September 16-20, 2019; Barcelona, Spain; Day #4 Highlights – Draft

Hola from Barcelona, where EASD 2019 is soon approaching the finish line. Be sure to catch up on all the learnings that have happened so far by perusing our highlights reports from:

Day #1 – Excitement abound for DAPA-HF; Commentary on SGLT-2s vs. GLP-1s on renal outcomes; Dr. Bergenstal on TIR; Dexcom decision support pipeline update

Day #2 – SUSTAIN 8 full results; inTandem post-hoc of BMI >27; Medtronic Envision Pro professional CGM – factory cal, fully disposable, Bluetooth; strong OpenAPS data; dT Solvable Problems

Day #3 – Results from VERIFY (Galvus early combo treatment) and CONCLUDE (Tresiba v. Toujeo); Dr. Raz on broader SGLT use; N=765 DCCT/EDIC participants wear FreeStyle Libre Pro – 9% meet time-in-range goal; NN smart pens in exhibit hall

Diabetes Therapy Highlights

1. DAPA-HF from the Endocrinologist’s Perspective: Dr. Silvio Inzucchi Breaks Down Divergent Metabolic Outcomes in People w/ and w/o Diabetes, Solidifying Glucose Independent Mechanism of CV Benefit

Yale’s Dr. Silvio Inzucchi presented metabolic outcomes from DAPA-HF, showing divergent effects in people with and without baseline type 2 diabetes and reinforcing the notion that cardiovascular benefits of SGLT-2s have little to do with glucose lowering effects.  As a reminder, DAPA-HF enrolled 45% of its participants with type 2 diabetes (3% were previously undiagnosed), with the rest having prediabetes or no diabetes status. Importantly, treatment effect in terms of the primary endpoint (CV death/HF hospitalization/urgent HF visit) was consistent across both subgroups of people with and without diabetes. Metabolic outcomes between these two groups told a different story, however. On A1c, people with diabetes saw a 0.2% drop compared to placebo at 8 months (the lower A1c drop than normally expected was attributed to lower eGFR in the patient population), while people without diabetes barely saw any difference at all with placebo (see graphs below). Weight outcomes told a similar story, with significant reductions in the diabetes group and nearly identical curves in the group without diabetes (see graphs below). Conversely, while effects on weight and A1c were different between the two groups, there were similar effects on blood pressure and hematocrit levels. Based off of these findings, Dr. Inzucchi posited that there may be a divergence of hemodynamic effects vs. metabolic effects in these two populations, with the possibility that consistent hemodynamic effects across subgroups could be driving the consistent cardiovascular benefit.

• Dr. Inzucchi answered the question that appeared to be on the minds of many endocrinologists in attendance: “Have we ‘lost’ this drug class to cardiologists?” Indeed, in light of the numerous positive CVOTs for the class, now along with DAPA-HF results, we’ve sensed (without question) heightened attention and enthusiasm from cardiologists regarding SGLT-2s. Dr. Inzucchi acknowledged this and tempered these fears, noting that (i) yes, SGLT-2s will see increased focus in cardiology because of their effects on heart failure regardless of diabetes status, but such usage will require collaboration with endocrinologists; and (ii) 90% of the patients without baseline HF that still experience positive CV effects with the class are primarily seen by endocrinologists, and it will be the job of endocrinologists to treat these patients with risk factors in order to prevent HF from developing (or presumably worsening). After all, heart failure remains an underrecognized complication of diabetes, as it represents the complication with the largest excess risk of mortality—even larger than MI and other ischemic events. It will be the job the endocrinologist to recognize risk factors for heart failure early on and have conversations with patients about initiating SGLT-2 therapy before heart failure develops. (Of course, we could imagine cardiologists having the same conversations.)

• Excitingly, Dr. Inzucchi hinted that future analysis of DAPA-HF will look at subgroup effects in participants with prediabetes, along with potential evaluation of diabetes prevention with SGLT-2s. When asked during Q&A about the prediabetes cohort in DAPA-HF and whether any future analysis of this subgroup will be conducted, Dr. Inzucchi answered in the affirmative and also seemed to indicate that Farxiga’s potential in diabetes prevention could also be studied. We’re particularly intrigued by this possibility of SGLT-2s in people with prediabetes, and whether such usage could represent an efficient diabetes prevention strategy that also exerts important CV and renal protective effects. It’s quite notable that DAPA-HF represents the largest study to date that has used SGLT-2s in people without diabetes, and we hope to see further analysis comparing Farxiga’s effect in this population vs. in people with diabetes.

2. EMPRISE: Additional Real-World Data Shows 26-34% Risk Reduction for Jardiance vs. GLP-1 Agonists in Drug Initiators with Type 2

Dr. Elisabetta Patorno (Brigham & Women’s Hospital) gave an engaging poster presentation on EMPRISE, Lilly/BI’s real-world evidence program for SGLT-2 inhibitor Jardiance, detailing a 26-34% reduced risk of hospitalization for heart failure (hHF) vs. GLP-1 agonists in patients with type 2 diabetes. These findings were fairly consistent with hHF outcomes from EMPA-REG, which found a 35% risk reduction (HR = 0.65; 95% CI = 0.50-0.85; p=0.002) compared to placebo, on top of standard of care. Importantly, Dr. Patorno pointed out that EMPRISE is an ongoing study, and the results presented today are only representative of the two-year endpoint and participant enrollment number. Study participants (n=52,824) were pulled from two commercial (Optum Clinformatics; Truven Health MarketScan) and one federal (Medicare fee-for-service) database, then 1:1 propensity-score matched between empagliflozin initiators and GLP-1 initiators. Patients were ~60 years old, 25% had a history of CV disease, and ~5% had a history of HF. In terms of GLP-1 agonists being used, 56% of patients were on liraglutide, 24% on exenatide, 16% on dulaglutide, and 4% on albiglutide. Primary outcomes were (i) hHF-broad, a heart failure discharge diagnosis in any position; and (ii) hHF-specific, a heart failure discharge diagnosis in the primary position. In the broad category, a 26% risk reduction was seen (HR=0.74; 95% CI: 0.60 to 0.90), while the more specific definition gave a 34% risk reduction (HR=0.64; 95% CI: 0.40 to 0.996). Due to the small number of events, further sub-analysis by type of GLP-1 could not yet be performed. Generally, we’re glad to see such robust results in what appears to be a lower risk population than was enrolled in EMPA-REG.

