June 22-26, 2018; Orlando, FL; Full Report – Diabetes Technology – Draft

In this report, we provide our full coverage of diabetes technology content from the ADA’s 78th Scientific Sessions. Talk titles highlighted in yellow denote the presentations and commentary that we found particularly notable. Titles highlighted in blue represent new coverage that wasn’t included in our daily highlights (day #1, day #2, day #3, day #4, or day #5). Note that some talks may appear in multiple sections.

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

Themes

Automated Insulin Delivery, Pumps, and Pens

CGM and BGM

Digital Health and Connected Care

Beyond A1c and Hypoglycemia

Exhibit Hall

…

Happy reading!

Themes

Diabetes Technology

Major New Devices Debut at ADA, Propelled by FDA and Accelerating Progress in CGM, AID, Apps, and Interoperability

• In a sign of diabetes technology’s accelerating pace, the FDA approved three major PMAs on the meeting’s eve: Medtronic’s MiniMed 670G pediatric indication (7-13 years), Senseonics’ Eversense 90-day implantable CGM, and Tandem’s Basal-IQ/Dexcom G6 (PLGS) – read coverage here, here, and here. One of those would have been a big day, but three in one day on the eve of ADA was truly remarkable! This complemented ADA debuts for DreaMed/Glooko’s Advisor Pro clinical decision support software (cleared a week before ADA), Insulet’s touchscreen Omnipod Dash touchscreen PDM (cleared in early June), Dexcom’s no-calibration G6 CGM (cleared in March), Medtronic’s Guardian Connect and Sugar.IQ app (approved in March), and of course Abbott’s FreeStyle Libre real-time (approved last September). To cap off the meeting with another approval on Monday, Medtronic announced a CE Mark for the MiniMed 670G/Guardian Sensor 3 in 7+ years, enabling the first hybrid closed loop launch outside the US in 10 EU countries this fall – and with pediatric approval from the start.

• Every diabetes device product theater we attended was absolutely packed at ADA, reflecting more interest than we’ve ever seen in diabetes technology. Companies are certainly raising their game, much to attendees delight – Insulet’s Omnipod Dash was widely cited as a terrific example of user-centric design, Dexcom’s G6 received praise for its strong no-calibration accuracy and new FDA iCGM pathway, Abbott’s FreeStyle Libre was frequently called a gamechanger among providers in attendance (especially related to cost and pharmacy access), Medtronic’s MiniMed 670G was lauded for delivering strong time-in-range (>70%) and securing a pediatric approval, and Senseonics’ Eversense and Tandem’s Basal-IQ were noted as excellent new options in AID and CGM.

• By any measure, the year-over-year progress in diabetes technology has been remarkable. At ADA 2017, there was only one standalone real-time CGM in the US (Dexcom G5); now there are four more with Abbott’s FreeStyle Libre real-time, Medtronic’s Guardian Connect, Senseonics’ Eversense, and Dexcom’s G6 entering the market! One year ago, there were zero factory-calibrated real-time CGMs in the US; now there are two with G6 and FreeStyle Libre. CGM interoperability was certainly of interest at ADA 2017, but there was no pathway besides the traditional, lengthy PMA route; this year, Dexcom marketed the first “integrated CGM” (G6) and Tandem showed why it matters with the unexpected early approval of Basal-IQ with G6 – the first iCGM-compatible pump and clear sign of things to come. We believe interoperability will increasingly become a key feature advantage, enabling companies to move faster, iterate more quickly, and integrate more seamlessly with the growing ecosystem of diabetes devices and apps. Korea-based SOOIL even announced plans to submit an open protocol pump to the FDA, hoping to commercialize a device tailored for the DIY community by ADA 2019. During the JDRF/NIH/HCT Closed-Loop Research night, Dr. Aaron Kowalski praised this move and shared strong enthusiasm for the non-profit’s Open Protocol initiative – this did not even exist at ADA 2017! As noted below, the DIY community had its strongest presence ever at ADA, and we fully expect the pace of technology to continue accelerating over the next year – particularly as the regulatory process becomes more streamlined with iCGM (and potentially iPump?), as FDA’s Digital Health PreCert program beta launches at the end of this year, as the DIY community pushes industry faster, and as more companies push innovation to software and move to consumer-grade hardware. Perhaps most of all, the higher level of competition will continue to drive the entire diabetes technology field forward!

Automated Insulin Delivery Commercial Progress from Medtronic, Insulet, Diabeloop, Tandem; Industry-DIY Start to Blend; Compelling Research from Cambridge Inpatient, McGill on Pramlintide, Bionic Pancreas

• In automated insulin delivery (AID), we saw progress on several commercial systems, alongside a greater presence for the DIY community and promising academic research on other hormones and populations. Highlights included Insulet’s Omnipod Horizon five-day hotel studies in 6+ years; Medtronic’s MiniMed 670G real-world data (including in former MDIs), a CE Mark, and a bold outcomes-based $25,000 guarantee for payers; FDA approval of Tandem’s Basal-IQ/Dexcom G6 PLGS system (launch in August) and the Control-IQ pivotal finally getting underway; and new Diabeloop data. Though Medtronic is still the only company to have an approved hybrid closed loop product on the market, momentum is certainly building for other companies’ products – even if they are quite a bit behind Medtronic. Though this ADA had no huge pivotal studies or long-term data, we expect to see much more by ADA 2019.     

  • Insulet presented three strong abstracts from five-day studies of the Omnipod Horizon hybrid closed loop in adults, adolescents, and pediatrics. The studies compared 96 hours of hybrid closed loop with a tablet computer and Dexcom G4 CGM to seven days of patients’ standard therapy. In adults, Horizon drove a robust +2.7 hour/day improvement in the range of 70-180 mg/dl (74% vs. 63%), a dramatic 63% reduction in time <70 mg/dl (-46 mins/day: 1.9% vs. 5.1%), and a one-hour per day reduction in time >250 mg/dl (-58 mins/day: 4.5% vs. 8.5%). In adolescents and pediatrics, the improvement was far more striking: impressive +3-4-hour/day improvements in time-in-range (!) and two-hour/day reductions in time >250 mg/dl, reflecting far more hyperglycemia in the younger age groups. Overnight outcomes looked similar or slightly better, depending on the group. In an encouraging sign for commercial use, Horizon users were in closed loop ~98% of time – and this was not even a commercial-ready platform. Insulet is being very methodical in the testing of its system, though the algorithm seems strong and the Omnipod’s tubeless form factor is unparalleled. We hope the company moves quickly to a pre-pivotal and to a pivotal (no timing shared). Presumably a ~2020 launch is still on track.

  • Medtronic presented a host of data describing MiniMed 670G’s US performance in the real world (100,000+ systems shipped in the US). The population-wide data have been positive enough for Medtronic to offer a bold $25,000 outcomes guarantee for payers. At the top of our list of real-world 670G outcomes were those in former MDI users (n=241). Relative to the pivotal trial and a bigger real-world 670G data set from 30,337 users, these 241 former MDI users had similar Auto Mode outcomes on the MiniMed 670G: 73% time-in-range (70-180 mg/dl), 2% time <70 mg/dl, and 25% time >180 mg/dl. The change from manual mode was also similar to the other groups: a +1.9 hour improvement in time-in-range (73% vs. 65%), no change in hypoglycemia, and an 8 mg/dl improvement in mean glucose. Notably, the MDI users spent 82% of the time in Auto Mode, slightly higher than the real-world users (79%) and closer to the pivotal’s 87%. Ideally, Medtronic would’ve also presented outcomes from the patients’ pre-670G MDI days – the transition we’re really interested in – though that would’ve admittedly been a much more difficult analysis to conduct.

    • In the biggest outcomes-based deal in pumps/CGM to date, Medtronic launched a 670G outcomes guarantee for payers: if there is a diabetes-related hospitalization or ER visit for a patient on the 670G, Medtronic will reimburse it up to a cap of $25,000 over a four-year period. Wow! We admire this move – the third public value-based contract for Medtronic Diabetes after deals with Aetna and UHC. This 670G guarantee shows Medtronic standing by its devices’ outcomes by putting its own money on the line! Will it put pressure on the rest of the field to propose similar arrangements? Will it change coverage dynamics in the US market? Will such guarantees become table stakes for competing in the future? Though 670G is on the market, Medtronic’s pipeline for improvements is moving slower than we expected even a year ago. Using payers as an additional battle ground could insulate 670G against competitive system with more compelling feature sets (e.g., auto bolusing in Tandem’s Control-IQ, no-fingersticks with Dexcom’s G6 integrated systems, potential smartphone control from Bigfoot and Insulet, etc.). 

    • Medtronic also finally announced CE Mark for the MiniMed 670G hybrid closed loop, with a launch to commence this fall in 10 European countries: Belgium, Denmark, Finland, Ireland, Italy, Netherlands, Spain, Sweden, Switzerland, and the UK. As noted above, it will launch with pediatric approval (7+ years) from the start.

  • Diabeloop 12-week home data (n=67) from a randomized crossover trial of its AID system (Cellnovo patch pump + Dexcom CGM + algorithm on handset) were very positive, with time in 70-180 mg/dl increasing by over two hours/day vs. open loop. Efficacy in the overnight period was very similar to that seen in the 24-hour data. Diabeloop still intends to obtain CE Mark and launch in France, the Netherlands, and Sweden in late 2018. Assuming it hits this goal, it is likely to be second to market OUS with hybrid closed loop, following the 670G.

  • Tandem’s Basal-IQ (predictive low glucose suspend) with Dexcom’s G6 received FDA approval the day before ADA started in users 6+ years. A US launch is expected in August, including a free software update for in-warranty t:slim X2 users. This will be the second approved AID system to launch in the US, following the 670G. Basal-IQ was not a major focus at this ADA or in Tandem’s booth, but we expect it will have a big presence at AADE.

    • Tandem didn’t have new data on its Control-IQ closed loop system (t:slim X2 + G6 + TypeZero) with auto correction boluses, but we learned that the pivotal opened enrollment as of June 21 - see the ClinicalTrials.gov post here. The plan is to enroll 168 participants, comparing six months of closed-loop to six months of sensor-augmented pump therapy (same devices, no automation). The study is expected to finish by “April 2019,” with a primary outcome of time-in-range and a remarkable number of secondary outcomes. As of two months ago, Tandem expected a rolling PMA submission in the second half of this year and a US launch in “1H19” – this is ambitious timing to be sure, though the precursor Basal-IQ has hit its timing with the added bonus of integrating with Dexcom’s G6 iCGM. Even if the timing slips, Tandem looks positioned to be second to market in the US with hybrid closed loop, and the first with a system that includes automatic bolusing.

• Brave moves from SOOIL Korea and Diabeloop to embrace and incorporate software from the OpenAPS community show how industry and the DIY movement don’t have to stand on opposite sides of the street. At DiabetesMine’s D-Data Exchange, SOOIL’s Mr. Justin Walker shared bold plans to submit the company’s smartphone-controlled Dana RS insulin pump to the FDA/CE Mark with an open communication protocol, and to also register a version of the open source OpenAPS algorithm. In other words, a pump tailored for the DIY community, which could radically change the innovation paradigm in the field. Meanwhile, Diabeloop will incorporate features inspired and at least partially designed by the OpenAPS community in its second-generation closed loop, expected to begin rolling out in early 2019. One of the modules, a fallback setting for when the patient is kicked out of closed loop, was “nearly copied and pasted” from OpenAPS, and the company would also like to include a variant of OpenAPS’ “eating soon” setting in the second-gen system. We will be keeping an eye on all of the regulatory questions that come with open protocol/DIY approaches: Will SOOIL actually do what it claims? Will it have to run a pivotal RCT with the OpenAPS algorithm running on its pump? Will the open protocol pump become a compelling feature that drives pump uptake? What if the OpenAPS algorithm is improved by a DIY developer on Github – could users of the SOOIL system adapt the updated algorithm? What kind of studies will needed to incorporate modes like “eating soon” into a commercial system (Diabeloop believes in silico simulation will be sufficient, at least in the EU)? Will other companies follow suit, leveraging and helping to scale the DIY community’s – presumably free – work?

• We also saw excellent academic research from the McGill (pramlintide + insulin), MGH/BU Bionic Pancreas (insulin-only vs. insulin/glucagon), Cambridge (inpatient fully closed loop in type 2s), and Harvard (MPC plus adaptive algorithms) groups. The work reflects an important academic community move to improve current hybrid closed loop with additional hormones, smarter algorithms, and new populations.

  • The JDRF-funded, 24-hour McGill study with the Class AP algorithm that Lilly has licensed, found bihormonal closed loop (insulin + pramlintide [amylin analog] infused in a fixed ratio) significantly boosted time-in-range to 86% vs. 74% with insulin alone (+3 hours/day). Notably, this improvement didn’t come at the expense of increased hypoglycemia or side effects, and was apparently driven by much-improved postprandial response. We thought this was among the most groundbreaking AID research at ADA – hopefully an amylin/insulin co-formulation will remain in the cards as a strong next-gen closed-loop addition!  

  • The Bionic Pancreas team presented head-to-head data comparing seven days of insulin-only closed loop to seven days of bihormonal (insulin/glucagon) closed loop to seven days of usual care. Bihormonal had a slight edge in the well-controlled study: mean glucose was 165 mg/dl on usual care, dropped to 148 mg/dl on insulin-only, and dropped further to 139 mg/dl on bihormonal. Similarly, time-in-range (70-180) was 60% on usual care, improving to 72% with insulin-only and up further to 79% with bihormonal. Median % time <54 mg/dl was very similar across the three arms and extremely low (<0.7% in all three arms). Overall, the Bionic Pancreas insulin-only outcomes look excellent and similar to other systems – with some potential user experience advantages around startup and meals – and the addition of glucagon seems to add a bit more efficacy and aggressiveness. (Of course, whether payers/patients will cover the cost of glucagon remains to be seen, but it is truly exciting to see the head-to-head research being done.)

  • The Cambridge closed loop group scored a major win at ADA, presenting positive fully automated inpatient type 2 closed loop data, which was simultaneously published in the NEJM. Overall, participants who received closed loop therapy (n=70) spent a significant 5.8 more hours in-range (100-180 mg/dl; 65.8% vs. 41.5%) and 6.2 fewer hours >180 mg/dl (23.6% vs. 49.5%) per day as compared to the control (n=66). Wow! We cannot wait to see closed loop make its way into the hospital.

  • Finally, a poster (Pinsker et al.) described the compelling results of Harvard’s eMPC (“enhanced MPC”) algorithm, set to be evaluated in protocol four of the IDCL as soon as 4Q18. The algorithm builds on a traditional MPC by adapting insulin response to the degree of confidence (“trust index”) it has in a certain action. With this additional software, n=15 participants saw time-in-range increase from ~75% on SAP to 88% on hybrid closed loop over a two-day study. Time <70 mg/dl was reduced five-fold, from 7.8% to 1.5%. Impressively, time-in-range was 88% both during the day and overnight, despite at least three meals per day with a minimum of 30 carbs/meal and one unannounced exercise session on the morning of day two. We look forward to the IDCL study to see if the trust index can maintain effectiveness over a longer time period, in the real-world, and in a bigger sample.

Moving from Beyond A1c Consensus to Validation, Headlined with TIR Data from DCCT

• The “Beyond A1c” movement has picked up tremendous momentum in just one year: from initial consensus on definitions at ADA 2017 to the start of validating CGM-based outcomes at this ADA! The headline presentation came from Jaeb’s masterful Dr. Roy Beck, who shared a never-before-seen time-in-range analysis of DCCT (seven-point fingerstick) data – it turns out that the correlation is very strong between time in 70-180 mg/dl and microvascular complications, a tremendous win in the fight to validate CGM metrics! Time in 70-180 mg/dl (>50%, 40%-50%, 30%-40%, <30%) had a stepwise correlation with retinopathy progression and development of microalbuminuria in the DCCT. Notably, each 10-percentage-point drop in time-in-range increased the hazard rate for retinopathy progression by 61% and for microalbuminuria by 53%. Side-by-side, the A1c and time-in-range graphs looked almost identical – wow! It is remarkable the correlation is this tight, given that seven-point daily fingersticks were only taken quarterly during DCCT (pre/post meals and pre-bedtime). If anything, Dr Beck pointed out, the correlation might be stronger with CGM, which would more accurately capture time-in-range after meals and overnight. We hope these data, once published (forthcoming), make a compelling case to regulators and payers that time-in-range endpoints are truly meaningful. And on a related note, that we do NOT need a modern-day DCCT to validate time-in-range.

• “Going Beyond A1c” mentions were the norm at ADA 2018, rather than an exception. About a year ago, we’d enthusiastically note any time a speaker referenced the importance of CGM endpoints, the limitations of A1c, the value of time-in-range and AGP, etc.; now, it is increasingly rare not to see such mentions, at least in device-related talks. (The diabetes therapy world is more A1c-centric, though drug presentations are increasingly sharing CGM data too – e.g., SGLT inhibitors in type 1 diabetes.) We saw Beyond-A1c talks from Dr. Rich Bergenstal (the new “GMI” calculator on Jaeb.org, replacing “estimated A1c” [eA1c]), Dr. Irl Hirsch’s group on misleading A1c’s (34% of patients in the study had discordant average glucoses not aligning with expected A1c), and even heard a mention from Dr. Helen Murphy on the limitations of A1c in pregnancy (while A1c levels vary predictably over the course of gestation in pregnant women with type 1 diabetes, mean blood glucose actually remains constant). Of course, all AID talks focused on time-in-range and related CGM endpoints, and several sub-analyses of CGM trials also highlighted A1c vs. CGM endpoints.

  • It’s not just researchers who have joined the beyond A1c movement – industry members are on board as well. For instance, Medtronic’s MiniMed 670G product theater emphasized the importance of time-in-range and the limitations of A1c, with Yale’s Dr. Jennifer Sherr and IDC’s Dr. Anders Carlson underscoring the value of time-in-range for both patients and providers. It’s a good sign to see industry promoting CGM-based outcomes – at the end of the day, a beyond-A1c climate would be more kind to their devices, systems, and apps (+ many drugs) than an A1c-centric one.

• It was highly encouraging to see the CGM consensus definitions widely used throughout ADA – 70-180 mg/dl for time-in-range, <70 and <54 for hypoglycemia, >180 and >250 for hyperglycemia, and coefficient of variation (SD divided by the mean). We also noticed emerging discussion of benchmarks, with many highlighting the MiniMed 670G pivotal trial’s >70% time-in-range as a goal to shoot for. Will benchmarking see more discussion in the next year, particularly as the validation movement expands? We think so, as it will be key for the field to define a clinically meaningful increase in time-in-range and reduction in hypoglycemia. Dr. Beck’s new DCCT/time-in-range data could certainly help on that front!

CGM as “Emerging Standard of Care,” and Not Just for Type 1s; Four CGMs Make ADA Debut; Convincing Professional CGM Outcomes as Use Cases Expand

• The absence of a major outcomes trial readout didn’t stop CGM from driving much of the technology advances (not to mention hallway buzz) at ADA – we wish we had counted how many times speakers framed CGM as the “standard of care” throughout the meeting. It was good timing too, as four – yes, four! – CGMs were making their ADA debuts: Dexcom G6, Abbott FreeStyle Libre real time, Senseonics Eversense, and Medtronic Guardian Connect. During the Q&A section of a Hypo-RESOLVE presentation, Sheffield’s Professor Simon Heller boldly forecasted that all type 1s will be on CGM in the next five years. (We think it will take longer, but doubling in the next five years certainly seems doable in the US and Europe.) Just three days earlier, Dr. Jim Chamberlain (St. Mark’s Diabetes Center) had predicted that CGM will replace BGM “completely” in the next five years, especially for those on intensive insulin therapy. Dr. Chamberlain’s comments were especially telling, given that they were directed towards primary care providers – as other thought leaders have continued to emphasize, integration of diabetes technology into primary care will be essential in truly expanding CGM penetration. Importantly, these sentiments were not exclusive to just those with type 1 diabetes – we heard stronger evidence on the value of CGM in pregnancy, in low-resource settings, and for type 2s, including those not on insulin. UW’s Dr. Irl Hirsch also described CGM as the “emerging standard of care” and underscored its relevance in type 1 diabetes, type 2 diabetes, and atypical diabetes. At a pre-conference event, he emphasized that his biggest surprise has been the value of CGM as a behavior change tool in type 2. We can’t wait to see this studied in a bigger way!

  • Of course, it’s also critical to remember that only ~0.5% of the world’s diagnosed diabetes population (by our estimate) currently wears CGM, and cost/reimbursement and integration into clinical practice will have to continue to improve in order for that number to increase drastically.

• We were very impressed by data demonstrating the value of professional CGM, especially in low-resource settings – we can’t recall ever seeing this much focus on outcomes and cost-effectiveness for clinic-owned CGM at a meeting! In particular, we were struck by an oral presentation detailing the successful implementation of an interim intervention technique (IIT) using Abbott’s FreeStyle Libre Pro in India. One HCP visit in the middle of a single 14-day blinded professional CGM session drove remarkable improvements in daily average glucose (from 191 to 137 mg/dl), time-in-range (+9 hours/day), time <70 mg/dl (-1 hour/day), and time >180 mg/dl (-8 hours/day) in 105 type 2 adults. These results show the incredible potential of “low-cost” intermittent CGM to change the lives of people with diabetes. (One FreeStyle Libre Pro costs ~$40, which is a non-trivial amount to many people around the world, but an amazing investment in long-term health if these intermittent-wear outcomes can be sustained.) On the opposite side of the spectrum, Dr. Rich Bergenstal et al. analyzed sequential FreeStyle Libre Pro use in the US, examining results from a first FreeStyle Libre Pro, and then a second, when the first and second were separated by ~five months. Results were encouraging, showing the intervention drove strong time-in-range improvements in high A1c users. Both studies reiterate what many in the field are beginning to accept: When used in a smart, intentional manner, CGM can be applied effectively for a variety of individuals in a variety of settings.

  • Henry Ford Health System’s highly-respected Ms. Davida Kruger described the profitability of professional CGM in her clinic. Reimbursement has long been a pain point for providers, so we are delighted to hear from multiple high-profile providers now that implementing CGM isn’t a money-losing proposition. Ms. Kruger’s clinic grossed ~$750,000 in annual revenue from >1,400 patients on professional and personal CGM – that in addition to how much more comfortable she felt treating her patients every day since she knew what exactly she should be treating. Similarly, at diaTribe/TCOYD’s Forum, we also heard from Dr. Carol Wysham, who will not see a new patient until they wear CGM at baseline – she wants the data available before the first visit. Wow! Of course, such integration requires some serious workflow modifications – how can other clinics learn from these successes to expand CGM access while keeping the bottom line in the black?

• We were very pleased to hear from numerous speakers that CGM is not just for adults with type 1 diabetes: Commentary held and outcomes showed that people with type 2 diabetes, adolescents/young-adults with diabetes, and pregnant type 1s can all benefit (and often in a cost-effective manner). Our very own Ms. Kelly Close echoed this sentiment Abbott-sponsored corporate symposium by asking what it would take to make CGM “the aspirin of diabetes.” Aspirin is widely-used in a very broad population, and it’s used in many different ways (for pain, CV protection, fever, swelling, etc.). Like with Aspirin, she argued, everyone with diabetes (and even prediabetes) could benefit from continuous glucose data at some point and in some form, it’s just about figuring out the right technology (e.g., type of sensor, level of automation), behavioral support (e.g., human coaching), and frequency/duration of use for each individual to maximize health and economic benefit.

  • The inimitable Drs. Anne Peters and Irl Hirsch shared fascinating CGM case studies during a pre-conference CGM workshop, emphasizing the benefits of CGM in type 2 diabetes. We were interested to hear from Dr. Peters that CGM benefits are “harder to show in trials,” especially in type 2s with less risks of lows. She also mentioned that there is little data on use of CGM in type 2s overall – how can we incentivize such studies so as to drive more robust clinical guidelines and recommendations?

  • Although adolescents and young adults are a traditionally tough-to-reach population, we saw encouraging RCT sub-analyses making a strong case for the efficacy of FreeStyle Libre in adolescents (13-17 year-olds; SELFY) and young adults (18-24 year-olds; IMPACT). Both populations achieved improvements in time-in-range, while teens also improved time >180 mg/dl and young adults also reduced time <70 mg/dl. FreeStyle Libre is still not yet approved for people under 18-years-old in the US, though management noted during the 1Q18 call’s Q&A that a pediatric claim was expected to be  filed with FDA before the end of the year.

  • If the compelling CONCEPTT RCT (CGM in type 1 pregnancy) data shared at EASD last summer weren’t convincing enough, Norwich Medical School’s Dr. Helen Murphy showed unpublished data denoting ~4.5 hours/day during which the CGM group had significantly lower glucose than the SMBG group. Even if the A1c advantage in the CGM group at 34 weeks (-0.2%) didn’t blow attendees away, 4.5 hours/day means a lot less fetal exposure to hyperglycemia! Moreover, she posited that use of CGM in pregnancy is likely to be cost-effective due to reduced neonatal hospital stay and fewer NICU admissions. If there’s anything we learned from the debate between Dr. Denice Feig and Dr. Elisabeth Mathiesen on CGM use in type 1 women during pregnancy, it’s that cost effectiveness may very well be the deciding factor for many providers and clinics. Dr. Feig noted that her team is currently investigating the cost effectiveness of CGM in pregnancy, but she believes that as more advanced, factory-calibrated CGMs come to market, CGM costs will only decline.

• Real-world data flooded ADA, continuing to demonstrate substantial improvements in glycemia with CGM use. Abbott’s FreeStyle Libre real-world data set has now reached an incredible 470,643 readers, amounting to 4.8 billion glucose readings, and the strong correlation between increased scanning frequency and decreased hypoglycemia continues to hold – a new analysis from ~238,000 of those readers showed that increased scanning also correlates with decreased glycemic variability. We also saw the first real-world data for Libre users in the US (n=7,979), suggesting that the consistently observed correlations in the EU between higher scanning frequency, lower mean glucose, and lower hypoglycemia, have safely arrived on American soil. Unpublished T1D Exchange data also demonstrated the A1c benefits of CGM in type 1 diabetes – we were especially impressed by the improvements achieved in children <13 years-old (0.9% lower A1c in CGM users than in non-CGM users). Importantly, this trend holds regardless of insulin delivery method. While all these improvements are undoubtedly heartening, we’re hoping penetration rises – T1D Exchange data shared last fall suggests CGM penetration in the US is at just 24%. We’re hopeful for expansion given that there are now four user-friendly, accurate CGMs are on the market, but there’s clearly much more work that needs to be done.

T2D Insulin Delivery Device RCTs (J&J One Touch Via, Insulet/Lilly U500 Omnipod) Positive Overall – A Good Omen for Market Adoption and Sustainable Businesses in T2D?

• ADA saw report-outs of two of the most highly-anticipated RCTs of novel insulin delivery devices for type 2 diabetes: J&J’s One Touch Via bolus-only insulin delivery device (formerly the Calibra Finesse) and Insulet/Lilly’s U500 Omnipod. We were shocked that One Touch Via didn’t drive greater A1c reductions than pens for type 2s starting on basal-bolus insulin therapy (-1.6% vs. -1.7%), but we believe the similar glycemic outcomes were in part due to the study’s design (more below), and there were highly encouraging preferences for One Touch Via expressed by both patients and providers. Plus, in real-world use, A1c reductions on that level with a device are quite rare to see! The U500 Omnipod did drive a greater A1c reduction than U500 MDI group (-1.27% vs. -0.85%), and with a much lower total daily dose (constant, whereas the MDI group went up by a concerning 50 units/day). This came at the expense of a slight increase in nocturnal hypoglycemia, though the absolute differences in events per year (~2-5) were tiny, and therefore not likely to be clinically significant. At the end of the day, the type 2 insulin delivery device/pump market is severely underpenetrated and high potential, though novel devices like Valeritas’s V-Go, Cequr’s PAQ, and One Touch Via have been slow to scale or even launch (V-Go is the only one on the market). We’d like to see many more type 2s with tubeless devices that make insulin more convenient and easier to use. Fully-featured pumps developed specifically for the type 2 population like Insulet/Lilly’s U500 and U200 Omnipods as well as BD’s “Swatch” patch pump should also help make a dent in outcomes and costs while retaining benefits like connectivity. The U500 Omnipod is expected to launch next year; we’ll wait for Insulet’s 2Q18 call for more specifics on FDA submission. One Touch Via’s future is still unclear in J&J’s hands – it is not part of the Platinum acquisition of LifeScan, and we believe this asset could fit in with many companies (e.g., BD, Insulet, Medtronic, Abbott, or even an insulin company who could prefill it). Given Via’s study outcomes and patient/HCP preferences, we think the real-world performance might be stronger than the RCT; we hope the device finds a future somewhere!

  • Though One Touch Via didn’t drive greater A1c reductions than the Novo Nordisk FlexPen, we are still very positive on the results for two reasons: (i) All enrollees used a pattern-based logbook, which combined SMBG values with a simple insulin adjustment algorithm, which may have confounded the study – the FlexPen group did far better than would be expected in real-world use; and (ii) Subjective reports from the full study population were almost invariably in favor of the OneTouch Via over the FlexPen, from both patients and providers – notably, 91% of providers preferred the One Touch Via over pens to advance type 2 patients from basal to basal-bolus insulin. That both patients and providers expressed such strong preference for Via is a likely sign that the device is less cumbersome, easier to use, more discrete, and easier to train than injections; even if it didn’t confer better glycemic outcomes, we suspect it would drive greater adherence (and by proxy, outcomes) in a real-world setting.

• The Cambridge closed loop group led by Dr. Roman Hovorka also presented compelling data from inpatient fully automated closed loop in type 2s. See the automated insulin delivery theme above for details.

New Era of Diabetes Apps? Decision Support, Decoding Data, Driving Glucose Prediction, Radically Convenient Coaching and Education

• Diabetes apps had a stronger presence at this ADA (in both orals and the exhibit hall), reflecting a shift to mobile software that will drive automatic data upload/analysis, remote monitoring, glucose prediction, decision support, and in-the-moment education. Our own Adam Brown summarized the state of the field in his presentation, Can Diabetes Apps Make Our Lives Easier? His slides, posted at diaTribe.org/Apps2018, have already been viewed nearly 500 times, reflecting plenty of interest in this still-young area – go download them if you are even vaguely interested in this area as he changed lots of minds on the arena. Adam highlighted two key bars for apps to cross – radically convenient tools and decision support aids – sharing more than 40 examples of diabetes apps that meet one or both bars. One of those, Medtronic/IBM Watson’s Sugar.IQ, had its formal launch at ADA, receiving the lion’s share of the Guardian Connect mobile CGM product theater. An oral on Sugar.IQ showed encouraging early launch data – +36 minutes/day in-range = 220 hours a year! – and we’ll be fascinated to see how much it can drive adoption of Medtronic’s standalone CGM. Medtronic plans to quickly follow (within the next year) with four-hour forward-looking hypoglycemia prediction, a widely anticipated feature. At Diabetes Mine’s D-Data Exchange, One Drop also announced plans to launch up to 12-hour forward-looking glucose prediction for type 2 non-insulin users (3Q18), a step into decision support for a population with very little data. Diabetes devices with companion apps are also becoming far more commonplace: Dexcom’s G6, Senseonics’ Eversense, and Insulet’s Omnipod Dash all launched at ADA, showcasing mobile apps that replace physical objects (CGM receivers) and/or capture data that previously required clunky desktop software and custom cables. While some providers remain overwhelmed at the thought of remote monitoring diabetes data, many seemed relieved at having any data to make decisions during appointments. On the diabetes education front, we were glad to see BD formally launch Briight, its AI chat and diabetes education app – see our complete review here. Plenty of coaching and remote care efforts – leveraging apps and connected devices – also shared new data at ADA, including Livongo, Onduo (Sanofi/Verily), Omada, One Drop, and Virta. None were huge outcomes studies of apps, but we were glad to see encouraging efficacy, engagement, and/or cost savings potential. Work is also refreshingly expanding on the food education front: (i) Ascensia’s diabetes challenge winner, Whisk, is an AI-driven food platform that will recommend foods and recipes based on blood glucose levels; and (ii) Medtronic launched an expanded partnership with Nutrino to link meal photos with professional iPro2 CGM responses in a one-page report.

Smart Pens/Pen Caps Inch Forward, Opening Up New Avenues of Research, and Informing Decision Support

• At ADA 2017, there was an impactful poster from Common Sensing, but no commercial movement from the smart pen landscape to speak of. This year, Companion Medical had its first ADA booth following US launch, Novo Nordisk showed off its NFC-enabled, reusable Novo Pen 6 in the exhibit hall, and we heard about lots of in-progress research. The field is still moving slower than we had hoped – arguably where CGM was about 3-4 years ago – but as the hardware and research gains momentum, we hope things show acceleration at future ADAs. Leaders in the field clearly recognize that pumps are not going to be a solution for everyone on insulin, and providers still have a significant gap in injection data that could be filled with smart pens. Though we only saw two possible dose capture form factors at ADA – durable pen and durable pen cap – we appreciate that numerous players are generating a wide variety of MDI dose capture options that will meet the needs of a variety of patient lifestyles. See our smart pen/cap competitive landscape.

  • Companion’s InPen (iOS only) has now been available in limited capacity in the US for ~six months, and the company celebrated the milestone with its first ADA booth. The booth showed off the new “Insights by InPen” reports, which we believe to be the first page in the US to show providers data from a smart pen in conjunction with an AGP-like glucose profile. We also learned that Companion has closed a Series C, and though it won’t comment on the size or investors yet, an SEC Form D filed on May 3^rd suggests that the total offering amount was ~$18.5 million.

  • At Novo Nordisk’s booth, tucked in a corner on the international side, we found the smart Novo Pen 6, which is currently piloting ~10 Swedish clinics (~1,000 patients) but now expected to expand “soon.” In its current form, the pen enables providers to upload insulin usage data to their Glooko/Diasend NFC pad within seconds, though we’d guess that Bluetooth communication is on the roadmap too. The pen’s dial also has a low-tech display on it, giving the size of the last dose and the time since the last dose.

  • Common Sensing’s smart Gocap pen cap is now being used by Good Measures’ diabetes management/coaching program, alongside a US beta launch and a pilot with One Drop and Innovation Health.

• There were just two exclusively smart pen/dose capture abstracts at ADA – both from the UVA, Stanford, Mt. Sinai group using NFC-enabled Novo Nordisk pens – but both give insight into the injection behavior of a well-managed group of type 1s. In one poster, over one in four meals had either a late or missed meal bolus, with 13% of total meals accompanied by a late bolus, and 14% of meals with no bolus whatsoever. Baseline A1c was provided, but there was a significant positive correlation between the number of missed meal boluses and A1c (there was no such correlation between late boluses and A1c). In the second poster, we were actually surprised to learn that the overall percent of pen injections that were primed was 80%. Percentage of missed prime doses did not correlate with A1c, but seemed to be more common with male gender and younger age. While these studies don’t necessarily give insight into the behavior of an entire population, they do show the power of injection dose capture, especially in tandem with CGM, as an educational tool. A provider will be able to assess patient’s dosing habits, and educate accordingly. As insulin companies assess their pipelines, we hope to see more investment in dose capture, data, and decision support – the outcomes potential could be just as big as improved new insulin formulations! 

• On day #1, we also saw a session dedicated to decision support, including new data and studies for type 1s on MDI. Through a ~four-week study of UVA’s decision support platform (with Dexcom CGM and Novo Nordisk connected pens), which included clinic-based meals and exercise, the system significantly improved a number of glucose parameters relative to standard of care: Primarily, coefficient of variation fell from 36% at baseline to 33% (34% -> 30% at mealtime), largely from less hypoglycemia, as time ≤70 mg/dl decreased by ~33 minutes per day (from 3.2% to 0.88%). In the study, the system recommended changes in therapy based on retrospective risk zones, adjusted boluses based on ratio of real-time insulin sensitivity to historical insulin sensitivity, and provided advice on how to handle exercise. OHSU’s Dr. Jessica Castle introduced the Helmsley Charitable Trust-funded, Jaeb-coordinated T1-DEXI pilot study. The pilot will enroll 60 individuals between ages 15-70 with type 1 diabetes, and collect one month of insulin, CGM, food, and physical activity data using Dexcom G5, DiabNext’s Clipsulin dose capture device (an unconventional choice), a Garmin activity tracker, and a custom app developed at OHSU for food photos and exercise logging. It will eventually inform a larger study of 300-500 subjects with a goal of building better exercise and food models for automated MDI titration. It is still early days for decision support for MDIs using dose capture, with no commercial products (apart from Companion’s bolus calculator). That said, this academic work, and especially pipeline systems from the likes of Lilly and Bigfoot, give us hope that many pen users will be able to benefit from closed loop-esque algorithms in the not-so-distant future.

