July 11-14, 2019; Keystone, CO; Full Report - Draft

Greetings from Keystone, where the 2019 iteration of Practical Ways to Achieve Targets in Diabetes Care (commonly known as ATDC) took place. Below, you’ll find our top highlights from the meeting separated into: (i) Diabetes Technology, (ii) Regulatory and Pipeline, and (iii) Access and Affordability, (iv) SGLTs in Type 1, and (v) Prevention and Pathophysiology. Read on!

Diabetes Technology Highlights

1. Dexcom Pipeline Updates: G6 Pro and New G6 Adhesive Filed with FDA; Direct-to-Watch “Should” be Available “in the Next Year”; G6 Medicare by End of Year, Along With Pregnancy Indications; New Market Updates

Dexcom EVP Mr. Andy Balo – who we learned has initiated all clinical trials performed at Dexcom as well as 142 US and OUS regulatory submissions (!) – provided a number of pipeline updates in closing a Tandem-Dexcom joint-sponsored dinner symposium: (i) the G6 Professional CGM was filed with FDA last week; (ii) Direct-to-Apple Watch should be available within the next year (i.e., before July 2020); (iii) a G6 Medicare launch is expected by the end of 2019; (iv) Dexcom hopes to gain indications for pregnancy by the end of 2019; and (v) Dexcom continues to work on new markets.

• The G6 Professional CGM was filed with FDA last week. As of ADA last month, this was slated to launch in early 2020. G6 Pro uses the same 10-day sensor (factory calibrated) but adds a fully disposable transmitter, an ability to run in blinded or real-time mode, a clinic-owned reader to download the data after 10 days (and import into Clarity), and the ability for users to get real-time CGM data through the same G6 app. (Based on the Pro transmitter, the G6 app will run in a simplified mode.) This is a serious upgrade from the outdated G4 Professional CGM, and it brings stronger competition to Abbott’s FreeStyle Libre Pro. According to Mr. Balo, the need to reuse G4 Professional transmitters was “one of the biggest hassles on the marketplace” – presumably because clinics had to track down and sanitize transmitters – and the disposable G6 model is far more convenient. This will also give Dexcom experience manufacturing disposable Bluetooth transmitters at scale, good experience ahead of G7.

  • G6 Pro is different from the "Pro-Q" retrospective-only product that was FDA cleared in November 2018. Dexcom told us Pro-Q is being used in pilot studies as a glucose recording device; G6 Pro is the commercial product and has both blinded and real-time modes. It was only at ADA that Dexcom made this distinction clear.

• In Mr. Balo’s words, Direct-to-Apple Watch “should be out in the next year. Hopefully I’m not talking about this at this time next year” (i.e., availability by July 2020, though it has yet to be filed with FDA). We haven’t heard a timing update since Keystone 2018, when CEO Mr. Kevin Sayer said the feature should come in 2019; this feature has been a very long road towards development, ever since Apple showed it on stage at its developer conference in June 2017. Mr. Balo noted that it has “taken some time” to ensure that the transmitter is compatible with Apple’s Watch independent of the operating system.

  • Dexcom is still likely to be first-to-market with a direct-to-watch CGM, and we’ll be interested to see what use cases and marketing advantages this offers. We don’t imagine direct-to-watch will drive new-to-CGM uptake, though it will offer convenience for use cases where a phone is a pain to carry – e.g., water sports, running, etc. As a sidenote, Senseonics noted recently that approximately 50% of those wearing Senseonics’ Eversense are new to CGM.

• Dexcom has several pregnancy studies in motion, and notably just finished one with arm-wear. This data is currently being analyzed, and pregnancy and arm wear indications are expected to be submitted to FDA “within the month.” The company hopes to eliminate the pregnancy contraindication by the end of the year, and “will also have an arm indication, which will be good for all of our products too” (no timing shared). Reversing the pregnancy contraindication will be huge, as far more women should be on CGM – the CONCEPTT study showed that it is cost saving (particularly in the neonatal outcomes category), so much so that NHS plans to offer CGM to all pregnant women with type 1 diabetes by 2020/2021!

• Dexcom is currently running four hospital CGM studies across the country, with the goal of completion by the end of the year: One in the ICU, two in med-surg, and one in pre-op. See our recent coverage of these, along with the hiring of FDA’s Dr. Stayce Beck. On the business case: “We’ve learned that, on a weekly basis, surgeries are cancelled because of high blood glucose levels. They can’t do the surgery – so they bring patient back, give them insulin, and the cost of the patient goes up…Pre-/post-op, we think this can reduce the cost of the patient, and reduce readmissions. If the glucose level goes up before they leave, what happens is they don’t leave – they get readmitted, sometimes back to the ICU.”

• Dexcom is currently gathering information on nursing homes and assisted living facilities. Collaborators at University of Michigan’s Frost School of Business report that people go to nursing homes “chronic,” then become “acute” and have to go back to the hospital. Dexcom has spoken to reps from some of these facilities, finding that “they don’t even really take fingersticks.” There is a clear opportunity for CGM in this setting and for far better outcomes. We salute Dexcom for its work in this important and often very overlooked (and not lucrative) population.  

• Regarding the type 2 non-intensive, prediabetes, and consumer (obesity/wellness) markets, Mr. Balo shared that Dexcom is in the process of exploring product options: “What is the proper design? It may not be G6.” Dexcom’s 1Q19 call shared the creation of a “new markets” division that will more aggressively focus on these areas. 

• There was no update on the lower-cost G6 transmitter, which is presumably still on track for a 4Q19 launch (per 1Q19). We wonder if this is a gating factor to launching G6 Pro.

2. Medtronic Again Developing Tubeless Patch Pump, along with All-in-One AID Patch (Pump, CGM, Algorithm) – Many Years Away; “Duo” Combined Infusion Cannula/CGM Sensor to have Minimum Seven-Day Wear

Keystone speakers showed new slides depicting two never-before-seen Medtronic AID pipeline products, extending past personalized closed loop to: (i) a tubeless patch pump; and (ii) an all-in-one patch pump with a built-in CGM and algorithm. Based on the slide, the patch pump and the all-in-one CGM/pump/controller patch appear to be multiple years away from launch (perhaps four or more) – and as with any pipeline product so far out, the company has not promised specifics and of course we know that sometimes pipeline products so far out do not always even launch – that’s business and we appreciate Medtronic being measured about these. Medtronic has had a patch pump in development for some time, mentioned at its 2010 and 2012 Analyst Meetings; even back in 2012, under different leadership (Chris O’Connell), it was positioned as in the “final design and review phase” and expected to launch in a short number of years. Management teams change, of course, as do priorities, as can regulatory guidance. Before its acquisition by Medtronic, MiniMed would not have been in a position to pursue product development at the pace Medtronic has and we are glad for its purchase by Medtronic in 2001. Even though this has been far from lucrative M&A short-term for Medtronic, the investment in CGM and the closed loop certainly has driven the broader field forward. Back on R&D, the pump patch was not mentioned in the company’s “every other year” R&D meetings in 2014, 2016, or 2018, and it hasn’t come in any of the numerous public pipeline updates this year –including JPM, ATTD, 1Q19, or ADA. Given how far any Medtronic patch pump is from 2019, how many interim products Medtronic must invest in, how likely the tech ecosystem is to change, and how often medical device companies can change their pipeline (particularly Medtronic, as it’s been at this longer than anyone), we’re paying closest attention to Medtronic’s near-term, one-to-two year pipeline: non-adjunctive Guardian Sensor 3 CGM (filed with FDA in June); the MiniMed 780G with automatic boluses (in its three-month pivotal study; launch by ADA 2020); the next-gen Guardian iCGM with day-one-only calibration and seven-day wear (in its pivotal study; FDA submission in April-October 2020); the Synergy fully disposable seven-day iCGM sensor (FDA submission in November 2020-April 2021), and Personalized Closed Loop (FDA submission in November 2020-April 2021). That’s quite a lot even in the next two years! What questions do we have? (i) what’s happening with the MiniMed 670G indication for 2-6 year olds; (ii) whether the 670G will add Bluetooth ahead of the 780G; (iii) when we’ll see a seven-day wear infusion set; and (iv) how interoperability will be prioritized. At ADA, Medtronic became an official Tidepool Loop partner via an in-development Bluetooth-enabled ACE Pump and Guardian Sensor 3 iCGM (both to be submitted to FDA).

• It’s no surprise that Medtronic is still mulling over a patch pump form factor, given the commercial marketplace: Insulet’s strong Dash platform (Horizon hybrid closed loop; Tidepool Loop; concentrated insulins with Lilly); Lilly’s DEKA pump; Roche’s Accu-Chek Solo patch pump (also a very long time in coming); and Tandem’s patch-pump-like t:sport (FDA submission in mid-2020). As for the all-in-one patch AID system, aligning patch pump wear time with sensor duration seems to be the biggest hurdle, though Medtronic will theoretically address this with the “Duo” infusion set/sensor first. Smaller, JDRF-funded companies like EOFlow (launch in Korea “by late 2021,” with US/EU to follow) and SFC Fluidics (FDA submission last expected in 2021-2022) are also pushing toward the all-in-one AID form factor. Medtronic is also working on a next-gen tubed pump (expected to be lower cost, 50% smaller). We’re very happy to see so many companies working on these discreet systems that are obviously of big appeal to so many patients.

3. Dr. Roy Beck on Racial/Ethnic Disparities in T1D Exchange: Racial A1c Disparities Exist Regardless of Income, but Devices Shrink the Gap; CGM Use in White 3x that in Blacks

Jaeb Center’s Dr. Roy Beck kicked off Thursday’s “Healthcare Disparities” plenary with a fascinating, and at times disheartening, discussion of racial/ethnic disparities in diabetes outcomes and technology adoption based on data from the US T1D Exchange (gathered from ~70 sites, ~30,000 type 1 children and adults). His full slide deck is worth a flip through, but we’ve included a couple of our top takeaways below.

• Mean A1c is highest in blacks and lowest in non-Hispanic whites, regardless of household income. While the shape of the population A1c vs. age curve is nearly identical in all three groups (black, Hispanic/Latino, and white non-Hispanic), with mean A1c peaking around 15-20 years, A1c levels for black patients were significantly higher across all ages. Specifically, mean A1c peaked at ~10.5% for black teens, about 1.5% higher than that of white, non-Hispanics and 1% higher than the peak for Hispanics. This difference in mean A1c across the three groups remained remarkably consistent – with blacks much higher and Hispanics and whites lower – even when accounting for household income differences. It was the same story for acute complications: Across all age groups, rates of severe hypoglycemia and DKA were much higher in black patients than Hispanic or white patients. One in five black adult patients (>26 years) in the dataset experienced an episode of severe hypoglycemia (defined as seizure or loss of consciousness) over a three-month period, nearly triple the rate of white and Hispanic patients. Dr. Beck made sure to note that T1D Exchange data reflected the top centers for diabetes care in the US, and that outcomes and disparities in the country as a whole were likely even worse.

  • When considering higher A1c levels in African American patients, Dr. Beck reminded the audience that estimated A1c is ~0.4% higher for black patients for a given CGM-measured mean glucose, per a recent study. The different A1c-mean glucose relationship in the two groups – perhaps a result of factors such as red blood cell lifespan – explains less than 30% of the disparities between black and white A1c, suggesting there are meaningful clinical differences in diabetes management. 

• Notably, diabetes devices appear to reduce racial disparities in A1c – i.e., the A1c gap by race is most pronounced in patients not using pumps or CGM, and greatly reduced when only CGM users are considered. This suggests that part of the racial gap in A1c is driven by differences in device access, and indeed Dr. Beck showed further data demonstrating a huge racial disparity in both pump and CGM adoption. This tech adoption gap was particularly marked in CGM: rates of CGM use in white patients was triple that of African American patients, regardless of income. Perhaps more telling, CGM use in white patients from families making <$50,000/year was higher than that of black patients from families making ≥$100,000/year (21% vs. 16%). Dr. Beck pointed to a 2015 study in Pediatrics that found providers waited one year longer to initiate pump therapy after diabetes diagnosis in black children than white children. Increasing device adoption in racial minorities – and reducing implicit provider biases to achieve this end – will be key to reducing racial outcome disparities. Dr. Beck emphasized the need for more investigation into the barriers to device adoption at both the provider and patient levels.

• While MDI patients consistently had higher mean A1c than pump users, CGM users had the lowest A1c levels, regardless of MDI/pump status. The benefits of CGM use on A1c was shown across all income groups (<$50,000, $50,000-$100,000, and >$100,000), with the greatest benefit – an average 1% absolute reduction – in the lowest income group. A1c was also lowest in CGM users across all insurance types (private, state/federal, and none), with the greatest reduction in uninsured users. In other words, the greatest improvements in A1c as a function of CGM use are observed in those who have the lowest access – if there was ever an argument to push for expanded Medicaid coverage and general access for diabetes technology, this is it.

• We first saw the most recent batch of T1D Exchange Registry data published in DT&T in February, and Dr. Beck mentioned that a “deeper dive” into the racial disparities would be coming later, also in DT&T.

4. Dr. Clements Gives Bright Spots from T1D Exchange Quality Improvement Collaborative: Five Key Habits to A1c Reduction Identified, CGM Use in 12-26 Year-Olds Up from 30% to 45%, Depression Screening Up from 10% to 70%

UMKC’s Dr. Mark Clements gave the first results we’ve seen from the T1D Exchange’s Quality Improvement Collaborative (QIC), an EMR data sharing platform of 10 pediatric and adult clinics serving 25,000 patients aimed at testing and sharing QI measures for diabetes care. Small changes are implemented in individual clinics, allowing successes to then be disseminated across the network and scaled. We think this could be highly impactful, given the concerning state of T1D outcomes and clear potential to share lessons learned.

• University of Michigan’s Dr. Joyce Lee used the QIC to identify five habits directly related to a lower A1c: (i) Taking four or more BGMs per day (or using CGM); (ii) taking 3 or more insulin boluses per day; (iii) reviewing blood glucose data between clinic visits; (iv) changing insulin doses between visits; and (v) bolusing before meals. The greatest proportion of participants in the cohort were following three of these habits, and a clear dose dependent reduction in A1c was found as more habits were undertaken. With this data in hand, the QIC and Dr. Lee are now exploring interventions aimed at increasing these habits.

• By sharing successful interventions, the aggregate percentage of 12-26 year-olds across six centers using sensors/CGMs increased from under 30% to 45% in six months, and the average prevalence of depression screening across seven sites increased from ~10% to ~70% in two years. However, the former hasn’t yet translated to better outcomes. Dr. Clements noted that the A1c data from the QIC is similar to that of the T1D Exchange Registry, peaking at just under 9.5% around 16-18 years of age, with no improvement over time. Moreover, although the total number of individuals using technology in the QIC increased from 2016 to 2018 – 2,524 to 3,184 using pumps and 2,445 to 3,463 using CGM – there was a slight increase in the A1cs of both groups in the same time frame, from 8.63% to 8.67% in pump users and from 8.77% to 8.89% in CGM users. That said, we see greater use of technology as the bedrock on which outcomes will eventually improve. With CGM tracings in hand, providers and patients will be better equipped to pinpoint the behaviors and interventions that minimize glycemic variation and improve time-in-range.

