Keystone 2018 (Practical Ways to Achieve Targets in Diabetes Care)

July 12-15, 2018; Keystone, CO; Days #1-2 Highlights – Draft

Executive Highlights

  • Two days of Keystone 2018 are now in the books, and we’ve nearly been blown off the mountain by tech updates and stellar therapy commentary alike. Below, you’ll find our top 12 highlights in tech and ten in drugs – stay tuned for two more days of coverage and check out our preview for what to expect.

  • Diabetes technology: The top highlights on the tech side came in industry updates. First, Medtronic Diabetes CMO Dr. Fran Kaufman shared that the first feasibility study for the advanced hybrid closed loop system with DreaMed (auto correction boluses) is nearly complete, and the company strives for the system to confer close to 100% time in auto mode, >80% time-in-range, lower targets for overall glucose and corrections, and improved meal handling. Nice! Elsewhere in the tech-sphere, we learned that a pediatric indication for Abbott’s FreeStyle Libre is under FDA review (it’s only 18+ years currently), and the limited launch of Omnipod’s new Dash PDM is just days away (on time with the “2H18” plan). On the commentary side, we heard lots of insights from one-year-plus of experience with 670G, Dr. Irl Hirsch’s plea for every patient to have at least one CGM experience, Dr. Bruce Bode’s enthusiasm for Medtronic’s standalone Guardian Connect mobile CGM, and more.

  • Diabetes therapy: On the drug side of things, it was commentary galore. Dr. Jay Skyler shared that he hasn’t prescribed metformin in 12 years (!), instead favoring GLP-1 as first-line therapy; Dr. Michael Bush echoed this call for earlier GLP-1 use, promoting the combined efficacy and patient-friendliness of Trulicity in a Lilly-sponsored breakfast. We also got to try out MannKind’s BluHale for ourselves, and we think this device (which monitors and graphs inhalation profile) could go a long way in improving prescriber comfort and confidence with Afrezza - that’s so great in our view since nearly 50% of people with diabetes are still not at their A1c targets while less than 30% of people with type 2 diabetes don’t yet take insulin (of any kind) and a majority likely don’t have “optimized” regimens. Dr. Steve Nissen posited that we’re reaching the end of placebo-controlled CVOTs for diabetes, and he also later advocated for less-intensive glycemic control for ICU patients (we don’t think there is substantial evidence for this and we’d love to see an updated NICE SUGAR now that the tools are actually available). Also check out a systematic takedown of the US healthcare/drug system from Dr. Irl Hirsch, with an eye toward cost-effectiveness, plus Dr. Robert Eckel on the current state and future of NAFLD/NASH therapy. On type 1, we have talks from Dr. Skyler on adjunct therapy (what role could GLP-1 play? SGLT? We continue to think both could play a significant role given the enormous unmet needs of people with type 1) and prevention trial design. Rounding things out was Dr. Ralph DeFronzo on the difference between microvascular and macrovascular diabetes (a hint: hyperglycemia vs. insulin resistance) – happy reading!

Greetings from 9,280 feet of elevation, where the air is thin, but the learning is thick at the annual Practical Ways to Achieve Targets in Diabetes Care Conference (ATDC; a.k.a. Keystone). Days #1 and #2 have been an absolute doozy, with industry updates and more KOL commentary than you can shake a stick at.

See our preview for a look at what’s in store over the weekend – we’ll be back with that coverage early next week!

Table of Contents 

Diabetes Technology

1. Medtronic Advanced Hybrid Closed Loop Feasibility Study Nearly Complete; Goal of ~100% Time in Auto Mode, >80% Time in Range, More Aggressive, Simplified Meals

A little over a year since the MiniMed 670G began to launch in the US, attendees were curious: “This is great! What’s in this next-gen system I hear so much about?” Medtronic Diabetes CMO took the question at an evening symposium, leading with news that the first feasibility study of the “advanced hybrid closed loop” system (670G + automatic correction boluses with DreaMed’s algorithm) recently finished. She didn’t go into the results but shared a number of features and goals for the system:

  • Keeping people in closed loop 100% of the time;

  • >80% time-in-range without changing the hypoglycemia profile in any significant way;

  • lower target, both overall (currently 120 mg/dl in 670G) and for corrections of elevated glucose (corrections not allowed below 150 mg/dl in 670G);

  • improving the meal experience (perhaps a simple meal announcement with no carb counting – “it’ll give you a little bolus and you’ll be on your way,” said the lead 670G engineer – sounds like the Beta Bionics iLet system); and

  • constraints on time in maximum and minimum basal dose extended (in 670G, 4 hours at maximum basal dose or 2.5 hours at no insulin leads to a likely exit from auto mode – right at the times when Auto Mode is most needed!).

We’re delighted to hear that most of these updates will do wonders to address three of the biggest pain points in the 670G. Per Medtronic’s June Analyst Meeting, launch of this system was pegged for 2+ years away (after April 2020). We’ve heard heaps of anecdotes about people getting frustrated with exits from auto mode, which frequently happen when the system’s needs aren’t tended to religiously. (We remind readers that this is a first-gen commercial product; improvement is guaranteed!) Many people also complain that they can’t set their target below 120 mg/dl and that they’d prefer to have a more aggressive algorithm; however, the logic behind getting a safe version of 670G out to market ASAP made perfect sense, and we’re eager to see how aggressive Medtronic can make the next iteration while maintaining safety. Regarding meals, we heard Yale’s Dr. Elizabeth Doyle at ADA say that “The success of carb counting is going to equal the success on [670G]” – carb counting is a burdensome and highly imprecise activity, so transitioning to a new meal announcement paradigm would be welcomed with open arms. (Indeed, we know some long-standing type 1s can do well without formal carb counting; what Dr. Howard Wolpert calls an “experiential bolus” – e.g., “I know this is usually a 3-unit meal.”) Preliminary simulation work with DreaMed on the advanced hybrid closed loop with earlier and automatic correction boluses seems to pave the way for abolishing or at least minimizing the impact of carb counting. At ATTD, we saw very encouraging simulations of the advanced algorithm – in one example, advanced hybrid closed loop cut the duration of a postprandial high after an unannounced meal in half, from six hours with 670G to three hours, while also diminishing its amplitude. Tandem’s Control-IQ system with TypeZero’s automatic correction bolus feature is on pace to launch in 1H19, with recruitment for the iDCL pivotal trial underway (see Tandem’s Control-IQ will not have qualitative meal announcement, so traditional carb counting will still be needed initially.

  • In a morning Medtronic symposium, the lead 670G engineer gave a similar response, albeit less specific: “In our current development of advanced algorithms, we’re really focused on usability, and also on more aggressive therapy. We’re going to be sneaking up on automatically bolusing for meals. If you are the type of person who has good control of your diet, with 670G right now, we are able to have people on low carb diets not bolus for meals, and they’re getting pretty good control. As we get more aggressive with therapy, you could maybe not bolus, go up to 220 mg/dl or 250 mg/dl, and come back down – if that’s good enough for you, you can do that. That’s the type of therapy we’re aiming for in the next generation. We’re also looking at simplifying the meal experience; perhaps a simple meal announcement without counting carbs, and a simple meal announcement, you just say ‘I’m eating.’ It’ll give you a little bolus, and you’ll be on your way with good control.

  • True to form, Dr. Kaufman challenged the diabetes field: “Let’s at least get to a point that when we have a true artificial pancreas, no child should die from a lack of insulin. When we don’t have to spend as much time individualizing doses for everyone, we can level the playing field globally. She pitched the challenge after describing her most recent visit to the Life for a Child clinic in Ethiopia: “They have no strips, people can only get premixed insulin, and they frequently run out so they often take half their dose – I’d say that happens half the time. I hope we’ll continue to work and say that what we have in the developed world will continue to get to everyone on planet within some reasonable time frame.” This experience sounded an awful lot like the situation we saw in Rwanda in March – all over the world, kids and adults with diabetes tragically do not have access to even the most basic healthcare resources. It shouldn’t take seeing it first-hand as Dr. Kaufman has to develop a passion for helping: What more can we do?

2. FreeStyle Libre Pediatric (4+ Years-Old) Claim Submitted to FDA, Under Review

At an Abbott-sponsored corporate symposium, we learned that a FreeStyle Libre pediatric claim (4+ years-old) has been submitted with the FDA and is currently under review – great to hear, as the device is approved down to four years-old in Europe and has shown to be safe and effective in this population (SELFY). This news confirms the 1Q18 call’s expectation that filing would occur this year. Abbott has three US pediatric FreeStyle Libre trials listed on, all in the recruiting phase: (i) an accuracy study (n=85 6-17-year-olds) slated for an August completion; (ii) a post-approval safety study (n=400 4-17-year-olds) slated for February 2021 completion; and (iii) an accuracy study (n=100 4-17-year-olds) slated for a July completion. It’s unclear whether any of the data from these studies has already been included in the FDA submission. Providers, patients, and parents alike have been clamoring for a US pediatric indication for FreeStyle Libre, with some endocrinologists admitting to off-label use – our fingers are crossed for a speedy approval, as Dexcom currently has the only standalone CGM in the US with a broad pediatric indication (down to age 2). Medtronic’s Guardian Sensor 3 (GS3) in the MiniMed 670G is now approved for use down to the age of 7, while the standalone Guardian Mobile (with the same GS3) is only initially approved for 14+ years.

3. Dash Limited Market Release to Begin “Within Days”; 2-6 Year-Old Horizon AID Study to Begin on Monday

As she did at an ADA product theater a couple weeks ago, Insulet SVP/Medical Director Dr. Trang Ly demoed the recently FDA-cleared Dash PDM and Omnipod, walking attendees through the startup, bolus, and pod activation steps, as well as the Display (user) and View (caregiver) apps. Once again, Dr. Ly left the audience marveling at the system’s simplicity and user-friendliness, which stems from Insulet’s extensive human factors work. Excitingly, the limited US market release is set to commence “within days,” right in line with the “2H18” expectation. During this six-month period, Insulet will ramp inventory and secure greater payer coverage (she touted Dash’s Medicare Part D pharmacy availability, and therefore ease of prescribing), and again targeted full US market release for early 2019, followed by a global expansion in 2H19. Dr. Ly mentioned in an offhand remark that Insulet is working on getting Omnipod to her native Australia – presumably that initial entry will entail the current generation, followed by Dash.

  • Of note, a study of the Horizon hybrid closed loop system in 2-6-year-olds, PI’d by Barbara Davis’ Dr. Greg Forlenza, will begin on Monday. At ADA, Dr. Bruce Buckingham presented strong results (+2.7 hour/day improvement in the range of 70-180 mg/dl) from a five-day, free-living hotel study of Insulet’s Omnipod Horizon hybrid closed loop in adults (n=11), and two posters detailed performance in adolescents and kids ages 6-12. Horizon drove remarkable +4.4 hour/day and +3.4 hour/day improvements in time in-range in adolescents and kids, respectively. It will be fascinating to see if performance in the highly challenging 2-6 year demographic compares similarly to that in the 6-12 years-old group (Medtronic is also in the midst of a 670G study in kids ages 2-6, following approval for 7-13 year-old use last month). Dr. Ly was eager to share with us that Insulet has “fine-tuned” the algorithm since the last studies.

  • As expected, Insulet took over Omnipod distribution from Ypsomed “in the last few days,” beginning July 1st, to be specific. Dr. Ly called it a “super-exciting time for the business,” emphasizing that the move puts the company in a better position to serve European patients, and in better position for geographic expansion. In line with this news, Insulet has expanded its Glooko data management partnership to the EU market, allowing users to upload pump data to Glooko+diasend. This partnership has been terrific for Insulet and we think it is very smart to use Glooko on the software side, which is “used at approximately 7,000 provider sites by over 1.5 million users in 23 countries” (15 languages!). Hard for Insulet to tackle that on its own!

  • Why does the Dash View feature allow 12 people to follow a given patient, and for a caregiver to follow 12 people? Dr. Ly first joked that a dozen was chosen because it had to be bigger than Dexcom’s allowed number of followers (five), but then shared that Insulet actually approached the National Association of School Nurses and asked how many kids with diabetes they typically care for – when one said 12, they went with it, as it sounded like a nice, round number. Talk about detail-oriented!

