July 13-16, 2017; Keystone, CO; Days #3-4 Highlights – Draft

Executive Highlights

Ciao, Keystone! Our team has descended from the nearly two-mile-high city back to sea level in San Francisco, where we put the finishing touches on our third and final highlights report from this year’s ATDC conference. Days #3-4 of Keystone 2017 featured a keynote lecture by FDA’s Dr. Courtney Lias on the roadmap for interoperability, three Dr. Bruce Buckingham appearances, and exciting updates from Dexcom on Medicare shipments and its strong pipeline. The meeting wrapped-up with one last plenary on day #4, all about the latest and greatest therapeutic tools for diabetes management. We heard from Dr. Simon Heller on next-gen basal insulins, from Dr. John Buse on SGLT-1/2 dual inhibitor sotagliflozin for type 1, and from Dr. Lori Sussel on rescuing residual beta cell function in type 1. You’ll find our top takeaways from days #3-4 below, divided into 10 diabetes technology highlights and six diabetes therapy highlights, plus additional coverage in the detailed discussion and commentary section of this report.

Wow, that was a lot of learning! We’ve done our best to capture all the major insights, opinions, and debates of Keystone 2017 in our highlights coverage from day #1, day #2, and days #3-4. Stay tuned for our full report, and mark your calendars for next year’s 28th annual ATDC conference back in Keystone, CO from July 12-15, 2018!

Diabetes Technology Highlights

1. Dexcom Senior Vice President of Global Access Dr. Claudia Graham gave the update we’ve all been waiting for: It has taken a while to implement, but Medicare shipments of G5 are up and running! However, it is slow at the beginning and the company is proceeding carefully. According to Dr. Graham, it will take Dexcom “several months” to get through the “thousands and thousands and thousands of patients in the queue.” Dr. Graham also confirmed that Share (remote monitoring) functionality will initially be disabled with the Medicare G5 system, to comply with CMS requirements – this is an absolutely insane CMS policy that CEO Kevin Sayer hopes to change within a year.

2. In a Dexcom pipeline update, VP of Advanced Technology Peter Simpson and CEO Mr. Kevin Sayer gave an update on G6 (pivotal trial wrapping up, FDA submission by the end of this September), first- and second-gen verily products (no timing changes), and the new Arizona factory’s impressive capacity (50 million sensors/year!).

3. Jaeb Center’s Dr. Roy Beck reviewed the plentiful evidence for CGM benefit in both type 1 and type 2 diabetes, concluding that CGM yields better glycemic outcomes than SMBG, and with the latest technology, is accurate enough to replace fingersticks. He also introduced upcoming studies of CGM in type 2 patients on basal insulin (MOBILE), in children under eight years old (SENCE), in adolescents (CITY), and in the elderly (WISDM).

4. FDA’s Dr. Courtney Lias proclaimed with an emphatic “Spoiler Alert!” that clinical trials are not chiefly responsible for slowing the commercialization of automated insulin delivery systems; rather, a lack of a framework for interoperability, and all that it entails, has hampered the field. We’re glad to see her continuing to hammer this message home. P.S. The team that handles BGM, CGM, sensor-enabled pumps, and AID devices at FDA? Comprised of just 10 people!!

5. BDC’s Dr. Gregory Forlenza presented un-published data from MiniMed 670G use in 10 of his pediatric patients in the continuation phase and at an ADA diabetes camp, as well as a look at how the system was optimized over time in the pivotal trial.

6. Dr. Bruce Buckingham prefaced his Dexcom symposium presentation on remote monitoring, non-adjunctive use/calibration, and extended G4 wear with trademark humor and energy, cautioning: “Almost everything I say will be off-label. Pediatricians by definitions are outlaws.”

7. Schneider Children’s Dr. Moshe Phillip provided a glance into the type of pump settings recommendations the Glooko-embedded Advisor Pro will provide, and highlighted the upcoming NextDREAM Consortium study to validate the software at six esteemed centers, which is now set to begin in September.

8. Dr. Bruce Buckingham gave a hard-hitting presentation on the sobering reality of hypoglycemia in type 1 diabetes, starting off with unpublished results on hypoglycemia during exercise released with Dr. Michael Riddell’s permission.

9. The great Dr. George Grunberger closed his second Keystone talk with a stirring call to action: “CGM is here to stay. It’s the standard of care for patients on intensive insulin therapy, and it’s high time we reach out and teach patients, providers, and payers how to use it.”

10. In a session on CGM advances, Dr. Bruce Buckingham discussed the DiaMonD and GOLD study results and detailed future sensors in developed for the US (Eversense, Glysens, Verily/Dexcom).

Diabetes Therapy Highlights

11. Dr. Simon Heller outlined the major advantages of next-generation basal insulins – Novo Nordisk’s Tresiba (insulin degludec) and Sanofi’s Toujeo (insulin glargine U300) – focusing especially on reduced hypoglycemia, an outcome beyond A1c that we cannot emphasize enough. This was an incredibly important talk, and made it well worth waiting for the last day of Keystone.

12. Dr. John Buse discussed SGLT-1/2 dual inhibitors, namely Lexicon’s phase 3 sotagliflozin which mitigates the DKA risk associated with current SGLT-2 inhibitors in type 1 diabetes. That said, he maintained that SGLT-2 inhibitors themselves can have profound benefits in type 1 diabetes care, and he offered practical recommendations for off-label use.

13. In the highly-anticipated last session of the conference (an event alluded to in Q&A since the very beginning of the meeting) UNC’s Dr. Richard Kahn posed a hard-hitting question: Why are so many people with diabetes “left behind” and not achieving their desired outcomes despite the abundance of available therapies and technologies?

14. Dr. Steven Nissen offered the cardiologist’s perspective on diabetes care, arguing that it has taken the field far too long to emphasize CV outcomes. He described the era of diabetes management pre-CVOTs as the “Dark Ages,” when endocrinologists and regulatory authorities were satisfied with a therapy based solely on its A1c-lowering efficacy, which he criticized as far too simplistic.

15. Past AACE president Dr. George Grunberger followed Dr. Nissen and shifted focus from macro to microvascular disease, where meticulous glucose control is the crux of preventing complications.

16. Representing the basic science wing of the Barbara Davis Center, Dr. Lori Sussel provided a fascinating overview of recent research on beta cell physiology and its implications for the future of type 1 treatment and cures. The most exciting breakthroughs in her view are the discovery of multiple beta cell subtypes and an alternate theory for type 1 pathophysiology whereby rather than beta cell death, islets are losing their cellular identity (a process that may be reversible).

Diabetes Technology Highlights

1. Dexcom Medicare Update: G5 Shipments Finally Up and Running; “Thousands and Thousands and Thousands” in the Queue, Could Take Several Months to Get Through Pipeline; Share Functionality (remote Monitoring) to Initially Be Disabled :(

Dexcom Senior Vice President of Global Access Dr. Claudia Graham gave the update we’ve all been waiting for: It has taken a while to implement, but Medicare shipments are up and running…slowly! According to Dr. Graham, Dexcom now has “thousands and thousands and thousands of patients in the queue, and we haven’t even started advertising…bear with us, it’ll take several months to work through all of the patients in the pipeline.” This progress report means that,  following the FDA’s non-adjunctive label approval for G5 in December, Medicare’s subsequent benefit category determination (DME, Part B) in January, and the announcement that coverage will be provided in March, “final” local coverage decisions and administrative logistics have been worked out (Dexcom projected in May that this would happen before the end of 2Q17). The company is now in the process of shipping G5 receivers that last for three years (~$236-$277 each), along with G5 sensors, 90-day G5 transmitters, 60 test strips/month, Ascensia glucose meters (partnered earlier this month), lancets, and control solution in a monthly bundled subscription model – the only such model in Medicare besides oxygen! – for ~$240/month (patients pay ~$50/month out of pocket).

• In unfortunate news, Share remote monitoring will indeed be disabled with G5 Medicare transmitters – at least for the time being. This is extremely unfortunate to hear and something we had feared following the CMS coverage decisions (using the smartphone means the system is non covered). She was visibly irritated by the policy, which states that even when seniors use their receiver, they are not allowed to use a smart phone to transmit their glucose values to caregivers.
• Dexcom is working aggressively to remedy the current situation. Said Mr. Sayer in Q&A, “We have literally plans A, B, C, and D in development if they won’t let us use the phone, because we believe remote monitoring is so important. In a year, we won’t be talking about this issue.” We assume this would require building cellular or WiFi into the receiver so that data can go to the cloud and a follower could see it remotely.
• Dr. Graham noted that 60 test strips will frequently not be enough for a 30-day month (~65-75 would be ideal), nor will four seven-day sensors – as the distributor, it is incumbent upon Dexcom to call the patient every 30 days to ensure that he/she has exactly what is needed for the month. (A reminder that Medicare opens up a new market, but also adds a lot of administrative work for Dexcom.) We love how hard Dexcom has pushed to get this done – management could easily say “our work is done” and leave the rest to suppliers/distributors, but they want to be absolutely certain that this goes well, given the arduous battle they fought to have this opportunity. Dr. Bruce Buckingham took to the microphone following Dr. Graham’s talk to assert that “Dr. Graham should be sainted.” Hear, hear! We look forward to hearing how many bundles have already been shipped and how many are in the queue.
• See our Detailed Discussion and Commentary below for more comments!

2. Dexcom G6 Pivotal Wrapping Up, to be Filed with FDA by End of September, on Track for 2018 Launch; Verily Products Timing on Track for 2018 and 2020-2021 launches; New Facility’s Crazy Capacity

VP of Advanced Technology Peter Simpson wrapped up Dexcom’s information-packed dinner with a look at Dexcom’s pipeline, with some help from CEO Mr. Kevin Sayer. The speakers touched on G6 (pivotal trial wrapping up, FDA submission by the end of this September), first- and second-gen verily products, G5 Mobile, a hint at strategic partnerships, and more.