Strong Remote Care Results for Prevention and Treatment of Type 2 Diabetes – Big Focus on Cost-Effectiveness

• ADA showcased a host of impressive data on remote care interventions, headlined by very strong results from Virta in prediabetes and Hygieia in a high-A1c type 2 insulin users. For the first time, we saw one-year data from Virta’s very-low-carb diet plus remote coaching in the prediabetes population: Of the 95 completers (82% retention), a staggering 61% moved from prediabetes to normoglycemia (A1c <5.7%) and not a single person progressed to type 2 diabetes. Moreover, body weight dropped an average of 29 lbs (11.5%) from baseline 111 kg (~245 lbs). These findings go along nicely with evidence from Virta’s controlled study in type 2 diabetes, which has demonstrated ~60% diabetes reversal (defined as an A1c <6.5% and elimination of all medications except metformin) at one year. Equally impressive was a Hygieia poster from its pilot with Blue Cross Blue Shield of Michigan (n=160 completers), demonstrating a remarkable 2.3% A1c drop at nine months from its insulin guidance service (specialty clinics + insulin titration + remote care + BGM). For individuals taking one or more expensive medication (GLP-1 agonists, DPP-4 inhibitors, SGLT- 2 inhibitors) at baseline, average prescription savings were $6,172 per patient-year. Across the entire population, prescription savings were $1,736 per patient-year – this notably doesn’t include savings from medical utilization.

• We also noted outcomes/cost-effectiveness posters from Omada, Livongo, and a 19-year telemedicine study from India! A Humana/Omada poster analyzed Medicare Advantage per member per month (PMPM) costs, finding that Omada’s 12-month digital diabetes program surprisingly didn’t impact total costs with statistical significance, though still drove cumulative savings of $1,298 per member (~$54 saved per member per month).  The authors note that pharmacy cost reductions typically precede medical cost reduction in this population, and propose that a longer-term follow-up may show more marked cost reductions. A separate poster showed that Omada’s intervention in a population of 107 employees greatly reduced prediabetes/diabetes prevalence. Meanwhile, a Livongo poster directly examining medical spending for Livongo users vs. matched non-Livongo users over 12 months. Intriguingly, the study found that a 10% increase in monthly Livongo usage rate was linked to a 2.1% decrease in medical spending, driven by a 2.9% reduction in spending on office-based services. Finally, the 19-year telemedicine intervention (n=414 type 2s) from Jothydev’s Diabetes and Research Centers – ~14.5 virtual consultations for education and medication titration plus ~two physical consultations per year – led to a 0.7% drop in A1c (baseline: 8.3%) and the prevention of vascular complications in a remarkable 94% of participants. There was no control or typical population data shared, but the results strike us as very positive, and more scalable than high-touch clinic-based care. 

• We expect to see digital diabetes care companies continue to take on more responsibility for patient outcomes, and payers/employers to gladly bring them on board to do so. In such an economy, companies will continue to compete on cost-effectiveness as much as they do on health and wellbeing outcomes. There was no data on the Diabeter approach– where a clinic takes on all aspects of and financial responsibility for a patient’s care – at ADA, though Medtronic has said it will look to expand the model into US type 2 clinics. How will this approach compare to that in which (i) payers contract with providers for narrow outcomes; and/or (ii) payers contract with exclusively-digital providers, but in-person care is delivered separately (e.g., Virta for diabetes reversal, Omada for weight loss, mySugr, Onduo, One Drop, etc.)?Will digital and in-person providers eventually have to merge, or at least collaborate, to optimally align incentives of care delivery?

Automated Insulin Delivery, Pumps, and Pens

Oral Presentations: Clinical Trials in Type 1 Diabetes

Safety and Performance of the Omnipod Hybrid Closed-Loop System in Adults with Type 1 Diabetes over Five Days Under Free-Living Conditions

Bruce Buckingham, MD (Stanford University, Palo Alto, CA)

Stanford’s Dr. Bruce Buckingham presented strong results from a five-day, free-living hotel study of Insulet’s Omnipod Horizon hybrid closed loop in adults (n=11, baseline A1c: 7.4%). The study compared 96 hours of hybrid closed loop with a tablet computer and Dexcom G4 CGM to seven days of patients’ standard therapy (27% MDI, 73% pump). Horizon drove a robust +2.7 hour/day improvement in the range of 70-180 mg/dl (74% vs. 63%, p=0.02), a dramatic 63% reduction in time <70 mg/dl (-46 mins/day: 1.9% vs. 5.1%, p=0.001), and a 47% reduction in time >250 mg/dl (-58 mins/day: 4.5% vs. 8.5%, p=0.01). Mean glucose was not significantly different between the groups – 150 mg/dl on closed loop vs. 156 mg/dl on standard therapy (p=0.46) – though with a larger study and broader population at baseline, perhaps this would have been significant. From our view, it doesn’t matter if “mean glucose” is significantly different – the more important point to us is this is why “mean glucose” is not the only metric to watch. The same was true for time >180 mg/dl, which trended in the right direction: 25% on closed loop vs. 32% on standard therapy (p=0.12). Overnight glucose outcomes also looked striking, with an even larger differential on time <70 mg/dl – an 88% reduction in hypoglycemia, from 5.7% to 0.7% (p=0.02), or 24 minutes less per night. As seen in the picture below, closed loop narrowed glucose variability across the entire day, and coefficient of variation came in at a strong 33% overall (vs. 37% on standard therapy), including a very impressive CV of 26% on closed loop overnight. (As a reminder, coefficient of variation is a metric of variability – standard deviation divided by mean – where <36% indicates stable glucose.) Generally this group was in strong control with usual care (more than the average person with type 1 diabetes), and Horizon still showed very significant results and compared quite favorably to the MiniMed 670G pivotal. Horizon users were in closed loop 97.5% of time, which was outstanding and hopefully a good sign for the commercial product (time in Auto Mode was 87% in the 670G pivotal and 79% in real-world 670G use). It was great to see the study include 27% MDI users – key for Insulet to include, given its target market.

• This was also a solid real-world challenge for the system, as the average meal size was 53 grams of carbs (largest meal was 200 grams of carbs!) and the 11 study participants engaged in a total of 43 exercise sessions averaging 54 minutes each (51% of which were moderate or high intensity). The algorithm targets 120 mg/dl, but participants could adjust their glucose set point temporarily anywhere between 100 mg/dl and 150 mg/dl – nice! A raised glucose set point was used for 37% of exercise sessions (mostly 140 or 150 mg/dl). Participants could also opt to use extended boluses, and 64% of participants chose to do so – a marker of how engaged this population is, perhaps even underselling the system’s benefit relative to baseline. Dr. Buckingham concluded that Horizon was safe and performed well over five days of use, and the team is planning further studies in 2-4 year olds. The 6-12 year old and adolescent data is included in the poster section below; it looked every stronger and we think this is a better comparison to the “average” outcomes in type 1 diabetes across the US. It’s unclear when Insulet plans to do its pivotal, but we’d guess 2019 is likely at this stage.

• To enable fast iteration, the team is still using a research platform: a tablet running the MPC algorithm and communicating to the Omnipod PDM and a separate Dexcom G4 receiver. The compelling commercial product vision remains an algorithm integrated into the tubeless pod and talking to the G6 directly – a full on-body closed loop ecosystem. The Dash PDM platform, adapted for Horizon, will allow users to interact with the system for meals, see G6 CGM values, view system status, etc.

• “The system handled exercise reasonably well…We had a guy who could run a 4-minute mile, and he wanted to run a few miles at a 6.5 minute pace. We didn’t have anyone that could run that fast with the tablet to keep up, so we had someone follow him on a bike with the tablet.”

Insulin-plus-Pramlintide Artificial Pancreas in Type 1 Diabetes—Randomized Controlled Trial

Ahmad Haidar, PhD (McGill University, Canada)

McGill’s Dr. Ahmad Haidar presented some of the most compelling hybrid closed loop data of ADA: use of pramlintide (amylin analog) on top of rapid acting insulin during a 24-hour, in-clinic, JDRF-funded study boosted time in 70-180 mg/dl by nearly three hours/day vs. insulin alone – wow! There were no concomitant increases in hypoglycemia or side effects. A small group, 27 type 1 adults (mean age 39 years, mean A1c 7.8%), each used single-hormone closed loop (rapid insulin), dual hormone closed loop (rapid insulin + pramlintide), and dual hormone closed loop (Humulin R + pramlintide), in randomized order, for 24-hours in clinic. Basal-bolus pramlintide was delivered through a separate pump at a fixed ratio of 6 ug per unit of insulin to mimic co-formulation, and to make a case that this could eventually be done with a one-chambered pump. On rapid insulin + pramlintide in the closed loop, patients spent a whopping 86% in-range, vs. 74% with rapid insulin alone and 68% with Humulin R + pramlintide. This translates to ~3 hours more in-range per day than rapid insulin alone, and ~4.5 hours more than with Humulin R + pramlintide. As expected with pramlintide, meals really drove the advantage, as shown in the quite compelling plots below. Mean glucose was also a significant 10 mg/dl lower on rapid insulin + pramlintide vs. insulin-alone and Humulin R + pramlintide (133 vs. 143 vs. 143 mg/dl). The same was true of coefficient of variation: 27% vs. 31% vs. 33%. Time >180 mg/dl was significantly reduced by 1.7 hours/day with rapid insulin + pramlintide compared to rapid alone (16.7% vs. 9.7%) There was no difference between rapid alone and rapid + pramlintide in terms of time <70 mg/dl – a great sign that improved time-in-range was not traded for more hypoglycemia. From an episodic perspective, there were 0.45 hypoglycemia events/day in the rapid insulin arm, 0.46 events/day with rapid insulin + pramlintide, and 0.7 events/day with Humulin R + pramlintide. It is still possible that pramlintide + rapid insulin caused greater time/depth of hypoglycemia than did rapid insulin alone, but the data presented is encouraging, as hypoglycemia is a risk with amylin. The swift PK/PD of the rapid acting insulin allows for what appears to be safe use in tandem with co-infused pramlintide, while the Humulin R-pramlintide combo seems less advisable – this is not surprising from our view as no matter how you look at it, that insulin is just less stable for most people with diabetes. Regarding side effects (nausea, headache, vomiting, bloating, and heartburn), there were no observable differences between the three arms, and all reported events were noted to be mild. This is a really big deal, as these are far better results in terms of side effect profile than taking pramlintide “straight” as a mealtime injection – presumably the fixed ratio and co-infusion helped minimize this. This preliminary data is highly encouraging, suggesting that amylin-insulin analog co-formulations could be very successful in closed loop for type 1s, especially because insulin is co-secreted with amylin in a non-diabetes beta cell. This study used the Class AP algorithm that Lilly licensed for its in-development hybrid closed loop – we can’t help but wonder whether Lilly will pursue this dual-hormone approach (after all, the company has said they are keeping an open mind with regards to closed loop system design). Lilly doesn’t have an amylin analog in its pipeline to our knowledge, but could well work with Adocia (who has an insulin/pramlintide coformulation), AZ (pramlintide, on the market), Novo Nordisk (long-acting amylin analog in phase 1), or Zealand/BI (long-acting analog). We’re not aware of other amylin analogs in-development. We would absolutely love to see amylin make a comeback – it is far more challenging to dose at meals (particularly figuring out how much less insulin to take) and we think the fixed-ratio co-formulation would appeal to many patients, especially given the single-chamber potential. Notably, AZ’s Symlin was originally used as “bolus” but some patients with old pumps used to pump Symlin and anecdotally had much better results. Some did both (basal and bolus) and this study suggests that works quite well. 

• During the day, rapid insulin + pramlintide conferred a remarkable 82% time-in-range, compared to 60% in the rapid insulin closed loop arm. This makes sense, as one of pramlintide’s key attributes is its ability to blunt postprandial excursions, and in this study it was co-infused with bolus insulin at meals (i.e., as a co-formulation would be). The improvement during the night was not significant (95% vs. 92%), though Dr. Haidar pointed out that there’s obviously not much room to improve from insulin alone. Daytime mean glucose with rapid insulin + pramlintide (143 mg/dl) bested that in the rapid insulin-alone arm (155 mg/dl). At night, the rapid insulin + pramlintide arm had slightly greater mean glucose than rapid insulin alone (118 vs. 111 mg/dl), though this difference was not significant. The only significant difference with Humulin R + pramlintide was a ~2 hour decrease time-in-range overnight relative to rapid insulin alone. In Q&A, a physician noted that the improvements came during the day, so why not just use pramlintide at mealtime as it is currently indicated? Dr. Haidar responded that if he had a dual-chambered pump, he wouldn’t choose to deliver at night, but the advantage of co-formulation is the user only needs one infusion site, one cartridge, etc. 

• Supplementing insulin with pramlintide in hybrid closed loop significantly improved postprandial response, particularly when the insulin was rapid-acting. The slide below depicts the incremental postprandial glucose change from baseline when glucose at the start of the meal was <90 mg/dl (upper left), >180 mg/dl (lower left), and between 90-180 mg/dl (right). The figures demonstrate how pramlintide blunts postprandial excursions in all three scenarios, and even causes an immediate decrease in glucose following a meal when initial glucose is >180 mg/dl. We assume 90 mg/dl was used because of 5 mmol/l (McGill is in Canada), or perhaps because the algorithm starts making tweaks at that level. Ideally, the ranges could also be reported in the now-accepted <70, 70-180, and >180.  

• In a post-trial survey, 19/26 (73%) individuals agreed or strongly agreed that they would use a co-formulation product of rapid insulin + pramlintide if it were on the market. Only three (~12%) disagreed or strongly disagreed. Further, 19/26 (73%) agreed or strongly agreed that rapid insulin + pramlintide made their sugar control more even or predictable (of course, this wasn’t exactly a placebo-controlled, blinded study). Meanwhile, 14/26 (54%) disagreed or strongly disagreed that they would use a co-formulation product of Humulin R + pramlintide, and 12/26 (46%) disagreed or strongly disagreed that Humulin R + pramlintide made their blood glucose more even or predictable. Yeah, Humulin R is a really hard product for many people with diabetes to use and we’d argue virtually all those with type 1.

• Close Concerns Questions: Was there a reduction in insulin dose during the pramlintide arms? (Dr. Haidar told us the data is not fully analyzed yet.) We assume insulin was reduced in the pramlintide arm, but look forward to seeing the data. Assuming there was a reduction, we wonder how much it was? How cost effective is pramlintide when used in a co-formulation? (We imagine that less insulin would certainly offset the cost of pramlintide but we can’t imagine short-term it’s cheaper – long-term of course there would be many potential benefits associated with greater time in zone.) Will any company pursue a single co-formulation to market? How much time did participants spend <70mg/dl? <54 mg/dl? Was there an overall reduction in time <180 mg/dl? Will Lilly pursue a pramlintide + rapid insulin algorithm? Would this have greater efficacy over insulin + glucagon? Is there any interest in triple therapy? : >

• Details on methods and logistics: The study used Dexcom sensors and Medtronic pumps with the patients’ usual rapid-acting insulins. Algorithms for basal insulin and basal insulin/pramlintide deliveries were identical, while bolus insulin and insulin/pramlintide deliveries were different across interventions. Meals and carbs were announced to the algorithms. Each intervention was preceded by a two-week run-in to titrate dosing, after which the participant came into the clinic for 24-hour evaluation including three self-selected meals and one snack.

Twelve-Week Home Use of Hybrid Closed-Loop Insulin Delivery System vs. Sensor-Assisted Pump Therapy in Adults with Type 1 Diabetes—Intermediate Results of the Multicenter Randomized Crossover Diabeloop WP7 Trial

Sylvia Franc, MD (Centre Hospitalier Sud-Francilien, France)

12-week home data from Diabeloop’s WP7 (n=67) randomized crossover trial of its AID system (Cellnovo patch pump + Dexcom CGM + algorithm on handset) were very positive, with time in 70-180 mg/dl increasing by 2.9 hours/day vs open loop (69% vs. 57%). Due to a fairly significant discrepancy in baseline A1c (7.9% in the open loop group, and 7.3% in closed loop), the authors also presented mean adjusted data, which still suggested a ~2 hour mean increase per day in time 70-180 mg/dl vs. open loop (67% vs. 59%). Because this is just an interim readout of the first segment of a real-world crossover trial – as requested by the French Health authorities, no less – the difference in initial A1c levels will come out in the crossover period. Adjusted for baseline A1c, the closed loop group spent 50 minutes less per day <70 mg/dl (1.6% vs. 4.9%), 10 minutes less per day <50 mg/dl (from a low base; 0.1% vs. 0.8%), and 30 minutes less per day >300 mg/dl (2.6% vs. 4.5%). We are curious what the differences were >180 mg/dl and over 250 mg/dl.  Efficacy in the overnight period was very similar to that seen in the 24-hour data. From a safety perspective, the open loop group experienced two severe hypoglycemia events, while the closed loop group had zero “due to the Diabeloop device,” and three “not linked to the Diabeloop device” – this was not further clarified. There was no DKA in either group. The slide below shows tightening and slight lowering of the glucose profile compared to the open loop control during both day and night, though differences in mean glucose and coefficient of variation (CV) were not significant: Mean glucose was 165 mg/dl in open loop vs. 161 mg/dl in closed loop; the improvement in CV from 33.8% in open loop to 31.8% in closed loop trended toward, but did not reach significance (p=0.062). The second “crossed-over” arm of WP7 is still ongoing.

• Up next, Diabeloop aims to obtain CE marking, complete the second half of WP7, commence WP8 (a study for European reimbursement), and evaluate the system in kids. As a reminder, the first iteration of Diabeloop’s automated insulin delivery system is at this point set to roll out in France, the Netherlands, and Sweden in late 2018. Though Cellnovo’s pump was used in WP7, we’re not sure if Cellnovo’s, Kaleido’s, or both patch pumps will be used in the initial launch. This system could be second hybrid closed loop on the EU market, as we assume the now-CE Marked MiniMed 670G launches (announced on the fourth day of ADA) will launch first (this fall in ten EU countries). Tandem has not given Basal.IQ timing for Europe. We also learned at this ADA from Ms. Dana Lewis that Diabeloop will incorporate features inspired and at least partially designed by the OpenAPS community in its second-generation closed loop, expected to begin rolling out in early 2019.

• The French Health authorities required Diabeloop to offer the telemedicine services of dedicated nurses for the purposes of the trial. A skeptical audience member asked if the improvements were due to the remote support or the closed loop, though we would assume both groups had access to the nurses. In the future, not all Diabeloop systems will come with the telemedicine contacts. 

Real-World Data from the MiniMed 670G System Commercial Launch—Patients Previously Using MDI Therapy

Scott Lee, MD (Medtronic Diabetes, Northridge, CA)

Medtronic’s Dr. Scott Lee shared interesting real-world data from former MDI users who transitioned to the MiniMed 670G (n=241), comparing their CGM data on 670G in Manual Mode vs. Auto Mode. As the table below shows, relative to the pivotal trial and a bigger real-world 670G data set from 30,337 users, these 241 former MDI users had similar Auto Mode outcomes on the MiniMed 670G: 73% time-in-range (70-180 mg/dl), 2% time <70 mg/dl, and 25% time >180 mg/dl. The change from manual mode was also similar to the other groups: a +1.9 hour improvement in time-in-range (73% vs. 65%), no change in hypoglycemia, and an 8 mg/dl improvement in mean glucose. Notably, the MDI users spent 82% of the time in Auto Mode, slightly higher than the real-world users (79%) and closer to the pivotal’s 87%. Dr. Lee clarified in Q&A that these are not pre-670G outcomes on MDI vs. 670G Auto Mode outcomes – the real question we’d want to see answered. Instead, these data reflect how someone on MDI did after transitioning to 670G, both in Manual and Auto Mode. The bigger question will be answered in Medtronic’s currently-recruiting outcomes study (goal of n=1,000). As a reminder, the 670G pivotal study required patients on a pump for at least six months (± CGM) – it’s great to see positive outcomes in MDI users, even if it’s a small fraction of users to start.

• Dr. Lee commented on his own practice, noting that many MDI users have had an easier time transitioning to the 670G than many pump patients: “In some ways, they have less baggage from wanting to control every single blood sugar and every single insulin delivery.” This is a great point and one we’ve been making for some time – and just another sign that the closed loop will certainly expand the pump market more than some may have thought – we’ve been calling the closed loop the “killer app” for pumps for ages and it’s fantastic to see more and more positive feedback for everyone that goes on it. The ROI for pump therapy certainly is improved, often dramatically, when CGM and closed loop automation are added – the comprehensive benefits will, of course, be seen even more over time as algorithms learn and adapt. A real homerun closed loop product for MDIs may well be Insulet’s Horizon, though over time Medtronic and others could get there with smaller and more intuitive versions of the 670G and other traditional pumps.

Oral Presentations: Innovations in Insulin Formulation and Delivery

Human Regular U-500 Insulin via Continuous Subcutaneous Insulin Infusion vs. Multiple Daily Injections in Adults with T2D—The VIVID Study—All Randomized Population

George Grunberger, MD (Grunberger Diabetes Institute, Bloomfield Township, MI)

Dr. George Grunberger presented positive data from the large, 26-week VIVID RCT of Insulet’s Lilly-partnered U500 Omnipod (vs. U500 MDI, both with Humulin) in type 2s (n=420) with high total daily doses (TDD) of insulin – the first ever major study of U500 insulin administered through a pump, to our knowledge. Enrollees were a mean ~57 years old, with BMIs of ~40, high A1cs (~8.75%), and TDDs of ~290 units. During the first week of the study, all participants injected U500 insulin three-times daily – one group continued with MDI for the subsequent 25 weeks, while the other switched to pumping U500 with Omnipod. By the end of the 26 weeks, both groups had significant A1c reductions with U500 insulin, but the Omnipod group’s decrease was significant greater by 0.42%; -1.27% vs. -0.85% (p<0.001). The groups’ A1c began to diverge as early as eight weeks (the first time it was measured per the protocol) – not surprising at all but surely heartening to the U500 manufacturers as well as patients lucky to be in this trial. Fasting plasma glucose reduction was also significantly greater in the Omnipod arm (~30 mg/dl decrease) vs. MDI (no net change) at 26 weeks. Further, the greater A1c reduction with the Omnipod was not accompanied by an increase in TDD by week 26, while the MDI group had increased its TDD by ~50 units, up to ~340 units per day – lower A1c, with less insulin on Omnipod= a home run in our book and great to see that the higher cost of pump delivery is offset by less insulin. Overall, there was no significant difference in documented symptomatic hypoglycemia <54 mg/dl (~15 events/patient/year) or ≤70 mg/dl (~35 events/patient/year), nor in severe hypoglycemia, which was low in both groups and presumably from significant insulin resistance in this population. (We assume these glucose events were measured via BGM, but it was not stated.) However, nocturnal hypoglycemia events/patient/year were significantly greater in the Omnipod arm, both at the <54 and ≤70 mg/dl thresholds – we’d point out in the graph below that the absolute difference looks very small (~2-5 events per year!) and therefore not likely very clinically significant though obviously any extra time in hypoglycemia isn’t a positive. One audience member asked during Q&A if the greater nocturnal hypoglycemia materialized because of large pre-sleep boluses; another questioned whether the basal rate needs to be reduced at night. While Dr. Grunberger showed that the bolus:basal ratio increased slightly from 50:50 to 53:47 in the Omnipod arm, it was unclear why they had greater levels of hypoglycemia. With further studies, occasional professional/real-time CGM, and greater education regarding titration, we believe the increase in hypoglycemia could be mitigated. Dr. Grunberger concluded the talk by flashing “U500 delivered in a dedicated pump could be a viable option for patients requiring high doses of insulin.” We’d agree that it could be and believe overall in inimitable Grunberger fashion, this was slightly understated. This long-in-progress study was a definite success, particularly because it filled in a large gap in existing literature – ~25% of U500 insulin is reportedly used in pumps designed for U100, and the safety and efficacy of this practice was previously unknown. We’re excited for very insulin-resistant people with type 2 diabetes to have a pump option where they won’t have to do arithmetic acrobatics in their head in the near future or use multiple syringes or pen doses. Insulet has long guided for this product to launch in 2019, and we believe this study proves its clear safety and efficacy. Overall, seems like lots of upside for Insulet as well as this patient group.

• We saw screenshots of the very well-designed U500 Insulet PDM at DTM 2017.

• A separate ADA poster (P-1011) detailed the same VIVID study, but examined a primary population excluding individuals on GLP-1s and SGLT-2s. Results were nearly identical to those in Dr. Grunberger’s presentation, with both groups seeing significant A1c improvements, but the pump group finishing the 26 weeks with a 0.44% greater A1c drop (1.3% vs. 0.86%). Nocturnal hypoglycemia (≤70 mg/dl) and severe hypoglycemia were slightly (and significantly) higher in the Omnipod arm, but in general the device still seems safe and effective in the GLP-1 agonist- and SGLT-2 inhibitor-using patients.

Fully Closed-Loop Glucose Control in Noncritical Care Settings—A Randomized, Controlled Two-Center Study

Lia Bally, MD, PhD (Inselspital Bern University Hospital, Switzerland)

Dr. Lia Bally presented very encouraging results from a multi-site, 15-day RCT (n=136) of the Cambridge group’s fully automated closed loop system in inpatient type 2 diabetes. In a huge win for the group and the closed loop field, the study was also simultaneously published in NEJM! Wow! To initialize the system – which consists of the Navigator 2 transmitter/receiver, Dana R insulin pump, and a tablet housing the algorithm – only body weight and estimated total daily dose are required. Overall, participants who received closed loop therapy (n=70) spent a remarkable 5.8 more hours in-range (100-180 mg/dl; 65.8% vs. 41.5%) and 6.2 fewer hours >180 mg/dl (23.6% vs. 49.5%) per day as compared to the control (n=66). Neither group spent any time <54 mg/dl (one could argue that the algorithm could be more aggressive to eliminate more hyperglycemia). Dr. Bally emphasized that these incredibly positive results were achieved by variations in insulin delivery (adjusted every 12 minutes), as the total daily dose did not significantly differ between groups. Standard deviation was also significantly lower in the closed-loop group (46 mg/dl vs. 59 mg/dl), as was mean glucose (154 mg/dl vs. 188 mg/dl). Importantly, the patient population was impressively diverse, admitted to the hospital for a variety of reasons including renal, cardiac, respiratory, and neurological complications. ~Two-thirds of the population were on a basal-bolus regimen. 98% of participants reported being happy to have their glucose levels controlled automatically, and 100% of participants would recommend use of the system to others. Dr. Bally concluded that the closed loop system promises to be a novel approach in managing inpatient hyperglycemia safely and effectively. These results add to similarly positive data from a smaller study (n=12) of the system presented at ADA 2016. As the only group, to our knowledge, currently running studies to bring closed therapy to the inpatient setting, we’re certainly rooting for Cambridge’s success. The fact that baseline time-in-range in a place where healthcare is supposed to be delivered is just ~40% is mind-blowing and highly concerning – particularly with recent research demonstrating that higher mean glucose in the cardiac perioperative setting significantly correlated with lower 30-day mortality. Dr. Bally mentioned during Q&A that future studies plan to actively involve the hospital staff.

Improvements in A1c and Time-in-Range in DIY Closed-Loop (OpenAPS) Users

Dana Lewis (OpenAPS, Seattle, WA)

Ms. Dana Lewis presented very impressive data from a retrospective, cross-over study of 20 do-it-yourself (DIY) users, comparing their glucose metrics four to six weeks before vs. after initiation of OpenAPS. DIY closed loop conferred improvements in every metric analyzed: time-in-range (70-180 mg/dl) increased from 76% (quite a high base) to 82% (+1.5 hours/day); time <70 mg/dl decreased from 6% to 4.5% (-22 minutes/day); and time >180 mg/dl decreased from 18% to 13% (-1.2 hours/day). On DIY, participants saw mean blood glucose decrease by 8 mg/dl (136 mg/dl to 128 mg/dl) and mean estimated A1c decline by 0.3% (from a very low base of 6.4%). Clearly, the improvements are made all the more impressive by the fact that the DIY users were very on top of their diabetes care regimen prior to going on closed loop. Although this study focused solely on glycemic outcomes, Ms. Lewis expressed the hope that all hybrid closed loop studies consider measures of quality of life, including frequency of clinical visits, number of manual interventions required, and sleep. We’re definitely impressed by these results – in the first ADA DIY oral, no less – and are delighted to see the DIY community releasing more data (the last data we saw from OpenAPS was in an ADA 2016 poster), lending even more credibility to this trailblazing community. Kudos to the ADA for choosing this as an oral! The posters section below includes more DIY data and updates – we thought it was great to see the “best” p-value beside overall time-in-range and time >180 mg/dl (sometimes reductions in hyperglycemia are not viewed the same way in importance as reductions in hypoglycemia).

• Ms. Lewis broke out overnight data, and to her surprise, found that greater improvements in glycemia were seen during the day. She postulated that there are so many tasks people with diabetes have to do during the day that OpenAPS can “benefit a lot more in helping us catch up with all the things we’re dealing with.” That logic makes sense – closed loop could feasibly help people better manage meals and exercise – though hybrid closed loop studies almost always show greater improvement overnight. They actually looked pretty equal to us: overnight time-in-range increased from 80% to 86% while daytime time-in-range increased from 74% to 80%. See below for the tabulated data.

• Interestingly, Ms. Lewis noted that a couple of people actually saw their time-in-range decrease on closed loop, as evidenced by the individual pre- and post- results shown below. She stipulated that this is likely due to variations in the set glucose target, as OpenAPS recommends setting a very conservative target (especially for beginners).

Oral Presentations: Preventing and Treating Hypoglycemia

Differential Effects of the Insulin-Only and Bihormonal Configurations of the Bionic Pancreas on Mean Glucose and Hypoglycemia during the Daytime and Nighttime

Courtney Balliro (Massachusetts General Hospital, Boston, MA)

This impressive, randomized, six-arm (!), crossover home study (n=23) compared use of the MGH/BU Bionic Pancreas in insulin-only and bihormonal (insulin/glucagon) modes to usual care. Each arm was seven days long; participants were studied with and without remote monitoring (the latter presented here); and the study used Dexcom’s G5 CGM, an iPhone 6S running the Bionic Pancreas control algorithm, and one or two Tandem t:slim pumps depending on the arm (Lilly glucagon was reconstituted daily). Notably, there were no restrictions on food, work, or activity; meal announcements were optional; and only body weight was used to initialize the system. The picture below shows the topline seven-day outcomes in each arm, which gave a slight edge to bihormonal: mean glucose was 165 mg/dl on usual care, dropped to 148 mg/dl on insulin-only, and dropped further to 139 mg/dl on bihormonal. Similarly, time-in-range (70-180) was 60% on usual care, improving to 72% with insulin-only and up further to 79% with bihormonal. Median % time <54 mg/dl was very similar across the three arms and quite low – 0.58% vs. 0.64% vs. 0.17%. In terms of projected A1c, only 30% of the usual care arm made it under 7% vs. 61% on insulin-only and a remarkable 91% on bihormonal; these corresponded to projected A1c’s of 7.2% on usual care, 6.8% on insulin only, and 6.5% on bihormonal. Day and night outcomes were also broken out, with the largest difference at night – bihormonal had an average glucose of 129 mg/dl and an impressive 88% time-in-range vs. 143 mg/dl and 76% time-in-range on insulin-only. Ms. Balliro strongly emphasized that overnight time <54 mg/dl was 0% on bihormonal vs. 0.7% on insulin-only – a point we felt was overemphasized, given the very low rates in both arms.

• In short, this was a very well-controlled, real-world, long-needed head-to-head evaluation of how the Bionic Pancreas performs in both modes. The insulin-only outcomes look excellent and similar to other systems (with some potential user experience advantages on meals, startup, and adaptation. The addition of glucagon seems to add a bit more efficacy and aggressiveness. (Insulin-only targeted 110 mg/dl, while bihormonal targets a slightly lower 100 mg/dl.) An open question remains whether payers and/or patients will cover the additional cost of glucagon for the added benefit. Still, we’re elated to see bihormonal being tested so rigorously, head-to-head against insulin and usual care!

• Notably, the bihormonal system was rated significantly higher for less worry about lows. There was no difference in self-reported hypoglycemia, though significantly more total carbs were used to treat hypoglycemia in usual care when compared to both bionic pancreas arms. We don’t think it’s worthwhile for some to proclaim this is “too expensive,” etc. – it’s too early to know how glucagon will be priced and distributed, especially because a Zealand cartridge of dasiglucagon may last for a week or more in the iLet (depending on how much is used).

• Beta Bionics expects an insulin-only pivotal trial to begin in the second half of 2019, while the bihormonal pivotal trial is expected to begin in late 2019 or early 2020. The team is beginning an outpatient bridging study this summer, testing the insulin-only version of the integrated iLet device – see our coverage from the IDE approval in May here. Following ADA, we learned that the first iLet dosing in adults using Dexcom CGM and Novo Nordisk’s Fiasp in PumpCart (prefilled) has already begun this month (July). The first children will also be dosed with the iLet using Dexcom CGM this month. In August, Beta Bionics will also dose the first adults with the iLet and Senseonics Eversense CGM.

• Ms. Balliro also noted that the Bionic Pancreas “knows nothing about a patient’s current insulin needs” – it initializes based on body weight and adapts over time. This could prove a significant startup/training advantage relative to other systems like the MiniMed 670G, and a significant advantage in adolescents with ever-changing insulin requirements.

• The study included 23 participants on insulin pump therapy for at least six months, and a current or previous CGM user. Participants had a mean age of 38 years, a mean A1c of 7.2%, and an average diabetes duration of 25 years.

Questions and Answers

Dr. Anne Peters: In the insulin-only arm, did they have to tell the system how many carbs they were going to eat?

A: There is no required input to the system in either arm – it’s an optional meal announcement feature. It gives 75% of the insulin it thinks you need. And it only asks for a qualitative measurement – “typical,” “larger than typical,” or “smaller than typical” meal. That may also be why there is a little more hypoglycemia in the insulin-only arm.

Q: How much difference was there in insulin use with bihormonal?

A: There was not a significant difference in total daily dose. The bihormonal arm did use a little more insulin, but insulin-only used the same as usual care. It was pretty typical for what we see in studies.

Q: What about precautions for exercise?

A: We haven’t had structured exercise in our studies. In these outpatient studies, we’re not going to tell participants what to do – we don’t specify or force them to do structured exercise. Studying structured exercise is not a priority in the development of the device at this point.

Oral Presentations: From Progression to Management in Type 1 Diabetes—What Is New?

Overnight to Early-Morning Glycemic Outcomes in Children Using the MiniMed 670G Hybrid Closed-Loop (HCL) System

Gregory Forlenza, MD (Barbara Davis Center for Diabetes, Aurora, CO)

Barbara Davis Center’s Dr. Gregory Forlenza shared unpublished data on overnight to early-morning glycemic outcomes in pediatrics (7-13-year-olds) with the MiniMed 670G/Guardian Sensor 3 hybrid closed loop. The results (n=105) were broken out from the three-month pediatric pivotal study presented at ATTD in February. The single-arm, multi-center investigation found the 670G significantly increased overnight (10PM – 7AM) time-in-range (70-180 mg/dl) from 57% to 71% (+1.3 hours). Overnight time <70 mg/dl also improved significantly, dropping from 5% to 2% (-16 minutes), and overnight time >180 mg/dl decreased from 39% to 27% (-1.1 hours). To us, these data sound like many more uninterrupted nights of sleep for children and parents, which is easily one of the most compelling outcomes of hybrid closed loop and far less risk longer-term (though admittedly better time in range is not yet formally associated with fewer long-term complications). Mean glucose declined from 166 mg/dl to 155 mg/dl. Impressively, the time during which no insulin was delivered increased a whopping 108 minutes – that’s nearly two hours! As Dr. Forlenza pointed out, this alone clearly demonstrates the benefit of matching insulin delivery to physiologic demand, which is impossible with a single set basal rate every night. We were delighted to see the 670G obtain FDA approval for 7-13 year-olds on the eve of this ADA, becoming the first hybrid closed loop system approved for pediatric use. Dr. Forlenza mentioned during Q&A that a post-market study is currently underway to investigate outcomes in CGM-naïve patients, as well as those transitioning from MDI – see it here (goal of n=1,000 participants). He expects the next round of post-market studies to focus on patients with a higher baseline A1c (baseline A1c in the pediatric pivotal was 7.9%). The seven-site, three-month 670G study (n=50) in children ages 2-6 is still ongoing.

