AACE 2018 (American Association of Clinical Endocrinologists)

May 16-20, 2018; Boston, MA; Day #3 Highlights – Draft

Executive Highlights

  • Diabetes therapy: CVOTs took center stage on day #3 of AACE, and Dr. Mikhail Kosiborod walked through CVOT evidence to (i) show cardioprotection as a class effect of GLP-1 agonists; and to (ii) suggest that GLP-1s will confer benefit in primary CV prevention (even though this has yet to be tested empirically). These are both current controversies (or at least areas of interst) in the diabetes field, and Dr. Kosiborod laid out very compelling arguments. He also joined a panel alongside Drs. Ralph DeFronzo and Vivian Fonseca discussing the ethics of placebo-controlled CVOTs. We were struck by one audience member comment: “I’d advise my mother against enrolling in SOUL for fear that she’d end up in the placebo group” in favor of simply prescribing Ozempic (Novo Nordisk’s once-weekly GLP-1 semaglutide). The evening prior offered two outstanding symposia: Novo Nordisk’s featured Drs. Guillermo Umpierrez, Bruce Bode, and Irl Hirsh on postprandial glucose control and the role of ultra-rapid-acting insulins plus basal/GLP-1 fixed-ratio combinations. Sanofi/Lexicon sponsored the other, and Drs. Paresh Dandona (a DEPICT investigator) and Satish Garg (an inTandem investigator) covered DKA risk mitigation for type 1s on an SGLT inhibitor – this was an absolutely fascinating conversation full of learning for us on ketone monitoring, dosing of these oral adjuncts (sotagliflozin and dapagliflozin), and possible cardioprotection in a type 1 population.

  • Diabetes technology: Outcomes beyond A1c was the theme du jour from tech KOLs. Dr. Irl Hirsch alluded to the glycemic targets he sets for patients, noting that Medtronic’s 670G pivotal trial is a good place to start – at a high level, he aims for mean glucose <140 mg/dl, <4% of time <70 mg/dl, CV ≤33%, and <40% basal dose. This topic is one of the big next steps in outcomes beyond A1c – now that glycemic bins have been defined, how much time should people optimally spend in them? Dr. Rich Bergenstal showed 670G outcomes from everyone in the US uploading to CareLink (n=~41,000), which were fairly similar to his cohort at Park Nicollet (n=138). He also gave a litany of 670G commentary and showed why A1c is not an adequate basis for making therapeutic decisions. We also heard pro CGM and beyond A1c thoughts from Drs. George Grunberger, Steve Edelman, and Etie Moghissi. On the outcomes side, we saw a poster with the first Valeritas V-Go study (we can remember) to report time-in-range improvements. 

AACE continued strong on day #3, which was the busiest for our team in terms of content on diabetes/obesity. See below for our top 13 highlights, including six on diabetes therapy and seven on diabetes technology.

We’ll have one more installment of AACE conference highlights for you, but in the meantime, enjoy these learnings from day #3, and see our previous reports from day #1 and day #2.

Table of Contents 

Diabetes Therapy Highlights

1. Dr. Kosiborod Suggests GLP-1 Agonists May Be Cardioprotective in Primary Prevention; Shares High Hopes for Positive REWIND Results; Reevaluates LEADER in Participants with vs. without Prior CV Events

Dr. Mikhail Kosiborod argued that “it doesn’t make sense physiologically” for GLP-1 agonists to give secondary CV prevention without also offering CV benefit to a primary prevention population. He headlined an early morning Novo Nordisk symposium, and referred to Victoza’s (liraglutide) CV indication, which only applies to type 2s with existing CV disease but not to those with risk factors for CV disease (the same goes for Lilly/BI’s Jardiance on the SGLT-2 front). Diabetes CVOTs like LEADER (for liraglutide) and EMPA-REG OUTCOME (for empagliflozin) have enrolled an enriched study population, i.e. patients with established CV disease, to drive event rate and show an agent’s CV effects in a shorter time frame (and arguably, a more reasonable time frame, given that CVOTs are an enormous investment for manufacturers). But Dr. Kosiborod noted that statins show similar relative risk reduction for CV events in both primary and secondary prevention populations (though of course absolute risk reduction is greater in secondary prevention, and number needed to treat is lower). He expects the same may be the case for GLP-1 agonists, particularly because their hypothesized mechanism of CV benefit involves decelerating the progression of atherosclerosis – Dr. Kosiborod emphasized that anti-atherosclerotic effects occur gradually, and would be beneficial at any stage of CV risk. It’s hard to imagine (or even to begin to imagine) FDA extrapolating CVOT results to lower-risk patients who weren’t extensively represented in the study population, but Lilly’s REWIND trial for Trulicity (dulaglutide) could provide empirical evidence to support a wider-reaching CV indication. Of all REWIND participants, 69% comprise the primary prevention cohort (vs. 19% in LEADER, 17% in SUSTAIN 6 for Novo Nordisk’s Ozempic, and 27% in EXSCEL for AZ’s Bydureon). Lilly has compensated for the lower event rate with a longer duration of follow-up (median 6.5 years vs. 3.5 years in LEADER, ~2 years in SUSTAIN 6, and 3.2 years in EXSCEL). Topline data is expected in 4Q18 and full results are expected at ADA 2019. Dr. Kosiborod shared his hope that REWIND results will be positive, indicating CV efficacy for dulaglutide, which evoked commentary from Dr. Ralph DeFronzo on AACE day #1. While LEADER subgroup analyses found a significant benefit on three-point MACE in the secondary prevention cohort (defined as >50 years of age with established CV disease; HR=0.83, 95% CI: 0.74-0.93) but not the primary prevention cohort (defined as >60 years of age with CV risk factors; HR=1.2, 95% CI: 0.66-1.67), Dr. Kosiborod pointed out that this discrepancy is no longer seen when the comparison is simply “history of CV events“ vs. “no history of CV events.” It was incredibly encouraging to hear Dr. Kosiborod’s endorsement of a possible primary prevention benefit to GLP-1s, because we’re eager to see these advanced agents prescribed to more patients, earlier in the course of disease. Indeed, some thought leaders have argued that all patients with type 2 diabetes should be considered high-risk for CV morbidity/mortality, and prescribing cardioprotective drugs in primary prevention could pack a bigger epidemiological punch by preventing or delaying the emergence of CV disease altogether.

  • SGLT-2 inhibitors were not the main focus of Dr. Kosiborod’s presentation, but we have high hopes that this class will also demonstrate compelling efficacy in primary CV prevention. The DECLARE CVOT for AZ’s Farxiga (dapagliflozin), expected to complete in July 2018 and to report in 2H18, features a primary prevention cohort reflecting 60% of the total study population (vs. >99% in EMPA-REG and 34% in CANVAS for J&J’s Invokana). AZ’s CVD-REAL suggests a cardioprotective class effect for SGLT-2s, and notably, featured a primary prevention cohort reflecting 75%-83% of the whole participant pool. We’d imagine that Dr. Kosiborod’s physiological reasoning that agents effective for secondary CV prevention should also support primary CV prevention would also apply to this class. Moreover, several KOLs have suggested that the main CV benefit of SGLT-2 inhibitors stems from heart failure risk reduction, and a much larger percentage of the type 2 diabetes population has heart failure (or at least subclinical heart failure/diastolic dysfunction) vs. a history of atherosclerotic events like stroke/MI.