• As a reminder, T1D Exchange recently announced partnerships with DreaMed and Glooko for the QIC. DreaMed will provide the FDA cleared DreaMed Advisor Pro to six member clinics to test its clinical decision support software for optimizing pump settings based on CGM data, while Glooko’s diabetes management platform will be used to incorporate anonymized diabetes device data into the QIC. We expect this could help validate time-in-range and Beyond A1c endpoints (e.g., linking time-in-range with hospitalizations, healthcare utilization); it could also help the field better understand the real-world impact of different technologies and allow more granular comparisons between products, patient groups, and clinics. 

• The QIC is actively recruiting additional US diabetes centers that specialize in treating type 1 diabetes, hoping to have 30 member sites by 2021. Current members of the QIC include:

  • Barbara Davis Center for Diabetes/Children’s Hospital Colorado

  • Baylor College of Medicine/Texas Children’s Hospital

  • Cincinnati Children’s Hospital

  • C.S. Mott Children’s Hospital, U of Michigan Hospitals-Michigan Medicine

  • The Joslin Diabetes Center

  • SUNY Upstate Medical University

  • Lucile Packard Children’s Hospital/Stanford

  • Nationwide Children’s Hospital

  • Penn Rodebaugh Diabetes Center, Penn Medicine

  • University of Missouri, Kansas City/ Children’s Mercy Kansas City

  • Wayne State University

5. Dr. Roy Beck: Time-in-Range Doesn’t Need to be Better Than A1c at Predicting Microvascular Complications, Just Equally as Good (Though BDC’s Dr. Polsky Believes it May be Better)

Given its clear advantages over A1c for daily diabetes management and retrospective review, Jaeb’s Dr. Roy Beck proclaimed that time-in-range (TIR) doesn’t need to be better than A1c as population-level predictor of complications, only equally as good. He reviewed results from a 2018 Diabetes Care paper that showed, convincingly, that lower TIR was strongly associated with higher rates of microvascular complications. Using data from the 7-point testing performed once every quarter in DCCT, Dr. Beck et al. calculated TIR, mean glucose, % of values >180 mg/dl, and % values >250 mg/dl. Risks of retinopathy and microalbuminuria were found to be strongly correlated with the SMBG-derived measure of TIR. When broken down into A1c bins (<7%, 7%-<8%, 8%-<9%, and >9%) and TIR bins (≥50%, 40%-<50%, 30%-<40%, and <30%), the rates of microvascular complications were strikingly similar for corresponding A1c and TIR bins (see figures below). For every 10% decrease in TIR, hazard ratios for retinopathy and microalbuminuria increased by 64% and 40%, respectively. At DTM 2018, Dr. Beck presented data showing how biochemical hypoglycemia – derived from the same data set – is associated with future severe hypoglycemia. With CGM measurements, rather than a quarterly 7-point profile, he postulated, the correlation between TIR and complications could be even stronger. BDC’s Dr. Sarit Polsky took this idea a step further in Q&A, proposing that – given all that we know about inter-individual differences in A1c attributable to factors like glycation rates and red blood cell turnover – TIR may one day be accepted as more predictive of future complications than A1c.

• Dr. Beck began his talk with A1c’s utility as a population-level metric: in large groups, individual variations in A1c for the same mean glucose cancel each other out, A1c has been a well-documented predictor of microvascular complications for more than two decades, and perhaps most importantly, it can be measured quickly and cheaply (vs. TIR which requires a minimum of the equivalent of 10 full days of CGM data). However, for individuals, the metric is limited. Dr. Beck expressed particular concern for clinical guidelines that call for particular actions (e.g. medication changes) when individuals do or do not reach a pre-specified A1c level.

  • A 2018 study, published in DT&T, showed that only 10-15 days of CGM data are needed to derive representative CGM profiles and glycemic values, including TIR. Currently, A1c is easier and faster to measure than TIR, but if TIR can be measured with just a couple weeks of CGM data (and even done remotely), it could be a huge step in getting TIR more widely adopted.

• The similarity between A1c and TIR correlations with microvascular complications may be surprising, given the relative weakness of the correlation between A1c and TIR. Dr. Beck showed results from a 2019 JDST paper (n=451) on the relationship between A1c and TIR. A measured TIR of 50% corresponded to an A1c of 7.9%, but with a 95% confidence interval of 6.6%-9.2%, indicating a rather wide spread. Dr. Beck did note that other studies are also arriving at similar correlations with TIR of 50% corresponding with A1c of ~8%.

• During the Q&A period, Dr. Beck made the case for using CGM in clinical trials for diabetes drugs. While it would take more work on the part of researchers and participants, Dr. Beck argued that having CGM data would provide valuable information on how a drug behaves, particularly after meals. CGM data could also capture potential increases in hypoglycemia from diabetes drugs, something that isn’t detected with A1c. We wonder if there is a role for NIH or philanthropic organizations in providing CGM for trials.

Questions & Answers

Q: With the consensus for time-in-range being 70-180 mg/dl, I could have 179 mg/dl for the entire day and get a TIR of 100%. Is that a concern at all for you?

Dr. Roy Beck: This is something that gets brought up a lot. Someone could have readings between 170-180 mg/dl all the time and another could have readings between 180-190 mg/dl all the time, and it’s going to look like they have huge differences. The fact is, having looked through hundreds of datasets, we haven’t seen that ever happen. Blood glucose changes throughout the day and that’s why AGP is useful. With AGP we can look at, say, just the post-prandial glucose measurements. In one of our studies, we’re actually just using the time-in-range two hours after a meal as our outcome.

Q: What is the ideal target for coefficient of variation (CV)?

Dr. Beck: This is just my opinion, but I don’t think it matters. It’s irrelevant, because you don’t treat CV. It is a pretty good indicator of hypoglycemia, but at that point just treat hypo- and hyperglycemia.

Prof. Thomas Danne: With blood pressure, we report the systolic and diastolic as, say, 120/80. It might help patients with high variability to report as time-in-range/hypo as something like 50%/20% or something.

Q: How would time-in-range, instead of A1c, help in a population setting for primary care? It seems like it adds more complexity on top of existing care.

Dr. Anne Peters: I don’t think we should argue A1c vs. TIR; they both clearly have utility. I actually think TIR can simplify care if you understand it. You don’t even really need TIR, just AGP. It tells you when to change therapy. If you put a CGM on a patient, you see they’re way out of range, and then realize you need to change something. A1c can’t do that, but TIR/AGP can. Ultimately, I think TIR will simplify care, but we have to teach it. In my clinic, all my patients love TIR.

Q: I work in a 90% Medicaid population clinic, and there’s just no way we can get CGM. We’re using the old-fashioned glucose log, then checking for values in and out of range. I also have my patients do 6-point testing.

Dr. Peters: Firstly, don’t despair. I work in a similar population. When I don’t have the same resources, I don’t use the same targets. Instead of trying to get A1c down to 6%, I look at what I can do using human insulins and fingersticks. With adherence, you can still help people do better. We need access for everybody, but we can still make progress with existing tools. 

Q: In clinical practice, time-in-range seems great. But, for trials, it seems arduous. What happens in a trial if you see disagreement between A1c and TIR? Which one do you trust?

Dr. Beck: In a clinical trial, A1c is an outstanding measure over a period of time, especially for reducing hyperglycemia. The variability across patients gets averaged out. But, you do learn a lot about a drug with TIR – especially in diabetes, where there’s risk for hypoglycemia. I think it would be useful for those developing drugs to have the CGM data. On the disagreement, it would be pretty unusual for TIR and A1c to disagree. One other value of A1c comes from the ease of performing a blood test, compared to wearing something in order to get TIR.

6. Dr. Jessica Castle: “We Need to Design [Closed Loop] Systems that Can Respond to Any Type of Exercise”

OHSU’s Dr. Jessica Castle reviewed some of the latest closed loop technology for handling exercise-induced dysglycemia before concluding that “we need to design systems that can respond to any type of exercise.” Beginning with currently available systems, she touted Tandem’s t:slim X2 Basal-IQ PLGS with Dexcom G6 and Medtronic’s MiniMed 670G, and looking into the not-too-distant future, she alluded to Control-IQ’s (Breton et al.’s 2017 ski study) and Omnipod Horizon’s (Forlenza et al., 2019) strong handling of exercise. She also detailed her team’s work with dual hormone closed loop informed by heart rate monitoring and accelerometry to automatically detect the start of exercise. In a 2018 outpatient crossover RCT published in Diabetes Care, the team showed that dual-hormone control reduced peri-exercise hypoglycemia in physically active adult type 1s while delivering time-in-range similar to that of the same system in insulin-only mode. An ADA poster from the same group (our coverage) with a similar system that now incorporated XeriSol liquid-stable glucagon and a glucose prediction algorithm to deliver small doses of glucagon at the outset of exercise if needed had the same outcome: Similar time-in-range as insulin-only, and a reduction in time <70 mg/dl during and four hours post exercise (2.9% vs. 8.2%). This glucose prediction algorithm, she noted, is designed to predict lows, but due to the sometimes-unpredictable glucose-raising effects of factors like anaerobic exercise and stress, the next step is to personalize closed loop response to exercise through machine learning. The OHSU group already has two grants along these lines. One is to use activity level as an input for an MPC controller and the other to look at context – GPS, calendar, past behavior, etc. – and estimate what the user is doing and how that may affect his/her glucose. For example, if a person has competitive tennis matches on Sundays and practices on Mondays, then the algorithm could tone down its aggressiveness on Mondays.

• UVA’s Dr. Marc Breton expressed during Q&A – and Dr. Castle agreed – that we may be missing the boat by focusing so exclusively on structured exercise in studies. OHSU’s Dr. Andrew Ahmann drove this point home with one anecdote: he had a patient who could run seven miles a day and never go low, but he experienced hypos every time he went outside to garden. The aforementioned OHSU grants could play a role here, and as Dr. Breton pointed out, possibly obviate the need for bihormonal systems. Dr. Castle agreed that we should push forward with insulin because it may be sufficient for most people, but others may still need glucagon.

• Dr. Castle shared that the T1-DEXI pilot study (n=49 type 1s, mostly adults), a deep dive on glycemia and exercise, is complete. This study collected one month of insulin, CGM, food, and physical activity data from participants. Data was collected with Dexcom G5, DiabNext’s Clipsulin dose capture device, a Garmin activity tracker, and a custom app developed at OHSU for food photos and exercise logging. Participants were randomized to complete two in-clinic and four home sessions of either aerobic, anaerobic, or high intensity interval exercise. Every seven days, providers review CGM and insulin data, and make insulin dose recommendations. This pilot will eventually inform the larger T1-DEXI study (n~600), with the objective of building exercise models to inform closed loop and decision support applications to reduce exercise related dysglycemia.

• The Q&A consisted primarily of providers asking Dr. Castle for her advice on specific cases, putting a finer point on how little we know about how to help patients manage glucose levels while exercising. People asked about: (i) the differences between type 2 and type 1 response to exercise (type 2s can release glucagon and have insulin resistance); (ii) taking pumps off during sports; (iii) managing glucose during exercise at elevation vs. closer to sea level; (iv) using Afrezza during exercise; and (v) the best way to optimize both blood glucose level and performance for competitive athletes.

7. “CGM vs. Pump: Which is More Important?” Dr. Andrew Ahmann Declares CGM Winner by TKO (Technical Knockout)

Using the “Thrilla in Manila” boxing match (Muhammad Ali vs. Joe Frazier) as an analogy for his “CGM vs. Pump: Which is More Important?” talk, OHSU’s Dr. Andrew Ahmann declared CGM the winner by TKO (Technical Knockout). Though the data convincingly points towards CGM as having greater effects on patients’ glycemic control, Dr. Ahmann did point out that insulin pumps will play an important role in maximizing benefits from CGM as part of AID systems. During the discussion, Dr. Ahmann suggested that CGM may be useful across the board, while pumps may be better-suited for more motivated patients; however, he noted that putting patients on pumps in some cases has resulted in A1c reductions so rapid he had to be wary of retinopathy.

• Looking at one meta-analysis (Yeh et al. 2012) of insulin pump studies, the use of pump reduced mean A1c by about 0.25% compared to MDI. When limiting the studies considered to ones using newer pumps and insulin analogs, pumps showed an average 0.3% A1c reduction compared to MDI (95% CI: -0.58% to -0.02%). Dr. Ahmann noted that removing a single study (DeVries et al. 2002) from the analysis would have completely eliminated all benefits seen by pump users (mean A1c reduction of 0.01%). Either way, he described the impact of insulin pumps on A1c as “underwhelming.”

  • Despite the underwhelming effect of pumps on glycemic control, Dr. Ahmann pointed at studies that showed higher patient satisfaction. In the 2017 Repose study, pump patients reported greater improvements in treatment satisfaction and lower “daily hassle” than the MDI group, despite non-significant differences in A1c.

• Though a much younger technology than pumps (pump technology was of drinking age before CGMs hit the scene), CGMs have been shown to improve glycemia and reduce hypoglycemia (and related anxiety). Dr. Ahmann noted that almost all CGM studies have shown outcome benefits to be positively correlated with CGM wear time (and with scanning frequency in the cases of flash CGMs), starting with 2008’s JDRF-funded CGM study. As newer CGMs have become much easier to use and convenient to wear, most studies show wear-rates >90% and A1c reductions of 0.5% or more. Dr. Ahmann highlighted Dexcom’s type 2 DIAMOND cohort, in which CGM conferred a ~6% increase in time-in-range (+1.4 hours/day) and ~0.8 hours less time >180 mg/dl. The study also found a slight increase (+0.25 hours/day) in hypoglycemia (<70 mg/dl).

8. HCT Program Updates from Dr. Agiostratidou: 105 Clinics Enrolled in Project ECHO with ~2,000 Patients; Child Device Use Up from 17% to 40% in 2 years, 90% Satisfaction with Dr. Gonzalez’s SMAs for Latinx Patients

Helmsley Charitable Trust’s Dr. Gina Agiostratidou expounded on two HCT-funded initiatives working to broaden PCP education about type 1 diabetes care and reduce disparities in care. The underpinning message of her talk: “We must vigorously pursue alternative healthcare models to support all people with type 1 diabetes. We have to take risks.”