Oral Presentation: Effects of Exercise on Metabolic Health in Type 1 and Type 2 Diabetes (With ADA Presidents' Select Abstract)

Reducing Basal Insulin 90 Minutes Before Exercise Protects Against Hypoglycemia Better than Insulin Suspension at Exercise Onset in T1D—The OmniTIME Results

Dessi Zaharieva (York University, Toronto, Canada)

York University’s Ms. Dessi Zaharieva presented the results of the OmniTIME study (Omnipod Type 1 diabetes Insulin Management & Exercise), which demonstrated that reducing basal insulin delivery from the Omnipod by 50%-80% 90 minutes before aerobic exercise better protects against hypoglycemia than pump suspension at exercise onset in 17 type 1s – no surprise there, given the lag time in insulin action! The OmniTIME study compared hypoglycemia during prolonged steady state exercise and in recovery using three different basal insulin reduction strategies: (i) pump suspension at exercise start; (ii) 50% basal insulin reduction, set 90 min pre-exercise; and (iii) 80% basal insulin reduction, set 90 min pre-exercise. Study participants were individuals with type 1 diabetes (n=17; mean A1c 6.5%; mean age 35 years; mean BMI 25 kg/m²; mean diabetes duration 14 years), who were on Insulet’s Omnipod (12 aspart and 5 lispro users). Over the course of the study, there was one hypoglycemic event each in the 80% and 50% basal reduction groups vs. seven in the pump suspension group. Mean change in blood glucose from start to end of exercise was -31 mg/dl (80% basal reduction) vs. -47 mg/dl (50% basal reduction) vs. -67 mg/dl (pump suspension). Ms. Zaharieva highlighted that the greatest drop in glycemia occurred with pump suspension at the start of exercise, with 41% of the group experiencing hypoglycemia. All three groups evaded drastic overnight lows – the participants’ diabetes seemed to be fairly well-managed at baseline, so it’s possible that they had enough experience with post-exercise nights to prevent excess insulin dosing. These results are largely confirmatory in our view, and the real key is translating them into widespread clinical practice. Of course, the effect of exercise on glucose levels is highly variable between individuals and within a given individual, where duration, intensity, environment, previous sleep, and other factors should be taken into account. What may be most important is to equip patients and HCPs with guidance on how to define their own exercise dosing strategies. (See Chapter 3 of Adam’s book, Bright Spots & Landmines for more on this topic.)

• Ms. Zaharieva, now a PhD student at York, knows a bit about exercising with type 1 diabetes – she won a bronze medal in Taekwondo at the World Championships in 2013! Check out Insulin Nation’s profile of her here!

Questions and Answers

Q: Did you look at fatty acids?

A: No, we didn’t, but it’s something we want to monitor.

Q: How do you see this evolving in the future? Are you hoping to have larger studies from which you can draw recommendations?

A: Yes, future research really needs to focus on this area. Exercise has not been focused on enough in type 1 diabetes. With closed loop systems now, we need to better address these issues in exercise. We’re seeing that it takes a long time to reduce insulin and it’s important to prevent that drop leading to hypoglycemia. So this could be a potential area to look at with new closed loop systems. I live with type 1 diabetes, and I know a lot of people are scared of exercise. We need to get it out there that exercise is possible and that we just need to do more planning.

Q: I was struck by the individual variability. You had one patient that had hypoglycemia no matter what you did. Any correlations as to why that person was like that?

A: We haven’t looked at that one participant individually. But if you experienced hypoglycemia in all three conditions, you may have had too much basal insulin. This just shows we can’t make a blanket statement for type 1 diabetes. There needs to be individualized change to make it work.

Q: Did you select the participants before?

A: No, it was randomized. We recruited based on flyers.

Q: I wonder if any patients supplemented carbs?

A: There’s always the option of carb supplementation. We’ve looked at ways of how to prevent hypoglycemia without carb addition. In pump suspension, it looks like additional carbs would be necessary.

Posters

OneTouch Via (Calibra Finesse): Comparing Patch vs. Pen Bolus Insulin Delivery in Type 2 Diabetes Using Continuous Glucose Monitoring Metrics and Profiles (73-LB); Optimizing Basal-Bolus Therapy in T2D: A Randomized Controlled Trial Comparing Bolus Insulin Delivery Using an Insulin Patch vs. an Insulin Pen (987-P); User- and Healthcare Provider-Reported Outcomes for a Wearable Bolus Insulin Delivery Patch (995-P)

ML Johnson et al.; RM Bergenstal et al.; M Peyrot et al.

A highly-anticipated trio of posters evaluating J&J’s OneTouch Via (three-day patch insulin bolus delivery device) vs. pen bolus delivery (Novo Nordisk’s FlexPen) in a 62-center, 44-week RCT of type 2s (n=278 with A1c ≥7.5%) ended with very positive results for both groups in terms of A1c outcomes and particularly positive (and better) results for Via group. Both groups experienced similarly robust reductions in A1c (~1.6%-1.7% drops!), but patients and providers alike preferred Via over the pen for implementing basal-bolus therapy, suggesting that the “wearable pen” device, which allows for simplified training and discreet insulin delivery, may well drive greater adherence in the real-world setting. Patients were randomized to use either OneTouch Via or the FlexPen when advancing to mealtime insulin, and all used a pattern-based logbook, which combined SMBG values with a simple insulin adjustment algorithm. Initial evaluation lasted for 44 weeks, after which the participants “crossed over” to the opposite therapy for four weeks. During a baseline period and prior to study visits at weeks 4, 12, 24, 36, and 44, participants recorded three days of 7-point SMBG values, and phone calls with subjects were conducted with patients five times during the first eight weeks to assist with self-titration. A CGM sub-study with Dexcom G4 Platinum was performed in 96 of the participants for seven days at baseline and again for seven days between weeks 22 to 24. Top-line, 44-week results showed that A1c decreases were significant in both groups, but not significantly different from each other: the FlexPen group saw their A1c drop an average of 1.7% (baseline: 8.7%) while those on Via saw A1c drop an average of 1.6% (baseline: 8.6%). In both groups, all of the improvement in A1c came in the first 24 weeks, then leveled off at ~7.0% (~83% of the reduction came by week 12), which is certainly a great result coming from an A1c over 8.5%. Both groups also saw significant reductions in fasting plasma glucose at 44 weeks (roughly -20 mg/dl from baseline of ~160-170 mg/dl), as well as significant reductions in 7-point SMBG (see below). In both groups, mean glucose decreased from ~200 mg/dl to ~140 mg/dl and TDD increased from ~50 units to ~130-140 units in both groups (they were under-insulinized, given the high baseline A1c to start). The only reported SMBG metric that significantly differed between the two study arms was coefficient of variation (CV) of mean daily blood glucose between days (an unconventional metric – not the same as within-day CV on AGP reports); both groups started at ~10.5%, and the Via group’s decreased to 9.4%, while the FlexPen group’s increased to 12% (p=0.02). There were no significant differences in hypoglycemia (defined as blood glucose ≤70 mg/dl), though at night, the pen group reported 74 episodes vs. 60 episodes in the Via group (NS, p=0.09). If anything, the study shows a very robust A1c reduction in both groups, but the edge goes by far to OneTouch Via in terms of patient and provider preferences (see below), particularly given that we know that insulin pen use tends to be far better in randomized controlled trial vs. real-world settings. While it was remarkable how well the FlexPen group did, we imagine lots of this stems from the randomized controlled trial setting – far better than we’d see in real-world use. What this also just reinforces to us is that insulin dosing is complicated and that having “real” input by HCPs is also probably really useful. Dosing this drug is really challenging in any setting, however, and we’d also so excited to see Via used with smart, adaptive dosing algorithms.

• For those in the IDC-run CGM sub-study (n=96), the story was the same at week 24: Both groups saw significant decreases in A1c, time in 70-180 mg/dl, alongside a significant increase in time <70 mg/dl, but none of the between group differences were significant. As a preliminary note, we assume that the G4 Platinum was used in its retrospective, blinded mode, but are not 100% sure. Based on the CGM metrics, both groups benefited tremendously by week 24 from just being randomized in the study, as demonstrated in the AGPs below (flatter, narrower, and more in-range!): In the Via group, time 70-180 mg/dl increased by 6.2 hours/day (48%->74%), time >180 mg/dl decreased by 7.2 hours/day (50%->21%), and time >250 mg/dl shrunk an impressive three hours/day! Variability, measured by standard deviation, decreased from 53 to 46 mg/dl (though when converted to coefficient of variation, it actually increased from 29% to 33%, which is still very stable). OneTouch Via increased hypoglycemia, as time <70 mg/dl and <54 mg/dl increased by a fairly large 50 minutes/day (1%->5%) and 13 minutes/day (0.2%->1.1%), respectively. The outcomes in the pen group were almost identical: Time 70-180 mg/dl increased by 7.9 hours/day (42%->75%), time >180 mg/dl decreased by 8.9 hours/day (57%->20%), and time >250 mg/dl shrunk by 4.5 hours/day! Variability, measured by standard deviation, decreased from 54 to 45 mg/dl (though again, when converted to coefficient of variation, it increased from 27% to 32%). For hypoglycemia, time <70 mg/dl and <54 mg/dl increased by 1 hour/day (1%->5%) and 14 minutes/day (0.2%->1.2%), respectively.

• Subjective reports from the full study population were almost invariably in favor of the OneTouch Via over the FlexPen, from both patients and providers. At 24 weeks, patch users were significantly more satisfied overall, particularly with the device’s ease of use. Relative to the pen group, they also reported improved discretion, lack of pain, ability to do things “spur of the moment,” social comfort, having insulin on them always, and likelihood of recommending. At 44 weeks however, they didn’t feel that they were more confident, had more peace of mind about managing diabetes, could fit diabetes management into their lives, or were taking positive steps with diabetes management relative to the pen group managing insulin – this is surprising to see, as we would have guessed a stronger response on these. Regarding quality of life, daily functions and diet restrictions were rated more highly in the patch group. In the subject preference survey – since both groups got to use both devices for at least four weeks – every response from both groups was significant in favor the patch (e.g., people felt more satisfied with the patch, preferred it, had to carry fewer supplies around, felt more freedom, and would recommend it over the pen). Both groups also said they wanted to switch from the pen to the patch, but the group that used the patch for 44 weeks followed by four weeks of the pen were significantly more likely to endorse this statement than was the group that used the pen for 44 weeks followed by the patch for four. 

  • One of the most important findings from the study, in our view, is that 91% of providers preferred the OneTouch Via over pens to advance type 2 patients from basal to basal-bolus insulin. In their survey, they endorsed that the device helped them transition patients to basal-bolus faster, facilitated a more gratifying relationship with the patient (though not reflected in the patient survey), will help overcome barriers to initiation, is easy to use and train. They also endorsed that they would prescribe the patch for MDI patients both at and not at goal. This was not a blinded study, so we acknowledge that there may be a bit of “whiz-bang” at play here – “wow, look how cool this device is, if I had diabetes I would love it!” or “this is a new device, I’m supposed to prefer it.” Still, getting providers on board, particularly with the endorsements that it is easy to train and could reduce clinical inertia, would do wonders for driving adoption in the type 2 insulin delivery device field.

  • With the connected pen market just beginning to ramp, we do wonder how the equation would’ve changed for both providers and patients if the comparator arm had been a smart pen/pen cap (same goes for overall glycemic outcomes). Might providers take the potentially-heightened hassle of a pen if they were able to view a patient’s mealtime insulin dosing behavior? Would that lead to more productive conversations and the ability to better support patients with titration and between visits? Of course, OneTouch Via could eventually have connectivity built in, though the beauty of the device is its simplicity (the only user interface is the buttons for bolusing) and its small, flat on-body profile – neither of those should be significantly compromised!

• OneTouch Via and Calibra Medical (manufacturer) are still part of J&J, though the company continues to pursue strategic options for it. We doubt that the device will remain as J&J’s standalone diabetes care offering, since J&J accepted Platinum Equity’s offer to acquire LifeScan for ~$2.1 billion earlier in June. We consider Via to have strong potential and the results from this study have very positive notes. The question remains, who could buy OneTouch Via? An insulin company could be a great fit – imagine a pre-filled? It would also make sense for a company like BD, Medtronic, or even Insulet to swoop in to add to bolster its insulin injection/delivery portfolio.

Safety and Performance of the Omnipod Hybrid Closed-Loop System in Adolescents (1376-P) and Children 6-12 Years (1377-P) with Type 1 Diabetes over Five Days Under Free-Living Conditions

Jennifer Sherr, Gregory P. Forlenza, Bruce Buckingham, Thomas A. Peyser, Joon Bok Lee, Jason B. Oconnor, Bonnie Dumais, Lauren M. Huyett, Jennifer E. Layne, Trang T. Ly, New Haven, CT, Aurora, CO, Palo Alto, CA, Billerica, MA

In tandem with adult data presented as an oral (207-OR; see above), Insulet presented these two compelling posters testing the Omnipod Horizon hybrid closed loop in 6-12 year olds (n=15; mean age: 10 years; baseline A1c: 8%) and adolescents (n=10; mean age: 14 years; baseline A1c: 8.2%). Both showed very impressive 3-4-hour/day improvements in time-in-range (!) and two-hour/day reductions in time >250 mg/dl, reflecting far more hyperglycemia in both age groups vs. the adult study. The studies compared 96 hours of free-living hybrid closed loop (tablet computer, Omnipod, Dexcom G4 Share CGM) to seven days of patients’ standard therapy (24% MDI, 76% pump). Meals were unrestricted (manual boluses, per usual care), and participants were encouraged to take part in at least 30 minutes of moderate physical activity per day. Overall, the outcomes remain very strong and should hopefully encourage Insulet to push fast towards a pivotal study!

• In n=10 adolescents, Horizon drove a remarkable +4.4 hour/day improvement in the range of 70-180 mg/dl (79% vs. 61%, p=0.01), four fewer hours per day >180 mg/dl (18.5% vs. 35%, p=0.02), a tremendous two-hour/day reduction in time >250 mg/dl (3.5% vs. 12%, p=0.02), and a non-significant 43% reduction in time <70 mg/dl (-27 mins/day: 2.5% vs. 4.4%, p=0.32). Mean glucose also declined 19 mg/dl – from 163 mg/dl on open loop to 144 mg/dl on closed loop – though the small sample size meant this barely missed statistical significance (p=0.08). Overnight glucose outcomes looked even better, with an even larger differential on all time-in-ranges (<70, 70-180, and >180) – see the first two pictures below. Wow! Adolescent participants were in closed loop for an average 98.5% of the time – very strong considering use of the investigational study devices. We are noticing extremely high “time in closed loop” with Insulet and wonder if that has more to do with the wearability of Insulet’s tubeless Omnipod or just strong Bluetooth communication in these hotel studies. Average meal size was 57 grams of carbs (max: 150 grams!); average exercise duration was 66 minutes/session (n=40); and participants increased the target to 150 mg/dl prior to ~1/3 of exercise sessions.

• In 6-12 year olds (n=15), Horizon similarly drove a massive +3.4 hour/day improvement in the range of 70-180 mg/dl (69% vs. 55%, p=0.003), three fewer hours per day >180 mg/dl (29% vs. 42%, p=0.007), a tremendous two-hour/day reduction in time >250 mg/dl (9% vs. 18%, p=0.03), and a non-significant 24% reduction in time <70 mg/dl (2.2% vs. 2.9%, p=0.23 – even if non-significant, it is still ten minutes a day!). Mean glucose declined a striking 22 mg/dl – from 178 mg/dl on open loop to 156 mg/dl on closed loop (p=0.02). Overnight glucose outcomes looked just as good as the 24-hour outcomes – see the second third and fourth pictures below. Similar to the adolescent data, participants were in closed loop for an average 98.3% of the time. Average meal size was 58 grams of carbs (max: 186 grams!); average exercise duration was 68 minutes/session (n=61); and participants increased the target to 150 mg/dl prior to 43% of exercise sessions. This is very strong hybrid closed loop performance in a very young population!

Adolescent Data:

6-12 Year Old Data:

Artificial Pancreas with Glucose Prediction Trust Index Improves Time in Target Glucose Range vs. Sensor-Augmented Pump Therapy (961-P)

JE Pinsker, AJ Laguna-Sanz, MM Church, JB Lee, LE Lindsey, C Andre, FJ Doyle, E Dassau

A Sansum/Harvard proof of concept clinical study found the addition of a trust index enhanced controller responsiveness to hyperglycemia and hypoglycemia during a 48-hour artificial pancreas session (88% time in range). The trust index essentially compares in real time how past artificial pancreas model predictions have differed from the measured CGM at each time point. Based on the generated prediction errors, either a high or low trust index is calculated, which modulates the intensity of the controller’s suggested insulin delivery. In hyperglycemia, a high trust index triggers the controller to deliver insulin more aggressively, whereas a low trust index causes the controller to pull back and deliver a more moderate dose. Likewise, during hypoglycemia, a high trust index allows the controller to reduce insulin delivery, with some basal insulin permitted if excursions are small and/or the predicted glucose is rising, whereas a low trust index results in pump suspension. Participants (n=15) underwent a one-week run-in phase with sensor augmented pump therapy, followed by a 48-hour artificial pancreas session with the addition of the trust index (Dexcom G4 CGM and Omnipod pump) in a supervised environment. Time-in-range (70-180 mg/dl) significantly increased from 75% to an impressive 88% (+3.2 hours/day) and time <70 mg/dl decreased significantly from 8% to 2% (-1.5 hours). Variability as measured by SD and CV also significantly decreased. Time >180 mg/dl increased, but the differences were not statistically significant. Overnight results showed similar improvements in time-in-range and time <70 mg/dl. Overnight reductions in time >180 mg/dl were also significant, halving from 17% to just 8%. Impressively, these improvements were achieved even in the setting of one hour of unannounced exercise. This was the very first proof of concept study to evaluate use of the trust index, and results clearly demonstrated a serious benefit over SAP. We’d love to see what kind of benefit this enhancement provides over the non-trust-index algorithm itself in a standard hybrid closed loop system. Harvard’s Dr. Frank Doyle III (a co-author of the poster) referenced this study during a presentation at ATTD – at the time, Dr. Doyle claimed that the impressive 88% time-in-range may set a “new high water mark” for his own work. We agree! We’d add that the SAP arm was doing extremely well at baseline in this study – 75% time-in-range, mean glucose of 135 mg/dl – so this study could highly underestimate the differential between the enhanced algorithm and standard of care.

Assessment of Infusion Set Survival of the Newly Developed Lantern Catheter in Type 1 Diabetes by Glucose Clamp Technique (89-LB)

A Ajsic, M Krasser, R Juliussen, P Schøndorff, M Heschel, T Póttier, T Augustin, T Pieber, G Treiber, and J Mader

In an important next-gen infusion set study, 16 type 1s wore Unomedical’s novel Lantern catheter set for seven days safely, though with a trend toward reduced insulin action and more hyperglycemia over time. We first saw the Lantern catheter at ATTD 2017 and again at ATTD 2018. The catheter includes several slits along the side to allow insulin to flow out of multiple places (e.g., for occlusion or kinking), and in new news, a next-gen version of Lantern was tested here with a “coating” to suppress the body’s foreign body response over seven days of wear. Euglycemic clamps were performed on days 1, 4, and 7 of Lantern infusion set wear following a subcutaneous bolus of 0.15 u/kg Humalog and compared to the same clamp on day 1 wear with a “standard infusion set. In the clamp studies, max glucose infusion rate was the same in the control and across days of Lantern use, but the differences appeared in time to max glucose infusion rate and area under glucose infusion curve (from 0->240 minutes). On days 1, 4, and 7, the time to max glucose infusion rate was ~111 minutes, ~57 minutes, and ~49 minutes respectively, indicating a progressively faster time to peak insulin action as length of wear increased (i.e., faster PK/PD). However, area under the curve from 0 to 240 minutes decreased from ~804 mg/kg (day 1) to ~799 mg/kg (day 4) and to ~596 mg/kg (day 7), indicating lower overall insulin action/glucose lowering over time. Indeed, the poster indicates a doubling in hyperglycemia as the study went on: median time >180 mg/dl was 20% on day 3, 36% on day 6, and 40% on day 7. It will be interesting to see if similar outcomes are seen in a larger scale trial under routine conditions – especially as Unomedical does a closed-loop study where a basal modulation system could theoretically adapt to the hyperglycemia. But if there is a tradeoff of more hyperglycemia and more insulin use by day 7, that will need to be considered carefully against the merits of longer wear. Extended wear is a holy grail for infusion sets, as it would greatly reduce the burden associated with pumping insulin and also open the door to a CGM-infusion set combo device that lasts up to seven days. (Medtronic hopes to launch such a set after April 2020, per its 2018 Analyst meeting.)

• This poster brought the first news that Unomedical has two versions of Lantern in development: (i) the original one we’ve seen at the past two ATTD’s, which has the addition of slits to a standard soft catheter for three-day wear; and (ii) “Coated Lantern,” which includes the slits along with the coating to enable extended wear.

• Lantern aims to address catheter occlusion, bending, and kinking via multiple slits on the side. For comparison, BD/Medtronic’s MiniMed Pro-set with FlowSmart has a single hole on the side to address occlusion, but encountered turbulence during its limited launch, and it’s unclear when it will be re-deployed. Last we heard a year ago, the set was slated to launch by September 2018, but we suspect this date may have been pushed back as the Pro-set was notably absent from 1Q18 and 2Q18 calls.

Time profiles of intravenous glucose infusion rates following a bolus of insulin lispro (0.15 IU/kg) administered with coated infusion set with Lantern Technology on Day 1, 4, and 7 of wear

Three OpenAPS Posters: “Autosensitivity” Feature (79-LB); OpenAPS in Korea (964-P); OpenAPS in Italy (993-P)

D Lewis et al. (79-LB); SB Choi et al. (964-P); V Provenzano et al (993-P).

The DIY diabetes community definitely had a stronger presence at this year’s ADA: with the first-ever ADA DIY symposium (see below) plus an OpenAPS oral presentation (see above) and three OpenAPS posters, the trailblazing community is finally sharing its outcomes and features more publicly. A poster (79-LB) co-authored by Ms. Dana Lewis and Scott Leibrand showed the benefits of a new OpenAPS feature, Autosens, which allows for real-time assessment of insulin sensitivity factor (ISF) and glycemic targets using the past 24 hours of data. The feature analyzes CGM data, comparing observed glucose changes to expected impact from insulin; the goal is to identify if a person’s sensitivity has changed (e.g., unexpected insulin resistance). In this analysis, Autosens was run retrospectively on a cohort of 16 individuals using OpenAPS (13 adults and 3 children <18 years) and identified one individual who tended towards insulin sensitivity and three individuals who tended towards insulin resistance. In other words, Autosens could be used for some meaningful real-time adjustments of insulin dosing – e.g., making an algorithm more aggressive when insulin resistance is observed. Of course, the feature could also be used in a retrospective analysis to understand patterns of changes. The authors suggested using the method to further explore the following areas: (i) growth/hormone-related sensitivity changes; (ii) circadian profiles and monthly variation; and (iii) the effects of insulin pump site changes. We love how the DIY community continues to be ruthless in its efforts to innovate, truly setting a new standard for the pace of algorithm improvement. These next-gen approaches could certainly make closed loop algorithms much more adaptive and personalized, changing with users’ insulin needs over time.

• We were also pleased to see two posters demonstrating the benefits of OpenAPS in international populations – one in Korea (964-P) and one in Italy (993-P). In the Korean poster, three months on OpenAPS (Dexcom G5 CGM with open protocol Sooil Dana R pump) drove notable outcomes in nine participants: time-in-range increased from 70% to 83% (+3.2 hours/day; p<0.001), time <70 mg/dl decreased from 5% to 3% (-24 minutes/day; p<0.004), time >180 mg/dl decreased from 25% to 13% (-2.7 hours; p<0.001), and a A1c dropped by 0.5% (baseline: 6.8%; p<0.001). Glycemic indices were also recorded for 13 patients over 3-6 months, showing 82% time-in-range, 5% time <70 mg/dl, and 14% time >180 mg/dl. Nice! The Italian poster (n=29) was similarly impressive: time-in-range increased from 69% to 86% (+4.1 hours/day; p<0.05), time <70 mg/dl decreased from 9% to 3% (-1.4 hours/day!; p<0.05), and A1c decreased by 0.9% (baseline: 7.2%; p<0.01). Time >180 mg/dl was halved from 22% to 11%, but was not found to be significant; clearly hypoglycemia was the key issue in this group. No adverse events were recorded. The poster does not provide details on the length of the follow-up period – still, the dramatic decrease in hypoglycemia alone makes a very compelling case for OpenAPS.

Late and Missed Meal Boluses with Multiple Daily Insulin Injections (992-P)

L Norlander, L Ekhlaspour, B Buckingham, L Hsu, S Loebner, G O’Malley, C Levy, D Lam, S Ogyaadu, S Anderson, J Robic, M Breton

A Stanford/UVA/Mt. Sinai poster showed that adults and adolescents (n=24) using smart pens in a study regularly miss or are late with mealtime boluses. While previously this kind of data would have been rendered largely “invisible,” the combo of CGM (Dexcom G5) and connected pens (Novo Nordisk NovoPen Echo Plus or NovoPen 5 Plus) helped investigators fill in the gaps in insulin dose history. Seven days of data were analyzed upon initiation of the devices followed by an additional seven days one month later. 1,173 meals in total were evaluated, identified by either the subject manually recording the meal through the TypeZero InControl Decision Support System app, or if the CGM read >70 mg/dl and there was a >70 mg/dl rise within two hours. A late meal bolus was defined when CGM increased >50 mg/dl from baseline prior to the insulin dose, and a missed meal bolus was defined by no insulin dose within two hours before the start of the CGM rise (see picture below). In total, 27% of meals had either a late or missed meal bolus, with 13% of total meals accompanied by a late bolus, and 14% of total meals with no bolus whatsoever. We’re willing to be that the rates of missed or late boluses in the general population – particularly when type 2 is included – may be higher than in this study. Importantly, although not surprisingly, there was a positive correlation between the percentage of missed meal boluses and A1c levels (p=0.02); there was no significant correlation between late boluses and A1c. It would be interesting to calculate how many missed/late boluses it would require for an individual to have any given A1c – this data was not provided, but has been shared in publications on pumps (Burdick et al., Pediatrics 2004).  We’d also be fascinated to see how missed boluses affect other outcomes like time-in-range and PROs like diabetes distress and treatment satisfaction. No significant differences were observed when comparing results between the first and last weeks of the study. We think this is an incredibly important trial, clearly demonstrating the value of connected pens and digital dose capture for injection users. Hopefully, as more connected pens launch, MDI data will facilitate more informed conversations between providers and patients and drive more efficient/effective insulin use.

Trends in Insulin Pen Priming (83-LB)

G O’Malley, S Ogyaadu, D Lam, L Norlander, J Robic, C Levister, S Anderson, L Hsu, S Lobner, L Ekhlaspour, M Breton, B Buckingham, C Levy

A study conducted by Icahn School of Medicine, UVA, and Stanford used the NovoPen 5 and NovoPen Echo to find that the frequency of insulin pen priming varies substantially among patients with type 1 diabetes (n=25). 16,135 pen records spanning 76 days of pen use per patient were analyzed and revealed that a surprisingly high 80% of overall injections were accompanied with a prime; however, the range of frequencies was quite large, varying from 2% to 99%. As evidenced below, priming frequencies were not evenly distributed – patients tended to either prime the majority of the time or very infrequently. Percentage of missed priming appeared to correlate with male gender and younger age (a statistical analysis was not provided) but did not correlate with A1c. Zooming out, this study highlights another valuable aspect of connected pens and dose capture, supplementing the poster above (where 27% of meals were accompanied by a late or missed bolus). Such data could be extremely informative in enhancing patient education and treatment, as well as in fostering useful discussion between providers and patients.

Functionality Evaluation of Investigational Continuous Subcutaneous Insulin Infusion (CSII) Set vs. Commercially Available Sets—Assessment of 3D Volume and Surface Area over Eight Days (980-P)

D Diaz, A Dinesen, A Khalf, G Eisler, C Loeum, M Torjman, P Strasma, and J Joseph

This swine study from Capillary Biomedical shows greater diffusion of a bolus of insulin + contrast agent with its prototype extended wear infusion set (“soft polymer material with multiple ports”) compared to commercially-available sets. All sets were inserted into the subcutaneous tissue of swine every other day for eight days – 70 ul of insulin and contrast were infused through each cannula five minutes prior to excision, when the specimens were frozen and imaged with Micro-CT (X-ray microtomography). Based on these in vivo animal results, the boluses delivered by investigational sets had both larger surface area and volume than the other sets. The authors suggest this result could translate to more consistent insulin absorption over time by reaching more vessels – the results have guided the design and development of a next-gen commercial infusion set. As a reminder, this research received a $1.5 million grant from JDRF in 2016 – at the time, Capillary Biomedical was targeting a first commercial device by 2020. Given how long it took BD to develop FlowSmart (MiniMed Pro-set) – which still hasn’t relaunched – this timing could be ambitious. In addition to the sprinkler-like multi-port design and extended wear, the set should have atraumatic insertion. Previously heating/vibration were expected, but we’d be surprised to see them in a commercially-viable product. We look forward to seeing longevity and PK/PD studies in humans.

Glucose Control Using a Standard vs. an Enhanced Hybrid Closed-Loop System: A Pilot Study (986-P)

B Paldus, MH Lee, HM Jones, SA McAuley, JC Horsburgh, KL Roem, GM Ward, RJ MacIsaac, N Cohen, PG Colman, AJ Jenkins, DN O’Neal

An Australian randomized crossover study (n=11) found that an enhanced version of the Medtronic MiniMed 670G with “enhanced correction insulin bolus reminders and modifications broadening insulin delivery parameters” reduced the number of alerts and closed-loop exits in pump-experienced type 1 adults, while also achieving non-significant improvements in time-in-range and average glucose. Type 1 adults were assigned standard- and enhanced-HCL systems in random order for one week each (with one week washout in between) in a supervised hotel setting. Challenges included two missed meal boluses, two high-glycemic index and two high-fat meals, and 40-minutes of moderate exercise. Alerts significantly decreased (nearly halved!) with enhanced-HCL (3.9 per week compared to 8.6, p = 0.01), and auto-mode exits decreased from 3.5 per week to none (n = 0.004). Importantly, such refinements did not come at the expense of glycemic outcomes. Time-in-range (70-180 mg/dl) was greater with enhanced-HCL than standard-HCL by one-hour per day (74% vs. 69%), although the difference was not statistically significant. Mean sensor glucose also improved with the enhanced-HCL, but again the differences were not significant (151 mg/dl vs. 160 mg/dl). There were no severe hypoglycemia or DKA events in either arm. Time <70 mg/dl (1.2% vs. 1.9%), time >180 mg/dl (24% vs. 29%), and time in auto mode (98.5% vs. 96.9%) all favored enhanced-HCL, though again not significantly. As the poster notes, this was an exploratory study with multiple challenges not powered to detect such differences. Accordingly, the glycemic, time in auto mode, and auto mode exit figures also can’t be compared to those seen in the real world. All participants expressed a preference for the enhanced-HCL during standardized interviews at study completion. Given that many 670G users cite copious alerts and closed-loop exits as primary sources of frustration, we believe the enhanced-HCL would be welcomed with open arms and also improve outcomes across the population of Medtronic hybrid closed loop customers.

• We’re not sure if this “enhanced-HCL" system is the same as the next-gen hybrid closed loop with DreaMed and automatic corrections boluses. At Keystone, Medtronic Diabetes Global CMO Dr. Fran Kaufman laid out goals for this future system (2+ years away): (i) Keeping people in closed loop 100% of the time; (ii) >80% time-in-range without changing the hypoglycemia profile in any significant way; (iii) lower target, both overall (currently 120 mg/dl in 670G) and for corrections of elevated glucose (corrections not allowed below 150 mg/dl in 670G); (iv) improving the meal experience (perhaps a simple meal announcement with no carb counting – “it’ll give you a little bolus and you’ll be on your way,” said the lead 670G engineer – sounds like the Beta Bionics iLet system); and (v) constraints on time in maximum and minimum basal dose extended (in 670G, 4 hours at maximum basal dose or 2.5 hours at no insulin leads to a likely exit from auto mode – right at the times when Auto Mode is most needed!).

Effect of Medicare (CMS) Insulin Pump Policies in T1D (974-P)

N Argento, J Liu, A Hickey, E Gautschi, A Mcauliffe-Fogarty

According to a T1D exchange survey, 59% of type 1 Medicare beneficiary insulin pump users (n=241) found their experience with Medicare to be worse or much worse than their experience with commercial insurance. Current Medicare policy requires that patients on insulin pumps have face-to-face visits with clinicians every quarter in order to continue receiving pump consumables. 56% reported that they had difficulties getting supplies, and 39% reported that they had changed some of their pump-related behaviors as a result of the Medicare policy. The most common issues for patients on Medicare-funded insulin pumps were delays in receiving supplies (reported by 30% of respondents), difficulty with paperwork (24%), problems scheduling a face-to-face visit with a provider every three months (18%), and difficulties informing suppliers that they had been seen for their face-to-face visit (11%). Of the 39% of respondents who reported that they changed their pump related behaviors due to Medicare policies, nearly half reported they had more erratic blood sugar (48%), increased anxiety or frustration (44%), or higher blood sugars (42%). Others reported more pain or irritation at sites (34%) and more scarring of sites (28%) – presumably related to extended set wear – more hypoglycemia (17%), and 15% were forced to receive emergency supplies from a local source. A similar poster from ADA 2017 from Drs. Nicholas Argento and Anne Peters suggested that the quarterly visits may be detrimental to patient safety. We’ll quote that same coverage from a year ago: “…it may be time for CMS to reconsider the quarterly visit policy for pump users.” To, this is a perfect opportunity for telemedicine to step in, particularly as the current administration’s CMS seems to be growing more comfortable with the idea of reimbursing providers for remote visits (see recent proposed Medicare Physician Fee Schedule).

Symposium: The Diabetes Do-It-Yourself (DIY) Revolution

The Diabetes DIY Technology Revolution—How Patients Are Changing the Innovation Pipeline and Care

Jason Wittmer, MD (Mercy Medical Center, Des Moines, IA)

In a persuasive, passionate talk, Dr Jason Wittmer shared his family’s personal experience with OpenAPS and urged industry to learn from the Do-It-Yourself (DIY) community - “Industry and DIY don’t have to stand on opposite sides of the street.” We loved his reflection on the history of CGM remote monitoring in 2013-2014 – at the time, the DIY community had developed Nightscout (John Costik), which ultimately pushed industry and FDA to approved remote monitoring solutions like Dexcom Share. Now in 2018, Dr. Wittmer believes that commercial hybrid closed loop (e.g., MiniMed 670G) is at a similar point as CGM remote monitoring was in 2014 – the DIY community, through OpenAPS and Loop, is years ahead of industry products. Indeed, Dr Wittmer shared some of the impressive DIY hybrid closed loop enhancements available in OpenAPS and Loop, including pump dosing directly from the phone and watch; engaging with the system through Siri, Google Assistant, and Alexa; unannounced meal compensation; personalized blood glucose targets; blood glucose predictions; Autosensitivity/ AutoTune (automatic analysis and adjustment of ISF, basal rates, carb ratio); remote web app monitoring and adding carbs (“Oops, we forgot to bolus him for breakfast”); dynamic carb absorption tracking (see below); incorporation of Fiasp’s pharmacokinetics into algorithms; automated exercise target setting using Google Calendar; and more! Dr. Wittmer, confirmed what we heard from SOOIL at Diabetes Mine earlier at ADA: the company plans to submit to FDA with an open communication protocol insulin pump, Dana RS – truly tailored for the DIY community with smartphone control. Dr. Wittmer said the company hopes to have it available in the US “by this time next year,” something we confirmed at SOOIL’s booth. Interestingly, he highlighted that SOOIL actually markets Dana RS outside the US as “compatible” with DIY – an awesome point that open communication protocols may indeed give companies a competitive marketing advantage. He made an impassioned plea for healthcare providers not to dismiss DIY systems as “unsafe,” since they are designed by patients and parents themselves (i.e., no one is more motivated for safety than the creators of these systems!). We hope all industry players hear the call and engage in some way with the DIY community – or at the very least, watch the systems from afar and learn what features to incorporate into next-gen products! “Industry and DIY don’t have to stand on opposite sides of the street,” he noted, “A lot of DIY diabetes is not motivated by trying to tinker with diabetes or program computers. We’re doing it because we have to do to serve an unmet need.” Well said!

• In fourth grade (before OpenAPS), Dr. Wittmer’s son visited the nurse 420 times during the school year; in sixth grade (with OpenAPS), he has visited the nurse five times – a 99% reduction! “This is life-changing stuff,” he emphasized, based on the quality of life improvements alone. Overall, his son has spent 85% of the past 152 days in 70-180 mg/dl.  Dr. Wittmer highlighted the OpenAPS Outcomes from ADA 2016, which were also headlined by significantly improved sleep and time-in-range.