2. Drs. Dandona and Garg Debate Optimal Ketone Monitoring for Type 1s on SGLT Inhibitor (How Often?); Recommend Lower Doses to Minimize DKA Risk; Speculate on Reasons for Higher DKA Risk in Pump Users

In a Sanofi/Lexicon-sponsored symposium on SGLT inhibition in type 1 diabetes, Drs. Paresh Dandona and Satish Garg offered their thoughts on ketone testing to mitigate DKA risk. Notably, they disagreed on the role of prophylactic monitoring but reached consensus on the importance of testing during high-risk events. Dr. Dandona shared that he asks patients with type 1 diabetes who start using an SGLT-2 inhibitor (currently off-label) to test their ketones once a week, and he asserted that all safety protocols and supplies used in clinical trials should be made available to real-world patients. He also noted that every type 1 who goes on an SGLT-2 inhibitor will see an increase in beta hydroxybutyrate: The key is to not let that ketogenesis progress to DKA. In Dr. Dandona’s view, SGLT inhibition puts patients in a ketogenic situation, and while progression to DKA is rare, individuals need to monitor both glucose and ketones diligently (and regularly). We’re inclined to align with this perspective, because knowing the immense quality of life impact of experiencing DKA, not to mention the financial burden when a patient is hospitalized, we’d want a broad monitoring strategy that catches all DKA episodes in their tracks. That said, Dr. Garg offered an interesting counterargument. He outlined his view that daily monitoring – or even regular checking – should not be mandated, but instead, patients should know to check ketones in specific situations, such as when they’re vomiting or dehydrated due to unrelated illness. In Dr. Garg’s estimation, risk of DKA is definitely higher on SGLT therapy, but it’s still small overall. He explained that it may be more valuable to identify the high-risk scenarios, such as illness and surgery, and he contended, “you can tell from the studies that there is no significant risk of DKA if you closely follow the patients, especially in high-risk situations.” We see how demanding daily or weekly monitoring of ketones, when the vast majority of measurements will be normal, may place undue burden (mentally, financially, and physically) on patients and could serve as an unnecessary barrier to accessing these agents. We can imagine a protocol wherein patients monitor ketones more regularly in their first several months on an SGLT inhibitor, but then wean off after consulting with their healthcare provider, so long as they demonstrate knowledge of what constitutes a high-risk event when they need to resume ketone testing. Much remains to be determined relative to safety protocols with both SGLT-2 and SGLT-1/2 dual inhibitors in type 1 diabetes, and it’ll be interesting to see if some degree of personalization can be incorporated, although we’d definitely stand by a “safety first” approach even for those who prefer to check ketones infrequently. Above all, we continue to see thorough patient education as the cornerstone of minimizing DKA risk with SGLT inhibitors in the real world.

  • Drs. Dandona and Garg wholeheartedly agreed that ketone testing in high-risk situations is critical. These include things as commonplace as missed meals, skipped insulin doses, or significant alcohol consumption, which can lead to euglycemic ketoacidosis. In our view, extensive and thorough patient and provider education is absolutely necessary. Patients must demonstrate that they understand when they’re at risk for DKA, and providers need to understand risk factors and treatment protocol; as Prof. Thomas Danne told us shortly after Sanofi/Lexicon’s sotagliflozin was submitted to FDA and EMA, treating DKA properly requires both glucose and insulin.

  • Dr. Garg called attention to the differences in study design between inTandem (for Sanofi/Lexicon’s sotagliflozin) and DEPICT (for AZ’s dapagliflozin) which preclude comparison of DKA event rates. He shared similar commentary in an NEJM response co-authored with Lexicon’s Dr. Paul Strumph. When “certain, definite, probable, or possible DKA” are considered from both trials, Dr. Garg showed how sotagliflozin 400 mg was associated with a 3% rate of DKA in inTandem 3 vs. 0.6% with placebo, and 10 mg dapagliflozin was associated with a 4% rate of DKA in DEPICT 1 vs. 2% with placebo. Ultimately, we very much think that comparing these two trials is irrelevant: They were designed and enrolled differently, and the field has reached a consensus that DKA risk is small but very real with either agent. Dr. Garg framed DKA incidence as relatively low in both clinical programs compared to a background DKA rate of around 5%-6% in insulin-treated patients with diabetes. To be sure, this could have more to do with the context of an RCT, in which patients almost always receive more attention and clinical care than usual. This feeds back to Dr. Dandona’s argument that support and education around DKA from inTandem and DEPICT needs to be translated for the real world.

    • To our knowledge, no one has measured DKA rates among people with type 1 diabetes taking SGLT-2 inhibitors off-label, at least not to any reasonable degree of accuracy. We expect DKA rates with these agents to be higher outside of clinical trials, and we’d be very curious to see this data. We certainly expect it to be collected if/when sotagliflozin and dapagliflozin are FDA-approved for use in type 1 diabetes.

  • Dr. Dandona advocated for the use of low doses of adjunct oral therapies in type 1 diabetes. He described how the differences in glycemic efficacy in DEPICT 1 and inTandem did not vary meaningfully by dose, whereas there was a clear dose-response relationship with DKA. As such, smaller doses would help minimize chances of developing harmful degrees of ketogenesis and eventually DKA, without much efficacy trade-off. In his practice, Dr. Dandona starts patients on lower doses of an off-label SGLT-2, has them monitor ketones, and then considers proceeding to a higher dose.

  • The panelists touched on the mechanistic underpinnings of the perceived increased risk for DKA in pump users. Dr. Garg explained that very few patients change their pump catheter every 2-3 days as instructed, and kinked catheters are common. The continuous infusion of a low dose of insulin given by a pump, Dr. Dandona elaborated, is inherently different from the twice-daily depot of a large basal insulin dose that patients get with MDI. With SGLT-2 inhibitors, the advantage of stability from CSII becomes a disadvantage: If the system malfunctions, the patient has no real reservoir of insulin left in the body, making him more likely to fall into DKA. Dr. Garg suggested that it would be interesting to see if next-gen basal insulins Tresiba (Novo Nordisk’s degludec) and Toujeo (Sanofi’s glargine U300), which are longer-acting and more stable, could reduce DKA in patients on MDI. Some thought leaders have cast doubt on whether pump vs. MDI is a relevant consideration, but Dr. Garg did note that a high proportion of patients enrolled in inTandem 3 (~40%) were on a pump, comparing this to 25%-30% of the patients he sees in his practice.

  • Speaking to a potential CV benefit of SGLT inhibitors in type 1 diabetes, Dr. Garg expressed doubt that a trial establishing this efficacy could ever be conducted: “It would have to be 20 years long.” Dr. Dandona referenced benefits on body weight and blood pressure, but noted that cardioprotection is still a relative mystery. Another panelist, Dr. Robert Henry, asserted that improvements in weight, systolic blood pressure, and LDL cholesterol might contribute to improved CV and renal outcomes and that both dapagliflozin and sotagliflozin maintain these benefits in type 1 diabetes. We would be incredibly keen on a CVOT of these agents in type 1 diabetes (which confers significant residual CV risk on its own), though we recognize the challenges of conducting this trial, as Dr. Garg alluded to. To this end, we wonder if real-world evidence could eventually shed light on this question.

