July 13-16, 2017; Keystone, CO; Day #1 Highlights – Draft

Executive Highlights

It’s that time of year again … the 27th annual Keystone conference (formally, Practical Ways to Achieve Targets in Diabetes Care) is underway! Our associate team joins ~550 total attendees traveling from 44 states to the nearly two-mile-high city of Keystone, Colorado (9,173 above sea level). This report details highlights from day #1 – five on diabetes technology and five on diabetes therapy.

We’ll have coverage from the rest of this meeting after the weekend. To whet your appetite, check out our Keystone 2017 conference preview: What’s coming up on the agenda for days #2-4? None other than Dr. Aaron Kowalski on the role of nonprofits in diabetes research (from the JDRF perspective), Dr. Steven Nissen sharing the cardiologist’s view on glycemic control, and talks by so many other renowned diabetes thought leaders.

Diabetes Technology Highlights

1. Medtronic Diabetes’ CMO Dr. Fran Kaufman hinted that there will be a next-gen 670G system with connectivity “in the future,” provided a comprehensive update on the 670G clinical trial program (7-13 year olds almost complete, 2-6 year olds started, 1,000-patient outcomes RCT has begun randomization), and broke out more granular data from the pivotal trial (prior CGM use vs. CGM naive, >50 years old). ~1,000 patients have used the MiniMed 670G at this point (including the pivotal + customer training phase), implying the Priority Access Program launch is still proceeding quite cautiously.

2. To kickoff this year’s Keystone Conference, Insulet hosted a panel discussion featuring Stanford’s legendary Dr. Bruce Buckingham, BDC’s Ms. Laurel Messer (University of Colorado, Denver, CO), Insulet’s Dr. Trang Ly, and our very own Mr. Adam Brown. The group explored the current state of closed loop systems, expectation management, and future must-haves. Naturally the conversation drifted to the MiniMed 670G and some of the pros/cons with the first-gen hybrid closed loop system. We also got a great update on Insulet’s trial progress.

3. Dr. Bruce Buckingham kicked off the first Keystone plenary with a whirlwind overview of hybrid closed loop systems, reiterating that nighttime is the low hanging fruit, severe hypoglycemia cost savings could be massive, and highlighting several recent team members that have recently gone to Google.  

4. Jaeb Center’s Dr. Roy Beck gave a balanced, very interesting overview of the commercial automated insulin delivery systems in development, pointed out strong data in Insulet’s recent Omnipod Horizon exercise study (n=12); called the speed of insulin the greatest barrier in the field (he was not too optimistic on Fiasp relative to something like Afrezza); and looked towards the future. 

5. Medtronic reps discussed updates to the CareLink platform and reminded attendees of the “suspend before low” (PLGS) function that is also embedded in the 670G.

Diabetes Therapy Highlights

6. Dr. John Buse presented a key argument for the use of A1c criteria to define prediabetes (5.7%-6.4%) rather than fasting plasma glucose (95-125 mg/dl) or an oral 2-hour glucose tolerance test (140-199 mg/dl): Notably, of these three metrics, A1c is the strongest predictor of CV disease.

7. ADA’s Chief Scientific, Medical, and Mission Officer Dr. William Cefalu spoke to the potential of metformin in diabetes prevention. He listed a number of unresolved issues in our understanding of prediabetes that should be considered as we move to implement largescale prevention efforts. To anchor his argument in favor of metformin, he pointed to 15-year results from the DPPOS (Diabetes Prevention Program Outcomes Study), showing that when A1c is used to define prediabetes rather than glucose metrics, metformin and lifestyle intervention appear to be equally effective, lowering type 2 diabetes incidence by 36% and 26%, respectively.

8. Dr. Richard Kahn pointed to the difficulty in preventing diabetes by lifestyle intervention: No program has shown the magnitude and long-term maintenance of weight loss needed to achieve clinically-meaningful health benefits, and results from RCTs rarely translate to the real world.

9. In a keynote address closing out the day’s agenda, Dr. Jay Skyler issued a compelling call for an aggressive combination therapy approach for the prevention of type 1 diabetes.

10. In an engaging Novo Nordisk-sponsored breakfast symposium, Dr. Anita Swamy (La Rabida Children’s Hospital, Chicago, IL) lamented the treat-to-fail paradigm of diabetes management, instead endorsing a DeFronzo-style initial combination therapy regimen of three to four agents that address as many members of the “ominous octet” as possible.

Diabetes Technology Highlights

1. Medtronic’s Dr. Fran Kaufman: Next-Gen 670G System with Connectivity Coming “In the Future”; 7-13 Year Old Study Almost Complete, 2-6 Year Olds Started; ~1,000 670G Users So Far

In an early morning talk, Medtronic Diabetes’ CMO Dr. Fran Kaufman hinted that there will be a next-gen 670G system with connectivity “in the future,” provided an update on the 670G clinical trial program, broke out more granular data from the 670G pivotal trial (prior CGM use vs. CGM naive, >50 years old), and reminded attendees of the next steps in Medtronic’s closed loop roadmap (including the addition of automatic correction boluses and eventually IBM Watson’s cognitive computing). ~1,000 patients have used the MiniMed 670G at this point, implying it has only reached a couple hundred additional people beyond the Customer Training Phase (~750 people) and pivotal study (124 participants). Many educators we talked to at this conference said the Priority Access Program rollout is going pretty slowly, in line with our expectations (one center has 200+ patients lined up to get on 670G). Dr. Kaufman shared that a next-gen 670G system with connectivity will be coming “in the future” – this is great to hear and key for staying ahead of the curve, since several other closed loop systems in development (e.g., Tandem, Insulet, Bigfoot, Beta Bionics) will have Bluetooth and smartphone display apps. Medtronic has never given specific timing on its Bluetooth-enabled pumps (Roche will provide the meter for them), though this feature should add remote monitoring and therefore pair nicely with pediatric indications. The company also will be in FDA discussions to obtain a non-adjunctive claim for the Guardian Sensor 3, a key feature for staying competitive with Dexcom (and possibly Abbott) and enabling automatic correction boluses with DreaMed. Dr. Kaufman also provided updates on the 1,000-patient 670G outcomes RCT, which has begun randomizing patients in the CSII cohort first, followed by the MDI cohort and then the sensor-augmented pump. Each will have a randomization for six months, followed by six months during which every patient uses 670G. People of all ages and durations of diabetes will be included in the trial, though the study cohort will be enriched for the Medicare population – Medtronic hopes this will help its case for reimbursement. On the pediatric side, the 670G study in 7-13 year-olds is almost complete (consistent safety profile to adults), while a study in 2-6 year-olds has already enrolled ten patients. Dr. Kaufman also presented sub-analyses and fascinating figures demonstrating the heterogeneity of nighttime insulin delivery – another reminder that the same basal rate every night with open loop makes staying in range extraordinarily difficult. 