• During Q&A, Dr. Patorno noted that additional results on MI and CV death in EMPRISE are to be reported at AHA 2019 in November.

• See past data from EMPRISE below:

  • CV efficacy and safety of empagliflozin vs. DPP-4 inhibitors (ADA 2019)

  • hHF benefit of empagliflozin vs. DPP-4 inhibitors (AHA 2018)

3. Prof. Philip Home’s Independent Commentary on CAROLINA: “Beautiful” Clinically Relevant Study; CAROLINA Published in JAMA this AM with Editorial from Dr. Wexler

Even for those who had seen the presentation of the CAROLINA CVOT (n=6,033) –showing non-inferiority on three-point MACE for Lilly/BI’s DPP-4 inhibitor Tradjenta (linagliptin) vs. the sulfonylurea glimepiride – at ADA 2019, independent commentary from Newcastle’s Prof. Philip Home made attendance at the European presentation well worth it. Following a series of talks on the study rationale, design, results, and significance from the same exact lineup present at ADA (Drs. Julio Rosenstock, Mark Espeland, Steven Kahn, Nikolaus Marx, Bernie Zinman, and Darren McGuire), Prof. Home took to the podium to offer his interpretation, which was largely positive: “Sadly, as I really love to criticize Julio [Dr. Rosenstock], I actually loved CAROLINA from the moment I heard of its conception. The reason really being because it was an active comparator study pitting linagliptin against an active comparator.” The bullets below delineate Prof. Home’s more specific takeaways. Note, too, that CAROLINA was published just this morning in JAMA alongside an editorial by Dr. Deborah Wexler asserting that CAROLINA represents “one of the most important findings to emerge from this set of trials [post FDA 2008] regarding the CV safety of SUs” and that “the perennial concern regarding the CV safety of SUs raised by the UGDP more than 40 years ago may finally have been assuaged by CAROLINA trial.”

• Prof. Home lauded the study for its clinical relevance. First, it compared the two most commonly-used second-line therapies (and most probable in those with mild CV risk). Second, the enrolled population was more representative of the general diabetes population: “Most CVOTs don’t apply to 80%+ of the people we see in clinics…This trial event rate [2.1 per patient-year] is quite close to the unselected population in the RECORD study” (see slide below). Third, it included a reasonably ethnically and geographically diverse population, unlike other recent CVOTs; he specifically pointed to the fact that 18% and 17% of participants were Asian and Hispanic/Latino, respectively. That said, just 5% of the population was black. Fourth, median 6.3-year follow-up with median drug exposure of 5.9 years was appropriate to detect adverse event for a therapy likely to be used long term, and the power was adequate to give smooth, “non-erratic” Kaplan-Meier curves, unlike you’d see in “say, EMPA-REG [OUTCOME].”

  • One instance in which the study may have strayed from clinical relevance, in his view, is in the at-first-glance concerning data on recurrent hypoglycemia. As Dr. Zinman presented, the rate ratio of a second blood glucose ≤70 mg/dl or severe hypoglycemia was 0.09 (95% CI: 0.07-0.11; p=0.0001) for linagliptin vs. glimepiride, confirming that recurrent hypoglycemia was significantly more common with the SU. Prof. Home reasoned that this data is not reflective of real clinical practice, since providers wouldn’t keep people on SUs after one year if they had had two events. (If only that were true in all geographies! We believe heaps of patients go on SUs and stay on them due to a variety of factors – even though of course what Professor Home says is what should be done.)

• “I believe results can be extrapolated to other DPP-4 inhibitors, I see no reason why not. For other SUs, we’re still uncertain.” The independent commentator at ADA, Dr. Matthew Riddle, aligned with Prof. Home on lack of generalizability to other SUs, but was more cautious about applying the linagliptin data to the DPP-4 inhibitor class. Dr. McGuire, too, agreed that this study does not assuage concern of all SUs: “It may be that others would behave similarly, but that is not a complete certainty.” All that said, he cautioned that we still haven’t demonstrated glimepiride’s safety in a high-risk population.

• Though he appreciated the design and premise of CAROLINA, Prof. Home lamented the fact that UGDP trial influenced the field so strongly for so long, quipping that such hesitation was “frankly silly” given the miniscule number of events in UGDP. He went on to cite plentiful evidence from UKPDS, ADOPT, RECORD, ADVANCE, and TOSCA.IT that SUs are safe from a CV perspective (if not hypoglycemia and weight). “Albeit not from a direct study like this beautiful one, we were already reasonably confident these drugs were safe.”

  • Dr. McGuire noted, on the other hand, that cardiologists have been very concerned about SUs, “probably more so than diabetologists,” in the wake of UGDP. “Never have 36 events caused so much confusion and consternation. Here we are 50 years later, finally with an answer.”