3. Ahead of SOUL for Novo Nordisk’s Ozempic, Opinion Leaders Debate Ethics of Placebo-Controlled CVOTs; Very Interesting Arguments from Drs. Fonseca, DeFronzo, Kosiborod, & Audience

Are placebo-controlled CVOTs still ethical, given the availability of agents with known CV benefit? Drs. Ralph DeFronzo, Vivian Fonseca, and Mikhail Kosiborod considered this question during an engaging panel discussion. The SOUL CVOT for Novo Nordisk’s Ozempic (semaglutide) is scheduled to begin mid-year. Notably, we’ve already seen positive CVOT data for semaglutide from SUSTAIN 6, but this pre-market CVOT wasn’t powered to show superiority, and it was ultimately too small and too short to support a CV indication on the Ozempic product label. The larger and longer post-market SOUL trial will have sufficient power to support a CV efficacy claim, but Dr. Fonseca characterized it as a questionable use of resources since the very compelling SUSTAIN 6 data all but guarantee that SOUL will also be positive. During Q&A, one audience member remarked, “I’d advise my mother against enrolling in SOUL for fear that she’d end up in the placebo group” in favor of simply prescribing Ozempic (which launched to US pharmacies in February 2018). Dr. DeFronzo agreed, suggesting that it’s a “waste” to perform another study given the vanishingly small probability of the SUSTAIN 6 results not being replicated, especially since liraglutide (“which is only two amino acids different!”) is already indicated for cardioprotection. Dr. DeFronzo also pointed out that the resources required to complete the SOUL CVOT, combined with a potential new CV indication, could drive up the price of the drug.

  • On the other hand, Dr. Kosiborod pointed out that while most acknowledge that SOUL is likely to be positive, it would show favoritism and set a potentially dubious precedent if FDA gave semaglutide a CV indication on the basis of the insufficiently-powered SUSTAIN 6 trial, which alone was “absolutely suggestive” but not definitive due to a small number of events. Dr. Kosiborod emphasized that “liraglutide data isn’t semaglutide data,” maintaining that it’s presumptive to ascribe cardioprotection to semaglutide before longer-term results are in. Moreover, we think SOUL will contribute other important knowledge to the field besides semaglutide’s CV effects: This larger, longer study could shed light on retinopathy risk with semaglutide, and could help determine whether the signal seen in SUSTAIN 6 was in fact due to early worsening phenomenon. However, Novo Nordisk will simultaneously be conducting a dedicated long-term study to assess retinopathy effects with semaglutide (part of the conditions for EMA approval), which brings us back to the potential “waste” of resources. There’s certainly a lot to unpack here, and we hope this conversation continues. 

  • As for whether or not it’s ethical for half of SOUL participants to be given placebo instead of a highly efficacious GLP-1 agonist, Dr. Kosiborod pointed out that sadly, the vast majority of people with diabetes aren’t on a GLP-1 to begin with and the placebo condition in SOUL will likely represent what’s standard of care (or better) for most places in the world. Nevertheless, we imagine this issue will become more important as access to advanced therapy classes improves over time, and we’re glad thought leaders are already talking about it. Dr. Kosiborod alluded to key challenges ahead, pointing out that active comparator trials are “much more complicated” to design, conduct, and analyze – and, we’d add, to standardize. He also claimed that active comparator studies can be much more expensive to conduct. While this may put strain on industry sponsors, we note that it also represents a long-awaited opportunity to compare CV outcomes with different diabetes agents head-to-head. Earlier at AACE, Dr. Benjamin Scirica characterized this as a hallmark of the next wave of CVOTs.

4. Drs. Umpierrez, Bode, Hirsch Talk Postprandial Glucose Control; Highlight Basal/GLP-1 Fixed-Ratio Combinations, Afrezza, & Fiasp; Future for Fiasp in Pumps in US?

Dr. Guillermo Umpierrez shared his own diabetes treatment algorithm, with an eye toward postprandial control, during a Novo Nordisk-sponsored dinner symposium. He alluded to Dr. Matthew Riddle’s 2017 editorial in Diabetes Care, which highlighted an unmet need for pharmacotherapies that address postprandial glucose excursions. To further this point, he discussed another study published in Diabetes Care showing that the most common type 2 diabetes drugs prescribed in the US are metformin (77% of first-line scripts), sulfonylureas (8% of first-line scripts and 46% of second-line scripts), and basal insulin (10% of first-line scripts and 17% of second-line scripts). Dr. Umpierrez emphasized that none of these agents – neither metformin, nor SUs, nor basal insulin – address postprandial hyperglycemia. Moreover, he established that postprandial hyperglycemia has more influence on A1c than does fasting hyperglycemia, especially when patients near target A1c around 7%. We’ve heard similar arguments promoting faster mealtime insulins elsewhere on the conference circuit, e.g. from Mr. Jerry Meece during a MannKind-sponsored product theater at AADE; indeed, Dr. Bruce Bode touted Afrezza’s very fast onset/offset later during this AACE symposium. After lifestyle changes and metformin, plus well-titrated basal insulin, Dr. Umpierrez explained that his ideal treatment algorithm would recommend a GLP-1 agonist, a basal insulin/GLP-1 fixed-ratio combination product, a DPP-4 inhibitor, or an SGLT-2 inhibitor, because all of these therapy classes act on post-meal glucose without excess risk of hypoglycemia. To be sure, we’ve noticed increasing competition for the rapid-acting insulin class from GLP-1s and SGLT-2s precisely for this reason, that these more advanced drugs address postprandial hyperglycemia without conferring additional hypoglycemia risk. Dr. Umpierrez suggested that Sanofi’s Adlyxin is the best GLP-1 option for patients with poor post-meal glucose control because lixisenatide is very short-acting. He elaborated on the benefits of basal/GLP-1 combos (“kind of another form of basal-bolus, but simpler because it’s one formulation”), and specifically called out Sanofi’s Soliqua (insulin glargine/lixisenatide) due to the short-acting lixisenatide component. Real-world uptake of basal/GLP-1 combos, also including Novo Nordisk’s Xultophy (insulin degludec/liraglutide) has been extremely (upsettingly) low; there are a number of reasons for this, including poor reimbursement and provider discomfort with the concept of fixed-ratio injection, but Dr. Umpierrez focused on labeling. He called it a “mistake” that FDA stipulated that patients have to fail on basal insulin or on lixisenatide/liraglutide before starting on the combination therapy. In his view, patients with high starting A1c should be considered for early combination treatment, so that hyperglycemia doesn’t persist as it does under the treat-to-fail paradigm.