• Project ECHO (Extension for Community Healthcare Outcomes) aims to educate PCPs about best practices to help underserved populations across the world. Drs. Michael Haller and Ashby Walker (University of Florida), and Drs. David Maahs and Nicolas Cuttriss (Stanford) headed the 18-month pilot grant to identify how PCPs can better care for adult and pediatric patients who lack access to endocrinologists. The majority of diabetes care happens in a PCP office (see maps below), making the dissemination of knowledge to these providers imperative. The program they developed uses technology (e.g., video chat) to leverage scarce resources, along with a web-based database to monitor outcomes. The ECHO model is different from traditional telemedicine, as the treating physician retains responsibility for managing the patient in a hub and spoke model. An expert hub team (e.g., Stanford) works with a community-based primary care team to ultimately reach patients. So far, the pilot has enrolled 37 non-diabetes primary care clinics in California serving ~1000 patients and 68 non-diabetes primary care clinics in Florida serving ~1000 patients. HCT is now in discussions to continue the ECHO program, aiming to evaluate outcomes at the patient, provider, clinic, and payor levels. Provided these demonstrate benefit, HCT hopes to install “super-hubs” to support national (“the goal”) and international (“the dream”) launches of Project ECHO.

  • ECHO’s design was informed by baseline surveys of PCPs and people with type 1 diabetes. Unsurprisingly, but very notably, PCPs lacked confidence, especially when it comes to CGM and pumps, yet 70% reported filling an insulin prescription for a person with type 1 in the last year. On the patient end, key barriers to care were cost of supplies, distance, long wait times for adult endocrinology visits, lack of attention to type 1 care in adult endocrinology visits, and a lack of any support groups for type 1 or networks therein.

• With a grant from HCT, Barbara Davis Center’s Dr. Andrea Gerard Gonzalez has implemented a culturally sensitive shared medical appointment program for the Latinx type 1 diabetes community. An underserved population, the mean A1c for Latinx patients at BDC was 9.0% vs. 8.4% for the non-Latinx population, and only 3.4% of Latinx patients had used a CGM. After the program, which included Spanish speaking providers, new culturally sensitive educational materials, and group setting visits to build community, there was a significant improvement in A1c, and technology use increased from 17% at baseline to 40% after two years among those under 12 years old. The satisfaction rate for the program was an impressive 90%, leading HCT to approve a grant to scale up the program. Dr. Gonzalez is now developing a manual of procedures to expand to four other sites, two in California and two in Florida.

9. Dr. Forlenza on 670G vs. Control-IQ within CARES Paradigm (Customizability, Simplicity); Underscores Randomization as Key to Reimbursement

BDC’s Dr. Gregory Forlenza applied Ms. Laurel Messer’s updated CARES paradigm to Tandem’s Control-IQ/Dexcom G6 hybrid-closed loop and Medtronic’s 670G, noting the respective strengths in customizability and simplicity. Standing for Calculate, Adjustment, Revert, Education, and Sensor, the CARES paradigm is one proposed method for assessing and helping a patient pick the best open- or closed-loop system (see below). Dr. Forlenza pointed to major differences in the “Adjustment” category between the two systems – namely, that insulin-to-carb ratios, basal rates, sensitivities, and exercise/sleep modes can all be adjusted on Tandem’s system, while the 670G is restricted to insulin-to-carb ratios, active insulin time, and temporary targets. As such, Control-IQ affords more customizability, while the 670G may be simpler from an HCP perspective. (Though Control-IQ is simpler for patients, since it requires no fingersticks and has no juggling between Auto and Manual Mode.) Of course, he acknowledged, one is not better than the other for every patient and having both options is advantageous to both the patient and provider.

• On Medtronic’s pivotal summary data for ages 2-75 from ADA 2019 (third slide below), Dr. Forlenza noted that “this is the kind of data I will write up and send to insurance companies when they say that closed-loop is ‘too experimental’ to cover.” However, he added that Tandem’s iDCL pivotal trial, by comparison, has the major advantage of randomization. According to Dr. Forlenza, the sustained improvement in TIR seen with Control-IQ (61% to 71% in six months vs. a steady 59% in the sensor-augmented pump group) carries more weight with insurers – “they have to pay for this stuff” (we see it as they “get” to pay for it – outcomes have been very positive) – because it was shown in a randomized cohort. By contrast, the MiniMed 670G has been tested in single-arm, three-month studies comparing two weeks of run-in to three months on closed loop.

10. Control-IQ Algorithm Studied in 50,000 Outpatient Days to Date – Expected to Triple in Next Year (Up to 410 Patient-Years); Dr. Breton Believes it Could Perform Better in Real World; Exploring Customizable Glucose Targets

UVA’s Dr. Marc Breton reviewed the latest data on the performance of Tandem’s Control-IQ (formerly UVA’s DiAs and then TypeZero’s inControl), including the Control-IQ IDCL pivotal trial and Freelife Kid AP studies that read out at ADA. Remarkably, the algorithm has now been studied in the outpatient setting for nearly 50,000 patient-days (137 patient-years – see diaTribe.org for Kelly’s experience in a trial in 2014 “UVA’s Overnight Makes for Great Dreams”), and this number is expected to triple by this time next year (see graph below – note the logarithmic Y-axis scale). Displaying a summary slide of performance across those two studies and two ski camp home phase studies from 2018/2019, he said, “Control-IQ performances are extremely stable – across age, across A1c, across study settings.” Indeed, time-in-range in these studies is stable at 71%-72% and time <70 mg/dl ranges from 1.1%-2.6%. And in the pivotal, user acceptance and trust is extremely high – the study has now had two extension periods, and participants are insisting to stay on the system. Dr. Breton believes the system may perform even better in the real-world: “It was the most open trial we’ve had – you just had to be type 1 and over 14 years. Nothing else. The clinician still had the right to exclude patients for safety reasons, but no subjected was rejected from the study…And [in the IDCL pivotal], there were minimal optimization of pump parameters in both groups at weeks 13 and 26. Investigators  were allowed to change basal rate, carb ratio, or correction factor for safety reasons at any time, but the protocol chair,  UVA’s Dr. Sue Brown was insistent in her desire to minimize the endocrinologists’ involvement as much as possible; to not let endocrinologists fine-tune the system. The results you’re seeing here are the system on its own with minimal help from the clinical team. It will be interesting to see what happens in the future studies when that conversation is reestablished properly.” This bodes well not only for when that conversation is reestablished, but also for busy endocrinologists and PCPs who may not know how or have time to tune parameters. The Control-IQ 6-13-year-old pivotal is “just starting.” As for the Freelife Kid AP study comparing nocturnal-only (dinner and overnight) with 24-hour use of Control-IQ for children in the real-world, an extension phase (n=120 type 1 children) will continue until week 36, with the nocturnal-only arm switching to 24-hour closed loop at week 18. Dr. Breton hopes to be able to present the findings early next year.

• Dr. Breton described a “cool feature” in the Control-IQ algorithm that allows users to “be in closed loop longer than they actually have CGM data” available. Any CGM data gap smaller than 20 minutes will not result in exit of closed loop. This is enabled by the system’s metabolic state estimation feature. Nice!

• Tandem/TypeZero have been exploring possibilities of having more or less algorithm aggressiveness depending on user preference – aka, customizable targets. In the current system, automated basal rate modulation during the day (including after meals) targets a range of 112.5-160 mg/dl. Automated basal rate modulation at night becomes more aggressive and tries to arrive at 112.5-120 mg/dl by morning.

11. Dr. Castle’s Perspective on Where We Are with Seven Aspects of Diabetes Burden; Predicts we Will See >90% T1 CGM Penetration in Her Career

OHSU’s Dr. Jessica Castle gave a phenomenally valuable, granular review of the state of diabetes burden (detailed almost verbatim in the bullets below). She expects that adoption of CGM in the type 1 community will exceed 90% during her career due to the falling cost and rising user-friendliness.

• Q: How are we doing with acute and chronic diabetes complications (and fear of developing complications)?

  • A: We are seeing improvements in time-in-range and reduction in hypoglycemia, along with new focus on patient-reported outcomes.

• Q: How are we doing on the cognitive load of diabetes-related decisions?

  • A: We are increasing automation, but no system is “set it and forget it” yet.

• Q: How are we doing on tasks, time, and discomfort (injections/device insertions, carb counting, fingersticks, etc.)?

  • A: No-calibration CGM devices (FreeStyle Libre and Dexcom G6) have been a huge step forward. But management still requires wearing an insulin pump or MDI, carb-counting/meal estimation are still required (but may be more automated in the future), and faster insulins or alternatives to subcutaneous insulin is likely required to complete AID systems.

• Q: How are we doing on alarms?

  • A: CGM accuracy has significantly improved, but alarms are still a burden for some. “Smart” alarms are in development – e.g., an alarm that would know not to sound when a person is high after a meal but already has a lot of insulin on board.

• Q: How are we doing on wearing of devices (and possible reactions)?

  • A: There are more options and better adhesives, but this is still clearly an issue. Dr. Castle added that adhesives are often patients’ biggest complaints about CGM, signaling a major shift from early devices where issues were more basic and deal-breaking (e.g., accuracy, false alarms). Device integration on the horizon could also lessen the burden of wearing multiple devices.

• Q: How are we doing on cost of drugs, devices, provider visits, etc.?

  • A: Insulin is way too expensive – biosimilar and generic insulins are now available but this has not resolved the issue. Lower-cost CGM devices are on the horizon.

• Q: How are we doing on hassle of refills and navigating insurance?

  • A: This remains a major issue. Dr. Castle hopes one day EHRs will help on this. We’d add that diabetes technology’s shift to the pharmacy should help a lot on this front.

Regulatory and Pipeline

1. Dr. Courtney Lias on a Digitally-Connected Diabetes Device Ecosystem: First, iCGM and ACE Pumps, “iControllers,” Decision Support Likely Up Next

In what’s become an annual Keystone tradition (see 2016, 2017, 2018), FDA’s Dr. Courtney Lias gave the Agency’s perspective on future directions for diabetes technology: standalone algorithms (“iControllers”), incentivizing and encouraging new, breakthrough devices, and increasing digital decision support. Device interoperability continues to be a major trend, as Dr. Lias reiterated the Agency’s vision for “an ecosystem of interoperable components that will lead to choices for people with diabetes, and the ability to customize tools that are best for them.” At ATTD 2019, Dr. Lias gave a talk expressing that the Agency should be able to create a class II iController pathway for AID systems; this would be the third, and final, piece to a completely interoperable AID system (along with iCGM and ACE pump). All of this has come together incredibly quickly (the first iCGM in 2018 and ACE pump earlier this year), a testament to the FDA’s willingness to listen to patients and providers and Dr. Lias’ leadership. Dr. Lias pointed to patient engagement, tech companies entering the medical device space (bringing their design-oriented mindsets and more competition to the field), and improvements in technology (e.g. batteries, sensors) as reasons for these rapid changes. Over the next few years, these new regulatory pathways could really start to make an impact for many people with diabetes, and, as Dr. Lias pointed out, it already has made an impact for some.

• Dr. Lias mentioned that pump manufacturers, particularly foreign manufacturers, are interested in bringing their devices to the US market (in many cases, through the new ACE pump pathway). This is a far cry from one of her concluding thoughts at Keystone 2017: “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.” Two years later and it’s hard to believe just how different US diabetes device regulation and the US commercial landscape are starting to look.

• Highlighting the difficulty of performing risk assessment and the importance of design, Dr. Lias told the story of a patient who set out to clean a BGM. It wouldn’t dry fast enough, so the person decided to microwave it – it blew up, resulting in a burn injury. It’s hard to predict every use case, but these stories remind us of the importance of human factors testing and education.

2. Dr. Lias Notes that FDA Can’t Directly Take Cost of a Device into Account, but Does Consider Potential Positive Effects (E.g., Increasing Accessibility, Market Competition); Again Underscores Need for Specificity in Surrogate Outcomes

FDA’s Dr. Courtney Lias gave a fast-paced 20-minute presentation touching on the FDA’s role in healthcare disparities, going beyond A1c, and nuances when weighing devices’ risks vs. benefits. She emphasized what she characterized as an often-forgotten part of the FDA’s role: to serve as “a referee,” ensuring equal treatment of manufacturers and that all stakeholders’ voices are heard. When evaluating a medication or product, the FDA only considers safety and effectiveness, not cost…directly. However, according to Dr. Lias, FDA does take into account positive externalities such as increasing accessibility or market competition. The FDA has also been working with payers to improve consistency of coverage for patients. In a very specific illustration of how this might play out, Dr. Lias described a real scenario in which two very similar devices were on the market, but because of a missing “key word” on the device’s labeling – a word that was apparently meaningful to a payer but not FDA – one device was covered by the payer and the other was not. Lastly, on the disparities and access front, Dr. Lias mentioned that the FDA has been encouraging manufacturers to consider pediatric populations when designing devices, particularly insulin pumps. While most device studies (at least initially) are run in adults, the FDA has pushed manufacturers to factor in potential pediatric users, especially when designing buttons and safety features. Indeed, due to a combination of FDA/patient pressure and industry initiative, manufacturers are pushing for very young indications. For example, Medtronic (670G) and Insulet (Omnipod Horizon) both expect to launch indications for use in kids ages 2+ in 2020, and Tandem is recruiting for a Control-IQ pivotal in kids ages 6+. 

• Dr. Lias mentioned the FDA’s desire to consider new “clinically meaningful endpoints,” whether objective or subjective. She brought up patient-reported outcomes (e.g. quality of life, functionality, stress, sleep, school/work performance) as examples of valuable, non-traditional endpoints. However, when using non-traditional health outcomes, she again emphasized, it becomes increasingly important to create very well-defined clinical endpoints. See our coverage Dr. Lias’ DTM remarks on the importance of defining surrogate outcomes, such as time-in-range, with a high degree of specificity.

3. Dr. Ratner on Hypoglycemia: FDA CDER Accepts Non-symptomatic <54 mg/dl (“Huge”), but Needs to See Evidence – Which We Have – that it is a Surrogate Measure; “Unrecognized Hypo the Iceberg Under the Surface”

After reviewing the evidence on hypoglycemia and the FDA drug division’s historical position on non-severe hypo, Georgetown’s Dr. Robert Ratner said that CDER recently “accepted” the <54 mg/dl consensus definition of hypoglycemia (“Level 2”), but now needs to see that this threshold is a good surrogate outcome. It’s unclear to us at this point if the FDA specified about the use of CGM vs. SMBG to quantify <54 mg/dl. Still, Dr. Ratner positioned this development as “huge and very positive.” The next step is to prove to FDA that time and/or episodes <54 mg/dl correlate with morbidity/mortality. Fortunately, he said, “We have that data. It’s growing, and we need to present it to the FDA in a cohesive way.” He then pointed to three such pieces of supporting evidence: (i) There is increased mortality associated with reported hypoglycemia in type 1 and type 2 diabetes (Hsu et al., Diabetes Care 2013; McCoy et al., Diabetes Care 2012); (ii) hypoglycemia is associated with adverse outcomes (total mortality, CV mortality, and MACE) in people with diabetes (Yeh et al., Acta Diabetol 2016); and (iii) higher rate of non-severe hypoglycemia is associated with higher rate of severe hypoglycemia, MACE, CV death, and all-cause mortality (Heller et al., ADA 2019 poster). To this list, we’d add a DCCT post-hoc analysis showing that fingerstick-measured hypoglycemia was associated with future severe hypoglycemia. This move on CDER’s part has come over a long period of time and we hope lead to greater use of CGM and comparison of time in hypoglycemia in diabetes drug trials moving forward.