• Prof. Home noted that the A1c trajectories were in line with expectations for both medications. In ADOPT and RECORD, he said, SUs were the most effective glucose-lowering drug early on, but they then lose their effect. Prof. Home noted that linagliptin had a similar impact on A1c over the first six to eight months of use as do metformin and rosiglitazone.

• Prof. Home who was a major advisor in the development of glimepiride, noted that the SU seems to have been titrated correctly in CAROLINA. By the end of the study, 66% of patients were on the maximum 4 mg/day dose, which he called “very reasonable.”  

• Echoing Dr. Zinman during his safety presentation, Prof. Home proclaimed: “I think the pancreatitis data puts the potential risks attributed to DPP-4 inhibitors to rest.”

• Prof. Home’s one complaint was that the reported lipid and A1c data stops at five years in the paper, whereas the CV, mortality, and other results extend to seven years. We’re not sure why this is the case, and the investigators did not respond. We’d like to see the additional two years of data as well – while some say they wouldn’t expect significant divergence, it’s hard to say what A1c data will show in terms of long-term durability without seeing it upfront and close. 

4. New REWIND Stroke Outcomes Data: Dulaglutide Reduces the Risk of Non-Fatal (But Not Fatal) Stroke and Protects Against Ischemia (But Not Hemorrhage)

Dr. Hertzel Gerstein unveiled new REWIND data, diving deep into stroke outcomes with Lilly’s GLP-1 agonist Trulicity (dulaglutide). Released only a few months ago at ADA 2019, the REWIND trial demonstrated a 12% relative risk reduction on 3-point MACE (HR=0.88, 95% CI: 0.79-0.99) with dulaglutide vs. placebo, with strong consistency between both the primary (69% of participants) and secondary (31%) prevention cohorts. Overall, dulaglutide showed a 24% relative risk reduction for fatal and non-fatal stroke (HR=0.76, 95% CI: 0.62-0.94; p=0.010). This effect seems to be driven primarily by nonfatal stroke (HR = 0.76, 95% CI: 0.61-0.95; p=0.017), as there was no significant effect of dulaglutide on fatal stroke (HR = 0.78, 95% CI: 0.47-1.30; p=0.34), and only a very small number of fatal strokes occurred in the trial. Furthermore, dulaglutide’s protective effects apply only to ischemic stroke (HR: 0.74, 95% CI: 0.59-0.94; p=0.012), not hemorrhagic stroke (HR: 1.05, 95% CI: 0.55-1.99). According to Dr. Gerstein, although the mechanistic insights that can be gleaned from CVOTs are limited, this finding suggests that dulaglutide’s protective effects relate to ischemia, or diminished blood supply, rather than ruptured blood vessels. The investigators additionally examined the impact of dulaglutide on “disabling stroke” – that is strokes with a modified Rankin severity score ≥3. Interestingly, dulaglutide showed an impressive 26% risk reduction for disabling stroke (HR: 0.74, 95% CI: 0.56-0.99; p=0.042), but the overall distribution of Rankin scores between the strokes that occurred in the dulaglutide and placebo groups were not significantly different (p=0.18). This indicates that while dulaglutide does reduce the risk of disabling strokes (and strokes in general) from occurring, if a stroke does occur the drug does nothing to reduce its severity. As Dr. Gerstein put it “there is a comparable distribution of stroke severity if you’re unfortunate enough to have a stroke, regardless of which drug you were on.” Finally, in exploratory subgroup analyses there appeared to be no heterogeneity in dulaglutide’s stroke-protective effects according to sex, BMI, A1c level, or absence vs. presence of existing CV disease (i.e. primary vs. secondary prevention).

• Highlighting a 2019 meta-analysis of all five reported GLP-1 agonist CVOTs, Dr. Gerstein pointed out that risk reduction for stroke was the highest of any MACE component (HR = 0.84, 95% CI: 0.76-0.93, p<0.0001). In fact, the only GLP-1 agonist not to show significant risk reduction for stroke was ELIXA (for Sanofi’s lixisenatide), although we’ve heard ample commentary that the entirely post-recent-ACS population enrolled in ELIXA was likely too sick for lixisenatide to confer any benefit (i.e., the population was ‘too far gone’). While the overall CV benefits of GLP-1 agonists are clear, Dr. Gerstein suggested that the especially potent effect on stroke risk reduction may point to a particular effect of GLP-1 agonists on the brain. (To this end, just yesterday we saw promising preclinical data on GLP-1 agonists for the treatment of neurodegenerative disease).

5. Prof. Stefano Del Prato’s Commentary on CONCLUDE Emphasizes Inconclusive Results of Tresiba vs. Toujeo Comparison Trials

Following presentation of results from CONCLUDE (Novo Nordisk sponsored trial of Tresiba vs. Toujeo in type 2s, see our coverage from EASD Day #3), Prof. Stefano Del Prato provided his independent commentary, underscoring the inconclusive nature of the trial and other sources of evidence comparing these two next-gen basals. Indeed, we’ve sensed that CONCLUDE has generated extensive debate in the community, as its non-significant primary endpoint but intriguing secondary endpoint results  have called into question (i) how exactly to interpret trials where the primary endpoint may not be met but secondary endpoints are; and (ii) how to reconcile these results with the BRIGHT study (Sanofi’s sponsored trial of Tresiba vs. Toujeo – obviously, of course, starting with sub-group analysis and recognizing the different populations). Another point of major discussion was the accuracy of the glucose meters used in CONCLUDE, which further complicate interpretation and analysis.