• Interestingly, there are now multiple do-it-yourself CGM receiver apps: XDrip+ (Android) and Spike (iOS; shared at Diabetes Mine on Day #1 of ADA). These offer wireless connections to the Dexcom CGM and FreeStyle Libre (if used with an added transmitter); extend use of the G5 transmitter beyond the three-month hard shutoff; allow for data entry and display; and add forward-looking predictive tools. DIY users have even developed technique to replace their Dexcom transmitter’s batteries, allowing for reuse. “These things cost a lot of money. I have great insurance, but for people self-funding and paying out of pocket, that could be the difference between having a CGM and not having one.”

The Data Behind DIY Diabetes—Opportunities for Collaboration and Ongoing Research

Dana Lewis (OpenAPS, Seattle, WA)

OpenAPS founder Ms. Dana Lewis drew a big crowd in the BIG hall – and boy, did she not disappoint! In a whirlwind talk, she described: (i) Diabeloop’s use of OpenAPS features/settings; (ii) four tools created by the DIY community (including the brand new OpenAPS Simulator, which allows patients to probe the effects of behavioral/algorithm changes); (iii) the “OpenAPS Data Commons” (where patients can donate data and researchers can use it for analyses – Stanford has already performed a machine learning study with it); and more! This movement has come such a long way from the days when Ms. Lewis was simply hacking her Dexcom to increase the volume of her alarm – or back in the day (2009) when she was a summer associate at Close Concerns!

• Ms. Lewis got permission prior to the talk to divulge that Diabeloop will be incorporating features inspired and at least partially designed by the OpenAPS community in its second-generation closed loop, expected to begin rolling out in early 2019. One of the modules, a fallback setting for when the patient is kicked out of closed loop, was “nearly copied and pasted” from OpenAPS, Diabeloop CEO Mr. Erik Huneker told us. The company would also like to include a variant of OpenAPS’ “eating soon” setting in the second-gen system (see below; it ramps basal insulin 45-60 minutes before eating, getting some insulin in early). Notably, it sounds like inclusion of these modules will only require an in silico study in Europe to demonstrate safety. The first iteration of Diabeloop’s automated insulin delivery system is still set to roll out in France, the Netherlands, and Sweden in late 2018 – he would not yet tell us whether Cellnovo’s, Kaleido’s, or both patch pumps would be used in the initial launch. To be clear, Diabeloop’s primary control algorithm is proprietary, not open source.

  • Ms. Lewis also disclosed that she is a consultant with Tandem, Roche, Lilly (pending), and willing to talk to more companies! We love to see the lines of industry and patient-led innovation blur, and believe that user-centric settings like “eating soon” could make a meaningful difference on the outcomes and burden fronts. We think as many companies as possible who could speak to her as possible would be terrific.

  • Also at ADA, SOOIL Korea announced plans to submit its smartphone-controlled pump to the FDA/CE Mark with an open communication protocol. Mr. Justin Walker also mentioned plans to register a version of the open source OpenAPS algorithm, implying it could connect to the pump via an app.

• Continuing with the flow of new announcements, Ms. Lewis introduced a new study, titled “OPEN” (Outcomes of Patients’ Evidence, with Novel, DIY AP tech). The study is supported by a grant from the EU Horizon 2020 RISE program, and will congregate an international consortium of traditional- and patient-experts (from Ireland, Denmark, Germany, the UK, Belgium, Australia, and the US). The description stresses knowledge-sharing and examination of clinical and quality of life outcomes for real-world DIY closed loop users, as well as further improvement of DIY automated insulin delivery technology and access. This study, along with an observational data collection study spearheaded by Helmsley Charitable Trust and JDRF, should help to legitimize the use of open source closed loop systems in the eyes of the clinical community, and perhaps even make a case to FDA for safety/efficacy.

• Ms. Lewis described four tools created in the DIY community, led by the never-before-seen OpenAPS simulator.

  • The new OpenAPS simulator generates a “baseline” glycemic profile for individuals, allows them to change preferences or the algorithm code (“What if I…”), and then simulates outcomes vs. baseline. Specifically, users can take their personal Nightscout URL and run “autotune” (below) on it. They then generate a baseline scenario and can ask any “what if...” question and observe the predicted effect on blood glucose. For example, a patient could ask “what would happen to my glucose profile if I didn’t set a temp target prior to exercising?” or “what if I went from open loop with just a CGM to DIY closed loop?” and receive an informed answer promptly. The biggest value of the simulator in Ms. Lewis’ mind is that it allows patients to make decisions about specific behaviors in light of their probable consequences – “it empowers us.” She provided two real-world, n=1 case studies:

    • In case #1, an adult enters carbs into his OpenAPS closed loop system, but never boluses for meals – what would happen if he stopped announcing the carbs? At baseline, he was spending ~90% time in-range, with an estimated A1c of 5.7%. The simulator predicted that he would lose ~2 hours per day in-range (now ~81%), and his A1c would increase to 6.1% (+0.4%). After seeing this projection, the man could decide whether those two extra hours in-range every day are worth the tradeoff of having to continually enter carbs. Now that is personalized medicine and informed decision making!

• In case #2, which was more of a validation, the simulation modeled expected outcomes in a teen who doesn’t announce meals or bolus. The actual outcomes are very close to those simulated ­– see the figure below.

• Dr. Boris Kovatchev described a similar tool called “digital twin” at EASD, though we’re not exactly sure how they are similar/different - .

• The “eating soon” mode aims for a smoother meal ride by dosing with a small amount of insulin 45 minutes to an hour before eating. This is not a pre-bolus, but instead a way for basal insulin to start ramping in preparation for a meal – the individual still takes a bolus at mealtime. Ms. Lewis said that delivering a small amount of insulin before a meal “prepares” the liver for the incoming food – this was corroborated by her sample glucose trace. The mechanism underlying the liver’s “preparation” is reportedly unknown, but Ms. Lewis hopes a researcher will look at why and how well it works. 

• “Autosensitivity”– allowing for real-time assessment of insulin sensitivity factor (ISF) and glycemic targets using the past 24 hours of data – and “Autotune” – which recommends changes in basals, ISF, and carb ratio by drawing from a much larger pool of data continue to improve through iteration. A late-breaker (see 79-LB above) from Ms. Lewis et al. shows how the autosensitivity feature performs retrospectively on 16 pumpers’ data – it picks up on changes to insulin sensitivity that a standard pump would have missed.

• A DIYer wrote a script to track changes in treatment intensity. Pre-OpenAPS closed loop, the module found that a family was performing 4.5 manual interventions (e.g., bolusing) per day; on OpenAPS, several years later, they are performing just 0.7 per day. Assuming comparable or better glycemic management, interventions per day is a really unique and effective way to look at diabetes burden – leave it to the patient community!

• After receiving countless requests from healthcare providers for her DIY data, Ms. Lewis devised a platform called “OpenAPS Data Commons” on Open Humans. The brilliant philosophy behind the Commons is to allow DIY users to donate data if they want to, and to allow researchers and members of the community to access and study it. The uploader simply takes the user’s NightScout URL, anonymizes data, and takes it into the data commons. 121 people have joined so far. This could be an extraordinarily rich data set, with countless parameters recorded from the hundreds (if not >1,000) DIY closed loopers in the world – as of the talk, Ms. Lewis estimates that there are over 710 DIYers all over the world, amassing 5.2+ million real-world closed loop hours. (We think this is probably a big underestimate, especially when Loop users are included – that group is at least over 1,000 we believe!) In her words, “that’s a lot of data. That’s a lot of knowledge.” No kidding! Researchers interested in accessing the data need only fill out a request form … and agree to two important stipulations: (i) Researchers have to share insights back with the community (even if it’s just “person #47 has frequent lows”; OpenAPS has an anonymous way to message person #47 and tell her so she can address the lows); and (ii) If the researcher does publish the data – hopefully in a timely manner – they will do so in an open manner. That means the manuscript could be published open access, or a copy of it could be provided to the community. That offer sounds tantalizing to data-hungry researchers, and it has already drawn a fair number who have reached out, though many are unfamiliar with OpenAPS. Ms. Lewis observed that, just as patients have a learning curve with OpenAPS, so do researchers who “may not be prepared to take full advantage of the rich data available. We (OpenAPS) are happy to help you understand the benefits and opportunities of using this data.” She also realized that some researchers with different skillsets (e.g., coding languages) may struggle to analyze the data, so she created open source tools to help people use the data (on Github). She tied it all up with a single slide: “We have the data…will you work with us?”

  • Stanford has already leveraged the OpenAPS data commons to assess the accuracy of a machine learning prediction models (pending publication). The investigators used three weeks of data in January 2017 from 14 de-identified DIY closed loopers. They then compared the glucose prediction accuracy of their machine learning models against the actual glucose readings from the DIYers. Ms. Lewis did point out that the comparison isn’t quite apples to apples because OpenAPS actually acts on the prediction every five minutes, but the ease of obtaining the OpenAPS data set clearly makes the assessment worthwhile.  

    • Ms. Lewis astutely noted that the pace of OpenAPS data aggregation presents a challenge for the traditional pace of academic studies. For example, the Stanford machine learning study was completed in January 2017, and it still hasn’t been published. Since the analysis, the OpenAPS code on which the publication is based has changed twice. This is why researchers are asked to share findings with the community as soon as possible.

    • Ms. Lewis listed other possible studies she’d like to see done with the OpenAPS data commons: (i) variability in insulin sensitivity in youth at different ages; (ii) assessment of sensitivity changes in pregnancy as well as other tools for managing insulin dosing during pregnancy; (iii) assessments of meal-time strategies; and (iv) insulin sensitivity changes and dosing adjustments around menstrual cycles. Talk about low-hanging fruits for data scientists looking to pump out quick, meaningful publications!

• This isn’t new news per se, but Ms. Lewis was introduced as a DIY patient innovator and PI of a new Robert Wood Johnson Foundation grant. The project is titled “Learning to not wait: Opening pathways for discovery, research, and innovation in health and healthcare,” and will entail creating a team of on-call data scientists to support research in the DIY community, exploring barriers to this type of research, and more.

The Front Lines of DIY Diabetes – State of the State and Closing the Loop

Joyce Lee, MD (University of Michigan, Ann Arbor, MI)

In a rousing talk, University of Michigan’s Dr. Joyce Lee acknowledged that safety and regulation are important, and despite this, depicted graphically just how much the DIY community has accelerated the pace of innovations. In the first figure below, the DIY community is above the timeline, while Dexcom, Tandem, and Medtronic are below the timeline – from the image, it’s fairly clear that since 2014, the DIY community has outpaced industry in a meaningful way, and likely even driven it forward. The second chart shows new Nightscout features over time: The user community has incorporated ~51 new features beyond “remote monitoring” and is in its eighth major software rollout; meanwhile, Dexcom has incorporated 13 novel features and five new platform rollouts in the same span. DIY is steps ahead industry and even driving industry forward, though of course DIY is only meeting the needs of a small fraction of patients and mass commercialization needs to be regulated and enable sustainable businesses. However, we were glad to see at ADA greater industry recognition of the DIY community, in part through this symposium, and partially through partnerships like those mentioned above (SOOIL, Diabeloop).

• Dr. Lee outlined three lessons that we have learned from the DIY community: (i) Individuals with diabetes and caregivers have expertise and increasing autonomy (“Individuals are willing and ready and able to push the needle forward, as we’ve been a little complacent as to how we’re delivering care”); (ii) We can’t ignore and/or underestimate the importance of social media for supporting and creating health (There are now 55,000 NightScout users globally, and people in the diabetes community seem to trust peers more so than providers for advice); and (iii) We must learn from the DIY community about large-scale health production created through human cooperation and collaboration (“Lots of contributions are being made, and unnoticed. You have a leadership team, a support team, a core development team. The code piece gives a window into the value of online collaboration and communication.”)

• University of Michigan will be undertaking a project this summer using the OpenAPS Data Commons focused on the relationship between a low carb diet and time-in-range. Through the Data Commons, patients can donate data and researchers can use it for analyses. In Adam’s experience, low carb has been the number one way to maximize time-in-range – he generally sees 75%-80% time in 70-140 mg/dl on Loop, and nearly 100% time in 70-180 mg/dl. (See his diaTribe column on time-in-range for more on this front.) We’re excited to see how these data turn out. We do wonder how Dr. Lee and co. are going to track diet (retrospectively) – patient estimates of carb counting entered into pumps? Manual tracking?

• Following the DIY talks, audience members including Drs. Lutz Heinemann, Kittie Wyne, and Grazia Aleppo engaged with Dr. Lee, Ms. Dana Lewis, and the other panelists about how providers can and should navigate DIY in light of concerns over liability. See the interesting exchange, which ended with Dr. Wyne proposing that a consensus on the legal aspects of DIY for providers be written, below.

Selected Questions and Answers

Q: What does the healthcare provider do? We have regulations. I want to help, but I also have to protect the patient and me.

Ms. Dana Lewis: You can answer questions and say, “I can’t prescribe or recommend,” but you can still listen to them and point them to places where they can get help. My hope is that through collaborations with companies, there will be approved versions of things that came out of the DIY movement.

Dr. Grazia Aleppo (Northwestern University, Evanston, IL): This is a wonderful situation – well sometimes, there may not be WiFi, no connection. Do I need to worry about going to manual mode basal rate? How do I assist them to stay safe if something is going wrong? We want to make sure that when they’re not in DIY they’re still ok.

Ms. Lewis: It’s not if it’ll break or turn off, but when. It’s designed to fall back safely to a standard pump situation and give an alert that you’re not looping, so pay attention. It’s important for companies to be aware, and to design for that too. Just because a product is on the market, that doesn’t mean we don’t have to have conversations about what happens if it breaks, because it will break. Thank you for your interest in supporting patients who are doing DIY.

Dr. Lutz Heinemann (Profil Institute, Neuss, Germany): In Germany, we had an outcry in the academic world about using such systems which are not medical products (approved). What are the legal aspects, and so on? I somehow missed this in the presentations.

Dr. Lee: As a clinician, I highly endorse the systems, their outcomes, the achievements of the systems. I suggest that patients use the Facebook group and talk to others who have built the systems. That’s where I draw the line. I don’t think as a clinician I can build the system for them. I can’t medically endorse it, but I can say “you can check this out, talk to the community.” People come in all the time with rigged-up stuff – it’s not about legality, it’s about a patient who makes a choice, then it’s our job as a clinician to support them in that decision.

Dr. Heinemann: I’m impressed by the quality of the results, so I’m in favor of DIY, but if something goes wrong – we have liability – what happens? You as a physician might be in a difficult position – we should discuss that.

Dr. Lee: Good point – there are a lot of good examples of this. Patients may not want to talk to their pediatricians about DIY because they’re worried they’ll disapprove. But we were trained with systems that are dinosaurs – I was trained on paper logbooks and NPH. Patients are pushing innovation. We need to listen. There’s risk in everything we do with insulin. I love this quote: “Who do I trust more a teen or a computer?” I trust the computer. I’m not sure someone could sue you for having a conversation about a choice a patient made about a device.

Dr. Lorenzo Sandini (Central Hospital of South Karelia, Lappeenranta, Finland): Devices are just between the sensor and pump – not one of those makes decisions without the will of a patient. It’s just mathematics, not black magic. It’s automation at its best, but it’s not replacing what a patient would do for themselves. We can’t try to pull them away from their decision, because the data is there to show it works better even if the person works manually, staying awake 24 hours a day to adjust basal insulin every five minutes.

Dr. Jason Wittmer (Des Moines, IA): FDA has taken a permissive approach on this. I met with all the reps yesterday. They are very interested in what the DIY community is doing. I prescribe things off-label every day. I do that for things with which I have no randomized controlled trial data. I do it for no other reason other than just because we do it that way and have done so for a long time, even though we have no trial data to stand on. Our data may be anecdotal, but it’s true. So, you didn’t prescribe it, but have the conversation, make an objective and genuine assessment of risks and benefits you see, and let the patient make their own decision.

Comment: Is there no practical way to get around it? If you’re worried about legality, set up a longitudinal, open research platform, and “enroll” people on DIY. There’s a bit more scope to do this without feeling the wrath of legality…it’s probably a way around this if you guys want to figure it out.

Dr. Kittie Wyne (Ohio State University, Columbus, OH): We spend all of our time trying to teach patients that they’re in charge and I’m here to help. I say, if the physician says, ‘I’m scared don’t do it,’ then find a new physician who will support you. The problem is that there are not many doctors who understand the words you’re saying. This symposium is a great place to teach people, and we need to get it further out in the open. Someone needs to put together a consensus, a series of papers to get the legality aspects out in open. When I put a prescription out for insulin, I write “as directed, up to 50 units.” And you go home and give yourself what’s best – I’m not making those decisions, you do! We need to push and get it further out in the open and help to make physicians understand they’re not at risk but we need to learn and help people with diabetes. Anything I can do to help you guys, I’m willing.

Special Event: JDRF/NIH/HCT Closed-Loop Research Meeting

Year in Review

John Lum, MS (Jaeb Center for Health Research, Tampa, FL)

In his opening remarks, Jaeb’s Mr. John Lum put the tremendous progress on closing the loop in perspective: at the first JDRF Closed-Loop meeting at ADA 2007 – 11 years prior to the day! – there were 12 invitees on the agenda, no products on the horizon, and a focus on control algorithms. Wow has the field come a long way! Mr. Lum summarized the tremendous year in closed loop with three excellent themes: (i) acceleration of US device approvals; (ii) progress with much larger, longer-duration closed-loop studies; and (iii) an increased emphasis on interoperability and openness. Specifically:

• Mr. Lum praised the amazing work of FDA’s Drs. Courtney Lias, Stayce Beck, and team, noting that the Agency is now a “facilitator” of the field – not a barrier. He highlighted FDA approval of FreeStyle Libre last fall (not part of closed loop devices yet, but will be with Bigfoot), Dexcom’s G6 in March, and three approvals the day before ADA 2018: Senseonics’ 90-day Eversense CGM, Tandem’s Basal-IQ/Dexcom G6 (PLGS), and Medtronic’s MiniMed 670G in 7-13 year olds.

• On the large study front, the six-month International Diabetes Closed Loop Study (iDCL protocol 3), which will serve as the pivotal trial for Tandem/TypeZero’s Control-IQ with Dexcom G6, opened enrollment as of June 21. The plan is to enroll 168 participants, comparing six months of closed-loop (t:slim X2/Control-IQ/G6) to six months of sensor-augmented pump therapy (same devices, no automation). See the ClinicalTrials.gov post here, which indicates the study is already recruiting at four centers. The primary outcome is time-in-range (70-180 mg/dl) – nice!

  • Mr. Lum also noted five other large closed-loop studies getting underway now/soon: (i) iDCL protocol 1 has completed enrollment and results with the smartphone-based TypeZero inControl algorithm will be shared at DTM in November; (ii) iDCL protocol 2, testing Senseonics Eversense with a Roche pump and TypeZero algorithm, will begin enrollment this fall; (iii) the Dan 05 study, testing Cambridge’s algorithm in a Medtronic pump/CGM is now underway; (iv) the FLAIR study, comparing the 670G to Medtronic’s Advanced hybrid closed loop (with DreaMed auto-bolusing algorithm) has finalized the study design and is set to begin in Q4 of this year (ClinicalTrials.gov page); and (v) the Bionic Pancreas team has large pivotal trials slated to begin in 2019 (insulin-only first; IDE approved in May for first iLet bridging study).

• Before handing it off to Dr. Aaron Kowalski, Mr. Lum noted the growing focus on device interoperability: JDRF’s Open Protocol AID Initiative (October 2017); Dexcom’s launch of a public API for third-party retrospective CGM data apps (September 2017); the clearance of Dexcom’s G6 as an iCGM with interoperability in mind (March 2018); and FDA approval of Tandem’s Basal-IQ with Dexcom’s G6 (June 2018). He also noted the April 23 JDRF/HCT workshop on interoperability – see our coverage here. Noting the interoperability progress with iCGM, Mr. Lum added, “We’d like to see the same thing happen with pumps.” We agree!

Update on JDRF Open Protocol Initiative

Aaron Kowalski, PhD (JDRF, New York, NY)

JDRF Chief Mission Officer gave a highly enthusiastic update on JDRF’s Open Protocol Initiative, which aims for seamless, secure, interoperable connectivity between diabetes devices and smartphone apps (e.g., Bluetooth) – both commercial devices and DIY efforts. In a room filled with academic investigators and industry, Dr. Kowalski was honest in sharing that he uses a DIY system (Loop) and absolutely loves it. He added that the DIY community is on the “bleeding edge” and is “doing amazing things,” and given the momentum, the DIY movement “is not stoppable…nor should it be stopped…” However, he noted, “it needs to be safer,” particularly given the secondary market for used Medtronic old pumps compatible with Loop/OpenAPS: “It just doesn’t feel right to me that we are operating under the table, with pumps that are being bought on Craigslist, eBay, and Facebook that are 7-8 years old.” To this end, JDRF has “invested significant effort in this area,” including working with industry, FDA, and understanding the legal/liability implications of diabetes devices approved with open communication protocols. He characterized April’s JDRF/HCT Interoperability workshop as very productive (“unbelievable in terms of dialogue…everyone spoke up”) and highlighted Dexcom’s G6 iCGM pathway as an “intriguing and exciting pathway forward…You could envision an interoperable ecosystem with iAlgorithms and iPumps in the future. That to me is very, very exciting,” he said, “and JDRF is going to be pushing very hard to see that become real. I think every company in the room, sensor, pump, and algorithm, has a whole new set of opportunities to build amazing solutions.” He concluded that “Interoperable systems are going to play a very important role in diabetes solutions in the future.” We agree and love this visionary, groundbreaking work to make devices safe, to harness the DIY community’s brilliance, and to provide a new innovation and regulatory paradigm for companies to compete in.

• Dr. Kowalski highlighted SOOIL on the pump side, who shared bold plans at this ADA to submit an open protocol Dana pump to the FDA and launch by ADA 2019 in the US. Its goal is to submit with an iPump de novo designation, following the path paved by Dexcom’s G6 iCGM. Can it execute on this vision? Will it be the first to get smartphone control of a pump approved?

• Dr. Kowalski was honest that liability is the biggest issue here for device makers, and JDRF has hired a law firm (Bowman & Brooke) to better understand the unique product liability challenges associated with interoperable automated insulin delivery – see our coverage of that from April. We think this is a potentially solvable problem with disclaimers and waivers – e.g., “I accept all risk by enabling DIY mode” – though the details are always the hard part with these sorts of things.

Mini Symposium: Understanding Hybrid Closed Loop Pump

What Patients Need to Know

Laurel Messer (Barbara Davis Center, Aurora, CO)

BDC’s Ms. Laurel Messer gave a tour de force overview of Medtronic’s MiniMed 670G from a patient perspective, concluding that “this is still diabetes, and behavior is still paramount…but this may be a quality of life game changer for many patients.” She drove home five main points:

(i) Insulin delivery fundamentally changes on hybrid closed loop. “It’s important patients understand the ‘hybrid’ part.” She suggested patients should know that basal rate is automated, that boluses are manual, that the concept of a basal rate goes away (“every five minutes there’s a new basal”), and how to use a pump in manual mode (since they will still be out of auto mode ~20% of the time in a good case).

• Ms. Messer wanted to dispel the myth that 670G “learns my diabetes”: “It doesn’t know when you’re going to eat, how much you’re going to eat, when you’ll be more insulin resistant, when you’ll have hormone secretion in the early morning, diurnal patterns, when you’ll exercise, if you did exercise, if you’re awake or sleeping. It is a blind system in many ways. It knows three things: total daily dose for the past two to six days, current glucose level, and insulin on board. That’s all – I get very nervous when patients say it needs to learn their body because it ends up being very different from their expectations.”

(ii) How/what insulin dose settings can be modified. She drove home that in auto mode, only active insulin time, insulin:carb ratio (ICR), and temp target (150 mg/dl) can be altered. She suggested that ICR should be more aggressive at each meal because the system will work on the back end to reduce basal rate if needed. When educating patients, Ms. Messer proposes teaching the temp target like a lower temp basal, to be used when the patient is more insulin sensitive – brilliant! She’s noticed that some users are frustrated that they can’t customize or override boluses or that boluses are “almost always less than a patient would expect,” but generally discouraged the practice of “phantom boluses” or “fake carbs,” where the user inputs carbs that they never took so the system will allow them to bolus or be more aggressive. “I think at times it’s appropriate – for example, with coffee in the morning, that’s one way patients can mitigate the problem of blood glucose rises without carbs. However, I’ve also seen times where it overloads the system and you can’t avoid hypoglycemia after fake carb boluses. It’s patient education. Glucose may not come down as fast as you’re used to, but as the system should overall increase time-in-range, you can build tolerance for the slower fall of hyperglycemia. Ask patients to see the forest through the trees, and to look at overall glycemic profiles – what ends up happening with fake carbs is their glycemic variability increases, and we’re not helping them very much. They can get used to coming down slower and see that they’re spending more time-in-range.”

(iii) Expectations on work load. Nothing surprising here from what we’ve previously heard: 670G requires an ongoing interaction, as patients need to check blood glucose and give meal boluses, calibrate their CGMs, and respond to alerts. There are also new system notifications like “BG required” and safe basal alerts (to notify the patient she is exiting auto mode). “In the pivotal study, people exited 5.6 times per week from auto mode (almost once a day). 95% of those exits were due to system requirements, and just 5% were due to the user. You need to tend to the system, follow alerts so it can keep automating doses.” (In our view, this is clearly a big area where Medtronic needs to improve the 670G. It’s also an area we expect more user-friendly systems to push Medtronic to improve on.) She also encouraged proper fingerstick technique – since she works with pediatrics and adolescents, she implores them to at least use the second drop of blood since she knows so few of them will ever wash their hands before testing.

• During Q&A, Ms. Messer called the just-FDA-approved t:slim X2 with Dexcom G6 PLGS system a “game changer” since it is the first AID system approved for use with a factory-calibrated sensor. “A system that doesn’t require fingersticks might make this technology work for people not able to have success on 670G. I’m very excited about it, and it’s great to see this competition.”

(iv) Learning curve. Ms. Messer has noticed that it can take “one to four weeks” to get used to the 670G – due to tweaks in ICRs and insulin action time and figuring out how to manage meals – so she asks patients to give it more than a month before making a judgement on how they like it. “After four to six weeks, they often say it’s changed their lives for the better.” Pointing out that most of the advice that patients seek and receive about 670G is how to trick it, Ms. Messer recommends using as directed to maximize results.

• We’d add, unfortunately, that Medtronic has a 30-day return policy on devices, meaning by the time someone decides if they like it after 4-6 weeks, they probably cannot return it. We wish Medtronic would extend this to at least 60-90 days; getting on a system like the 670G is a huge decision, and unfortunately one patients cannot test drive before getting it. Given the new 670G Outcomes Guarantee for payers (lasting four years – the length of a pump warranty), we’d love to see a Patient Satisfaction Guarantee extended to at least 60 days. Progress on this front could really reduce the fear factor of getting a new pump and perhaps even increase adoption on MDIs – “If you don’t like it, return it after 90 days – no questions asked!” Insulet is the most ahead on this with its free demo pods. Over time, Insulet could even expand to a free pod trial with smartphone control from a free app – meaning no hardware startup cost to try before buying.

• Though the 670G technically requires only two days of manual mode before it allows users to enter auto mode, patients at BDC are typically in manual mode for one to two weeks before transitioning to auto mode. That way the user can learn the CGM/calibration routine and optimize insulin dosing so the system will have total daily doses better optimized for Auto Mode start . Ms. Messer recommends telling patients to give themselves as much insulin as (safely) possible in the week before hybrid closed loop begins – “the more insulin the patient gets without hypoglycemia, the better Auto Mode will do for hyperglycemia.”

(v) Behavioral tips for patients:

• Stay in auto mode as often as you can. Regarding boluses, “Give ALL, give BEFORE.”

• Give correction doses with any blood glucose >150 mg/dl.

• Hyperglycemia often leads to hybrid closed loop exits, so avoid as much as possible.

• You still have to calibrate the sensor and respond to system alerts. “This is especially pertinent now that there are factory-calibrated sensors out there. It’s s a tradeoff that needs to be pointed out.”

What Clinicians Need to Know

Elizabeth Doyle, DNP (Yale University, New Haven, CT)

Referring to the MiniMed 670G, Yale’s Dr. Elizabeth Doyle stated: “The success of carb counting is going to equal the success on this system.” We’ve heard similar sentiments in the past – the basal:bolus ratio on 670G often shifts in favor of bolus to better cover meals, while the basal rate can decrease on the back end to compensate. Of course, if this is to happen safely and effectively, good carb counting practices are necessary. Dr. Doyle suggested referring new 670G starts to a nutritionist. (We’d point out that in someone eating low-carb, the opposite is often true – far more insulin delivery shifts to basal on hybrid closed loop. On Loop, for instance, Adam typically sees 85% as basal delivery and only 15% as bolus.) Dr. Doyle also noted that Yale has used 670G successfully in some type 2 patients who are “totally insulin-dependent” (just on insulin, after not seeing success with orals). We’d be very interested to see 670G outcomes data from this population of type 2s. Dr. Doyle also emphasized the goal of 80% time in auto mode, noted very positive experiences with the system in her clinic, and excitedly announced to the room that FDA had approved 670G for use in kids ages 7-13 years old (the session came the day after the announcement).

• Dr. Doyle shared an impressive case study of a 57-year-old type 1 man who was treated at Yale, noting that following a reluctant start, she saw “such a change in this man. He was far happier, and less obsessed. He and his wife would agree that the changes in his quality of life were unexpected. He gave over a lot of his anxiety to the pump – he realized he just needed to let it do its job.” At baseline, this patient had a history of hypoglycemia and some hypoglycemia unawareness, checked his blood glucose 10.4 times per day, and had an A1c of 6.2%. He was a part of the 670G Priority Access Program and found the pump and CGM “absolutely fantastic” – still, while in manual mode, he was taking 13.4 boluses per day, checking blood glucose 11.8 times per day (despite the CGM), and had many lows. After starting hybrid closed loop, he was averaging fewer than eight boluses per day, spent 86% of time in-range, and only 2% below 70 mg/dl. His A1c has leveled at 6.3%, and he’s now only checking his blood sugar 3.6 times per day (basically what’s needed for calibration). While not everyone gets up to 86% in-range, this remarkable anecdote shows how hybrid closed loop can drastically improve the outcomes and quality of life, especially for someone working hard and already getting an A1c <7%.

Symposium: Artificial Pancreas and Decision Support Approaches

Continuous Glucose Monitoring-Based Decision Support for Type 1 Diabetes

Marc Breton, PhD (University of Virginia, Charlottesville, VA)

UVA’s Dr. Marc Breton fleshed out promising pilot data from a study of a CGM-based decision support system based on the UVA DiAs platform in 24 type 1 adults (16 pumpers, 8 on MDI). [Dr. Bruce Buckingham flashed the summary figure from this pilot – shown below – at DTM]. In its current form, the system is composed of three main modules: automated treatment parameters adaptation (recommends changes in therapy based on retrospective risk zones), insulin sensitivity-informed bolus calculator (adjusts boluses based on ratio of real-time insulin sensitivity to historical insulin sensitivity), and exercise advice (determines whether patient should snack/wait or proceed with exercise depending on age, insulin on board and initial glucose). Through the ~four-week study, which included clinic-based meals and exercise, the decision-support system significantly improved a number of glucose parameters relative to standard of care: Primarily, coefficient of variation fell from 36% at baseline to 33% (34% -> 30% at mealtime), largely from less hypoglycemia, as time ≤70 mg/dl decreased by ~33 minutes per day (from 3.2% to 0.88%). Time in range (70-180 mg/dl) and time above 250 mg/dl did not significantly change from standard care, a surprising finding (potentially because of the small study size). The plot below of “quality” glucose – average vs. time below 70 mg/dl – does show a much tighter band in the treatment group towards less hypoglycemia and a lower average. Notably, this decision support system is in the midst of a three-month, n=120 study at Stanford, UVA, and Mt. Sinai to explore applications in MDI – we first heard about this study, which is using Novo Nordisk-supplied smart insulin pens, at DTM 2017. Two posters from this study group related to smart pens are described above. The team will also conduct studies to parse out the impact of each decision support module during 2018, attempt to untangle the impact of information without support vs. prescription (“take X units of insulin”) and their acceptability to patients, and extrapolate the system to type 2s. 

• Dr. Breton provided granular data from the in-clinic days assessing the effectiveness of the decision support in helping participants handle exercise and meals. When individuals exercised two hours post-meal, the system (red) appears to greatly reduce hypoglycemia; three hours post exercise, it appears to prevent early hypoglycemia, though may put the user at slightly higher risk of hypoglycemia starting ~40 minutes after beginning activity. For sessions when individuals began exercise with blood glucose <180 mg/dl, Dr. Breton remarked that the glucose traces resemble those from hybrid closed loop – significantly less hypoglycemia, and significantly less variability. Regarding postprandial excursions, the decision support system improved patient handling of high fat/protein meals, without significantly impacting handling of meals overall or large meals. In the postprandial excursions slide below, a distribution to the left (lower area under the curve) is favorable – higher red bars on the left indicate lower and/or shorter duration hyperglycemic excursions after a meal. We’ve never seen meal data plotted as histogram of AUC; a simple time-in-range metric might have summarized it better?

• The other modules to be included in the decision support system are: Bedtime advice, average glycemia indicator, and hypoglycemia prediction (up to three hours in advance). We’re not sure if these are incorporated into the system currently in evaluation at UVA, Mount Sinai, and Stanford.

  • Dr. Boris Kovatchev, Dr. Breton’s mentor, showed impressive data from the hypoglycemia prevention module at DTM.

Advanced Treatments for Type 1 Diabetes—Adapting to Meals and Exercise

Jessica Castle, MD (Oregon Health & Science University, Portland, OR)

OHSU’s Dr. Jessica Castle gave an update on her team’s work with closed loop and automated decision support to improve adaptation, including the Helmsley Charitable Trust-funded, Jaeb-coordinated T1-DEXI pilot study. The pilot study will enroll 60 individuals between ages 15-70 with type 1 diabetes, and collect one month of insulin, CGM, food, and physical activity data. Data will be collected with Dexcom G5, DiabNext’s Clipsulin dose capture device (an uncommon choice), a Garmin activity tracker, and a custom app developed at OHSU for food photos and exercise logging. The app asks patients to take photos before and after eating, estimate carbohydrate content, and provide a qualitative estimate of how much protein and fat is consumed. Participants will be randomized to complete two in-clinic and four home sessions of either aerobic, anaerobic, or high intensity interval exercise. Every seven days, providers will review CGM and insulin data, and make insulin dose recommendations. This study will eventually inform a larger study of 300-500 participants (!) with a goal of building better exercise and food models for automated insulin delivery and decision support.

• Dr. Castle detailed a couple of other current and future OHSU studies in the domains of dual-hormone closed loop and decision support:

  • The team is in the midst of a three-day outpatient study with in-clinic aerobic exercise (n=20) assessing a dual-hormone closed loop system consisting of Insulet Omnipod and Dexcom G6. Notably, this study is using Xeris glucagon, which Dr. Peter Jacobs previously told us is stable for the duration of the study and does not need to be changed. The system will also be informed by OHSU-developed hypoglycemia prediction and postprandial hyperglycemia prevention algorithms.

  • “DailyDose” is a smartphone-based decision support system for type 1s on MDI therapy that provides bolus, basal, and behavioral recommendations. The application was co-developed with University of Toronto’s Dr. Joe Cafazzo and funded by the Helmsley Charitable Trust. No further details were shared on this platform, including study timing, blood glucose input (CGM, SMBG, both?), other data inputs, etc.

• Dr. Castle quickly reviewed three OHSU-developed algorithms, two for hypoglycemia prevention, and one for minimizing postprandial hyperglycemia. The first hypoglycemia prediction algorithm is a heuristic algorithm that uses heart rate and glucose at the start of exercise to help patients predict hypoglycemia risk in real time. The second hypoglycemia prediction algorithm is a “random forest algorithm,” meaning that it runs through a group of decision trees and the outcome of each tree is averaged to get a higher predictive accuracy. Inputs for this second, more complex algorithm, include glucose and heart rate at the start of exercise, insulin on board, average TDD, sex, weight, height, and BMI. The algorithms were trained on 154 exercise observations (some closed loop, some open loop), and were validated in 90 individuals: The first heuristic algorithm more accurately predicted hypoglycemia than the decision tree algorithm (87% vs. 80% accuracy), driven by greater specificity. Regarding meals, the OHSU team is developing ALPHA – adaptive learning postprandial hypo-prevention algorithm.