• There were no tangible updates on the company’s next-gen closed loop roadmap, other than a continued commitment to add DreaMed’s Fuzzy Logic algorithm for automatic correction boluses and eventually IBM Watson to move “towards personalized closed loop.” The slide suggested that Medtronic will look at additional sensor input signals and sensor/infusion set failure detection – we’ve seen this for a while now but there has not been further granularity on launch timing or product form factor. Still, great to see Medtronic is not resting on 670G, since there are many companies on its heels.
• See our Detailed Discussion & Commentary section below for more from this packed talk.

2. Drs. Ly & Buckingham, Laurel Messer, and Adam Brown Discuss Closed Loop Successes and Pain Points, Expectation Management, Wish Lists

To kick off this year’s Keystone Conference, Insulet hosted a panel discussion featuring Stanford’s legendary Dr. Bruce Buckingham, BDC’s Ms. Laurel Messer (University of Colorado, Denver, CO), Insulet’s Dr. Trang Ly, and our very own Mr. Adam Brown. In front of ~100 attendees, the panel explored the current state of closed loop systems, expectation management, and future must-haves. Dr. Buckingham led off, calling this an “exciting, amazing, revolutionary, and historic time in terms of control and management. With devices, we’ve consistently introduced more burden, and this is the beginning of coming down the other side.” We don’t undervalue the significance of such a remark coming from a pioneer of the field who has as close to a birds-eye view as anyone we know! Even with such optimism, the panelists all acknowledged that there is no shortage of difficulties associated with the use of a new hybrid closed loop system – Ms. Messer went so far as to refer to the first month of using the 670G as “more work” and a “learning curve.” That is to say, overnight control is remarkable when patients are in Auto Mode, but they still need to calibrate and count carbs during the day, and 670G users still spend an average of ~20% of time in manual mode. It’s crucial that they retain their diabetes self-management skills for these nearly five hours a day when they won’t have the luxury of automated insulin delivery. Similarly, panelists mused on the fact that after using the system for a while, patients (and parents) are not used to adjusting basal rates anymore – all of the clinicians on stage emphasized the importance of holding onto this skill. Moreover, if patients drift out of Auto Mode, they revert to their previous pre-programmed basal rates in open loop – if these are not consistently updated (the 670G doesn’t), they may be inappropriate. To properly manage expectations, Dr. Buckingham recommended underselling hybrid closed loop and having patients be pleasantly surprised if and when they see positive outcomes. Adam highlighted the importance of managing patient behavior (e.g., to avoid manual dosing/hypoglycemia correcting on top of what the system is doing), the amazing overnight impact (even in those in tight control), and why automated insulin delivery may change the conversation around diabetes tech and even enable CGM uptake. See our Detailed Discussion and Commentary for the panelist’s closed loop wish-lists and our favorite quotable quotes!

• Prior to the panel, Dr. Buckingham gave a terrific overview of Insulet’s Omnipod Horizon clinical trial progress to date, which has tested the system in 82 patients (adults and pediatrics) over 3,384 hours. Dr. Buckingham highlighted improvements as studies have gone on, with mean glucose coming down from 161 mg/dl in the first study to 136 mg/dl in the most recent exercise study in adults; hypoglycemia at 2% or less; and time-in-range typically 70% or above (“these are the targets we want to see”). Dr. Buckingham was elated to see Insulet studying a broad range of ages, as pediatric approval from the start is a major goal. The team will start a five-day hotel study on Sunday in 8-11 year olds, offering a first look at adaptivity in an outpatient setting. He also noted the “Amazing” on body ecosystem approach, as users will remain in closed loop even when the Dash PDM is out of range (the pod has the algorithm built in and will directly talk to the Dexcom CGM transmitter). We agree this is truly compelling! Finally, he couldn’t help but comment on powerhouse Dr. Trang Ly’s commitment to testing this product widely and getting it to market quickly: “I don’t think anyone could have gotten the number of trials through FDA as she has.”

3. Dr. Buckingham: Hybrid Closed Loop’s Overnight Awesomeness (Especially Severe Hypoglycemia); Tech Giants Poaching Stanford Team

Dr. Bruce Buckingham kicked off the first Keystone plenary with a whirlwind overview of hybrid closed loop systems, reiterating that nighttime is the low hanging fruit and suggesting that integration into consumer devices “may be a giant that we just haven’t even seen yet.” On the former, he reminded attendees that in the DCCT, 55% of severe hypoglycemia occurred during sleep; in 65,000 days and nights of 670G wear (178 patient-years), there have been exactly zero severe hypoglycemic events in Auto Mode, when T1D Exchange data dictates that there would otherwise be at least 12. From a health economics standpoint, the CDC estimates that there are ~300,000 ER visits in adults each year for hypoglycemia – with an average cost of $1,161 per visit, hybrid closed loop systems could save the system quite a bit of money. With every system that’s been tested to date, Dr. Buckingham said, glucose values stay very close to target overnight, as there are few to no perturbations from food, exercise, stress, etc. Most payers have seen the value of the 670G (including Aetna, Cigna, Humana, UHC, Sutter Health), but Anthem Blue Cross Blue Shield has yet to offer coverage (Medtronic is reportedly confident they will give in). After bemoaning the all-too-common pain points of infusion sets/sites and proper CGM calibration technique – and pointing to Bigfoot, who just partnered with Abbott, as he proclaimed that it would be a huge plus to have a sensor in a system not requiring calibration – Dr. Buckingham turned to integration of closed loop systems into consumer devices (Apple, Android products). He was operating in purely speculative territory, but noted that Apple hired DIY automated insulin delivery app Loop developer Mr. Nate Racklyeft and one of Dr. Buckingham’s former fellows. Similarly, Google recently hired another two nurses who used to work with Dr. Buckingham. Of course, none of these individuals are allowed to speak about their work, but Dr. Buckingham suspects they may be working more diligently in diabetes. Of course, Apple announced on stage in June that the next version of Apple Watch will receive CGM data straight from Dexcom’s transmitter, and HealthKit will be updated this fall with basal/bolus doses. Meanwhile, Google (Verily) is developing next-gen bandage-like CGM with Dexcom, in addition to its Novartis and Sanofi (Onduo) partnerships (but no recent updates on the latter two). We wonder if either company has more in store in diabetes technology!