• Prof. Del Prato took the audience through a thought experiment while trying to comprehend results from CONCLUDE: assuming that results do indicate a lower hypoglycemia risk with Tresiba, could there be a valid explanation for this finding? One possible explanation could be varying PK/PD profiles for both of these agents which could help differentiate their hypoglycemia effects. Prof. Del Prato reviewed PK/PD studies for Tresiba and Toujeo, with conflicting results in this space preventing a clean conclusion one way or the other. Some studies have shown better lower day to day variability with Tresiba, whereas others have demonstrated lower daily fluctuations with Toujeo. He noted that there exist tremendous amounts of intrinsic variability in PK/PD studies, with many variables (ranging from different primary endpoint to different bioavailability’s and metabolic effects) complicating interpretation and comparison. Without clear results in these studies, Prof. Del Prato finds it hard to connect evidence to the claim that CONCLUDE’s results indicate a lower hypoglycemia risk for Tresiba.

• Could past data (namely the BRIGHT trial) help contextualize CONCLUDE results? Important differences between the two trials make conclusions difficult. Prof. Del Prato emphasized study population differences between CONCLUDE and BRIGHT (see table below). The two trials had different designs, different endpoints, different populations in terms of age, baseline A1c and eGFR, diabetes duration, and more. Real-world evidence between the two agents do not help to differentiate them either: both DELIVER D+ (lower rates of hypoglycemia with Toujeo) and CONFIRM (showed lower rates of hypoglycemia with Tresiba) suggested different conclusions – again, the population being examined makes a big difference. Although Prof. Del Prato elicited laughs from the audience by joking at the start of his presentation: “You would expect that a trial named CONCLUDE would be more conclusive with its results …” it is certainly true that there’s so much that is different with the trials!

• We very much appreciate the tremendous scientific and technical learnings that have been generated with CONCLUDE and BRIGHT in comparing Tresiba vs. Toujeo, particularly in different populations. Both of these insulins are highly efficacious when compared to their predecessors, and yet are still not reaching the totality of patients that can truly benefit most from them. Intra-class competition aside, we’re hoping for more investment in access and reimbursement for both of these therapies.

6. AZ’s 38-Nation Observational DISCOVER Trial Sheds Light on Global Heart Failure Trends

New data from AZ’s observational, >15,000 person, 38-nation DISCOVER trial reveals trends in the global prevalence of heart failure. DISCOVER is an observational longitudinal study of people with type 2 diabetes initiating second-line glucose-lowering therapy; to date, data has been collected for 2 years. This study determined that nearly 2% of the DISCOVER population had heart failure at baseline. The people with HF tended to be older (67 vs. 57 years), have a longer diabetes duration (8.2 vs. 5.5 years), and a greater prevalence of coronary artery disease (45% vs. 6%) and chronic kidney disease (23% vs. 4%). Furthermore, 60% of the DISCOVER population had the obesity-associated and harder-to-treat HFpEF. Heart failure guru Dr. Mikhail Kosiborod has emphasized the uptick in HFpEF in epidemiological studies, in line with the global obesity epidemic. This is a question from a clinical point of view, as none of the therapies approved for HFrEF have meaningful effects in HFpEF. Over the course of two years follow-up, 2.6% of DISCOVER participants developed heart failure, but this proportion rose to 15.3% for the subgroup of participants with baseline coronary artery disease. The speaker, Dr. Suzanne Arnold (University of Missouri, Kansas City, MO), elaborated that among these incident heart failure events, 74% were diagnosed in the outpatient setting and 56% in the absence of coronary artery disease. (Dr. Arnold commented that in real-world clinical practice  most heart failure diagnoses are made in an outpatient setting, so other trials that assess only HF hospitalizations are likely underestimating the true incidence.) One major limitation of the DISCOVER study is variation in screening for diabetes complications between the different nations involved, not to mention heterogeneity in the patient populations enrolled from each nation. That said, we admire the richness of this data set, and are troubled by the comparatively high rates of heart failure in such a relatively young population (mean=56.6 years old) with a relatively short duration of diabetes (mean=5.7 years).

• The ambitious DISCOVER program is proving to be a wealth of information about global trends in diabetes complications. At ADA 2019 a separate DISCOVER analysis revealed that people with type 2 diabetes initiating a second-line glucose-lowering therapy had a strikingly high prevalence of microvascular (27%) and macrovascular (15%) complications after two years, and people with macrovascular complications at baseline were at an extremely high likelihood for recurrent events (OR=382, 95% CI: 2.47-5.90).

7. Post-Hoc Analysis of DURATION-8 Trial Hints at NAFLD/NASH Benefits with GLP-1/SGLT-2 Combination Therapy

A post-hoc analysis of the DURATION-8 trial showed improvements in liver fat and fibrosis with combination therapy of the GLP-1 agonist exenatide and SGLT-2 inhibitor dapagliflozin. As a reminder, DURATION-8 compared exenatide/dapagliflozin combination therapy against both monotherapies alone, finding significantly greater A1c reductions and weight loss with two agents vs. one over two years follow-up. At baseline, >81% of the DURATION-8 population (mean age 55 years; mean A1c 9.3%; mean BMI 33 kg/m²) had fatty liver/steatosis and >9% had liver fibrosis (a more severe stage of the disease). In participants on exenatide/dapagliflozin combination therapy, the prevalence of patients with biomarker scores suggestive of fatty liver/steatosis fell by 6.1% to 10.5% after 28 weeks of therapy. Also at this time point, the proportion of patients with liver biomarker scores indicating the presence of advanced fibrosis fell 4% for those on the combination therapy regimen.