Automated Adjustment of Basal-Bolus Insulin Therapy Using Run-to-Run Control and Case-Based Reasoning

Pau Herrero-Viñas, PhD (Imperial College London, London, United Kingdom)

Imperial College London’s Dr. Pau Herrero-Viñas detailed the Advanced Bolus Calculator for Diabetes Management (ABC4D), an insulin bolus calculator that uses Run-to-Run (R2R) control and Case-Based Reasoning (CBR) to automatically adjust patients’ insulin:carb ratios depending on the post-prandial outcomes of the previous week. The ABC4D patient-facing app and clinician platform was tested in a small pilot study (n=10) of adults with type 1 diabetes on MDI. Participants used the ABC4D smartphone app for six weeks in their home environment, returning to the clinical research facility weekly for data upload, revision, and adaptation of the CBR case base. Although Dr. Herrero-Viñas alluded to a trended reduction in postprandial hypoglycemia and improvements in other glycemic outcomes, no changes were found to be significant. Still, he was encouraged by the results, believing the non-significant outcomes to likely be the result of the small sample size. Since the study’s publication in 2016, Dr. Herrero-Viñas and his team have designed an updated ABC4D app intended to be more user-friendly: all features are accessible on the patient’s smartphone (i.e. no need to go to the clinic for data download or algorithm revision), and CGM integrates directly with the bolus calculator to account for metrics like rate of glucose change. A non-inferiority, crossover RCT investigating the new version is currently being planned in a larger study population – Dr. Herrero-Viñas displayed a picture of the user app with a Dexcom G6, but we’re not sure if that is the only CGM that can be integrated. Dr. Herrero-Viñas’s team is also planning to incorporate the bolus calculator into a hybrid closed loop system called “Biap.” The slide introducing Biap depicted Dexcom’s G6 and Tandem’s t:slim X2 pump, although these devices were not directly named during the presentation – this name is new to us! An open-label, randomized study (n=20) with three arms (sensor-augmented pump vs. Biap with standard bolus vs. Biap with adaptive bolus) is in the works.

Symposium: Joint ADA/JDRF Symposium – Current Management of Type 1 Diabetes in Youth – What Are the Options?

Insulin Therapy in Youth with Type 1 Diabetes Mellitus – Multiple Daily Injections or Continuous Subcutaneous Insulin Infusion?

Hood Thabit, MD, PhD (Manchester University Hospitals NHS Foundation Trust, Manchester, England)

Drawing on data from an amalgam of sources including the T1D Exchange (T1DX), German/Austrian DPV registry, Western Australia Children Diabetes Database (NDPA), Manchester’s Dr. Hood Thabit concluded that pumps are generally advantageous in comparison to MDI for youth with type 1 diabetes, though choice of insulin delivery must be individualized to maximize treatment satisfaction and quality of life. Further, he called for greater research into the psychological aspects of insulin therapy in youth to increase adherence with both pumps and pens. When data from the three registries was compiled, pumps only seemed to confer a modest improvement in A1c levels. However, pump users have significantly lower frequency of DKA and severe hypoglycemia events. Pumpers in their registry experienced 16% fewer DKA events than their injecting counterparts, and, according to a cited study, pumpers were 30% less likely to experience a severe hypo. Based on data from the Swedish National Diabetes Register, all-cause mortality is also 27% lower for pump users, and, with respect to psychosocial outcomes, Dr. Thabit acknowledged that results have been mixed, but that treatment satisfaction and diabetes self-efficacy have both shown to be higher in those using pumps compared to injections. Dr. Thabit noted that these data all point to pumps as superior, but qualified this data-driven conclusion: At the end of the day, choosing an insulin therapy for youth must align with individual and parent preferences, and it is the doctor’s responsibility to educate them on the benefits and drawbacks of each while keeping the long-term goals of the patient in mind. (To this we would add cost, which can be make or break for many families.) Particularly at such a young age – mean A1c is highest among 13-25 year-olds with type 1 diabetes according to the T1D Exchange and NDPA – Dr. Thabit underscored that establishing a solid therapeutic regimen is important, as early complications can be significant in terms of metabolic imprint and future CV risk. Therefore, whichever therapy gives the greatest likelihood of achieving positive outcomes, based on the patient’s lifestyle and preference, should be used. Hear, hear! We add that the data highlighted by Dr. Thabit was largely correlational, and the preferable outcomes in pump users could have more to do with the type of person likely to start pump therapy (selection bias), and less to do with the mode of insulin delivery itself. This all the more underscores Dr. Thabit’s conclusion that a patient should deliver insulin whichever way he/she is most likely to use and succeed with – a teen on the swim team may well prefer to use injections so as not to be singled out as “different” for being tethered to a pump!

• Dr. Thabit briefly touched on novel automated insulin delivery systems and smart pens as bright spots in the future of diabetes technology for youths with type 1 diabetes. Systems that make insulin delivery easier and safer will certainly help youths feel more confident managing their diabetes – and more often doing so without much conscious thought – but barriers like on-body burden, aesthetics, and stigma will still be at play! A increasingly concerning issue could be infusion sets and subcutaneous delivery over years – as younger and younger patients pump insulin for decades, the field will need to move towards better infusion that drives less scarring and lower risk of lipohypertrophy.

Symposium: Innovative Strategies to Improve the Inpatient Experience

Patients’ Own Diabetes Devices―To Wear or Not to Wear?

Donna Jornsay (Mills-Peninsula Medical Center, Burlingame, CA)

Mills-Peninsula Medical Center’s Ms. Donna Jornsay spoke broadly about the use of insulin pumps in the inpatient setting. In order to remain on an insulin pump, patients often must sign written agreements provided by institutions with the following requirements: patients must be alert and oriented, not be in critical condition or suicidal, be able and willing to self-manage the pump, be willing to have BG values tested with the hospital meter, have personal pump supplies with them (which the hospital may be lacking), be willing to have BG values tested with the hospital meter, be willing to sign appropriate paperwork/forms, and be willing to have an endocrinology/diabetes educator consult. In addition to providing the insulin for pumps and the blood glucose meter in inpatient settings, the hospital must have the model of the pump and relevant contact information readily available. Such cautionary measures are especially important to consider due to potential issues in the operating room arising from infusion site placement, X-rays which can interfere with proper functioning, and staff with limited pump knowledge/experience. In clinical settings, pumps may also be problematic for those with poorer health and who depend on feeding tubes or other apparatuses. Because patients with diabetes are more likely to be admitted for inpatient care, Ms. Jornsay pointed out, it’s important that the knowledge gap surrounding insulin pumps is bridged between care providers and patients, especially for pediatric patients, many of whom use 10ths or 100ths of an insulin unit as given by pumps. The policies outlined by Ms. Jornsay seem doable, which is a good thing in our view, as a capable/aware person self-managing his/her diabetes in the hospital seems more likely to lead to optimal glycemic outcomes. Every time the Cambridge group presents new inpatient closed loop data (as they did at this ADA, in type 2s), we are amazed at the poor quality of glycemia in the control group – 412% time-in-range, with nearly 50% spent >180 mg/dl in that study! Hospital diabetes management and the degree to which patients should be in charge of regulating their glucose in that setting are areas requiring more research, particularly in light of signals that mean glucose correlates with lower mortality in the cardiac perioperative setting (assuming this finding is somewhat generalizable to the other inpatient scenarios).

• Ms. Jornsay noted that few institutions have policies on patients’ own CGM devices, there have been no FDA approvals for CGM use on inpatients, and patients wearing CGMs may be unwilling to have blood sugar checked in clinics. There is an n=244 VA hospital CGM study underway, and Dexcom has expressed interest in exploring its no-calibration G6 in the inpatient setting, but there have been no commercial gains on this front to date – this is a tough and expensive and slow-moving market, and there is plenty of runway in the personal CGM market to tackle already. Nova Biomedical’s StatStrip hospital BGM just became the first fingerstick monitor to be cleared for use in critically ill patients – will this clearance kick off more glucose monitoring innovation in the hospital setting?

Product Theaters

The Omnipod Dash System—Simplifying Insulin Delivery for Your Patients and Practice (Presented by Insulet)

Trang Ly, MBBS, PhD (Insulet Corporation, Boston, MA)

Insulet’s Omnipod Dash product theater offered a full, detailed demo at the new touchscreen PDM and paired mobile secondary display apps. Based on what we saw, Insulet’s Dash has set a new industry standard in the user experience of insulin pumps, even relative to Tandem’s t:slim X2: from the big things (fully integrated food library, simple home screen centered on IOB, very large color interface) to the small things (graphical display of basal insulin, excellent scroll wheels), Insulet has done an awesome job on this product and the attention to detail really shows. We especially loved the startup screens that guide the user through setting up the PDM – Dash has a warm feel, and Dr. Trang Ly’s demo had the entire pump parameters set up from scratch in just a few minutes. We imagine healthcare providers will love how much easier the PDM is to train, and the advantage of automatic upload to Insulet-provided Glooko (when connected to WiFi) will be big. Insulet has laid a great foundation for the future goal to dose directly from the phone. The launch timeline is the same in the US (six-month limited market release in 2H18, full launch in early 2019); President Shacey Petrovic confirmed the plan for a 2H19 Dash launch outside the US.

MiniMed 670G System—Increasing Time In Range with Automated Insulin Delivery (Presented by Medtronic)

Jennifer Sherr, MD, PhD (Yale University School of Medicine, New Haven, CT), Anders Carlson, MD (International Diabetes Center, Minneapolis, MN)

During a Medtronic MiniMed 670G product theater, we were delighted to hear Yale’s Dr. Jennifer Sherr and IDC’s Dr. Anders Carlson emphasize the importance of time-in-range and the limitations of A1c. As Dr. Sherr put it, “time-in-range really provides clinicians with actionable insights, whereas A1c tells me where you are vs. a target.” Dr. Carlson cited the recently published (Runge et al., Clinical Diabetes) paper to highlight that patients care about time-in-range – surveyed participants with both type 1 and type 2 diabetes ranked time-in-range very highly in a list of factors they care about in their diabetes management. Dr. Carlson pointed to the real-world 670G data from nearly 33,000 patients shared at ADA (see above) to assert that a time-in-range of at least 70% can be “reasonably expected” with appropriate use of the 670G. Several thought leaders are interested in defining goals for glycemic outcomes like time-in-range, and many have advocated for using the 670G pivotal data as a place to start. Dr. Sherr wrapped up the discussion by sharing a case study of an eight-year-old patient on the 670G for the very first time. After she had spent years struggling to manage glucose overnight and facing extreme glucose variability during the day, Dr. Sherr finally decided to put her on the 670G (off-label at the time). Once the patient transitioned to auto-mode, Dr. Sherr recalled the tearful mother simply saying: “we just sleep.” We’re delighted that the 670G is finally approved in those as young as seven, and that Tandem also just received approval for its G6-integrated t:slim X2 Basal.IQ (PLGS) system, which will also give parents more sleep and peace of mind.

Diabetes Mine D-Data Exchange

SOOIL to Submit to FDA with Open Protocol Dana Pump Designed Around Needs of DIY Community; Potential OpenAPS Inclusion?

Justin Walker (SOOIL, New Zealand)

SOOIL Korea’s Justin Walker shared bold plans to submit the company’s smartphone-controlled Dana RS insulin pump to the FDA/CE Mark with an open communication protocol – an interoperable pump to meet the wishes of DIY users. He also mentioned plans to register a version of the open source OpenAPS algorithm (of Dana Lewis/Scott Leibrand fame), implying in Q&A it would sit in a phone app and communicate with the Dana pump and an iCGM. SOOIL actually submitted interest to JDRF’s Open Protocol Initiative (which Roche is also part of), and this bold plan could be a big asset in the US market if the company can figure out the regulatory path – an open “iPump” that patients can buy brand-new for use in systems like Loop and OpenAPS could be compelling! The timing on FDA submission was a bit unclear, though SOOIL hopes to release the open protocol pump “shortly” – it had a few minor design updates, but looked mostly similar to the already-available Dana RS. As we’ve previously noted, Dana RS is the only pump to our knowledge with approved smartphone control apps for iOS and Android. The pump is no-frills, but has become a popular choice in the DIY community outside the US, as the direct smartphone control means no communication relay device or “rig” is needed (e.g., AndroidAPS users simply run the closed loop app on the phone and wear the Dexcom and Dana RS pump). Dana’s RS pump is CE Marked and Korean FDA approved and available in both Europe and Asia. Interestingly, SOOIL has been building pumps since 1979 and its move to smartphone control was driven by the DIY community, including Mr. Walker’s own experience.

Medtronic Diabetes ADA Analyst Briefing

MiniMed 670G Performance Guarantee

In a huge update at Medtronic’s one-hour ADA Analyst Briefing (slides, webcast), it is launching a bold 670G outcomes-based guarantee for payers: “If there is a diabetes-related hospitalization or ER visit for patients on the 670G, we’ll reimburse it – up to a cap of $25,000 over a four-year period.” President Hooman Hakami was very impassioned in presenting the slide below: “Based on the outcomes with the  670G … and the compelling data we see from the millions of patient-days, we’re ready to stand behind the system through a business model…There is no other diabetes company that can say this or do this. No other company has the benefit of a system that actually keeps a patient in the right glycemic control. There is no other company that has the data we do that demonstrates the value of that control. This is something we think the community is going to embrace. It’s a further indication of our commitment to value-based-healthcare, and not just driving fee-for-service – but actually selling outcomes.” He later added that “Medtronic is trying to change the conversation. Today’s paradigm, where companies offer a product, they are compensated for the product immediately, but the healthcare system lives with a promise that things will get better because of utilization. ‘Pay me for this, trust me, it’s going to drive better outcomes and lower costs.’ That’s a broken dynamic. We want to change that. Instead, we say, ‘I’m going to prove it to you; if you buy this, we feel so strongly that this was designed for outcomes, we’re going to stand by it.” Wow – this is a very bold and important move for the field, and Susanne Winters (a VP at Medtronic) said in Q&A that payers are “very interested in this,” especially to have predictability around costs. Will we see more preferred arrangements like the UHC/Medtronic deal?

MiniMed 670G CE Mark

Medtronic also announced CE Mark for the MiniMed 670G hybrid closed loop, with a launch to commence this fall in 10 European countries: Belgium, Denmark, Finland, Ireland, Italy, Netherlands, Spain, Sweden, Switzerland, and the UK. Notably, Medtronic obtained approval for 7+ years, meaning it will have pediatric approval from the start in Europe. (In the US, 670G approval for 7-13 years just came on Thursday.) The EU approval comes ~21 months following the September 2016 US approval for adults – a testament to FDA’s remarkable leadership in automated insulin delivery. (It’s also possible Medtronic waited to secure the pediatric label and to get manufacturing ready before submitting in Europe.) The 670G EU launch will also bring the improved Guardian Sensor 3 outside the US for the first time; currently, Medtronic uses Enlite Enhanced in the 640G and standalone Guardian Connect mobile CGM. Following 670G’s EU rollout, Medtronic will add Guardian Sensor 3 to Guardian Connect outside the US. This 670G EU launch puts Medtronic ahead of competitors on AID: Diabeloop is planning for a late 2018 EU launch of its hybrid closed loop in only three countries (see Day #3); Insulet is taking over direct Omnipod distribution on July 1, though Dash is not expected in Europe until 2H19 and AID with Omnipod Horizon is further out; and Tandem will make its first EU sales in 2H18 following CE Mark of t:slim X2/G5, but there is no international timing on its AID systems, Basal-IQ or Control-IQ.

Pipeline Updates: Non-Adjunctive CGM, No-Calibration CGM

The ADA Briefing also gave some new pipeline updates not shared at the early June Analyst Meeting: a PMA for a non-adjunctive CGM claim will be filed “this year” (unclear if calendar or fiscal); a clinical trial for a Medtronic no-calibration real-time CGM will happen this year with a planned launch within two years (presumably Harmony); talks are ongoing with the FDA to run a trial for a seven-day wear infusion set, with a goal to start before the end of this fiscal year (by April 2019); Medtronic’s Envision Pro (formerly iPro3; launch by April 2020) will be blinded, fully disposable, no-calibration, and Bluetooth-enabled (single-use data transfer, like Libre Pro). Management confirmed the next iteration of 670G will add Bluetooth, with “eventually smartphone control” (the Analyst Meeting implied those could come together).

Guardian Connect Pricing and Telehealth/Remote Monitoring

We also heard that Guardian Connect will be priced somewhere between Dexcom G5/G6 and FreeStyle Libre. Management reiterated the power of Sugar.IQ and that Medtronic has “1,000” people resources in the US that it can leverage to drive Guardian Connect (we assume this is reps + clinical).

Sheri Dodd, the new VP of the Type 2 (Non-Intensive) Diabetes Business spoke for the first time in a diabetes presentation, sharing plans to combine the telehealth/remote monitoring in Medtronic Care Management Services with professional CGM. She’ll continue to lead both businesses. Her slides talked about risk stratification and figuring out the best tools to care for patients with comorbidities – definitely a stronger Medtronic move into broader type 2/prediabetes care.

CGM and BGM

Oral Presentations: Glucose Monitoring—Advances, Pitfalls, and Clinical Relevance

Efficacy of a Novel Interim Intervention Technique (IIT) with Retrospective Flash Glucose Monitoring to Improve Glycemic Control

Akshay Jain, MD (LMC Diabetes and Endocrinology, Toronto, Canada)

Dr. Akshay Jain (LMC Diabetes & Endocrinology) detailed the successful implementation of an interim intervention technique (IIT) using Abbott’s FreeStyle Libre Pro in India. Given limited resources, the investigators sought to determine if one intervention during a single 14-day blinded professional CGM session could drive patient outcomes – think of this approach as a cheaper alternative to running one professional CGM, contemplating, then running another. 105 type 2 adults with A1c >7% – on oral agents and/or insulin – and no previous history of CGM use were recruited. During the first visit, participants were evaluated by an endocrinologist and initiated with FreeStyle Libre Pro. Participants returned within one week for a data download (i.e., halfway through the 14-day sensor wear). Individualized dietary and pharmacotherapy modifications were discussed, and patients were shown how specific food choices correlated with their glycemic excursions. Less than a week later, participants returned for a final evaluation once the 14-day wear was up. Incredibly, in just 14 days, all outcomes improved significantly: Daily average glucose dropped from 191 mg/dl t0 137 mg/dl (an ~1.3% estimated A1c/GMI reduction); time-in-range increased from 42% to 80% - a whopping +9 hours/day!; time <70 mg/dl decreased from 6% to 1% (-1 hour/day); and time >180 mg/dl decreased from 52% to 18% (-8 hours). Importantly, ITT seems to be particularly effective in reducing hypoglycemia. A sub-analysis of 27 patients identified to have recurrent episodes of hypoglycemia pre-IIT revealed the intervention drove a significant increase in time-in-range (66% to 87%; +5 hours) and a dramatic decrease in time <70 mg/dl (21% to 2%; -4.5 hours), without any significant increase in hyperglycemia. These results are downright spectacular, and show the incredible potential of low-cost intermittent CGM used in clever ways to change the lives of people with diabetes in resource-poor (and rich) locales. Could this sort of intervention be deployed at the level of primary? We wonder how durable the improvements were, or if patients would require follow-up CGM sessions to regain habits learned in the initial 14 days. Dr. Jain did mention that many participants managed to maintain their dietary recommendations because of the visual impact from the retrospective CGM tracings.

• How many of the patients were on MDI? How many were just on basal insulin? This question won’t exit our minds, as we look at the incredible 80% and 87% time-in-range figures. Our instinct is to say that most of the patients were on orals, but then again, some were experiencing five hours of hypoglycemia every day (or is this Libre Pro overreporting hypoglycemia, as the FDA label indicates?). One possible explanation is that many of the individuals were over-prescribed insulin or sulfonylureas at first and were down-titrated by their physicians.

• Dr. Jain acknowledged that many patients were prevented from signing on for a repeated trial due to financial constraints – in India a doctor’s visit alone ranges from $5-$15, and the FreeStyle Libre Pro costs $40 (we believe  the patient is responsible for the entire cost). Based on the huge improvement in glycemia, we see this as an approach ripe for philanthropic investment (after confirmatory trials). What if, for example, every pregnant woman had a single Libre Pro and clinic visit early on in pregnancy to set her on a path to avoid hyperglycemia? How much good for public health would this be!

Real-World Assessment of Sugar.IQ with Watson—A Cognitive Computing-Based Diabetes Management Solution

Huzefa Neemuchwala, PhD (Medtronic Diabetes, Northridge, CA)

In a highly-anticipated oral, Medtronic Global Head of AI and Digital Health Dr. Huzefa Neemuchwala shared the latest batch of Sugar.IQ data from 256 530G/Enlite + MiniMed Connect users between April-August 2017. Relative to baseline metrics, Sugar.IQ conferred a 36-minute/day improvement in time-in-range (+2.5%-points), a 30-minute/day decrease in time >180 mg/dl, and a 6-minute/day decrease in time <70 mg/dl (all statistically significant). Though these are fairly modest by closed loop standards, the Medtronic press release notes the impact: 36 minutes per day translates to over nine additional days per year in-range! Participants in the study also experienced 1.22 fewer high episodes (>180 mg/dl for >120 minutes) per month and 0.95 fewer low episodes (<70 mg/dl for >20 minutes) per month. Dr. Neemuchwala noted that all of the users benefitted from 530G’s low glucose suspend feature through the baseline and experimental arms of the study, likely masking what could’ve been a more marked improvement at the low end of the glucose range. The implication is that MDI users who will be using Sugar.IQ in the commercial launch (with Guardian Connect) may see greater reductions in hypoglycemia; we agree. During the course of the 31+ patient-years of use, Sugar.IQ generated 655 insights related to hypoglycemia and 699 related to hyperglycemia. Notably, 231 of the 256 (~90%) users recorded at least two weeks of data, demonstrating a solid pattern of engagement, though we wonder what the engagement curve looks like after that point (at least ~30% of all engagement with the app came within those first two weeks). For context, the time-in-range data is almost identical to the 33-minute-improvement in an n=136 subset of users, presented at DTM – time in hypoglycemia and hyperglycemia have never been presented before, and episodes of hypoglycemia/hyperglycemia have previously been reported as percent reductions, so direct comparisons aren’t possible. For the limited learning launch of a novel app, we consider these data to be very encouraging, and we see big potential for Sugar.IQ to drive Medtronic’s Guardian Connect CGM marketing and outcomes in the real world. More details and screenshots from the presentation – including findings from giving some of the 256 Sugar.IQ users Fitbits, plus the glucose prediction feature in the pipeline – below! Medtronic and IBM Watson posted a press release immediately following the presentation, announcing the commercial availability of Sugar.IQ with the Bluetooth-enabled standalone Guardian Connect CGM; see our coverage of the launch from prior to ADA here.

• Medtronic randomly gave 134 Sugar.IQ users Fitbits during the course of the study, and discovered that glucose responses to meals and activity vary greatly. Each participant wore CGM for 50 days, wore a Fitbit for 47 days, and manually logged food for 12 days. Not surprisingly, investigators found that when people had a recent meal and then exercised for a median 45 minutes beginning with glucose in-range, there was an assortment of glycemic responses: (i) In 799 cases, people remained flat; (ii) In 371 cases, glucose rose slightly (mean ~140 mg/dl) toward the end of the exercise session; (iii) In 237 cases, glucose went above range (mean ~200 mg/dl) at four hours; and (iv) In 163 cases, glucose went above range (mean ~220 mg/l) at two hours. This data demonstrates the importance of personalization, as glucose responses vary greatly. We’re curious if a given individual is likely to stick to the same response to exercise or jump between the four “bins” depending on internal and external variables, and how big an impact the type of exercise has on glucose response (it could be that those who go above range are doing intense anaerobic workouts or under bolused).

• Dr. Neemuchwala showed that users “like” insights more over time, a testament to Sugar.IQ’s ability to tailor them. In the graph below, the first insights are liked by ~80% of users, while by the 18^th insight, ~100% of users like them. We found the examples of helpful vs. not-helpful tips (as “liked” or “disliked” in the app) to be particularly interesting. Users found “I see that on days when your sensor glucose is [between 80-100 mg/dl] at dawn, you tend to stay in target more all day. I like this pattern!” and “I see that you tend to spend more than 80% time in target [between 3-6 pm]” helpful. On the other hand, “I notice that [between 3PM-6PM], you tend to stay in target longer when you’ve spent [more than 75%] time in target over the previous three hours. Way to keep up a positive trend!” and “I notice that when you take [1-2] bolus(es) [between 6PM-9PM], you tend to spend more time in range. That’s what I like to see.” At first glance, there is not a drastic difference between the two sets of insights, particularly because they are all pats-on-the-back, but the differing responses could be chalked up to: (i) the most useful insights don’t require user action, while the least useful insights ask the user to keep up a good behavior or (ii) the least useful insights are more specific and less simple. It could also be that the n=256 users comprise too small a data set, and therefore the different ratings are an artifact. Regardless, it’s awesome to see this be a big focus of research and thought. Dr. Neemuchwala stated, “We don’t want to overload the individual with a lot of insights. It’s not hard to come up with them, they’re basically just correlations. The hard part is to figure out which part to deliver at what time. If I’m driving and then I go low, telling me that is more relevant than telling me what happened three days ago.” He noted that the logic behind who gets what insight and when is “fairly complex and rule-based”; it depends on glycemic acuity (how urgent is the insight?), starts simple and then works up to more complex insights, tailors to user preference and feedback, avoids repeating insights, and is internally consistent with its insights (i.e., not contradicting itself).

• Dr. Neemuchwala gave two awesome case studies of Sugar.IQ in action. In the first, a woman discovered a “problem” meal – “natural peanut butter spread low sodium” – tracked the meal using the glycemic assist feature, and learned how to manage it (image below). In the second case study, a 63-year-old male who had had type 1 for 41 years found out through Sugar.IQ that his meals were hyper-loaded with carbs on Fridays (40-55 grams per meal), which was causing him to spend 71% of the four hours following a bolus above range. After receiving the insight, the man began logging lower carb meals on Fridays (20-30 grams per meal) and reduced his time spent above range following a bolus to 39%. These cool anecdotes show the power of Sugar.IQ, but crucially depend on users manually entering food data – how will patients engage with the app in the real-world, beyond the limited learning launch? We’re excited to see – we could imagine many coming back to the app repeatedly to see what’s new!

• Dr. Neemuchwala showed new mock-up screenshots illustrating how the Hypoglycemia Prediction feature (within four hours) could look in-app. We love the sample alert, which reads: “Now 110 mg/dl. You are EXPECTED to experience a low within the next 4 hours.” As for prediction accuracy, prediction accuracy is impressive with 90% for two hours in advance, and 84% for four hours in advance, on average (based on CGM data from 10,000 type 1 users; measured as ROC AUC). These numbers seem like they would align with Medtronic’s prior claim of ~90% hypoglycemia prediction accuracy. Per the recent analyst meeting, Sugar.IQ with Hypoglycemia Prediction is set to launch within the next year (by April 2019).

• During an analyst meeting earlier this month, Medtronic announced that it would launch forward-looking glucose prediction for MDI users “beyond” April 2020 – Dr. Neemuchwala detailed “Glucose prediction research,” which we assume to be the same feature. The machine-learning model was trained on CGM data from 60 type 1s, tested on CGM data from 20 type 1s, and validated on CGM data from another 20 type 1s. Every five minutes, the glucose prediction algorithm predicts glucose four hours into the future. In the first set of data, researchers pitted the algorithm’s glucose predictions for the next four hours (from a set moment in time) against the actual readings from modern CGMs – on the Clarke Error Grid, the algorithm was able to lace 95.57% of all estimations in zones A and B, indicating good predictive power – though we wish Zone A had been broken out, since it likely wouldn’t meet where Zone A of current CGMs are at. Of course this feature won’t look too foreign to the DIY community, who already uses algorithms to prognosticate future glucose levels, but we’re excited to see Medtronic investing in bringing this game-changing technology to the masses. One Drop separately announced a 12-hour forward-looking glucose prediction for fingerstick users with type 2 not on insulin; see below for more on that.

First Assessment of the Performance of an Implantable CGM System through 180 Days in a Primarily Adolescent Population with Type 1 Diabetes

Ronnie Aronson, MD (LMC Diabetes and Endocrinology, Toronto, Canada)

Dr. Ronnie Aronson (LMC Diabetes & Endocrinology) presented full results from a study of Senseonics’ 180-day Eversense XL CGM conducted in a primarily pediatric population (n=30 adolescents, 6 adults), headlined by a very strong MARD of 9.4% vs. YSI. The mean age in the study was 17 years overall; the average adolescent age was 14 years, and the average adult age was 32. Importantly, accuracy showed a “nice consistency,” with no degradation over the very impressive six-month duration (see below). 83% of the readings were within 15 mg/dl (in hypoglycemia) or 15% (in hyperglycemia) of the reference value, and 93% of readings were within zone A of the consensus error grid. Participants demonstrated a median transmitter wear time of 23 hours/day, showing very strong adherence to the unique form factor in a young population. As Dr. Aronson pointed out, given that the transmitter is easily removed, the high wear time provides strong evidence that the device is comfortable and perceived as useful by young patients. To this end, 82% of participants “agreed” or “highly agreed” that the Eversense XL was easy to use, and 90% “agreed” or “highly agreed” that the mobile app was easy to use. The vast majority of participants (90%) indicated that they liked the ability to display glucose readings on a smartphone – as a reminder, Eversense is smartphone-only display (no receiver). Importantly, no serious adverse events or infections were reported. A few mild skin reactions occurred but were resolved within two to eight weeks of sensor removal; one patient experienced a limited skin reaction to the adhesive. Preliminary results presented at ATTD showed 78% of the sensors made it to the full six months without having to be extracted. Of the eight sensors that were removed prior to 180 days: one participant withdrew consent one day after insertion; one participant’s sensor was removed at 93 days due to connectivity issues; and there were six early sensor replacement alarms after day 130. We’re excited to see the first signs of expanding Eversense into pediatrics, as we think this population could massively benefit from the unique features and form factor of the six-month wear sensor. It’s unclear where the Health Canada regulatory submission process stands (for adult and pediatric), as well as whether Senseonics intends to pursue the 180-day indication directly in this geography. 

• As a reminder, the 90-day Eversense was just approved by the FDA on the eve of ADA for adjunctive use in adults (18+ years) with two fingerstick calibrations/day. The 180-day Eversense XL received its CE mark in September and is now available in all OUS markets except South Africa. Senseonics plans to have the entire existing OUS userbase converted to the XL by the end of the year and expects to begin recruitment for a US XL clinical trial this summer. The jump from 90-day implantation to 180-day implantation (plus potential “reduced calibration” in the US) could be very meaningful for the value-add here over traditional CGM, particularly younger users.

Symposium: Role of Continuous Glucose Monitoring in Diabetes Treatment

Continuous Glucose Monitoring—Overview, Key Learnings to Date, Appropriate Candidates, and Special Circumstances

Irl Hirsch, MD (University of Washington, Seattle, WA) and Anne Peters, MD (University of Southern California, Los Angeles, CA)

In back-to-back talks, Drs. Anne Peters and Irl Hirsch shared fascinating CGM case studies, with a greater-than-usual emphasis on the benefits of CGM in type 2 diabetes. Noted Dr. Hirsch, “CGM can be very helpful in type 2 diabetes.” Two striking cases stood out: professional CGM helping to identify an inaccurate Medicare-provided BGM (!) in one of Dr. Hirsch’s 85-year-old type 2s, and CGM-driven food choice changes resulting in multi-point A1c reductions. Many of Dr. Hirsch’s patients with type 2 diabetes on CGM “have made major changes in lifestyle,” one of the “most interesting things” he’s seen so far.

Dr. Peters added, “I put CGM on anybody I can – prediabetes, type 2 diabetes, type 1 diabetes – both professional and real-time CGM. It helps me manage my patients.” Dr. Peters noted that CGM’s benefits are “harder to show in trials,” especially in type 2s with less risk of lows. Indeed, Abbott’s REPLACE study in type 2 insulin users with a high A1c did not drive as much efficacy as many had hoped. Dr. Peters added that there is little data on use of CGM in type 2s overall – and especially in those not on insulin – which drives less robust clinical guidelines recommendations. (Of course, this will improve as the products are used and designed in smarter ways for type 2, as software helps drive lifestyle change (e.g., meal photos + CGM), and more studies happen.) On the product front, both offered quite balanced views on what’s available, including mentions of Dexcom’s G6 (“quite accurate” in Dr. Peters’ early experience; read diaTribe’s test drive), Senseonics’ Eversense (approved before ADA), MiniMed 670G in 7-13-year-olds (the pre-ADA approval was “a big deal, a real big deal,” in Dr. Hirsch’s view), and Tidepool (Dr. Peters: “I love Tidepool for looking at data, it makes my life so easy”). Abbott sponsored the session with an unrestricted educational grant, and though many of the cases focused on FreeStyle Libre, it was balanced on studies and honest product commentary. See below for highlights on some pretty unique cases – 670G in down syndrome, Medicare G5, recurrent DKA from daily marijuana use to treat neuropathy, death from severe hypoglycemia while on CGM – and more.

• Several audience members mentioned FreeStyle Libre’s accuracy in Q&A, questioning the 9.7% MARD and noting over-reading hypoglycemia and sensor errors. Most of the questioners seemed to be from outside the US, which makes us wonder about the one-hour warmup outside the US vs. 12 hours in the US – it’s likely the US 9.7% MARD with a 12-hour warmup meaningfully improves real-world accuracy. Drs. Hirsch and Peters also rightly pointed out that a meter is a not a gold standard, making “true” accuracy impossible to measure in real-world use. We completely agree!

• Professional CGM identifies an inaccurate offshore meter: An 85-year-old man, with type 2 diabetes with an average glucose of 185 mg/dl, checking with fingersticks 2.5 times/day, and a lab-measured A1c of 11.8%. When he was put on CGM, the A1c was consistent with the CGM-measured average, far higher than the meter average. Dr. Hirsch discovered the Medicare-provided meter was reading consistently low!

• Use of real-time FreeStyle Libre in type 2 diabetes. A 68-year-old man (retired surgeon) with type 2 diabetes for 20 years; on basal-bolus insulin, metformin, and empagliflozin; an A1c around 9% for 10 years; and coronary artery disease, hypertension, and dyslipidemia. After three months of FreeStyle Libre, his A1c dropped 1.6% (baseline: 9%) to an average glucose of 153 mg/dl and time-in-range of 72%. Dr. Hirsch: “I see this over and over. What did he do? He changed his food. He wasn’t eating bagels and drinking juice and cereals. Nobody is more excited about new medications than me. But as we’ve moved to new diabetes medications, we’ve lost sight of fact that diet and exercise has such a huge role” This has also proved true anecdotally as we’ve shared Adam’s book, Bright Spots & Landmines, with many type 2s and people with prediabetes.

• MiniMed 670G use in a type 1 with down syndrome. Dr. Hirsch showed a case of using hybrid closed loop in a type 1 with down syndrome. Time in Auto Mode on the 670G was only 67% of the time; “I want him over 80% of the time, and would really like over 90% of the time.” Dr. Hirsch noted the 46% bolus, 54% basal split, offering the ability to strengthen the insulin:carb ratio (more aggressive). In his clinic’s experience, 670G users need only 40-45% as basal insulin. Dr. Hirsch also shortened the active insulin time, as we’ve heard others do. We were so moved by Dr. Hirsch’s commitment to this person with such significant special needs.

• Medicare-covered senior moves to Dexcom G5 once covered: This individual was on CGM and always had A1c’s in the 7%-range, but aged out when he went on Medicare – driving an A1c rise to 8.5%. On the trace from fingerstick data, he ran high overnight, given fear of lows without CGM. Dr. Peters got him on the Dexcom G5 Medicare as soon as it was available. He is now down to an average glucose of 146 mg/dl, with a much higher percentage in range and a higher basal rate overnight (more safety). Most importantly, he says, “I feel better. I feel like I’m in more control.”

• Recurrent DKA from daily marijuana use to treat neuropathy (in hospital at least once a month). Dr. Peters shared the puzzling case of someone experiencing DKA recurrently, with no clear cause. She put the patient on CGM and ultimately concluded it wasn’t directly diabetes-driven DKA; rather, DKA episodes came from vomiting after daily use of marijuana to treat peripheral neuropathy. Once the patient stopped smoking, the vomiting stopped, the DKA ceased, and A1c came down to 6.8% This served as a cautionary tale for the audience: “It is helpful to look at old records and listen to patients.” What a puzzle!

