IDF (International Diabetes Federation) Congress

December 4-8, 2017; Abu Dhabi, UAE; Day #2 Highlights – Draft

Executive Highlights

  • On the therapy side, we attended back-to-back symposia with results from the real-world DISCOVER study (offering a global picture on type 2 diabetes care across 38 countries and nearly 16,000 patients) and from the DiRECT trial (which found that calorie restriction and substantial weight loss can produce diabetes remission). Dr. J Hans DeVries confirmed that Novo Nordisk is continuing work on oral insulin to bring down cost of goods (the current formulation required ~350x the amount of insulin as a subcutaneous injection) and that Merck has advanced a new glucose-responsive insulin into phase 1. Although not part of IDF, Novo Nordisk’s once-weekly medicine Ozempic got approved in the US and despite the higher cost to manufacture, will be priced the same as leading weekly GLP-1s.
  • In technology, a highlight of the day was a comparative CGM accuracy study from MGH’s Dr. Steve Russell: Senseonics’ Eversense beat out Dexcom’s G5 and Abbott’s FreeStyle Libre Pro in a six-week outpatient Bionic Pancreas study. Based on these results, Beta Bionics is exploring the option of adding Senseonics’ Eversense to its system, supplementing Dexcom. We also have highlights from our own Adam Brown’s talk on diabetes tools and apps (“What’s New, What Works, and What Do Patients Want”) – get his slides here – along with Dana Lewis’ DIY perspective from a #WeAreNotWaiting debate.

Hello from Day #2 of IDF 2017 in Abu Dhabi! See our top highlights from today below, click here for our day #1 highlights, and click here for our conference preview.

Table of Contents 

Diabetes Therapy Highlights

1. Initial Results from DISCOVER: Mean A1c for Global Type 2 Diabetes Population is 8.4%, Complication Rate High Even in People with Short Duration of Disease; Dr. Kosiborod Points to Signs of Widespread Inertia in Screening and Treatment

An afternoon symposium featured initial results from the DISCOVER study, a massive effort in real-world data collection spanning 38 countries and enrolling 15,992 patients with type 2 diabetes. The primary objective of this program, according to China’s Dr. Linong Ji, is to describe diabetes management patterns and disease evolution over three years for people initiating a second-line glucose-lowering therapy. Dr. Mikhail Kosiborod presented key findings from the program so far (the three years of planned follow-up is ongoing), and he emphasized a few key messages: (i) mean worldwide A1c for the type 2 diabetes population is far above goal; (ii) global complication rate is high, despite this population having a relatively short duration of diabetes (4-7 years, 5.6 years on average); and (iii) there’s evidence for broad treatment inertia, given that people starting a second diabetes drug displayed high A1c and frequent complications. First off, <80% of participants had a recorded A1c assessment, which Dr. Kosiborod described as an important finding in itself, revealing gaps in screening and monitoring. A1c screening rate was lowest in Africa at 57%, followed by Southeast Asia at 61%, ranging up to the Eastern Mediterranean with 94%. When screening was defined as A1c or fasting plasma glucose evaluations, the rate increased, but was still <90% globally. During later Q&A, Dr. Kosiborod pointed out that patients in some parts of the world have to pay out of pocket for screening, delivering this powerful quote: “This has been an eye-opening experience for someone who practices in the US. What we think of as basic measures, this is something that people pay out of pocket for in large parts of the world, so it’s not surprising that we see low rates of screening.” Mean A1c across the entire global study population was 8.4%, from a high of 8.7% in the Eastern Mediterranean to a low of 8.1% in Europe. Nearly one-third of the population had A1c ≥9%, while 55% had A1c ≥8%, and only 15% met an A1c target <7%. This is highly indicative of treatment inertia, Dr. Kosiborod explained, because major guidelines recommend second-line therapy initiation after ~three months not at glycemic goal, and it’s disheartening to see so many individuals waiting until they surpass an A1c of 9% before adding a second agent. The prevalence of microvascular complications (CKD, albuminuria, retinopathy, retinal laser photocoagulation, autonomic neuropathy, peripheral neuropathy, or erectile dysfunction) was 18% globally, ranging from 13% in the Americas to 22% in Southeast Asia – presumably the “real number” is higher and awareness is low – we’re in the midst of seeking out opinions on this. Macrovascular complications were defined broadly (coronary artery disease, stroke, transient ischemic attack, carotid artery stenting, carotid endarterectomy, peripheral artery disease, diabetic foot, amputation, defibrillator use, or heart failure) and appeared in 13% of the overall study population, ranging from 10% in Africa and the Americas to 18% in Europe – these figures struck us as particularly low given that nearly a third of these patients had an A1c over 9% although being relatively newly diagnosed may account for relatively fewer complications. Dr. Kosiborod’s main takeaway from this outcomes data was that a large proportion of patients relatively early in the course of diabetes present with microvascular disease, macrovascular disease, or a combination of both. We already see meaningful implications of this finding: As one example, many thought leaders have argued that diabetes CVOTs are hardly generalizable to the real-world diabetes population because they enroll very high-risk patients, but DISCOVER suggests that the real-world diabetes patient is actually quite likely to have existing micro or macrovascular disease. We also hope that HCPs see them as “hardly generalizable” because it’s difficult to compare CVOTs given different enrollment criteria, etc but we are not sure about this. Dr. Kosiborod listed four correlates for both micro and macrovascular complications: (i) male gender, (ii) older age, (iii) smoking, and (iv) a history of hypoglycemia (we find this last one particularly notable, in the movement for glycemic outcomes beyond A1c). He spoke to the “remarkable opportunity for aggressive risk factor screening and modification early in the disease process.” Indeed, while one objective of this program was to reveal the true global picture on diabetes, the researchers also anticipate many actionable insights, and the need for better screening (of glycemia, of micro and macrovascular risk factors) is one of them. See our detailed discussion and commentary section below for our full coverage of this DISCOVER symposium, also featuring Drs. Kamlesh Khunti (who reviewed baseline characteristics as well as the most common first- and second-line treatment options) and Naveed Sattar (who provided the discussant).

2. DiRECT Trial Sends 46% of Intervention Participants into Diabetes Remission; Intensive Weight Loss Regimen Includes ~800 Calories/Day for 3-5 Months, Cessation of Anti-Diabetes and Anti-Hypertensive Pharmacotherapy