• Dr. Buckingham will be departing Keystone on Sunday at 5 am to be back in California to initiate a five-day AirBnb study of Insulet’s Omnipod Horizon closed loop system at 9 am. The study will enroll 6-12 year-olds and investigate use of Insulet’s system in an outpatient setting for the first time. In an inpatient feasibility study in this population (presented in an ADA poster), overall mean glucose was 157 mg/dl, with 70% time in range (70-180 mg/dl), 2% time <70 mg/dl, and 28% time >180 mg/dl – we wonder if these encouraging outcomes will translate to the AirBnb setting. Under the leadership of Dr. Buckingham’s disciple, Dr. Trang Ly, Insulet is cranking through clinical studies at a far faster clip than Tandem, Insulet, Beta Bionics, and others. We love seeing this commitment and Dr. Buckingham emphasized that the rubber really hits the road when systems move out of simulation modeling and into people.

4. Dr. Roy Beck on Commercial Automated Insulin Delivery Systems in Development, Speed of Insulin, Future Areas of Research

Jaeb Center’s Dr. Roy Beck gave a balanced overview of the commercial automated insulin delivery systems in development, noting that he uses Close Concerns’ AID competitive landscape to stay up to date on products/studies coming from Medtronic, Tandem, Bigfoot, Insulet, Diabeloop, Beta Bionics, Cellnovo, Animas, Cambridge, Roche, and Inreda – we’ll be updating this piece soon given this week’s Abbott/Bigfoot partnership (pivotal in 2018) and Beta Bionics’ delayed timing shared at FFL (pivotal studies to start in 2H18).

• Dr. Beck was impressed with a recent Insulet Omnipod Horizon exercise study (n=12 adults), noting strong 85% time in range and mean glucose of 136 mg/dl. For this particular study, time below 70 mg/dl was just 1.5% and time in range overnight was a whopping 94%. Of course, these results look better than prior studies since they occurred in the exercise setting, but nonetheless is shows Insulet is challenging its algorithm in important settings. (We also saw this topline data at the company’s product theater at ADA.)
• On a more somber note, Dr. Beck noted that the Animas’ artificial pancreas program is temporarily on hold “until strategic options for the company have been determined.” He considers the Hypoglycemia-Hyperglycemia Minimizer product with Dexcom quite “promising” (it’s been in development for over five years) and hopes it will eventually push through. We’ll listen for updates on J&J’s diabetes technology business on next week’s 2Q17 earnings call.
• Dr. Beck claimed that the greatest barrier to closed loop systems is insulin delivery, citing the need for improved rapidity of prandial insulin onset. Current insulin kinetics make meal announcement and manual insulin bolusing necessary for optimal time-in-range. Dr. Beck noted that, to date, companies developing faster injected insulins have achieved only modest outcomes that are still far from what is needed. Indeed, he highlighted Novo Nordisk’s upcoming FIasp (under FDA review, with a decision expected in 3Q17), comparing the PK/PD to ultra-fast Afrezza and current rapid-acting analogs – on that scale, FIasp basically overlapped with current analogs and looked far slower than Afrezza. He believes that the “solution” may not be possible with subcutaneous insulin delivery; rather, intraperitoneal delivery has more potential. With IP delivery, the onset and delivery of insulin are far more rapid, as is the offset, reducing the likelihood of hypoglycemia. While IP delivery is certainly not without its risks and extreme expense at this point (detailed at length at April’s JDRF-Helmsley Charitable Trust Closed Loop Intra-Peritoneal Infusion Workshop), Dr. Beck discussed the demonstrated benefits ranging from tighter glycemic control, to restoration of glucagon response, to higher portal plasma insulin levels. Quoting Montpellier’s Dr. Eric Renard, Dr. Beck observed that patients who try IP delivery never want to go back to subcutaneous. Still, he cautioned that there is a ton of work to be done in the field, demanding some serious research commitment in both time and dollars.
• To wrap up the discussion, Dr. Beck highlighted what he considers to be the future directions of closed loop systems: He finds sensors to be “pretty good already” but predicts stronger accuracy during times of rapid change (key for systems responding to both post-meal and exercise changes) , longer wear duration, less calibration, reduced lag time, miniaturization, and implantable models on the horizon. Other future improvements include utilization of machine learning to personalize algorithms (we see very high potential here), multi-hormone systems with pramlintide, and multiple sensor inputs (such as exercise).

5. Medtronic CareLink Updates and 670G “Suspend Before Low” Reminder

Medtronic’s Ms. Nancy Lea discussed updates to the CareLink data management platform, which includes three new reports: assessment and progress, weekly review, and the meal bolus wizard, all of which can be generated regardless of the amount of data collected. The assessment and progress report includes percentile comparisons using current and historical data, time in range, exit information, and a slew of statistics such as time in auto mode, sensor wear, daily dose information, and average blood glucose. The weekly review report provides much of this information but over the course of a week, diving in a bit more with an active insulin curve and dated pump suspension information. Lastly, the meal bolus wizard facilitates observation of trends by collecting data at three time points: one hour pre-bolus, time of bolus, and three hours post-bolus. We find these views to be very user-friendly and a welcome addition to the platform, and we appreciate Medtronic’s commitment to listening to providers to incorporate the most helpful features. We were also reminded that the MiniMed 670G comes with a “suspend before low” (PLGS) feature, which was highlighted by Medtronic’s Ms. Kelsey Winger – this is recommended by Medtronic when not in auto mode. The setting induces pump suspension when sensor glucose is at or within 70 mg/dl of the set low limit and is predicted to fall an additional 50 mg/dl within 30 minutes. We especially like how the feature occurs without alerts (for a max of 2 hours), freeing up some much-needed headspace and correcting a huge criticism of the MiniMed 530G. We assume most patients will use Auto Mode and then have this feature turned on for Manual Mode – the predictive suspend in the MiniMed 640G has been very well received in Europe, though Medtronic leapfrogged this in the US to get straight to the MiniMed 670G. Ms. Winger emphasized that the 670G is designed to gradually adjust blood glucose and to avoid rollercoasters, but is not capable of correcting missed meals or boluses.