4. Sugar.IQ for 670G “Very Soon” After Bluetooth 670G, “Lots New with Sugar.IQ in the Coming Months”; Patient Perspective on Sugar.IQ

Medtronic’s Ms. Michelle Shaw, with some input from Data Science & Innovation Senior Manager Mr. Pratik Agrawal, gave a high-level vision for what Sugar.IQ will become, with a promise for “lots new with Sugar.IQ in the coming months.” While the current version only works with Guardian Connect and only takes into account data related to insulin, glucose, and food, the future version will move toward leveraging a broader “digital ecosystem,” including data on mood, sleep, location, medical/claims, exercise, CRM, and biometrics. During Q&A, Mr. Agrawal chimed in to note that the team is working intently to bring Sugar.IQ to MiniMed 630G and 670G pump users, suggesting one will be available “very soon” after the Bluetooth-enabled 670G launches. Per the recent Analyst Day, Bluetooth capabilities are expected within the very broad window between now and April 2020. Still, since the new product appears to simply add Bluetooth to the existing pump hardware, a more near-term launch wouldn’t be surprising to us, especially because this would probably just be secondary display of pump data like Omnipod Display. In a follow-up conversation, Mr. Agrawal also mentioned internal deliberation about the possibility of rolling out a Sugar.IQ report, which would supplement real-time guidance with retrospective, broader trend-based insights possibly delivered by email or push notifications (á la Dexcom Clarity), which could be particularly valuable to drive conversations with providers. We asked if this would be something like the iPro2 professional CGM clinician facing software, which proposes possible causes for observed patterns – e.g., Oral medication(s) too high or incorrectly timed? Basal insulin injection in evening(s) too high or missed? – to which Mr. Agrawal said yes, only more specific and personalized. We love the idea of giving patients a guided opportunity to sit down and dig into their data – not everyone will take advantage all of the time, but as we heard from Dr. Rich Bergenstal at ENDO, optimal CGM use requires both “fast” (real-time adjustments) and “slow” (retrospective) thinking – and doing the latter can be exhausting and hard to remember and frustrating to engage with. Medtronic’s Dr. Huzefa Neemuchwala presented the latest Sugar.IQ data in an oral presentation at ADA, showing a 36-minute/day improvement in time-in-range (+2.5%-points), a 30-minute/day decrease in time >180 mg/dl, and a 6-minute/day decrease in time <70 mg/dl (all statistically significant) relative to baseline metrics. He also presented interesting findings from Fitbit integration with Sugar.IQ, plus the glucose prediction feature in the pipeline.

  • Ms. Shaw, who has lived with type 1 diabetes for 30 years and now uses Sugar.IQ, positioned the app as a way to calm the constant thought bubbles and questions swirling around the minds of people with diabetes: “Is my blood glucose higher than normal this morning? Should I bolus now or later? I should not have had the extra glass of wine last night. Now what should I do?” (Note: it does not give insulin dosing advice, at least in this version.) She emphasized that Sugar.IQ won’t inundate patients with updates – a major development priority – most people only get ~one insight per day. However, she added that the more a user engages, the more personalized and useful information they can learn. She said she started out skeptical about Sugar.IQ – only natural, as someone who has managed diabetes without the app for 30 years! The more she used it, the more she started to love seeing the insights, particularly the congratulatory insights (e.g., “You’ve only had one nighttime low in the past month. Whatever you’re doing is working very well!”). She described this positive feedback as a “high five, pat on the back, and kudos,” and said it motivates her to continue doing what she’s doing…and to continue engaging with the app.

    • Ms. Shaw confessed that she was the subject of the peanut butter glycemic assist case study we heard at least twice at ADA. (This was a cool story but we thought it was a bit overshared at Scientific Sessions.) She loves peanut butter and eats it by the spoonful. When she was first diagnosed, she was told that peanut butter was a freebie and didn’t require a bolus. When she went on Sugar.IQ, she decided to follow peanut butter, and quickly realized that it not only spiked her blood sugar, but also kept her above 180 mg/dl even four hours later. A few days after receiving the insight, she’s now able to successfully stay in target after eating peanut butter. She even receives motivational messages reflecting the improved peanut butter handling: “In the last week, you spent 6% less time above target than in the week before.” Nice! As Medtronic grows the Sugar.IQ user base, we hope they share insights about how people learn to handle their favorite, glycemic assist-worthy foods, and whether those learnings are at all generalizable to others.

    • Ms. Shaw receives Sugar.IQ notifications and insights on her Apple Watch, and so does her husband! Pushing valuable insights to the wrist is an even easier way to diminish diabetes burden and help people recognize retrospective patterns without having to “log in” to a web portal. This could be especially useful for the hypoglycemia prediction feature, which would be useful to receive ASAP. As a reminder, Guardian Connect and Sugar.IQ do not have dedicated Apple Watch apps like Dexcom’s G5/G6 – both only show the notification that appears on the phone’s lock screen.

5. Dr. Irl Hirsch: “Everybody with Diabetes Needs to Have at Least One CGM Experience”; Several Patients Have Reduced A1c by 1.5% Through Diet Alone Following Libre Wear; “Glycation Gap” -> “A1c Discordance”

The great Dr. Irl Hirsch (University of Washington) delivered a passionate talk at an Abbott-sponsored event on the importance of CGM for a wide range of patients with diabetes. He emphasized during Q&A that “everybody with diabetes needs to have at least one CGM experience for 10-14 days at some point, whether it’s the Pro (professional) or non-Pro (consumer-owned, real-time).” Dr. Hirsch finds CGM to be particularly useful in helping understand whether his patients tend to run on the high or low side of the average glucose range associated with a given A1c – as he explained, an A1c of 7% means the average glucose could be anywhere between 126-184 mg/dl. To this end, Dr. Hirsch underscored A1c as “more of a crude biomarker” of glucose control, characterizing the practice of treating patients’ A1c rather than their glucose as “bad medicine.” As an illustration, Dr. Hirsch provided a case example of a patient whose estimated A1c failed to match the laboratory value, the reason being a reduction in red blood cell survival associated with mitral valve prolapse (a new favorite case study of Dr. Hirsch’s that always seems to leave the audience in awe). Dr. Hirsch has been most surprised in how Abbott’s FreeStyle Libre has helped changed his patients’ lifestyles, regardless of whether they take insulin or not. He referenced several of his patients who have reduced their A1c by 1.5% through diet alone following CGM use. In fact, Dr. Hirsch boldly asserted that “wearing the Libre has had more of an impact on patients’ glucose control than meeting with dieticians.” However, Dr. Hirsch reminded attendees that these outcomes are accompanied with a caveat: Providers need to have sufficient time with their patients to download CGM data and identify useful patterns and trends. He referenced his presentation at AACE, during which two-thirds of the room indicated that they are allotted <15 minutes with each patient – we don’t necessarily see this changing in the near-term as the supply of endocrinologists declines and the demand of people with diabetes climbs, so we’re hoping for continued strides in the domain of decision support to help providers get more out of their time with patients. (Or, for better data management and pattern recognitions that allows patients to help themselves.) Ultimately, Dr. Hirsch asserted that CGM is “now the standard of care,” claiming that 80% of his own patients are on CGM and that “nobody with type 1 diabetes should be without this” – hear, hear!

  • During Q&A, Dr. Hirsch provided additional color on the distinction between A1c discordance and the glycation gap. He explained that there is no current agreement on the exact definition of a glycation gap but reasoned that “certainly a one-point difference” between estimated and actual A1c would be considered a gap. However, Dr. Hirsch emphasized that the field is no longer using this term and is instead moving towards “A1c discordance,” which implies an average glucose value 15% outside the mean value as determined by the ADAG trial – his team presented very interesting work at ADA finding that 34% of 1,039 patients had discordant average glucoses. Still, he advocated for a new consensus on the term, acknowledging that the field is “just learning about this.”

6. Dr. Robert Slover: Camper on 670G Had 100% Time-in-Range for 24 Hrs!; Granular 670G 7-13 Year-Olds Pivotal Data; “New Paradigm” in Diabetes Toward Time-in-Range, Variability

Barbara Davis Center’s Dr. Robert Slover gave a whirlwind talk, discussing 670G performance in peds and touting a “new paradigm” in diabetes management focused on time-in-range and glycemic variability. He overviewed a slew of data, including amazing anecdotal results from a diabetes camp and granular results from the 670G pivotal in 7-13-year-olds.

  • The overnight Camp Colorado has 280 kids with diabetes, 40 of which were on the 670G. Reading from a folded-up piece of paper he drew from his pocket a second earlier, Dr. Slover shared that one of those campers had an entire 24-hour day on hybrid closed loop with 100% time-in-range. There were audible gasps from the audience. He added that a lot of others had many days with 85%-90% time in target, remarking “if you’ve ever been to a diabetes camp, that’s darn impressive.”

  • While 36% of enrollees in the 7-13-year-old 670G pivotal study had A1c ≤7.5% at baseline, this number grew to 51% in hybrid closed loop. 88% of subjects with A1c >7.0% at the beginning had dropped A1cs by the end of the study, and all without a single severe hypoglycemia. Overnight, time-in-range in the 7-13-year-old cohort increased from 57% to 71% (+1.3 hours), while time ≤70 mg/dl was halved from 4.9% to 2.4% (-14 minutes). Said Dr. Slover, “these are adolescents and children, folks – this is really amazing to me.” Distinguishing between the two groups, Dr. Slover noted that it was much easier to run the trial with younger children than with the older ones, chalking it up to “a difference in physiology (hormones),” as younger kids were “cooperative.” A trial in 2-6-year-olds is still under way. 

    • A year following the pivotal study, the Barbara Davis Center’s adolescent cohort has maintained a mean A1c of ~7.5%. It is excellent to see the study results continue to hold in the real-world (though, admittedly, BDC is one of the best pediatric diabetes clinics in the world).

  • Dr. Slover is ready for a “new paradigm” in diabetes management because he’s tired of seeing the Rocky Mountain Range on every sheet people bring in: “You’ve all seen patients who have 20% or 25% hypoglycemia – that’s not a good quality of life…Quality of life really does change with improved glucose management. It defines the daily experience of living with diabetes. I talked years ago at a Keystone Conference about walking the tightrope between hypoglycemia and hyperglycemia – isn’t it wonderful to live in a time when we can achieve tight glucose control without the risk of severe hypoglycemia or hypoglycemia?”

7. Dr. Irl Hirsch on Basal Insulin Overnight Goals: -20 mg/dl to +20 mg/dl for Type 1s, Possibly <60 mg/dl for Type 2s; High BeAM Factor as Indicator of “Over-Basalizing”

UW’s Dr. Irl Hirsch asserted that “it’s clear we have been over-basalizing,” recommending the BeAM factor as a helpful metric in assessing basal insulin in both type 1 and type 2 diabetes. He was surprised to see that only a handful of attendees in the room had heard of the BeAM factor, which is calculated as the difference between bedtime glucose and morning glucose (e.g., BeAM for 150 mg/dl bedtime glucose and 120 mg/dl morning glucose would be +30 mg/dl). A positive BeAM factor indicates glucose declining overnight, whereas a negative BeAM factor indicates glucose rising overnight. Higher BeAM levels are linked with higher A1cs and contribute to postprandial hyperglycemia. To that, we’d add that waking up high also makes the entire next day much, much harder. As Dr. Hirsch explained, many patients with high A1cs are thus “clearly over-basalized,” explaining the drop in overnight glucose (negative BeAM), and usually should be on more prandial insulin. As for specific BeAM goals, Dr. Hirsch referenced a study in type 2s (n=1188) showing an association between a BeAM level >60 mg/dl and A1c levels >7%, suggesting that perhaps the goal for type 2 patients should be a BeAM level <60 mg/dl. For type 1 patients, Dr. Hirsch noted that he looks for BeAM levels between -20 mg/dl and +20 mg/dl in his own clinic. However, he believes the ultimate goal should be a BeAM of zero – assuming an in-range bedtime reading – although he typically sees BeAM levels above 50 mg/dl. We wonder what the correlation would be between overnight glycemic variability and BeAM – how necessary is it to evaluate both metrics?

  • Through a serious of case examples, Dr. Hirsch illustrated the need to perform a rigorous patient history in addition to viewing insulin and glucose data. In one case, he discovered that a patient’s high BeAM factor of 126 mg/dl was mainly due to consuming 80% of his calories at dinner. Similarly, another patient had a BeAM factor of -44 mg/dl due to her habit of snacking before bed. Ideally, Dr. Hirsch explained, providers would be able to look at pump downloads alongside CGM data. Dr. Hirsch reiterated previous remarks from his presentation at AACE that “our biggest gap in managing patients” is the lack of insulin data for MDI users, which he estimates comprises about 65%-70% of type 1s in the US and “almost all” of the type 2s.

8. Patients Who Spend >80% in Auto Mode on 670G Should Achieve >70% Time-In-Range; Considerations for Exercise, High Fat/Carb Meals, and Steroids

Medtronic’s Ms. Nancy Lea outlined best practices for the MiniMed 670G system, highlighting that patients who spend at least 80% of the time in auto mode should reach glycemic goals of >70% time-in-range (70-180 mg/dl), <3% in hypoglycemia (55-70 mg/dl), and <1% in severe hypoglycemia (<55 mg/dl) – of course, we’d prefer to see the consensus definition of the second level of hypoglycemia (<54 mg/dl). A key contributor to meeting these goals is of course consistent sensor wear, which Ms. Lea asserted should be at least 85% of the time. To increase the likelihood of a successful transition to the 670G, Ms. Lea strongly advised that patients spend six to seven days in manual mode prior to initiating auto mode, despite the official 48-hour required warmup – the Barbara Davis Center stretches this range to one-two weeks. She also noted the importance of comparing auto mode basal delivery to the pre-programmed basal rates, suggesting that, in the presence of a large discrepancy, providers might consider adjusting the manual mode basal rate to match auto basal delivery. After all, even with 80% in Auto Mode, patients are spending nearly five hours a day in manual mode. In a perfect world – and likely in future generations – AID systems will automatically update manual settings in the background. Not surprisingly, Ms. Lea cautioned against the practice of inputting “fake carbs” to artificially make the 670G system more aggressive. Several attendees indicated that they have encountered patients who will attempt to “beat the system” in this way. The field is still fairly mixed on the practice –  at AACE, UW’s Dr. Irl Hirsch described fake carbs as “really important” on the 670G, and many providers acknowledge that patients are going to add fake carbs and should therefore be taught to do so safely. Still, Ms. Lea emphasized the importance of helping patients understand that more insulin won’t necessarily help drop glucose faster, but rather, will push glucose down lower and to potentially dangerous levels. We think the debate is a false one – it’s not that “fake carbs” are good at bad. What matters is higher time-in-range; if someone can judiciously use fake carbs in the 670G to get more time-in-range, it has value. Still, a system that requires lying to it to achieve tighter control is definitely sub-optimal; we imagine many will appreciate the more advanced hybrid closed loop discussed in #1 above.