Dr. Mike Lean presented highly-anticipated one-year data from the DiRECT trial (n=298), reporting that 24% of participants receiving intensive weight loss intervention (36 out of 149) achieved ≥15 kg (~33 lbs) weight loss, and that 46% of this treatment arm (68 individuals) achieved diabetes remission. No member of the control group achieved that level of weight loss (p<0.0001 for comparison), though 4% (6 out of 149) did achieve diabetes remission with a mean 5% body weight reduction (p<0.0001 for comparison). In DiRECT, intensive weight loss intervention was a combination of withdrawal of anti-diabetic and anti-hypertensive drugs, total diet replacement with an ~800 calorie/day meal plan for three-five months, stepped food reintroduction for two-eight weeks, and structured support for long-term weight maintenance. Average weight loss for the intervention group was 14.5 kg (~30 lbs) after meal replacement, and average weight regain at 12 months was ~2 kg (~4.4 lbs). In those losing ≥15 kg, a striking 86% achieved diabetes remission. These results have been discussed some time, with talks describing the aims of DiRECT at Diabetes UK 2017 and EASD 2017, and we must say, we did not necessarily expect such encouraging success from the trial – we will be very interested to see longer-term results, of course. An intervention delivered by modestly-trained usual care providers that eliminates current and future need for pharmacotherapy is very intriguing, particularly from a health economic perspective as well as a health equity perspective – we also would’ve been interested in understanding what might’ve been possible with a combination approach with metformin and other easy drugs to prescribe and take that will be going generic over the next decade, like SGLT-2 inhibitors (particularly since some in this class now have established cardioprotection). We’ll be curious to see cost-savings projections for these 149 intensively-treated patients, and Dr. Lean announced that economic analyses are underway. That said, this protocol will be difficult for many to follow – Dr. Lean shared that 32 participants dropped out due to the intervention’s demands, or, presumably, lack of appeal. Ultimately, it only seems applicable in diabetes that has been diagnosed relatively recently – Dr. Roy Taylor shared in discussion that shorter duration of diabetes predicted responders vs. non-responders in a subgroup of DiRECT. Overall, Dr. Lean later clarified that people with longer duration of diabetes who lose a substantial amount of body weight can achieve diabetes remission, although the best results appear in individuals with shorter disease duration. All participants were from the UK, so it’s unclear how calorie reversal would affect diabetes status in more diverse populations (not that the UK isn’t diverse). Two- and three-year data from DiRECT are still being collected, and we look forward to future readouts. All in all, we did find these to be inspiring results – in the words of Dr. Lean, “as good as or even better than results from bariatric surgery.” We don’t expect mainstream diabetes remission clinics to pop up anytime soon, but this does seem to be a fruitful area of research, particularly given increased interest in “intermittent” fasting, etc. Notably, one-year DiRECT findings were simultaneously published in the Lancet online, alongside a positive editorial: “Remission of type 2 diabetes: mission not impossible.

3. Sanofi Symposium Spotlights Head-to-Head Comparisons of Toujeo vs. Tresiba in the BRIGHT Study (First Head-to-Head RCT) and LIGHTNING & DELIVER-D Studies (Part of Company’s Impressive Real-World Evidence Campaign)

A Sanofi-sponsored symposium was rich with real-world data on next-gen basal insulin Toujeo, as speakers dissected the results of the LIGHTNING study and the DELIVER-D study, two retrospective observational analyses comparing the safety/efficacy of Toujeo vs. Novo Nordisk’s Tresiba based on electronic medical records and claims data. The LIGHTNING study (which boasts the title of largest real-world comparative study in diabetes to-date) evaluated EMR data from April 2015 to December 2016 from >130,000 adults with type 2 diabetes who switched from one basal insulin to another. In this real-world setting, rates of severe hypoglycemia were significantly lower in participants who switched to Toujeo vs. those switching to Lantus (3.6 vs. 9.7 events/100 patient-years, p=0.009) or Levemir (3.6 vs. 15.1 events/100 patient-years, p=0.002), and were comparable vs. those switching to Tresiba (3.4 vs. 5.3 events/100 patient-years, p=0.37). The risk reduction for severe hypoglycemia with Toujeo vs. Lantus and Levemir occurred without any compromise to glycemia, as A1c did not change significantly after any basal insulin switch. Notably, this is just the tip of the iceberg for the LIGHTNING project (we first heard about this exciting endeavor at EASD), which is expected to produce more data over the coming months utilizing machine learning techniques to define the patient segments in which Toujeo provides the greatest hypoglycemia differentiation vs. comparators, and to evaluate the medical cost-savings for these populations. Similarly, the DELIVER-D study evaluated data from a different EMR source, evaluating 2,893 people with type 2 diabetes who switched their basal insulin prescription to Toujeo or Tresiba from March 2015 to December 2016. After adjusting for baseline hypoglycemia, the overall hypoglycemia event rate was equivalent between Toujeo and Tresiba arms (0.50 vs. 0.51 events per person per year, p=0.88), as was the event rate for hypoglycemia associated with hospitalization (0.17 vs. 0.18 events per person per year, p=0.82). Furthermore, both Toujeo and Tresiba produced a significant 0.5% improvement in A1c over six weeks of follow-up, and participants on Toujeo and Tresiba were equally likely to attain the A1c targets of <7% (13% vs. 16%, p=0.24) and <8% (44% vs. 45%, p=0.92). Presumably, these real-world findings bode well for Sanofi and for the Toujeo business. Still, we can’t imagine DELIVER-D would have the same commercial influence as a hypoglycemia claim for Tresiba (approved in the EU, decision on US label expected in 1Q18) based on the RCT DEVOTE. Still, the health economic analysis from LIGHTNING could well have substantial impact, unveiling cost-savings in specific patient segments, which should be appealing to payers.

  • To complement this real-world data, the symposium also covered the recent BRIGHT study, the first RCT to compare the next-generation basal insulins Toujeo and Tresiba head-to-head; full results are expected in 2018. The study randomized people with type 2 diabetes (n=929) to Toujeo or Tresiba for 24 weeks; all participants were naïve to insulin therapy with an A1c above goal on oral agents or a GLP-1 agonist. Sanofi also announced topline findings in a press release yesterday, underscoring that the BRIGHT study met its primary endpoint of similar A1c reductions with Toujeo vs. Tresiba. Secondary endpoints include the percentage of participants experiencing adverse events, the rate of hypoglycemia episodes, and a variety of patient-reported outcomes from the Diabetes Treatment Satisfaction Questionnaire (DTSQ). We have long speculated about how the two next-generation insulins stack up against one another, so needless to say we are eagerly awaiting the readout of this first head-to-head RCT. Still, we must keep in mind that this is a Sanofi-sponsored trial, so we’ll have our eyes peeled for any elements of study design that might favor Toujeo over Tresiba or the opposite. Our interest is especially piqued for the secondary endpoint of hypoglycemia: To-date, Tresiba appears to have the edge on this front, boasting an impressive hypoglycemia benefit vs. Lantus in both the DEVOTE trial and the SWITCH 1 and 2 trials, but it is impossible to be sure since Tresiba and Toujeo have never been directly compared in the same RCT. From our view, it’s critically important not to miss the forest for the trees we believe both insulins are much better than predecessors Lantus and Levemir and both would be major advantages.

4. Updates on the Insulin Pipeline: Novo Nordisk Hasn’t Given Up on Oral Insulin, Merck Forging Ahead with Glucose-Responsive Insulin