• Mr. Sayer confirmed that Dexcom plans to submit the impressively accurate G6 sensor by the end of September, keeping the planed 2018 launch timing on track. This sensor has one calibration/day after startup, 10-day wear, a 30% thinner wearable than G5, and no acetaminophen interference – a very meaningful improvement and excellent to see Dexcom keeping up with the timeline laid out in 1Q17! We assume this FDA review wouldn’t last more than a year, meaning a launch is actually possible in the first half of 2018. Mr. Sayer also shared that the pivotal trial (n=300+; 30,000+ matched pairs) will wrap up next week – we can’t wait to see how the data lines up with the pre-pivotal data shared at DTM (n=49) – an MARD of 8.1% on one calibration per day after startup (10% on day one!), with excellent potential for a no-calibration algorithm. Mr. Sayer stated that the leap from G5 to G6 will be almost as big as Seven+ to G4 – considering that many consider G4 to be the first viable consumer CGM, these are strong words! Wow!
• Mr. Simpson hinted that the first-gen CGM with Verily is very well into development, and we can expect to hear more about it soon, and the second-gen CGM with Verily is still in the feasibility stage (animal and human studies) with a lot of work to do. The timeline hasn’t changed for either version: Gen-1 is slated for a 2018 launch, and gen-2 for a 2020-2021 launch. As a reminder, the first-gen CGM leverages G6 sensing technology, but is smaller by 30%, single use disposable, zero-calibration, 14-day wear, and is inserted with an auto-applicator. The second-gen CGM is the diameter of a penny and the thickness of two, also has a single use disposable, zero-calibration, 14-day wear, has a new auto-deployment system, and features new electronics/sensing technology. In our view, the potential for these small, simple, and cost-effective sensors in type 1 and type 2 can’t be overstated. IN an update from the January JPM presentation, the investor deck now notes that the cost reduction is targeted for the second-gen Verily sensor.

• Dexcom is leasing a new facility in Flagstaff, Arizona, capable of manufacturing 50 million sensors per year – this is the first quantification we can recall hearing of this facility’s capacity. For context, Dexcom has only made 10-12 million sensors cumulatively to this point. This is a serious investment, as the company will pour $70 million into just the duct and air flow work – wow! This is another step on the quest to make CGM ubiquitous – Mr. Sayer stressed that he has every intention of making sensors simple and cost-effective enough to be used everywhere. This includes prediabetes, a ways down the road: “That’s a strategic issue we battle internally with resources. This is a great tool, but the system we have today won’t fix it. We need the new products…Once we get through type 2 diabetes, I think you’ll see prediabetes after.”
• Dexcom’s post-market study of G5 Mobile for non-adjunctive use is “just ramping up now.” The comprehensive post-market study to ensure the safety of non-adjunctive use was condition of FDA approval – debated heavily at the July Advisory Committee meeting. We’re not sure what form this study will take, but have little doubt it will show safety and effectiveness (especially given the T1D Exchange REPLACE BG study).
• Mr. Simpson also spoke about G5 Mobile for Android and the Android Wear app (approved last month and already launched to somewhat mixed reviews), Apple opening up Bluetooth communication allowing direct flow of Dexcom CGM data to the upcoming WatchOS4 without a phone nearby (announced last month ahead of the Watch OS4 update this fall – we’re not sure when Dexcom will launch this, but it’s a highly compelling user experience win in our view), and the new G5 touchscreen receiver (seen at ADA; it’s a form factor step back from the current receiver, but it has to be designed to last three years for the Medicare DME bundle). This company does not rest!

3. Dr. Roy Beck Concludes CGM Ready for Near-Total Replacement of SMBG in All Type 1s and Type 2s on Basal-Bolus; Upcoming Studies to test CGM in Type 2s on Basal, Peds, Adolescents, and Elderly

Jaeb Center’s Dr. Roy Beck reviewed the plentiful evidence for CGM’s benefits in both type 1 and type 2 diabetes. He concluded that CGM yields better glycemic outcomes than SMBG, and with the latest technology, is accurate enough to replace fingersticks. In his widely cast literature review, Dr. Beck briefly touched on Dexcom’s DIaMonD study, the GOLD study, Abbott’s REPLACE and IMPACT studies (FreeStyle Libre in type 2 and type 1, respectively), and REPLACE-BG (a T1D Exchange study testing non-adjunctive use with Dexcom’s G4). While he did glance over the overwhelmingly positive glycemic outcomes from these trials, on every data slide, wear time and treatment satisfaction/quality of life were highlighted, sending a clear message that CGM drives glycemic benefits, but only when it is worn – in our view, now it’s about bringing this technology to a wider population, dropping the cost, and simplifying users’ wear experience. The field has come far from the poorly-tolerated GlucoWatch! All of these studies, which came out in just the past three years, indicate to Dr. Beck that CGM could be near-total replacement for all patients with type 1 diabetes and patients with type 2 diabetes on MDI or pump therapy. “Near-total” refers to the need for calibrations (except for FreeStyle Libre), and even with factory-calibrated sensors, he would like to see the ability to re-calibrate in scenarios when CGM values look off. (We expect this will be an increasing point of debate, and it will be fascinating to see how FreeStyle Libre and next-gen Libre comes to market in the US, especially with Bigfoot’s automated insulin delivery system.) How about replacement in type 2s on just basal insulin? There is no substantive clinical data to turn to, so the Jaeb Center and Dexcom will embark on a clinical study (the MOBILE study) soon to fill this knowledge gap. Nice! Dr. Beck believes that CGM, in its current form, is unlikely to be widely adopted by type 2 patients not taking insulin, though future simple, unobtrusive, low-cost sensors (e.g., Dexcom-Verily’s second-gen sensor) could change this. We like the potential of CGM providing behavioral feedback – even in people with prediabetes – perhaps at diagnosis and a few times per year. Before wrapping up, Dr. Beck announced a series of upcoming studies to evaluate CGM in understudied age groups (conducted by the T1D Exchange clinic network): The SENCE trial of CGM in children under eight years old (funded by Helmsley Charitable Trust), the CITY trial of CGM in adolescents (starting soon, expected to be funded by the Helmsley Charitable Trust), and the WISDM trial of CGM in the elderly (starting soon, looking at the interaction between cognitive issues and CGM efficacy, funded by JDRF and the Helmsley Charitable Trust).

• Two recent studies confirm that CGM is non-inferior, if not superior, to SMBG in terms of accuracy. Ekhlaspour et al. (JDST 2017) found that the MARD of a number of FDA-cleared BGMs ranges from 5.6% (Ascensia Contour Next) to 20.8% (Trividia Sidekick) - nine of them have a MARD >10%! Meanwhile, Dexcom’s G5, Medtronic’s Guardian, and Abbott’s FreeStyle Libre range from 9.0%-12.3% MARD. In a Diabetes Technology Society study, 18 BGMs were each tested in three sub-studies among 1,035 participants. In order to pass the assessment, >90% of tested meters had to read standard blood glucose within 15% of reference for blood glucose >100 mg/dl and within 15 mg/dl for blood glucose <100 mg/dl. The results were rather disturbing: Only 6/18 tested had passing grades in all three sub-studies (Ascensia Contour Next, Roche Accu-Chek Aviva Plus, Walmart ReliOn, Agamatrix CVS Advanced, Abbott FreeStyle Lite, and Roche Accu-Chek Smart View); 5/18 passed in two of the three sub-studies; 3/18 passed in one of the three sub-studies; and 4/18 failed in all three sub-studies! These studies will be crucial for the education of providers in the near future – there was audible shock in the room when the Ekhlaspour data was on the screen, and we’re sure that at least some are now convinced that CGM is not less accurate than BGM (see the full results here). We’re glad to see some of the accuracy baggage dissipating from early-gen CGMs that did not deliver.
  
  • Even if CGM isn’t more accurate than BGM, Dr. Phillip reminded that trend arrows are just as valuable, if not more, than point accuracy. He referenced an old study from Europe where a sensor with MARD 19.7% caused an A1c drop of 1%, the same as seen in the recent DIaMonD study with the much more accurate Dexcom G4. This was a point that Adam and others made at last year’s Dexcom-FDA meeting on non-adjunctive use of CGM – trend arrows provide far more valuable information than point-in-time fingersticks, even if the point accuracy is not as good as the most accurate meters (slides here).  
• Drs. Jeremy Pettus and Bill Polonsky tackled the question of which type 2s should be on CGM in an entertaining debate at ADA. They reached the same conclusion as Dr. Beck, though the debate of CGM in type 2s on basal swung convincingly to the “pro” side – we wonder if this will be supported by data from the MOBILE study.

4. FDA’s Dr. Courtney Lias Preaches Benefits of Interoperability (Less Regulatory Burden, Fewer/Smaller Clinical Trials, Patient Choice) and Hints at Upcoming Discussions to Address Challenges

FDA’s Dr. Courtney Lias proclaimed with an emphatic “Spoiler Alert!” that clinical trials are not chiefly responsible for slowing the commercialization of automated insulin delivery systems; rather, a lack of a framework for interoperability, and all that it entails, has slowed the field. We’re glad to see her continuing to hammer this message home, which we’ve heard from her at a number of conferences in the past year (e.g., ADA 2017, NIH Workshop, AADE 2016, ADA 2016). Number one on Dr. Lias’ list of things that bog down progress toward launch is contracting – she noted that some companies may have to work with three to four others to put together a closed loop system (e.g., CGM, pump, BGM), and the requisite effort put into determining and finalizing who is responsible for what in every scenario, what modifications need to be made in the future, post-market responsibilities, and performance requirements are astronomical. In her experience, contracts can slow things down by months to years at a time: “I only have an impression, which is that there’s a vastly larger sum of time and money spent developing and maintaining contracts than spent dealing with us.” She’s under no illusion that we can expect for every company to develop or acquire their own sensor, pump, algorithm, and BGM (for sensor calibration, if necessary) – on the contrary, she would rather see the development of communication specifications for automated insulin delivery components. Such a framework would make empirical design and testing – the process of developing a system and iterating through tweaks after small studies – a thing of the past. This kind of interoperability would enable safer, more nimble automated insulin delivery innovation, diminish regulatory and clinical study burden, and create a thriving ecosystem of component parts. In essence, devices must be optimized to work together, not in silos. The creation of transparent performance specifications – something she asked for the community’s input on, perhaps with an eye on updating the 2012 AP guidance – “may allow for open development and open approval of automated insulin delivery systems. I can’t tell you how many submissions that would eliminate.” She later mentioned that FDA is working to facilitate open discussion toward solving these challenges – we look forward to this discussion moving to tangible specifics, as the vision and benefits seem obvious for all stakeholders. She concluded her talk by explaining the urgency for a mindset shift in artificial pancreas development: “Companies are struggling and peering over the edge of the cliff – there’s no market for pumps, and people are not using devices because of real resource challenges. We don’t want regulatory burden or contracts to be the reason they fall off the cliff. We need to bridge the gap and have resources be put to what matters the most.” So, in simple answer to her speaking charge: “Will guidelines for artificial pancreas approvals be different in the future? It has to be yes.” Dr. Lias, our hats go off to you, and we hope interoperability moves forward in a meaningful way. By the way, please vote for Dr. Lias, Beck, and the FDA AP team here – they deserve to win a Service to America Medal!

• See our Detailed Discussion and Commentary below for more, including a terrific Q&A!