Diabetes Therapy Highlights

6. Dr. Buse on Catching CV Risk Early, in Prediabetes

Dr. John Buse presented a key argument for the use of A1c criteria to define prediabetes (5.7%-6.4%) rather than fasting plasma glucose (95-125 mg/dl) or an oral 2-hour glucose tolerance test (140-199 mg/dl): Of these three metrics, A1c is the strongest predictor of CV disease. According to one meta-analysis, there was no association between fasting plasma glucose in the 95-125 mg/dl range and clinical complications, or between 2-hour glucose in the 140-199 mg/dl range and clinical complications, but there was a significant correlation between A1c in the 5.7%-6.4% range and CV events. More specifically, this A1c-based prediabetes diagnosis led to a 2x risk of CV disease. All-cause mortality was also significantly increased for people with an A1c in the prediabetes range vs. those with an A1c <5.7%. Dr. Buse underscored that choosing the right screening tool will be essential in population-level diabetes prevention – and he added that A1c seems best. The purpose of identifying prediabetes is to delay or prevent long-term diabetes complications, especially CV complications, given that CV disease is the leading cause of death for a diabetes patient population. We agree wholeheartedly with this sentiment, since the growing emphasis on prediabetes is not merely to predict type 2 onset, but to allow for earlier intervention and improved health outcomes. CV risk reduction is becoming a more central, critical component of diabetes management, and this extends to prediabetes management. That fasting plasma glucose and 2-hour glucose don’t show a significant correlation with future CV events makes them less useful diagnostic tools in a clinical context. Dr. Buse highlighted that A1c is most appropriate in real-world clinical settings because it can be done in a non-fasting state. It also catches people at risk for diabetes in their 20s and 30s, earlier in the history of disease progression: Individuals in this age group aren’t necessarily thinking about chronic disease prevention, but if they schedule a doctor’s visit for a broken ankle, they’ll likely be more amenable to an A1c screen vs. a fasting glucose test. Moreover, Dr. Buse mentioned that 2-hour glucose tolerance tests are rather cumbersome for most medical settings. His talk (in conjunction with Dr. William Cefalu’s and Dr. Richard Kahn’s, also covered in this report) revealed lingering questions that may be standing in the way of maximally effective diabetes prevention efforts. At the top of this list: the field needs consensus on a prediabetes definition, and we hope to see agreement on a diagnosis that correlates with outcomes. We also imagine this consensus would further initiatives to get FDA to recognize prediabetes as a disease, so that therapies like metformin can be indicated for prediabetes and prescribed with an aim toward prevention.

7. Dr. Cefalu Advocates for Metformin in Diabetes Prevention

ADA’s Chief Scientific, Medical, and Mission Officer Dr. William Cefalu spoke to the potential of metformin in diabetes prevention. He listed a number of unresolved issues in our understanding of prediabetes that should be considered as we move to implement largescale prevention efforts. To anchor his argument in favor of metformin, he pointed to 15-year results from the DPPOS (Diabetes Prevention Program Outcomes Study), showing that when A1c is used to define prediabetes rather than glucose metrics, metformin and lifestyle intervention appear to be equally effective, lowering type 2 diabetes incidence by 36% and 26%, respectively. In contrast, the fasting plasma glucose-based diagnosis showed lifestyle intervention to be more effective, reducing incidence by 27% after 15 years vs. 18% with metformin, but accumulating evidence suggests that A1c may be a better catch-all for people at high risk for new-onset type 2 diabetes. Dr. Cefalu described metformin as a cost-effective solution as well, echoing Dr. John Buse’s remarks from WCPD 2016 – it would cost a little over $4 billion to treat the entire US prediabetes population with metformin vs. >$1.2 trillion to treat this group with high-dose liraglutide (Novo Nordisk’s Saxenda) and ~$1.4 trillion to treat this group with intensive lifestyle intervention as provided by the DPP (although lower-cost iterations of the program are also proving effective). He concluded that we should invest in both lifestyle modification strategies and metformin as diabetes prevention tactics. He also reviewed SCALE data on Saxenda and ACT NOW data on TZD pioglitazone for diabetes prevention, explaining that while both agents effectively delayed type 2 diabetes onset, each therapy came with side-effects that should be carefully considered in making treatment decisions for the individual patient. Pioglitazone, for example, was associated with weight gain and edema, which could be counterproductive to the goal of improving overall health outcomes. In concluding his presentation, Dr. Cefalu suggested that there is little disagreement in the diabetes field that metformin should be used in some stage of prediabetes, alongside strategies to control other risk factors like lipids, blood pressure, and smoking. This was music to our ears, and we can’t wait to see this take root in real-world clinical practice, with HCPs screening for prediabetes and prescribing metformin where indicated. That said, unresolved issues remain in disseminating pharmacotherapy far-and-wide for diabetes prevention. Dr. Cefalu elaborated on a few:

• How should we define prediabetes? Recent evidence favors A1c over fasting plasma glucose or an oral glucose tolerance test, but until there is standardization of diagnosis, it may be difficult to get prediabetes recognized as a disease so that FDA and other regulatory bodies can approve therapies for this indication. Dr. Cefalu also raised the question of how to efficiently screen for prediabetes in under-resourced areas.
• Can pharmacological agents maintain effectiveness over the long term? In general, the timescale of RCTs is an issue in prevention research – these studies require very long-term follow-up to determine if any given intervention is truly preventing disease onset. On the other hand, there is certainly value even in delaying diabetes onset, given the lower rates of micro and macrovascular complications that result.
• Is our goal in prevention to treat hyperglycemia, or to alter the natural history of disease? The latter calls for pharmacotherapies that intercept the process of beta cell dysfunction (which is a commonly-cited argument in favor of TZDs).
• How do we implement lifestyle interventions over long-term periods? Dr. Cefalu referred to Dr. Richard Kahn’s presentation (coverage below), highlighting that we need to help patients lose weight and keep it off in order for lifestyle strategies like those in the DPP to be truly effective.