Dr. J Hans DeVries shared exciting details on Novo Nordisk’s oral insulin and on Merck’s glucose-responsive insulin, suggesting that both represent promising future innovations in diabetes therapy. Shortly after Novo Nordisk discontinued its oral insulin candidate 338 in 3Q16, management attributed this decision not to lack of efficacy in phase 2 trials, but to cost of goods. Dr. DeVries confirmed this in his presentation, showing how you need ~350x the amount of insulin in an oral capsule to achieve the same pharmacokinetics as a subcutaneous insulin injection (absorption is significantly slower through the GI tract – still, this surprised us). He highlighted key advantages of oral insulin, which are pretty obvious – namely, that oral administration is more patient-friendly than injectable administration, and that it’s actually more physiologic since subcutaneous insulin bypasses hepatic clearance and therefore presents at much higher concentrations in the blood – but underscored that these benefits come at a high price for the manufacturer. On the (very) bright side, Dr. DeVries announced that Novo Nordisk is likely working to make the absorption process more efficient to reduce ~350x to a more reasonable level, and a more reasonable cost of goods (this was also unsurprising given that otherwise it would be impossible to commercialize). Overall, we’ve gathered from management at the company that they haven’t shut the door on oral insulin despite the 3Q16 discontinuation, given how transformative this could be for people with diabetes, and we were thrilled to hear further assurance of this from Dr. DeVries. He also reviewed preclinical results on Merck’s glucose-responsive insulin (GRI) candidate, presented on a poster at ADA 2017 (study in eight dogs). Notably, Merck discontinued its phase 1 glucose-responsive insulin MK-2640 in 2Q17, and this data was on another agent which doesn’t yet appear on the company’s pipeline page. According to Dr. DeVries, however, Merck has initiated phase 1 clinical studies of this other GRI, which will hopefully be presented at a scientific meeting in 2018. This was a second piece of good news on the diabetes pipeline, since we weren’t sure if Merck would invest further in glucose-responsive insulin following the discontinuation of MK-2640. Both oral and glucose-responsive insulin are challenging therapeutic propositions, but the payoff of these drugs could be tremendous, reducing injection burden (and daily burden of diabetes), improving quality of life, and enhancing the overall quality of diabetes care available to patients in the real world. These candidates are early-stage, and it’s important to note that neither actually appears in the public-facing pipeline from Novo Nordisk or Merck, so we’ll try to temper our enthusiasm for now.

5. Dr. Vilsbøll Distinguishes Long-Acting vs. Short-Acting GLP-1s on CV Benefit; Advocates for Widespread Use of Long-Acting GLP-1 Agents Based on NNT; Shares Evidence for Anti-Atherosclerotic Mechanism of Cardioprotection

The great Dr. Tina Vilsbøll argued that a GLP-1 agonist needs to be continuously available/long-acting to have a significant impact on CV disease, which could explain the mix of positive and neutral CVOT results for the class. Sanofi’s Adlyxin (lixisenatide) is short-acting – and the agent showed no CV benefit in ELIXA – while Novo Nordisk’s liraglutide (Victoza) and semaglutide (Ozempic; just FDA-approved!) are long-acting, and demonstrated CV efficacy in LEADER and SUSTAIN 6 respectively. The REWIND study for Lilly’s Trulicity (dulaglutide) will shed more light on this hypothesis (expected to complete in July 2018), as dulaglutide is another long-acting GLP-1. Interestingly, Novo Nordisk CSO Dr. Mads Thomsen has speculated that REWIND will report neutral findings on the primary outcome of three-point MACE because of the low baseline A1c (mean ~7.3%), relatively short duration of diabetes (mean ~10 years), and large size primary prevention cohort (only ~31% of study participants had established CV disease at baseline). That said, if Dr. Vilsbøll’s proposed explanation is correct, might we see significant risk reduction in the secondary prevention cohort of REWIND on par with LEADER and SUSTAIN 6? Dr. Vilsbøll made clear that there are key molecular differences between liraglutide, semaglutide, and dulaglutide, even though they all have greater homology to human GLP-1 vs. exendin-4-based exenatide and lixisenatide. Novo Nordisk has been promoting this molecular variation between human GLP-1-based agents and exendin-4-based agents for some time, and while this opinion has generated a fair amount of controversy in the field (with some thought leaders downright rejecting it), there is inherent logic in Dr. Vilsbøll’s argument: You should get more benefit when a drug is active all the time. To Dr. Thomsen’s point, trial design is clearly an important factor in whether a CVOT will demonstrate benefit. Dr. Vilsbøll noted that EXSCEL for AZ’s Bydureon (exenatide once-weekly) met its primary endpoint when patients in the placebo group taking SGLT-2 inhibitors were removed from analyses, but added that this was not pre-specified, which needs to be taken into account in discussing the results. Another valid claim, though we do think it’s meaningful that EXSCEL might have shown CV efficacy with some tweaks to the CVOT design, as this keeps very open the possibility of a cardioprotective class effect for GLP-1 agonists. To this end, we should also mention that neutral trials like ELIXA and FREEDOM-CVO for Intarcia’s ITCA 650 were designed purely as safety studies.

  • Dr. Vilsbøll made a compelling case for widespread use of long-acting GLP-1 agonists with number needed to treat (NNT) data. She shared that the NNT to prevent one CV death, non-fatal MI, or non-fatal stroke was 66 patients for three years with liraglutide and 45 patients for two years with semaglutide. While these numbers may seem high, Dr. Vilsbøll aptly pointed out that the NNTs for statins and ACE inhibitors are actually higher, but those medications are still prescribed nearly ubiquitously.
  • Dr. Vilsbøll also presented preclinical evidence supporting an anti-atherosclerotic mechanism for the CV benefit seen with GLP-1 agonists. In mouse models, liraglutide has been shown to inhibit early atherosclerotic lesion development, and both liraglutide and semaglutide have been shown to reduce aortic plaque lesion area in mice on a high fat, high sugar diet. While these experiments could never be done in humans, observational data can provide marked insight. In a study of atherosclerotic plaques collected during carotid endarterectomy from patients without diabetes (n=30), with diabetes who have never used incretins (n=28), and with diabetes who were current incretin users (n=24), significant differences were seen in plaque stability. Inflammation (TNF-alpha) and oxidative stress (nitrotyrosine) were significantly reduced in current incretin users vs. never users, though both groups were higher than controls. Current incretin users also had significantly higher collagen content than never users; both were lower than controls. While this is a small, non-randomized, non-controlled group, it’s not the first suggestion we’ve heard for the anti-atherosclerotic effects of GLP-1 agonist products, and it contributes valuable mechanistic knowledge to our understanding of this class and its groundbreaking CV benefits.  

6. Should SGLT-2 Inhibitors Be First-Line Therapy for Diabetes? Drs. Scheen & Schernthaner Sound Off

In front of a packed room, Drs. André Scheen and Guntram Schernthaner debated a provocative question currently facing the diabetes field: Should SGLT-2 inhibitors be first-line therapy? Dr. Scheen spoke first, taking the “pro” stance, highlighting the clinical benefits to SGLT-2 inhibitors, the limitations of metformin (to say nothing of SFU’s!), and the limitations of incretin-based drugs (that part was more surprising). Dr. Schernthaner followed, arguing that there’s insufficient evidence for SGLT-2 inhibitors to replace metformin as the first-line therapy recommendation in diabetes treatment guidelines, though he supported the use of SGLT-2 agents as second-line therapy for people with established CV disease.

