months, there were similar relative risk reductions in MACE for SGLT-2s and GLP-1s in people with and without frailty, but larger absolute reductions in frail people

SGLT2 Inhibitors Reduce Cardiovascular and Renal Risks Compared with GLP-1 Receptor Agonists and DPP4 Inhibitors in Patients with Diabetic Kidney Disease: A Systematic Review and Network Meta-Analysis

Hongwei Cao, Tao Liu, Li Wang, Qiuhe Ji

 

  • Systematic review and meta-analysis comparing the effects of SGLT2-s, GLP-1s, and DPP4-s on CV and renal outcomes in patients with DKD
  • SGLT-2s reduced CV and renal risk compared to the other drugs
  • DPP-4s did not reduce CV and renal risks compared to placebo
  • GLP-1s did better than placebo but worse than SGLT-2s
  • No significant difference in CV deaths between the three different treatment groups

Novel Therapies

SURMOUNT-1 full results show tirzepatide confers up to 23% weight loss on average in people with obesity; majority of participants on tirzepatide 10 mg or 15 mg achieved 20% weight loss; 95% of participants with prediabetes reverted to normoglycemia

In a room filled to the brim, the full SURMOUNT-1 results of Lilly’s dual GIP/GLP-1 agonist tirzepatide in obesity were presented to rapturous applause. Full results showing tirzepatide’s groundbreaking weight loss were simultaneously published in the New England Journal of Medicine, along with an editorial titled “Shifting Tides Offer New Hope For Obesity” by Tufts University’s Dr. Clifford Rosen and Dr. Julie Ingelfinger. Also, see Lilly’s press announcement. While tirzepatide is not yet approved for obesity, it received FDA approval for type 2 diabetes in May 2022. These full results follow the release of topline results in April 2022.

SURMOUNT-1 was a double-blind and placebo-controlled study evaluating once-weekly tirzepatide (5 mg, 10 mg, and 15 mg) compared to placebo in a cohort of patients with obesity without diabetes (n=2,539) over 72 weeks. At baseline, 41% of the patient population had prediabetes, and those with prediabetes are currently undergoing an additional two-year treatment period. Today’s presentation only includes findings from the initial 72-week treatment period. Full results showed that SURMOUNT-1’s coprimary endpoints were met: on tirzepatide 5 mg, 10 mg, and 15 mg doses, (i) participants experienced significant 16%, 21%, and 23% weight loss, respectively, compared to 2% weight loss with placebo, and (ii) 89%, 96%, and 96% of participants achieved ³5% weight loss, respectively, compared to 28% of participants on placebo.

Notably, tirzepatide now becomes the first drug to show greater than 20% weight loss in a phase 3 trial for obesity, which led today’s discussant, Dr. Lee Kaplan (Massachusetts General Hospital), to conclude, “We are in a new era of treating obesity. That is one of the most important conclusions of what we have seen today.” To summarize the remarkable effect of tirzepatide, Dr. Ania Jastreboff (Yale) shared a quote from one of her trial patients: “It’s just as easy to lose weight as it ever was to gain weight.” See below for full details of the study. 

  • Study design. SURMOUNT-1 randomized 2,539 patients from nine countries to four treatment groups: (i) tirzepatide 5 mg (n=630); (ii) tirzepatide 10 mg (n=636); (iii) tirzepatide 15 mg (n=630); and (iv) placebo (n=643). SURMOUNT-1 included adults with either a BMI ³30 kg/m2 or a BMI³27 kg/m2 with previously diagnosed hypertension, dyslipidemia, obstructive sleep apnea, or cardiovascular disease. People taking weight loss medications, with type 1 or type 2 diabetes, or with a prior or planned surgical treatment for obesity were excluded from the study. Looking to discontinuation rates, 73.6% of participants in the placebo group completed the study period on the study drug, while completion rates on the study drug within the tirzepatide treatment groups ranged from 83.6% to 85.7%.  Discontinuation rates due to adverse events in the tirzepatide treatment groups ranged from 4.3% to 7.1%, compared with 2.6% of the placebo group.
  • Baseline characteristics. Across treatment groups, baseline characteristics remained relatively consistent. The mean age of participants was 45 years, the mean body weight was 105 kg, and the mean BMI was 38 kg/m2. Almost 95% of participants had a BMI ³30 kg/m2, and 60% of participants had Class 2 obesity (BMI 35-39 kg/m2) or Class 3 obesity (BMI ³ 40 kg/m2), which Dr. Sean Wharton (Wharton Medical Clinic, Toronto, Canada) said was representative of patients in the clinic. Similar to other obesity trials, most participants identified as female (67%) and identified as White (70%), while only 11% identified as Asian, 9% as Native American, 8% as Black. On a positive note, 48% identified as Hispanic or Latino, which is more representative than other trials. Dr. Wharton also presented an overview of participants from the United States, which comprised 44% of the study population, noting that US participants were representative of the broader US population. Specifically, 80% of US participants identified as White, 14% identified as Black, 2.5% identified as Asian, and 27% identified as Hispanic or Latino. Baseline clinical characteristics included normal systolic and diastolic blood pressure, a waist circumference of 114 cm, and 41% prevalence of prediabetes. Furthermore, two thirds of participants had one or more comorbidity, most commonly hypertension (32%).
  • Weight loss outcome. As shown in the first figure below, among people who stayed on therapy, tirzepatide 5 mg led to 16% weight loss, tirzepatide 10 mg led to 21.4% weight loss, and tirzepatide 15 mg led to 22.5% weight loss, compared to 2.4% weight loss on placebo (efficacy estimand, or on treatment values). In terms of absolute weight change, from a baseline of 231 lbs., participants on tirzepatide 5 mg, 10 mg, and 15 mg lost 35 lbs., 49 lbs., and 52 lbs., respectively. Dr. Jastreboff highlighted the early treatment response after just four weeks in all three tirzepatide groups. She also highlighted the similar weight reduction for tirzepatide’s 10 mg and 15 mg doses. As shown in the second figure below, the majority of participants on tirzepatide 10 mg or 15 mg achieved ³20% weight loss, and remarkably over a third of participants on tirzepatide 10 mg or 15 mg achieved ³25% weight loss. As with all obesity therapies, there was heterogeneity in patients’ weight loss to tirzepatide, as depicted in the third figure below. Still, nearly all participants on tirzepatide 15 mg (98%) lost weight, whereas 67% of participants on placebo lost weight.
    • Two methods of analyzing efficacy: What is the treatment-regimen estimand and the efficacy estimand? The treatment-regimen estimand is the treatment difference regardless of adherence to treatment (i.e., in trial efficacy) and indicates tirzepatide’s efficacy across all the patients who received it. The efficacy estimand is the treatment difference if all patients remained on treatment for the entire 72 weeks (i.e., on treatment efficacy) and indicates tirzepatide’s efficacy in patients who took tirzepatide as prescribed. Dr. Carel de Roux (University College Dublin, Ireland) explained that while the treatment-regimen estimand is what regulators want to see, as a clinician he is more concerned with the efficacy estimand to understand what patients will experience if they consistently take tirzepatide as prescribed. The efficacy and treatment-regimen estimands were similar, indicating “a well-conducted trial with a very acceptable retention rate,” according to Dr. Nadia Ahmad (Lilly Senior Medical Director, tirzepatide Obesity Program).

  • Cardiometabolic outcomes. All prespecified cardiometabolic measures improved with tirzepatide. 40% of participants had prediabetes, and astoundingly, more than 95% of those with prediabetes reverted to normoglycemia during the primary trial period – more detailed analysis of tirzepatide’s effect on prediabetes will be available after the two-year extension phase. Even though SURMOUNT-1 did not enroll people with diabetes, participants on tirzepatide saw an A1c reduction of -0.4% to -0.5% from a baseline of 5.6%. Furthermore, compared to placebo, tirzepatide 15 mg led to almost a five times greater reduction in waist circumference. Additionally, tirzepatide (three doses pooled) resulted in a three times greater reduction in fat mass than in lean mass. Pooling all tirzepatide doses, fasting insulin decreased by nearly 47% with tirzepatide, compared to a 10% decrease with placebo. Triglycerides, liver enzymes, systolic blood pressure, and diastolic blood pressure also improved significantly with tirzepatide. Dr. Jastreboff noted that the improvement in blood pressure was not dependent on the high magnitude of weight reduction, as blood pressure leveled off as body weight continued to drop.
  • Safety results. Overall, the safety and tolerability of tirzepatide was consistent with GLP-1s approved for obesity (liraglutide and semaglutide), characterized by an increased rate of nausea, diarrhea, and constipation. While 26% of participants on placebo discontinued treatment due primarily to lack of treatment efficacy, 14%, 16%, and 15% of participants on tirzepatide 5 mg, 10 mg, and 15 mg discontinued treatment, respectively. Dr. Sriram Machineni (University of North Carolina Chapel Hill) reported that the primary reason for tirzepatide discontinuation was gastrointestinal (GI) adverse events, though he noted drug discontinuation due to GI events was not frequent and was reported in <5% of participants on tirzepatide (see chart below). He emphasized that GI events were transient, mostly mild to moderate, and occurred most frequently during the dose escalation period (see figure below). About a quarter of those on tirzepatide 5 mg and a third of individuals on tirzepatide 10 mg and 15 mg reported nausea. Notably, there was not an increase in adverse events between the 10 mg and 15 mg doses. Other adverse events associated with tirzepatide included transient hair loss (expected with weight loss therapy), dizziness (largely due to hydration), injection site reactions, and decreased appetite, which Dr. Machineni considered an on-target, positive effect as opposed to an adverse event. Tirzepatide was also associated with a slight rise in heart rate of 0.6, 2.3, and 2.6 beats per minute for the 5 mg, 10 mg, and 15 mg doses, respectively, and a slight increased frequency of gall bladder disease, consistent with other GLP-1s for obesity.
    • During a press conference today, Dr. Jastreboff elaborated on when drug discontinuations most often occurred and how to mitigate GI side effects. To mitigate adverse events and facilitate sustained, consistent use of tirzepatide (or GLP-1s), Dr. Jastreboff recommended that patients experiencing GI side effects talk with their clinician about handling those side effects, perhaps by reducing treatment dose. Dr. Jasteboff also recommended that patients do not eat past fullness, be prepared to eat smaller amounts but more frequently, and identify foods that may exacerbate GI side effects. She added those strategies can be challenging since the social pressure to eat is immense, so she emphasized the importance of open conversations between clinicians and patients to best prepare patients for taking tirzepatide or incretin-based therapies. 

Discontinuations due to adverse events

 

Placebo (n=643)

Tirzepatide 5 mg (n=630)

Tirzepatide 10 mg (n=636)

Tirzepatide 15 mg (n=630)

% of drug discontinuations due to adverse events

3%

4%

7%

6%

% of drug discontinuations due to gastrointestinal adverse events

0.5%

2%

4%

4%

  • Discussion. Discussing the implications of these results, Dr. Louis Aronne (Weill Cornell Medicine) said that these results, along with the >12% weight loss associated with semaglutide 2.4 mg, are paving the way for a new obesity treatment paradigm. He explained that the old treatment paradigm was to ignore obesity and focus on treating obesity’s complications since lifestyle interventions and obesity pharmacotherapy were not sufficiently effective to impact obesity complications. However, with the advent of therapies yielding >10% weight loss (semaglutide 2.4 mg and tirzepatide), or what Dr. Kaplan called “3rd generation anti-obesity medications,” clinicians can now treat obesity first and then treat other cardiometabolic factors if they require additional management. In other words, SURMOUNT-1’s results support the shift of obesity treatment upstream so that obesity treatment entails weight loss and complication prevention, as opposed to prioritizing piecemeal treatment of multiple complications. As Dr. Jastreboff discussed at the press conference, SURMOUNT-1 indicates that patients should be treated for obesity earlier, and perhaps treating obesity head on can prevent type 2 diabetes. Looking ahead, Dr. Aronne said we need outcome studies for tirzepatide, widespread insurance coverage for obesity medications, scalable systems to enable access to comprehensive medical treatment, and the further development of new compounds to tackle obesity through multiple mechanisms.

    • Highlighting that even for tirzepatide there is large variability in response, Dr. Kaplan said, “We will not solve everyone’s obesity by treating everyone with tirzepatide.” As a result, he suggested seven approaches of prioritizing who should receive tirzepatide or other obesity medications: (i) by degree of obesity or amount of excess body fat; (ii) by overall clinical severity of obesity, not just amount of adipose tissue; (iii) by overall cost of care and anticipated cost savings with effective therapy; (iv) by specific types of obesity; (v) prioritizing obesity that interferes with effective care of another disease; (vi) prioritizing patients who have received previous obesity treatment, such as bariatric surgery; and (vii) prioritizing patients highly response to obesity pharmacotherapy by prediction or empiric evidence.
    • The reflections from Dr. Aronne and Dr. Kaplan, along with the NEJM editorial, raise important questions on tirzepatide and obesity treatment more broadly: Does tirzepatide reduce cardiovascular complications? How will GI effects change over time and in the real world? Could tirzepatide be used in intervals? What is tirzepatide’s mechanism of action, and to what extent does GIP agonism on top of GLP-1 agonism contribute to tirzepatide’s remarkable weight loss? Since there is variability in people’s response, even for tirzepatide, which patients will respond best to tirzepatide, and who may need other or additional weight loss therapies/interventions? What other mechanisms of weight regulation can we exploit to potentially achieve even more weight loss through pharmacotherapy? Looking ahead, we are excited to see additional color on tirzepatide’s benefits from the remaining three SURMOUNT trials, which are expected to reach primary completion in mid-2023.

Dr. Juan Pablo Frias (Velocity Clinical Research) presented our first-ever Lilly-sponsored symposium on Mounjaro (tirzepatide) to a standing room only crowd – the excitement for the dual GIP/GLP-1 agonist is palpable. Dr. Frias’s presentation was dedicated primarily to the use and safety of Mounjaro, backed by evidence from the SURPASS clinical trials program. Mounjaro was approved by the FDA in May and is currently indicated as an adjunct to diet and exercise to improve glycemic control in adults with type 2 diabetes. Following the impressive results from SURMOUNT-1, also presented this morning at ADA, we feel as though the field may be on the edge of a new dawning of diabetes management. We simply cannot wait for the launch of Mounjaro later this month and to see this incredible new therapy in the hands of patients with diabetes – and, in future, people who may identify as people with pre-diabetes, or even as people with normoglycemia.

  • Mounjaro is a first-in-class GIP/GLP-1 dual agonist. The novel drug operates by selectively binding to and activating both incretin receptors, which are the targets for native GIP and GLP-1 hormones. GIP is responsible for two-thirds of the incretin effect in healthy people compared to one-third for GLP-1 suggesting that GIP generates a more significant impact on insulin secretion than GLP-1. Ultimately, the dual agonist increases insulin sensitivity, enhances insulin secretion, decreases food intake, slows gastric emptying, and decreases glucagon levels.
  • Notably, Dr. Frias shared data from SURPASS-2 – the head-to-head comparison trial with semaglutide, manufactured by Novo Nordisk, a major competitor of Lilly. As a reminder, SURPASS-2 was a 40-week, phase 3 trial that randomly assigned 1,879 patients, in a 1:1:1:1 ratio, to receive tirzepatide at a dose of 5 mg, 10 mg, or 15 mg or semaglutide at a dose of 1 mg. Tirzepatide conferred statistically significant A1c and body weight reductions from baseline compared to semaglutide in all three treatment arms (5 mg, 10 mg, and 15 mg) by efficacy estimand (efficacy prior to discontinuation of study drug or initiating rescue therapy for persistent hyperglycemia). Results are shown in the table below. It was a notably decision by Lilly to directly highlight the head-to-head trial with its closest GLP-1 competitor, instead of other SURPASS programs comparing tirzepatide to placebo, metformin, or insulin. While body weight and glucose outcomes did seem to show tirzepatide outperforming semaglutide, we would be curious to hear more on patient reported outcomes between the two medications given the high potential for GI side effects. More to the point on this, we’re also just eager to learn more about which kinds of clinicians are able to help PWD titrate optimally or which are less schooled at this – titration is absolutely critical to good starts on new therapies, particularly outside of clinical trials

 

5mg tirzepatide

10mg tirzepatide

15mg tirzepatide

1mg semaglutide

A1c reduction

-2.1%

-2.3%

-2.5%

-1.9%

Body weight reduction

-8.5%

-11%

-13%

-7%

Percent of participants achieving A1C <7%

85%

89%

92%

81%

Percent of participants achieving A1C <5.7%

29%

45%

51%

20%

  • Mounjaro is safe and generally well-tolerated, though gastrointestinal side effects are the most commonly reported adverse events, similar to other incretins. In SURPASS 2, between 40-46% of participants on tirzepatide reported GI-related adverse events, depending on the dose, on par with 41% in semaglutide. That said, only 3-4% of participants discontinued this treatment because of the GI events suggesting that benefits of glycemic control and weight loss may have outweighed the costs. Notably, Dr. Frias spent significant time detailing the increased risk of thyroid c-cell tumors. Although this hasn’t presented itself as a major concern in clinical trials, early studies in mice found that tirzepatide increased these tumors in a dose-dependent and treatment-duration dependent manner. Mounjaro is contraindicated in people with a personal or family history of medullary thyroid carcinoma, as well as in pediatric patients, pregnant women, and patients with pancreatitis.
  • Dr. Frias offered a bit more insight into potential dosing for patients starting Mounjaro. Recall that dosing starts at 2.5 mg, a maintenance dose that patients stay on for four weeks, before having the option to increase to 5 mg, 7.5 mg, 10 mg, 12.5 mg, and 15 mg doses. Dr. Frias noted that Lilly will have “lots of samples” for interested patients to try out and that eligible patients with commercial insurance may pay as little as $25/month with the Mounjaro Savings Program. For patients for whom this impacts, this could be a huge win for them, if they qualify, particularly as the field appears to anticipate the sticker price could approach nearly $1,000/month. For all the benefits, from a systems perspective, this therapy is bringing a lot of value – for anyone who is underinsured, of course, this may be extremely challenging.

Dr. Julio Rosenstock advocates for provocative new paradigm reverting type 2 diabetes from day one; calls for simultaneous initiation of combination therapy to achieve normoglycemia, followed by maintenance therapy as needed inspired by an oncologic model of aggressive intervention

On the last day of ADA 2022, Dr. Julio Rosenstock (Velocity Clinical Research and Univ. Of Texas Southwestern Medical Center, TX) gave a provocative talk in which he pushed for a new definition for diabetes remission or reversion  calling for normoglycemia as a goal in type 2 diabetes treatment. In August 2021, a consensus report published by a panel of international experts convened by the ADA set the definition for type 2 diabetes remission as an A1c <6.5% for at least three months without the use of glucose-lowering drugs. Dr. Rosenstock criticized this definition, saying the requirement to not use glucose-lowering drugs is not translatable to clinical practice. He said even with disease-modifying therapies – therapies that improve beta cell function and insulin resistance – maintenance therapy will be required, but normoglycemia with maintenance therapy is an admirable goal. He suggested re-defining diabetes remission using the oncology conception of remission, which includes maintenance therapy during the post-remission period. Specifically, Dr. Rosenstock pushed for “short-term intensive use of drugs targeting multiple pathways and achieving meaningful body weight loss to attain a reversion to normal glycemic metric with maintenance therapy as needed.” In practice, he said, this means simultaneous initiation of combination therapy, and he suggested maintenance therapy could involve lower doses of drugs or even intermittent therapy. The first-line combination therapy he recommended was metformin with an SGLT-2 and/or a GLP-1. Dr. Rosenstock acknowledged that the 2022 ADA Standards of Care allow for the first-line SGLT-2 or GLP-1 treatment prior to metformin initiation, but he expressed skepticism that more personalized but still sequential treatment recommendations were sufficient to lead to better glucose control and less clinical inertia. The ADA Standards of Care recommends an SGLT-2 or GLP-1 for people with ASCVD or indicators of high ASCVD risk, HF, or CKD, but Dr. Rosenstock said that all people with diabetes are at high risk for ASCVD, HF, and CKD. Therefore, he said that SGLT-2s and/or GLP-1s should be standard components of first-line therapy in type 2 diabetes regardless of an individual’s A1c in pursuit of achieving diabetes remission. Dr. Rosenstock concluded, “Tirzepatide has moved the goalpost for type 2 diabetes management towards attaining diabetes reversal or remission!” 

  • Highlighting how normoglycemia is an achievable goal in type 2 diabetes, Dr. Rosenstock cited results from the STEP program for GLP-1 semaglutide 2.4 mg and the SURPASS program for GIP/GLP-1 tirzepatide. For example, STEP-2 showed that 77% of participants on semaglutide 2.4 mg achieved an A1c £6.5% and 85% achieved an A1c <7%. In STEP-1, of the participants who had prediabetes at baseline, 84% of those on semaglutide reverted to normoglycemia, compared to 48% of those on placebo. Turning to tirzepatide, Dr. Rosenstock characterized data from tirzepatide’s SURPASS program as “unparalleled.” He said the particularly SURPASS-1, which enrolled drug naïve patients with type 2 diabetes, could change the paradigm of type 2 diabetes treatment. SURPASS-1 found that, among participants on tirzepatide 15 mg, 92% achieved an A1c <7%, 86% achieved an A1c £6.5%, and 52% achieved an A1c <5.7%. This result underscores the potential tirzepatide, which was approved to treat type 2 diabetes in May 2022, to bring a significant number of people who receive it into normoglycemia, even as first-line therapy. Turning towards the SURMOUNT-1 results of tirzepatide in obesity, Dr. Rosenstock exclaimed, “Wow!” He was particularly impressed that of the 41% of the study population that had prediabetes, 95% of people reverted to normoglycemia with tirzepatide. He concluded that tirzepatide is a very powerful tool that we should use to not just drive type 2 diabetes remission but also to prevent type 2 diabetes.
  • Dr. Rosenstock also shared clinical evidence supporting the positioning of SGLT-2s and GLP-1s as first-line therapy. Citing a study he published in 2016 in Diabetes Care (n=1,364), Dr. Rosenstock showed that first-line combination therapy of metformin and SGLT-2 empagliflozin led to significantly greater A1c reductions over 24 weeks compared to empagliflozin alone and metformin alone. Furthermore, the PIONEER-1 study (n=703) showed that once-daily oral semaglutide as first-line therapy in people with type 2 diabetes led to significant A1c and bod weight reductions compared to placebo. In fact, 64% of participants on semaglutide 14 mg achieved an A1c £6.5%. Finally, Dr. Rosenstock cited the AWARD-3 study (n=807) to show that once-weekly injectable dulaglutide confers significant A1c reduction and body weight reduction compared to metformin in early stage type 2 diabetes.  
  • Dr. Rosenstock also listed six goals for simultaneous initiation of combination therapy to facilitate earlier achievement of glycemic targets: (i) the therapies should exhibit complementary actions/mechanisms; (ii) glycemic control on combination therapy should be better than with any individual component; (iii) Combined doses for each therapy may be lower than monotherapy doses; (iv) side effects should not be increased, and ideally should be mitigated; (v) treatment management should be simplified and may improve the consistency of medication use; and (vi) the cost of combination therapy should be lower than the sum of costs for individual therapies.
  • While Dr. Rosenstock mentioned that meaningful body weight loss drives improvements in glycemic control, in a preceding talk, Dr. Jennifer Green (Duke University) noted that weight loss is only partially responsible for improvements in insulin resistance As Dr. Jennifer Green (Duke) noted in an earlier talk during the same session as Dr. Rosenstock, tirzepatide has been shown to improve both beta cell function and insulin resistance. Citing results from SURPASS-1, for which Dr. Rosenstock was first author, Dr. Green explained that tirzepatide 15 mg led to a 23% reduction in a measure of insulin resistance, HOMA2-IR, over 40 weeks, compared to a 15% rise seen with placebo. However, she added that weight loss is only partially responsible for this improvement in insulin resistance. Citing a poster presented at EASD 2019, Dr. Green said that weight loss with tirzepatide was responsible for about 22-28% of improvement in insulin resistance, depending on tirzepatide dose, indicating that while weight loss is a significant factor for reduced insulin resistance it is not the only factor. Still, underlining the significance of weight loss, Dr. Rosenstock shared that losing 10-15 kg in body weight is associated with a two-to-three-fold higher likelihood of reverting to normoglycemia. We should note also that about 10% of people with type 2 diabetes do not have overweight or obesity, so while weight loss is clearly important it will not address everyone’s type 2 diabetes. 
    • In another talk preceding Dr. Rosenstock’s, Dr. Rodolfo Galindo (Emory University) emphasized that A1c reduction is not enough for patients with type 2 diabetes – weight loss is also important. Though neither Dr. Rosenstock no Dr. Galindo explicitly prioritized GLP-1s or other incretin therapies, like tirzepatide, over SGLT-2s, the weight loss benefits of incretins took the limelight in this session. In reviewing the SURPASS program, along with tirzepatide’s A1c reduction and weight loss benefits, Dr. Galindo reiterated that with the advent of newer cardiorenal protective therapies, A1c reduction alone is not “good enough” for people with type 2 diabetes. He said that the ideal type 2 diabetes therapy not only controls glycemia but also has a low risk of hypoglycemia, improve beta cell function, controls body weight, controls lipoproteins, controls hypertension, and reduces cardiorenal complications. Clearly, the standard for type 2 diabetes therapies has been raised with the advent of SGLT-2s and GLP-1.

Tirzepatide demonstrates significant reduction on albuminuria, eGFR slope, and a composite kidney outcome compared to insulin glargine in an exploratory analysis of SURPASS-4; consistent benefits regardless of baseline SGLT-2 use

In a packed room, Dr. Hiddo Heerspink (University Medical Center Groningen, Netherlands) presented a prespecified exploratory analysis of SURPASS-4, indicating that tirzepatide may provide kidney protection. In May 2022, GIP/GLP-1 dual agonist tirzepatide was approved to treat type 2 diabetes based on the SURPASS program. SURPASS-4 investigated tirzepatide vs. insulin glargine in adults with type 2 diabetes and increased CV risk. Dr. Heerspink pointed out that a substantial portion of the SURPASS-4 population had chronic kidney disease, with 80% of participants having an eGFR below 60 mL/min/1.73 m2, 28% with microalbuminuria, and 8% with macroalbuminuria. This exploratory analysis (n=1,995) found that tirzepatide led to significant benefits across three kidney endpoints compared to insulin glargine: (i) a 32% reduction in urine albumin-to-creatinine ratio (UACR); (ii) a reduce rate of eGFR decline by 2.21 mL/min/1.73 m2 per year; and (iii) a 42% relative risk reduction on a composite kidney endpoint (macroalbuminuria, 40% eGFR decline, end stage kidney disease, or renal death). Dr. Heerspink interpreted these results to indicate that tirzepatide may provide kidney protection, adding that these “striking” and “pronounced” results should not be interpreted as suggesting that insulin has harmful kidney effects. Notably, tirzepatide’s benefits on UACR and stabilization of eGFR were consistent regardless of SGLT-2 use, indicating the potential for combination therapy to further prevent kidney dysfunction.

  • Tirzepatide reduced UACR and eGFR slope across high-risk kidney-specific subgroups. Compared to insulin glargine, tirzepatide conferred significant reductions in UACR greater than 40% in three high-risk subgroups: (i) people with micro- or macroalbuminuria (n=706); (ii) people with moderately or severely reduced kidney function (n=335); and (iii) people at high risk for kidney-related outcomes (n=184). Additionally, in the same three subgroups, tirzepatide reduced the rate of eGFR decline. 
  • Tirzepatide led to an initial eGFR drop, followed by eGFR stabilization, reflecting similar kidney protective effects as SGLT-2s. The initiation of SGLT-2s is accompanied by a well-established acute drop in eGFR due to glomerular tubular feedback that is followed by a reduced decline in eGFR slope, ultimately reducing the rate of CKD progression. Additionally, as Dr. Heerspink noted, discontinuing SGLT-2 treatment has been associated with a rise in eGFR, indicative of glomerular hyperfiltration. As shown in the graph below (blue indicates tirzepatide; white indicates glargine), patients on tirzepatide experienced a similar eGFR profile: an acute eGFR drop after 12 weeks, stabilization of eGFR, and a rise in eGFR after treatment is discontinued. Dr. Heerspink noted that mechanistic studies of tirzepatide are ongoing, so he was not able to make direct mechanistic comparisons between SGLT-2s and tirzepatide.

  • Dr. Heerspink shared a few potential mechanisms by which tirzepatide may exert kidney benefits. He noted that a reduction in A1c can lead to an acute decline in eGFR, which may explain tirzepatide’s acute eGFR drop. He also suggested that GLP-1 agonism may lead to natriuresis (excretion of sodium in urine), in turn leading to tubular glomerular feedback, similar to SGLT-2s. He acknowledged there may be other ways tirzepatide reduces albuminuria, such as reducing endothelial dysfunction. He added that tirzepatide has been shown to reduce blood pressure, which can lead to reduced albuminuria, though Dr. Heerspink added that mediation analyses indicate blood pressure only plays a minor role in albuminuria reduction. On the particular impact of GIP agonism, Dr. Heerspink said that a clear explanation would require head-to-head studies of GIP agonism, GLP-1 agonism, and dual GLP-1/GIP agonism.

Closer look at six (!) novel, early-stage incretin therapies highlight the incredible development and progress driving field forward

This packed oral presentation session highlighted several early-stage, novel incretin therapies currently in development for the treatment of type 2 diabetes and obesity. We were pleased but not surprised to see the number of combination agents being studied, given the impressive effects of tirzepatide that are likely conferred, at least in part, by the dual GIP/GLP-1 receptor agonism. The studies are predominantly smaller phase 1 trials investigating the safety and tolerability, basic clinical efficacy, and pharmacokinetics of the compounds. GI-related side effects were the most commonly reported adverse events in all of the studies, which was not unexpected given this is consistent across the rest of the incretin class. We’ve highlighted some of the featured compounds, alongside topline results, in the table below. Please note that this table is not meant to compare the studies to one another (infeasible given the completely different study designs) and is simply intended to collect the topline results from of the most exciting incretin-based therapies in development that were presented in this session.

Compound Name

Class

Mode of delivery

Doses

Study pop. (n=#)

Trial length

Change in body weight

Sponsor

Notable Comments

Pemvidutide

GLP-1/Glucagon

Injectable

1.2 mg, 1.8 mg, 2.4 mg

People with over-weight or obesity (n=34)

12 weeks

-10%

Altimmune

Longer half-life (110 hrs), delayed time to peak concentration, no titration needed; Initiated 48-week phase 2 MOMENTUM trial in obesity in April 2022

Dapiglutide

GLP-1/GLP-2

Injectable

1.o mg, 2.25 mg, 3.5 mg, 6.0 mg

Healthy subjects (n=40)

4 weeks

-4%

Zealand

Body weight loss was dose-dependent; predictable PK profile; First mentioned in March 2021

LY3502970

GLP-1

Oral

2 mg, 6 mg, 16 mg, 24 mg

Healthy subjects (n=92)

4 weeks

-4%

Lilly

Decreased excursions in oral glucose tolerance tests; dose proportional PK response supporting once-weekly dosing; first heard about LY3502970 when it was moved to phase 1 trials in 4Q19

LY3437943

GIP/GLP-1/Glucagon

Injectable

5 rising dose cohorts

Patients with type 2 diabetes (n=72)

12 weeks

-10%

Lilly

A1c decreased by 1.56% on the highest dose; first triple-g agonist that we’ve seen; expected to confer “bariatric surgery like weight loss”; phase 2 trial expected to complete in 2H22

LY3305677

(Mazdutide)

GLP-1/Glucagon

Injectable

Cohort 1: 1.5 mg to 4.5 mg; ascending dose

 

Cohort 2: 2 mg to 10 mg; ascending dose

Healthy subjects and patients with type 2 diabetes

(n=24)

Cohort 1: 12 weeks

 

Cohort 2: 16 weeks

Cohort 1: -3.0%

 

Cohort 2:

-12.7%

Lilly

In patients with type 2 diabetes, Mazdutide significantly reduced HbA1c and body weight from baseline

Danuglipron

GLP-1

Oral

40 mg, 80 mg, 120 mg

Japanese participants with type 2 diabetes

(n=28)

8 weeks

-2.63%

Pfizer

Danuglipron reduced HbA1c, plasma glucose levels, and body weight, as well as showed a favorable safety profile

Post-hoc analysis of SURPASS-4 finds that tirzepatide confers better sustained glucose control than insulin glargine

Professor Ewan Pearson (University of Dundee, Dundee, UK) presented a post-hoc analysis of the SURPASS-4 trial investigating the durability of A1c control of tirzepatide compared to insulin glargine. Professor Pearson framed type 2 diabetes as a progressive disease, where A1c and fasting glucose levels tend to rise over time, even when patients are on anti-diabetes medications. Fast progression of diabetes is associated with younger age of T2D onset, higher body mass, high levels of triglycerides, and low levels of high-density lipoprotein. Recall that SURPASS-4 compared tirzepatide to insulin glargine in a 52-week trial of 2,002 patients with type 2 diabetes and high CVD risk, and showed that tirzepatide conferred significant reductions in A1c, body weight, and triglyceride levels. The aim of Professor Pearson’s post-hoc analysis was to evaluate the sustainability of glycemic control after reaching A1c targets at week 52. Participants continued to be followed after the 52-week primary endpoint, for a median duration of 85 total weeks, though some participants had data through 104 weeks of follow-up.

  • The results show tirzepatide confers significantly better long-term glycemic control compared to insulin glargine. The table suggests that not only are participants on tirzepatide able to achieve lower A1c levels, they are also able to sustain these lower levels for a longer period than participants on glargine. Notably, glargine failed to meet the criteria for sustained control, defined as an increase of ≤0.2% in A1c from Week 52.

Results for A1c <7%

 

TZP 5mg (n=274)

TZP 10 mg (n=281)

TZP 15 mg (n=283)

Insulin glargine (n=862)

% with A1c <7% at week 52

81%

86%

90%

51%

Of those participants, % with A1c <7% post-week 52

76%

79%

80%

66%

Mean A1c change between week 52 and post-week 52 of participants meeting <7% at week 52

0.14%

0.08%

0.10%

0.27%

Odds ratio of achieving and sustaining A1c <7% compared to placebo

3.2

4.6

5.5

--

Results for A1c <6.5%

 

TZP 5mg (n=274)

TZP 10 mg (n=281)

TZP 15 mg (n=283)

Insulin glargine (n=862)

% with A1c <6.5% at week 52

67%

74%

81%

32%

Of those participants, % with A1c <6.5% post-week 52

75%

80%

83%

66%

Mean A1c change between week 52 and post-week 52 of participants meeting <6.5% at week 52

0.14%

0.08%

0.09%

0.26%

Results for A1c <5.7%

 

TZP 5mg (n=274)

TZP 10 mg (n=281)

TZP 15 mg (n=283)

Insulin glargine (n=862)

% with A1c <6.5% at week 52

24%

33%

42%

3%

Of those participants, % with A1c <6.5% post-week 52

75%

78%

84%

52%

Mean A1c change between week 52 and post-week 52 of participants meeting <6.5% at week 52

0.06%

0.13%

0.10%

0.51%

Odds ratio of achieving and sustaining A1c <7% compared to placebo

3.9

5.8

8.4

--

Tirzepatide confers superior A1c reductions compared to placebo for patients <65 years and ≥65 years

Dr. Carol Wysham (University of Washington) presented a pre-specified analysis of the SURPASS clinical trial program assessing the effect of tirzepatide on participants older and younger than 65 years old. Given that tirzepatide was a major theme of the conference, we are pleased to see this subgroup analysis assessing the efficacy of the dual GIP/GLP-1 agonist in different age groups to understand tirzepatide’s effects across diverse populations. Patients across the five SURPASS trials were randomized to tirzepatide 5 mg, 10 mg, 15 mg, or placebo/active comparator and followed for 40 weeks (SURPASS-1, SURPASS-2, and SURPASS-5) or 52 weeks (SURPASS-3, SURPASS-4). Notably, the results from this age-based analysis were consistent with the primary study results: tirzepatide conferred superior reductions in A1c compared to placebo/active comparator at all three doses in both people <65 and ≥65 years old. As expected, the most commonly reported adverse events associated with tirzepatide were mild-moderate and gastrointestinal-related and typically occurred during the titration period.

  • The effect of tirzepatide in older adults is especially interesting given the different treatment targets for this population. For example, the 2022 Standards of Care recommend that older adults with multiple coexisting chronic illnesses, cognitive impairment, or functional dependence should have less stringent glycemic goals (such as A1C <8.0–8.5%) and also notes that “glycemic goals for some older adults might reasonably be relaxed as part of individualized care.” With that said, the primary concern in older adults regarding glycemic control is related to hypoglycemia, and since tirzepatide does not put patients at increased risk for hypoglycemia, this shouldn’t affect its use in the older adult population. There are also additional considerations regarding weight loss in older adults, as we heard about at ENDO 2022. Dr. Reina Villareal (Baylor College of Medicine) noted that some clinicians are concerned that weight-loss can reduce bone and muscle mass, worsening osteopenia and sarcopenia. However, pharmacotherapy combined with exercise and other behavioral interventions can lead to healthy, sustained weight loss. We would have been curious to see a similar subgroup analysis showing body-weight outcomes and are hopeful that this data will be shared at upcoming conferences.

Posters – Novel Therapies

Title

Authors

Details + Takeaways

Tirzepatide as Monotherapy Improved Markers of Beta-Cell Function and Insulin Sensitivity in People with Type 2 Diabetes (SURPASS-1)

Clare Lee, Huzhang Mao, Vivian Thieu, Melissa K. Thomas

  • A post hoc analysis of SURPASS-1 exploring changes in biomarkers of beta-cell function and insulin sensitivity after tirzepatide monotherapy
  • After 40 weeks, proinsulin/C-peptide ratios (markers of insulin processing, beta-cell stress, and beta-cell function) significantly improved with tirzepatide vs. placebo (-47-49% vs. -0.1%, respectively)
  • Tirzepatide significantly improved insulin sensitivity, as shown by increases in adiponectin compared to placebo (16-23% vs. -0.2%)

Higher Weight Loss Is Associated with Improved Quality of Life in Patients with Type 2 Diabetes—SURPASS Program

Kristina Boye, Vivian Thieu, Maria Yu, Helene Sapin

  • Explored the relationship between weight loss and patients’ quality of life using pooled data from the SURPASS clinical trials, regardless of treatment and dosing
  • Greater weight loss correlated with greater improvements in quality of life, including physical activity performance, self-perception, and treatment satisfaction as assessed by Ability to Perform Physical Activities of Daily Living (APPADL), Impact of Weight on Quality of Life-Lite Clinical Trials (IWQOL-Lite-CT), Impact of Weight on Self-Perceptions (IW-SP), and EQ-5D-5L and Diabetes Treatment Satisfaction Questionnaire change (DTSQc)
  • All three weight loss categories (≥5%, ≥10%, ≥15%) saw improvements in self-reported quality of life

Psychosocial Impact of Severe Hypoglycemia and Perceptions of Nasal Glucagon in Young Adults with Type 1 Diabetes

Caitlin S. Kelly, Huyen Nguyen, Weixiu Luo, Katherine S.M. Chapman, Jiat Ling Poon, Levenia A. Baker, Wendy Wolf, Magaly Perez-Nieves, Beth Mitchell

  • Assessing the perceived impact of a severe hypoglycemic emergency and nasal glucagon on social engagement and distress among young adults with T1D (n=364)
  • Majority of emerging adults perceived treating severe hypoglycemia as distressing, but not all individuals reported lower social engagement because of severe hypoglycemia
  • 31% of participants reported their freedom to engage in social activities improved/greatly improved with use of nasal glucagon; 67% reported no change.
  • Greater feelings of preparedness and protection were reported with nasal glucagon than with emergency glucagon requiring constitution

Effects of Finerenone in Patients with CKD and T2D Are Independent of HbA1c at Baseline, HbA1c Variability, and Duration of Diabetes

Janet B. McGill, Rajiv Agarwal, Stefan Anker, George Bakris, Gerasimos Filippatos, Betram Pitt, Luis M. Ruilope, Andreas L. Birkenfeld, Luiza Caramori, Meike Daniela Brinker, Amer Joseph, Andrea Z. Lage, Robert Lawatscheck, Charlie Scott, Peter Rossing

  • RCT (n=13,026) of finerenone vs. placebo on CV and kidney outcomes evaluating effects by baseline A1c, A1c variability, and diabetes duration
  • A 1% increase in mean residual A1c was correlated with a 20% increased risk of a CV event and a 36% increased risk of a kidney event
  • Reduced risk for CV and kidney outcomes with finerenone were standard across all baseline categories

Safety, PK, PD, and Efficacy of Weekly Dosing Supaglutide in Patients with T2D

Yue Zhou, Xin Jiang, Yan-Ru Lou, Anran Ma, Jia Li, Ying Zhao, Gerald J Prud’Homme, Qinghua Wang

  • A 7-week RCT (n=40) of supaglutide (1, 2, 3, and 4 mg) vs placebo on fasting glucose and fasting insulin/c-peptide in patients with T2D
  • Supaglutide reduced fasting blood glucose and body weight; A1c decreased by 1.3%; fasting insulin/c-peptide increased
  • Half-life of supaglutide found to be ~207 h (8.5 days) with median maximum dosage found in bloodstream at 60-84 hours after injection
  • With positive safety profile, supaglutide has potential to be a novel alternative therapy for T2D and metabolic disorders

A Randomized Clinical Trial of Safety, Pharmacokinetics, and Pharmacodynamics of Weekly Dosing Supaglutide in Healthy Chinese Subjects

Yue Zhou, Yan-Ru Lou, Yijing Liao, Anna Shao, Zhihong Wang, Yaojing Jiang, Qiaoli Cui, Ying Zhao, Jia Li, Gerald J. Prud’Homme, Dalong Zhu, Qinghua Wang

  • 14-week RCT (n=48) evaluating safety, pharmacokinetics, pharmacodynamics, and potential immunogenicity of supaglutide vs placebo in healthy subjects
  • Supaglutide was well-tolerated in healthy subjects with mild to moderate GI symptoms as doses increased, and no anti-supaglutide antibodies were developed
  • Results supported supaglutide’s potential as an alternative GLP-1 therapy

A Phase 1, Double-Blind, Placebo-Controlled Multiple Escalating Dose Study of RGT-075 Novel Small-Molecule Oral GLP-1 Receptor Agonist in Adults with Type 2 Diabetes

Mark A. Pirner, Jing Lin, Feng Liu, Lili Yao, Marjorie E. Zettler, David J. Valacer

  • Four-week RCT (n=36) of varying doses of RGT-075, a small molecule oral GLP-1, vs. placebo on safety (primary), pharmacokinetics, A1c , FPG, CGM-TIR, mixed-meal tolerance test (MMTT), and body weight (BW) in patients with T2D
  • RGT-075 cohort showed improved A1c, FPG, CGM-TIR, MMTT, and BW
  • Limitations of trial included a short duration and high baseline variability

Indirect Treatment Comparison (ITC) of Three Ready-to-Use Glucagon Treatments for Severe Hypoglycemia: Baqsimi, Gvoke, and Zegalogue

Marga Gimenez, Kamlesh Khunti, Kristen Syring, Levenia A. Baker, Suresh Chenji, Rebecca J. Threlkeld, Yu Yan, Munehide Matsuhisa

  • Systematic literature review comparing Baqsimi, Gvoke, and Zegalogue on proportion of participants achieving treatment success, maximum blood glucose, and treatment-emergent adverse events
  • All treatments conferred high and comparable success (>98%) for adults and children with diabetes
  • Statistically significant faster time in achieving treatment for Baqsimi vs. Gvoke

Experiences with Nasal Glucagon from People with Diabetes and Caregivers, Most Mentioned ‘Feeling Safe’: A Qualitative Study

Beth Mitchell, France G. Sowell, Paul Williams, Jiat Ling Poon

  • Assessing the experiences of people with diabetes (n=22) and caregivers (n=7) with nasal glucagon (NG) use and delivery
  • 93% of participants felt safe having NG, 90% were comforted by NG as a treatment option, and 86% were confident in others’ ability to use NG

Greater Improvement in Insulin Sensitivity Per Unit Weight Loss with Tirzepatide Compared with Selective GLP-1 Receptor Agonism

Kieren J. Mather, Andrew Mari, Jing Li, Tim Heise, J. Hans de Vries, Shweta Urva, Tamer Coskun, Zvonko Milicevic, Axel Haupt, Melissa K. Thomas

  • 28-week RCT of tirzepatide 15 mg vs. placebo vs. semaglutide 1 mg on relationship between weight loss and insulin sensitivity improvement in patients with T2D
  • Greater improved insulin sensitivity per unit weight loss seen with tirzepatide 15 mg use vs. semaglutide 1 mg

Glycemic Control with Tirzepatide in People with Type 2 Diabetes by Baseline HbA1c = 8.5% or >8.5%

Grazia Aleppo, Christophe de Block, Joshua A. Levine, Elisa Gomez-Valderas, Brian D. Benneyworth

  • Assessing glycemic control with tirzepatide by baseline A1c of ≤8.5% or >8.5%
  • A1c reductions from baseline ranged from 1.6-2.1% in the baseline A1c ≤8.5% subgroup and 2.7-3.5% in the baseline A1c >8.5% subgroup
  • All 5 SURPASS studies had consistent A1c reductions regardless of baseline A1c with tirzepatide vs. placebo

Tirzepatide Induces Weight Loss in Patients with Type 2 Diabetes Regardless of Baseline BMI: A Post Hoc Analysis of SURPASS-1 through -5 Studies

Anita Kwan, Juan M. Maldonado, Hui Wang, Neda Rasouli, John Wilding

 

  • Analysis of SURPASS trial data to determine whether tirzepatide-induced weight loss depends on baseline BMI for patients with T2D
  • Although the amount of weight lost varied with baseline BMI, tirzepatide still significantly reduced body weight across all groups
  • The amount of weight reduction depended on tirzepatide dosage

Change in Body Weight from Baseline with Tirzepatide: Sex Subgroup Analysis of the SURPASS Studies

Arian W. Plat, Neda Rasouli, Jennifer Peleshok, Helene Sapin, John Wilding

  • Subgroup analysis of data from SURPASS trial by sex, to examine how sex affects weight loss with tirzepatide vs. placebo for patients with T2D
  • All tirzepatide doses caused greater weight loss than placebo, regardless of sex

Tirzepatide Improves Multiple Aspects of Beta-Cell Function

Andrea Mari, Jing Li, Shweta Urva, Tim Heise, J. Hans De Vries, Edward J. Pratt, Robert J. Heine, Melissa K. Thomas, Zvonko Milicevic, Kieren J. Mather

  • Double-blind trial of beta cell function in people with T2Don 15mg tirzepatide vs. 1mg semaglutide vs. placebo
  • Tirzepatide significantly improved insulin sensitivity, insulin secretion, and glucose sensitivity over semaglutide and placebo
  • Tirzepatide significantly decreased glucose levels more than semaglutide and placebo did

The Dual Glucagon and Glucagon-Like Peptide-1 Receptor (GCGR/GLP-1R) Agonist BI 456906 Reduces Body Weight in Diet-Induced Obese Mice Based on Food Intake Reduction and an Increase in Energy Expenditure

Tina Zimmermann, Leo Thomas, Tamara Baader-Paglerr, Peter Haebel, Eric Simon, Wolfgang Rist, Ingo Uphues, Heike Neubauer, Robert Augustin

  • In vitro and in vivo pharmacological characterizations of BI 456906, a long acting GCGR/GLP-1 dual agonist, in diet-induced obese mice
  • Second messenger signaling was measured in vitro, and food intake, gastric emptying, and glucose tolerance were measured in vivo to determine GCGR/GLP-1 receptor activation
  • BI 456906 reduced energy intake and increased energy expenditure, achieving significantly greater weight loss outcomes than semaglutide  and demonstrating potential as an obesity therapy

A First-in-Human Study of RGT-075, a Novel, Orally Bioavailable, Small-Molecule GLP-1 Receptor Agonist, in Healthy Adult Subjects

Mark A. Pirner, Jing Lin, Feng Liu, Lili Yao, Marjorie E. Zettler, David J. Valacer

 

  • Assessing safety, tolerability, and pharmacokinetics of RGT-075, a novel oral GLP-1, in healthy volunteers. (n=56)
  • Single doses between 15-280 mg were safe and well-tolerated. Further studies will evaluate the use of RGT-075 as a T2D therapy

Glycemic Effect of Tirzepatide by Duration of Diabetes

Christophe De Block, Chantal Mathieu, Helene Sapin, Jacek Kiljanski, Jennifer Peleshok

  • Subgroup analysis of 5 SURPASS clinical trials to determine tirzepatide efficacy across 3 baseline duration of T2D: <5 years, 5 to 10 years, >10 years
  • Tirzepatide significantly and consistently lowered A1c across all groups, regardless of diabetes duration

Glycemic Variability of Tirzepatide vs. Insulin Degludec in People with Type 2 Diabetes Using Continuous Glucose Monitoring (SURPASS-3 CGM)

Richard Bergenstal, Amy Bartee, Meltem Zeytinoglu, Ross Bray, Sheryl Allen, Katelyn Brown

  • Using CGM to compare glycemic variability in people with T2D on tirzepatide vs insulin degludec (n=243)
  • After 52 weeks of treatment, individuals on tirzepatide had significantly lower glycemic variability than those on insulin degludec; variability was similar to control patients without T2D
  • Subjects on tirzepatide also spent more time in the target range (25% longer than the competitor) and less time in hyper/hypoglycemia

Relationship between Body Weight Change and Glycemic Control with Tirzepatide Treatment in People with Type 2 Diabetes

Sue Pedersen, Guillermo E. Umpierrez, Francesco Giorgino, Ángel Rodríguez, Vivian Thieu, Helene Sapin, Laura Fernandez Lando, Chrisanthi A. Karanikas, Jacek I. Kiljanski

 

  • Post-hoc analysis of SURPASS trials to determine the relationship between A1c and body weight reductions with tirzepatide
  • Across trials, 87-97% of patients experienced A1c reduction with weight loss
  • A1c and body weight changes were significantly correlated in SURPASS 2, 3, 4, and 5 (5mg only)

Patients with Type 2 Diabetes Reach Glycemic Targets Faster with Tirzepatide Compared with Semaglutide and Titrated Insulin Degludec

Kevin M. Pantalone, Adie Viljoen, Rodolfo J. Galindo, Xuewei Cui, Ruth Huh, Laura Fernández Landó, Hiren Patel

  • Evaluating time required to achieve glycemic control in SURPASS 2 (tirzepatide vs. semaglutide 1mg, n=1,878) and SURPASS 3 (tirzepatide vs. insulin degludec, n=1,437)
  • Glycemic targets were compared at 40 weeks of treatment and 52 weeks of treatment
  • Subjects on tirzepatide achieved A1c <7% 4 weeks faster compared to semaglutide 1 mg and insulin degludec; body weight loss ≥5% was achieved 12 weeks faster with tirzepatide 10 mg and 15 mg than semaglutide 1 mg
  • Hypoglycemia risk was significantly reduced in the SURPASS 3 condition compared to the competitor, but not in the SURPASS 2 condition

Effects of a Novel Long-Acting GIP/GLP-1/Glucagon Trireceptor Agonist, HISHS-3001, on HbA1c, Body Weight, and Lipid Metabolism

Vinod Burade, Adolfo Garcia-Ocana, Richard E. Pratley, Guy A. Rutter, Tina Vilsbøll, Bernard Thorens, Rajamannar Thennati

  • Evaluating efficacy of HISHS-3001, a novel glucagon tri-receptor agonist, compared to tirzepatide in T2D mouse model (3 groups: tirzepatide n=8; HISHS-3001 n=8; saline n=8)
  • HISHS-3001 was significantly more effective at lowering body weight, controlling glucose homeostasis, and decreasing blood glucose, and was effective at lower concentrations than tirzepatide

Trends in Out-of-Pocket Cost for Glucagon 2010-2020

Margaret Zupa, Robert Feldman, Jing Luo

  • Examining out-of-pocket costs for glucagon from 2010-2020 to characterize its contribution to low glucagon uptake
  • Out-of-pocket costs were $0 for most individuals on Medicare and less than $40 for most commercially insured individuals, demonstrating that cost is an unlikely barrier to uptake

Glucagon Prescription Rates in Insulin-Treated Patients with Diabetes at a Teaching Outpatient Clinic

Tiba Abdulwahid MD, Jacob Lloyd MD, Karen Selk DO

  • A retrospective observational study of 111 patients with diabetes in a teaching outpatient clinic to determine if glucagon is under-prescribed
  • Only 5 patients (4.5%) were found to have been prescribed glucagon, compared to a 12.9% prescription rate for patients who had seen an endocrinologist, potentially due to a clinical knowledge gap

DA-1726, a Balanced GLP1R/GCGR Dual Agonist, Effectively Controls Both Body Weight and Blood Glucose

Tae-Hyoung Kim, Il-Hun Jung, Kyumin Kim, Boram Lee, Mi-Kyung Kim, Yuna Chae

  • Evaluating pharmacological effects of DA-1726, a novel long-acting oxyntomodulin, for use as a potential GLP-1/glucagon receptor agonist
  • DA-1726 showed enhanced activation of GLP-1/glucagon receptors, and increased body weight loss (through decreasing food intake and increasing energy metabolism) compared to WT control and semaglutide
  • DA-1726 provided effective glycemic control without inducing hypo/hyperglycemia at fasting states

Insulin Therapies

Lilly’s once-weekly basal insulin Fc (BIF) confers similar A1c reduction (-1.2%) and Time in Range improvement (+2.5 hours/day) compared to insulin degludec in phase 2 study of patients type 2 diabetes; similar safety profile between both treatments

Dr. Juan Pablo Frias (Velocity Clinical Research, CA) presented results from a phase 2 study comparing Lilly’s once-weekly basal insulin Fc (BIF) vs. daily insulin degludec in insulin-naïve patients with type 2 diabetes (215-OR). Participants (n=278) had type 2 diabetes treated with metformin alone or metformin with a DPP-4 (~13%), with an SGLT-2 (~20%), or with both a DPP-4 and SGLT-2 (~9%). BIF and degludec dose adjustments were based on fasting blood glucose in a treat-to-target methodology. All patients underwent 14-day blinded CGM with Abbott LibrePro prior to baseline, 12 weeks, and 26 weeks. Over 26 weeks, from a baseline A1c of ~8%, BIF led to a significant 1.20% A1c reduction, similar to degludec’s significant 1.26% A1c reduction (treatment difference = 0.06%; 95% CI: -0.11%, 0.24%). Furthermore, similar proportions of patients achieved an A1c <7% on BIF and degludec – 62% and 69%, respectively. Participants on both BIF and degludec spent more than 18 hours/day in Range (>75% Time in Range): compared to baseline, those on BIF spent +2.5hr/day in Range, reaching 76% at 26 weeks, and degludec improved Time in Range by +2.4 hours/day to 77% at 26 weeks. BIF and degludec led to a similar reduction in Time Above Range (-3 hours/day to ~12%). While both treatments led to an increase in Time Below Range, BIF had a numerically lower rise in Time Below Range than degludec (+1.1 hours/day [5%] vs. +1.7 hours/day [9%], respectively). Dr. Frias said that overall BIF and degludec demonstrated similar safety profiles, with similar body weight gain (~2 kg) and similarly low rates of hypoglycemia; there were no cases of severe hypoglycemia. However, BIF was associated with a higher rate of systemic hypersensitivity reaction than degludec (4.2% vs. 0%). Dr. Frias concluded that BIF achieved excellent glycemic control with no concerning safety effects, supporting its development in phase 3 trials. BIF has also completed a phase 2 trial in people with type 1 diabetes, and we are awaiting these results.  