• Severe hypoglycemia death while on CGM. Dr. Peters shared this tragic case of a type 1 with “unbelievable variability … no matter how much we tried, we could not get her to give insulin in a way that was rational.” Her husband, who also has type 1 (they met each other at diabetes camp) took on a lot of her management, Dr. Peters said. The woman had three to four severe lows every week, and when her husband to leave for an overnight trip, the woman passed away from severe hypoglycemia while sleeping. Dr. Peters showed the CGM trace – a long period low, followed by a massive spike over 400 mg/dl (presumably a last-ditch counter-regulatory response), followed by death and a period with flat-line low glucose. Reflected Dr. Peters, “Obviously we cannot prevent all severe hypoglycemia and death. Probably one to two times per year I have a patient that dies from severe hypoglycemia. This isn’t trivial – over-dosing on insulin. I cannot stop them. Even this woman, whose husband had taken over, could not leave her for long.” We salute Dr. Peters for being so candid about what must be the absolute low point of any diabetes specialist’s job.

• Libre vs. Dexcom – how do Drs. Peters and Hirsch decide what to prescribe? For somebody who has clear hypoglycemia unawareness, Dr. Hirsch prefers Dexcom for the alerts/alarms and sharing (especially at night). He noted that “Nobody has ever compared hypoglycemia outcomes between the two,” though we’d note Dr. Nick Oliver’s iHART CGM study did do this – with Dexcom coming out on top, unsurprisingly in our view. Dr. Peters does like alarms for those with lows, but lets patients decide – they are the ones wearing the device, she says. She added that part of it is coverage – Libre is much less expensive for those paying out of pocket, but in California, it is “less well covered.” If patients are paying out of pocket, she uses Libre because it is less expensive.

Practical Considerations for Implementing Continuous Glucose Monitoring in the Clinical Setting

Davida Kruger (Henry Ford Health System, Detroit, MI)

In a morning CGM session, Henry Ford Health System’s Ms. Davida Kruger and IDC’s Dr. Rich Bergenstal discussed the economics, infrastructure, and process of implementing CGM in a clinic. Ms. Kruger shared that her clinic grossed $750,000 in revenue from >1,400 CGM data interpretations (both personal and professional). She also discussed some of the measures that her clinic takes to enable such a workload.

• Ms. Kruger’s clinic is a CGM data interpretation powerhouse: In 2017, it analyzed CGM data from >1,400 patients (~30 average starts per week) on personal and professional CGM, grossing $750,000 in revenue (and with a positive net profit). Wow! These stats really put a face to Dr. Etie Moghissi’s claim at AACE that CGM is “not a money-losing proposition” – Ms. Kruger even pointed out that endocrinologists can get paid more for CGM than for thyroid procedures, which is why it is important to stay on top of billing. Ms. Kruger herself sees 15-18 complicated patients every day, each on MDI or a pump – if they’re not, and just taking metformin, for example, then Ms. Kruger simply doesn’t have bandwidth to see them. This sort of complexity and volume hearkens to the importance of infrastructure for professional CGM workflow, which Ms. Kruger discussed in detail: Her practice has 50 Dexcom G4 Platinum and 50 FreeStyle Libre Pro systems, assigned dedicated resources of people and (organized) space, sign-out books for devices, secretarial support, a dedicated cleaning staff, EMR templates/smart sets for both Libre Pro and Dexcom G4 Platinum initiations, and set procedures to maximize efficiency (e.g., patient starts device in the clinic, drops off/mails the device 7-14 days later, medical assistant downloads the data, NPs do most of the interpretation, and then calls the patients). We would love to see experienced teams from clinics with established professional CGMs like Henry Ford’s do rounds at clinics looking to begin their own program – establishing this workflow is non-trivial and certainly went through a great deal of trial-and-error. Ms. Kruger also had a fair number of notable quotes in her talk; see the following bullets.

  • “For my money, if I put professional CGM on someone, I want them to see the data. It’s not my diabetes; give the patient more data!” This was wonderful to hear and so inspiring. She later specified that she lets the patient decide and will often recommend real-time CGM for people engaged in their diabetes and will benefit from seeing the data (also because the only real-time professional option available, G4 Platinum, requires fingerstick calibrations). At the end of the day, she pointed out that Libre Pro is great for some patients, while G4 Platinum is great for others: “When Libre Pro became available, we thought our Dexcom needs would go down, but it just doubled our number of total systems. The people we didn’t think Dexcom was right for could go on Libre Pro.”   

  • “In 2018, reimbursement should not be the thing holding you back from wearing CGM.” This was great to hear from someone on the front lines (who also happens to be a research star)! She cited Medicare coverage of therapeutic CGM, and hopes for Medicaid coverage in her native Michigan soon. The latter is quite spotty and something Dr. Anne Peters brought up – California still doesn’t have statewide Medicaid coverage for CGM although over 30 states do overall.

    • Regarding Medicare coverage, she recommended documenting requirements for CGM coverage – MDI/pump, four SMBG/day, etc. – in the charts of Medicare-age patients, even before requesting CGM. As a follow-up, she strongly advised against requesting CGM from Medicare until the patient has met the coverage criteria, because there’s a chance that after a first denial, coverage won’t be granted in the future.

  • “In my clinic, we move people from professional to personal – we want them to know what they’re getting so we do professional first.”  We love this approach, and wonder how many other clinics are reliably using it! It also suggests professional CGM will be a funnel for personal CGM, a big advantage for Abbott right now. Will Dexcom launch a professional version of the current G6 or wait for the disposable gen one with Verily? We can’t wait to learn more on this question.

  • “Though not approved by FDA, we use CGM in pregnancy or when women are trying to become pregnant.” Given that one in six experience hyperglycemia during pregnancy (this source is said to be IDF Diabetes Atlas – that sounds like a very low statistic given the number of women with gestational diabetes alone, to say nothing of high birth weight infants), we firmly believe that every pregnant woman should be on CGM – this would help detect and more aggressively treat hyperglycemia as soon as possible in pregnancy (as well as hypoglcyemia and severe hypoglcyemia for those on insulin).

• In the midst of Q&A, Ms. Kruger also relayed speculation that Dexcom professional CGM will move to no fingersticks. Though we are not surprised by this news – given the no-cal G6 – Dexcom has never discussed its professional CGM plans publicly (and, in fact, almost never talks about professional CGM or its pipeline). Ms. Kruger didn’t say more on when this iteration could launch, nor on the hardware configuration. It is possible that she could be referring to a professional CGM based on G6 hardware – with a reusable transmitter and disposable sensor – though we find it more likely that Dexcom would wait for the Verily gen one sensor (fully-disposable, potentially 14-day wear, with the G6 sensing platform). According to Dexcom’s 1Q18 call in May, the gen one Verily CGM is in “validation and verification.” A clinical trial of 14-day G6 wear in 2H18 could open the door for both an expanded indication for G6 (from 10- to 14-day wear), and for the gen one Verily sensor. There hasn’t been a firm timing update on Verily gen one in quite a while – it was last slated for a late 2018/early 2019 launch as of JPM in January. However, the more recent February call said timing depended on G6’s FDA path – G6 has now received clearance from FDA, but a 14-day G6 study will need to be conducted first. We look forward to hearing public updates from Dexcom on its next professional CGM play, since this side of the company’s portfolio hasn’t been updated for the better half of a decade (in stark contrast to the personal CGM side), while Abbott seems to be doing well with Libre Pro, and Medtronic is investing significantly here.

  • A factory-calibrated, 14-day, fully-disposable Dexcom professional CGM would have a similar profile to and could be very competitive with Abbott’s FreeStyle Libre Pro. Dexcom’s future professional product might still be toggle-able between real-time and blinded and might still have alarms, which would differentiate it from Libre Pro. At the same time, Abbott is working on a real-time CGM with alarms to incorporate into Bigfoot’s Loop automated insulin delivery system, and we could envision them building out a personal and professional platform off of this base too. The biggest opportunity for Dexcom in advancing its professional CGM pipeline is to reduce cost, which many attending CDEs/NPs cited during Q&A as a reason for putting their patients on Libre Pro. Ms. Kruger noted that her clinic receives transmitters for free because of the bulk of their purchases (and can often stretch them for up to a year), but G4 sensors still cost ~$20 more each than Libre Pro sensors – the lower-cost Verily platform could push Dexcom closer to parity with Abbott’s pricing and change prescribing dynamics significantly.

  • An OUS launch of Medtronic’s Envision Pro (rebranded from “iPro 3”) is expected by April 2020 per the company’s recent analyst meeting. Medtronic’s ADA Analyst Briefing said this will be blinded, fully disposable, no-calibration, and Bluetooth-enabled (single-use data transfer, like Libre Pro.

Best Practices for Interpreting Continuous Glucose Monitoring Data

Richard Bergenstal, MD (International Diabetes Center, Minneapolis, MN)

Dr. Bergenstal outlined IDC's nine steps for interpreting CGM data side-by-side with a patient (see bullets below). He also overviewed a number of CGM rules of thumb, consensus guidelines, and hot topics: (i) At least 14 days of data is desirable for making treatment decision; (ii) the goal is to get glucose profiles “narrow, flat, and in-range”; (iii) the abbreviation GMI (glucose management indicator) has been proposed to FDA and the diabetes community as a replacement for eA1c – “stay tuned to see if it ultimately flies”; (iv) MiniMed 670G pivotal data – 72% time-in-range, 6.9% A1c, 3% <70 mg/dl, 25% >180 mg/dl, is a good starting point for targets in his view (this is new “broad” guidance from Dr. Bergenstal – see what 15 other diabetes experts think in a late 2017 diaTribe article “CGM and Time in Range: What Do Diabetes Experts Think About Goals?”); (v) There is a ~0.4% A1c drop for every 10% increase in time-in-range – “you need to know about that correlation, because patients will ask for it” (this is also new); and (vi) Aim for coefficient of variation (CV) ≤36% (Monnier et al.) or standard deviation (SD) < mean glucose divided by three. See the following bullets for IDC's process of navigating an AGP report with a patient, and a sample marked-up AGP below:

• 1: Check for adequate data (as close to ~14 days as possible). “Don’t spend a lot of time if there’s no data.”

• 2: Mark up the AGP, noting factors that may affect the management plan. “Just print it out and write some stuff on it. Like to know a little bit – how old is person, do they have type 1 or type 2 diabetes, for how long, do they have known CVD, how much do they weigh (that helps with dosing insulin), and their eGFR. Include these little things right in front of me.” Dr. Bergenstal also asks the patient when they wake up, eat breakfast, lunch, dinner, snack, and put medications and other contextual information under the profile.

• 3: Ask the patient, “What do you see?” Listen. Notably, every speaker in the session underscored this approach. Said Dr. Bergenstal, “Usually you walk in and they’re already talking about it – they know why they went high at lunchtime.”

• 4: Look for patterns of low glucose levels. Dr. Bergenstal usually “treats the cloud,” referring to the AGP’s lines denoting the lower 10% decile and the upper 90% decile of glucose. In 15-minute visits, he typically bases therapeutic adjustments on this AGP picture (especially if the 10/90 lines dip low), and will only verify with the daily snapshot reports if he has time.

• 5: Look for patterns of high glucose levels.

• 6: Look for areas of wide glucose variability. Dr. Bergenstal boiled variability down to timing and amount. Timing refers to insulin/meals, weekend/weekday, snacks, exercise, and stress; amount refers to insulin, the insulin:carb ratio, and exercise intensity. “Anything to reduce the variability makes the other therapeutic changes more effective.”

• 7: Compare to past AGP and reinforce successful strategies. “We never go back enough – we can show that we changed something and it worked. Maybe it also brought up a new issue, but we’re treating that now.”

• 8: Get an action plan.

• 9: Copy the AGP for the patient and the EMR.

Posters

Fourth-Generation Glucose Sensor with an Advanced Algorithm Featuring No Calibrations and Optional Calibration Modes of Operation (936-P)

Taly Engel, Peter Ajemba, Jeffrey Nishida, Keith Nogueira, Benyamin Grosman, Andy Tsai, Yunfeng Lu, Ashley Sullivan, Andrea Varsavsky, Northridge, CA

This poster shares the first early data on Medtronic’s factory calibrated CGM development efforts in an n=31 study. Accuracy of a fourth-gen sensor and new algorithm was tested for “up to 7 days,” comparing algorithms requiring two fingerstick calibrations per day (current approach), one calibration per day, and zero calibrations. The data were evaluated retrospectively, and 80 sensors were tested in each group. Overall MARD vs. SMBG values was 10.9% with no calibrations, 9.2% with one calibration, and 9.3% with two calibrations. %20/20 was also a solid 87% without calibration, 92% with one calibration, and 91% with two calibrations per day. Day 1 MARD was 12.7% with no calibrations, 11.8% with one calibration, and 12.0% with two calibrations. Mean sensor life was only six days, a clear area for Medtronic to improve on. See the complete data in the table below. The no-calibration 20/20 accuracy does not look good enough to meet the iCGM standard at this point, as we’d assume the lower-bound of the %20/20 confidence interval dips below 87%.

• We’d note several critical study caveats: (i) it only included n=4 people with type 1, n=17 with type 2, and n=10 without diabetes – critical context, since fewer glucose fluctuations in the latter populations would improve accuracy (% of points in different glucose bins was not shared); (ii) The accuracy comparator was SMBG – not YSI – though over 2,000 paired points were included in each arm; and (iii) the data were evaluated retrospectively.

• That said, these data certainly suggest Medtronic has an R&D pathway to a no-calibration CGM to compete with Dexcom’s G6 and Abbott’s FreeStyle Libre. Like G6, Medtronic’s no-calibration algorithm will accept an optional calibration – e.g., in cases of sensor inaccuracy. The poster also notes that no factory calibration code is needed, an improvement over Dexcom’s G6 and on par with Abbott’s FreeStyle Libre. Medtronic technically calls the system “self calibrating,” as its approach compensates for various factors (e.g., sensor-to-sensor variability, subject-dependent sensitivity changes over time) and combines multiple independent sensor glucose calculation units.

• This appears to be the Harmony sensor mentioned at Medtronic’s 2018 Analyst Meeting – expected to launch within the next two years (by April 2020). At its ADA briefing, Medtronic said this sensor will enter a clinical trial this year. From here, Medtronic will need to demonstrate strong accuracy vs. YSI, test in broad glucose ranges, ensure manufacturing can dramatically limit sensor-to-sensor variability, and ideally get to 10-day sensor life.

• Study methods and calibration details: “An advanced algorithm was developed for a fourth-generation CGM system to determine system performance with the elimination of BG calibrations. The new algorithm uses electrochemical impedance spectroscopy and other measurements to self-calibrate the sensor and adjust CGM system sensitivity. Self-calibration adjusts for potential sensor-to sensor variability, as well as subject-dependent sensitivity changes, over time. It was designed to not require self-monitored (SMBG) input, but to accept calibration at any time to optimize CGM performance. It contains multiple independent sensor glucose (SG) calculation units that implement a distinct model of interstitial glucose that accounts for differences in response caused by sensor-dependent and time-dependent nonlinearities. The design approach is particularly useful during the post-sensor insertion period and during period of hypoglycemia. Each dynamic model analyzed was selected for its unique pattern of strengths, which complement each other to augment overall performance. The output of each model was combined to provide a final SG value and a measure of confidence in the calculated value. Analytical accuracy (MAD) and relative difference (MARD) and percentage agreement rates relative to an SMBG reference were retrospectively evaluated in 31 subjects who wore a sensor on the arm connected to a glucose sensor recorded for up to 7 days.”

Accuracy Evaluation of the WaveForm Cascade CGM System Vs Dexcom G5 Sensors (81-LB)

M Rebec, E Anderson, R Dutt-Ballerstadt, A Haider, A Janez

An AgaMatrix poster depicted results from a head-to-head comparison of the WaveForm Cascade CGM vs. the Dexcom G5. Seven-day MARD vs. YSI comparator were similar between the Cascade system calibrated once-daily and G5 calibrated twice-daily: 10.4% and 11.0%, respectively. Cascade MARD at 10 days was 10.9%, showing sustained accuracy beyond seven days. Consensus error grid analysis over seven days of the Cascade showed that 90% of data points were in zone A, similar to 93% for G5. The seven-day study was conducted in Slovenia and included only 15 individuals (some with type 1, some with type 2; breakdown not specified), so it’s hard to know if they will hold up in a larger study. Still, these early results suggest Cascade may meet the ~9%-11% MARD that has now become industry standard in CGM. The poster also includes longer 14-day data from clinical studies currently underway. Average MARD vs. YSI over 14 days is 11.9%, with individual per-day in-clinic MARD vs. YSI reported at 12.3% on day 1; 10.6% on day 4; 11.2% on day 7; 11.8% on day 10; and 13.3% on day 14. The poster notes that further clinical studies to be included with CE mark submission are planned for 3Q18-4Q18. An EU launch is expected for 1Q19. This timeline is delayed by about a quarter from that shared in January, which planned for a CE mark and launch in 2018. At the time, Waveform also hoped for clinical trials in support of PMA submission to begin “as early as 2019,” with an IDE for the US planned by the end of 3Q18; we assume that is back by at least a quarter at this point.

• Earlier in June, WaveForm licensed two OHSU hybrid closed loop control algorithms, as well as OHSU’s smartphone and cloud-based drug delivery platform iPancreas, in a non-exclusive agreement. WaveForm has reportedly begun testing a hybrid closed loop system consisting of Cascade, the OHSU algorithms, and a pump in May, with testing of its own “vertically integrated’ system (i.e., owning all the components) slated for 2019.

Associations between A1C and Continuous Glucose Monitoring-Derived Glycemic Parameters (899-P)

P Calhoun, T Johnson, J Welsh, T Walker, D Price

A picture is worth a thousand words with this poster. Data compiled from 530 Dexcom CGM users (455 type 1s) in the last two weeks of outcomes trials – DiaMonD Phase 1, DiaMonD Phase 2, Replace-BG, and HypoDE – provides a look at the relationship between A1c and time in ranges. Based on the figures, mean glucose, time >180 mg/dl, and time 70-180 mg/dl correlate strongly with A1c, while time <70 and <54 mg/dl do not as strongly, confirming that A1c is more a measure of hyperglycemia than hypoglycemia. As for specific numbers, median time-in-range for subjects with A1c <6.5% was nearly 75% (~6% of time <70 mg/dl), while time-in-range for subjects with A1c ≥8.5% was <40% (~1% of time <70 mg/dl, >20% of time >250 mg/dl). Patients with high A1cs did spend less time <54 mg/dl, but the median time was <20 minutes/day for all groups. As the blue box on the right points out, >90% of subjects had an A1c value <7.0% with: (i) mean glucose <140 mg/dl; (ii) >80% time 70-180 mg/dl; or (iii) <2% of time >250 mg/dl. These could offer great benchmarks for CGM goals – more granularity to help patients lower their A1c. For more on the relationship between A1c and CGM-derived metrics, see Dr. Roy Beck’s 54-page slide deck from a talk he gave at The diaTribe Foundation’s ATTD 2018 Art Walk in Vienna – this is the single most comprehensive resource exploring these relationships that we know of!

Cost Calculation and Adherence to ADA Recommendations Based on a Flash Continuous Glucose Monitoring System for People with T1DM or T2DM Using MDI Therapy (69-LB)

R Hellmund

An interesting analysis from Abbott concludes that FreeStyle Libre is cheaper than SMBG for people on MDI, whom the ADA recommends take 6-10 fingersticks per day. Assuming $0.60-$1.42 per strip (based on manufacturers’ list prices and a discounted price for strips) and adherence to ADA’s fingerstick frequency recommendations, per-person-per-month (PPPM) pricing for SMBG ranges anywhere from $108-$426. Assuming each FreeStyle Libre sensor costs $36, discounting the cost of the reader, and incorporating 0.5 fingersticks per day for type 1s (based on IMPACT) and 0.3 per day for type 2s (based on REPLACE) with the same pricing assumptions as above, FreeStyle Libre costs $117-$129 PPPM for type 1s and $113-$121 PPPM for type 2s. Both of these tallies are slightly higher than SMBG when the assumption is six tests per day and $0.60 per strip, but given the benefits of historical data and trend arrows, these additional $5-$9 PPPM are well worth the investment in our view, assuming there is the possibility for a patient to pay (obviously in many cases this is simply not possible). It is possible to obtain strips at a cost of $0.22 each without insurance (e.g., Accu-Chek SimplePay program), which would put PPPM costs between $40-$66, significantly lower than estimations derived from the poster’s estimations. The authors also looked to the IMPACT and REPLACE RCTs to calculate an “effective cost per scan.” Assuming 15.1 scans per day in IMPACT and 8.3 per day in REPLACE, the effective cost per scan in type 1 would be $0.24 and in type 2 it would be $0.43 (as noted, Libre glucose monitoring also gives the user arrows and eight hours of historical data). Our back-of-the-envelope calculations show that 16.4 scans/day would be necessary to achieve an effective cost per scan of $0.22. The author notes that the high rate of scanning seen in IMPACT and REPLACE have been replicated (if not exceeded) in large, real-world data sets (see Abbott’s symposium in this report), and conclude that Libre CGM is cost-effective relative to SMBG in people on MDI. The acquisition cost of SMBG may be lower than that cited in this paper, the numbers still suggest that SMBG and FreeStyle Libre are at minimum cost-comparable in many users, without even considering the decrease in hypoglycemia demonstrated by FreeStyle Libre nor the long-term benefits in the form of (probable) diminished complications.

A Meta-analysis of Real-World Observational Studies on the Impact of Flash Glucose Monitoring on Glycemic Control as Measured by HbA1c (72-LB)

A Seibold, S Ells, C Schlaeger, Z Welsh

This Abbott poster presented results from a meta-analysis of 17 studies (including IMPACT, REPLACE, SELFY, etc.) reporting a mean effect of FreeStyle Libre use on A1c of -0.56% (95% CI: -0.76, -0.36). The analysis incorporated data from a remarkable 1,338 participants (type 1: n=1,112; type 2: n=226), including children, adolescents and adults, from real-world observational studies reporting longitudinal A1c data using Abbott’s FreeStyle Libre. The trials were weighted by inverse of variance observed (more consistency in A1c effect => higher weighting in the meta-analysis). The investigators found substantial heterogeneity between trials, much of which could be explained by differences in baseline A1c. The missing link in the meta-analysis is outcomes beyond A1c, and especially hypoglycemia. The IMPACT study of FreeStyle Libre in type 1s seems like a major drag on FreeStyle Libre’s effectiveness in this analysis, since patients using Libre saw mean A1c rise from 6.7% to 6.9% (+0.2%; the greatest increase of any of the studies presented), given the profound level of hypoglycemia they had at baseline (more than three hours per day). However, the “quality” of A1c was improved markedly: Patients using Libre spent ~49 minutes fewer per day <55 mg/dl (a 50% reduction; p<0.0001) and ~33 minutes fewer per day <45 mg/dl (a 60% reduction; p<0.0001). It was a similar story in the REPLACE study of FreeStyle Libre in type 2s with a high baseline A1c, as both Libre and control SMBG users had mean A1c reduction of 0.3% - a finding attributed in part to differences >65 years old.

• No significant differences were observed based on length of study, type of diabetes, or children vs. adults. The latter observation is particularly important, as it suggests that both adults and children can equally benefit from the A1c improvements associated with the FreeStyle Libre. Abbott has submitted a pediatric claim for FreeStyle Libre to FDA; Libre is already approved down to four years-old in Europe.

Budget Impact Analysis of Self-Monitoring of Blood Glucose vs. Flash-Continuous Glucose Monitoring in Intensive Insulin Users with Diabetes Type 2 Covered by Medicare and Medicaid (142-LB)

M Stueve and Y Zoellner

A poster from a J&J-employed lead author determines that the reimbursement costs incurred to CMS for FreeStyle Libre is 12x-greater than that from SMBG at a fingerstick frequency of 3.7 times/day – it suggests that reimbursement should be focused on the type 2s that will benefit the most, and that “on cost-effectiveness grounds, SMBG remains the rational and obvious choice.” The poster bases calculations off a target population of 663,742 adults with type 2 diabetes on intensive insulin therapy and a patient-paid co-pay of 20% for all devices. The cost of SMBG to CMS is $180 per patient per year (~$0.50/day), the poster claims, compared to $2,156 incurred per FreeStyle Libre user per year (a difference of $1,976 per year, or $5.41 per day). At these figures, CMS would spend the same amount covering 12 patients on SMBG for every one on FreeStyle Libre. The poster also says the technologies would break even at a consumption of 44 SMBG strips per day. This analysis makes a good point about the relative cost differences between the technologies, and we’d acknowledge that REPLACE didn’t show an improvement in the primary outcome of A1c reduction. That said, REPLACE did show improvements in hypoglycemia (including time <45 mg/dl), suggesting that FreeStyle Libre could lead to huge cost savings from (i) a reduction in hospitalizations and (ii) a lesser degree of fear of hypoglycemia, and therefore more aggressive diabetes management. We are not all that surprised to see this analysis, especially since FreeStyle Libre’s pricing at ~$8 per day via CMS reimbursement for therapeutic CGM is nearly double its ~$4-$5 retail price per day. But at the end of the day, people on insulin (and insulin secretagogues) should be equipped with the technology to see where their glucose has been and where it’s going – this will surely benefit CMS in the short- and long run. Perhaps the bigger question is who should use SMBG – and how often – as standard of care moves to CGM …

Sustained CGM Use in Low Income Youth Following Insurance Coverage (1390-P)

P Prahalad, B Buckingham, D Wilson, and D Maahs

It’s well-documented that cost and reimbursement are two of the biggest barriers to CGM adoption, so what happens when low-income youth finally get access to CGM? According to this retrospective study, they stick with it. 78% of lower socioeconomic status youth who started on CGM continued to use it at six months, and those individuals who stuck with it wore it consistently (13.3 days out of a 14-day period). These data come from the first 40 children at the Stanford pediatric diabetes clinic to be approved by California Children’s Services (CCS; a supplemental state medical insurance for low income children with chronic medical conditions). The children had a mean age of 12 years and a diabetes duration of six years, a surprisingly high 65% were on pumps, 67% were ethnic minorities, and 15% were non-English speakers. Notably, of the eight children who stopped wearing CGM, two were due to lapses in insurance coverage, meaning only six dropped due to personal preference. Mean A1c was stable at 8.2% in those who continued at six months (no change), and those who continued spent just over an hour per day (4.3%) <70 mg/dl. Skeptics might point out that Stanford is one of the best places for pediatric diabetes care in the world and can offer top-notch clinical and psychiatric support for children and adolescents starting on CGM – this is true, of course, but patients are still on their own outside the clinic, and thus we still find the data quite illuminating overall. On-body form factor and the hassle(s) of wearing devices are still a barrier for some, but the number one factor remains the ability to pay and obtain insurance coverage. We hope to continue to see great work supporting this point, as well as counteracting the notion that the status with which someone was born should influence the quality of medical care they receive.

Real-World Flash Glucose Monitoring in a Developing Country (75-LB)

LE Calliari, Y Xu, S Jangam, G Hayter, T Dunn

This Abbott poster compared glycemic outcomes associated with FreeStyle Libre use in Brazil (n=8,979 readers) vs. worldwide (n=237,747 readers), confirming that use of the device in a developing country yields similar behavior and results as in the rest of the world. Daily scan rate was significantly (but only slightly) higher in Brazil than the average for all countries (13.5 scans/day vs. 13.2 scans/day; p = 0.002). Aligning with previous FreeStyle Libre real-world analyses, increased scanning frequency was linked with improved glycemic outcomes. In Brazil, those who had the lowest number of scans/day (4.2 scans/day) spent significantly less time in-range (70-180 mg/dl; 13.5 hours/day), compared to the highest average scan group (34.2 scans/day) who spent 16.2 hours/day in range (p<0.001). Worldwide data were very similar. The trend held for time-in-hyperglycemia (>180 mg/dl), which decreased from 9.2 to 6.4 hours/day (p <0.001) between scanning frequency groups in Brazil, and time in hypoglycemia (<54 mg/dl), which decreased from 33.3 to 30.0 minutes/day (p <0.001) in Brazil – the lack of a clinically meaningful impact on the latter make us wonder how time <54 will change once next-gen Libre adds alarms? Comparable hyperglycemia and hypoglycemia data were observed worldwide. The results translated to an estimated A1c decrease from 7.7% to 6.8% in Brazil. The Santa Casa de Sao Paulo School of Medical Sciences and Abbott investigators concluded that flash glucose monitoring is as effective in Brazil, as it is in the other primarily highly-developed countries that have previously been studied.

Predicting Future Glucose Fluctuations Using Machine Learning and Wearable Sensor Data (738-P)

C Panagiotopoulos, A Hayeri

An algorithm developed at Vancouver’s BC Children’s Hospital was able to predict glucose 30- and 60-minutes in advance with strong accuracy based on CGM and Fitbit (step count + heart rate) data in a small pilot study with eight pediatric type 1s. The algorithm was trained on two weeks of patient-specific data, validated for another week, and then evaluated for prediction accuracy, as measured by Clarke Error Grid (CEG) analysis, over the next five weeks. During this period, predictions 30-minutes into the future were very accurate: On average, 98% of values fell within regions A or B of the CEG (83% in region A), indicating that they “will not lead to inappropriate treatment.” Prediction accuracy deteriorated, but was still robust in the 60-minute prediction horizon: 95% of values still fell within Regions A and B, though with a lower ~64% in zone A. The model is currently undergoing further testing as the app DiaBits for iPhone (4.1/5 stars on App Store from 35 ratings), with an Android app coming soon. From the looks of the screenshot below, Diabits also tracks current and predicts future insulin and carbs “on board,” similar to the DIY Loop app. Based on the screenshot below, we assume the algorithm predicts future glucose using meal and insulin input.

• One Drop and Sugar.IQ also discussed glucose prediction features at ADA. At DiabetesMine’s D-Data Exchange, One Drop shared bold new plans to launch a forward-looking 12-hour glucose prediction for type 2 users not on insulin in 3Q18; it is based on limited fingerstick data. The feature may eventually support people on insulin as well. As for Medtronic/IBM Watson’s Sugar.IQ app, forward-looking glucose predictions for MDI users are in-development, with a launch slated for “beyond” April 2020. Of course the DIY community has been predicting future glucose and acting on it for some time now, but we’ll be glad to see commercial products rolling this important feature in. We think it could be quite significant for driving behavior change!

A Wireless, Integrated, Extremely Miniaturized Continuous Glucose Monitoring System (930-P)

M Mujeeb-U-Rahman, M Nazari, M Sencan

A poster detailed high-level accuracy (in rats), warmup (in rats), and longevity (in vitro) data of Integrated Medical Sensors’ extremely small (smaller than a sesame seed at 0.1mm x 0.6mm x 3mm) implantable CGM. With one calibration per day and what appears to be <25 total readings in the “normal to hypoglycemic range” in rats, day one MARD was 8.6%, day three MARD was 8.6%, and day five MARD was 9.9%. In vivo results also demonstrated a <two-minute lag time, as well as a minimal warmup (results began lining up with SMBG at ~11 minutes, though the poster claims “no” warmup time). Finally, in vitro testing shows the device has consistent sensitivity (glucose oxidase-generated current) past 30 days, and possibly up to 50 days – the in vivo accuracy data showed MARD inched up to 9.9% by day five, so extended in vivo data will be needed to support this assertion. We’re also not sure when human testing will commence. See more on Integrated Medical Sensors from ATTD.

• The (projected) ~$1/day implant can be inserted using a custom injector and removed using a “simple procedure under local anesthesia.” The device is powered by an external wearable transmitter with RFID, and the transmitter uses Bluetooth to communicate with a smartphone reader (i.e., similar form factor to Senseonics). Wound healing response and histopathology studies indicate that the device is safe and biocompatible.

Use of the New Smartphone Application for Blood Glucose Monitoring (BGM) with Information-Motivation-Behavioral Skills (IMB) Model Has an Impact on Diabetes Control Parameters (925-P)

S Pardo, S Zhuplatov, J Wallace, T Bailey

In a six-week study of n=46 insulin users (76% type 1, 39% pumpers), Ascensia’s Contour Next One BGM and updated app proved usable (89% of subjects successfully synced meter blood glucose value meters with the app) and decreased fructosamine levels (which correlate with blood glucose over the prior two-three weeks), though it did not significantly decrease A1c (7.7% at baseline and completion), weight, BMI, or self-reported mean daily insulin. The updated app – available in 24 countries as of May – incorporates glucose pattern-recognition, reminder plans, and a glucose-insulin-carb overlay view. It uses the Information-Motivation-Behavioral Skills (IMB) model of health behavior change to help the user achieve behavior changes that promote glycemic management. Use of the app caused fructosamine levels to drop significantly over the six weeks (from 333 μmol/l to 323 μmol/l; p<0.0076), and post-hoc analyses showed: (i) a decrease in the median number of “high” blood glucose measurements per day from ~six at the beginning to ~three at the (p<0.0003; “high” threshold not specified); (ii) a statistically significant (but clinically insignificant) increase in the mean fingersticks per week from ~4.3 to ~4.4 per day (p<0.001); and (iii) no statistical difference in pre-meal, post-meal, or fasting glucose levels. Further on usability, 98% were able to access and interpret blood glucose displays, and 100% were able to access and use the Smart Reminders feature in the app. That fructosamine levels decreased while A1c stayed constant in the short study suggests that a longer study might be required to see the full glycemic benefits of the app in this insulin-using population. We’d be interested to see the app evaluated in other populations such as type 2s on basal insulin or orals, and possibly along with CGM! Ultimately, the results of the study showed limited clinical impact, though the primary endpoint – usability of the app and its new features – was met. The Contour Diabetes app currently has 2.7/5 stars on Google Play from 822 ratings, and 2.5/5 stars on the App Store from 179 ratings.

Adults with Type 1 Diabetes (T1D) Using Continuous Glucose Monitoring (CGM) Report Disease Has Little Impact on Daily Functioning–T1D Exchange (932-P)

L Fan, C Garey, J Liu, B Mitchell, J Bispham, A McAuliffe-Fogarty

In a real-world, prospective, 28-day, Lilly-sponsored study of n=98 type 1 CGM users (mean A1c: 6.8%) recruited from the online patient community myGlu.org and T1D Exchange Registry, metrics of glycemic variability were compared to patient-reported outcomes: insulin treatment satisfaction questionnaire (ITSQ) and the work productivity impairment and activity impairment questionnaire (WPAI). A long-desired study, we would have hoped the data could show how negatively treatment satisfaction and productivity is affected by glycemic variability. Strangely, that is the opposite of what the study found. Insulin treatment satisfaction surprisingly increased slightly as measures of variability and hypoglycemia deteriorated (i.e., more variability had slightly higher treatment satisfaction). As presented in the poster, this relationship held for coefficient of variation (CV; standard deviation divided by mean blood glucose), low blood glucose index (LGBI), percent time <70 mg/dl, and daily hypoglycemia area under the curve (AUC) – see first picture below. Interestingly, glycemic variability metrics were not associated with productivity. These results are obviously counterintuitive to us: Ask most people with diabetes if they prefer to be in-range with low variability and little hypoglycemia, and they’ll say “yes.” The counterintuitive data could reflect a few factors: (i) the cohort, as demonstrated by low A1c, 100% CGM use, 88% pump use, and little baseline variability represents a very engaged subset of the type 1 population, with low glycemic variability (CV: 36%), high baseline treatment satisfaction already, and things mostly in control; (ii) the study population is willing to accept slightly more variability and hypoglycemia to stay at A1c goal (time <70 mg/dl was 6% per day); (iii) given the population, perhaps there wasn’t enough difference between days to pick up a correlation between variability and treatment satisfaction; and (iv) the ITSQ measure of treatment satisfaction may not be sensitive enough to pick up changes or may not be the right metric. Still, we’d love to see more studies of this type – linking CGM metrics to PROs – particularly in populations with higher A1c levels, those with more glycemic variability (CV >50%), and those naïve to CGM.

• We’d love to follow up with the cohort to: (i) ask them about their views on A1c and time-in-range; (ii) track them over time and see whether changes in time-in-range correlate more closely to a wider range of PROs. We’d also be fascinated to see a day-by-day analysis, with PRO metrics overlaid on CGM modal graphs – when you have THIS kind of day, what is your satisfaction, productivity, etc.?

  • In a way, this could be a backdoor approach to validate time-in-range – show that variability metrics correlate with quality of life, treatment satisfaction, productivity, etc. This didn’t work here, but we’re still hopeful that future studies could do so. Of course, Dr. Roy Beck’s presentation on the strong correlation between in-range glucose readings and microvascular complications in the DCCT could also go a long way on the validation front.

