IDF (International Diabetes Federation) Congress

December 4-8, 2017; Abu Dhabi, UAE; Days #3-4 Highlights – Draft

Executive Highlights

Greetings from Abu Dhabi, where our team is spending this week for IDF 2017. This report details top highlights from days #3-4 of the meeting, organized into four sections: (i) diabetes therapy, (ii) hypoglycemia, (iii) diabetes tech, and (iv) big picture topics.

Today we learned of yet another diabetes partnership for IBM Watson Health – the company and Sanofi delivered four new analyses based on real-world data and advanced analytics. We go in-depth on two of these below: What factors predict success on basal insulin/GLP-1 combination therapy? What are the chances that someone will reach goal on basal insulin if he/she hasn’t after six months? We also enclose coverage of a new post-hoc analysis of EXSCEL, which hit superiority when excluding placebo-treated patients with drop-in SGLT-2 inhibitors or SUs. A new meta-analysis strongly implied a cardioprotective class effect for GLP-1 agonists, and we attended brilliant talks by Drs. Rury Holman and Jaakko Tuomilehto on prediabetes as a CV risk state.

You’ll notice we include a special “hypoglycemia” category in this report. It’s not every day that we’re able to pull out five substantive highlights on this topic – kudos to IDF for elevating the dialogue around outcomes beyond A1c like hypoglycemia. The major development was news from EMA that new diabetes drug guidance will be published in January, including updated hypoglycemia definitions in line with international consensus. Also on Dr. Bart Van der Schueren’s panel, Kelly and Prof. Simon Heller engaged in a frank Q&A on the most pressing questions in the beyond-A1c movement. Download Kelly’s slides here and her talking points here.

On the tech side, short-but-sweet: An encouraging in-patient study of adjunctive dapagliflozin in fully closed loop, and an MDI sub-analysis from Abbott’s IMPACT outcomes trial of FreeStyle Libre in type 1s.

Last but not least, we heard from the incoming and outgoing IDF presidents on the 8th edition of the Atlas and on the need to address diabetes in underdeveloped regions globally. Novo Nordisk and IDF also announced interim results from their survey on CV risk awareness among diabetes patients worldwide. Data collection will continue through March 2018, and we look forward to full results at next year’s EASD gathering in Berlin.

We’ll be back after the weekend with one more installment of IDF highlights, after a long flight home to San Francisco! Here’s a great reading list until we’re back:

IDF 2017 Day #1 Highlights (Dr. Verma on SGLT-2s for heart failure/first-line; Dr. Polonsky’s takeaways from EMOTION study)

IDF 2017 Day #2 Highlights (Eversense vs. G5 vs. Libre in Bionic Pancreas study; Adam on CGM, clever insulin delivery, coaching/remote care; New data from DISCOVER and DiRECT)

IDF 2017 Conference Preview

Table of Contents 

Diabetes Therapy Highlights

1. Sanofi + IBM Watson Health Present Four Studies Using Real-World Data to Answer Clinical Qs; Predictors of Success with Insulin+GLP-1 Combo? Chances of Basal Insulin Success After 12 Months?

Sanofi and IBM Watson Health presented four (!) studies at IDF demonstrating the power of real world data and advanced analytics to answer meaningful clinical questions. In one such poster, researchers predicted A1c outcomes with basal insulin and GLP-1 agonists, and in another, evaluated the probability of bringing A1c below 7% in type 2 patients using basal insulin. (Sanofi sent out a press release on the latter). Both are described in more detail below, and really highlight the insanely big data sets IBM Watson can tap into. Big data mining is the first step toward predictive analytics and importantly, clinical decision support. Some of the findings were both valuable and immediately applicable to clinical practice: (i) historical trends in factors such as A1c, systolic blood pressure, and BMI, as well as seemingly-extraneous variables like residential population density predict treatment success in type 2s starting a free-dose combination of basal insulin + GLP-1 agonist; and (ii) individuals not reaching target on basal insulin within 12 months of initiation are unlikely to do so during the remainder of the two-year study. We hadn’t realized IBM Watson and Sanofi were partnered – this could easily have fallen into the Sanofi/Verily (Onduo) partnership, given the machine learning applications. We’re glad to see Sanofi investing more in digital and intrigued already by this first series of analyses – Watson has access to millions of patient medical records, so it can conduct large-scale cohort studies to better understand how treatments work and how they can be best employed...in record time and at a remarkably low cost. Notably, these are exactly the types of analyses that we expected from the Novo Nordisk-IBM Watson Health partnership announced two years ago. News on that front has been absent, but presumably the two are moving forward with internal projects, and we believe Glooko, Novo Nordisk and IBM Watson Health will ultimately work together in bringing intelligent digital solutions to market.

  • (P-0488) “Using advanced machine learning to predict HbA1c control with basal insulin and glucagon-like peptide-1 receptor agonist.” In this busy poster, lead author Dr. Kyu Rhee (IBM Chief Health Officer) and colleagues used advanced machine learning techniques on the IBM Explorys dataset (n=~108,000 T2D patients on insulin+GLP-1 combo therapy from EMRs of 39 major integrated care systems in US) to identify clinical and nonclinical factors affecting response to basal insulin+GLP-1 combo therapy. 3,300 of the 7,600 patients (43%) included in the study with valid A1c results achieved “successful A1c control,” defined as A1c ≤7% or ≥1% drop in A1c over at least a 180-day continuous treatment without DKA or hyperosmolar syndrome. Pictured below are the top 20 predictive factors of success after intensification to the combo therapy, and the authors singled out a few: (i) A trend of decreasing systolic blood pressure in the year prior to the start of the therapy; (ii) Decreasing BMI or a high standard deviation of BMI; (iii) Increasing A1c in the year prior to treatment; (iv) Patients in higher population density areas saw greater success; and (v) Patients whose last treatment was ≤50 days prior to combo therapy saw greater success. Interestingly, variability of body surface area and body weight in the year prior to the start of the period are the strongest predictors of success – the directionality is not specified, i.e., is low standard deviation positively correlated with success, or high? We imagine the latter, given the BMI trend noted above – though neither were discussed in the poster. The authors were careful to point out that this was a retrospective, population-level study that can’t be validated at the individual level, and called for future studies using large real-world clinical datasets with long term data to explore additional factors. There is clearly much work to be done, but we love the data driven approach to precision care – and the fact that all of the inputs are already in the EMR means that providers wouldn’t have to take on more work to reap the benefits should algorithms be implemented. The vision here is very compelling: IBM Watson could recommend a treatment at the point-of-care, given a patient’s recent history and what it knows about other patients. This is what Watson does in oncology and we see very obvious, high potential value in type 2 diabetes prescribing. Moving forward, we’d like to see other medicines included in the analysis, and hopefully outcomes beyond A1c one day (if they ever make it into EMRs). What if the patient and provider could together choose the suite of outcomes that matters to them, and then have the algorithm churn out the best drug to help that patient meet her goals? Clearly, we’re excited about the possibilities…
    • We are dying to know how much this analysis costs to perform. It seems like a no-brainer to use cognitive computing to get to the heart of what treatment is best for what patient and when, and at the advantage of not having to enroll a single patient or pay for a single drug. We also wonder what the partnership between Sanofi and IBM Watson looks like – how does it compare to the Novo Nordisk and Medtronic partnerships? Is the Sanofi iteration more of an R&D effort, while the others are more commercial?
    • The authors emphasize that the results can’t be validated at the individual level, but we wonder just how predictive they are. If a patient who had decreasing systolic blood pressure, decreasing BMI, and lives in an area of high population density, what is the probability she will respond well to basal insulin+GLP-1 therapy? What if she lives out in the countryside (low population density), but has all the other things in place – what is the probability that she will respond now?

  • (P-0489) “Real-world observational study to evaluate probability of achieving glycaemic control in patients with type 2 diabetes.” Ochsner’s Medical Center’s Dr. Lawrence Blonde et al. analyzed clinical EMR data from 5,936 type 2 diabetes patients in the IBM Explorys dataset – each had intensified treatment from an oral to a first basal insulin prescription with or without oral continuation. They assessed time to reach “glycaemic control” (A1c ≤7%) and percentage of patients reaching A1c ≤7%. From a baseline of mean ≥9.2% (depending on medication subcohort; a high baseline probably reflecting clinical inertia), ~20% of patients had reached goal after six months of treatment; at 12 months, ~29% had reached goal. After six months, the probability that patients would achieve target within the next three-month period dropped considerably: 10.9% in months 6-9, 5.7% in months 9-12, and just 1.9%-3.5% over the next year (see below). The results are not limitation-free, but taken as is, there are two main, potentially antagonistic, takeaways: (i) The data suggest that, if target A1c hasn’t been met as early as 6-12 months after basal insulin initiation, alternate therapeutic strategies should be investigated, or (ii) insulin doses are not being up-titrated aggressively enough. To the first point, there are now fixed-ratio combination therapies that show much greater efficacy and ease of use with less hypoglycemia and less weight gain, and to the second point, the data remind us that insulin is still an incredibly difficult drug to dose safely and starting conservative makes a lot of sense – assuming the doses are quickly and continuously updated if glucose does not get to target. We see tremendous potential for basal insulin titration software to empower physicians and patients, make them feel more secure using insulin, and thereby reduce clinical inertia. We would also love to know which insulins were used by the patients in the study – is goal achievement more likely with newer insulins?