26-week Glucose Metrics

Basal Insulin Fc (BIF)

Insulin Degludec

A1c

6.8%

6.7%

Time in Range

76%

77%

Time >180 mg/dL

13%

11%

Time <70 mg/dL

5%

9%

Time <54 mg/dL

1%

2%

Fasting Plasma Glucose

118 mg/dL

113 mg/dL

  • BIF has already initiated phase 3 trials, making it the second farthest in development once-weekly insulin, behind Novo Nordisk’s insulin icodec. Once weekly BIF began its first phase 3 trial (QWINT-3) in type 2 diabetes in March 2022, which is expected to complete in April 2024. The QWINT program will consist of four trials investigating BIF in type 2 diabetes and one trial in type 1 diabetes, all of which are expected to initiate this year and complete by 2024. The once-weekly insulin furthest into development, Novo Nordisk’s insulin icodec, has released topline results from three phase 3 trials: ONWARDS 1, ONWARDS 2, and ONWARDS 6. For more on once-weekly insulins, see our insulin competitive landscape.

Once-weekly insulin icodec is non-inferior to insulin glargine in hypoglycemia frequency, magnitude, and physiological response

Dr. Thomas Pieber (Medical University Graz, Austria) presented data on the hypoglycemia frequency and physiological response to double or triple doses of once-weekly insulin icodec compared to once-daily insulin glargine. As Dr. Julio Rosenstock (UT Southwestern) pointed out during the Q&A, this as a “brilliant study design addressing the most common misconceptions about insulin icodec.” Namely, there could be reason to believe that once weekly basal insulin could lead to more severe episodes of hypoglycemia, but Dr. Pieber’s study dispelled this concern. The once-weekly dosing has the potential for improved patient experience, as well as more consistent use, persistence, and clinical outcomes. The consistent rates of hypoglycemia between insulin icodec and the once-daily glargine are very encouraging ultimately meaning patients can reap the benefits of once-weekly insulin without any added hypoglycemic fear.

  • The randomized, open-label, two-period crossover trial enrolled 43 individuals with type 2 diabetes. On average, patients were 56 years old, had a BMI of 28.5 kg/m2, had a diabetes duration of 13.5 years, and an A1c of 7.2%. Following a run-in period, patients were randomized between insulin icodec (once-weekly for six weeks) or IGlar U100 (once-daily for 11 days). This was followed by a wash-out period before patients were crossed over to the other treatment. In each of the treatment periods, patients were twice induced with hypoglycemia: once with a double dose and once with a triple dose.
    • Patients were treated with IV glucose or IV insulin to attain euglycemic blood sugar levels at 100 mg/dL after they received either the double or triple dose of insulin. To induce hypoglycemia, IV glucose was stopped and development of hypoglycemia observed. Hypoglycemia symptoms and counter regulation was evaluated at 70 mg/dL, 54 mg/dL, and 45 mg/dL, after which patients were treated to restore normoglycemia.
  • Notably, icodec demonstrated non-inferiority on hypoglycemic events compared to glargine, and was, in some cases, associated with lower rates of hypoglycemia than glargine. The table below describes the percent of patients in each trial group who dropped below the stated hypoglycemic thresholds. The “mean nadir” is the average blood sugar following the stated dose. The p-value is assessing whether there is any statistically significant difference in hypoglycemia between the icodec and glargine groups.

 

Double Dose

Triple Dose

 

icodec

glargine

p-value

icodec

glargine

p-value

<54 mg/dL

40%

36%

p=0.63

53%

70%

p=0.14

<45 mg/dL

5%

7%

p=0.63

3%

25%

p=0.03*

Mean nadir

58 mg/dL

59 mg/dL

p=0.07

56 mg/dL

52 mg/dL

P<0.001*

  • On the physiologic response, hypoglycemic symptoms were similar between icodec and glargine. Responses in glucagon and growth hormone also tracked comparably between the two groups. Interestingly though, icodec induced a greater response among counterregulatory hormones like adrenaline and cortisol, and non-adrenaline responses also trended in that direction. Although there was much discussion during Q&A as to why this might be the case, there was no conclusive answer. Ultimately, perhaps the most important finding is that outside the induced hypoglycemic episodes, there were no severe hypoglycemic events, nor were there any severe adverse events among patients taking once-weekly icodec.

Jam-packed symposium highlights latest developments in glucose-responsive, oral, and once-weekly insulins; Dr. Michael Weiss shares unpublished data on his lab’s unimolecular, chemically modified glucose-responsive insulin

In three back-to-back sessions, Dr. Michael Weiss (Indiana University), Dr. George Grunberger (Grunberger Diabetes Institute, Michigan), and Dr. Harpreet Bajaj (LMC Diabetes & Endocrinology, Canada) provided an overview of the pipelines for glucose-responsive, oral, and once-weekly insulins. 2022 is a momentous year for insulin, marking the 100th anniversary of the first insulin treatment. Given this, we have heard many speakers on the conference circuit provide reflections on the evolution of insulin, many of whom have also published these reflections in review articles (ex. Med, Nature Medicine, and the Journal of Endocrinology). Among the three innovative insulin classes in development – glucose-responsive, oral, and once-weekly insulins – once-weekly insulins are the farthest along in development, with Novo Nordisk’s insulin icodec having released topline data for multiple phase 3 trials and Lilly’s basal insulin Fc (BIF) having initiated phase 3 trials. Still, Dr. Weiss and Dr. Grunberger made clear that progress is being made on glucose-responsive and oral insulins. Currently, researchers are pursuing several creative approaches, and in fact, Oramed’s oral insulin ORMD-0801 is currently in two phase 3 trials: ORA-D-013-1 and ORA-D-013-2, which are expected to complete in September 2022 and November 2023, respectively. Notably, Dr. Weiss shared unpublished data on his group’s glucose-responsive insulin that demonstrated significant changes in activity based on glycemic conditions in mice. All the speakers commented that these “third generation” insulins, as Dr. Weiss called them, have the potential to improve patients’ consistency of insulin use and quality of life, along with potentially reducing clinical inertia and improving glycemic management.

  • Dr. Weiss briefly reviewed classes of glucose-responsive insulin before diving into positive, unpublished data on his group’s modified insulin molecule that demonstrates glucose-responsive activity. While the aim of oral and weekly insulins is not necessarily to achieve superior efficacy compared to currently available insulins, Dr. Weiss explained that the clinical goals for glucose-responsive insulins is to reduce hypoglycemia, mean glucose levels, and glycemic variability, while also enhancing consistency of treatment use and quality of life. He listed five types of glucose-responsive insulins: (i) mechanical (ex. CGM-coupled insulin pumps and closed loop systems); (ii) polymer-based (glucose-responsive lectins and matrices); (iii) bio-inspired carriers (albumin and GLUT channels); (iv) metabolic clearance targeting (mannose receptor); and (v) unimolecular glucose-responsive insulins (his lab’s approach). Due to time constraints, he directed people to his 2021 review on smart insulin delivery technology in Diabetologia
    • Sharing unpublished data, Dr. Weiss explained the mechanism of his groups single-molecule glucose-responsive insulin (GRI). He characterized the design of his group’s GRI as “mechanism-based mimicry of induced fit.” In other words, the drug mimics the induced fit mechanism that native insulin uses to bind to the insulin receptor. In contrast to native insulin, the GRI contains two modifications: (i) an N-terminal glucose-binding element, and (ii) a C-terminal ligand that binds the N-terminal glucose-binding element in the absence of glucose. When the C-terminal ligand binds to the N-terminal glucose-binding element, the GRI cannot activate the insulin receptor. Under high glucose conditions, glucose binds the N-terminal glucose-binding element, instead of the C-terminal ligand, allowing for the GRI to bind and activate the insulin receptor. Dr. Weiss published a proof-of-concept of this technology in the Proceedings of the National Academy of Science in July 2021. As shown below, Dr. Weiss’ unpublished data of glucose clamp studies in mice, the GRI (blue bars) demonstrated a marked reduction in activity during euglycemic and hypoglycemic conditions compared to insulin lispro (red bars). Dr. Weiss believes the simplicity of this GRI could be viable for long-acting, rapid-acting, once-weekly, and oral insulin.

  • Dr. Grunberger highlighted seven types of oral insulin in the pipeline. Oral insulin administration allows for intestinal insulin absorption, and the insulin is sensed by the liver first before going out to general circulation; in contrast, he explained, subcutaneous administration leads to hyperinsulinization of the entire body starting with peripheral circulation and ending with liver sensing. Dr. Grunberger then reviewed seven types of oral insulin: (i) epithelium microenvironment-adaptive nanoparticles; (ii) oral delivery of a D-DNP peptide + insulin hexamer; (iii) imine-linked covalent organic framework nanoparticles; (iv) orally delivered bile acid polymer nanocarriers of insulin; (v) glucose-responsive oral insulin delivery; (vi) swallowable device with an injection system (ex. SOMA robotic pill); and (vii) a luminal unfolding microneedle injector. Looking to oral insulins further along in development, Dr. Grunberger highlighted insulin tregopil, which completed a phase 3 trial in 2019 but failed to confer a significant A1c reduction over 24 weeks. Tregopil is now undergoing a phase 1/2 trial in type 2 diabetes. Farthest along is Oramed’s ORMD-0801, which initiated two phase 3 trials in January 2021: ORA-D-013-1 and ORA-D-013-2, which are expected to complete in September 2022 and November 2023, respectively. Concluding his talk, Dr. Grunberger said, “It’s too early to tell whether this is a field of dreams or the future.”
  • Dr. Bajaj (LMC Diabetes & Endocrinology) reviewed key milestones in the development of once-weekly insulins, highlighting Lilly’s basal insulin Fc (BIF) and Novo Nordisk’s insulin icodec. After contending that a once-weekly insulin would significantly improve consistency of treatment use, glycemic control, and quality of life for patients, Dr. Bajaj dove into the history of BIF. Notably, phase 1 trials have shown that BIF has a half-life of 17 days, and phase 2 trials presented at ADA 2022 found that BIF conferred a similar A1c reduction and Time in Range improvement compared to insulin degludec, though BIF was associated with a lower occurrence of hypoglycemia. BIF has initiated a phase 3 program, consisting of five QWINT trials, four in type 2 diabetes and one in type 1 diabetes. On insulin degludec, Dr. Bajaj briefly reviewed phase 2 results, showing that people switching from daily basal insulin to once-weekly insulin spent +1.9 hours/day in Range compared to insulin glargine. While full results from icodec’s phase 3 ONWARDS trials have yet to be released, Dr. Bajaj highlight that positive, topline results have been released for ONWARDS 1, ONWARDS 2, and ONWARDS 6. Dr. Bajaj concluded his thorough review with a few outstanding questions, calling on the need for more clinical data regarding the impact of once-weekly insulin on glycemic control, dosing algorithms, and treatment adherence to better understand how once-weekly insulin can be used in practice.

InRange study analysis finds comparable between-day and within-day Glycemic Variability (GV) with Toujeo and Tresiba; participants not achieving within-day GV target (≤36%) at baseline achieve average target at 12 weeks with both basal insulins

Building on the initial InRange readout at ATTD 2022, a poster in the poster hall offers detailed glucose variability outcomes for people with type 1 diabetes randomized to Toujeo (Gla-300) or Tresiba (IDeg-100) (105-LB). As a reminder, the InRange study was the first study comparing second-gen basal insulins to use Time in Range as a primary endpoint and determined that Toujeo is non-inferior to Tresiba regarding both Time in Range and glycemic variability (based on total coefficient of variation (CV)) outcomes in patients with type 1 diabetes. Specifically, at 12 weeks, participants achieved a Time in Range of 52% and 55% and total glycemic variability of 41% and 39.9% in the Toujeo and Tresiba arms, respectively (neither were significantly different). Expanding on total glycemic variability data presented at ATTD 2022, this poster reported within-day and between-day glucose variability in the Toujeo (n=172) and Tresiba (n=171) treatment groups. While total glycemic variability is generally the metric reported (and is the metric included in AGP reports), it is a product of within-day and between-day glycemic variability and can obfuscate differences in these measures. Because of this, the researchers expanded their analysis to look specifically at within-day and between-day glycemic variability.

  • Both within-day and between-day glycemic variability were not statistically significantly different between the two treatment groups. This was true regardless of whether patients met the consensus target for glycemic variability (≤36% CV) at baseline. At week 12, the average between-day glycemic variability was 17% and 18% for Toujeo and Tresiba, respectively (statistically equivalent). Likewise, at the study endpoint, within-day glucose CV was 33.5% for those using Touejo and 34.37% for those using Tresiba, which were also statistically equivalent. Notably, those who were not achieving the within-day glycemic variation target of ≤36% at baseline did, on average, achieve this goal after 12 weeks of Toujeo or Tresiba, ending with an average CV of 33.7% and 35.4%, respectively.
  • During a Sanofi-sponsored symposium, Dr. Jeremy Pettus (UCSD) discussed the InRange study and highlighted benefits of the Toujeo Max Solostar (Gla-300). He noted that although there was no significant difference in Time in Range, glycemic variability, rates of hypoglycemia, or HbA1c at week 12 between Toujeo and Tresiba, the results are important, as they counter the sentiment sometimes expressed in “the community” (as he put it) that Toujeo is not as good as Tresiba. Furthermore, he championed the patient convenience that the Toujeo Max Solostar pen offers as an added benefit beyond the glycemic outcomes. In particular, he highlighted Toujeo’s 900-unit capacity and 56-day shelf life, meaning that the pen would last long enough even for patients on a low dose and that no insulin would be wasted. Dr. Pettus emphasized that “this matters to patients.” He also noted Sanofi’s focus on increasing accessibility for Toujeo, the 91% coverage rate among commercial patients, and the savings program for uninsured patients.

Phase 3 PRONTO-Peds trial finds that ultrarapid insulin lispro is safe and well-tolerated compared to lispro in children and adolescents with type 1 diabetes

Dr. R Paul Wadwa (University of Colorado) presented results from the phase 3 PRONTO-Peds study showing that ultrarapid insulin lispro is non-inferior to insulin lispro in children and adolescents with type 1 diabetes. While we were surprised, offhand, that it wasn’t superior to lispro, this was a strong result. The study randomized 716 pediatric patients to three trial arms: (i) double blind ultrarapid insulin lispro injected 0-2 min prior to meals (mealtime); (ii) double blind lispro injected 0-2 min prior to meals (mealtime); or (iii) open-label ultrarapid insulin lispro injected 0-20 minutes after the start of the meal (postmeal). Participants remained on pre-study basal insulin (degludec, detemir, or glargine) and were followed for 26 weeks to assess the primary endpoint – change from baseline A1c. Mealtime ultrarapid insulin lispro was non-inferior to mealtime lispro with an estimated treatment difference of -0.02% (95% CI: -0.17, 0.13) in the double-blind trial and -0.02% (95% CI: -0.20, 0.17) for postmeal ultrarapid insulin lispro compared to mealtime lispro. In terms of secondary outcomes, postprandial glucose was lower with ultrarapid insulin lispro both one hour after breakfast (p<0.001) and one hour after dinner (p=0.006) though not enough to prompt a different A1C. Notably, there were no major significant differences in rates of hypoglycemia nor adverse events between the treatment arms. Participants in the mealtime ultrarapid lispro group did have slightly higher rates of injection site reactions (7.9%) leading to two study discontinuations compared to the postmeal ultrarapid group (2.9%) and mealtime lispro group (2.7%); though all reactions were deemed mild or moderate in severity.

  • As a reminder, ultrarapid insulin lispro was initially approved by the FDA in June 2020 for use in adults with either type 1 or type 2 based on the PRONTO-T1D and PRONTO-T2D studies under the brand name Lyumjev. In August 2021, the FDA approved an expanded indication for the ultra-rapid acting insulin Lyumjev in insulin pumps. Recall that one of the major advantages of Lyumjev is that it’s indicated for use within 20 minutes of eating, which is a major benefit for children and adolescents who may not be able to exactly dose their insulin prior to each meal given the unpredictability of daily life. Thus, the positive results from PRONTO-Peds suggest that a pediatric indication could be forthcoming, which could ultimately lead to greater quality of care for this young patient population.
  • At ATTD 2022, there was significant support and excitement for Lyumjev.  There, Dr. Andreas Liebl suggested that “nearly every person with [MDI-dependent diabetes] would benefit from Lyumjev.” Moreover, the data from dQ&A suggest that Lyumjev significantly outperforms Humalog, the comparator in the PRONTO trials, and Fiasp, the other available ultra-rapid acting insulin on the market. dQ&A panel participants consistently rank consistency, amount of glucose control provided, and time to onset as the top three metrics that PWD view as the most important for mealtime insulin.

Insulin aspart is a safe and effective bolus insulin for pregnant women with type 1 diabetes

Dr. Elisabeth Mathiesen (University of Copenhagen) shared data suggesting that there are no significant differences in maternal, perinatal, or neonatal outcomes between insulin aspart and other bolus insulins. Dr. Mathiesen framed her work by noting that there has historically been some concern that insulin analogues like aspart cause congenital malformations when taken by pregnant women; thus, there is significant incentive to better understand safety outcomes in this population. The EVOLVE study was a prospective multinational cohort trial of 1840 pregnant women with type 1 diabetes. The post hoc analysis used propensity match scoring to assess differences in major congenital malformations, perinatal, and neonatal death between women on insulin aspart and other bolus insulins. While the insulin aspart group was smaller than other bolus insulins, the other baseline characteristics were similar across groups, with a diabetes duration of between 15 and 18 years, an A1C of 7%, and a BMI of 25 kg/m2. The results show that there is no significant difference in major congenital malformations, perinatal, and neonatal death, nor are there differences in maternal major hypoglycemia, abortion, preterm delivery, preeclampsia, or a large for gestational age at birth babies. Thus, Dr.  Mathiesen concludes that insulin aspart is an effective bolus insulin for use during pregnancy.

Posters – Insulin Therapies

Title

Authors

Details + Takeaways

Safety, Pharmacokinetics, and Pharmacodynamics of Once-Weekly Basal Insulin Fc (BIF) in Japanese Patients with Type 2 Diabetes: A Phase 1 Open-Label, Randomized, Active-Controlled Trial

Risa Nasu, Tomonori Oura, Kenji Ohwaki, Makoto Imori, Kenichi Furihata

  • Phase 1 study assessing safety, pharmacokinetics, and pharmacodynamics of BIF vs once-daily insulin degludec in Japanese patients with T2D (n=28) over 6 weeks
  • Patients with T2D on multiple once-weekly doses of basal once-weekly insulin (BIF) were well tolerated with no clinically significant safety differences from insulin degludec
  • Treatment-emergent adverse event incidence was low for both BIF (n=4; 22.2%) and insulin degludec (n=2; 20.0%) and found to be unrelated to BIF treatment
  • BIF treatment was associated with improved glycemic control
  • Long periods of hypoglycemia were not observed with BIF treatment

Efficacy and Safety of Ultrarapid Lispro (URLi) vs. Humalog (Lispro) in Chinese Pediatric Patients with Type 1 Diabetes: A Subpopulation Analysis of PRONTO-Peds Study

Wei Gu, Yuxin Yang, Yuan Yuan, Feihong Luo

  • Evaluating the safety and efficacy of ultrarapid insulin lispro vs lispro in Chinese pediatric patients with T1D over a 26-week treatment period (URLi n=7; lispro n=11).
  • Mealtime and post-meal URLi and lispro increased A1c by 1.1% and 1.2%, respectively for the Chinese subpopulation, which was similar to the overall population

Impact of Hypoglycemia on Insulin Titration and Fasting Plasma Glucose in Basal Insulin-Treated Type 2 Diabetes: A Subanalysis of the SoliMix Trial

Francesco Giorgino, Julio Rosenstock, Robert Ritzel, Oguzhan Deyneli, Augstina Alvarez, Elisabeth Souhami, Lyde Melas-Melt, Rory J. McCrimmon

  • Utilizing data from the SoliMix trial, comparing iGlarLixi (n=443) with premix basal insulin aspart U30 (n=444) in adults with T2D, the trial aimed to understand if hypoglycemia influenced insulin titration and fasting plasma glucose reductions
  • Patients with T2D using Premix BIAsp 30 had more frequent hypoglycemia than with iGlarLixi, resulting in lower insulin doses that reduced insulin titration and increased FPG and A1c

Advancing Therapy in Basal Insulin-Treated Type 2 Diabetes: Exploratory Analysis of the SoliMix trial by Baseline HbA1c, Insulin Dose, and BMI

Philip Home, Rory J. McCrimmon, Julio Rosenstock, Matthias Blüher, Katrin Pegelow, Lydie Melas-Melt, Khier Djaballah, Francesco Giorgino

  • Analyzing the SoliMix trial, which found lower ADA Level 1 and 2 hypoglycemia incidence and event rates, greater A1c reductions, and greater body weight benefit with iGlarLixi vs premix BIAsp 30 in adults with T2D
  • Results of SoliMix were analyzed by baseline A1c, insulin dose, and BMI, and results were found to be consistent across baseline A1c and BMI, but not insulin dose

Advancing Therapy in Basal Insulin-Treated Type 2 Diabetes: Exploratory Analysis of the SoliMix Trial by Baseline Age, Diabetes Duration, and Renal Function

Philip Home, Julio Rosenstock, Francesco Giorgino, Matthias Blüher, Khier Djaballah, Katrin Pegelow, Lydie Melas-Melt, Rory J. McCrimmon

  • Results of the SoliMix trial were analyzed by baseline age, diabetes duration, and renal function
  • No differences in treatment effect were seen across subgroups for the lesser insulin dose increments with iGlarLixi vs. BIAsp 30 (p>0.10)
  • Insulin doses ≥30 units were associated with lower rates of hypoglycemia
  • Results were unaffected by baseline age, diabetes duration, and renal function across

Greater HbA1c Reduction with iGlarLixi vs. Insulin Glargine and Lixisenatide Regardless of Levels at Screening: Subgroup Analysis of the LixiLan-O Asia Pacific Trial

WenYing Yang, Dong Xiaolin, Guoye Yuan, Ming Liu, Jianzhong Xiao, Shenghong Gu, Lijuan Chen, Elisabeth Souhami

  • 24-week LixiLan-O-AP RCT (n=878) found that iGlarLixi led to greater A1c improvement than insulin glargine (iGlar) or lixisenatide (Lixi) alone in Asian Pacific adults with T2D advancing from 1-2 oral glucose-lowering drugs
  • Analysis found better glycemic control with iGlarLixi than with iGlar or Lixi alone regardless of A1C level at screening

Glucose Variability with Second-Generation Basal Insulin Analogs Glargine 300 U/mL and Degludec 100 U/mL, Evaluated by CGM in People with T1D—The InRange Randomized Controlled Trial

Richard M. Bergenstal, Steven Edelman, Pratik Choudhary, Thomas Danne, Eric Renard, Jukka WesterBacka, Bhaswati Mukherjee, Pascaline Picard, Valerie Pilorget, Tadej Battelino

  • InRange RCT (n=343) compared the impacts of glargine-300 and insulin degludec-100 on total glucose coefficient of variation in adults with T1D for 12 weeks using 20-day CGM profiles
  • Gla-300 and IDeg-100 have similar within-and between-day glucose variability for people with T1D

Ultrarapid Lispro (URLi) Improved Postprandial Glucose (PPG) Excursion vs. Humalog (HL) in Predominantly Chinese Adult Patients with Type 2 Diabetes (T2D)

Jian Zhou, Si Chen, Jie Cheng, Li Shen, Jiankun Zhu, Ying Lou, Yuqian Bao, Weiping Jia

  • Study evaluating the efficacy of ultrarapid insulin lispro (URLi) (n=395), a novel formulation of insulin lispro, against Humalog (n=200) in predominantly Chinese adult patients with T2D
  • URLi was superior to humalog in controlling 1-hour (4.8 mmol/L vs. 5.6 mmol/L, p<0.001) and 2-hour post-prandial glucose (PPG) excursions (6.2 mmol/L vs. 7.4 mmol/L, p<0.001) at Week 26
  • URLi vs. humalog in a basal-bolus regimen demonstrated noninferior glycemic control as assessed by A1c and superior control on PPG excursions

Patient and Physician Experience of Hypoglycemia during Basal Insulin (BI) Titration in Type 2 Diabetes (T2D) in the U.S.

Stewart B. Harris, Kamel Mohammedi, Monica Bertolini, Valery Walker, Maureen H. Carlyle, Fang L. Zhou, Jochen Seufert, John E. Anderson

  • Study investigating the physician and patient perspectives of hypoglycemia during basal insulin (BI) titration by mailing a survey to adults (n=416) with T2D and physicians (n=386) who treated ≥30 patients with T2D
  • Most physicians reported discussing hypoglycemia signs/symptoms (93%) and how to titrate BI in response to blood glucose (BG) levels (81%) with patients
  • Among patients who experienced hypoglycemia (49%), 57% felt very/extremely confident titrating basal insulin during hypoglycemia. Only 35% met fasting blood glucose targets

Physician Perspectives and Experiences with Basal Insulin Titration in Type 2 Diabetes: A U.S. Cross-Sectional Survey

Stewart B. Harris, Jochen Seufert, Monica Bertolini, Valery Walker, John C. White, Fang L. Zhou, Kamel Mohammedi, John E. Anderson

  • Study investigating the physician experience with basal insulin titration through a one-time mailed survey
  • Out of the 386 respondents, 86% of physicians explained titration to all/most of their patients new to basal insulin
  • Only 27% of physicians expected patients to self-manage titration; 60% managed titration for them, and 13% did both
  • Most (84%) physicians had in-office titration education by healthcare providers
  • The majority expressed concerns about patients’ abilities to follow titration algorithms (79%), and to monitor blood glucose effectively (66%), as well as lack of engagement in the titration process (72%)

Effect of Adding Computerized Insulin Dose Adjustment Algorithms (CIDAA) to a Remote Patient Monitoring (RPM) Program on A1c Levels

Juvairiya S. Pulicharam, Josh Davidson, Mayer B. Davidson

  • Study contextualizing the effects of adding Computerized Insulin Dose Adjustment Algorithms for individuals enrolled in a remote patient monitoring program
  • There were no differences among baseline A1Cs but significant differences from baseline
  • A1C levels fell twice as much without increases in hypoglycemia in poorly controlled T2D patients taking insulin whose providers received recommendations for dose changes based on the algorithm every 2-3 weeks compared with remote patient monitoring alone or usual care

Barriers and Concerns Regarding Initiating Insulin Therapy among People with Type 2 Diabetes

Julia Stevenson, Emily Xu, Erik Monroy-Spangenberg, Emily Ye, Richard Wood

  • Questionnaire-based study investigating the perceptions and barriers towards insulin therapy among people (n=1939) with T2D
  • Nearly three quarters (73%) cited worries that once they start taking insulin, they would have to take it for the rest of their life
  • Over half (58%) felt that starting insulin would mean they were not adequately taking care of their diabetes, and 44% felt upset when their doctor first brought up insulin therapy
  • Concerns about weight gain (49%), hypoglycemia (48%), and hesitancy toward injections (43%) were also common
  • Only 22% felt confident that taking insulin would improve their diabetes management, and 69% believed they could improve their diabetes management without insulin
  • The most common preferred steps instead of starting insulin were altering diet (70%) and exercising more frequently (64%); 43% would rather start taking a new oral drug, while fewer would try a non-insulin injectable medication (22%)

Glycemic Control in People with Type 2 Diabetes (PWT2D) Switching from NPH to Insulin Glargine 300 U/mL (Gla-300): REALI Pooled Database

Dirk Müller-Wieland, Nick Freemantle, Riccardo C. Bonadonna, Celine Mauquoi, Gregory Bigot, Mireille Bonnemaire, Pierre Gourdy, Didac Mauricio

  • Analysis of REALI database to determine the effectiveness of Gla-300 in people with T2D switching from human insulin
  • A1c markedly improved by 0.9% and fasting plasma gluocse decreased by 35 mg/dL after a 24-week Gla-300 therapy
  • People with T2D, previously uncontrolled on basal insulin, benefited from switching to Gla-300 in terms of A1c improvement, and this was especially observed in those previously treated with human insulin

Association between Analogue Compared with Human Insulin and the Outcomes of Mortality, Hospitalization, MACE, and Hypoglycemia in Hemodialysis Patients with Type 2 Diabetes: The ARO Research Initiative

Thomas Ebert, Nosheen Sattar, Marni Greig, Claudia Lamina, Marc Froissart, Kai-Uwe Eckardt, Jürgen Floege, Florian Kronenberg, Peter Stenvinkel, David C. Wheeler, James Fotheringham, Sheffield

  • Multi-year study comparing outcomes to analogue and human insulin therapy for patients with T2D in medical care facilities across European countries
  • Patient data was gathered for 3 years and analyzed to generate Cox-proportional hazards, MACE, and confirmed hypoglycemic events
  • Patients with T2D on hemodialysis (n=713), compared to human insulins (n=733), analogue insulins were associated with better clinical outcomes, although hypoglycemia rates were increased

Digital-Tool-Supported Basal Insulin (BI) Titration: Real-World Effectiveness of My Dose Coach (MDC) in People with Type 2 Diabetes (T2D) in Colombia

Ana Maria Gomez, Alex Ramirez Rincon, Karen Feriz, Carlos F. Salamanca, Liliana P. Silva

  • Study evaluating real-world outcomes for people with T2D on basal insulin (BI) therapy who registered for the My Dose Coach (MDC) application—an app that helps people adjust to their BI doses
  • Total of 505 MDC users from Colombia were included and upon analysis, revealed that 81% users achieved individualized fasting blood glucose targets and reached this target in 10.3 days, on average
  • 14% of users experienced a hypoglycemic event
  • People with T2D from Colombia using MDC were able to successfully titrate their BI doses and achieve FBG target with low risk of hypoglycemia

Real-World Performance of IDegAsp over Six Years in Asian-Indian Population

Jothydev Kesavadev, Banshi D. Saboo, Arun Shankar, Gopika Krishnan, Anjana Basanth, Asha Ashik, Ashwin David, Sunitha Jothydev

  • Multi-centric, retrospective observational study evaluating the long-term efficacy and safety of IDegASP (co-formulation of ultra-long-acting insulin Degludec and rapid acting insulin Aspart) in T2D
  • 535 individuals with T2D previously on insulin regimen were switched to IDegAsp over a five-year period and saw significant improvements from observed baseline in A1c
  • IDegAsp offers a less complex, more physiological, safe, and effective solution for glycemic control

Real-World Persistence, Adherence, Health Care Resource Utilization (HRU), and Hypoglycemia in People with Type 2 Diabetes (T2D) Continuing the 2nd-Generation (2nd-Gen) Basal Insulin (BI) (Glargine 300 U/mL [Gla-300]) vs. Switching to a 1st-Generation (1s

Steven Edelman, Jennifer D. Goldman, Daniel C. Malone, Ron Preblick, Kovida Munaga, Xuan Li, Jasvinder Gill, Sumana Gangi

  • Retrospective observational study to compare persistence, adherence, health care resource utilization, ER visits, and hypoglycemia among adults with T2D who continued with 2nd-gen basal insulin Gla-300 vs. those who switched to a 1st-gen basal insulin
  • Continuing on Gla-300 was associated higher persistence and significantly lower hypoglycemic events (26.2 vs. 42.8/100 person-years) and ER visits (all-cause: 100.5 vs. 146.8 per 100 person-years) vs. switching to a 1st-gen basal insulin

Treatment Satisfaction and Health Status in People with T2D Treated with Insulin Glargine 300 U/mL (Gla-300): Patient-Reported Outcomes (PRO) from ATOS Study

Niaz Khan, Amir Tirosh, Anil Bhansali, Hernando Vargas-Uricoechea, Stewart B. Harris, Aude Roborel De Climens, Maria Aileen N. Mabunay, Mathieu Coudert, Valerie Pilorget, Gagik R. Galstyan

  • One-year observational study investigating patient satisfaction and health changes in the ATOS trial of 3,931 insulin-naive patients on Gla-300 via PRO questionnaires (DTSQ and EQ-5D-3L)
  • Mean age was 57.5 years, duration of diabetes was 10.1 years and baseline A1c was 9.3 ±1.0%
  • Treatment satisfaction improved over time (DTSQs score of 21.7 at baseline to 29.8 and 31.3 at Month 6 and 12, respectively) and perceived frequency of hyperglycemia decreased over 12 months
  • EQ-5D-3L results showed that the proportion of people with better health status increased over time
  • Overall, results showed that initiating Gla-300 in insulin-naïve people with T2D across multiple geographic regions improved treatment satisfaction and health status

Treatment Satisfaction in People with T2D Switched from Basal Insulin (BI) to Insulin Glargine 300 U/mL (Gla-300): Patient-Reported Outcomes from ARTEMIS-DM Study

Khalid Alrubeaan, Emma G. Wilmot, Aude Roborel De Climens, Maria Aileen N. Mabunay, Valerie Pilorget, Baptiste Berthou, Gustavo Frechtel, Bipin Sethi

  • Compared treatment satisfaction (using ITSQ) in patients who switched to Gla-300 in Asia vs. Middle East Africa vs. Latin America in the ARTEMIS study (n=362)
  • Mean ± SD ITSQ total score increased from baseline 76.30 ± 15.63 to 83.51 ± 13.29 and 84.42 ± 13.15 at weeks 12 and 26, respectively
  • Improvements were noted for all ITSQ domain scores, especially glycemic control scores (baseline: 65.83; mean change from baseline to Week 12: 15.12 and Week 26: 16.06)
  • Despite similar mean baseline scores (73.08 to 79.31), LS mean change in ITSQ total score was numerically greater in Latin America (10.86) and Middle East & Africa (10.40) compared with Asia (1.56) at Week 26
  • These results suggest that switching to Gla-300 in people with T2D uncontrolled on BI led to improvement in treatment satisfaction and specifically in glycemic control in a diverse population from multiple geographic regions

Improving Decision Making by Clinicians for Initiation of Insulin in Adults with Type 2 Diabetes Using Simple Python Program

Om J. Lakhani

  • Developed python code computer program to support insulin initiation (dosage and type) in insulin-naïve T2Ds based on EHR data from 500 patients
  • Insulin was initiated in 112 of 500 patients by the ten volunteer clinicians, while the computer recommended insulin initiation in 163 patients
  • There were no cases in which the clinician-initiated insulin, and the computer recommended against it
  • The computer program led to a significant 28% improvement in insulin initiation (p=0.037)

Trends in Insulin Use and Glycemic Control in U.S. Adults with Diabetes, 1988-2020

Siddharth Venkatraman, Elizabeth Selvin, Michael Fang

  • Investigated the relationship between insulin or oral diabetes medication use and A1C level using NHANES data
  • Examined 8,658 non pregnant adults with diabetes, who were >20 years old
  • Glycemic control improved among users of oral medications (p<0.001) but was unchanged among insulin users (p=0.87)
  • Mexican American insulin users had a decrease in glycemic control (25.1% in 1988-94 to 9.9% in 2013-20, p=0.004) and had poorer glycemic control, overall, as compared to their non-Hispanic White or Black counterparts

Progression of Position of Basal Insulin in the Treatment of Type 2 Diabetes – A Real-World Analysis

David Schapiro, Alexandra Meeks, Dongju Liu, Felicia Gelsey, Rattan Juneja, Magaly Perez-Nieves, Ahong Huang

  • Retrospective study to assess whether clinicians were prioritizing GLP1-s and SGLT-2s for CV and renal benefits in 6 million+ patients with T2D
  • The proportions of patients starting basal insulin who had already started on GLP-1 and SGLT-2 grew from 14.8% and 11.4% in 2015 to 25.2% and 20.5% in 2019
  • Clinicians started other patients on GLP-1s and SGLT-2s earlier and earlier relative to starting basal insulin

Real-World Persistence, Adherence, Health Care Resource Utilization (HRU), and Hypoglycemia in People with Type 2 Diabetes (T2D) Continuing the 2nd-Generation (2nd-Gen) Basal Insulin (BI) (Glargine 300 U/mL [Gla-300]) vs. Switching to a 1st-Generation (1s

Steven Edelman, Jennifer D. Goldman, Daniel C. Malone, Ron Preblick, Kovida Munaga, Xuan Li, Jasvinder Gill, Sumana Gangi

  • Retrospective observational study to compare persistence, adherence, health care resource utilization, ER visits, and hypoglycemia among adults with T2D who continued with 2nd-gen basal insulin Gla-300 vs. those who switched to a 1st-gen basal insulin
  • Continuing on Gla-300 was associated higher persistence and significantly lower hypoglycemic events (26.2 vs. 42.8/100 person-years) and ER visits (all-cause: 100.5 vs. 146.8 per 100 person-years) vs. switching to a 1st-gen basal insulin

Treatment Satisfaction and Health Status in People with T2D Treated with Insulin Glargine 300 U/mL (Gla-300): Patient-Reported Outcomes (PRO) from ATOS Study

Niaz Khan, Amir Tirosh, Anil Bhansali, Hernando Vargas-Uricoechea, Stewart B. Harris, Aude Roborel De Climens, Maria Aileen N. Mabunay, Mathieu Coudert, Valerie Pilorget, Gagik R. Galstyan

  • One-year observational study investigating patient satisfaction and health changes in the ATOS trial of 3,931 insulin-naive patients on Gla-300 via PRO questionnaires (DTSQ and EQ-5D-3L)
  • Mean age was 57.5 years, duration of diabetes was 10.1 years and baseline A1c was 9.3 ±1.0%
  • Treatment satisfaction improved over time (DTSQs score of 21.7 at baseline to 29.8 and 31.3 at Month 6 and 12, respectively) and perceived frequency of hyperglycemia decreased over 12 months
  • EQ-5D-3L results showed that the proportion of people with better health status increased over time
  • Overall, results showed that initiating Gla-300 in insulin-naïve people with T2D across multiple geographic regions improved treatment satisfaction and health status

Type 1 “Cures” and Adjunctive Therapies

Low-dose empagliflozin increases Time in Range by +3.1 hours/day in patients with type 1 diabetes on AID with no DKA events in 14-day crossover RCT

Dr. Melissa-Rosina Pasqua (McGill University) shared data showing that low-dose empagliflozin increases Time in Range and reduces insulin requirements in patients with type 1 diabetes on AID. This is an important topic of conversation given the dearth of adjunctive glucose therapies available to patients with type 1 diabetes. Although the positive glycemic, cardiovascular, and renal benefit of SGLT2s is undisputed in people with type 1 diabetes, there is an associated increase in risk of diabetic ketoacidosis that has limited its use in this population. Notably, low-dose empagliflozin – 2.5 mg compared to the 10 mg tablets that are sold for type 2 diabetes – has been identified as an exception, with no associated DKA risk. Thus, this trial utilized lower doses to assess whether the SGLT-2 inhibitor could improve Time in Range, insulin dose, and hypoglycemia for patients on AID.

The double-blind, crossover study enrolled 24 participants with A1cs between 7-10% who were randomized to three groups: (i) empagliflozin 2.5 mg; (ii) empagliflozin 5 mg; and (iii) placebo. Participants were followed for 14 days. On average, the patients were 33 years old, had a long duration of diabetes (21 years), a baseline A1c of 8.1%, and BMI of 28.8 kg/m2. In order to monitor ketone levels, participants’ ketone levels were assessed each morning for the 14 days that they were enrolled in the trial.

  • Low-dose empagliflozin significantly improved Time in Range by +2.6-3.1 hours/day relative to placebo. Specifically, at 14 days, Time in Range was 59% in the placebo arm, 72% in the empagliflozin 2.5 mg arm, and 70% in the empagliflozin 5 mg arm. This is a particularly impressive improvement given that these participants were on AID across all arms, meaning that this is the additive benefit of SGLT-2. Notably, there did not appear to be a dose dependent effect across empagliflozin doses, with 2.5 mg and 5 mg delivering similar benefit. As is usually the case, the primary driver in Time in Range improvements came primarily from decreases in Time Above Range. Time Below Range (hypoglycemia) was already quite low, likely due to use of the AID system, but it is worth noting that the 5 mg empagliflozin arm did see slightly higher rates of daytime hypoglycemia (+7 minutes/day). However, participants were still well below the threshold of 4% Time Below Range that is recommended in the practice guidelines. These positive CGM outcomes were accompanied by a reduction in insulin requirements for those on the low-dose SGLT-2.

Time in Range Outcomes at 14 days

 

Placebo

Empagliflozin 2.5 mg

Empagliflozin 5 mg

Time in Range

59%

72%

70%

Daytime Time in Range (6h00-24h00)

58%

71%

68%

Nighttime Time in Range (0h00-6h00)

61%

74%

75%

Time Above Range (≥180 mg/dL)

40%

27%

28%

Time Below Range (≤70 mg/dL)

1.o%

0.8% (p=ns)

1.5%

Insulin Dose Outcomes

 

Placebo

Empagliflozin 2.5 mg

Empagliflozin 5 mg

Insulin doses (units/day)

59

54

53

Basal insulin dose (units/day)

34

30

31

Bolus insulin dose

26

24 (p=ns)

23

  • There were no serious adverse events reported in the study, nor were there any episodes of diabetic ketoacidosis. This is a major win for empagliflozin and adds further evidence for the low risk of DKA at smaller doses. Rates of other side effects were low with increased urination, increased thirst, and nausea reported at slightly higher rates in the 5 mg group than 2.5 mg empagliflozin or placebo.

 

Placebo

Empagliflozin 2.5 mg

Empagliflozin 5 mg

Morning ketone level (mmol)

0.15

0.15 (p=ns)

0.17 (p=ns)

Number of days with ketone levels ≥1.5 mmol

1

2

1

  • Dr. Pasqua noted that the size and scope of the trial may limit its external validity. She noted that a larger, longer study is required to ensure long-term efficacy and safety of SGLT-2 use in a type 1 population. Also, the careful ketone monitoring required increased surveillance that is not entirely applicable to the real world. Dr. Pasqua believes that a continuous ketone monitor, the likes of which Abbott announced earlier this week, could improve safety among these patients.

Six-month results from real-world STEMT1 trial (n=18) suggest that once weekly semaglutide 1 mg improves weight loss and glycemic control in people with type 1 diabetes

A poster by Dr. Jonathan Mertens (University of Antwerp, Belgium) and colleagues demonstrated promising results for once-weekly semaglutide 1 mg on weight change and metabolic control in people with type 1 diabetes in the STEMT1 trial (751-P). In this real-world study, 55% of participants were male with a mean age of 46 years, mean diabetes duration of 30 years, and mean A1c of 7.4%. At baseline, mean BMI was 33 kg/m2 and 80% of subjects had obesity. Among 18 participants, mean bodyweight change over six months was -8.5 ± 7.8 kg, ranging from +1.5 to -24.7 kg. Relative weight loss of ≥5% was achieved in 60% of participants, while weight loss of ≥10% was achieved by 40% of participants. These weight loss findings align fairly closely with those seen in the SUSTAIN FORTE trial in type 2 diabetes, in which 51% of participants on semaglutide 1 mg achieved ≥5% weight loss and 23% of those on semaglutide 1 mg achieved ≥10% weight loss over 40 weeks. In the STEMT1 trial, three participants experienced mild weight gain, ranging from 0.2 kg to 1.5 kg. In terms of glycemic control, mean A1c reduction was -0.3 ± 0.7%, though six subjects (30%) did not show a reduction in A1c. Approximately one-third (35%) of participants achieved an A1c reduction of ≥0.5%; we note that this reduction is from an already low baseline A1c of 7.4%, bringing participants almost to the ADA’s generally recommended target A1c of 7% for adults with type 1 diabetes. Relative reduction in total daily insulin dose was 13.5%, though data on absolute insulin dose reductions or the percentage of participants achieving insulin dose reductions was not shared. There was a moderate correlation between weight loss and A1c reduction (r=0.52, p=0.020). Notably absent from the poster were data on safety and tolerability, though the poster notes that GI tolerance was evaluated at every contact during follow-up. Overall, the researchers concluded that adding once weekly semaglutide 1 mg in people with type 1 diabetes was safe, well-tolerated, and resulted in promising effects on weight, glycemic control, and total daily insulin requirement. We’re encouraged to see positive results from the first trial of blockbuster type 2 diabetes therapy Ozempic in type 1 diabetes, particularly in people with a long duration of diabetes and good baseline glycemic control. We look forward to seeing longer term results from the STEMT1 trial, and we hope this will inspire further research evaluating GLP-1s in type 1 diabetes, such as the SEMA-AP study assessing semaglutide as an adjunct to closed-loop therapy.

Real-world retrospective analysis of SGLT-2 treatment in type 1 diabetes finds no DKA events among 134 people over 222 patient-years

Dr. Elisabeth Stougaard (Steno Diabetes Center Copenhagen, Denmark) shared results from a Denmark quality assurance project that found no cases of diabetic ketoacidosis among 134 people treated with SGLT-2s (396-P). For background, in 2019 dapagliflozin was approved as an adjunctive therapy to treat type 1 diabetes in Europe and Japan, but in October 2021 AstraZeneca voluntarily withdrew dapagliflozin’s European type 1 diabetes indication. In the US, no SGLT-2s are indicated to treat type 1 diabetes due to concerns regarding SGLT-2-associated diabetic ketoacidosis (DKA). This retrospective observational study surveyed nine centers in Denmark that served 10,500 adults with type 1 diabetes and found that 134 adults were treated with SGLT-2s. Among the 134 adults and 222 patient-years of observation, this study found that no participants experienced a DKA event. Since this was a quality assurance project, Dr. Stougaard’s team was able to gather only minimal data, covering age (mean 51 years old), sex (72% female), SGLT-2 treatment duration (mean one year; range from six months to 2.4 years), and DKA frequency. Dr. Stougaard lamented that she could not provide a thorough analysis of SGLT-2 treatment effect in type 1 diabetes since her team did not have access to data on BMI, insulin pump use, SGLT-2 efficacy, non-DKA SGLT-2 side effects, or how these 134 people were selected for SGLT-2 treatment. Dr. Stougaard noted that in Demark people with type 1 diabetes who are offered an SGLT-2 are carefully selected by an attending physician, receive training regarding the risks of the treatment, and must demonstrate an understanding of how to mitigate and manage DKA. Dr. Stougaard concluded that this analysis indicates that SGLT-2 treatment may be safe in people with type 1 diabetes if they are carefully selected and instructed. While this real-world result is certainly positive, the lack of context surrounding participants’ full treatment regimen and baseline characteristics limits the generalizability of this finding to SGLT-2 treatment in type 1 diabetes more broadly.

  • Broadly, clinical trials of SGLT inhibitors in type 1 diabetes indicate that SGLT inhibitor use is associated with an increased DKA incidence. One notable exception was the EASE-3 trial, which used a 2.5 mg dose of empagliflozin, and did not result in an elevated DKA risk compared to placebo. Moreover, a 2019 meta-analysis of 10 randomized controlled trials (total n=5,961) found that SGLT inhibitors in people with type 1 increased the risk of DKA compared to placebo (RR=3.11; 95% CI: 2.11-4.58), except for 2.5 mg empagliflozin (RR=0.67; 95% CI: 0.11-3.95). The meta-analysis found that SGLT inhibitors at low doses were associated with a slightly lower risk of DKA (RR=2.90; 95% CI: 1.64-5.12) than at high doses (RR=3.36; 95%: 1.90-5.95). In this real-world retrospective analysis in Denmark, we would have appreciated seeing information on people’s SGLT-2 treatment dose and concomitant therapies to provide more context to the absence of DKA events.  

Trial

DKA incidence in treatment group

DKA incidence in placebo group

EASE Program (empagliflozin) (n=1,946)

---

---

EASE-2 and EASE-3 pooled (10mg empagliflozin)

5.9% (n=491)

1.8% (n=484)

EASE-2 and EASE-3 pooled (25mg empagliflozin)

5.1% (n=489)

1.8% (n=484)

EASE-3 (2.5mg empagliflozin)

1.7% (n=241)

2.5% (n=241)

DEPICT Program (dapagliflozin) (n=1,646)

---

---

Dapagliflozin 5mg

4.0% (n=548)

1.1% (n=532)

Dapagliflozin 10mg

3.5% (n=566)

1.1% (n=532)

Canagliflozin trial (n=351)

---

---

Canagliflozin 100mg

4.3% (n=177)

0% (n=177)

Canagliflozin 300mg

6.0% (n=177)

0% (n=177)

inTandem1 (sotagliflozin) (n=792)

---

---

Sotagliflozin 200mg

3.4% (n=263)

0.4% (n=268)

Sotagliflozin 400mg

4.2% (n=262)

0.4% (n=268)

inTandem2 (sotagliflozin) (n=782)

---

---

Sotagliflozin 200mg

2.3% (n=261)

0% (n=258)

Sotagliflozin 400mg

3.4% (n=263)

0% (n=258)

New CGM data + additional metrics shared on first two patients dosed in VX-880 islet cell replacement trial; patient 1 is insulin independent + 99.9% Time in Range at Day 270

Dr. James Markmann (Massachusetts General Hospital) presented groundbreaking results from the VX-880 trial investigating stem cell-derived pancreatic islet cell replacement therapy in people with type 1 diabetes. Dr. Markmann shared the impressive data from the first two patients in the study, who have been treated with just half of the expected trial dosage, per the study protocol. The 90-day results for the first patient dosed were announced in October showing a 91% decrease in daily insulin dose and reduction in A1c from 8.6% to 7.2%, sending shock waves through the diabetes community and beyond – see the New York Times and Good Morning America pieces. Results for the first patient, shown in tables below, have continued to improve so much so that at Day 270, he has been insulin independent for 60 days with a Time in Range of 99.9% (!!). This is the first demonstration of insulin independence achieved following treatment with a stem-cell derived islet therapy. While results for patient two are also trending in a positive direction, including a 30% reduction in insulin usage, they pale a bit in comparison to those of patient one. We are curious to know, especially as more patients receive the treatment, what patient characteristics are associated with a greater response. Overall, though, we continue to be so impressed by Vertex’s work in this area and look forward to seeing more data as it becomes available.