5. Data from MiniMed 670G in BDC Pediatrics, at ADA Camp; Increased I:C Aggressiveness in Pivotal Correlated with Increased Time in Range

BDC’s Dr. Gregory Forlenza presented un-published data from MiniMed 670G use in 10 of his pediatric patients in the continuation phase and at an ADA diabetes camp, as well as a look at how the system was optimized over time in the pivotal trial. 10/12 BDC pediatric patients (baseline A1c: 8.0%) enrolled in the pivotal trial opted to continue on the system – at three months, A1c had fallen to 7.0%, but this rose to 7.3% at six months, 7.5% at nine months, and 8.1% at 12 months (essentially back to baseline). This drift upwards is a bit surprising to us, since we’d assume these are a pretty engaged group. (We also wonder what happened with hypoglycemia, time in range, and patient-reported outcomes – Dr. Forlenza told us that groups at BDC, Stanford, and Yale have started a 670G Pediatric Collaborative and will be investigating this over the next few months). Dr. Forlenza made the outstanding point that some of these patients are going from age 14 to 16: “they’re supposed to get worse. I consider it a mild victory that their A1cs didn’t go up to 9%.” Prior to the ADA camp, Drs. Forlenza and team were concerned about the transition from Cheeto- and video game-filled days (not at camp) to active days (at camp), considering that the 670G operates based on the insulin requirements of the past six days. They made some preemptive tweaks to the pumps when the kids arrived at camp, and were happy to see that it worked – everyone had the blue shield displayed prominently on their pump home screens, indicating they were in Auto Mode. Relative to the six days of home 670G use, six days of 670G use at camp was very successful: Mean blood glucose was roughly the same (158 mg/dl at home vs. 156 at camp), percent of time in closed loop improved by about an hour per day (82% to 88%), as did time in range (68% to 71%) and time >180 mg/dl (30% to 27%). Time in hypoglycemia stayed flat at a very low 2%. As expected, total daily dose at camp plummeted by a remarkable 26% (from 51 to 37 units per day), indicating how much more active the kids were.

• A paper on optimizing hybrid closed loop in pediatrics, authored primarily by Ms. Laurel Messer and the 670G Pediatric Collaborative, will be published soon. Previewed data indicated that mean TDD, mean basal/autobasal insulin, mean bolus insulin, % basal/bolus split, and mean 24-hour basal program did not significantly vary over the course of the pediatric pivotal trial, indicating that glycemic improvements were likely unrelated to these parameters. Rather, the analysis found that significantly more aggressive carb:insulin ratios over time correlated with improved time in range. We’d note that this is a sneaky way to compensate for the 670G’s more conservative daytime dosing – give more manual bolus upfront and let the system shut off insulin if necessary. This is a viable strategy if patients are bolusing, though probably not a solution if patients are missing boluses. 
  • ​Another plot showed, as expected, that time in Auto Mode correlated well with time in range – another reason that clinicians (and Medtronic) will need to focus on keeping patients in Auto Mode for as much time as possible. 
• Dr. Forlenza emphasized the importance of updating basal rates for situations when patients get exited from auto mode: “One of my patients grew like two feet in two years since he started the study – he started as a pre-pubertal 14 year old, and now he’s a 16 year old man.” Insulin requirements change drastically in this time. To keep basal rates updated, Dr. Forlenza takes the total daily dose, divides it by 24, and multiplies by 0.8 to be cautious, then goes up or down depending on the time of day (more insulin resistant in the morning, exercise in the afternoon, etc.). As a reminder, the MiniMed 670G does not continuously update the background manual mode pre-programmed basal rates, meaning clinicians/patients will still need to do this. 
• Dr. Forlenza had a number of quotable quotes putting the 670G in the broader closed loop context and reflecting on patient experiences: 
  
  • “The 670G is analogous to Zack Morris 1980s cell phone – it doesn’t have games or widgets – but in the 1980s, it was just amazing that you could be talking on the phone while not touching a wall. My kid would say “you were excited about this?” The system has constraints on it, but does amazing things relative to where we are right now.”
  • “The coolest thing we see with artificial pancreas systems is the shattering of the paradigm of trading between lower A1c and less hypoglycemia – minimizing variability does both.”
  • “Glycemic spread, standard deviation is the real measure of frustration. Patients always come in and say, ‘I just feel like it’s out of control, I’m always 300 or 50.’ If you reduce that, you improve quality of life.”  
  • “Telling people ‘2.5 more hours in range’ is much more impactful than saying ‘7% more.’ And 45 minutes per day less time less than 70 mg/dl is huge, anyone with diabetes knows 45 minutes feels like two hours at less than 70 mg/dl.”
  • “Adolescents are more difficult than adults – everyone in this room knows that. We have a much harder job then the group next door.”

6. Dr. Buckingham: Remote Monitoring in Peds Crucial, Non-Adjunctive CGM Use to Mitigate Risk of Mis-Calibration with Fingersticks; Data on Extended G4 Wear

Dr. Bruce Buckingham prefaced his Dexcom symposium presentation on remote monitoring, non-adjunctive use/calibration, and extended G4 wear with trademark humor and energy, cautioning: “Almost everything I say will be off-label. Pediatricians by definitions are outlaws.” Dr. Buckingham highlighted the ability of remote monitoring to significantly cut down parental stress and deemed it essential for children under six, although setting up boundaries for a sharing relationship can be problematic when expectations don’t align – see more on this topic from diaTribe’s panel discussion at FFL. He was particularly excited about non-adjunctive use of CGM, citing the REPLACE-BG study as the first major trial to strongly support non-adjunctive use without sacrificing glycemic control. He recommended continuing fingersticks during the first 12 hours of a new sensor and waiting four hours before using the sensor again after taking acetaminophen, stipulating that it’s important to remain realistic about non-adjunctive use. Dr. Buckingham sees non-adjunctive use as helpful in mitigating risk of mis-calibration, as it requires only two calibrations per day (typically done when waking up and before bed, times where hands are more likely to be clean). Still, he emphasized the importance of washing hands before calibration, referencing one study that demonstrated 50% higher blood glucose values after peeling fruit (even with use of an alcohol wipe!). Echoing Dr. George Grunberger’s talk from earlier in the day, Dr. Buckingham called for consistent CGM reporting and was pleased by Dexcom’s addition of the standard AGP to Clarity, appreciating the ability to delve into each day’s data and clearly view patterns and variability. Still, he’s well aware that most patients almost never look at their download but do view their real-time data 10-15 times per day in hopes of achieving a “no-hitter” (100% time-in-range). He finds trend arrows absolutely critical in attaining this goal, but warns that algorithms pairing trend arrows with insulin dosing adjustments have not been validated and are only meant as rough guidelines ­­– this is a very new field and we expect this will improve markedly over the coming years as decision support comes online for both pumpers (automated insulin delivery) and injectors.

• Dr. Buckingham presented new data investigating extended sensor wear for the Dexcom G4 CGM. Welsh et al. (2017) demonstrated what Dr. Buckingham considered to be nice accuracy over the course of normal seven-day + seven-day consecutive wear times (n=79 pediatric; n=51 adult). Another study examined Dexcom G4 wear for three weeks (n=20): Interestingly, adhesive failures leveled off by day 15, but sensor failure continued to occur increasingly (see picture below). For those sensors that survived, accuracy remained flat over the 21-day period. These data explain why so many people are comfortable using sensors for longer than their intended duration (and why Dexcom CEO Mr. Kevin Sayer loves Germany – “they’re all analytical, they take it off in seven days”). As a reminder, G6 will be 10-day wear and is technically capable of 14-day wear; we don’t believe a mandatory shutoff will be required, but presumably Dexcom will discuss this with the FDA.

7. Sample Recommendations from DreaMed’s Advisor Pro; Pivotal Study to Begin in September

Schneider Children’s Dr. Moshe Phillip provided a glance into the type of recommendations the Glooko-embedded Advisor Pro will provide, and highlighted the NextDREAM Consortium study, which is now set to begin in September (delayed slightly). Within the Glooko platform, providers can see the patient’s current therapy alongside the suggested adjustments – for pumpers, there are currently pages for basal rates, carb ratio, correction factor, and active insulin time. In addition to very specific suggested adjustments (e.g., change basal rate from 12pm-3pm to 0.85 u/hr), the Advisor will rationalize with 400+ unique notes describing detected patterns relating to behavior and insulin dosing. Examples include: “Due to a pattern of postprandial hyperglycemia, the carb ratio was updated to deliver more insulin”; “Your patient tends to skip boluses at lunch”; or “Your patient skips boluses every second weekend of the month.” This sort of pattern detection is very exciting, hard for busy clinicians to do, and refreshingly specific. Indeed, Dr. Phillip estimates that optimizing pump settings – including poring over 25+-page CareLink reports (which he doesn’t read in full) – requires 10-20 minutes of valuable provider time for each patient. This clinic time can be liberated for discussion of things that really matter to people with diabetes – food, exercise, behavior changes, foot exams, etc. Dr. Phillip then turned to the clinical validation of the Advisor Pro, a job for the NextDREAM consortium, which will conduct a multi-center study beginning in September (back from the late July/early August timing provided at AACE 2017). Ljubljana, Schneider Childrens, Auf Der Bult, Yale, Barbara Davis, University of Florida, and Joslin comprise the impressive Consortium – Dexcom will provide the G5 for the study. As a reminder, Advisor Pro was non-inferior to esteemed physicians in a pilot study (MED-Logic Advise4U; presented at ATTD). In the words of Dr. Phillip: “We are happy. It’s just another team member.” Yes, another team member, but one that can sift through tomes of data, scale time-pressed providers, and drive better outcomes.

8. Dr. Buckingham: Initial Glucose Levels Predictive but Not Always Protective of Exercise-Induced Hypo; Advice to Adolescents: “Never Sleep Alone”

Dr. Bruce Buckingham gave a hard-hitting presentation on the sobering reality of hypoglycemia in type 1 diabetes and how to prevent it, beginning with unpublished results on hypoglycemia during exercise from Dr. Michael Riddell. Dr. Riddell (whom we last heard speak at this year’s ADA on exercise and the artificial pancreas) found starting glucose levels can be predictive but not always protective of blood glucose following aerobic activity (see data below; red bars represent individuals who dropped below 70 mg/dl). In the Dskate Camp study (also below), some individuals saw blood glucose drop immediately after while others saw it rise. These data show striking variation in individual responses to exercise, indicating a need for personalized plans and the use of a sensors with decision support or sensor-augmented pump/hybrid closed-loop systems. Dr. Buckingham also warned against the dangers of hypoglycemia unawareness, which he’s seen in the majority of “dead-in-bed” cases. To those with this condition, Dr. Buckingham recommended “never to sleep alone” and joked that he offers this advice to all of his patients going off to college. Humor aside, hypoglycemia unawareness is reported in an estimated ~20%-30% of type 1 adults, and unfortunately was not found to reverse after five weeks on closed loop in a study referenced by Dr. Buckingham. However, a separate study (Ly et al., Diabetes Care, 2011) found that CGM use improved the autonomic response to hypoglycemia in adolescents with type 1 diabetes. Dr. Buckingham praised remote monitoring and real-time CGM as critical in mitigating the risks associated with hypoglycemia unawareness, referencing one of his own camp studies of 7-16 year-olds (n=60), which demonstrated benefits of remote monitoring using the G4 sensor (he mentioned that the experience of monitoring alongside Dr. Buckingham convinced Dexcom’s Mr. Andy Balo to rapidly push Share through the pipeline!). However, there is a lack of research investigating remote monitoring and more work in this field is needed.