• Dr. Ratner reminded the audience that “unrecognized hypoglycemia is the iceberg below the surface that we haven’t been recognizing.” Accordingly, CDER has traditionally been reluctant to “deal with hypoglycemia” because the data they see – from clinical trials – is not reflective of the general diabetes population. Trials in many cases exclude the hypo unaware, those with a history of severe hypoglycemia, elderly individuals (who often have long diabetes durations and are therefore at a higher risk of hypoglycemia), and individuals with multiple morbidities. Studies are also sufficiently short that a true perception of hypoglycemia prevalence is hard to assess. Furthermore, hypoglycemia is underestimated in the real world because providers may not ask, patients may not tell/remember, treatment of severe hypoglycemia by EMS is seldom reported to the patient’s clinician, many with type 2 (and some with type 1) have limited knowledge of hypoglycemia and its treatment, and blinded CGM studies have shown that self-reported hypoglycemia is extremely underreported. Henriksen et al. showed in a 2018 that 74% of all events <72 mg/dl were asymptomatic in a group of type 1s; at ADA 2019, a Polish group (Szymanska-Garbacz et al.) showed that 40% of type 2s experienced hypoglycemia over a 5-6 day period with blinded CGM, and with a majority of participants having asymptomatic nighttime events.

4. Dr. Jessica Castle Details What’s New in the Type 2 Diabetes Pipeline: Oral Semaglutide, Tirzepatide, MEDI0382, Faster-acting and Once-weekly Insulins, Chiglitazar and Glucagon Receptor Antagonists

In a lightening round session on new type 2 therapies under investigation, Dr. Jessica Castle blitzed through the latest and greatest in incretins, insulin, and more. Much of Dr. Castle’s presentation consisted of new data released at this year’s ADA. We are so excited to see continuous innovation across the landscape:

• In incretin therapies, Dr. Castle focused on Novo Nordisk’s GLP-1 oral semaglutide. She highlighted oral semaglutide as a very promising option for patients with a fear of needles but noted that the required fasting and timing restraints might pose some barriers. So far, oral semaglutide has shown comparable lowering of A1c and weight loss to liraglutide and empagliflozin, however, it was unable to demonstrate superiority vs. placebo on the primary outcome of three-point MACE in PIONEER 6 (though it did trend strongly in that direction with a hazard ratio of 0.79; 95% CI: 0.57-1.11; p<0.001 for non-inferiority; p=0.17 for superiority).

• In dual agonists, Dr. Castle shared data on both tirzepatide, a once-weekly injectable GIP/GLP-1, and MEDI0382, a once-weekly injectable GLP-1/glucagon receptor agonist. 2018 data showed that more participants on tirzepatide achieved target A1c <7% than those on dulaglutide, though this was somewhat expected considering that tirzepatide combines GIP agonism with GLP-1. Tirzepatide also conferred an average 11.3 kg (~25 lbs) of weight loss at the highest 15 mg dose (wow!). However, there was a significant discontinuation rate of 24.5% primarily due to adverse GI-related events (compared to 11.1% for dulaglutide). More recently, a poster at ADA 2019 indicated that slower titration meaningfully reduces the severity of GI side effects and treatment discontinuation with tirzepatide. Turning to MEDI0382, a phase 2a trial (n=64) was completed in 2018 comparing 200mg MEDI0382 to placebo for 41 days. Compared to placebo, MEDI0382 demonstrated superior change in fasting plasma glucose (-2.8 mmol/L vs. -1.1 mmol/L, p=0.0001), change in A1c (-0.9% vs. -0.6%, p=0.0004), and bodyweight (-3.84 kg vs. 1.70 kg, p=0.0008). Read more about MEDI0382’s significant reduction of hepatic fat content in subjects with type 2 diabetes here.

• In insulin therapies, Dr. Castle spoke about Novo Nordisk’s Insulin 287 (a once-weekly basal insulin), Lilly’s ultra rapid lispro (URLi), Adocia’s ultra-rapid BioChaperone lispro, and Novo Nordisk’s oral insulin (I338). Multiple pharmacokinetic and pharmacodynamic studies have been completed (but not reported) for Insulin 287 in comparison to placebo and once-daily insulin glargine. Additionally, Novo Nordisk is interested in combining Insulin 287 (175 U) with once-weekly semaglutide (0.5 mg), with a phase 1 clinical trial currently underway. Looking at URLi, new data from PRONTO-T2D was released at ADA 2019. The results showed that ultra-rapid-acting URLi was superior to Humalog in controlling 1- and 2-hour post-prandial glucose excursions during a mixed-meal-test through 26 weeks, with no significant difference in average A1c. While the overall rates of documented and nocturnal hypoglycemia (<54 mg/dL) were nearly identical, URLi conferred significantly higher rates of postprandial hypoglycemia between 1-2 hours after a meal (0.7 events per patient year vs 0.3, p<0.001) and between 2-4 hours post-meal (1.0 events per patient year vs. 0.7, p=0.04). In a study on 36 participants with type 1 diabetes, Adocia’s BioChaperone lispro demonstrated improved postprandial glucose excursion in comparison to insulin lispro. Lastly, Novo Nordisk’s oral insulin demonstrated very similar outcomes to insulin glargine. However, the extremely large dosage that was needed has been deemed “not commercially viable” to produce, prompting discontinuation.

• In other therapies, Dr. Castle highlighted Roivant’s glucagon receptor antagonist RVT-1502 (formerly Ligand’s LGD-6972) and Chipscreen Biosciences’ PPAR pan-agonist chiglitazar. For the former, Dr. Jeremy Pettus presented phase 2 results at ADA 2018, in which the candidate showed dose-dependent A1c reductions in people with type 2 diabetes (n=166) over 12 weeks. From a baseline ~8.2%, A1c dropped by a mean 0.15% in the placebo arm (n=41), by 0.9% in the 5 mg RVT-1502 arm (n=43), by 0.9% in the 10 mg arm (n=40), and by 1.2% in the 15 mg arm (n=42), with all p<0.0001 vs. placebo. However, proof-of-concept trials for the exciting candidate have been slowed by Roivant with no timeline given. At this year’s ADA, Dr. Weiping Jia unveiled results from two separate phase 3 trials of non-TZD PPAR pan-agonist chiglitazar vs. placebo and sitagliptin that demonstrated significant and sustained A1c reductions and a well-tolerated safety profile. In the CMAP trial comparing chiglitazar to placebo, both the 32 mg and 48 mg doses achieved superiority on A1c lowering at 24 weeks – the higher dose gave a very notable 1.5% A1c drop, compared to a 0.5% A1c reduction in the placebo group. In the CMAS trial, which used Merck’s DPP-4 inhibitor Januvia (sitagliptin) as an active comparator, chiglitazar achieved non-inferiority on A1c at 12 and 24 weeks (see tables below). Very notably, chiglitazar had a clean safety profile; adverse events were balanced between treatment, sitagliptin, and placebo groups in the two trials.

5. Future Advances in CGM: Smaller, Cheaper, and Integrated Into AID Systems – Dr. Viral Shah (BDC) and Dr. Andrew Ahmann (OHSU)

BDC’s Dr. Viral Shah gave a look at future directions in CGM technology – to be smaller, cheaper, and more integrated – now that CGM accuracy has matched that of BGMs. Both Dr. Shah and Dr. Andrew Ahmann (OHSU) presented slides from Dr. Roy Beck, showing the impressive progression in CGM accuracy over time – pushing to MARDs below 10% and closing in on the accuracy of the best BGMs. Drs. Ahmann and Shah also reviewed results from the REPLACE-BG study (n=226) that demonstrated that insulin dosing off CGM is safe and effective.

• Dr. Ahmann highlighted the ADA Standards of Care, which recommends patients on intensive insulin therapy check glucose 6-10 times/day. In his clinical practice, he generally says four is the minimum, and not that many patients even get to that – this is yet another reason why CGM has so much runway to drive outcomes in insulin users, since many people are not even getting four glucose data points per day! Further, Dr. Ahmann also told the story of a primary care physician who doesn’t even use the blood glucose data that his patients do record (we imagine they are time-constrained).

• When discussing the costs of CGM vs. SMBG, Dr. Ahmann noted that he, and many other clinicians, often have “no clue” what costs really are for patients. Dr. Ahmann commended the affordability of Freestyle Libre (and flash glucose monitoring, in general) as a cost-effective option for patients without insurance or with high-deductibles. Based on a pretty old 2011 study (Dr. Ahmann noted that CGM technology has changed immensely since then), the incremental cost-effectiveness ratio (ICER) from CGM was about $45,000/QALY for people with type 1 diabetes. More recent CGM cost-effectiveness studies from 2017 and 2018 have found even more favorable levels $25,000-$35,000/QALY. Obviously with lower-cost devices, longer sensor wear, no fingersticks, and decision support, these cost-effectiveness analyses should continue to improve – showing better outcomes at a lower price point.

• Both Drs. Shah and Ahmann pointed to the future of CGM devices as being smaller, cheaper, and integrated as part of closed loop systems. Dexcom/Verily’s next-gen G7 (coming late 2020/early 2021) was mentioned as an example of a smaller, easier-to-wear device. Dr. Ahmann noted that Freestyle Libre has already achieved an impressively small form factor, though it lacks real-time CGM capabilities. (The under-FDA-review FreeStyle Libre 2 iCGM will bring alarms and fill some of this gap.)

• In discussing Senseonics’ pipeline, Dr. Shah brought up the potential for Eversense to be used as an on-demand flash CGM for type 2 diabetes with no transmitter needed (slide below). He described the possibility for the sensor to be toggled between real-time and flash CGM. For patients with type 2 diabetes, the sensor could serve as an “on-demand” source for glucose data whenever desired. In our recent interview with CMO Dr. Fran Kaufman and CEO Tim Goodnow, this was positioned as a follow-on to the 365-day wear.

Questions and Answers (Panel)

Q: I’m here from Nightscout and I was wondering if any of you could touch on how you approach DIY systems in your practice?

Dr. Satish Garg: It’s a relevant topic, given the recent warning from FDA about DIY systems. Under our society, it’s not appropriate for providers to prescribe something off-label, especially with the risk involved. In clinic, we notify patients they are not FDA recommended/approved. However, it is ultimately each person’s choice. Some of our employees are actually using these systems, as well. We strongly discourage providers talking with patients about it off the bat, but if patients ask, it is our responsibility to go over the pros and cons. We’re hoping they will get the FDA’s blessing soon and it won’t be an issue in the future.

Dr. Andrew Ahmann: I don’t discourage patients from using DIY. All my patients on it are doing well. I’m very cautious to tell them it’s for people who are already into tech because they’re taking on a lot of responsibility. We don’t have the risks fully defined. If you look at the stuff on social media, they aren’t reporting a lot of the negatives and there could even be deaths we don’t know about.

Dr. Satish Garg: There were two episodes of severe hypo in Europe with people using Loop. There’s no backup or anybody you can call because you’re hacking these pumps. As a physician, I don’t want to be legally responsible.

Sonya Walker: As a nurse educator at BDC, it’s one of the many “risky” things people are doing. Taking insulin is dangerous. We see DKA frequently. As clinicians, if someone tells us they’re doing something that is not what we would recommend, e.g. keto diet, our goal is not to strictly forbid it. We want to help our patients be safe. They’re going to go home and do what they do, and we’re not following them 24/7. Our goal is not to dictate exactly what they can do, but whatever they choose to do, we try to help them do it safely.

Dr. Jessica Castle: I would echo Satish’s comments. Thank goodness for Tidepool hopefully making this commercially available, so that we can use in-warranty pumps and have a supported app. It’s no surprise that these DIY systems exist, people were sick of waiting. I applaud the efforts people have made, but I agree there are risks involved. I’m glad Tidepool is making possible to do it above the table.

Dr. Viral Shah: We are not discouraging patients to use Loop, but we also don’t promote it.

Access and Affordability

1. Dr. Aaron Kowalski on JDRF’s Mandate for Equal Access: Emphasizes Insulin Costs (“Every One of the Players are at Fault”) and Coverage2Control Campaign

Newly minted JDRF President and CEO Dr. Aaron Kowalski provided an impactful overview of JDRF’s work on equal access for people with type 1 diabetes. Overviewing the advocacy work that JDRF has worked on, Dr. Kowalski highlighted JDRF’s Online T1D Health Insurance Guide, the Coverage2Control campaign, JDRF’s outreach to large employers, and its work on insulin affordability.

• On insulin affordability, Dr. Kowalski underscored “insane list price increases” and noted that “every one of the players are at fault here.” He said that patients paying $300/vial of insulin is simply unacceptable, and that JDRF has been working “very aggressively” with all involved stakeholders to achieve progress. Furthermore, Dr. Kowalski described a recent meeting that JDRF had along with ADA and the various insurers, insulin manufacturers, and PBMs. As expected, there was a lot of finger pointing between all of the various stakeholders, but Dr. Kowalski did note that insulin manufacturers “have an interesting story to tell—they can show data that their profitability has not changed much over the past decade, which speaks to the system driving the problem.” Indeed, data has emerged showing that although list prices have astronomically increased, the net prices that funnel back to manufacturers have not meaningfully changed over the past decade.

  • Notably, Dr. Kowalski briefly commented on just-released news that the Trump Administration will no longer pursue its initial proposal to eliminate rebates in government plans. Although JDRF has not released an official position statement on this news, Dr. Kowalski emphasized that JDRF continues to see this rebate system as a “major part of the problem.”

• Regarding JDRF’s Coverage2Control campaign, Dr. Kowalski emphasized that “people with diabetes and their care providers should be deciding the best care for them, not insurers.” This sentiment was met with a hearty round of applause from the packed room, and many commended Dr. Kowalski for forcefully driving home this point. He commented that “choice remains a hugely important component to us” and detailed JDRF’s ongoing work to ensure patient choice with CGMs, pumps, and insulin.

• “One of the things I learned while working on policy is that it’s easy to yell and be angry. What we are trying to do is be pragmatic about what we can do – the real things that we can do that can start to break down these challenges for patients.” Dr. Kowalski channeled the immense frustration that often takes place when trying to impact health policy and emphasized that JDRF is trying to take a savvy and realistic approach toward progress.

