American Diabetes Association 79th Scientific Sessions

June 7-11, 2019; San Francisco, CA; Day #4 Highlights

Executive Highlights

  • Hello from ADA 2019 in San Francisco!

  • In diabetes therapy, we’ve now seen two out of three major CVOT readouts coming at ADA 2019, as Lilly/BI’s DPP-4 Tradjenta (linagliptin) was strongly neutral vs. sulfonylurea glimepiride on three-point MACE (HR=0.98, 95% CI: 0.84-1.14) in CAROLINA. Neutrality was also found on the contentious components of CV death (HR=1.00, p=0.99) and hospitalization for heart failure (HR=1.21, p=0.18). Notably, Dr. Darren McGuire argued that cardiologists should be reassured by the CV safety of glimepiride in this population (extension to other SUs is uncertain), and even asserted that FDA should revisit the product-label warning for CV mortality for this specific SU. Glimepiride conferred significant increases vs. linagliptin on hypoglycemia, severe hypoglycemia, and hospitalization for hypoglycemia for the entirety of the study’s 6.3 year median follow-up – the “real world” is no doubt worse. We’d love to see patients have a chance to use medicines that are not associated with hypoglycemia, to say nothing of weight gain, which although in this randomized group, was neutral in the SU group to the DPP-4 inhibitor group, we’d love to see RWE (real world) on this. And, indeed, although it is surely true that SUs work fine for some patients, as a population approach, we would like to see other alternatives made available even though more than less reassurance on the CV front is certainly very welcome.

    • A day after Lilly’s GLP-1 Trulicity gave a 12% risk reduction on three-point MACE in REWIND, consistent across primary and secondary prevention cohorts, we’ve gathered opinions from key opinion leaders on the results. While REWIND adds to the existing evidence base for cardioprotection as a GLP-1 class effect, some remain unconvinced about use in those without established CV disease. See below for thoughts from Drs. Alice Cheng, Naveed Sattar, Francesco Giorgino, Charles Alexander, Zachary Bloomgarden, Philip Home, Dan Drucker, Sanjay Kaul, and David Harlan. Indeed, there are very notable takeaways for those with established CV disease, of course.

    • A post-hoc of 21-year STENO-2 results found very significant reductions in stroke with multifactorial intervention (blood pressure, lipids, glycemia) vs. standard-of-care. We continue to believe that STENO-2, which found an extra year of life expectancy gained for each year near targets, is undervalued in terms of the evidence it offers for validating targets and preventing costly complications, despite being a relatively small population (n=160).

    • Renal post-hoc analyses of AWARD VII (Trulicity vs. insulin glargine) and EMPA-REG (CVOT for SGLT-2 inhibitor Jardiance) furthered our understanding of renoprotection with novel agents. Trulicity reduced risk of a composite renal endpoint of (i) >40% eGFR decline; and (ii) ESRD while renal benefit was maintained in nonproteinuric CKD with Jardiance – an extremely prevalent group within the diabetes population.

    • Also today, new EMPRISE data supported the real-world CV efficacy and safety of Jardiance vs. DPP-4s, Drs. D’Alessio and Wilt debated ADA (<7%) vs. ACP (7%-8%) A1c targets, and a concerning CVOT meta-analysis found no CV benefit with novel agents in black study participants.

  • In diabetes technology:
    • CITY and SENCE results read out testing Dexcom’s G5 CGM in younger age groups over six months. In CITY (n=153), CGM drove a 0.4% advantage on A1c and a +1.7 hour/day benefit on time-in-range in a challenging type 1 adolescent (14-25 years) population (baseline time-in-range: 37%) over six months. We were elated to see such positive results alongside extremely high wear time in this population; of course, more work is still needed, as participants still spent 13 hours per day in hyperglycemia. The SENCE study (n=143) similarly tested CGM in 2-8 year olds with a high A1c (8.2%) and low time-in-range (~40%). CGM + a family behavioral intervention drove a 43 minute/day gain in time-in-range (70-180) vs. no change in the CGM-only and BGM groups. Both CGM groups saw significant improvements in time <70 mg/dl and significant reductions in severe hypoglycemia. CGM wear was very high - ~90% of the groups were wearing the G5 for >6 days per week at the end of six months.

    • Dr. Rich Bergenstal presented an exciting oral on individualized A1c, which accounts for variations in red blood cell turnover between individuals. The team found that 70% of individualized A1c values differed from their lab A1c by over 0.5%. As long as A1c is still being used on an individual level – to which someone aptly questioned: “why not just kill A1c all together?” – Dr. Bergenstal proposed that an individualized A1c be implemented, personalized by the patient’s RBC turnover.

    • In big data, at an Insulet dinner symposium, Dr. Trang Ly highlighted three ADA posters that display real-world GMI (glucose management indicator) data from nearly 40,000 Omnipod users. We also saw an interesting UK audit on FreeStyle Libre.

    Greetings from day #4 of ADA 2019, where the last full day of talks is in the books! This highlights report contains our top takeaways in therapy, including full CAROLINA results, a compilation of insightful KOL thoughts we’ve collected on REWIND, and a stroke post hoc of the 21-year (!) follow-up from STENO-2, and CITY & SENCE results testing Dexcom's G5 CGM in pediatrics, adolescents, and young adults over six months. And – day #5 will bring results from the PIONEER 6 CVOT for Novo Nordisk’s oral semaglutide and new analyses from the CREDENCE (renal outcomes for J&J’s Invokana) and CARMELINA (CVOT for Lilly/BI’s Tradjenta) trials.

    Table of Contents 

    Diabetes Therapy Highlights

    1. CAROLINA Finds No Significant CV Differences with Tradjenta vs. SU Glimepiride; Substantially Lower Hypo Risk with the DPP-4, but No Meaningful Shown Disparity on A1c or Weight

    In the second of three major CVOT readouts at ADA 2019, full results from the CAROLINA CVOT (n=6,033) for Lilly/BI’s DPP-4 inhibitor Tradjenta (linagliptin) vs. the sulfonylurea glimepiride were presented. The sponsors had previously announced non-inferiority of the DPP-4 to glimepiride on the primary three-point MACE outcome, and Dr. Nikolaus Marx today announced a hazard ratio of 0.98 (95% CI: 0.84-1.14, p<0.0001 for non-inferiority, p=0.76 for superiority) – a very neutral finding indeed, and one that brings clarity to decades of concern and confusion over the cardiovascular safety of glimepiride specifically though not necessarily to the class. Notably, the biggest area of interest going into the presentation was hypoglycemia, and investigators identified substantially lower risk on all types of hypoglycemia with linagliptin vs. glimepiride – find the full details below. Comparatively, the treatment differences on A1c and body weight were less meaningful, leading trial presenters to conclude that when it comes to second-line oral therapies, DPP-4s remain preferred to glimepiride in all cases save those where cost is of highest concern. Below, you’ll find far more detail on the nine-part, two-hour symposium, and you can also take a look at our updated CVOT landscape.

    • Introduction and study background. Dr. Julio Rosenstock set the scene for CAROLINA, outlining the current place of DPP-4s vs. SUs in the ADA/EASD treatment algorithm. Currently, DPP-4s are recommended in the case that hypoglycemia minimization or weight management are key concerns, while SUs remain a recommended option only when cost is a critical issue. While SUs are cheap, supported by extensive clinical experience, and carry evidence for improved glycemic control and microvascular benefit from UKPDS, he said, they also carry the well-known baggage of hypoglycemia risk, weight gain, limited durability, and – critically – lingering concern over long-term CV safety, dating back to the UGDP trial with tolbutamide in 1970. Since then, all SUs have carried a label warning in the US for a possible increased risk of CV mortality. Despite their de-prioritization, he emphasized, SUs continue to be widely used. Thus, CAROLINA was undertaken to give definitive evidence on the relative merits (CV and hypoglycemia safety, as well as efficacy) of linagliptin vs. glimepiride as add-ons to metformin in early type 2 diabetes.

    • Study design, statistical plan, and baseline characteristics. Dr. Mark Espeland presented the study design, statistical plan, and baseline characteristics. CAROLINA was an international trial, with 607 sites in 43 countries, and the first DPP-4 inhibitor CVOT to use an active comparator in sulfonylurea glimepiride. Over 10,000 patients were screened, had a 2-4-week placebo run-in period, and then were randomized to either linagliptin or glimepiride. For those on glimepiride, a titration phase lasting up to 16 weeks was conducted to ensure proper dosing in a safe and effective manner. CV risk inclusion criteria included having at least two of the following factors: previous vascular disease, evidence of vascular related organ damage, older than 70 years, or at least 2 CV risk factors. In total, 3,023 patients were in the linagliptin group and 3,010 in the glimepiride group. Mean age was 64 years and mean A1c in both groups was ~7.2%. On a broad level, this patient population reflected those typically seen in clinical practice who are relatively early in type 2 disease progression, making CAROLINA broadly applicable. The median duration of diabetes for the population was 6.3 years, 42% had established CVD, 0% were on insulin, and 83% were on metformin.

    • Metabolic and body weight outcomes. Dr. Steven Kahn presented metabolic and weight outcomes. Consistent with reported action for the drug class, glimepiride showed a dramatic A1c reduction in the first 16 weeks of treatment, then this effect waned back to and above the baseline level before the end of the study. By the end of the study, the linagliptin group had only a slightly lower A1c than the glimepiride group (see graph of change over time below). There were no significant differences in the new introduction of diabetes drugs post-baseline in either group (see table below), and the time to initiate such drugs was not significantly different, either. On body weight, linagliptin was associated with a significant reduction of 1.5 kg (95% CI: 1.8 kg to 1.3 kg weight loss) when compared to glimepiride at the end of the trial, although it is notable that the glimepiride group also achieved weight loss when compared to baseline (~0.7 kg). No significant between-group differences were seen in SBP, DBP, LDL-C, HDL-C, or triglycerides.

    New introduction of Glucose-lowering medications post-baseline:




    Any glucose lowering therapy



    Non-insulin therapies






    • Cardiovascular, mortality, and hospitalization outcomes. Dr. Nikolaus Marx shared CV outcomes with the eager crowd, wasting no time in revealing the strong neutrality on 3-point MACE (HR=0.98, 95% CI: 0.84-1.14, p<0.0001 for non-inferiority, p=0.76 for superiority). Event rates for each arm were equal at 2.1 per 100 patient-years, with 362 events in the glimepiride arm compared to 356 in the linagliptin arm. This result was consistent across all subgroups of previous vascular disease, age, sex, prior anti-diabetic treatment, background metformin, A1c, and baseline eGFR. When hospitalization for unstable angina was added to create a 4-point MACE – the original primary endpoint for the trial – results remained neutral (HR=0.99, 95% CI: 0.86-1.14, p=0.87). None of the components achieved significance in either direction, with non-fatal stroke showing the strongest trend in favor of linagliptin (HR=0.87, 95% CI: 0.66-1.15, p=0.34). Non-significance was particularly noteworthy for the CV death component (HR=1.00, 95% CI: 0.81-1.24, p=0.99), in light of the doubled risk for this endpoint with the sulfonylurea tolbutamide in the UGDP. However, a trend in favor of linagliptin on non-CV death (HR=0.82, 95% CI: 0.66-1.03) did shift the point estimate on all-cause mortality (HR=0.91, 95% CI: 0.78-1.06, p=0.23). Lastly, given the complex history of DPP-4 inhibitors and hospitalization for heart failure (saxagliptin was associated with a concerning signal in the SAVOR-TIMI trial on this endpoint), it was reassuring to see no increased risk vs. glimepiride, though the point estimate did favor the latter (HR=1.21, 95% CI: 0.92-1.59, p=0.18). To be sure, a signal was certainly not expected after linagliptin came out clean as a whistle on HHF vs. placebo in CARMELINA (HR=0.90, p=0.2365). Altogether, with no significance on any endpoint and near perfect unity on 3-point MACE and CV death, Dr. Marx concluded linagliptin and glimepiride are virtually identical on CV outcomes.