• Dr. Buckingham ended his discussion with these final words of wisdom about how to avoid hypoglycemia: “Get a sensor, or sensor-augmented pump, or hybrid closed loop system – probably what you expected from me.”  This is exactly what we expected – see our preview if you don’t believe us : > Day-to-day variability in insulin sensitivity makes it nearly impossible to prevent glucose excursions (both up and down) without real-time feedback, ideally supplemented with automation.
  
  • To highlight the value of real-time CGM, Dr. Buckingham discussed a tragic example of a driver who got behind the wheel with a 108 mg/dl reading. 30 minutes later, the driver crossed the divider and killed a mother in another vehicle – his glucose reading was 20 mg/dl. Had the driver had been wearing a real-time CGM with trend arrows, this preventable death could likely have been avoided.  
• According to a recent 2017 study published in Pediatric Diabetes, parents of children with diabetes three years old or younger express significant fear of hypoglycemia at night. Not surprisingly, the data also suggest highly significant (p<0.001) fear at night amongst parents of pump users. Concerningly, Dr. Buckingham found that 71% of parents fail to recognize when their child is low – imagine how impossible this would be for a teacher in a crowded school setting! CGM and remote monitoring need to become the standard of care in all insulin users, and especially in the pediatric population.

9. Dr. George Grunberger: “CGM is here to stay. It’s the standard of care for patients on intensive insulin therapy, and it’s high time we reach out and teach patients, providers, and payers how to use it”

The great Dr. George Grunberger closed his second Keystone talk with a stirring call to action: “CGM is here to stay. It’s the standard of care for patients on intensive insulin therapy, and it’s high time we reach out and teach patients, providers, and payers how to use it.” He emphasized the need for actionable CGM data and a standardized outcomes reporting system, citing his commonly-stated goal to make CGM reports as universally understandable as an EKG. Given the variety of glucose targets and hypoglycemia thresholds found on different devices, Dr. Grunberger praised Dexcom’s recent addition of the standardized, one-page Ambulatory Glucose Profile to its Clarity platform as an important step towards standardization (Dexcom joined Abbott, Roche, and Glooko/Diasend as key adopters). He also noted that A1c misses the mark when it comes to actionable data, calling for clinicians to move past the A1c-centric culture and instead focus on expanding time in range. We loved his shout out to this Friday’s Glycemic Outcomes Beyond A1C workshop – he will not be in attendance due to a prior conflict, but plenty of thought leaders, including AACE representative Dr. Zachary Bloomgarden, will be. Dr. Grunberger emphasized that, while clinicians should be responsible for analyzing CGM data alongside their patients, CDEs are critical in ensuring that patients can interpret CGM data in real time before device initiation – that includes education on arrows/trends, best practices for wear and calibration, and dosing off a CGM (he reviewed the DirecNet, Edelman/Pettus, and Scheiner methods). According to Dexcom’s Dr. Claudia Graham, people pick up on CGM use pretty quickly, usually within 45 minutes. The recently published AACE/ACE guidelines derived from a consensus conference echoed Dr. Grunberger’s recommendations, focusing on patient and provider education and comprehensive, standardized analysis. On the importance of provider education, Dr. Grunberger gravely noted: “This is not for the fainthearted, providers make life and death decisions and it’s really important that they understand glucometrics. If you’re not interested, send your patient to someone who is.” Dr. Grunberger also discussed ATTD CGM guidelines slated for publication soon. The ATTD group recommends a minimum of 14 consecutive days of CGM data for an actionable report and suggests five threshold buckets to interpret time in range (<54 mg/dl, <70 mg/dl, 70-180 mg/dl, >180 mg/dl, >250 mg/dl). We very much appreciate all of the efforts that have been put into consensus meetings; Now, we turn our attention to implementation, getting CGM routinely used in drug trials, incorporated into regulatory decisions, and used widely in clinical practice to make therapeutic changes.

10. Dr. Buckingham Offers DIaMonD and GOLD Commentary, Optimism for Future Sensors

In a session on CGM advances, Dr. Bruce Buckingham discussed the DiaMonD and GOLD study results and highlighted several future sensors in development for the US. On the landmark studies of CGM in MDI, Dr. Buckingham was particularly impressed with sensor wear in the DIaMonD study, calling attention to the incredible 93% wear time amongst type 2 adults after six months. Type 1 subjects demonstrated equally impressive usage, with an average wear of 6.7 days/week at six months. (We’d note that patients who didn’t wear the CGM were screened out of the study – a smart design move on Dexcom’s part.) According to Dr. Buckingham, one of the major benefits indicated by the DIaMonD results is the ability to record glucose levels overnight – this is obviously a huge gap in SMBG! Importantly, the study demonstrated no significant differences due to age, education, or numeracy, and Dr. Buckingham hopes this will drive high levels of Medicare adoption (now that Dexcom G5 is reimbursed and shipping, see above!). The GOLD data reinforced findings from the DIaMonD study, proving that patients on MDI can and will wear CGM on a daily basis to the effect of reductions in both A1c and hypoglycemia. As Dr. Buckingham rather bluntly commented, “you need to use the system for it to work…you sort of expect that.” Following Dr. Buckingham’s presentation, Dr. Aaron Kowalski vocalized what the majority of the room was likely thinking: “I’m bubbling with frustration that we’re still talking about getting people on CGM. The evidence is irrefutable.” As for up-and-coming sensors, Dr. Buckingham shared optimism for implantables (Senseonics, Glysens), as well as smaller/less expensive sensors like the Dexcom/Verily and FreeStyle Libre (for which he sees a “real-time future” given last week’s major Bigfoot agreement).

• Dr. Buckingham expressed excitement regarding the Senseonics Eversense implantable CGM (under FDA review) and the in-development Glysens CGM. So far, the fluorescence-based Eversense has been tested in 500 patients for over 100,000 days. While the quick in-office insertion procedure is relatively simple, it will be fascinating to see how clinical adoption of this goes – endocrinologists are not used to procedure. Dr. Buckingham also mentioned the Glysens implantable, which is calibrated monthly, and cautioned that it’s been hard to find published results – the one study he did discuss published in IEEE Biomedical Engineering showed a MARD varying monthly from 8.1%-28% (see picture below). Still, he sees a future for Glysens, commenting that “if people are looking for an implanted sensor that lasts for a year, this seems to be one of the ones coming down the road.” This company has not moved as quickly as some others, though the long implant time does lengthen testing and development; we’re not sure at what point we’ll see a Glysens pivotal trial or commercialization. 

Diabetes Therapy Highlights

11. Dr. Heller on Benefits of Insulin Degludec over Other Basal Insulin Analogs (Hint: Look at the Hypo Data!)

Dr. Simon Heller outlined the major advantages of next-generation basal insulins – Novo Nordisk’s Tresiba (insulin degludec) and Sanofi’s Toujeo (insulin glargine U300) – focusing especially on reduced hypoglycemia, an outcome beyond A1c that we cannot emphasize enough. Dr. Heller shared current statistics indicating that people with type 1 diabetes experience one severe episode of hypoglycemia per patient-year, while people with type 2 experience one severe episode per three patient-years, on average. Studies using CGM show that unrecognized hypoglycemia is common, occurring in 63% of patients with type 1 and 47% of patients with type 2 diabetes. Moreover, 74% of these unrecognized cases happen at night. Another study Dr. Heller presented found that 54% of hypoglycemia is nocturnal, and 83% is unrecognized. He emphasized that nocturnal hypoglycemia is associated with a nearly 10x increase in bradycardia, which is one of many reasons that the complication is so dangerous. While Sanofi’s Lantus (insulin glargine U100), a current standard of care, is associated with a 14% decrease in nocturnal hypoglycemia vs. NPH insulin (Dr. Heller pointed out that this finding led to a most-welcome change, promoting widespread use of Lantus), Tresiba does one better. Novo Nordisk’s SWITCH 1 study in type 1 diabetes reported an 11% risk reduction for the primary endpoint of severe or blood glucose-confirmed symptomatic hypoglycemia with Tresiba vs. Lantus (p<0.0001) as well as a 36% risk reduction for severe or blood glucose-confirmed symptomatic hypoglycemia overnight (p<0.0001). SWITCH 2 used the same trial design but enrolled participants with type 2 diabetes, and found a 30% risk reduction for the primary endpoint with Tresiba vs. Lantus (p<0.0001), plus a 42% risk reduction overnight (p<0.0001). Full results from the DEVOTE CVOT comparing Tresiba vs. Lantus were presented at ADA 2017, and insulin degludec showed a remarkable 40% risk reduction for severe hypoglycemia vs. insulin glargine in a much larger sample size (p<0.001). Even more impressive was the data on nocturnal hypoglycemia: Risk for a severe low overnight was reduced by 53% with insulin degludec vs. insulin glargine (p<0.001). The EMA approved a label update for Tresiba to include the SWITCH 1 and 2 hypoglycemia data in 1Q17, and we’re eagerly awaiting an FDA decision on this same label revision by September 2017. Dr. Heller described how regulatory authorities were reluctant to include hypoglycemia as a known benefit of the degludec formulation due to lack of double-blind studies, so SWITCH 1 and SWITCH 2 were designed to address this. Given the persistence of severe and nocturnal hypoglycemia as diabetes complications, this benefit is a major advantage of insulin degludec therapy, as it could address one of the great looming health and health economic consequences of diabetes (hypoglycemia has a distinct adverse impact on patient quality of life, and is also very costly for the healthcare system). Indeed, our fingers are very crossed that this important benefit is reflected on the Tresiba product label in the US, soon.