  • Dr. Bruce Bode reviewed ultra-rapid-acting insulin options, including MannKind’s inhalable Afrezza, Novo Nordisk’s next-gen Fiasp (faster-acting aspart), and from the pipeline, Lilly’s ultra-fast lispro (in phase 3) and Adocia’s BioChaperone Lispro (phase 3-ready). During Q&A, he explained that Onset 5 evaluating Fiasp in pumps is not on the product label (presented at ATTD 2018). According to Dr. Bode, FDA initially said Fiasp could be approved for pump use if Novo Nordisk conducted similar studies to those carried out for NovoLog (insulin aspart), but later changed its mind and requested more data. This led to Onset 5 – “hopefully Novo Nordisk will submit that, or should submit it to get approval for Fiasp in pumps,” Dr. Bode stated. Interestingly, Fiasp is approved for pumps in Europe, but not in the US or Canada.

  • Dr. Bode also described how Afrezza is the only product in this ultra-rapid-acting class that has shown significant hypoglycemia benefit in phase 3 trials. In our view, less frequent hypoglycemia (as well as improved time-in-range) could be the most valuable benefits of these more advanced mealtime insulin products. To this end, MannKind will present STAT results at ADA 2018 (the study compared time-in-range with Afrezza vs. NovoLog using CGM). While the emergence of SGLT-2s, GLP-1s, and basal/GLP-1 combos sometimes has us wondering whether growth prospects are limited for the rapid-acting insulin class – this was our takeaway from Dr. Umpierrez’s talk, and certainly the glycemic-dependent mechanism of these newer therapies gives them a distinct edge – we can’t deny that there are people with type 1 diabetes (and some type 2s) who will need to bolus at mealtime, and who need faster insulins. For these patients, reducing uncertainty and fear of hypoglycemia around meals is key. We see clear potential for Afrezza, Fiasp, and eventually Lilly’s ultra-rapid-acting lispro and Adocia’s BC Lispro to offer these advantages around meals, and we’re eager to see data supporting hypoglycemia and time-in-range benefits in the real world. Dr. Bode also noted the importance of ultra-rapid-acting insulin for closed loop systems.

  • Dr. Irl Hirsch emphasized the importance of glycemic outcomes beyond A1c. In order to get all stakeholders to accept these alternate endpoints in diabetes, he acknowledged the need for data linking glycemic variability to hard outcomes. After all, DCCT results demonstrating how more aggressive A1c reductions lead to lower incidence of microvascular complications (specifically retinopathy) were fundamental in establishing A1c as a core metric. Dr. Hirsch presented one interesting study from Australia (n=1,604 adolescents with type 1, published in Diabetes Care). Between 1990-1994 and 2005-2009, use of MDII/CSII in this cohort climbed from ~18% to nearly 90%. Mean A1c was more or less constant through this time span, but retinopathy declined substantially, from >50% in 1990-1994 to ~14% in 2005-2009. Although we’re limited in what we can conclude from this one study, Dr. Hirsch commented on the “definite possibility” that introduction of mealtime insulin/smaller postprandial spikes reduced retinopathy risk. He also discussed early evidence linking severe hypoglycemia to macrovascular complications, specifically CV events. For example, a post-hoc analysis of DEVOTE in type 2s found that participants faced significantly elevated risk for CV death and all-cause death following an episode of severe hypoglycemia. Dr. Hirsch outlined pathophysiological mechanisms by which both hyper and hypoglycemia contribute to oxidative stress, inflammation, and heightened susceptibility to arrhythmias – in other words, the greater the glucose fluctuations, the greater the chance of experiencing diabetes complications. Prof. Simon Heller has spoken about this at a number of diabetes meetings; for more, see our CODHy 2018 report. We’d of course love to see a large outcomes study that definitively connects glycemic variability to adverse outcomes, as this could really move the diabetes field to accept time-in-range and other glycemic endpoints beyond A1c. But in the meantime, we believe there’s already sufficient evidence about how time-in-range impacts patient quality of life, and we’re desperate for regulators, payers, manufacturers, HCPs, and patients alike to put more stock in glycemic outcomes beyond A1c.

Dr. Mikhail Kosiborod outlined the evidence in favor of CV class effects for both GLP-1 agonists and SGLT-2 inhibitors. Key to his argument is the definition of “class effect.” Dr. Kosiborod was careful to clarify that class effects don’t imply that every GLP-1 agonist or every SGLT-2 inhibitor is exactly the same, but rather, that outcomes with members of the same class trend in the same direction. He suggested that this is exactly what we’ve seen for both of these advanced diabetes drug classes, a viewpoint shared by many opinion leaders (with some notable exceptions, for instance, Dr. Sanjay Kaul at CMHC West).

  • On the GLP-1 front, Dr. Kosiborod emphasized differences in CVOT design to explain the heterogeneity of results. Only LEADER (for Novo Nordisk’s Victoza) and SUSTAIN 6 (for Novo Nordisk’s Ozempic) found statistically significant risk reduction for three-point MACE (non-fatal MI, non-fatal stroke, or CV death) with the GLP-1 molecule (liraglutide or semaglutide) vs. placebo. Nonetheless, Dr. Kosiborod pointed out that EXSCEL (for AZ’s Bydureon) may in some ways be interpreted as supporting a GLP-1 class-wide benefit, arguing that it does more to support a cardioprotective class effect than it does to refute it. For ELIXA (Sanofi's Adlyxin), he stated that the trial had “exactly the design you would use if you were trying to demonstrate neutrality and nothing else.” Indeed, ELIXA featured a very high-risk patient population (recent acute coronary syndrome) and a very short follow-up in order to accelerate the trial’s completion (during an era when neutral CVOTs were the norm). Without a longer-term trial, Dr. Kosiborod believes that lixisenatide’s CV effects remain an open question (with the possibility for CV benefit). As for EXSCEL, exenatide just barely missed the threshold for significant CV superiority (HR=0.91, 95% CI: 0.83-1.00, p<0.001 for non-inferiority, p=0.06 for superiority). Dr. Kosiborod ascribed statistical disadvantages to the trial’s pragmatic design and low adherence. He called attention to a meta-analysis presented by EXSCEL investigator Dr. Angelyn Bethel at IDF, indicating that GLP-1 agonists as a class have given a 10% relative risk reduction (HR=0.90, 95% CI: 0.82-0.99, p=0.033) for three-point MACE and a 12% relative risk reduction for all-cause mortality (HR=0.88, 95% CI: 0.81-0.95, p=0.002) across the four reported CVOTs (LEADERSUSTAIN 6ELIXA, and EXSCEL). This meta-analysis, in particular, supports a cardioprotective class effect for GLP-1s in line with Dr. Kosiborod’s definition of outcomes trending in a similar direction. This point was also brought up in the discussion of the EXSCEL paper, published in NEJM.

    • Notably, while highlighting variation in trial design, Dr. Kosiborod acknowledged the hypothesis that molecular differences might explain diverging CVOT results for GLP-1s – but he does not believe this is likely. Some have argued that human GLP-1-based molecules (liraglutide, semaglutide, dulaglutide) are cardioprotective, while exendin-4-based agents (exenatide, lixisenatide) are not; we most recently heard this perspective from Exeter’s Dr. David Strain at CODHy Asia Pacific. Increasingly, it seems like thought leaders are rejecting the notion of molecular distinction contributing to major differences in CV efficacy, but certainly this hypothesis still has some fierce supporters. In general, the debate over cardioprotection as a GLP-1 class effect continues strong, far from being settled. In our view, as well as Dr. Kosiborod’s, EXSCEL does more to support a GLP-1 CV class effect than it does to refute it, but we fully expect this discussion to continue moving forward, and our understanding may evolve.