2. Post-Hoc Analysis of EXSCEL Points to Drop-In Medications in Placebo Arm (SUs, SGLT-2 Inhibitors) as Possibly Reducing Detectable Effect Size for Three-Point MACE, Leading to Narrow Miss on Superiority

EXSCEL investigator Dr. Angelyn Bethel presented new analyses excluding drop-in medication use in the placebo group. When SGLT-2 inhibitors, SUs, or “any medication” were excluded, AZ’s GLP-1 agonist Bydureon (exenatide once-weekly) gave a significant reduction in three-point MACE. Session chair Dr. J Hans DeVries reminded the audience that EXSCEL reported neutral CV results at EASD, although the p-value was “marginal” for CV superiority (HR=0.91, 95% CI: 0.83-1.00, p=0.06 for superiority). The new results announced at IDF indicate that concomitant medication use in the placebo arm of EXSCEL – which was at the discretion of the usual care provider, and only prohibited GLP-1 agonists – may have had small effects on risk, theoretically reducing the detectable effect size. Indeed, metformin, SUs, insulin, DPP-4 inhibitors, SGLT-2 inhibitors, and GLP-1 agonists all had at least 5% drop-in rates in the placebo group. Analyses censoring these patients at time of drop-in were used to estimate new hazard ratios and 95% confidence intervals. When censoring sulfonylurea use out of the placebo arm, the hazard ratio point estimate for three-point MACE became 0.89 (95% CI: 0.81-0.98, p=0.022). For SGLT-2 inhibitors, the hazard ratio post-censoring became 0.91 (95% CI: 0.83-1.00, p=0.046). And for “any medication,” the hazard ratio post-censoring became 0.88 (95% CI: 0.80-0.98, p=0.022). Thus, the magnitude of the hazard ratio/risk reduction was not changed, but statistical significance did increase, with a shift to the left of unity in favor of Bydureon. Results on all-cause death were consistent with primary results (HR=0.88, 95% CI: 0.77-0.97, p=0.016): Significant effects were seen from censoring SUs (HR=0.84, 95% CI: 0.75-0.95, p=0.007), SGLT-2 inhibitors (HR=0.86, 95% CI: 0.77-0.97, p=0.016), and any medication (HR=0.84, 95% CI: 0.73-0.96, p=0.009), but also from censoring DPP-4 inhibitors, insulin, GLP-1 agonists, and GLP-1 agonists + SGLT-2 inhibitors. While EXSCEL was not properly powered for these analyses, which assume that new medications were continued once started (Dr. Bethel identified this as a limitation before diving into the post-hoc findings), we do find the results compelling. EXSCEL’s narrow miss for superiority has been attributed, by many thought leaders, to the pragmatic nature of the trial (the intervention was delivered in usual care settings), a larger primary prevention cohort (27% of participants), a lack of patient-friendliness in dosing Bydureon (lengthy mixing/reconstitution was required with the single-dose kits), and the wide range of concomitant drug use allowed – this last suspicion is at least partially validated by these results. This post-hoc analysis joins one presented by Dr. Robert Mentz at AHA 2017, showing that Bydureon’s effects were consistent across baseline risk quintiles. We eagerly await more looks at the data to come over the next few years, which will hopefully clarify the benefits of exenatide across a diverse spectrum of clinical scenarios. 

3. Meta-Analysis of GLP-1 Agonist CVOTs Implies Cardioprotective Class Effect on Endpoints of MACE, CV Death, All-Cause Mortality; Encompasses ELIXA, LEADER, SUSTAIN 6, and EXSCEL

Oxford’s Dr. Angelyn Bethel also presented a meta-analysis of four reported CVOTs for a GLP-1 agonist, suggesting significant class effects on three-point MACE (non-fatal MI, non-fatal stroke, or CV death), CV death, and all-cause death. The results were simultaneously published online in the Lancet Diabetes & Endocrinology. For three-point MACE, a 10% relative risk reduction was calculated (HR=0.90, 95% CI: 0.82-0.99, p=0.033). Tests for heterogeneity among trials were moderate (Q-test p=0.11, I2=50%), indicating that the meta-analysis was statistically appropriate and this effect was likely robust for the GLP-1 agonist class. Of note, the ELIXA CVOT for Sanofi’s lixisenatide used a primary outcome of four-point MACE (also including hospitalization for unstable angina) – Dr. Rury Holman thus excluded ELIXA and presented a meta-analysis of LEADER (Novo Nordisk’s liraglutide), SUSTAIN 6 (Novo Nordisk’s semaglutide), and EXSCEL (AZ’s exenatide) in his commentary at EASD, reporting a relative risk reduction of 12% for three-point MACE across the three longer-acting agents (HR=0.88, 95% CI: 0.81-0.95, p=0.002). All-cause death was reduced by 12% across the four trials vs. placebo (HR=0.88, 95% CI: 0.81-0.95, p=0.002, Q-test p=0.63, I2=0%), and CV death was reduced by 13% vs. placebo (HR=0.87, 95% CI: 0.79-0.96, p=0.007, Q-test p=0.43, I2=0%). We’ve come to think of EXSCEL for AZ’s Bydureon (exenatide once-weekly) as confirming CV safety while hinting at CV efficacy for GLP-1 agonist agents, and this meta-analysis supports this overarching message, though of course the limitations of meta-analysis must be considered in interpreting the results. By nature, a meta-analysis can be driven by particularly strong – perhaps outlier – results in one direction or the other, but on the whole, we’re encouraged by what Dr. Bethel called fair heterogeneity among the trials.

  • Safety endpoints of particular interest were also positive. The odds ratio was 0.93 for severe hypoglycemia with any GLP-1 agonist product vs. placebo (95% CI: 0.74-1.18), 0.90 for pancreatitis (95% CI: 0.63-1.28), and 0.83 for pancreatic cancer (95% CI: 0.33-2.11). There was a neutral effect overall on fatal and nonfatal MI (HR=0.94, 95% CI: 0.86-1.03), fatal and nonfatal stroke (HR=0.87, 95% CI: 0.75-1.00), hospitalization for heart failure (HR=0.93, 95% CI: 0.83-1.04), and hospitalization for unstable angina (HR=1.09, 95% CI: 0.90-1.32). Notably, where SGLT-2 inhibitors are showing efficacy in reducing heart failure for people with diabetes, GLP-1 agonists are firmly demonstrating safety. At ESC 2017, Dr. Naveed Sattar mentioned the very low likelihood that any company will launch a dedicated study of a GLP-1 agonist in heart failure, in contrast to the recently-initiated outcomes trials of SGLT-2 inhibitors Jardiance (Lilly/BI’s empagliflozin) and Farxiga (AZ’s dapagliflozin) toward a heart failure indication.

4. Meta-Analysis Shows Superior A1c-Lowering Efficacy for Novo Nordisk’s GLP-1 Semaglutide (Now Approved as Ozempic!) vs. SGLT-2 Inhibitors

A Novo Nordisk poster compared GLP-1 agonist semaglutide vs. SGLT-2 inhibitors, finding more powerful A1c-lowering efficacy in type 2 diabetes for the GLP-1 – which, coincidentally, was just FDA-approved as Ozempic. This meta-analysis encompassed 19 phase 3 studies (n=13,329), all of which enrolled participants with type 2 diabetes inadequately controlled on metformin, and all involving randomization to either semaglutide (0.5 mg or 1 mg) or an SGLT-2 inhibitor (canagliflozin 100 mg and 300 mg, dapagliflozin 2.5 mg, 5 mg, and 10 mg, and empagliflozin 10 mg and 25 mg). As illustrated in the figure below, both doses of semaglutide showed greater A1c reductions than any of the SGLT-2 inhibitors at both 26 weeks and 52 weeks. Interestingly, dapagliflozin showed the most modest effect on A1c, with slightly greater A1c-lowering efficacy for canagliflozin and empagliflozin on an absolute numeric basis (although this analysis was not meant to show statistical differences between different SGLT-2 inhibitors, and this wasn’t reported). In line with semaglutide’s superiority in reducing A1c from baseline, both doses of the agent were statistically superior to all SGLT-2 inhibitors in achieving the target A1c of ≤7% at 26 weeks. In fact, further mathematical modeling demonstrated that semaglutide has between 70%-80% higher probability of achieving this A1c goal than any of the SGLT-2 inhibitors. Novo Nordisk’s clinical program for semaglutide has included a number of head-to-head comparisons, and Ozempic has now shown superior glycemic efficacy to AZ’s GLP-1 agonist Bydureon (exenatide once-weekly), to Lilly’s GLP-1 agonist Trulicity (dulaglutide once-weekly), to Merck’s DPP-4 inhibitor Januvia (sitagliptin), and to Sanofi’s basal insulin Lantus (insulin glargine) in RCTs, in addition to this meta-analysis pointing to advantages over SGLT-2 inhibitors. Demonstrating superiority to some of the most commonly-used diabetes drugs and to some of the most advanced is no small feat, bolstering Ozempic’s already-impressive clinical profile. That said, we note that this poster should not diminish our impression of the clinical benefits of SGLT-2 inhibitors, which set a high bar for semaglutide to top.

  • This meta-analysis also evaluated weight loss with semaglutide vs. SGLT-2 inhibitors. Semaglutide 1.0 mg was significantly more efficacious in reducing body weight relative to all SGLT-inhibitors at 26 weeks (~11 lbs with semaglutide vs. ~4-8 lbs with an SGLT-2 inhibitor), but lost significance for the comparison with respect to dapagliflozin 5 mg at 52 weeks. At ~8 lbs decline in body weight, semaglutide 0.5 mg also had a favorable weight loss profile at both 26 weeks and 52 weeks, but only achieved statistically significant superiority with respect to dapagliflozin 2.5 mg and empagliflozin 10 mg (both at 26 weeks). Notably, semaglutide is being developed for a dedicated obesity indication, and the phase 3 STEP program is scheduled to begin in 1H18. In our view, these results indicate the weight loss efficacy of SGLT-2 inhibitors as well, but there’s no doubt from clinical data collected so far that semaglutide comes with a meaningful weight loss benefit. Novo Nordisk also plans to conduct a CVOT of semaglutide in obesity; this would be the first study to establish medical obesity as a chronic disease. 