• Dr. Heller concluded that while insulin degludec currently seems to provide the most robust and clinically-significant reductions in severe hypoglycemia in type 1 and type 2 diabetes vs. insulin glargine U100, similar trials with Sanofi’s Toujeo (insulin glargine U300) are underway. We’ve heard from thought leaders that Tresiba offers a flatter PK/PD profile for less daily glycemic variability, and that the flexible dosing of Tresiba makes it an even more convenient therapy vs. Toujeo. That said, we know the value of having multiple products with high efficacy on the market. Moreover, we underscore that many people with diabetes aren’t yet accessing either next-gen basal insulin product, so we see more than enough room for both Tresiba and Toujeo to do well commercially and enhance diabetes care broadly in type 1 and type 2.

12. Dr. Buse on Sotagliflozin, First Adjunct Oral Therapy for Type 1 on the Horizon

In the final plenary of the meeting, Dr. John Buse discussed SGLT-1/2 dual inhibitors, namely Lexicon’s phase 3 sotagliflozin, which mitigates the DKA risk associated with current SGLT-2 inhibitors in type 1 diabetes. Safety concerns surrounding DKA and euglycemic DKA in particular (when the absence of hyperglycemia makes it more difficult for patient/provider to identify and treat the complication) have been a major roadblock in the clinical development of SGLT-2 inhibitors for a type 1 indication. According to Dr. Buse, patients with type 1 diabetes taking an SGLT-2 inhibitor off-label face a 5%-10% risk for DKA over one year of treatment, and that’s “scarily high.” That said, he pointed to a recent study which found that 67% of DKA episodes attributed to SGLT-2 inhibitor therapy had an obvious precipitating event, something like alcohol intake, reduced insulin dose, or dehydration. He maintained that SGLT-2 agents can have profound benefits in type 1 diabetes care, as long as the patient is trained to check ketones, stay hydrated, consume carbs with full doses of insulin if feeling unwell, and proactively hold the SGLT-2 therapy ahead of potential precipitants (i.e. a hike that will cause major energy expenditure and calorie burn). We appreciated this commentary and Dr. Buse’s practical recommendations for off-label SGLT-2 inhibitor use – we continue to believe that DKA risk can be effectively managed in type 1 patients who want to include an SGLT-2 inhibitor in their medication regimen, and we see massive need for better adjunct therapies in type 1 diabetes. Lexicon’s sotagliflozin is closest-to-market for this indication (the company scheduled a meeting with the FDA in 2Q17 to discuss a New Drug Application). Dr. Buse announced with distinct excitement that presentations of inTandem1 and inTandem2 at ADA 2017 were the first phase 3 studies to report for any oral type 1 therapy. He reviewed phase 2 and phase 3 results from the inTandem clinical program, showing sotagliflozin to be effective in lowering A1c without increasing hypoglycemia, reducing bolus insulin dose, promoting weight loss, and decreasing both fasting and postprandial glucose. Importantly, DKA event rates were very small across all inTandem trials, and were mostly in the setting of insulin pump therapy. Dr. Buse interpreted this as a strong safety profile for the drug candidate, one that supports continued development and eventual commercialization. We certainly hope that SGLT-2 inhibitors may one day be an adjunct therapeutic option for people with type 1 diabetes, but near-term, we echo Dr. Buse’s enthusiasm for sotagliflozin and the promise it holds to sidestep DKA-related concerns from providers and regulators.

• Lilly/BI and AZ have ongoing phase 3 programs for their SGLT-2 inhibitor products in type 1 diabetes. Lilly/BI’s EASE-2 and EASE-3 trials for Jardiance (empagliflozin) are expected to complete in October 2017 and September 2017, respectively. AZ’s DEPICT 1 and DEPICT 2 for Farxiga (dapagliflozin) are scheduled to complete in August 2017 and April 2018, respectively. These completion dates are coming up soon, and we’re incredibly eager to see the data. Our fingers remain oh-so crossed that empagliflozin and dapagliflozin demonstrate clinically-meaningful efficacy alongside a highly-manageable safety/tolerability profile among study participants with type 1 diabetes. That said, these companies face multiple investment opportunities for their SGLT-2 inhibitor products – type 2 diabetes, type 1 diabetes, chronic heart failure, chronic kidney disease, etc. We understand that pursuing a type 1 indication may take the backburner for a heart failure indication, given that both Lilly/BI and AZ have initiated studies of their SGLT-2 agents in people with chronic heart failure with or without diabetes. J&J reported phase 2 data for Invokana (canagliflozin) in type 1 diabetes at ADA 2016, but the future of this clinical development project is uncertain. The company is also conducting a dedicated study of Invokana in diabetic kidney disease (CREDENCE, expected to complete in June 2019), and will have to evaluate options in the aftermath of CANVAS full results, which showed significant CV risk reduction but also a nearly two-fold increase in lower limb amputations with canagliflozin vs. placebo.

13. Dr. Kahn Asks Why People with Diabetes Are Getting “Left Behind”

In the highly-anticipated last session of the conference (an event alluded to in Q&A since the very beginning of the meeting) UNC’s Dr. Richard Kahn posed a hard-hitting question: Why are so many people with diabetes “left behind” and not achieving their desired outcomes despite the abundance of available therapies and technologies? He argued that the problem has less to do with whether we are using the “right” drugs or the most advanced devices, and is rather a consequence of our incomplete understanding of behavior and how to change it. Commonly-cited reasons for a high percentage of diabetes patients not reaching target A1c include the cost of diabetes therapy and provider disincentives – lack of awareness of diabetes treatment guidelines, insufficient time to spent with patients, and sheer lack of bandwidth in primary care clinics, where most people receive their diabetes care. But Dr. Kahn pointed out that the issue runs much deeper. Even in the setting of clinical trials where participants receive world-class care, frequent visits and calls, free drugs and supplies, and more of everything if needed, there are non-responders who are “left behind” and don’t reap the benefits of the therapy under investigation. In the DCCT and ACCORD trials, for instance, a sizeable proportion of participants could not get their A1c down to the starting goal of 6% despite strong motivation and the best possible care. He further remarked that A1c didn’t budge after implementing CGM for 15% of participants in the DIaMonD study, and the LEADER trial’s North American participants fared 20% worse than their European counterparts on CV outcomes with liraglutide despite controlling for baseline demographics, disease characteristics, A1c, and body weight (a significant talking point at the recent FDA Advisory Committee meeting for Victoza). In short, “there are people for whom these therapies and technologies don’t work.” Instead, he focused on the fact that there is much left to learn about behavior: How can we best increase motivation, improve rational thinking, and streamline decision-making? These challenges loom even larger against the backdrop of persistent patient disincentives like the cost of therapy, appointment hassles, stress, comorbidities, aversion to a complicated treatment regimen, and the list goes on. Dr. Kahn closed with an impassioned call-to-action, highlighting our “urgent need of robust and comprehensive systems of care” and value-based healthcare that better aligns the incentives of patients and providers as a starting point. We were left feeling the magnitude of this persistent challenge in diabetes care, a fitting note on which to end this conference, and one that reminds us why we come to work every day.

14. Dr. Nissen: CVOTs Brought Diabetes Care Out of the “Dark Ages”

Dr. Steven Nissen offered the cardiologist’s perspective on diabetes care, arguing that it has taken the field far too long to emphasize CV outcomes. Merely lowering glucose is far too simplistic as a treatment goal, because ultimately, we want to improve outcomes. Dr. Nissen described the era of diabetes management pre-CVOTs as the “Dark Ages,” when endocrinologists and regulatory authorities were satisfied with a therapy based solely on its A1c-lowering efficacy (he provided a memorable visual of two ostriches burying their heads in the sand, one representing FDA, the other representing diabetologists). A meta-analysis of the ACCORD, ADVANCE, and VADT trials found no significant association between a modest ~1% A1c decline and CV outcomes: The hazard ratio was 0.97 for stroke (in favor of glucose-lowering) and 1.03 for coronary heart failure (trending against glucose-lowering). Dr. Nissen reminded the audience that large diabetes CVOTs employ a glycemic equipoise design, meaning patients are treated toward the same target A1c in the treatment and placebo arms. In EMPA-REG OUTCOME (for Lilly/BI’s SGLT-2 inhibitor empagliflozin), LEADER (for Novo Nordisk’s GLP-1 agonist liraglutide), and CANVAS (for J&J’s SGLT-2 inhibitor canagliflozin) alike, A1c was ruled out as an explanation for cardioprotection. Instead, there must be something inherent to each therapeutic molecule conferring CV benefit, and Dr. Nissen issued a wake-up call to regulators and guideline-writing committees to recognize the value of cardioprotection in diabetes management. We were pleased to see Jardiance (empagliflozin) and Victoza (liraglutide) recommended to type 2 diabetes patients at high CV risk in the ADA’s 2017 Standards of Care, and the FDA approval of a new CV indication for Jardiance was a big step in this post-Dark Ages era of CVOTs. We’re hopeful that the FDA will follow-through with a new CV indication for Victoza as well, following last month’s Advisory Committee meeting.

• Pushing for even more forward progress, Dr. Nissen suggested that one next step for CVOTs might compare a GLP-1 agonist vs. an SGLT-2 inhibitor, or might investigate the additive CV effects with co-administration of both agents. AZ’s DURATION-8 is the first major clinical trial to evaluate this particular combination: Participants in the treatment arm received GLP-1 agonist Bydureon (exenatide once-weekly) alongside SGLT-2 inhibitor Farxiga (dapagliflozin), and quite notably, achieved statistically significant A1c decline and weight loss vs. either monotherapy. That said, Dr. John Buse has suggested that there isn’t enough evidence right now to assume an SGLT-2 inhibitor and a GLP-1 agonist would have additive CV benefit, especially since they have differential effects on glucagon.
• Dr. Nissen further argued that TZD pioglitazone is under-utilized in diabetes care, based on CV outcomes data. A secondary outcome in the PROactive study showed a statistically significant 16% risk reduction for three-point MACE (non-fatal MI, non-fatal stroke, and CV death) with pioglitazone vs. placebo (HR=0.84, p=0.027), which Dr. Nissen characterized as “very similar to what was reported with empagliflozin and liraglutide.” More recently, the IRIS trial showed a statistically significant 24% risk reduction for stroke or MI associated with low-dose pioglitazone in patients without diabetes but exhibiting insulin resistance (p=0.007). We also can’t ignore the fact that pioglitazone is generic, and thus lower-cost vs. empagliflozin or liraglutide currently. In Dr. Nissen’s words, “pioglitazone ought to be considered.”