• Several individual GLP-1 agonists and SGLT-2 inhibitors have been investigated in NAFLD/NASH, but to our knowledge this is the first large-scale examination of GLP-1/SGLT-2 combination therapy. The study author, Dr. Cristian Guja (Carol Davila University of Medicine and Pharmacy, Bucharest, Romania) underscored that DURATION-8 wasn’t powered to investigate NAFLD/NASH outcomes and didn’t include any histological examination or imaging of the liver, so these findings are by no means definitive and must be validated in a dedicated prospective trial. That said, we find the results very encouraging. More than half of people with type 2 diabetes and obesity are estimated to have NAFLD, so it is a very attractive proposition for diabetes combination therapy to double as therapy for liver disease. Of course there is also no shortage of novel agents in development for NAFLD/NASH in the ever-expanding competitive landscape.  

8. Prof. Filip Knop’s Minkowski Lecture Flips Understanding of Hyperglucagonemia on Its Head; Alpha Cell Insulin-Resistance Hypothesis May be Wrong

University of Copenhagen’s Prof. Filip Knop wowed with a Minkowski Prize Lecture during which he presented his team’s convincing work that challenges conventional thinking about hyperglucagonemia in type 2 diabetes. While the traditional explanation for hyperglucagonemia is a combination of impaired insulin secretion and the alpha cell growing resistant to the glucagon-suppressive effect of  insulin, Prof. Knop put a “big fat question mark next to these explanations,” which he doesn’t believe. Rather, he proposed dividing hyperglucagonemia into fasting hyperglucagonemia and post-absorptive hyperglucagonemia, which his group’s data suggests are the result of distinct processes. In his well-supported framework, elevated fasting plasma glucose is a result of a disruption of the liver-alpha cell axis while post-absorptive hyperglucagonemia may be caused by gut-derived glucagon secretion. The liver-alpha cell hypothesis hinges on the idea that liver steatosis induces glucagon resistance at the level of ureagenesis, resulting in a rise in the concentration of circulating “glucagotropic” amino acids, which then signal the alpha cell to secrete more glucagon. The hypothesis has been supported by a number of physiological studies, as well as transcriptomics, which showed that genes associated with amino acid transport are downregulated in steatotic livers, as are enzymes important in ureagenesis. Taken together, steatosis may contribute to hyperglucagonemia and hyperglycemia in type 2 diabetes, and NAFLD/NASH therapies may have positive externalities on hyperglucagonemia (our speculation). As for the post-absorptive state, researchers noted that while healthy controls can suppress glucagon in response to OGTT and isoglycemic IV infusion, type 2s do not have the ability to suppress the glucagon response during the OGTT. This suggested that hyperglucagonemia in the post-absorptive state could be a result of something going on in the gut. Prof. Knop and colleagues elegantly showed that – contrary to canon – glucagon may be secreted by gut cells, and this secretion may be regulated differently from that from pancreatic alpha cells. Prof. Knop said that he’s not yet sure how these new insights could be used at this point, but they present potential therapeutic targets (in type 1 and type 2 diabetes).

Diabetes Technology Highlights

1. Dr. Anne Peters on the upcoming ADA-EASD Joint Position Statement on Diabetes Apps

The extraordinary Dr. Anne Peters (Keck School of Medicine) outlined key points from the soon-to-be-published ADA-EASD joint position statement on diabetes apps, which was put together by ADA-EASD Diabetes Technology Group’s (AEDTG) Drs. Alexander Fleming, John Petrie, Rich Bergenstal, Reinhard Holl, Anne Peters, and Lutz Heinemann. The full statement will be published in Diabetes Care and Diabetologia soon, offering a summary of the benefits, challenges, and recommendations in this fast-evolving area. This follows similar ADA-EASD technology statements on CGM (2017) and pumps (2014). According to Dr. Peters, “The aim of what we did was to look at the current landscape of available diabetes digital health technology, the practices of regulatory agencies and organizations, and then go through the next steps that we believe should be taken.” The statement will be broad-reaching, including guidance for regulators, industry, professional organizations, researchers, funders, healthcare providers, and patients. See below for some of Dr. Peters’ comments, and we’ll alert readers when the statement is online.

On Obstacles Facing Health Apps

• “Nutrition apps seem simple right? They’re databases of carbs, fat, protein, and calorie content. People use them, they count what they’re eating, they count their carbs, but they’re not all the same. In fact, I’m not sure that all of them are written by people who really know the carbohydrate, fat, and protein content of certain foods. When I start comparing them, I find differences. Clinically, I don’t know how to help a patient choose which app is best.”

• “[Physical activity apps] are really good motivationally, but … I see patients come in feeling great with their Fitbit and their steps, but six months later, it’s in a drawer and they’re not using it. To me, everything about diabetes management is for the long haul – you can’t just do it and then just quit it.”

• “[Glucose-monitoring apps] are incredibly important and helpful when people look at the data. But, even with these sorts of data-displaying apps, you have to ask the question, ‘Who gets to set the target levels?’ I have patients who target glucose levels that are too low and end up with too much hypoglycemia. I have patients who are way out of control and I want to actually step them down more slowly, so I need glucose-monitoring apps that let me help patients figure out what their individualized goals and targets should be.”

• “I remember for [my son’s] entire childhood, all I did was walk into his room and he was playing another video game. It wasn’t like he was playing five at the same time, he was addicted to one. I asked him to give me a list of all the games he played, and he said, ‘Mother, there are so many I can’t possibly tell you.’ This, of course, disturbed me greatly … These apps, these guys who are experts in gamification, make us interested for a while. Then you reach the peak of the mountain, you get the prize, you beat everybody, and then you’re bored, so you move onto your next app. I think we need to be mindful of the long-term benefit we’re offering patients and how we can make it sustain their interest.”

On “Digiceuticals” (Prescription Apps)

• “I really like these, I think they can work. These are one step more advanced: they require a physician’s prescription, at least in the US. They’re regulated like drugs or medical devices. I can prescribe these to a patient who I think needs one.”