    • Adverse events and hypoglycemia analyses. Dr. Bernard Zinman dove into highly-anticipated hypoglycemia results; unsurprisingly, glimepiride produced significant increases on every hypoglycemia metric vs. linagliptin. 37.7% of patients experienced any form of hypoglycemia in the glimepiride arm vs. 10.6% in the linagliptin arm, resulting in a hazard ratio of 0.23 (95% CI: 0.21-0.26, p<0.0001). Considering only moderate (≤70 mg/dL) and severe (requiring assistance) hypoglycemia, linagliptin conferred a sizable 82% risk reduction (HR=0.18, 95% CI: 0.15-0.21, p<0.0001), and on the serious and costly outcome of hospitalization due to hypoglycemia, a significant, 93% risk reduction (HR=0.07, 95% CI: 0.02-0.31, p=0.0004), despite only 29 events occurring (27 with glimepiride, 2 with linagliptin). For all metrics, the slope of the Kaplan-Meier curve with glimepiride outpaced that of linagliptin, suggesting they would have further separated with a longer follow-up. Increased hypoglycemia with glimepiride was consistent across all subgroups, except when stratified by prior sulfonylurea or glinide use, for which an increased risk of hypoglycemia was found regardless of treatment group in those with prior use (p<0.0001). All other safety outcomes, including cancer and pancreatitis, were neutral.

    • Hypoglycemia was prevalent and consistent throughout the trial in the glimepiride arm, and across all glimepiride doses. This includes the titration period, in which many participants were on lower doses, demonstrating that hypoglycemia is dependent only on sulfonylurea presence, not on dose.

    • The high incidence of hypoglycemia in the glimepiride arm conferred positive secondary metabolic efficacy outcomes for linagliptin, including significantly more people achieving an A1c ≤7.0% without need for rescue medication and with weight gain <2% (OR=1.29, 95% CI: 1.11-1.48, p=0.0004). When moderate/severe hypoglycemia was added to this endpoint, the difference became more pronounced (OR=1.68, 95% CI: 1.43-1.96, p<0.0001). Of course, these endpoints were built to favor linagliptin, considering that weight gain and hypoglycemia (and therefore greater likelihood of requiring rescue medication) are known class effects of sulfonylureas.

    • While CAROLINA was not designed to test the relationship between hypoglycemia and CV risk, the results do not demonstrate a correlation between the two. However, many questions still have to be answered before this could be definitively concluded. For example, when did CV events occur in relation to hypoglycemia events in the trial (right after, right before, etc.)? Could the effect between the two be longer-term than the 6.3 year follow-up? Was there increased CV risk for those with multiple hypoglycemia events?

    • Clinical implications for the cardiologist. Dr. Darren McGuire shared that the cardiology community has also, for some years, been concerned about the CV safety of SUs, welcoming the clarity that CAROLINA will bring (presumably it would be viewed as clarity on glimepiride, but not on the entire class). While he compared the CAROLINA population to that of CARMELINA to highlight the relatively lower risk of the former, he did challenge the notion that that CAROLINA enrolled a “low risk” population – the event rate of 2.1/100 person years, he said, is a CVD equivalent. That said, Dr. McGuire welcomed the data as the first-ever RCT without substantial limitations assessing the CV safety of SUs, pointing particularly to the result on all-cause mortality (HR=0.91, 95% CI: 0.78-1.06), including a perfectly neutral result on CV death (HR=1.00, 95% CI: 0.81-1.24). When looking to extrapolate glimepiride’s CV safety based on only RCT evidence (see image below), Dr. McGuire noted apparent heterogeneity. In summation, he argued, cardiologists should be reassured by the CV safety of glimepiride in this population (extension to other SUs is uncertain), and even asserted that FDA should revisit the product-label warning for increased CV mortality for this specific SU. We’re not sure which body would submit this.

    • Clinical implications for the diabetologist. Dr. Bernard Zinman turned to hypoglycemia, A1c, and weight outcomes in his discussion of clinical implications, first and foremost emphasizing the importance of hypoglycemia outcomes due to their impact on cost, hospitalizations, fall injuries (particularly in the elderly), fear of hypoglycemia (and subsequent under-treatment), quality of life, and potential CV risk. To be sure, he said, hypoglycemia remains the primary concern with SUs: Relative to glimepiride, linagliptin was associated with a 77% lower risk of any hypoglycemia (95% CI: 0.21-0.26), an 82% lower risk of moderate or severe hypoglycemia (95% CI: 0.15-0.21), an 85% lower risk of severe hypoglycemia (95% CI: 0.08-0.29), and a 93% lower risk of hospitalization due to hypoglycemia (95% CI: 0.02-0.31) – see numbers needed to treat below. All in all, more than one-third of the glimepiride group had hypoglycemia. When it came to A1c, Dr. Zinman noted that many PCPs like to use SUs because of the early, dramatic A1c reduction they give. However, he qualified this against potential risk of hypoglycemia with rapid drops in glucose. On weight, Dr. Zinman was similarly underwhelmed by the treatment difference. When it comes to the ADA/EASD treatment algorithm and clinical practice, he summarized, CAROLINA does not impact those with CVD. And in those without CVD, it reinforces the need to avoid SUs in those where hypoglycemia is a compelling concern (some questioned this but we felt it was very helpful – some patients are far more at risk than others), but supports their CV safety when cost is a major issue. How “a major issue” is determined is another question.  

    • Independent commentator. Dr. Matthew Riddle delivered the independent commentary, describing CAROLINA as an “excellently designed and conducted” trial that achieved exactly what it set out to do. He commended the study for answering a question of major clinical significance and for statistically and clinically clear outcomes for the main endpoints. On this front, he noted that previous studies of sulfonylurea vs. active comparators (ADOPT, ADVANCE, and TOSCA.IT) all failed to show increased CV risks with sulfonylurea, but all had important limitations that hindered their generalizability.

      • Interestingly, Dr. Riddle provided a very rough risk comparison of data for hospitalizations for heart failure or hypoglycemia, two outcomes of high clinical significance. On hospitalizations for hypoglycemia, the linagliptin group had only two events, compared to 27 for the glimepiride group. On hospitalizations for heart failure, the linagliptin group had 112 events vs. 92 for the glimepiride group. In summing both of these totals, Dr. Riddle showed that the glimepiride group only had five more total events. He conceded that the conclusions are obviously limited for this very rough analysis, but broadly speaking, the low numbers seen for such a big trial speaks to the safety of both of these drugs.

      • Dr. Riddle emphasized the heterogeneity of the sulfonylurea class and cautioned against extrapolating this data to other sulfonylureas. Indeed, KOLs have often underscored a very meaningful heterogeneity within the sulfonylurea before, with glimepiride and gliclazide (the latter only available OUS), to our understanding, the better agents within the class. Dr. Riddle also cautioned against applying this data for linagliptin for other agents in the DPP-4 inhibitor class as well. We were somewhat surprised about this caution.

    • We note that before Dr. Riddle provided his commentary, Dr. Julio Rosenstock again took the stage to offer his take home messages from the trial. He noted that CAROLINA was in part designed in response to a 2012 article from Dr. Steven Nissen, who lamented the fact that millions of patients were receiving sulfonylureas without high quality evidence regarding their CV safety. He deemed this “continued darkness” unacceptable, and CAROLINA was therefore designed to “make the dark ages brighter” by generating data in this space. Dr. Rosenstock provided three main points from the trial: (i) CAROLINA supports the use of linagliptin before glimepiride, other than in relation to cost considerations (which are a very valid concern short-term – longer-term, of course, costs related to hypoglycemia will effectively be ignored); (ii) CV safety should no longer be a consideration in the decision-making process for selecting between these two agents (though not between the classes); and (iii) CAROLINA reaffirms current clinical recommendations to choose an agent after metformin based on proven cardiovascular benefit, which neither linagliptin or glimepiride provide. Presumably, were a doctor to choose an agent based on hypoglycemia outcomes wouldn’t be out of the question, but there are major considerations here.

    2. KOLs Sound Off on REWIND: Comments on GLP-1 Class Effects, Renal Data (“Mildly Disappointing”?), Clinical Practice Influence, and More

    On how REWIND fits into existing GLP-1 CVOT data, and whether cardioprotection in primary prevention might be a class effect:

    • Dr. Alice Cheng: There are important differences between GLP1 molecules in terms of half-life, derivation, size etc.  The effects on secondary prevention seem consistent for the GLP-1 class. The story is less clear cut for the CV benefits for primary prevention as a class effect because of the small numbers in the other studies (LEADER, SUSTAIN-6).  I would like to see more data to be more confident. 

    • Dr. Naveed Sattar: One suspects it is true for several GLP-1 agonists - there may be subtle differences between the main GLP-1s, but for now, the broad effects and outcome benefits seem more similar than different.  That said, guidelines will continue to recommend those GLP-1s which have proven significant CVD benefit

    • Dr. Mikhail Kosiborod: I believe this is a class effect for long-acting GLP1 RAs. Whether others will be motivated to pursue trials that include large proportions of lower risk patients with type 2 diabetes remains to be seen. Such trials are challenging to conduct and complete – since (as can be seen in REWIND, and as expected) the event rates are substantially lower than in the higher risk patients, and required duration of follow up, therefore, is substantially longer (as these are event driven trials).
    • Dr. Francesco Giorgino: I do not see why it should not be a class effect, provided that adherence to treatment and time exposure to the drug is high (e.g., >80%).
    • Dr. Charles Alexander: Most prescribers are likely to see the positive results from REWIND as a class effect. There is no reason to believe that the results are unique to dulaglutide.  I would not expect to see additional outcomes trials initiated since they are so expensive and labor-intensive.  Once a company has accumulated outcome results sufficient to satisfy FDA, they will not have any incentive to launch another one at this point. 

    • Dr. Zachary Bloomgarden: The long-acting GLP-1RA now appear to have a class effect of reducing adverse CV outcome - this is good news!

    • Dr. Philip Home: REWIND places another piece in the CV protection jigsaw. But clearly the findings of all the GLP1-RA studies taken together show many inconsistencies, likely as a result of chance.  The finding of no interaction for MACE between CV and CV risk populations is encouraging, but not consistent with the other CVOTs, and thus not conclusive. Accordingly, this does not allow the conclusion that we should be prescribing GLP-1RAs for CV protection in people who do not already have CV disease.