15. Update on Microvascular Complications with Dr. Grunberger

Past AACE president Dr. George Grunberger followed Dr. Nissen and shifted focus from macro to microvascular disease, where meticulous glucose control is the crux of preventing complications. In contrast to CV outcomes, which seem to depend on much more than A1c, Dr. Grunberger explained that “microvascular complications are a direct consequence of hyperglycemia.” In reviewing the latest guidelines on retinopathy, nephropathy, and neuropathy, he showed how time and time again optimizing blood glucose (as early as possible) is at the core. Only ~3% of people with newly-diagnosed diabetes present with nephropathy, for example, and ~50% of cases of diabetic nephropathy occur in people with diabetes duration ≥20 years, when longer exposure to high blood sugar has done its damage. Dr. Grunberger positioned the Steno-2 study as a paradigm of what diabetes care providers should be doing to minimize microvascular disease – treat glucose, lipids, blood pressure, and smoking all at once. We recently learned the 21-year results from Steno-2 at ENDO 2017, and indeed, multifactorial intervention was associated with a median increase in survival of 7.9 years and a 45% risk reduction for all-cause death (HR=0.55, p=0.005). Four-year data from Steno-2 was enough to demonstrate clinically-meaningful reductions in microvascular disease, according to Dr. Grunberger.

16. The Latest in Beta Cell Science: Beta Cell Heterogeneity and Islet Plasticity

Representing the basic science wing of the Barbara Davis Center, Dr. Lori Sussel provided a fascinating overview of recent research on beta cell physiology and its implications for the future of type 1 treatment and cures. The most exciting breakthroughs in her view are the discovery of multiple beta cell subtypes and an alternate theory for type 1 pathophysiology whereby rather than beta cell death, islets are losing their cellular identity (a process that may be reversible):

• Recent findings demonstrate that “not all beta cells are created equal.” Beta cells come in four subtypes, distinguished by different gene expression profiles and patterns of basal and glucose-stimulated insulin secretion – this finding was published in Nature Communications in July 2016, and beta cell heterogeneity was a key theme at Levine-Riggs 2017. Subsequent studies have revealed that people with type 1 diabetes have a different ratio of these beta cell subtypes vs. the general population. This raises important questions regarding whether an individual’s distribution of beta cell subtypes plays a causal role in determining type 1 diabetes risk, whether these different beta cell subpopulations serve unique functional roles, and whether certain beta cell subpopulations are more susceptible to the autoimmune response than others, Dr. Sussel explained. This research is still very early-stage, but it underscores that beta cells are more complex than was originally appreciated, and this must be factored into any beta cell replacement strategy.
• Dr. Sussel’s own research is focused on beta cell plasticity, the idea that different mutations can cause beta cells to differentiate into other cell types, or to produce other hormones besides insulin. Her team demonstrated that deletion of the Nkx2.2 gene decreases the glucose responsivity of beta cells by causing them to lose glucose transporters. Instead of dying, these mutated beta cells begin to change identity, producing somatostatin instead of insulin. Dr. Sussel pointed to evidence showing that a similar loss of beta cell characteristics is happening in people with diabetes. Investigation of pancreatic islets in both type 1 and type 2 diabetes shows increased somatostatin levels, an indication that beta cells are losing their normal insulin-secreting characteristics. The discovery of beta cell plasticity challenges the traditional model of type 1 diabetes in which the primary defect is beta cell death, suggesting instead that the loss of beta cell function may be due to the cells changing identity. The natural extension of this finding, according to Dr. Sussel, is a new approach to type 1 therapy involving the repair – rather than replacement – of these cells via gene therapy.  We’d be very excited for a therapy in this vein to next join the competitive landscape for type 1 diabetes therapies and cures.

Detailed Discussion and Commentary

Patient/Provider Question and Answer Panels

Pediatric Patient Panel

Drew; Gavin; Yanci

The patient/provider discussion panels are a beloved and highly-anticipated tradition at Keystone, and this year certainly didn’t disappoint. Several Barbara Davis Center patients took the stage to discuss their unique stories on life with diabetes. The pediatric panel, featuring 16 year-old Drew (who participated in the 670G pivotal studies), eight year-old Gavin (who was diagnosed with type 1 diabetes two years ago through the TEDDY study), and 18 year-old Yanci (who participates in the Barbara Davis Center’s Latino clinic), touched upon the impact of diabetes technology and the ways in which this chronic disease affects family dynamics and life at school. We capture the power of this session best through a collection of the most quotable quotes.

• On Having Diabetes at School
  
  • Gavin: “It’s kind of helpful that everyone in my class knows cause then they know that if I don’t hear a beep and they do then they can tell me about it.”
  • Drew: “I wouldn’t say that many people know about my diabetes – I don’t talk about it unless it’s absolutely needed. My close friends know as a safety net so that they can get help if I start acting strangely. I’m in high school and people aren’t the nicest. I’ll answer them honestly if they ask about my pump or meter, I’ll have a discussion, but I don’t like to advertise it.”
  • Yanci: “I just graduated from high school this May. Mostly everybody knew about my diabetes. When I was younger people made such a big deal of it because it was so rare in this small school. What I most remember is “do the shots hurt?” “Can you eat this?” I know good people, and I don’t have to hide my diabetes. What I didn’t like was the pity – “you’re so little, you don’t deserve this.” Half the school knew I had diabetes because I would go to the nurse to take my medicine. I’m pretty open about it.
• On Family Dynamics
  
  • Drew: “I think my parents did treat me differently for a year or two after being diagnosed. I would go to sleepovers before, and I didn’t go to one the week after I was diagnosed and I felt that pattern of worry for a year or two after being diagnosed […] After I was diagnosed it wasn’t something I really cared about and my mom handled most of it for me, but when I got on the 670G I started to pay attention and learn how to do everything myself for my treatment.”
  • Gavin’s mom: “We’ve had some struggles with site changes and we don’t want him to feel like he has no control over what’s going on in his life, so we explain and discuss that he has to have insulin. He decides if he wants to do a site change or change and go back to shots for a while. We want him to be in control and try to make sure we all understand what’s going on.”
  • Yanci: “The only time we’ve had conflicts are when I have to check my blood sugar. I used to try to ignore it and we would talk about it that later that day, and they had expected me to check 4-6 times and I would just say that I would but then not check. Since I was little I’ve done my own shots. Now they just have me check in front of my mom, or sister, or dad, or they follow me on their phones.”
• On New Technology
  
  • Yanci: “The Dexcom does make me more aware of my highs because it beeps every 5 minutes and is so annoying. So this helps me bring my blood sugars down much quicker. I can just look at my phone and take action really quick.”
  • Drew: “The biggest suggestion I can give from personal experience with the 670G is that it isn’t an end-all, be-all solution. It’s a treatment tool like any other. Just remember that. You need to use the tool correctly.”
  • Gavin’s mom: “Worry is all relative. After he was diagnosed, Dexcom – or a CGM in general – was on our mind. If I had my way we would be finger poking every five minutes for my peace of mind, so CGM was very important. It comes into perspective when we need to recalibrate and can’t get readings for 2 hours. We just had a sensor expire when he was pretty low, which brings to mind how much we rely on it when we aren’t in those good numbers.”

Practical Review of Therapeutic Continuous Glucose Monitoring in Clinical Practice and the Future of Dexcom Technology (sponsored by Dexcom)

CGM Use in Seniors: Clinical Outcomes and Reimbursement


Claudia Graham, PhD (Dexcom, San Diego, CA)

Dexcom Senior Vice President of Global Access Dr. Claudia Graham gave the update we’ve all been waiting for: It has taken a while to implement, but Medicare shipments are up and running…slowly! According to Dr. Graham, Dexcom now has “thousands and thousands and thousands of patients in the queue, and we haven’t even started advertising…bear with us, it’ll take several months to work through all of the patients in the pipeline.” This progress report means that,  following the FDA’s non-adjunctive label approval for G5 in December, Medicare’s subsequent benefit category determination (DME, Part B) in January, and the announcement that coverage will be provided in March, “final” local coverage decisions and administrative logistics have been worked out (Dexcom projected in May that this would happen before the end of 2Q17). The company is now in the process of shipping G5 receivers that last for three years (~$236-$277 each), along with G5 sensors, 90-day G5 transmitters, 60 test strips/month, Ascensia glucose meters (partnered earlier this month), lancets, and control solution in a monthly bundled subscription model – the only such model in Medicare besides oxygen! – for ~$240/month (patients pay ~$50/month out of pocket). Dr. Graham noted that 60 test strips will frequently not be enough for a 30-day month (~65-75 would be ideal), nor will four seven-day sensors – as the distributor, it is incumbent upon Dexcom to call the patient every 30 days to ensure that he/she has exactly what is needed for the month. (A reminder that Medicare opens up a new market, but also adds a lot of administrative hassle for Dexcom.) We love how hard Dexcom has pushed to get this done – management could easily say “our work is done” and leave the rest to suppliers/distributors, but they want to be absolutely certain that this goes well, given the arduous battle they fought to have this opportunity. Dr. Bruce Buckingham took to the microphone following Dr. Graham’s talk to assert that “Dr. Graham should be sainted.” Hear, hear! We look forward to hearing how many bundles have already been shipped and how many are in the queue.

• Dr. Graham got heated when describing “one of the dumbest thing [she’s] ever seen”: Even when Medicare patients use their receiver, they are not allowed to use their smart phone to share or view CGM data – if they use their cell phone, even as they use their receiver, they are ineligible for reimbursement. Accordingly, for the time being, the only way for Dexcom to ensure compliance is to turn off the functionality of the app for Medicare transmitters. This is something that Dexcom will be implementing in short order. This is extremely unfortunate to hear and something we had feared. She was visibly irritated by the policy.
  
  • Both Mr. Sayer and Dr. Graham understand the value of remote monitoring, and are working aggressively to remedy the current situation. Said Mr. Sayer in Q&A, “We have literally plans A, B, C, and D in development if they won’t let us use the phone, because we believe remote monitoring is so important. In a year, we won’t be talking about this issue.”
  • Submitted but yet-to-be published data shows the benefits of remote monitoring in seniors. The study found that over 43% of older adults using the Dexcom Share Cloud had at least one follower. Compared to those with no followers, these patients had slightly higher glucose values, but they used their sensors more regularly and had 14% fewer sensor glucose readings <70 mg/dl. Data like this underscores the insanity of depriving this population of remote monitoring.
• To Dexcom’s utter surprise and joy, former CMS Acting Administrator Mr. Andy Slavitt signed an administrative ruling (like a presidential executive order), one of only 18 in history, mandating that therapeutic CGM be covered. “It caught us off guard, we didn’t think it would happen that fast,” commented Dr. Graham. Once the administrative ruling, including a clause that everything needs to be included in a bundle, came through, Dexcom spent “about $8 billion” (jokingly) in consultants trying to figure out implementation.
  