  • Cardioprotection as a class effect of SGLT-2 inhibitors is less controversial, following positive results in all relevant CVOTs to-date (CANVAS and EMPA-REG OUTCOME). Dr. Kosiborod referred to DECLARE for AZ’s Farxiga (dapagliflozin) as a “coming attraction” to “settle the record” on CV class effects for SGLT-2s. With 60% of DECLARE participants starting the trial without pre-existing CV disease, this study also promises valuable insight on primary vs. secondary CV prevention with dapagliflozin. DECLARE additionally incorporates heart failure into a co-primary endpoint, which Dr. Kosiborod argued should be the case for every agent in the class. The significant risk reduction for heart failure hospitalization in EMPA-REG and CANVAS has drawn attention to this under-recognized and under-treated condition, despite it being one of the deadliest complications of diabetes. Dr. Kosiborod expressed enthusiasm about ongoing trials of SGLT-2 inhibitors in heart failure with or without diabetes (the EMPEROR program for Lilly/BI’s Jardiance and Dapa-HF for AZ’s Farxiga), to further illuminate the range of benefit for these agents. The real-world CVD-REAL study (for which Dr. Kosiborod was lead investigator) has provided an intriguing suggestion of lower risk for heart failure and all-cause mortality with SGLT-2s across a broad spectrum of risk.

6. Dr. Einhorn’s Warning: We Risk Confusing HCPs by Over-Emphasizing Within-Class Differences for GLP-1s; Urges Providers to Prescribe Whichever is Reimbursed for Each Patient

Moderating a Novo Nordisk symposium on GLP-1 agonists, Dr. Daniel Einhorn cautioned against becoming overly preoccupied with intra-class differences. He attributed different clinical trial results to potential variations in study design, while also noting that it is possible that there are meaningful molecular differences – this is an argument we’ve heard often in the context of EXSCEL (the CVOT for AZ’s Bydureon) and a cardioprotective class effect for GLP-1s. Dr. Einhorn added that in over-emphasizing within-class differences, we risk sending a confusing message to HCPs that only the GLP-1 agonists with positive CVOT data are worth prescribing. All things being equal, he would prescribe the GLP-1 that has the FDA indication for the right patients with CV disease or high CV risks – i.e. liraglutide. He underscored, however, that only certain GLP-1 products may be “in formulary” for certain patients, and prescribing exenatide or dulaglutide as opposed to liraglutide (to-date, the only GLP-1 agonist with a CV indication on its product label) is still a very strategic choice for a patient’s CV health. This echoed commentary from Dr. Ralph DeFronzo on AACE day #1: “All of these drugs are fantastic, and all you need to do is make sure your patient is on one of them,” he stated, emphasizing that any GLP-1 is better than no GLP-1. Dr. DeFronzo has even authored a comment piece on SUSTAIN 7 (head-to-head of Ozempic vs. Trulicity) detailing similar views. We ultimately agree with these opinions and we were pleased to hear this commentary come up several times at AACE. While we see merits to head-to-head research, it’s important to send a clear message throughout the diabetes community that different products may be the best choice in different clinical circumstances, and that “best” choice also depends on patient preferences. It was intriguing to see Ozempic’s superior glycemic and weight loss efficacy vs. Trulicity in SUSTAIN 7, but (i) this was only one study (sponsored by Novo Nordisk at that) and (ii) Trulicity could be the favorable option depending on clinical circumstance (as just one example, some patients may prefer the easy-to-use dulaglutide autoinjector over the semaglutide pen). We also appreciated the acknowledgement of how access affects therapy choice for so many people with diabetes.

Diabetes Technology Highlights

1. Dr. Irl Hirsch’s Patient Targets: Mean Glucose <140 mg/dl, <4% of Time <70 mg/dl, CV ≤33%, Basal <40% of Total Insulin; Providers with <15 Min/Patient Shouldn’t Prescribe Hybrid Closed Loop

UW’s Dr. Irl Hirsch noted a particular need to define targets for glucose ranges, recommending the Medtronic 670G pivotal trial data as “a good place to start.” His own personal goals include helping his patients achieve a mean glucose <140 mg/dl (it was ~150 mg/dl in the 670G pivotal), time <70 mg/dl of <4% (it was ~3% in the 670G pivotal), and a CV (coefficient of variation) of <33%. Most importantly, he works to decrease basal insulin, which he finds can substantially help reduce nocturnal hypoglycemia. He noted that often <40% of total insulin should be delivered as a basal dose and that many patients require a more aggressive bolus. In fact, he described fake carbs as “really important” on the 670G (i.e., shifting more insulin to bolus), a shift from the reservations he shared in March regarding the practice of falsely adding carbs into the bolus calculator to artificially make the 670G more aggressive. We have heard from other KOLs that the 670G pivotal may be a good place to start, though it was interesting to get a glimpse at Dr. Hirsch’s personal goals. Consensus goals likely won’t be reached until the field analyzes time in range profiles of patients considered to manage their diabetes with a variety of tools – e.g., should time-in-range of 70% be set as a goal for someone on MDI? He maintained that he is not interested in getting rid of A1c as a metric entirely; rather, he called for “everybody’s guidelines to focus on time-in-range, time-above-range, and time-below-range” (obviously, for these to gain traction, more people need to be on CGM).

  • Dr. Hirsch advised providers with limited time per patient to avoid prescribing hybrid closed loop systems, as their use requires especially rigorous data analysis. He noted that the “most benefit” from CGM data is obtained from looking at a patient’s daily modal day profiles, although acknowledged that for those who only have 15 minutes to spend with their patients, such analysis is virtually impossible. As we’ve come to expect, nearly two-thirds of the audience indicated that they had 15 minutes or less for each appointment. Many attendees also indicated that they have several patients who “mix technology” – in other words, a patient might use a Dexcom CGM with a Medtronic pump. Dr. Hirsch admitted that this practice is unfortunately “really common” and argued that “ideally everyone would have the pump and CGM integrated,” as being able to see CGM and pump data side-by-side is critical for appropriate dosing. He mentioned that such broad integration is “not the case, at least not yet,” potentially alluding to the integrated CGM (iCGM) regulatory pathway recently created by the FDA. (Tidepool can also combine Medtronic pump and Dexcom CGM data.) Tandem’s t:slim X2 is the only Dexcom-integrated pump right now (G5), and G6 integration is expected by the end of the year (per Tandem 1Q18). Following substantial interest exhibited at the JDRF/HCT workshop on interoperability, it’s possible that a parallel integrated pump class may also become a future reality.

  • During a case study, audience members were surprised to learn that mitral valve prolapse can speed up red blood cell turnover, artificially suppressing A1c (the patient’s eA1c was much higher than lab A1c!). This largely unknown effect is just one of the many examples of the limitations of A1c, and elicited gasps from the packed room. Dr. Hirsch exclaimed: “I think it’s really important!”