    • Dr. Dan Drucker: REWIND provides further support for the cardioprotective actions of the long-acting GLP-1 class. While each trial population, design, and study drug is unique, it is highly reassuring to again see a reduction in MACE, now in a lower baseline A1c population with less pre-existing CV disease.

    Interpreting exploratory renal analysis from REWIND:

    • Dr. Sanjay Kaul: The renal data are interesting exploratory results which are driven by the most prevalent, but arguably the least important component (microalbuminuria). Of note, the choice of eGFR decline >30%, while amplifying the event rate, did not enhance the chance of success. Had they chosen eGFR decline >40% or >50% in the composite, there would be a larger effect size on the renal composite, albeit with lower frequency of events. Thus, softening the criterion for a win by lowering eGFR decline threshold might not be a successfully strategy.
    • Dr. Naveed Sattar: As I understand it, the main effect was on albuminuria but not on eGFR. I think to prevent deterioration on harder renal outcomes, SGLT-2 inhibitors have an advantage and one imagines more clinicians are more impressed by the SGLT-2 inhibitor effects on renal outcomes.   

    • Dr. Dan Drucker: The renal data, while perhaps mildly disappointing, may reflect the low rate of pre-existing renal disease and more studies of different populations are warranted. 

    • Dr. Charles Alexander: The renal results were good across the board, but not as good as the SGLT-2 inhibitor results (e.g., CREDENCE) where you see amelioration of the decline in eGFR. REWIND showed continued decline in eGFR despite the positive renal outcomes.  Subsequently, when treating a patient with diabetic nephropathy, the preference will continue to be SGLT-2 inhibitor.  The data in CREDENCE is very convincing that the previously certain inevitable progression of diabetic nephropathy can be aborted by a SGLT-2i.  Although GLP-1 RA improved renal outcomes in REWIND, it did not do as well as SGLT-2i.

    • Dr. Mikhail Kosiborod: The renal data are encouraging, but are primarily driven by albuminuria, rather than “hard renal endpoints” – similar to what we have seen before in GLP1 RA trials. The effects on meaningful reduction in eGFR need to be confirmed in dedicated renal outcome trials. Until then, the renal benefits of GLP1 RAs remain to fall short of the dramatic benefits we have seen with SGLT-2i,
    • Dr. Zachary Bloomgarden: The renal data also indicates a class effect, reducing albuminuria and additionally with modest benefit in slowing the rate of fall in GFR.

    • Dr. Francesco Giorgino: It looks like the effect is on albuminuria largely, as for other GLP-1RA. I am a little surprised that baseline eGFR did not impact on the renal outcomes.
    • Dr. Philip Home: The most intriguing exploratory results reported in REWIND is the much greater statistical power for change in creatinine for reduction of 50% compared to 40% or as pre-specified here 30%. This suggests we should re-examine some of the data for GLP-1RAs and SGLT2 inhibitors using these discriminants, as we may be missing an important message. But again, chance may be misleading us. 

    • Dr. Alice Cheng: As for the renal data, the majority of the benefit was driven by the urine ACR which was seen with other GLP1RA studies so there is consistency. 

    On the robustness of REWIND data:

    • Dr. Sanjay Kaul: The effect size on the primary endpoint is modest and statistically not robust (p=0.03). These results fall short of the “substantial evidence” criterion required for approval based on a single trial. A couple of extra events in the dulaglutide arm or a couple of fewer events in the control arm would overturn statistical significance. The fragile data are particularly prone to missing data. I did not see sensitivity analyses accounting for missing data.

       Unlike LEADER where the point estimate was a HR of 1.20 and SUSTAIN-6 and EXSCEL where the point estimate was a HR of 1 and 0.99, respectively, the point estimate in REWIND is a HR of 0.87. .  Guidelines appropriately restrict the use of approved GLP-1 RA for mitigating CV risk to patients with established CV disease. I would have liked to see a greater treatment benefit in the cohort with established CV disease. In contrast with semaglutide that showed a 4.5% absolute risk reduction in SUSTAIN-6 in patients with established CV disease, the absolute risk reduction in REWIND was 2.4%. 

    • Dr. Zachary Bloomgarden: The benefit on cardiovascular outcomes is somewhat modest - certainly less than that of statins and of BP-lowering treatment. The NNT is likely not to be very impressive [Editor’s note: the NNT is 60 for those without established CVD, and 18 for those with established CVD].

    On how REWIND may impact clinical practice:

    • Dr. Alice Cheng: These results are definitely going to influence clinical practice. If you had asked me back in 2015 if any of our diabetes therapies would reduce CV outcomes in a majority primacy prevention population, I would have said that is wishful thinking. Now, it is reality. I hope this will shift treatment paradigms to first focus on outcome-based therapies that happen to lower A1c then if still needed, add other therapies to further lower A1c. 

    • Dr. Naveed Sattar: Evidence adds to now strong belief that GLP-1 receptor agonists lessen vascular disease risk in diabetes patients whether or not they have existing CVD - this new evidence will continue to push towards these agents to be used sooner in algorithms and clinical care but of course the higher costs means this will be a slow process for now. Budgets are not boundless so higher risk patients will likely be targeted first. That the results show benefit despite excellent baseline A1c further supports an effect which is not mediated only by glucose.  

    • Dr. Francesco Giorgino: I think the issue of the CV risk level of the patient should be reconsidered on the basis of the REWIND data. In the context of several-year treatment and more stringent glucose control without hypoglycemia GLP-1RAs seem to be protective also for individuals without CVD. 

    • Dr. Mikhail Kosiborod: The results of REWIND are consistent with what we have seen before, and confirm the beneficial effects of GLP-1 RA, as a class, on major adverse cardiovascular events in patients with Type 2 diabetes with established atherosclerotic cardiovascular disease (ASCVD). The effects of dulaglutide on these ASCVD events were modest, but statistically significant. The REWIND trial also extends these beneficial effects to patients with Type 2 diabetes that are in a “high risk primary prevention” cohort, as it had the greatest proportion of such patients across all the large GLP1 RA CVOTs. This is clinically important, as the majority of patients with T2D do not yet have established ASCVD.
    • Dr. Charles Alexander: I don't think that REWIND will change ADA guidelines (especially given the recent ADA update) and don't think that it will change prescriber behavior beyond what they were already doing or going to be doing based on other recently reported CV outcomes trials (e.g, CREDENCE, LEADER, etc.).  That said, it was a great result for Lilly as they now have a GLP-1 RA that shows superior CV outcomes like the results for liraglutide and semaglutide. The DPP-4 inhibitor class cannot compete with the SGLT-2i/GLP-1 RA results and you will see that class being used less and less especially after oral semaglutide is approved.

    • Dr. Zachary Bloomgarden: I would like the results to influence clinical practice, since I happen to believe that glucose control matters, and therefore that these drugs should play a role throughout the natural history of type 2 diabetes.

    • Dr. David Harlan: Being a clinician scientist who is, when it comes to the patients I serve, concerned first, last, and always in doing all that I can to promote their health -- this study just strengthens my belief (based on good data) that nearly all patients with T2D with suboptimal glycemia control and/or other factors like worsening obesity should be considered candidates for a GLP1R based therapy.

    Interpreting individual MACE component effects:

    • Dr. Philip Home: REWIND again emphasized the importance of the GLP-1RAs in stroke reduction.  This is important as the SGLT2-blockers show no effect at all on stroke.  Furthermore, in the REWIND population stroke was nearly as frequent as MI as an endpoint, and a cause of more deaths. 

    • Dr. Sanjay Kaul: The reduction in nonfatal stroke (the only component that was significantly reduced) is consistent with semaglutide in SUSTAIN 6 and PIONEER 6, but not with albiglutide in HARMONY. 

    • Dr. Zachary Bloomgarden: The CV data is strongest for stroke prevention - very interesting, and this also confirms some of the other GLP-1RA data. The mortality data is suggestive, again, not a strong effect

    • Dr. Charles Alexander: I was also surprised that nonfatal stroke was the endpoint that was driving the statistical significance of the MACE composite endpoint in this study. CV death was next and nonfatal MI had the least effect on the composite endpoint and was not different between the 2 groups.  That is very strange.  Other GLP-1 RA studies have shown benefit for MI.

    Disclosure: Dr. Kosiborod was an investigator in the REWIND Trial, received research grants from AstraZeneca and Boehringer Ingelheim, and is a consultant for AstraZeneca, Boehringer Ingelheim, Janssen, GSK, Merck (Diabetes), Novo Nordisk, Sanofi, and Intarcia

    3. 21-Year STENO-2 Follow Up Finds 69% Reduction in Stroke with Multifactorial Intervention vs. Standard-of-Care; +1 Year of Life Expectancy for Each Year on Intervention; Undervalued Study Now Officially Finished

    Steno Diabetes Center’s Dr. Peter Gæde shared new post-hoc findings from the STENO-2 study (n=160), showing that multifactorial intervention (simultaneous glucose, blood pressure, and lipid lowering) reduced the risk for stroke by 69% vs. standard of care (p=0.004), while conferring an extra 7.9 years of life expectancy (p=0.002) and 8.1 years without a 3-point MACE event (p=0.005). Stroke results were simultaneously published in Diabetologia; survival data first surfaced in 2016. After 21 years, including the 8-year intervention period and 13-year follow-up, only 9 of the 80 patients in the intervention arm had a stroke (11%) vs. 21 in the standard-of-care arm (26%). The secondary endpoint, transient ischemic attack (often called a “ministroke”), was reduced by a similarly impressive 92% (p=0.019), though event rates were low. When combined with stroke, composite endpoints including TIA, CV death, all-cause death, or subsequent stroke were all significantly reduced (see table below). Based on these long-term, consistent, and compelling results, we maintain that STENO-2 is undervalued by the diabetes community in terms of validating targets and preventing costly complications, despite being a relatively small study (see below the table for more).

    Intriguingly, patients in the intensive-therapy group survived a significantly shorter period of time after a stroke, on average, than those in the standard-of-care arm; the median survival in the former was 0.3 years compared to 3.1 years in the latter (p=0.015). Dr. Gæde posited that strokes can occur either as a macrovascular entity – the more prevailing etiology – or as a microvascular complication, since similar deposits which cause vascular occlusion in the kidney can also be seen in the brain. This would suggest that signs of kidney disease or other microvascular complications might be associated with an increased risk of stroke in STENO-2, and indeed, one of the two baseline risk factors to be significantly associated with stroke was reduced kidney function. For each 10 mL/min/1.73m2 higher GFR at baseline, there was a 20% reduction in risk of stroke (HR=0.80, 95% CI:0.68-0.95, p=0.012). To this end, Dr. Gæde hypothesized that the multifactorial intervention in STENO-2 more effectively prevented microvascular strokes than more serious macrovascular strokes (though there was likely a significant benefit on both), thereby explaining the shorter life expectancy after an event in the intervention arm.