  • Yes: Dr. Scheen started his answer with a question of his own – what are the main objectives in treating type 2 diabetes? In the age of diabetes CVOTs, there’s certainly been a shift toward greater emphasis on outcomes, on micro and macrovascular risk reduction. Lowering A1c is one component of this, but a glucose-centric view of diabetes is no longer acceptable among thought leaders, Dr. Scheen asserted and he posited that SGLT-2 inhibitors show the greatest efficacy in micro and macrovascular risk reduction. There is strong evidence from EMPA-REG OUTCOME for Lilly/BI’s Jardiance (empagliflozin), from CANVAS for J&J’s Invokana (canagliflozin), and from CVD-REAL, the AZ-sponsored real-world study evaluating outcomes with all SGLT-2 inhibitors, including Farxiga (dapagliflozin). Dr. Scheen speculated that the class may show benefits in primary CV prevention as well, based on the fact that only 13% of CVD-REAL participants had established CV disease at baseline. A recent post-hoc analysis of CANVAS also found that heart failure and renal outcomes were reduced with canagliflozin in the study’s primary prevention cohort. Moreover, SGLT-2 agents have demonstrated profound weight reduction and blood pressure-lowering. Dr. Scheen alluded to the common safety argument in defense of metformin as first-line (i.e. this agent has been on the market for ~60 years and its safety/tolerability is thus well-established), but suggested that SGLT-2 inhibitors are very safe as well, with a manageable risk for genital mycotic infections in females (he didn’t mention the CANVAS amputation signal, which Dr. Schernthaner picked up on during his remarks). Metformin’s positioning as first-line treatment for diabetes is largely based on UKPDS, Dr. Scheen explained, which should be challenged by the current data on SGLT-2 inhibitors. These agents weren’t on the market at the time of UKPDS, but have churned out impressive outcomes data in the past couple years. He also pointed out that baseline A1c in UKPDS was relatively low, at 7.2%, which isn’t the case for the average real-world patient today – and a higher starting A1c is all the more reason to use a more effective therapy as first-line. Dr. Scheen acknowledged the powerful efficacy of GLP-1 agonists, but highlighted their ~three-fold cost vs. SGLT-2 inhibitors as well as their injectable administration, which would deter some people from starting on a GLP-1 as their initial medicine for diabetes.
  • No: Dr. Schernthaner refuted that SGLT-2 inhibitors show superior glycemic effects vs. metformin in all cases, and in fact, he called into question the durability of A1c-lowering with SGLT-2 agents. He pointed out that real-world treatment discontinuation is more common for SGLT-2 inhibitors (25%) compared to metformin (only 8%), GLP-1 agonists (20%), and DPP-4 inhibitors (17%), which suggests safety/tolerability concerns, from genital mycotic infections, to dehydration, to DKA, bone fractures, and amputations. Notably, the FDA has only issued a black box warning for lower limb amputations on the Invokana label, but the EMA has applied the same warning across the class. Cost was Dr. Schernthaner’s most compelling argument, in our view: The price of generic metformin is <$50/day vs. ~$4,800/day for an SGLT-2 inhibitor and ~$9,300/day for a GLP-1 agonist. Still, with SGLT-2s going generic in the next decade and DPP-4s even sooner, we don’t think cost should be the primary determinant, particularly since the cost of the drugs is dwarfed by the costs of complications they could avoid. As we’ve learned in our explorations of why sulfonylureas still appear in treatment algorithms, guidelines cannot ignore cost considerations, since the vast majority of type 2 diabetes patients face some sort of access/affordability issues in their medical care. To this end, we’d love to see a cost-savings analysis on first-line intervention with SGLT-2 inhibitors, because annual hospitalizations for heart failure also run an astronomical bill. We appreciated Dr. Schernthaner’s balanced view (he even highlighted the increased longevity associated with metformin in some studies of people without diabetes), and his ringing endorsement of SGLT-2 inhibitors as second-line treatment in patients with established CV disease. We’re glad to hear that SGLT-2 therapy for CV protection is not as controversial as the first-line debate, considering the promising support for a class effect from EMPA-REG, CANVAS, and CVD-REAL, not to mention the high unmet need in addressing residual CV risk in people with diabetes.

7. Dr. Ghosal on Basal Insulin/GLP-1 Combos: ~80% of Participants on Xultophy Reach A1c <7% without Weight Gain or Hypoglycemia – “Not Something We’ve Been Able to Achieve with Traditional Molecules” – Highlights Impressive DUAL VII Results Comparing Xultophy vs. Basal-Bolus

Dr. Samit Ghosal captured the benefits of basal insulin/GLP-1 combination therapy by focusing on the composite endpoint of proportion of patients who reach A1c goal <7% without weight gain or hypoglycemia. A meta-analysis published in 2012 (by first author Dr. Bernard Zinman) found that 40% of patients (n=1,513) randomized to 1.8 mg liraglutide (Novo Nordisk’s GLP-1 agonist Victoza) achieved this metabolic composite vs. 33% of patients (n=1,077) on 1.2 mg liraglutide, 15% of patients (n=225) on basal insulin glargine (Sanofi’s Lantus), and only 8% of patients randomized to placebo (n=505). Summarizing data from the DUAL program on Novo Nordisk’s Xultophy (insulin degludec/liraglutide fixed-ratio combination), Dr. Ghosal showed how 80% of participants randomized to the combination product achieved A1c <7% without weight gain or hypoglycemia – he added that this is “not something we’ve been able to achieve with traditional molecules.” This metabolic composite has clear clinical relevance, as glucose control, weight control, and risk reduction for hypoglycemia are all key components of diabetes management. Dr. Ghosal cited ADA, EASD, and AACE guidelines, all of which recommend treating to target A1c while also prioritizing weight loss (or at the very least, avoiding weight gain) and minimizing hypoglycemia. Moreover, weight gain and fear of hypoglycemia are common factors contributing to low medication adherence, to treatment dissatisfaction, and to poor quality of life for people with diabetes. By establishing the value of this composite endpoint, and by demonstrating that we now have an emerging therapy class that shows greater benefit on this endpoint than any drug that has come before, Dr. Ghosal made an extremely compelling case for basal insulin/GLP-1 fixed-ratio combination products (the class also includes Sanofi’s Soliqua, a fixed-ratio of insulin glargine/lixisenatide). By and large, thought leaders have expressed enthusiasm for Xultophy and Soliqua, and Dr. John Buse went so far as to suggest that Novo Nordisk’s IDegLira may be “the most effective anti-hyperglycemic agent on the planet.” Still, commercial uptake of these products has been sluggish to-date (mainly due to perceptions of combos and Novo Nordisk’s decision to price the basal/GLP-1 combo very high and to emphasize the specific components of Tresiba and Victoza more than the combo), and we can only hope that improved reimbursement and greater HCP familiarity with fixed-ratio combinations leads to a steeper incline in volume/sales for the class in 2018. Dr. Ghosal emphasized the cost issue, explaining that >90% of his patients in India pay out of pocket for drugs, which keeps basal insulin/GLP-1 combinations (or even co-administration of the two agents separately) out of reach.

  • Dr. Ghosal called DUAL VII, comparing Xultophy vs. basal-bolus therapy (insulin glargine/insulin aspart), “the most ambitious study” in this clinical program, and then proceeded to review the positive results. Novo Nordisk simultaneously issued a press release highlighting this data. Full results from DUAL VII were a major highlight at this year’s ADA Scientific Sessions, when Dr. Liana Billings reported non-inferior A1c-lowering, half-size total daily insulin dose, greater weight loss, and less hypoglycemia with Xultophy vs. basal-bolus. At IDF, Dr. Ghosal specifically called out the “practically negligible” GI side-effects with Xultophy compared to a standalone GLP-1 agonist. Patient-reported outcomes from the trial also favored Xultophy over a basal-bolus regimen: After 26 weeks of the study, 85% of people in the Xultophy arm were “very” or “extremely” willing to continue treatment vs. 68% of people in the basal-bolus arm, and mean improvement on diabetes management, treatment burden, and compliance (quantified by the Treatment-Related Impact Measure-Diabetes questionnaire and the Short Form Health Survey 36 v2) was significantly greater with Xultophy vs. basal-bolus therapy.