  • Dr. Hirsch described his “biggest gap” as being unable to view insulin dosing history for his MDI patients. He specifically noted Lilly’s efforts to develop smart pens – at the Lilly Diabetes Summit, we received confirmation that a connected smart pen leveraging the disposable KwikPen is under development (it was not shown) – it is unclear what form factor this will take (e.g., sleeve, cap, dial, or integrated into the disposable pen). Lilly has also invested in Companion Medical’s InPen, a reusable smart pen, which initiated a limited US launch in December. Companion’s InPen is just the first US domino in a category that we expect to explode in the very near future. Novo Nordisk has an NFC-enabled pen in pilots, Common Sensing is in a beta launch and piloting with Sanofi, and Bigfoot (Timesulin) has its own dose capture cap slated to launch in 2020 with auto-titration and next-gen FreeStyle Libre. We agree with Dr. Hirsch that dose capture for injectors is a massive need. The value of making this invisible data visible was best captured in a Joslin/Common Sensing poster at ADA, where the connected Gocap identified missed/under-bolusing, extra/over-bolusing, and basal doses taken outside the scheduled time window. On the subject of missing data, Dr. Hirsch also noted that if an individual goes more than eight hours without scanning FreeStyle Libre, then the data goes unrecorded – while he didn’t explicitly note frustration, such gaps can pose a challenge, especially in the late overnight/early morning period. We assume Abbott’s planned next-gen improvements, especially continuous communication, will address this.

  • Dr. Hirsch underscored that despite AGP standardization, differences between system screens still exist. He repeatedly emphasized that the Libre AGP trace depicts the 10th/90th percentiles, while the Dexcom AGP shows 15th/75th percentiles. Dexcom Clarity is actually a bit more nuanced than Dr. Hirsch implied: the Clarity “Data” tab shows the 15th/75th percentiles in the modal day, though the official “AGP” report page in Clarity shows both the 15th/75th and 10th/90th percentiles. Still, his point stands: Knowing the degree to which glucose readings are true patterns vs. outliers is very key in safe titration. He also noted that the Libre AGP no longer shows the estimated A1c, which he found a useful comparator to measured A1c (of course, this will theoretically return to the display shortly as “GMI” – see Dr. Bergenstal’s talk below).

2. Park Nicollet 670G and National 670G Cohort Outcomes Very Similar; 100,000 670Gs Shipped, 70,000 Patients Trained; Dr. Bergenstal 670G Commentary – Fake Carbs, Dawn Phenomenon, Special Circumstances

IDC’s Dr. Rich Bergenstal displayed real-world 670G outcomes from all patients uploading data to CareLink so far (n=40,824) and from his own clinic (n=138), concluding: “It’s holding up across the country.” Fascinatingly, though patients at Park Nicollet have access to Dr. Bergenstal’s team of 670G experts (who helped conduct the three-month US pivotal study), their time spent in auto mode and outcomes therein are not noticeably different from the national cohort. In the overall, ~41,000 strong group of patients who have uploaded to CareLink, people spend ~76% of time in auto mode, and while in auto mode, time “71-180” mg/dl is ~72%, time ≤70 mg/dl is ~2%, and time >180 mg/dl is ~26%. In the Park Nicollet group, people spend 78% of their time in auto mode, and while in auto mode, time 71-180 mg/dl is ~75%, time ≤70 mg/dl is ~3%, and time >180 mg/dl is ~23%. (Side note: We’re disappointed to see Medtronic continuing to forget to use the consensus definition for time-in-range, 70-180 mg/dl.) Time in hypoglycemia is more than halved in both groups from baseline – from 4% to 2% nationally and from 6% to 3% at Park Nicollet. The bigger discrepancy can be seen when users are in manual mode, where the national cohort spends ~62% of their time in range, while the Park Nicollet spends an impressive ~68% of their time in range. Given that the sample sizes and population composition are different, direct comparisons require caution, but the data are certainly instructive and suggest hybrid closed loop can help level the playing field, delivering improved outcomes for those who don’t necessarily have access to the top tech-savvy diabetes care teams in the world. Above all, as we noted in the exhibit hall where we first saw the national data, real world results continue to resemble that of the pivotal – with the clear room for improvement on time in auto mode, which is currently hovering around ~18 hours per day. We note that, while there are just under 41,000 users in this national data set, Medtronic has confirmed that over 100,000 MiniMed 670G systems have actually been shipped, and over 70,000 people have been trained on the system. The incredible side-by-side data was obtained through the new MiniMed Outcomes app, which continually updates and allows providers to compare the glycemic data of their patients on 670G with the national 670G average. We LOVE this benchmarking!

  • Dr. Bergenstal offered myriad 670G commentary – much of which we have heard before – through the course of the talk.

    • “Dawn phenomenon is a main indication in my mind – anyone on alternative basal rates on a pump should think about a hybrid closed loop system, which is basically anyone on a pump.” He referenced a poster from ATTD 2018 in which 66% (82/124) of patients in the pivotal study were found to have dawn phenomenon. Those with dawn phenomenon had improved time-in-range (70-180 mg/dl) during the hybrid closed loop phase compared to the run-in phase from 12-6 am (77% vs. 68% in-range), 12-3 am (73% vs. 67% in-range), and 3-6 am (82% vs. 69% in-range).

    • “670G is built around safety, and probably could be more aggressive in spots.” This is a diplomatic statement from Dr. Bergenstal! We’ll be interested to see how much uptake more aggressive systems see, particularly the Tandem/TypeZero/Dexcom system with automatic correction boluses (slated for 1H19 launch).

    • “Guess what: 70% of the time, inputting fake carbs was a mistake. You just go low later. Just let the system work.” Dr. Irl Hirsch shared the term “fake carbs” – falsely adding carbs into the bolus calculator to artificially make the 670G more aggressive – at ENDO Fellows, where he admitted to having reservations about the practice though he recognized that many patients are doing it anyway.

    • “I’m not sure 50:50 basal:bolus ratio is wrong with other forms of therapy, but it’s not the expectation with closed loop.” Dr. Bergenstal has long been a proponent of setting more aggressive insulin:carb ratios since the basal rate can decrease or shut off to compensate for higher prandial doses. We are fans of moving away from “best practices” for things like basal: bolus ratio, since they can differ so much by system and by diet – e.g., Adam’s basal:bolus ratio on Loop is 80% basal, 20% bolus.

    • “On 670G, there’s quite a bit of interaction with the system, but it is worth it. On a pump and sensor, you’re working hard and not necessarily seeing results. Now you’re working hard, and it’s paying off.”

    • “Pediatric endos say that the prospect of getting kicked out of auto mode is the best tool for adherence. Once you get in system, you want to stay in the system. You try to beat the system, like a game, you try to stay in auto mode. You start to figure out that if you take insulin 10 minutes earlier, you don’t get kicked out. Whatever works.”

      • This brings up an interesting point: how much of the 670G’s efficacy is from the core system automation vs. its conservative design, alarms, and user burden demands? We’ll be fascinated to compare these outcomes to other commercial systems that should have less conservative design.