    N (Intervention)

    N (Standard-of-care)


    95% CI














    Stroke or TIA






    Stroke + CV death






    Stroke + all-cause death






    Patients with >1 stroke






    • By and large, we think the results of the STENO-2 study are undervalued for both clinical and economic reasons, especially when considered alongside dismal estimates for meeting targets for glycemic control, blood pressure, or LDL cholesterol. With an 8-year intervention period conferring ~8-years extra life expectancy in the follow-up, the implication is that for each year you keep near target for lipids, blood pressure, A1c, and weight, you will gain an entire year of life expectancy. Wow! However, Dr. Gæde dishearteningly noted that only 14.3% of people with diabetes in the US met these goals in a long-term survey from 1999-2010. On day #1 of ADA 2019, Dr. Mikhail Kosiborod presented a contemporary registry analysis that found a “soberingly poor” 6% of participants with clinical ASCVD and either, (i) LDL ≥70 mg/dL or (ii) on a PCSK9 inhibitor, was on optimal medical therapy. From an economic standpoint, although these therapies and interventions (STENO-2 also included behavioral approaches) are costly in the short-term, Dr. Gæde noted that cost is neutral in the long-term due to prevention of complications, according to a presentation by his son at ADA 2018. Unfortunately, issues of adherence and therapeutic inertia are proving tough nuts to crack, but we’re hoping with collaboration from all stakeholders, the needle will begin to move.

    • After 21 informative years, Dr. Gæde confirmed that the STENO-2 follow-up is over, with over 50% of participants now having passed away. According to him, the remaining participants seem to be immortal, with mortality rates in both interventions flattening out around the 18 year mark (see below).  

    4. Drs. D’Alessio and Wilt Debate ADA vs. ACP A1c Targets

    In a lively morning session, Duke’s Dr. David D’Alessio on behalf of the ADA and Dr. Timothy Wilt, an author of the highly-controversial ACP guidelines, debated their organization’s respective guidelines on glycemic targets in diabetes. Dr. Wilt defended ACP’s guideline to aim for an A1c of 7%-8% in most patients (highly controversial throughout virtually the entire diabetes community), while Dr. D’Alessio argued in favor of a more stringent <7.0% target. Within this debate there was genuine agreement about the need to personalize A1c targets for each individual patient, but deep-rooted differences in opinion regarding the interpretation of the existing clinical trial data on glycemic control.

    • Reprising points made in a similarly-framed debate on ACP’s guidelines at ENDO 2019, Dr. Wilt defended the 7%-8% glycemic target on the grounds that available data doesn’t support meaningful improvements in outcomes with tighter glycemic control. He discussed ACCORD, ADVANCE, UKPDS, and VADT: By his characterization, two of these showed no benefit (ADVANCE, VADT), one mixed or small benefit (UKPDS – insulin/SU and metformin arms), and the last showed harm (ACCORD). While we agree that the lesson of ACCORD is not to blindly lower A1c, we also feel that all of these trials would be designed very differently if conducted with modern medicine and technology. It also remains an open question whether they achieved a big enough glycemic divide to show any difference: Is there no benefit simply because it has not been demonstrated in type 2 diabetes? Nevertheless, Dr. Wilt did aptly point to a fairly large gap in the RCT evidence, and he further cited the harm that lower A1c targets can cause – namely, higher healthcare burden, costs, and increased adverse events.

    • Financial burden was a major factor in Dr. Wilt’s estimation of the costs of aggressive A1c lowering vs. its benefits. Citing annual retail prices as high as $6,600 for SGLT-2 inhibitor empagliflozin and $12,844 for GLP-1 agonist liraglutide, he underscored the sheer inaccessibility of these therapies to average patients. He further argued that the CV protective effects of these agents don’t outweigh the steep prices: Based on three-year NNT values from their respective CVOTs, each individual CV event avoided with empagliflozin costs $1.2 million, and this rises to over $2 million with liraglutide.

      • While we hesitate to define an upper limit for how much should be spent to avoid CV events, Dr. Wilt’s points about the exorbitant costs of diabetes care are well-taken. That said, Dr. D’Alessio countered that worry over the cost of diabetes therapies is “orthogonal” to the conversation about A1c targets. “I can get patients to 7% using cost-effective drugs,” he noted, “and there are people on GLP-1 agonists, SGLT-2 inhibitors, and next-gen insulins who still have poor control.” He elaborated that patients’ financial situations should be considered a “fifth vital sign” when designing a therapeutic regimen: “The stress of trying to make ends meet is detrimental to health, and of course we need to take that into consideration in order to do our jobs.” We interpret this as a call for the wider primary care and internal medicine community to receive more comprehensive training about the use of older diabetes therapies for patients who cannot afford newer, easier-to-use therapies; although achieving a <7% A1c is likely easier with newer agents, it is certainly not impossible with older tools. 

    • Dr. D’Alessio’s defense of the ADA’s <7% A1c target hinged on the practical implications of glycemic targets: “When guidelines get loose, practice gets even looser.” In his view, clinical inertia is the biggest risk with the ACP’s new guidelines. “When I aim for a 7% A1c, I’m happy with 7.3%. If our target is 8%, will we then be happy with 8.3%? Or 8.5%?” We agree that the ACP’s elevated target A1c range could easily allow patients to drift into a dangerously high >8% A1c range. Dr. D’Alessio furthered this point with the valuable observation that for all the ACP’s emphasis on the gaps in evidence to support the benefit of a <7% A1c, absolutely no evidence exists to validate the usefulness of an 8% A1c cutoff.

    • On a more granular level, Dr. D’Alessio outlined three flaws in the rationale for ACP’s A1c guidelines:

      • The recommendation is based mainly on a limited interpretation of ACCORD, generalizing it to the entire diabetes population without consideration of treatment regimen;

      • The ACP guidelines don’t take into account newer drugs with better safety profiles, and ignore the benefits of these agents borne out in recent CVOTs; and

      • The ACP’s recommendation is overly prescriptive and restricts personalized care.

    • Finally, Dr. D’Alessio emphasized the importance of tight glycemic control for avoiding microvascular complications. While it can be argued that glycemic control is not as influential for the prevention of macrovascular complications, findings from UKPDS and DCCT clearly support the view that lower A1cs confer a lower risk of retinopathy and kidney disease – conditions which, unlike widespread cardiovascular complications such as stroke and MI, are “diabetes-defining” and a unique cause for concern among people with diabetes specifically.

    • Overall, we view this ongoing debate as an educational illustration of the gaps in the evidence for diabetes management. Dr. D’Alessio pointed out that of the countless RCTs focused on diabetes, the patient populations represented capture only ~40% of the actual diabetes population. Therefore, a majority of patients don’t have a “home” within the design of any trials in the literature, and the best course of action becomes more and more subjective and up to the clinician’s interpretation. If nothing else, we hope the continuing conversation over ACP’s controversial guidelines serves as a call for more research to better capture the diabetes population in all its diversity.

    5. AWARD-7 Post-Hoc Demonstrates ~Halving of Risk on ≥40% eGFR Decline and ESRD with Dulaglutide vs. Insulin Glargine, Driven by Macroalbuminuria Subgroup

    Dr. Katherine Tuttle presented a post-hoc analysis of the AWARD-7 trial, demonstrating that 1.5 mg dulaglutide significantly reduced risk of a composite renal endpoint of (i) >40% eGFR decline and (ii) ESRD when compared to insulin glargine, and that this was driven primarily by the macroalbuminuria subgroup. In a time-to-first-event analysis of the primary composite endpoint, 1.5 mg dulaglutide was shown to significantly reduce risk (HR=0.45, 95% CI: 0.20-0.97, p=0.04) while the lower 0.75 mg dose dulaglutide did not reach significance (HR=0.79, 95% CI: 0.41-1.5, p=0.47) – see the survival curve below. These results were primarily driven by participants in the macroalbuminuria subgroup, as event rates in the microalbuminuria and no albuminuria subgroups were much lower (see table below). Of course, this is somewhat expected given the short duration of follow-up; we would expect that longer follow-up would help glean insights into potential effects in the microalbuminuria subgroup as well (Dr. Tuttle said as much during the ensuing Q&A session). These results join renal findings from the REWIND CVOT for Trulicity presented yesterday; at present, Lilly has no plans to initiate a renal outcomes trial.

    • As a reminder, initial results from the AWARD-7 trial showed that decline in eGFR was significantly reduced with both doses of dulaglutide when compared to glargine at both 26 and 52 weeks. While eGFR declined by an average of 1.9 units in the insulin glargine group at 26 weeks, eGFR only declined by 0.1 and 0.4 units in the 1.5 mg and 0.75 mg groups, respectively. Baseline eGFR was around 38 ml/min/1,73 m2 in all three treatment arms.

    • A related post-hoc on AWARD-7 was just published in DOM (also led by Dr. Tuttle) showing that the lesser decline in eGFR associated with dulaglutide treatment was not influenced by body weight loss.

    Primary Outcome Results


    Insulin glargine

    Dulaglutide 0.75 mg

    Dulaglutide 1.5 mg

    Proportion of all study participants with composite outcome (>40% eGFR decline or ESRD)




    Proportion of all study participants with >40% eGFR decline




    Proportion of all study participants with ESRD events




    Proportion with composite outcome in macroalbuminuria subgroup




    Proportion with composite outcome in microalbuminuria subgroup




    Proportion with composite outcome in no albuminuria subgroup




    6. EMPA-REG Post Hoc Finds Consistent CV and Renal Benefits in Non-proteinuric DKD; Comparison of Proteinuric DKD to CREDENCE Bodes Well for Renal Benefit in Higher-Risk Population

    Yale’s Dr. Silvio Inzucchi provided an intriguing post-hoc of the EMPA-REG CVOT, suggesting that the CV and renal benefits with SGLT-2 inhibitor Jardiance are maintained in those with CVD and nonproteinuric DKD. Results were simultaneously announced by both Lilly and BI. Specifically, no significant p-values for interaction were found between the sub-population with eGFR <60 mL/min/1.73m2 and UACR ≤300, and the rest of participants. The only intriguing trend, as highlighted by Dr. Inzucchi, was on CV death, which trended less in favor of Jardiance than in other EMPA-REG patients. No new safety concerns were raised with empagliflozin in this population. In total, 1,290 (18%) people in the study qualified for the post-hoc, and the cohort was older, had a longer time since diagnosis, and was more female than other EMPA-REG enrollees (all p<0.0001). For context, the cohort represented here is more substantial in real life. According to Lilly’s press release, more than 500 million people globally have CKD, up to 40% comorbid with diabetes, the majority of which is nonproteinuric (~100 million people worldwide, by our calculations).

    • Prior to the sub-analysis, Dr. Inzucchi briefly compared EMPA-REG to the landmark CREDENCE trial, which has thrust SGLT-2 inhibitors into the CKD prevention and treatment spotlight of late. Including only those subjects who would have been eligible for CREDENCE, he pointed out that the (very rough) comparison between the two suggests strong renal protection with Jardiance (and as a class effect, we might add) in this higher-risk renal population (eGFR between 30 and 90, UACR >300). However, we will need full results from the CREDENCE-equivalent EMPA-KIDNEY trial to draw definite conclusions. Dr. Inzucchi reminded the audience that this trial is enrolling a broad CKD population, including both proteinuric and nonproteinuric CKD as well as people without diabetes (and also those with type 1!), to give a more complete picture of the renal protection conferred by Jardiance.