Baseline Characteristics

 

Patient 1

Patient 2

Age 

64 years

35 years

Gender

Male

Female

Duration of diabetes

42 years

11 years

BMI (kg/m2)

21.3

22.1

Results - Max C-peptide (pmol/L)

 

 

Patient 1

Patient 2

Baseline

undetectable

undetectable

Day 90

560

202

Day 150

1148

--

Day 270

523

--

Results – A1c

 

Patient 1

Patient 2

Baseline

8.6%

7.5%

Day 90

7.2%

6.7%

Day 150

6.7%

7.1%

Day 270

5.2%

--

Results – Insulin usage (units/day)

 

Patient 1

Patient 2

Baseline

34

26

Day 90

2.9

18.3

Day 150

2.6

18.2

Day 270

0

--

Results – Time in Range

 

Patient 1

Patient 2

Baseline

40%

36%

Day 150

81%

52%

Day 270

99%

--

  • Safety results for VX-880 are generally quite positive. Patient 1 reported 25 total adverse events that were primarily mild to moderate in severity. The most serious event was a rash associated with the immunosuppression regimen, and an episode of dehydration requiring hospitalization, both of which were resolved. The patient also experienced six severe hypoglycemic events in the perioperative period, which were ultimately not found to be associated with VX-880. Patient 2 experienced 12 adverse events, all of which were mild to moderate in severity and were not related to VX-880.
  • Recall that VX-880 is an investigational allogenic stem-cell derived, fully differentiated, insulin-producing, islet cell therapy – the first of its kind. Patients who are eligible for the study must have impaired hypoglycemic awareness and frequently reported severe hypoglycemia. Although this certainly isn’t everyone with type 1 diabetes, it is an important first step in assessing the efficacy of the cells before they are used in a broader T1D population. VX-880 is delivered by infusion directly into the hepatic portal vein, whose protocol was established based on the transplantation of cadaveric islet cells. Of course, one major thing to remember about VX-880 is that it does require an immunosuppression regimen to protect the stem cells from the immune system. While Vertex is also working on a cells + device encapsulation program that would erase the need for immunosuppression, at this point, patients must take the immunosuppression regimen into account when deciding to enroll in the trial.
  • Dr. Markmann did acknowledge that the study is currently on clinical hold by the FDA but didn’t say much more on the subject. The hold, which was announced in May, occurred because of “insufficient information to support dose escalation,” after the third patient in the trial received the full dose, per the study protocol. The trial is still running in Canada and continues to advance this research forward while Vertex works to resolve the FDA’s concerns in the US.

Preliminary phase 1/2 results for Sernova’s Cell Pouch device for islet transplantation: three patients remain insulin independent for two years, six months, and three months; Cell Pouch implantation was well-tolerated in 6/7 patients

Dr. Piotr Bachul (University of Chicago) presented preliminary results from Sernova’s phase 1/2 trial of its Cell Pouch for islet encapsulation in people with type 1 diabetes and hypoglycemia unawareness. This presentation was accompanied by a Sernova press announcement. Preliminary positive results from the first five patients in the trial was previously presented at the 2021 American Society of Transplant Surgeons Winter Symposium. In this study, Sernova’s Cell Pouch was implanted without islets, and after three weeks immunosuppression was introduced. After three to four weeks of immunosuppression, islets were transplanted into the Cell Pouch. After six months, patients received a second islet transplantation into the Cell Pouch, and after one year, patients were eligible for a supplemental intraportal islet transplantation to provide additional clinical benefit. Currently, of the seven patients who initially enrolled in the trial, six of whom received the Sernova Cell Pouch with islet transplantations and are still enrolled. The remaining patient received the Sernova Cell Pouch, but developed a post-procedure infection that resulted in device excision and study withdrawal. As shown in the table below, the first three patients (Patients A, B and C) have demonstrated persistent C-peptide levels during follow-up, confirming islet engraftment into the Cell Pouch and partial islet function. The three patients became insulin independent after one supplemental intraportal islet transplantation, maintaining insulin independence at the three-month, six-month, and two-year follow-ups. Furthermore, these three patients have achieved an A1c in the normoglycemia range of about 5%. The other three patients are awaiting additional islet transplantations but are still insulin dependent (see more on Patients D, E, and F below). Notably, Dr. Bachul said that no patients experienced opportunistic infections or unexpected adverse events. Ultimately, Dr. Bachul concluded that islet transplantation into the Cell Pouch plus supplemental intraportal islet transplantation led to significant clinical benefit, including sustained insulin independence for patients with hypoglycemia unawareness.  

Patient

Time Insulin Independent

A1c

C-peptide

Patient A

2 years

5.0%

Positive stimulated serum C-peptide after first and second Cell Pouch islet transplantations

Patient B

6 months

5.2%

Positive fasting serum C-peptide 12 months after second Cell Pouch islet transplantations

Patient C

3 months

5.2%

Positive stimulated serum C-peptide after second Cell Pouch islet transplantation

  • Patients D, E, and F developed antibodies after islet transplantation into the Cell Pouch, indicating immune rejection, but Patient D demonstrated that the autoantibodies did not prevent benefits of islet transplantation. Dr. Bachul explained that this immune rejection occurred during periods of suboptimal immunosuppression due to gastroenterocolitis (infection/inflammation of the digestive system) for Patients D and E and inconsistent use of immunosuppression for Patient F. Patients D, E, and F did not demonstrate endogenous insulin product at the three-month follow up. However, Patient D has now received a supplemental intraportal islet transplant. One month after this intraportal islet transplantation, Patient D has demonstrated positive C-peptide levels, a daily insulin dose reduction from 30 units to 11 units, and a 2% A1c reduction to 6.7%. Patients E is awaiting an intraportal islet transplantation and Patient F is awaiting a second islet transplantation into the Cell Pouch. 
  • To improve outcomes from Sernova’s previous phase 1/2 trial at the University of Alberta, the University of Chicago researchers modified multiple aspects of the implantation procedure. The Cell Pouch was implanted under the anterior rectus fascia (abdominal muscle) instead of the subcutaneous space to improve vascularization and limit the risk of seroma (accumulation of fluid under the skin). Immunosuppression was introduced three to four weeks prior to transplant to achieve optimal immunosuppression at the time of transplantation. To minimize the risk of post-procedure infection, the surgical procedure was modified to use two shorter incisions instead of one long incision. To minimize islet competition for oxygen, this trial only transplanted highly pure islet mass and reduced the islet mass to 3,000 islet equivalents (IEQ, a measure of islet volume) from 5,000 islet equivalents. Additionally, islets are suspended in patients’ own plasma instead of media, and the implantation procedure was done under general anesthesia rather than local anesthesia. 
    • Even during the study, researchers further optimized the implantation procedure. They found that suspending islets in patients’ plasma made the islet suspension too “sticky” and “jelly-like,” so instead they suspended islets in patients’ serum, which does not contain clotting factors. Additionally, islet concentration was reduced to avoid islet overcrowding, which resulted in greater stimulated C-peptide.
  • Since engraftment and C-peptide improved with lower islet concentration, Sernova has developed a new higher-capacity Cell Pouch to accommodate greater islet volume at a lower concentration. Specifically, the higher-capacity Cell Pouch allows for 50% more volume than the current Cell Pouch and has a >50% increase in internal surface area to allow for greater vascularization. Subsequent patients enrolled in this phase 1/2 trial will receive this higher capacity Sernova Cell Pouch.

MAP-T1D study find that small pancreas size and altered shape predicts progression from stage 2 to stage 3 type 1 diabetes

Dr. Jordan Wright (Vanderbilt) presented data from the MAP-T1D study showing that in individuals with positive autoantibodies, smaller pancreas volume predicts progression to stage 3 type 1 diabetes. The study assessed longitudinal pancreas MRIs in 41 multiple autoantibody-positive individuals (stage 1 or stage 2) across three TrialNet Centers. Participants were followed with an MRI every 6-12 months for four years, during which seven participants progressed to stage 3 clinical diabetes. Notably, while most patient characteristics were similar across progressors and non-progressors, participants who progressed to stage 3 diabetes were significantly younger than the other groups. Perhaps more importantly for the study, these participants had a significantly lower pancreas volume index at baseline. However, the pancreas size did not significantly change over time for the progressors or the non-progressors, suggesting that pancreas size is a relatively stable risk marker of clinical diabetes. Also, participants who progressed had a slightly different pancreas shape, with larger surface area to volume ratio and shorter principal axes, which could be assessed as an additional risk factor. Dr. Wright hopes to continue this research by understanding how pancreatic volume relates to other predictive factors, such as genetic risk scores and oral glucose tolerance tests.

Posters – Type 1 “Cures” and Adjunctive Therapies

Title

Authors

Details + Takeaways

Bioprinted Allogeneic Islet-Containing Implants Normalize Blood Glucose Control in Diabetic Rat Models without Immune Suppression

Valerio Russo, Reza Jalili, Yang Yu, Rishima Agarwal, Sheng Pan, Navid Hakimi, Kaushar Jahan, Emily M. Wilts, Shogo Ida, Spiro Getsios, Timothy J. Kieffer, Sam Wadsworth

  • Aspect Biosystems delivered bioprinted rat islet tissues encapsulated with immunoprotective shells to diabetic rats
  • Implants established normoglycemia for > 90 days in immunodeficient rats and for > 30 days in immunocompetent rats, demonstrating viability without immune suppression
  • Future studies will involve scaling up to larger animal models, and ultimately, patients with T1D

Identification of Sex-Specific Cytokine and Viral Infection Profiles in Type 1 Diabetes Patients

Khyati Girdhar, Alejandra Pina, Jelena Momirov, Mark A. Atkinson, Jason Ladner, Emrah Altindis

  • Comparative analysis of viral infection and cytokine profiles in patients with and without T1D
  • Sex-specific differences found in cytokine profiles indicated that disease progression could be sex-dependent
  • Greater % seropositivity of enterovirus C, influenza A, and orthopneumovirus was found in patients with T1D
  • Correlation analysis of cytokines with age revealed different age-based changes in cytokine profiles among people with T2D, suggesting impaired immune response

Plasma-Induced Signatures Measured at Type 1 Diabetes (T1D) Onset Correlate with Post-onset Beta-Cell Function

Amina Bedrat, Shuang Jia, Mark Roethle, Kristen Dew, Susanne M. Cabrera, Chien-Wei Lin, Martin Hessner

  • A sensitive bioassay was used to characterize diversity in T1D and analyze samples of 566 participants with new-onset T1D across six trials
  • Through weighted gene correlation network analysis to generate clustered modules of variation of co-expressed transcripts, which reflected differences in inflammatory and regulatory bias and differentiated subjects by post-onset beta cell function
  • These results support the existence of distinct T1D endotypes

Deep Immune Phenotyping of Type 1 Diabetes by Machine Learning

José Antonio Vera-Ramos, Barbara Prietl, Laurin Herbsthofer, Verena Pfeifer, Pablo López-García, Martin Stradner, Thomas Pieber

  • A machine-learning workflow was used to compare T1Ds with rheumatoid arthritis and systemic lupus erythematosus (other autoimmune diseases)
  • The program identified a subtype of T cells — marked by high CD127 levels and low CD25, CD161, and FoxP3 levels — that is elevated in patients with T1D

Learning from Immune Privileged Tissue Stem Cells to Generate Hypoimmunogenic Islets

Judith Agudo

  • Current methods of transplanting pancreatic islets lack optimal efficacy in resisting immune attack
  • This poster proposed mimicking the mechanism of an immune privileged stem cell from the skin and muscles, which was found to have resisted T cell attack without the use of a physical barrier
  • Identified a sustainable cell-intrinsic mechanism of reducing antigen presentation and avoiding natural killer cell responses  

Real-Time In Vivo Analysis of Beta-Cell Autophagy in Autoimmune Diabetes

Olha Melnyk, Charanya Muralidharan, Michelle M. Martinez Irizarry, Justin Crowder, Amelia K. Linnemann

  • Beta-cell autophagy in mice observed through an injectable beta-cell-selective autophagy biosensor
  • External pancreatic imaging alongside chloroquine infusion after three weeks showed expected autophagic flux in response to treatment in mice without diabetes, but a lack of effective autophagic flux response in mice with prediabetes
  • Results demonstrated a potential contribution of beta-cell autophagy defects in the pathogenesis of T1D

Population-Based Optimization of Metabolic Monitoring for Autoantibody-Positive Individuals at Risk for Type 1 Diabetes

Colin Orourke, Cate Speake, Alyssa Ylescupidez, Christine Bender, Sandra Lord

  • Developed a model to optimize metabolic monitoring schedules for autoantibody-positive individuals, with the goal of reducing DKA incidence and encouraging early T1D diagnosis
  • The model, built with OGTT data from 6,193 autoantibody positive individuals, capped undiagnosed time to <6 months (a 2-month reduction from an evenly spaced schedule) and utilized fewer visits (only 13, as compared to 34)
  • Results suggest that conducting half the number of metabolic monitoring visits usually done in research studies is likely to be effective in reducing the population incidence of DKA at diagnosis of T1D

Differentiation of PTPN2 Knockout Pluripotent Stem Cells into Stem-Cell–Derived ß-Like Cells Leads to Increased Stimulation of Diabetogenic TCR T Cell Transductants

Julia Q. Matuschek, Holger A. Russ, Taylor M. Triolo, Aaron W. Michels, Kristen Mcdaniel, Maria S. Hansen, Shane Williams, Roberto Castro-Gutierrez, Ali Shilleh

  • Explored the specific role of PTPN2, a known high risk T1D gene, in beta cell interactions with diabetogenic T cells
  • PTPN2 knockout and wild type human pluripotent stem cells were differentiated into stem-cell derived beta-like cells (sBCs) and exposed to diabetes stress conditions and an HLA-peptide T-cell receptor
  • PTPN2 knockout sBCs showed both elevated expression of HLA Class I molecules (linked to T cell recognition) and increased stimulation of T cell transductants
  • Results highlight a potential explanation for the contribution of PTPN2 to the autoimmune response in T1D

Treatment Algorithms, Strategies, and Guidelines

Management of Hyperglycemia in Type 2 Diabetes: Draft 2022 ADA-EASD Consensus Report emphasizes holistic care, weight management, and personalized therapeutic regimens

Closing out the day, an esteemed group of KOLs from the US and Europe joined forces to present the joint ADA/EASD Consensus Report on the Management of Hyperglycemia in Type 2 Diabetes. As consensus co-chair Dr. Stefano Del Prato explained, these consensus recommendations build on the 2018 statement on the management of hyperglycemia in type 2 diabetes and subsequent 2019 update to the 2018 document outlining best practices in type 2 diabetes management. The consensus group reviewed clinical literature from the last three years and met for a three-day virtual meeting as well as a two-day in-person workshop to develop the consensus report. Overarchingly, the consensus group advocated in favor or “holistic, person-centered” type 2 diabetes care across the four components of (i) weight management, (ii) glycemic management, (iii) cardiovascular risk management, and (iv) cardio and renal protection. Across these four arenas, there was substantial emphasis on the importance of personalized goals based on individual people’s risk factors and priorities. We were especially impressed by the emphasis on weight management as a clinical factor with equal relevance to glycemic management that we see as representative of the ongoing paradigm shift toward obesity and weight management as intimately related aspects of diabetes management for people with type 2 diabetes. Additionally, this draft consensus report, like its 2018 predecessor, includes a robust “implementation” component providing key questions and information that must be answered for clinicians to effectively implement these best practices into their clinics. The draft consensus report will be available for review here and is now in an open comment period until June 21, 2022 at 11:59 pm ET (all comments can be sent to adacomments@diabetes.org). Following the review of submitted comments, the report will be published in September 2022 at the Annual Meeting of the European Association for the study of Diabetes.

  • Multiple speakers including Dr. John Buse (University of North Carolina) and Dr. Vanita Aroda (Brigham and Women’s Hospital) highlighted the recognition in the guidelines that glycemic management and weight management are two co-primary goals for the type 2 diabetes patient. On the glycemic management side, Dr. Aroda noted that, while metformin has traditionally been considered first-line therapy, combination therapy should be considered to increase the durability of the treatment effect and improved glycemic outcomes. Notably, the ADA/EASD guideline flowchart provides a color-coded ranking of agents based on glycemic control and weight management efficacy. Under the glycemic management category, dulaglutide, semaglutide, tirzepatide, insulin, and combination therapies are listed as those with the highest efficacy. On the weight loss side, semaglutide and tirzepatide are listed as the therapies with greatest efficacy. Echoing the 2022 ADA Standards of Care, Dr. Aroda highlighted the importance of combination therapy to better achieve and maintain glycemic and weight loss goals. She described increasing evidence and rationale for this approach, including the following factors: (i) increased durability of glycemic effect; (ii) potential to address therapeutic inertia; (iii) simultaneous target of multiple pathophysiologic processes characterized by type 2 diabetes; (iv) potential impact on medication burden, adherence, and treatment persistence; and (v) complementary clinical benefits.
  • Dr. Aroda discussed the use of GLP-1s and SGLT-2s to provide cardiorenal risk reduction in people with or at risk of ASCVD and CKD, highlighting the role of combination therapy to best improve outcomes. The ADA/EASD guidelines align with the 2022 ADA Standards of Care, which recommend the use of a GLP-1 or SGLT-2 for people with established ASCVD or multiple risk factors for ASCVD and those with CKD. Dr. Apostolos Tsapas (Aristotle University of Thessaloniki, Greece) emphasized that the cardioprotective effects of GLP-1s and SGLT-2s are consistent regardless of metformin use and baseline A1c, according to several large meta-analyses. Meanwhile, Dr. Jennifer Green (Duke University) noted that the kidney benefits of SGLT-2s are consistent across the spectrum of eGFR values and in people with and without macroalbuminuria. Dr. Green also highlighted consistent CV and kidney benefits with SGLT-2s and GLP-1s regardless of age (both ≤ and ≥65 years). While these recommendations do not come as a surprise given the large body of evidence from many CVOTs, we are pleased to see these newer anti-hyperglycemic medications continuing to be prioritized in treatment algorithms and gaining recognition for their cardiorenal benefits. For both SGLT-2s and GLP-1s, Dr. Aroda noted that the literature search revealed increased confidence surrounding safety issues of interest. In particular, Dr. Aroda noted that there is greater clinical expertise anticipating and addressing GI side effects of GLP-1s, as well as greater recognition of the importance of education to address what tend to be mild to moderate temporary side effects. This is certainly welcome news, and we hope that the ADA/EASD guidelines’ emphasis on the safety of agents will lead to further uptake among all specialists in the field of cardiometabolic health.
    • Notably, the guidelines include a section on a new class of agents: glucose-dependent insulinotropic polypeptide receptor and GLP-1 receptor agonists – namely, tirzepatide. Overall, the guidelines acknowledge tirzepatide’s very high glycemic efficacy and high weight loss, with a low risk of hypoglycemia and unknown cardiorenal effects (trials in progress). We’re pleased to see tirzepatide incorporated into the guidelines so quickly and we expect that other professional societies will follow suit, given the highly impressive weight loss seen in the SURMOUNT-1 trial as well as the recent FDA approval under brand name Mounjaro.
  • Dr. Nisa Maruthur (Johns Hopkins University) and Dr. Sylvia Rosas (Joslin Diabetes Center) provided practical tips for the implementation of these guidelines, emphasizing the importance of integrated, individualized care. We were pleased to learn that the guidelines emphasize the role of Diabetes Self-Management Education and Support (DSMES), with Dr. Rosas stating that providers should embrace DSMES as an aspect of diabetes care that is as important as other components of the treatment plan, including pharmacotherapy. Overall, Dr. Maruthur highlighted three key domains to really “make it work” with the new guidelines. First, providers must consider delivery arrangement – how, where, and by whom care is delivered, as well as coordination of care and communication technology. Second, Dr. Maruthur highlighted governance arrangements, or the infrastructure to support the implementation of these guidelines, including training, certification, and quality of practice. Third, overall implementation strategies are paramount and must include consideration of the health system, health care setting, and health care workers themselves.
    • Diving deeper into proactive care, Dr. Maruthur highlighted the recommendation to consider initial combination therapy, especially in those with high A1c at diagnosis and those with youth-onset type 2 diabetes. The guidelines also specifically urge providers to avoid clinical inertia, with Dr. Maruthur reminding attendees that each clinic visit is an opportunity to re-evaluate health behaviors, medication use, and side effects of therapeutic agents. When further glycemic control is needed, the guidelines recommend incorporating additional agents with complementary mechanisms of action. At the same time, the guidelines also acknowledge the issue of pill burden and suggest the use of fixed-dose combinations and/or treatment deintensification in certain patients (such as frail older adults and those with hypoglycemia-causing medications).

New Joint Statement from the ADA and KDIGO on the management of diabetes and CKD: SGLT-2, metformin, RAS inhibition, and statin as first-line therapy for type 2 diabetes and CKD; GLP-1 for additional glycemic control and finerenone for additional albuminuria lowering

Dr. Peter Rossing (Steno Diabetes Center Copenhagen, Denmark) presented the joint statement from the ADA and KDIGO on the management of diabetes and CKD, emphasizing that both organizations are highly aligned in their recommendations.  To our knowledge, the joint statement has not yet been published. Dr. Rossing prefaced his presentation by saying that guidelines and consensus statements are an important way to educate people about breakthrough therapies and provide guidance on how and when to use those therapies. He said that a writing group consisting of ADA and KDIGO representatives convened to compare and contrast ADA and KDIGO recommendations since, as he said, “The more we provide the same guidance, the more likely we are to be successful in implementation.” He noted that the 2022 ADA Standards of Care were updated in May 2022 and that the 2022 KDIGO guidelines will be published later this year (see the draft 2022 KDIGO guidelines), and both sets of guidelines reflect the guidance in this joint statement. As shown below, Dr. Rossing shared the joint treatment algorithm for people with CKD and diabetes (note green is for type 2 diabetes only and light blue is for type 1 and type 2 diabetes). Dr. Katherine Tuttle (University of Washington) previously shared this algorithm at AACE 2022. For people with type 2 diabetes, the joint statement recommends an SGLT-2, metformin, RAS inhibitor, and a statin. Dr. Rossing specifically highlighted the new SGLT-2 eGFR initiation threshold of ³20 mL/min/1.73 m2. He explained that this new threshold is a result of the EMPEROR trials, which enrolled participants with eGFRs ³20 mL/min/1.73 m2, along with data from DAPA-CKD, CREDENCE, and data indicating that SGLT-2s generally have similar risks and benefits (aside form A1c lowering) across a wide range of eGFR values. Notably, the current guidelines recommend GLP-1 addition only if a patient’s individualized glycemic target is not met since there has been no kidney outcomes trial for a GLP-1 – the results of the FLOW trial of semaglutide in CKD (expected completion August 2024) may eventually change the positioning of GLP-1s. Overall, Dr. Rossing emphasized that both the ADA and KDIGO recommend a foundation of lifestyle modification followed by therapies for further control of hyperglycemia, albuminuria, blood pressure, or lipids. He concluded, “These two guidelines are very well aligned. We agree. We have seen the same evidence. So please, go out and do it.” 

  • On screening for CKD in people with diabetes, the joint statement emphasizes the importance of screening both urine albumin-to-creatinine ratio and eGFR. The joint statement recommends screening yearly starting at diagnosis for those with type 2 diabetes and starting five years after diagnosis for those with type 1 diabetes. A positive CKD result is defined as persistent UACR ³30 mg/g, persistent eGFR <60 mL/min/1.73 m2, or other evidence of kidney damage. After a positive result, clinicians should evaluate possible temporary or spurious causes, consider using cystatin C and creatinine to more precisely estimate GFR, and initiate treatment if kidney abnormalities are persistent. Dr. Rossing also shared a heat map of UACR and eGFR indicating risk of CKD progression, suggested frequency of visits, and whether a referral to a nephrologist is needed.

  • Dr. Rossing reviewed the statement’s recommendations on lifestyle modification and education. The joint statement recommends maintaining protein intake of 0.8 g of protein/kg/day for those with diabetes and non-dialysis CKD, whereas those on dialysis should consume between 1-1.2 g of protein/kg/day. In terms of physical activity, the statement recommends 150 minutes of moderate-intensity activity per week. On education, Dr. Rossing shared key objectives for a structured self-management educational program, which the joint statement recommends all people with diabetes and CKD take.  

  • Dr. Rossing also shared additional information-packed diagrams from the joint statement to guide clinicians in selecting and adjusting medications. Dr. Rossing shared a detailed diagram summarizing the CKD, ASCVD, HF, glucose-lowering, hypoglycemia, and weight effects of different antihyperglycemic therapies (second figure below). He also shared a chart depicting how to adjust different medication based on a patient’s eGFR.

Prevention or delay of type 2 diabetes in the 2022 Standards of Care: The importance of weight loss, targeted and individualized care, and consideration of social determinants of health

A rockstar panel of KOLs discussed the development of the 2022 ADA Standards of Care with a focus on Section 3: Prevention or Delay of Type 2 Diabetes and Associated Comorbidities. Dr. Florence Brown (Joslin Diabetes Center) described the development of the ADA Standards of Care, highlighting that the guidelines are a Living Document based on the latest evidence through a comprehensive literature search – in fact, the ADA updated its 2022 Standards of Care just last week to reflect recent research in therapies for CKD and HF. Earlier this year at the ADA Clinical Update Course, ADA CMO Dr. Robert Gabbay emphasized the importance of making the Standards of Care as actionable and accessible to HCPs as possible – we continue to commend the ADA for its focus on access and transparency and its commitment to continually updating the guidelines as new therapies and technologies are approved. Throughout today’s session, speakers emphasized the importance of weight loss and individualization of care to prevent progression from prediabetes to type 2 diabetes. The available options for weight loss continue to expand with the approval and launch of incredible weight loss seen with incretin therapies like dual GIP/GLP-1 agonist tirzepatide (recently approved in type 2 diabetes as Mounjaro). We continue to hope these newer diabetes pharmacotherapies will be incorporated further upstream in the treatment algorithm to prevent the development of type 2 diabetes and its associated comorbidities.

  • Dr. Vanita Aroda (Brigham and Women’s) discussed individualizing care goals for patients at high risk of type 2 diabetes, emphasizing language changes in the 2022 Standards of Care that allow for more targeted interventions for people at highest risk of progressing from prediabetes to type 2 diabetes. Previously, the Standards of Care called for monitoring for the development of type 2 diabetes in those with prediabetes at least once per year. This year’s guidelines continue to generally recommend annual monitoring but suggest that the regularity is modified based on individual risk/benefit assessment. Dr. Aroda suggested that this more targeted recommendation is designed to help providers better identify people at highest risk of progressing to type 2 diabetes and provide interventions to these individuals earlier on. In terms of lifestyle behavior change, the 2022 Standards of Care specifically recommend referring adults with overweight/obesity at high risk of type 2 diabetes to an intensive lifestyle change program, which is much more specific than the previous recommendation to refer anyone with prediabetes to such interventions.
  • Dr. Scott Kahan (National Center for Weight and Wellness) reviewed pharmacotherapy for diabetes prevention, highlighting that metformin has the strongest evidence base and demonstrated long-term safety for diabetes prevention. Recall that metformin conferred a 31% reduction in progression to diabetes in the DPP; participants saw an 18% continued reduction in both the 10- and 15-year follow-up studies. Subgroup analyses of the DPP revealed that people with younger age, higher BMI (>35), higher baseline fasting plasma glucose (>110 mg/dL), higher baseline A1c (>6.0%), and a history of gestational diabetes (GDM) were more likely to benefit from metformin therapy. In particular, Dr. Kahan emphasized metformin’s strong prevention for women with a history of gestational diabetes, who saw a 40% reduction in progression to diabetes in the DPP. Echoing Dr. Aroda, Dr. Kahan stressed the importance of weight loss for diabetes prevention, emphasizing that participants who achieved all three goals in the DPP (weight, exercise, and dietary fat intake) saw a whopping 90% reduction in progression to type 2 diabetes. We are, of course, eagerly awaiting data on the potential of newer incretin therapies, such as tirzepatide and semaglutide, to prevent progression from prediabetes to diabetes, given the encouraging preliminary results from the SURMOUNT-1 study as well as the finding in STEP-5 that 80% of participants with prediabetes reverted to normoglycemia with semaglutide treatment.
  • Dr. Jane Reusch (University of Colorado) discussed the bigger picture in diabetes prevention, stating that we are “outstripping our ability to take great care of people at risk of diabetes” by not focusing more urgently on prevention. Overall, she argued that prevention or delay of diabetes is urgent because diabetes shortens the life span, citing a 2015 study published in JAMA that found an increased risk of diabetes complications with earlier age of onset. In particular, youth-onset type 2 diabetes carries an especially high burden of complications (as was shown in a 2021 NEJM publication). Noting that obesity is the #1 modifiable risk factor for progression to diabetes, Dr. Reusch highlighted the Diabetes Prevention Program (DPP) as a medical intervention that incorporates intentional dietary changes and physical activity to reduce the risk of diabetes. Borrowing Nike’s slogan “Just Do It,” Dr. Reusch emphasized that the greatest benefit comes from transitioning from a sedentary lifestyle to doing some form of physical activity and stated, “just getting out of the seat is the first step.” Turning to social determinants of health, she urged providers to treat people realistically based on their unique context, including individual circumstances and barriers to optimal engagement in diabetes prevention and management. The 2022 Standards of Care therefore call for comprehensive assessment of social determinants of health, including food insecurity, housing insecurity/homelessness, financial barriers, and social capital/social community support to inform treatment decisions, with referral to appropriate local community resources. We were pleased to hear Dr. Reusch draw attention to Dr. Felicia Hill-Briggs’ Scientific Review on social determinants of health and diabetes (published in Diabetes Care in November 2020).

Dr. James Januzzi presents the ADA’s consensus report on heart failure; SGLT-2s take center stage for those at-risk for HF and those with clinical HF

Dr. James Januzzi (Harvard) presented an overview of the ADA’s consensus report on heart failure published on June 1, 2022, which he co-authored: “Heart Failure: An Underappreciated Complication of Diabetes. A Consensus Report of the American Diabetes Association.” Dr. Januzzi noted that the ADA’s consensus report was endorsed by the American College of Cardiology, which itself released the 2022 ACC/AHA/HFSA Guidelines on heart failure earlier this year. We are thrilled to see the ADA and ACC working together to amplify guidance on how best to management HF in people with diabetes and prediabetes. The 16-page consensus report covers HF epidemiology, HF diagnosis and clinical stages, management of HF in diabetes, and opportunities and challenges for multidisciplinary care, and knowledge gaps. In his talk, Dr. Januzzi highlighted the four stages of HF, as set out in the ACC/AHA/HFSA guidelines and the University Definition and Classification of HF task force: (i) stage A defined as those at high risk for HF; (ii) stage B or pre-HF defined as those with structural heart disease without symptoms; (iii) stage C or symptomatic HF; and (iv) stage D defined as advanced HF with persistent symptoms despite therapeutic treatment. Based on these stages, Dr. Januzzi walked through the ADA consensus report’s recommendations for HF management.

  • Dr. Januzzi discussed using biomarkers to identify stage B (at-risk) HF since an echocardiogram can be difficult to obtain in a timely manner. Dr. Januzzi specifically highlighted the value of measuring natriuretic peptides (NT-proBNP) to identify people with type 2 diabetes at high risk for hospitalization for heart failure (hHF). He explained that NT-proBNP sharply increases six months prior to hHF in people with diabetes. He acknowledged that albuminuria, which many endocrinologists already measure, is itself a good predictor for heart failure. While Dr. Januzzi pushed against only measuring albuminuria to predict CV risk since NT-proBNP is a superior predictor for CV risk, noting that albuminuria is additive but insufficient to take the place of NT-proBNP, we wonder if this extra information about albuminuria is still likely yet another reason that HCPs may be taking greater care or even pushing with patients to get this regular screening done more … regularly!
  • On the treatment of diabetes in HF, the ADA consensus report says that SGLT-2s are “an expected element of care” in all people with diabetes and symptomatic HF (stages C or D). Furthermore, the report recommends prioritizing SGLT-2 use in people with stage B or pre-HF. If additional glycemic control is needed, the report suggests initiating GLP-1, metformin, or both, followed by insulin if needed. Dr. Januzzi emphasized that DPP-4s and TZDs are not recommended for people with stages B-D heart failure because DPP-4s have no demonstrated CV benefit (and may increase HF risk) and TZDs have been generally shown to increase CV risk. Beyond glucose-lowering drugs, the report recommends ACEIs/ARBs for stages A/B and notes that finerenone may reduce DKD progression, which in turn may prevent HF. For stages C/D, the report is well-aligned with the 2022 ACC/AHA/HFSA Guidelines on heart failure, recommending the four pillars of heart failure with reduced ejection fraction (RAAS inhibitors, beta-blockers, MRA, and SGLT-2s) and SGLT-2s for heart failure with preserved ejection fraction. 
  • Emphasizing the importance of multidisciplinary care for people with diabetes and HF, Dr. Januzzi highlighted the ADA’s recommendations for referring patients to a cardiovascular specialist. For patients with stage A HF, the report suggests referral when guidance is needed on the management of risk factors or on the assessment of CV risk. For stage B, the report suggest referral when it would be helpful for risk assessment, determination of the cause for structural heart disease, and the initiation of treatment. For stages C and D, the report states that a referral should be made for all aspects of care, including intensive diagnostic evaluation, therapy initiation and titration, and longitudinal follow-up.

Follow-up analysis from GRADE: Significantly higher discontinuation rates of non-insulin medications compared to insulin glargine (p<0.001); Dr. Rasouli suggests higher prominence of therapeutic inertia for insulin intensification than for insulin initiation; what ‘GRADE’ does glargine receive? Dr. Rasouli argues B+

During an afternoon session, Dr. Neda Rasouli (University of Colorado) presented data from a follow-up analysis of the head-to-head GRADE study. As a reminder, GRADE enrolled 5,047 patients with type 2 diabetes who were randomized to basal insulin glargine, DPP-4 sitagliptin, sulfonylurea glimepiride, or GLP-1 liraglutide, and full data from the study were presented at EASD 2021. During today’s session, Dr. Rasouli shared data showing that participants who permanently discontinued one of their assigned medications during the trial were less likely to discontinue insulin, vs. non-insulin medications (p<0.001). Specifically, 13% of those on glargine discontinued (n=22 of 169), compared to 20% of those on glimepiride (n=50 of 248), 17% of those on liraglutide (n=37 of 218), and 16% of those on sitagliptin (n=32 of 203). We’re curious to better understand why participants were more likely to continue use of insulin glargine compared to the other medications. Compared to glimepiride, glargine was associated with a reduced rate of severe hypoglycemia (p=0.05), but glargine was also associated with the highest percentage of participants seeing a ≥10% weight gain, with those on liraglutide significantly less likely to gain ≥10% weight (p<0.0001). Notably, Dr. Rasouli presented data showing that in the glargine-treated group, ~55% of participants who were supposed to progress to insulin aspart per the protocol never progressed to insulin aspart. In contrast, approximately two thirds of participants in the non-insulin treated group progressed to insulin therapy as outlined in the study protocol. Dr. Rasouli interpreted these data as an indication that therapeutic inertia was more prominent for insulin intensification in the glargine group than for insulin initiation in the three non-glargine groups.

  • Dr. Rasouli finished her presentation by offering a letter grade for insulin therapy as seen in the GRADE study. She offered a detailed breakdown of her grading criteria: (i) since only 30% of participants maintained A1c <7% at year four, the drug earned a “C” in this category; (ii) since 65% of participants maintained an A1c ≤7.5% at year four and outperformed those in the other three non-glargine arms, the drug earned an “A” in this category; (iii) since there was a low risk of severe hypoglycemia, significantly better than glimepiride, but still nonzero, the drug earned a “B” in this area; (iv) since 13% of participants gained >10% weight, the drug earned a “C” in this area; and (v) since the mean participant quality of life score was a 65, which was similar to other groups, the drug earned a “B” in this area. Putting it all together, Dr. Rasouli gave glargine overall a B+, saying that she is awaiting “smart” insulins that operate without hypoglycemia, require less frequent injections, and have no need for titration. Certainly, we were curious for some mention of Tresiba and Toujeo. It’s worth noting that there was significant disagreement in the audience about the B+ grade from the audience, with one member even emphatically stating that glargine gets an “F.” We were very surprised by this and while we like debates, we felt that sentiment was a stretch by any definition.

Precision Medicine

Heterogeneity blurs the lines between type 1 and type 2 clinical diabetes; precision medicine aims to use this heterogeneity to provide optimal diabetes care

Dr. Richard Leslie (Blizard Institute, Queen Mary University of London, UK), Dr. William Hagopian (Pacific NW Research Institute), Dr. Maria Redondo (Baylor College of Medicine, Texas), and Dr. Paul Franks (Lund University Diabetes Centre, Sweden) closed out ADA 2022 with a deep dive into what some consider to be the future of diabetes care – precision medicine. The discussion focused largely on heterogeneity in diabetes, including the underlying genetic risk factors that result in improper insulin secretion or resistance. The primary thesis of the presentation is that although some cases of diabetes can clearly be differentiated into “type 1” or “type 2” diabetes, based on common phenotypes, there are a number of cases that have overlapping phenotypes that make them more difficult to classify.

  • Moreover, distinction between type 1 and type 2 diabetes isn’t quite as clear cut as some may think. Defining biomarkers – BMI, c-peptide production, and autoantibodies – are not perfectly correlated with disease. Additionally, as both Dr. Redondo and Dr. Franks noted, the phenotypic subclassification can change over time, making it a somewhat unreliable source of diagnosis. For instance, C-peptide levels, a commonly used measure of insulin function, may appear normal in type 1 diabetes during the honeymoon period, and autoantibodies may come back as negative in ~5% of people with clinical type 1 diabetes and positive in ~5-10% of people with clinical type 2 diabetes.
  • Not only did the presentations raise in our minds that the classifications for diabetes may be a bit on the broader side, we were also surprised to learn how complex the clustering based on phenotypes can be – wow! For example, a 2022 study (n=726) “Four groups of type 2 diabetes contribute to the etiological and clinical heterogeneity in newly diagnosed individuals: An IMI DIRECT study” assessing the presentation and underlying pathophysiology of type 2 diabetes used a soft-clustering method to characterize newly diagnosed T2D based on 32 clinical variables. The study identified four cluster profiles based on the biomarkers that represented dysfunction patterns across combinations of type 2 diabetes. In this study, the cluster associated with obesity, insulin resistance, dyslipidemia, and impaired β cell glucose sensitivity corresponds with the fastest disease progression and highest demand for anti-diabetic treatment.
  • While someday the goal might be to use this cluster identification to provide personalized treatment, that is likely a long way off according to the experts. Dr. Franks said that it is thought that when patients are not forced by trial design into a cluster, just 30% of people can be sorted into a genetic cluster, suggesting that the majority of patients do not fit into any defined archetype. It may actually be somewhat lower than this upon listening to a few more of the parameters, it turns out; specifically, the lack of diversity in these phenotypic and genetic clustering trials may call into question the degree to which external generalizability of these studies makes sense. For example, the aforementioned study excluded genotypes of non-European descent. We learned on Day #1 of ADA that only 2% of genome-wide association studies come from people of African descent and that the majority of variants identified in European genotypes were not found to be associated risk in African genotypes, and vice versa. Thus, before turning to this precision medicine approach, there must be significantly more work done to make sure the genotypic and phenotypic research that is being used to identify clusters is more representative of the overall population.
  • We appreciated Dr. Franks’ perspective suggesting that precision medicine is not a revolution, but rather an evolution that will be used to supplement the current standard of clinical care. Precision medicine aspires to be a safer and more efficacious, accessible, and equitable way of delivering care to help patients achieve their optimal outcomes, and we look forward to watching it grow from these early days to someday reach that goal.

Drs. John Dennis and William Cefalu discuss precision diabetes medicine, highlighting individualized prediction models for SGLT-2 vs. DPP-4 usage in type 2s without cardiorenal risk and roadmap of ongoing research

Drs. John Dennis (University of Exeter, UK) and William Cefalu (NIDDK and Louisiana State University) discussed practical applications of precision diabetes medicine in clinical practice as well as the rationale and roadmap for this approach to diabetes pharmacotherapy. Both Dr. Dennis and Dr. Cefalu emphasized that, while the 2022 ADA Standards of Care provide individualized recommendations for the use of newer anti-hyperglycemic agents in people with type 2 diabetes and established CVD or CKD, the guidelines offer few recommendations beyond metformin use in people without these complications. In fact, approximately two-thirds of people with type 2 diabetes in the UK fall into the latter category for whom there is no clear guidance after metformin. For these patients, the guidelines instruct providers to select therapeutic agents based on glycemic efficacy, adverse events, comorbidities, and cost – a variety of reasons that are individualized to the patient’s priorities, but none of which based on the patient’s specific genetic, physiologic, and phenotypic factors. Both speakers emphasized that the current ADA/EASD guidelines provide recommendations based on average treatment effects from clinical trials, with Dr. Dennis noting that people with type 2 diabetes with lower cardiorenal risk have largely been excluded from these trials. 

  • Dr. Dennis discussed a 2020 study (published in Diabetes Care) investigating the use of individualized prediction models to optimize treatment with either an SGLT-2 or DPP-4 for people with type 2 diabetes but without CVD or renal disease. The study employed a model based on five routine clinical features (A1c, age, BMI, eGFR, and ALT) to target treatments based on A1c reductions, weight change, and tolerability. Notably, the predicted differences in A1c were reproducible in clinical trials, revealing a ≥5 mmol/mol reduction for 40% of those on SGLT-2s. Meanwhile, a modest DPP-4 benefit was predicted and observed for approximately 15% of participants. The analysis revealed powerful subgroup differences for treatment discontinuation, finding that participants receiving a DPP-4 had a 50% lower risk of treatment discontinuation compared to those receiving an SGLT-2. Dr. Dennis noted that this may be particularly relevant for older patients, suggesting that this subpopulation is likely to experience fewer discontinuations on DPP-4s compared to SGLT-2s. Overall, this precision medicine approach provides accurate estimates of relative A1c benefit of different drug classes and differences in discontinuation risk with low-cost implementation, since it relies on routine clinical features. Looking ahead, models for all the major drug classes after metformin can offer a more comprehensive approach to precision diabetes medicine. Dr. Dennis shared a beta-version of a calculator that uses these routine clinical parameters to provide individualized estimates of likely achieved glycemic control (A1c reduction), weight change, and early treatment discontinuation for DPP-4 and SGLT-2 therapy.
  • On long-term outcomes, Dr. Dennis advocated an individualized approach for the use of SGLT-2s in heart failure based on the combination of RCT data and individualized prediction models to derive absolute, rather than relative, estimates of benefit. In an ongoing study of nearly 63,000 patients initiating an SGLT-2, SU, or DPP-4, integrating the QDiabetes-Heart Failure risk calculator (a UK model to predict individualized HF risk) with trial evidence from SGLT-2 CVOTs revealed that median absolute risk of heart failure and median absolute risk benefit were fairly low at 2.3% and 0.5%, respectively. Notably, the distribution plots for both analyses were right-skewed, with a number of patients at much higher risk of developing heart failure. However, when patients were divided into four quartiles based on heart failure risk, there was a clear difference in absolute HF benefit with SGLT-2s: the three lowest-risk quartiles showed minimal evidence of benefit, whereas the highest risk quartile had a marked reduction in heart failure with SGLT-2s vs. DPP-4s. Ultimately, Dr. Dennis advocated augmenting RCT data with real-world modelling to calculate the absolute CVD risk benefits of SGLT-2s and GLP-1s in order to better inform and individualize pharmacotherapy.
  • Dr. William Cefalu (NIDDK and Louisiana State University) described precision medicine as an all-encompassing approach that seeks to match the right drug to the right patient at the right time to obtain the best clinical outcome. Dr. Cefalu noted that although marked heterogeneity exists in terms of clinical presentation of diabetes, risk factors, outcomes, and disease course, treatment strategies are not targeted to specific genetic, physiologic, and phenotypic features. As one example, there are numerous biomarkers of diabetes, but current recommendations focus on glucose as the sole biomarker for treatment decisions. Dr. Cefalu discussed a landmark Scandinavian study (published in The Lancet in 2018) that used cluster analysis of six key variables (GAD antibodies, age, BMI, A1c, HOMA2-B insulin secretion, and HOMA2-IR insulin resistance) to identify diabetes subtypes, which were associated with clinical outcomes in subsequent analysis. Indeed, these clinical subtypes have been validated in many studies across the globe, including in DEVOTE, LEADER, and SUSTAIN-6 in a 2020 publication in Diabetes Obesity and Metabolism and in the ORIGIN trial in a 2022 publication in Diabetologia. Going a step further, the IMI-RHAPSODY study conducted genetic (n=12,828), metabolomic (n=2,945), lipidomic (n=2,593), and proteomic (n=1,170) analysis in people with type 2 diabetes in the Netherlands, revealing differences in key protein levels by disease subtype. For instance, people with Severe Insulin Resistant Diabetes (SIRD) had elevated branch-chain amino acid levels, which are associated with diabetes risk, alongside lower levels of proteins involved in the glucose translocation pathway. Meanwhile, those with Mild Diabetes with High HDL (MDH), the mildest diabetes phenotype, had lower levels of branch-chain amino acids and higher levels of proteins involved in insulin signaling. Overall, Dr. Cefalu highlighted that studies such as IMI-RHAPSODY and other efforts through the NIH’s Precision Medicine Initiative will help us address the heterogeneity of diabetes. Ultimately, he suggested that precision medicine will lead to a re-classification of diabetes via identification of subtypes through validated biomarkers, which will lead to improvements in prevention, diagnosis, and management.

Dr. Anna Gloyn highlights the role of CALCOCO2 and PAM gene variants in reducing insulin content in type 2 diabetes, likening the type 2 diabetes genetic landscape to a massive jigsaw puzzle

Stanford’s Dr. Anna Gloyn gave the Outstanding Scientific Achievement Award lecture, titled “Mining the Genome for Gold—Drilling Down on Mechanisms for Pancreatic Islet Cell Dysfunction in Diabetes.” Dr. Gloyn framed her lecture with a jigsaw puzzle analogy, likening the challenge of understanding the genetic causes of type 2 diabetes to solving a 100,000-piece puzzle. While approximately 350 signals for type 2 diabetes have been identified in the human genome, few of these variants are located in the coding region of the genome, hampering efforts to drill down deeper into their mechanisms in type 2 diabetes. Dr. Gloyn explained that these variants are like individual snowflakes in a puzzle scene and it’s hard to know where exactly they are found – are they snow on trees? People? Buildings? Recent genome-wide association studies have provided some clarity, such as an ambitious project by Oxford’s Dr. Antje Grotz to identify all of the genes that regulate insulin content in beta-cells. The genome wide screen revealed 580 genes that influence insulin content in human beta-cells. Intersection of these 580 genes with 336 genes predicted to underly type 2 diabetes yielded 20 genes implicated in type 2 diabetes risk. Dr. Gloyn focused on one of these 20 genes, the CALCOCO2 gene, which has been implicated in the innate immune response but previously had no known role in beta cells. Research by Stanford’s Dr. Yingying Ye demonstrated that loss of CALCOCO2 leads to reduced numbers of insulin granules, especially immature granules. A closer look at beta cell morphology revealed large vacuolated structures following CALCOCO2 silencing; these structures contained remnants of cell machinery, which is consistent with CALCOCO2’s known autophagic activity. Ultimately, this finding suggests that loss of CALCOCO2 leads to increased autophagy, which targets recently synthesized insulin granules, leading to a reduction in insulin content. Returning to the jigsaw puzzle analogy, Dr. Gloyn concluded her lecture stating that we now have identified most of the puzzle pieces and have a good idea of how to assemble the puzzle, but we just need to figure out what the snowflakes do – in other words, how the individual genetic variants contribute to type 2 diabetes.  

  • Dr. Gloyn also discussed research based on the T2D-GENES multi-ethnic genome-wide association study (n=13,000), which found that variants in the PAM (peptidylglycine alpha-amidating monooxygenase) gene are associated with reduced granular insulin content and secretion as well as a reduced response to GLP-1. For background, PAM is the only enzyme in the body that can create amide groups on glycine-extended peptide hormones, which increases the biological potency or half-life of these hormones. The T2D-GENES study identified two independent coding variants in the PAM gene associated with type 2 diabetes risk. Further research by Oxford’s Dr. Soren Thomsen (published in Nature Genetics in 2018) found that silencing of PAM in human beta-cells leads to reduced glucose-stimulated insulin secretion and reduced insulin content. Importantly, these defects in insulin secretion and content were recapitulated in primary tissue from people who carried the PAM T2D-risk alleles. In addition to PAM’s direct effects on reducing granular insulin content and secretion, Dr. Gloyn hypothesized that PAM may lead to reduced incretin response via reduced amidation of GLP-1. Confirming this, a study using Oxford BioBank data found a ~50% reduction in PAM activity in the serum of carriers of PAM T2D-risk alleles vs. controls. Carriers also had higher levels of GLP-1 but no differences in the incretin effect, suggesting GLP-1 resistance. Indeed, an analysis of the DIRECTGLP, GoDARTS, and PRIBA studies found that only 0-10% of carriers reached their A1c targets on GLP-1s compared to 30% of non-carriers, suggesting that the PAM T2D-risk alleles reduce an individual’s response to GLP-1. Based on these findings, Dr. Gloyn recommended that people who carry the PAM T2D-risk allele should be treated with SGLT-2s, not GLP-1s. We’re excited to keep learning how research from genome-wide association studies can inform precision medicine for type 2 diabetes to better individualize treatment plans and lead to better outcomes.

Applying precision medicine to better understand the heterogeneity of diabetes diagnosis & treatment

Dr. Maria Jose Redondo (Baylor College of Medicine) advocated for the use of precision medicine to address the complexities of diagnosing and treating heterogeneous diabetes. There is significant overlap in the criteria defining type 1 and type 2 diabetes, which makes distinguishing between the two increasingly difficult. Dr. Redondo highlighted the mounting evidence that a combination of genetic and environmental factors is driving this heterogeneity within and between diabetes types.

  • Dr. Redondo proposed that providers should identify all clinically significant diabetogenic mechanisms that are playing a role in a person’s diabetes and may determine clinically significant outcomes, such as the trajectory of insulin needs. Classical markers for the different forms of diabetes, such as autoantibodies, BMI, and c-peptide, are not singularly correlated with one type of disease. These overlapping features also change over time, making it difficult to rely on these markers to diagnose a patient. Dr. Redondo gave some examples of difficult cases to diagnose, such as a person with evidence of only mild islet autoimmunity and signs of insulin resistance. Several frameworks have been developed to explain the heterogeneity, including a spectrum of diabetes model and the threshold hypothesis, which are shown in the figure below, can be combined to understand diabetes.

  • Dr. Redondo proposes precision medicine as possible treatment solution given that current classifications do not guide treatment nor prevention. Most people with type 1 diabetes need insulin but notably, there are exceptions, such as slow progressive autoimmune diabetes, which does not require insulin immediately after diagnosis. Generally, people with type 2 diabetes do not need insulin early in the diagnosis, but there are exceptions to this as well and some people with type 2 diabetes progress faster to insulin dependence. Dr. Redondo proposed pathophysiology-based treatments, a form of precision medicine, to address the intricacies of overlapping disease types.
  • Dr. Redondo also presented data about heterogeneity in pediatric patients. Many characteristics of pediatric diabetes are age dependent, such as appearance of islet autoantibodies, and the progression through pre-clinical stages of type 1 diabetes. Other characteristics at the onset of disease are also age-dependent, with children less than seven years old having unique immunologic characteristics and more dramatic loss of beta-cells. In addition, they have differential clinical characteristics, which are reviewed here. Notably, there is a strong genetic determinant contributing to the age of onset and residual beta-cell function in type 1 diabetes. There are also differences in type 2 diabetes and age of onset. For pediatric type 2 diabetes, patients typically progress to insulin deficiency faster than adult-onset patients. A diagnosis of type 2 diabetes during childhood or adolescence is associated with more frequent and severe complications, which is made worse by the shortage of drugs available for children with type 2 diabetes. We are excited to follow precision medicine closely as it becomes a part of diabetes prevention, diagnosis and treatment.

Precision medicine can be used to detect indicators of diabetes in childhood and adolescence

Dr. William Hagopian (Pacific Northwest Diabetes Research Institute, Washington) discusses potential reliable indicators for the early onset of diabetes, including DR4, early-age related CBV infection, and IAA appearance. Dr. Hagopian first outlined monogenic diabetes (MODY), a less common type of diabetes caused by a change or mutation to a single gene. MODY can in some cases be caused by mutations to immune system genes. Although MODY can be confused with classic polygenic type 1 diabetes, due to its similar activity in hindering the pancreas’s ability to produce enough insulin, the underlying cause of disease differs. Because of the potential risk of misdiagnosis, Dr. Hagopian argued that precision medicine can and should be used to diagnose diabetes. He outlined the precision medicine biomarkers that may be used to identify type 1 diabetes. 