  • Ms. Lea detailed useful strategies for addressing several special cases while on the 670G, including exercise, high fat/high carb meals, and steroid use. She also described tactics for patients experiencing the dawn phenomenon and for those post-partum and breastfeeding.

    • For exercise, Ms. Lea underscored the importance of differentiating between aerobic and anaerobic activity. Aerobic activity is fairly straightforward, often addressable with a pre-exercise switch to a temp target of 150 mg/dl at least an hour before. Anaerobic activity, she said, requires a bit more trial and error. Ms. Lea described a patient who stays in-range by initiating a temporary target (150 mg/dl) 30 minutes prior to weight-lifting, and then reverting to the 120 mg/dl target mid-workout. Ms. Lea also noted that providers should be sure to thoroughly assess patient behaviors surrounding exercise – for example, patients who tend to eat a snack before activity will trigger a rise in the auto basal and may experience a low later. We think expectations management is also critical – exercise can easily drop OR raise glucose 100 mg/dl in a short time, a level of variability not closed loop can cope with for now. Of course, if the pattern is reliable, it can be planned for, experimented with, and managed!

    • For high fat/high carb meals, Ms. Lea assured attendees that entering carbs prior to eating can be effective for many patients. However, she suggested that for those struggling to achieve post-prandial glycemic control, patients may want to consider entering 50% of their carbs before the meal and the remaining half one to two hours later. In this way, they can mimic something that looks like a dual wave bolus. We LOVE that on Loop users can set the absorption time of the meal – Adam uses 4-5 hours for most of the low-carb, high-fat meals he eats and it works beautifully.

    • For medications like steroids that increase insulin resistance in the short-term, Ms. Lea advised that some patients be transitioned back into manual mode with a temporary higher basal rate. Importantly, she recommended that patients be kept in manual mode five to six days post-steroid use to allow the 670G to reset (since it sets aggressiveness based on total daily dose from the past six days). She also repeatedly emphasized the need for patients to turn the suspend before low (PLGS) feature on whenever they move into manual mode. A representative from Medtronic R&D noted during Q&A that the next generation system will “most likely” address the issue (though he didn’t say outright that it would automatically send people straight from auto mode to PLGS). The rationale for making people manually turn on the suspend before low feature was to prevent dangerously long periods of suspension – that is, patients who get kicked out of auto mode during hypoglycemia after experiencing >2.5 hours without insulin delivery in closed loop could feasibly continue without insulin for a total of 4.5 hours. That’s not a bad thing so long as the glucose values are low and trustworthy, but it is a valuable feature to mitigate risk of DKA in the case that the sensor is falsely reading low.

    • For the dawn phenomenon, Ms. Lea suggested that patients either check their blood glucose upon waking and deliver a correction bolus or administer a meal bolus at least 15-20 minutes before eating breakfast. However, she noted that while some patients may struggle with managing the dawn phenomenon initially, many find the issue to resolve as the total daily dose and other settings are optimized over time. In fact, Dr. Rich Bergenstal has previously suggested that dawn phenomenon is a “main indication” for hybrid closed loop. 

    • For post-partum and breastfeeding patients, hypoglycemia tends to be more frequent, requiring strategic system adjustments. Ms. Lea emphasized that 670G is currently off-label for use in pregnancy but acknowledged that occasionally patients will become pregnant on the 670G and proceed through labor with the system. Staying on label, she later explicitly said they should go into manual mode during pregnancy (Dr. Irl Hirsch recently asserted that use of 670G in pregnancy is ill-advised because it is too conservative). Ms. Lea recommended that patients be brought back into auto mode at least seven days following delivery. Post-partum, she said that setting a temp target of 150 mg/dl one-to-two hours prior to breastfeeding, snacking without bolusing, and suspending the pump 30-60 minutes before breastfeeding can be helpful in avoiding hypoglycemia associated with breastfeeding. For “cluster feeders” (i.e., babies who require nursing throughout the day), Ms. Lea recommended that providers consider keeping a temporary target of 150 mg/dl continuously, necessitating a reset every 12 hours.

Selected Questions and Answers

Q: I’ve experienced Medtronic calling a family after looking at their CareLink and talking about potentially adjusting a patient’s basal. We do pediatrics and they did say you need to call the provider. What kind of automatic follow-up is Medtronic doing? I’m not saying it’s bad, just – wow.

Ms. Nancy Lea: You might be referring to our Clinical StartRight program. It involves clinicians evaluating CareLink reports if given permission by the patients. They might flag something and reach out to the patient. They’re definitely not changing settings and always are referring back to providers.

Editor’s Note: This is only the beginning of a larger trend we expect to see – companies taking on more direct care. With Medtronic’s acquisition of the Diabeter clinic and its new MiniMed 670G Performance Guarantee, Medtronic is equipping itself to make direct changes and incentivized to do so through risk/outcomes-based business models.

Q: Any data on swimmers? I know they have to disconnect from the pump.

Ms. Lea: There’s no data or clinical trials on the issue. There are different strategies. I know we have a lot of pediatric patients who swim and a lot will disconnect. The recommendation is always that you suspend the pump. You don’t want insulin dribbling out and not being taken into account.

Comment: I’ve had success swimming laps if the pump is secured right over where the transmitter is. It’s not having to travel through water so I have been able to keep it on.

9. Dr. William Polonsky on Quality of Life Improvements from Omnipod Dash Share Feature; Working on a Booklet of Data Sharing Etiquette for Patients/Caregivers

At an Insulet-sponsored corporate symposium, UCSD’s Dr. William Polonsky reminded attendees that it is important for technology to not only drive glycemic outcomes, but also to improve diabetes quality of life. As Dr. Polonsky explained, the two are inextricably linked: If a therapy improves quality of life, patients are more likely to adhere to that therapy for longer – that said, we are unaware of any diabetes study proving this link or if it is even possible to do such a size (i.e., randomizing people to high vs. low quality of life therapies and tracking their adherence?). Dr. Polonsky cited a study showing that diabetes distress is by no means a rare occurrence: 39% of type 1s (n=390) and 35% type 2s (n=503) scored “quite high” on the diabetes distress scale. He asserted that technology can help to reduce diabetes distress by: (i) helping patients feel more in control; (ii) helping patients feel safer; and (iii) helping patients feel freer. Dr. Polonsky is especially excited to see how Insulet’s recently FDA-cleared Omnipod Dash system – set to commence a US limited launch in a matter of days – will help patients feel freer by allowing for the sharing of real-time pump data with caregivers. With the Dash share feature, Dr. Polonsky believes patients will be able to “offload the burden of diabetes for a little bit,” granting them the treasured “safe glucose getaway” by ensuring them that a loved one is able to look out for them. Enabling patients to take “safe glucose getaways” can have a major impact on diabetes quality of life – Dr. Polonsky mentioned that the second most common complaint amongst people with diabetes is the “24/7 problem” of feeling unable to escape the massive responsibility associated with diabetes. While Dr. Polonsky regards the Dash share app as “one of the neat, really cool features that I’m excited about,” he cautioned that sharing is “truly a double-edged sword” (particularly when it comes to adolescents). To this end, his team is working together on preparing a booklet of sharing etiquette – we can’t wait to get our hands on this! At Friends for Life, the Insulet booth reps said parents are extremely excited about viewing Omnipod insulin data remotely. Some lamented that Omnipod insulin data won’t go into Dexcom Share, but we expect that over time.

  • To illustrate the complex considerations that surround diabetes data sharing, Dr. Polonsky provided the following sentiments from a Dexcom Share user: “I have SO MANY unresolved feelings about this. I love that I can have people helping me, but I also hate when my type 1 diabetes messes spill out into other people’s lives. Also, when I’m dealing with a low, and I’m already a bit frazzled by it, the last thing I need is my phone beeping from worried loved ones asking if I’m OK. However, what if I’m in real trouble and I need their help? This hasn’t happened yet, but I know it could. And then, their help is exactly what I’d want.

  • Dr. Polonsky is also eager to see how the Omnipod Dash system provides insight into patients’ bolusing habits. He emphasized that for those with type 1 diabetes in particular, bolusing is “really critical.” One study in pump users showed that missing just two boluses over a two-week period can make an “enormous difference in glycemic control.” While bolus data have previously been fairly invisible until recently, Dr. Polonsky explained, by recording each bolus in the cloud, Dash will help both providers and patients “actually know what’s going on.” As more systems like Dash, not to mention connected pens and dose captures devices like Companion Medical’s InPen and Common Sensing’s Gocap, come to market, we can only imagine how such data will improve quality of care. In just one example, a Stanford/UVA/Mt. Sinai poster presented at ADA found that adults and adolescents (n=24) using smart pens in a study had late or missed boluses for 27% of their meals – such data is excellent fodder for productive patient-provider conversations.

10. 670G Dinner Panel – Insights from >One Year of Hybrid Closed Loop on the Market from Drs. Fran Kaufman, Robert Slover, Greg Forlenza, & Justen Rudolph

We attended a star-studded Medtronic dinner symposium, where we heard from CMO Dr. Fran Kaufman, BDC’s Drs. Robert Slover and Greg Forlenza, and Montana’s Dr. Justen Rudolph on their experiences with MiniMed 670G. As expected – since this group has some of the most experience with the system and the conversation was driven by Dr. Kaufman – this was one of the most insightful panels on 670G and hybrid closed loop in general that we’ve ever heard. It’s great to see the field drilling down on best practices with 670G, along with benefits, drawbacks, good candidates for use, expectation management, and more. We also heard from a rep introducing the panel that there are now >50,000 people in auto mode, which actually struck as low since Medtronic shared in May that it had already shipped >100,000 systems and >70,000 people had been trained.

  • Dr. Forlenza suggested that the behavioral modification effects of 670G are almost as important as the automated insulin delivery. At first, this seems like  a stretch, but other panelists agreed. Dr. Rudolph laid out the three-fold behavior change he’s seen with the system: (i) It turns people into good pre-meal bolusers, because they want to stay in auto mode (Dr. Forlenza noted that at camp, kids actually wanted to pre-bolus because they enjoy staying in-range, in auto mode, and receiving the ever-positively-reinforcing “blue shield of success”); (ii) It changes the conversation around the bedtime snack. He explained that the habit of eating a bedtime snack had been hammered into his DCCT, NPH-era patients, but they had found that eating a snack before bed while wearing a 670G led to overnight hyperglycemia; and (iii) It changes the conversation around how to treat lows. People accustomed to long-acting insulin have to realize that they have usually have less insulin on board when they go low with 670G: “Having that conversation and reminding patients that the system hasn’t been giving insulin for an hour, two hours – if you over-treat like you’ve been doing in past, you’ll be in the 200s right away. So reinforce that rule of 15 – 10, 15 grams of carbs for hypos. In some respect, you’re going back to the basics of diabetes education.” (On Loop, Adam often needs only 4-8 grams of carbs when he is low and has been suspended; anything more puts him into hyperglycemia.) Answering the question last, Dr. Slover simply noted that “people don’t get lazy with this” the way he’s noticed they do with standalone pumps. Wow is this great news for the field, assuming it continues as closed loop reaches a broader population. This likely hearkens back to Dr. Rich Bergenstal’s point that the amount of work won’t change on 670G, but the same amount of work will result in better outcomes.

  • Dr. Kaufman said she’s “most proud” of the way that 670G can change and “de-escalate” the diabetes conversation between parents and children. Instead of forever asking, “what’s your blood glucose? Did you bolus? …”, parents have told her that they now only want to see that time-in-range is good and the blue shield, indicating that the child is in auto mode. “They don’t care about the number, if the blue shield is there, everything must be going ok.” Dr. Slover added that parents at first want remote monitoring in their systems, but he tries to convince them that they can really trust the system so they won’t need remote monitoring as much as they once did. Of course, the ability to see if a child has a blue shield without nagging them will also be a huge win – per Medtronic’s June Analyst Day, a Bluetooth-enabled pump is slated to launch between now and April 2020. 

  • Expectation management, even before receiving the pump, seems to be the name of the game for getting patients to stick with and derive benefit from 670G. Dr. Rudolph summed up the proposition of the pump pithily: “It’s not that you’ll be in range 100% of the time, but it’ll help you prevent lows, wake up with good blood glucose the next morning. And help you overcome highs during the day. If it does all that, the system’s doing great.” Dr. Kaufman refers to the first month on the system as “boot camp – which is never fun, but once you get through that, you’re better off.” Dr. Forlenza tells his patients that when they start the device, it’ll be a degree of new learning, and that’s normal: “You’ll get frustrated with it, and there may be a time when you want to quit. Power through, when you get to the other side, blood glucose will be better with less work with fewer lows.” Dr. Rudolph simply tells patients they’ll want to throw it out the window in the first month. This is another big area where these systems can improve – providing a truly delightful experience from the start.