      • Dr. Inzucchi made a point of stating that nonproteinuric DKD, the population studied in the sub-analysis above, was excluded from CREDENCE despite this pathophysiology being widely prevalent. However, these patients were not excluded from the CANVAS CVOT, which found a nominally significant 40% reduction on a renal composite outcome with Invokana vs. placebo.

    7. New EMPRISE Data Supports Real-World CV Efficacy and Safety of Jardiance vs. DPP-4s

    Dr. Elisabetta Patorno (Harvard and Brigham and Women’s) presented expanded results from EMPRISE, Lilly and BI’s ~35,000-patient real-world evidence program comparing SGLT-2 inhibitor Jardiance (empagliflozin) to DPP-4 inhibitors. See all effectiveness outcomes presented in the figure below. Note that the composite CV outcome is hospitalization for MI or stroke or death; the extended CV outcome also includes hospitalizations for heart failure or coronary revascularization. As a reminder, Lilly/BI released the first results from EMPRISE last November, before presenting in-depth heart failure outcomes in a poster at AHA 2018. The outcomes presented by Dr. Patorno today build on an already-announced 44% relative risk reduction on hospitalization for heart failure with Jardiance in this real-world dataset. On the whole, some of the outcomes presented below might be considered disappointing. However, Dr. Patorno qualified these data as an interim assessment, as EMPRISE is still ongoing and the number of events (on both safety and efficacy) is small – more precision will be achieved in the future. Review the outcomes from the original EMPA-REG OUTCOME trial here.

    • Dr. Patorno also presented safety outcomes from EMPRISE, which are well-aligned with findings from EMPA-REG OUTCOME, supporting the reliability of the CV findings as well as the real-world safety of Jardiance. Data for four key adverse events was shown:

      • Bone fracture: EMPRISE identified a trend in favor of dapagliflozin on bone fracture, with a HR of 0.81 (95% CI: 0.41-1.57). In EMPA-REG, the HR was 0.94 (95% CI: 0.73-1.21).

      • Lower-limb amputation: There was a slight trend favoring placebo at HR=1.14 (95% CI: 0.64-2.04), compared to HR=1.00 (95% CI: 0.70-1.44) in EMPA-REG.

      • DKA: The point estimate most favoring placebo in both EMPRISE and EMPA-REG was on DKA; in EMPRISE, HR=1.74 (95% CI: 0.84-3.58), while in EMPA-REG, HR=1.92 (95% CI: 0.21-17.17). Importantly, both CI’s are wide.

      • Acute kidney injury: Both point estimates significantly favor placebo, at 0.64 for EMPRISE (95% CI: 0.43-0.95) and 0.58 for EMPA-REG (95% CI: 0.38-0.90).

    • EMPRISE utilized three US claims databases (Optum, 65 million lives; MarketScan, 150M lives; and Medicare, 55M lives) from August 2014-September 2019 to assess the efficacy, safety, cost, and utilization of Jardiance vs. DPP-4 inhibitors in new users (within the past year; to avoid immortal time bias). Investigators propensity score matched patients 1:1 across >140 covariates, including baseline demographics, CV conditions and medications, and indicator of diabetes severity. DPP-4s were chosen given their similar placement in treatment algorithms. Follow-up continued until treatment discontinuation or switching to comparator, occurrence of an outcome of interest, nursing home admission, death, plan disenrollment, or the end of study period. See patient characteristics here:

    8. Worrisome CVOT Meta-Analysis Finds No CV Benefit for GLP-1s and SGLT-2s in Black Study Participants

    An unsettling meta-analysis of recent CVOTs found no cardioprotective benefit for GLP-1 agonists or SGLT-2 inhibitors in black and African American people with type 2 diabetes. Across major GLP-1 agonist and SGLT-2 inhibitor CVOTs, these next-gen therapies delivered no significant relative risk reduction (RR=0.93, 95% CI: 0.70-1.23) in black study participants. This held true for GLP-1 agonists alone (RR=0.96, 95% CI: 0.61-1.53) and SGLT-2 inhibitors alone (RR=1.00, 95% CI: 0.47-2.14). This study included seven CVOTs: On the SGLT-2 inhibitor front EMPA-REG OUTCOME (empagliflozin) and CANVAS (canagliflozin), and on the GLP-1 agonist front LEADER (liraglutide), SUSTAIN-6 (semaglutide), EXSCEL (exenatide), HARMONY (albiglutide), and – very impressively – REWIND (dulaglutide), which was released only yesterday! (Major congratulations to Dr. Mishriky for updating these results so quickly!). The DECLARE trial (dapagliflozin) was not included because outcomes were not reported by race in the published results.

    • Dr. Mishriky was careful to underscore that this meta-analysis has limitations – namely a small population of black participants enrolled in CVOTs (only ~5% of the study population in most CVOTs), a correspondingly low number of events in black participants (ranging from 106 events [LEADER] to just 12 events [SUSTAIN-6] across the included trials), variation in trial design and population between different CVOTs, and the fact that the trials were not powered to specifically examine racial differences. In this way, the results can be interpreted less as an admonition that the cardioprotective benefits of GLP-1 agonists and SGLT-2 inhibitors don’t extend to the black population, and more as a call for dedicated trials to assess whether they do.

    • We couldn’t agree more with Dr. Mishriky’s conclusion that there is an “urgent need” for additional studies evaluating new diabetes therapies in the black population specifically. During subsequent Q&A, Dr. Mishriky issued an important reminder on the epidemiological front that black people with type 2 diabetes have an especially high need for therapies that reduce cardiovascular risk. This population is at higher risk for developing cardiovascular disease than the general diabetes population, and the black population in the US has experienced some of the most modest improvements in cardiovascular mortality over recent decades compared to other groups. To say nothing of population-level differences in socioeconomic status and healthcare access, there are also physiological differences that may disadvantage black individuals when it comes to diabetes outcomes – Dr. Mishriky pointed out that the sickle cell trait confers additional risk of CV and kidney disease, for instance. Dr. Mishriky pointed out that it is well understood that certain cardiovascular therapies such as ACE inhibitors are less effective in black individuals, and it would be extremely unfortunate from a public health perspective if the same was true for the next generation of GLP-1 agonists and SGLT-2 inhibitors. We hope this meta-analysis causes manufacturers to reconsider the designs of future CVOTs to more directly assess this important question.

    Diabetes Technology Highlights

    1. CITY Trial of Dexcom G5 CGM in Teens (14-25 Years) Shows 0.4% A1c Advantage, +1.7 Hours/Day in-range; CGM Group Still Spending 13 Hours >180 mg/dl

    Joslin’s Dr. Lori Laffel presented very positive results from the CITY trial (n=153), one of the very first interventions to demonstrate efficacy and safety of CGM vs. BGM use in type 1 adolescents (14-<25 years). The six-month, multicenter, randomized controlled trial found use of the Dexcom G5 conferred a 0.4% A1c advantage in favor of CGM over BGM (p=0.01; baseline: 8.9%). As Dr. Laffel pointed out, these results are particularly impressive, given that the general trajectory for adolescents is to see an increase in A1c. Moreover, more than twice as many in the CGM group as compared to the BGM group achieved an A1c reduction ≥0.5% (44% vs. 21%, p=0.005) and over four-times as many participants in the CGM group vs. the BGM group saw an A1c reduction of ≥1% (25% vs. 6%, p=0.003). There was also a non-significant greater proportion of those in the CGM group who achieved an A1c <7.5% as compared to the BGM group (18% vs. 11%, p=0.11).

    The CGM group also saw a 1.7 hour/day advantage vs. BGM on time-in-range (70-180 mg/dl) (p<0.001). Data from the CGM group were matched to time points when the BGM group wore blinded CGM (7 days at 13 weeks; 14 days at 24 weeks). Those in the CGM group increased time-in-range from 37% to 43% (+1.7 hours) vs. no meaningful change in the BGM group (36% to 35%). While the increased time in range is promising, Dr. Laffel emphasized that adolescents on CGM were still spending the majority of the day >180 mg/dL – clearly, there is much work to be done in helping this population. (We imagine that intuitive, minimal effort AID systems will be the real key, along with nutritional approaches that adolescents might be willing to do.) Time in hyperglycemia also saw improvements with adjusted differences of ~1.4 fewer hours/day (p=0.009) spent >180 and a notable ~1.2 fewer hours/day spent >300 mg/dl (p<0.001). Again, although the improvement is encouraging, Dr. Laffel pointed to the just-published international recommendations for spending <25% time above 180 mg/dl. In CITY, the CGM group declined from 58% to 54% time >180 mg/dl, meaning that even with the benefit of CGM, adolescents still spent over half the day in hyperglycemia. However, this was favorable compared with the BGM group that began the study at 59% time >180 mg/dl and after 6 months, time >180 mg/dl was 61%. Time spent in hypoglycemia (<70 mg/dl, <54 mg/dl) was minimal but still showed improvements for CGM: adjusted group differences were ~10 minutes/day (p=0.02) and ~6 minutes/day (p=0.002), respectively. CITY is another incredibly positive study demonstrating the (obvious) benefits of CGM over BGM in type 1 diabetes – even for such a challenging patient population. Still, the data are a reminder that much work is needed to reduce hyperglycemia in adolescents. Additionally, Dr. Laffel highlighted the importance of working with payers to ensure that all adolescents have access to CGM.

    • Dr. Laffel found it “remarkable” that over two-thirds of the CGM group were using CGM at least five days/week by the end of the six-month study – the highest CGM use observed for adolescents in a study to date. Study retention was “extraordinary,” which Dr. Laffel credited mainly to the staff. 96% of the CGM group and 90% of the BGM group completed follow-up. It is interesting to compare the rate of CGM use to the WISDM CGM participants (type 1s ≥60 years), the majority (81%) of whom used CGM seven days/week by six months. In CITY, just 41% used CGM seven days/week. Although elderly patients have been misconstrued to lack tech savviness, it’s clear that once trained they may stick to CGM at a higher rate than adolescents. Perhaps not surprisingly, CITY participants exhibited far greater use of the Dexcom G5 mobile app, with 81% using the app at six months; to compare, just 30% of WISDM participants did the same. Notably, 50% of CITY participants used the G5 mobile app with the Share feature turned on (only ~10% of WISDM participants did so). As Dr. Laffel noted, while there may be a tendency to assume adolescents don’t want to be involved with their family, the results suggest that with the proper training, adolescents may not mind sharing some of their data. The vast majority of participants (98%) used their CGM for insulin dosing by week 26; indeed, participants made use of this feature almost immediately, with 92% using their CGM for dosing at the second week.