5. Post-Hocs of ACE from Dr. Holman: Bayer’s Glucobay (Acarbose) Demonstrates Compelling Efficacy in Diabetes Prevention in On-Treatment Analysis and Analysis of Reversion to Normoglycemia

Dr. Rury Holman presented new analyses of ACE, the CVOT of Bayer’s alpha-glucosidase inhibitor Glucobay (acarbose) which reported neutral CV outcomes vs. placebo but significant risk reduction for new-onset type 2 diabetes in a population with prediabetes in China (n=6,526). Full results from the trial were presented at EASD 2017 in Lisbon, and Dr. Holman expanded on these prior data with a new on-treatment analysis and a new reverse analysis – that is, how many individuals were able to revert to normoglycemia? After five years, 59% of acarbose-treated patients presented a normal result on a glucose tolerance test vs. 54% of placebo-treated patients, corresponding to a favorable rate ratio of 1.26 (95% CI: 1.13-1.40, p<0.0001). This lends further support to the notion that acarbose could have applications in prediabetes, following the initial finding of an 18% relative risk reduction for new-onset diabetes over five years (HR=0.82, 95% CI: 0.71-0.94, p=0.005). An on-treatment analysis for new-onset diabetes showed “even more impressive results,” according to Dr. Holman, with a rate ratio of 0.79 (95% CI: 0.68-0.92, p=0.0017). Notably, participants enrolled in ACE had a prior history of CV events, and delaying/preventing type 2 diabetes in this population could be especially important because further progression of hyperglycemia would greatly exacerbate their CV risk (a diabetes diagnosis confers similar risk for premature death as having a previous MI). ACE represents a successfully completed outcomes study in a prediabetes population, which should also set an example for the field as we push for more clinical research into effective strategies for earlier intervention – not only to avoid diabetes, but also to avoid hard outcomes like stroke, MI, hospitalization for unstable angina or heart failure, and CV death. J&J announced plans for a prediabetes CVOT of SGLT-2 inhibitor Invokana (canagliflozin), though we’ve heard no updates on this project since the company’s 3Q16 update, while Novo Nordisk just recently announced plans for an obesity CVOT of GLP-1 agonist semaglutide (now FDA-approved for type 2 diabetes as Ozempic). Dr. Holman established the high unmet need for prevention efforts in China, where almost half of the adult population has prediabetes – that’s ~500 million people! We’d very much like to see pharmacotherapy utilized for diabetes prevention in China and elsewhere, and to this end, we’ll be curious to see if ACE results extend to a more diverse, global population outside of China, as prediabetes is certainly a worldwide epidemic by this point.

  • During Q&A, Dr. Holman shared his view that prediabetes is indeed a CV risk marker. Several sessions at IDF have focused on CV risk reduction for people even before they progress to diabetes (see our coverage of a talk by Dr. Jaakko Tuomilehto in this report), and we wonder if this interest from thought leaders foreshadows additional clinical trials in prediabetes populations. We see tremendous value in these investigations, especially when they collect data on outcomes more so than biomarkers, because there’s no reason we should be waiting until patients exhibit multiple CV risk factors before we intervene to prevent CV morbidity/mortality.

6. Dr. Tuomilehto: Don’t Wait for Onset of Diabetes – “Impaired Glucose Tolerance is Already a Category of Increased CV Risk”; Argues for Two-Hour Postprandial Glucose as Best Metric for Gauging CV Risk in Prediabetes; Reviews DECODE & EUROSPIRE IV Results

Dr. Jaakko Tuomilehto, a highly-esteemed expert in diabetes prevention, showed how postprandial glucose (measured by a two-hour post-meal test) may be the best predictor of CV outcomes in people with prediabetes. The overarching takeaway from his talk was that prediabetes itself confers CV risk – many of these individuals will have a CV event before onset of type 2 diabetes, so we should be intervening earlier with CV risk reduction strategies. Dr. Tuomilehto focused most of his presentation on the DECODE study (n=55,000 people across 22 sites and 11 countries in Europe), but also called out recently-published EUROSPIRE IV (n=4,004), in which two-hour postprandial glucose was significantly correlated with CV outcomes in a population with coronary artery disease but no diabetes, while neither A1c nor fasting plasma glucose showed a statistically significant association with MACE (non-fatal MI, non-fatal stroke, CV death, or hospitalization for heart failure) over two years. In DECODE, researchers found a linear relationship between two-hour postprandial glucose and all-cause mortality. If the risk of death with known diabetes is mapped to a hazard ratio of 1.00, then a two-hour postprandial reading <95 mg/dl was associated with a hazard ratio of 0.47, a reading between 95-112 mg/dl was associated with a hazard ratio of 0.50, and this increased steadily until a reading >200 mg/dl was associated with a hazard ratio of 0.92. No such linear relationship was seen between fasting plasma glucose and all-cause mortality. For example, fasting glucose <85 mg/dl mapped to a hazard ratio of 0.58, fasting glucose between 85-105 mg/dl mapped to a slightly lower hazard ratio of 0.52, and a fasting glucose between 105-110 mg/dl mapped to a slightly higher hazard ratio than that, at 0.56. Dr. Tuomilehto also broke down DECODE results by different categories of death. When impaired glucose tolerance was defined by a two-hour meal test, the rate ratio for CV death was 1.34, and this was barely changed when also adjusting for fasting plasma glucose (1.32). The rate ratio was 1.26 for fatal stroke, and this was barely changed to 1.21 when also adjusting for fasting plasma glucose, suggesting that the latter has little to no effect in linking prediabetes hyperglycemia to mortality outcomes. Notably, UNC’s Dr. John Buse presented a contrasting opinion at Keystone 2017, arguing that A1c is a better predictor of CV events for people with prediabetes, rather than fasting or two-hour glucose. Italy’s Dr. Antonio Ceriello spoke right after Dr. Tuomilehto during this IDF symposium, and also pointed to some conflicting evidence, ultimately concluding that it’s still “an open question” whether postprandial glucose is an independent risk factor for CV disease. That said, he acknowledged that evidence is accumulating, and indeed, with a growing emphasis in the field on prediabetes and prevention, we’re hoping that further analyses and meta-analyses will elucidate answers here.

  • Above all, Dr. Tuomilehto, Dr. Buse, Dr. Ceriello, and numerous other thought leaders agree that prediabetes should be considered a risk state for CV morbidity/mortality. As Dr. Ceriello put it, “CV risk mitigation is becoming a central goal in diabetes management, and maybe it should also be the goal of prediabetes management.” This consensus is very important. The correlation with macrovascular outcomes could compel payers and regulators to recognize prediabetes as a disease. This would make it easier to get drugs indicated for prediabetes, and that might prompt more real-world HCPs to treat for prevention, while also encouraging people at-risk for diabetes to seek care sooner. Turning to a Finnish study (n=504), Dr. Tuomilehto reported a CV event rate of 8 per 1,000 patient-years across the cohort with prediabetes, which rose only to 9.3 events per 1,000 patient-years when looking at all participants with prediabetes or diabetes. Again, the key takeaway was that CV risk is already substantially elevated for the individual with hyperglycemia even if it’s not quite at the level where we define type 2 diabetes. “There are people with impaired glucose tolerance who are developing CV disease before they develop diabetes,” Dr. Tuomilehto explained, and added “this is a category where we should focus interventions, because it’s where things happen.”  

7. Could DPP-4 Inhibitors Play a Role in Slowing Atherosclerosis? Possibly, If Used Earlier and Longer, Says Dr. Watada

Dr. Hirotaka Watada presented two Japanese studies demonstrating beneficial effects of alogliptin (Takeda’s DPP-4 inhibitor Nesina) and sitagliptin (Merck’s Januvia) in reducing carotid intima-media thickness (cIMT) vs. placebo in patients with type 2 diabetes and no established CV disease. He pointed to a potential CV benefit when these agents are used early in the course of disease – which, coincidentally, is when thought leaders recommend them the most. While the SAVOR-TIMI, EXAMINE, and TECOS studies of Onglyza (AZ’s saxagliptin), Nesina, and Januvia all found a resoundingly neutral impact of these DPP-4 inhibitors on three-point MACE (despite mixed and controversial results on heart failure), Dr. Watada presented a series of compelling studies that aim to reopen the book on cardioprotection for the class. While the aforementioned CVOTs only enrolled high-risk patients or those with established CV disease, who would tend to have advanced atherosclerosis, Dr. Watada published two papers last year evaluating DPP-4 inhibitors earlier in the course of diabetes, both using cIMT as a measure of atherosclerosis. The SPEAD-A trial (n=322) randomized patients with type 2 diabetes not in glycemic control and free from apparent CV disease to two years of either alogliptin or conventional treatment, measuring cIMT at intake and one and two years. At 102 weeks, the treatment effect of alogliptin was -0.031 mm on mean IMT (95% CI: -0.058 to -0.005, p=0.019), -0.065 mm on right max-IMT (95% CI: -0.114 to -0.016, p=0.01), and -0.070 mm on left max-IMT (95% CI: -0.114 to -0.014, p=0.008). Similarly, the SPIKE trial (n=274) enrolled patients with type 2 diabetes not in glycemic control and on insulin, without apparent CV disease, and randomized them to either sitagliptin or placebo. At 102 weeks, the treatment effect of sitagliptin was -0.029 mm on mean IMT (95% CI: -0.090 to -0.016, p=0.005) and -0.065 mm on left max-IMT (95% CI: -0161 to -0.014, p=0.021). Right max-IMT was not significantly reduced with sitagliptin. Thus, two DPP-4 inhibitors have been shown to attenuate the progression of atherosclerosis in common carotid arteries vs. placebo in patients with type 2 diabetes without established CV disease. These effects are all the more important when one considers the increase in IMT seen in the placebo group. That said, the beneficial reductions in atherosclerosis with alogliptin and sitagliptin have not been directly linked to improved macrovascular outcomes, making the clinical significance unclear. While these are relatively short and small trials (compared to a full CVOT) in a very homogeneous Japanese population, we are encouraged to see evidence supporting the early use of DPP-4 inhibitors for CV benefit. Moreover, this trial could have particular implications in Japan, where DPP-4 inhibitors are often-prescribed. Also to his point, Dr. Watada mentioned a cohort study (n=~104,000) in Taiwan, published in 2016, showing a significant CV benefit with sitagliptin. We wouldn’t be surprised if trial design were a big factor in demonstrating benefit here, and we’ll keep our eyes peeled for more on this front.