8. Basic Science Galore: Impact of AgRP vs. POMC Neurons on Body Adiposity; Age-Dependent Increases in Body Weight; Hypothalamic Inflammatory Pathways

UCSF’s Dr. Allison Xu and University of Ulsan College of Medicine’s Dr. Min-Seon Kim presented riveting data investigating the role of the brain in metabolic regulation. While still very early-stage, these exciting developments may point to new therapies, promising highly specific drug targets. This important research also underscores the substantial complexity of obesity ­–  as Dr. Xu noted, our bodies fiercely defend an adiposity set point via myriad pathways, making it extremely difficult to lose weight. Read on for some of our top takeaways of the session.

  • Dr. Xu presented fascinating data on the cellular mechanisms underlying body adiposity, highlighting AgRP neurons as likely first responders to short-term metabolic changes, and POMC neurons as engaging in long-term energy homeostasis alterations. Following consumption of a high-fat diet, changes in AgRP neurons were quickly exhibited in mice, with severely reduced leptin signaling observed at just two days; POMC neurons took longer to reveal an impact, demonstrating only moderate decreased signaling (albeit significant) at two weeks. As a reminder, leptin is a key hormone in metabolic homeostasis that promotes feelings of fullness. Here, the brain is shown to ignore these signals in response to a high-fat diet, prompting the mice to continue eating. 
  • Dr. Xu also detailed pathways contributing to age-dependent increases in body adiposity. In mice, both body fat and leptin levels increase with age under normal feeding conditions. This struck Dr. Xu as odd –if leptin is increasing, which ostensibly signals the individual to feel full, shouldn’t body fat decrease? Dr. Xu reasoned there must be some counter-regulatory mechanism antagonizing leptin’s anorexigenic action. Interestingly, in middle-age mice, AgRP neurons are observed to concentrate to specific targets in the brain, a process known as innervation. POMC neurons are among these targets of age-specific AgRP innervation, and as a result are shown to decrease firing frequency. The same phenomenon occurs in young mice on a high-fat diet, suggesting diet-induced obesity to operate via a similar pathway as age-dependent adiposity.
  • Dr. Kim discussed how microglia, astrocytes, and macrophages contribute to hypothalamic inflammation, known to be associated with diet-induced obesity by resulting in impairment of leptin and insulin signaling. Microglia are rapidly activated in the hypothalamic arcuate nucleus following a high-fat diet, suppressing POMC neuronal function. Inhibiting microglia action prevents weight gain in mice on a high-fat diet, strongly suggesting their role in diet-induced obesity. While microglia show increases at three days due to a high-fat diet, macrophages are also present at higher concentrations in the arcuate nucleus of diet-induced obese mice, but lag slightly behind, taking roughly two weeks to exhibit enhanced proliferation. Astrocytes increase quickly in the hypothalamus in response to a high-fat diet. By suppressing the macrophage response in diet-induced obese mice, hypothalamic leptin resistance is improved, resulting in decreased food intake and body weight.

Diabetes Technology Highlights

1. Bionic Pancreas + Senseonics’ Eversense? MGH & Beta Bionics Exploring Additional Sensor Option Based on Head-to-Head Accuracy vs. G5, Libre Pro; Positivity on Eversense from Dr. Russell

In a Roche symposium, MGH’s Dr. Steven Russell shared new comparative accuracy data from a six-week outpatient Bionic Pancreas study. Each patient wore Senseonics’ Eversense, Dexcom’s G5, and Abbott’s FreeStyle Libre Pro. Eversense recorded an MARD of 14.8% vs. 16.8% for Dexcom’s G5 vs. 18% for Abbott’s FreeStyle Libre Pro – each patient wore all three sensors, with CGM values compared to fingersticks taken via Nova StatStrip Xpress (n=829 paired points; p<0.008 vs. Eversense). When data was only compared with five-minute sample intervals, Eversense had a similar 15.1% MARD vs. 16.9% for G5 (n=2,277; p<0.0001 vs. Eversense). Based on this strong Eversense performance – it was statistically superior to G5 and Libre Pro – MGH and Beta Bionics are now “exploring” the device as an “additional sensor option,” supplementing Dexcom. FDA discussions are just beginning, but the hope is to include Dexcom’s G6 and Senseonics’ Eversense in the Bionic Pancreas pivotal trial (per the last update, this is expected to start at the beginning of 2019). If it comes to fruition, Beta Bionics could be the only AID system that offers two different companies’ sensors. 

  • This “Monitoring Study” will be presented as an oral at ATTD 2018 – the trial actually tested the Bionic Pancreas in both insulin-only and bihormonal configurations with vs. without remote monitoring. (This sensor comparison was a side benefit.) The team didn’t have access to FreeStyle Libre real-time for this study, which is why Libre Pro was used. Per the US label, Libre real time has slightly better accuracy (MARD: 9.7%) than Libre Pro (MARD: 12%), as the real-time version has a 12-hour warmup. For a six-week study, we assume the number of paired points was so low because fingersticks were only taken a couple times per day for device calibration – i.e., not for frequent sampling like in a traditional CGM accuracy study.
  • Beyond the head-to-head comparisons, the other obvious takeaway is how far real-world CGM performance can differ from clinical trial performance – Eversense had an 8.8% MARD in its US pivotal vs. 14.8% here; G5 has a 9% MARD in its label vs. 16.8% here; and FreeStyle Libre Pro has 12% in its label vs. 18% here. This presumably reflects dirty fingers and other real-world issues that may be avoided in best-case-scenario accuracy studies.
  • It was notable to hear Dr. Russell’s positivity on Eversense – he highlighted the accuracy, no acetaminophen interference (Dexcom will have this fixed soon but doesn’t yet), adhesive/wear flexibility, and simple outpatient insertion procedure. Dr. Russell said he has inserted the second-most Eversense sensors in the world (according to Senseonics), and once he has the skin site scrubbed and draped, it takes him “less than two minutes” to implant a sensor. (As a reminder, it requires lidocaine, a small incision in the upper arm, and is closed without sutures using a Steri Strip bandage. It can be done in an office visit.) Dr. Russell likes that the Eversense on-body transmitter adhesive can be changed every day, a plus for patients plagued by skin reactions or adhesive failures. Moreover, since the Senseonics transmitter doesn’t actually have to be adhered to the skin – it just has to be over the sensor implantation site – it is possible to hold it in place with a “sleeve” and not use any adhesive at all. We hadn’t realized this point and like the idea! Dr. Russell concluded that Eversense will be another great CGM option on the US market, pending approval – as a reminder, an FDA advisory committee is expected in 1Q18 (see Senseonics 3Q17). In Europe, Roche and TypeZero are using the 180-day version of Eversense in their AID system, expected to start its pivotal study in 2018.

2. Adam Brown’s Three Tech “Cs” That Are Delivering Outcomes and/or Reducing Burden: CGM, Clever Insulin Delivery, and Coaching/Remote Care

Our own Adam Brown reviewed the latest in diabetes tools and apps, condensing areas of promise into three Cs: CGM, Clever insulin delivery, and Coaching/remote care. Download a PDF of his slides here, and see the bullets below for more depth on each of the areas, as well as Adam’s six-tem shopping list of where diabetes tech still needs to improve. Adam grounded his presentation with a couple visual examples to demonstrate what a great diabetes tool or app might do – (i) turn a mountainous, curvy road on a foggy day into a flat, straightway road on a sunny day; and (ii) turn that feeling of blind floundering in the darkness to confidently walking down a well-lit path. In short, he said, diabetes tools and app will add value to patients if they deliver outcomes and/or reduce burden – ideally, they will do both. By both outcomes and burden, Adam emphasized this isn’t just glycemia (A1c, time-in-range), but a broader box that might include weight loss, mindset, energy, better sleep, cost, time, finger pokes/injections, and frustrating conversations. He noted that, while there has been a steady progression toward better outcomes thanks to better diabetes technology, they have often come with more burden (e.g., pumps and CGMs). “I hope we’re entering the era of better outcomes with increasingly less burden.” Amen!