  • Dr. Graham joked that Christmas 2016 was the worst in her life. Word of FDA approval of a non-adjunctive (insulin dosing) claim – the primary hurdle with Medicare – came in on December 20. Dr. Graham immediately tried to get in touch with CMS, but many of the key players were on vacation. All her team did for the next few days was try to reach CMS.  They knew they had a deadline, because on January 20, a new administration would take over.
• According to CEO Mr. Kevin Sayer, obtaining Medicare reimbursement and getting it up and running was “a bigger mess than [Dr. Graham] can possibly describe.” Part of the reason for the headache was the timing of Medicare approval – Mr. Sayer told investors on January 11^th at JP Morgan that it wouldn’t come until Spring 2018, and it came the very next day [commencing a months-long scramble at Dexcom]. Mr. Sayer also noted that Dexcom loses money when it ships the first bundle to patients, taking on a level of risk – the company is exploring how to manage that risk. See our 1Q17 coverage for more on how the business model changes under Medicare – less upfront and more ongoing revenue.
• Dexcom will make training as simple as possible for patients, with instruction on how to use the receiver, a special training video, and a phone call. If the patient can’t be reached by a phone call, then trainers will be sent to train the patient one-on-one. Wow! The training protocol is also expected to evolve – the kit that Dexcom ships in a month is expected to look very different from the current version.

Dexcom CGM Technology Now and in the Future


Peter Simpson (Dexcom, San Diego, CA) and Kevin Sayer

VP of Advanced Technology Peter Simpson wrapped up Dexcom’s information-packed dinner with a look at Dexcom’s pipeline, with some help from CEO Mr. Kevin Sayer. The speakers touched on G6 (pivotal trial wrapping up, FDA submission by the end of this September), first- and second-gen verily products, G5 Mobile, a hint at strategic partnerships, and more.

• Mr. Sayer confirmed that Dexcom plans to submit the impressively accurate G6 sensor by the end of September, keeping the planed 2018 launch timing on track. This sensor has one calibration/day after startup, 10-day wear, a 30% thinner wearable than G5, and no acetaminophen interference – a very meaningful improvement and excellent to see Dexcom keeping up with the timeline laid out in 1Q17! We assume this FDA review wouldn’t last more than a year, meaning a launch is actually possible in the first half of 2018. Mr. Sayer also shared that the pivotal trial (n=300+; 30,000+ matched pairs) will wrap up next week – we can’t wait to see how the data lines up with the pre-pivotal data shared at DTM (n=49) – an MARD of 8.1% on one calibration per day after startup (10% on day one!), with excellent potential for a no-calibration algorithm. Mr. Sayer stated that the leap from G5 to G6 will be almost as big as Seven+ to G4 – considering that many consider G4 to be the first viable consumer CGM, these are strong words! Wow!
• Mr. Simpson hinted that the first-gen CGM with Verily is very well into development, and we can expect to hear more about it soon, and the second-gen CGM with Verily is still in the feasibility stage (animal and human studies) with a lot of work to do. The timeline hasn’t changed for either version: Gen-1 is slated for a 2018 launch, and gen-2 for a 2020-2021 launch. As a reminder, the first-gen CGM leverages G6 sensing technology, but is smaller by 30%, single use disposable, zero-calibration, 14-day wear, and is inserted with an auto-applicator. The second-gen CGM is the diameter of a penny and the thickness of two, also has a single use disposable, zero-calibration, 14-day wear, has a new auto-deployment system, and features new electronics/sensing technology. In our view, the potential for these small, simple, and cost-effective sensors in type 1 and type 2 can’t be overstated. IN an update from the January JPM presentation, the investor deck now notes that the cost reduction is targeted for the second-gen Verily sensor.

• Dexcom is leasing a new facility in Flagstaff, Arizona, capable of manufacturing 50 million sensors per year – this is the first quantification we can recall hearing of this facility’s capacity. For context, Dexcom has only made 10-12 million sensors cumulatively to this point. This is a serious investment, as the company will pour $70 million into just the duct and air flow work – wow! This is another step on the quest to make CGM ubiquitous – Mr. Sayer stressed that he has every intention of making sensors simple and cost-effective enough to be used everywhere. This includes prediabetes, a ways down the road: “That’s a strategic issue we battle with internally with resources. This is a great tool, but the system we have today won’t fix it. We need the new products…Once we get through type 2 diabetes, I think you’ll see prediabetes after.”
• Dexcom’s post-market study of G5 Mobile for non-adjunctive use is “just ramping up now.” The comprehensive post-market study to ensure the safety of non-adjunctive use was condition of FDA approval – debated heavily at the July Advisory Committee meeting. We’re not sure what form this study will take, but have little doubt it will show safety and effectiveness (especially given the T1D Exchange REPLACE BG study).
• Mr. Simpson also spoke about G5 Mobile for Android and the Android Wear app (approved last month and already launched to somewhat mixed reviews), Apple opening up Bluetooth communication allowing direct flow of Dexcom CGM data to the upcoming WatchOS4 without a phone nearby (announced last month ahead of the Watch OS4 update this fall – we’re not sure when Dexcom will launch this, but it’s a highly compelling user experience win in our view), and the new high-reliability G5 touchscreen receiver (seen at ADA; it’s a form factor step back from the current receiver, but it has to be designed to last three years for the Medicare DME bundle). This company does not rest!

Keynote Address

Will Guidelines for Future Artificial Pancreas Approvals be Different?


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

FDA’s Dr. Courtney Lias proclaimed with an emphatic “Spoiler Alert!” that clinical trials are not chiefly responsible for slowing the commercialization of automated insulin delivery systems; rather, a lack of a framework for interoperability, and all that it entails, has slowed the field. We’re glad to see her continuing to hammer this message home, which we’ve heard from her at a number of conferences in the past year (e.g., ADA 2017, NIH Workshop, AADE 2016, ADA 2016). Number one on Dr. Lias’ list of things that bog down progress toward launch is contracting – she noted that some companies may have to work with three to four others to put together a closed loop system (e.g., CGM, pump, BGM), and the requisite effort put into determining and finalizing who is responsible for what in every scenario, what modifications need to be made in the future, postmarket responsibilities, and performance requirements are astronomical. In her experience, contracts can slow things down by months to years at a time: “I only have an impression, which is that there’s a vastly larger sum of time and money spent developing and maintaining contracts than spent dealing with us.” She’s under no illusion that we can expect for every company to develop or acquire their own sensor, pump, algorithm, and BGM (for sensor calibration, if necessary) – on the contrary, she would rather see the development of communication/interoperability specifications for automated insulin delivery components. Such a framework would make empirical design and testing – the process of developing a system and iterating through tweaks after small studies – a thing of the past. This kind of interoperability would enable safer, more nimble automated insulin delivery innovation, diminish regulatory and clinical study burden, and create a thriving ecosystem of component parts. In essence, devices must be optimized to work together, not in silos. The creation of transparent performance and communication specifications – something she asked for the community’s input on, perhaps with an eye on updating the 2012 AP guidance – “may allow for open development and open approval of automated insulin delivery systems. I can’t tell you how many submissions that would eliminate.” She later mentioned that FDA is working to facilitate open discussion toward solving these challenges – we look forward to this discussion moving to tangible specifics, as the vision and benefits seem obvious for all stakeholders. She concluded her talk by explaining the urgency for a mindset shift in artificial pancreas development: “Companies are struggling and peering over the edge of the cliff – there’s no market for pumps, and people are not using devices because of real resource challenges. We don’t want regulatory burden or contracts to be the reason they fall off the cliff. We need to bridge the gap and have resources be put to what matters the most.” So, in simple answer to her speaking charge: “Will guidelines for artificial pancreas approvals be different in the future? It has to be yes. We will make these changes because if we don’t, we won’t see these systems become available.”  Dr. Lias, our hats go off to you, and we hope interoperability moves forward in a meaningful way. By the way, please vote for Dr. Lias, Beck, and the FDA AP team here – they deserve to win a Service to America Medal!

• Dr. Lias emphasized that interoperability standards would not mandate open system design, transmissions of proprietary data, or prevent secure data access, but could involve the specification of defined mandatory and optional input and output data elements. The upside of interoperability is undeniable, as it could expand and simplify collaboration, streamline commercialization following component updates for single-manufacturer systems (e.g., if Medtronic wanted to use a next-gen CGM for 670G), and effectively promote innovation.
  
  • To give some examples, Dr. Lias said that the specifications CGM manufacturers currently use do not optimally enable interchangeable use because there is no way to know how different parameters of performance change across patient populations or device generations, and CGM developers don’t often think beyond MARD in understanding performance. Similarly, CGM manufacturers don’t currently consider how accurate a BGM needs to be for optimal performance – instead, they have to back-calculate, which creates a delay in selecting BGM partners (the threshold Dr. Lias frequently hears for BGM accuracy is “if it’s cleared,” but she’s the first to point out that there are sadly a lot of cleared meters that aren’t accurate).
  • She also mentioned that not all pumps are created equal in terms of automated insulin delivery adaptability – some have existing suspension/predictive algorithms (how will external algorithms interact with this software?), and the fundamental properties of pumps (dosing precision, bolus delivery speed, etc.) are not well understood.
• Dr. Lias acknowledged that there are different challenges for unique paths to market. Systems can be developed fully within one company, making things relatively easy (e.g., MiniMed 670G, though it relies on a partnered Ascensia BGM); a pump company can buy an algorithm and license a CGM (e.g., Insulet, Tandem); an academic investigator can use a commercially available CGM and pump (e.g., Dr. Roman Hovorka’s Cambridge group); or an algorithm can make a new pump and license the CGM (e.g., Beta Bionics, Bigfoot). Within many of these paths, there are unique challenges, but each must nail down BGM, CGM, pump, post-market responsibilities, performance requirements, clinical investigators, right of reference, and device modifications/updates. It’s clear how a hiccup in one of these areas can severely delay progress!
• Solving the challenges laid out by Dr. Lias will streamline clinical evaluation of new devices. Doing so could eliminate the need for new studies using different CGMs, as well as the need for clinical evaluation of modular or updated automated insulin delivery systems. At the end of the day, a more comprehensive understanding across AID algorithms and how they interact with component devices will reduce the need for clinical data to determine safety. The FDA device division, since the 2012 guidance, has been open to clinical endpoints for these systems beyond A1c, including time in range and improved quality of life.
  
  • “At the moment, Medtronic will be doing a huge post-market study, but I really don’t want years and years of postmarket studies to be the forever goal of all of the companies. The 1,000-patient, four-arm Medtronic study just began enrolling, is the biggest and most complicated run by the company to date (at least in diabetes), and won’t read out until 2021 according to ClinicalTrials.gov. While this study will be hugely informative (following just a three-month, single-arm pivotal trial powered for safety), particularly for payers, we’re glad to hear that not every company will be expected to pour millions of dollars into similar studies.
• In Q&A, Dr. Lias shared that her team, which regulates and evaluates the postmarket performance of BGM, CGM, sensor-enabled pumps, and automated insulin delivery devices is comprised of just 10 people! We knew FDA was under-resourced, but holy moly! Dr. Lias is very proud of her team, and rightfully so – they are doing a heroic job in a field that continues to move and innovate faster from a higher base.
  