  • Regarding starting 670G without prior experience on sensors and/or pumps, Dr. Kaufman reported that a number of clinicians say it’s “easier to learn than to de-learn and then re-learn.” Dr. Forlenza echoed this point – for people going to straight on to 670G, it’s the first thing they know, so it’s not that it’s a big change from a previous MDI regimen. Supporting data was presented at ADA, where Medtronic’s Dr. Scott Lee showed that 241 former MDI users’ outcomes on the MiniMed 670G in auto mode were similar to those seen in the broader real-world 670G cohort (30,337 users).  

  • When asked who would be a good candidate for 670G, two varieties of answers cropped up: Dr. Forlenza specifically pointed to young children and kids going off to college (after all, he is a pediatrician). Dr. Rudolph said people with fear of lows and elderly people at high risk of hypoglycemia are good candidates (after all, he sees a lot of older patients in his practice). On the other hand, Drs. Slover and Kaufman responded with what the former called a “broad swath approach.” He astutely pointed out that it depends how you define success: “The art of diabetes care is to get people to do better. The truly ideal patients, who are motivated, and their parents are motivated, you know they’ll be stunningly successful on this. But I’ve seen other people who have not done very well previously, and they do surprisingly better than they’ve ever done before. They wouldn’t be in target 90% of the time necessarily, but maybe for the first time ever, they had a drop in A1c, and maybe they’re in target some of the time.” Dr. Kaufman added that she’ll never get some of her patients to 70% time-in-range, but she’d be happy with 50% from some – one of her patients started with 12% time in-range in manual mode.

    • Who shouldn’t necessarily be on the system? Dr. Kaufman said it’s like anything else, a bell distribution, where 670G probably wouldn’t be the best option for “someone who does nothing, nor for someone doing so many everythings all the time.” Dr. Forlenza zoomed in on the less-engaged side of the bell curve, noting that someone who doesn’t check blood glucose or give basal or bolus insulin could “figure out a way to be unsuccessful with MDI, pump, or hybrid closed loop.” Dr. Slover chimed in that patients with mental illness and those “so poor they’re just focusing on how to live” would not be the best candidates.

  • We were excited to hear panelists, unprompted, discuss glycemic goals for their patients. Dr. Slover tells his patients to aim for standard deviation of <60 mg/dl. Drs. Slover and Rudolph (and most providers who share opinions on this topic) look for 70%+ time-in-range. Remarkably, 80% of patients in Dr. Forlenza’s practice wear CGM, which gives him ample opportunity for the following conversation: “Sometimes you have a patient where it’s a perfect match: 7% A1c and CGM 150 mg/dl and time in range 70%. But if there’s a discrepancy, ask patients which they feel most accurately reflects their glycemia over the past few weeks, and most of the time, they say the CGM metrics and graphics best capture what they’re feeling.”

    • We’ve heard an easier-to-remember standard deviation (SD) algorithm from Dr. Irl Hirsch on multiple occasions: Three times the SD should be lower than the mean glucose. Given that mean glucose on 670G in the pivotal study was ~150 mg/dl, Dr. Slover’s target is actually more relaxed than Dr. Hirsch, who would say SD should be <50 mg/dl in the average 670G patient. All that said, the field is shifting toward coefficient of variation (CV) as the key measure of variability, which is SD/mean BG and adjusts for the tight correlation between SD and mean (higher mean = higher SD). Monnier et al. published a paper in Diabetes Care last year identifying a CV of 36% as a threshold to distinguish between stable (<36%) and unstable glycemia (>36%), though Dr. Hirsch prefers the slightly easier-to-remember benchmark of <33%.

  • Prompted to give an elevator on hybrid closed loop, Dr. Slover provided an interesting perspective on choosing a pump system in the current landscape: “For many patients, the choice comes down to this, and I get this question all the time: Do I do something with a sensor and an unrelated pump that is maybe easier now because I’ve heard that this 670G kicks you out of auto mode, or do I put the time investment into a hybrid closed loop system? For me, trying to be measured with the patients and explore the pros and cons, I feel that 670G is a pathway toward achieving control. It’s not just a system that tells you something, then you do something, this is something I’ve been dreaming about my entire career, something that will help you achieve better control. You will be tied into whatever pathway you choose because pumps are warranted for a number of years, but to me, the elevator speech is it’s just worth it to get on this kind of a system and finally be in tight control without fear of hypoglycemia, even though that may take more work than another pathway might take.”

    • This point was well-taken, but we view the choice as more complex: Since there is a 4-5 year pump reimbursement cycle, someone who starts on the 670G now is making it very difficult to get on a different system – systems that will be more user-friendly, not require calibration, and potentially lend to better glucose management – than the current Medtronic system. In just one example, Tandem aims to have a hybrid closed loop with auto boluses, remote updating/monitoring, and zero required calibrations in 1H19 – there could be six hybrid closed loop systems on the market by the time 2020 rolls around, Medtronic’s next-gen version included. We wonder if Medtronic will offer free remote software updates for current 670G users to get a better algorithm; we certainly hope so. 

11. Dr. Bruce Bode Enthusiastic About Guardian Connect; Suggests 80%-90% of T1s will Eventually Be on CGM

Dr. Bruce Bode expressed a great deal of enthusiasm over Medtronic’s much-improved Guardian Connect Standalone Mobile CGM, which began shipping in the US last month. Introducing the sensor with a flurry of opining interspersed, Dr. Bode commented on the “remarkable” 98.5% detection of lows (within the 15-minute period before or after your blood sugar goes below 70 mg/dl), as well as the 8.7% MARD with upper arm wear and 3-4 calibrations per day (9.1% with 2-3 cals/day). He also noted that the sensor is 80% smaller than prior Medtronic sensors (thanks in large part to removed tubing around the sensor), and said it’s easier to insert. In his experience, children (“the only group that will tell you what they’re really thinking”) say they like Guardian Connect because they can’t feel it when they put it in – “it’s almost painless.” Dr. Bode also pointed to the new adhesive, which should reduce skin irritation, and extended wear of up to seven days, though he added that some people may only get 5.5 or 6.5 days of wear – a major downside relative to 10 days for Abbott and Dexcom. During a demo of the Guardian Connect iOS app, Dr. Bode provocatively stated: “People say, ‘why not go to Abbott’s Libre, you don’t have to prick your finger?’ With Libre, it’s very difficult to get contextual data from the patient. As a clinician, we want to see what the insulin:carb ratio is, if the person exercises, what happens. And the electronic logbook helps dramatically. If you just have CGM without the contextual data, it won’t help you at all.” This feels a bit hyperbolic to us – CGM is useful whether the contextual data (food, exercise) is there or not. The real question is how much more useful CGM becomes with added contextual data vs. the burden of entering that contextual data. Guardian Connect gathers contextual data manually in the app, though over time presumably it will at least read exercise data from Apple Health. Medtronic has certainly made huge strides since the older-gen Enlite sensor, but many of the advantages Dr. Bode pointed to have been matched or bested by other systems. Accuracy is not so differentiated across the board (Senseonics and Dexcom sensors are similarly accurate, particularly when taking into account the calibration burden of Guardian Connect); Dexcom’s new auto-inserter and insertion for FreeStyle Libre are certainly better experiences than Guardian Connect; wear time is greater in other systems; etc. To us, the major value-add in Guardian Connect is Sugar.IQ, which will continue to build out and nudge people toward making smarter, better, and timelier decisions around their diabetes care and general behavior.

  • Dr. Bode displayed data from real-world use of Guardian Connect in Europe, showing that 258 patients using Guardian Connect for at least nine months from January 2017-January 2018 spent 61.4% of time 70-180 mg/dl, and just 4.1% of time <70 mg/dl. Medtronic’s Dr. Ohad Cohen showed similar data at ATTD in February, but with a smaller sample size. Dr. Bode followed this slide with data from a larger cohort (n=3,720) who had uploaded at least five days of sensor glucose data to CareLink, looking specifically at their utilization of predictive alerts. The mean sensor glucose limit for high alerts was 210 mg/dl, and 70 mg/dl for lows, and users asked to be alerted to highs and lows on average 15 minutes and 17.5 minutes before the event, respectively. Those who had alerts enabled avoided hyperglycemia excursions 39% of the time vs. 10% of the time in those who had alerts disabled, and the percent of excursions that lasted >60 minutes was nearly cut in half (63% without predictive alerts, 33% with predictive alerts). It was a similar story on the low end: Predictive alerts helped users avoid 60% of lows, compared to just 33% avoidance of lows for people who had disabled alerts, and the percent of low excursions that lasted >60 minutes was cut from 24% with alerts disabled to just 7% with alerts enabled. Dr. Bode concluded: “Obviously, if you really want to get the best out of the CGM, and you’re not attached to a pump, all you need to do is set these high and low alerts –  it works very well.” We’d love to see a Big Data comparison of alerts and percent of excursions avoided across CGMs – Medtronic, Dexcom, and Senseonics. It is clear alerts can change near-term behavior, and the nuance over the coming years will be in making alerts smarter, more accurate, more personalized, and more predictable.

  • In one of his opening slides, Dr. Bode cited BCC’s projection that the CGM market will surpass the pump market by 2022 (when it could be worth $2.7 billion). Given current trends, we might expect to see this crossover happen sooner than in four years. In 2017, we estimated CGM sales grew 46% YOY to $1.7 billion (including Libre), while we estimate pumps fell 4% YOY to ~$2.2 billion. There’s no doubt that CGM will continue to exhibit rapid 30%+ growth as more options become available and the field learns how to better deploy it in wider populations. If CGM grows another 30% this year and next year (including Libre), it will exceed $2.8 billion in sales; so the key variable is what happens to the pump field over the next few years. We do expect that pump sales will return to growth this year, driven by Medtronic/Tandem/Insulet success and other new products – there could be up to six AID systems on the US market by 2020, and the type 2 pump market could take off with the entries of BD’s patch pump, Insulet’s U500/U200 Omnipods, not to mention continued uptake of simpler delivery devices like Valeritas’ V-Go and (should they launch) Cequr’s PAQ and J&J’s OneTouch Via. Of course, many of these pumps will be either literally driven by CGM or be prescribed in tandem with CGM to optimize outcomes.

    • Dr. Bode suggested that 80%-90% of type 1s will eventually be on CGM. According to recent T1D Exchange data, this number stands at 24%, so there’s a lot of work to be done, but we hope to see the day where most, if not all, people on insulin, are on CGM, in addition to large groups of people with prediabetes and type 2 diabetes. We love hearing leading key opinion leaders say this, as the message to other providers is: prescribing CGM is clearly standard of care; are you prescribing it to most of your patients?

12. First Randomized Controlled Nutritional Intervention Study Assessing Impact of Diet on Hyperglycemia During Pregnancy “About to Start”; CGM to Help “Define and Diagnose” GDM

Norwich Medical School’s Dr. Helen Murphy announced the upcoming DiGest trial as the first randomized controlled, double-blinded nutritional interventional study to assess the impact of diet on hyperglycemia during pregnancy – using CGM! While there is an alarming dearth of data on gestational diabetes in general, one area that is particularly lacking in evidence, Dr. Murphy explained, is determining the optimal diet for improving maternal and neonatal outcomes associated with gestational diabetes. She pointed to a relatively small study (n=400) from her own clinic, which found that women with excessive gestational weight gain had increased risk of delivering large for gestation age (LGA) babies as well as instrumental delivery (i.e. assisted vaginal delivery). While Dr. Murphy acknowledged that the scope of the study limits its generalizability, she believes it is sufficient to support DiGest, which is “just about to start.” Following gestational diabetes diagnosis, women will be randomized at 30-weeks gestation to either a 1200 kcal/day or 2000 kcal/day diet. Importantly, the study will ship all meals directly to participants’ homes, eliminating a major barrier to adherence. Both meals will adhere to the Mediterranean diet and will be “nutritionally balanced” for gestational diabetes and pregnancy, composed of 50% carbohydrate, 25% protein, and 25% fat – as we understand it, calorie content is the only variable being assessed and that is an unfortunate design (see our view below). At 30 weeks and 36 weeks, women’s weight, CGM, food diary, and accelerometry data will be collected. At the final visit (six weeks postpartum) participants’ weight and oral glucose tolerance will be recorded. Dr. Murphy did not provide the primary outcome of the trial, nor did she elaborate on the CGM data – will participants wear real-time CGM throughout pregnancy/intermittently or use blinded CGM intermittently? Either way, we’re glad to see a study dedicated to better understanding gestational diabetes and hope future studies tests things the carbohydrate-insulin model of obesity (see Dr. David Ludwig’s great JAMA piece on this here).

  • Put differently, this trial is testing a diet of 150 grams of carbs per day (1200 kcal, 50% carbs) vs. a diet of 250 grams of carbs per day (2000 kcal/day, 50% carbs.) When the lower-carb AND lower-calorie group wins – which it should by a longshot – what is the scientific takeaway? To truly answer the question beyond saying “eat less,” we wish the arms being tested were matched calorie for calorie, but testing a lower-carb approach (e.g., 10%-15% carb, 20% protein, 65% fat) in one arm vs. Mediterranean in the other arm. Low-carb, high-fat is highly effective in reducing hyperglycemia and insulin utilization in many individuals, and often does not even require calorie restriction to hit those metrics.