  • Dr. Hagopian said that islet autoantibodies, type 1 diabetes genetic risk score, and serum C-peptide are usually very useful precision medicine tools to identify type 1 diabetes, but they cannot discern diabetes type in all cases. In unusual presentations or with equivocal biomarkers, other factors may be useful to consider such as age, ethnic and environmental background. Dr. Hagopian showed that in terms of age, autoantibodies appeared early in life even in individuals with late childhood diagnosis of diabetes. The older a person is at hyperglycemic onset, the slower the autoimmune process typically is which leads up to that onset. This means that the individuals’ immune response had longer to develop in those who got diabetes later in childhood and they may have a different islet autoantibody profile by the time of onset, such as less frequent insulin autoantibodies.
  •  In addition, Dr. Hagopian also noted that the immune system matures and changes as childhood proceeds, potentially explaining the different histological endotypes described by Morgan and Richardson. For example, children who are younger than seven years at diagnosis have a greater numbers of CD8+T cells and B-cells in their islets, with fewer and sicker residual beta cells at onset. On the other hand, children who are teenagers have fewer immune cells in their islets and more and healthier remaining beta cells, representing a more slowly progressing disease.

  • In the TEDDY study, time to event analysis was used to test early probiotic exposure (<27 days) and risk of IA. Results were most clear in those with the highest T1D genetic risk (HLA genes of both the DR3 and DR4 type) perhaps because more diabetes cases occurred with this genotype. In any case, it was highly significant that early probiotics were associated with fewer cases of islet autoantibodies. It makes biological sense since very early on, the gut microbiome is just forming and may be easier to modify than later, even later in infancy. To infer causation for the protection which bacteria in the probiotics seem to confer, however, will require a randomized controlled trial. 
  • Additionally, Dr. Hagopian showed that the Insulin autoantibodies (IAA) and Glutamic Acid Decarboxylase autoantibodies (GADA) are often the first islet autoantibodies to appear in young individuals destined to develop clinical diabetes. Respiratory infections were statistically associated with the appearance of islet autoantibodies in the TEDDY Study. Persistent Coxsackievirus (also called Enterovirus B) was found more often in the stool of young children developing islet autoantibodies, and this virus can cause respiratory infections. These findings, when taken together, may explain these observations. Those children with Coxsackievirus-associated type 1 diabetes most often have Insulin Autoantibodies (IAA) first, and most often have at least one HLA gene of the DR4 type. It is unknown if a different virus or a non-viral exposure triggers the GADA first form of diabetes but it seems to associate with the HLA DR3 gene.

Heterogeneity in diabetes can be partially addressed through precision medicine: Dr. Paul Franks reviews the merits and limitations of current stratification methods

Dr. Paul Franks (Lund University, Sweden) shined light on the need for precision medicine in treating diabetes and offered examples of ways in which this could be approached. To justify the importance of precision medicine, Dr. Franks highlighted the LookAHEAD trial, which demonstrated significant heterogeneity among responses to treatment in individuals with adult-onset type 2 diabetes. For instance, participants in the intensive lifestyle intervention group who had an early onset of type 2 diabetes experienced the lowest rates of adverse cardiovascular events; meanwhile, participants in the intensive lifestyle intervention group with poor glycemic control at baseline experienced the highest rates of adverse cardiovascular events. Dr. Franks emphasized that the field must shift from considering the average treatment effect alone to instead individualizing treatment processes, as broad generalizations about the population as a whole may not be the optimal solution. Noting that precision medicine seeks to reduce medical errors through optimizing disease characterization, Dr. Franks highlighted the fundamental principle of precision stratified medicine — to predict individual responses to treatment and allocate individuals to treatments that maximize efficacy and minimize side effects. 

  • Group-level stratifications in diabetes have been done both genetically and phenotypically. A 2019 study combined DNA sequence variants associated with a predisposition to diabetes along with an array of molecular and physiological factors to develop individual genetic risk scores. The scores were then used to predict the onset of type 1 and type 2 diabetes and create subclasses based on the risk of developing diabetes. Although there were several limitations to this study, such as a lack of calibration to an individual’s ethnicity, Dr. Franks regarded it as one of the pioneering papers in precision medicine for diabetes. Another study considered phenotypic clinical features including age at diagnosis, BMI, and HbA1c to develop five clusters, each of which was associated with a higher risk of a certain complication (e.g., retinopathy).
  • Dr. Franks noted key limitations of current stratification techniques. He emphasized that although studies have begun to formulate subclasses of diabetes, we cannot yet assign individuals to categories using prior studies, as this would require new studies involving detailed repeated-measures assessments for that individual. Additionally, most research, such as the DIRECT Study — which utilized 32 phenotypes to identify four archetypes — only uses discrete clusters. However, individuals tend to shift through clusters over time due to factors including aging, evolving health states, and changing treatment regimens. To address this issue, a recently published study utilized the DDRTree algorithm to create spatial maps of varying courses of disease and complications; these maps were then used to determine how complications may align with variations in drug responses, subsequently creating phenotypic subtypes based on resulting probabilities. The merit in this study, as Dr. Franks highlighted, was its use of later-stage stratification on a continuum rather than the use of discrete subclasses. Despite incredible progress on the precision medicine front, Dr. Franks concluded by noting that while classifications can inform treatment responses on a group level, an accurate, individual level categorization system has yet to come to fruition.

“Stop binarizing diabetes,” Dr. Richard David Leslie suggests; studies show patients with diabetes are not truly as binarized as their diagnoses imply

Dr. Richard David Leslie (University of London, UK) argued that physicians should "move beyond" diagnosing and treating patients with such a great focus on type 2 vs. type 1 classifications. To support this claim, Dr. Leslie presented three themes: (i) moving beyond binarization; (ii) moving beyond classification; and (iii) moving into therapy. Diabetes experts often equate juvenile-onset diabetes to insulin-dependency and insulin-intensive treatment, while equating adult-onset diabetes to non-insulin dependency and low or no insulin treatment. However, studies show these binary characterizations may not be representative of an individual’s true underlying disease mechanism.

  • Diagnostic tests and classifications show a continuous spectrum exists between the two types of diabetes. Clinicians commonly diagnose patients based on glycated hemoglobin levels, with very defined ranges used to diagnose the types of diabetes. Recall that a normoglycemia is considered to be ≤5.7%, a level of 5.7% to 6.4% indicates prediabetes, and a level of ≥6.5% indicates diabetes. However, glycated hemoglobin levels can vary by genetics and iron and glucose metabolism, and Dr. Leslie states that classifying diabetes based off specific ranges of glycated hemoglobin can be harmful as it can exclude cases at risk of both macrovascular and microvascular disease. Furthermore, diabetes is a progressive disease, and over time, patients categorized as "non-insulin dependent" may require insulin treatment. On the other hand, patients categorized as “insulin-dependent” may not require insulin treatment depending on adjunctive therapy, diabetes technology, or changing metabolism. Further, age at onset can be misleading as a classifier as most cases with type 1 diabetes are diagnosed as adults. Such data suggest that type 1 and type 2 diabetes represent extreme ends of a continuum rather than two logical binary categories with implications for the disease natural history and treatment.

  • Dr. Leslie proposed that we instead focus on individualizing therapy to each patient. Given the evidence supporting a diabetes continuum, diabetes diagnoses, therapies, technology, and research studies should be considered accordingly. Dr. Leslie argued it is more important to treat the specific individual case, employing clinical features and laboratory-based biomarkers, rather than a specific diabetes classification.

Glucose Monitoring – BGM and CGM

IMMEDIATE RCT: Non-insulin-using type 2s see 0.3% A1c reduction and 2.3 hour/day Time in Range improvement with FreeStyle Libre and diabetes education vs. diabetes education alone at 16 weeks

Dr. Ronnie Aronson (LMC Healthcare, Canada) presented results of the IMMEDIATE Study (95-OR), a multi-site RCT evaluating FreeStyle Libre and diabetes self-management education (DSME) vs. DSME alone in people with type 2 diabetes not on insulin (n=116, age 58, 10 years of diabetes, baseline A1c 8.6%). All participants had baseline A1c values ≥7.5%, had had type 2 diabetes for at least six months, and had no prior CGM use. Participants were on ≥1 non-insulin diabetes therapies, including metformin (98%), sulfonylurea (50%), SGLT-2s (39%), DPP-4s (45%), and GLP-1s (31%). Those who received FreeStyle Libre and DSME (n=58) saw a +2.3 hour/day Time in Range improvement at 16 weeks relative to those only receiving DSME (n=58) (76% vs. 66%; p=0.01). Baseline CGM metrics were not reported. This relative Time in Range improvement was not affected by the number of therapies, GLP-1 use, diabetes duration, or scanning frequency. Additionally, the FreeStyle Libre + DSME arm saw a -0.3% greater A1c improvement than those in the DSME only arm, falling 0.9% (8.4% to 7.6%) in the FreeStyle Libre group compared to -0.5% (8.6% to 8.1%) in the DSME only group (p=0.048). On the psychosocial front, those who used FreeStyle Libre saw improvements in their GMSS-based glucose monitoring satisfaction scores, rising from 3.4 to 3.9 (p<0.01), while those receiving DSME alone did not see an improvement. The average number of therapies that participants were on did not change from baseline to 16 weeks (2.5 therapies/participant on average in both groups), suggesting that the improvements seen in the FreeStyle Libre arm were likely due to behavior changes. These results offer two important takeaways: (i) CGM – even one without alarms like Libre “1” – improves glycemic outcomes for people with type 2 diabetes not on insulin even when therapies are not changed; and (ii) diabetes education is incredibly valuable, as those who received solely education still saw improvements in PROs and A1c. Currently, the IMMEDIATE Study is in its 16-week extension phase, during which those receiving DSME only switched over to use FreeStyle Libre while those already on FreeStyle Libre continued FreeStyle Libre use. We hope that when the IMMEDIATE Study is complete and is published, its findings will encourage changes in practice guidelines, payer decisions, and prescribing behavior. To us, it is clear that CGM is valuable across a wide range of people with diabetes, and we’re very moved to see researchers investing in building out the evidence base to support this belief.

 

FreeStyle Libre + DSME (n=41) at 16 weeks

DSME alone (n=41) at 16 weeks

Baseline-adjusted between-group difference at 16 weeks

p-value for baseline-adjusted difference

A1c

7.6%

8.1%

-0.3%

0.048

Time in Range

76%

66%

+2.3 hours/day

0.009

Time in tight range (70 mg/dl-140 mg/d)

50%

40%

+2.0 hours/day

0.042

Time above range

21%

31%

-1.9 hours/day

0.04

Time below range

1.9%

3.0%

-19 minutes/day

0.22

Time <54 mg/dl

0.6%

0.9%

-4 minutes/day

0.55

CV

27%

28%

0.7%

0.65

  • Beyond Time in Range and A1c, time above range and time in tight range (70 mg/dL-140 mg/dL) improved significantly more with FreeStyle Libre/DSME vs. DSME alone. Specifically, those using CGM spent two more hours/day in tight range (70 mg/dL-140 mg/dL) compared to those who only received DSME when adjusted for baseline (p=0.04). Likewise, those using CGM saw a 1.9 hour/day reduction in time above range compared to those on DSME alone (21% vs. 31%, respectively, at 16 weeks, p=0.04).
  • Although several outcomes did not differ between the groups, both those using Libre and those receiving DMSE only achieved the targets. Hypoglycemia was “rare” in both groups, and on average, both groups achieved the targets for <4% time below range and <1% time <54 mg/dL at 16 weeks. Both groups had an average <0.1 events/day of hypoglycemia. There were no significant differences in glycemic variability (% CV), but both were below 36%. Although the Libre group saw greater improvements in diabetes treatment satisfaction, there were no differences in adherence (ARMS-D), diabetes-related distress (DDS), and self-efficacy (SCPI) between those using FreeStyle Libre/DSME vs. DSME alone, although both groups saw improvements across these measures from baseline.

International Diabetes Center has directly integrated FreeStyle Libre data into Epic EHR for >60% of patients using CGM; UCLA’s Dr. Juan Espinoza argues CGM-EHR integration can eventually enable population-level management; Dr. Rich Bergenstal’s vision for CGM-EHR integration: “Why not have a dashboard of Time in Range across the country?”

During a professional-interest group session, Drs. Rich Bergenstal (International Diabetes Center) and Juan Espinoza (UCLA) highlighted developments in the growing push to integrate CGM data directly into the electronic medical record. As a reminder, the International Diabetes Center (IDC) easily had one of the top CGM-related sessions at ADA 2021, with Dr. Amy Criego (IDC) giving an exciting demonstration of IDC’s successful integration of Abbott LibreView data into Epic EHR. At the time, we could already tell this integration was going to be a big win for healthcare providers at IDC, as they no longer would have to navigate between multiple software platforms when reviewing data for patients using Abbott’s FreeStyle Libre CGM. Dr. Espinoza, too, made waves in the field at the recent Hospital Diabetes Meeting 2022, presenting on an interoperability framework for CGM data and explaining how the recently formed iCoDE Project (Integration of Continuous Glucose Monitor Data into the Electronic Health Record) is working toward technical standards and best practices for CGM-EHR integration. Kicking off the session, Dr. Espinoza shared the same interoperability framework as he did at the Hospital Diabetes Meeting earlier this year, and then pivoted to a framework with 32 questions and barriers that stand in the way of integrating CGM data into the EHR (see the picture below). Broadly, this framework illustrates how feasible solutions for CGM integration into the EHR will require interdisciplinary, multi-stakeholder approaches that include HCPs, senior practice leadership, industry representatives, IT engineers, funders, and PCPs. Dr. Espinoza said that because he is an informaticist, he would let Dr. Bergenstal continue and give the perspective on this subject as a provider who has successfully navigated this complex scheme of barriers.

  • In a hugely exciting and impressive update, Dr. Bergenstal shared that the IDC has successfully integrated CGM data directly into Epic EHR for >60% of their patients using CGM. According to Dr. Bergenstal, after the IDC successfully integrated LibreView into Epic EHR, it began a pilot trial (n=30) to test the integration at scale. However, Dr. Bergenstal explained that of these initial 30 patients, only one (!) successfully had their CGM integrated into Epic, with the other 29 failing due to issues such as legal name/birthdate mismatches, lack of consistent CGM use, issues during the linking process, clinic disruptions, and using the Libre reader as opposed to one’s smartphone. Dr. Bergenstal said that the IDC has taken a lot of time since to think about how to improve the success rate and realized that Epic’s MyChart app is a fantastic (albeit imperfect, perhaps) way to contact patients ahead of their clinic visit and prime them for CGM-EHR integration (see an example message below). Overall, while many barriers remain (e.g., getting existing CGM users to sign a data sharing agreement to permit CGM-EHR integration), Dr. Bergenstal explained that the IDC has been successful moving past this pilot trial – we certainly see this is the case given that IDC has succeeded in >60% of their CGM patients. Dr. Bergenstal closed by saying that of the 32 barriers Dr. Espinoza listed, IDC has covered about 26 of them, with only ~six left to tackle.

  • Dr. Bergenstal reiterated his core thesis from the Time in Range Academy’s inaugural session, saying that EHR integration can give clinicians more time to focus on action as opposed to analysis. If CGM data is automatically funneled into the EHR, Dr. Bergenstal envisioned that providers could walk into their clinic, see a patient that is coming in later during the day, and then immediately get a treatment recommendation for that patient because their CGM data was passed through some kind of treatment algorithm stored in the HER. In his example, Dr. Bergenstal specifically referenced the IDC’s novel C2GM therapy adjustment guide that he introduced at ATTD 2022.
  • Dr. Espinoza stressed that while phase “one” of the CGM-EHR movement involves data integration, the second phase involves thinking about how CGM-EHR data can power population-level insights. Dr. Espinoza explained that the scale of this next phase requires some kind of “staging domain” that has access to the larger bounty of CGM data in EHRs around the nation and can analyze and visualize the data so that we can derive meaningful insights from it. Dr. Bergenstal mused that if Johns Hopkins has a COVID-19 dashboard that visualizes nationwide and global COVID-19 cases, “why not have a dashboard of Time in Range across the country?” Our Associate team was giddy with excitement at this innovative vision. Dr. Bergenstal explicitly called on stakeholders to lay the groundwork to make this vision a reality, saying: “we encourage Epic, Cerner, MEDITECH, and Allscripts along with CGM Industry to embrace direct integration of CGM data (followed by smart pens and AID).”
  • Dr. Espinoza explained that the next iCoDE meeting will be on June 22, 2022, with the project set to conclude later this year in October/November. Importantly, Dr. Espinoza stated that “if we’re going to democratize” the ability for practices to integrate CGM data into the EHR, then “the technical part can’t be the hard part, because that’s a disservice to our patients, and not everyone has the same set of resources.” Encouragingly, Dr. Espinoza stated that important stakeholders have been a part of the iCoDE process, including but not limited to Medtronic, Abbott, Dexcom, Glooko, Tidepool, and the FDA. As he put it, the findings of the project will serve as the “termini” to guide future implementation and dissemination, and he made a commitment on behalf of the consortium to be as transparent as possible.

UK’s ABCD National Audit data (n=3,250) finds that FreeStyle Libre use is associated with reduction in resource utilization and hypoglycemia; plus, real-world validation of the association between achieving Time in Range >70% and improving A1c, diabetes distress, and hypoglycemia awareness

As part of the afternoon’s oral presentation session, Dr. Harshal Deshmukh (Hull York Medical Center, UK) read out a data-rich analysis of the UK’s ABCD National Audit evaluating improvements with FreeStyle Libre initiation and the correlation between Time in Range and glycemic and psychosocial outcomes (90-OR). The researchers pulled data from 16,427 FreeStyle Libre CGM users (96% type 1s) from 107 NKS UK hospitals, 3,250 (47%) of whom had follow-up Time in Range data. Among these 3,250 users, the mean follow-up was 7.6 months. The study’s primary aim was to test whether meeting the consensus target for Time in Range ≥70% was associated with A1c, hypoglycemia awareness, diabetes-related distress, and resource consumption; however, it also enabled the evaluation of the impact of FreeStyle Libre initiation on hypoglycemia and resource utilization. Although not the primary aim of the study, we were glad to see that the study showed that FreeStyle Libre use was associated with even more reductions in resource use and hypoglycemia. Compared to the year prior to FreeStyle Libre initiation, there was a 67% reduction in hospital admissions due to hypoglycemia, a 63% reduction in hyperglycemia/DKA-related hospital admissions, and an 85% reduction in paramedic callouts in the seven months following FreeStyle Libre initiation. There were no hospital admissions due to hypoglycemia among those with Time in Range values >50% after initiating FreeStyle Libre use. The reductions in DKA-related hospital admissions and paramedic callouts were independent of Time in Range. On hypoglycemia, the percentage of participants with impaired awareness of hypoglycemia (IAH) fell from 28% at baseline to 18% after FreeStyle Libre initiation. Impressively, over half (55%) of people with impaired hypoglycemia awareness at baseline restored their hypoglycemia awareness after using FreeStyle Libre (Gold score fell <4). The two factors that were associated with improved hypoglycemia awareness after initiating FreeStyle Libre were Time in Range (p=0.004) and duration of diabetes (p=0.001). The number of severe hypoglycemia events fell from 140 events/month in the year prior to FreeStyle Libre initiation to 48 events/month during the seven months following CGM initiation. The percentage of participants with severe hypoglycemia events also fell from 14% to 5%.

  • Of those with follow-up data, a minority (38%) had a Time in Range >70%. When adjusted for age, gender, baseline A1c, and diabetes duration, a follow-up Time in Range >70% was associated with a -1.3% A1c improvement, a 0.4 reduction in diabetes-related distress (evaluated by DDS2), and a 0.3 reduction in hypoglycemia unawareness (evaluated by Gold Score) relative to those with Time in Range values <50%. Even if participants weren’t achieving the consensus target for Time in Range, having a Time in Range value >50% still offered glycemic and quality of life benefits: a Time in Range ≥50% was associated with a 0.8% A1c reduction and a -0.3 reduction in diabetes-related distress relative to those with Time in Range values <50%. This offers the reminder that we’d always like people with diabetes to receive – that even if a PWD is unable to achieve the consensus target, moving toward that target from any base also enables significant benefits.
  • Although achieving the consensus Time in Range target was associated with positive outcomes, the use of the consensus target was low: of the 19% of participants who had follow-up Time in Range data, only 38% were using >70% as their personal Time in Range target. Those that used the consensus target were mostly similar to those who used a Time in Range target other than ≥70%: they had similar A1c values prior to initiating FreeStyle Libre (8.2% vs. 8.1%), were similarly aged (41 years vs. 42 years) and primarily White (77% vs. 88%), had similar BMIs (26 vs. 25), had similar diabetes durations (17 years vs. 18 years), and had similar baseline diabetes distress and hypoglycemia awareness scores. Dr. Deshmukh drew on this finding suggest that further work is necessary to increase awareness and use of the international consensus Time in Range target across the board.
  • The study offered insights into those who are most likely to achieve glycemic targets with FreeStyle Libre CGM: those who achieved a Time in Range >70% with FreeStyle Libre had lower A1c values, lower diabetes distress, and greater hypoglycemia awareness at baseline (pre-FreeStyle Libre) compared to those not achieving the target. Specifically, those achieving a Time in Range >70% with FreeStyle Libre had baseline A1c of 7.7% vs. 7.9% among those who achieved a Time in Range of 50%-70% vs. 9.2% among those whose Time in Range after using FreeStyle Libre was <50% (p<0.001 for trend). The rates of diabetes distress and hypoglycemia unawareness were generally moderate to low across the board, although there was a slight and significant trend between lower Time in Range and higher distress (p<0.001) and hypoglycemia unawareness (p=0.01).

Cost-effectiveness analysis finds that rt-CGM is cost-effective in insulin-using people with type 2 diabetes in Canada with ICER of CAN$18,523 relative to BGM

As part of the morning’s oral presentation session on health economics, Mr. John Isitt (Vyoo Agency) presented a Dexcom-sponsored analysis evaluating the cost-effectiveness of rt-CGM vs. BGM in insulin-using people with type 2 diabetes in Canada (134-OR). Overall, the analysis found that the incremental cost-effectiveness ratio of rt-CGM vs. BGM was CAN$18,523/QALY gained, which meets the Canadian willingness-to-pay threshold of CAN$50,000/QALY gained. There was a 67% of rt-CGM being cost-effective based on this threshold and a 30% probability of rt-CGM being cost-saving relative to BGM. The researchers used the Kaiser real-world data (n=36,080) that was published in JAMA in June 2021 in their cost-effectiveness analysis. As a reminder, the study was a retrospective cohort study of adults with type 2 on insulin therapy (mean age 64.5, baseline A1c 8.3%), which found that those who used rt-CGM saw a significant -0.56% A1c reduction relative to those on BGM at 12 months. The researchers also inputted the hypoglycemia and hyperglycemia benefits of rt-CGM into the model; these include a reduction in severe hypoglycemia events per 100 person-years, a reduction in hyperglycemia/DKA per 100 patient-years, and a quality-of-life benefit associated with no longer needing to use daily and frequent fingerstick testing . These clinical benefits, as well as the annual costs of rt-CGM and BGM (CAD$3,588 and CAN$1,096, respectively) were inputted into the IQVIA CORE diabetes model. It’s worth noting that only direct medical costs were used.. Furthermore, some of the clinical data inputted into the model was based on older rt-CGM devices (2014-2019), which the authors suggested “likely led to a more conservative approach in the current analysis.”

  • rt-CGM use was associated with a +0.95 improvement in QALYs compared to BGM (9.971 vs. 9.021). Total lifetime direct costs were CAN$17,603 higher with rt-CGM than with BGM (CAN$207,466 vs. CAN$189,863), which were associated with the increased annual cost of rt-CGM vs. SMBG. However, costs related to cardiovascular, renal, ulcer/amputation/neuropathy, and ophthalmic complications, as well as costs related to severe hypoglycemia and DKA, were all lower with rt-CGM vs. BGM.
  • As expected, lowering the cost of rt-CGM dramatically improved the ICER of rt-CGM vs. BGM. Were the cost of rt-CGM 25% lower, the ICER for rt-CGM vs. BGM would improve to an incredible CAN$5,026/QALY. rt-CGM became cost-saving relative to BGM if the cost of rt-CGM was reduced by 50%. Discounting future costs and outcomes from the base case of 1.5% to 0% reduced the ICER from CAN$18,523/QALY to CAN$16,853/QALY, while an increasing the discount of future costs and outcomes  to 3%  increased the ICER to CAN$ 20,172.
  • The analysis suggests that earlier initiation of rt-CGM in people with type 2 diabetes on insulin may make rt-CGM more cost-effective and potentially even cost-saving relative to BGM. As noted above, the clinical data that the analysis was based on was taken from the Kaiser study published in JAMA in which the average participant age at initiation of rt-CGM was 64.5 years. The researchers found that if the age of rt-CGM initiation was reduced to 50, the ICER would drop to CAN$7,651/QALY and that if rt-CGM started even earlier at age 40 or 30, rt-CGM would be cost-saving relative to BGM. While these results suggest that younger rt-CGM initiation would increase the cost-effectiveness of rt-CGM in type 2s on insulin, many people with type 2 diabetes don’t initiate insulin until later in disease duration; we look forward to seeing more research on the degree to which further delay of insulin initiation is positive or negative longer-term (in both early-onset and late-onset type 2 diabetes. Insulin at this stage is far cheaper than many drugs used “earlier” – the degree to which cardiovascular and kidney health improves in addition to glycemic health deserves far more attention, and we’ll be back on this after some consulting with our experts.

FreeStyle Libre 3 US accuracy trial (n=100 adults and children): MARD of 7.9% and ±20/20% of 93% vs. YSI (n=6,836 paired points); only 120 CGM-YSI pairs <70 mg/dL

In a late-breaking ePoster (76-LB), Abbott presented additional accuracy data (n=100) for the FreeStyle Libre 3 system, reporting a topline MARD of 7.9%. Abbott also published a press release on its website showing the sensor’s performance. As background, we saw data from a US accuracy trial (n=100 adults and children with 6,845 paired points) with FreeStyle Libre 3 at ATTD 2022, in which the topline MARD was 7.8% and the system had a ±20/20% of 93% compared to YSI. As always for the “current day,” there are lots of opinions on accuracy. The field has seen values ranging from 7.8% shared at ATTD 2022 to 7.9% in the system’s FDA clearance press announcement to 9.2% from the system’s CE-Marking in September 2020 – from our view, all the marks are positive, and anything under 10%, particularly given the value of the arrows, is top-rate for any CGM. It seems fairly straightforward from our view that since submissions for different regulatory agencies will have different sets of date, MARD wouldn’t be the same for each CGM. Additionally, as we understand it, MARD can also vary depending on where the “average” numbers are in a clinical data set – this makes it very hard to compare “head-to-head” accuracy data among different CGMs. To be sure, clinical data will likely continue to be different for FDA clearance and CE Mark, which will continue to impact  MARD data.

Ultimately, these MARDs demonstrate a strong accuracy profile that meets FDA special controls for iCGM clearance as well as non-adjunctive labeling. In any case, this prospective, non-randomized, single-arm, multicenter study assessed the performance of FreeStyle Libre 3 compared to YSI without glucose manipulation in 100 type 1s and 2s who were on insulin therapy and ages four and up (56 were ≥18 years old, 39 were 6-17 years old, and five were 4-5 years old). Participants wore two sensors (one on the back of each upper arm for precision and accuracy, respectively) for 14 days. For those ≥six years old, venous YSI measurements were collected over three eight-hour in-clinic sessions; for those aged four through five, BGM was used as a comparator during one four-hour in-clinic session. Overall

  • In the study, 93% of matched CGM-YSI pairs met the ±20/20% overall agreement rate. This agreement rate, which, as we understand it, is generally a more robust metric than overall MARD, places the FreeStyle Libre 3 sensor accuracy right around that of Dexcom G7. Again, for reference, Dexcom G7 reported a ±20/20% agreement rate of 95% and an overall MARD of 8.2% for upper arm wear and a ±20/20% agreement rate of 93% and an overall MARD of 9.1% for abdomen wear.

Cohort

Within ±15%/±15 mg/dL

Within ±20%/±20 mg/dL

Within ±40%/±40 mg/dL

Number of CGM-YSI pairs

MARD

Adults (18+ years)

89.1%

94.7%

99.7%

2,068

7.6%

Peds (6-17 years)

83.9%

89.7%

99.1%

4,768

8.7%

Overall (6+ years)

87.6%

93.2%

99.5%

6,836

7.9%

  • By day, FreeStyle Libre 3 demonstrated the strongest accuracy on days nine through 12 with the weakest accuracy on days seven and eight of sensor wear. While this sounds slightly surprising, since historically, the first couple of days are the least accurate, the MARD on the earliest days was close to that of Days 7 – 8 (8.5% vs. 9.1%, respectively). On days nine through 12, the MARD dropped significantly, to 6.6%, increasing to 7.1% through days 13 and 14. The trend of increasing accuracy after day one is common among CGMs and seen in all systems currently on the market.

Wear period

Within ±20 mg/dL and within 20% of reference

MARD

Beginning (Days 1 - 3)

92.3%

8.5%

Early middle (Days 7 - 8)

90.8%

9.1%

Late middle (Days 9 - 12)

95.8%

6.6%

End (Days 13 - 14)

94.5%

7.1%

Overall

93.2%

7.9%

  • Across glucose ranges, FreeStyle Libre 3 demonstrated strong accuracy and generated the most accurate readings in the high ranges of >250 mg/dL and 181 mg/dL – 250 mg/dL. Specifically, sensor MAD (mean absolute difference) was 16.1 mg/dL in the range of <54 mg/dL and 7.9 mg/dL in the range of 54-69 mg/dL. When in Range (70-180 mg/dL), MARD was 8.5%. Turning to non-severe hyperglycemia, MARD was impressively low at 6.4% in the range of 181 mg/dL – 250 mg/dL. While we find this data encouraging overall as accurate CGM data in both hypo and hyperglycemia is especially important to inform treatment decisions, hyper-accuracy is not quite as important as it used to be in the “before-arrow” era. Some have noted that there are very few paired points in the hypoglycemia range; we imagine some of this may be because the average A1c was higher going into the study than in seen in other studies. As there are now “arrows” that serve as customizable alerts that prevent very dangerous hypoglycemia in most cases, compared to BGM, we are not overly concerned about this at this point although we realize this puts us in a different spot than those who are overall purists. That said, we of course would also say that higher accuracy at all glucose levels is better, all else equal – from our view, how much “alarm” is necessary at various levels with lower accuracy continues to be controversial and we look forward to learning more on this front and particularly what impact this may have on AID.  

Glucose Level

Number of CGM-YSI Pairs

MAD

MARD

<54 mg/dL

15

16.1 mg/dL

n/a

54-69 mg/dL

105

7.9 mg/dL

n/a

70-180 mg/dL (in Range)

4,669

n/a

8.5%

181-250 mg/dL

1,640

n/a

6.4%

>250 mg/dL

407

n/a

5.0%

Overall

6,836

n/a

7.9%

Interim analysis from Medtronic’s fully-disposable Simplera CGM pivotal: MARD of 10.2% in adults, 10.7% in youth arms, and 10.1% in youth buttock; overall, 20%/20 mg/dL agreement rate of 91% in adults and 88% and 89% in youth arm and buttock

We finally got our first look at Medtronic’s next-gen CGM Simplera (f.k.a. Synergy) in the poster hall of ADA 2022 (672-P). As background, Simplera is Medtronic’s next-gen fully disposable, no-calibration CGM, is 50% smaller than Guardian Sensor 3 and Guardian 4. Based on the timeline shared in Medtronic’s 1Q22 call in May, Simplera is slated for FDA submission “this summer,” which is a slight delay from the previous expectation for submission by the end of April 2022. CE-Mark submission is also expected this summer. The data presented in this ADA poster is interim analysis of the Simplera pivotal trial, which will support FDA submission and includes participants from the US and China; however, this interim analysis only includes those from the US sites, as data collection at those sites completed first in November 2020. Since then, Medtronic has been completing the China-based portion of the study.

The interim analysis included 121 adults (age 45, A1c 7.3%, diabetes duration of 19 years) and 120 children (age 11, 8.2% A1c, diabetes duration of six years) from 13 sites in the US. Nearly all pediatric participants had type 1 diabetes (only 1 had type 2 diabetes), but the adult cohort was more mixed (60% type 1s and 40% type 2s). Adult participants wore two CGMs total on their arms, and pediatric participants wore 2-3 sensors on their arm and buttock. YSI was used as reference for adults and pediatrics over the age of seven, while SMBG served as reference for children six and under. During the seven day trial, adults participated in four in-clinic days, while pediatrics participated in two in-clinic days. Glucose manipulation was included as part of the in-clinic days for some participants, though it’s unclear what percentage had manipulations. In total, 15,388 paired points were generated from the arms of adults, 8,627 paired points from the arm of youth, and 7,781 paired points from the buttock of youth.

 

Overall MARD

MARD (70-180 mg/dL)

MARD <70 mg/dL

MARD >180 mg/dL

Overall 20%/20 mg/dL agreement rate

Adult arm

10.2%

10.1%

13.7%

8.7%

91%

Youth arm

10.7%

10.7%

15%

9.4%

88%

Youth buttock

10.1%

10.1%

16.1%

7.8%

89%

  • MARD values were 10.2% for adults, 10.7% for youth arm, and 10.1% for youth buttock measurements. For glucose measurements <70 mg/dL, MARD values were 13.7%, 15%, and 16.1% for adult, youth arm, and youth buttock, respectively; and for values >180 mg/dL, MARD was 8.7%, 9.4%, and 7.8% for adult, youth arm, and youth buttock, respectively. Overall, the 20%/2o mg/dL agreement rate was 91% for adults and 88% and 89% for youth arm and buttock, respectively.
  • Though Medtronic has never stated intentions to submit Simplera as an iCGM, it is worth comparing this interim pivotal trial data against the accuracy standards specified by iCGM special controls. Of course, differences in the way special controls are written and the way results were presented in the poster make an exact assessment difficult. In general, Simplera appears to be near or above these accuracy targets and an impressive step-up from Medtronic’s previous sensor, although we’d note that this is still below the accuracy data we’ve seen for Dexcom and Abbott CGMs (example: Dexcom G7 preschool accuracy data presented at ATTD 2022).

Performance Standard (lower-bound of 95% CI)

Standard Met?

Relevant Interim Pivotal Trial Result (Adults)

Euglycemia: >70% within ±15% for 70-180 mg/dl

Unknown

No relevant result presented

Euglycemia: >99% within ±40% for 70-180 mg/dl

Unknown

No relevant result presented

Overall: >87% within ±20% over full device measuring range

Likely Yes

Overall 20%/2o mg/dL agreement rate was 91%

Hypoglycemia: >85% within ±15 mg/dl for <70 mg/dl

Likely Yes

Agreement rate of 90.1%

Hypoglycemia:>98% within ±40 mg/dl for <70 mg/dl

Unknown

No relevant result presented

Hyperglycemia:>80% within ±15% for >180 mg/dl

Likely Yes

Agreement rate of 87.6%

Hyperglycemia: >99% within ±40% for >180 mg/dl

Unknown

No relevant result presented

  • In addition to being accurate, Simplera was shown to be safe. Among both adults and youth, there were no severe hyperglycemia events, DKA, or severe hypoglycemia events. There was one non-device-related severe adverse event in both adults and youth.
  • Adult participants and the parents of pediatric participants were quite satisfied with the sensor in terms of insertion, onboarding, and stability during activities. On average, both adults and parents “strongly agreed” that the sensor insertion process was easy and exceeded their expectations and “agreed” that the process was faster and more comfortable than inserting their previous sensor. Adult participants “strongly agreed” that they could insert the sensor on their own, and parents “agreed” that their children could insert the sensor on their own. On onboarding, adults and parents “agreed” that the sensor was easier to learn than previous sensors used, potentially because of the zero calibrations required with the system.

Two Dexcom-sponsored posters on Dexcom G6 in type 2 diabetes: Retrospective EHR analysis finds large (at least 0.8%) A1c reduction across type 2s not on insulin, on basal-only, and on MDI; observational prospective study finds that 35% of non-bolus-using type 2s see ≥5% improvement from high baseline

Two Dexcom-sponsored posters on CGM in type 2 diabetes caught our eye in the poster hall. In particular, we were impressed by a large retrospective EHR analysis of Dexcom G6’s impact on glycemic outcomes in primary care-treated type 2 diabetes in 13 health centers between August 2015 and September 2020 (687-P). The analysis included a broad range of people with type 2 diabetes (n=458, age 61), including 64 on diabetes medications but not insulin (14%), 51 on basal-only therapy (11%), and 343 on bolus insulin with or without basal insulin (75%). Across the full cohort, the vast majority (84%) were White, half were on commercial insurance, and half were women. Two-thirds of the cohort had baseline A1c values >7.5% (n=306) with the remaining third having an A1c ≤7.5% at baseline. Those with baseline A1c values ≤7.5% did not see a significant A1c reduction with G6 from baseline to 3-9 months of follow-up (p=0.123), although we wonder if CGM metrics were evaluated whether there may have been an improvement in Time in Range and in time below range. While those with baseline A1c values ≤7.5% did not see a significant improvement with G6, those with baseline A1c values >7.5% saw an average -0.9% A1c reduction from baseline (18 months prior to G6 initiation) to three to nine months after initiating G6 (p<0.001). Notably, those with baseline A1c values >7.5% in all three subgroups (no insulin, basal-only, and bolus insulin users) saw significant improvements in their A1c with G6: non-insulin users saw a 1.3% A1c improvement (p<0.01), those on basal-only therapy saw a 1.6% A1c improvement (p<0.001), and those on bolus insulin saw a 0.8% A1c improvement (p<0.001). Fascinatingly, this suggests that non-insulin users may benefit even more from G6 (on an A1c basis) than those on bolus insulin, contrary to popular belief.

Baseline A1c >7.5%

 

Baseline A1c

Follow-up A1c

A1c Change

p-value

Non-insulin users

9.3%

8.1%

-1.1%

0.01

Basal-only insulin users

10%

8.5%

-1.6%

<0.001

MDI users

9.2%

8.5%

-0.8%

<0.001

Baseline A1c ≤7.5%

 

Baseline A1c

Follow-up A1c

A1c Change

p-value

Non-insulin users

6.5%

6.5%

0%

0.98

Basal-only insulin users

6.7%

6.8%

+0.1%

0.63

MDI users

6.8%

7%

+0.2%

0.11

  • Elsewhere in the poster hall, another Dexcom-sponsored poster (669-P) displayed the results of a 12-week observational study evaluating Dexcom G6 in 150 non-bolus-insulin-using adults with type 2 diabetes (i.e., those on basal-only therapy or no insulin). Participants averaged age 57 and were more representative of the US type 2 population than many studies (18% Black, 10% Asian, and 32% Hispanic). Overall, participants did not see a significant improvement in Time in Range from week one to week 12; however, average Time in Range was already quite high in week one (84%), and this was maintained out to 12 weeks (82%). Among the 35% of participants (n=53) who saw a >5% Time in Range improvement, Time in Range improved 3.8 hours/day from 72% at week one to 88% at week 12. This group also saw reductions in weekly severe hypoglycemia events and weekly hyperglycemia events. Particularly stark Time in Range improvements were seen during the daytime: Time in Range improved by +6.5 hours/day in the morning (67% to 94%), by +4.8 hours/day in the afternoon (68% to 88%), and by +5 hours/day in the evening (63% to 84%). It’s worth drawing attention to the design of this study: comparisons were not made between baseline and week 12 but rather week one of G6 use and week 12 of G6 use, meaning that this data enables insights into how users are benefitting from CGM learnings over time.

Investigator-initiated “Dexcom G6 Intervention Study” hospital RCT (n=162): CGM prevents recurrent inpatient hypoglycemia vs. BGM while improving Time Below Range; overall, CGM and BGM drive statistically equivalent glycemic outcomes; 64% of participants identify as Black

In a late-breaking poster (140-LB), Dr. Elias Spanakis (University of Maryland) and co-investigators presented topline results from the “Dexcom G6 Intervention Study” (n=162), a multicenter RCT comparing the glycemic control achieved through insulin adjustment by point-of-care (POC) testing compared to insulin adjustment from CGM readings in hospitalized type 1s and type 2s on intensive insulin therapy. This Dexcom-sponsored, investigator-initiated study builds on a previous investigation by Dr. Guillermo Umpierrez (Emory University), which assessed Dexcom G6 compared POC testing in detecting hypoglycemia and hyperglycemia (read here for more from Dr. Umpierrez). The Dexcom G6 Intervention Study is one of the several trials underway assessing Dexcom G6’s safety and efficacy in the hospital, and this trial readout arguably marks the most impactful data we’ve seen on the in-hospital use of a Dexcom CGM since the company received FDA Breakthrough Designation for a hospital-specific CGM system in March 2022. This RCT was conducted at Grady Memorial Hospital, Emory University, and the University of Maryland and included 162 participants over 18 years old on basal-bolus insulin regimens with an anticipated length of hospitalization over three days. Across the point-of-care (n=79) and rt-CGM cohorts (n=83), participants were predominantly male (63% and 58%, respectively), Black (66% and 64%, respectively), and had type 2 diabetes (91% and 88%, respectively). In both cohorts at baseline, the mean A1c was 9.5%, the mean diabetes duration was roughly 15 years, and approximately 50% of each cohort was on insulin (10% – 15% of each cohort on oral agents). The primary outcome of the study was the difference in hypoglycemia (<70 mg/dL) between the rt-CGM and POC arm, with key secondary endpoints including Time in Range (70 – 180 mg/dL), Time Below Range (<70 and <54 mg/dL), Time Above Range (>180 and >250 mg/dL), mean glucose, recurrent hypoglycemia, and recurrent nocturnal hypoglycemia.

  • There was no statistically significant difference between the glycemic control and Time in Range outcomes between the rt-CGM and POC cohorts. In the POC study arm, Time in Range was 49% at the conclusion of the study, compared to 55% in the rt-CGM arm (p=0.14). Of course, baseline values would have been hugely helpful to better understand whether CGM drove a greater improvement in Time in Range compared to the POC arm, as well as to understand more generally how many more hours per day participants were spending in Range. Similar to Time in Range, there were no statistically significant differences in Time Above Range, Time Below Range, and mean glucose across the two cohorts. While it’s disappointing not to see improved glycemic management in the CGM cohort, especially since this trial was an RCT, we also imagine that the complexities inherent in managing diabetes in the hospital, as well as navigating the changing clinical conditions in the hospital, could be in part responsible. From our view, it’s still notable that there was a trend toward higher Time in Range in the CGM arm compared to the POC arm, even if it is not statistically significant. Indeed, the results still suggest that the inpatient use of Dexcom G6 is safe and effective in guiding insulin adjustment in the hospital.
  • Dexcom G6 was incredibly successful at preventing recurrent hypoglycemia in the hospital compared to POC testing, both overall and at night. For users with ≥one hypoglycemic event, there were significantly fewer hypoglycemic events per patient for rt-CGM compared to POC overall (1.8 vs. 3.0, respectively; p=0.04) and at night (1.2 vs. 1.9, respectively; p=0.02). Similarly, Dexcom G6 led to an improved Time Below Range compared to POC testing for those with ≥one hypoglycemic event during the course of the study, overall (1.9% vs. 5.5%, respectively; p=0.02) and at nighttime only (1.3% vs. 4.3%, respectively; p=0.004). Impressively, these results suggest that the inpatient use of Dexcom G6, beyond being safe and effective in the hospital, can also significantly reduce diurnal and nocturnal hypoglycemia compared to POC glucose testing. We must admit, we’re wondering how these results will inform Dexcom’s approach to building an in-hospital CGM system.

 

Overall

Night

CGM

POC

p-value

CGM

POC

p-value

Hypoglycemic Events (events/patient)

1.8

3.0

p=0.04

1.2

1.9

p=0.02

Time Below Range

1.9%

5.5%

p=0.02

1.3%

4.3%

p=0.004

  • Importantly, the population in the Dexcom G6 Intervention Study represented a very diverse population from a racial/ethnic perspective. We’re especially glad to see the diverse sample given that this is an RCT studying CGM in a population that, from our view, is still a nascent setting for CGM use. The study population contained predominantly (64%) Black individuals (n=105 of 162), followed by White individuals (n=42 of 162, 26% of cohort), Hispanic individuals (n=12 of 162, 7% of cohort), and then self-reported “other” individuals (n=3 of 162, 2% of cohort). Additionally, the demographic breakdown of the respective study arms was representative of the larger cohort itself.

Retrospective analysis of Canadian private payer claims data finds that adults with type 2 diabetes (n=373,871) are 86% to 181% more likely to progress their diabetes therapy if using FreeStyle Libre rather than BGM

In the poster hall (680-P), a retrospective Canadian private payer claims analysis shows that type 2s who used FreeStyle Libre CGM were more likely to intensify their diabetes therapy treatment than those on BGM, suggesting that FreeStyle Libre use may facilitate earlier treatment intensification and improve therapeutic inertia. The massive analysis included 373,871 adults with type 2 diabetes in Canada with data from May 2018-April 2021. The majority (63%) were on existing diabetes therapy at the beginning of the analysis. Using claims data, participants were classified into levels of therapy progression each month; the treatment categories included: (1) no diabetes drug therapy; (2) single oral agent; (3) two oral agents; (4) three oral agents; (5) four or more oral agents; (6) injectable GLP-1 (with or without oral agents; (7) basal insulin (with or without oral agents); and (8) MDI insulin (with or without oral agents.

  • Overall, those who used FreeStyle Libre were 86% to 181% more likely to progress their treatment than those on BGM. FreeStyle Libre users who were naïve to diabetes therapy at the beginning of the analysis and didn’t use insulin in the analysis period were 86% more likely to intensify their treatment than their BGM-using counterparts (relative risk over 24 months = 1.86). This difference expanded further when looking at: (i) those naïve to diabetes therapy who used insulin during the analysis period (112% more likely to intensify treatment if used FreeStyle Libre; HR=1.86); (ii) those who were on existing diabetes therapy and didn’t use insulin in the study period (103% more likely to intensify treatment if used FreeStyle Libre; HR=2.12); and (iii) those who were on existing therapies at baseline and used insulin in the analysis period (181% more likely to intensify treatment if used FreeStyle Libre; HR=2.81). These trends were true across all treatment categories in both those who were treatment naïve and those who were on an existing treatment, as is shown in the figures below.
  • These results are fascinating, in particular because clinical trials using CGM in people with type 2 diabetes (e.g., the MOBILE study) haven’t always shown treatment intensification with CGM use. (We would also emphasize that various studies have various variable components and they shouldn’t be directly compared without this acknowledgement.) Findings from this research make us hopeful that in the real world, CGM is being used to guide therapy decisions; we expect that the new C2GM algorithm from International Diabetes Center and Dr. Richard Bergenstal and team will further support CGM-guided treatment decision-making in type 2 diabetes.

 

 

Relative risk of treatment intensification at any point over 24 months (FreeStyle Libre CGM user vs. BGM user)

Naïve to diabetes therapy at start of analysis period (n=137,110)

Did not use insulin during the analysis period

1.86

Use insulin during the analysis period

2.12

On existing diabetes therapy at start of analysis period (n=236,761)

Did not use insulin during the analysis period

2.03

Used insulin during the analysis period

2.81

p-value for all relative risk outcomes <0.001

Pilot data from Senseonics’ 365-day implantable CGM demonstrates MARD of 9.8% (n=31,482 paired glucose measurements)

In a poster presentation (665-P), Senseonics presented new data from an extension of the PROMISE study evaluating the accuracy and longevity of its Eversense CGM out to 365 days finding an overall sensor MARD of 9.8% as compared to BGM. Of note, this data is not from the company’s 365-day sensor (IDE submission for the 365-day device is expected in 2Q22), but from 30 sensors used in participants in the PROMISE study (read out at ATTD 2021) that had modified sensor chemistry to extend the glucose-binding lifespan of the device. For the first 180 days of this study, participants underwent YSI glucose comparisons per the protocol of the PROMISE study. For the full 365-day duration of the study, participants also conducted 4-7 BGM measurements per day and these data were used as comparator points to assess accuracy after 180-days. There were a total of 31,482 paired sensor and BGM glucose values for an overall sensor MARD through 365 days of 9.8%. Using BGM comparator values, sensor MARD through 180-days was 9.5% (n=20,901 paired measurements) and sensor MARD from days 181-365 of 10.4% (n=10,607 paired measurements). While accuracy does appear to have decreased in the extended life of the sensor, we are very impressed by the overall MARD of 9.8% for the full 365-day use and will be interested to see how this compares to accuracy from Senseonics’ 365-day sensor once that pivotal trial begins, which, as of the company’s 1Q22 update, is expected in 4Q22.

  • Senseonics’ 365-day device performed well across a range of glucose values with the highest degree of 365-day accuracy in the hyperglycemic range of 301-350 mg/dL with a MARD of 8.8%. That said, the system also demonstrated strong accuracy in the hypoglycemic ranges of 40-60 mg/dL and 61-80 mg/dL with MAD values of 9.9% and 9.5%, respectively.
  • In the PROMISE study, Senseonics’ 180-day sensor was found to have a MARD of 9.1% when compared to YSI which is slightly lower than the MARD of 9.5% as compared to BGM that is reported through the first 180 days of these data. However, when compared to YSI, MARD through 180-days decreased to 8.1% reminding us of the importance of study design and the challenges of comparing MARD and other accuracy data across CGM systems. We also want to note that when Eversense E3 received FDA approval in February, Senseonics shared that the data used in the device’s FDA submission demonstrated a MARD of 8.5%. While we are so appreciative of all the work CGM manufacturers have done to bring down the MARD of CGMs from the mid 20% range of early devices to <10% in currently available systems, we do continue to wonder what impact these smaller differences in MARD will have on patient and provider experiences using the devices and continue to see non-adjunctive labeling, which Eversense E3 has, as the hallmark of CGM clinical utility. There is clearly lots of competition in the field – we are keen, of course, for special life circumstances like pregnancy to be getting easier and easier sensors to use.

First data out of UnitedHealth Group’s Level2 type 2 management program: Those with CGM data at follow-up see -0.8% A1c/GMI reduction from 7.7% to 7.0%; 70% of those with baseline A1c values 7%-9% and 94% of those with baseline A1c values >9% see ≥0.5% improvement

The poster hall offered a first look at the outcomes achieved with UnitedHealth Group’s Level2 type 2 diabetes management program, overall displaying strong engagement and glycemic results among those who participated and wore CGM (695-P). As a reminder, Level2 is UnitedHealth Group’s type 2 diabetes remission program that launched in July 2020 following a successful pilot. The intervention provides eligible members a Dexcom G6 continuous glucose monitor, a Fitbit activity tracker, smartphone app-based alerts, personalized clinical coaching, and virtual specialist consultations. With 230,000 employer-sponsored UnitedHealthcare members with type 2 diabetes as eligible to enroll, we believe Level2 is the largest type 2 diabetes program offering CGM to-date, which makes this data readout particularly significant for the broader digital coaching arena.

This retrospective assessment of engagement and glycemic outcomes over 26 weeks after Level2 enrollment included 7,886 participants who joined Level2 in January-July 2021 (average age 55). Based on medical and pharmacy claims data, most (85%) had been on one or two diabetes therapies over the year prior to enrolling in Level2, 18% had filled prescriptions for three diabetes medications, and 7% had filled prescriptions for four or more diabetes medications. About half were on metformin, 28% were on a GLP-1, 22% were on SGLT-2s, 22% were basal insulins, 11% were on rapid-acting insulin, 21% were on a sulfonylurea, 10% were on DPP-4s, and 5% were on TZD. Participants averaged a Diabetes Complications and Severity Index (measure of existing and at risk diabetes-related complications; scale 0-2) of 0.7, suggests some but not severe complications.

  • Engagement was high in the 26 weeks following Level2 initiation. Over the 26 weeks following enrollment, participants averaged 175 days of enrollment and 11 coaching interactions. Over two-thirds had at least one coaching interaction over 26 weeks with 60% having more than four interactions, suggesting that participants were highly engaged in the program. Two-thirds (n=5,886) wore a CGM at least once. Characteristics associated with coaching visit frequency and CGM wear included a younger age and nephropathy-related claims history.
  • Of the two-thirds of participants who wore CGM at least once, 7% had baseline A1c data and sufficient CGM data at week 12-13 and 25-26 (n=378). These participants were included in a subanalysis evaluating change in glycemic control using baseline A1c and follow-up GMI, which is a limitation of the study given that A1c and GMI are correlated, but not equivalent, which likely impacts the results. Overall, GMI/A1c improved by an adjusted -0.8% in these CGM wearers, falling from 7.7% at baseline to 7% at 26 weeks. Those with baseline A1c values <7% (n=125) saw an increase in A1c/GMI (+0.3% from 6.3% to 6.6%), potentially due to a reduction in hypoglycemia; those with baseline A1c values of 7%-9% (n=184) saw a nonsignificant trend toward a -0.3% A1c/GMI improvement; and those with baseline A1c values ≥9% (n=69) saw a whopping 2.6% A1c/GMI improvement from 10.1% to 7.5%. Just over half (55%) of these participants saw a significant ≥0.5% A1c/GMI improvement. This increased to 70% among those with baseline A1c values 7%-9% and to 94% among those with baseline A1c values >9%.