8. Novo Nordisk’s Next-Gen Tresiba (Insulin Degludec) Still Significantly Reduces Hypoglycemia Risk After Adjustment for A1c Fluctuations, and Regardless of Whether or Not Patients Achieve Target A1c ≤7%, According to SWITCH Post-Hoc Analysis

A new post-hoc analysis of the SWITCH 1 and 2 trials found that Novo Nordisk’s Tresiba (insulin degludec) was associated with fewer hypoglycemia episodes vs. Sanofi’s Lantus (insulin glargine) for people with type 1 and 2 diabetes alike, regardless of A1c. The results, presented on a poster and also highlighted in a company press release, demonstrate that the estimated rate ratios for severe, nocturnal, and overall hypoglycemia for people with type 1 and type 2 diabetes in the SWITCH program (all of which favor the next-gen Tresiba over current standard of care Lantus) were identical after statistical adjustment for fluctuations in A1c throughout the trial, as illustrated in the table below. These rate ratios reflect a significant risk reduction for hypoglycemia with Tresiba across all categories for people with type 1 diabetes in the SWITCH 1 trial, and for overall and nocturnal hypoglycemia for people with type 2 diabetes in the SWITCH 2 trial (risk reduction for severe hypoglycemia trended in favor of Tresiba, but did not reach statistical significance). Furthermore, the analysis indicates that there was less hypoglycemia with Tresiba regardless of whether participants had an A1c above or below the target of 7%. For people with type 1 diabetes and an A1c ≤7%, observed rates of hypoglycemia were numerically lower with Tresiba vs. Lantus (overall [22 vs. 26 events per patient-year of exposure], nocturnal [2 vs. 4], and severe [0.6 vs. 0.9]) and the same held true for their counterparts with A1c >7% (overall [22 vs. 22], nocturnal [3 vs. 5], and severe [0.8 vs. 0.9]). This was also the case for people with type 2 diabetes and A1c ≤7% (overall [1.6 vs. 2.8], nocturnal [0.5 vs. 1.0], and severe [0.09 vs. 0.1]) as well as A1c >7% (overall [2.1 vs. 2.5], nocturnal [0.6 vs. 0.9], and severe [0.01 vs. 0.06]). Hypoglycemia and fear of hypoglycemia are major barriers to initiation and adherence in the context of insulin therapy – by reducing this risk, Tresiba represents a major win for patients. It’s certainly reassuring to know that Tresiba’s hypoglycemia benefit is consistent regardless of A1c. These results add to the impressive body of evidence around hypoglycemia risk reduction for this next-generation basal insulin. To this end, a hypoglycemia claim for Tresiba has been approved in the EU and an FDA decision on a similar revision to the US label is expected in 1Q18, based on the SWITCH studies as well as the landmark DEVOTE trial.

9. Saxa/Dapa Combo Demonstrates Superiority vs. SU Glimepiride on A1c, Body Weight, & Hypoglycemia in One-Year DapaZu Trial

A poster comparing combination therapy with AZ’s SGLT-2 inhibitor Farxiga (dapagliflozin) + DPP-4 inhibitor Onglyza (saxagliptin) vs. glimepiride pointed to the inferior glycemic efficacy, significant weight gain, and significant hypoglycemia risk associated with sulfonylureas. This study, DapaZu, was completed in March 2017 according to ClinicalTrials.gov, randomizing 939 patients on background metformin to dapagliflozin, dapagliflozin + saxagliptin, or glimepiride for 52 weeks. The average participant was 57-59 years-old with an A1c of 8.3%, fasting plasma glucose of ~190 mg/dl, body weight between ~202-213 lbs, and BMI of 33 kg/m2. After one year of treatment, the combination approach showed superior A1c reductions vs. glimepiride (1.2% vs. 1%, p<0.001), while dapagliflozin monotherapy was non-inferior to glimepiride on this endpoint (0.8% vs. 1%). Fasting plasma glucose declined by a mean ~35 mg/dl in the combination arm vs. ~25 mg/dl in the SU arm (p<0.001), and standalone dapagliflozin again showed non-inferiority. Both dapagliflozin-containing regimens did significantly better on weight loss compared to the sulfonylurea: Body weight dropped ~8 lbs for patients on dapagliflozin only, while patients on glimepiride gained ~4 lbs on average over 52 weeks (p<0.001). The magnitude of weight loss was just barely smaller in the dapagliflozin/saxagliptin group, at ~7 lbs, but this was still superior to weight gain with glimepiride (p<0.001). Hypoglycemia was much more common in the SU arm of the trial, with symptomatic events <70 mg/dl occurring in 216 of 312 patients (a whopping 69%) – only 7 of 312 people in the combination group (2%) and only 1 of 313 people in the SGLT-2 only group (<1%) experienced this endpoint. In our view, the weight gain and hypoglycemia associated with sulfonylureas are enough to deter patients/HCPs from this drug class, and we note that glimepiride is actually one of the “better” agents in the class according to thought leaders. Dr. Irl Hirsch has suggested that glimepiride confers the least hypoglycemia risk of all sulfonylureas, and still 69% of study participants taking it had one or more episodes. On the flip side, DapaZu underlines the weight loss efficacy of SGLT-2 inhibitors, to say nothing of the profound A1c-lowering that occurs without excess hypoglycemia. As head-to-head data vs. sulfonylureas accumulates (we saw a fair amount at EASD this year), we’re hoping that more patients are switched to a safer, more effective diabetes therapy. SUs remain the most-often prescribed second-line diabetes drug in the US and are also very commonly prescribed globally. This poster was particularly intriguing to us because both SGLT-2 inhibitors and DPP-4 inhibitors have been proposed as alternative first-line agents (though metformin is the recommended first-line currently, 8% of first-line diabetes prescriptions in the US were for sulfonylureas in 2016).

  • Sulfonylureas have also been associated with attenuated glycemic efficacy over the long term, as they spur beta cell burnout. For this reason, we imagine that longer study duration may have revealed significantly superior A1c-lowering for dapagliflozin vs. glimepiride as well as dapagliflozin/saxagliptin vs. glimepiride. Two ongoing outcomes studies promise to illuminate this longer-term impact of SU treatment: (i) Lilly/BI’s CAROLINA CVOT compares DPP-4 inhibitor Tradjenta (linagliptin) head-to-head vs. glimepiride, and is expected to complete in March 2019, while (ii) the NIH-funded GRADE study compares outcomes between basal insulin, DPP-4, GLP-1, and SU, and the last patient visit is expected around April 2021.
  • AZ has also developed Qtern, a fixed-dose combination of dapagliflozin/saxagliptin. The product is available in Europe, though it has not yet launched in the US despite FDA approval in March 2017.

10. How Did Merck/Pfizer’s SGLT-2 Ertugliflozin Fare Across Six Clinical Trials? Poster Presentation Shows Superior A1c Reductions, Weight Loss, Blood Pressure Decline; Highly-Anticipated FDA Decision by Year-End

A meta-analysis covering six phase 3 studies of Merck/Pfizer’s SGLT-2 inhibitor ertugliflozin was presented on a poster. All part of the VERTIS clinical program, these studies enrolled 4,395 adults with type 2 diabetes in total. Across VERTIS MONO, VERTIS MET, VERTIS SITA, VERTIS SITA2, and VERTIS FACTORIAL, ertugliflozin demonstrated superior A1c reductions vs. its comparator. The only exception was VERTIS SU, the focus of an oral presentation at EASD by Dr. Brett Lauring, in which 15 mg ertugliflozin demonstrated non-inferiority vs. glimepiride on the primary A1c endpoint after 52 weeks (as we note in our coverage of an AZ poster above, sulfonylureas are thought to cause beta cell burnout over the long term, leading to attenuated glycemic efficacy – were VERTIS SU a longer trial, it may have produced a different result, perhaps superiority for the SGLT-2 inhibitor). Weight loss was consistently greater with ertugliflozin vs. its comparator in all six trials, as was systolic blood pressure decline. Safety findings were as expected, with no major imbalance in adverse events with the exception of genital mycotic infections occurring more often in the ertugliflozin group. The timing of this meta-analysis is noteworthy, as an FDA decision on ertugliflozin is expected by end of year (Merck/Pfizer’s New Drug Application was accepted for active review this past March). The candidate’s clinical profile seems to match what we know about the SGLT-2 class, with profound benefits to A1c, body weight, and blood pressure. The VERTIS CV study is ongoing to evaluate CV safety (or possible cardioprotection), with an expected completion date of October 2019.