    Adjusted Difference









    Time in Range (70-180 mg/dl)




    +1.7 Hours/Day




    Time >180 mg/dl




    -1.4 Hours/Day




    Time >300 mg/dl




    -1.2 Hours/Day




    Time <70 mg/dl




    -10 mins/day




    Time <54 mg/dl




    -6 mins/day




    Coefficient of Variation (CV)








    • Dr. Laffel highlighted the decrease in hypoglycemia (<70 mg/dl) for the CGM group, particularly overnight. She found it “reassuring” to see a corresponding increase in time-in-range (70-180 mg/dl) over 24 hours, again especially overnight, further supporting that the CGM group’s decline in hypoglycemia did not come at the expense of increased hyperglycemia. There were no significant differences in severe hypoglycemia or DKA events, and no device-related adverse events were reported. There was one related adverse event of dizziness due to blood draw, and one study withdrawal due to pregnancy.

    • Patient-reported outcomes showed “substantial benefits” of CGM compared to BGM including: (i) higher glucose monitoring satisfaction; (ii) greater perceived benefits of CGM and SHARE; (iii) lower perceived barriers of CGM and SHARE; and (iv) improved diabetes technology attitudes. There were no differences in diabetes distress or sleep quality, despite concerns that CGM may increase burden and disrupt sleep from alarms.

    • CITY participants from both groups are currently being followed for an additional six months using the Dexcom G6. Participants in the CGM group will be assessed for the durability of use and sustainability of glycemic benefits. It will also be interesting to see how the move from G5 to G6 improves clinical outcomes, and how much the BGM improves with the jump straight to G6. Those in the BGM group will be evaluated for glycemic benefits and preferences regarding CGM alerts/alarms.

    • Participants could not have used CGM in the past three months and were required to complete the run-in phase consisting of ≥200 CGM hours in at least eight days out of the 14-day period. 54% of participants overall used a pump, comprising 49% of the CGM group and 59% of the BGM group. All participants received initial training at screening for the baseline placement of masked CGM (Dexcom G4 Pro), and real-time CGM training was delivered at randomization, two-weeks, and six-weeks. Complexity of instruction increased with each visit, beginning with CGM basics and scaling to using CGM to minimize highs and lows. A CGM benefit handout was also provided at each visit.

    2. SENCE CGM Trial In 2-8 Year Olds: G5 CGM + Family Behavioral Intervention Improves TIR (+43 mins/day), Reduces Time <70 and >300; CGM Reduces Severe Hypoglycemia!

    Dr. Linda DiMeglio (Indiana University) presented positive time-in-range, hypoglycemia, and severe hypoglycemia results from the six-month SENCE study (n=143), testing use of the Dexcom G5 in 2-8 year olds. The 14-site trial randomized participants to three groups: CGM + a family behavioral intervention (FBI; n=50) vs. CGM-only (n=44) vs. BGM-only (n=49). FBI consisted of five visits (weeks 1, 3, 6 13, 19) with additional training on using and living with CGM, using CGM away from home, CGM burnout, and more – an excellent design to see the value of added structured education on top of the usual CGM training. Time-in-range (70-180 mg/dl) was low at baseline in all three groups – 38%, 41%, and 41% – consistent with the baseline A1c of 8.2% and a younger age group (median age: ~7 years old). The CGM groups used the Dexcom G5 non-adjunctively, with access to Dexcom Share.

    • By the end of six months, CGM+FBI drove a +1 hour/day improvement in time-in-range (+4%) vs. identical declines in the CGM-only and BGM groups (-1%, -12 minutes/day). The adjusted difference was +43 minutes/day in-range in favor of CGM+FBI, which just missed statistical significance (p=0.11, p=0.12). An expanded analysis only looking at 19-26 weeks – after the last FBI session – did show a statistically significant advantage on time-in-range, with CGM+FBI improving time-in-range by 1.5 hours from baseline compared to declines in the other groups.

    • A1c did not change significantly in any group from the 8.2% baseline – likely reflecting the improvement in hypoglycemia (see below) and extreme hyperglycemia, which averaged out to no A1c change.

    • Relative to BGM, both CGM groups saw significant improvements in time <70 mg/dl – a -1.5% reduction (-21 minutes/day) with CGM+FBI and a -2.4% reduction with CGM alone (-34 minutes/day) vs. a 1.2% increase in the BGM group (+17 minutes/day) (p<0.001). Both CGM groups nearly eliminated time <54 mg/dl (0.5%, or 6.5 minutes per day at six months), whereas the BGM group was still spending a full 2% of the day – 32 minutes – below 54 mg/dl (p<0.001). The hypoglycemia benefits were especially pronounced overnight – the BGM group had significant variability in overnight time <70mg/dl, whereas both CGM groups were consistently under 5% time <70 mg/dl across the entire day and night.

    • CGM also showed a significant benefit on severe hypoglycemia (requiring third party assistance): there were 0 severe hypoglycemia events in the CGM+FBI group, 1 event in the CGM only group, and five events in the BGM group (p=0.03 for CGM vs. BGM).

    • CGM+FBI also reduced time >180 and >300 mg/dl, covered in the table below. Of note, however, both CGM groups still saw more than three hours per day spent >300 mg/dl at six months – clearly automation and more education are needed to reduce pediatric hyperglycemia.

    • In line with the WISDM (Day #3) and CITY trials (highlight #1 above), CGM use was very high - ~90% of the CGM groups were wearing the Dexcom G5 for >6 days per week at the end of the six-month study. Non-adjunctive use was also high: Over 85% of the parents in the CGM groups were dosing insulin without fingersticks at six months. Just under half of families were using Dexcom Share at six months.

    • Dr. DiMeglio concluded that “These data have important clinical care implications,” given the link between time in hyperglycemia and neurocognitive development and function. There was not a lot of commentary on the added of CGM+FBI vs. CGM alone, though the former did seem to be more beneficial.


    + FBI

    -> 6 Mos


    -> 6 Mos


    -> 6 Mos

    P-Values for Adjusted Differences

    CGM+FBI vs. BGM
    CGM Only vs. BGM
    CGM+FBI vs. CGM Only


    38% -> 42%

    +1 hr/day

    41% -> 40%

    -12 mins/day

    41% -> 40%

    -12 mins/day


    Time >180

    57% -> 55%

    -30 mins/day

    53% -> 57%

    +1 hr/day

    54%-> 54%

    No Change


    Time >300

    21%-> 16%

    -1.2 hrs/day


    -12 mins

    15%-> 17%

    +30 mins


    Time <70

    4%-> 2%

    -21 mins/day

    4%-> 2%

    -34 mins/day

    4%-> 5%

    +17 mins/day


    Time <54

    1.5%-> 0.5%

    -15 mins/day

    1.4%-> 0.4%

    -14 mins/day

    1.3%-> 2.2%

    +13 mins/day


    • On the PRO front, CGM+FBI was superior to both groups on the Problem Areas in Diabetes (PAID) scale, a sign that the educational program has some psychosocial benefits. Other PROs – Diabetes Technology Questionnaire, Hypoglycemia Fear Worry Subscale – also favored BGM over CGM.

    • Baseline characteristics: median age was 6 years in the CGM groups and 7 years in the BGM group. Mean A1c was 8.2% in all three groups. Pump users were a minority in this study – 30% of the CGM groups and 45% of the BGM group. ~11% of the CGM groups and 14% of the BGM group had previously used CGM prior to the study. More than one-third of study participants did not have private health insurance, and the majority of parents did not have a bachelor’s degree or higher. One third of participants were came from minority populations.    

    • Participants wore a masked Dexcom G4 Pro (Software 505 algorithm) during screening, and the BGM group wore the G4 Pro at four time points throughout the study.

    3. Dr. Rich Bergenstal’s Personalized A1c Based on Red Blood Cell (RBC) Turnover; Incorporation of Mean RBC Turnover Can Adjust A1c by 0.5%!

    IDC’s Dr. Rich Bergenstal described how A1c depends on red blood cell (RBC) lifespan, explaining that A1c may be misrepresented by varying rates of RBC turnover. The higher the RBC turnover, the shorter the RBC life, lowering the length of RBC exposure to glucose and ultimately artificially depressing A1c. However, glucose exposure remains exactly the same in other cells, meaning that the A1c measurement is not accurately describing the degree of glucose exposure that other cells sensitive to hyperglycemia (eye, kidney and nerve)  experience. To determine the individualized RBC turnover rate, Dr. Bergenstal’s team modeled the prospective A1c according to baseline A1c and glucose history derived from dense CGM data. The RBC turnover rate can then be calculated such that it aligns each individual’s CGM and A1c measurements. Analysis was performed on 58 individuals with diabetes and required a minimum six-month CGM dataset with 80% coverage and no gaps >12 hours. A1c was tested at 0, 3, and 6 months. RBC turnover rates were found to differ substantially among patients, with a mean rate of 1.18%/day. 70% of individualized A1c values differed from their lab A1c by over 0.5% when the RBC turnover was adjusted to 1.18%/day. As Dr. Bergenstal pointed out, such misrepresentation can be dangerous, with overestimation causing exposure to hypoglycemia by unwarranted escalation of anti-diabetic therapies, and underestimation leading to undertreatment and exposure to higher risk of complications. Dr Bergenstal pointed out that this kinetic model for calculating the individualized or personalized A1c takes into account mainly the calculated RBC lifespan but also additional factors regarding the calculated rate of transport of glucose into the RBC (which in his 10 minutes he did not have time to elaborate on). As long as A1c is still being used on an individual level (to which one audience member aptly questioned: “why not just kill A1c all together”) Dr. Bergenstal proposed that an individualized A1c be implemented, personalized by the patient’s RBC turnover.

    4. Omnipod Dash Product Theater: Real-World Data from ADA Posters – T1 Mean GMI=7.8%, T2 Mean GMI=7.5%; User Base by Age; Dr. Forlenza’s Excitement over Horizon (Can’t Wait to Deploy in Camp)

    During an Insulet product theater, Medical Director Dr. Trang Ly highlighted real-world Omnipod outcomes data from Glooko, separated by diabetes type and age. Among the 1,427 adult Omnipod users (1042-P) who self-identified as type 2s, the average GMI (glucose management indicator; formerly “eA1c”) was 7.5%, and 75% of them have A1c <8%. Said Dr. Ly, “I think most adult endocrinologists would be pretty happy having a clinic where three of four patients have an A1c under 8%.” Among the 20,243 Omnipod users with type 1 diabetes (1043-P and 1357-P), mean GMI is 7.8%, and 63% have A1c <8%. Insulet intentionally plotted its users’ data on the same axis as the infamous T1D Exchange Registry A1c plot. Based on a direct comparison, which has some obvious limitations (e.g., the type 1 Omnipod users not in this analysis are those who do not download), it would appear that Omnipod users may have lower A1cs (particularly in the adolescent ‘hump’) than the rest of the population. Dr. Ly also shared some granular data about the age of Insulet’s base in this data set: 28%-30% are under 18 years old, 25% are >50 years old, and 14% of adults on Omnipod are type 2s (most are in the 50-70 age range). Before handing the podium over to Dr. Forlenza, Dr. Ly closed with an overview of the Dash portfolio: Insulet is busy filling the “thousands of pre-orders” for Dash in the just-started full market release, patients are “flying through training” for the product, Display and View secondary display apps are “coming soon,” the Lilly-partnered U500 pump is under FDA review, Dash was the first pump partner of Tidepool Loop, and the Horizon AID system will launch with the option of direct phone control (only on Samsung devices at first).