2. Patient/Provider Panels Illuminate Incredible Cost of Diabetes, Difficulties with High Deductible Plans; Pediatrics Panel Highlights Judgment and Patient Experience

Each year, Keystone includes two parallel panels of patients with type 1 diabetes, adult and pediatric, and these are always a valuable opportunity to gain insight on the day-to-day of living with diabetes. This year’s adult session, despite being scheduled for discussion on cost, eye disease, kidney disease, and hybrid-closed loop, focused almost entirely on cost. However, the discussion did show near unanimous support for time-in-range over A1c, and panelists shared their preferences for data sharing. In the pediatric panel, we heard on what can be improved in clinics, advice to new families, what providers can do to reduce judgement, and diabetes celebrities as role models. We hope in the future to see more patients with type 2 in the panels. It was not clear what access various panelists had to Patient Access Programs.

Quotable Quotes – Adults

Eric, Quentin, Jay, Olivia, Hal, Dr. Peter Gottlieb, and Dr. Satish Garg

On Personal Diabetes Costs and Insurance Plans

• “I have type 1 diabetes, was diagnosed at age 6, and now I’m 53… I didn’t know what bad health insurance was until I got to my new job. I have high deductible plan - $4,000 per year – and contribute to an HSA at $4,300 per year that I don’t use because my financial advisor said I’ll need it when I’m older. 90 days of Humalog out of pocket for me is $1,475, 90 days of Tresiba is $1,423, BD pens are $152.75, and One Touch strips are $320. I’ve also been trying to go on Dexcom, and they’re pricing my start-up costs at $1,100, and my maintenance at $300 per month.” - Eric

• “I was diagnosed with type 1 at age 8, and I’m now 48… I’m on the 670G now, and my kidney numbers are almost back to normal after being on pumps for 10 years. I have similar costs to Eric. I have insurance but my insulin is $400-$450 per month, my 670G is $150-$300 per month, test strips are $50 per month. I don’t have great insurance; I’m in the same boat as Eric. I worked at Douglas County and thought I had bad insurance until I got a new job.” - Quentin

• “I was diagnosed with type 1 at 17 … My insurance costs were my own before the Affordable Care Act – I had similar insulin costs to Eric and Quentin. I’m now on Medicaid, which covers everything for me. I still get my test strips on eBay for 20 cents per piece. I test 6x per day. My A1c is pretty regular since seeing Dr. Gottlieb – around 6.5%-6.9% - and I’m on Lantus and Humalog. My biggest issue is that I can do the exact same activities and eat the same carbs, but sometimes my insulin acts like it’s not working at all – I have to correct pretty heavily – and the next day it’s the opposite. I don’t know why.” - Jay

• “I’m now 36, and I’ve had diabetes since I was 23 … I’m fortunate to work at the Barbara Davis Center, where the University of Colorado system covers 100% of diabetes costs with no copays. I don’t pay anything for pumps, sensors, strips, or insulin other than my employer insurance costs. My sister lives in London and has socialized healthcare and also pays nothing for supplies – she gets her strips covered, but they’re counted, and she’s on a pump and the Libre. She has to fight for it though. She has to prove why her life is better with those things, and she struggles getting the right amount of strips per day. My other sister who lives in GA doesn’t have insurance and goes under a program to get Levemir and Humalog for $18 per month. She tests 7-8 times per day and doesn’t have a sensor.” - Olivia

• “I was diagnosed in 1984 and work at BDC as a PA… I think the expectations set on people with diabetes by friends, family, providers, and coworkers make us think we can achieve perfection. There is no perfection in diabetes. We balance. We learn. Diabetes is a curveball… What are the answers? I think CGM is one. For me, I would prefer bouncing off a low rather than a high because the time to correct is much quicker. Perhaps the rapid acting insulins coming down the road may help. Right now, the alarms don’t understand what I see. They go off constantly and annoyingly. I would love a suspend for rising alarms – diabetes tech companies, are you listening? Waiting is the hardest part.” - Hal

• “I’m a Federal employee with the “Cadillac” of insurance plans – Blue Cross Blue Shield Federal. My premium is $400 for my family per two weeks, and I pay $400-$500 on top of that each month for diabetes supplies (pump, CGM, strips, and supplies). It seems like a lot, but it’s not that much compared to Eric and Quentin.” - Audience Member

• “I’m a CDE and have had diabetes for 25 years. I work two jobs to pay for medicine, and I get insurance through one. My premium is almost $800. My pump supplies cost $2,500 for three months, and insulin is $400 per month. That’s the nature of things. I can only go to BDC once per year. I have to take off work, hire a babysitter, and drive far. The bill is $400 because it is out of network, but I believe in having a good endocrinologist. I usually take patients with me, and we try to group our visits. One thing I’ve found helpful is the samples. Getting that extra one insulin vial per month has meant putting off that $400 copay until insulin comes in. I live in the middle of the mountains, and no reps come to visit. But every month on the 7^th I order samples not for myself, but for my patients in the donut hole, or those who are above the Medicaid limit but can’t afford insurance.” - Audience member

• “I’m a family care practitioner, and I meet my family’s $7,000 deductible by mid-year due to insulin. My premiums alone are $996. I think about going on Medicaid, but then certain providers won’t see me.” - Audience Member

• “If I had been forced to go on Obamacare [and not been eligible for Medicaid], it would’ve been cheaper for me to pay the fine they were imposing for not having insurance and pay for everything out of pocket.” - Jay

• “This experiment [BDC paying for 100% of diabetes supplies for employees] started five years ago. UC hoped to avoid many sick days for employees, so it created a foundation. First, the employer enforces that BCBS or Kaiser pay for supplies at a higher rate, and then the remainder is covered by the Foundation. I believe it’s about $20 million per year being put into the fund. They’re now planning to analyze data after five years for sick days and quality of life. In fact, HCT has been looking into this.” - Dr. Satish Garg in response to an audience member asking how BDC pays for 100% of diabetes supplies for employees

On How to Balance Retirement, Life Insurance, and Diabetes Costs

• “I don’t know when I will retire, but I feel like I have a normal lifespan in front of me. I will try to retire in my mid 60’s. My financial advisor said to sock as much into my HSA as possible, because it’s tax free. Now I have enough where I can invest money from the HSA to help it grow, but once I’m on Medicare, will that be enough to pay for supplies? I also may not be putting as much into 401k because I’m diverting money to my HSA. I may not be living life I want to because I’m trying to take care of myself and save money at the same time.” - Eric

• “I have life insurance through work right now. I don’t know that the premiums will be feasible after retirement, so I have it for now and not later I guess.” - Eric

• “Maybe disability insurance more important than life insurance. If you’re saving in a 401k, then that will balance out retirement, but the average American family doesn’t have enough money to last a week if they don’t work. The average American family is downright broke.” - Hal

• “My plan has an HSA too, but the problem is I go over my deductible every year. I’ve got four kids and need to get them through school, so I need life insurance for the next eight years, but then I’m not too sure what I’ll do.” - Quentin

• “I’m self-employed and have no 401k. Work till I die is my plan.” - Jay

On Getting Insulin from Outside the US

• “I have considered it. I went to Mexico on a church mission with my wife and paid $27 for a bottle of Lantus there. It’s $340 in the US. If I wasn’t on Medicaid, I would’ve gone to Mexico twice per year and bought a six-month supply. It doesn’t make sense why Mexico can sell the same insulin for so much cheaper.” - Jay

•  “If you knew how much it costs to make a bottle – less than one dollar – you’d be shocked. Insulin companies know it jolly well.” – Dr. Garg

• “I have bought insulin from Canada.” - Quentin

• “I’m getting my pen needles from Canada for $25/box. They’re more expensive here even through my insurance.” - Eric

• “Why do Americans subsidize the rest of the world’s cost for pharma? I think we need to get political and make them answer these questions.” - Hal

On Using Time-in-Range Instead of A1c

• “I think it would be beneficial for me at least. I could be 300 for two days and 40 for the next four and it would balance out in my A1c. Time-in-range is more important because I can’t cheat it.” - Quentin

• “I agree. I think patients are so focused on A1c that they’re not looking at the bigger picture. I tend to run a higher A1c of 7.5%-8.5%. I do a lot of exercising and don’t like to go low when I’m exercising. I run a 150-200 mg/dL range most of the time. If you could get me closer to 180, I’d say I feel better.” - Olivia

• “I think all of the metrics that we as providers use to judge diabetes management become something that people with diabetes feel embarrassed or offended by. It doesn’t matter if its time-in-range or A1c. It doesn’t matter how we’re valued, we, as people with diabetes, feel judged. I don’t care whether its A1c or time-in-range, we will feel judged.” - Hal

• “I do agree that you will be judged directly or indirectly. But it is changing from something that is an abstract average to a daily measurement – the blood sugars you check every day. With time-in-range, we might be talking on the same wavelength in a doctor’s visit. That’s the whole point. I agree, unfortunately, that there is a judgement factor when visiting the doctor, but that’s something we must go beyond.” - Dr. Garg

• “I think time-in-range is more empowering to the patient. It’s easier for me to hear 85% or 25% in range. I know what I need to do improve. I can game A1c, but I can’t game time-in-range.” - Quentin

• “I agree – 6.5% is a specific number, but 70-180 is a whole range of numbers and a lot easier to maintain than correcting to 100 every day, for example.” - Olivia

On How to Motivate People with Type 1

• “Something that has helped is telling patients that diabetes has changed in the past 50 years so much that you can really live a near normal life. You can never have a normal life, but you can expect to live near to normal length and that sometimes gets the message across. I’m not sure if that’s for all people but something may click … Our job is to reinforce that advances have been made.” - Dr. Garg

• “My pediatric doctor told me that I should expect to live until 42. It wasn’t until I met Dr. Garg, and he told me that I could live a near normal lifespan when it made sense to take care of myself. It was empowering.” - Eric

On Type 1 Diagnosis

• “We’re always talking about type 1 in kids, but 50% of people today who develop it are over the age of 20. It was common, but we weren’t looking. We would argue that screening would be helpful.” - Dr. Peter Gottlieb

• “Here’s a takeaway for all of your practices. 15-20% of people diagnosed with type 2 are actually antibody positive. We need to keep that in mind, but insurance may not pay for testing. If it were my kids, I would get them screened because the biggest advantage is preventing DKA. In a nutshell if cost is not a barrier, that should be the way to go.” - Dr. Garg

On Sharing Data

• “I have no concerns because all [companies] are going to see is blood sugar levels and A1cs. I don’t think I would have any issues. Again, I can’t afford CGMs because it’s out of pocket, so I will just be buying strips. I also don’t have a smart phone.” - Jay

• “I think there’s concern if you can tie data back to the patient. I think it’s good for research to have the access to the data points as long as its deidentified. Especially for people looking for a job, employers may consider whether one person is healthy while another has diabetes. Making decisions based on that data is concerning for me.” - Quentin

• “The more people that have access to this knowledge, it is going to make things better for people with diabetes. A lot of people have negative thoughts, but I tend to be more positive. I’m a more positive person overall. I can’t imagine where having more information would make things worse for PCPs. Inside BDC we are very lucky to have access to a lot of data, and it helps us tremendously. To see what can happen with that data is amazing.” - Olivia

• “If [companies with access to data] only focus on healthcare, this might be useful. Suddenly now the PCP and other providers are out in the communities, and they might be empowered through these tools.” - Dr. Garg

Quotable Quotes – Pediatrics

Panelists: Killian (age 4), Audrey (age 13), Gavin (age 19), Aria (age 17)

On What Can Be Improved in Health Clinics and Care

• “Little activities or books for siblings would be good. I know when we go, we have to take our other brother who doesn't have diabetes, and he gets bored, so activity or coloring books would be great for him.” - Audrey

• “From a mom standpoint, maybe more education type stuff for the kids. Classes for when they’re transitioning from different stages of life or what to do in different scenarios. I don't know if that would work for these little guys, but as a parent, I would love that.” – Audrey’s mom

On Relationship to Food

• “I would generally say that diabetes didn’t change how I viewed food, but using a pump changed a lot of that. Before I had a pump, I had to choose what food was worth another shot, but now with the pump, I can bolus whenever and eat what I want.” - Aria

• “Our whole family changed pretty drastically. We pay a lot more attention to other kids who aren't type 1. The hard part is that Killian reacts really heavily to any juice –at every birthday party, there's Capri Sun and Kool Aid, and as a result we’ve now found non-sugar/sugar free stuff that we bring on our own” -Killian’s Mom

On Advice to New Families

• “Try to get on technology as fast as you can. That’s been the biggest thing that has helped me.” - Aria

On What Providers Can Do to Reduce Judgement and Improve Patient Experience

• “It’s nice to know once in a while from providers if they have diabetes too. Let us know you’re not perfect too and you mess up just like patients do.” – Gavin

• “I think it’s important to stress that there are no necessarily ‘good’ or ‘bad’ numbers – you have to learn to deal with what to do with your care based on the numbers. It’s important to relate that to patients.” – Aria

On What Providers Can do To Make Parents Feel Less Judged

• “Start out by telling us what we are doing really well, and then move to what we need to work on. I’m always personally focusing on how bad I am doing, so if you start out by telling me what I’m doing right, it goes a long way.” – Killian’s mom

• “Every parent gets really nervous one week before the appointment because it’s like a report card. Sometimes we work really hard and it pays off. And sometimes we have a yucky three months that aren’t as good. Understanding and realizing this is always good. Our doctor always makes me feel like an incredible mom even if I’m actually not!” – Audrey’s mom

On Diabetes Celebrities They Look Up To

• “Nick Jonas” – Audrey

• “Nick Jonas. Seeing him in all the movies and singing and acting, I never would have guessed he has type 1. It’s very impressive.” – Olivia

• “My friends with type 1. I look to them for help. If I run out of insulin I trade with them sometimes.” – Gavin

3. Dr. Anne Peters on Dealing with Disparities in Healthcare: “My Answers are Wrong,” Solutions Have to Come From Patients Themselves

Dr. Anne Peters (USC) closed out Keystone 2019 with what Dr. Satish Garg called the “most important talk of the conference,” lessons from her experiences with underserved patients on the east-side of Los Angeles. Like she shared at ADA, Dr. Peters underscored that in the American healthcare system, patients are often taken care of as children, but once they become adults, their diabetes care becomes optional or falls through the cracks. With the type 1 diabetes patients Dr. Peters sees in east-side LA, complications often occur by age 30 and many don’t live to age 50. For these patients, it can take years to improve glycemia, if at all.

• Dr. Peters presented a DIY Diabetes Survival Kit wish list – she called for a judgement-free diabetes “help line” that anyone could call to get information or help. Every clinic with type 1 patient should have a cupboard with extra supplies, something that could help fill in reordering gaps and wouldn’t have a huge impact on companies’ profitability according to Dr. Peters. Lastly, she talked about the need for providers, educators, manufacturers, and researchers to reach out to patients. Even when she received funding to bring some of her underserved patients to conferences, they didn’t feel comfortable riding an airplane or fitting in. Taking better care of these populations will require everyone to be more proactive in meeting with these patients, she stressed.