  • Adam spent most of his talk on CGM’s increasing evidence base, user base, and published outcomes data. He showed strong adoption curves for Dexcom CGM (sales) and Abbott’s FreeStyle Libre (users), and covered recent studies showing strong health and cost benefits – but estimated that <0.5% of people with diagnosed diabetes globally are using CGM. Assuming roughly 15% of people with diagnosed diabetes are on insulin worldwide, he continued, then just ~2-3% of insulin users are using CGM. (For the initial calculation, Adam estimated from recent Abbott and Dexcom user base updates and Medtronic sales data that ~0.7-1.0 million people are on CGM – the denominator, 212 million people diagnosed worldwide, came from the 8th IDF Atlas.) Strikingly, this <0.5% penetration says nothing of the 212 million undiagnosed and 352 million people with impaired glucose tolerance, all of whom could potentially benefit from CGM as well. CGM clearly has a massive runway, which Adam characterized as “both exciting and daunting.” With an eye toward access and affordability, he suggested that professional CGM (FreeStyle Libre Pro) is a great option in lower resource settings, and that the smaller/less costly Dexcom/Verily sensors (in-development) “will really expand the market in a big way.” With data supporting clinical and economic outcomes, burden and price coming down, growing payer interest (see remote care below), and data getting easier to understand (with AI, such as Medtronic/IBM Watson’s Sugar.IQ), adoption should continue to improve.
    • To support the health and financial benefits of CGM, Adam pointed to a slew of recently published or presented data: (i) Abbott’s IMPACT and REPLACE studies, plus real-world data from now 200,000+ users (DTM); (ii) Dexcom’s DIaMonD and GOLD studies (here and here); (iii) A Belgian reimbursement study from EASD demonstrating strong decreases in hospitalizations and work absenteeism with CGM; (iv) the CONCEPTT study of CGM in pregnancy – the first study to indicate potential for improvements in non-glycemic health outcomes from CGM use,” in this case costly neonatal outcomes such as lengthy NICU admissions and neonatal hypoglycemia; and (v) the IN CONTROL study showing CGM reduces incidence of severe hypoglycemia in people with hypoglycemia unawareness.
  • Within “clever insulin delivery,” Adam (i) exclaimed that he is “really excited” for smart pens to provide injection data; (ii) shared enthusiasm for the power of insulin titration to continuously update doses; and (iii) highlighted a slew of commercial closed loop systems in development. For each, he displayed slides with graphic versions of Close Concerns’ competitive landscapes – smart pens/pen attachments (led by Companion Medical, which is set to launch this month in the US); insulin dose titration; and automated insulin delivery. After alluding to a Common Sensing poster from ADA with overlaid CGM and injection data, Adam noted that smart injection devices are not widely available at the moment, but “could be even bigger than closed loop with pump and CGM.” No doubt that connected pens and decision support will be more prevalent, and we look forward to finding out how far smarter open loop can go. How will patients segment between MDI+advice and pump+CGM hybrid closed loop? Will anyone run a head-to-head study once commercial systems are out? (Bigfoot/Abbott and Lilly/Dexcom both plan to have a system in each category.)
  • The big question with coaching and remote care, according to Adam, is can it scale? We’re not sure yet. He discussed some of the solid outcomes demonstrated by Livongo, mySugr, One Drop, and Virta, cautioning that much of the data is from observational, real-world studies where there is not control group. He argued that we DO need these studies, though payers will likely want to see real-world data and RCTs (more in Q&A). Do the reductions translate to the broader population? Can these interventions scale? Crucially, can they get reimbursed? One good sign, Adam said, is that payers are getting interested – in just the past couple of weeks, Dexcom and UnitedHealth Group announced an unprecedented pilot of CGM and coaching in 10,000 type 2s, Onduo and Blue Cross Blue Shield announced a pilot of Onduo’s (“Virtual Care Clinic”) in three southern states, and Sanofi and Innovation Health (Aetna+Inova Health Systems) launched a pilot with Common Sensing and One Drop. Do these payer-based coaching setups have a better chance of reaching more people? How will the companies support the needs of the patients? How does Dexcom/a UnitedHealth coach interact with the provider that a patient has traditionally seen? There are a lot of questions in this nascent area, but closing the gaps between infrequent clinic visits is seemingly a no-brainer success; at this point, it’s about establishing successful workflows, incentive structures, risk-stratification, and program tailoring.
  • Despite progress in the three Cs, Adam identified six areas in which we still need massive innovation: (i) Cost and access globally – most diabetes tech is limited to developed countries; (ii) more simplicity, seamless, “it just works – magic!” experiences; (iii) More medication dosing guidance including oral medications – e.g., “this patient should take a higher metformin dose” or “this one should add a GLP-1 agonist to her regimen”; (iv) more implementation in different systems and patient populations (especially type 2s); (v) Improving providers’ efficiency, scaling their  expertise and limiting administrative hassle; and (vi) designing for more delightful, bright spots moments “where you’re like ‘Yeah – awesome!’” Adam related a story about how he had his glasses prescription refilled at home using the brilliantly innovative Warby Parker Prescription Check app –  the app interfaced with his computer, guided him with videos, helped him measure the distance, and allowed him to complete an eye-chart-like test by swiping on his phone. Those “WOW, this is awesome!” moments rarely happen in diabetes tech, but they could!
  • Drawn in by the “you might also like” section of the online Wall Street Journal, Adam recently stumbled upon an article about drones delivering blood to remote locations in Rwanda by a startup called Zipline. With this system, delivering blood takes 10-15 minutes, whereas it used to take three hours. What if we used this technology to deliver insulin or diabetes supplies? On day #1 of the conference, we chatted with a doctor now based in Fiji, who told us that people were scattered across 100 islands in the nation and many times without refrigeration … Brian pointed out that drone delivery would make a huge difference, especially in extreme cases like Fiji! Adam wondered how other mainstream consumer tech could similarly be leveraged in diabetes to great effect (e.g., 3-D printing, VR, etc.).

Questions and Answers

Q: Lot of things need to be done before we progress in diabetes technology – evidence generation, reimbursement issues…the results you showed are often not enough. With observational studies, there are lots of confounders. How do we collect more evidence in medicine?

Adam: We could do a whole session on that. There are a couple challenges in tech. CGM was plagued for a number of years because studies took so long to do that devices were outdated by the time the study was published. It colored the field a lot. Now we’re starting to see a mix of RCTs that take time to do but are published (DiaMonD, GOLD, IMPACT, REPLACE), and real-world data. That may be one reason why diabetes technology is starting to see more reimbursement, especially in the EU – we’re seeing this nice mix of both data. It’s a really good sign to me, a canary in the coal mine (in a good way), that payers are now testing CGM and remote care and insulin titration in pilots. Reimbursement will naturally always lag behind new stuff; I think that’s okay, but it does put pressure on companies to build solid evidence as soon as possible. And I think they get it. I’m not an expert on the right balance between real world data and RCTs, and I also don’t know what payers think about this question.