 

Baseline A1c

Final GMI at week 25-26

A1c change (* for p<0.05)

Percentage of participants who saw >0.5% reduction

All included in subanalysis (n=378)

7.7%

7.0%

-0.8% *

54%

Baseline A1c <7% (n=125)

6.3%

6.6%

+0.3% *

9%

Baseline A1c 7%-9% (n=184)

7.8%

7.0%

-0.8%

70%

Baseline A1c >9% (n=69)

10.1%

7.5%

-2.6%

94%

  • As noted above, this is the first data we’ve seen from Level2 since it launched in July 2020. In the announcement of that launch, UnitedHealth shared some tidbits from its pilot trial (n=790), which are in line with the results shared in this ADA 2022 poster. Specifically, UHG’s press announcement for the full launch shared that in the pilot, “certain” participants achieved a clinically meaningful reduction in A1c, and those with baseline A1c values >8% saw on average a >1% reduction. Though not quantified, “some Level2 participants” achieved type 2 diabetes “remissions” (A1c <7% and no longer required medication). In the pilot, Level2 eliminated the need for more than 450 prescriptions (~0.6 medications/participant).

“As complexity increases, likelihood of success decreases”: Dr. Tom Martens presents actionable CGM-driven primary care treatment algorithm for patients with type 2 diabetes on MDI therapy

Dr. Thomas Martens (International Diabetes Center) presented a novel CGM-driven insulin treatment algorithm for people with type 2 diabetes on MDI therapy. Dr. Martens’ presentation built upon that of his colleague Dr. Rich Bergenstal from ATTD 2022 during which Dr. Bergenstal presented the C2GM algorithm (CGM Clinician Guided Management) for CGM-driven basal insulin therapy in people with type 2 diabetes. As a reminder, the C2GM algorithm is a simple three-step process to use CGM data, specifically Time in Range and Time Below Range, to help providers, especially those in primary care, avoid clinical inertia and optimize glycemic management. In today’s presentation, Dr. Martens presented the second iteration of the C2GM algorithm tailored to patients with type 2 diabetes on MDI therapy (for full details on how Drs. Martens and Bergenstal developed the algorithm and why a Time Below Range cutoff of >2% was chosen, see our coverage from ATTD 2022). We remain incredibly impressed by the work conducted by Drs. Martens and Bergenstal, as well as the rest of their working group, to develop these two treatment algorithms as making CGM data actionable appears to be one of the next hurdles in improving glycemic management for people using the technology. As Dr. Martens explained, there are an estimated ~3 million people in the US with type 2 diabetes on MDI therapy, many of whom receive care in primary care and may be struggling to achieve glycemic management targets. In this vein, Dr. Martens acknowledged the challenges of optimizing MDI therapy, which he described as a “complicated” therapy, saying “as complexity increases the likelihood of success decreases” indicating that for many people with type 2 diabetes and their providers, the complicated nature of MDI therapy may hinder their ability to achieve improved glycemic management. To address this, Dr. Martens and his colleagues developed the “CGM Guided Background and Mealtime Insulin Adjustment for Type 2 Diabetes” algorithm to help guide MDI therapy for type 2s receiving care in primary care. Similarly to the C2GM algorithm presented at ATTD, this algorithm includes three steps: (i) determining if the patient has comorbidities that could be mitigated with a GLP-1 or SGLT-2; (ii) identifying the % Time in Range and % Time Below Range based on CGM data; and (iii) using Time in Range and Time Below Range cohorts to adjust insulin and glycemic management therapeutics. Specifically, for patients with type 2 diabetes on MDI therapy, Dr. Martens outlined the following clinical adjustment guidelines:

  • Time in Range >70% and Time Below Range <2%: Continue regimen to optimize current therapy and reinforce lifestyle changes and taking insulin as prescribed with recommended follow-up in 3-4 months.
  • Time in Range >70% and Time Below Range >2%: Address hypoglycemia and if patients are experiencing hypoglycemia overnight reduce background insulin dose. If patient is low at one specific time point during the day, decrease the mealtime dose directly preceding the low. If patient experience hypoglycemia consistently during the day, decrease all background and mealtime insulin doses. For insulin reductions, Dr. Martens laid out recommendations based on the percentage of time patient spent below range: (i) if Time Below Range is 2%-4%, reduce total insulin dose by 5%; (ii) if Time Below Range is 4-12%, reduce total insulin dose by 10%; and, if Time Below Range is >12%, reduce total insulin dose by 15%.
  • Time in Range ≤70% and Time Below range <2%: Address hyperglycemia and consider adding a GLP-1 agonist. Otherwise, if Time in Range is <50%, increase total daily insulin dose by 10% and redistribute insulin dose to 50-50 between background and meal-time insulin dose. If Time in Range is 50-70%, but patient has postprandial glycemic increases of >50 mg/dL, increase the mealtime bolus for that meal by 10%. Finally, if glucose is high overnight, increase background insulin by 10%.
  • Time in Range ≤70% and Time Below Range >2%: Address hypoglycemia and refer the patient to a diabetes educator to move into a team-based management plan. With a focus on hypoglycemia, all background and mealtime insulin should be reduced according to the percentage of Time Below Range: (i) if Time Below Range is 2%-4%, reduce total insulin dose by 5%; (ii) if Time Below Range is 4-12%, reduce total insulin dose by 10%; and, if Time Below Range is >12%, reduce total insulin dose by 15%.

  • When discussing clinical care for patients with type 2 diabetes on MDI, Dr. Martens emphasized the role of team-based care in driving improved outcomes. Dr. Martens was especially adamant about this point when it comes to the management of type 2 diabetes in primary care to ensure that CGM data is turned into clinical action by either physicians or educators to help people with type 2 diabetes achieve improvements in glycemic management. Given the complexity of MDI therapy, Dr. Martens also discussed how team-based management can take the burden of diabetes management off of primary care providers and transition some of the responsibility to diabetes educators who may be able to spend more time with patients or have greater comfort helping patients manage diabetes. While we certainly agree with Dr. Martens assertion that team-based management can help patients achieve stronger glycemic management, we are also aware that there any many providers in primary care who may not have access to diabetes educators once again highlighting the importance of straightforward tools like the C2GM algorithm that can allow primary care providers to adjust their patients’ diabetes treatment regimen quickly, easily, and safely.
  • Dr. Martens also provided tips for the effective implementation of CGM-driven insulin titration regardless of whether patients are on basal-only or MDI insulin therapy. Specifically, Dr. Martens outlined six tips for providers including:
    • Changing therapy based on patterns of hypo and hyperglycemia instead of isolated outlier events;
    • Including conversations around diet and nutrition if patients consistently see rises in postprandial glucose that could be indicative of high carbohydrate foods and beverages;
    • Considering challenges patients may face when trying to follow their insulin regimens (e.g., missed or rationed insulin doses, injecting the wrong kind of insulin, incorrect timing of meal-time boluses, lipohypertrophy at the site of injection, and insulin stacking);
    • Recognizing potential causes for sudden increases in Time Above Range (e.g., use of expired or improperly stored insulin, acute infection/illness, initiation of steroid therapy, or rationed insulin);
    • Considering whether excessive alcohol intake may be related to hypoglycemic episodes; and
    • Considering a potential misdiagnosis of type 2 diabetes that is actually undiagnosed type 1 diabetes.

Full readout from Stanford’s “4T” study: New-onset pediatric type 1s with early CGM initiation achieve 0.6% lower A1c vs. pilot 4T cohort due to lower A1c target (<7% vs. <7.5%); 67% of participants achieve A1c <7% from mean baseline of 12.3%

Dr. Priya Prahalad (Stanford) presented data from Stanford’s “4T” study of newly diagnosed pediatric type 1s, showing that reducing the A1c goal from <7.5% in the pilot 4T study to <7% in the full study led to a relative reduction in A1c. As a reminder, the four Ts in the “4T” pilot study name are teamwork, targets, technology, and tight control. We saw the readout of the pilot study data at ATTD 2022, showing that early CGM use and timely interventions led to a significant increase in the percentage of people with A1c <7% one year after diagnosis, from 28% to 53%. In the new study (study 1) that was read out today, youth with newly diagnosed type 1 diabetes between June 2020 and February 2022 were offered CGM initiation in the first month of diagnosis, along with remote patient monitoring (4T study 1 cohort, n=133). The use of RPM allowed Stanford researchers to develop a “Timely Intervention for Diabetes Excellence” (TIDE) tool, which takes CGM data from patients, analyzes the data, and presents results in a dashboard that makes it easy to identify patients that may need more or less help. The investigators compared outcomes in this cohort against the outcomes from the pilot cohort (n=135), as well as against historical data from people not offered CGM close to diagnosis (n=272). Mean A1c at diagnosis was similar in the 4T and Pilot 4T cohorts (12.3% and 12.2%, respectively) but was lower in the historical cohort (10.7%).

  • Tighter A1c targets in the new 4T study 1 (<7% vs. <7.5% in the pilot study) led to a 0.6% relative A1c reduction compared to the pilot 4T program. While the A1c improvement was already 0.3% at six- and nine-months post-diabetes diagnosis, it’s quite impressive to see that the relative A1c improvement increased to -0.6% at one year. Notably, a greater percentage of youth achieved an A1c target of <7% at 12 months in the new 4T study 1 compared to the pilot (67% vs. 50%), without any significant increase in hypoglycemia. Considering the cohort’s high baseline A1c, these results are staggering.

  • Dr. Prahalad reiterated commentary from ATTD 2022 that a second study with 4T has already begun and includes AID education between months 1-3 after diagnosis to encourage early AID use. 4T Study 1 had an exercise sub-study and the investigators also plan to scale the exercise sub-study to Spanish-speaking families and to share the approach with other clinics. Additionally, this approach has been scaled to patients with established diabetes.

Two posters (n=3,346) demonstrate association between CGM use among adult type 2s and younger age, insulin use, and higher A1c; overall CGM uptake remains low at 12%

Two late-breaking posters at this year’s ADA (62-LB, 63-LB) focused on demographic factors associated with CGM use among adults with type 2 diabetes demonstrating an association between younger age, insulin use, and higher A1c and CGM uptake. Data from a retrospective EHR analysis of adults with type 2 diabetes in the Vanderbilt University Medical Center (n=26,841) found that 3,310 adults with type 2, representing 12% of the total type 2 population at Vanderbilt, had been prescribed a CGM between 2018-2021 and that these adults had an average age of 57, an average A1c of 8.1% and 88% were on insulin therapy. These figures compared to an average age of 63, average A1c of 7.2, and 44% use of insulin therapy among adults with type 2 who were not on CGM during this same time period (p=0.001, p<0.001, p<0.001, respectively). Additionally, the investigators found that for each month of 2021, 110 new patients were prescribed CGM demonstrating that CGM uptake is continuing to grow in this population. Similar data on demographic factors associated with CGM use among type 2s in primary care at Vanderbilt University also found that from a sample of 109 adults with type 2 diabetes, 36 (33%) used CGM and CGM use was associated with younger age, higher baseline A1c, and insulin use. Specifically, CGM users had an average age of 55 compared to 60 among non-users (p=0.002). CGM users also had a higher average baseline A1c at 9.1% compared to non-users with an average baseline A1c of 8.3% (p=0.048). Finally, a higher percentage of CGM users were also prescribed insulin therapy 69% compared to 36% of the non-CGM using population (p=0.002). Across both studies, gender, race, and ethnicity were not found to be associated with CGM use. We would be very curious to learn more about whether insurance status or private versus public insurance or deductibility level is related to CGM use.

Quality improvement initiative in Philadelphia PCP office increases CGM prescriptions while improving provider familiarity and comfort with prescribing CGM; patients initiating CGM see 1.7% A1c reduction from 10.0% to 8.3%

Dr. Raashi Mamtani (Thomas Jefferson University) presented outcomes from a quality improvement initiative mean to increase the utilization of CGM in a Philadelphia-based primary care practice. We found this project to be hugely important, as dQ&A tells us that while only 9% of type 1s (n=2,283) primarily receive their diabetes care from a PCP (73% from endocrinologists, 16% from other types of providers), a whopping 56% of type 2s (n=3,149) primarily rely on PCPs for diabetes care (25% from endocrinologists, 14% from other types of providers). However, PCPs often face significant barriers to prescribing diabetes technology, given the inherent complexity in doing so for those who are unfamiliar. Turning to the study, Dr. Mamtani explained that the goal was to increase: (i) provider familiarity and comfort with CGM use; (ii) the number of CGM prescriptions in the clinic; and (iii) the overall glycemic control of the patient population, as measured by A1c. The investigators conducted a pre-intervention survey with PCPs to assess baseline familiarity with CGM use and understand barriers to CGM use and prescription, and then deployed a targeted intervention that focused on provider education and organized office workflow suggestions for prescribing CGM and educating patients. The primary outcome was the change in the number of clinic CGM prescriptions, and secondary outcomes included provider-reported changes in knowledge and comfort with CGM use, as well as change in A1c after three to six months of the intervention.

  • The number of patients with CGM prescriptions increased from 41 at baseline to 101 at six months, and the percentage of patients who actually filled their CGM prescriptions increased from 63% of those with prescriptions at baseline to 80% at six months. Additionally, mean patient A1c (n=10 of the 60 who were newly prescribed during the study) dropped 1.7%, from 10.0% to 8.3%, which is an impressively significant change that, from our view, speaks to the power of CGM to improve glycemic outcomes. The study is ongoing, and the investigators plan to continue tracking A1c values and document additional improvements. Of course, we’d hope that the investigators consider tracking Time in Range, as this would provide more granular results that would better help us understand the root causes of glycemic improvement.
  • The intervention improved provider familiarity with CGM, from 76% of providers responding with a “1” or “2” at baseline (“1” = not at all familiar, “5” = very familiar) to 70% of providers responding with a “4” or “5” at month six. At baseline, commonly cited barriers to CGM use included: (i) cost and insurance coverage; (ii) provider education and knowledge of device data; and (iii) support staff training. It’s hugely encouraging to see that the multifactorial intervention could, for the most part, reduce the impact of these barriers, likely owing to the deft and intentional design of the intervention.

Dr. Natalie Bellini on the importance of making CGM data actionable advocating in favor of implementing the DATAA and C2GM treatment models into clinical practice

Dr. Natalie Bellini (R&B Medical Group) highlighted the importance of clinician action on patient CGM data to improve & optimize outcomes and presented two models for interpreting and action on patient data. Despite tremendous improvement in the technologies available for PWD and an increasing amount of individualized data points collected by these devices, Dr. Bellini highlighted how population-level diabetes control and treatment has stalled. However, Dr. Bellini noted that the sheer quantity of data provided by CGMs can feel overwhelming and make clinical interpretation and action challenging. To overcome this,  Dr. Bellini advocated in favor of clinical treatment algorithms and best practices for CGM data interpretation specifically dsicussing the DATAA and C2GM (CGM Clinician Guided Management) models, both of which demonstrate the power of CGM data to create treatment adjustments and tangible improvements for PWD.

  • Dr. Bellini presented the DATAA Model for CGM data interpretation as a patient-centered and collaborative framework for clinicians and shared her own experience using the DATTA model in her clinical practice. As a reminder, the model, developed by Dr. Diana Isaacs, outlines a framework for clinicians to follow with the following steps: (i) Download Data; (ii) Assess for Safety; (iii) Time in Range; (iv) Areas to Improve; and (v) Action Plan. As Dr. Bellini explained, this five-step approach allows HCPs to directly collaborate with their patients to review their data and create an optimal care plan for their goals. The first step of downloading data allows a clinician to get a snapshot of a patient’s trends; Dr. Bellini suggests that clinicians check a patient’s (i) active wear time; (ii) average glucose; (iii) Time in Range; (iv) GMI %; (v) CV %; and (vi) pattern identification. Dr. Bellini next emphasized the importance of assessing safety and incidences of hypoglycemia, which may not be captured by the Time in Range bar graph and therefore requires an investigation of daily trends. Moving on to a review of a patient’s Time in Range information, Dr. Bellini stressed the value of focusing on the positives and encouraging behavior that patients are already doing that has demonstrated higher Time in Range. Aditionally, Dr. Bellini urged providers to avoid labeling days as “good” or “bad,” and instead work to understand the context behind glycemic patterns. Finally, Dr. Bellini explained that by examining areas in which the patient could improve, the DATAA model allows clinicians to provide advice on sustainable changes that may have big glycemic impacts, such as exercise. Using these “areas for improvement,” Dr. Bellini expressed that providers and patients can work collaboratively to create an action plan which can include lifestyle changes, CGM optimizations, and medication adjustments.
  • Dr. Bellini also highlighted the C2GM model as a fast and simple method to use CGM data to inform therapeutic adjustments. As a reminder, the C2GM model allows clinicians to use Time in Range and Time Below Range metrics to categorize patients in a treatment algorithm table that corresponds to the patient’s CGM data. Crafted by Drs. Rich Bergenstal and Thomas Martens (International Diabetes Center), this treatment algorithm, specifically targeted toward primary care, provides simple and actionable steps based on patient CGM data. Walking through the four treatment groups of the C2GM algorithm, Dr. Bellini reminded attendees that the vast majority of primary care providers do not have access to diabetes educators and specialists for their patients and that resources like the  C2GM algorithm can help them provide more individualized care.

Dexcom- and Medscape-sponsored symposium: KOLs muse on CGM in the hospital, CGM for type 2s, and the future of glucose monitoring

During a bright-and-early morning on the New Orleans Bayfront, Dr. Satish Garg (Barbara Davis Center) moderated a Dexcom- and Medscape-sponsored symposium entitled “What’s New in Continuous Glucose Monitoring.” With nearly 100 live attendees and over 200 virtual audience members, the session was packed despite being held at 6:00 am! Dr. Garg kicked off the session by commenting on the declining population-level glycemic control during the 2010s, as outlined in a paper that he co-authored in DT&T (Foster et al., 2019). Concerningly, the data show that only a quarter of people with diabetes at leading health centers are achieving an A1c <7%, and data from NEJM reinforce that this decline has been accompanied by a “leveling off” of diabetes prescriptions (Fang et al., 2021). Dr. Garg highlighted these data to stress that while conversations about diabetes technology often center on highly advanced systems, most of the ~150 million insulin-treated people with diabetes in the world do not use CGM or pumps, and we should recognize that affordability, availability, knowledge, and implementation all remain challenges preventing technology uptake.

  • Following Dr. Garg’s introduction, Dr. Francisco Pasquel (Emory University) presented a timeline tracing the genealogy of in-hospital CGM use. Shown below, Dr. Pasquel’s depiction of CGM in the hospital splits the timeline into a “pre-COVID” and “COVID” era, showing that after the onset of the pandemic, there was a steep acceleration in the push to bring CGM into the hospital, culminating with the FDA’s Breakthrough Designation for Dexcom’s in-hospital CGM system. Of course, Dr. Pasquel noted that as real-world studies and RCTs begin to come to light showing that CGM has a huge promise in the hospital, we must also remember that CGM is really hard to implement inside the hospital because doing so requires aligning all institutional stakeholders to secure approval. This sentiment reflects many conversations we engaged in at the Hospital Diabetes Meeting 2022, where we learned that “it takes a village” to successfully manage inpatient insulin therapy. Reinforcing this sentiment, Dr. Pasquel pointed to a publication in the Journal of Diabetes Science & Technology (Galindo et al., 2020) that contains important considerations for HCPs trying to initiate CGM in the hospital. Closing out his presentation, Dr. Pasquel shared results from a small study performed at the ICU at Grady Memorial Hospital in Atlanta (n=9) that he presented last year at ADA 2021. In this study, hospitalized patients were monitored using Dexcom G6 and insulin was administered via Glucommander. While Dexcom G6 proved mostly safe and effective in the hospital, Dr. Pasquel pointed out there were a few problem areas when clinical conditions created difficult circumstances (e.g., mechanical interference). Nonetheless, Dr. Pasquel’s takeaways were quite positive, noting that the number of bedside encounters and PPE utilization were significantly reduced with inpatients still achieving adequate glycemic control.

  • Dr. Katharine Barnard-Kelly (Barnard Health) followed with a presentation on using rt-CGM in type 2s. Dr. Barnard-Kelly began by noting that when people cite the massive healthcare expenditure related to diabetes in the US and worldwide, that it’s important to remember how these financial costs mask the personal costs that diabetes poses to people with diabetes and their families. The burden of diabetes, Dr. Barnard-Kelly argued, is reflected in the suboptimal glycemic outcomes achieved by many people with diabetes domestically and abroad. Turning to CGM specifically, Dr. Barnard-Kelly said that BGM is “increasingly inadequate” compared to CGM in its capacity to provide sufficient information to make informed decisions regarding lifestyle and treatment decisions. For type 2s specifically, Dr. Barnard-Kelly explained that CGM is “increasingly accessible and effective” for type 2s on insulin, and that users generally are quite satisfied while reporting reduced A1c and improved Time in Range. That is not to say that CGM is without woes, as Dr. Barnard-Kelly noted how wearing a CGM means drawing visibility to one’s condition, making them vulnerable to diabetes stigma, which she argued “remains very offensive and very problematic.” Dr. Barnard-Kelly emphasized that there is increasing data showing that CGM is beneficial for people outside of tech-savvy, well-controlled type 1s: (i) Martens et al., 2021 show how CGM can help basal-only type 2s in the MOBILE study; (ii) Cowart et al., 2021 review how CGM can help type 2s not on insulin; and (iii) Ruedy et al., 2017 analyze the DIAMOND trial to show that individuals ≥60 years old can also benefit from CGM. Dr. Barnard-Kelly closed by touching on Dexcom’s Type 2 Help Study, the first data from which was presented at ATTD 2022 – we look forward to seeing additional results presented later on at ADA!
  • Dr. Garg closed with a summary of next-generation CGM systems, while musing on the future of CGM. To start, Dr. Garg reviewed accuracy data for the three CGMs that are available either in the US and/or Europe: (i) Eversense E3; (ii) Dexcom G7; and (iii) FreeStyle Libre 3. Dr. Garg highlighted Time in Range as a key metric that is enabled by CGM and touched on data associating Time in Range with diabetes complications, including recent publication in the Journal of Clinical Endocrinology and Metabolism that we saw at ATTD 2022. In particular, Dr. Garg focused on conceptualizing how industry stakeholders should focus on standardizing download reports for people with diabetes on MDI using connected pen systems and for those using AID systems, which we think definitely has a potential to make data analysis easier for providers, similar to the Ambulatory Glucose Profile (AGP) Report for CGM. Besides mentioning some next-generation CGM systems (including Medtronic’s Simplera and Dexcom’s 15-day G7 sensor), Dr. Garg closed out the session by highlighting a retrospective Dexcom study published in February 2021, showing that although Dexcom G6 users in wealthier zip codes had higher Time in Range values and saw a greater improvements during lockdown, all socioeconomic groups benefitted from using CGM, which speaks to the potential for this life-changing technology to help broad swaths of the diabetes community achieve better outcomes.

Novel Glucose Pattern Insights Report (GPIR) associated with 3X increased likelihood of hypoglycemia identification and treatment among primary care providers compared to AGP; Dr. Eugene Wright advocates for simple and actionable CGM data to improve provider satisfaction and patient outcomes

In an Abbott-sponsored symposium on CGM data use and interpretation, Dr. Eugene Wright (Charlotte Area Health Education Center) discussed the ability of CGM data to enhance clinician satisfaction and experience. Specifically, Dr. Wright shared his view that CGM data, when presented in an understandable and actionable format, can help clinicians get value out of visits with their patients ultimately ideally leading to improved outcomes. Recognizing that CGM systems produce large amounts of data that can be challenging for clinicians, to interpret and act on, especially those in primary care who may be less familiar with CGM data, Dr. Wright presented data from a late-breaking poster (79-LB) demonstrating that a novel Glucose Pattern Identification Report (GPIR) can help clinicians more quickly identify trends in patient data to then inform safer therapeutic adjustments. The GPIR is a similar data report to the updated AGP that we saw Dr. Bergenstal present at DTM 2021 in which Time below Range is clearly identified to help providers identify hypoglycemic episodes and adjust treatment regiments accordingly. However, unlike the updated AGP, the GPIR also includes a “considerations for the clinician” box that highlights important lifestyle and medication questions for providers to assess when meeting with their patients. Additionally, the GPIR does not include the daily glucose profiles that are included at the bottom of AGPs. In this study, 35 non-specialist primary care providers were given 10 GPIRs and AGPs (AGPs in the 2019 format prior to the recent 2021 color-coded update) of the same patient data with five demonstrating patterns of hypoglycemia, three demonstrating patterns of hyperglycemia, one with both hypo- and hyperglycemia, and one demonstrating strong glycemic management. Participating providers were randomized to either evaluate GPIRs or AGPs first followed by a cross-over phase during which providers evaluated the other type of data report. Of note, providers only had 60 seconds to identify the single most-important clinical change they would adopt and were unaware that the same patient data was used to generate pairs of GPIRs and AGPs allowing for a comparison of their clinical decisions based on each report format.

  • Dr. Wright noted that providers were three times more likely to identify and treat hypoglycemia when evaluating patient data in the GPIR compared to AGP. Additionally, Dr. Wright shared that the decision clinicians made based on GPIR data were 50% less likely to increase hypoglycemia (p<0.005) and 50% less likely to prolong hypoglycemia (p<0.005) compared to decisions based on AGP information. These trends were most prevalent in the five GPIR and AGP reports demonstrating patterns of hypoglycemia and the study did not find a significant difference in therapeutic choices and risks between GPIR and AGP informed treatment for the cases including patterns of hyperglycemia or patterns of both hypo and hyperglycemia.
  • GPIR use was associated with faster identification of hypoglycemia compared to AGP interpretation. Specifically, when using the GPIR, providers identified patterns of hypoglycemia in ~40 seconds compared to ~48 seconds when using the AGP (p<0.05). While this may not seem like a clinically significant difference, Dr. Wright expressed his view that any time savings and simplification that can be enabled in primary care is an important step to give patients and providers the adequate time to discuss diabetes management and concerns the patient may have.
  • This study adds to a growing body of treatment algorithms and tools for CGM data interpretation that can be leveraged by providers to help drive improved outcomes and patient and provider satisfaction with CGM systems. At ADA this year, we heard from Dr. Natalie Bellini (R&D Medical Group) who provided an overview of the DATAA model and C2GM treatment algorithm walking attendees through how to best use these tools to inform clinical decisions. Building on Dr. Rich Bergenstal’s initial presentation of the C2GM algorithm for people with type 2 diabetes on basal-only therapy at ATTD 2022, Dr. Thomas Martens presented a second iteration of the treatment algorithm for people with type 2 diabetes on basal-bolus insulin therapy earlier at ADA. We are optimistic that these treatment algorithms will become available for providers to use with their patients in the near future and commend the exceptional work that has gone into helping providers best make use of CGM data to inform treatment decisions and help improve patient outcomes.

Posters – Glucose Monitoring

Title

Authors

Details + Takeaways

System Performance of an Integrated Blood Glucose Monitoring Device Developed to Reduce Common Barriers to Checking Glucose

Michael Tomasco, Judith M. Davis, Paul D. Reynolds, Raul Escutia, John N. Perry, Tim O’Donnell, Robb P. Hesley

  • Accuracy study of Intuity Medical’s POGO BGM system for type 1s and 2s
  • Untrained participants (n=285); mean age 58 years; 50% female; and 85% White
  • 95.3% of results were within ±15 mg/dL or 15% of the reference reading and 99.3% of results were within ±20 mg/dl or 20% of the reference reading

Nursing Perspectives on Hospital Use of Dexcom Continuous Glucose Monitor in Patients with known COVID-19 During Insulin Infusion

Jillian Pattison, Kathleen M. Dungan, Elizabeth Buschur, Matthew C. Exline, Laureen G. Jones, Casey May, Molly Mcnett, Keaton Smetana, Eileen R. Faulds

  • Anonymous survey of ICU nursing staff perceptions around CGM use in insulin-requiring patients
  • Survey taken by 51 ICU nurses after one and a half years of standardized protocol use
  • ICU nurses reported that CGMs: (i) were accurate (92%); (ii) reduced their workload (94%); (iii) provided safer patient care (88%); and (iv) were preferable to only using BGM (94%)

Targeting Equity in Type 1 Diabetes (T1D) Technology – Effect of Practice Transformations on Provider Prescribing Patterns of Continuous Glucose Monitors (CGMs)

Priyanka M. Mathias, Lakshmi Priyanka Mahali, Shivani Agarwal

  • Retrospective analysis; studied impact of “practice transformations” on CGM prescription rates across racial-ethnic groups
  • Participants (n=1,357) with type 1 diabetes; mean age 38 years; 45% Hispanic, 30% non-Hispanic Black, 12% non-Hispanic White; 74% publicly insured
  • CGM prescription rates increased by 54% (!), from a baseline of 15% to 69% in two years (p<0.001); equal improvements observed within racial-ethnic groups

Lower A1c with Real-Time CGM than with Intermittently-Scanned CGM after One Year – The CORRIDA LIFE Study

Lucie Radovnická, Aneta Hásková, Quoc Dat Do, Eva Horova, Vendula Navrátilová, Ondrej Mikes, David Cihlár, Christopher Parkin, George Grunberger, Martin Prazny, Jan Soupal

CamAPS FX users achieved a 0.4% lower A1c than those on SAP at 6.6% and 7.0%, respectively

  • Twelve-month, real-world, non-randomized CORRIDA LIFE trial assessing real-time CGM versus intermittently scanned CGM
  • 191 type 1s were followed for 52 weeks using either Dexcom G5 or G6 (rt-CGM arm, n=81) or FreeStyle Libre 14-Day (is-CGM arm, n=110); mean age 40 years; mean diabetes duration 16 years
  • Participants on rt-CGM achieved an 0.6% A1c reduction compared to is-CGM arm (p=0.0001), at (7.1% vs. 7.7 from baselines of 8.0% vs. 8.2%, respectively)
  • Participants in rt-CGM arm spent less time in clinically significant hypoglycemia (<54 mg/dL) compared to the is-CGM group (0.9% vs. 2.3%, respectively; p<0.0001)

Ambulatory Glucose Profile Informs Better Treatment Decisions for Type 2 Basal-Insulin Patients

Eileen Huang, Mohamed Nada, Eugene Wright, Jr.

 

 

  • Prospective trial examining whether glycemic management improved with use of Ambulatory Glucose Profiles (AGP)
  • The study enrolled 105 type 2s with A1cs between 7% and 10% and on basal insulin; mean age was 64 years; mean duration of diabetes was 17 years; mean baseline A1c of 8.2%
  • Utilizing the AGP, 94% of patients had therapy change recommendations based on AGP results; 40% of the time the change was an increase in basal insulin
  • Majority of patients (97%) reported a better understanding of their treatment plan after reviewing AGP with their PCP

How the Incidence of Anxiety and Other Psychosocial Factors Impact Quality of Life in Parents of Children with Type 1 Diabetes Who Used a Continuous Glucose Monitor (CGM) – A Systematic Review

Bailee C. Sawyer, Leanne M. Hutson, Amanda P. Ciprich

 

  • Systematic review of how anxiety and other psychosocial factors impact quality of life for parents and their children with type 1 diabetes on CGMs
  • Review of seven studies published between 2012-2021
  • CGM use was associated with improved glycemic management across all studies, seen through decreased A1c levels and hypoglycemia episodes
  • Studies collectively demonstrate that CGM use in kids can shape anxiety and quality of life for parents, leading investigators to call for increased focus on psychosocial factors

Association between Daytime vs. Nighttime Mean Glucose and Time in Range with A1c in Adults with Type 1 Diabetes

Viral Shah, Timothy B. Vigers, Laura Pyle, Halis K. Akturk, David C. Klonoff

  • Real-world study on the relationship between nighttime vs. daytime CGM metrics and A1c levels in adult type 1s (n=340; diabetes duration > 2 years; use of Dexcom G6 for greater than six months)
  • Higher A1cs correlated with increased mean sensor glucose and reduced Time in Range for daytime and nighttime (p<0.001 for both); no difference between daytime and nighttime Time in Range or mean glucose across A1c levels (p=0.08 and p=0.42, respectively)

The Effects of Glucose Reporting Tools on Therapeutic Decision Making – A Comparative Reading Study with Primary Care Providers

Matthew T. Novak, Gary Hayter, Eugene Wright, Kurt Midyett, Jr., Howard Wolpert, Naunihal Virdi

  • Assessing the effectiveness of a novel CGM-based Glucose Pattern Insights Report (GPIR), highlighting “most important patterns” (MIPs) of clinical relevance
  • Non-specialized primary care providers (n=35) were given GPIRs or standardized glucose report; therapy recommendations between two arms were compared via a two-round crossover analysis
  • When using GPIR, 99% of providers correctly addressed hypoglycemia when matched providers using standardized glucose reports did not, indicating that GPIR may aid in identifying and treating hypoglycemia vs. standard care

At Home, Monitoring Device for Prevention and Interception of Diabetic Ketoacidosis Events in Children

Sophie L. Edgar, Alex Paunescu, Ming Dong

 

  • Development of a single-use sensing pad that fits in a standard diaper and connects to a Bluetooth enabled reusable device to alert caretakers of early indicators of diabetes and diabetic ketoacidosis
  • The device can successfully recognize glucose levels at 50, 100, and 250 mg/dL; all levels of ketones (0, 15, 40, 80, & 160 mg/dl) can be distinguished by the system

Sustained impact of switching from Intermittently Scanned to Real-Time Continuous Glucose Monitoring in Adults with Type 1 Diabetes: 24-month results of the ALERTT1 Trial

Margaretha M. Visser, Sara Charleer, Steffen Fieuws, Christophe De Block, Robert Hilbrands, Liesbeth Van Huffel, Toon Maes, Gerd Vanhaverbeke, Eveline L. Dirinck, Nele Myngheer, Chris F. Vercammen, Frank Nobels, Bart Keymeulen, Chantal Mathieu, Pieter Gillard

  • Partial crossover extension of ALERTT1 trial to examine the long-term benefits of switching from intermittently scanned CGM (is-CGM) to real-time CGM (rt-CGM)
  • Participants (n=254) had an average age of 43 years; mean A1c of 7.4%
  • After switching from is-CGM to rt-CGM, participants spent +2.8 hours/day Time in Range, from a baseline of 52% to 64% at twelve months (p<0.0001) and remained stable through twenty-four months (p<0.0001)

Improved Glycemic Control and Continuous Glucose Monitoring (CGM) Utilization: a comparison of real-time CGM and intermittently-scanning CGM

Katia Hannah, Poorva Nemlekar, David A. Price, Gregory J. Norman

 

  • Retrospective analysis of CGM use in naïve participants on either rt-CGM (n=272; average age=47 years; 47% female) or is-CGM (n=467; average age 51; 38% female)
  • Adjusting for baseline, rt-CGM users were nearly twice as likely to reach an A1c below 7% (OR=1.97, p<0.01)
  • Hospitalizations from severe hypoglycemia decreased from 40 to 13 (p<0.01) after initiation of rt-CGM, compared to a nonsignificant decrease from 42 to 32 in is-CGM arm (p=0.239)

Lag Times in a Seventh Generation Continuous Glucose Monitoring System

Xiaohe Zhang, Jennifer L. Reid, Tomas C. Walker, John Welsh, Andrew Balo

  • Pivotal accuracy study (n=316) of CGM lag times for Dexcom G7
  • Lag time of two minutes provided the lowest MARD of 8.6% and the highest ±20%/20 mg/dL agreement rate of 94.4%
  • A lag time of ≤0 min was observed in 33% (n=204) sensors, where interstitial glucose precedes blood glucose

Continuous Glucose Monitoring in Primary Care: Explaining Characteristics Associated with CGM Prescription

Tamara Oser, Tristen Hall, Meredith K. Warman, Melissa K. Filippi, Brian Manning, Elisabeth Callen, L. Miriam Dickinson, Leann C. Michaels, Donald E. Nease, Sean Oser

  • Mixed-methods sequential explanatory study of US primary care clinicians, with a survey portion (n=632) followed by semi-structured interviews (n=55)
  • Clinicians greater than 40 miles from an endocrinologist were more likely to prescribe CGM to patients than those within 10 miles (p=0.03, OR= 1.9)
  • Residents were less likely than attending physicians to prescribe CGMs (p<0.001, OR= 0.3)
  • Clinicians with greater than 50% of patients on Medicare had greater confidence using CGM to manage diabetes than those with less than 25% of their patients on Medicare (p<0.01)

Effect of Calibration Frequency on Accuracy of a 180-day Implantable CGM System

Arnaud E. Jacquin, Lujain Al-Khawi, Samanwoy Ghosh-Dastidar, Katherine S. Tweden, Francine R. Kaufman

  • Data from the PROMISE study was run through an algorithm to calculate mean absolute relative difference (MARD) of sacrificial boronic acid CGMs
  • Participants (n=43) from eight clinic sites wore sensors for 180 days
  • The sensors maintained an overall MARD of <9.0% across glucose ranges and <9.6% across all days of sensor wear through 180 days

Frequency Scanning Correlates Not Only with Glycemic Indices but Also with Fear of Hypoglycemia in Type 1 Diabetes Patients using is-CGMs

Jerzy Hohendorff, Przemyslaw W. Witek, Michal Kania, Maria Sudol, Katarzyna Hajduk, Adam Stepien, Katarzyna Cyganek, Beata Kiec-Wilk, Tomasz Klupa, Maciej Malecki

  • Examination of the correlation between CGM-derived metrics and fear of hypoglycemia in type 1s using is-CGM
  • Participants (n=77) with type 1 diabetes; 50% on MDI and 50% on pumps; 75% female; mean age of 34; average scanning frequency was 14 scans/day
  • Higher scanning frequency was associated with better performance across CGM-derived metrics (e.g., GMI, Time in Range, Time Below Range, and Time Above Range) and less fear of hypoglycemia (p<0.05 for both)

CGM Access Is the Main Barrier to TIR Use Among Providers

Julia Kenney, Jacqueline Tait, Andrew Briskin, Erik Shoger, Richard Wood, Anne L. Peters

  • Online provider survey (n=303) on discussing diabetes management with patients and the concomitant challenges
  • For providers who utilize Time in Range (n=234), 80% identified access to CGM as the main drawback of Time in Range and 72% (n=289) reported cost as a downside of CGM use
  • Providers who do not utilize Time in Range (n=64) reported, on average, that increased access to CGM would convince them to use it; 83% of non-users think Time in Range would be useful in their diabetes care

Lower Peak Glucose and Increased Time in range (TIR) in a CGM-Wearing T2D Population Not Taking Fast-Acting Insulin Shows Value of Real-Time CGM (rt-CGM) as a Behavior Change Tool

Margaret A. Crawford, Daniel R. Cherñavvsky, Katharine Barnard, Xiaohan (Aria) Wang, Paul Genge, Kenneth Greenawald, Michelle Tressler

  • Observational, multi-center trial on CGM use in type 2s not taking fast-acting insulin
  • Participants (n=150) wore CGMs for twelve weeks; mean age 54 years; 54% female; 32% Hispanic, 18% Black, and 10% Asian
  • The entire population did not see significant improvements in Time in Range (+24 minutes/day from baseline of 82% to 84% at twelve weeks) while on a CGM (p>0.05), but this probably reflected very well-controlled glycemia at baseline
  • Thirty-five percent of participants improved their Time in Range by greater than 5%; these glucose responders (n=53) spent +3.6 hours/day Time in Range, from a baseline of 72% to 88% at twelve weeks

Benchmarking Continuous Glucose Monitor Use across T1D Exchange Quality Improvement Pediatric Clinics

Ashley Garrity, Ann Mungmode, Kristina Cossen, Mary Pat Gallagher, Alexis J. Feuer, Daniel Brimberry, Priya Prahalad, Mary L. Scott, Grace Nelson, Osagie Ebekozien, Donna Eng

  • Examination of T1D-Exchange Quality Improvement (T1DX-QI) Collaboration Clinic Reports from July 2020 to June 2021
  • Across T1DX-QI clinics (n=17) the percentage of participants utilizing CGM ranged from 23% to 85%; median A1cs ranged from 7.4% to 9.2%
  • No statistically significant correlation between CGM use and median A1c levels in T1DX-QI clinics

Difference in CGM Estimated Hemoglobin A1c in Adults with Type 1 and Type 2 Diabetes

Maamoun Salam, Ryan Bailey, Peter Calhoun, Janet B. Mcgill, Roy Beck, David A. Price

  • Investigation of the relationship between mean glucose (measured via CGM) and A1c in type 1s (n=199) and 2s (n=175); data compiled from MOBILE, DIAMOND, and WISDM trials
  • Estimated A1c was slightly higher for same mean glucose value in: (i) participants with type 2 diabetes compared to those with type 1 (p=0.005); (ii) non-Black type 2s compared to non-Black type 1s (p=0.07); and (iii) Black type 2s compared to non-Black type 2s (p=0.005)
  • Correlation coefficients for mean glucose vs. A1c were 0.76 for type 1s and 0.84 for type 2s

Racial and Ethnic Disparities in CGM Use among Adults with Diabetes

Samia M. Aljedaani, Ayesha S. Siddiqui, Nazia Raja-Khan

  • Cross-sectional study investigating differences in CGM use across races in adults with diabetes
  • Only 2% of participants (n=73,210 total) were using CGMs
  • In the type 2 cohort, (i) 2.0% of White; (ii) 1.2% of Black; 1.1% of Hispanic; and (iv) 1.9% of Asian participants utilized CGM; average A1c was 7.8% in those using CGM, compared to 7.9% in those not using CGM
  • Among type 1s, (i) 9.1% of White; (ii) 6.8% of Black; 7.6% of Hispanic; and (iv) 8.9% of Asian participants utilized CGM; average A1c was 8.4% in those using CGM, compared to 8.4% in those not using CGM

Costs and Cost-Effectiveness of the FreeStyle Libre System vs. Blood Glucose Self-Monitoring in Patients with Type 2 Diabetes on MDI Insulin Treatment in Israel

Dan Greenberg

 

  • Real-world data on the incremental cost per quality adjusted life year (QALY) gained when using FreeStyle Libre in type 2s; assuming a willingness to pay threshold of $43,600 per QALY
  • FreeStyle Libre use is associated with cost increase of $8,444 and a 0.26 QALY; generating an ICER of $32,493 per QALY gained
  • ICER/QALY falls below willingness to pay, indicating that FreeStyle Libre is cost-effective

TBR (Time Below Range) in Routine Clinical Practice: A Retrospective Analysis of Routine CGM in Outpatient Care

Purvi M. Chawla, Manoj S. Chawla

 

  • Single center analysis of is-CGM in type 1s and 2s (n=227)
  • Participants with an A1c <7% spent the most time below range compared to those with A1c ≥7% (p=0.0016), indicating highest risk for hypoglycemia
  • Participants with an A1c between 7% and 9% spent nearly 10% of their time Below Range (p=0.0008)

Tracking Type 2 Diabetes Progression Using CGM Systems

Michele Schiavon, Chiara Dalla Man

 

  • An assessment of a novel method to track progression of type 2 diabetes; utilizes CGM data instead of the Oral Minimal Model to inform the Disposition Index (i.e., the mathematical product of insulin sensitivity times the amount of insulin secreted in response to blood glucose levels)
  • The method was tested on early (n=100) and advanced (n=100) stage type 2s who underwent a mixed meal challenge (75g of carbohydrates) while wearing a CGM
  • The CGM-based model was correlated with the Oral Minimal Model (p<0.001)
  • DI (disposition index) was significantly lower in the advanced disease group compared to the early-stage group (p<0.001) using both OMM and novel method, indicating a potential to track progression throughout disease stages noninvasively

Comparing Rates of Sensor-Detected Hypoglycemia and Patient-Reported Hypoglycemia by Awareness of Hypoglycemia Using Continuous Glucose Monitoring: the Hypo-METRICS trial

Patrick Divilly, Gilberte Martine-Edith, Zeinab Mahmoudi, Natalie Zaremba, Uffe Soeholm, Ulrik Pedersen-Bjergaard, Rory J. Mccrimmon, Bastiaan E. De Galan, Eric Renard, Simon R. Heller, Mark Evans, Julia K. Mader, Stephanie A. Amiel, Pratik Choudhary, Alexander Seibold

  • Preliminary results comparing rates of sensor-detected hypoglycemia (from CGM data) and patient-reported hypoglycemia (flash glucose monitoring) in type 1s
  • Participants (n=129) were defined as either possessing normal or impaired hypoglycemia
  • Neither Time in Range (59% vs. 62%; p=0.18) or Time Below Range (4.6% vs. 6.0%; p=0.44) were different between groups
  • During sleep and awake periods, rates and duration of sensor-detected hypoglycemia and patient-reported hypoglycemia were not significantly different (p>0.05)

Inpatient Glycemic Control and Glucose Variability by Continuous Glucose Monitoring in Older Adults with Type 2 Diabetes

Thaer Idrees, Rodolfo J. Galindo, Maria A. Urrutia, Iris A. Castro-Revoredo, Emmelin Marie Moreno, Alexandra L. Migdal, Georgia M. Davis, Priyathama Vellanki, Maya Fayfman, Francisco J. Pasquel, Limin Peng, Guillermo E. Umpierrez

  • Data from three prospective CGM clinical trials with type 2s in non-acute hospital settings; glycemic metrics were compared between older adults (≥60 years, n=103) and younger adults (<60 years old, n=160)
  • Participants had a mean age of 52 years; 78% Black
  • Older adults spent +1.9 hours/day in Range, compared to younger adults (59% vs. 51%; p=0.017)
  • Multivariate analysis of older adults revealed higher glycemic variability during hospital stays compared to younger adults, potentially due to confounding variables such as higher A1c levels and higher glucose levels on admission

Parental Education and Family Income Associated with Increased CGM Use in Adolescents with Type 1 Diabetes

Angelee Parmar, Sarah S. Jaser, Karishma Datye

 

  • Survey-based analysis of parents (n=150) with children (adolescent type 1s) using CGM
  • 81% of parents indicated their adolescent children currently utilized CGM; no significant differences in adolescent CGM use across races
  • Adolescents with a family annual income >$90,000 had higher CGM use (88%) than lower-income families (72%) (p=0.018)
  • Overall, significant predictors of CGM use were parental education and income

In Patients with Type 2 Diabetes Mellitus (T2DM), Continuous Glucose Monitoring System (CGMs) Usage Does Not Alter Multiple Measures of Glycemic Variability

Yee Liong Lee, Ayesha Mehfooz, Qi Wang, Lisa S. Chow

 

 

  • Single-center RCT (n=40) of type 2s using unblinded (FreeStyle Libre) or blinded (FreeStyle Libre Pro) CGM for sixteen weeks
  • No significant difference across all CGM-derived metrics (e.g., mean glucose, standard deviation, Time in Range, and GMI) between the blinded and unblinded cohorts at baseline and at sixteen weeks

Automated Insulin Delivery, Pumps, and Pens

Omnipod 5 in type 2s 34-week extension phase: former MDI (n=12) users maintain eight-week improvement in glycemic management with A1c of 8% and Time in Range of 58%; former basal-only users (n=10) see remarkable +7.9 hour/day improvement from baseline in Time in Range to 65% with average A1c of 7.5%

Data from the extension phase of Insulet’s feasibility study of Omnipod 5 in type 2 diabetes (769-P) demonstrated continued improvements in glycemic management through 34 weeks with participants (n=22) achieving an average A1c of 7.7%. As a reminder, Insulet presented initial feasibility data from 24 adults with type 2 diabetes using Omnipod 5 at ATTD 2022 and demonstrated strong improvements in glycemic management for patients previously on basal only or MDI therapy. After the initial eight-week study, 92% of participants (n=22) continued into the extension phase, and saw improvement in glycemic management from an average A1c at eight weeks of 8% to an average A1c at 21 weeks of 7.7% (p<0.05). As in the Omnipod 5 in type 2s feasibility study, participants were either former basal-only (n=10) or MDI users (n=21) and had an average age of 61 and an average A1c of 9.4% at baseline. As we noted at ATTD 2022, these results represent some of the first data on AID systems in people with type 2 diabetes and we are encouraged to see participants achieve strong markers of glycemic management across both A1c and Time in Range. Elsewhere at ADA 2022, we’ve seen other encouraging data on the use of AID among adults with Type 2 Diabetes from both Control-IQ and MiniMed 780G users, and we remain optimistic that these systems may become available to more patients with type 2 diabetes in the future.

  • For former MDI users, Omnipod 5 drove significant improvements in both A1c and Time in Range. Specifically, A1c decreased from 9.4% at baseline to 8.1% in the eight-week feasibility study (p<0.05). Through 34 weeks, former MDI users achieved an A1c of 8%, in-line with results from the initial eight-week study. Former MDI users also saw improvements in Time in Range from 43% at baseline to 61% at eight weeks (p<0.05), and 58% through 34 weeks representing a massive increase of 3.5 hours/day in range. These are certainly impressive improvements, and we suspect anyone who has had a 9.4% A1C can attest to 8.1% being truly different and better. Though we will without question continue to hear questions about what is needed to be done to help drive even higher Time in Range among patients with type 2 diabetes (T2D) using Omnipod 5 to help them achieve consensus targets for 70% Time in Range, clearly, there is not a single answer, nor are there even several answers. Stress, poverty, food environments, relationships, social determinants of health, and on it goes … these are just some of the areas that represent barriers to higher time in range as well as, of course, the absence of the right therapeutic interventions. That last is among the most vexing to us since this is so easily changeable and represents proactive rather than reactive care or, worse, absence of care. We also point out, therapeutic inertia isn’t just the wrong medicine, which is bad enough (say, SFUs rather than SGLT-2s or GLP-1s or dual agonists), but often therapeutic inertia is just the wrong dosing – like, only 10 units of Lantus prescribed when prandial insulin is also needed, or maybe just a higher dose of Lantus or Toujeo or Tresiba. 
  • Speaking of basa, for former basal-only patients, Omnipod 5 also drove significant improvements in both A1c and Time in Range. Specifically, A1c decreased from 9.5% at baseline to 8.1% in the eight-week feasibility study (p<0.05) to 7.5% through 34 weeks. Former basal-only patients also saw significant improvements in Time in Range through 34 weeks from 31% at baseline to 57% at eight weeks (p<0.05) to 65% at 34 weeks representing a remarkable overall increase of 7.9 hours/day in range. This is one of the largest absolute increases in Time in Range we have seen among patients initiating AID therapy and demonstrates the ability of AID to drive remarkable outcomes among patients with type 2 diabetes. Additionally, unlike prior MDI users, patients previously on basal-only therapy continued to experience improvements in Time in Range out to 34 weeks and we are hopeful to see further data on Omnipod 5 in this population to see what measure of further Time in Range improvements can be achieved – it is no question that some can.

CREATE RCT comparing OpenAPS and SAP demonstrates +2.4 hours/day improvement in Time in Range for participants on AID to 71% after six months; no severe hypoglycemia or DKA episodes among OpenAPS users

During a highly anticipated oral presentation, Dr. Martin de Bock (Otago University, New Zealand) readout topline results from the six-month CREATE trial, the first-ever RCT comparing open-source AID with sensor-augmented pump in type 1s (n=97). Dr. de Bock gave a huge nod to the DIY community and #OpenAPS movement founder Ms. Dana Lewis, who recently spoke excellently at ATTD 2022 on how the similarities between DIY AID systems and commercial systems “outweigh differences.” Dr. de Bock mused back to ADA 2018, when it was clear that the real-world evidence base for DIY AID systems was impressive in highly motivated, well-controlled individuals; however, many questioned the generalizability of real-world data, and thus was the genesis of the CREATE (Community deRivEd AutomaTEd insulin delivery) RCT. The primary endpoint of the open-label, multicenter trial was Time in Range between the open-source AID and SAP arms during the final two weeks of the six month study period, and key secondary endpoints included other CGM-derived metrics, A1c, safety outcomes, and psychosocial impact. Participants between seven and 70 years old who were on insulin pump therapy for ≥six months prior to screening were included, and they needed to have an A1c <10.5% and Wi-Fi access. Participants randomized to the OpenAPS cohort (n=44 of 97) used the AnyDANA-loop system, consisting of: (i) the OpenAPS 0.7.0 algorithm in a locked Android smartphone; (ii) a Dana-i insulin pump; and (iii) a Dexcom G6 CGM; there was no need for Riley Link, and the system had unannounced meals set to “always on.” The investigators allowed participants to set a glucose target between 90 mg/dL and 117 mg/dL, and during activity, participants could opt for a 144 mg/dL target. At baseline, the population had impressive glycemic outcomes, with a mean A1c of 7.7% and 7.5% for adults (≥16 years) and children (<16 years), respectively, and a mean Time in Range of 62% and 56% for adults and children, respectively.