11. Dr. Frias Touts Benefits to Soliqua (Insulin Glargine/Lixisenatide) in Sanofi-Sponsored Symposium; ~33% of Soliqua-Treated Patients Achieve A1c ≤7% without Weight Gain or Hypoglycemia

During a Sanofi-sponsored symposium, Dr. Juan Frias reviewed key data on fixed-ratio combination Soliqua (basal insulin glargine/GLP-1 agonist lixisenatide), highlighting the drug’s benefits to A1c, adherence, patient quality of life, and the composite outcome of A1c ≤7% without weight gain or hypoglycemia. He discussed post-hoc analyses of the phase 3 LixiLan program showing dramatic A1c reduction for Soliqua-treated patients with baseline A1c ≥9%. From a mean of 9.5% at baseline, this cohort reached a mean 7% A1c with Soliqua vs. 7.6% with Lantus (insulin glargine) after 30 weeks. Soliqua demonstrated superior glycemic efficacy vs. component monotherapies in all subgroups regardless of baseline A1c, but Dr. Frias suggested that HCPs should particularly consider the fixed-ratio combination product for their patients with very high starting A1c. He positioned Soliqua as a simpler option for treatment intensification vs. basal-bolus therapy, given the single injection. In his slides, he included graphs displaying the inverse relationship between complexity of medication regimen and adherence: Patients prescribed one dose/day take 79% of doses on average (which we point out is already rather low), and adherence declines to an abysmal 51% when patients are prescribed four doses/day. Dr. Frias established the importance of adherence for long-term clinical outcomes in diabetes. Drug tolerability is another critical aspect to patient quality of life, and he explained how GI side-effects (nausea, vomiting, diarrhea) are much milder with insulin glargine/lixisenatide together vs. lixisenatide alone, because Soliqua allows for very gradual dose escalation of the GLP-1 component. He emphasized the decreased frequency of hypoglycemia and the neutralized weight gain with the fixed-ratio combination vs. basal insulin monotherapy, which are extremely important outcomes beyond A1c. Considering all these factors together, Dr. Frias shared that ~33% of Soliqua-treated patients achieved A1c ≤7% with no weight gain or severe hypoglycemia. He seemed quite impressed with this number, although Dr. Samit Ghosal announced on day #2 of IDF that ~80% of patients on Novo Nordisk’s fixed-ratio combination Xultophy (insulin degludec/liraglutide) achieved the same composite. In response, Dr. Frias emphasized the different definitions of hypoglycemia used in the LixiLan program for Soliqua (<70 mg/dl) vs. the DUAL program for Xultophy (<56 mg/dl) – on the surface, this would lead to more recorded hypoglycemia in the LixiLan trials, and Dr. Frias cautioned against over-comparison between “apples and oranges.” To be sure, there’s been no head-to-head study comparing Soliqua and Xultophy, and there’s a lack of standardization across these clinical trials. Moreover, we don’t think the focus should be on in-class competition at all, but rather whole class growth, since sales of both Soliqua and Xultophy have trended below expectations since launch in January and May 2017, respectively. We did find it notable that both Dr. Frias and Dr. Ghosal underscored the effects of basal insulin/GLP-1 fixed-ratio combos on the composite endpoint of A1c ≤7% without weight gain or hypoglycemia, as this could be considered the sweet spot for this emerging therapy class.

12. Dr. Lutz Heinemann on Efforts to Enhance Insulin Absorption Speed – Pharmaceutical and Mechanical Approaches

Dr. Lutz Heinemann covered current efforts to accelerate insulin absorption and action. He reviewed Adocia’s BioChaperone insulin lispro, shown to result in a 61% reduction in postprandial glycemic excursions, as well as Novo Nordisk’s Fiasp (faster-acting aspart), which he believes has a particularly bright future. He was also encouraged by Lilly’s ultra-rapid insulin, just moved into phase 3, which uses a substance known to treat pulmonary hypertension. While he was optimistic regarding these attempts to improve insulin formulation, he was not so positive about advances in mechanical delivery systems. Insuline’s InsuPad, a disposable patch for insulin injections, automatically delivers localized heat to the injection site, aiming to increase internal blood flow and accelerate insulin absorption. The patch received its CE mark back in 2008, but has been off the radar for a few years now. Dr. Heinemann noted its initial commercial success, but confessed he is unsure of InsuPad’s future – as far as we know, the device has not been cleared in the US, and we imagine wearing something else on the body for incremental benefit is a deal-breaker for most. Intradermal application (as opposed to traditional subcutaneous injection) via 1-1.5mm microneedles has also been shown to induce enhanced insulin absorption. BD cancelled this project, though there are some such applications in development within academia level. Dr. Heinemann also noted data suggesting that “insulin spreading” results in faster absorption, referencing a study demonstrating nine injections of two units to act faster than one injection of 18 units. This is a product of larger surface area:volume ratio, and a piece of the rationale behind the sprinkler-type infusion set cannulas. A needle-free system promoting insulin spreading has been tested, but was ultimately not brought to market due to an unpleasant noise upon injection. While the lack of tangible progress for patients is unfortunate, we’re seeing more investment in this area than we have historically. For now, we can only preach good practices: Site rotation, 4-mm needle, etc. 

  • Dr. Heinemann also highlighted his study aiming to visualize lipohypertrophy via thermography. The prevalence of lipohypertrophy varies significantly, ranging from 16%-60%. Dr. Heinemann believes this is likely because lesions are often not visible, requiring palpation to detect. Lipohypertrophy can pose substantial challenges, as insulin injected into affected skin areas exhibits decreased absorption rates, translating to higher postprandial glycemic excursions. Dr. Heinemann hopes his approach will improve identification of lipohypertrophy by distinguishing cooler areas in the patient’s body.
  • In the same session, Dr. Jothydev Kesavadev reviewed previous, current, and future insulin delivery systems. He was particularly excited about Medtronic’s Advanced Hybrid Closed Loop (previously called “690G”) with Dreamed’s Fuzzy Logic, referring to the promise of automated correction boluses as a “dream come true.” Dr. Kesavadev was also very encouraged by Bigfoot’s hybrid closed loop efforts, underscoring the benefits of the company’s auto-titration system (“Inject”) to MDI users. He also spoke highly of Beta Bionics’ system, which he considers to be highly effective and especially user-friendly, as it requires minimal user input and is easy to set up (just body weight). In addition to advances in delivery devices, Dr. Kesavadev also covered new insulins, noting that Fiasp will be the insulin of choice for both pumps and hybrid closed loop systems.

Hypoglycemia Highlights

1. New EMA Diabetes Drug Guidance Out In January, BIG Update to Include <54 as “serious, clinically important hypoglycemia”; <70 also included, aligns with International Consensus

University of Leuven’s Dr. Bart Van der Schueren announced that EMA’s updated diabetes drug development guidance will be out for public comment in January. In a major victory, he gave a preview of the hypoglycemia section, which has been updated since his EASD talk to align with the consensus definitions of <54 mg/dl, <70 mg/dl, and severe hypoglycemia. (At EASD, the slide included less than or equal to.) It’s great to see EMA is now consistent with the two just-published Diabetes Care statements from ATTD and JDRF et al. – “It’s very important that we align throughout the world.” Dr. Van der Schueren clarified the exciting Beyond-A1c implications in Q&A: “Previously, any additional treatment had to show a change in A1c. Now if we include hypoglycemia into the guidance, the idea is to foster development of therapies that would lead to less hypoglycemia. Therefore, we would recognize that in the label. That also means you could get approved on that basis, even if you didn’t show a difference on mean glycemic control.” What a major victory! Indeed, <54 mg/dl is characterized as “Serious, clinically important hypoglycemia,” a clear sign EMA believes it matters and might approve a therapy on this basis. (Take note, FDA CDER!) Dr. Van der Schueren added in Q&A that time-in-range is not in the updated EMA drug guidance, as there is debate on what the high threshold should be. (Perhaps some think >180 mg/dl is not serious enough?) In other words, based on the current draft guidance, time-in-range data could not be included in a label. Still, we’re glad to see hypoglycemia is firmly in place, and we hope EMA receives many positive comments in January on these thresholds and the benefits of using CGM to characterize drugs. Perhaps advocates will also recommend time in 70-180 mg/dl as a meaningful endpoint to be included in the final guidance. We also hope this EMA guidance encourages the FDA to update its own outdated diabetes drug guidance.

  • Specific patient-reported outcomes (PRO) instruments are not specified in the updated EMA guidance. Dr. Van der Schueren said “they are not currently robust enough,” and there is obviously no consensus on which instruments should be included. We believe this is a solvable problem, but one that will demand as much commitment as the drive to standardize CGM thresholds (and perhaps more, since there seems to be less clarity in that field than in CGM).
  • For context, EMA released its last diabetes drug guidance in 2012 (which included CGM already), and is already updating it five years later. Meanwhile, FDA’s 2008 diabetes drug development guidance (still “draft”) will have its ten-year anniversary in March! Boy have the drug and device fields changed a lot since then…

2. Key Beyond A1c Q’s: What are the right benchmarks? What about standardizing PRO Instruments? What research is needed?

The selected Q&A below addressed some of the key issues in the Beyond A1c movement: benchmarking outcomes, patient-reported outcomes (PROs), and future research. This came from the all-star hypoglycemia session, on which other highlights appear directly above and below this one.

Kelly Close: It was great to hear you talk about fostering development of therapies that reduce hypoglycemia. What about PROs? What instruments should be used? Are they sensitive enough?

Dr. Van der Schueren: We knew we needed to include endpoints not directly related to glycemic control. Then, when we started to look at PROs used by companies – and which to use –it became a very traumatic experience. We decided we would do it on a case by case basis. Dulaglutide (Lilly’s Trulicity) included a PRO in section 5.1 of its summary of product characteristics (page 16 here). But we are not able to tell companies what PRO to use in their clinical development program. (Although the hypoglycemia fear survey has been widely used.) We should move in the next couple of years to make this happen. I think it’s very important that treatments improve quality of life of patients – this is crazy that’s it’s not there.

Dr. Heller: We know PROs are very affected by the way medication is sold to them. If you don’t have a double-blind study, there are serious questions over PROs. It would be interesting to hear a regulatory comment on that.

Dr. Van der Schueren: We would only include PROs from placebo controlled studies.

Kelly: We’ve come to conclusions on how hypoglycemia can be measured and monitored. I’m hopeful about this for Pros. It’s going to require some investment in better understanding it.

Q (JDRF Australia): What single trial, analysis, or study, could be the most influential to bring this topic further into the public eye?

Dr. Van der Schueren: We should perform a DCCT with glycemic variability in it, confirming that glycemic variability has an influence on harder endpoints. This will validate it more for regulatory agencies. That’s one of the first things I would prioritize.