    • BDC’s Dr. Greg Forlenza, who has been running studies and working with Insulet on Omnipod Horizon for the past four years, reviewed the body of evidence to date – see yesterday’s IDE3 toddler study writeup for a tabular summary. Dr. Forlenza confirmed that the pivotal trial will take place later this year (4Q19, per Insulet’s last update). Throughout the talk, his excitement regarding the system was palpable (and at times explicit). He specifically called out the system’s stellar overnight control (“I can’t wait to deploy this at diabetes camps”), ability to handle missed meal boluses (“Those of us that take care of adolescents and young adults, we know they’re going to do this experiment in real life at least a few times a week”), and ability to handle exercise (“I get so excited about these results every time I show them. Every single patient was 0% hypoglycemia overnight”).

    5. Biochemical Hypo from Blinded and Real-Time CGM Effectively Predicts Subsequent Severe Hypo Events; No Surprise, but Potential Risk Flagging Mechanism?

    Through a retrospective, pooled analysis of DIaMonD (CGM in type 1 MDIs) and HypoDE (CGM in type 1 hypo unaware MDIs) data, Prof. Norbert Hermanns et al. showed that hypoglycemia metrics from real-time CGM, masked CGM, and SMBG predict future severe hypoglycemia events with high sensitivity and specificity. These biochemical markers of future severe hypo risk could be useful for risk stratification and human/automated intervention, though we suspect that more analyses will be needed to verify Prof. Hermann’s proposed risk cutoffs (see tables below), possibly even in a prospective study. In addition, the findings are not applicable beyond the study populations, including type 2s and type 1s not on MDI. At the end of the day, the concept of “hypo begets hypo” holds true, and it’s not all that surprising that those with greater exposure to hypoglycemia are at higher risk of future assistance-requiring events.

    • The researchers first interrogated whether hypoglycemia metrics in the period of blinded baseline CGM wear (two weeks in DIaMonD, four weeks in HypoDE) could predict severe hypoglycemia in the first two weeks of the control arm (i.e., no CGM use). Unsurprisingly, yes: percent time <70 mg/dl and <54 mg/dl, number of events <70 mg/dl and <54 mg/dl, and LBGI all predicted future severe hypoglycemia events (AUC ROCs of 0.71-0.75). Coefficient of variation, however, narrowly missed significance.

      • See the following table for Prof. Hermanns’ recommendations for threshold values at which one should suspect future severe hypoglycemia based on blinded CGM– we weren’t sure how these were decided. The sensitivity and specificity of the cut-off is detailed in the right-most columns.





    % <70 mg/dl




    % <54 mg/dl




    # events <70 mg/dl




    # events <54 mg/dl








    Glycemic variability (CV)




    • Biochemical hypoglycemia in the first two weeks of the HypoDE treatment period (real-time CGM wear) also successfully predicted severe hypoglycemia frequency in the subsequent two weeks. Percent time <70 mg/dl, number of events <70 mg/dl and <54 mg/dl, LBGI, and coefficient of variation all predicted future severe hypoglycemia events (AUC ROCs of 0.69-0.76). Percent time <54 mg/dl missed significance, which we find very strange; we would’ve expected this metric to be among the best predictors.

      • See the recommended cut-off values below for real-time CGM. Notice that the risk cut-offs with real-time CGM are far more stringent than those for masked CGM (above). For example, if someone is on a blinded CGM, one should be concerned about subsequent severe hypoglycemia when they spend ≥7% below 70 mg/dl, but when they are on real-time CGM, one would be concerned when they spend ≥3% below 70 mg/dl. (The latter aligns well with the consensus recommendation to have <4% of values <70 mg/dl.) This is presumably because the comparison is not apples to apples. In the latter condition, researchers were using real-time data to predict severe hypo events during real-time CGM use – hypoglycemia will naturally be lower in both conditions, so more than ≥3% connotes risk for a severe event. In the former scenario, blinded CGM is used to predict events for patients on SMBG – biochemical hypoglycemia is presumably far greater in this condition.





    % <70 mg/dl




    % <54 mg/dl




    # events <70 mg/dl




    # events <54 mg/dl








    Glycemic variability (CV)




    • SMBG-measured hypoglycemia – percent of readings <70 mg/dl and <54 mg/dl, LBGI, and coefficient of variation – also successfully predicted future severe hypoglycemia with AUC ROCs between 0.71-0.78.

    6. Randomized Crossover Trial (n=17) Finds FreeStyle Libre Accuracy to Decline During Hypoglycemia and Acute Exercise; MARD of 29.8% During Exercise

    Medical University of Graz’s Dr. Othmar Moser presented results from a small (n=17 type 1s) secondary outcome analysis of a single-center, randomized, open-label crossover trial showing reduced FreeStyle Libre accuracy during hypoglycemia and acute exercise. Participants engaged in four, 23-hour inpatient phases consisting of glycemic challenges (standardized meals) and a 45-minute exercise session. Venous blood samples and capillary samples from the fingertip and earlobe were taken to compare against FreeStyle Libre readings. Overall MARD was 12.9% (n=1,114); in hypoglycemia, MARD increased to 31.6% (n=79) and in hyperglycemia, MARD declined to 8.7% (n=331). During exercise, MARD climbed to 29.8% (n=475) reaching 45.1% during hypoglycemia (n=70). Dr. Moser highlighted “quite impressive” overnight MARD of 8.5% (n=282). As one audience member pointed out, some of these sample sizes are quite low, given that MARD is recommended to be calculated with at least 2,000 glucose pairs. While Dr. Moser acknowledged that the sample size is not ideal, he countered that previous exercise studies evaluating “all the devices” use the same method and that this study could thus be compared to existing literature. Dr. Moser concluded that FreeStyle Libre performs well during low to moderate rates of glucose change, but exhibits weaker accuracy around hypoglycemia and acute exercise. Indeed, during exercise, the rate of change in glucose was 2.7 mg/dl/min, as compared to the nocturnal rate of change, found to be just 0.12 mg/dl/min. Of course, this is true of all CGMs – not just FreeStyle Libre. Dr. Moser suggested adjunctive blood glucose measurements may be helpful during periods of high glucose rate of change.

    • The Clarke Error Grid analysis revealed substantial differences in accuracy during exercise. The overall analysis found 66% of readings to fall within Zone A, while only 26% of readings fell in Zone A during exercise. Alarmingly, during exercise 52% of readings fell within Zone B and 22% of readings fell within Zone E.

    7. UK-Wide Audit of 4,709 Type 1 Patients on FreeStyle Libre Find 0.6% A1c Reduction (Baseline 8.2%) At Six Months; Declines in Hospital Admissions for Hyperglycemia/DKA and Hypoglycemia

    Dr. Harshal Deshmukj presented results from the Association of British Clinical Diabetologists’ UK-wide audit of the FreeStyle Libre, which included data from 4,709 type 1 patients across 114 centers. The data revealed a significant median 0.6% A1c reduction (median baseline A1c: 8.2%) at six months. Not surprisingly, those with a baseline A1c >8.9% achieved a larger median A1c reduction with FreeStyle Libre of 0.8% (median baseline A1c: 9.4%). Notably, hyperglycemia and DKA admission dropped from 7.3% to 1.9% over the study period, as did hypoglycemia-related hospital admission, which declined from 2.7% to 0.5%. 79% of patients reported that FreeStyle Libre use enabled reduction of time in hypoglycemia, and 31% of patients reduced their rate of hypoglycemia. 39% of patients reduced their nocturnal hypoglycemia. Impressively, 9% of patients reported reversal of hypoglycemia unawareness as measured by the Gold Questionnaire. The reduction in hypoglycemia unawareness is particularly notable given that FreeStyle Libre does not include alarms – i.e., the richness of real-time CGM data and patterns might help patients regain awareness. Diabetes distress score decreased significantly and just 1% of patients discontinued FreeStyle Libre at six months, mostly due to the lack of reimbursement. Dr. Deshmukj also provided a preliminary analysis to discern predictors for an A1c response: baseline BMI, A1c, and age topped the list, followed by scans/day, structured training, and gender.

    • We were interested to see data showing reasons for FreeStyle Libre use. 30% of users selected the replacement of SMBG as reason for starting FreeStyle Libre, while 25% selected having a high A1c. 17% indicated frequent hypoglycemia and 16% fear of hypoglycemia. Patients were able to select multiple reasons, hence the sum to more than 100%.

    Exhibit Hall

    Diabetes Therapy


    Allergan’s booth was notable for its “NASH lab”, a series of NASH gross anatomy and histopathology stations aimed at improving provider understanding of fatty liver disease and its stages on a cell and organ level. As a reminder, Allergan has cenicriviroc, a CCR2/CCR5 inhibitor, in phase 3 for NASH – the AURORA trial in F2-F3 fibrosis due to NASH is still recruiting. Primary completion is anticipated in October 2021, but topline results to potentially support regulatory submission should come in 2020. Allergan also has a phase 2 combination study of CVC ongoing with Novartis’ FXR agonist tropifexor. Read more about Allergan’s diabetes-related pipeline here.


    Amarin made its first appearance at an ADA with a well-attended booth in the exhibit hall, mirroring its continually increasing presence in the diabetes space following the emergence of highly-positive REDUCE-IT CVOT results for Vascepa. Since those results, we’ve also seen booths from Amarin at AHA, ACC, and ENDO, with considerable amounts of interest at each meeting. As a reminder, REDUCE-IT enrolled nearly 60% of its trial population as those with type 2 diabetes, and we see enormous potential for Vascepa to make a difference for those with diabetes and CV risk. The ADA apparently does as well – it made the surprising but deserved move of updating its 2019 Standards of Care to recommend that Vascepa be considered for patients with diabetes and ASCVD or other CV risk factors (i.e., primary or secondary prevention), who are on a statin but have high triglycerides (125-499 mg/dl). Even more encouragingly, FDA just granted priority review to Amarin regarding its sNDA requesting a CV indication for Vascepa. Sales representatives were not able to comment on either of these pieces of positive news, and could not give further detail on how HCPs are reacting to the novel data seen in REDUCE-IT. Nonetheless, seeing how large the diabetes cohort was in the REDUCE-IT trial and the ever-growing importance of CV risk reduction in diabetes care, we imagine that interest from endocrinologists will grow.


    Amgen’s sizable, front-and-center booth was dedicated entirely to PCSK9 inhibitor Repatha (evolocumab) and its CV indication for the prevention of heart attack, stroke, and coronary revascularization in patients with established CV disease. Throughout panels describing vignettes of patients who fit the clinical profile for Repatha, and the drug’s clinical data on LDL-lowering and cardioprotective benefits,  a recurring message was “lower together” – reflecting Amgen’s goal of lowering LDL levels and cardiovascular risk for patients, as well as hassle and cost. On cost, a prominently-placed panel highlighted Repatha’s 60% reduced price for Medicare Part D and commercial insurance, bringing the average monthly list price to $450 – ostensibly on par with Victoza, Jardiance, and Invokana. In addition to its usual espresso bar, we were extremely pleased to partake in Amgen’s A1c testing and cholesterol testing offerings. Judging by the foot traffic, this was certainly a hit among ADA exhibit hall-goers.