8. Dr. Kahn’s Pessimistic Take on Lifestyle Intervention for Prediabetes

Dr. Richard Kahn pointed to the difficulty in preventing diabetes by lifestyle intervention: No clinical trial has shown the magnitude and long-term maintenance of weight loss needed to achieve clinically-meaningful health benefits, and results from RCTs rarely translate to the real world. Community programs have not replicated the results of RCTs, and there’s limited evidence on how delaying diabetes onset prevents adverse outcomes. To-date, the most successful program aimed at diabetes prevention according to Dr. Kahn has been the Diabetes Prevention Program (DPP), in which participants achieved a 58% reduction in incidence after 2.8 years, due almost entirely to weight loss. However, he pointed out that this doesn’t mean we can truly “prevent” diabetes, since if the DPP were ended years later, the proportion of cases prevented would be far less. Indeed, longer-term follow-up out to 15 years found incidence reduced by a smaller 27%, and the DPP investigators reported that lifestyle intervention only delayed the onset of diabetes by a mean four years. Moreover, Dr. Kahn explained that in the first year of the DPP, the lifestyle arm achieved a mean weight loss of ~7.4%, and from that low point, weight regain occurred throughout the trial even while the intervention was still in full-force. Dr. Kahn displayed similar data from other lifestyle trials, ultimately emphasizing that weight loss is incredibly difficult to maintain, due to the seeming imperative of the human body to return to its highest-ever weight. Dr. Kahn also mentioned that the lifestyle intervention in the DPP consisted of many strategies to help participants lose weight, such as frequent coaching, free meal replacements and health club memberships,  frequent contact with a variety of support staff, personal trainers and even other material incentives, all of which are unrealistic for real world programs due to cost and limited resources. He presented a host of evidence to show that community programs have not come close to obtaining the weight loss reported in the DPP. He further argued that since no trial has shown any profound clinical benefit, we should not expect such benefits from real-world interventions that achieve lower levels of weight loss. Finally, he showed data to suggest that one has to lose ~4%-5% of starting body weight and keep it off for many years in order to see a delay in diabetes onset, for whatever that delay is worth (in his view, there is no clinical benefit). Dr. Kahn maintained that we should still encourage weight loss, because there are a small minority of individuals (~5%) who can lose a substantial amount of weight and keep it off. Unfortunately, the field currently lacks ways to identify these ideal candidates for lifestyle intervention. There’s no arguing that the massive prediabetes epidemic is intimidating (86 million people in the US alone), and that large-scale diabetes prevention is a tall task, but we maintain a more optimistic view on lifestyle intervention. Medicare is slated to begin DPP reimbursement in 2018, and the AMA has called upon private payers to cover the lifestyle program as well. The CDC is beginning to recognize virtual DPP platforms, which will make the program more scalable and bring down cost. We do think it’s important to note the value in delaying new-onset diabetes for any number of years – this lowers cost on the healthcare system (and research has projected cost-savings with a DPP-like lifestyle intervention). We’ll be interested to hear what Dr. Ann Albright (Division of Diabetes Translation, CDC, Atlanta, GA) has to say about implantation of the NDPP at AADE 2017.

9. Dr. Skyler Calls for Aggressive Combination Therapy Approach to Type 1 Diabetes Prevention

In a keynote address closing out the day’s agenda, Dr. Jay Skyler issued a compelling call for an aggressive combination therapy approach for the prevention of type 1 diabetes. Specifically, he proposed that a successful approach might combine preservation of beta cell health with simultaneous action on multiple components of the immune system: anti-inflammatory therapy targeting innate immunity, immunomodulatory therapy targeting adaptive immunity, therapy to drive regulatory immunity, and antigen-based therapy directly regulating beta cell immunity. To set the stage, Dr. Skyler gave an overview of the rather pessimistic history of type 1 diabetes prevention trials: the literature on primary prevention (preventing disease onset), secondary prevention (halting disease progression after the onset of autoimmunity), and tertiary prevention (arresting the progression of disease symptoms) all shows a very limited impact for the majority of type 1 prevention interventions. Why have such a wide variety of prevention approaches – from nicotinamide to oral insulin to GAD vaccine – all failed to show a meaningful impact? Dr. Skyler suggested that the answer to this question lies in the complexity of type 1 diabetes pathogenesis. He argued that so many biochemical pathways contribute to this process that salvaging the beta cell will require intervention in multiple targets simultaneously, hence his combination therapy proposal. Since Dr. Skyler’s first espousal of this view in a 2015 Diabetes Care article, two trials have been initiated to investigate the efficacy of this aggressive combination therapy approach. The first is Dr. Skyler’s own DIPIT (Diabetes Islet Preservation Immune Treatment) trial, which utilizes anti-TNF as an anti-inflammatory agent, ATG (anti-thymocyte globulin) as an immunomodulator, interleukin-2 and GCSF (granulocyte colony stimulating factor) to drive regulatory T-cell action, and the GLP-1 agonist exenatide to preserve beta cell health. Similarly, a study led by the University of Alberta’s Dr. James Shapiro – posted on ClinicalTrials.gov only last month – utilizes etanercept and anakinra as anti-inflammatory agents, alemtuzumab as an immunomodulator, plerixafor as a regulatory immune agent (which mobilizes stem cells for beta cell repair), and the GLP-1 agonist liraglutide to preserve beta cell health. While both these investigations pique our interest, as always with type 1 cures, we’re careful to manage our expectations. Prevention studies need to be quite long to show convincing efficacy, so this combination therapy cocktail is still far from being readily available. Moreover, “cure” is loosely-defined in the type 1 context and means different things to different people. For some, independence from daily insulin or prevention of diabetes complications (including hypoglycemia) would represent a functional cure, while others view a cure more rigorously as the prevention of beta cell destruction.  