Dr. Heller: We are looking at a changing landscape. BGM was introduced in ~1977. In the next five years, in type 1 diabetes, patients are not going to participate in regulatory trials because they have to measure glucose with fingersticks. Many in Europe will be using CGM. The large companies developing these products, have failed to use CGM to date. I know that because I participate in those trials. They have sub-studies where a few people use it. Those data are never published. It’s extraordinary. They don’t have the experience, and many of the centers don’t know how to use CGM. For me, it’s taking the new tech and incorporating it into regulatory studies that EMA and other regulators need. We can look at glycemic variability. For me that is a fundamental thing we need to do.

Adam Brown: What should the benchmark goal be for time spent <70 mg/dl? Should it be <3%? <5%? <10%?

Dr. Heller: We don’t know. We need more research. I think the critical thing is the between 70 and 55 mg/dl range. If you spent 5% of the time at that level (55-70), there is very little evidence that you’ll impair responses. I think it’s around the 5% mark. Watching people use Dexcom G5, they see how much hypoglycemia they have at night – it’s transformative. It’s why I think CGM is so helpful. I think the 5%-10% mark, and for under 55 mg/dl, we should be stopping it completely. More than 2-3% would be very concerning. For 55-70 mg/dl, I’m not so worried – lower than that is the issue. Above it, 5% is where we might aim.

Adam: If we need more research, what would that trial look like, to help us understand a proper benchmark for time in hypoglycemia?

Dr. Heller: If people drop under 55 mg/dl, they develop unawareness, which drives risk of severe hypoglycemia. With large numbers and very good CGM, we can begin to measure what the consequences are of clinical episodes in that range. In the lab, we could even measure awareness thresholds. It would be great to know whether 5% between 55-70 adversely impact awareness thresholds – I don’t know. But let’s be clear, in five years’ time, in type 1 diabetes, everybody will be using CGM in the trials. There is a huge amount we will learn.

3. Dr. Simon Heller – Severe Hypoglycemia Has Not Improved in Recent Years, Underreported in Clinical Trials vs. Real-World Practice

In a compelling talk on impaired hypoglycemia awareness, Dr. Simon Heller emphasized a major gap – rates of severe hypoglycemia have not fallen in clinical practice, despite therapeutic advances. He emphasized the clinical trials like STAR-3 underestimate the current real-world burden of severe hypoglycemia – see the slide below, where the rate of severe hypoglycemia in recent studies is still double that of DCCT’s intensive group and roughly 10x the level seen in STAR-3. In a 2012 Denmark Study and a 2007 UK Hypoglycemia Study, frequency of severe hypoglycemia was more than 1 episode per patient-year, compared to 0.62 in DCCT’s intensive groups. In those two studies, 22%-46% of type 1s were affected by severe hypoglycemia, a staggering number. (Assuming 1/event/patient-year is true in the US, that means people with type 1 diabetes alone experience ~1.5 million severe hypoglycemia episodes per year in the US – to say nothing of type 2 diabetes.) Dr. Heller noted impaired awareness of hypoglycemia affects 20-25% of adults with type 1 and ~10% of people with insulin-treated type 2 diabetes. These individuals have a 3-6 fold greater risk of severe hypoglycemia. Dr. Heller emphasized that for society, hypoglycemia is a much more important problem in type 2, given the greater number of events – we agree and think this point deserves far more attention. He lamented that “rates of severe hypoglycemia in our clinics have not really changed,” though Dr. Heller was optimistic that structured education (e.g., DAFNE), new therapies (insulin analogs), and new technology (CGM, AID) can reduce risk.

4. Kelly Close on Hypoglycemia: Hypoglycemia is a Huge Problem – Let’s Start Answering The Important Questions with CGM

To open the hypoglycemia session, our own Kelly Close shared a wide-ranging perspective on the field (download her slides here and her talking points here): (i) hypoglycemia is still a huge problem worldwide, as shown in global studies like HAT (Khunti et al., Diabetes Obes Metab. 2016); (ii) there are many faces to a 7% A1c – time-in-range gives us critical information about quality of A1c; (iii) recent evidence suggests A1c levels are not related to hypoglycemia (Munshi et al., J Diabetes Complications 2017); and (iv) CGM is THE tool to help us answer meaningful questions related to hypoglycemia. She recapped the compelling real-world data we saw at EASD from the Belgian experience with CGM – 12 months of CGM use dramatically improved days in the hospital for hypoglycemia/DKA (-67%) and days of work absenteeism (-53%). Still, Ms. Close lamented how much is spent on diabetes every hour – over $150 million globally! – and wondered how much of this $1.3 trillion annual expense (Bommer et al., Lancet Diabetes & Endocrinology 2017) could be avoided with smarter investment in CGM, drugs that reduce the risk of hypoglycemia, and better research. She highlighted the dQ&A patient priorities data presented at several venues this year (ADA 92-LB, AADE, Outcomes Beyond A1c Workshop), and posed several research questions for the Beyond A1c movement:

  • Which drugs have the most and least hypoglycemia?
  • Is it possible that time-in-range or hypoglycemia might influence adherence?
  • How much time do insulin users spend in hypoglycemia at work? How does this impact their productivity?
  • Is there a link between hypoglycemia and long-term complications or mortality? What about time-in-range?

5. Severe hypoglycemia and CV Complications – Correlation or Causation? Legends Debate…Audience Sides with Causation

A debate between Prof. Brian Frier (University of Edinburgh) and Prof. Markolf Hanefeld (Technical University Dresden) left most thinking that the association between severe hypoglycemia and macrovascular complications is causal (rather than correlational), despite a lack of bona fide evidence. Unfortunately, at this point, it seems that the field doesn’t have the data to make a judgement call – Prof. Frier suggested that an expensive, labor-intensive prospective study using simultaneous CGM and Holter monitoring for EKG would be the only way to firmly establish causation. (We’d note Dr. Irl Hirsch tried this in the FLAT-SUGAR pilot study, but showing big differences in glycemic variability/hypoglycemia for the same A1c was challenging.) Prof. Frier believes a trial would actually reveal that the answer to the debate probably lies “somewhere in the center ground,” and he acknowledged that many of the large studies that show an association between hypoglycemia and CV mortality are just as likely to show that hypoglycemia is an index of poor cardiovascular prognosis as hypoglycemia being the direct cause of the adverse outcomes. University of Dundee’s Prof. Rory McCrimmon told us separately that groups will be getting together to discuss large trials with CGM, which we’d certainly welcome! We’ve noticed some advocating for a modern-day DCCT with CGM, while others think such a trial would be a waste of time. Would such a trial really move the needle? What would the randomization arms look like? The cost and size of such a long-term study would have to be weighed against the opportunity cost of other things that could be funded. This may be the prime opportunity for a registry study, perhaps in the context of a payer who could link claims and glucose data. Would it not be in a payer’s interest to randomly select a number of patients in its population to wear CGM, and then track hypoglycemia (and time in zone and glycemic variability) in concert with CV events? (Perhaps Dexcom/UHC’s pilot in type 2 will do just this.)

  • Prof. Frier plainly stated that the relationship between hypoglycemia and adverse vascular events is complex and trials thus far haven’t been conclusive, but made the case for causality based on secondary analyses from RCTs and physiological/electrical evidence. On the trial front, he pointed to VADT (severe hypoglycemia was the strongest predictor of CV death, above previous CV event and A1c); ACCORD (patients with severe hypoglycemia had higher mortality, though only in the non-intensive group); the Edinburgh Type 2 Diabetes Study (severe hypoglycemia associated with CV events and mortality); and a six-study meta-analysis from 2013 finding that severe hypoglycemia was strongly associated with higher risk of CVD, and comorbid severe illness couldn’t explain this higher risk. The problem of course is that all of this evidence is totally circumstantial, and one could make a strong argument for occurrence of severe hypoglycemia being a marker for frailty and vulnerability to poor health outcomes. To indirectly bolster his point of view, Prof. Frier went the mechanistic route, pointing out that hypoglycemia has many detrimental effects on vasculature, including endothelial dysfunction, hemodynamic changes/rhythm abnormalities/heart rate variability, blood coagulation abnormalities, and inflammation. These effects, he went on, last for several days, which could explain difficulties in attributing sudden death directly to severe hypoglycemia – it may be setting the patient up for a CV event days later. Further, there is evidence that hypoglycemia is associated with pro-arrhythmogenic changes in EKG. Finally, since many people with diabetes already have established CV disease, Prof. Frier proposed that severe hypoglycemia may often just be the smoking gun that causes death, exacerbating cardia and autonomic dysfunction.
    • Prof. Frier laid out a number of limitations to the available data with respect to searching for a link between hypoglycemia and adverse CV events: (i) Evidence comes from secondary analyses of RCTs recruiting high-risk patients; (ii) Large outcome trials can’t prove causality between hypoglycemia and macrovascular events; (iii) Cardiac events are rare, severe hypoglycemia is uncommon in many individuals, and hypoglycemia is under-reported (he dismisses claims about hypoglycemia and CV outcomes in DEVOTE for this reason).
  • To counter Prof. Frier’s argument, Prof. Hanefeld cited numerous studies showing increased (non)-severe hypoglycemia along with neutral or improved CV outcomes. In the DCCT, there was a three-fold higher incidence of severe hypoglycemia in the intensive arm, but this arm also had 30% CV risk reduction. There were similar stories in UKPDS and ORIGIN (the latter saw neutral effects of CV events despite more mild and severe hypoglycemia). In DPP-4 inhibitor trials SAVOR TIMI 53, EXAMINE, and TECOS, CV outcomes were all neutral despite less hypoglycemia in the DPP-4 arms. In DEVOTE, there was a clear benefit for Tresiba with respect to severe hypoglycemia, but there was no difference in CV outcomes or A1c. Finally, in the ARIC study (of older individuals), in the one year since a hypoglycemic event, risk was actually higher for cancer mortality than it was for coronary heart disease, heart failure, and CV mortality – in other words, severe hypoglycemia may be a sign of declining health, particularly in the elderly.
  • In hallway chatter, University of Dundee’s Prof. Rory McCrimmon told us he believes this debate may have been viewing the relationship too simplistically. First of all, he doesn’t believe that the effects of non-severe hypoglycemia were discussed sufficiently. Second, he thinks that the context within which the hypoglycemia occurs, and overshooting the target range on the recovery from hypoglycemia may be more relevant factors than hypoglycemia itself – a great point! At EASD, he discussed how hypoglycemia may be deleterious only in the context of hyperglycemia, and today he proposed that hyperglycemia after a bout of hypoglycemia may be damaging. He alluded to literature suggesting that people with hypoglycemia but no diabetes (hyperglycemia) actually live longer. One thing’s for sure, without the CGM data, we won’t know for sure the impact of any “real-time” glycemic metric on overall or CV-related mortality.