• “I’ll tell you my experience, but I have no answers.” To illustrate the importance of having solutions come from patients rather than clinicians, Dr. Peters shared a particularly compelling story of one of her patients who works as a mechanic. He was performing fingersticks just once or twice a week. When Dr. Peters met with him, she realized that his hands were so brutally callused (from his day job) that it was extremely painful and difficult for him to take a fingerstick – the photograph was shocking. Her immediate instinct was to prescribe him a CGM; however, he shared that he has to “slither” on the ground between trucks during his job, which presented a sensor adhesion challenge. Walking between trucks instead was not an option, as the boss would fire him for taking too much time to get between the trucks. The ultimate solution came from the patient – he realized that he could wear the sensor under a lumbar support band that he already used at work, preventing the sensor from being ripped out. It was just one example demonstrating the level of troubleshooting that can occur to effectively integrate technology into patients’ lives.

• Dr. Peters called getting devices to her patients “the bane of my life,” even with her wealthier west-side LA clinic patients. For underserved patients, the issues grow. Pictures of devices are always shown on white people. When signing up patients for devices, most forms require patients’ emails – something most of her east side patients do not have. Dealing with customer support on the phone is a major obstacle for Spanish-speaking patients. Lastly, even when patients are able to access devices, they would might get stolen or lost if they are home-delivered in unsafe neighborhoods. Dr. Peters told us about one of her “typical” patients who, despite working hard at maintaining a constant supply, didn’t have supplies for her insulin pump about one-third of the time.

• Unpublished results from the Helmsley-funded STEPP-UP study showed that tailored educational materials for diabetes devices (i.e., lower reading level, fewer numbers) could be effective for lower-literacy patients. While glycemic parameters weren’t significantly affected, diabetes knowledge increased, while distress and depression decreased. She did note that fear of hypoglycemia increased greatly.

• Ultimately, improving care for everyone with diabetes will require changes in and outside of healthcare – e.g., immigration. Dr. Peters noted that many patients have presented with significantly worse glycemia in recent weeks due to fear of impending ICE raids.

4. Dr. Anne Peters Explains Rationale Behind Controversial ACP Guidelines; Human Insulin > Analog Insulin for Some Type 2s in Terms of Cost Savings

Dr. Anne Peters gave a valuable talk on guidelines, explaining differences between AACE/ADA/EASD/ACP suggestions and commenting on how the choice between human and analog insulin can be leveraged to drive healthcare cost savings. Differences in diabetes care guidelines have become a hot topic over the past few years – see our coverage on the topic from recent meetings, including ADA Postgrad 2019, AACE 2019, and ENDO 2019.

• Dr. Peters voiced several critiques of AACE guidelines for type 2 diabetes care. First, she noted lamented lack of diversity on the writing committee for the guidelines: out of 20 authors, only one woman was included, and all authors were over the age of 60. Dr. Peters urged for more integration of different voices in the future. Moreover, the AACE guidelines advocate for an A1c target at or below 6.5%, and Dr. Peters asserted that there is actually no clear RCT evidence that supports this target. The AACE A1c algorithm also leaves out several patient groups, as Dr. Peters pointed out; it does not consider what the target should be for (i) those with a serious illness and at low risk for hypoglycemia; and (ii) those without a serious illness but at high risk for hypoglycemia. Dr. Peters then went on to comment on the guidelines’ questionable approach to integrating data (it sometimes cites review articles instead of the original source publication) and approach to CGM (statements are not necessarily corroborated by clinical evidence yet).

• Dr. Peters attempted to explain the rationale underlying the ACP’s decision making for its guidelines, which have been the source of much controversy (target A1c of 7%-8%) in the diabetes community. She noted that the ACP is more focused on macrovascular events rather than microvascular events. Consequently, its guidelines do not advocate for lower A1c targets due to a lack of evidence linking lower targets to improved macrovascular outcomes. Indeed, evidence from ACCORD, ADVANCE, UKPDS 33 and 34, and VADT do not robustly support the notion of reduced macrovascular risk with lower A1c beyond 7%-8% (although we do note that these studies were completed more than a decade ago and did not include newer, more efficacious glucose-lowering agents). Since the release of ACP’s stance, we’ve seen ample criticism of their recommendations and methodology—look no further than last year’s CMHC 2018, where several KOLs candidly sounded off on these guidelines.

• On the question of using human vs. analog insulins in people with type 2, Dr. Peters asserted that “this is an area where we need to have a serious discussion about costs vs. benefit.” She briefly pointed to a wealth of data on the subject, which indicate (i) small A1c differences; (ii) small difference in nocturnal hypoglycemia; (iii) small to no difference in severe hypoglycemia; and (iv) no difference in ED visits or hospitalizations for hypoglycemia between usage of analog vs. human insulin (see first slide below). As a result, Dr. Peters stated that the conversion from analog to human insulin in some type 2 patients should be feasible, safe, effective, and (very importantly) a major cost saver. On this point, she provided an intriguing case study of how clinicians might improve healthcare value by considering the currently high costs of many diabetes pharmacotherapies. The case: a type 2 with an A1c of 6.3%, currently on metformin, glipizide, sitagliptin, and insulin glargine (total annual cost: $10,358). Dr. Peters detailed how clinicians might “deprescribe” to improve value and lower costs here. The patient could be discontinued on sitagliptin, have their glargine switched to NPH, and may still meet an A1c of 7.3%. The new cost? $821/year.

  • For more commentary on the important benefits vs. drawbacks of using human insulin, see our piece on a JAMA viewpoint published in 2017 from Drs. Irl Hirsch, Matthew Riddle, and Kasia Lipska on the subject.

5. Dr. Elizabeth Seaquist on Cost of Hypoglycemia: New DOM Study Finds Severe Hypoglycemia Associated with >$10,000 in Annual Direct Medical Costs Compared to Matched Controls w/o SH

In an in-depth overview of the costs of hypoglycemia, Dr. Elizabeth Seaquist touched on an interesting and recently published study in DOM regarding direct medical costs of severe hypoglycemia. The population-based, retrospective cohort study involved 23,000 people with diabetes in Hong Kong, half of whom had a recorded severe-hypoglycemia (SH) event, and another propensity score matched-controls half without SH. Direct medical costs in the years before, during, and after the first SH event were determined by summing up the costs of health services utilized within the respective year. It was seen that patients with SH had direct medical costs of nearly $11,751 in the year of their SH event, compared to $878 in the matched controls group without SH. These increased medical costs were due to increased outpatient clinic visits, increased emergency room visits, and increased nights of hospitalization. Of course, as a retrospective cohort study, this study cannot solely link severe hypoglycemia to the increase in cost. Nonetheless, these data further underscore the need for a comprehensive approach to reducing rates of hypoglycemia and severe hypoglycemia, given the enormous burden these events can place on healthcare systems (let alone the impact they make on the individual patient level).

• In a separate talk, Dr. Seaquist reviewed the definition of hypoglycemia and evidence-based models for reducing risk in both type 1 and type 2. She noted that hypoglycemia is classified from level one to three, with level three corresponding to severe hypoglycemia; these cut-offs guide regulators, companies, and providers in creating standards for diabetes management.

  • For type 1 diabetes, evidence-based methods to reduce hypoglycemia include structured education programs, using threshold suspend/hybrid closed loop pumps, CGM and new insulins (i.e. degludec), and islet transplantation. Dr. Seaquist began the discussion emphasizing that “education is critical,” and discussed results from the DAFNE structured education program. In DAFNE, participants lowered their A1c, though not necessarily to target, and gained awareness of hypoglycemia through the structured education program in flexible insulin dosing. Elsewhere, Dr. Seaquist reviewed that patients on a threshold-suspense pump had a 38% reduction in mean AUC for nocturnal hypoglycemic events and less time spent with sensor glucose under 70 mg/dl compared to the control. Results from the SWITCH 1 trial also showed that patients taking insulin degludec had fewer mean hypoglycemic episodes compared to insulin glargine U100.  

  • For type 2 diabetes, evidence-based methods include focusing on therapies that aren’t associated with hypoglycemia (such as GLP-1s and SGLT-2s instead of sulfonylureas) and using CGM and next-gen basal insulins. On the last, Dr. Seaquist recommended either Novo Nordisk’s Tresiba (insulin degludec) or Sanofi’s Toujeo (insulin glargine U300) for patients at greatest risk because of their association with lower hypoglycemia rates.

Q: How many episodes of hypoglycemia does it take for patients to recover [from impaired awareness] so they don’t keep getting hypoglycemia? Are there guidelines that can help patients prevent events?

A: I don’t have a concrete answer, but I think impaired awareness of hypoglycemia can happen and change at any point based on a patient’s glycemia in the days preceding an event. Studies on the persistence of hypoglycemic events have shown blunted responses in patients as late as day 5 after the event, so we need to teach patients that one episode of hypoglycemia is going to have an impact. This is why we felt so strongly about level one being 54 mg/dl- even though patients feel no symptoms, their regulatory response is altered once their blood glucose dips below this value. Patients need to be cautious, and I would like for them to be on sensors so they can receive real-time warnings.

SGLTs in Type 1

1. Adjunct Pioglitazone Therapy to Mitigate SGLT-related DKA Risk in Type 1? Dr. Ralph DeFronzo Hints at “100% DKA Risk Reduction” in JDRF-Sponsored Trial

On the subject of SGLTs in type 1 that emerged during a Q&A session, Dr. Ralph DeFronzo excitingly and somewhat provocatively asserted that DKA risk can be mitigated “100%” by using a novel approach currently being tested in a JDRF-sponsored trial (now posted to ClinicalTrials.gov). Details were sparse, but Dr. DeFronzo appeared to indicate that adding off-label pioglitazone to dapagliflozin treatment in type 1s results in no elevated levels of ketones in any participants in the trial, and therefore no DKA events either. We picked up the following details from ClinicalTrials.gov regarding the study: 120 type 1s are enrolled, with the study design including a 4-week run in, then all participants put on dapagliflozin for 12 weeks, followed by randomization to either placebo or pioglitazone for 16 weeks. The primary outcome measure will be change in A1c and secondary outcomes include plasma ketone levels and changes in insulin dose. The ramifications of this study could be huge, seeing how DKA is the major drawback associated with adjunct SGLT therapy in type 1 diabetes. Dr. DeFronzo appeared quite confident in the results of the study but was notably met with caution from fellow panelist Dr. Jay Skyler, who asserted that “it’s important to finish the study before making these conclusions.” Dr. Satish Garg also briefly commented on the trial, calling it a “double whammy of off-label therapies,” referring to how both pioglitazone and dapagliflozin are being used off-label in the trial. We’ll keep our ears perked for any further details on this front. To our knowledge, data on pioglitazone in type 1 diabetes is relatively sparse: we found one poster from ADA 2005 that tested pioglitazone in type 1 adolescents and found limited efficacy and another review of its use in type 1.

• We’re curious if this pioglitazone-trial might be related to another JDRF-sponsored trial on DKA mitigation that we first heard about from Dr. Bruce Bode at AACE 2019 – that trial involved using ketone monitors, and Dr. Bode claimed that there were zero DKA events with use of SGLT inhibitors and ketone monitors in the trial. We’re excited to see such a focus from the diabetes community on developing novel methodologies for mitigating DKA risk, especially given how enthusiastic patients are about the potential of SGLTs in type 1.

2. Dr. Anne Peters on SGLTs and DKA Risk: “I Can’t Prevent DKA Unless the Rest of the World Understands the Problem as Well”; Provides Her Personal Prevention Protocols

On SGLTs in type 1 and DKA risk, Dr. Anne Peters advocated for broader education of the entire healthcare community in order to truly mitigate risk for patients. She shared a particularly impactful story regarding one of her patients, underscoring that although patient and provider education is a crucial first step, it’s also important that other involved groups (such as ED physicians) are well informed on the issue. Her patient was an ideal candidate for adjunct SGLT inhibitor use—on a pump and sensor, A1c in the low 7s, well-informed about SGLTs, and vigilant about checking ketones with a ketone monitor. She had successfully used SGLT adjunct therapy for quite some time before experiencing a few DKA events: First, a pump failure after a long flight led to elevated ketone levels that eventually put her in the ICU with a serious case of DKA. Unfortunately, the ICU unit continued to lower her dextrose and insulin infusions after seeing her glucose levels decrease, which put her back into DKA. Frustrated with the situation, Dr. Peters took it upon herself to call the ICU department, saying “I am the world’s expert on this issue—just listen to me and do what I say, and we can solve this problem.” Dr. Peters advised the team to increase the dextrose drip and increase insulin administration, which eventually lead to resolution of her DKA. This experience lead Dr. Peters to conclude: “Even with the best job that I can do, I can’t prevent DKA unless the rest of the world understands this as well.” Indeed, DKA events associated with SGLT use can be tricky for those who are not familiar with the phenomena, as blood sugar levels are not as elevated as in traditional DKA cases. On a related note, at ENDO 2019 we heard results from a small medical record study that highlighted the prevalence of euglycemic DKA and its impact on delaying diagnosis and treatment of DKA in hospital settings. Much of this appears to be due to a lack of education regarding SGLT related DKA events. On this front, Dr. Peters is taking action herself, as she will be teaching at this year’s upcoming Emergency Department Physicians meeting to inform this group about what to do when low glucose levels and seen in tandem with high ketone levels (“euglycemic DKA”). We tip our hats to Dr. Peters’ unwavering commitment and initiative in this emerging landscape.

• Interestingly, Dr. Peters challenged the general line of thinking that DKA rates in the real world will inevitably be higher than those seen in clinical trials for SGLTs in type 1. She pointed out several flaws in the DKA protocols used across the three programs (inTandem for sotagliflozin, DEPICT for dapagliflozin, and EASE for empagliflozin), calling these protocols “not adequate” and noting that none of these protocols truly emphasized DKA prevention. Dr. Peters’ biggest qualms with these clinical trial protocols? – (i) baseline A1c ranges for participants were too high (she noted that A1c’s in the double digits are already a risk factor for DKA and therefore should not be candidates for SGLTs) and (ii) most trials did not have an exclusion criterion based off of baseline BHB levels. Attendees appeared glad to hear this optimism regarding lower DKA rates in the real world from Dr. Peters and virtually all stakeholders that we’re in touch with sound very hopeful about seeing how further education of all stakeholders will help in driving this positive outcome. See figure below for the comparison between the clinical trial DKA prevention protocols and Dr. Peters’ own suggestions.