• “It’s not just diabetes that has digiceuticals. There are all sorts of others. Can’t sleep? Try Sleepio. Daylight is about worry and anxiety…Propeller for treating COPD/asthma. Ginger.IO has behavioral support, and Reset/ResetO for addiction. So, you can imagine a person might have three or four programs they’re supposed to be following. At night, they pull out their insomnia app; during the day, the look at their anxiety app, their diabetes app, and then you start to get app overload.”

• “What happens if [advice from] your insomnia app starts arguing with your anxiety app? Then you could get into real trouble … What we really want to do is have effective tools that last over the long haul.”

• “There’s not a lot of [clinical evidence] that looks at these apps in terms of long-term benefit … It’s hard to judge apps the way we judge a real pharmaceutical, because apps aren’t frozen in time. In fact, they evolve over time. We want them to evolve over time, so we have to study these as they evolve and see what the benefit is for patients.”

2. Nemaura Medical’s 24-Hour, Non-Invasive sugarBEAT CGM “Is Not Vaporware,” Launch in UK Delayed to 2020; De Novo Submission to FDA Aims for class II (510k) regulation

UK-based Nemaura Medical, maker of the 24-hour, non-invasive CGM sugarBEAT, made its first EASD appearance with a symposium Thursday night hosted by CEO Dr. Faz Chowdhury and Global Commercial Officer Dr. Fred Schaebsdau. The system was CE Marked in June. Skeptics in the room were addressed early on, with Dr. Schaebsdau stating, “It’s not vaporware” while rolling up his sleeve to display the sugarBEAT on his upper arm. Later on, Dr. Schaebsdau called the device a “real-time CGM, full stop,” which has a claimed MARD of 11.9%, once-daily fingerstick calibration, and adjunctive CE-Marking for adults. Like all CGMs, sugarBEAT combines the glucose measurement with a “predictive algorithm” that produces an estimated blood glucose value. Dr. Chowdhury claimed that this may help reduce lag-time seen between interstitial and blood glucose levels. The device has a warm-up time of “20-25 minutes” and transmits glucose readings to a smartphone every five minutes using a rechargeable Bluetooth transmitter. The app will also allow for absolute and predictive alarms (“up to 20 minutes” in advance). The app will not have remote monitoring in its first generation. Dr. Chowdhury did show a sugarBEAT Gen II on a product pipeline slide, which it hopes to bring to pediatric and gestational diabetes markets. sugarBEAT will be launched starting with the UK “in 2020,” already a delay from the June plan to launch in “3Q19.” The device will initially be “self-purchase” (cash pay), as Nemaura has not secured reimbursement. In the past, the company has claimed the device will be at a price “comparable to the cost of using glucose meters and strips.” On the US side, the company claimed to have submitted sugarBEAT to the FDA in July under a de novo application, with the goal of being regulated under the 510(k) pathway. According to Dr. Chowdhury, the company has more than enough cash to get through UK launch and beyond.

• Nemaura intends to use the 24-hour wear-time to market “flexibility” for type 2 diabetes market. The company’s focus will be on “non-intensive medical applications,” which they deemed to include weight management, people with prediabetes, and people with type 2 diabetes not on intensive insulin therapy. To discuss clinical applications of sugarBEAT, the company brought in UK-physician Dr. Iain Cranston, who emphasized how underserved the type 2 population is with respect to “data-led” personal disease management. Further, Dr. Cranston noted that type 2s on “simpler” insulin regimens (i.e. basal-only) have lower day-to-day glucose variability and only need a few days of data to see patterns. Lastly, Dr. Cranston touched on the need for much more research into optimal uses of short-term CGMs and its relation to “consensus-accepted outputs” (e.g., AGP).

• During the “Patient user experience” portion of the symposium, a senior product engineer from Nemaura shared that she actually discovered she had type 2 diabetes after seeing abnormal traces when testing sugarBEAT on herself. While she had no symptoms, her diabetes status was confirmed by an oral glucose tolerance test. Another story highlighted someone with type 2 diabetes on basal insulin since 2013. She was recently prescribed Trulicity and now wears sugarBEAT as a tool for retrospective therapy adjustment.

• Nemaura is advertising the sugarBEAT CGM as non-invasive, safe, hypo-allergenic, environmentally friendly, flexible-use, and affordable. In a memorable demonstration of sugarBEAT’s environmentally-friendliness, Dr. Schaebsdau noted sugarBEAT didn’t have an inserter or excessive sensor packaging, as he threw a FreeStyle Libre inserter and Dexcom packaging into a trash can. During Q&A, CEO Dr. Chowdhury shared that the device had been worn for a collective ~1,200 days by ~200 people, with no reported allergic or skin reactions. He credited this to sugarBEAT’s silicon adhesive (as opposed to acrylic adhesives used in longer-wear CGMs), which he described as less “tough and aggressive.”

3. Basal-IQ Real World-Data in Over 8,000 Users: Time in Hypoglycemia (<70 mg/dl) Reduced from 3% to 1.8% After Upgrading; Corresponding TIR Bump from 61% to 62%

Halfway through EASD, Tandem announced publication of expanded real-world data from Basal-IQ (predictive low glucose suspend) in n=8,123 users in DT&T. The retrospective analysis collected data from t:connect (August 31, 2018 to March 14, 2019) in users with at least 21 days of system use. In the full data set (n=8,123), time-in-range (70-180 mg/dl) was 59% on Basal-IQ, with a median 1% time with glucose levels <70 mg/dl (mean: 1.7%) and just 0.1% time <54 mg/dl – excellent hypoglycemia elimination! A remarkable 31% of patients on Basal-IQ never had a CGM reading <50 mg/dl. Tandem also looked at two interesting subgroups: (i) a pre-post group that allowed a comparison of the impact of adding the Basal-IQ algorithm (n=1,371); and (ii) users who had at least nine full weeks of PLGS system data (n = 3,563).