  • Dr. Murphy asserted that CGM is “going to be increasingly useful to help define and diagnose gestational diabetes.” She believes it may help to clarify some of the controversies surrounding gestational diabetes like diet and screening practices. Moreover, she is interested in seeing how CGM informs treatment decisions – while Dr. Murphy admitted that women with gestational diabetes will have “quite different glycemic profiles” from those with type 1 or type 2 diabetes, CGM data may help focus treatment strategies on whether specific patients require diet and/or insulin intervention. Dr. Murphy also called for a repeat of the landmark HAPO trial (n=23,316), which found glucose measures to be significantly stronger predictors of maternal and neonatal outcomes than A1c – but this time using CGM. We’d be very curious to see such a trial – we hope investigators in the room were listening!

  • We were particularly interested to learn that for women with a previous gestational diabetes delivery, every 1 kg (2.2 lb) increase in weight prior to the next pregnancy is associated with a 40% increase in type 2 diabetes risk! Dr. Murphy noted that sharing this statistic with patients can be an especially impactful motivator. Given that not only women with gestational diabetes but also their offspring are at a greater risk of developing type 2 diabetes, it’s clear that any serious diabetes prevention effort must include a focus on maternal health and gestational diabetes reduction. How can we better support women with a history of gestational diabetes to ensure healthy future pregnancies? How can we improve screening efforts to prevent gestational diabetes entirely? Certainly CGM must be part of the solution!

Diabetes Therapy Highlights

1. Continuing the CANVAS Conversation: Dr. Skyler Casts Doubt on Risk/Benefit Profile of Canagliflozin, Questions Metformin’s Utility as First-Line Therapy

In a review of CVOTs, Dr. Jay Skyler offered a few provocative tidbits on how these trials should impact clinical practice, expressing serious reservation about the risk/benefit profile of canagliflozin and revealing that he has not prescribed metformin in 12 years. In discussing CANVAS, Dr. Skyler focused on the ~doubled risk for lower extremity amputations seen in the canagliflozin group (HR=1.97, 95% CI: 1.41-2.75) as a substantial factor to consider in clinical practice; he also emphasized that a fair number of the total 187 amputations seen in the trial were actually major, at the ankle or above (29%). Moreover, we note that the hazard ratios for minor (HR=1.94, 95% CI: 1.31-2.88) vs. major (HR=2.03, 95% CI:1.08-3.82) amputations were very similar, which we take as a sign that the overall signal represents a genuine risk. Dr. Skyler also presented incident rates per 1,000 patients over five years based on CANVAS outcomes; over such a period, canagliflozin would result in 23 fewer patients with a MACE event, 16 fewer patients hospitalized for heart failure, and 17 fewer patients with adverse renal events. However, canagliflozin would also give 15 more patients with amputations vs. placebo, a risk/benefit profile that gave Dr. Skyler deep pause toward prescribing canagliflozin. Over a year after CANVAS read out at ADA, the debate over safety and efficacy marches on; while our sense is that there is virtual consensus that the amputation signal from CANVAS is real, there has been a difference of opinion about potential class effects and how HCPs should change their prescribing practice in response. Some have argued that more vigilant monitoring and patient selection against amputation risk factors would minimize any negative impact, as the signal in CANVAS was driven by very high-risk patients. On the other hand, some – including Dr. Skyler – have questioned whether Invokana should be prescribed at all, given other members of the class have not shown amputation risk. That said, others have also questioned whether differences in data collection have impacted signal detection; to this end, Dr. Skyler underscored the continued need for CVOTs to be completed on every new drug in this class in order to answer the outstanding questions about class effects for SGLT-2 inhibitors. Indeed, we’re very excited for DECLARE on dapagliflozin to read out later this year, and for VERTIS-CV on ertugliflozin to complete in September 2019; additional studies of SGLT-2s in heart failure and kidney disease, in patients with and without diabetes, should further chip away at the amputation quandary.

  • Turning to metformin, Dr. Skyler emphatically stated, “Those of you walking around thinking metformin prevents cardiovascular disease – you’re deluding yourselves,” arguing for broader, earlier use of GLP-1s. He argued that studies showing sucheffects are “statistical flukes,” pointing to meta-analyses showing uncertain effects at best. UKPDS is the most commonly-cited trial in favor of a CV benefit for metformin, and we’ve heard increasing criticism that this analysis is based on 342 patients and should not be considered robust evidence. As a result, Dr. Skyler revealed that he has not prescribed metformin to his type 2 patients in over 12 years; instead, he relies on GLP-1s as first-line therapy and then adds an SGLT-2 as needed, also commenting that he thinks cardioprotection with these agents extends to low-risk, primary prevention patients if used long enough. Dr. Skyler actually asserted that out-of-pocket costs for these classes haven’t been a big problem for his patients, but he did point to Dr. Ralph DeFronzo’s inclusion of pioglitazone and metformin in his “Fab Four” of cardioprotection as very affordable generic options.

  • Dr. Skyler also called for modern clinical trials to conclusively determine the role of metformin in cardiovascular protection. This sentiment was echoed by Dr. Steven Nissen during Q&A, though the session’s speakers were not optimistic such a trial would ever be funded. As Dr. Eckel pointed out, however, the VA is funding VA-IMPACT,  a CVOT of metformin in ~8,000 patients with prediabetes that should go a long way in aiding both our understanding of metformin and the treatment of prediabetes as a disease. Of course, we would love for such an affordable drug to offer cardioprotection, but if metformin truly is not cardioprotective, we would love more conclusive data to this end. There’s no doubt too many people spend too much time only taking metformin, and any data that can drive a big shift in treatment algorithms would be welcome to us.

2. Dr. Michael Bush Promotes Trulicity’s Patient-Friendliness and Advocates for Expanded Use of GLP-1s as First Injectable

In an intimate Lilly-sponsored breakfast symposium, Dr. Michael Bush praised Trulicity’s ease of use and efficacy, calling for its expanded adoption as an earlier treatment option in type 2 diabetes while acknowledging the ongoing shift in diabetes treatment algorithms. Commenting on the sheer glucose-lowering effectiveness of Trulicity, Dr. Bush excitedly remarked about how he can “give a patient a shot on Wednesday and tell them that their blood glucose levels will be better by Thursday.” There’s no doubting Trulicity’s glucose-lowering efficacy; in AWARD-6, the GLP-1 gave non-inferior A1c reductions vs. Novo Nordisk’s market-leading Victoza (liraglutide) (-1.4% for both, p<0.0001 for non-inferiority), and the product was shown to be superior to both Sanofi’s Lantus and Merck’s Januvia on A1c lowering and weight loss. To this end, Dr. Bush called for the expanded use of Trulicity as an earlier treatment, to be added on to metformin, in type 2 diabetes, pointing to clinical inertia as a barrier to initiation of injectable therapy. Indeed, Dr. Bush supports the use of Trulicity (and presumably other GLP-1s) in the place of insulin, stating, “I don’t think your first shot [as a patient] should be insulin.” Instead, injectable GLP-1s such as Trulicity offer a lower-hassle, just-as-effective alternative to insulin injections that patients are more likely to adopt and adhere to; glucose-lowering with GLP-1s is at least as good as with basal insulin, and they also offer weight loss and carry no hypoglycemia risk (the main drawback remains GI side effects). This thinking certainly reflects a trend that has gained momentum of late; at ADA 2018, the ADA/EASD consensus committee on type 2 diabetes management formally recommended GLP-1 agonists as first-line injectable therapy over basal insulin, except in cases of catabolism and extreme insulin deficiency. While patients and providers will certainly still be limited by out-of-pocket costs, we’re very encouraged to see GLP-1s become widely considered the ideal first injectable, and we think this can go a long way in driving higher and faster uptake of GLP-1s, and maybe even improving reimbursement. Additionally, Dr. Bush emphasized that patients are much more likely to adopt Trulicity because of its ease of use; he cited survey data showing that 99% of patients felt that Trulicity was easy or very easy to use and that 97% of injection-naïve patients were willing to continue using the Trulicity auto-injectable pen, reinforcing our understanding that patients are often thought to be far more resistant to injectable therapy than they really are – another driver of clinical inertia. Generalizing from this point, Dr. Bush asserted that pens like Trulicity are “a model for how we will treat all metabolic diseases in the future...patients recognize that pen devices are effective, easy to use, and not such a big deal.”

  • All of this said, we couldn’t help but ask Dr. Bush about SUSTAIN 7 results, which demonstrated the superiority of Novo Nordisk’s once-weekly Ozempic (semaglutide) on A1c lowering (1.8% vs. 1.4%) and weight loss (~10-14 lbs vs. ~5-7 lbs) vs. Trulicity; Ozempic launched in the US February. Dr. Bush acknowledged the positivity of these results for Ozempic but also emphasized that Trulicity will continue to find success with patients because of the product’s ease of use, with a well-liked IDEO-designed autoinjector; patients are familiar and satisfied with Trulicity’s effectiveness and ease of use. Indeed, while the comparison between Trulicity and Ozempic is difficult not to make, we think the class and patients will benefit if we can focus on the unique strengths of each product: Remarkable patient friendliness and relatively very good efficacy for Trulicity, and tremendous efficacy and a fairly easy-to-use FlexTouch pen for Ozempic. Patient goals and preferences – and formulary status – will undoubtedly guide clinical decision making. Moreover, Dr. Bush uncertainties on pricing and access, due to Ozempic’s recent market entry, as another contributing factor for patients and HCPs to weigh. Certainly, this will remain an important factor through 2018 at least, but Novo Nordisk has been positive about how formulary access is progressing. Ultimately, we expect Ozempic to drive more growth in the GLP-1 market than anything, and we see more than enough room for multiple successful products given the currently low utilization of the class – only 7% of second-line prescriptions go to the class.

  • Interestingly, Dr. Bush also mentioned that if approved, Novo Nordisk’s oral semaglutide could shift patient preferences away from Trulicity, because of the general ease of use of oral medications. So far, we’ve seen data from five phase 3 trials of oral semaglutide: PIONEER 1, PIONEER 2, PIONEER 3, PIONEER 4, and PIONEER 7. Perhaps most notable, PIONEER 4 demonstrated non-inferiority to Victoza on A1c lowering and weight loss. We’re on the edge of our seats in waiting for results from PIONEER 10, which is comparing oral semaglutide head-to-head with Trulicity, and we would speculate that oral semaglutide will give comparable efficacy to Trulicity. That said, much remains to be determined as to pricing and marketing strategy for oral semaglutide, so it’s hard to make many guesses about competition at this point. We actually think Trulicity and oral semaglutide could appeal to separate patient populations: There is some inflexibility in fasting requirements for oral semaglutide (a once-daily pill), while Trulicity (a once-weekly injection) is much for flexible. As with Ozempic, we’re excited about the potential of oral semaglutide to continue expanding the GLP-1 market, as the first-ever oral GLP-1 would be a truly disruptive addition to the class

3. Hands-on with MannKind’s BluHale: Tracking Afrezza Inhalation with a New Connected Device, Currently Under Distribution as a Clinician Training Tool

In what Dr. Anne Peters later deemed, “the most unusual and fun first lecture I’ve ever seen,” Dr. Timothy Bailey led attendees of MannKind’s morning workshop through a hands-on demonstration of BluHale, the company’s new Bluetooth-enabled inhalation monitor for their ultra-rapid inhalable insulin Afrezza. According to Dr. Bailey, the most common concern among patients taking Afrezza has been whether they’re inhaling the full dose of Afrezza; indeed, we recently learned that there’s actually no indication from the inhaler or cartridges to signal to a patient that they’ve inhaled the full dose. In response, MannKind has developed BluHale, a small, Bluetooth-enabled device that can be attached to the Afrezza inhaler and uses a connected app to display a inhalation profile (below), alerting a patient if they haven’t successfully dosed the cartridge. We’ve seen concept images of the device before and were very excited to test it for ourselves. Our take? We found the device extremely intuitive and smooth to use; the automatic connection of the device to the app was faster than we might have expected, and both the inhalation and orientation sensors were precise with virtually no lag time – overall, a very user-friendly experience. The BluHale device easily clicks onto the inhaler, is turned on with the push of a button, and is charged through a micro-USB port (shown below; we’re not sure about how long a charge lasts). BluHale automatically connects to an app, which is now available on both the iOS and Google Play stores – the user interface could be a little prettier, but is quite intuitive and clear nonetheless. The app has a sectioned display, with the inhalation profile of the patient on the top half of the screen and the orientation of the inhaler on the bottom. The inhalation effort graph contains a beige guide profile; to receive the full dose, patients must inhale above the guide for two seconds, then the graph will turn green; if inhaled improperly, the graph turns red. Notably, the inhalation sensor on the device can be used without connecting to the app – if the user has properly inhaled their insulin, a small green light on the device flashes (or red light, if the insulin was inhaled improperly). While we, of course, weren’t actually inhaling any powdered insulin, we did find that meeting BluHale’s inhalation metrics was easier than not; the required inhalation was not strenuous and only took two seconds before the green light flashed. That said, our sense from previous conversations is that more of the difficulty arises from holding the powder in your lungs, and the experience is likely different when actually dosing the insulin. Altogether, we applaud MannKind for this foray into tech and think BluHale serves its purpose quite well, and we’re optimistic that the device could help overcome some HCP resistance to Afrezza.