Diabetes Technology Highlights

1. Adjunctive Dapagliflozin Increases Time in Range by 4.3 Hr/day in 24-Hour Inpatient Dreamed Full Closed Loop Study

A poster presented findings from a double-blinded, randomized controlled cross-over trial (n=15) showing that AZ’s SGLT-2 inhibitor dapagliflozin (DAPA; Farxiga) significantly increased overall and postprandial time in range (70-180 mg/dl) combined with the DreaMed system in full closed loop (FCL) mode. 15 young adults with type 1 diabetes (mean age: 19 years-old; baseline A1c: 8.3%) used the “DreaMed Substance Administration System” (we haven’t heard this branding before) for 24 hours and randomized to receive either 10 mg DAPA or placebo twice. Two unannounced mixed meal tests were performed. Average time in range increased an impressive 4.3 hours/day in the DAPA group compared to the placebo, a notable 18-percentage point boost (68% vs. 50%; p<0.001)! The curve below (pink=placebo, blue=DAPA) depicts statistically significant reductions in postprandial glycemic excursions, decreases in glycemic variability, and a lower mean blood glucose in the DAPA group. These are good outcomes, though since this was full closed loop, the meal excursions were quite big (>250 mg/dl), and the placebo group was at a big disadvantage with the current speed of insulin. Indeed, time in range during the day (7AM-7PM) was 42% with DAPA vs. just 19% for placebo, a ~2.7 hours gain in the DAPA group (p<0.001). There was no significant difference in time in-range overnight, when the system alone kept people in range an impressive average of 91% of the time. However, it is worth noting that mean time in range in the DAPA group was 100%. Time in hyperglycemia (>180 mg/dl) was the big driver of DAPA’s benefit, significantly decreasing by nearly four hours relative to control (29% vs. 45%; p<0.001). No increases in hypoglycemia or serious ketosis were observed – there appears to have been very little hypoglycemia at baseline. Further, total insulin dose decreased by ~30% with DAPA use, from a total daily dose of ~39 units to 28 units. The strong in-patient data led the esteemed investigators to conclude that DAPA is a safe and effective adjunction to FCL, benefiting time in range. We would add that a diminished dose of insulin is very desirable as it can lead to less weight gain, and the SGLT-2 inhibitor can cushion post-prandial excursions on fully closed loop, making up for the speed of insulin. Of course, removing the need to announce meals will also do much to decrease hassle and burden – carbohydrate counting and bolusing still take up headspace for those on hybrid closed loop systems. We’re glad to see such promising results in such a short study and hope to see the data replicated in a longer outpatient study moving forward. Should that happen, we’re also intrigued by commercial opportunities – is there a path forward for a device-pharma partnership to package an SGLT-2 inhibitor and a closed loop system?

2. FreeStyle Libre IMPACT (T1D) MDI Sub-Analysis: ~1.5 hours fewer per day <70 mg/dl

Dr. Raimund Weitgasser presented an in-press sub-analysis from Abbott’s IMPACT RCT of FreeStyle Libre in low-A1c type 1s, demonstrating MDI participants (n=167, or 75% of the full study participants) saw significant reductions in hypoglycemia after 26 weeks using the sensor. He didn’t break out numbers, but based on a figure (below), we estimate that patients in the experimental arm dropped from ~3.5 hours/day below <70 mg/dl at baseline to ~1.9 hours/day <70 mg/dl, a ~46% reduction translating to ~1.5 hours fewer in hypoglycemia every day. The SMBG comparator arm saw no such decrease. Time below 55 mg/dl (~30 min less per day) and 45 mg/dl (~40 min less per day) also shrunk considerably in the MDI cohort using FreeStyle Libre, while remaining stagnant in the SMBG group. Scanning rate was sustained above 15/day for the duration of the study, replacing SMBG almost entirely. These results are in line with the overall study population (pumpers and MDIs; n=221 total), where patients using FreeStyle Libre spent ~1.2 hours fewer per day <70 mg/dl relative to controls, ~49 minutes fewer per day <55 mg/dl (a 50% reduction) and ~33 minutes fewer per day <45 mg/dl. We look forward to seeing more results – A1c, time-in-range, hyperglycemia, data by time of day, etc. – published in Diabetologia soon. At this point, this analysis adds to the evidence base in favor of CGM in type 1s on MDI, along with the Dexcom DIaMonD and GOLD trials, which were published side-by-side in JAMA in January. There’s no question about it – the field is on a roll to combat the historical “pump first, CGM second” mindset. Ultimately, companies need to influence HCPs to change their prescribing, and then influence payers that CGM is a better investment than a pump for many patients.

Big Picture Highlights

Novo Nordisk and IDF announced interim results from their co-sponsored “Taking Diabetes to Heart” survey: One-third of respondents considered their CV risk to be low (“minor” or “marginal”), and one in six have never talked to a healthcare provider about how their type 2 diabetes confers risk for CV disease. The 27-question online survey, open to everyone with type 2 diabetes, found that while 52% of people with diabetes learn about CV disease before diagnosis, 12% never learn about it, 7% learn only when diagnosed with CV disease, and 14% learn several years after diabetes diagnosis. Further, while 27% of patients discuss CV disease with their provider at time of diabetes diagnosis, 16% never have, 5% did several years after diagnosis, and 3% did only when diagnosed with CV disease. These results represent an overall lack of awareness of the overlap between diabetes/CV disease, as well as subpar patients/provider conversations with respect to CV disease and risk factors. In a panel convened to discuss the results, Professor Eduard Montanya (University of Barcelona, Spain) shared that patients in clinic frequently ask about eye problems and weight, but not about CV risk, even though it is most likely to lead to death (indeed, CV morbidity remains the leading cause of mortality for the diabetes patient population, and thought leaders are issuing passionate wake-up calls that diabetes should be considered a CV disease in and of itself). According to Professor Montanya, the focus on microvascular complications is partially responsible for unawareness around CV risk. As we learned in examining the CANVAS results on J&J’s SGLT-2 inhibitor Invokana, there are some complications (e.g. amputations) that affect people more viscerally than others (e.g. the possibility of a future stroke or heart attack). We didn’t find the Novo Nordisk/IDF results entirely surprising, especially in light of data from a similar Lilly/BI-sponsored survey conducted in the US (66% of people with type 2 diabetes were unaware of their elevated risk for CV death). In some ways, the interim results from the global survey are even better than we may have expected, though we recognize the potential for selection bias, with more engaged patients than average (and those with internet access) completing the voluntary exercise. These analyses were based on 943 responses from 32 countries, but 83% of respondents are from Denmark, followed by 6% from the US, 5% from Belgium, and 1% from the UK. That said, statistician Dr. Nafeesa Dhalwani noted that analyses including and excluding Denmark gave very similar results. Data collection will continue through March 2018 (the survey was launched in September 2017), and full results are expected at EASD 2018, followed by a full report in October 2018.

2. IDF President-Elect Prof. Nam Cho Reviews 8th Edition of IDF Atlas; MENA Estimates

  • The latest edition of the IDF Atlassee our coverage here – was presented today in a high-level session kicking off with remarks from the newly-elected IDF President Prof. Nam Cho. Prof. Cho highlighted that the values in the IDF Atlas are still estimates, with gaps in data collection and quality prohibiting the release of more accurate figures. Although he did not provide a specific timeline or action plan, Prof. Cho mentioned that he hopes improvements in diabetes research and epidemiology will eventually lend itself to the Atlas using actual values. We suspect that such developments will require significant energy and time and likely will be reserved as a long-term project – enhanced data capture via EHR and other digital tools are only just starting to gain ground in some high-income countries. How much value would actual figures add, anyway? The most valuable aspects of the Atlas are in showing trends and acting as a periodic thermometer for how the world is doing in diabetes care. Prof. Cho also stressed the major discrepancies that exist in healthcare expenditure related to diabetes, as well as the startling rates of undiagnosed diabetes. As a reminder, one in two people with diabetes is undiagnosed, with a large proportion (85 million people, or 40%) residing in the Western Pacific – the IDF region with the greatest number of people currently living with diabetes. The complete 8th edition of the IDF Atlas is now available in English, Spanish, French, and Arabic for free download here, or see our detailed coverage here. Dr. Alireza Estaghamati (Tehran University of Medical Sciences) provided additional color to estimates for the Middle East and North Africa (MENA) region. MENA is estimated to have the second highest diabetes prevalence following North America. Currently, there are 39 million people living with diabetes in MENA, predicted to skyrocket to 82 million by 2045. While prevalence increases with age, females are more affected across every age group and experience higher rates of mortality. There are likely several reasons for the gender disparity: In addition to cultural factors and education inequity, hyperglycemia in pregnancy (HIP) is a probable contributor. Only three countries (Iran, the UAE, and Qatar) in MENA have data sources for gestational diabetes, estimating 3.8 million live births to be affected by HIP, for which the age-adjusted prevalence is nearly 18% overall and 17.3%, 37.1%, and 25.5% respectively by country. Dr. Estaghamati called for more collaboration, cooperation, strategies, and dedicated, focused plans to combat diabetes in the region.
    • Prevalence of undiagnosed diabetes in the MENA region, estimated to be 49%, is on par with the worldwide average. Major discrepancies exist, with people in Pakistan and Egypt least likely to know they have diabetes. Substantial variation across countries is also seen for impaired glucose tolerance testing, as well as diabetes-related healthcare expenditure, accounting for a staggering $21.3 billion overall or 17% of the total spend in MENA. This level of spend is obviously too high (compared to a diabetes-free world), but it is shockingly just over 6% of what the US spent in 2017 ($348 billion).
  • Algeria, Saudi Arabia, and Morocco are the MENA countries with the highest number of children and adolescents (ages 0-19 years) with type 1 diabetes, totaling 42,500, 35,000, and 31,800, respectively. Interestingly, none of these countries are amongst those with the highest number of adults with diabetes, although Saudi Arabia does have one of the highest age adjusted comparative diabetes prevalences in MENA at 17.7%.