  • At six months, OpenAPS users had a mean adjusted treatment effect of +3.4 hours/day Time in Range compared to the SAP arm (71% vs. 55%; p<0.001). Specifically, the AID arm achieved +2.4 hours/day Time in Range from a baseline of 61% to 71% at six months. Comparatively, those in the control arm saw a 46 minute/day decrease in Time in Range, from 58% at baseline to 55% at the conclusion of the study. Notably, there was a fourfold difference in the percent of participants meeting consensus targets for Time in Range (60% of those in the AID arm vs. only 15% of those in the control group). Additionally, there was no treatment effect by age interaction detected (p=0.56).
  • Participants using OpenAPS also saw decreases in A1c and mean glucose. Specifically, for children and adults A1c fell by 0.5% and 0.7% from baseline, respectively, to 7% in both cohorts. Mean glucose also declined by an average 22 mg/dL among OpenAPS users from baseline and there were no severe hypoglycemia or DKA events. Excitingly, among OpenAPS users, the system was in automode 94% of the time throughout the study.
  • As with commercially available AID systems, the largest improvements in glycemic management were seen overnight with both children and adults achieving nighttime Time in Range of >80%. Additionally, as with commercial AID systems, improvements in glycemic outcomes were seen immediately within three weeks of system initiation and maintained throughout the duration of study.

Real-world Control-IQ data from massive cohort (n=20,314 type 1s and 2s) demonstrates 71% Time in Range after three months; largest glycemic improvements among patients with baseline GMI >8%; type 2s on Control-IQ (n=960) see +2.2 hours/day Time in Range to 72%

Dr. Boris Kovatchev (University of Virginia) presented new real-world data from a massive cohort of Control-IQ users (n=20,314) demonstrating an overall average Time in Range of 71% after during three months on the system. Patients included in this analysis had at least one month of Basal-IQ data available followed by at least three consecutive months of Control-IQ data for a total of four months of data and had either type 1 or type 2 diabetes. On Basal-IQ, the cohort achieved an average of 58% Time in Range with a mean GMI of 7.5%. However, after only a few days on Control-IQ, the cohort experienced a significant increase in Time in Range of 3.1 hours/day to 71% (p<0.001) with an average GMI of 7.1% (p<0.001). This improvement was then sustained for the duration of the study with patients seeing an average Time in Range of 70% and GMI of 7.1% after three months on Control-IQ. While this improvement is certainly impressive, Dr. Kovatchev raised the question of why, if patients see such substantial improvements in glycemic management so quickly after starting AID systems, they do not continue to see improvements after one month. While Dr. Kovatchev recognized that we don’t currently know what may be limiting the achievable Time in Range on AID, he recognized that it was likely a combination of behavior and physiology. Addiitonally, Dr. Kovatchev discussed the challenges associated with insulin action time as a potentially limiting factor in the future development of AID or even fully closed-loop systems. At ATTD 2022, Dr. Kovatchev discussed a similar topic looking to the future of AID and suggesting that faster insulin action time or combination with additional therapeutics, namely SGLT-2s, may be able to drive even greater improvements in glycemic outcomes for patients using AID.

  • By baseline glycemic management, patients with higher GMIs (>8%) saw the greatest increases in Time in Range from baseline at +5.4 hours/day. Specifically, patients with a baseline GMI >8% (n=4,064) had an average Time in Range of 33% that, after one month on Control-IQ increased to 56% and was maintained at 54% after three months (p<0.001). While this improvement in Time in Range from Basal-IWQ to Control-IQ is certainly impressive, patients are still falling somewhat short of consensus targets for 70% Time in Range, suggesting (some may say reinforcing) there may be additional factors impacting their diabetes management that should be addressed help drive improved outcomes and mitigate long-term complications. Also of note, among this study cohort, patients with baseline GMI of 7.4%-8% also saw a significant 3.6 hour/day increase in Time in Range (p<0.001), but still fell short, on average, of consensus targets at 68% Time in Range after one month of Control-IQ and 66% Time in Range after three months of Control-IQ, once again suggesting that patients often need more than and AID algorithm to achieve glycemic management and Time in Range Goals. While the largest improvements in Time in Range were seen among patients with higher baseline GMI, Dr. Kovatchev did make sure to note that even among patients with baseline GMI ≤6.9% and an average Time in Range of 81% (n=4.451), Control-IQ was able to drive improvements in glycemic management with patients experiencing a 50 minute/day increase in Time in Range to 84%.

  • Across all age cohorts, patients saw improvements in Time in Range following initiation of Control-IQ. The largest improvements in Time in Range were seen among pediatric and adolescent users of Control-IQ. Specifically, Control-IQ users ages 0-18 (n=4,636) saw a 3.5 hour/day increase in Time in Range from 50% on Basal-IQ to 65% on Control-IQ (p<0.001). Young adult patients ages 19-25 also saw a remarkable improvement in Time in Range of 3.2 hours/day from 55% on Basal-IQ to 68% after one month on Control-IQ. Similar to the data stratified by baseline GMI, these results indicate that while Control-IQ can drive impressive improvements in Time in Range, AID alone may not be enough to help all patients achieve consensus glycemic targets.
  • Among a smaller cohort of real-world Control-IQ users with type 2 diabetes (n=960), Time in Range improved by 2.2 hours/day from 63% on Basal-IQ to 72% after a few days on Control-IQ (p<0.001). Excitingly, this is the largest cohort of type 2s on AID that we have seen and confirms the assumption that people with type 2 diabetes on basal-bolus insulin therapy can benefit from using AID systems.
  • Dr. Kovatchev presented new data on changes in patient bolusing behavior before and after initiating Control-IQ demonstrating that, as expected, patients initiated significantly fewer boluses once using Control-IQ. As Dr. Kovatchev explained, this result is expected due to the correction bolus feature of the Control-IQ algorithm that provides correction doses without the need for patients to manually input them. Interestingly, patient manual-bolusing behavior ranged from four to six manual boluses per day while on Control-IQ, which could align well with individual eating patterns suggesting that patients only needed to input boluses ahead of meals and snacks. Dr. Kovatchev also shared that, as far as he is aware, this is some of the first data demonstrating behavior changes (i.e., fewer user-initiated insulin boluses) following initiation of an AID system, which he also argued is indicative of the potential AID systems have to reduce the burden of diabetes management on patients.

Feasibility study shows fully-closed loop version of MiniMed 780G with meal gesture detection software Klue sees consistent Time in Range with no meal announcements relative to MiniMed 780G with manual carbohydrate counting; second feasibility study to evaluate updated version is underway

In one of the most exciting tech presentations of the day, Dr. Anirban Roy (Medtronic) presented the outcomes of a feasibility trial assessing the efficacy of version of MiniMed 780G that drives bolus dose decisions from meal gesture detection software Klue rather than carbohydrate counting or meal announcements. In the study (NCT04964128), MiniMed 780-naïve adults with type 1 diabetes (n=17, ages 18-75) engaged in two five-day study periods: (i) at-home use of MiniMed 780G (with Klue disabled) with carb-counting/pre-meal boluses; and (ii) in-clinic use of MiniMed 780G with Klue disabled and no carb-counting/pre-meal boluses/meal announcements. In both periods, participants engaged in one standardized meal test per day for five days: (i) large meal with low carbs; (ii) small meal with high carbs; (iii) normal size meal with medium carbs; (iv) normal size meal with high carbs; and (v) normal size meal with low carbs. Beyond these five standardized meals, eating was not regulated. Overall, the study found no significant differences in Time in Range and time above range between Klue and manual carb counting and a slight but significant time in hypoglycemia benefit of Klue both during the daytime and during the full 24-hour day.

Before digging into the results further, we’ll provide some background on the Klue app and how Medtronic used the gesture data it provides to drive insulin bolus dosing. As a reminder, Klue was acquired by Medtronic in December 2019 and is an Apple watch-based app that detects and differentiates eating and drinking hand gestures based on motion sensors. Specifically, when the watch is worn on the dominant hand, the app use gyroscope and accelerometer data to determine the probability of an eating or drinking hand-gesture in real-time. These detected gestures are translated into carbohydrate counts using gesture-to-carb mapping, which correlates duration of meal and number of gestures with carbohydrate count. These Klue-based carb counts are then converted into insulin dosing decisions by the meal wizard algorithm in the pump.

 

MiniMed 780G with Klue but no meal announcements

MiniMed 780G with carb counting

p-value

Time in Range

76%

79%

0.4

Time <70 mg/dL

2%

3.9%

0.02

Time <54 mg/dL

0.1%

0.4%

0.009

Time >180 mg/dL

22%

17%

0.4

Time >250 mg/dL

4.9%

5.7%

0.9

Glycemic variability (CV)

35.8%

35.2%

0.8

Time in Auto Mode

97%

96%

0.5

  • Time in Range was consistent between Klue-enabled MiniMed 780G without meal announcements and MiniMed 780G with manual carbohydrate counting/meal announcements (76% vs. 79%; p=0.41). This noninferiority in Time in Range was accompanied by a significant improvement in time below range and time <54 mg/dL with KLUE relative to manual carbohydrate counting. Specifically, time <70 mg/dL fell from 4% with manual carbohydrate counting to 2% with Klue (p=0.02), and time <54 mg/dL fell from 0.4% with manual carbohydrate counting to 0.1% with Klue (p=0.009). Looking at daytime specifically, Time in Range was consistent (77% vs. 73%; p=0.3), and time <70 mg/dL was 35 minutes/day lower with Klue relative to manual carbohydrate-counting (1.6% vs. 4.1%; p=0.007). There was a non-significant trend toward increased daytime time >180 mg/dL with Klue than with carb counting (26% vs. 18%; p=0.09), but further research would be necessary to confirm that. There were no differences in time above range, time >250 mg/dL, glycemic variability, or time in Auto Mode between the Klue period and the manual carbohydrate counting periods. Together, these results suggest that Klue enables noninferior Time in Range outcomes with improved time in hypoglycemia relative to manual carbohydrate counting, all while reducing user burden by removing the need to count carbohydrates.
  • While overall the glycemic outcomes were similar with Klue and carb counting, Klue did not perform well in high-carbohydrate meal tests. Looking at the four hours after each meal test, Time in Range was similar for MiniMed 780G with Klue and MiniMed 780G with carbohydrate counting after low- and medium-carbohydrate meals regardless of size. However, when tested in high-carbohydrate meals, Klue performed significantly worse in both the small or normal size meal tests (66% vs. 48% in small, high-carb meal and 64% vs. 27% in normal-size, high-carb meal).
  • Understandably, the audience had many questions about the safety guardrails in place, particularly those that prevent overdosing of insulin. Responding to Dr. Trang Ly’s (Insulet) questions about false positives, Dr. Roy noted that while there were some false positives, they usually came randomly and were seldom in a group. To avoid false positive-driven insulin dosing, the algorithm did not react to the first gesture coming in but rather “waited until it crossed a threshold” at which it was considered a meal. As a further safety measure, the boluses delivered were smaller than the full carbohydrate count estimated by Klue, which may explain why the system did not respond well to high-carb meals. Dr. Pratik Choudhary (University of Leicester, UK) also had concerns about no-carb meals; however, Dr. Roy assuaged such concerns noting that there were “other fail-safe measures” beyond the gestures, which we presume may be based on glucose sensing, although that’s speculative.
  • Based on these findings, the researchers are working to further improve the Klue-enabled system in several ways. As noted above, Klue can distinguish eating and drinking gestures. In the version used in this feasibility study, only eating gestures were used to dose insulin. Looking ahead, the Medtronic team intends to add insulin dosing following drinking gestures followed by a rise in glucose levels to address glycemic changes following the consumption of high-carb drinks (e.g., soda) while avoiding false-positive for no/low-carb drinks (e.g., water). Separately, given that Klue did not perform well in high carb meal tests, the researchers are working to make Klue more responsive to high-carb meals. An updated version of the Klue-enabled MiniMed 780G system with these improvements is being tested in a second feasibility study in Israel, which is already underway. Dr. Roy stated that the preliminary data looks “very encouraging.”
  • These findings bring further progress toward the effort to develop fully closed loop systems. Other key players in this effort are UVA, which is developing RocketAP with enhanced meal detection software, McGill, which is taking multiple routes including co-administration of insulin and pramlintide, CamDiab, whose CamAPS HX does not require meal announcements or carb counting, and Beta Bionics, whose iLet system (while not fully closed loop) requires only simple meal announcements (smaller than usual, usual for me, larger than usual). There are of course others (including those like Tandem who have stated intentions to have a fully closed loop system but haven’t offered specifics publicly), but these are those at the top of our mind and whose efforts have been made public through publications and presentations.

Two-year results from CLOuD RCT suggest AID initiation from type 1 onset does not limit C-peptide deterioration but does significantly improve glycemic control relative to standard care (+3.4 hours/day in Range and -1% A1c at two years)

In an action-packed session dedicated to insulin delivery innovations, Dr. Charlotte Boughton (University of Cambridge) read out the results of the CLOuD Study, an RCT evaluating the impact of AID therapy on type 1 diabetes progression and glycemic outcomes when initiated within 21 days of diagnosis. Participants (ages 10-16, average 12) had an average baseline A1c of 10.6% and about 30% had DKA at diagnosis. Upon enrollment, participants engaged in a mixed meal tolerance test (MMTT, n=101) and then were randomized to either CamAPS FX (n=51) or standard therapy (n=46). Four participants withdrew after the MMTT, four withdrew in the CamAPS FX arm, and eight withdrew in the control arm, leaving 47 and 40 participants in the CamAPS FX and control arms, respectively, in the intention to treat analysis. Every three months, participants had an A1c test and wore a masked Libre Pro CGM, and at six months, one year, and two years, participants engaged in MMTT to assess c-peptide levels. Although the study is continuing through to 48 months (four years), the analysis that Dr. Boughton shared included the first 24 months of data. Overall, there was no difference in c-peptide levels at twelve or 24 months between the CamAPS FX and standard care groups, and both saw reductions in C-peptide levels from baseline. Although there was no difference in C-peptide levels, participants in the CamAPS FX group saw a significant improvement in glycemic outcomes relative to those in the standard care arm, which are described further below. On safety, there was no significant difference in severe hypoglycemia events, DKA events, serious adverse events, or other adverse events. Based on these results, Dr. Boughton suggested that: (i) improving glycemic control is not enough to slow the deterioration of C-peptide levels; and (ii) AID is feasible and beneficial when initiated soon after type 1 diabetes onset. This study is hugely significant, as even if it doesn’t achieve its primary outcome pertaining to C-peptide levels, it provides support for the use of AID technology as soon as possible in type 1 diabetes.

 

Baseline

12 months

Mean adjusted difference

24 months

Mean adjusted difference

 

CamAPS FX

Standard Care

CamAPS FX

Standard Care

 

CamAPS FX

Standard Care

 

Time in Range

74%

72%

64%

54%

+2.4 hours/day

64%

49%

+3.4 hours/day

A1c

10.7%

10.5%

6.9%

7.3%

-0.4%

6.9%

8%

-1%

Time <70 mg/dL

9%

11%

6%

5%

+13 minutes/day

11%

7.5%

+40 minutes/day

  • Compared to the standard care arm, those on CamAPS FX spent 2.4 more hours/day in Range at one year and 3.4 more hours/day in Range at two years, when adjusted for baseline. Both groups saw deteriorations in Time in Range from baseline to one year, falling from 74% to 64% in the CamAPS FX group and from 72% to 54% in the standard care arm. However, the CamAPS FX group saw less deterioration in Time in Range and spent an additional 2.4 hours/day in Range compared to those receiving standard care when adjusted for baseline (64% vs. 54%; p=0.02). Furthermore, at two years, those on CamAPS FX maintained their one-year Time in Range of 64% while those on standard care further deteriorated to a Time in Range of 49%, expanding the baseline-adjusted between group difference to 3.4 hours/day. At baseline, one year, and two years, time below range was higher than is generally seen with AID systems, but Dr. Boughton attributed this to the use of masked Libre Pro to assess glycemic outcomes, which is known to overestimate hypoglycemia.
  • When adjusted for baseline, the CamAPS FX group saw a -0.4% relative A1c improvement compared to the standard care arm at one year, which expanded to a -1.0% relative improvement at two years. Unlike Time in Range, both groups saw A1c improvements from baseline to one year, falling from 10.7% to 6.9% in the CamAPS FX group and from 10.5% to 7.3% in the standard care arm. However, at two years, the CamAPS FX group maintained their A1c of 6.9% while those receiving standard care saw their A1c rise to 8.0%. Only 56% of the standard care arm achieved an A1c <7.5% at one year, which fell to 39% at two years, despite high technology uptake (43% were on a pump and 68% on a CGM by year two). This compares to 78% in the CamAPS FX group at one year. This suggests that CamAPS FX might prevent the deterioration in glycemic control seen in children as their type 1 diabetes progresses.

Packed Omnipod 5 product theater offers new insights on security of smartphone control, iOS smartphone control efforts, and next-gen CGM integrations; Dr. Viral Shah calls for unbiased AID discussions with all people with type 1 diabetes

The Monday morning Omnipod 5 product theater was packed with attendees eager to learn about the newest AID system on the market from Dr. Trang Ly (Insulet Medical Director), Dr. Grazia Aleppo (Northwestern), and Dr. Viral Shah (Barbara Davis). Dr. Ly kicked the session off with a review of the system’s key features and offered several updates to what we previously knew, which we discuss further in the bullets below. Dr. Aleppo then reviewed the most recent Omnipod 5 data in type 1 diabetes, including the 15-month extension pivotal data across children and adults and 12-month preschool data that was presented earlier at ADA 2022. Dr. Shah closed out the session by discussing his belief – and the consensus report guideline – recommending that all people with type 1 diabetes are offered AID, particularly if they are not meeting targets, have challenges with hypoglycemia or glycemic variability, or if their diabetes is substantially reducing their quality of life. He refuted the sentiment that people who have high A1c values or are not following their treatment plan shouldn’t be offered technology, calling the belief a “concept of the past.” Dr. Shah’s discussion fit well with the Insulet and Tandem symposiums at EASD 2021 and the increasing education about the particular value of AID technology in those struggling with managing their diabetes.

  • In what was news to us, Dr. Ly shared specifics around the smartphone control interface and security. Using the slide shown below, Dr. Ly highlighted that the smartphone control is quite secure and requires a security step (e.g., facial or fingerprint recognition) to access the smartphone control feature. She noted that this is beneficial for a parent whose child might be using their phone or a person whose phone might accidentally turn on and open apps in their pocket to make sure these accidents don’t result in unintentional insulin dosing. Following Dr. Ly’s explanation, she showed a video of an Insulet employee’s interaction with their Android-based Omnipod 5 app, bolus calculator, smartphone control feature to dose insulin – it’s exciting to see what will hopefully soon become the norm in AID technology! Per Dr. Ly, there are currently seven Android models for which Omnipod 5 smartphone control is available (in line with ATTD 2022 commentary). Tandem also has FDA approval for smartphone bolus control (in limited launch), and this is also in the works at Medtronic, although it’s further off.

  • Preemptively addressing what was sure to be a question, Dr. Ly stated that Insulet is “well underway” in its efforts to integrate with Dexcom G7 and FreeStyle Libre CGMs and to develop iOS smartphone dosing control. Today’s presentation also offered a glimpse of what the apps could look like when integrated with these next-gen CGMs and on an iOS device. Dr. Ly’s discussion of these efforts offered two notable learnings: (i) iOS smartphone control is in human factors testing, an update from February 2022 when iOS was still in the technical development phase; and (ii) the slide specifically states FreeStyle Libre 2. The latter is particularly notable because Insulet has never once (as far as we’re aware) stated which FreeStyle Libre CGM it intends to integrate into its AID system. Previously, in February, Ms. Petrovic expressed her belief that an integration with a FreeStyle Libre CGM is “likely to require a submission,” the work for which is already underway. Given that FreeStyle Libre 2 has a contraindication against use in AID technology, integration would also require removal of that contraindication.

  • Dr. Ly noted that all prescribers can now prescribe their patients Omnipod 5, but that these orders currently must be placed a select group of pharmacies. This was certainly exciting for the rapturous audience and is in line with the commentary we heard during Insulet’s 1Q22 call in early May, when management stated that it was ramping up its limited market release from hundreds of users to thousands of users by enabling anyone to prescribe through a limited distribution network.
  • Like at ATTD 2022, Dr. Ly highlighted the Omnipod 5 Simulator App (Google PlayApple Store), which enables potential users and prescribers to gain a better understanding of the app, of the training/setup process, and of the smartphone control feature. She also showcased provider learning tools, including the Connect with a Peer program, which allows clinicians to learn directly from the experiences of Omnipod 5 clinical trial investigators and to ask them questions, as well as the Omnipod Connect tool, a new HCP online portal that allows clinicians to learn about Omnipod 5 technology at their own pace.

Omnipod 5 pivotal trial extension phase: glycemic improvements maintained out to 15 months with pediatric Time in Range of 65% and adult Time in Range of 73%

Insulet presented new data from participants in the Omnipod 5 pivotal trial (759-P) demonstrating that improvements in A1c and Time in Range were maintained out to 15 months. As a reminder, results from the Omnipod 5 pivotal trial were presented at ENDO 2021 and demonstrated an average Time in Range of 70% for children and 74% for adults up from 52% and 64% at baseline (p<0.05), respectively. We saw additional nine month data from this population presented at EASD 2021 demonstrating that these improvements in Time in Range were maintained, and this new data out to 15 months continues this trend. Specifically, among children ages 6-13 (n=110), Time in Range at months 10-12 and months 13-15 was 67% and 65%, respectively, roughly in-line with the 67% Time in Range achieved in the first three months of the pivotal study. Adult participants also maintained their improvements in glycemic management from the Omnipod 5 pivotal trial with average Time in Range at months 10-12 of 73% and months 13-15 of 73%, in-line with the 74% Time in Range achieved in the first three months of the pivotal trial.

  • Participants also maintained A1c improvements out to 15 months. Among children, A1c at 12 and 15 months was 7% and 7.2%, respectively. This is roughly in-line with an A1c of 7% after the three-month pivotal among pediatric participants. For adults, A1c at 12 and 15 months was 6.8% and 6.9%, respectively, which was also in-line with an A1c of 6.8% after the three-month pivotal.

  • As we’ve seen for many AID systems, the largest improvements in A1c were seen among patients with baseline A1c >8% and these improvements were largely maintained at 15 months. Specifically, among children with a baseline A1c >8% (n=38), 12- and 15-month A1c was 7.5% and 7.7%, respectively, which was in-line with an A1c of 7.6% from the three-month pivotal. For adults with a baseline A1c >8% (n=20), 12- and 15-month A1c was 7.4% and 7.6%, which was in-line with an A1c of 7.6% after the three-month pivotal.

Omnipod 5 preschool pivotal: Nine-month extension phase (n=80) shows sustained outcomes following three-month pivotal; participants spend +2.5 hrs/day in Range and achieve 0.5% A1c reduction at one year vs. baseline

Following the readout of Omnipod 5’s preschool pivotal results at ADA 2021, Dr. Daniel DeSalvo (Baylor) presented results from the extension phase of Omnipod 5’s preschool pivotal trial (ages 2-<6). Insulet also announced the data via a press release on their website. As a reminder, while Insulet’s Omnipod 5 was cleared by the FDA in January for ages six and up, the preschool indication for Omnipod 5 remains under active FDA review. Per CEO Ms. Shacey Petrovic during Insulet’s 1Q22 earnings call, the company expects the FDA to clear Omnipod 5 for children two and older sometime “this year,” and when this happens, the expanded indication would be a major coup for Insulet and for young children with diabetes, for whom the pod form factor is well-suited. Very impressively, Dr. DeSalvo explained that all participants (n=80) who participated in Omnipod 5’s preschool pivotal proceeded to the extension phase of the trial, which ran for nine months after the three-month pivotal study.

  • The impressive glycemic improvements seen in the pivotal trial were sustained through the entire extension study, with participants achieving +2.5 hours/day Time in Range at month 12 compared to baseline (68% vs. 57%, p<0.05). Participants saw almost all their glycemic improvements, on average, during the first three months of system use, with Time in Range increasing to 68% at month three and then staying flat through month 12 of the study. Similar patterns were observed for other CGM-derived metrics and for A1c. At month 12, participants spent -3.5 minutes/day Time Below Range and -2.3 hours/day Time Above Range compared to baseline (p<0.05). A1c improved by -0.5% from 7.4% at baseline to 6.9% at month 12 (p<0.05), and 49% of participants achieved an A1c <7% after 12 months of Omnipod 5 use compared to only 31% of participants at baseline.

 

Baseline

AID Months 1-3

AID Months 4-6

AID Months 7-9

AID Months 10-12

A1c

7.4%

6.9%

7.0%

7.1%

6.9%

Time in Range

57%

68%

68%

68%

68%

Time <70 mg/dL

2.2%

1.9%

2.1%

2.4%

1.9%

Time >180 mg/dL

40%

30%

29%

30%

30%

Mean Glucose (mg/dL)

171

157

157

158

159

  • The percentage of preschoolers achieving consensus targets for glycemic control was sustained from the pivotal phase to the extension phase. At the end of the extension phase, 42% of participants had a Time in Range >70%, compared to 44% at three months, and only 18% at baseline. On Time Below Range, 85% of participants had spent <4% of time below 70 mg/dL at month 12, compared to 81% at month three, and 71% at baseline. Putting these metrics together, the percentage of participants met the composite endpoint of <4% Time Below Range and >70% Time in Range improved from 11% at baseline to 30% at month three and to 32% at month 12. Notably, Dr. DeSalvo explained that if the Time in Range target was lowered to >60%, then the percentage of people meeting Time in Range improved from 46% at baseline to 78% at month 12. We are very pleased to see the strong sustained glycemic improvements achieved by preschool-aged children using Omnipod 5, as we understand that this population can be an especially challenging one when it comes to managing diabetes.

  • Unsurprisingly, Omnipod 5 was incredibly successful at night (midnight – 6 am), driving preschoolers to achieve +1.2 hours/night in Range at one year compared to baseline, with Time in Range rising from 58% to 78% (p<0.05). Dr. DeSalvo explained that participants also saw improvements in Time Above Range (-1.1 hours/night to 20%) and Time Below Range (-1.7 minutes/night to 1%). The system was also incredibly safe, with no episodes of severe hypoglycemia or DKA recorded throughout the entire extension phase.

Medtronic MiniMed 780G with Guardian 4 CGM drives strong real-world glycemic outcomes among early adopters (n=7,346) with average Time in Range of 73%

In a poster presentation (760-P), Medtronic presented real-world data from some of the first MiniMed 780G with Guardian 4 CGM users across 21 European countries who had at least 10 days of sensor glucose data available in CareLink (n=7,346) demonstrating average Time in Range of 73%. Across this population, patients spent an average of 92% of time in auto mode with only 0.12 exits from auto mode per day. This represents a significant improvement from the frequent auto mode exits that were traditionally seen with Medtronic’s 670G system, largely due to the calibration requirements of the Guardian Sensor 3 CGM. Turning to other glycemic data from these early adopters, patients had an average GMI of 7% and an average sensor glucose of 153 mg/dL. While daytime Time in Range was reported at an already impressive 73%, overnight Time in Range was even higher at 81%. We are encouraged to see this high overnight Time in Range as we typically see the big value of AID systems come from their ability to smooth overnight glycemic values and help patients wake up in range and it is exciting to know that these features are maintained in real-world use. That said, we would note that this dataset includes the first 7,000 users of MiniMed 780G and Guardian 4, which may not always represent the larger population of people with diabetes. However, with such strong glycemic outcomes, we can’t help but be encouraged for more people to gain access to MiniMed 780G and we hope to see these outcomes maintained across users, regardless of when they start using the system.

  • Stratified by age, MiniMed 780G was associated with strong glycemic management for users ≤15 years old (n=2,159) as well as those >15 years old (n=4,786). Specifically, among users pediatric users, average Time in Range was 71% overall and 80% overnight. Patients in this cohort also had an average GMI of 7%. Among MiniMed 780G users >15 years old, average Time in Range was 74% with 80% overnight Time in Range and an average GMI of 6.95%. Those who used the optimal settings (a two-hour active insulin time and 100 mg/dL target) achieved even better outcomes (Time in Range of 79% and GMI of 6.8%).
  • A small cohort of MiniMed 780G users (n=59) had type 2 diabetes and also demonstrated strong glycemic management while using the system. Specifically, patients with type 2 diabetes had a mean sensor glucose of 152 mg/dL, an average GMI of 6.95%, and a Time in Range of 75%. These data add to a growing body of evidence supporting the use of AID systems in people with type 2 diabetes on basal-bolus insulin therapy.

More from the pediatric insulin-only iLet pivotal: Children with baseline A1c values >9% see whopping +7.4 hour/day Time in Range with iLet relative to standard care (56% vs. 25%); 75% of youth previously on AID feel iLet is better

As part of the mid-day symposium on the insulin-only iLet pivotal, Dr. Laurel Messer (Barbara Davis) and Dr. Jill Weissberg-Benchell (Lurie Children’s Hospital) read out pediatric glycemic results and patient reported outcomes, respectively. The results that the two researchers shared built upon those read out by Dr. Steven Russell (Massachusetts General Hospital) at ATTD 2022. During ATTD 2022, Dr. Russell reported that when adjusted for baseline, children on iLet (n=112) saw a 0.5% A1c improvement and +2.4 hour/day Time in Range improvement relative to children receiving standard care (n=53). Specifically, in the pediatric iLet group, A1c fell from 8.1% to 7.5% while the standard care arm saw no change from 7.9% at baseline to 13 weeks. Likewise, those in iLet saw Time in Range improve from ~47% to 60% within the first four weeks, which was maintained out to week 13. The standard care arm’s Time in Range improved only from 48% to 50%. Today’s symposium offered further details into the glycemic outcomes of study subgroups, including those with baseline A1c values >9% and those on different insulin delivery methods at baseline, and patient-reported outcomes for the overall pediatric cohort.

  • Perhaps the most impressive outcomes shared during today’s symposium were the glycemic improvements seen in children who had a baseline A1c >9%. These participants averaged a 9.7% A1c and 22%-28% Time in Range at baseline. Those who used iLet saw a whopping -2.1% A1c improvement and spent +7.4 hours/day in Range relative to those in the standard care arm (p<0.001 for all baseline-adjusted between group comparisons). As Dr. Laurel Messer noted, these are incredible improvements, particularly given that these are children who will have diabetes for the rest of their lives and improving outcomes at a young age can dramatically improve their long-term quality of life and complication risk.

  • Youth who switched from MDI to iLet saw the greatest A1c improvements, although all insulin delivery subgroups saw A1c improvements from baseline to 13 weeks with iLet, and none saw significant improvements if they stayed on standard care. Specifically, children on MDI at baseline saw a 1.0% A1c decline after 13 weeks of iLet use (8.4% to 7.4%) compared to a -0.5% decline in those who had previously been on pump therapy (8.0% to 7.5%) and a -0.3% A1c decline in those who had previously been on a different AID system (8.0% vs. 7.7%).
  • Because pediatric participants’ psychosocial wellbeing was already very good at baseline, no significant PRO treatment effects were found. According to Dr. Weissberg-Benchell, this was true of measures for distress, fear of hypoglycemia, positive expectancies, treatment satisfaction, well-being, and perceived health status, although no specific figures were reported. Although improvements in PROs were not reported, 75% of youth who had previously been on an AID system reported that iLet was better or much better than their previous system, and 67% stated that they would refer to remain on the iLet system. Similarly, 74% of parents whose children transitioned from another AID system to iLet rated iLet as better or much better than the previous system, and 66% stated that they’d prefer that their children stay on iLet. Parents also reported a significant improvement in treatment satisfaction with iLet relative to standard care (p=0.001).
  • Today’s symposium also offered further insights into the makeup of the pediatric population studied. Specifically, Dr. Messer shared that the pediatric cohort was even more diverse than was the overall trial. As a reminder, the overall iLet trial (including adults and youth) was one of the most diverse AID clinical trials we’ve ever seen with over a fourth of participants non-Hispanic Black (10%), Hispanic or Latinx (10%), or another non-White race (6%). Today, we learned that the pediatric cohort was particularly diverse, with 35% of participants non-White (10% non-Hispanic Black, 15% Hispanic or Latino, and 10% other). This is hugely important given the disparities in diabetes technology use and outcomes that are particularly well-documented in children with type 1 diabetes.
  • Teens who began the trial with high A1cs showed a significant decline in fear of hypoglycemia. Additionally, youth who began the trial with high A1cs showed a significant decline in diabetes distress and parents who began the trial with high diabetes distress had children with a significant decline in A1c. Last, families with less financial resources showed a significant improvement in diabetes distress.

More from the adult cohort in the insulin-only iLet pivotal: Nearly half achieved >0.5% A1c improvement and a quarter a >1% A1c improvement; A1c improvements seen across income level, race/ethnicity, education level, and insulin delivery method subgroups; plus, positive PRO results

As part of the symposium dedicated to the insulin-only iLet pivotal, Dr. David Kruger (Henry Ford Health Systems) and Dr. Jill Weissberg-Benchell (Lurie Children’s Hospital) shared new glycemic and psychosocial data from the cohort of adults on iLet with aspart/lispro. As a reminder, we saw the overall A1c and Time in Range data at ATTD 2021 during the initial insulin-only iLet pivotal readout. During that session, Dr. Steven Russell (Massachusetts General Hospital) reported that adults on iLet (without Fiasp) saw a 2.6 hour/day Time in Range improvement relative to those on standard care, consistent with the overall sample, and saw their Time in Range improve from ~56% at baseline to 69% at 13 weeks, an improvement observed within the first day. The standard care arm saw a slight improvement as well from ~53% to 58% at 13 weeks. Adults (excluding those on Fiasp) also saw a 0.5% A1c improvement when on iLet vs. standard care (p<0.001), with the iLet group seeing their A1c fall from 7.6% at baseline to 7.1% at 13 weeks while the control group declined only slight from 7.6% to 7.5%. Today’s session built on these overall results and offered subanalyses stratified by baseline insulin delivery method and by demographic variables, overall showing improvements across a wide range of people with type 1 diabetes.

  • Nearly half (43%) of those in the iLet arm saw a >0.5% A1c improvement at 13 weeks compared to only 17% in the standard care arm (p<0.001). Likewise, nearly a fourth of participants (23%) in the iLet arm improved their A1c by >1% compared to only 4% in the standard care arm (p<0.001). As can be expected, those with higher baseline A1c values saw greater A1c improvements with iLet, and those with baseline A1c values >9% saw a ≥2%improvement. The greatest Time in Range improvement was seen for the individual with a baseline A1c value of 12%, who saw a >4% A1c decline with iLet, a massive improvement in only three months.

  • When stratified by insulin delivery type, those on MDI at baseline saw the greatest A1c improvements with iLet compared to those who were on a pump without automation or a different AID system at baseline. Specifically, those on MDI at baseline saw a -1.0% A1c reduction (8.2% to 7.2%; no change in standard care comparators), and those on pump therapy or an AID system at baseline saw a 0.3% A1c improvement (7.3% to 7.0%; no change in standard care comparators).

  • Because the study’s intentional diversity, the researchers were able to stratify the adult results by race/ethnicity, education level, and income. Across the board, iLet improved A1c, and there was no significant improvement with standard care. However, the degree of improvement was higher in those who were non-Hispanic non-White vs. non-Hispanic White (-1.1% vs. -0.4%), in those with less than a Bachelor’s degree vs. those with at least a Bachelor’s degree (-0.9% vs. -0.4%), and in those with an income <$100,000 vs. those with incomes ≥$100,000 (-0.7% vs. -0.4%). That said, those in the groups with greater A1c improvements started at a higher baseline A1c, which likely explains much of this difference.

  • On patient-reported outcomes, adults on iLet saw significant improvements in diabetes distress, hypoglycemia fear, and mental wellbeing relative to those in the standard care arm (p<0.001; measured via Diabetes Distress Scale, Hypoglycemia Fear Survey, and WHO-5, respectively). The improvements in wellbeing and in diabetes distress with iLet are consistent with the findings of studies evaluating MiniMed 670G and Control-IQ; however, this is the first study to report improved fear of hypoglycemia with an AD system. Notably, 60% of those who had previously used an AID system reported that iLet was better or much better than their prior system.

Exploratory analysis of insulin-only iLet pivotal shows minimal glycemic improvement with Fiasp vs. aspart/lispro: No A1c differences and only slight +29 minute/day Time in Range improvement

Dr. Steven Russell offered a first look at the exploratory analysis comparing iLet with aspart/lispro (n=103) vs. with Fiasp (n=113). During the ATTD 2022 readout, Dr. Russell presented data comparing the iLet with Fiasp arm to the standard care arm (-0.5% baseline-adjusted between-group difference, p<0.001), but we had not previously seen a direct comparison of Fiasp vs. aspart/lispro with iLet. Before delving into the results, Dr. Russell noted that the algorithm was not changed for use with Fiasp, and that the same PK parameters were used for aspart, lispro, and Fiasp. Fiasp was provided to participants in pre-filled glass cartridges while those using aspart or lispro manually filled their glass cartridges. Overall, those who used iLet with Fiasp did not see much improvement in glycemic outcomes or PROs compared to those using iLet with aspart or lispro. When adjusted for baseline, there was no difference in the A1c improvement seen by the groups (0.0%; p=0.67). Furthermore, the distribution of A1c values at six weeks and at thirteen weeks was identical between the two groups. There were also no significant between-group differences in: (i) mean glucose (2 mg/dl lower with Fiasp; p=0.1); (ii) proportion achieving a ≥5% Time in Range improvement (p=0.49); (iii) median time <54 mg/dl (p<0.001 for noninferiority); (iv) total daily insulin dose (p=0.73); (v) severe hypoglycemia events per 100 patient years (p=0.83); (vi) DKA events (none in either group); and (vii) patient-reported outcomes, including scores for diabetes distress, hypoglycemia confidence, hypoglycemia fear, treatment satisfaction.  The only outcomes that were significantly different were related to Time in Range and are detailed below.

  • Mean Time in Range was 29 minutes/day higher in those using iLet with Fiasp compared to those using iLet with aspart/lispro (p=0.005), and the distribution of participants’ Time in Range values at 13 weeks was slightly better in the Fiasp group compared to the aspart/lispro group. The slight – but significant – Time in Range improvement with Fiasp was solely observed during the day when those using Fiasp achieved a 70% Time in Range at 13 weeks compared to 67% for those using aspart/lispro. Of course, this makes sense given that Fiasp has been shown to be most beneficial around mealtimes.

  • Significantly more participants in the Fiasp arm achieved a Time in Range >70% compared to those in the aspart/lispro arm (58% vs. 47%, respectively; p=0.01). However, there was no significant differences in the proportion of participants achieving ≥5% or ≥10% Time in Range improvements between the two iLet arms. Specifically, 80% of those in the Fiasp arm and 74% of those in the aspart/lispro arm achieved a ≥5% Time in Range improvement (p=0.5). These figures fell to 74% and 68%, respectively, when looking at a ≥10% Time in Range improvement (p=0.5). Even though there was no difference between the groups in these outcomes, both iLet groups saw positive outcomes – in particular, it’s positive to see that regardless of the insulin used, the majority of participants using iLet saw a ≥10% Time in Range improvement.

12-month CLIO results demonstrate sustained glycemic improvements; substantial 0.7% A1c improvement among prior MDI users (n=225) to 7%

New data from the CLIO study presented in a poster (761-P) demonstrated sustained improvements in glycemic management out to 12 months among Control-IQ users (n=1,107). As a reminder, the CLIO study is a real-world investigation of glycemic outcomes among adults using Control-IQ. Participants who uploaded Control-IQ data and had ≥75% active CGM time over the course of the 12-month study were included in this analysis. Among this CLIO study population (n=882), patients had an average age of 41 and the majority had previously used insulin pump therapy before switching to Control-IQ. Similarly to earlier CLIO data on three and nine month outcomes, prior MDI users (n=225) saw the largest improvements in glycemic management over the 12-months from a baseline A1c of 7.7% to a 12-month GMI of 7.0%. For prior pump users, glycemic improvements were less pronounced going from a baseline A1c of 7.2% to a 12-month GMI of 7.1% (p<0.0001). Evaluating both groups, we are impressed to see similar glycemic outcomes regardless of prior therapy demonstrating the ability of AID to be effective across patient populations. We are very excited to see that patients have largely maintained their early improvements in glycemic outcomes while using Control-IQ with consistent outcomes throughout the 12-month study and we are looking forward to additional real-world data as it becomes available.

  • All ago cohorts of patients demonstrated improvements in glycemic management from baseline to 12-months. The largest absolute improvement was seen in adults ages 31-45 previously on MDI (n=65) who experienced a 1% decrease in A1c from 8% at baseline to a GMI of 7% at 12-months (p<0.0001).

Bigfoot Unity real-world data (n=49) demonstrates strong glycemic improvements from baseline A1 of 8.5% to 3-month GMI of 7.5%; largest improvements in glycemic outcomes among patients with baseline A1c ≥8%

Late-breaking data (68-LB) from real-world users of the Bigfoot Unity connected insulin pen system (n=49) demonstrated strong improvements in glycemic management following three months of use. As a reminder, Bigfoot Unity received FDA approval in May 2021 and became available in the US on a clinic basis in June 2021 providing users with a FreeStyle Libre 2 CGM and connected insulin pen cap to provide insulin dosing recommendations based on clinician input. In this study, 49 adults with diabetes across 11 clinical sites with at least 50% of CGM data in the first two weeks and third month of system use were included and the data were analyzed to assess the impact of Bigfoot Unity on glycemic management. At baseline, patients had an average A1c of 8.5% which quickly dropped with an average GMI of 7.4% after only 14-days on Bigfoot Unity. This decline in GMI was then maintained out to three months with patients achieving an average GMI of 7.5% between days 61-90 of Bigfoot Unity use. These rapid gains  in glycemic management as early as two weeks following system initiation closely mirrors data from AID systems that have demonstrated rapid improvements in glycemic management followed by a leveling off of improvements. Excitingly, while this was three month data, the poster did reference that 47 of the 49 patients included here have continued using Bigfoot Unity out to six months.

  • Time in Range remained relatively consistent across the three-month duration of this analysis at 62% in the first two weeks and 59% from days 61-90. While participants did not see an increase in Time in Range across this real-world evaluation, we are curious what baseline values prior to Bigfoot Unity use may have looked like. More specifically, given that patients experienced a marked decrease from baseline A1c to GMI by week two, we are curious if a similar phenomenon may be in effect with Time in Range that may have been missed in this data set due to the timing of Time in Range data collection. We assume so and are eager to check in on this.
  • Among patients with a baseline A1c ≥8% (n=26), for whom the average A1c was 9.5%, Bigfoot Unity was associated with a substantial reduction in GMI down to 7.5% at three months. We are very encouraged to see this data demonstrating that Bigfoot Unity has the potential to improve glycemic outcomes among higher risk patients. Of note, this theme is not unique to Bigfoot Unity, but something we continue to see across diabetes devices reinforcing that there is no “ideal patient” for diabetes technology, and that all patients who want to use technologies have the potential to see glycemic improvements. 

Beta Bionics’ insulin-only iLet pivotal 13-week extension phase: Transition from standard care to iLet results in -0.6% A1c drop and +2.9 hour/day Time in Range improvement; pediatric participants using iLet with Fiasp in extension phase see similar results as those using iLet with aspart/lispro in main study phase

Beyond the readout of the full adult, pediatric, and adult Fiasp insulin-only iLet pivotal results on Friday, the poster hall also featured data from the extension phase of the pivotal trial, during which those who were randomized to standard care had the opportunity to use the insulin-only iLet system for 13 weeks (98-LB). Notably, in the extension phase, pediatric participants (n=48; ages 6-17 years, mean A1c 7.8%) used Fiasp while adults (n=42; age ≥ 18 years, mean A1c 7.5%) used lispro/aspart. Overall, A1c fell 0.6% from 7.7% at baseline to 7.0% at 13 weeks (p<0.001), in line with the benefit seen in the iLet group during the main study. Likewise, Time in Range improved +2.9 hours/day from 52% to 65% (p<0.001), similar to the Time in Range seen in the main study. Those who transitioned to iLet in the extension study saw a slight but significant reduction in time <70 mg/dL, falling seven minutes/day from 2.5% to 2.1% (p=0.02). There was no significant difference in time <54 mg/dL (p=0.24). On safety, there were two severe hypoglycemia events, both of which occurred in the same adult who had had two events during the preceding RCT while on MDI, and one DKA event in a pediatric participant which was related to an infusion set failure. Overall, these results are in line with those seen in the group randomized to use iLet in the pivotal study and confirm the glycemic benefit of the insulin-only iLet system.

  • The results were generally similar between adult and pediatric participants. Both saw a -0.6% A1c improvement and a ~2.9 hours/day Time in Range improvement (2.8 hours/day and 2.95 hours/day, respectively). However, the reductions in glycemic variability and time <70 mg/dL were only significant in the pediatric age group (-1.7% CV, -12 minutes/day <70 mg/dL).
  • The extension phase enables a comparison of iLet with Fiasp vs. aspart/lispro in children with diabetes. While hard to compare head-to-head, the results (shown below) signal that the improvements seen in the pivotal and in the extension phase were generally similar, although slightly greater improvements were seen in the extension phase period. However, these between-group differences may not have been significantly different; the poster did not state the statistical significance.

 

Pivotal pediatric cohort on iLet with aspart/lispro (n=112)

Extension pediatric cohort on iLet with Fiasp (n=47)

A1c change

-0.5%

-0.6%

Time in Range change

+2.4 hours/day

+3.0 hours/day

Time <54 mg/dL change

-35 sec/day

-2 minutes/day

Mean glucose

-15 mg/dL

-19 mg/dL

T1D Exchange analysis (n=10,653) finds that AID use is correlated with improved A1c and less hypoglycemia compared to other insulin delivery/sensing methods across age cohorts; 60% of those on AID achieving A1c <7%; those not using CGM are >2x more likely to experience severe hypoglycemia events compared to those on AID

T1D Exchange’s Dr. Kellee Miller read out results from analysis from a massive cohort of T1D Exchange participants (n=10,653) assessing technology use across age groups and the association between technology use and glycemic outcomes (289-OP). The analysis included participants (ages 1-91) who completed a questionnaire between January 2020 and December 2021. Participants averaged age 38, were 69% female, 88% non-Hispanic White, and mostly on private insurance (73%), had diabetes for an average 17 years, and had an average A1c of 7.3%. The vast majority were on CGM (86%) and pump (71%) technology, which of course, means that this cohort is not representative of the general US type 1 diabetes population. Based on self-reported insulin delivery and glucose sensing method, participants were stratified into: (i) AID (DIY looping, Control-IQ, and MiniMed 670G/770G); (ii) predictive low glucose suspend technology (Basal-IQ, MiniMed 640G); (iii) sensor-augmented pump therapy (CGM+pump); (iv) pump (pump+BGM); (v) CGM+MDI; and (vi) MDI (MDI+BGM).

  • Across all age groups, sensor-augmented pump was the most used technology, followed by MDI+CGM. Across cohorts ages <13 (youth), 13-17 (teens), 18-29 (young adults), 30-55 (adults), and >55 (older adults), at least a third of participants were on sensor-augmented pump therapy. A third of youth (ages <13) were on MDI+CGM, which fell to 19% in teens and adults, to 17% in young adults, and to 21% in older adults. AID use averaged ~11%-17% of the cohort, and PLGS use ranged from 7% in youth to 16% in young adults and adults. Very few participants were on pump therapy without CGM (2%-7% of each age cohort) or on MDI without CGM (5%-10% of each age cohort).
  • In line with previous research, those using more automated technology had lower A1c values. Overall, 60% of those on AID had an A1c <7%. When adjusted for age, those on AID had significantly lower A1c values and were significantly more likely to achieve an A1c <7% relative to those on sensor-augmented pump therapy, on pump therapy without CGM, on MDI+CGM, and on MDI+BGM. This difference was particularly shocking when comparing those on AID with those on pump+BGM or MDI+BGM; those on MDI+BGM and on pump+BGM were 88% and 71% less likely to achieve an A1c <7% compared to those on AID, a dramatic difference in outcomes. There was not a significant difference in the mean A1c and proportion achieving an A1c <7% between those on AID and those on PLGS systems. This trend of improved outcomes with greater automatic was true across all age groups, but was particularly stark among youth, teens, and young adults. For example, among youth, those on AID had an average self-reported A1c of 6.6% whereas those on MDI without CGM had an average self-reported A1c of 10.6%. Likewise, among youth, 73% of those on AID had an A1c <7% whereas only 15% of those on MDI without CGM had an A1c <7%.

  • Severe hypoglycemia was less common in those using AID compared to any other insulin delivery and sensing method. Overall, a fourth (25%) of participants reported a severe hypoglycemia event in the past year. Compared to those on AID, those on PLGS systems were 40% more likely to have ≥1 severe hypoglycemia event in the past year (p<0.001) and those on pump+CGM or MDI+CGM were 30% more likely to have experienced at least one several hypoglycemia event (p<0.001 for both). Even more stark, those using BGM with pump therapy or MDI were more than two times more likely to have experienced at least one severe hypoglycemia in the past year compared to those on AID (OR=2.1 and 2.3, respectively; both p<0.001). Concerningly, 50% of adults (ages 30+) not using CGM experienced at least one severe hypoglycemia event – this is just yet another reinforcement of the importance of CGM technology and AID technology in preventing life-threatening events.

Four-month pediatric and adolescent Control-IQ real-world study (n=59): participants see +2.2 hours/day Time in Range; only 5% of participants had Time in Range ≤70% at baseline; significant psychosocial improvements in parents and participants over course of study

Dr. Caroline Zuijdwijk (Children's Hospital of Eastern Ontario) presented data from a four-month real-world study of pediatric type 1s (n=59, ages 6-18) using Control-IQ, demonstrating significant psychosocial and glycemic improvements. Dr. Zuijdwijk explained that over 98% of participants used CGM prior to the study, with the vast majority of participants also using Basal-IQ predictive low glucose suspend technology. The mean pump duration of the sample was ~two years, and nearly 70% of participants reported having total household incomes of >$100,000 and also having a university or postgraduate degree. Thus, while we do find this data to be worthy of note, it’s important to understand that the study’s population is not representative of the vast majority of people with diabetes – including those who live in developed countries and have access to new diabetes technology, but also including the many people with diabetes in low-and-middle-income countries for whom diabetes technology is often out of reach.

  • Over the course of the study, Time in Range improved 2.2 hours/day (from ~55% to ~64%). The percentage of individuals achieving a Time in Range ≥70% improved from roughly 5% at baseline to roughly 18%. Approximately 14% of the cohort achieved a GMI <7%. While these improvements in glycemic control are certainly significant, we also are curious to better understand the barriers that prevented people from achieving a Time in Range >70% and a GMI <7% (over 14 days, 30 days, and 90 days).
  • Participants and their parents saw notable improvements across four psychosocial patient-reported outcome measures. Parents saw a significant improvement in hypoglycemia fear behavior and worry subscores (p<0.001 and p=0.01, respectively), as well as in INSPIRE scores (INsulin Dosing Systems: Perceptions, Ideas, Reflections, and Expectations), the latter of which reflects improved attitudes toward AID and its impact on their children (p<0.04). However, parents did not achieve a low enough INSPIRE score to be considered a “perceived benefit” relative to baseline expectations. Parents also reported an improvement on the WHO-5 wellbeing index (p=0.002). Meanwhile, youth reported improved scores on the DIDS device satisfaction (p<0.01) and diabetes impact (p<0.001) scales, as well as on the hypoglycemia fear worry subscale (p=0.003). From our view, it’s highly interesting to see that parents in this population saw improvements across more psychosocial outcome measures compared to their children.