• Despite evidence that diabetes significantly reduces productivity on a population level, glucose variability did not show any association with work productivity impairment or activity impairment in this study. Although these results are not generalizable to the general population since the study participants are likely more motivated and engaged with their diabetes than most, it does suggest that technology can help minimize interference with daily life.

Three DarioHealth Posters in Type 2s: Reduction in Hyperglycemia (76-LB); Decrease in Severe Hyperglycemia (77-LB); Decline in Average Blood Glucose (78-LB)

Y Hershcovitz, S Dar, E Feniger

A trio of late-breaking DarioHealth posters revealed promising reductions in hyperglycemia and average blood glucose over one year among type 2 diabetes users of the all-in-one Dario BGM device (headphone jack meter + lancing device + strip vial) and app. Patient data was retrospectively analyzed from the Dario cloud database during the entirety of 2017. In line with the move to population management and selling outcomes (with strips as one tool), Dario recently announced the launch of tiered subscription plans, offering patients access to “specialists” (coaches) plus unlimited strips and other supplies.

• 76-LB: A cohort of 17,156 type 2 adults saw a 19% reduction (from 28% to 23%) in proportion of “high blood glucose readings” (180-400 mg/dl). At the same time, the proportion of “normal range readings” (80-120 mg/dl) increased by 11% (from 26% to 29%), suggesting that the decrease in hyperglycemia did not come at the expense of increase hypoglycemia. Interestingly, the most significant shift occurred after one month of usage (a 14% reduction in high values) followed by relatively stable measurements over the following months. Using 80-120 mg/dl is quite a tight range, and we wonder what results would have looked like using the standard 70-180 mg/dl for “normal range readings.” The study was also notably lacking any analysis of statistical significance, though given the high n, it’s likely the results were significant.

• 77-LB: Of 225 type 2 adults who “continuously” measured their blood glucose with the Dario BGM for the entirety of 2017, the proportion of high readings (180-400 mg/dl) reduced from 23% to 19% over the course of the year. Importantly, the proportion of severe hyperglycemia readings (>400 mg/dl) decreased by 58% (from 1% to 0.4%). No statistical analysis was provided to determine if the decreases were significant.

• 78-LB: Among 238 type 2 adults classified as “highly engaged” (>1 fingerstick/day – arguably a low bar) and “high risk” (average blood glucose level >180 mg/dl in the first 30 days on Dario), average blood glucose was 225 mg/dl at baseline, decreasing by approximately -15 mg/dl at three months (-7%), -25 mg/dl at six months (-11%), and -32 mg/dl at 12 months (-14%). Of the 180 users (76% of the total) who did decrease their blood glucose average over the year (i.e., a responders analysis), mean glucose reductions were larger at -23 mg/dl (-10%) at three months, -37 mg/dl (-16%) at six months, and -55 mg/dl at 12 months (-24%) from a baseline of 228 mg/dl. Again, these results are encouraging, though we wish the investigators had provided statistical significance indicators.

Oral Presentations: From Progression to Management in Type 1 Diabetes—What Is New?

Timing of CGM Initiation in Pediatric Diabetes—The CGM TIME Trial

Margaret Lawson, MD (Children’s Hospital of Eastern Ontario, Ottawa, Canada)

Dr. Margaret Lawson (Children’s Hospital of Eastern Ontario) presented results from the JDRF-sponsored CGM TIME trial, demonstrating higher CGM adherence in pediatrics when CGM is initiated at the same time as pump therapy as opposed to after a six-month delay. The five-site, randomized controlled trial (n=144) evaluated the number of hours per 28 days that pump-naïve children (5-18 years) with type 1 diabetes used CGM over a span six-months. The study excluded those who had used CGM for more than 50% of the time during the six months prior to the study. Those randomized to simultaneous initiation of CGM and pump therapy spent an additional 2.8 hours/day wearing CGM during the final month of the study than those in the delayed group. Significant differences in wear-time between the two groups were observed consistently and (modestly) increasingly, beginning in month one. There were no significant differences in A1c, despite better CGM adherence in the simultaneous start group (though it’s possible hypoglycemia and hyperglycemia might have been lower in this group). Females exhibited significantly higher adherence overall. Dr. Lawson noted that there were unexpected differences in adherence by study site. In fact, she mentioned during Q&A that in two of the five sites, “CGM was not used very well at all,” despite implementing standardized education and initiation protocols. Dr. Lawson is interested in further investigating the differences between sites to potentially identify predictors of CGM success. We wonder how CGM adherence might change if pump therapy was initiated six months later – especially as the field seems to be shifting toward a “CGM-first” mindset, with growing MDI studies. Wearing a sensor would theoretical give pump-naive patients a better understanding of their diabetes before handing them a tool to fine tune insulin delivery.

• We’re not entirely sure why the investigators decided to include participants who had used CGM in the past, rather than a completely naïve population. Was it to investigate the behavior around trying CGM for a second time after low initial adherence? Of course, behavior in an RCT is inherently flawed, so researchers might be better off studying real-world observational data. Perhaps it also had to do with a shortage of enrollment with stricter exclusion criteria – CGM is increasingly common in young children with type 1, at least in the T1D Exchange.

Data, Data, Everywhere and Not a Pattern to Be Found (Sponsored by the Helmsley Charitable Trust)

Ambulatory Glucose Profile (AGP) – The Picture Says It All

Mary Johnson (Park Nicollet International Diabetes Center, St. Louis Park, MN)

IDC’s Ms. Mary Johnson informed the audience of further momentum in AGP adoption, as IQVIA, Medtronic (officially), Senseonics, and Ypsomed have joined the ranks of existing licensees – Abbott, Dexcom, Glooko/Diasend, and Roche. Wow, what a victory for standardization in the field to see more players come on board, and we’d note that many companies who haven’t officially adopted the one-page standard AGP and glucose metrics are still displaying glucose in similar fashions. It seems that a tipping point has come and passed, as all major CGM companies, plus entities in data management and pumps, have signed on, all in the past two years (the first inked contracts were announced at ADA 2016). Now that the standardized one-page intuitive display is becoming ubiquitous, we look forward to next steps taking hold: Widespread education about best interpretation practices (as outlined by Dr. Bergenstal at ADA) and progress in establishing benchmarks for CGM metrics (time-in-range, hypoglycemia, glycemic variability). Many researcher/clinicians have highlighted the MiniMed 670G pivotal trial’s >70% time-in-range as a possible goal to shoot for, and a Monnier et al. paper suggested ≤36% coefficient of variation represents “stable” glucose (though Dr. Irl Hirsch and others prefers ≤33%, which is more simply stated as “SD should be less than one-third of the average).

• Ms. Johnson emphasized the long journey toward a standardized CGM reporting system over the last six years and the broad support for the concept from organizations like the ADA, AACE, JDRF, Helmsley Charitable Trust, AADE, ENDO, and ATTD – indeed, we covered the HCT/IDC expert panel in 2012, followed by journal publications in 2013. She then offered a detailed walk-through of the AGP in CGM, pump, closed loop, and SMBG views, all of which are preferred by both patients and clinicians as an improvement over previous CGM reporting interfaces, according to Ms. Johnson. She emphasized that patients appreciate AGP’s simplicity, effectiveness, and the fact that all of the needed information is found on one page. As one caregiver put it, “For my daughter to see the graph vs. just numbers was a great validation of what is happening from her choices.” On the clinician side, providers appreciated the clean design and efficiency of AGP. One rave review: “The glucose statistics, ambulatory glucose profile, and daily glucose profile are fantastic! It saved me a lot of time to quickly identify the patterns, trends, and issues.”

• Ms. Johnson concluded her talk with practical tips for systematic interpretation of AGP. The International Diabetes Center has created a clinician tool with a standard process to guide clinicians in interpretation: (i) check for adequate CGM data (e.g., at least 14 days); (ii) mark up the AGP with patient information and timing of insulin and food intake; (iii) ask the patient what they see in the AGP; (iv) look for patterns of lows; (v) look for patterns of highs; (vi) look for areas of wide glucose variability; (vii) compare to a past AGP and reinforce successful strategies; and (viii) agree on an action plan with the patient.

Professional Continuous Glucose Monitoring – Avoiding Pitfalls and Optimizing Outcomes

Patricia Knutsen (Washington University School of Medicine, St. Louis, MO)

Ms. Patricia Knutsen offered a glowing endorsement of professional CGM use: “The most fun I’ve had in 30 years of diabetes care is using CGM. It’s the best thing we can do for patients. I almost feel like I’m cheating because it’s gotten so easy now. It’s probably the best thing that’s happened in my career.” Compared to her experience analyzing written glucose logs in the past, the first time she worked with an Ambulatory Glucose Profile (AGP), Ms. Knutsen described feeling like she had “died and gone to heaven.”

• She especially recommended the use of professional CGM in a handful of distinct patient populations:

  • Patients who are interested in CGM, but not quite willing to take the plunge. For instance, these patients may be wary of adding another wearable to their diabetes regimen on a long-term basis. Ms. Davida Kruger, in fact, said that every patient she sees interested in CGM tries professional CGM first – “we want them to know what they’re getting so we do professional first,” she said.

  • Patients who don’t qualify for a personal CGM. This may be driven by insurance barriers.

  • Patients who are trying to identify causes of specific issues in their diabetes management. For example, patients who feel that they are hyperglycemic all the time and respond by giving more and more insulin to no avail, and possibly with risk of hypoglycemia.

  • Elderly patients with a caregiver.

• In addition to the ease of use of professional CGM and the patient benefits, Ms. Knutsen repeatedly underscored that professional CGM insertion, patient education, and interpretation are all billable services and provided detailed descriptions of all of the CPT codes related to professional CGM. The advent of Abbott’s FreeStyle Libre Pro really elevated the enthusiasm surrounding professional CGM to new levels. Words like “affordable,” “profit,” “cheating,” and “easy” are regularly used to describe Libre Pro’s implementation in clinics, and with improvements in Medtronic and Dexcom professional CGMs in the pipeline, usage and demonstration of benefit will only grow in lockstep. We hope to see a study comparing professional vs. real-time CGM in type 2 diabetes – who benefits from which one, and at what cadence should each be deployed (e.g., quarterly, 24/7, etc.).

Blood Glucose Meter – Interpreting Clinical Data

Alison Evert (University of Washington Neighborhood Clinics, Seattle, WA)

Ms. Alison Evert took the stage to discuss how healthcare providers can best work with patients with diabetes to encourage regular SMBG use, and how to best interpret this data for making treatment recommendations. She stressed the utility of BGMs in providing an important, more time-relevant measure besides A1c in driving therapeutic changes, but emphasized that data must be properly collected and organized to be useful. According to Ms. Evert, blood glucose should be a vital sign used for people with diabetes – a tenet that is actually the mission statement of Medtronic’s type 2 diabetes (“non-intensive”) business unit. Absent in her discussion was exactly which patients she believes should be monitoring, and how frequently. For example, data in type 2s not on insulin has been mixed – e.g., results from the SMBG study suggest that fingersticks are extremely advantageous from an A1c reduction perspective, while the Monitor study showed no benefit of testing on A1c nor on quality of life.   

• Ms. Evert noted that simple benchmarks for measurements such as mean glucose level and standard deviation can be used to make easy rules for patients to interpret data. She has patients target a standard deviation less than half the mean blood glucose (SD x 2 < mean BG level) – a looser benchmark (CV<50%) than that proposed by people like Dr. Irl Hirsch, who aims for SD x 3 < mean BG level (CV<33%). Dr. Hirsch has said that SD x 2 < mean glucose is the minimum amount of variability without having excessive hypo or hyperglycemia, though the entire rule only really works for glucose means between 120-180.

• Ms. Evert also encourages patients to take quick pictures of the food that they eat so they have food data on their smartphone, which they will obviously bring to their appointments. We’re all in on this approach, especially since the phone automatically time stamps the photos. This is the whole philosophy behind apps like Meal Memory (no longer available, but Dexcom hired developer Doug Kantor), the one currently rolling out in Onduo’s launch, and one in used in a Helmsley Charitable Trust-funded study at OHSU – we think the overlay of food photos with blood glucose traces will be incredible for driving behavior change!

Symposium: Joint ADA/JDRF Symposium – Current Management of Type 1 Diabetes in Youth – What Are the Options?

Monitoring Glycemia in Youth with Type 1 Diabetes Mellitus—Meter Blood Glucose, Continuous Glucose Monitoring, or Both?

Gregory Forlenza, MD (Barbara Davis Center, Aurora, CO)

The always-engaging Dr. Gregory Forlenza expressed some frustration with ADA’s conservative CGM guidelines, which he says underrepresent CGM’s utility in patients beyond those who are not meeting glycemic targets and those with hypoglycemia unawareness and/or frequent hypoglycemic episodes. If one does not fall into either of these categories, he remarked, then insurance companies often decline coverage. Based on our read of the ADA 2018 Standards of Care document, we were pleased to see a strong recommendation for using CGM in type 1 adults (ages 18+, improved from 25+ in previous iterations) not at glycemic target, though we fear (and Dr. Forlenza’s account confirms) that some will take the corollary – CGM isn’t recommended for people who are at target – to be a highly unfortunate interpretation. CGM can certainly be useful for anyone with diabetes – whether they are at target or not. The document also gives the statement, “CGM may be a useful tool in those with hypoglycemia unawareness and/or frequent hypoglycemic episodes,” a grade of “C” for strength of evidence – an incredibly weak vote of confidence, given the mounting data. Digging deeper, ADA gives use of CGM in MDI- and pump-using children and adolescents a grade of a “B,” a very positive update following DiaMonD, GOLD, and other studies. (ADA also endorsed CGM in people ages 65+ and gives a positive mention of CGM in “selected patients with type 2 diabetes,” though also claims that “Data indicate similar A1C and safety with the use of CGM compared with SMBG.”) We agree with Dr. Forlenza that the guidelines are limited and need to improve, though as the evidence base grows, that should evolve in a positive direction – access obviously remains a huge issue in the meantime, though stronger guidelines could certainly help with this. And even once coverage is obtained, said Dr. Forlenza, a new set of limitations arises: Youth are often opposed to the feeling and aesthetic of wearing a device or may experience skin reactions and adhesive allergies. As such, despite Dr. Forlenza’s confidence in CGM as a superior glucose monitoring method, he feels that the decision between CGM and BGM for a younger patient with type 1 diabetes is one that must be made as a team, with the patient, family, and provider all involved. These barriers do of course loom, though an encouraging ADA poster from the Stanford pediatric group showed that, when low-income youth finally get access to CGM, they stick with it.

• Despite the advantages conferred by CGM over BGM, Dr. Forlenza stressed that BGM will remain relevant until access issues are resolved. With many countries lacking access to even the most basic diabetes technology, ubiquitous CGM use worldwide remains many years in the future. Of course, we’re heading in the right direction with competition on the manufacturer side to drive down cost – the fact that more than 1 million globally are on CGM now (>800,000 on FreeStyle Libre, >270,000 on Dexcom, likely >150,000 on Medtronic). We hope that access will remain at the top of industry’s priorities in a world that is increasingly stratified by access to medical equipment and care.

• A large majority of Dr. Forlenza’s presentation was dedicated to dispelling the myth that CGM is only good for temporal trends, while BGMs give you the accurate moment-to-moment blood glucose value. He cited studies along the lines of DTS’ BGM Surveillance Program, showing that accuracy of many BGMs on the market is lacking considerably – many have higher MARDs than CGMs! That being said, he maintained that the best BGMs currently remain more accurate than the best CGMs, although CGM accuracy has been rapidly improving with each generation over the past 15 years (and the value of trend arrows, alarms, and historical data more than make up for slightly lower point accuracy). At the end of the day, Dr. Forlenza firmly believes that CGM is a better therapy choice for young patients with type 1 diabetes than BGM. Symposium: Tech and Teens

To Share or Not to Share, Do Teens Care?

Korey Hood, PhD (Stanford University, Stanford, CA)

Dr. Korey Hood presented the results of several behavioral surveys in which teens and young adults with type 1 diabetes reported the highest average A1cs, levels of diabetes distress, and number of barriers to using beneficial diabetes devices such as insulin pumps and CGMs; young adults also reported the lowest rates of CGM and insulin pump usage, perhaps contributing to (or more likely correlating with) their poorer outcomes. In a self-selecting survey (n=1,503), adults with type 1 diabetes who participated in the T1D Exchange clinic registry and had opted to provide an e-mail address to be contacted for research studies identified which of 19 barriers they experience using diabetes devices, their level of diabetes distress on a scale from one to six, and their attitudes towards diabetes technology and technology in general. The mean self-reported responses are tabulated below:

Age Group	A1c	Diabetes Distress	CGM Use	Pump Use	# of Barriers to Using Diabetes Devices
18-25 years old	8.21%	2.29	26%	64.1%	4.25
26-34 years old	7.44%	1.91	40%	69.4%	4.27
35-50 years old	7.37%	2.00	44.5%	77.3%	3.72
50+ years old	7.17%	1.89	48%	68.9%	3.12

• The primary modifiable barriers (not including cost or insurance coverage) to device use for all participants were the hassle of wearing devices all of the time (47%), dislike of having diabetes devices on their bodies (35%), and dislike of how diabetes devices look on their bodies (26%) – as outlined in the group’s excellent 2016 Diabetes Care paper. In order to increase diabetes device uptake in young adults, Dr. Hood proposed determining an individual’s device readiness based on their age, diabetes distress, and technology attitudes, to match each patient with personalized diabetes technology suggestions. In this model, a device profile would be generated for a patient based on quiz responses, and general suggestions, such as which device to use and how the patient should be educated, would be given. Dr. Hood gave even more context on the topic of barriers to device use and the psychology of diabetes at Friends for Life 2018. Dr. Tanenbaum also gave an excellent talk focused on clinician device-readiness profiles at ADA 2017.

• Dr. Hood emphasized that increased levels of diabetes distress seen in younger patients are potentially due to body image. According to the study, teenagers identified that not liking when people notice them wearing a CGM is a primary barrier to use. As such, Dr. Hood suggested rules of engagement with teens that have diabetes including calm discussion, active listening, and the establishment of windows when they do not have to share information.

• Individuals ages 35-50 and >50 years had the most positive attitudes toward general technology and diabetes technology, respectively. With younger generations often thought of as more tech-friendly, it was interesting to see that those aged 18-25 had the second lowest general technology attitude score (25.22) and the lowest diabetes technology attitude score (21.74). Those aged 35-50 had the highest general technology attitude score (25.65), and the oldest age group, those >50 years old, had the highest diabetes technology attitude score (22.24).

Corporate Symposium: FreeStyle Libre Flash Continuous Glucose Monitoring System – Clinical, Real-World, and Patient Perspectives (Sponsored by Abbott)

Insights from Real World Use of Flash Continuous Glucose Monitoring

Ramzi Ajjan, MD (University of Leeds, UK)

University of Leeds’ Dr. Ramzi Ajjan gave an early preview of a number of Abbott’s ADA poster haul (many of them late-breaking), plus the latest real-world data recorded from a remarkable 470,643 FreeStyle Libre readers (!) from September 2014 to April 2018! The real-world data sets, demonstrating improvements in glycemia with use of FreeStyle Libre, particularly with increased scanning frequency, always impress, but we were perhaps most blown away by the late-breaker (74-LB) that looked at the efficacy of FreeStyle Libre Pro.  

• Abbott’s FreeStyle Libre real-world data set has now reached 470,643 readers – 4.8 billion glucose readings – and the correlation between increased scanning and decreased hypoglycemia still holds! Estimated A1c ranges from 8.2% in those scanning ~4-6 times per day to ~6.9% for those scanning 20 times per day (and a bit lower for the ~5% of people scanning 20+ times per day). Time <54 mg/dl drops as scanning frequency increases, with an improvement of about 10 minutes per day from least scans to most scans. This is far and away the biggest set we’ve seen, as the company presented data from 250,000+ most recently at ATTD. As of April, there were 650,000+ FreeStyle Libre users globally, implying Abbott has data from an impressive 72% of users!

• The first real-world data for the US (n=7,979) suggest that the correlations between scanning frequency and reduced A1c/hypoglycemia hold true on American soil. By the looks of it (no statistical analysis), Americans may derive slightly greater benefit from more scanning beginning at ~15 scans/day. Dr. Ajjan said there are a few possible explanations for the differently-shaped curve plotting daily scans vs. time >54 mg/dl, though didn’t delve into any. It may have to do with better hypoglycemia management at baseline in the US, or simply an early adopter population in the US that is different from the average global user.

• (74-LB). Dr. Rich Bergenstal et al. conducted an analysis on sequential FreeStyle Libre Pro use in the US (n=2,836 users), using clever methodology: They looked at the results from a first FreeStyle Libre Pro, and then a second, when the first and second were separated by a mean of five months. The authors broke the study population up by baseline estimated A1c (eA1c) from the first sensor: One group had baseline eA1c >7.5%, and the other had baseline eA1c ≤7.5%. The table below summarizes the glycemic outcomes (all significant). Boiled down, individuals with baseline eA1cs/GMI >7.5% saw a 0.9% reduction in A1c, increased time in range (+2.5 hours/day), and increased (but not exorbitantly) time in hypoglycemia (+30 mins). For this group, the mean tradeoff is an additional 2.6 hours in range in exchange for an additional half hour below 70 mg/dl every day – depending on the patient and how deep that time in hypoglycemia goes (<54 mg/dl?), Libre Pro is a good proposition. (It could also be over-measuring lows, as the FDA label warns about.) On the flip side, those with baseline eA1cs/GMI ≤7.5% saw decreases in hypoglycemia (-30 mins), a decrease in time in range (-1 hour per day) and a slight increase in eA1c (+0.5%). The tradeoff here is ~30 minutes less per day in hypoglycemia, at the expense of ~two hours more >180 mg/dl – once again, this could be a good proposition for some patients, depending on their baseline magnitude of hypoglycemia and hyperglycemia though overall the tradeoff seems less worth it in the second case (time in range went the wrong direction, after all). All in all, however, we’d classify these results as encouraging, as they use a light Libre Pro intervention to drive strong time-in-range improvements in high-A1c users. For those with a low eA1c, hypoglycemia is reduced a bit, but the benefit seems fairly small in this study.

	Sensor 1 eA1c >7.5% (n=1,681)	Sensor 1 eA1c ≤7.5% (n=1,155)
Absolute Change in eA1c (Sensor 1 -> Sensor 2)	-0.9% (9.6% -> 8.7%)	+0.5% (6.4% -> 6.9%)
Absolute % Change in time in 70-180 mg/dl (Sensor 1 -> Sensor 2)	+11% (33% -> 44%)	-6% (73% -> 67%)
Absolute % Change in time <70 mg/dl (Sensor 1 -> Sensor 2)	+2% (2% -> 4%)	-2% (8% -> 6%)
Absolute % Change in time >180 mg/dl (Sensor 1 -> Sensor 2)	-13% (65% -> 52%)	+8% (19% -> 27%)

• (71-LB) This late-breaker shows that coefficient of variation – standard deviation divided by mean glucose, times 100 – decreases in near linear fashion as scanning frequency increases in a population of 237,747 real-world FreeStyle Libre users. Patients start deriving benefit from scanning, with respect to glycemic variability, at ~7-8 scans/day, and the benefit improves from there. Curiously, CV is high at 41% at the low-frequency scanning end of the spectrum, but only decreases to ~34% even in people looking at their glucose ~41 times per day. Since each blue dot represents 5% of the population, 95% of the patients in the below tracing appear to have CV >36%.  Monnier et al. consider CV <36% to be stable, and Dr. Irl Hirsch aims for <33%. From the Abbott 1Q18 call, we know that ~67% of global FreeStyle Libre users are type 1 – still, such high variability in the bulk of the population indicates the need for decision support and insulin automation.

• As a sidebar, we couldn’t reconcile another figure in the presentation depicting time in hypoglycemia and hyperglycemia broken out by % CV (≤36% and >36%) with the above graph. The graphs show that individuals with CV ≤36% spend 90 minutes less per day <70 mg/dl (54 vs. 144 min/day) and 2.5 hours less per day >180 mg/dl (7.5 vs. 10 hours/day). That makes sense – less hypo/hyperglycemia with lower glucose variability. However, the graphs claim that 33% of the users (n=78,597) have CV ≤36% – this doesn’t agree with the above graph, which points to just 5% of the population having CV ≤36% (n=~12,000). We’re not sure which numbers are correct, but either way, the data show the benefits of FreeStyle Libre and having a low glycemic variability.

• (70-LB) Dr. Ajjan presented a real-world analysis of hypoglycemia broken out by day vs. night. Corrected for length of time (since daytime is longer than nighttime), users spend a greater portion of nighttime ≤54 mg/dl. This is not shocking, as users can scan during the day and do something about lows, but at night they do not receive alarms with the current device. As expected, higher scanning frequency does translate to less time ≤54 mg/dl in both time frames.  

Use of Flash Glucose Monitoring and the Ambulatory Glucose Profile in Clinical Practice

Irl Hirsch, MD (University of Washington, Seattle, WA)

UW’s Dr. Hirsch emphasized that scanning frequency is paramount to success on FreeStyle Libre, sharing that he advises his own patients to scan at least 10 times/day. Based on real-world correlational data presented, 10 scans/day correlate roughly to an eA1c/GMI of 7.5%, ~30-minutes/day below 54 mg/dl (FreeStyle Libre has been noted to over-read hypoglycemia), and a CV of 40%. 10 scans per day doesn’t seem to be associated with ideal outcomes, though it’s worth noting that: (i) these correlations are based on the worldwide user base (which has different characteristics than Dr. Hirsch’s clinic); (ii) the hypoglycemia and glycemic variability data could be overestimates; and (iii) Kellee Miller et al. showed that 10 fingersticks per day – with no historical nor trend data – begets a ~7.2% A1c…we have to imagine those who scan 10 times per day in Dr. Hirsch’s clinic do better than the real-world correlations suggest, and we would love to see such a benchmark analysis! Why did he choose 10 times/day? Theoretically, data from his practice supported the recommendation. (In presentations to healthcare providers and patients, Adam has also noted that with 42 factors that affect blood sugar, staying in range on intensive insulin therapy really does need 10+ data points per day or if a patient can’t afford this, a very disciplined approach to diabetes management with little variation.)

• For Dr. Hirsch, one of the most important uses of CGM is understanding the meaning behind a patient’s A1c. To this end, he finds the ambulatory glucose profile (AGP) absolutely critical, while the additional glucometrics and individual daily details are also helpful. Through a series of illustrative case examples, Dr. Hirsch highlighted the limitations of A1c, demonstrating how a low value can often mask high levels of hypoglycemia; moreover, a low A1c may be artificially suppressed from lesser-known factors like the presence of mitral valve prolapse. Dr. Hirsch estimates that ~50% of the diabetes patients at UW clinics use CGM (notably, this is much higher than the 24% national uptake reported in the T1D Exchange and quite high given the poor payer geographic location he is in). He asserted that CGM is important for type 1 diabetes, type 2 diabetes, and atypical diabetes, characterizing the technology as the “emerging standard of care.” Hear, hear!

SELFY & IMPACT: Improved Clinical Outcomes in Teens and Young Adults

Kurt Midyett, MD (St. Luke’s Hospital, Overland Park, KS)

Dr. Kurt Midyett (St. Luke’s Hospital) presented compelling RCT sub-analyses in adolescents (13-17 year-olds; SELFY) and young-adults (18-24 year-olds; IMPACT), making a strong case for the efficacy of FreeStyle Libre in a traditionally tough-to-reach patient population. SELFY data in adolescents (n=24; mean age: 15 years-old) demonstrated time-in-range (70-180 mg/dl) to significantly improve from 10% to 11% (+1.2 hours) after eight weeks of FreeStyle Libre use. Teens also achieved a significant 0.7% A1c reduction (baseline: 8.3%), and decreased time >180 mg/dl from 13% to 11% (-1.7 hours). Importantly, these results were not at the expense of hypoglycemia, as no significant differences in time <55 mg/dl were observed (time <70 mg/dl was not provided). Dr. Midyett was most excited about the frequency of scans, which participants averaged nearly 10/day – at baseline, teens were performing fingersticks on average six times/day, showing a marked increase in engagement with their diabetes data. IMPACT data in young adults (n=11, mean age: 21 years-old) was similarly impressive, showing time-in-range (70-180 mg/dl) to increase from 51% to 63% (+2.9 hours) at six months of FreeStyle Libre use. Time <70 mg/dl dropped significantly from 11% to 9% (-0.5 hours) and participants averaged 11 scans/day. Dr. Midyett emphasized that adolescents and young adults feel under control with the FreeStyle Libre and find the device to fit well with their lifestyle – an “important psychological factor” when attempting to convince patients to wear CGM. Ultimately, Dr. Midyett is thrilled to see his teen and young adult patients, who had previously resisted managing their diabetes, using FreeStyle Libre to engage with their data and dramatically improve outcomes. FreeStyle Libre is not yet approved for people under 18-years-old in the US, though we might speculate the pediatric slant of the Abbott-sponsored session could be preparation for a near-term approval.

Patient Perspectives on Flash Glucose Monitoring

Kelly Close, MBA (Close Concerns, San Francisco, CA)

Our very own Ms. Kelly Close posed a simple question in Abbott’s FreeStyle Libre symposium: How can we make CGM the aspirin of diabetes? Download her slides here. Though attendees were initially boggled, she explained that aspirin is a widely-used, mass market product (50% of US adults took it in 2012), and it has a plethora of use cases (CV protection, pain, fever, cold, swelling, and even colorectal cancer prevention). When abstracted, those properties sound a lot like CGM, or at least a utopian vision of CGM. While CGM is not yet a mass-market product – Ms. Close estimated that over one million people wear CGM today globally, a fraction of all who could benefit from it – it could be, and it, too, has a plethora of uses. These applications range from real-time CGM, professional CGM, closed loop, using CGM with human coaching, using CGM with contextual text prompts to help the user identify the cause of glucose excursions, and even a mail-in professional CGM analysis service. The list went on, and then Ms. Close turned to the other variable: Time. She argued that the dichotomous view of CGM – real-time CGM worn 24/7 vs. professional CGM as much as a payer will cover – is an over-simplistic view and might be selling the technology short. She showed a neat plot depicting a slew of possible permutations of real-time and professional CGM – mixing and matching the two and using them at different times. Our favorite was “the holiday plan,” where real-time CGM is just used from October through December. Though reimbursement is improving (Libre is now reimbursed in at least 28 countries), any intervention less-intense than 24/7 real-time CGM would still be health-promoting, while reducing costs. [Look no further than an incredible Indian FreeStyle Libre Pro study from ADA day #1.] The answer to increasing CGM’s penetration and making it a mass-market product like aspirin, said Ms. Close, is to better understand patient heterogeneity and how each patient could best use CGM and be supported to derive the best outcomes. She argued that the field’s understanding of behavioral science – an enabler of effective CGM interventions –  is in its infancy, but improving every day, and that we have reached a “tipping point” in the technology as it grows smaller, cheaper, more discrete, and easier to use. Ms. Close’s talk left the audience with food for thought: The question is no longer, “should the patient use CGM?” but rather “how can we best match the patient to the type of CGM that is best for their care?” While she acknowledged that this question is idealistic, since so many patients are still unable to access even BGMs, she said it should nevertheless be what we strive for, since “everyone deserves to be using CGM (either real-time or intermittent).”

Corporate Symposium: The Rhythm of CGM (Sponsored by Dexcom)

Panel Discussion

Moderated by Dr. Max Gomez (CBS News Medical Correspondent, New York). Panelists: Lori Laffel, MD (Joslin Diabetes Center, Boston, MA); Nick Oliver, MD (Imperial College, London, UK); Davida Kruger (Henry Ford Health System, Detroit, MI); Patricia Gaye Knutsen (Barnes Jewish Hospital Diabetes Center, Saint Louis, MO).

This excellent panel provided attendees with a wide-ranging conversation on CGM, including reflections on its benefits, future, and barriers to use. All of the speakers predicted that CGM will become the standard of care over the next several years, that CGM will become smaller and easier to use, and that we will (and must) become more adept at using CGM data to benefit patients. On the reimbursement front, Ms. Davida Kruger happily stated that “overall coverage is good” with Medicare covering therapeutic CGM for intensive insulin users, as well as almost all private insurance – though of course, Medicaid still does not in 17 states. Drs. Lori Laffel and Nick Oliver discussed barriers to CGM use, including patients who prefer simplicity (though a multiple fingersticks per day sounds more complex than CGM!) and patients who get alarm fatigue. They emphasized the importance of positive first experiences with the technology as the key to long-term success and adherence. That encompasses anything from setting appropriate expectations, to making sure the timing is right (e.g., don’t start a patient on CGM right after a major life change), and setting lax alarm thresholds to start, especially for those who are worried or complained about them. All panelists were adamant that just about everyone could benefit from CGM usage, and that there are no absolute contraindications to its use. The panel was also in consensus about the improvements in quality of life that people with diabetes (and especially parents of children with diabetes) enjoy once they start using CGM, particularly with having less fear of nocturnal hypoglycemia (“parents can finally sleep again”) and less work performing numerous fingersticks. They also discussed how CGM empowers patients to more easily connect behaviors to glycemic outcomes, describing it as a “light going on” that inspires patients to improve their self-management. Dr. Oliver also noted that alerts provide immediately actionable prompts, which can be life-saving.

Quotable Quotes

• “The frontier now is not the device or how to use it. It’s what to do with the data.” – Ms. Kruger

• “Every technology and development before required more work from patients – pump changes, fingerstick calibrations, constant worry. But now for the first time, advances can take away some of the work of diabetes.” – Dr. Laffel

• “Once they’re on it, they can’t imagine what they’d do without it.” – Ms. Knutsen

• “Sometimes I feel like we’re cheating.” – Ms. Knutsen

• “First use is the best time to learn to do it well.” – Dr. Laffel

• “Real-time data is a behavior change agent. Retrospective data is not.” – Ms. Knutsen [Editor’s note: We agree that real-time CGM is highly preferred for driving behavior change, assuming access is possible. Retrospective review – identifying trends related to days of the week, times of the day, meals, etc. – can be valuable too as a behavior change tool, but can be harder to extract.]

• “I think that in the next few years the mounting evidence of benefits of CGM will make it the standard of care.” – Dr. Oliver

Diabetes Mine D-Data Exchange

FDA’s Dr. Courtney Lias on iCGM pathway, AID integration

Courtney Lias, PhD (FDA, Silver Springs, MD)

FDA’s Dr. Courtney Lias reviewed the exciting integrated CGM (iCGM) 510(k) pathway that came with Dexcom’s G6 clearance, including clarification on how a pump company will integrate multiple iCGMs – e.g., Tandem. Her comments echoed JDRF/HCT’s AID Interoperability Meeting, but they took on additional relevance after pre-ADA approval of Tandem’s t:slim X2/Basal-IQ as the first pump with iCGM compatibility (only Dexcom’s G6 for now). We’ve been wondering: if Senseonics obtains iCGM designation for Eversense, what is the process for Tandem integrating that sensor into Basal-IQ? Dr. Lias clarified that when an iCGM is cleared in the future, Tandem would evaluate if they want to claim use with it, perform whatever communication protocol/development work is needed, ensure its AID algorithm will work with the iCGM, and add the new iCGM to their labeling. Notably, the FDA will NOT have to get a new submission for this iCGM addition – Tandem can add new iCGM sensors to their pump without regulatory submission at all! In other words, it won’t quite be plug-and-play, as the onus is on the pump company to integrate the iCGM device and get the labeling – and in this case, Tandem would have to do some work, since G6 is a no-calibration system, Eversense requires two per day, and the sensor wear lengths are obviously 10 days vs. 90 days. (Presumably Senseonics and Tandem would have to work together in some capacity, but would not need a complicated contract and no coordinated PMA submission.) Dr. Lias emphasized that the iCGM path does not require factory calibration, and it does not mandate that a company connect with other systems – e.g., if an iCGM company wants to remain closed and doesn’t want to connect to another company, it doesn’t have to.

• Dr. Lias concisely summarized the benefits of the iCGM pathway, which aims to accelerate innovation, reduce regulatory hassle, and enable patient choice. Separating the iCGM from the AID or other compatible device system will make upgrades/modifications more efficient and reduce the need for duplicative regulatory submissions. It will also allow connected systems to be updated more quickly and with a predictable process. Notably, it works for many different business models, as companies can choose to be open or closed. Last, it will reduce the contract hassle that stalled many pump/CGM regulatory efforts in the past. For more complete comments, see our previous piece on iCGM here and Dr. Lias’ JDRF/HCT talk here.

• Related to the FDA De Novo clearance of DreaMed’s Advisor Pro software for optimizing pump settings, FDA has posted the Special Controls for the new category of “Insulin Therapy Adjustment Device” – read them here. They are much less specific than the iCGM special controls, mainly focused on properly documenting inputs and outputs, software validation and verification, etc.  