10. Dr. Anita Swamy Swaps Treat-to-Fail Paradigm for Early Combination Therapy

In an engaging Novo Nordisk-sponsored breakfast symposium, Dr. Anita Swamy (La Rabida Children’s Hospital, Chicago, IL) lamented the treat-to-fail paradigm of diabetes management, instead endorsing a DeFronzo-style initial combination therapy regimen of three to four agents that address as many members of the “ominous octet” as possible. She began her remarks on a personal note: Half of her father’s 12 siblings have type 2 diabetes, as do all 10 of her mother’s siblings, and Dr. Swamy herself has prediabetes. This background explains why she has dedicated her endocrinology practice exclusively to diabetes. Dr. Swamy described how the treat-to-fail paradigm led to inadequate glycemic control for so many of her relatives, being prescribed a mild treatment regimen only to have it repeatedly intensified as their glucose control and diabetes complications inevitably worsened. Instead, from the time of diagnosis, Dr. Swamy advises her patients that “this isn’t your grandma’s diabetes.” She prescribes multiple therapies right off the bat to give patients their best chance at optimal health outcomes. Dr. Swamy noted that in her experience, this more aggressive approach leads to a more positive mindset in her patients, who avoid the disheartening cycle of repeatedly added therapies after repeatedly worsening glycemia. She highlighted two therapies in particular that have helped improve patient quality of life in her practice: Novo Nordisk’s GLP-1 agonist Victoza (liraglutide) and next-generation basal insulin Tresiba (insulin degludec). She characterized Victoza as offering “more bang for your buck” compared to other diabetes drugs –appetite suppression, weight loss, decreased blood pressure and fasting plasma triglycerides, decreased insulin resistance, and preservation of beta cell function – a reason she generally prefers GLP-1 agonists over SGLT-2 inhibitors. On Tresiba, Dr. Swamy emphasized the product’s flexible dosing, noting how this makes adherence easier for people who find it challenging to take their insulin at the same time every day (college students, shift workers, busy parents, etc.). She closed by urging the audience, “rather than throwing your patients the most common drug, think about what drugs would benefit them the most” – both in terms of ease of use and addressing the underlying pathophysiology of diabetes.

• Although not mentioned in Dr. Swamy’s prepared remarks, we also add that recent CVOT data provides additional evidence for the benefits of these agents on the diabetes complications front. The LEADER trial, first presented at ADA 2016, demonstrated that Victoza can reduce the risk of CV events in high-risk patients (the basis of a proposed label update for the drug), and the DEVOTE trial, hot-off-the-press from ADA 2017, demonstrated significant reductions in severe hypoglycemia with Tresiba vs. Sanofi’s Lantus (insulin glargine). The importance of these outcomes beyond A1c for patient quality of life and long-term prevention of diabetes complications can’t be overemphasized.

Detailed Discussion and Commentary

Hybrid Closed Loop: Advancing Your Practical Knowledge of the MiniMed 670G System (Sponsored by Medtronic)

Clinical Evidence

Fran Kaufman, MD (CMO, Medtronic Diabetes, Northridge, CA)

Medtronic Diabetes’ CMO Dr. Fran Kaufman hinted that there will be a next-gen 670G system with connectivity “in the future,” provided a comprehensive update on the 670G clinical trial program (7-13 year olds almost complete, 2-6 year olds started, 1,000-patient outcomes RCT has begun randomization), and broke out more granular data from the pivotal trial (prior CGM use vs. CGM naive, >50 years old). ~1,000 patients have used the MiniMed 670G at this point, implying the Priority Access Program launch to 630G users is still proceeding quite cautiously.

• A next-generation Medtronic closed loop system with connectivity will be coming “soon.” We knew connectivity was coming ever since Medtronic signed a deal with Roche to provide Bluetooth-enabled BGM for future Medtronic pumps (announced in February). Dr. Kaufman immediately followed her comment with a precautionary “but soon could mean anything, right?” On a recent Roche Accu-Chek Guide webinar, remarks implied that integration with a MiniMed pump could occur as soon as the middle of 2018, but this was not confirmed by Medtronic. We wonder if this next-gen pump will simply be a 670G with Bluetooth built in, if Dr. Kaufman was referring to the advance hybrid closed loop with DreaMed, or something in between. Whatever form it takes, this is a huge win – many view connectivity and remote monitoring as must-haves, especially in pediatric populations. It will also help Medtronic staying ahead of the curve, since several other closed loop systems in development (e.g., Tandem, Insulet, Bigfoot, Beta Bionics) plan to have Bluetooth and smartphone display apps.
• Medtronic’s expansive 670G clinical trial program includes two pediatric studies (7-13 years old and 2-6 years old) and a 1,000-patient, randomized post-approval study.
  
  • The 7-13 year-old study is almost complete and data will be ready “soon.” This is on track with timing for a mid-summer readout shared by Dr. Robert Vigersky at ENDO – he also suggested that data would be submitted to FDA by the end of the year.  The trial enrolled 110 patients, 64 of which are in a continued access phase. Dr. Kaufman didn’t share glycemic outcomes data, but she did say that in a total of 16,200 patient-days (36 patient-years), the safety profile is equivalent to that seen in previous studies.
  • The 1,000-patient RCT – “the biggest and most complicated trial [Medtronic] has ever done” – has begun the process of randomization. As a reminder, this multi-national trial (US, Canada, and the EU) will randomize patients to 670G, CSII, MDI, or SAP for six months, followed by six months during which every patient uses 670G. People of all ages and durations of diabetes will be included in the trial, though the study cohort will be enriched  for the Medicare population – Medtronic hopes this will help its case for reimbursement. The only people excluded from participation are those on dialysis, those who have had a severe cardiac event in the past year, and those recently diagnosed who are still in the honeymoon phase. According to CT.gov, this trial isn’t expected to read out until mid-to-late 2021. We hope interim readouts are possible, since the technology will have moved a long way four years from now!
  • There are currently 10 patients enrolled in the 2-6 year-old 670G trial. The plan is to enroll 25 patients in the 5-6 year-old group and 25 in the 2-4 year old group. Great to see this moving quickly, since many parents would probably welcome a better night of sleep. The system is currently approved for ages 14+, and requires patients to have a minimum daily dose of eight units, though we assume this is because it has not been tested in these populations; that said, perhaps some modifications will need to be made for a safer device in these super tough populations.
  • A Stanford study examining the clinical startup of 670G is listed on clinicaltrials.gov as “not yet recruiting.” Investigators aim to track 100 patients’ time in auto mode and time in range over a one year period to assess the clinical approach to starting the system. This could lead to improved guidelines for initiation. We love this, since it’s a major concern of several providers we’ve talked to.
• Dr. Kaufman shared that ~1,000 patients have now used 670G, implying it has only reached a couple hundred additional people beyond the Customer Training Phase (~750 people) and pivotal study (124 participants). Members of the pivotal trial continued access phase have now given Medtronic 57,000 patient-days (158 patient-years) worth of trial and real-world date to learn from. The safety profile in all patients is strong, hashing out to roughly one to two episodes of severe hypoglycemia per 100 patient years (that’s just two to four in the entire continued access phase population). Dr. Kaufman pointed out that most episodes have occurred in open loop (e.g., bolusing for a meal and forgetting to eat), but Medtronic still tracks and reports these outcomes because they are part of the system – even the most conscientious patients don’t spend 24 hours a day in auto mode. 
• Medtronic remains in talks with FDA regarding a non-adjunctive (replacement) claim for the Guardian Sensor 3 CGM, which would eliminate the need to check blood glucose with a fingerstick before meals when using the 670G system. This is excellent to hear, though with a recommended 2-4 calibrations per day, we wonder if the sensor’s accuracy on fewer calibrations will need to improve before this claim is granted. As a reminder, Guardian Sensor 3 has an adjunctive claim in the 670G, since it modifies basal but is not approved for taking boluses. This claim would (i) reduce user burden (though presumably not every patient performs fingersticks before every single meal); (ii) might open the door for Medicare coverage for the whole system; and (iii) knock down a key step for automatic correction boluses in Medtronic’s planned Advanced Hybrid Closed Loop.
• The immediate next-generation, dubbed “advanced hybrid closed loop” incorporates DreaMed’s Fuzzy Logic algorithm to assist with post-prandial corrections. It will be pitted against the 670G in the NIH-funded FLAIR trial beginning in late 2017.
• In Dr. Kaufman’s view, competition from the field is “exciting,” as it keeps everyone striving to do better. She went on to add that competition helps payers realize that the field is “real.” The 670G has been received well by most payers – she characterized the one holdout (Anthem) as “problematic,” though hoping that they will eventually get on board.
• Much of the presentation was devoted to new sub-analyses from the pivotal study, demonstrating 670G’s benefits, especially overnight. The most compelling analysis was that comparing the 670G outcomes of CGM-naïve patients (n=78) with experienced users (n=78). Experts have voiced that patients with difficulty wearing CGM may not have the best 670G experiences, but this analysis found no difference in time in range between the two groups during the pivotal study. The data wasn’t shared, but Dr. Kaufman added that there is similarly no impact of prior pump use (and some believe it may be easier to initiate pump-naïve patients because there’s less need to re-teach bad habits).
  