Q: I don’t have a specific question, but can you talk about barriers related to medicolegal issues? What about providers that don’t trust algorithms?

Adam: Whenever there’s new technology, many are scared of what they don’t understand. There was recently a really cool New York Times Magazine article, about whether AI can explain itself. The idea with AI is you feed it data sets, it learns, and then it can make smart decisions. But for a court of law or a healthcare provider, you have to explain how the decision was arrived at. So there’s a big debate in AI in healthcare: If the algorithm can’t explain how it arrived at a decision, where does that leave us? This NYT piece covered the beginnings of this field called “explainable AI” (XAI for short). One of my favorite examples in the piece was an expert on XAI, who said “the solution to XAI is just more AI.” Some researchers are now lashing two algorithms together, one that makes the decision, the other that watches that algorithm and explains the decision making. I think whenever tech is new, many people are naturally scared of it. We hear it at digital health conferences too as it relates to radiology. The IBM Watson computer can read image scans pretty accurately and far quicker than a human – what does that mean for radiology? Well, perhaps it changes the way people work – just like the industrial revolution and the agricultural revolution – I hope tech will allow providers to do less monotonous tasks. What if an Amazon Echo could listen to the consultation and capture what was said? Providers do a lot of mundane stuff, and tech should help with that. There’s a big question on liability I don’t know the answer to – what about when a company takes care of a patient through one of these remote care models? Who’s accountable in that system, and who do I listen to – the company or my own HCP? I don’t know. 

3. We Are Not Waiting vs. Why We Are Waiting – Dana Lewis on Big Value of DIY vs. Parental Caution (Annabel Astle) to Use an Unapproved System

OpenAPS’ Dana Lewis made a persuasive “pro” case for the #WeAreNotWaiting movement, citing the enormous glycemic and quality of life benefits from do-it-yourself (DIY) automated insulin delivery systems. She noted over 500 people worldwide are using various DIY “Loop” systems (likely an underestimate), which totals ~2.8 million hours of closed loop experience in her estimate – an order of magnitude larger than the ~267,000 hours that could be gathered in a single company’s 124-patient, three-month pivotal study. She shared a number of moving DIY user stories from social media, including a very compelling side-by-side retina scan – one DIY user’s retinal swelling declined markedly after starting on closed loop (picture below)! Ms. Lewis emphasized the “overwhelming health benefits” of these systems (A1c, time-in-range, sleep, stress, diabetes burden), something she couldn’t justify keeping to herself. She has now gone ~ 105 days without taking a bolus, a testament to continuous improvements in the OpenAPS algorithm (e.g., “eating soon mode”). Ms. Lewis concluded her part of the debate with a good point – DIY systems are not for everyone, but patients “now have a choice” on whether to use them (i.e., it’s an individual benefit/risk decision). Indeed, many will continue to use OpenAPS and Loop, even with the MiniMed 670G on the US market and other commercial systems coming. A broader AID portfolio will be great for patient choice, and we’ll be fascinated to see if the field evolves to a mix of commercial and Open Protocol-enabled AID systems.

  • In Q&A, Ms. Lewis highlighted out the Bluetooth-enabled Dana insulin pumps (R and RS), which are being used by many Android APS users in Europe and Asia – these allow for direct Android smartphone app communication to the pump (no communication relay bridge), an even bigger user experience win. Ms. Lewis lamented that in the US, out-of-warranty Medtronic pumps are what DIYers are forced to use – she hoped the JDRF Open Protocol AID Initiative will encourage more pump companies to include Bluetooth for interoperability with DIY systems. We hope so too, as this community is incredibly innovative and will enhance the value of pumps and CGMs with excellent software. As Ms. Lewis said, “We’re going to keep doing this (DIY) – we would all love to use an in-warranty pump and we’d be happy to give you money to do so.”
  • Ms. Lewis highlighted the dramatic shrinkage in communication hardware form factor (see picture below), and noted an incredible ecosystem of software around OpenAPS – e.g., Auto Tune to optimize pump settings, a user can “text” their pancreas with Siri and inform it of upcoming events (“getting ready to speak”, running, eating soon.), etc.
  • Ms. Annabel Astle, mother of a 14-year-old girl with diabetes, explained why she is happy to wait (counter to the #WeAreNotWaiting DIY movement): Risk aversion and a desire to guarantee continuity of care for her daughter. Ms. Astle isn’t 100% risk averse, she pointed out – her daughter wore a Dexcom sensor off-label as a child, and today wears it on her arm. But she calculates the risk of off-label CGM to be extremely small, since insulin isn’t being directly delivered off-label (though CGM probably informs dosing). With DIY automated insulin delivery systems, however, she perceives higher risk, and she said it comes down to being a parent: “You need to be aware of a number of things when caring for a child with diabetes – if she were an adult, it’d be very different. The decision would be hers, and I would support her either way.” Certainly, choosing how to deliver a dangerous drug to your child is an enormously difficult decision, so we understand the desire to stick “on label” and proceed with caution – of course, to have diabetes is to be forced to accept significant baseline risk, and a DIY system in the hands of many will reduce that risk (especially overnight). According to Ms. Lewis, there have been no adverse events reported to date for OpenAPS users related to under- or over-delivery of insulin, though admittedly, there is no single responsible party that tracks them. But as Ms. Lewis pointed out, the systems are widely used in very young pediatrics, with set up done by a mix of tech-savvy and non-tech-savvy parents. As for continuity of care, Ms. Astle pointed out that she can’t guarantee a supply chain of parts if the off-warranty Medtronic pump or computer breaks, and she wonders how going back to “analog” care from Looping would impact her daughter’s emotional wellbeing (especially if she’s happy with her current regimen the moment). The question is a complex one: Is this a “better to have loved and lost…” situation, or is it better to maintain the current level of satisfaction if things seem fine? This status quo bias is one that may impact diabetes technology uptake for several years –“I don’t want to try to something new or switch, because what I have works.” Ms. Astle also likes that her daughter has products with warranties, rigorous testing, guarantees, and assurances – a fair point and one out-of-warranty Medtronic pumps and DIY algorithms cannot attest to at this point. Again, this is a very difficult choice for a parent to make. Many find DIY automated insulin delivery systems to be very safe (basal modulation only, and lots of constraints can be set by the user), but not everyone will be comfortable without regulatory approval. This adoption cycle is natural in any market!

Retinal Scan – Before and After DIY Closed Loop

Text My Pancreas – Software Ecosystem Example

The Burden Liftoff, Especially for Parents

Big Picture Highlights

1. Is There Hope for Diabetes Prevention? How Promising is Precision Medicine Based on Genetics? Call for Systems-Level Change

A fascinating discussion – during which speakers and audience members discussed hope (or lack thereof) for diabetes prevention, an interesting take on precision medicine, and the need to change systems via legislation – followed a forum on the function of GPs in diabetes. We’ve provided the entire transcript below. After a Ugandan provider asked the panel if there was reason to be optimistic despite climbing diabetes rates even in the best care settings, conversation turned to systems, policy, and the detrimental effects of modernization. VU University Medical Center’s Dr. Petra Elders seemed to place the onus on society to find ways to do better, while chair Dr. Ali Khalil focused on “business-minded multi-national companies” like McDonalds and Coca Cola, and a “macro environment where we involve government and push them very strongly against certain things that we know as professionals is actually causing disease…” Dr. Elders proposed that even in countries, like Uganda, where elaborate diabetes care networks haven’t been established, that society-level prevention strategies may be a solution. The whole discussion is worth a read, including Dr. Elder’s comment that she doesn’t expect much from precision medicine based on genetics – while this may help a few, it is not entirely feasible at this point in time outside of academia, and she believes broader strokes are sufficient for most people (exercise sufficiently, “eat less than your predecessors”). Of course, determining the appropriate strategies to promote a healthier lifestyle for an individual by definition requires some sort of precision medicine, but “deep phenotyping” and behavioral analysis could be even more informative than genomics. However, when it comes to picking the right medication, we still believe that –omics will have a lot to offer.