Posters – AID, Insulin Pumps, and Smart Pens

Abstract Title

Authors

Details + Takeaways

Safety and Glucose Regulation by the Bionic Pancreas without Continuous Glucose Monitoring Input

C. Balliro

  • iLet continues to automate insulin delivery without CGM input based on manual BGM values and learned patterns from prior CGM use
  • 49.5-hour test of iLet without CGM input following iLet use in 13-week pivotal (n=54)
  • Participants entered BGM values every two hours while awake and once overnight (9-10x/day)
  • No severe hypoglycemia or DKA
  • Time in Range fell back to baseline value (52% at baseline to 69% in RCT to 54% with BGM-driven iLet)
  • No significant difference in hypoglycemia and hyperglycemia metrics based on blinded CGM

Glycemic Control Using Recommended Settings in youth and Adults with Type 1 Diabetes: MiniMed 780G System with Calibration Free Guardian 4 Sensor Results

R. Vigersky, B. Bode, R. Brazag, B. Buckingham, A. Carlson, K. Kaiserman, M. Kipnes, D. Liljenquist, A. Philis-Tsimikas, C. Pihoker, R. Pop-Busui, J. Reed, J. Sherr, D. Shulman, R. Slover, J. Thrasher, X. Chen, M. Lui, T. Cordero, M. Vella, A. Rhinehart, J. Shin

  • 176 participants transitioned from MiniMed 780G with Guardian Sensor 3 to MiniMed 780G with the Guardian 4 Sensor for three months
  • No severe hypoglycemic or DKA events
  • Time in Range was 75% overall (n=119), 73% in participants aged 7-17 years (n=65), and 77% in adult participants (n=54)
  • No clinically significant change in Time in Range during MiniMed 780G use with Guardian 4 vs. 780G with Guardian Sensor 3

Utility and Safety of backup Insulin Regiments generated by the Bionic Pancreas – A Randomized Study

Nelly Mauras

  • Two to four day transition phase following 13-week pivotal RCT trial of insulin-only iLet
  • Participants in the IO-iLet group were randomized 1:1 to pre-study insulin regimen (n=148) or iLet regimen (n=149)
  • MDI users in iLet guidance arm had 42% Time in Range vs. 44% Time in Range for those adhering to their pre-study insulin regimen
  • Insulin pump users using iLet guidance had 61% Time in Range compared to 55% Time in Range for those adhering to their pre-study insulin regimen
  • Participants who transitioned back to an AID system had 61% Time in Range in the iLet guidance group and 63% Time in Range for the pre-study insulin regimen group

Evaluating the Impact of Quality Initiatives on Improving Insulin Pen and Needles Prescriptions on Hospital Discharge in Patients with Type 2 Diabetes

Bobak Moazzami, Patricia Hwang, Shanza Ashraf, Amanda Van Prooyen, Hasan F. Shabbir, Maria Klimenko, Erika Brechtelsbauer, Ethan Molitch-Hou, Maureen A. Hinds, Mary Wald, Ram Jagannathan, Rodolfo J. Galindo

 

  • Retrospective analysis of adults discharged from Emory Hospitals on insulin pens (n=8097) with and without an associated pen needle prescription
  • After QI project, the proportion of patients without an associated prescription for pen needles decreased by 31% vs. pre-intervention
  • After the intervention, 90-day hospital readmission rates decreased by 1% per quarter (p<0.005); no significant change in 30-day hospital readmission rates

ADA Presidents’ Select Abstract: Impact of Overnight Blood Glucose on Next-Day Functioning in T1D

Elizabeth Pyatak, Donna Spruijt-Metz, Stefan Schneider, Jill P. Crandall, Anne L. Peters, Haomiao Jin, Shivani Agarwal, Loree T. Pham, Aina Ali, Claire J. Hoogendoorn, Gladys Crespo-Ramos, Heidy Mendez-Rodriguez, Pey-Jiuan Lee, Valerie F. Ruelas, Rose Basile, Jeffrey S. Gonzalez

  • Participants wore a blinded CGM and accelerometer for 10-14 days
  • During study, participants completed surveys and cognitive tasks 5-6 times per day
  • Dynamic structural equation model was used to evaluate the within-person impact of overnight blood glucose on the following day’s functioning, while controlling for the prior day’s functioning
  • Time <70 mg/dL, time >250 mg/dL, and %CV predicted changes in next day function
  • Time <70 mg/dL was associated with poor sustained attention; more time >250 mg/dL predicted more fatigue and lower step count (all p<0.05)
  • High %CV associated with poor sustained attention, more fatigue, and less engagement in demanding activities (all p<0.05)

A Randomized Controlled Trial to Alleviate Carbohydrate Counting in Type 1 Diabetes with Automated Fiasp and Pramlintide Closed-Loop Delivery

Elisa Cohen, Michael Tsoukas, Julia E. Von Oettingen, Jean-François Yale, Natasha Garfield, Michael Vallis, Nikita Gouchie-Provencher, Adnan Jafar, Milad Ghanbari, Emilie Palisaitis, Joanna Rutkowski, Laurent Legault, Ahmad Haidar

  • Adults (n=15) and adolescents (n=15) used (i) Fiasp and placebo with carbohydrate counting; (ii) Fiasp and pramlintide with meal announcement; and (iii) Fiasp and placebo with meal announcement for two weeks
  • Prior to first arm, participants had a one-week run-in with automated Fiasp (single pump) delivery and carbohydrate counting; mean Time in Range was 71% in adults and 64% in adolescents
  • In adults, mean Time in Range was 65% on Fiasp and placebo with carbohydrate counting, 71% on Fiasp and pramlintide with meal announcement, and 64% on Fiasp and placebo with meal announcement
  • In adolescents, mean Time in Range was 51% on Fiasp and placebo with carbohydrate counting, 55% on Fiasp and pramlintide with meal announcement, and 46% on Fiasp and placebo with meal announcement
  • Data suggest fiasp and pramlintide delivery may alleviate carbohydrate counting without degrading glucose control

A Randomized Crossover Trial to Compare Automated Insulin Delivery with Carbohydrate Counting or Simplified Qualitative Meal-Size Estimation in Type 1 Diabetes

Ahmad Haidar, Laurent Legault, Marie Raffray, Nikita Gouchie-Provencher, Adnan Jafar, Marie Devaux, Milad Ghanbari, Rémi Rabasa-Lhoret

  • Three week randomized non-inferiority crossover trial (n=30) of “iPancreas” system with OHSU dosing algorithm with carbohydrate counting vs. “iPancreas” with qualitative meal-size estimation (low, medium, high, or very high carbohydrate)
  • Low, medium, high, and very high carbohydrate (CHO) meals were defined as <30g CHO, 30g – 60g CHO, 60g – 90g CHO, and >90 g CHO respectively
  • Non-inferiority margin was set to 4% Time in Range
  • Qualitative meal-size estimation Time in Range was 3.6% lower than full carbohydrate counting (74%), so non-inferiority of qualitative estimation was not confirmed

Long-Term Glycemic Control in Adult Participants Using Control-IQ Technology: Real-World Evidence

Rishi Graham, Harsimran Singh, Gabriel Alencar, Lars Mueller, Michelle L. Manning, Kirstin N. White, Alex Wheatcroft, Karen Carmelo, Eliah S. Aronoffspencer, Stephanie Habif, Jordan E. Pinsker

  • Patients wore t:slim X2 insulin pump with Control-IQ (n=1,107) for 12 months
  • Majority of participants were former pumpers (n=882) vs. MDI (n=225) before starting Control-IQ
  • Median GMI at 12 months was significantly lower compared to baseline A1c (7.0% vs. 7.2%, respectively) for all users (p<0.0001)
  • Participants aged 18-30 (GMI of 7.2% and 6.9% for former pumpers and MDI, respectively) and 46-64 (GMI of 7.0%) achieved greatest improvements from baseline (p<0.0001)
  • Older adults (>65 years old) showed the lowest GMI at study end (6.9%)

Comparing Dual-Hormone and Single-Hormone Automated Insulin Delivery System on Nocturnal Glucose Management among Pediatric People Living with Type 1 Diabetes: A Pooled Analysis

Zekai Wu, Virginie Messier, Maha Lebbar, Rémi Rabasa-Lhoret

  • Analysis of pooled data from 3 open-label, randomized, controlled, crossover studies comparing dual-hormone and single-hormone AID in pediatric patients (8-17 years old) with type 1 diabetes
  • Glucose management observed overnight (midnight – 6 am) for 246 nights (123 nights for single-hormone AID and 123 nights for dual-hormone AID)
  • Time in Range during single-hormone treatment (76.0%) was lower vs. dual-hormone treatment (83.3%; p<0.05)
  • Time Below Range was 4% and 2% for single-hormone and dual-hormone AID, respectively (p<0.05)
  • Time in severe hyperglycemia (>250 mg/dL) was 8.5% and 3.0% for single-hormone and dual-hormone AID, respectively (p<0.05)

Real-World use of the MiniMed 780G Advanced Hybrid Closed-Loop (AHCL) System with the Guardian Sensor 3

Robert A. Vigersky, Javier Castañeda, Arcelia Arrieta, Toni L. Cordero, Andrew S. Rhinehart, John Shin

  • Glycemic outcomes from real-world users of MiniMed 780G with Guardian Sensor 3 and Guardian 4 Sensor (n=3.213)
  • Time in Range for GS3 (74%) vs. G4S (73%) during MiniMed 780G use was comparable
  • Number of daily self-monitoring blood glucose measurements with the Guardian 4 Sensor (0.7) was lower than with Guardian Sensor 3 (3.0; p<0.001)
  • Number of closed-loop exists was lower with Guardian 4 Sensor (0.13) compared to Guardian Sensor 3 (0.16; p<0.001)

A Random-Order, Double-Blinded, Placebo-Controlled Study of the Bionic Pancreas in the Bihormonal vs. the Insulin-Only Configuration Using Two Different Glucose Targets in Adults with type 1 Diabetes in a Standardized exercise Setting

Luz E. Castellanos, Courtney A. Balliro, Jordan Sherwood, Mallory Hillard, Rajendranath Selagamsetty, Hui Zheng, Firas El-Khatib, Edward Damiano, Steven J. Russell

  • Adults type 1s (n=20) wore either the insulin-only or bihormonal iLet system with 2 glucose targets (110 and 130 mg/dL) for three days at home
  • On the last morning of each arm, fasted subjects exercised on a stationary bike for ~30 minutes
  • For the 130 mg/dL glucose target, there were no hypoglycemic events (plasma glucose <60 mg/dL)
  • For the 110 mg/dL glucose target, three insulin-only iLet subjects had hypoglycemia; no bihormonal configuration subjects had hypoglycemia (p=0.25)
  • No carbohydrates were required for hypoglycemia with 110 mg/dL or 130 mg/dL targets for both insulin-only and bihormonal configurations

Glycemic Outcomes across Total Daily Insulin Doses with the Omnipod 5 Automated Insulin Delivery System (AID) among People with Type 1 Diabetes (T1D) Ages 2 to 70 Years

Melissa Schoelwer, Bruce W. Bode, Anders L. Carlson, Amy B. Criego, Gregory P. Forlenza, Ruth S. Weinstock, David W. Hansen, Bruce A. Buckingham, Sanjeev N. Mehta, Lori M. Laffel, Jennifer Sherr, Carol J. Levy, Irl B. Hirsch, Sarah A. Macleish, Daniel Desalvo, Viral Shah, Anuj Bhargava, Thomas C. Jones, Grazia Aleppo, Rachel E. Gurlin, Trang T. Ly, Omnipod 5 Research Group

  • Pivotal study of patients with type 1 diabetes wearing the Omnipod 5 system for three months
  • Main study recruited adolescents and adults (ages 6-70) with type 1 diabetes for >6 months (n=241)
  • Patients ages 2-6 were a part of the preschool study (n=80)
  • No association was found between total daily dose (TDD) during standard therapy (ST) and Time in Range during AID (p=0.29)
  • Time in Range improved between 1.7 hrs/day and 3.2 hrs/day (p<0.05) for all six cohorts, split by TDD during ST

Glycemic Outcomes with the Omnipod 5 Automated Insulin Delivery System (AID) Stratified by Baseline Hypoglycemia Risk among People with Type 1 Diabetes (T1D) Ages 2 to 70 Years

Gregory P. Forlenza, Bruce W. Bode, Anders L. Carlson, Amy B. Criego, Sue A. Brown, Ruth S. Weinstock, David W. Hansen, Bruce A. Buckingham, Sanjeev N. Mehta, Lori M. Laffel, Jennifer Sherr, Carol J. Levy, Irl B. Hirsch, Sarah A. Macleish, Daniel Desalvo, Viral Shah, Anuj Bhargava, Thomas C. Jones, Grazia Aleppo, Lauren M. Huyett, Trang T. Ly, Omnipod 5 Research Group

  • Three month use of Omnipod 5 AID system in main pivotal study (n=241) and preschool pivotal study (n=80)
  • Participants sorted into three groups based on baseline hypoglycemia risk with standard therapy: (i) Time Below Range £1% (n=113); (ii) Time Below Range >1% to £4% (n=132); and (iii) Time Below Range ³4% (n=75)
  • Participants in subgroup aged 6-<14 had lower hypoglycemia risk vs. those in 2-<6 year old and 14-70 year old cohorts (41%  of participants aged 6-<14 were in £1% Time Below Range group and 44% were in>1% to £4% Time Below Range group at baseline)
  • Those with the greatest hypoglycemia risk during standard therapy (Time Below Range ³4%) had the greatest reduction in Time Below Range on AID from 6% to 3% (p<0.05); Time in Range increased 2.2 hrs/day from 65% to 74% (p<0.05)
  • Those with lowest hypoglycemia risk during standard therapy (Time Below Range £1%) had the greatest increase in Time in Range (+4.1 hrs/day from 51% to 68%; p<0.05)

Changes in Ambulatory Glucose Profile (AGP) in Patients with Type 1 Diabetes Mellitus after Switching from Sensor-Augmented Insulin Pump Therapy to a Do-it-Yourself Artificial Pancrease System: A Retrospective Data Analysis of Real-World Data

Ichael Mueller-Korbsch, Lisa Fruehwald, Antonia-Therese Kietaibl

  • Retrospective data analysis of AGP from type 1s on Dexcom G6 who switched from SAP therapy to a DIY AID system (n=25)
  • Participants underwent four weeks on SAP before switching to DIY AID for four weeks
  • Time in Range increased by 2.4 hrs/day from 74% at baseline to 84%; no change in hypoglycemia frequency
  • Time Above Range fell 2.4 hrs/day from 21% to 11% (p=0.001); time in severe hypoglycemia fell slightly from 5% to 2%
  • Users of open-source AID systems with automatic correction boluses and basal rate adjustments spent +4.6 hrs/day, on average, in Range vs. people on systems only modifying basal rate
  • Time in severe hypoglycemia was cut in half when using systems with autoboluses and basal rate adjustment vs. basal rate adjustment only (0.8% vs. 0.4%)

Treatment Satisfaction and Fear of Hypoglycemia in the ADAPT Study: A Six-Month, Randomized Controlled Trial Comparing an Advanced Hybrid Closed-Loop System to Multiple Daily Injections (MDI) with Intermittently Scanned CGM (isCGM)

Simona De Portu, Pratik Choudhary, Jens Kroeger, Charles Thivolet, Mark Evans, Roseline Ré, Linda Vorrink De Groot, John Shin, Aklilu Habteab Ghebretinsae, Javier Castañeda, Ohad Cohen, Adapt Study Group

  • Comparison of treatment satisfaction and fear of hyperglycemia in adult type 1s on MiniMed 780G vs. MDI users on is-CGM in the ADAPT study
  • A1c decreased an additional 1.4% in the AID arm compared to the MDI+is-CGM cohort (p<0.001); no difference in hypoglycemia observed
  • Mean Diabetes Treatment Satisfaction Questionnaire score increased after six months of AID (30) compared to MDI+is-CGM (22; p=0.0003)
  • Perceived frequency of hypoglycemia decreased by 10 points and 3 points from baseline in the AID and MDI+is-CGM group, respectively

Combining Meal-Bolus Reduction and Exercise Announcement Eliminates the Risk of Hypoglycemia during Postprandial Exercise in people Living with Type 1 Diabetes Treated with Automated Insulin Delivery Systems

Marie Raffray, Rémi Rabasa-Lhoret, Étienne Myette-Côté, Joséphine Molveau, Marie Devaux, Zekai Wu

  • Participants were type 1s on AID for 42 days (n=13)
  • Participants performed two 60-min exercise sessions; one 60-min post meal (60EX) and one 120-min post-meal (120EX)
  • Standardized meal was given at 8 am with a 33% insulin bolus reduction and exercise was announced to the AID algorithm through the end of the exercise
  • Time in Range during exercise was 61% and 47% for the 60EX and 120EX groups, respectively
  • Time Below Range during exercise was 0.4% and 0.0% for the 60EX and 120EX groups, respectively

Impact of Six Months of Hybrid Closed-Loop Use on Sleep in Youth with Type 1 Diabetes and Their Parents

Erin C. Cobry, Angela J. Karami, Timothy B. Vigers, Laura Pyle, Emily Jost, Lisa J. Meltzer, R. Paul Wadwa

  • Pediatric type 1s (aged 2-17) on Control-IQ (n=39) were enrolled with parents (n=39) in an observational study of sleep outcomes after AID initiation
  • PSQI improved to 5.4 from 6.3 points at baseline after three months (p<0.05); a PSQI score >5 indicates poor sleep
  • Pediatric Sleep Disturbance and Sleep Related Impairment (PROMIS) scores improved from 55.0 for sleep disturbance to 53.2 and 53.5 at three months and 6 months, respectively; PROMIS sleep-related impairment scores improved from 51.3 baseline to 50.1 at both three months and six months

Effect of Optimized Settings during Six Months of Real-World MiniMed 780G Advanced Hybrid Closed-Loop (AHCL) Therapy use in Chile

Matias Castro, Camila Nicole Carrasco Bonilla, Joana C. Leon Macedo, Maria Fernanda Gutierrez, Zeudi C. Valera Lopez, Andrea Daghero

  • Real-world glycemic outcomes of individuals living in Chile after six months of MiniMed 780G use (n=37)
  • Sensor glucose data collected 14 days pre-780G and 180 days post-780G initiation
  • Time in Range increased +1 hour/day, from 74.1% to 78.4% (p<0.01)
  • Users with active insulin time setting of 2-hours achieved the highest Time in Range (80%) and lowest Time Below Range (2.6%)

Progressive Accelerations of Insulin Exposure over Seven Days of Infusion Set Wear

Timothy S. Bailey, Jasmin R. Kastner, Poul Strange, Lei Shi, Keith A. Oberg, Jeffrey I. Joseph, Paul J. Strasma, Douglas B. Muchmore

  • Capillary Bio extended-wear infusion set (CBX) was worn by adult type 1s (n=7) for seven days
  • Study assessed insulin pharmacokinetics over one week of wear time of CMX versus control (teflon angled infusion set)
  • Insulin exposure accelerated as infusion set wear time extends; no meaningful differences between CBX and control infusion set performance

Comparison of Six Months Experience on Hybrid Closed Loop MiniMed 670G System and Advanced Hybrid Closed Loop MiniMed 780G System Using a Structured Initiation Protocol in Children and Adolescents with Type 1 Diabetes

Goran Petrovski, Judith Campbell, Fawziya Alkhalaf, Khalid Hussain

  • Children and adolescent type 1s on MDI started AID – either MiniMed 670G (n=30) or MiniMed 780G (n=34) – using the same structured protocol for initiation
  • Time in Range increased from 47% at baseline to 72%, 74%, and 76% at month one, two, and three, respectively, for MiniMed 670G users
  • Time in Range increased from 42% baseline to 75%, 80%, and 80% at month one, two, and three, respectively for MiniMed 780G users
  • Six months post initiation, the number of exits from closed loop mode was 3.8/week for MiniMed 670G and 0.6/week for MiniMed 780G, potentially due to fingerstick requirement for Guardian Sensor 3

Evaluating the Feasibility of Incorporating Nonglucose Signals in Automated Insulin delivery System

Ayan Banerjee, Ravinder Jeet Kaur, Isabella Zaniletti, Mei Mei Church, Shelly K. Mccrady-Spitzer, Donna Desjardins, Sandeep Gupta, Yogish C. Kudva

  • JDRF and HCT-funded study of electrodermal activity as non-glucose signal for AID algorithm
  • Electrodermal activity power spectral density (EDA PSD) found to be a good indicator for pharmacological stress
  • Positive change in EDA PSD was associated with salivary cortisol (p=0.015)
  • Positive change in EDA PSD correlated with hydrocortisone dosage; notable because current AID systems achieve suboptimal glucose control when hydrocortisone is used

Evaluation of Long-Term Glycemic Outcomes by Ethnicity in Adults with Type 1 Diabetes Using Control-IQ

Rishi Graham, Harsimran Singh, Lars Mueller, Michelle L. Manning, Gabriel Alencar, Kirstin N. White, Alex Wheatcroft, Karen Carmelo, Ravid Katchalski, Eliah S. Aronoffspencer, Jordan E. Pinsker, Stephanie Habif

  • Single-arm, longitudinal study evaluating real-world use of the t:slim X2 with Control-IQ in American type 1s over 12 months (n=1,045)
  • Study enrolled White (n=921), Latinx (n=74), Black (n=34), and Asian (n=16) adults
  • Black adults reported highest median baseline A1c (8%) followed by Latino adults (7.3%)
  • GMI improved at 12 months for all participants; GMI reduction in Black, Latind, White, and Asian participants was 0.7%, 0.2%, 0.2%, and 0.1%, respectively
  • Time in Range after starting Control-IQ was 73%, 70%, 65%, and 72% for White, Latinx, Black, and Asian participants, respectively

Real-World Use of Do-It-Yourself Artificial Pancreas Systems in Adults with Type 1 Diabetes

Sandra Amuedo, María Antequera, Sharona Azriel

  • Retrospective analysis of glycemic outcomes using DIY AID systems in adults with type 1 diabetes
  • Participants used either AndroidAPS with Accu-Chek Insight insulin pump/Dexcom G6 or FreeStyle Libre 2 CGM; or, participants used Loop with MiniMed Paradigm and Dexcom G6
  • Time in Range after 12 months of using DIY AID systems was 88% compared to 70% at baseline (difference of 4.3 hrs/day; p<0.001)

Use of the inControl Algorithm for Automated Insulin Delivery during Childbirth: A Case Series

Lois E. Donovan, Julie Mckeen, Denice Feig

  • Five women with type 1 diabetes using the inControl algorithm continued using AID during childbirth and postpartum
  • Only 2 of the 5 participants needed treatment for neonatal hypoglycemia; both cases had one episode corrected with feeding
  • Precautionary reductions in insulin dosing made prior to childbirth were associated with avoidance of hypoglycemia during childbirth and in the early postpartum period
  • Mild hyperglycemia resulted in some cases at childbirth, likely resulting from overly cautious reductions in insulin dosing

Digital Health, Telemedicine, and Decision Support

Five-year Virta outcomes (n=122): Participants achieve 8% weight loss (257 lbs to 237 lbs) and 0.3% A1c improvement (7.5% to 7.2%); significant medication deprescription even though cohort A1c falls above ADA target; 20% achieve “diabetes remission"; additional hepatic, renal, cardiovascular, and inflammatory benefits

In the ADA 2022 poster hall (832-P), Virta Health presented five-year results from participants in its very low carbohydrate intervention including nutritional ketosis (VLCI) that is delivered via continuous remote care (CRC). As a reminder, Virta published two-year outcomes data in 2019, showing that 7% of the cohort had achieved “complete diabetes remission” (A1c<5.7% without medication), 18% achieved “partial remission” (A1c<6.5% without medication), and 54% achieved “diabetes reversal” (A1c <6.5% and no medications apart from metformin). In the five-year study, a subset of patients with type 2 diabetes originally enrolled in the two-year non-randomized, controlled trial continued to receive CRC emphasizing VLCI as part of a three prospective follow-up period. Specifically, 169 of the 200 patients (85%) in the two-year trial proceeded to the five-year trial, with 122 patients (72%) completing the full five years. Similar to the two-year study, the investigators assessed changes in weight and glycemia from baseline, while also analyzing medication changes among prescribed users by medication class.

  • At five years, users achieved an 8% reduction in weight and a 0.3% reduction in A1c. Specifically, participants’ weight decreased from a mean 116 kg (~257 lbs) at baseline to 108 kg (~237 lbs) at five years (p<0.001). Additionally, A1c improved from 7.5% at baseline to 7.2% at five years (p<0.05). As a point of comparison, during Virta’s two-year analysis of VLCI, A1c improved by 0.9% (from 7.7% to 6.8%) and weight improved by 12 kg (from 115 kg [254 lbs] to 103 kg [227 lbs]). While the weight loss and A1c improvements are laudable, the final A1c at five years was above consensus target goals of <7% - while some expressed curiosity about the sustainability of Virta’s VLCI’s two-year outcomes as a population approach, certainly we can see that the strength of the model with the Virta Health Clinic has helped some populations enormously.
  • The percent of Virta users prescribed a diabetes medication significantly dropped at five years from 85% at baseline to 71%. Specifically, the percentage of participants taking sulfonylureas decreased from 27% at baseline to 5% at five years, along with those taking insulin (26% to 13%) and SGLT-2s (11% to 3%). It’s worth noting that despite less medication use, A1c still impressively managed to improve over time. However, we’re curious to understand the rationale behind the massive deprescription of SGLT-2s, especially given that this medication class has been shown to be reno- and cardioprotective. Additionally, with the A1c at five years still above 7%, we also question whether insulin deprescription was warranted in this cohort, but we’d need to see more granular data to fully understand the rationale behind the decrease in insulin prescriptions. We are indeed gratified to see that Virta’s coaches focused on prescribing GLP-1, as shown in the graph below. Overall, the total number of diabetes medications prescribed from baseline to year five as reduced by 47%, and excluding metformin, 60% were eliminated (we imagine this was hugely driven by an elimination of sulfonylureas and DPP-4s, along with SGLT-2s and insulin to a lesser degree).

  • In a separate poster (1176-P), Virta shared that 20% of the individuals completing its five-year study achieved diabetes remission (A1c<6.5%, no meds for ≥3 months). The percentage of individuals meeting ADA (A1c <7%) and HEDIS (A1c <8%) targets remained constant over the study, but the percentage of individuals meeting AACE targets (A1c <6.5%) and Virta targets (A1c <6.5%, no anti-diabetes meds or metformin only) improved significantly from 23% to 42% and from 12% to 33%, respectively (both p<0.001).

  • During an oral presentation session on Day #3, Dr. Caroline Roberts (Virta Health) also presented additional renal, cardiovascular, inflammatory, and hepatic outcomes from Virta’s five-year cohort (shown below). Overall, Dr. Roberts emphasized that the biomarker data demonstrate the long-term safety of Virta’s VLCI for type 2s. Virta’s VLCI led to improvements in lipid and inflammatory markers despite the fact that it involves a high fat diet, which Dr. Roberts found to be noteworthy since some believe that long-term ketosis could increase cardiovascular risk. Additionally, Dr. Roberts highlighted the (relatively) stable renal function of the cohort, as evidenced by no change in eGFR, with possible improvements in people with stage three CKD (additional studies would need to be conducted to confirm this hypothesis). Dr. Roberts explained that the stability of eGFR during Virta’s VLCI given that the ADA’s Standards of Medical Care caution against high fat diets for those with a high renal disease risk.

First data out of UnitedHealth Group’s Level2 type 2 management program: Those with CGM data at follow-up see -0.8% A1c/GMI reduction from 7.7% to 7.0%; 70% of those with baseline A1c values 7%-9% and 94% of those with baseline A1c values >9% see ≥0.5% improvement

The poster hall offered a first look at the outcomes achieved with UnitedHealth Group’s Level2 type 2 diabetes management program, overall displaying strong engagement and glycemic results among those who participated and wore CGM (695-P). As a reminder, Level2 is UnitedHealth Group’s type 2 diabetes remission program that launched in July 2020 following a successful pilot. The intervention provides eligible members a Dexcom G6 continuous glucose monitor, a Fitbit activity tracker, smartphone app-based alerts, personalized clinical coaching, and virtual specialist consultations. With 230,000 employer-sponsored UnitedHealthcare members with type 2 diabetes as eligible to enroll, we believe Level2 is the largest type 2 diabetes program offering CGM to-date, which makes this data readout particularly significant for the broader digital coaching arena.

This retrospective assessment of engagement and glycemic outcomes over 26 weeks after Level2 enrollment included 7,886 participants who joined Level2 in January-July 2021 (average age 55). Based on medical and pharmacy claims data, most (85%) had been on one or two diabetes therapies over the year prior to enrolling in Level2, 18% had filled prescriptions for three diabetes medications, and 7% had filled prescriptions for four or more diabetes medications. About half were on metformin, 28% were on a GLP-1, 22% were on SGLT-2s, 22% were basal insulins, 11% were on rapid-acting insulin, 21% were on a sulfonylurea, 10% were on DPP-4s, and 5% were on TZD. Participants averaged a Diabetes Complications and Severity Index (measure of existing and at-risk diabetes-related complications; scale 0-2) of 0.7, suggests some but not severe complications.

  • Engagement was high in the 26 weeks following Level2 initiation. Over the 26 weeks following enrollment, participants averaged 175 days of enrollment and 11 coaching interactions. Over two-thirds had at least one coaching interaction over 26 weeks with 60% having more than four interactions, suggesting that participants were highly engaged in the program. Two-thirds (n=5,886) wore a CGM at least once. Characteristics associated with coaching visit frequency and CGM wear included a younger age and nephropathy-related claims history.
  • Of the two-thirds of participants who wore CGM at least once, 7% had baseline A1c data and sufficient CGM data at week 12-13 and 25-26 (n=378). These participants were included in a subanalysis evaluating change in glycemic control using baseline A1c and follow-up GMI, which is a limitation of the study given that A1c and GMI are correlated, but not equivalent, which likely impacts the results. Overall, GMI/A1c improved by an adjusted -0.8% in these CGM wearers, falling from 7.7% at baseline to 7% at 26 weeks. Those with baseline A1c values <7% (n=125) saw an increase in A1c/GMI (+0.3% from 6.3% to 6.6%), potentially due to a reduction in hypoglycemia; those with baseline A1c values of 7%-9% (n=184) saw a nonsignificant trend toward a -0.3% A1c/GMI improvement; and those with baseline A1c values ≥9% (n=69) saw a whopping 2.6% A1c/GMI improvement from 10.1% to 7.5%. Just over half (55%) of these participants saw a significant ≥0.5% A1c/GMI improvement. This increased to 70% among those with baseline A1c values 7%-9% and to 94% among those with baseline A1c values >9%.

 

Baseline A1c

Final GMI at week 25-26

A1c change (* for p<0.05)

Percentage of participants who saw >0.5% reduction

All included in subanalysis (n=378)

7.7%

7.0%

-0.8% *

54%

Baseline A1c <7% (n=125)

6.3%

6.6%

+0.3% *

9%

Baseline A1c 7%-9% (n=184)

7.8%

7.0%

-0.8%

70%

Baseline A1c >9% (n=69)

10.1%

7.5%

-2.6%

94%

  • As noted above, this is the first data we’ve seen from Level2 since it launched in July 2020. In the announcement of that launch, UnitedHealth shared some tidbits from its pilot trial (n=790), which are in line with the results shared in this ADA 2022 poster. Specifically, UHG’s press announcement for the full launch shared that in the pilot, “certain” participants achieved a clinically meaningful reduction in A1c, and those with baseline A1c values >8% saw on average a >1% reduction. Though not quantified, “some Level2 participants” achieved type 2 diabetes “remissions” (A1c <7% and no longer required medication). In the pilot, Level2 eliminated the need for more than 450 prescriptions (~0.6 medications/participant).

Glooko mobile app and remote patient monitoring associated with 0.5% improvement in A1c compared t0 traditional care; largest A1c improvements in users with baseline A1c ≥8.5%

Dr. Mark Clements (Chief Medical Officer, Glooko) presented exciting data on the role of Glooko’s digital health platforms on diabetes outcomes. Specifically, Dr. Clements discussed new data demonstrating improvements in glycemic outcomes following the use of Glooko’s mobile health application. While presenting this data, Dr. Clements stressed that the ability of digital technologies to enable data sharing across platforms and between patients and providers can be a strong strategy to improve outcomes for patients and reduce provider burnout and stress levels.

  • Glooko’s mobile-enabled remote patient monitoring (RPM) with coaching (n=80) was associated with 0.5% lower A1cs with an average of 7.8% at three months compared to an average A1c of 8.3% among patients receiving traditional care (n=82). Specifically, in this investigation, adults with type 2 diabetes on any type of glucose monitoring were randomly placed into either a mobile application-enabled RPM group or a control group receiving traditional care without RPM. In this presentation, Dr. Clements outlined interim three-month data read and the fully study will continue for another twelve weeks. In the study, the intervention group was remotely monitored via Glooko’s patient facing app on a weekly basis for glycemic and SMBG indicators. These patients received telephonic coaching on diet, exercise, & medication as needed through the mobile application. Content provided to patients was determined by glycemic and engagement triggers. The control group performed basic self-management without coaching or monitoring. Among the RPM and traditional care arms, average baseline A1c were 8.6% and 8.7%, respectively. After three months, patients in the RPM arm achieved lower A1cs, on average, at 7.8%, compared to the traditional care arm with an average A1c of 8.3% (p=0.013). Among the subgroup of patients with a baseline A1c greater than 8.5% even larger improvements in glycemic outcomes were seen. Specifically, among this cohort in the RPM arm, patients had a baseline A1c of 9.5% which was reduced to 8.2% after three months (p< 0.005). Patients in this cohort in the traditional care arm also saw improvements in glycemic outcomes from 9.4% at baseline to 9% at three weeks (p=0.05). Dr. Clements did note that the clinical implementation of these results remains to be determined, including identifying the optimal “dose” of virtual coaching to help patients achieve improved outcomes.
  • Dr. Clements also presented data on the correlation between patient disengagement from Glooko’s patient-facing mobile application and declines in glucose management. Dr. Clements explained that despite downloading and utilizing an application, not all mobile health application users remain engaged. This study (n=472) examined whether a drop off in app use was linked to poorer diabetes outcomes by comparing glucose-related outcomes before and after a people with diabetes discontinued use of Glooko’s patient-facing mobile app. Data from prior Glooko app users who had stopped interacting with the app, but subsequently uploaded device data into Glooko at a clinic visit after at least six weeks of inactivity were included in this analysis. Following app discontinuation, patients’ readings in range decreased by 1.7% over a 12-week period (p=0.03; baseline and 12-week values not provided). Over the same 12 weeks, the percent of readings above range for patients who discontinued app use increased by 3% (p=0.02). Dr. Clements noted that this data indicates disengagement with a mobile health application is predicative of worsening diabetes management. Glooko is working to integrate predictive capabilities into its patient-facing application, to help clinicians identify patients who may need encouragement to reengage. Dr. Clements expressed his hope that this reengagement will lead to improvements in glycemic management among patients who may otherwise be a risk for disengagement and subsequent declines in glycemic outcomes. Dr. Clements seemed excited by the potential opportunity to leverage this type of data to predict other factors for clinicians, such as disengagement from CGMs.
  • During the same session, Dr. Carla Demeterco-Berggren (Rady Children’s Hospital) presented a case study on the implementation of Glooko’s RPM platform. Dr. Demeterco-Berggren stressed the ability of Glooko’s services to decrease the workload of providers at her diabetes center. Dr. Demeterco-Berggren’s diabetes center implemented Glooko during the pandemic, when the switch to telehealth placed increasing burdens on the transfer of CGM data from patients to providers. Dr. Demeterco-Berggren emphasized that the platform improved remote patient monitoring, standardized the downloading of data, decreased the mislabeling of reports, and overall, increased the efficiency of providers who were then able to spend more time with patients interpreting and acting on device data. Her hospital was able to use Glooko’s integration products, which allowed for easier access to Glooko data reports in patient EHRs.

Virta in Veterans (n=254): 7% sustained weight reduction (from 241 lbs to 223 lbs) and 0.6% A1c reduction (from 8.1% to 7.5%) at two years (both p<0.0001)

In the ADA 2022 poster hall (834-P), Virta Health presented two-year results from Veterans (n=254) participating in its very low carbohydrate intervention including nutritional ketosis (VLCI) that is delivered via continuous remote care (CRC). We’re very moved to see Virta’s continued focus on Veterans, dating back to the company’s 90-day pilot study from Veterans’ Day 2019 showing that 84% of veterans achieved an A1c <6.5% or reduced A1c by at least 1% after 90-days with an average 5% weight loss. Indeed, veterans constitute a large segment of people with diabetes. In this study, 425 veterans were initially enrolled, and 59% were retained at two years. Similar to other studies from Virta, the investigators assessed changes in weight and glycemia from baseline, while also analyzing medication changes among prescribed users by medication class. At baseline participants: (i) had a mean age of 58; (ii) were 12% female and 62% non-Hispanic White; (iii) had a mean A1c of 8.1%; (iv) had a mean BMI of 35 kg/m2; and (v) were taking a mean of 1.6 medication classes excluding metformin.

  • At two years, users achieved an 7% sustained reduction in weight and a 0.6% reduction in A1c. Specifically, participants’ weight decreased from a mean 241 lbs at baseline to 223 lbs at two years (p<0.0001). Additionally, A1c improved from 8.1% at baseline to 7.5% at two years (p<0.0001). As a point of comparison, during Virta’s two-year analysis of VLCI in a general cohort, A1c improved by 0.9% (from 7.7% to 6.8%) and weight improved by 12 kg (from 115 kg [254 lbs] to 103 kg [227 lbs]). While the weight loss and A1c improvements are laudable, the final A1c at five years was above consensus target goals of <7%.
  • The percent of Veterans prescribed a diabetes medication beyond metformin after using Virta’s VLCI dropped by 41%, from 100% at baseline to 59% at two years. Virta’s VLCI enabled medication deprescription of diabetes medication across all drug classes. Specifically, after two years, 42% fewer participants required insulin. We’re curious to understand the rationale behind the massive deprescription of SGLT-2s, especially given that this medication class has been shown to be highly renal- and cardioprotective. Additionally, with the A1c at two years still above 7% for some of the cohort, we also wonder about the degree to which insulin de-prescription was warranted in everyone in this cohort, but we’d need to see more granular data to fully understand the rationale behind the decrease in insulin prescriptions. Certainly the weight gain and hypoglycemia typically associated with any kind of insulin use is negative by virtually any definition.

  • In a separate poster (932-P), Virta shared that 7% of the Veterans completing its two-year study achieved diabetes remission (A1c<6.5%, no meds for ≥3 months). The percentage of individuals meeting ADA (A1c <7%), AACE (A1c ≤6.5%) and HEDIS (A1c <8%) targets remained constant over the study, but the percentage of individuals meeting Virta targets (A1c <6.5%, no anti-diabetes meds or metformin only) improved from 0 individuals at baseline to 26 individuals at two years.
  • During an oral presentation session, Virta’s Dr. Michelle VanTieghem shared data showing that Virta’s VLCI improved several renal, hepatic, and cardiovascular biomarkers among Veterans at two years. These data are shown in the picture below. Interestingly, Dr. VanTieghem did note that Virta’s study had a 54% retention rate, which some audience members proposed may have been due to the strict nature of Virta’s diet (baseline of <30 g of carbs/day). However, during a separate correspondence with Virta, we learned that many other experts in the diabetes industry characterized this performance as “very good,” and Virta told us separately that the individuals leaving their program were doing “remarkably well” at the time of exit.

ADA/EASD Precision Medicine Lecture: Dr. David Kerr argues digital health can meaningfully advance precision medicine through digital phenotyping, while cautioning that broadband internet access remains a “super” social determinant of health

During an ADA/EASD session on precision medicine, Dr. David Kerr (Sansum Diabetes Research Institute) argued that digital health “has a great deal to offer precision medicine.” Dr. Kerr started by exploring what he called the “two Achilles’ heels of precision medicine and digital health”: (i) precision medicine is often concerned about adapting one’s approach to healthcare based on quantifiable elements, which can lead one to overlook parameters that are not yet quantifiable; and (ii) digital health presumes widespread connectivity to the internet, which, Dr. Kerr emphatically stressed, is not ubiquitous even in 2022. Dr. Kerr then argued that the contributions of “nature” and “nurture” to one’s health outcomes make diabetes a uniquely difficult problem in precision medicine: (i) on nature, precision diagnoses can be driven by subtypes using ones genomics, metabolomics, epidemiology, ancestry, geography, and clinical features, which can power precision prediction via subgroup analysis, and ultimately power personalized therapeutics – matching the right therapy to the right person at the right time; and (ii) on nurture, Dr. Kerr argued that health care delivery dictates 10% of outcomes, environment contributes 5%, genetics 30%, and social determinants of health (SDOH) a whopping 55%. Putting it all together, Dr. Kerr remarked that “diabetes is a wicked problem” where one must balance economic interests from multiple stakeholders, which disproportionately impact groups in line with health inequities and leads to “ripple” effects whereby certain realities might not seem detrimental in the moment, but over time can lead to disastrous (or, on the flip side, beneficial) outcomes. As an example, Dr. Kerr motioned to the COVID-19 pandemic, in which minorities have significantly worse health outcomes as the global health emergency has continued to progress.

  • Dr. Kerr was fascinated with the idea of using “phenotypes” to enable precision medicine. Specifically, Dr. Kerr was fascinated with using various glucose metrics (e.g., A1c, Time in Range, or even the novel Glycemic Risk Index) along with data from lifestyle (e.g., exercise, diet, sleep, stress, mood) to paint a holistic, integrated picture of someone (i.e., their “phenotype”). As a conceptual model, Dr. Kerr noted that phenotyping would need to incorporate one’s social phenotype as well – that is, some encapsulation of social determinants of health. Dr. Kerr highlighted two studies that have begun chipping away at the clinical utility of phenotypes: (i) Barua et al., 2022: investigators showed that across three distinct groups (“at risk” for diabetes, pre-diabetes, and type 2 diabetes), there are three distinct average glucose spike patterns after breakfast; and (ii) in unpublished data, Barua et al. show that across the three aforementioned groups, the “dawn phenomenon” (i.e., insufficient insulin for morning endogenous cortisol) becomes more prominent down the line. Essentially, Dr. Kerr believes that by creating a comprehensive phenotype, informed by several of these “glucotype” or other lifestyle parameters, one could create meaningful cohorts of individuals to inform treatment decisions. Dr. Kerr also touched on machine learning and just-in-time adaptive interventions as additional tools to drive personalized self-management interventions and precision engagement, respectively.

  • Dr. Kerr cautioned that digital inclusion remains a “super” social determinant of health. Pointing to his own paper (entitled “A persisting parallel universe in diabetes care within America’s capital”) in the Lancet eClinicalMedicine, Dr. Kerr highlighted how areas with higher concentrations of Black individuals have not only lower access to supermarkets and higher rates of poverty, but also, significantly lower access to broadband internet. Dr. Kerr used this example to caution that broadband internet access disparities are still very real and that they no doubt will affect the degree to which individuals can receive “personalized care” or “precision medicine” in the future.

Six-month RCT (n=199 type 2s) of Twin Health diabetes reversal program: 84% achieve reversal, defined as A1c <6.5% with no glucose-lowering pharmacotherapy for three months

Dr. Shashank Joshi (Lilavati Hospital, India) presented data from a six-month RCT in India (n=199 type 2s) evaluating the efficacy of Twin Health’s “Whole Body Digital Twin” diabetes reversal program. As a reminder, Twin Health’s AI-enabled Whole Body Digital Twin integrates CGM data, fitness trackers, and other testing (connected scales, blood pressure monitors, etc.) to inform precision nutrition and medication management to drive chronic metabolic disease reversal and prevention. The company closed a whopping $140 million Series C in October 2021 and has also presented a small analysis from an RCT at ADA 2021, showing that 39% and 57% of participants (n=33) achieved full (A1c <5.7%) and partial (A1c >5.7% and <6.4%) diabetes remission, respectively, after four months.

  • Dr. Joshi explained that at six months, 84% of participants achieved “diabetes remission,” defined as “return[ing] to an A1c level of less than 6.5%, with the absence of glucose-lowering pharmacotherapy, for at least three months. This result is undoubtedly impressive, and we’ll be waiting to see whether these outcomes can be sustained at one year. The cohort also achieved notable improvements in A1c, decreasing by 3.3%, from 9.0% at baseline to 5.7% at four months (p<0.001). This improvement is comparable to outcomes from Twin’s four-month RCT, in which participants saw a 3.1% reduction in A1c from baseline (from 8.7% to 5.6%; p<0.0001). Dr. Joshi also explained that at six months, participants saw significant improvements in biomarkers across metabolic disease (e.g., cholesterol, inflammation, NAFLD, etc.). Dr. Josh explained that the one-year trial data should become available to investigators in August 2022, and so we look forward to hopefully seeing these outcomes at an upcoming conference.
  • Twin Health is currently recruiting patients for a US-based clinical trial at the Cleveland Clinic. The trial will examine the efficacy of Twin Health’s diabetes reversal program on a broader segment of the population, including those with a longer duration of diabetes, older age, and A1c as high as 11.5%. We are excited to see the outcomes from this trial with a more representative patient population, and hope that Twin Health will also aim to increase enrollment of underrepresented racial and ethnic populations in the US who can often be left out of clinical trials.

Posters – Telemedicine and Remote Monitoring

Title

Authors

Details + Takeaways

Role of Telehealth in Improving Health Care Quality in Type 2 Diabetes in Rural Community-Based Clinical Practice

Supraja Gururaj, Amir Asfa, Leslie C. Hanley, Yogesh R. Yadav, Randall T. Bashore, Rita Basu, Ananda Basu

  • Quality improvement study examining the impact of telehealth endocrinology consultation on glycemic outcomes for type 2s in a rural community-based clinic
  • Participants (n=27) with A1c >8% and on at least one diabetes-related drug; average age 60 years; 12 participants received professional CGM for two weeks to guide treatment plan and 15 received standard of care
  • Participants receiving telehealth achieved a 1.0% lower A1c from a baseline of 9.3% to 8.3% at six-months (p=0.03) and +5.8 hour/day in Time in Range
  • No statistically significant difference in Time Below or Above Range for the telehealth group
  • Standard of care group saw no differences in A1c or Time in Range

Using Telehealth to Deliver Diabetes Self-Management Education and Support (DSMES) in Rural Communities

Prem Sahasranam, Frank Gavini, Sandeep Sodhi, Theresa Garnero, Jacqueline Thompson, Carolyn M. Salinas, Stormie S. Baxter

  • Study assessing My Diabetes Tutor, a DMSES program, as a convenient means to access diabetes education in rural communities
  • Participants achieved -1.2% A1c improvement, from 8.3% at baseline to 7.1%; time of study is not listed on poster
  • Decrease in LDL (-9.4 points); improvement in blood pressure (average decrease of 10 mmHg); significant weight loss (75% of participants lost ≥7 lbs)

Feasibility and Acceptability of a Family-Based mHealth Intervention in low-Income Chinese Families with Type 2 Diabetes

Lu Hu, Shen Cheng, Nadia Islam, Judith Wylie-Rosett, Bei Wu, Naumi Feldman, Antoinette Schoenthaler, Olugbenga Ogedegbe, Nan Jiang, Kosuke Tamura, Omonigho Bubu, Mary Ann Sevick

  • Acceptability and feasibility of a family-based mHealth intervention for low-income Chinese immigrant families with diabetes; mean age of 55; 96% had limited English proficiency; 70% had less than a high school education
  • Patient-family partners (n=11) watched two videos per week for twelve weeks; survey completed at twelve weeks to assess acceptability; feasibility evaluated via retention and video watch rates
  • At three months, retention rate was 91%; mean video watch rate was 77%; lack of free time was most common barrier to adherence
  • High satisfaction scores (9.1/10 for participants and 10/10 for family members)

Feasibility of a Remote Clinical Trial in People with Type 2 Diabetes – Findings from the MOTIVATE T2D Trial

Andrew P. Davies, Katie Hesketh, Jonathan Low, Victoria S. Sprung, Helen Jones, Ali M. Mcmanus, Matthew Cocks, Motivate Team

  • RCT evaluating the feasibility of a remote clinical trial for type 2s (n=62)
  • Participants were sent self-testing kit at baseline and post-intervention
  • Participants lived in the UK (n=41) or Canada (n=21), and were 46% male and 86% white
  • Remote testing was feasible though only 50% of participants who consented to the study completed the entire study, meaning that retention might be an issue

Evaluating the Effectiveness of a Novel Form of Online Cognitive Behavioral  Therapy for Patients with Type 2 Diabetes and Comorbid Anxiety or Depression

Ann G. Hayes, Ana Catarino, Shaun Mehew, Andrew Blackwell

  • Cognitive behavioral therapy (CBT) intervention for type 2s with a mental health disorder
  • Participants (n=102) received online, diabetes-specific CBT; synthetic control group received standard online CBT
  • Treatment resulted in significant decrease in PHQ-9 (from 15.1 to 6.7), DDS (from 3.4 to 2.5), and GAD-7 (from 11.1 to 5.3) scores; maintained at six-month follow up
  • Recovery rates of 76% compared to an average 62% in the control group

A Personalized Retrospective Continuous Glucose Monitoring (CGM) Report Improves Engagement and Glycemic Control in a Remote Monitoring Diabetes Program (RMDP)

Robert J. Ellis, Tejaswi Kompala, Kevin Weng, Robert J. Brooks, Atousa Salehi, Roberta James, Hau Liu

  • Investigating the efficacy of Teladoc’s “CGM-powered Insights Report” within Livongo for Diabetes
  • Participants (n=1,105) include type 1s (n=436) and 2s on insulin (n=464) and not on insulin (n=205); mean age of 46 years; 55% male; mean diabetes duration of 12.3 years
  • Participants opened reports 49% of the time; participants who opened the reports were more likely to subsequently utilize other Livongo features (p< 0.05)
  • Opening more reports was correlated with improvements in Time in Range (+2.7%) and mean sensor glucose (-2.57 mg/dL)

Screening, Selecting, and Training Peers to Delivery Mental Health Support via a Virtual Care Platform to Adults with Type 1 Diabetes Living in Rural and Remote Communities

Tricia S. Tang, Kai Wai Yip, Lauren Moore, Gerri Klein

  • Study on the process to screen, select, and train peers to provide diabetes distress interventions to support adults living with type 1 diabetes in geographically marginalized settings
  • Peer supporter candidates (n=65) underwent a screening process (online personality assessment and interview); participants who passed (n=52) underwent a six-hour online peer support training on: (i) identifying internal motivation, boosting resilience, and expressing empathy; (ii) exercising mindfulness, sitting with difficult emotions, and exploring diabetes distress; and (iii) practicing active listening, avoiding giving advice
  • 46 of 52 participants successfully passed the program, demonstrating that a rigorous process is feasible to screen, select, and train individuals for this purpose

Telemedicine Intervention is Effective in Changing Psychological and Behavioral Correlates of Weight Loss in the REAL HEALTH Diabetes Trial

Janaki Vakharia, Linda M. Delahanty, Tanayott Thaweethai, Chu Yu, Deborah J. Wexler

  • The REAL HEALTH diabetes trial, a three-arm randomized comparative effectiveness trial examining a lifestyle intervention program for weight loss in adult type 2s; sought to evaluate whether in-person (n=69) or telehealth (n=72) delivery of the intervention would impact outcomes
  • Participants across both groups had a mean age of 62 years; 45% male; 79% white
  • At six months, there was no significant difference in outcomes between in-person and telehealth cohorts
  • Average weight loss of 5%; 0.5% A1c reduction from a baseline of 7.7% to 7.2% at six months; also, improved psychological and behavioral outcomes

Telehealth Results in Increased Visit Frequency and Lower Diabetes Distress in Young Adults (YA) with Type 1 Diabetes

Jennifer Raymond, Jennifer L. Fogel, Mark W. Reid, Elizabeth Salcedo-Rodriguez, D Steven Fox, Elizabeth Pyatak

  • A 15-month, four-arm randomized controlled trial that investigated the adaption of the CoYoT1 Care model for a diverse, publicly insured group of young adult type 1s at a large urban hospital
  • Participants who attended study visits via telehealth attended more sessions than those attending in person (p=0.006)
  • Telehealth cohort saw significant reduction in A1c (-0.8%) vs. in-person cohort (p=0.048) and had stable levels of diabetes distress versus worsening levels among in-person cohort (p=0.02)

Virtual Peer Groups (VPG) for Adolescents and Young Adults (AYA) with Type 1 Diabetes (T1D): Patient-Reported Importance of Specific Features

Daniel I. Bisno, Elizabeth Pyatak, Mark W. Reid, D. Steven Fox, Jennifer L. Fogel, Elizabeth Salcedo-Rodriguez, Jaquelin J. Flores Garcia, Alejandra Torres Sanchez, Jennifer Raymond

  • CoYoT1 to California is a fifteen-month, randomized controlled trial for young adult type 1s, examining the effect of introducing bi-monthly virtual peer groups led by a young adult with type 1 diabetes
  • Participants receiving CoYoT1 (n=40) were 40% female; 53% Latinx; and 72% publicly insured
  • Seeing peers use diabetes technology and finding support in same-aged peers with type 1 diabetes was the most impactful self-reported facet of CoYoT1 for participants
  • 75% of participants found the virtual peer groups to be “extremely” or “very” valuable