  • Another sub-analysis looked at the >50 year-old pivotal trial cohort (n=31): This group saw significant one-hour per day improvements in time between 70-180 (run-in: 71.6%; study: 75.7%), an ~50% reduction in time ≤70 mg/dl (run-in: 6.5%; study: 3.0%), an ~50% reduction in time ≤50 mg/dl (run-in: 1.1%; study: 0.5%), a 0.3% drop in A1c (baseline: 7.2%), and improvements in glycemic variability. We assume this data set is too small for Medtronic to go for Medicare coverage before the 1,000-patient RCT.
  • A fascinating slide quantified the variability in nighttime insulin delivery in the pivotal trial that we have qualitatively heard so much about. Unsurprisingly, SD and CV of insulin delivery skyrocketed in auto mode (each increasing by a multiplier of ~eight compared to the run-in period!), and the duration of no-insulin delivery approached two hours/night in auto mode vs. just four minutes during the run-in. The pie chart on the left really helps to visualize the heterogeneity of delivery – ranging from 0-100% of max Auto Mode Delivery, and this is without a change in the total nightly dose! Concluded Dr. Kaufman, “we can’t possibly recapitulate this ourselves.” Time in range improved by an impressive nine percentage points over night (67%->75%), hypoglycemia was halved (6% to 3%), and hyperglycemia declined from 27% to 22%. A particularly strong benefit was observed between 3 am and 6 am (prime dawn phenomenon time).
    • We hope future cuts of the date begin using the consensus time-in-range endpoints of <70 mg/dl and 70-180 mg/dl. Though 71-180 mg/dl and <70 mg/dl is not a big difference, having consistency would be ideal.

• The Medtronic team goes into CareLink every week for signals that 670G behaves differently in the real world from how it did in the pivotal study. Based on the most up-to-date data (below), time in auto mode, time in ranges, and mean sensor glucose values still strongly resemble that seen in the pivotal trial. If anything, as operators gain more experience, it may be that time in target is actually creeping up (72% in the pivotal, 74% in data from May, and 76% now). 

• Dr. Kaufman shared that, according to CareLink, patients are only really using the 150 mg/dl temporary exercise target ~2% of the time (~50 minutes per day, assuming everyone uses it with equal frequency). This isn’t surprising to us, since getting benefit out of this temp target during exercise requires planning ahead – setting the temp target at least an hour before exercise, which is hard to remember to do in real life. We assume many are simply eating before exercise and letting Auto Mode run, or perhaps others are going back to Manual Mode.
• Dr. Kaufman shared strong views on terminology related to the automation of insulin delivery: “Artificial pancreas is what we all dreamed of, but no one has the same vision of what an artificial pancreas is. I tell people on a plane I work on artificial pancreas, and they say, ‘Oh really, does that cure cancer?’ No, it’s something that has virtually no user interaction, but that’s not where we are, and we don’t want patients to feel over promised with these current systems. We still use a lot of different terms, and I think we’ve been able to confuse a lot of different people with them.” We think Medtronic has done a good job of using hybrid closed loop and being very clear about what the 670G does and does not do; popular media and JDRF, on the other hand, continue to use “artificial pancreas.”
• Medtronic still plans to use IBM Watson’s cognitive computing in future closed loop system (two iterations down the line). According to a briefly-displayed roadmap slide (below), this system will move Medtronic “towards personalized closed loop” by incorporating PID, MPC, and DreaMed’s Fuzzy Logic algorithms, along with Watson’s capabilities. They will together be leveraged to improve interface and meal dosing and recognize patterns, and the slide also suggests that Medtronic will look at additional sensor input signals and sensor/infusion set failure detection. We can’t wait to hear more about the capabilities of the pattern recognition feature – we could see it being highly beneficial in adapting to recurring event-based discrepancies in insulin requirements (e.g., menstruation, soccer practice every weekday at 3, fatty meals Friday at lunch). The timeline for this latest system wasn’t disclosed, but it will presumably require communication from a smartphone to gather additional data and leverage IBM Watson. Medtronic partnered with IBM in April 2015 to improve diabetes care through cognitive computing, analytics, and Big Data, and closed loop algorithms have always been on the agenda for this partnership. When explaining the long-term trajectory of the artificial pancreas program, Dr. Kaufman proclaimed that once the loop is fully closed, then she’ll retire.