• Dr. Peters also reviewed the various protocols that have been published regarding DKA risk mitigations with adjunct SGLT use. These include the STICH Protocol published by Dr. Satish Garg and colleagues, the International Consensus document spearheaded by Dr. Thomas Danne, and the recently published STOP DKA protocol authored by Drs. Irene Hramiak, Bernie Zinman, and others. After reviewing these protocols, Dr. Peters provided her own personal approach for using SGLTs in type 1. See second figure below:

3. Dr. Thomas Danne Pushes for SGLT Use to Treat Type 1 Diabetes and Encourages HCPs to Look at Long-term CV Benefits Rather Than Acute DKA Challenges

In a rousing session on adjunctive therapies, Dr. Thomas Danne reiterated his whole-hearted recommendation for the use of SGLT inhibitors in patients with type 1 diabetes, encouraging attendees to consider the longer term. He gave sobering statistics showing that, on average, girls who develop type 1 diabetes before the age of 10 lose 17.7 years of life (95% CI: 14.5-20.5), while boys lose 14.2 years (95% CI: 12.1-18.2), primarily due to cardiovascular disease. To this end, SGLTs could be an extremely promising adjunct to insulin, with many KOLs backing the hypothesis that class-wide cardioprotection is independent of diabetes type. To be sure, the rise of SGLTs in type 1 has led to increasing, very interested calls for a full CVOT in type 1 diabetes – which, doubtless, we’d love to see. As a reminder, three phase 3 clinical trials have been completed including (i) EASE for empagliflozin; (ii) DEPICT for dapagliflozin; and (iii) inTandem for sotagliflozin. While sotagliflozin and dapagliflozin have been approved for type 1 usage in Europe, there are currently no pending FDA approvals in the U.S (sotagliflozin was given a CRL by the FDA in March 2019).

Dr. Danne also reported that the clinical trial patients he spoke to were “very, very thrilled from day one” of using SGLTs and noticed a sudden difference in glycemic variability and predictability. Indeed, a Lexicon-sponsored poster at EASD 2018 shared findings from a phone survey of those enrolled in the inTandem studies after study completion, offering strong patient support for the benefits conferred by SGLTs. Patients using the treatment explained how the greater glycemic stability offered by sotagliflozin yielded an improved sense of well-being and reduced levels of burden and stress. Across the 5,000 participants in EASE, DEPICT, and inTandem, there was a moderate 0.3-0.5% decrease in A1c, accompanied by a more meaningful 3 kg drop in weight. Dr. Danne also pointed out that SGLTs do not lead to an increase in hypoglycemia like many other type 1 therapies, and counterintuitively, seem to have lower rates of hypoglycemia at lower A1c.

• Dr. Danne’s presentation was understandably focused on SGLTs (given his vast experience in this area) – we look forward in the future to hearing broader discussion of other possible combination therapies. Early in the presentation, he did briefly review the potential benefits of metformin, amylin analogues, GLP-1 agonists, and DPP-4 inhibitors as well; however, we note there was no mention of the potential for GLP-1 agonists to preserve beta cell health despite the fact that it was major focus of Dr. Jay Skyler’s talk on the same subject last year. While we certainly hope that the FDA moves forward with the approval of SGLT usage for type 1 diabetes, especially considering that some HCPs are already using the therapy off-label, we hope that continued innovation across the landscape is encouraged since no single therapy can be a silver bullet. 

• As always, DKA was highlighted as the most critical risk. To put this drawback into perspective, however, Dr. Danne weighed DKA fatality against long-term CV and renal failure: there was only one reported DKA fatality from the 5,000 participants, while 30-40% of the participants would develop cardiovascular or kidney disease later in life. Indeed, many people with type 1 are moved to start SGLT inhibitor treatment because of cardiovascular and renal benefits observed with these therapies in type 2 patients – on this front, we continue to advocate for a full-fledged type 1 CVOT to more clearly elucidate potential benefits (as we have since LEADER – we’d love to see a CVOT looking at GLP-1 and well as SGLTs).

• Dr. Danne strongly advocated for SGLT approval in type 1 diabetes in order to improve education. Throughout the presentation, he continuously stressed that proper DKA education and risk communication are key to safely using SGLTs. Dr. Danne emphasized that in his mind, this education can best be done with on-label usage rather than off-label. See Dr. Danne’s proposed education system below.  

• Importantly, Dr. Danne also acknowledged the “psychological barrier” preventing HCPs from prescribing SGLTs to those with type 1. He explained that HCPs can feel hesitant to prescribe SGLTs because of the immediate repercussions and fault posed by DKA; often, HCPs might feel better choosing to withhold SGLTs because patients are unlikely to come back to a provider in 20 or 30 years with CV or kidney disease asking why they were not put on the medication. However, Dr. Danne urged HCPs to focus on the long-term CV/renal benefits rather than the possible acute harm. It seems clear that DKA can certainly be prevented (perhaps through Dr. Peters or Dr. DeFronzo’s suggestions above), although it’s not possible to know about every case and we look forward to seeing more about the experiences of patients in the EU. 

Q: You showed in one slide that there was significant reduction in hypoglycemia. Is that a real effect of the drug or more so due to education?

A: We don’t stress the hypoglycemia reduction, but there was no increase in hypoglycemia in low A1c where you would expect it to happen. So, there was good news in all type 1 studies.

Q: The risk of fractures is 3x higher in those with type 1 diabetes, probably due to hyperglycemia and a problem with bone accrual. SGLT-2s do increase urinary calcium release, so if you’re giving SGLTs to a child for 10-20 years, what will happen to bone health?

A: Clinical trials are desperately needed. I just read on Twitter that NICE has recommended dapagliflozin to everyone who uses insulin above 0.5 units/kg, but whether the select patients include pediatrics is not yet known. My personal feeling is that I’m pretty sure it’s not going to be a problem

Q: You mentioned some CGM sensors will be able to simultaneously measure ketones continuously or semi-continuously?

A: Yes, it’s being discussed, but we haven’t seen any data yet. Currently, there are no trials for ketone sensors.  

Prevention and Pathophysiology

1. To CVOT or Not: Dr. Jay Skyler Speaks on Clinical Relevance of CVOTs and Wonders Aloud About Future of Trials

In a yearly CVOT update, Dr. Jay Skyler highlighted the ambiguous future of CVOTs in terms of their design and purpose and called into question whether the clinical significance achieved in these trials are worth their immense costs. Dr. Skyler peppered the audience with multiple issues challenging the current status quo of CVOTs, including trial design, the question of class effects vs. drug specific effects, the difficulty of performing CVOTs in people without established CVD (i.e., generalizability), and true clinical meaningfulness. As data that confirms cardiovascular safety and benefit for diabetes pharmacotherapies continues to roll out, there is considerable uncertainty regarding where CVOTs should and will go in the future, especially considering that no CV safety signal has yet been raised in any trial. For more on the debate, see last October’s FDA Advisory Committee meeting on the CVOT mandate. In particular, Dr. Skylar’s talk picked up on themes from Keystone 2018, where Dr. Skyler, along with Drs. Ralph DeFronzo, Robert Eckel, and Steve Nissen, noted the approaching end of the “placebo-controlled era” for CVOTs now that some therapies have been associated with a cardiovascular benefit. Indeed, we saw the first readout of an active-comparator CVOT at ADA 2019 in CAROLINA, which found no significant difference between DPP-4 inhibitor Tradjenta (linagliptin) and sulfonylurea glimepiride on the 3-point MACE, despite major differences in hypoglycemia (see more in the bullets below). In this year’s presentation, Dr. Skyler stated that he is unsure if (i) placebo-controlled trials, (ii) active comparator trials, or (iii) imputed placebo control analyses will be the most useful method moving forward. In the last option, the analysis would take a “transitive property”-like approach, allowing a drug to be proven statistically superior to placebo if it was proven non-inferior to another drug statistically superior to placebo. In terms of CVOTs on patients without CVD or patients with prediabetes, Dr. Skyler pointed out the difficulty with obtaining adequate event numbers, as these trials would require long durations and large participant numbers. As the original era of CVOTs comes to a close, with VERTIS-CV on ertugliflozin to finish in September 2019, we look forward to seeing how the conversation surrounding CVOTs and possible class effects continue to evolve.

• In a particularly impassioned conclusion, Dr. Skyler questioned whether or not CVOT results are truly clinically meaningful by citing absolute risk reduction data. Although he acknowledged that CVOTs have added important information to our knowledge base, Dr. Skyler noted that absolute risk reduction across 3P-MACE CVOTs compared to placebo is only ~1%, whereas relative risk reductions of 7-26% difference have been seen. Considering that CVOTs have enrolled more than 144,000 participants and collectively cost more billions (!!) of dollars (at $300-$500 million each), Dr. Skyler “scratches his head” at whether or not these efforts are truly clinically meaningful.

  • Further commentary on this point during Q&A contributed to whether the CVOT mandate is still necessary at all in today’s environment. JDRF’s Dr. Aaron Kowalski said that the CVOT mandate is a major barrier for diabetes companies: “A number of companies have contemplated not staying in diabetes, and one of the major barriers is the CVOT mandate.” Dr. Kowalski wondered aloud whether a different system—one in which if a safety signal is not detected early, then post-marketing studies could be used in lieu of a full-blown CVOT—could be used instead of lower the bar for more therapies to enter the market.

• On a different note, we also picked up on Dr. Skyler’s hesitation to “resurrect” sulfonylureas following the CAROLINA CVOT that compared DPP-4 inhibitor linagliptin vs. glimepiride. Trial results showed non-inferiority of the DPP-4 to glimepiride on the primary 3P-MACE outcome (HR=0.98; 95% CI: 0.84-1.14, p<0.0001 for non-inferiority, p=0.76 for superiority) and provided reassuring data concerning the cardiovascular safety of glimepiride. Regarding these results, Dr. Skyler acknowledged that some KOLs have called CAROLINA the “resurrection of sulfonylureas,” but seemed to remain less convinced himself. We point to significantly elevated rates of hypoglycemia associated with glimepiride, consistent with well-known worries with the sulfonylurea class, as reason for caution.

2. The Future of Diabetic Nephropathy: Dr. Ralph DeFronzo Emphasizes Potential of SGLT-2 Inhibitors and Advocates for Off-Label Usage

Dr. Ralph DeFronzo highlighted positive renal outcomes data for SGLT-2 inhibitors and predicted that “this class will have a very unique place in the treatment of diabetic nephropathy.” Even though no SGLT inhibitor yet has a formal indication for use in treatment of nephropathy (Invokana is currently under priority review from FDA for such an indication), Dr. DeFronzo noted that “the evidence is so strong with SGLT-2 inhibitors that they should be used off-label in combo with ACEs/ARBs.” On this point, he highlighted that SGLT-2 inhibitors hold the most promise in preventing diabetic nephropathy because of their ability to mitigate the “Ominous Octet” (a new iteration of his well-known Ominous Octet for type 2 diabetes pathophysiology) that leads to disease progression (see below). As comprehensive treatment of chronic kidney disease in patients with diabetes includes prevention of (i) progressive renal disease, (ii) macrovascular complications, and (iii) microvascular complications, SGLT-2 inhibitors can be effective in targeting all three domains. DeFronzo even went as far to say that SGLT-2 inhibitors are his “drug of choice, [and] even better than the angiotensin converting enzymes (ACEs)/ angiotensin II receptor blockers (ARBs) [for nephropathy].” He highlighted the 39% RRR in composite new onset of macroalbuminuria, doubling of serum creatinine, and death due to renal events from the EMPA-REG trial for empagliflozin, as well as the 30% RRR (95% CI: 0.59-0.82, p=0.00001) on the primary renal endpoint of ESKD, doubling of serum creatinine, and renal or CV death from the CREDENCE trial. Nevertheless, although he is a strong believer in the SGLT-2 class based on current data, Dr. DeFronzo urged for future trials to consider more diverse factors that can lead to diabetic nephropathy. The efficacy of SGLT-2 inhibitors is based on the drug class’s ability to reduce six factors of the Ominous Octet, with emphasis on their ability to block tubular hypotrophy and reduce podocyte drop-out/albuminuria. Close watchers of the field appear very eager to see what these new markers of kidney disease mean for future outcome trials, as well as for the development of diverse therapies that effectively target common comorbidities. In the meantime, be sure to check out the diabetic nephropathy competitive landscape for current trends.

• Dr. DeFronzo is less enthusiastic about the potential for GLP-1 agonists in the diabetic nephropathy space. Indeed, this seems to be consistent with the broader diabetes community, as renal results from REWIND trial (15% RRR, 95% CI: 0.77-0.93, p=0.0004 on a composite renal outcome) tempered enthusiasm from KOLs. Like others, Dr. DeFronzo pointed out that this finding was solely driven by reductions in macroalbuminuria, which according to him, is not the best marker of diabetic kidney disease. He emphasized this with REWIND’s lack of significant findings in sustained GFR decline

  

   30% (HR: 0.89, 95% CI: 0.78-1.01, p=0.07) and renal replacement therapy (HR: 0.75, 95% CI: 0.89-1.44, p=0.39).

• Dr. DeFronzo also overviewed all of the various types of interventions currently used in diabetic nephropathy. Specifically, these interventions include anti-hypertensives, ACEs/ARBs, intensified glycemic control, and low-protein diets. Antihypertensive therapy has been shown to decrease the rate of GFR decline and intensified glycemic control through pumps and artificial pancreases has decreased macroproteinuria. ACEs/ARBs have also shown promising results, but according to the hemodynamic theory of diabetic nephropathy, it’s possible that SGLT-2 inhibitors can better prevent the progression of nephropathy. With this, protein restriction (to 40-60g/day) has also been discredited because of its low impact in patients who are already on ACEs/ARBs.

3. Dr. DeFronzo Unambiguously Backs Drugs Over Lifestyle for T2D Prevention; Triple Combo Pio/Met/GLP-1 Most Effective (and ~$35/month); ADA Prediabetes Criteria “Basically Useless”

UT San Antonio’s Dr. Ralph DeFronzo voiced his unequivocal support for pharmacologic therapy over lifestyle intervention in preventing type 2 diabetes, backing a triple combination of pioglitazone/metformin/GLP-1 agonist as the best option. On its own, the evidence supporting each individual component is compelling: Metformin reduced prediabetes progression to type 2 diabetes by 31% in DPPOS, pioglitazone conferred an astounding 72% reduction (HR=0.28, 95% CI: 0.16-0.49, p<0.00001) on type 2 diabetes incidence vs. placebo in Dr. DeFronzo’s ACT NOW trial (n=602), and Novo Nordisk’s GLP-1 agonist Victoza (liraglutide) reduced the prevalence of prediabetes by ~90% vs. placebo in a 2009 study of people with obesity (n=564). The latter two results were particularly noteworthy, as TZDs and GLP-1 agonists both address the underlying pathophysiology of diabetes by preserving beta cell function – the key component to diabetes prevention, according to Dr. DeFronzo. Metformin, by comparison, does not correct pathophysiology (“It basically does nothing”), but he chose to include it in the triple combination therapy because “everyone likes it.” Further, despite Dr. DeFronzo “single-handedly bringing metformin to the United States,” it’s “number four” on his list of good diabetes drugs (presumably behind GLP-1 agonists, TZDs, and SGLT-2 inhibitors – a class he invented). Regardless, results of the triple combination are difficult to argue with. In Dr. DeFronzo’s own STOP DIABETES trial (n=422) comparing lifestyle change, pioglitazone/metformin dual therapy, and the triple combo therapy, annual incidences of type 2 diabetes were 4.1%, 1.7% (p<0.001 vs. lifestyle), and 0% (p<0.0001 vs. lifestyle), respectively. Moreover, six year results from the EDICT trial of the same triple combo therapy in newly diagnosed diabetes (presented at ADA 2018) corroborated these findings, with mean A1c remaining at 5.8% in the treatment group vs. 6.7% in the conventional therapy arm (p<0.0001), after both groups started with baselines between 8.7%-8.8%. Despite these numerous positive readouts, we’ve noted continuing debate on when to initiate combination regimens – this is so frustrating given other therapeutic areas like cancer where barely anyone (!) starts on monotherapy. For example, the writing committee for the new ADA/EASD consensus statement on type 2 diabetes management did not find sufficient evidence to recommend starting all patients on a combination from the get-go. Dr. DeFronzo acknowledged this discrepancy, beginning his presentation with a bold statement: “I can solve the problem [of type 2 diabetes prevention], but we need to use the right drugs, not what the ADA says.”