    • “I’m a big fan of the 670G reports. I love the before and after. At a glance, I don’t even have to say anything to the patient. ‘Are you doing better? Where are you doing better?’

      • His one gripe with the reports was that the hypoglycemia ranges are 51-70 mg/dl and 40-50 mg/dl, counter to the consensus ranges (<70 mg/dl and <54 mg/dl). Either showing his cards or making a request, Dr. Bergenstal said “I’m sure they’re going to change to 54 mg/dl from 50 mg/dl soon for the lower range on CareLink.” We’d add the same for time-in-range, which should be 70-180 mg/dl!

  • He also reviewed special cases of 670G use: Exercise, illness, gastroparesis, (and Dr. Irl Hirsch chimed in on pregnancy).

    • “Exercise is tough. Be careful about taking a lot of carbs before – if you take carbs, auto basal will kick in. It might be better to take fewer carbs, let the system work, and take carbs along the way if needed.”

    • During illness, I’ve had most people stick with the system. Some people’s glucose drops, in which case we use the 150 mg/dl temp target. But most people’s glucose shoots up, and sometimes they need to drop into manual mode [to better manage that insulin-resistant state].”

    • “670G has not been studied in gastroparesis. Whatever worked before – if you split boluses, used dual waves, whatever – then do that in auto mode too. Basal rates really save you a lot when you guess wrong. I think it’ll be a big help in gastroparesis though. Avoid the quick bolus with 670G – most say the faster the better, but you may want to set that back for gastroparesis.”

    • “We’re all in agreement that 670G is too conservative – once we get word that family planning is starting and the patient is off contraception, we go straight into manual mode.” – Dr. Irl Hirsch

3. Dr. Bergenstal: A 6.7% A1c is NOT a 6.7%; AGP is 90% About Profiles, 10% About Metrics; “eA1c” -> “GMI” (Glucose Management Indicator)

Dr. Bergenstal encouraged attendees to print out Ambulatory Glucose Profiles from 30 type 1s and 30 type 2s in order to see the fallibility of A1c. He demonstrated the exercise with 12 randomly-pulled type 1 profiles from his own clinic. There were two people with A1cs of 6.1%, and while one of them appears to be doing “amazing,” another regularly spends ~2.5 hours under 40 mg/dl overnight! He zoomed in on three patients who all had A1cs of 6.7%:

  • 83% time-in-range (70-180 mg/dl), with 1% time <70 mg/dl and CV of 26%;

  • 69% time-in-range, with 6% time <70 mg/dl and CV of 42%; and

  • 51% time-in-range, with 9% time <70 mg/dl and CV of 53%).

  • His staff filled him in on the unsurprising kicker: The strongest profile was from a patient on 670G, the second strongest was on pump therapy, and the most challenged was on MDI (this doesn’t rule out correlation, but is an interesting data point). This was a brilliant demonstration of the fallibility of A1c for guiding therapeutic decisions.

  • When Dr. Bergenstal analyzes AGPs, ~90% is about the glucose profile and ~10% is about the metrics. This makes sense to us and seems like the best way to key in on areas for improvement – when are highs, lows, and variability occurring, and why might those patterns be happening? (Those cannot be unpacked from overall metrics alone, which are not separated by time of day.) We wonder how many providers would agree with this breakdown. We hope to see decision support software continue to improve such that Dr. Bergenstal’s wisdom is built right into AGP and informs likely courses of action.

  • One of the first questions from the audience was along the lines of, “what the heck happened to eA1c (estimated A1c) on the Abbott reports?” Dr. Bergenstal told the story of how there has been a lot of confusion over the discrepancies between eA1c and lab-measured A1c, so FDA asked companies to remove it until they come up with a different name. He and Dr. Roy Beck have proposed “Glucose Management Indicator” (“GMI”), which elicited laughter from the crowd. “You got a BMI, so why not a GMI?” FDA has reportedly mandated that the new name not include “estimated,” “A1c,” or “index,” which has led Dr. Bergenstal, Dr. Roy Beck, and others to settle on GMI. This was first announced at The diaTribe Foundation’s Beyond-A1c ATTD event, and we expect a forthcoming publication with more details.

4. Dr. Grunberger Knocks A1c as Go-To Diabetes Metric, Emphasizes Patient CGM Education; How Can Endos Stay Relevant in Data-Heavy Diabetes Landscape?

The very busy Dr. George Grunberger gave a high-level tech update for clinicians, placing emphasis on outcomes beyond A1c, the need for patient education, and the challenge facing endos to remain relevant as diabetes technology becomes more sophisticated. He skinned the A1c cat (so to speak) in many ways throughout the talk, pointing out that: (i) time in-range(s) is what patients actually experience, not A1c; (ii) A1c is not really that important for differentiating inhaled insulin from other products (rather, we believe, it should be reduced hypoglycemia, blunted postprandial hyperglycemia, less glycemic variability, etc.); (iii) anyone can bring down A1c, but technology is the only way to do it safely; and (iv) every sort of provider – plus payers and patients – need to be educated on how to read and interpret standardized glucose reports (glucose profiles and metrics). Dr. Grunberger is a key member of the Beyond A1c Writing Group, which published an e-Letter in Diabetes Care today (!) entitled, “Need for Regulatory Change to Incorporate Beyond A1C Glycemic Metrics.” This must-read takes learning from the July 2017 Outcomes Beyond A1c conference, co-organized by The diaTribe Foundation.

  • Dr. Grunberger shared that he still has patients who perform 10, 15, or even 20 fingersticks per day to make sure that their blood glucose matches their CGM reading. He commented, “I’m sure they never leave the house because they need to make sure it matches. But it’s not the same medium (i.e., plasma vs. blood), and you have lag time, so you need to set their expectations properly.” This “obsession” can be damaging as he’s seen some patients decide that their CGM is “a piece of crap” so they don’t continue using it. He emphasized that patients must be taught to focus on the trends and patterns in their blood glucose, and less so on discrete glucose values. This also makes an argument for factory-calibrated CGM, because if there is no BGM comparator, then there is no discrepancy to “obsess” over. Reminding patients that many popular BGMs are highly inaccurate (see DTS’ BGM Surveillance Program and this 2016 study from Dr. Steve Russell’s group) – and worse in the real world when compounded by dirty hands. With factory calibration and continued improvements in CGM accuracy, we expect the subjective experience of using CGM to improve enormously. 

  • The challenge facing providers now is not only to learn how to interpret CGM data, but also how to continue to get paid for it, according to Dr. Grunberger: “Any techie could figure out a way to do this, so why do we need doctors? How do we make sure that we stay in the forefront, that we are the ones that figure out how to use this technology to improve our patients and lives and also survive as a profession?” This is obviously not an either/or – even if CGM manufacturers and remote data analytics companies took on most of the data analysis for CGM, patients would still need a licensed provider to further contextualize data, prescribe medications, support them psychologically, and help to establish and adjust treatment plans. That said, an oft-heard (and justified) grievance from endocrinologists is that they aren’t paid enough for all of the work that they do, so they do need to make sure that they are reimbursed at adequate levels for their data reviews and treatment guidance.