  • BluHale also includes a three-dimensional sensor to track the inhaler’s orientation in space, a vital but sometimes overlooked component of Afrezza dosing. As Dr. Peters later said, “There are an amazing number of ways that patients can do something simple wrong [with Afrezza]” and we’re glad MannKind looked beyond inhalation alone.  The app displays the inhaler’s orientation on the bottom half of the screen, and any rotation of the inhaler beyond the recommended guidelines results in a large, red caution signal for the user, signaling that they should level the device before inhaling (below). Rotation of the device can result in a decrease in the amount of insulin actually inhaled, and CEO Dr. Michael Castagna also noted that full inversion of the loaded inhaler would result in loss of the insulin-bound powder and the need for a new cartridge. This could be a drawback (or at least an annoyance) for some patients; should the user accidentally drop or invert the inhaler a few times in a given month, they may find themselves running low on insulin. However, we don’t see this as a major issue; it’s more important that patients be aware that orientation affects the dosing process.

  • MannKind is currently distributing BluHale as a training tool in clinics, with no current timeline for release to the general public; while we would love to see the device become available to patients, we strongly believe it will have a bigger impact via clinicians. Big picture, our sense is that many patients love Afrezza; the reduced injection burden is a positive for many, but more than anything, it’s a remarkably fast insulin – as reflected on the product’s FDA-approved label. On the other hand, many clinicians have been quite resistant to prescribe Afrezza. To be sure, the concept of inhaled insulin comes with a lot of baggage: Pfizer’s Exubera was very difficult and cumbersome to use, and Afrezza itself carried REMS for patients with chronic lung disease until April – now to get the education out that this is no longer. Also very importantly, reimbursement for Afrezza was low initially – it did improve somewhat from what we understand, but many HCPS still do not realized this. Thus, to date, the product’s commercial performance has reflected this poor utilization; the product sold only $3.4 million in 1Q18. Could BluHale help more prescribers become comfortable with the concept of inhaled insulin and feel confident that they know how the device is working? We’re optimistic that it can, though it’ll likely take a lot of legwork from Mannkind’s sales force, and we can’t forget about the access barrier.

  • Later updates to BluHale may include data downloading and personal insulin logging. In speaking with MannKind’s Dr. Castagna following the workshop, we grew optimistic that these updates would be included in future iterations of the app and we’ll be eager to have this verified in the field; he pointed to InPen’s affordable dose capture platform, which includes provider access and a dose calculator, as a good model from which BluHale could draw. This would, of course, be a useful update only when BluHale goes beyond the clinic and into patient hands, and for now, we’re very excited to hear that MannKind already has remote monitoring and connected care in their sights since as noted the major value at this stage will be for HCPs. Indeed, we think many patients will only see real value in BluHale – and be potentially willing to pay for the device – with these updates. To us, it does seem unlikely that patients will want to carry, charge, and connect a device to their inhaler if all they receive is confirmation that they dosed insulin. However, should later editions of the app include the ability to share dose data with providers or upload to other programs, BluHale would certainly offer real value and convenience to patients.

4. CVOTs of the Future: Dr. Nissen Says We’re Approaching the End of the “Placebo-Controlled Era,” Thought Leaders Sound off on Trial Design

During a particularly timely Q&A, Drs. Ralph DeFronzo, Robert Eckel, Steve Nissen, and Jay Skyler shared their thoughts on CVOT endpoints and how these trials should evolve in the modern era. Most notably, Dr. Nissen asserted that the field is “rapidly approaching the end of the placebo-controlled era,” and the paradigm is running out of gas. Because we now have two classes of diabetes drugs that lower cardiovascular risk, he says, we need to take new agents and do comparison trials to see if we can do even better. He pointed to CV medicine, where ACE inhibitors were known to be good for heart failure; when Novartis’ Entresto came along (see recent background here), the FDA told Novartis they had to do a head-to-head with ACEis – and Entresto beat them, changing heart failure treatment for the better. He thinks new diabetes therapies should be asked to show that they’re at least as good, if not better than, SGLT-2s and GLP-1s. There’s also an ethical component at play; Dr. DeFronzo pointed to an analysis showing that, when SGLT-2 inhibitor use in EXSCEL’s placebo group was controlled for, Bydureon met its primary endpoint. Going forward, he expects drop-in use of cardioprotective diabetes therapies will only become more common in CVOT placebo arms, so how do we design ethical CVOTs that can still show what they’re meant to? Is it even ethical to have a placebo arm, when we know we can lower CV risk in the high-risk populations that enroll in these trials?

  • The panel also discussed whether the traditional three-point MACE (CV death, non-fatal MI, non-fatal stroke) was the best or most appropriate endpoint for diabetes CVOTs. Dr. Skyler pointed out that, in CVOTs that have reported so far, stroke is often different as an outcome; is MACE realistic, or should endpoints be split apart? Dr. Nissen is of the opinion that, particularly in EMPA-REG OUTCOME, the stroke data is unreliable due to low event numbers, a problem that extends to other trials. He explained, however, that the endpoint components were chosen as “hard, irrevocable outcomes.” A certain number of patients will come into the hospital with heart failure, be treated, and go home, but it’s not a “permanent injury” in the way stroke and MI are, and he’s against combining irrevocable endpoints with reversible hospitalizations in a composite endpoint.

  • The topic of a type 1 diabetes CVOT came up, and panelists thought it was unlikely that one would ever be conducted. In fact, conference organizer Dr. Satish Garg actually argued that it would be more or less unethical, because patients would have to be studied for so long, and it’s wrong to withhold a potentially lifesaving treatment for 20+ years. There is, of course, no approval at this stage for these classes (GLP-1 and/or SGLTs) and we would never assume that the CVOT would be a pre-approval requirement. The panel was generally in agreement that the CV safety demonstrated in type 2 can reasonably be extended to a type 1 population; while benefit is another story, Dr. Nissen felt quite confident saying GLP-1s and SGLT-2s presented no CV threat to other patient populations. That to us is not the question – from our view, a major question is CV benefit (we assume benefit or neutrality). Importantly, as Dr. Skyler pointed out, Lilly/BI’s EMPEROR program for Jardiance in heart failure is enrolling patients with type 1, and we’ll be very interested in these subgroup analyses.

5. Dr. Irl Hirsch Picks Apart US Healthcare System, Cost-Effectiveness in CVOT Era, & Argues for Human Insulin in T2s (Though He Prefers Analogs When Possible)

Every Dr. Irl Hirsch lecture on healthcare economics forces us to take a step back and ask ourselves, “wow, where to begin?” Friday’s was no exception, as he (with some help from Dr. Jay Skyler) ranted about the US healthcare system, discussed cost-effectiveness in the CVOT era with a focus on empagliflozin, and again advocated for greater use of human insulin over analogs in type 2 diabetes where necessarily. From our view it’s obvious that NPH is better than no insulin although many HCPs don’t know how to prescribe it so from a stability perspective, it will be dangerous in some patients and we’d much rather see more work on access -  Lantus for example is available to many patients at $99 a bottle as is the first mealtime biosimilar, Admelog! Alluding to a just-published Diabetes Care analysis of diabetes spending, he pointed out that in 2013, $59 billion of the $101 billion spent on diabetes in the US came from retail pharmaceuticals, and pharmaceutical spending has grown at a much greater rate than ambulatory care, nursing, inpatient services, and emergency services – “We’re running out of money, it’s obvious where all that money is going.” We would point out that $101 billion is a fraction of the $1.3 trillion spent globally on diabetes, a majority of which is undoubtedly US-based. He wrapped his talk up with two closing statements: (i) “I don’t want to read about any more deaths from people not getting their insulin,” – we know the world agrees; and (ii) “We all need to learn more about medical economics. We need to start from scratch if we don’t understand it, and teach it to our students and fellows, because this, whether we like it or not, is now part of medicine.” We agree on this as well and would like to see more on the $101 billion as well as $1.3 trillion. 

  • To illustrate one aspect of the complexity of the US health system, Dr. Hirsch showed the below graphic from the recent Insulin Access and Affordability Working Group statement in Diabetes Care. He made three primary observations on the figure. First, PBMs are the “pituitary of the entire system,” since everything has to feed back through them (not the patient!). Second, he was upset that the figure didn’t include clinicians as part of the supply chain in any respect. And third, there are so many players in the chain because everyone wants a piece of the profit, he said – we would point out, however, that many hospital executives believe that no one on the hospital end is profiting from people with diabetes even though this is a large part of spending (we’d love to see economics from the University of Washington on people with diabetes in the system – heart attacks, strokes, dialysis, etc. – this could be a great starting point for a case study and we’d love to see the investment made in early, appropriate therapy and to better understand where this is done best.)

  • Pointing to another figure from the same paper, Dr. Hirsch said that insulin companies get “a lot of heat”, and some of it’s deserved, but not all of it. From 2007 to 2016, Sanofi’s list price for Lantus has shot up by 252%. However, the devil is always in the details -  in the same time frame, the net price has only increased by 57%. While these figures absolve pharmaceutical companies to some degree, Drs. Hirsch and Skyler re-focused the microscope on them. Dr. Hirsch noted that the pharmaceutical industry is a very lucrative one, averaging 20% profit margin – this is, of course, also largely driven by therapeutic areas outside diabetes. In the last several years, of course, multiple pharmaceutical and biotech companies have exited diabetes including Amgen, BMS, Gilead, GSK, Novartis, Takeda, and many others. Meanwhile, the doctors noted that hospitals aim for 1% net margins (editor’s note - we think “aim” may be overstated), and Dr. Hirsch added that his is not clearing that low bar – we’d love to see how people with diabetes are impacting that and we’re not sure that putting people with diabetes on NPH is favorable from that perspective though we do understand it is sometimes necessary due to payers.

  • “The problem is the amount of profit, and this is where the pharmaceutical industry absolutely doesn’t want to talk,” Dr. Hirsch said. We would again suggest that 25% net margin may not be achieved across the board – most companies do not report profit by therapeutic area. Dr. Skyler pointed at PBMs for jacking up prices by adding little value while skimming large profits, and then – to grand applause – proclaimed that the pharma industry “should be ashamed for allowing that to continue, because they could go straight to pharmacies and cut out the middlemen.” We do agree little is understood about the value from PBMs and we are very excited to see what happens in new models that don’t require PBMs - Intarcia, Amazon/Berkshire/JPM are a couple of examples we look so forward to watching! Given the gross price discrepancies between insulin in the US and other countries, multiple stakeholders clearly need to change their tactics for the sake of the lives of millions of Americans. To illustrate his point, Dr. Hirsch shared that he recently meandered into a pharmacy in Spain and found a box of insulin degludec (Novo Nordisk’s Tresiba) pens in Spain for five euros (this is obviously well known and has been often cited – whether Spain and countries like it would be able to fund R&D that has enabled innovation that Dr. Hirsch uses in some patients is another question). For some perspective, when Tresiba was approved, the list price at our local pharmacy for the same box was ~$526; of course, copay cards and reimbursement bring out-of-pocket costs down, but the point stands that the US system of high list prices and corresponding discounts remains a question – who would fund innovation that eventually result in generics is also a question – presumably experts have been happy to have seen new classes ushered in that will eventually provide generics for the world.

  • Dr. Hirsch has found that patients in the pacific northwest frequently drive to Canada because they can get a vial of insulin for $30, which is less than their co-pay in the US if they were to go through their insurance – this has also been oft-cited and again we would point out that if Canada could fund innovation (with Spain) that would be terrific but it simply cannot be done with the countries that have cheap prices.

  • Importantly, Dr. Hirsch noted that 15.5% of people in the US were uninsured as of March 30th, and this number has been inching up every quarter since the 2016 election – that’s an entire other conversation that we’d love to see get more attention.

  • Dr. Hirsch shared and endorsed a number of concerns about and missed opportunities in the Trump administration’s drug pricing “Blueprint”, originally published in JAMA and well worth a read. The Viewpoint pokes holes in a number of the Blueprint’s positions (Dr. Hirsch’s commentary in parentheses): that list prices are already available to consumers; it may be too late for PBM regulation since they are merging with payers and employers to negotiate directly with drug companies; emphasis on cost-effectiveness only works when payers make difficult decisions to not cover a drug; and outcomes-based contracting has historically not worked due to rebates having easily-achievable, short-term outcomes (“Medtronic is now working to show that its tech can show better outcomes – it’s an experiment, we’ll see how it works”). The authors write that they would’ve liked to see (i) CMS given stronger authority to negotiate prices in Medicare Part D plans (“this has been a bugaboo for a long time”); (ii) authorization of the importation of generic drugs from well-regulated markets outside of the US; and (iii) prohibit the use of prescription drug coupons that incentivize patients to demand costly branded products when bioequivalent generics are available (“these coupons really do hurt everyone, including pharmaceutical companies in the end”).