  • According to Dr. Lias, her team receives 20,000-30,000 adverse event reports per year for diabetes devices. They don’t go through each by hand, but have developed internal tools to help look for spikes in signals, after which they can look for patterns of underlying cause and reach out to manufacturers as they see fit.
• Dr. Lias opened her talk with measured enthusiasm: “670G is here, and we’ve seen great outcomes, but a patient is not a patient is not a patient. What works for one won’t always meet the needs of others.” To illustrate her point, she displayed a photo of two young boys, her sons, sitting side-by-side in a shopping cart at Sam’s Club. While the younger looks content and delighted to be doing as his big brother does, the older has an expression that screams “Mom, get me out of here.” The message of the adorable example is clear: People with diabetes need choices!

Questions and Answers

Q: Recently, most artificial pancreas teams have published a combined statement about guidelines and reporting methods (Maahs et al., Diabetes Care 2016) – would this be helpful and have an impact in your analysis?

A: Standardized ways of analyzing device results do help – we’ve been able to work with manufacturers ahead of time to find out how we would want to see data analyzed, so it does help to have communication, and standardization will help. Do I think it will speed up the process as much as some of these things I’ve talked about? No. It’s more a matter of saving a couple of weeks of work, vs. months to years.

Dr. Jay Skyler (University of Miami, FL): You mentioned a number of different components, but not insulin. They’re getting faster – do you think automated insulin delivery development teams will need to compare their pumps with newer, faster insulins separately?

A: If you understand how your algorithm reacts to certain things, sometimes you might be able to reduce the types of necessary evaluations. For example, a fewer number of subjects in a study looking at PK/PD profile of insulin in a pump, if you understand some of the underlying properties. More intentional design, it’s hard to do up front, but the more we understand about algorithms and devices, the more we’ll be able to respond to different developments.

Q: How many people actually review all of these systems? And do you have a post-market mechanism to monitor repeated problems? And third, a lot of the studies involve almost what I would call “professional people” who are in the trials, professional trial applicants who are so used to the products – is there an effort to limit that?

A: We do talk to companies about expanding the type of patients in trials, and we do try to get people who are naïve to CGM and pumps, so we do try to get at that. At the moment, Medtronic will be doing a huge post-market study, but I really don’t want years and years of post-market studies to be the forever goal of all of the companies. In my group, we regulate BGM, CGM, sensor enabled pumps, and automated insulin delivery devices, and we’re about 10 people who work very hard – I’m very proud of them. We also do post-market evaluation of devices. We get a very high volume, and sadly, diabetes is a dangerous disease, and there are a number of event reports. We get 20,000-30,000 adverse event reports per year for those devices. We’ve developed internal tools to help us look for spikes in signals, and look for the pattern or underlying cause. So, we do see information about things that go wrong – sometimes the reports are useful. 

Q: Is there a way that patients on automated insulin delivery can give you direct feedback?

A: We do have the ability for people to call us, and for patients to send us reports of information. It’s called Med Watch.

Q: There are other medical devices – pacemakers, defibrillators – that I imagine to be as or more complex, but insulin pumps are patient-managed – is there a different level, stricter criteria of risk-benefit?

A: I don’t see that difference. Differences in devices are supposed to reflect the differences in the scenario. A surgeon puts in a pacemaker and the patient goes on his/her way, but people with diabetes, if they don’t have pumps, are at risk of severe hypoglycemia or long-term complications, so we try to titrate the level of scrutiny that’s appropriate for the risk. For type 1 and type 2 diabetes, you can tolerate a lot of risk in the device, because the baseline risk is so high. With a pacemaker or defibrillator, you don’t have all those companies working together. The problem here is the complexity of the system.

Q: What was the influence, positive or negative, of the so-called “hackers” of these devices?

A: One thing we learned from groups like that is the passion behind people who want to have better tools. We learn from them. When people really understand what they’re doing, really know what the algorithm is doing, there’s a little less risk. We do have some concern about people who have to have others set it up for them. We’re looking for a win-win solution – can patients get what they need safely? We’re trying to balance that, and it’s a difficult balance. We try to keep their perspective in mind.

Mr. Brandon Arbiter (Tidepool, San Francisco, CA): It’s so wonderful to see you at this conference and see the progress at FDA. One of my takeaways is that it sounds like FDA is not the thing slowing down development, but the cooperation between two to four companies that need to come together to produce one system. Any visibility into how much time in this system is spent behind closed doors arguing over contracts vs. how much time is actually the regulatory process?

A: You definitely would have to talk to the companies…I only have an impression, which is that there’s a vastly larger sum of time and money spent developing and maintaining contracts than spent dealing with us. The regulatory process right now is the way it is because of the way our framework of one company handling everything is working. If we had specifications, then we wouldn’t need this – each company could be held accountable individually, no need for contracts. Right now, regulatory is in the way because of the way the systems are being developed.

Dr. Satish Garg: What if down the road, oral drugs for type 2 got approved for type 1 diabetes, would that require additional studies for automated insulin delivery?

A: It’s hard to tell what studies would be necessary, a lot of it has to do with understanding. If the manufacturer knows that someone on this drug needs different precautions, that can be managed. The goal is to find most efficient path possible through good discussion and communication, and that’s the key to everything.

Diabetes in Special Populations – Pediatric

Early Diagnosis of T1D Through Antibody Screening

Andrea Steck, MD (Barbara Davis Center, Aurora, CO)

Dr. Andrea Steck delivered an engaging talk on the etiology of type 1 diabetes, offering a comprehensive review of what we know so far about the natural history and causes of type 1, and discussing current trials. People with type 1 diabetes generally progress from genetic risk to immune activation, immune response (single autoantibody development), and then through four stages of the disease: (i) ≥two autoantibodies with normal glucose tolerance, (ii) abnormal glucose tolerance, (iii) clinical diagnosis, and (iv) long-standing type 1 diabetes. ADA diagnostic criteria require fasting plasma glucose >126 mg/dl, 2-hour glucose >200 mg/dl, or A1c >6.5%. The natural history of diabetes can be monitored via biomarkers of immune response to beta cells (autoantibodies, t-cell response) and via biomarkers of metabolic changes (oral glucose tolerance test, A1c, CGM). In the Barbara Davis Center’s DAISY study, children age 0-11 who were either a first-degree relative of someone with type 1 diabetes or who screened positive for high-risk HLA antibodies were followed for development of islet autoantibodies. Participants were provided with education, glucose meters, and test strips. Among children who progressed to type 1 diabetes, those in the DAISY trial were diagnosed at a lower-onset A1c vs. other children in the community not enrolled in the study – 7.2% vs. 10.9%, respectively. Dr. Steck described this as an encouraging result, given that earlier intervention affords more opportunity to treat hyperglycemia and prevent long-term complications of diabetes. Another positive finding from DAISY was that many children were diagnosed outside the context of DKA or a hospitalization. Dr. Steck stated that in Colorado, 40%-50% of new-onset type 1 diabetes is diagnosed during a DKA episode – being able to avoid this is an important step in advancing screening and treatment for pediatric type 1 diabetes. Of particular note, DAISY used CGM to mark early glycemic changes. Dr. Steck shared one particularly striking anecdote of a DAISY participant who wore a CGM leading up to diabetes onset. Six months before onset, 14% of blood glucose readings were >140 mg/dl and A1c was 5.8%. Two months before onset, 44% of readings were >140 mg/dl and A1c was 6.8%. At onset, 53% of readings were >140 mg/dl and A1c was 6.1%. CGM data allowed this patient to begin insulin therapy with only Humalog (Lilly’s insulin lispro) at dinnertime, as his providers could see this is where he was having the highest blood glucose readings. Not that we needed more convincing, but this illustrates the critical value of CGM during all stages of diabetes to help people achieve their optimal glycemic control, to help personalize treatment strategies, and to give patients the best possible diabetes management for long-term prevention of complications.

• The TEDDY study, an international, multi-site trial that screened participants at birth for high-risk HLA genes and followed-up with antibody screening, also showed lower rates of DKA at time of diabetes diagnosis. Comparatively, rates of DKA at diagnosis for children <two years-old were significantly lower in TEDDY vs. SEARCH and the Swedish, Finnish, and German registries. This trend remained statistically significant for children <five years-old vs. SEARCH and the German registry. In an analysis of 43 TEDDY cases matched to community controls, TEDDY children had significantly lower A1c, age, rates of DKA, and hospitalization at diagnosis, and only 50% showed symptoms at diagnosis compared to almost 100% of community controls. C-peptide levels among TEDDY participants were also higher on average throughout the first year post-diagnosis. Ultimately, Dr. Steck suggested that earlier diagnosis is key to better long-term outcomes (starting with less DKA and fewer hospitalizations).

Diabetes Care Challenges in Minority Populations

Andrea Gonzalez, MD (Barbara Davis Center, Aurora, CO)

Dr. Andrea Gerard Gonzalez highlighted the Barbara Davis Center’s newly-established Latino Program for type 1 diabetes, which has improved diabetes outcomes in this harder-to-reach population. She set the stage with a description on the elevated risk for poor metabolic control and diabetes complications in minority populations – a phenomenon she noted is not well understood, but likely has a great deal to do with differences in health literacy and suboptimal communication between HCPs and people from different backgrounds. The Barbara Davis Center’s Latino Program was established to address this issue. More than 700 Latino-Hispanic families affected by type 1 diabetes receive care from this program, which features culturally-sensitive educational materials, Spanish-speaking HCPs, and regular group clinic appointments designed to build a strong sense of community and foster open discussion of diabetes care challenges. This approach works: Dr. Gonzalez shared that after two years, participants in the program experienced significant reductions in A1c: -1.2% in the first year (p=0.01) and -0.6% in the second year (p=0.007) from a baseline of 9%. The population also saw a significant increase in insulin pump use. These are impressive findings, and we would love to see more examples of this kind of culturally-sensitive and community-driven approach to diabetes care. Beyond reaching minority populations in this way, we agree with Dr. Gonzalez that it is crucial to further elucidate the causative mechanisms that place minority populations and groups from low socioeconomic status at higher risk of poor diabetes outcomes in the first place. This is certainly one of the more challenging systems-wide issues in diabetes care.  

-- by Ann Carracher, Abigail Dove, Brian Levine, Payal Marathe, Maeve Serino, Adam Brown, and Kelly Close