Rural Caregiver Perceptions of an Occupation-Based Coaching Telehealth Intervention to Improve Child Diabetes Management and Quality of Life

Vanessa D. Jewell, Katie J. Funk, Alexis Currie, Julia Shin, Emily Knezevich, Andrea D. Valdez, Maggie Bunsness

  • Investigating perceptions of a twelve-week occupational-based telehealth coaching intervention by rural caregivers of children with type 1 diabetes
  • Eight caregivers who had previously received the intervention were interviewed about their experiences with the intervention and their child’s diagnosis
  • Salient themes across interviews included: (i) satisfaction with diabetes management and psychosocial support; and (ii) hope for childhood normalcy
  • From the caregivers’ experience after diagnosis, the themes emerging were: (i) occupational deprivation and decreased wellbeing; (ii) longing for connection with social supports; and (iii) desire for knowledgeable, relatable, and accessible providers

Improve Glycemic Control from a Remote Patient Monitoring Program across 19 Primary Care Practices Integrating Home Monitoring, Mobile Health, Nutrition/Lifestyle Coaching, and Telemedicine for Diabetes Management

Li Wang, Hang Xu, Zhiyu Liu, Nathan D. Wong

  • Participants across nineteen primacy care practices were enrolled in iHealth Unified Care to examine the efficacy of the program, which integrates RPM with lifestyle coaching for type 1s and 2s
  • 691 participants were enrolled with a mean age of 62 years; mean BMI of 30 kg/m2; and 47% female
  • Participants achieved a 0.6% lower A1c, from a baseline of 7.4% to 6.8% at three months; A1c was flat at six months (p<0.01 for both)

Impact of Social Determinants of Health (SDoH) on Participation in a Remote Monitoring Diabetes Program (RMDP) to Improve Glycemic Control

Tejaswi Kompala, Wei Lu, Stefanie L. Painter, Roberta James, Atousa Salehi, Hau Liu

  • Retrospective analysis of SDOH within users of Livongo for Diabetes
  • Participants’ (n=1,308) mean age was 51 years; 54% female; 13% Black; 5% Asian; 12% Hispanic; and 84% were type 2s
  • After adjusting for participant characteristics, medication usage, and diabetes duration, SDOH were neither associated with high program utilization nor glycemic management

Disparities in Telemedicine Use among Pediatric Patients with Type 1 and Type 2 Diabetes

Karen Chan, Carla Demeterco-Berggren

  • Survey-based study on the use and barriers surrounding telemedicine services in diabetes care; 39 participants had only in-person services; 39 participants utilized telemedicine appointments
  • Reported reasoning for not attending a telemedicine visit included: (i) preference for in-person care (44%); (ii) not being offered a telemedicine visit (18%); (iii) technology issues (13%); and (iv) scheduling conflicts or forgetting about appointments (18%)
  • 86% of participants who had attended telemedicine visits reported they were “very satisfied” and 97% thought their telemedicine visit was better or the same as their in-person visit

Use of Telemedicine for Type 1 Diabetes Care in the T1D Exchange Quality Improvement Collaborative (T1DX-QI) in 2021

Joyce M. Lee, Emma L. Ospelt, Ann Mungmode, Osagie Ebekozien, Meenal Gupta, Faisal Malik, Naomi R. Fogel, Siham Accacha, Susan Hsieh, Meredith Wilkes, Anna Neyman, Francesco Vendrame

  • Pediatric (n=24) and adult (n=9) diabetes clinics completed a survey about their rates of telemedicine usage and clinic-level processes to integrate these visits into established workflows
  • Telemedicine represented 38% of all visits in 2020 and 20% in 2021; 52% reported that 11%-25% of their visits occur virtually; 73% of clinics reported having a pre-visit preparation workflow but 64% did not have a staff member dedicated to supporting telemedicine visits
  • No major differences in telemedicine use between pediatric and adult clinics; largest barriers to telemedicine for providers were: (i) patient internet access; (ii) patient health disparities; and (iii) access to device data

Implementation of Remote Continuous Data Monitoring within a Clinical Setting for the Management of Type 2 Diabetes Mellitus

Milena Caccelli, Yousef Said, Joena Fe D. Mojado, Carolyn Palsky, Raiza Colodetti, Ihsan Almarzooqi, Ali Hashemi

  • Retrospective, investigational study of GluCare program in type 2s
  • Participants utilizing GluCare (n=20) achieved a 2.4% A1c reduction at three months; superior (p< 0.001) to A1c decrease in control group (n=20)
  • Participants in treatment group also demonstrated improvements in BMI, cardiovascular risk scores, and triglycerides compared to the control group (p=0.046, p=0.001, p=0.018, respectively)

Utilization of Glucose Management Team-Conducted Remote Patient Visits Decreases Hospital Readmission for Patients with Diabetes

Allison Johns, Catherine E. Price, Joseph A. Aloi, Cynthia Burns

  • Examination of readmission rates for patients who received post-discharge telemedicine visits
  • Participation in scheduled telehealth visit (n=49 of 88) resulted in a 4% 30-day readmission rate and a 22% 90-day readmission rate
  • Odds ratio for a ninety-day readmission was 2.7 for those who did not attend a post-discharge telemedicine visit, compared to those who did (p=0.042)

The Association between Patient Characteristics and the Efficacy of Remote Patient Monitoring and Messaging

Johannes Ferstad, Priya Prahalad, David M. Maahs, Emily Fox, Ramesh Johari, David Scheinker

  • Retrospective study on the association between CGM metrics and remote patient monitoring messages; participants (n=135) clustered into “more responsive” or “less responsive” groups
  • Both clusters achieved an average +1.4 hours/day Time in Range when receiving messages relative to +0.7 hours/day Time in Range for those not receiving messages

Telehealth Benefits among Latinx Adolescents and Young Adults (AYA) with Type 1 Diabetes (T1D)

Jaquelin J. Flores Garcia, Mark W. Reid, Elizabeth Pyatak, Jennifer L. Fogel, D Steven Fox, Elizabeth Salcedo-Rodriguez, Jennifer Raymond

  • In this two-by-two RCT, participants (n=68) received CoYoT1 Care or standard of care either in-person or via telehealth
  • Participants receiving CoYoT1 telehealth care achieved reduced A1cs (p=0.007) and reduced physician-related distress (p=0.02) compared to increased A1cs and distress in the in-person standard of care group
  • Latinx participants receiving CoYoT1 via telehealth achieved a mean 1.0% (!) lower A1c compared to the in-person CoYoT1 group and a 0.90% lower than the standard of care telehealth group (p=0.003); non-Latinx participants demonstrated no significant difference in A1c or distress when engaging via telehealth or in-person

Access to Telemedicine and the Number of Medical Visits Pre-and Post-Lockdown in Latin American Children with Type 1 Diabetes

Valeria Hirschler, Julie Pelicand, Angela M. Figueroa Sobrero, Diana S. Gonzalez, Edit R. Scaiola, Patricia Bocco, Paola M. Pinto Ibarcena, Carlos M. Del Aguila, Carolina Andrea Ramirez Trillo, Ailín I. Mac, Siliva Lapertosa, Claudia Molinari

  • Examining pre- and post-pandemic medical visits for pediatric type 1s in three Latin American countries
  • Participants (n=217) had an average age of 13 years; diabetes duration of 6 years; 64% of participants had access to telemedicine
  • Number of medical visits was significantly higher in participants via telemedicine in 2020 (mean 7.0 visits) and 2021 (mean 5.4 visits) compared to those without telemedicine (2.6 and. 1.9 visits in 2020 and 2021, respectively)
  • The number of medical visits post-lockdown (5.3) was not significantly different than those in 2018 (5.1)

INNOVATIONS: Improving Care of Patients with Type 1 Diabetes Mellitus through Utilization of Telemedicine and Outreach

Lisa E. Rasbach, Virginia Purrington, Deanna Adkins, Robert Benjamin

  • Three-month data from a quality improvement project providing weekly consultations with DCESs and monthly virtual visits with NPs
  • Youth participants (n=29) with a mean age of 14 years; 65% female; 88% non-white
  • No difference in A1c levels at three months vs. baseline; no difference in hospitalization rates compared to two-year historical data
  • Baseline A1c levels negatively correlated with diabetes comprehension scores (p=0.03)

An Interactive Capacity Planning Dashboard for Algorithm-Enabled Telemedicine-Based Diabetes Care

Annette Chang, Michael Z. Gao, Johannes Ferstad, David M. Maahs, Priya Prahalad, Ramesh Johari, David Scheinker

  • Design of an interactive dashboard for CGM-derived metrics to improve population-level management via algorithm-enabled telemedicine
  • Iterative design based on user feedback (n=277); 53% on public insurance; 35% non-Hispanic White
  • Dashboard contains: (i) seven modifiable parameters; (ii) table illustrating capacity and demand within clinic; and (iii) three alternative capacity plans for the clinic
  • Modest reductions in average time needed to review patient data through the standardization of workflows, producing large improvements in clinic capacity

Is Glycemic Control Better in Adults with Type 1 Diabetes when Office Visits Are Supplemented with Telehealth Visits?

Elena Toschi, Atif Adam, Christine Slyne, Lori M. Laffel, Medha Munshi

  • Retrospective analysis of EHRs from type 1s pre- and post-pandemic to determine whether telehealth supplementation is correlated with improved glycemic management
  • Participants (n=1,820) were stratified by age: younger group (n=1,210) was designated as people 40-64 years old, and the older group (n=610) was designated as adults >65; whole population had CGM use rate of 53% at baseline 
  • Both groups achieved lower A1c levels after hybrid care, with the younger group achieving a 0.2% lower A1c than the in-office control (p=0.005) and the older group achieving a 0.2% lower A1c vs. control, both from a baseline of 7.6%

Posters – Digital Health, Decision Support, and Algorithms

Title

Authors

Details + Takeaways

Health Care Professional (HCP) Perspectives on Support Tools for Diabetes Devices

Laurel H. Messer, Timothy B. Vigers, Halis K. Akturk, Gregory P. Forlenza, Kelsey B. Huss, Angela J. Karami, Emily Malecha, Sean Oser, Sarit Polsky, Laura Pyle, Viral Shah, R. Paul Wadwa, Tamara Oser

  • Survey of PCPs (n=115), pediatric endos (n=36), and adult endos (n=89) assessing the relative importance of support tools to increase patient engagement with diabetes devices
  • HCPs ranked “how useful each tool would be in clinical practice” via visual analog scale (VAS)
  • 27% (n=31/115) of PCPs were in rural settings; only 3% (n=1/36) of pediatric endos and 5% (n=4/89) of the adult endos practiced in a rural setting
  • HCPs broadly ranked insurance coverage tools and online data platforms most useful
  • PCPs ranked automated insulin dosing and consultations with experts as more useful vs. endocrinologists
  • HCPs assigned high importance to EHR data integration (mid-80s VAS scores)

Feasibility of CGM Use in Basal-Bolus Insulin Therapy with an Electronic Decision Support System in Hospitalized Patients

Kujtim Bytyqi, Petra M. Baumann, Daniel A. Hochfellner, Tina Poettler, Amra Simic, Peter Beck, Julia K. Mader

  • Study evaluating feasibility of using CGM to guide clinical decision making for hospitalized patients on basal-bolus therapy
  • 30 adults type 2s wore CGM during inpatient stay to assess benefit of CGM vs. BGM
  • Total daily dose suggestions were lower with CGM vs. BGM (23 IU vs. 27 IU, respectively) due to higher detection of hypoglycemia
  • Mean daily glucose levels were lower with CGM (124 mg/dl) vs. BGM (147 mg/dl)
  • Time in Range was 4.0 hours/day lower for those using CGM (57%) vs. BGM (74%)
  • 36 hypoglycemic events could have potentially been avoided if CGM was used in BGM-only users due to insulin adjustment

Importance of Data Distribution on Deep Learning Model for Glucose Prediction

Hector M. Romero-Ugalde, Alice Adenis, Laurent Daudet, Maxime Louis, Yousra Tourki, Erik Huneker

  • Analysis of the impact of glycemia distribution on a deep learning model for glucose prediction in type 1s using Diabeloop DBLG1 AID system (n=139)
  • Fully connected neural network model uses historical glycemic values for training
  • Hyperoptimized deep learning model was a better predictor of individual-level glycemic control when individual glycemic distributions had smaller deviations from hyperoptimized model

Improvement of Glycemic Control in People with Type 2 Diabetes Using Basal Insulin (PWT2D-BI) and Continuous Glucose Monitoring (CGM) -U.S. Study Using Machine Learning (ML)

Chanadda Chinthammit, Alan J.M. Brnabic, Magaly Perez-Nieves

  • Retrospective analysis studying association between glycemic control and CGM utilization in people with type 2 diabetes on basal-only therapy (n=5,934)
  • Majority of participants were in the control group and were CGM-naïve before and during the study (n=5,683)
  • Intervention cohort included patients with continuous use of CGM for at least six months (n=251); some participants kept wearing CGM out to 12 months (n=129)
  • CGM users in six-month cohort experienced a statistically significant 0.9% reduction in A1c from 8.7% at baseline to 7.8%; 12-month users saw 1.0% reduction from 8.4% at baseline to 7.4%; control group saw 0.3% A1c reduction from 8.2% at baseline to 7.8%

Deep Learning Model for Predicting Hospital Readmission among People with Diabetes

Ameen Abdel Hai, Mark Weiner, Anuradha Paranjape, Alice Livshits, Jeremiah Brown, Zoran Obradovic, Daniel J. Rubin

  • Deep learning model using EHR data developed to predict readmission risk among people with diabetes
  • Compared to those not readmitted, patients who were readmitted had lower BMIs, higher number of prior encounters, fewer mean days since last encounter, lower hematocrit/glucose/A1c values, and higher creatinine/LOS
  • Readmitted patients were more likely to be discharged to a nursing/rehab facility
  • Deep learning recurrent neural network (RNN) model had higher accuracy (AUC=0.81) than multi-layer perceptron model (AUC=0.75) or logistic regression model (AUC=0.77)

The Potential Value of Clustering-Based Type 2 Diabetes Subgroups for Guiding Intensive Treatment: A Maximum Cost-Based Comparison with Threshold-Based Classifications

Xinyu Li, Anoukh Van Giessen, James Altunkaya, Roderick Slieker, Joline Beulens, Ewan Pearson, Petra J.M. Elders, Talitha Feenstra, Jose Leal

  • Study evaluating the added value of treatment intensification targeting cluster-based subgroups, threshold-based subgroups, or a “homogenous” view of type 2 diabetes
  • Patients from the Hoorn Diabetes Care System cohort (n=2,935) were divided into five cluster-based subgroups: (i) severe insulin-deficient (SIDD); (ii) severe insulin-resistant (SIRD); (iii) mild obesity-related (MOD); (iv) mild (MD); and (v) mild with high HDL-cholesterol (MDH) diabetes; four risk threshold-based subgroups were used
  • United Kingdom Prospective Diabetes Study Outcomes Model was used to simulate lifetime health outcomes; gains from hypothetical treatment scenarios based on clinical guidelines were compared to standard care and expressed in incremental quality-adjusted life-expectancy (QALE) and complication costs
  • Severe insulin-deficient and mild obesity-related subgroups had the lowest QALE (7.90 and 9.07, respectively); MOD interventions found to be cost effective in US and UK based on willingness to pay
  • Threshold-based classification better stratified patients with high and low QALE vs. clustering-based subgroups
  • Targeting LDL and BMI in addition to A1c led to a tenfold increase in incremental QALE vs. A1c alone

External Validation of the BRAVO Diabetes Model Using Data from the EXSCEL Study

Yixue Shao, Shu Niu, Lizheng Shi, Vivian Fonseca, Hui Shao

  • Apply the “building, relating, assessing, and validating outcomes” (BRAVO) model to patient-level data to predict outcomes of EXSCEL CVOT
  • BRAVO diabetes model is a person-level discrete-time microsimulation model, predicting the progression of diabetes based on individuals’ sociodemographic and clinical characteristics and treatments
  • Study concluded BRAVO model could successfully predict first occurrence of non-fatal MI, stroke, HF, revascularization, and all-cause mortality

Improving Decision Making by Clinicians for Initiation of Insulin in Adults with Type 2 Diabetes Using Simple Python Program

Om J. Lakhani

  • Python program retrospectively deployed on EHR data (n=500); ten clinicians then used the program on 500 other individuals
  • Program incorporated evidence-based recommendations for insulin initiation in insulin naïve, outpatient, non-pregnant, adult type 2s
  • The program recommended insulin initiation in 164 patients, but clinicians only initiated insulin in 112 patients; there were no cases in which the clinician initiated insulin but the computer recommended against it
  • When used prospectively, HCPs initiated insulin in 159 of the 164 patients that the computer recommended; clinicians manually overrode five patients
  • Insulin initiation increased by 28% when program was used (p=0.037)

Multicenter application of EFR (Endocrine Final Recommendation) Note Embedded in EMR Improves the Accuracy of Diabetes Discharge Regimen in Hospitalized Patients with Diabetes

Jacob Lloyd, Tiba Abdulwahid, Joel Levin, Diana Pinkhasova

  • Retrospective chart review evaluating accuracy of diabetes discharge regimen for patients with Endocrine Final Recommendation (EFR) note (n=40) vs. patients for whom EFR not utilized (n=40)
  • Accuracy of diabetes regimen at discharge was higher when EFR note was utilized vs. when not (88% vs. 60%; p=0.011)
  • Most common type of treatment regimen error was incorrect insulin type/dosing (accounted for 76% of all errors); such errors were less prevalent in the EFR group relative to control (8% vs. 33%; p=0.012)

Improvements in Glycemic Control following Use of an Established mHealth Program among Individuals with Diabetes

Gretchen Zimmermann, Aarathi Venkatesan, Kelly Rawlings, Richard S. Frank, Caitlyn Edwards

  • Retrospective study on the effectiveness of Vida mHealth program on A1c in patients with baseline A1c >8% and >9% (n=1,023)
  • A1c declined after enrollment in mHealth: in baseline >8% A1c group, -1.4% from 9.8% to 8.4% (p<0.001); in baseline >9% A1c group, -1.9% from 10.8% to 8.9% (p<0.001)

Feasibility and Acceptability of a Family-Based mHealth Intervention in Low-Income Chinese Families with Type 2 Diabetes

Lu Hu, Shen Cheng, Nadia Islam, Judith Wylie-Rosett, Bei Wu, Naumi Feldman, Antoinette Schoenthaler, Olugbenga Ogedegbe, Nan Jiang, Kosuke Tamura, Omonigho Bubu, Mary Ann Sevick

  • 11 Chinese American patient-family partner adult dyads were recruited to watch 2 brief mHealth intervention videos for 12 weeks via WeChat; dyads mostly had less than high school education, limited English proficiency, and annual household income <$25,000
  • 77% of videos were watched over the 12-week study; retention rate of 91%; acceptability was high, with lack of time cited as most common barrier to watching videos
  • Participants ranked satisfaction with the intervention 9.1/10; family members ranked satisfaction 10/10

Significant Changes in Dietary Behaviour, Glycemic Control, and Weight after Participation in Personalized Glycemic Response–Based Diabefly Program

Ritika Verma, Mohammed A. Khader, Ruchira M. Ranadive, Sweta Budyal, Amit A. Saraf, Vivek Raskar, Ashish Sarwate, Mitali V. Joshi, Shilpa Joshi, Arbinder K. Singal

  • Analysis of type 2s (n=108) in 90-day personalized glycemic response (“PGR”)-based coaching program, Diabefly
  • Program associated with improved glycemic control: mean A1c reduced by 2.0% from baseline of 8.8% to 6.8% (n=66; p<0.001)
  • Weight reduction of 3.4 kg from mean baseline weight of 76.4 kg to 73 kg (n=66, p<0.001); BMI reduced by 1.2% from baseline of 27.34 kg/m2 (n=66; p<0.001)
  • Program led to significant chances in dietary behavior: daily protein intake increased by 10% (p<0.001); daily fat intake reduced by 8% and daily carb intake reduced by 10% (p<0.001)

Significant Weight Reduction and Improvement in Glycemic Control among People with T2D after Participation in Diabefly Digital Therapeutics Program

Ritika Verma, Mohammed A. Khader, Forum Malde, Tejal Lathia, Snehal R. Tanna, Sneha Kothari, Piya Thakkar, Santosh B. Thakur, Saimala Guntur, Arbinder K. Singal

  • Study of type 2s using Diabefly program for 90 days (n=172)
  • Mean A1c was reduced by 1.7% from a baseline of 8.7% (p<0.001)
  • Weight and BMI were significantly reduced by 2.6 kg and 0.9 kg/m2 from a baseline of 75.9 kg and 27.5 kg/m2 respectively (p<0.001 for both)
  • Among participants (n=42) with complete fasting blood sugar and postprandial blood sugar data, a significant mean reduction by 51 mg/dL and 90 mg/dL from a baseline of 164 mg/dL and 228 mg/dL was observed, respectively (p<0.001 for both)

CGM Attitudes and Adoption among People with Type 2 Diabetes Using One Drop

Lindsay Sears, Steven D. Imrisek, Jamillah R. Hoy-Rosas, Lindsey M. Lavaysse, Matthew Lee, Matthew R. Chapman, Jeff Dachis

  • Survey of health attitudes and behaviors of type 2s paying for One Drop membership (n=171) in November 2021
  • 96% of participants found One Drop’s health logging features helpful; 95% found the glucose analysis and reports helpful; 92% reported logging their blood sugar daily/weekly
  • 90% of participants were familiar with CGM devices but 83% of that group had never tried one at baseline; 73% of those familiar with CGM were open to starting it
  • Reasons for openness to adopting CGM (n=107) included valuing continuous feedback (28/107), relief from fingersticks (15/107), convenience (12/107), and improved diabetes management (12/107); 12 participants expressed concerns about cost and reimbursement

Effectiveness of Breathe Well-Being Diabetes Reversal Program in Achieving Diabetes Remission

Pawan K. Goyal, Surajeet K. Patra, Aditya Kaicker, Rohan Verma, Seema Goel, Saleha Rehman, Rupali Jangid, Venugopal Madhusudhana

  • Study of 181 individuals type 2s in India participating in a study to evaluate effectiveness of Breathe Well-Being Diabetes Reversal Program (BDRP) on achieving diabetes remission
  • Individuals were sorted into three cohorts: cohort 1 used BDRP and doctor-prescribed medication (n=60); cohort 2 used BDRP, doctor-prescribed medication, and a stress reduction module (n=61); cohort 3 (the control) only had access to doctor-prescribed medication (n=60)
  • 47% of cohort 1 had high (>70%) medication adherence (n=11) or medium (50%-70%) medication adherence (n=17); 54.1% of cohort 2 had high (n=13) or medium (n=20) medication adherence
  • 89% of cohort 1 and 91% of individuals in cohort 2 with high or medium adherence had medication withdrawal at mean 19 months from baseline
  • Diabetes remission at 19 months with A1c <6.5% and total medication cessation was achieved for 79% and 82% of cohort 1 and 2 with high/medium adherence, respectively

Hypertension Control among People with Diabetes Using a Self-Management Multicondition Digital Platform

Yifat Hershcovitz, Michal Tamir, Omar Manejwala

  • Retrospective analysis of DarioHealth users in hypertension stage 1 (systolic >130 mmHg or diastolic >80 mmHg) upon app initiation (n=2,554)
  • 39% of users (n=99) moved to a lower hypertensive stage after six months of use (p<0.001)
  • Blood glucose reduced by 15% on average over six months for users who started at “high-risk” levels (n=306), from 232 mg/dL at baseline to 198 mg/dL

Persons with High-Risk Diabetes, Depression, and Stress Using a Digital Health Platform Experience an Improvement in Glycemic Management

Yifat Hershcovitz, Michal Tamir, Marilyn D. Ritholz, Omar Manejwala

  • Retrospective data analysis of DarioHealth users who activated the app during 2019-2021 and self-reported stress and depression in the app questionnaire and were defined as “high risk” due to BG >180 mg/dL (n=491)
  • After one year of utilizing the app, high-risk users reduced blood glucose by 13% (from average of 234 mg/dL at baseline to 204 mg/dL; p<0.001)
  • Subgroup of high-risk type 2s (n=379) reduced average blood glucose by 14% (from 233 mg/dL at baseline to 201 mg/dL)

Real-World Evidence of Arabic Digital Therapeutics for People Living with Type 2 Diabetes in Qatar

Noor Jandali, Rahma Saad

  • People with type 2 diabetes in Qatar (n=633) used Droobi Health, an Arabic digital behavior change program, for three months
  • Mean A1c reduction for program completers was 0.9% from baseline of 9.7% (p<0.001)
  • Male participants had a higher reduction in mean A1c (-1.0%%; p<0.001) than female participants (0.7%; p<0.001) after three months

User Acceptability of Ecological Momentary Assessment to Report Hypoglycemia over Ten Weeks Using a Novel Smartphone Application in the Hypo-METRICS Study: Hypoglycemia Measurements, Thresholds and Impacts

Natalie Zaremba, Gilberte Martine-Edith, Patrick Divilly, Zeinab Mahmoudi, Uffe Soeholm, Melanie M. Broadley, Frans Pouwer, Bastiaan E. De Galan, Ulrik Pedersen-Bjergaard, Rory J. Mccrimmon, Eric Renard, Simon R. Heller, Mark Evans, Julia K. Mader, Alexander Seibold, Stephanie A. Amiel, Pratik Choudhary, Hypo-Resolve Consortium

  • Type 1s and 2s (n=514) using Hypo-METRICS app were studied using ecological momentary assessments (EMA) over 10 weeks to determine the physical, psychological, and economic impact of hypoglycemia; overall completion rate was 88%
  • Number of self-reported hypoglycemic events/week for type 1s (n=241) was 2.2; type 2s (n=273) reported an average of 1.0 hypoglycemic events/week

Blood Glucose Levels in High-Risk Type 2 Diabetes Users of a Digital Therapeutic Platform by Race/Ethnicity

Yifat Hershcovitz, Tamar Gershoni, Omar Manejwala

  • Retrospective study of DarioHealth users with type 2 diabetes that were active between 2019 and 2021
  • Over a year, average blood glucose was reduced by 14% across all users
  • Across ethnic groups, average blood glucose reduction of: (i) 14% for White participants; (ii) 15% for Latinx participants; (iii) 15% for Black participants; and (iv) 15% Asian participants
  • No difference in average blood glucose level between groups was found at month 12 (p=0.751)

Clinical Outcomes of Euglycemic DKA with a Computerized Insulin Algorithm: Descriptive Analysis of a Nationwide Cohort

Jordan Messler, Priyathama Vellanki, Robert Booth

  • Assessed safety and efficacy of Glucommander in treating euglycemic DKA (euDKA)
  • 533 patients with euDKA were treated with Glucommander (28% type 1s, 33% type 2s, 39% unknown)
  • Time to bicarb >18 mEq/L (i.e., time to resolve euDKA) was 14 hours with a median length of hospital stay of 3.2 days

Impact of Culturally Appropriate Diabetes Care amongst Asian Americans with Type 2 Diabetes during the COVID Pandemic

Atif Adam, Ka Hei Karen Lau, Hetal Shah, Marc Gregory Yu, George L. King

  • Retrospective EHR review of type 1s who had visits pre-pandemic (April 2019 – December 2019) and during pandemic (April 2020 – December 2020)
  • Patients were selected from two clinics: General Adult Diabetes Clinic (GADC) and Asian American Diabetes Initiative Clinic (AADI)
  • During the pandemic, average number of visits doubled per person: from 2.9 to 4.7 appointments for AADI (n=533), and from 2.3 to 4.2 appointments for GADC (n=4,419)
  • Non-native English speakers had 45% lower odds of having an A1c ≤7% in GADC clinics vs. AADI clinic

Improvement in Glycaemic Control and Cost of Diabetes Care Using Diahome, a Smartphone Application

Arun Raghavan, Arun Nanditha, Krishnamoorthy Satheesh, Priscilla Susairaj, Ramachandran Vinitha, Dhruv Nair, Santhosh Jeyaraj, Vajpayee Sharad, Ambady Ramachandran

  • Comparison of glycemic control in type 2s using Diahome diabetes management app (n=91) vs. nonusers (n=82)
  • After four months, app users saw a significant decrease in FPG (156 mg/dL vs. 129 mg/dL; p<0.02)
  • A1c reduction was greater for in app users (-1.0%, from 8.2% to 7.2%) vs. non-users (-0.5%, from 7.8% to 7.3%; p=0.004 for interaction)
  • Cost of treatment per visit was reduced from $112 to $57 when using the app

Time in Range and Beyond A1c

A1c and GMI are not good estimates for mean glucose in nondiabetes range nor at high mean glucose levels (>250 mg/dL) based on analysis of massive (n=2,842 people with diabetes) dataset

Dr. Viral Shah (Barbara Davis) read out an important study analyzing the correlation between GMI, A1c, and CGM-derived mean glucose (88-OR). As background, Dr. Shah reminded the audience that the mean A1c of the trial that was used to develop the GMI calculation was 7.3% and that we currently have little understanding of whether the calculation is accurate at low or high A1c values, in those with prediabetes, or in those without diabetes. In the analysis read out by Dr. Shah, the researchers pooled data (n=2,842, 92% type 1s, mean A1c 7.7%, mean age 46) from the Barbara Davis Center adult clinic (type 1 diabetes for at least two years and using Dexcom G6 for six months), from a study of CGM in nondiabetes, the FLAIR study, the MOBILE study, the CITY study, the iDCL DCLP3 RCT, the WISDM study, the REPLACE-BG study, the DIAMOND type 1 and type 2 studies, and the HypoDE study. The researchers applied two models to the data: (i) a simple linear regression similar to that which was used for the original GMI calculation; and (ii) a non-linear generalized additive model (GAM). Overall, the analysis found that A1c and GMI are inaccurate measures of mean glucose in those without diabetes and that its accuracy substantially declined at glucose values < 150 mg/dL and >250 mg/dL. Based on these findings, Dr. Shah suggested that: (i) GMI should not be used in people without diabetes; and (ii) “we should take all equations and metrics with a grain of salt” and that there is no perfect metric. In response to questions about whether we should stop using A1c based on this data, Dr. Shah agreed that mean glucose is a more valuable metric when available but noted that A1c is tied to long-term outcomes, which means that it still is valuable, and that no one metric – whether that’s A1c, GMI, or Time in Range – offers enough information on its own. He implored all to use multiple metrics when assessing an individual’s glycemic management, and long-term complication risk to drive treatment decisions. On this, there appears to be increasing agreement in the field could not agree more – it’s all about using data to drive personalized, collaborative diabetes management.

  • The analysis found no correlation between mean glucose and A1c in people without diabetes. Based on data from 153 people without diabetes, A1c had very weak correlation with mean glucose, suggesting different glycation rate in different people even at near normal glucose level. GMI is not appropriate to use in people without diabetes as it would not provide accurate estimate of A1c. This is significant, particularly as CGM use expands into broader populations, including into wellness spaces, as it suggests that people without diabetes may see a GMI far higher than their actual mean glucose, which could cause a false positive if used for diabetes diagnosis.

  • In people with diabetes, A1c and GMI were well correlated with mean glucose between ~6.5% and ~9%. In fact, all three models – the original GMI model, the new linear model, and the nonlinear GAM model – fit the data well between glucose levels of 150 mg/dL and 250 mg/dL. However, below and above this range, none of the models worked well. Even within the range from 150 mg/dL to 200 mg/dL, people with the same mean glucose value could have vastly different A1c values. For example, three people with a mean glucose of 175 mg/dL could have an A1c of 6%, 8%, and 10%, as was seen in this dataset.

  • Across all glucose ranges, a large proportion of participants saw ≥0.4% discrepancy between A1c and GMI. Specifically, 40% of those with mean glucose values ≤150 mg/dL, 46% of those with mean glucose values 151-200 mg/dL, 61% of those with mean glucose values 210-250 mg/dL, and 61% of those with mean glucose values >250 mg/dL had a ≥0.4% discrepancy between their A1c and GMI. As Dr. Shah pointed out, this is particularly significant given that the A1c/GMI change that has been deemed clinically significant is 0.4%.

“It’s complicated”: Dr. Irl Hirsch discusses clinical utilization of glycemic metrics overviewing the benefits and limitations of A1c, Time in Range, and GRI

Dr. Irl Hirsch (University of Washington) discussed the pros and cons of three existing and proposed assessments of glycemic management: A1c, Time in Range, and GRI (Glycemia Risk Index), ultimately concluding that none are perfect, but can be used collectively to help inform diabetes management. While there has been much conversation around the benefits of A1c versus Time in Range and vice versa, this was the first session in which we also heard GRI discussed. We heard from Dr. Hirsch on the topic of A1c vs. Time in Range at ATTD 2022 when he expressed his support for the use of CGM-derived metrics to inform the clinical and personal management of diabetes. Instead of focusing on the comparison between Time in Range and A1c in today’s presentation, Dr. Hirsch took it in a different direction discussing the shortcomings of A1c, reminding attendees that Time in Range without Time Below Range is “an incomplete number,” and providing examples of patients for whom GRI may not accurately predict glycemic risk. That said, across these three metrics, Dr. Hirsch also highlighted how they can help guide clinicians’ decision-making to help drive improved outcomes for people with diabetes. However, in order to use these metrics to help improve diabetes outcomes, Dr. Hirsch advocated for increased provider education on the limitations of A1c and the development of consensus targets for Time in Range, Time Below Range, and GRI across a variety of patient populations so that clinicians can use the information effectively.

  • Starting with A1c, Dr. Hirsch emphasized the inability of A1c to identify and alert clinicians to a patient’s risk of hypoglycemia. At ATTD 2022 we saw data demonstrating that patients with higher A1cs (9-10%), actually experienced a greater risk of severe hypoglycemia, likely due to increased glycemic variability, compared to patients with lower A1cs. Given this, Dr. Hirsch emphasized the lack of granularity in A1c as a key limitation of the metrics and also shared that the only reason he conducts A1c tests in his clinic anymore is “because of insurance” and that he very rarely uses it to inform his clinical management of patients with diabetes. While we believe there is still value to A1c, especially in terms of risk assessment for long-term complications, we agree with Dr. Hirsch that it lacks the level of detail needed to direct highly-informed and personalized diabetes care.
    • In a separate presentation on the relationship between Time in Range and micro and macrovasulcar complications, Dr. Roy Beck (Jaeb Center) addressed a question on whether it is mean glucose or a “non-glycemic factor that A1c is a marker for” that is associated with long term complications. Wow! While Dr. Beck shared his view that it is mean glucose that drives the long-term complications of diabetes, we found this to be a fascinating question hinting at an often-unspoken component of the debate on the utility between A1c and Time in Range in which the additional information included in these metrics may in and of itself play a role in long-term complications and outcomes. Given this potential, we see even more reason to use A1c and Time in Range together to help guide diabetes management and create a fuller picture of each individual’s glycemic profile.
  • Discussing Time in Range, Dr. Hirsch was adamant in his view that Time in Range must be considered in conjunction with other CGM-derived metrics, most importantly, with Time Below Range. To provide an example of why using Time in Range and Time Below Range together is so important, Dr. Hirsch asked the audience if they would be happy with a patient achieving 80% Time in Range, to which most attendees responded positively. However, Dr. Hirsch then followed-up by asking how many attendees would be happy with a patient achieving 80% Time in Range, but spending the remaining 20% of time <70 mg/dL to which the audience expressed substantial concern leading Dr. Hirsch to further emphasize the importance of using Time in Range alongside additional CGM metrics. Similarly, Dr. Hirsch discussed the benefit of Time in Range alongside CGM-derived GMI instead of A1c, which he argued provides a more accurate assessment of mean glucose and thus serves as a better predictor of long-term complication risk. That said, Dr. Hirsch did recognzie that research in this arena, while growing, is not as robust as the evidence correlating A1c and complications and expressed his view that dedicating further resources to understanding the relationship between, Time in Range, GMI, and outcomes will be critical to expand the acceptance of Time in Range in the clinical management of diabetes.
    • During Q&A, Dr. Elizabeth Holt (LifeScan) noted that, due to the design of AGPs, patients and providers almost never see isolated Time in Range data and more frequently see the “Time in Range Bar” outlining Time in Range alongside Time Below Range. Given this presentation of the data, Dr. Holt took slight issue with Dr. Hirsch’s assertion that Time in Range is an “incomplete number” instead suggesting that when considered holistically, the Time in Range bar can help providers quickly and easily identify both Time in Range and Time Below Range. We certainly appreciate Dr. Holt’s point and feel that both Dr. Holt and Dr. Hirsch were highlighting similar perspectives on the importance of evaluating glycemic management holistically to both identify areas where patients are doing well and also make adjustments as necessary to help patients achieve improved glycemic management and outcomes.
  • Finally, on GRI, Dr. Hirsch used case studies to provide examples of where this novel metric, while an interesting and potentially useful assessment of both hypo- and hyperglycemia exposure, may not accurately represent risk at an individual level. Across three cases, two in older adults and one in a young adult hoping to become pregnant, all of whom had concerningly high GRI values but A1cs close to target ranging from 6.8% to 7.5%, Dr. Hirsch demonstrated times when there may be additional factors outside of consensus guidelines for hypo- and hyperglycemia exposure that should be considered. Specifically, among older adults with comorbidities such as CKD, Dr. Hirsch explained that hyperglycemia is much less of a concern than hypoglycemia and that GRI may overestimate the risk posted from hyperglycemia exposure. Based on this assessment, Dr. Hirsch argued that while he views GRI as “an important new metric” for “adult, otherwise, non-frail patients” it may not have as many applications for older patients or those with comorbidities. These are important perspectives, and while they may result in less “standardization” and more “customization,” which is also more time-consuming in practices, it also is the personalization that is so helpful to patients.
    • During Q&A, many audience members raised questions regarding the GRI in clinical practice – it’s always fantastic to see so many patients eager to learn here. Of note, Dr. Tadej Battelino called into question how actional GRI is as providers will need to reference both a patient’s GRI and their AGP in order to see exactly how much hypo and/or hyperglycemia they are experiencing and make clinical decisions accordingly. Similarly, other attendees raised concerns around whether introducing a new metric would confuse patients and providers to which Dr. Hirsch responded that he does not envision GRI making it into primary care for a long time, and that currently he sees this as another way to visualize glycemic data alongside A1c and Time in Range and may be of special interest to some patients and less relevant to others. From our perspective, we are unsure if GRI will have clinical utility to help inform treatment, but we do think it can be an interesting and valuable tool to help patients track changes in glycemic management over time and could also prove useful for evaluating large population-level datasets along hypoglycemia and hyperglycemia exposure parameters.

FEEL-T1D study finds no correlation between time in consensus range and short-term wellbeing but does see correlation between time within 50 mg/dL of mean glucose and short-term wellbeing

In a session on the correlations between CGM metrics and daily psychosocial outcomes, Dr. Elizabeth Pyatak (University of Southern California) read out the results of the Function and Emotion in Everyday Life with T1D (FEEL-T1D) study, which examined whether people with diabetes “feel better” in their “habituated range” or in the consensus target range of 70-180 mg/dl. Based on the study protocol, a person with diabetes’ “habituated range” is 50 mg/dL below and above a person’s mean glucose value based on 14 days of blinded Libre Pro 2 CGM data. Glucose levels below 50 mg/dl from the patient’s average are considered “below habituated range,” whereas blood glucose levels higher than 50 mg/dl above the patient’s average are considered “above habituated range.” Throughout the 14-day study, patients wore a blinded Libre Pro CGM (reprocessed using Libre2 algorithm) and took 5 to 6 surveys daily assessing physical, mental, and cognitive well-being. Dr. Pyatak presented only a subset of the data (n=103), as data collection has just wrapped up and has not been fully analyzed.

  • The preliminary data suggest that people report better short-term well-being when in their “habituated range” but saw no improvements when in the consensus range (70 mg/dL-180 mg/dL). Although there was no correlation between being in the consensus range and psychosocial outcomes, participants reported significantly lower levels of fatigue and pain, recorded decreased negative affect, and demonstrated faster reaction times when in their habituated range for three hours.

  • Dr. Pyatak concluded by emphasizing that time spent in a patient’s “habituated range” is the most important indicator of momentary well-being. Dr. Pyatak indicates that these results are novel in that they are “near-real-time” and allow for a greater understanding of within-person changes at a “granular level” compared to more static aggregate time periods used in previous studies. Dr. Pyatak stressed that clinicians must consider the trade-off between a patient’s “habituated range” and consensus range for creating individualized target ranges that weighs daily functioning against risk for longer-term complications. Dr. Pyatak hopes this data results in more tailored interventions that optimize well-being at an individual and real-time level while also addressing the long-term complication risks associated with spending time outside the consensus range.

Posters – Time in Range and CGM-Derived Metrics

Title

Authors

Details + Takeaways

Time in Range as a Clinical Outcome – Results of a Longitudinal Analysis of the Literature and Clinical Trials

Pranav M. Patel, Richard Abaniel, Natasha Dogra, Marie Frazzitta, Naunihal Virdi

  • Longitudinal review of literature and clinical trials on Time in Range
  • Time in Range was reported in 1,561 publications and 401 clinical trials between 2012 and 2021; twelve-fold increase in publications from 2017 to 2021; 240 clinical trials involving Time in Range expected between 2021 and 2025

Ambulatory Glucose Profile Informs Better Treatment Decisions for Type 2 Basal-Insulin Patients

Eileen Huang, Mohamed Nada, Eugene Wright, Jr.

  • A perspective, non-randomized, single-arm, multicenter study on using the AGP report to inform treatment decisions in type 2s on basal-only therapy
  • Participants (n=105) used Professional CGM (FreeStyle Libre Pro) for up to 15 days; providers were given training on AGP interpretation
  • Based on AGPs, 94% of patients were given a treatment adjustment; most common change (40%) was an increase in basal insulin doses
  • Most patients (97%) had a clear understanding of the rationale behind their treatment change
  • AGP provided new insights as reported by providers in the study, namely: Time Above Range (84%) and overall glucose variability (58%); these insights suggest that CGM can help providers drive changes in treatment plans

Association between Daytime vs. Nighttime Mean Glucose and Time-in-Range with A1c in Adults with Type 1 Diabetes

Viral Shah, Timothy B. Vigers, Laura Pyle, Halis K. Akturk, David C. Klonoff

  • Real-world study on the relationship between nighttime vs. daytime CGM metrics and A1c levels in adult type 1s (n=340; diabetes duration >two years; use of Dexcom G6 for greater than six months)
  • Higher A1cs correlated with increased mean sensor glucose and reduced Time in Range for daytime and nighttime (p<0.001 for both); no difference between daytime and nighttime Time in Range or mean glucose across A1c levels (p=0.08 and p=0.42, respectively)

Fear of Hypoglycemia, Hypoglycemia Confidence, and Time in Range in Adolescents and Young Adults with Type 1 Diabetes

Sruthi Menon, Daniel Desalvo, Madhav Erraguntla, Kalyan Pamidimukkala, Siripoom Mckay, Carolina Villegas, Marisa E. Hilliard

  • Exploratory study wherein young adult participants (n=20, ages 13-21) completed questionnaires assessing fear of hypoglycemia and hypoglycemia confidence, which were matched with CGM data over twelve weeks
  • Fear of hypoglycemia was associated with less Time Below Range (<70 mg/dL, p=0.05); no other correlations were statistically significant

Long-Term Health Benefit and Economic Return of Time in Range (TIR) Improvement in Individuals with Type 2 Diabetes

Khalid Alkhuzam, Lizheng Shi, Vivian Fonseca, Yongkang Zhang, Jingchuan Guo, Hui Shao

  • Simulation modeling study to quantify the long-term health benefits and cost effectiveness of improving Time in Range for people with type 2 diabetes
  • The model assumed no recurrence of cardiovascular disease and constant Time in Range throughout each simulation
  • The estimated annual spending on the treatment necessary to improve Time in Range from less than 50% to (i) 51-70%; (ii) 71-85%; and (iii) >85% was estimated to be (i)$5,604; (ii) $8,246; and (iii) $9,067, respectively

Awareness of Time in Range Varies by Type of Health Care Provider

Julia Kenney, Andrew Briskin, Jacqueline Tait, Erik Shoger, Richard Wood, Anne L. Peters

  • Provider survey (n=303) on discussing diabetes management with patients
  • Diabetes educators (n=106) were more likely to be aware of (96%) and use (94%) Time in Range than endocrinologists (n=98; 92% and 88%, respectively)
  • Only 56% (!) of primary care providers (n=99) were aware of Time in Range and only 46% utilize it in clinical practice
  • Endocrinologists were more likely to use Time in Range to make treatment decisions (87%) than primary care providers (69%),  emphasizing opportunities for increased education and training for primary care providers

TBR (Time Below Range) in Routine Clinical Practice: A Retrospective Analysis of Routine CGM in Outpatient Care

Purvi M. Chawla, Manoj S. Chawla

  • Single center study on the use of is-CGM as a proxy for CGM in type 1s and 2s (n=227; unclear how many type 1s vs. type 2s)
  • Participants were categorized based on A1c: 18% were <7%; 23% were between 7% and 8%; 34% were between 8% and 9%; and 26% were ≥9%)
  • Those with A1c levels below 7% spent the most amount of Time Below Range at 4.6 hours/day (p=0.0016); participants with A1c levels between 7% and 9% spent an average of 2.3 hours/day Below Range
  • No differences in CGM-derived metrics were found between male (n=132) and female (n=95) participants or between different treatment regimens (i.e., insulin-only (n=25); oral meds-only (n=55); or oral meds + insulin (n=146))

Will Choice of Treatment Influence TIR in T2D with Diabetic Complications?

Jothydev Kesavadev, Banshi D. Saboo, Arun Shankar, Gopika Krishnan, Geethu Sanal, Anjana Basanth, Sunitha Jothydev

  • Study of association between CGM-derived metrics and diabetes complications in type 2s (n=1,218; mean age 53; 69% male)
  • Moderate (r=0.6) correlation between A1c and estimated A1c values; 68% of participants achieved a target Time in Range of >70% without complications; 70% of participants with vascular comorbidities achieved >50% Time in Range
  • For those with vascular complications, the use of analogue basal-bolus regimens was associated with Time in Range >50% and Time Below Range <1% (p<0.05 for both)

Time in Range and Time Below Range in Insulin-Treated Older Adults with Type 2 Diabetes

Silmara A.O. Leite, Michael Silva, Ana C.R. Lavalle, Murilo Bastos, Maria C. Bertogy, Suelen C. Vieira, Guillermo E. Umpierrez

  • Prospective observational cohort study on the use of CGM in older type 2s
  • Participants (n=66; >65 years; on insulin therapy; A1c between 7% and 9%) were placed on FreeStyle Libre is-CGM for six weeks
  • Participants achieved a mean +43 minutes/day in Range, from a baseline of 67% to 70%
  • Mean Time Below Range decreased by 29 minutes/day from 4% at baseline to 2% at six weeks

Acceptance of Decision Support Recommendations Improves Time in Range for People living with Type 1 Diabetes

Jessica R. Castle, Alejandro Z. Espinoza, Nichole S. Tyler, Leah M. Wilson, Clara M. Mosquera-Lopez, Joseph Pinsonault, Robert Dodier, Sos M. Oganessian, Deborah Branigan, Virginia Gabo, Jae H. Eom, Joseph El Youssef, Katrina Ramsey, Taisa Kushner, Kerri Winters-Stone, Joseph A. Cafazzo, Peter G. Jacobs

  • Participants with type 1 diabetes (n=25) used DailyDose, a mobile app-based insulin decision support system, for eight weeks
  • Time in Range increased by +1.5 hours/day per week when more than half of the application’s recommendations were accepted (p=0.001); however, cohort-level Time in Range changes were not statistically different from a baseline of 47% to 51% after eight weeks of DailyDose use (p=0.25)

Correlation of Derived Time in Range (dTIR) and Time in Range (TIR) in People with Type 2 Diabetes (T2D) Treated with IDegLira (IDL), Degludec, or Liraglutide: A post-hoc analysis of the DUAL I Trial

Athena Philis-Tsimikas, John M. Dcruz, Ramsathish Sivarathinasami, Christophe De Block

  • Retrospective analysis of type 2s with uncontrolled diabetes on oral antidiabetic drugs (n=260) assessing correlation between derived Time in Range (dTIR; a calculated estimate of Time in Range) and actual Time in Range
  • Strong correlation seen between dTIR and Time in Range at baseline (n=207, r=0.88); similar correlation between change in Time in Range and change in dTIR from baseline to the end of treatment (n=137, r=0.77)
  • dTIR may be a useful proxy for Time in Range when CGM data are unavailable

School-Day Routines Positively Impact Time in Range for Children with Type 1 Diabetes

Christine March, James K. Lutz, Kwonho Jeong, Linda M. Siminerio, Scott D. Rothenberger, Elizabeth Miller, Ingrid Libman

  • Cross-sectional analysis of CGM-derived metrics from children during school hours across virtual and in-person instruction modalities
  • Youth with >70% CGM use were included (n=209); mean age of 10 years; 96% white; mean A1c of 7.5% and 64% on insulin pumps at baseline
  • During virtual instruction, participants spent +12 minutes/day in Range vs. in-person schooling (52% versus 51%; p<0.001); throughout virtual schooling, participants spent nine less  minutes/day Above Range compared to during in-person schooling (48% vs. 47%; p<0.001)

Big Picture of Diabetes Technology

Tech doesn’t fix all: Hypoglycemia and glycemic management challenges persist despite tech use in T1D Exchange cohort; insulin delivery technology does not mitigate glycemic fluctuations during exercise in T1-DEXI analysis

Despite exciting advancements presented in oral presentation sessions on glucose monitoring and insulin delivery, these sessions also provided reminders that diabetes technology is not a panacea. In the oral presentation session on glucose monitoring, Dr. Jeremy Pettus (UCSD) read out results from a survey of adults in the T1D Exchange (n=2,044, average age 43, 6.9% A1c) who had had diabetes for at least two years, which assessed A1c, impaired awareness of hypoglycemia (IAH), and severe hypoglycemia events (92-OR). The analysis showed that although these outcomes improved with CGM and AID, there were still gaps in glycemic management and hypoglycemia among tech users. Elsewhere, in the oral presentation session on insulin delivery, Dr. Michael Rickels (University of Pennsylvania) read out analysis from the T1-DEXI study on factors showing that advanced insulin delivery technology (e.g., AID) did not mitigate glycemic fluctuations during exercise (292-OR). Together, these findings are a clear reminder that although massive strides have been made with diabetes technology, current technology does not solve all diabetes management challenges, and further innovation – as well as expanded investment in education and support – are necessary to improve diabetes outcomes for all.

  • The T1D Exchange survey analysis clearly illustrated the glycemic benefits of CGM and AID technology, but also showed that even with this advanced technology, people with diabetes still are not achieving targets and are experiencing hypoglycemia-related challenges. The analysis included 169 non-CGM users and 1,875 CGM users, of whom 339 (18%) used MDI, 574 (28%) used pump therapy, and 953 (47%) used an AID system. Dr. Pettus described these participants as “the best of the best” – those that are part of the T1D Exchange network, who are plugged in with endocrinologists, and who are almost entirely (92%) using at least CGM technology. While in many cases this would be a detriment to the study design, Dr. Pettus framed it as a benefit, as it allows for a look at how these high-resource, tech-using folks with low A1c values are doing, noting that if they’re struggling, others are very likely struggling as well. Overall, outcomes were better for those using CGM and more advanced insulin delivery methods but were still not without issues. A majority (60%) of BGM users were not achieving an A1c <7%, and while better in those using diabetes technology, 40% were still not achieving an A1c <7% on CGM and 36% were not achieving this goal on AID technology. Furthermore, among those who provided CGM data (n=926, 652-P), about 40% of CGM users and 30% of AID users did not achieve a Time in Range >70%. Turning to hypoglycemia, the average number of severe hypoglycemia events (SHEs) saw similar outcomes across groups with the mean number of SHEs lower in those using CGM (1.14/year) than BGM (1.83/year) but still occurring in those using CGM and in those using AID systems (0.82/year). Likewise, while significantly lower among CGM users than BGM users, the proportion of participants experiencing ≥1 and ≥2 SHE in the prior year was still high among CGM users (19% and 11%, respectively). Among BGM users, these figures were 31% and 25%, respectively. Results for impaired hypoglycemia awareness were even more disheartening, with no difference in IAH between those on BGM vs. CGM and no differences between insulin delivery methods among CGM users. Across the board, about 30% of participants had impaired hypoglycemia awareness, signaling that this is a major issue regardless of diabetes technology use.
<

Data from entire study (92-OR)

 

BGM users

CGM users

AID users (subgroup of CGM users)

Proportion with A1c >7%

61%

40%

36%

Proportion with impaired hypoglycemia awareness

26%

31%