• In the pre-post group, time spent in hypoglycemia (<70 mg/dl) went from 3% to 1.8% after users updated their pumps to add Basal-IQ; that comes out to ~17 extra minutes/day and a 45% relative risk reduction in hypoglycemia, and they were already on CGM at baseline. With the reduced hypoglycemia, time-in-range increased slightly from 62% to 63% in the pre-post group, while mean glucose remained stable at 168 mg/dl. Time spent <54 mg/dl dropped from 0.3% to 0.1% after using PLGS. Hypoglycemic episode frequency (<54 mg/dl for at least 15 minutes) was reduced from 0.11/day to 0.03/day (or one event every 9 days to one event every 30 days) after initiating Basal-IQ. See more pre-post outcomes in the table below, and see the full outcomes in the paper.

• On average, patients recorded 50 days using Basal-IQ and had the feature on ~98% of the time; age ranged from 6-90 years old, 50% were female, and 96% had type 1 diabetes.

• The reductions in hypoglycemia in real-world use exceeded reductions in the controlled pivotal study (read out ATTD 2018). In the pivotal, Basal-IQ delivered a 19 minute/day reduction in time <70 mg/dl, though from a higher baseline of 4.5%.

• In a related talk at EASD itself, Dr. Timothy Jones of Perth Children’s Hospital reviewed a variety of studies highlighting how CGM alone, CGM and pump, and CGM and pump with insulin suspension (threshold or predictive) can reduce hypoglycemia in patients with hypoglycemia unawareness. Additionally, these studies overwhelmingly found that alarms could reduce time in hypoglycemia regardless of insulin delivery method (i.e., MDI or pump). Notably, Dr. Jones noted a greater inclusion of higher-risk hypoglycemia unaware patients in randomized controlled trials over time – this . Among the notable studies referenced by Dr. Jones were HypoDE testing CGM in  MDIs with impaired hypoglycemia awareness (read out ATTD 2018), IN CONTROL testing CGM in MDI or pump users with impaired hypoglycemia awareness (read out at EASD 2016), and ASPIRE testing the MiniMed 530G with threshold suspend (Bergenstal et al., 2013).

4. SEARCH for Diabetes in Youth Study (n~7600) Shows DKA Prevalence in Kids Increase from 35% in 2010 to 41% in 2016

Wake Forest’s Dr. Elizabeth Jensen presented new data from the five-center SEARCH for Diabetes in Youth study, showing that the prevalence of DKA at type 1 diagnosis in kids <20 years old has increased from 35% in 2010 to 41% in 2016 (p=0.01). The sample of youth onset type 1 cases (n~7,600) represented a racially and economically diverse sample, allowing Dr. Jensen to explore prevalence changes in subgroups. All age groups saw a general increase in DKA prevalence, as did all racial/ethnic groups, all groups regardless of insurance status, and all sites regardless of clinical site. However, Dr. Jensen identified a trend (p=0.16) toward a greater uptick in South Carolina relative to the other sites (Barbara Davis Center, Cincinnati, Seattle Children’s, and Kaiser Permanente). Notably, accounting for shifting demographics in the SEARCH cohort – younger age at diagnosis and greater proportion of Hispanics – did not substantively change estimates observed for the association between calendar year and DKA. Dr. Jensen speculated that the concerning increase in DKA at diagnosis could be related to the US recession of 2008-2010 (decreased use of preventive care services) or the changing landscape of healthcare in the US, such as Medicaid expansion (notably, South Carolina was the only state of the five study sites that did not expand Medicaid, and also saw a potentially greater increase in prevalence). This also aligns with the general trend in type 1 diabetes of worsening outcomes in the teen/adolescent age group.

• In the subsequent talk, Sanofi’s Dr. Angela Ibald-Mulli described DKA risk factors that were identified in the German DPV type 1 diabetes registry using a “subgroup discovery algorithm.” There were a number of significant differences between those in the group that had DKA and those who didn’t: those who had DKA were younger (mean age 15 vs. 29 years), had higher A1c (9.3% vs. 8.0%), were younger at diagnosis (9 years vs. 15 years), had shorter disease duration (6 vs. 10 years), were more likely to have a “migration background” (19% vs. 12%), exercised less (0.76 vs. 1.5 supervised physical activity units per week), and had more frequent medical visits (42 vs. 23). The algorithm selected a number of “profiles” that placed individuals at high DKA risk: A1c >8.87%; age between 6-10 years; age between 11-15 years; diagnosis of nephropathy; DKA at onset; presence of hypoglycemia with coma; presence of severe hypoglycemia; diagnosis of thyroiditis; and no short-acting insulin. 89% of the patients with a DKA event in the DPV had at least one of these profiles; more importantly, <5% belong to 3+ profiles, but 20% of those that do have a DKA. This data should help to better stratify patients according to risk and avoid the extremely preventable acute complication of DKA. 

Exhibit Hall

Diabetes Technology, Take 2

See our Day #3 report for technology booths from Abbott, A. Menarini (WaveForm CGM), Dexcom, LifeScan, Insulet, Medtronic, Novo Nordisk, Roche, and SOOIL.