  • “What has happened up until now is payers haven’t yet taken CVOTs into thinking for how to use a drug. It’s all about this class, whatever the best rebate or bundle is, that’s what we’ll use, without thinking about the patient with a toe amputated – why do I want to use canagliflozin because it’s on the damn formulary. It makes no sense!” The first part of the quote refers to the much greater cost-effectiveness in terms of cost per 1% A1c drop of metformin, SUs, and TZDs when compared to branded DPP-4s, SGLT-2s, and GLP-1s. But now that we have positive CVOTs to look to, there needs to be a recalibration on the part of payers and providers; how does CV protection (not to mention lesser risk of hypoglycemia due to glycemic-dependency) play into the equation? As at ATTD, Dr. Hirsch reviewed cost-effectiveness analyses of empagliflozin (Lilly/BI’s Jardiance) in Turkey, Italy, Greece, and the UK (the official EMPA-REG cost analysis). In all four, researchers deemed empagliflozin “extremely” cost-effective, well below established thresholds. Lacking at ATTD was a cost-effectiveness analysis of empagliflozin in the US, where Dr. Hirsch noted the retail cost is 4-10x higher than in other countries (e.g. Canada, Spain, UK). Fortunately, one was presented in poster form at ADA a couple weeks ago. In that poster (1294-P), empagliflozin was found to cost $55,489/QALY compared to standard of care ($2,394 more cost-effective than canagliflozin, using CANVAS results). Using a threshold of $50,000/QALY, empagliflozin is not cost-effective in the US, though the authors of the poster said it was cost-effective using a threshold of $100,000, reflecting how subjective “cost-effective” can be. Notably, if empagliflozin were to cost as much as it did in other countries, its cost-effectiveness could be in the range of $6,000-$14,000/QALY – extremely cost-effective by any developed nations’ criteria. The last line of Dr. Hirsch’s quote was clearly a call to think more critically about personalizing care for individuals, rather than creating and abiding by a narrow, exclusive formulary – one could argue that the payer has quite a bit to gain from switching up the approach. We often hear clinicians put a caveat on any treatment recommendation they make – “but it all depends on their insurance” – and our first reaction is that it’s unfortunate to have patient choice restricted; but could payers actually be doing harm by restricting HCPs’ ability to personalize care, particularly when we’re talking about a costly side effect like amputations?

    • Since the EMPA-REG OUTCOME study enrolled 100% of patients with existing CVD, Dr. Hirsch doesn’t believe that empagliflozin should necessarily be used in patients with type 2 diabetes and no pre-existing CVD. “At end of day, we need to understand the granularity of these CVOTs. I don’t think the majority of practitioners understand the granularity of the data. It’s complicated. There’s a lot of it, and a lot more coming.” That said, we’ve also heard prominent thought leaders, including Dr. Mikhail Kosiborod and – at this very meeting – Dr. Skyler, assert that SGLT-2s and GLP-1s would likely offer CV protection in a primary prevention population, if used for long enough. However, that raises a new question, particularly relevant to the issue of cost-effectiveness: For how long would low-risk patients have to take these to see an effect? Our sense is that the cost effectiveness would be lower in this use case, but we also imagine the magnitude of the benefit to be gained over 20 years of GLP-1 agonist use could be great.

  • As at ATTD, Dr. Hirsch relayed analyses showing that insulin analogs are highly cost-effective in type 1s, but not at all in type 2s – in type 2s, the health economics evidence points to human insulin, but prescription rates haven’t followed. The big difference in the presentation of the data was that instead of comparing the cost-effectiveness of analogs in type 2 diabetes ($425,913/QALY gained) to that of a pediatric heart transplant, he instead compared it to a bone marrow transplant. “The evidence for the use in type 2 diabetes is not convincing. People are shaking their heads right now – why? These are the data.” He showed calculations from an additional 2009 publication. This study assessed the probability that given analogs would be cost-effective in both type 1s and type 2s, arriving at similar findings to the first paper. For example, the study found that NovoLog in type 2s had a <30% chance of being cost-effective vs. regular human insulin at a threshold of $50,000/QALY (in type 1s, there was a ~50% probability of being cost-effective at the same threshold). “So,” Dr. Hirsch posed, “which would you rather use? Human or insulin analog?” Other factors to consider: Both types reduce blood glucose the same, and prandial analogs reduce hypoglycemia relative to regular insulin in type 1, but not in type 2; in both type 1 and type 2, basal analogs reduce hypoglycemia relative to analogs (though just a 6.5% absolute risk reduction). In a recent JAMA paper, Yale’s Dr. Kasia Lipska et al. showed in 25,489 Kaiser Permanente members starting basal insulin, there was no statistical different in hypo-related ED visits or hospital admissions in those given NPH vs. those given basal analogs (admittedly, a crude endpoint). Still, human insulins accounted for just ~5% of US insulin prescriptions in 2017, down from ~10% in 2009. Dr. Hirsch, himself, says he doesn’t like to use human insulins despite the evidence (we’ve heard him praise new insulins, especially next-gen basals), though he does when needed due to issues of coverage or affordability.

Selected Questions and Answers

Dr. Larry Hirsch (BD): Irl, always a great talk and perspectives. When we look at the cost of drugs in the US vs. almost everywhere else, it’s dramatically different. The reality is that the vast majority of products in the current therapeutic armamentarium come out of the US. They come out of companies based here, even if their headquarters are in Europe – the R&D facilities and things are here in the US. I just worry that if we take an extremely broad, blunt approach on drug pricing that the ability to produce these new innovations will be impaired – I’m no apologist for the pharma industry. For full disclosure, I worked at Merck for 18 years, so I have an idea of what it takes to bring a drug to market. We take CVOTs for granted, but each costs in excess of $60-$100 million. I love your perspective, but I’m just worried that we’ll end up with a blunt approach to drug pricing and we’ll rue the day we do that.

Dr. Irl Hirsch: I didn’t show you the data, but you are in a very lucrative industry, Dr. Hirsch, and you’ll have a great retirement, probably better than most people in this room. Your industry averages 20% profit margin – not-for-profit hospitals need to be at least on the positive side of the ledger to maintain viability, and I’ll tell you we’re not even hitting it. The problem is the amount of profit, and this is where the pharmaceutical industry absolutely doesn’t want to talk, and that has to do with profit margins.

Dr. Larry Hirsch: I work in the device industry, where the margins are far lower.

Dr. Jay Skyler: The problem when talking about profits is, what value do PBMs add to the system for the profits they take out? Without all the rebates and other things, prices could go down further, and the other problem is the Group Purchasing Organizations (GPOs) who work for hospitals and keep the prices up. They take so much off the top. The pharmaceutical industry should be ashamed for allowing that to continue, because they could go straight to pharmacies and cut out the middleman.

6. Dr. Eckel Emphasizes Importance + Effectiveness of Weight Loss in NAFLD/NASH, Recommends Pioglitazone + Vitamin E as Currently-Available Therapies

Renowned Dr. Robert Eckel delivered a thorough overview of the NAFLD/NASH therapeutic landscape, emphasizing the diversity of potential treatment options in development and placing hope in combination therapies. Speaking to the lacking of approved NASH therapies, Dr. Eckel began his talk by proclaiming that his “whole talk [was] about unapproved drugs” and that HCPs today are simply “shooting from their hips” in trying to treat this burgeoning disease. In terms of proven therapies for the treatment of NAFLD and NASH, Dr. Eckel emphasized, above all, the importance of HCPs encouraging their patients to lose weight. He cited data showing that increased weight loss is directly associated with higher proportions of NAFLD resolution in patients; in a similar vein, more weight loss has also been shown to lead to greater regression of NASH. While one study demonstrated nearly 100% resolution of NAFLD in patients who lost >10% weight from baseline over 12 months, Dr. Eckel tempered expectations for this result by reminding the audience just how rare such dramatic weight loss really is. In terms of current and readily available (albeit off-label) treatments, Dr. Eckel commented on pioglitazone (which he prescribes for his patients with NAFLD/NASH) and Vitamin E, both of which have been shown in the PIVENS study to improve steatosis and increase the rate of NASH resolution. Vitamin E treatment is especially intriguing, considering how readily accessible and affordable it is for most patients, and we note that AASLD’s 2018 Guidelines recommend Vitamin E as first-line therapy. Dr. Eckel also briefly touched on the potential to use GLP-1s in NASH, citing a meta-analysis showing positive effects in terms of reducing steatosis and lobular inflammation. We are especially keen on the idea of GLP-1 use for NASH treatment, seeing as it has tremendous weight loss effects, along with potential benefits on inflammation, glucose metabolism, insulin resistance, and lipids, all of which could potentially contribute to NAFLD/NASH pathogenesis. Be sure to check out our NAFLD/NASH competitive landscape for more information on the growing number of candidates in development.

  • Dr. Eckel outlined the three main contributing factors to NAFLD/NASH pathogenesis (lipotoxicity, fibrogenesis, and inflammation) and underscored that combination therapies could be effective because of their ability to simultaneously treat all of these factors. Dr. Eckel particularly highlighted Gilead’s selonsertib, tentatively expected to be the first NASH therapy to hit the market, as well as the company’s other two therapies, currently being tested in combinations with selonsertib. The use of combination therapies in NASH remains a hot topic, and our sense is that the field very much believes combining multiple therapeutic targets will be necessary to maximize effectiveness of NASH treatment – to this end, we note that companies are already testing combination therapies, despite the lack of any approved NASH monotherapy.  We love the idea of moving toward targeted therapies, and in many therapeutic areas, monotherapy (!) would never be considered – in our view, doing the research and getting (even more) data on the value of combination therapy would be very helpful since many doctors do not seem to see the value or do not have the time needed to learning how to prescribe it most effectively. 

7. Dr. Skyler Underscores the Need for Adjunctive Therapy in Type 1, Calling for Dedicated Investigation of DKA/SGLT Relationship + Early GLP-1 Use for Beta Cell Preservation

Dr. Jay Skyler advocated for investigation and use of GLP-1 agonists in type 1 diabetes, particularly for beta cell preservation early in the course of disease, also emphasizing the lack of robust data on SGLT inhibitor use and DKA incidence, making clinical best practices uncertain. In just the last few weeks, we’ve observed renewed excitement for the role of GLP-1 agonists in type 1, on the heels of very positive data for liraglutide in type 1 from Dr. Paresh Dandona’s clinic. To demonstrate the potential of GLP-1s in type 1 diabetes, Dr. Skyler reviewed positive data on exenatide (AstraZeneca’s Bydureon) and liraglutide (Novo Nordisk’s Victoza) in type 1, where both have demonstrated efficacy on improving A1c, reducing insulin dosage, and promoting weight loss. Somewhat flabbergasted, Dr. Skyler wondered why Novo Nordisk stopped development of liraglutide in type 1, emphasizing his belief that liraglutide could be a successful adjunctive therapy.  During Q&A, an informal poll of the audience reinforced our understanding that GLP-1s are, indeed, commonly prescribed to patients with type 1. But as we recently learned from Novo Nordisk’s North American CMO Dr. Todd Hobbs, the company stopped development due to modest overall results and significant heterogeneity in the ADJUNCT ONE  and ADJUNCT TWO trials. Apparently, the difference in how people respond and need for very individualized insulin adjustments makes it difficult to design a trial that meets regulatory guidance; an audience member added that Novo Nordisk was startled by the number of unexpected hypoglycemia and DKA events, leading Dr. Skyler to highlight the need for expert use of these as adjuncts. He recommended utilizing them in conjunction with CGM, to get insulin adjustments right, and argued that trial sites should have been selected more carefully for expertise in type 1. All of this said, Dr. Skyler seemed, by far, the most excited about the potential for GLP-1 agonists to be used early on, soon after diagnosis, to preserve beta cell health and extend the “honeymoon period” of residual insulin secretion. Indeed, Novo Nordisk does have an ongoing phase 2 trial (n=304) of liraglutide +/- anti-IL 21 therapy in newly-diagnosed type 1, expected to complete in March 2019. Extending beta cell life can offer a longer period of better glycemic control, theoretically improving long-term outcomes but also helping patients feel better in the short-term, and we see no reason why a patient couldn’t continue on liraglutide for an extended period. Victoza going generic will only make this easier, and we love the idea of a generic that can help people with type 1 lose weight, need less insulin, spend more time in range, and maybe even reduce CV risk; we’d also love to see academia, at least, continue to investigate GLP-1 in type 1 so that more providers can use this safely, off-label.

  • On SGLTs in type 1, Dr. Skyler decried the lack of hard data on preventing DKA with SGLT inhibitor use in type 1, suggesting a potential role for lower doses in optimizing the risk/benefit profile for patients. SGLTs for type 1 could hardly have been a more prominent theme at ADA 2018, reflecting the deep excitement within the field for a first-ever oral adjunct for type 1. Indeed, Dr. Skyler echoed this enthusiasm, commenting on these agents’ utility in reducing glycemic variability and A1c levels; however, he inevitably turned to the increased risk of DKA they carry in type 1, commenting that there is still no solid consensus among HCPs and thought leaders on the specifics of preventing this complication. Dr. Skyler lamented the simultaneous lack of data and surplus of opinions on the topic in calling for a randomized, controlled trial to concretely look into the subject. Indeed, there is a fundamental lack of evidence that patient or provider education would help at all, and we would love to see Sanofi/Lexicon, AZ, and Lilly/BI partner on developing and testing a risk mitigation program. Another issue that needs investigating? Whether lower-dose SGLTs could offer the similar benefit as higher doses while presenting no significant DKA risk. Discussing preliminary results from EASE-2 and EASE-3 that were presented at ADA 2018, Dr. Skyler was intrigued by the lack of increased DKA risk with 2.5 mg empagliflozin (1/4 of the lowest dose available for type 2), which also maintained significant glucose and weight loss benefits vs. placebo. Dr. Skyler wondered whether “we might look to prescribe lower doses in type 1 patients compared to what we do with type 2 patients” in order to reduce DKA risks, though we note that efficacy in the inTandem program for sotagliflozin and the DEPICT program for dapagliflozin indicate a likely trade-off in efficacy with lower doses and no numbers have been released on the EASE program (to be presented at EASD 2018). Additionally, he pointed out that sotagliflozin was developed – and doses selected as such – for type 1 from the get-go, so exploring a lower dose presents some confusion.