• To date, Medtronic and IBM Watson have been busy developing and slowly rolling out the Sugar.IQ pattern recognition app (first demoed at Health 2.0 last September). At ADA, Medtronic Head of Innovation Dr. Huzefa Neemuchwala shared the first data from the “limited learning launch” phase of the Sugar.IQ app with Watson – now tag-lined, “Intelligent Diabetes Assistant App.” Relative to baseline metrics (one month prior), a small group of Sugar.IQ users experienced a solid 37-minute/day improvement in time-in-range, an 11% reduction in sustained hypoglycemia, and an 8% drop in sustained hyperglycemia. We assume the gating factor for this launch is Guardian Connect, the company’s standalone mobile CGM that will compete with Dexcom’s G5. 

Corporate Symposium: The Artificial Pancreas: Expectations, Experiences and Evolution (Sponsored by Insulet)

Trang Ly, MBBS FRACP PhD (Insulet Corporation, Billerica, MA); Bruce Buckingham, MD (Stanford University, Palo Alto, CA); Laurel Messer (University of Colorado, Denver, CO); Adam Brown (Close Concerns, San Francisco, CA).

To kick off this year’s Keystone Conference, Insulet hosted a panel discussion featuring Stanford’s legendary Dr. Bruce Buckingham, BDC’s Ms. Laurel Messer (University of Colorado, Denver, CO), Insulet’s Dr. Trang Ly, and our very own Mr. Adam Brown. In front of ~100 attendees, the panel explored the current state of closed loop systems, expectation management, and future must-haves. Dr. Buckingham led off, calling this an “exciting, amazing, revolutionary, and historic time in terms of control and management. With devices, we’ve consistently introduced more burden, and this is the beginning of coming down the other side.” We don’t undervalue the significance of such a remark coming from a pioneer of the field who has as close to a birds-eye view as anyone we know! Even with such optimism, the panelists all acknowledged that there is no shortage of difficulties associated with the use of a new hybrid closed loop system – Ms. Messer went so far as to refer to the first month of using the 670G as “more work” and a “learning curve.” That is to say, overnight control is remarkable when patients are in Auto Mode, but they still need to calibrate and count carbs during the day, and 670G users still spend an average of ~20% of time in manual mode. It’s crucial that they retain their diabetes self-management skills for these nearly five hours a day when they won’t have the luxury of automated insulin delivery. Similarly, panelists mused on the fact that after using the system for a while, patients (and parents) are not used to adjusting basal rates anymore – all of the clinicians on stage emphasized the importance of holding onto this skill. Moreover, if patients drift out of Auto Mode, they revert to their previous pre-programmed basal rates in open loop – if these are not consistently updated (the 670G doesn’t), they may be inappropriate. To properly manage expectations, Dr. Buckingham recommended underselling hybrid closed loop and having patients be pleasantly surprised if and when they see positive outcomes. Adam highlighted the importance of managing patient behavior (e.g., to avoid manual dosing/hypoglycemia correcting on top of what the system is doing), the amazing overnight impact (even in those in tight control), and why automated insulin delivery may change the conversation around diabetes tech and even enable CGM uptake.

• The panelists culminated their discussion by detailing a few items at the top of their closed loop wish-lists: Dr. Buckingham focused on system adaptability for different patient personalities; Ms. Messer emphasized remote monitoring and clarity on how patients are to handle insulin dosing during the transition from Auto to Manual mode; and, lastly, Mr. Brown called for customizable glycemic targets, use of the most up-to-date CGM and a pathway to quick sensor/transmitter upgrades (i.e., let’s avoid what happened with the Animas/Tandem/Dexcom G4), and the ability to remotely update pump software (“we need to get out of the model of a four-year pump cycle”). Both Ms. Messer and Mr. Brown highlighted the need for Auto Mode to ideally inform the pre-programmed open-loop basal rates, but Dr. Ly confirmed this will not be a feature in Insulet’s first Horizon system (it could introduce risk).
• Our favorite quotes from the panel discussion:
  
  • “This is an exciting, amazing, revolutionary, and historic time in terms of control and management. With devices, we’ve consistently introduced more burden and this is the beginning of coming down the other side.” – Dr. Buckingham
  • “Getting patients on closed loop systems will be more work, not less work. As a provider, you need to learn and work with the patients. There will be more alarms, more hassles during the day, and more to keep the system in closed loop. I recommend having patients come in with fairly low expectations and have them be happy when things go well. Undersell this – let them know all the things they will have to do and remind patients that the payoff is great control at night and possibly during the day, but not without work.” – Dr. Buckingham
  • “The first month on 670G is the hardest. I always tell patients not to make a decision on this until one month later. It’s a learning curve unlike anything else.” – Ms. Messer
  • “There’s no question that CGM is a learning curve accelerator, and since I wear one, I expected little benefit from wearing a closed loop system. But I cannot manage my diabetes while I’m sleeping, so closed-loop has really benefitted me too. It’s also clear that we need more options, because people aren’t doing well enough.” – Mr. Brown responding to an audience question that wearing and learning from CGM is likely just as good as wearing hybrid closed loop
  • “The reason to have closed loop is to get rid of variability, have more time in range, and to have a great night sleep. It should control your glucose overnight, which is such a relief to families, to young kids, and to all patients.” – Dr. Buckingham
  • “If I’m not running closed loop, I go to bed with one pre-programmed basal program regardless of what happened that day. When you go on closed loop, you realize that is insane.” – Mr. Brown
  • “The amazing thing about this, I think, is that you’re going to have a G5 or G6 sensor talking to the Pod and you can leave that PDM in your car and still be modulating your basal rate 24/7, eliminating the need to carry your handheld with you.” – Dr. Buckingham
  •  “One really exciting thing is that CGM is really coming of age, which is critical for closed loop – allowing for less calibration and better form factor. Investigators used to be forced to use clunky research devices, and now we’re finally seeing a commercial landscape where companies are building real products around the research that patients can actually use.” – Mr. Brown
  •  “The next step is to handle meals so we can get rid of carb counting – how great would it be if people with diabetes could just get up and start eating? It won’t be here in a year, but I think we’ll see it in three-to-five years.” – Dr. Buckingham
  • “Remote monitoring is a must for all pediatric patients. This should be obvious.” – Ms. Messer
• “Companies are realizing they can’t keep charging more for their products. They either need to improve outcomes if they want to charge more, or charge less.” – Mr. Brown responding to a question on how expensive new diabetes technology is.

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