Diabeloop

Diabeloop’s understated booth featured its DBLG1 algorithm on a smartphone between a Dexcom G6 CGM and Kaleido patch pump. Diabeloop’s algorithm was CE Marked last November, and it had hoped to launch earlier this year; at this point, we believe it is still in fundraising/pre-launch mode, since the company has not announced a commercial launch Kaleido is only available in the UK (per its booth reps; see below). A demo of the Diabeloop’s controller (a locked down Android phone) showed the last CGM value and trend arrow on the lock screen in place of the time – a convenient way to check system status. The now 60-person team at Diabeloop is currently working on studies in children ages 6-13 and developing an algorithm that doesn’t require meal announcements. The company has already met with the FDA and intends to submit its algorithm to the FDA through an “iController” pathway, assuming one emerges. The pivotal trial will take place in France using DANA pumps, though usability tests will take place in the US.

EOFlow

Fresh off announcing a distribution partnership with A. Menarini Diagnostics, EOFlow had a big, bright booth near the center of the Fira Barcelona Exhibit Hall. The company has three projects in the pipeline (i) EOPatch, a disposable, 3.5-day patch pump, (ii) EOPatchX, which will bring in hybrid closed loop capability based on the InControl algorithm licensed from TypeZero (i.e., the same progenitor for Tandem’s Control-IQ), and (iii) EOPancreas, the company’s “dream,” an all-in-one patch with a CGM sensor and algorithm all built into the patch pump. The EOPatch has already “soft-launched” in South Korea. With the new agreement with A. Menarini, EOFlow now boasts two CGM partners: POCTech and A. Menarini. Dexcom could be considered a pseudo partner, as it does own TypeZero. Notably, we learned that SOOIL has been able to get smartphone-controlled pump dosing through South Korea’s regulatory agency and EOFlow expects to do be able to do the same in the future. Lastly, the “dream” EOPancreas is now planned for 2022. Merging the wear-times of CGMs and infusion sets/patch pumps will be a major challenge, though EOPancreas does have backing from JDRF through a two-year grant and received a Breakthrough Device designation from the FDA in May.

Infinovo

New to us, UK-based CGM manufacturer Infinovo displayed its Glunovo i3 CGM System, which reps claimed would receive a CE Mark in June and begin rolling out in Europe starting in January. The CGM system has a 14-day sensor life, three-year transmitter life (which apparently doesn’t require charging), and a reported MARD of 9.5% vs. YSI (n=78 people with diabetes). It requires two fingerstick calibrations per day. The no-receiver system sends data directly to Android and iOS apps with alarms. A representative also mentioned that the company hopes to get FDA approval at some point in the future, but no tangible dates were provided. As with the rotating cast of unproven CGM manufacturers, we’re approaching this one with a healthy dose of skepticism, but we’re always glad to see more competition.

iSens, Inc.

The iSens booth representative highlighted the dual glucose and ketone meters, CareSense Dual and Expert. These products are cleared in over sixty countries with price varying by geography. The CareSense Dual connects to a mobile application via Bluetooth and identifies glucose/ketone trends. We’re keeping an eye on ketone meters – and also continuous ketone monitors, perhaps manufactured by Abbott, Percusense, or others – as enthusiasm for SGLT inhibition in type 1 swells.

Kaleido

Kaleido’s vibrant, “Instagrammable” booth with a spinning prize wheel stood out clearly from the back corner of the Exhibit Hall. The company just launched its pumps in the UK last week and has ~600 users now in the Netherlands – still quite early for a patch pump company, as proving manufacturing scale and supply reliability is critical. (Cue Cellnovo, who shut down earlier this year.) The company’s new CEO, Franz Cromme, does bring med tech experience, including five years at British giant Smith & Nephew. The sizeable booth featured a large round table with all of Kaleido’s colored pump offerings, though we note that the marketing and in-booth devices border on misleading, as the pump requires a short infusion set – see the website here. Beyond expanding in Europe, the company ambitiously hopes to bring its pump to the highly competitive US market, taking advantage of the ACE pump pathway.

POCTech

China-based POCTech was once again present at EASD boasting a new, 14-day wear CGM. The device has already been submitted for CE Mark Approval. According to representatives, the new 14-day CGM would be priced lower than Dexcom’s G6. POCTech already has a 7-day, one-calibration sensor that is CE Marked, according to its website; it is “ready” for distribution in Europe. The rep said neither of POCTech’s CGMs have launched in its native China, where getting regulatory approval is “more difficult than CE Mark, but not as hard as FDA.” No plans to enter the US market are on the horizon. 

Tandem

Tandem returned to the EASD Exhibit Hall for its second time ever, showing off Basal-IQ with the t:slim X2 and Dexcom integration. In June, Tandem began launching Basal-IQ in a “couple of countries,” and more international launches are expected in 2H19; a rep did not specify where these would occur. Getting Tandem Device Updater set up in Tandem’s European markets is well underway and expected to complete by the end of the month. On the US side, many are, of course, excitedly waiting for the launch of the Control-IQ hybrid closed loop (under FDA review; launch expected in 4Q19). 

Terumo

Japanese med-tech giant Terumo was on the floor at EASD showing off its brand-new patch pump. The three-day wear patch pump was just launched two months ago in Japan and “not-yet CE-Marked” stickers were placed on the company’s demo devices. The device has been submitted for CE Mark and is expected to launch in EU markets “next spring.” Like Roche’s Accu-Chek Solo, Terumo’s pump has a three-day disposable portion that connects to a piece with reusable electronics. In our walkthrough of the device, company reps seemed especially excited about the device’s simple one-button insertion process, though we note it is more complicated than Insulet’s all-in-one, single-piece Omnipod with automated insertion. Data from the pump’s dedicated controller handset can be transferred to a smartphone through NFC and the Terumo is currently working on finding CGM partners.

--by Albert Cai, Ursula Biba, Ani Gururaj, Rhea Teng, Abigail Dove, Brian Levine, Martin Kurian, and Kelly Close