    • UCSD’s Dr. Tim Bailey discussed new reimbursement billing codes for CGM starts and collection/interpretation of glucose data on Day #1.

UCSD’s Dr. Steve Edelman and Marina Diabetes Center’s Etie Moghissi, in separate symposia, discussed their approaches of getting CGM on patients as soon as possible post-diagnosis. Dr. Edelman said that no matter what age a patient is diagnosed with type 1 diabetes, he or she should have a real-time CGM within 24 hours (“I don’t believe in blinded CGMs for type 1, but that’s a matter of debate”). In a slight departure from Dr. Edelman’s philosophy, Dr. Moghissi gives every patient, regardless of diabetes type, a professional CGM at baseline, and then gives them an appropriate treatment regimen plus real-time CGM. In type 1, we can see both approaches having merit: Dr. Moghissi’s in that she can properly titrate insulin doses prior to the behavioral effects of real-time CGM and without patients being overwhelmed by too much data off the bat, and Dr. Edelman’s in that CGM can be useful and lifesaving for someone newly-diagnosed and help them immediately start climbing the learning curve. Slight practice disagreements aside, both esteemed physicians preached the benefit of CGM as a tool, throwing around descriptors like “life-changing” and “standard of care” frequently.

  • Dr. Edelman and Dr. Moghissi also agreed that there is no substitute for sitting down with patients and reviewing glucose profiles. According to Dr. Edelman, “you don’t have to be a rocket scientist to understand data from CGM. The best thing to do is sit down with them and go through it. It takes time. I also like to download their best day, and ask them what they did that day” (Bright Spots!). Dr. Moghissi argued that the educational value of sitting down with patients and explaining graphs is “amazing,” and criticized the idea of having patients “mail the sensor in to be analyzed.” Face-to-face education is likely the best way to teach and incite behavior change, but with many endocrinologists working on shoestring time budgets as is, we also see tremendous value in mail-in analysis services – either provided by the physician, the manufacturer, or a third party.

  • On an encouraging note, Dr. Moghissi said that, with new reimbursement billing codes, prescribing CGM is not a money-losing proposition: “You can now charge for putting a personal sensor on, and you can charge the interpretation code every time the patient comes in. It’s not a money-losing proposition. The value that it brings to my practice and to your practice is priceless.” This is encouraging to hear and we hope to hear more HCPs share this at conferences!

6. Medicare Considering Elimination of Prior Auths for CGM & Pumps Through Red Tape Relief Project

Good news on the prior authorization (PA) front: According to Galileo Consulting Group’s Ms. Jill Rathbun, the Ways and Means Health Subcommittee’s Medicare Red Tape Relief Project is considering eliminating PAs for certain drugs and devices (including CGMs and pumps). As background, the Project is the government’s ambitious undertaking to eliminate documentation and laborious steps in Medicare that don’t enhance patient care. PAs are still very common for Medtronic and Dexcom sensors – FreeStyle Libre is exempt since it is distributed through the pharmacy channel – causing a lot of providers a lot of undue headache. Lifting the PA requirement would massively reduce barriers to prescription and perhaps even increase penetration (our speculation). According to a recent AMA Survey of 1,000 physicians, 64% report waiting at least one business day for PA decisions, and 30% report waiting three business days or longer; 92% said the PA process delays patient access to necessary care, and 78% say it can lead to a patient abandoning a recommended course of treatment. A medical practice completes a mean ~29 PA requirements per physician every week, which takes nearly 15 hours to process – yikes! Hopefully, some of this burden will be alleviated in diabetes, especially for CGM. UnitedHealthcare took a big step by eradicating CGM PAs this past January – we imagine CMS could turn to UHC data as a “test case” to demonstrate the productivity benefits and patient outcomes from doing so.

7. Two Valeritas V-Go Posters: (i) Time-in-Range (80-140 mg/dl) Up Two Hours/Day (n=7); (ii) A1c Down 1.3% (Baseline: 9.8%); Distribution Agreement Signed for Germany/Austria

In a Valeritas-sponsored, five-week pilot study (n=7 individuals with type 2 diabetes) conducted in a real-world practice, the 24-hour, basal-bolus insulin patch delivery device (V-Go) significantly increased time-in-tight-range (80-140 mg/dl) from 25% to 33% (up ~two hours/day) vs. a seven-day run-in period. Time <80 mg/dl significantly decreased with V-Go use from 8.7% to 6.6% (-30 minutes/day), as did insulin total daily dose (TDD), which significantly decreased by a remarkable 39% (97 units/day to 59 units/day). Only one of the seven patients did not see an improvement in time-in-range. See AGPs of three of the participants in the picture below. All patients reported that they were less likely to miss mealtime insulin doses with V-Go and indicated higher treatment satisfaction and ease of use as compared to MDI, testaments to the device’s discretion and simplicity. Eligible patients wore blinded professional CGM for an initial seven-day period, after which they switched to V-Go. Following 30 days of V-Go use, a second blinded CGM evaluation was conducted and a patient questionnaire administered. To be eligible for the study, participants had to be diagnosed with type 2 diabetes and on an insulin regimen of >3 injections/day. While it’s encouraging that Valeritas used time-in-range as an outcome (the first time, to our knowledge; historically Valeritas’ trials have tracked A1c and TDD), we hope to see the company use the now-consensus glycemic buckets in the future, including in-range (70-180 mg/dl) and hypoglycemia (<70 mg/dl). It’s important to note the potential biases of a very small sample size, but the results remain promising for V-Go and suggesting many more could benefit from this device than currently are.

  • A separate three-month study (n=60 individuals with type 2 diabetes) conducted across multiple endocrine clinics within Southern California Kaiser Permanente Group found that V-Go use significantly reduced A1c by 1.3% from a high baseline of 9.8% over the course of three months. Insulin total daily dose (TDD) also decreased significantly, falling by an impressive 33% (-24 units/day; baseline: 72 units/day). Prior to initiating V-Go, 60% of patients had A1c >9%; by the end of the study, this percentage was reduced to just 22%. The proportion of patients with A1c <8% increased from just 5% to 35%. This is very promising data in a very tough population! As with the above study, time-in-ranges would ideally be tracked in the future to add further valuable context to the robust A1c-lowering.

  • As a side note, Valeritas just today announced a distribution agreement with MED TRUST, an Austrian-based medical device manufacturer and multi-national distributor specializing in diabetes equipment, for the commercialization of V-Go in Austria and Germany. Per the press release, Austria and Germany combined have over eight million adults with diabetes, making for  a large target market. Similar to Valeritas’ other international distribution agreements in Puerto Rico, Italy, Australia, and New Zealand, Valeritas will retain responsibility for product development, regulatory approval, quality management, and manufacturing, while MED TRUST will oversee sales, marketing, customer support, and distribution activities. No launch timing was given for Austria and Germany; an international launch “at some scale” in Australia and New Zealand is expected before the end of 3Q18 and an Italian launch expected after the summer. It’s great to see Valeritas finally prepping for international roll outs after having been CE marked for quite some time.


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