3. Outgoing IDF President Dr. Shaukat Sadikot Urges HCPs to “Come Down from Ivory Towers,” Calls for More Investment in Underserved Areas of the World

Outgoing IDF President Dr. Shaukat Sadikot delivered a compelling keynote address, urging that we shift discussions of diabetes care “from ivory towers to on-the-ground realities.” Drawing upon decades of work in rural, underserved communities in India, Dr. Sadikot suggested that the current discourse around diabetes focuses too much on developing new medicines and updating treatment guidelines, and not enough on action that will directly help people with diabetes – 80% of whom live in low- and middle-income countries where the newest diabetes drugs are out of reach and major treatment guidelines are hardly applicable (this certainly seems to be the case based on an IDF survey of care centers from 90 different countries – see below for our coverage of this). According to Dr. Sadikot, people in this field must understand that “diabetes is not solely a matter of medicine, medicine, medicine,” because there are many other factors that determine health, including socioeconomic conditions and their effect on healthcare access and health literacy. To drive this message home, Dr. Sadikot told a story from his early years of medicine: After hosting a session in a rural Indian village to educate people about diabetes-friendly dietary recommendations, a man approached him and said “Doctor, thank you for telling me what to eat. My problem is that I don’t know whether my family will have anything to eat today.” He surmised, “You’re not a diabetologist unless you understand that it’s about much more than blood glucose.”

  • “We are here to fight a war against diabetes, and PCPs and diabetes educators are on the front lines.” Beyond empowering patients and becoming champions of policies to lessen healthcare disparities, Dr. Sadikot also emphasized the importance of empowering PCPs, nurses, and educators, who provide care to a majority of the world’s people with diabetes. Under Dr. Sadikot’s leadership, the IDF has developed the “Best of IDF” program, which provides free educational content on diabetes to providers. Dr. Sadikot also spearheaded the IDF School of Diabetes, which offers intensive six-month courses in diabetes management tailored for diabetes educators, PCPs, and specialists. Currently, the programs are free to anyone living in a low-income country or working for a government health center.
  • Dr. Sadikot acknowledged that diabetes is one of the most challenging public health issues of our time, deeply entangled with complex issues of poverty and disparities in access to healthcare. Thus, to meaningfully confront the diabetes epidemic will require a more expansive approach than our traditional efforts for scientific advancement on top of already-excellent therapies, he argued. He pointed out that within the past decade, every international organization has come out with academic guidelines and advisories, dozens of new diabetes medications have hit the market, and several diabetes conferences occur every month – and yet the fact remains that the majority of people with diabetes are not at optimal control, and most are not even near optimal control. We have tools to manage diabetes, but what’s missing is widespread education and empowerment, in Dr. Sadikot’s view.
  • This emphasis on the different stories of diabetes in developed vs. developing regions of the world is certainly important and appropriate for an international meeting like IDF. Every sentiment in Dr. Sadikot’s remarks rang true, although we’re not sure everyone would agree that our diabetes toolkit is stocked with already-excellent therapies. In other words, even the most advanced treatments available today are not to the point where they’re lowering the burden of diabetes and maximizing patients’ quality of life (so that they can live as they would without diabetes). Slowing innovation, or investing less in it, is not the answer from our perspective. That said, while manufacturers continue to improve diabetes treatment to be safer, more effective, and easier to use, we also need all stakeholders (including industry) to lend a hand in promoting screening/monitoring (for diabetes, for complications) and education in harder-to-reach pockets of the globe.

4. IDF Survey Finds Marked Disparities in Access to Diabetes Medications Across High-, Middle-, and Low-Income Countries; Gap Between Availability (Is This Drug in the Supply Chain?) and Provision (Will My Payer Pay?)

IDF’s Dr. Belma Malanda illuminated stark disparities in access to diabetes medicines/supplies around the globe by presenting survey data from a spectrum of IDF-member care centers across 90 countries. Availability (as defined by presence in the supply chain) of basal insulin, rapid-acting insulin, metformin, and sulfonylureas exceeded 90% in high-income countries, ranged from 75%-95% in middle-income countries, and ranged from only 40%-50% in low-income countries. Provision of these medicines (whether by a governmental agency, private insurance, or another source) was lower across all geographies, but followed the same expected pattern: 70%-80% in high-income countries, 50%-60% in middle-income countries, and 10%-20% in low-income countries. The gap between availability and provision reflects a worldwide disconnect between the presence of these medicines and access to them. What’s more, this survey didn’t even include advanced therapy classes such as DPP-4 inhibitors, SGLT-2 inhibitors, or GLP-1 agonists, and we can only imagine that provision for these drugs comes in at even lower percentages. Dr. Malanda discussed barriers to insulin access in particular. Lack of diabetes education emerged as the greatest challenge for insulin access in high-income and low-income countries alike, with 45% of the surveyed IDF-member care centers ranking this as a major barrier. Notably, diabetes education is an even greater challenge in low-income countries, ranked as such by 70% of care centers, though this is outmatched by the high cost of insulin, ranked by 80% of surveyed care centers. These statistics are hard to stomach, but Dr. Malanda noted that the reality is probably even worse, since only IDF-member centers were surveyed, leaving out responses from less-engaged areas (even less likely to have diabetes education, availability, or provision). Overall, these findings underscore the sheer magnitude of the systemic barriers preventing people with diabetes from accessing the essential medicines they need. As we continue to grapple with IDF’s latest statistics on the size of the global diabetes epidemic, it is clear that meaningful improvement will require seismic shifts on the health policy front. Dr. Malanda left us with a compelling call-to-action: the first step in addressing disparities is high-quality data, which is the only way we’ll fully understand the scope of this problem to develop precise and maximally-effective strategies for improving access.

5. Google Searches for “Kim Kardashian” ~Infinitely Higher than for “Diabetes Prevention”, but Lower than For “Diabetes”

Imperial College London’s Dr. Josip Car found the lack of patient education and empowerment worldwide to be particularly striking. To this end, he showed a Google search trend (recreated and adapted below) demonstrating substantially (infinitely?) higher and more dynamic interest in “Kim Kardashian” (in blue) compared to “diabetes prevention” (in red) over the past year. Interestingly, a deeper analysis of the Google trends for the term “diabetes prevention” revealed that the most searches derive from Nigeria, followed by the Philippines, India, and the US. Considering diabetes prevalence continues to grow, it’s especially disappointing that interest in prevention has remained so stagnant, though we can’t say we’re surprised that a socialite garners more public interest than something few people are even aware of before it’s too late for them to take advantage. Our own analysis shows that the term “diabetes” (in yellow) is actually quite well-represented in the world’s searches (more so than Ms. Kardashian), and significantly more common than “flu” (green) and “Zika” (purple). This is, in a microcosm, the apex of the problem: Humans are reactive, and only take interest in something when they are directly affected by it. We can’t expect everyone to become a hypochondriac and Google everything that they could possibly be afflicted with every day, but we certainly do need a global shift in mindset toward prevention.

  • Dr. Car was not hopeless regarding prevention, advocating for a population health approach to promote health and prevent disease from cradle to grave – a population level approach is necessary, he said, because 83 countries are already below the WHO’s recommended threshold of 23 skilled healthcare professionals for every 10,000 individuals. The strategy will require comprehensive mobilization of resources, including active participation from schools, organizations, communities, cities, regions, and national governments. Traditional population health interventions focus on leveraging communication and education to: (i) change norms; (ii) establish standards; and (iii) manage outliers. The rise of smartphone use for healthcare may serve to enhance these methods – Dr. Car noted that among smartphone owners, 77% of 18-29 year-olds, 68% of 30-49 year-olds, and 39% of 50+ year-olds used their phone for information about a health condition. Dr. Car believes there’s huge opportunity for real world evidence (RWE) from apps and sensors to help move the needle, yet stipulated that the current promise of technology, particularly smartphones, to improve healthcare is just that – promising, but not yet a reality. He expects the greatest challenge will not be to make big data meaningful, but instead to integrate data streams into routine clinical care.
  • Dr. Car highlighted the importance of app accreditation to provide clinical assurance regarding quality and safety, increasing the likelihood of promotion and adoption by healthcare professionals and patients. He detailed a much-discussed UK study published in BMC Medicine reviewing 46 insulin dose calculator apps, which unfortunately identified only one (unnamed) app as completely issue-free. 44 apps had issues affecting data input, 42 failed to validate numeric inputs, 22 violated “basic clinical assumptions,” and only 14 provided the formula used to suggest the insulin dose. We’re reminded of a talk from this year’s EASD meeting, during which a predominately international panel expressed concern regarding the regulation of mHealth apps. While of course we encourage a healthy dose of skepticism, we’d note that there are a few apps that are delivering outcomes and publishing in peer-reviewed settings, though the majority of these are not just an standalone app now, but an app plus coaching, a connected device, and a subscription package.

 

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