• Some members of the audience appeared skeptical of the triple therapy from a cost standpoint – a concern which Dr. DeFronzo immediately dismissed in Q&A. According to him, pioglitazone and metformin can be obtained from the pharmacy for $5 each, and all insurers are required to have a GLP-1 agonist on their formulary for a $25 copay, which can be problematic in the Medicare population but is “not a problem for any commercial insurance.” He added that the triple combination therapy is effective regardless of which GLP-1 agonist is used because of the class’ incredible efficacy, making the flexibility according to formulary access possible: “The worst GLP-1 is a great drug, the next is a fantastic drug, the next is a phenomenal drug, and the best is an out-of-this-world drug. These should be first line therapy in all patients in my opinion.”

• Dr. DeFronzo maintained that lifestyle therapy is still needed and can be the most important therapy if weight loss is achieved. However, that goal is easier said than done, especially in the long-term, and providers should first and foremost make sure that people with prediabetes and type 2 diabetes don’t gain additional weight. Moreover, he noted that even among those that achieved successful weight loss in the DPP, 40-50% still progressed to type 2 diabetes, underscoring a remaining unmet need after lifestyle-conferred weight loss in prediabetes.

• Dr. DeFronzo marked ADA’s current criteria for diagnosing prediabetes (2-hour plasma glucose of 140-199 mg/dL or fasting plasma glucose of 100-125 mg/dL or A1c 5.7-6.4%) as “basically useless.” Instead, he opts for a 1-hour plasma glucose greater than 155 mg/dL – a metric which “far surpasses” ADA’s measures, and one which he will defend in an upcoming article in Endocrine Reviews. Notably, he and his coauthors used this measure alongside the Matsuada index of insulin sensitivity (“a more pathophysiological metric for prediabetes”) to stratify the patient population by risk level in STOP DIABETES (see main paragraph above). Despite all participants having prediabetes according to the ADA’s guidelines, the “low-risk” population identified by this new classification system had an annual type 2 diabetes incidence of just 0.6% - over 1% better than the pioglitazone/metformin dual therapy group – thereby supporting the method’s validity.

4. Dr. Kathleen Dungan: Fixed-Ratio Combos, Better Injection Systems, Patient-Centered Decision Making Improve Adherence; GLP-1, SGLT-2, DPP-4 Persistence > Metformin

Dr. Kathleen Dungan provided intriguing data on medication adherence and persistence as part of a talk on practical aspects of clinical care. She first pointed to a 2017 Diabetes Care study that found a large gap between predicted changes in A1c based on RCT data and real-world outcomes with GLP-1s and DPP-4s (see below). According to the authors, approximately 75% of this treatment gap can be attributed to less than ideal adherence. Building off this point, Dr. Dungan pivoted to a 2018 study database study out of Hungary that tracked persistence with new glucose lowering therapies over two years. At 24 months, metformin was surprisingly seen to have the lowest adherence (39%), followed by GLP-1s (52%), SGLT-2s (57%), and DPP-4s (57%). Some believe that metformin’s affordability and top place in treatment algorithms lends itself to use among most, if not all, people with type 2 diabetes, including those who are less engaged in their diabetes management – we also know the drug can be incredibly hard to dose and is associated for many patients with GI side effects, unlike more recently developed compounds. While providers may be more likely to prescribe novel, more expensive therapies to those patients that are most engaged in their own diabetes care (we suspect it mainly has to do not with this, simply with their formularies for those with insurance), patients will still experience side effects. We also find it very notable that there is only a marginal difference between novel oral therapies (SGLT-2s and DPP-4s) and GLP-1s. Intriguingly (but not surprisingly), the study found that fixed combination therapies (e.g., GLP-1/basal insulin, SGLT-2/DPP-4, SGLT-2/metformin) had a higher persistence than add-on therapy. The evidence certainly backs this finding: DUAL VII for Novo Nordisk’s GLP-1/basal insulin fixed ratio combination Xultophy found that the dual therapy was less burdensome to patients than basal-bolus therapy in terms of dose adjustments, number of injections per day, and total daily insulin dose. To this end, Dr. Dungan provided several potential strategies to improve medication persistence: (i) decreasing pill burden with fixed-dose combinations; (ii) better injection systems (such as patient friendly pens instead of vials); and (iii) shared decision making between patient and provider that empower the patient in their own care.

• Dr. Dungan’s talk had us thinking of Dr. Steve Edelman’s presentation from ADA Postgrad 2018, where he proclaimed medication adherence to be the most important problem in type 2 diabetes. Dr. Bill Polonsky also spoke on the topic at that same meeting, arguing that while low medication adherence is often attributed to forgetfulness, it may instead be because patients who don’t take their medications are either not convinced of their value or are discouraged by the fact that taking medications makes them feel unhealthy. We see immense potential for added emphasis on shared decision making—as Dr. Dungan advocated for—in helping to address this aspect of adherence.

5. Dr. Mark Anderson Reviews the Successes of Type 1 Immunotherapy, Backs Combo Therapy for the Future; Calls for Screening Beyond Relatives of Those with Type 1

While anti-CD3 therapy has now been shown to delay onset of type 1 diabetes, Dr. Anderson posited that future treatments will likely involve beta cell replacement and multi-pronged approaches targeting both T and B cells to delay c-peptide loss. There’s no doubt that the difficulties in type 1 diabetes intervention are multifactorial: It’s difficult to predict who will develop type 1 for early prevention, and the larger toxicity involved with more aggressive therapy in the later stages of prevention can do more harm than good. That said, the field has certainly come a long way: The type 1 cure and prevention landscape has evolved from toxic immunosuppressants in the 80s to now including one therapy (Provention’s teplizumab) with evidence-based clinical potential in delaying type 1 diagnosis. Looking to the future, Dr. Anderson contended that compounds that modulate the CD80/86 and CD20 pathways, such as TrialNet-sponsored abatacept and rituximab, respectively, may also delay disease progression by delaying c-peptide loss. In the grander scheme, he believes that drugs that these drugs, which target the adaptive immune system (i.e., “learned” immunity to pathogens), will be the future of type 1 cures and preventions, rather than anti-inflammatory drugs like canakinumab.

• Combination therapy may be able to prevent, delay, or “even reverse” type 1 diabetes. Though no studies are yet underway, researchers “are planning” to test therapies that target adaptive immune cells (anti-CD3 or -CD20) in combination with other potential immune system modulators, such as oral insulin, T-reg therapy, or GAD (see below). 

• Dr. Anderson asserted that type 1 diabetes prevention efforts should extend beyond first degree relatives. Although first degree relatives are at a 10-fold higher risk for developing type 1 diabetes than the general population, about 90% of new onset type 1 occurs in families without any history of the disease. Doubtless, there is a clear need for screening more people for type 1 than just relatives, but we’re unsure on what the requirements for screening should be, or if the healthcare infrastructure exists to equitably screen a large population nationwide as it stands. 

6. Dr. Ahmann Attributes Improving Complication Rates to Evolving Targets, SoC; GLP-1s, SGLT-2s Important Opportunities to Further Close Gap; DKA and ESRD on the Rise as Macrovascular Outcomes Decline

Dr. Andrew Ahmann highlighted evolving treatment targets and the ADA Standards of Care as reasoning for declining rates of complications in diabetes, acknowledging that we still have room for improvement. Pointing to a 2019 Diabetologia paper, he optimistically noted worldwide decreases in both mortality as well as lower extremity amputations in both type 1 and type 2 diabetes from 1990-2015. He attributed these improvements, well as those in MI and stroke seen over the same time period, to (i) A1c targets and monitoring; (ii) recognition and focus on overall CV risk – blood pressure control, statins and LDL control, antiplatelet therapies, reduced smoking; and (iii) the benefits of ACE inhibitors and ARBs. As evidence, intensive therapy on all of these measures conferred a 53% relative risk reduction vs. conventional therapy on the 5-point MACE (CV death, nonfatal MI, nonfatal stroke, revascularization, and amputation) and a 7.9 year increase in life expectancy in the STENO-2 study. Even more conservative goals, argued Dr. Ahmann, can make a major difference in life expectancy. In fact, a 2018 Primary Care Diabetes study using NHANES data found that those who maintained a less intensive goal (A1c<9%, BP<160/110, and non-HDL <190) lived four years longer on average than those who didn’t (an obvious question here is what other confounders there may have been). However, despite this progress, the relative rates for those with diabetes vs. those without diabetes of MI (1.8), stroke (1.5), amputation (2.7), and end stage renal disease (6.1) suggest great differences remain. To this end, he pointed to cardioprotective SGLT-2 inhibitors and GLP-1 agonists, reduced hypoglycemia with CGM and closed-loop, and Amarin’s Vascepa (now included in the Standards of Care) as future opportunities to close the gap.

• Despite the improvement in macrovascular outcomes, DKA and ESRD remain on the rise. While mortality rates for the former have decreased steadily since 2000 in the US, an uptick in the overall rate was seen starting in the early 2010’s. On one hand, this could be due to the introduction of SGLT-2 inhibitors (J&J’s canagliflozin was approved by FDA in March 2013), which have a known association with DKA in off-label use for type 1 diabetes. On the other hand, we firmly believe that this risk could be managed, especially if a member of the class were to be approved, allowing for widespread dissemination of educational materials – a view also espoused by multiple KOLs. Of course, the drugs would not have to be approved for major educational material to be approved. As a reminder, while Sanofi/Lexicon’s sotagliflozin and AZ’s dapagliflozin have been approved for type 1 in Europe (for BMI ≥27 kg/m²), there are currently no pending FDA approvals in the US (sotagliflozin was given a CRL by the FDA in March 2019 and AZ received one in July 2019). With respect to kidney disease, for which no new therapy has been approved in 20 years, SGLT inhibitors have emerged as the next great hope forward. While no candidate yet has a formal indication for use in treatment of nephropathy (Invokana is currently under priority review from FDA for such an indication), Dr. Ralph DeFronzo asserted on day 1 of Keystone that “the evidence is so strong with SGLT-2 inhibitors that they should be used off-label in combo with ACEs/ARBs.” 

7. Dr. Ratner Affirms Diagnosis of Type 1 Using Autoantibodies, Firmly Believes “Type 2 Diabetes is a Beta Cell Disease”

In an engaging lecture on new diabetes classifications, Dr. Robert Ratner underscored the importance of moving past dated classification paradigms towards ones that meaningfully inform the natural history or treatment of the disease. Dr. Ratner focused on the differentiation of diabetes by pathophysiology, natural history, and prognosis – ideas highlighted in his 2017 paper in Diabetes. In regards to type 1 diabetes, Dr. Ratner focused on the JDRF and ADA’s 2015 staging system, which divides disease progression into (i) multiple islet autoantibodies with normoglycemia and no symptoms; (ii) multiple islet autoantibodies with dysglycemia and no symptoms; and (iii) symptomatic type 1 diabetes. These newer classifications establish that type 1 diabetes can be diagnosed independently of blood glucose, solely relying on autoantibody positivity. Data shows that virtually 100% of children with two or more islet autoantibodies will develop symptomatic type 1 diabetes by the age of 20, prompting Dr. Ratner to ask, “why wait to make the diagnosis?” The staging system establishes a framework that can be used to better inform regulation, reimbursement, and clinical care. In addition, he hopes that the framework will also further encourage enrollment in clinical trials that examine disease natural history and prevention.  

• Pivoting to type 2, Dr. Ratner pronounced that “type 2 diabetes is a beta cell disease as opposed to an insulin resistance disease.” To back up his claim, he cited that although well over 400 genes have been implicated in type 2 diabetes, only four of those genes (PPARG, FTO, IRS-1, and HMGA2) are not beta cell genes. In addition, when examining the natural history of type 2 diabetes, rises in postprandial glucose and fasting glucose are preceded by a decrease in insulin level caused by beta cell failure. While practically all individuals with obesity have insulin resistance, the 20% that develop type 2 diabetes are determined by whether or not their beta cells can compensate.

• To end, Dr. Ratner presented data on the Andia Cohort, which identified a new 5-group classification system for type 2 diabetes. The cohort was divided into those with (i) Severe Autoimmune Diabetes (LADA); (ii) Severe Insulin Deficiency Diabetes; (iii) Severe Insulin Resistance Diabetes; (iv) Mild Obesity-Related Diabetes; and (v) Mild Age-Related Diabetes. Interestingly, these classifications carried through to many other populations, as determined by a cluster study performed on the ADOPT study population. This system would again be of interest when considering how to treat each phenotype. Broadly speaking, Dr. Ratner firmly believes that any classification should reflect pathophysiology, provide guidance for future research, and most essentially, provide a roadmap for personalized management.

8. Dr. Kathleen Dungan Gives Overview of Current State of Adjunctive and Combination Therapies for Type 2 Diabetes

Dr. Kathleen Dungan emphasized the promise of combination therapies for their ability to improve A1c while minimizing weight gain and hypoglycemia. She presented a series of meta-analyses that showed that combination therapies should be effective based on complementary mechanisms of action and that additive effects often create better overall control in patients who are higher above target at baseline. Cost, patient lifestyle, risk of hypoglycemia, weight gain, CV benefit, and other side effects all influence the efficacy of combination therapies. Because of this, health professionals need to match pharmacology to physiology when thinking about possible drug combinations:

• Even when a promising combination is discovered, a decision must be made between early combination or sequential therapy. Early combination therapy may address pathophysiology and promote better glucose control, durability, and adherence, while sequential therapy is cheaper and easier to access an individual response from. However, it is unclear how sequential therapy affects glucose control and durability long term. A recent meta-analysis focusing on triple therapies revealed that early combination therapy with metformin can promote greater A1c reduction, but also cause hyperglycemia depending on the participant’s baseline state.