    AZ occupied a sizable booth, front and center to the exhibit hall entrance. The forward-facing half was dedicated to SGLT-2 inhibitor Farxiga, GLP-1 agonist and autoinjector Bydureon BCise, and antiplatelet Brilinta. A wall-sized screen flashed data readouts for all three agents, interspersed with motion blur photos taken at a booth in the corner of AZ’s space. These photos have become a staple since ENDO 2019 and AACE 2019, playing into the theme of early initiation and inertia AZ has keyed into for Farxiga especially. The back side of the booth was dedicated to general education about the connections within cardio-renal-metabolic disease. Three touchscreen monitors were available to passers-by to dive into the pathophysiology of these diseases, which we take as part of the company’s preparation for a potential CV indication for Farxiga based on positive results from the DECLARE CVOT (currently under review at EMA and FDA). We wonder whether advertising will at all pivot toward heart failure – the component of the positive co-primary outcome that drove superiority in DECLARE – seeing as Farxiga is poised to become the first diabetes drug with this indication.


    Bayer’s presence at ADA 2019 – including both this exhibit and an informational product theater – were entirely focused on DKD, as in the past. The company is setting the stage for oral finerenone, its phase 3 mineralcorticoid receptor antagonist currently under investigation in the FIGARO-DKD (n=7,437) and FIDELIO-DKD (n=5,734) trials, which are expected to complete in July 2021 and May 2020, respectively. As Bayer can’t yet promote the candidate, its booth featured educational messaging on the connection between diabetes-related kidney disease and CV risk and noted that patients with DKD are almost three times as likely to die from cardiovascular causes than those with type 2 alone. We’re glad to see Bayer continue to display interest in DKD, given the recent success of SGLT-2 inhibitors have been balanced against mixed results and discontinuations for other candidates (namely, AbbVie’s atrasentan). See our nephropathy competitive landscape for an overview of other candidates in development.


    BI and Lilly hosted one of the more intriguing (and largest) exhibit hall setups at ADA (similar to its booth at ACC 2019) – visitors could climb to the second story of the booth and use virtual reality headsets to go on tours through various scenic spots in San Francisco (and New Orleans). Facts about Jardiance’s effect on CV outcomes were peppered throughout these virtual tours. As always, promotional stands prominently featured the 38% relative risk reduction on CV death achieved by Jardiance in the EMPA-REG OUTCOME study. We also saw informational exhibits for Trulicity, Basaglar, and Lilly’s entire suite of diabetes products. We’re glad to see this dedicated presence in the exhibit hall from BI/Lilly, especially after Lilly has somewhat scaled back its own exhibit hall presence at diabetes conferences over the past year or so.


    Genfit’s small booth focused on the company’s efforts in NASH, including its work to improve NASH patient care and education. Indeed, the company’s NASH awareness campaign (begun in France in March 2017) led to the launch of a US NASH Education Program, which seeks to improve medical education around the disease while also raising general awareness among patients, professionals, and families. The company’s leading NASH candidate, phase 3 elafibranor (a dual PPAR alpha/gamma agonist), is currently under investigation in the RESOLVE-IT trial (n=2,000) for F1-F3 fibrosis due to NASH. While full primary completion is expected in December 2021, an interim analysis – that could potentially support approval – is anticipated in 2019. Additionally, Genfit’s booth advertised NIS4, the company’s 4-biomarker-based blood test algorithm to identify NASH and fibrosis, licensed to LabCorp in January 2019. Genfit plans to submit to FDA for marketing authorization in 2020.


    Gubra hosted a small but well-staffed booth, featuring materials promoting its CRO services for preclinical work in diabetes-related CV and renal disease. Notably, Gubra is involved in a number of collaborations that weren’t promoted in its ADA booth (likely due to their early-stage nature).

    In June 2018, Gubra joined a partnership with JDRF to develop a new glucose-responsive insulin (GRI). Additionally, just last month Gubra expanded its partnership with BI, to develop peptide poly-agonists for the treatment of obesity. This is Gubra and BI’s second collaboration, following a  partnership penned in November 2017 for the development of peptide compounds that regulate food intake.


    Along with regular promotional materials for SGLT-2 inhibitor Invokana, Janssen’s booth had a sizable section dedicated to education on the overlap between CKD and type 2 (slogan: “How stable is a future with CKD and type 2 diabetes?”). Of course, this outsized promotion of CKD in type 2 makes sense in light of landmark CREDENCE results showing Invokana’s impressive benefit in reducing renal risk. Although sales representatives were not able to comment on Invokana’s potential renal indication (a possible first among diabetes drugs!) or FDA’s recent decision to grant priority review to review such an indication, we believe this compound will receive these indications.


    Merck boasted a commanding booth at this year’s ADA, with one wing dedicated to DPP-4 inhibitor Januvia (sitagliptin) and the other to Pfizer-partnered SGLT-2 inhibitor Steglatro (ertugliflozin) – this was by far the largest Merck booth and most notable promotion of Steglatro that we’ve seen. The booth had an open floor plan, interspersed with touch screens highlighting recent clinical data for these two products, plus patient vignettes to illustrate the clinical profiles that best fit these different therapies. Touch screens for Januvia were focused on the recently-completed CompoSIT-R and CompSIT-I trials, which investigated Januvia (i) in patients with mild renal impairment vs. Farxiga and (ii) in patients initiating insulin. Both studies were positive: Januvia provided superior A1c reductions to Farxiga in the former, which makes sense given the glucose-lowering mechanism of SGLT-2 inhibitors is strongly dependent on renal function (of course, SGLT-2s offer cardioprotection and renal-protection that DPP-4 inhibitors do not). On Steglatro, results supported the continued use of Januvia as a patient adds basal insulin to their diabetes treatment regimen.

    Novo Nordisk

    Novo Nordisk brought a small city to ADA with two booths: a larger one at the entrance to the exhibit hall advertising its entire suite of novel diabetes therapies and technology as well as a smaller one focused on promoting discussion around GLP-1 use in type 2 diabetes. The former was sectioned into six areas:

    • For next-gen basal insulin Tresiba, an augmented reality game with iPads whereby users pointed the camera at a swimming pool model and guided their racer down the lane by answering questions about the DEVOTE CVOT: for example, its size (n=7,637), significant 40% relative risk reduction in overall rate of hypoglycemia vs. Lantus, non-inferiority on MACE, and improved glycemic stability;

    • For Ozempic, an ~10 foot tall billboard with results from SUSTAIN 7, demonstrating superiority over Lilly’s Trulicity on A1c (~0.4%) and weight (~5 lbs) over 40 weeks – we note Novo Nordisk was also keen on promoting these results at ADA 2018;

    • For NovoLog, a small demonstration of the NFC-enabled NovoPen Echo Plus and its ability to upload data to Glooko/Diasend;

    • A one-person booth on Cornerstones4Care, Novo Nordisk’s jointly-branded digital tool with Glooko offering personalized diabetes support, including automated blood sugar and exercise tracking, trend recognition, charts and graphs, reminders, food and medication databases, and educational information on diet, exercise, and lifestyle choices;

    • Another one-person booth on NovoCare, the company’s patient access program (we love this focus on access and affordability at an exhibit hall booth – an all too rare occurrence);

    • And last but not least, a 1 km stationary biking challenge put together by Team Novo Nordisk. We arrived just in time to see one person take the top spot on the leader board with a time of 1:34! Let us know if you beat them.

    The smaller booth advertised Novo Nordisk’s new website,, which is dedicated to explaining GLP-1 pathophysiology in type 2 diabetes and treatment options, for both patients and HCPs.


    We were pleased to see JDRF’s Network for Pancreatic Organ Donors with Diabetes (nPOD) on the scene at this year’s ADA exhibit hall. The small booth featured promotional materials describing nPOD’s work collecting donor pancreas/islet cells from the existing National Organ Donor Registry. This fills an important unmet need in the type 1 diabetes research arena, given the shortage of donor islet cells and the reality that data collected from animal models is often not applicable to humans. Over a decade old, the nPOD initiative has a long history of facilitating cutting-edge work from these tissue samples. As we learned at February’s nPOD meeting, the organization now has 238 projects in 21 different countries, has collected over 50,000 samples, has created over 24,000 histology images that are available online, and has directly supported over 200 publications – with these statistics, nPOD founder the legendary Dr. Mark Atkinson has argued that nPOD is the biggest type 1 diabetes project in the world. See our report from nPOD’s 2019 annual meeting for more on this inspiring project.


    At 8,000 square feet, Sanofi’s booth dominated the exhibit hall. Thematically, the product promotion was unified by its “Your Type” campaign, comprised of four pillars – medicine, education, integrated care, and access – geared to “support patients by offering personalized care options for everyone living with diabetes.” However, what impressed us most what the booth’s emphasis on education, patient support, and affordability programs. This included prominent advertisement of Sanofi’s Insulins Valyou Savings Program and copay card offers. We noted that the Valyou program was advertised as “for patients without insurance” – we’ll be interested to see if Sanofi releases data on who utilizes this program, as patients with high insulin needs and/or on high-deductible plans may find this to be the cheapest option for them. That said, reps emphasized that, for the majority of patients, $99/month exceeds what their monthly supply of insulin would cost otherwise. 

    A front corner of Sanofi’s booth featured the company’s beyond A1c and type 1 focused campaign, which debuted at AACE 2018 and was revived there in April. The corner calls for attendees to “join the movement” and underscored its mission with key stats: 60% of adult type 1s have overweight or obesity, the average type 1 has 2 episodes of symptomatic hypoglycemia each week, and even patients meeting A1c goals can spend up to 9 hours out of range each day. Moreover, ~75% of adult type 1s aren’t achieving an A1c <7% – and that number actually seems to be rising. To be sure, this booth aims to familiarize HCPs with the outcomes beyond A1c where SGLT inhibitors are very useful. However, Sanofi/Lexicon also recently received a CRL from FDA in response to their NDA for sotagliflozin in type 1 diabetes. On the other hand, the novel candidate has been approved for type 1 in Europe.


    TrialNet had an informational booth set up in the exhibit hall. We sensed excitement, seeing how it was a huge ADA for TrialNet – it’s phase 2 sponsored trial of teplizumab made a splash with positive results, making it the first trial to meet a clinical endpoint of delaying type 1 diagnosis. TrialNet plans to further study the candidate and follow-up with study participants (regardless of whether they were or were not diagnosed with diabetes in the trial – we love to see this commitment to longer term follow-up).


    Xeris brought its largest booth ever (that we know of) to ADA 2019, likely in anticipation of the mid-conference, June 10 PDUFA date for liquid-stable glucagon autoinjector, Gvoke HypoPen. However, just five days before a decision was to supposed come, Xeris announced that FDA extended the PDUFA date by three months, to September 10, following submission of requested data that constituted a Major Amendment. However, CEO Mr. Paul Edick exuded optimism at the booth, as he did in the conference call held to discuss the delay, reiterating that the last 8-10 products to have received such a delay were approved. While reps will not be able to roll out their sales campaign at ADA, it was apparent from the size of the booth and team that the company will be all-in once approval is achieved.

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