8. Dr. Skyler Advocates for Better Primary Endpoints in Diabetes Prevention Trials, Gives His Key Elements of Successful Clinical Trials

As the head of the landmark DPT-1 trial, Dr. Jay Skyler offered his insights into successful clinical trial design for diabetes prevention therapies, keying in on primary endpoint identification and clinical trial therapy selection. Only three therapies for type 1 autoimmunity have progressed to phase 3 clinical trials to date, each of which, according to Dr. Skyler, has taught us a “good, bad, or ugly” lesson. The “good” lesson underpinned a main takeaway of Dr. Skyler’s talk: primary endpoints in diabetes prevention therapy trials should be carefully structured. In one of the few phase 3 trials, the primary endpoint (which was negative) was a composite of A1c and insulin dose; Dr. Skyler felt this was an inappropriate conversion of two continuous variables to an acute dichotomous variable and didn’t accurately measure the delay of type 1 diabetes. In his view, preserved C-peptide levels, which was positive in the trial, would have been a superior primary endpoint and supported the efficacy of the therapy; in his view, the the researchers missed out on a huge opportunity simply by misidentifying their primary endpoint. Moreover, primary endpoints should be based on all of the data that is collected in a study – case in point: the currently accepted measure of C-peptide levels is the two-hour area-under-curve (AUC) from a four-hour mixed-meal-tolerance-test (MMTT); Dr. Skyler’s response – why not use the AUC of the whole four hours? He gave several examples in which this extension would have made a significant difference to the outcomes of immunosuppressant clinical trials. The “bad” lesson, according to Dr. Skyler, related to inconsistent design of progressing clinical trials; in another of the few phase 3 clinical trials for type 1 autoimmunity, researchers used 1/16 of the dose that was used in previous trials in the hope of completely eliminating side-effects. Unsurprisingly, this also eliminated all beneficial effects of the therapy. While conservative guidelines such as these are common practice in clinical trials (as the safety of the patient is on the line), Dr. Skyler stressed that researchers searching for diabetes prevention treatments need to have, “the courage to study interventions with potential adverse side-effects.” With over 400 therapies having been found to prevent or delay type 1 diabetes in the NOD mouse (albeit with several of them being trivial, such as cage height), researchers should not only consider the ease and safety of a therapy when selecting it for clinical trials, but also weigh the risk/benefit to individuals and society. It’s important to note that Dr. Skyler was not advocating for reckless abandon in clinical trial selection; he was suggesting further risk/reward analysis as a means to move the field forward. In his own work, Dr. Skyler combines this risk/reward evaluation with two criteria to identify candidate drugs for type 1 diabetes autoimmunity: (i) success in animal model(s), typically the NOD mouse; and (ii) clinical use in other autoimmune diseases. Lastly, the “ugly” lesson, which Dr. Skyler did not dwell on, came from the last phase 3 clinical trial for type 1 autoimmunity, which was found to be fraudulent. Scientific integrity was a consistent theme throughout the presentation.

  • Dr. Skyler identified statistical power and robustness as two of his key elements for a successful clinical trial. He underscored that conclusions and clinical trial decisions cannot be drawn from small, pilot studies; for example, an underpowered study conducted by the great Dr. George Eisenbarth showed statistically significant delays in the development of type 1 diabetes when those at risk were given preemptive insulin. However, Dr. Skyler’s famous DPT-1 study, which was adequately powered and enrolled, dispelled this myth. On the topic of robustness, Dr. Skyler identified current standards of significance to be less than sufficient. In his view, “replication is at the heart of scientific endeavor,” and p-values of 0.05 which give only a 50% chance of replication are arbitrary and outdated; instead, Dr. Skyler suggested that a stricter p-value of <0.001 should be the goal. While we are not mathematicians nor experts in trial design, we wonder what the effect of this change in standard would have had on the trial landscape in a post-hoc analysis. For example, none of the p-values for superiority over placebo in the four positive CVOTs thus far (EMPA-REG OUTCOME for empagliflozin, CANVAS for canagliflozin, LEADER for liraglutide, or SUSTAIN-6 for semaglutide) would have met this standard. While a new p-value threshold would not change the objective efficacy of any of these therapies, we find it interesting the incredible implications of these “arbitrary” cut-offs on our understanding of drugs and their effects.

9. Dr. Steven Nissen Challenges the Utility of Intensive Glycemic Control for ICU Patients, Endorses ACP Guidelines Calling for Moderate Glucose Control

In a talk attempting to reconcile conflicting clinical trial evidence on the use of intensive glycemic control in ICU patients, Dr. Steven Nissen (Cleveland Clinic, Cleveland, OH/USA) criticized this treatment practice and rather supported more moderate glycemic control treatment protocols. Hyperglycemia remains a significant issue for patients in ICU, as high glucose levels have been associated with significant increases in viral infections, total number of hospital days, and mortality rates—however, Dr. Nissen repeatedly emphasized that such data does not indicate a causal relationship between hyperglycemia and these outcomes. Dr. Nissen reviewed the surgical ICU Leuven study, which seemed to indicate benefits of maintaining blood glucose levels no higher than 110 mg/dL in terms of lowering mortality rates in patients with intensive glycemic control (4.6% vs. 8.0% for intensive vs. conventional, p=0.04). However, the follow-up medical ICU Leuven study did not show such benefits, and instead demonstrated significant increases in hypoglycemia (18.7% vs. 3.1% for intensive vs. conventional, p <0.001), which Dr. Nissen characterized as “significant evidence for significant harm caused by this type of treatment.” In light of the conflicting results of these two studies, Dr. Nissen discussed the NICE-SUGAR study, which contradicted the Leuven studies and demonstrated a statistically significant level of harm for intensive glycemic treatment in terms of mortality rates (27.5% vs. 24.9% for intensive vs. conventional, p=0.02) and no positive results in terms of secondary endpoints. Consequently, Dr. Nissen aired his opinion that intensive glycemic treatments should not be pursued for patients in ICU, and that a conventional treatment range of 140-180 mg/dL should be targeted.

  • Dr. Nissen endorsed ACP guidelines which recommended not using intensive insulin therapy and instead targeting blood glucose levels of 140 to 200 mg/dL. He seemed to criticize outdated ADA/AACE guidelines that were initially vague and simply advocated for “individualizing care” without any specific guidelines. However, we do note that the ADA and AACE have since updated their position on the issue which now more closely align with the ACP recommendations. This point was echoed by Dr. Irl Hirsch (a contributor to the ADA/AACE guidelines) in the Q&A session following the talk, where he stated that “we are more in agreement than disagreement…the ADA and AACE numbers are similar to yours.” In our view, it’s important that hospitals attempt to adhere to these less stringent control guidelines, as it seems as if many hospitals have disregarded the idea of glycemic control after the results of the NICE-SUGAR study. We believe that good glycemic control is still essential and must be done responsibly in order to reduce hypoglycemic events.

  • Dr. Irl Hirsch also expressed a desire for the NICE-SUGAR trial to be redone with CGM usage. Dr. Hirsch lamented the fact that NICE-SUGAR (completed in 2009) was done before the widespread advent of CGMs, proclaiming that “I think that if we could do a multi-center trial with CGMs, I think we could see [positive] differences [with intensive glycemic control], because the hypoglycemia really did us in [in the original NICE-SUGAR study].” What Dr. Hirsch said clearly resonated with the audience – hypoglycemic events could be better avoided with CGM use, and then it would be possible to decouple the effects of hypoglycemia from intensive glycemic control in contributing to outcomes for ICU patients. We are very eager to see such results and urge thought leaders to pursue this idea.  

10. Dr. Ralph DeFronzo Argues that Cardiovascular Disease is Well Underway Long Before Clinical Presentation of Diabetes, Pointing to the Pathophysiology of Insulin Resistance; Calls for Greater Crossover Between Cardiology and Diabetology

In a scientific tour de force, Dr. Ralph DeFronzo attributed type 2 diabetes, CV disease, and a host of other cardiometabolic diseases to underlying insulin resistance, arguing that type 2 diabetes is actually two diseases with two distinct pathogenic sequences and clinical presentations: the microvascular and the macrovascular. We find this paradigm quite useful; while the pathogenesis of microvascular complications is well-established as a sequela of serious and chronic hyperglycemia, Dr. DeFronzo explained that hyperglycemia has a present but very weak impact on the pathogenesis of atherosclerotic and other cardiovascular complications. In UKPDS, lowering A1c had no impact on CV complications. This, Dr. DeFronzo outlined, is because CV complications in patients with type 2 diabetes actually arise from the same insulin resistance that contributes to the development of diabetes: CVD is not secondary to type 2; rather, both are secondary to insulin resistance – as are visceral obesity, hypertension, dyslipidemia, NAFLD/NASH, inflammation, hyperinsulinemia, and more. To this end, Dr. DeFronzo actually conducted his own, unpublished meta-analysis (n=26,964) of patients with acute MI and/or coronary syndrome who weren’t known to have diabetes and prediabetes. He found that, when those patients came to the hospital, 22% were found to have diabetes and 32% to have prediabetes. Perhaps more poignantly, he also found that the annual incidence of recurrent CV events in those with newly-discovered glucose intolerance was 12% for diabetes, 9% for prediabetes, and 5% for non-diabetes. In Dr. DeFronzo’s assessment, this points to an incredibly strong tie between subclinical diabetes and longstanding CVD, and he further demonstrated that lean type 2 diabetes, obesity with normal glucose-tolerance, essential hypertension, hypertriglyceridemia, coronary artnery disease, and impaired glucose tolerance are all associated with high insulin resistance, manifested as impaired glycogen synthesis. On a molecular level, he explained, the blocked insulin signaling pathway (i.e. insulin resistance) that is commonly associated with diabetes – broken IRS-1/PI3K/Akt signaling – also blocks the generation of nitric oxide, linked diabetes and CVD right from the beginning. What’s more, when that pathway is broken, insulin’s activity as a growth factor is magnified by resulting hyperinsulinemia, giving rise to excessive MAPK signaling, which has a downstream effect on inflammation, atherosclerosis, and vascular smooth muscle cell growth. Indeed, Dr. DeFronzo has actually been advancing the insulin resistance hypothesis for over two decades, but fundamental, important questions remain with respect to the path forward: How do we treat the biological underpinnings of insulin resistance and cardiometabolic syndrome, would should an earlier intervention target, and how can we better understand the diverse etiologies and phenotypes that arise from insulin resistance?

  • Addressing the “ticking clock hypothesis,” Dr. DeFronzo asserted that problems with coronary artery disease start when you’re born (or at least, very young), citing this as support of the insulin resistance hypothesis. He attributed this time-course largely to genetics, presenting evidence that, in the normal-glucose-tolerant offspring of two parents with diabetes, insulin resistance is already present, which Dr. DeFronzo takes to mean there are people nearly destined to have diabetes later in life, who already have the molecular components of the disease decades before its manifestation. He describes this phenomenon as “sitting in a sea of insulin resistance syndrome.” In contrast to macrovascular disease, however, the “ticking clock” for diabetes and microvascular complications doesn’t start until much later, with the onset of chronic hyperglycemia.

  • Dr. DeFronzo’s key practical takeaway is that cardiologists and diabetologists need to take greater interests in each other’s work, particularly in an era in which diabetes drugs are indicated for reduced CV risk. In his observation, cardiology doesn’t appreciate metabolic aspects and endocrinology doesn’t appreciate CV aspects, but patients would benefit if they did. Indeed, this call to action has been echoed by a number of thought leaders over the past year or two, as we’ve noticed an ever-increasing presence of diabetes data and commentary at cardiology meetings. In fact, during the plenary Q&A, Dr. Robert Eckel made an impassioned call for the creation of a new medical subspeciality in cardiometabolic medicine that would pick and choose skills from endocrinology and cardiology to train experts in diabetes, obesity, lipids, and associated CV complications. This piqued our interest; there’s no doubt that endocrinology is facing a serious shortage of physicians, and in the face of today’s diabetes epidemic, does it make sense for every endocrinologist to be trained both diabetes and other endocrine disorders? Further, Dr. DeFronzo reiterated his recently-coined “Fab Four” (metformin, pioglitazone, GLP-1 agonists, and SGLT-2 inhibitors) as the four therapies worth using in patients with type 2 diabetes – and the four therapies that have been shown to offer cardioprotection, really driving cardiology and diabetologists together. Though Dr. DeFronzo has been critical of over-reliance on metformin and many (including himself) have cast doubt on the robustness of evidence from UKPDS indicating a CV benefit with the therapy, it’s uniquely ingratiated in treatment algorithms and even Dr. DeFronzo isn’t ready to kick it out – though he had also previously stipulated that it should always be combined with a more efficacious drug.


-- Ann Carracher, Martin Kurian, Brian Levine, Peter Rentzepis, Maeve Serino, Adam Brown, and Kelly Close