Standing Ovation for Dr. Anne Peters Access Work at D-Data Exchange

Anne Peters, MD (USC, Los Angeles, CA)

Dr. Anne Peters received a standing ovation for her tireless, two-year crusade against the State of California to obtain a CGM for a homeless type 1 with diabetes complications and significant hypoglycemia. The story reflects the tireless, relentless champion that Dr. Peters is. Notably, we learned that Dr. Peters and colleagues have produced guides to help low-literacy individuals learn about diabetes technology. At the end of her talk, she dramatically held up a huge shopping bag full of thumb drives with low-literacy manuals for diabetes tech; we’re excited to dig into them and will report back with more.

Symposium: Using Continuous Glucose Monitoring and Smart Devices to Control Glucose when It Matters Most

Intensive Glycemic Treatment during Pregnancy

Helen Murphy, MD (University of East Anglia, Norwich, United Kingdom)

Norwich Medical School’s Dr. Helen Murphy showed unpublished data from the CONCEPTT RCT of CGM in type 1 pregnancy (n=215), and commented: “We can categorically say that CGM is associated with small improvements in glycemic control, and a large reduction in neonatal complications. If you’re not convinced by those outcomes, we can argue to payers that the use of real-time CGM in pregnancy is likely to be cost-effective.” Yes! On the unpublished results front, Dr. Murphy presented a very functional data analysis, a more sophisticated statistical technique which allows researchers to parse out where glucose levels are significantly different between the control and intervention groups throughout the day. Overall, glucose is significantly lower in the CGM group ~4.5 hours per day, with marked reductions in the post-breakfast and the post-lunch periods (during sleep, the CGM group actually appears to run slightly higher). This analysis is illuminating, since Dr. Murphy and other experts believe hyperglycemia correlates better with fetal outcomes – 4.5 hours per day is quite meaningful. On the other end of the spectrum, Dr. Murphy noted that some people have been disappointed that CGM was “seemingly ineffective” in reducing hypoglycemia. As a counter, she pointed out that women in both arms actually had far less hypoglycemia than pregnant women in previous trials. In fact, to generate enough statistical power to show a blunting of hypoglycemia in the study, CONCEPTT would’ve had to enroll 1,000 women! Overall, Dr. Murphy was extremely positive on the study, particularly in that the results were robust across all 31 international clinical sites with varying degrees of technological expertise, and that the reductions in length of neonatal hospital stay and NICU admissions could likely pay for the full-term of real-time CGM. As a reminder, Medtronic’s older Guardian CGM drove a significant reduction in the incidence of large for gestational age, fewer NICU admissions lasting 24+ hours, fewer incidences of neonatal hypoglycemia, and one-day shorter length of hospital stay. These outcomes stemmed from what Dr. Murphy said “you might consider modest changes in maternal glycemia”: The primary A1c endpoint showed a small -0.2% A1c advantage for CGM at 34 weeks. However, mothers on CGM spent a significant 100 more minutes/day in range (68% vs. 61%), 72 fewer minutes/day in hyperglycemia (27% vs. 32%), and a non-significant ~14 fewer minutes per day in hypoglycemia (3% vs. 4%). We were surprised that Dr. Murphy didn’t comment on the reduced usability and accuracy of the older Medtronic CGM vs. newer versions on the market as a possible driver of less impressive glycemic outcomes. 

• Dr. Murphy announced a multicenter, pivotal RCT (AiDAPT – Automated insulin Delivery Among Pregnant women with Type 1 diabetes) in 124 pregnant women, set to begin in October. The 124 participants will be randomized to an automated insulin delivery group and a control group on standard insulin pump/pen therapy, and the primary outcome is a narrowly-defined time-in-target (70-140 mg/dl). Key secondary outcomes include obstetric and neonatal outcomes, patient-reported outcomes and interviews, and importantly, cost-effectiveness. The data follow up on feasibility study CLIP_03 (Closed loop in Pregnancy_03; n=16), which was published in NEJM in 2016, and the follow-up day and night study CLIP_04, which Dr. Murphy called “more impressive” than the very strong CLIP_03 results. In CLIP_04, day and night time-in-range (70-140 mg/dl) did not change significantly vs. SAP (62% vs. 60%), but hybrid closed loop did drive significant reductions in two levels of hypoglycemia and the number of hypoglycemia events. Specifically, time below ~63 mg/dl decreased from 2.7% to 1.6% (-16 minutes/day; p=0.02), time below ~50 mg/dl decreased from 0.5% to 0.2% (-4 minutes/day; p=0.03), and number of hypoglycemia events dropped from 12.5 on SAP to eight on closed loop (p=0.04). Additionally, 27 of the 32 women from the CLIP studies opted to retain their systems through labor and delivery. At a high level, 24 hours prior to delivery, time-in-range was 82%, time below target was 0%, there were no hypoglycemia events >20 minutes, and mean glucose was ~124 mg/dl. 48 hours post-delivery (1432-P at this year’s ADA), the women spent an impressive 83% time in-range, 2.5% in hypoglycemia, had 1.5 hypoglycemia events >20 minutes, and a mean glucose of ~130 mg/dl. There were no severe hypos. These are very strong results, and Dr. Murphy added that the women loved closed loop – “we struggled to get the systems back from them after they gave birth.” It’s heartbreaking from our view that they had to give the systems back – everyone with type 1 (in particular) deserves CGM as standard of care.

• On a slide labeled, “Limitations of HbA1c in Pregnancy”, Dr. Murphy proclaimed “we can, in type 1 pregnancy, move beyond A1c.” She showed a plot of A1c levels varying in a predictable manner over the course of gestation – a long, slow dip from week ~nine to week ~25, followed by a swifter uptick through week ~30. The only problem is that mean blood glucose within each A1c line was constant – at a constant mean glucose, A1c drops and then rises in a predictable and apparently consistent fashion. According to Dr. Bruce Buckingham, this is due to greater erythrocyte turnover as the fetus grows – the mother produces more and more blood cells, skewing the cell count younger and therefore with less time to accumulate glucose tags (and therefore show up on lab A1c). Plus, Dr. Murphy emphasized that there is “absolutely no difference in CGM accuracy during pregnancy, and that’s shown across a variety of sensors. 9%, 10% MARD is more than accurate enough. As a reminder, Abbott’s FreeStyle Libre, Dexcom’s G5/G6, and Medtronic’s Guardian Sensor 3 are not approved for use in pregnancy in the US, though notably, Abbott began promoting a pregnancy indication in 18 European countries last summer.

• We were intrigued and excited to hear Dr. Murphy frame many of the benefits of CGM and closed loop in pregnancy as equalizers to reduce center-to-center variability (“modifiable variation,” she called it) and improve outcomes on the population level. This is exactly the kind of thinking and framing that could lead to payer buy-in!

Closed-Loop Systems for Type 1 Diabetes in Youth—Children Are Not Just Small Adults

Jennifer Sherr (Yale University School of Medicine, New Haven, CT)

Yale’s Dr. Jennifer Sherr shared unpublished T1D Exchange data showing the A1c benefits of CGM in type 1 diabetes. In the youngest sub-group (<13 years-old), A1c is 0.9% lower in CGM users than in non-CGM users (7.8% vs. 8.7%). In adolescents to young adults (13-26 years old), A1c was 0.8% lower in CGM users than in non-users (8.3% vs. 9.1%). Finally, in adults over 26 years old, A1c was 0.5% lower in CGM users than in non-users (7.4% vs. 7.9%). This trend holds regardless of insulin delivery method (injections vs. pump) – a finding Dexcom has used for several years now to emphasize CGM works regardless of insulin delivery method. Of course, this data is correlational, and could reflect some degree of higher engagement, tech-savviness, or socioeconomic status in CGM users. Still, the data point to an encouraging trend, which confirms outcomes seen in a plethora of CGM studies. See our report of Dr. Roy Beck’s talk at the T1D Exchange 2017 meeting for trends of CGM use in the Exchange. The most important T1D Exchange metric, in our view, is trended CGM penetration – the update from last fall put it at 24%, though we’d expect that to grow with four great CGMs in the US in 2018.

• Dr. Sherr also showed that use of pumps has increased from 57% to 63% since 2010. Of course, this is the T1D Exchange, so broader use in type 1 is lower. This increase has been largest in individuals under six years-old, but is relatively constant across age groups. We hope to see CGM penetration this high in five years, especially as more pumps build in automation and CGMs continue to improve in form factor, connectivity, and cost.

Symposium: Diabetes Is Primary

What Every PCP Needs to Know about Diabetes Technology in 2018 and Beyond

James Chamberlain, MD (St. Mark’s Hospital & Diabetes Center, Millcreek, UT)

In a very practical session aimed towards educating primary care providers on current diabetes tech offerings, Dr. Jim Chamberlain (St. Mark’s Diabetes Center) emphasized that “CGM is here to stay” and has “taken over the type 1 diabetes world.” In fact, Dr. Chamberlain opined that CGM will likely replace BGM “completely” in the next five years, especially for those on intensive insulin therapy. While we firmly believe CGM is the new standard of care and are elated to see uptake increase in those with type 1 and type 2 diabetes, things will need to accelerate quite precipitously in the next five years to completely replace BGM – even in the US. T1D Exchange CGM penetration stood at 24% as of data last fall at the best US type 1 centers, though with Libre and Guardian Sensor 3 CGM penetration is likely higher now. Dr. George Grunberger recently estimated at AACE that ~95% of people with diabetes are not on CGM, and expansion into developing nations will not be easy. To this end, we’re hopeful to see more patients at least using connected BGM so as to facilitate data download. Dr. Chamberlain provided the example of the One Touch Verio Flex and One Touch Reveal App. When analyzing data, Dr. Chamberlain advised attendees to consider not only the average blood glucose and A1c, but also variability and time-in-range, denoting a “good” standard deviation as one-third to one-half of the average blood glucose (Dr. Irl Hirsch frequently cites one-third as the desired upper limit, in line with Monnier’s work on a CV <36%.) He briefly emphasized the limitations of A1c, made especially clear by CGM data. Dr. Chamberlain noted that he has a “handful of patients” using Companion Medical’s InPen, which he jokingly characterized as “a poor man’s pump” (this is actually a big positive, considering InPen’s ability to deliver a pump’s benefits in a lower-cost package). He was especially excited about InPen’s ability to track insulin on board, as well as the 0.5 unit dose increments. We can’t wait for more smart pens and caps to hit the market and scale, and are pleased to hear that providers are similarly enthusiastic. Dr. Chamberlain briefly detailed the now two (!) automated insulin delivery systems approved in the US – Medtronic’s 670G (now approved in those 7+ years-old) and Tandem’s t:slim X2 pump with Basal-IQ and Dexcom’s G6 iCGM (approved in those 6+ years-old). Dr. Chamberlain mentioned that the 670G requires “quite a bit of training” and acknowledged the practice of adding “fake carbs” as “fine, it’s where we’re at right now.” He also noted that he has two patients on DIY systems, admitting that they are “doing pretty well.” We were glad to see a session dedicated to informing primary care providers on diabetes technology – given that the majority of people with diabetes do not receive their care from endocrinologists or diabetologists, many thought leaders advocate for better incorporation of technology in the primary care setting. 

• Dr. Chamberlain, who has type 1 diabetes himself, has worn Dexcom’s G6 and Abbott’s FreeStyle Libre. He had to calibrate the G6 a few times during the first couple of days and encouraged providers to warn their patients that some initial calibration may be necessary. He found the Libre “very comfortable” and emphasized that patients must be informed that the system lacks alarms. Read diaTribe’s test drive of Dexcom’s G6 here.

Symposium: Should All Pregnant Women with Type 1 Diabetes Use Continuous Glucose Monitoring from Planning Pregnancy until after Breastfeeding?

Pro and Con

Denice Feig, MD (Mt. Sinai Hospital, New York, NY), Elisabeth Mathiesen, MD (Center for Pregnant Women with Diabetes, Denmark)

In an equal parts humorous and insightful debate, University of Toronto’s Dr. Denice Feig and the Center for Pregnant Women with Diabetes’ Dr. Elisabeth Mathiesen went head-to-head on whether all pregnant women with type 1 diabetes should use CGM from planning pregnancy until breastfeeding. Dr. Feig took the pro-argument, commenting to a round of applause: “Maybe this topic shouldn’t be debated because the evidence is clear.” Dr. Feig focused mainly on the compelling CONCEPTT data, showing CGM use in pregnant women to significantly increase time in-range (+100 minutes/day) and decrease time in hyperglycemia (-72 minutes/day). She acknowledged that 80% of participants reported problems with the sensor, although the study used Medtronic’s older Guardian CGM, and she maintained that more recent RCTs outside of pregnancy have demonstrated improved use and satisfaction with use with other, newer CGM systems. Though not the primary endpoint of CONCEPTT, the significantly improved neonatal outcomes were the headline: a 50% reduction in large-for-gestational-age (LGA) newborns, as well as reductions in hypoglycemia requiring IV dextrose and NICU admittances lasting >24 hours. To this end, Dr. Feig believes CGM is likely to prove cost-effective, which her team is currently investigating. She noted that nine months of CGM use in the UK costs ~£1,500, the same as the expenses associated with just one day spent in the NICU. With more advanced CGMs not requiring fingersticks, she pointed out, CGM costs will only decline. Moreover, she asked the audience: “What price would you place on your baby to avoid a prolonged stay in the NICU?” Needless to say, we and many others were moved. As her final argument, Dr. Feig provided a quote from a CONCEPTT participant, which really does say it all: “It’s like being completely blind and then having somebody open your eyes… Managing my sugars without CGM is like driving a car blind-folded.”

• Dr. Feig also made a strong case for CGM use postpartum and during breast-feeding. During the postpartum stage, women require ~54% of their pre-pregnancy insulin dose, while at the same time experiencing marked glucose variability. While breast-feeding, women have to intake more carbs to prevent hypoglycemia. For all these reasons, Dr. Feig asserted that CGM should be maintained through breast-feeding. As for those planning pregnancy, Dr. Feig acknowledged that the CONCEPTT arm investigating CGM use in women planning pregnancy did not find the A1c reduction to be significant; however, there was a clear positive trend, and she believes this arm was underpowered. Ultimately, she argued that there is an enormous amount of literature supporting the use of CGM in non-pregnant patients, so why not provide the added benefit during pregnancy planning.

• Dr. Mathiesen presented a counter-argument to CGM use in pregnancy. As her main evidence, she cited a 2013 study (n=154; Secher et al.) evaluating intermittent, real-time CGM in women with type 1 and type 2 diabetes during pregnancy. Participants were randomized to wear CGM for six days at a time at weeks 8, 12, 21, and 33 of pregnancy in addition to routine care, which included seven fingersticks/day, or to routine care only. A1c and hypoglycemia events were comparable in both groups, with a tendency towards greater rates of LGA and preterm delivery in the CGM group. In fact, when Dr. Mathiesen examined data from the per protocol users of CGM exclusively (as there were high rates of poor adherence), there were even higher rates of LGA among CGM users. However, Dr. Feig addressed this study during her presentation, reminding attendees that intermittent use prevents reaping the full benefits of CGM data, and only 60% of participants followed the protocol, leading to a small, underpowered sample size. Moreover, she finds A1c to be a “lousy measure of glycemia during pregnancy,” preferring time-in-range, as it more accurately captures “what the fetus is facing.” Dr. Mathiesen also noted that she was shocked to see that very few participants in the study wanted to remain on the CGM. We’re not too surprised though, given that participants still had to perform seven fingersticks/day and only wore CGM for six days at a time somewhat sporadically.

• Dr. Mathiesen also focused on the cost of CGM, pointing out that the expenses associated with CGM use in 20 women are equivalent to the salary of one nurse. She also asserted that there are extra costs due to increased doctor/nurse time – however, as one audience member noted, CGM tends to massively reduce provider time, as blood glucose readings are easily accessible and can be more quickly evaluated. Another of Dr. Mathiesen’s concerns was that CGM readers do not show glucose values from the last three to seven days, which are necessary when looking to change a pregnant patient’s insulin dose. She remarked that to view such data, patients have to upload readings to a “computer program” and one would “have to be a nerd to do that.” This sounds like a strawman to us – to view seven-day retrospective data from most BGMs, a “computer program” would also be necessary. She also advocated for more real-world data and more than one RCT before making such a costly investment. However, as Dr. Feig opined during her rebuttal, “why wait for real-world data when there is such a good and definitive RCT? Our patients deserve it.” Hear, hear!

Selected Questions & Answers

Comment: That was awesome, but it didn’t change my mind. Most of my patients who are pregnant with diabetes are using CGM in one form or another and I thought it was interesting that you [Dr. Mathiesen] said it adds time to visits. In my clinic, it saves me incredible time. It probably saves five minutes.

Comment: I’ve found CGM and mainly flash CGM to be so instructive to my patients in pregnancy for the timing of insulin, modifying diet … it’s almost been no-brainer. My real-life experience is people with CGM just want to continue.

Q: What about CGM use in type 2 diabetes during pregnancy?

Dr. Denice Feig: I certainly think that it’s possible all patients will benefit but we might need very large numbers to show that. It’s certainly worth looking into.

Dr. Elisabeth Mathiesen: It has to be cost-effective. The prices of sensors have to come down and CGMs have to be simpler to use.

Dr. Feig: Helen Murphy had an uneducated patient who could hardly read or write and she just followed the CGM arrows and did very well.

Q: If you couldn’t give sensors to all of women during pregnancy, who would you give it to?

Dr. Mathiesen: I’d give to women with the highest risk of hypoglycemia and to the very poorly regulated patients.

Dr. Feig: If I couldn’t use it for everyone, I’d do the same. But I’d prefer to give it to everyone.

Q: Do you think there is a place for using CGM in gestational diabetes?

Dr. Mathiesen: Save the economy, go for type 1 diabetes first.

Dr. Feig: I would agree, but I think there’s still a place for research. We’re assuming they won’t necessarily benefit. Type 1 diabetes would be my priority but I wouldn’t rule it out for GDM.

Product Theaters

Introducing the New Dexcom G6—Simple, Accurate, Effective (Presented by Dexcom)

Keri Leone, MS and Jake Leach (Dexcom, Inc., San Diego, CA)

Dexcom’s standing-room-only G6 Product Theater provided details on how Dexcom factory calibrates the sensor and a look at the two main pivotal G6 studies – now published in DT&T here (n=262, main pivotal study, manual applicator) and here (n=62 auto-applicator study). The latter was interesting, as Dexcom is marketing a G6 overall MARD of 9.0% – including 9.8% in adults and a strikingly low 7.7% in pediatrics – based on the results of the smaller automatic applicator study (n=62). In the actual G6 user guide, table 1-A on accuracy pools the manual applicator (n=262) and automatic applicator (n=62) studies together (n=324), arriving at a slightly higher MARD of 9.8% overall, 9.9% in adults, and 9.6%-9.9% in pediatrics. Dexcom is marketing the auto applicator data (table 10), as it is more reflective of the commercial product and probably a “truer” reflection of the accuracy users would see. This makes perfect sense, though it is odd to us that the data supporting this main marketing claim is buried in table 10 (while the pooled data in table 1-A is not a focus). Some of the DT&T publications’ stats – including %20/20 and %15/15 – did not line up exactly with the G6 product label. Is this due to FDA analysis or some statistical difference (e.g., median vs. mean)? For example, the FDA label says 93.5% of points are within 20/20 in the applicator study, while the publication shares a higher 93.9% - that’s a small difference but strange nonetheless and we’re not sure which to use... In any case, G6 is obviously a highly accurate factory-calibrated CGM, now codified in published studies and verified in our real-world experience so far (see diaTribe test drive). We include pictures from the talk below alongside the user guide tables.

• All G6 sensors have their individual performance assessed in the factory. Each sensor has a set of specific characteristics, and Dexcom looks to bundle these characteristics and assign a specific sensor code to each bundle. The information is converted to a four-digit sensor code printed on the bottom of each sensor. Mr. Leach showed an example histogram of sensors (see below), with different edges of the distribution getting different sensor codes. He noted the sensor’s manufacturing already results in a tight distribution with not much variability between sensors, but the codes ensure better sensor-to-sensor performance and accuracy and less outlier sensor performance. Since diaTribe’s test drive and our initial Closer Look piece, the photo scan of the G6 sensor code is working flawlessly and performance continues to be very strong.

   

• “The first truly zero fingerstick CGM system that meets the FDA’s most recent iCGM standard.” VP Jake Leach emphasized the iCGM standard a few times, highlighting the Thursday’s FDA-approved Tandem t:slim X2 with Basal-IQ/G6 and how iCGM will “accelerate innovation and reduce regulatory burden for integrated systems. All of our partners in the insulin space are able to take advantage of these new standards.”

• Mr. Leach showed a slide with the Verily gen one sensor – confirming 14-day-wear, factory calibration plans, and a single-use transmitter. No launch timing was shared, and interestingly, the penny-sized gen two sensor was not shown – usually it appears on the same slide. We’re not sure if this is worth reading into, or simply caution in front of the HCP audience.

Introducing a New Era in Glucose Monitoring—The Eversense Continuous Glucose Monitoring System (Presented by Senseonics)

Katherine Tweden, PhD (VP Clinical Sciences, Senseonics), Steve Edelman, MD (UCSD, San Diego, CA), Tim Bailey, MD (AMCR Institute, San Diego, CA), and Tobias Schulte (Eversense User, Germany)

In its first-ever US product theater, Senseonics drew a standing-room-only crowd (three-deep!) to learn about the freshly approved 90-day Eversense implantable CGM, on-body transmitter, and Android/iOS/Apple Watch apps. As we noted just before ADA, Senseonics has FDA approval with an adjunctive label, two fingerstick calibrations per day, 90-day wear, and for 18+ year-olds. Features mentioned in this product theater that we did not emphasize: the mobile app is approved for both Apple iOS/Watch and Android (a big win for Senseonics and underemphasized in the marketing); the transmitter lasts one year, takes only ~5 minutes to charge (e.g., while showering), lasts a little more than a day on a full charge, and battery status can be viewed on the phone; the Eversense Now app for remote monitoring is also approved for sharing data; and sensor lag time is under eight minutes. Following insertion, the sensor does have a 24-hour warmup (no data), something we mentioned at the March Advisory Committee but did not mention in our FDA approval write-up; given the 90-day wear time, this is only relevant four days a year. A highlight of the product theater was hearing from Tobias Schulte, an Eversense user from Germany on his sixth sensor (three sensors inserted in each arm). Eversense has had a hugely positive impact on his life as an endurance athlete – he’s run 89 marathons! Tobias cited the on-body vibration alerts, the discretion (“No one realize any more that you have diabetes”), the ability to take the transmitter off and “have a proper shower…completely free of equipment,” the no-irritation sensor adhesive (he’s worn over 400 and never noticed any redness or side effects). Most importantly, “The perceived difference between a marathon before the diagnosis and today [with Eversense] has become quite small.” As we heard at the March Advisory Committee, Drs. Steve Edelman and Tim Bailey did a great job framing the need for additional CGM options, noting only ~30% of type 1s on CGM in the US – this was cited as T1D Exchange but not via an exact stat (3/10 colored stick figures). (To us, 30% is likely an overestimate of general US type 1 population’s CGM usage, but that could change this year with four (!) accurate CGMs available.)

The Guardian Connect Continuous Glucose Monitoring System—Utilizing Predictive Alerts and Data Backed Insights to Help with Diabetes Management (Presented by Medtronic)

Huzefa Neemuchwala, PhD, MBA (Medtronic Diabetes, Northridge, CA) and Tim Bailey, MD (AMCR Institute, San Diego, CA)

Medtronic’s Guardian Connect CGM product theater focused mostly on the paired Sugar.IQ app’s insights and next-gen pipeline – including a look at future plans for healthcare provider insights and an MDI dosing assistant powered by IBM Watson. The dosing insight noted very specifically, “Your next dose is going to be 2 units at 5:50 PM. Input carb: 80G. insulin type: Short-acting.” Said Medtronic Head of Digital Health Solutions/AI Dr. Huzefa Neemuchwala, “We were first to bring hybrid closed loop to market, and we plan to be the first to bring closed Loop to MDI users.” Now that’s a line in the sand! This would of course provide a Medtronic answer to what Bigfoot and Lilly (among others) are doing to close the loop for people on injections with apps, CGM, and smart pens. (Medtronic has not announced what its hardware plans on the smart pen front.) It also aligns with Medtronic’s new business unit, MDI Solutions, led by Laura Stoltenberg. No timing was shared on the dosing assistant, though we know from the early June Analyst Meeting that it is targeted to launch within two years (by April 2020). Meanwhile, our first-look at Sugar.IQ-driven healthcare provider insights looked fantastic. Near-term, a provider dashboard would show similar insights as patients will see on Sugar.IQ (i.e., unique days, unique repeated habits, meal responses, rapid glucose changes, bolus habits). However, next-generation HCP insights will give clinical decision support driven by IBM Watson, advice that a provider can simply “Deliver” to the patient from their dashboard or “Skip”– e.g., “increase insulin sensitivity factor from 2 to 2.2” or “Increase Bolus Wizard target glucose value from 90 to 95 mg/dl.” Nice! (No timing shared, but this could be the “intelligent therapy recommendations” expected to launch within two years.) Dr. Neemuchwala also mentioned the “Inner Circle” CGM gamification/engagement app a few times in quick passing, but didn’t share more details or specific timing; as of the Analyst Meeting, this is expected to launch within the next year (by April 2019), allowing users to redeem points for more time-in-range. Dr. Neemuchwala shared two other notes on Sugar.IQ: (i) the company’s goal to push a new version of Sugar.IQ every three months; and (ii) plans for Sugar.IQ to fit into its larger strategy on value-based diabetes care. Otherwise, most of the product theater was review from Friday’s oral presentation, what we saw at ATTD on Guardian Connect, and things we’ve previously seen on Guardian Connect/Sugar.IQ – see our report on the launch from a few weeks ago.

Digital Health and Connected Care

Oral Presentations: Nutrition 2018—New Data, New Perspectives

Continuous Remote Care Model Utilizing Nutritional Ketosis Improves Type 2 Diabetes Risk Factors in Patients with Prediabetes

Amy McKenzie, PhD (Virta Health, San Francisco, CA)

Virta’s Dr. Amy McKenzie added to the evidence on Virta Health’s continuous remote care + ketogenic diet intervention, sharing mightily impressive one-year data in the prediabetes population (n=116) for the first time: Of the 95 completers (82% retention), 61% (n=58) moved from prediabetes to normoglycemia (A1c <5.7%), and not a single person (!) progressed to type 2 diabetes. Virta’s intervention is a low-carb/high-fat diet (to induce nutritional ketosis) combined with tech-enabled remote care, and the study was non-randomized, un-controlled, and conducted in collaboration with Indiana University Health Arnett. Drilling deeper, the other outcomes from this study looked very strong (see below for corresponding photos):

• Body weight dropped an impressive average of 29 lbs (11.5%) from baseline 111 kg (~245 lbs). 70% of completers lost >7% of body weight; only ~4 individuals gained weight, and one lost ~40% of starting weight! It is truly remarkable to see the inter-individual weight variability in the graph below – this suggests a very wide response to the ketogenic diet. Of course, on average it produces great results, and we’re just as interested in why some people respond vs. not.

• A1c decreased by 0.3% from a very low baseline (5.9%). (i) 80% of completers decreased A1c, and 12% saw increases; (ii) 61% of completers regressed to normoglycemia (A1c <5.7%); (iii) no completer progressed to type 2 diabetes (A1c ≥6.5%). We’re surprised to see some A1c increases – was this from not following the program or from a biological response?

• Fasting plasma glucose (FPG) dropped from 109 to 100 mg/dl, 51% of completers had FPG drop below 100 mg/dl at one year.

• Systolic blood pressure (SBP), diastolic blood pressure (DBP), triglycerides (TG), and HDL-C all significantly improved at one year. SBP dropped from 130 to 124 mmHg; DBP dropped from 83 to 80 mmHg; TG dropped from 150 to 107 mg/dl; and HDL-C increased from 52 to 58 mg/dl. LDL-C was not reported. In the type 2 study, there was a mean ~10% increase in total LDL-C, though the more nuanced LDL partitioning (e.g., particle number and their relative sizes) changed in a positive direction – and some studies suggest this is a more sensitive CVD risk factor than LDL-C alone.

• These prediabetes management/diabetes prevention results go along nicely with evidence from Virta’s controlled study in type 2 diabetes, which has demonstrated ~60% diabetes reversal (defined as an A1c <6.5% and elimination of all medications except metformin) at one year. Taken together, an over-simplified calculation suggests that Virta may be able to bring 60% of type 2 enrollees into prediabetes at one year, and then bring ~60% of those individuals (36% of the initial population) into the normoglycemic range – for an employer, a two-year intervention that takes 36 out of 100 people off of their diabetes medications, makes them feel better and more productive, and lowers their risk of complications seems like a very solid proposition! We’re not sure that Virta will ever offer a “spectrum-like” continuous program as we just laid out, if they will aim to deliver separate diabetes remission and DPP, or something else – in fact, the company’s stated goal to reverse type 2 diabetes in 100 million people by 2025 says nothing about plans to address prediabetes (though we would love to see more companies with unique approaches enter the space!). Notably, both type 2 and prediabetes studies will again report results at two years and five years the type 2 study is already >2.5-years in, and we assume the prediabetes study is as well.

• Only 56% of the completers reached a mean blood beta-hyodroxybutyrate concentration of ≥0.5% mmol/L over the year of treatment (the goal of Virta’s ketosis-inducing diet), meaning 44% did not. Virta has not yet examined the outcomes between the two subgroups, but we think parsing them would be extremely informative. One sentiment regarding Virta goes something like, “I’m not sure about the ketogenic diet, but I’m all in on the remote continuous care.” This was brought up today, as a Pennington Biomedical researcher asked during Q&A if patients would see the same outcomes on a Mediterranean or low-fat diet if they were interacting with coaches an average of three-times per day (as they did in this study). We would at least like to see Virta conduct a study where it positions remote continuous care without a one-track diet recommendation as a control group. Given our own anecdotal experience with low-carb, high-fat, it is extremely effective, assuming someone knows how to navigate situations and titrate medications – hence the remote care.

• Throughout the talk, Dr. McKenzie put Virta’s data into the context of typical, CDC-identified diabetes progression and the results of the landmark DPP study. At a high level, every year, CDC expects 5-10% of people with prediabetes to progress to diabetes, and another 5-10% to regress to normoglycemia; Virta’s data blows this population-wide trend out of the water, with 61% regression and 0% progression. In the DPP study, participants who received the lifestyle intervention lost 7% body weight, and those who received metformin lost 3% body weight at one year – Virta’s program led to 11.5% body weight loss (notes: people in the DPP started at a lower average weight of ~207 lbs; and Virta’s intervention is hands on for a year, while the DPP lasts 24 weeks). In the DPP, the risk of progressing to type 2 diabetes was 56% lower for those that regressed to type 2 diabetes; 61% reverted to normoglycemia in this study.

Questions and Answers

Q: I imagine these people restricted their carbs…

A: When they started the study, we recommend they consume <30 grams of carbs per day. They’re mostly getting their carbs from starchy vegetables.

Q: Outstanding data. Virta has the ketogenic diet, and continuous remote care. I’ve seen data presented before, that the average daily connection for patients and remote coaches is three times per day, maybe more than that. That’s a lot. What if you did the Mediterranean diet or a low-fat diet when you have that much interaction.

A: Great question. I have no idea. And that’s exactly what we have, it’s an average of ~three interactions with coaches per day. In terms of glucose and A1c reduction, the interaction component is definitely very interesting.  

Brian Levine (Close Concerns): 44% of the group didn’t have beta-hydroxybutyrate levels ≥0.5 mmol/L – what were their outcomes?

A: Mean beta-hydroxybutyrate, over the course of a year. No, we didn’t look into that, but that’s an excellent question.

Acute Insulin Secretory Effects of a Classic Ketogenic Meal in Healthy Subjects

Simona Bertoli, MD, PhD (University of Milan, Italy)

A brief oral presentation demonstrated that a classic ketogenic meal (2% carbs, 88% lipids, 10% protein) requires just 12% as much insulin as a classic Mediterranean diet meal (60% carbs, 30% lipids, 10% protein). Wow! In people without diabetes, the Mediterranean meal caused glucose to increase from 90-110 mg/dl in the first hour, and then decrease to and below baseline for the remaining three hours (measured by frequent SMBG); with the ketogenic meal, glucose remains stable for a half hour and then drops in the second hour post-meal. To cope with the Mediterranean meal, insulin shoots up 12-fold in the first 20 minutes post meal, and then remains elevated – with a ketogenic meal, insulin levels increases 2-fold in the first 20 minutes and stays there. C-peptide levels follow similar trends. Overall, total insulin secreted in the three post-meal hours was 8x greater in the Mediterranean group than the Ketogenic group. University of Milan’s Dr. Simona Bertoli, the presenter, posed future research questions: (i) What is the effect of significantly lower circulating levels of insulin in infants, children, and adults? (ii) Could ketogenesis be effective for patients with insulin resistance and/or secretory defects? See the highlight on Virta’s prediabetes data above for more on the ketogenic diet coupled with remote care.

Symposium: Online and Mobile Support – Wading through the Noise

Can Diabetes Apps Make Our Lives Easier?

Adam Brown, BS (Close Concerns, San Francisco, CA)

Our own Adam Brown answered the question, “Can Diabetes App Make Our Lives Easier?” in a tour-de-force, state-of-the-field talk – view the slides at diaTribe.org/Apps2018 or download the PDF here. The talk aligned on two key features Adam believes will characterize truly useful apps: (i) radically convenient, frictionless tools; and (ii) compelling decision support. Adam then proceeded to call out 40 different examples of apps with exciting prospects – either currently out or in development. Here’s a summary of the three areas he discussed:

• Are people using diabetes apps now? Which ones? Adam began by sharing fascinating dQ&A panel data on use of apps to manage diabetes right now – weekly diabetes app use is variable but generally low, ranging from 13% in type 2 diabetes (n=2,776) to 29% in in type 1 adults (n=1,592) to 56% in parents/type 1 kids (n=135). The most popular apps in the dQ&A panel include Dexcom G5, Share, and Clarity (CGM); One Touch Reveal, Contour Diabetes, and Accu-Chek Connect (BGM); mySugr, One Drop, Glooko, and Glucose Buddy (Data); Apple Watch Activity, Apple Health, and Fitbit (Exercise); myfitnesspal and Calorie Kind (Food); and pharmacy apps (e.g., Walgreens, CVS).

• What makes an app useful? In part two, Adam stepped back from diabetes and considered what makes any consumer app useful – a framework to define how a diabetes app could make life easier. Drawing analogies to transportation, music and movies, and shopping and mobile payments, Adam called out Lyft and Uber; Netflix, Spotify, Pandora; and Amazon, Square Cash, and Venmo as examples of radically convenient, frictionless tools – apps that replace physical objects and high-hassle processes (cars, taxis, parking; video store rentals, home DVD/CD collections, per-song purchases; in-person shopping and cash/ATMs), save time through a dramatically better experience, and save money. The second category of useful consumer apps, Adam argued, offer compelling decision support – Google Maps and Waze. These bring personalized, real-time, responsive, better-than-a-human help in making important decisions.

• Moving to the talk’s main title – Can diabetes apps make our lives easier? – Adam provided 40 examples of radically convenient tools and/or compelling decision support. The tables below cover the examples in this last third of the talk. Download the slides at diaTribe.org/Apps2018 for pictures and the whole talk’s flow.

Radical Convenient Tools – App Examples	Why useful?
Mobile CGM: Dexcom G5/G6, Guardian Connect, FreeStyle LibreLink, Senseonics Eversense.	No CGM receiver: replace physical objects Automatic, frictionless CGM data upload to cloud
Data apps: Glooko and Tidepool	Replace high-hassle data downloading with frictionless web apps (J clinics!) – less confusing cable snags, no single-device software platforms.
Connected BGM: One Touch Reveal, Accu-Chek Connect and mySugr, Contour Diabetes, One Drop.	Replace high-hassle cable downloading with automatic upload and analysis – Notably, Adam showed connected meter penetration in the dQ&A panel is up 6.5x between 2016-2018 – now at 10% of the panel, and there’s been a particularly big increase in the past quarter. Is it reaching a “tipping point?”
Frictionless data insights: Dexcom Clarity mobile notifications	Weekly emails/notifications on patterns/CGM data are automated, with no need to log into a web portal. Insights/patterns are pushed to user, changing paradigm of diabetes data.
Bluetooth-enabled insulin pumps/apps are coming: Insulet Omnipod Dash, Tandem t:slim X2, Medtronic MiniMed 670G.