Q: I practice in Uganda, where our dream is that, if we had the resources that you do in the Northern European setting, we could stop diabetes. Even in your environment, diabetes is still going up. Is there hope? Do we still have a chance of ever bringing diabetes down, if we’re going up even in the best environment?

Dr. Petra Elders (VU University Medical Center, Amsterdam, Netherlands): Since you’re here and I’m here and there are people in the audience, we still have hope. It has something to do with patient awareness – we as doctors can prevent complications. Guide glycemic control, blood pressure, lower lipids, prevent complications, and have a multidisciplinary approach to treat complications. But we’re past this. Now the public has to do something. It’s a societal question, whether we as mankind are able to stop eating horrible foods. And if we don’t have to move to survive, we prefer to sit on our chairs. We need a new view and new solutions, I think that’ll take a long time.

Dr. Ali Khalil (Imperial College London Diabetes Center, Abu Dhabi, UAE): The big issue is the elephant in the closet – the big problem is not with the physicians, and it’s not with the population. It’s the business-minded multi-national companies – McDonalds, Coca Cola – despite our best efforts, they are able to advertise their products better than we can advertise ours (healthy eating and exercise). Unfortunately, we’re going in the wrong direction. Now we don’t even have to get out of chairs to get food because its delivered right to our doors. The problem is even being propagated now. 30 years ago when I went to China, there were bikes everywhere. Then eight years ago, I didn’t see any bikes in Beijing. They’re coming back again, but that whole issue of modernization, what we call modernization, that’s plaguing us.

Dr. Elders: And maybe even in the countries where diabetes care still has to be established, maybe even finding societal solution in reducing consumption there might be a solution, but still small steps and we have a lot left to do.

Dr. Ed Fisher (UNC, Chapel Hill, NC): In your presentation, precision medicine touched on a whole number of things, many related to phenotype. Talk about precision medicine in US focus mostly on genotype directed care. Can you talk a little more about that?

Dr. Elders: As a GP, I’m realizing that the gene pool is not changing so much over time, but I’m also realizing that we are having an epidemic of diabetes. To be very honest, I don’t expect much from genetic research. Maybe a little for small select segments, but for the broad population, things will probably remain very practical – exercise sufficiently, eat less than your predecessors. Sorry, GPs usually give very flat answers. One thing we can make a lot of headway in is patient input – in western countries, we tend to send out a lot of input, but in activating patient input, getting patients to take care of their own issues, we have a long way to go.

Dr. Fisher: Both of your presentations point to importance of understanding systems of care rather than thinking healthcare will be solved by individuals trying harder and being smarter. We are trained as professionals to think that we as individuals can make a difference, but it’s important to realize that we need systems to make us more effective.

Dr. Khalil: I came to UAE from Canada. It was very different when I came here. We were part and parcel of people who started programs here, developing specialized medicine for diabetes or diabetic centers because of the high prevalence here. The sad part is a lot of it starts at youth – we’re seeing type 2 diabetes at the ripe age of 25. Not even caught up with complications yet, but they’re about to come. As members of a medical community, we need to push legislation really, much more so than working on our own environment. We need to have a macro environment where we involve government and push them very strongly against certain things that we know as professionals is actually causing disease and problems for indigenous populations who had never had to deal with onslaught of different foods and changes in their environment. That’s the biggest challenge in developing world. We need to develop some sort of structured methodology, but that works separately in different areas. It’s a whole different mental status in the Netherlands and the UAE. Understanding the population and getting through to them is primary. But aside from that, it’s not always the public’s fault. Given the choice between high calorie food that you can chow down in five minutes and having to wait three to four hours for a smaller meal, most will choose the faster option, and face the consequences later. In the UAE, we set up a lot of these legislations through regulatory bodies to educate the population in a much stronger way as opposed to leaving that and going with specialized centers.

2. Training GPs to Perform Retinopathy Screening Reduces Ophthalmologist Referrals (Better Resource Allocation!)

Javeriana University School of Medicine’s Prof. Pablo Aschner explained how training GPs in his clinic to screen people with diabetes for eye disease cut down on ophthalmologist referrals significantly. While the cost of a retinal camera is certainly a major investment, Prof. Aschner emphasized that the value gained in the long-term far outweighs the initial financial burden: Among those with a diabetes duration less than one year, only 30% had abnormal retinal readings and were referred to an ophthalmologist, seriously cutting down on visits. As diabetes duration increased, so, too did the frequency of abnormal readings requiring referrals, leading Prof. Aschner to suggest that those with a diabetes duration of more than ten years should be sent directly to an ophthalmologist. Prof. Aschner was prompted to advocate for early screening by less-specialized providers by a finding that 11% of newly diagnosed individuals at a diabetes care center in Colombia already had retinopathy – compared to 0% at a center in Canada. Upon seeing these results, he recommended immediate screening up on diagnosis for type 2 patients, and screening beginning five years post-diagnosis in type 1 diabetes. We are equally concerned by the 11% rate of retinopathy at diagnosis, which uncovers a larger problem; perhaps earlier glucose tolerance or fasting glucose tests are required so people can be diagnosed and complications can be prevented. Early retinopathy screening is an investment; so, too, is earlier diabetes screening (of course, we recognize this is easier said than done). Apart from reduced ophthalmologist usage (cost) and therefore better resource allocation to those who really need it, we wonder if Prof. Aschner’s work has led to a decrease in retinopathy prevalence in his clinic.

  • We were surprised to see data reflecting substantial global variation in diabetic retinopathy awareness. A meta-analysis of 15 articles revealed that among those with diabetes, between 29% and 84% were aware diabetes could affect vision. Awareness among general practitioners was also found to be variable, with 44% of general practitioners in Myanmar lacking awareness of vitreous hemorrhage and retinal detachment as possible diabetes complications, and only 34% of physicians surveyed in the Latin American Congress on Diabetes reporting correct referral of type 1 patients for eye exams. Prof. Aschner emphasized the need for education, highlighting rampant levels of misinformation and lack of information.

Detailed Discussion and Commentary

The DISCOVER Study Program

Rationale and Importance

Linong Ji, MD (Peking University Diabetes Center, China)

Dr. Linong Ji provided an introduction to the real-world DISCOVER study, which began in 2014 (the first participant was enrolled in December that year) and completed recruitment in July 2016. The last patient visit is expected mid-2019, with a mean follow-up of three years for all type 2 diabetes patients enrolled. Dr. Ji presented the primary objective of DISCOVER – to describe diabetes management patterns and disease evolution over three years for people initiating a second-line glucose-lowering therapy – and also reviewed secondary aims – to assess clinical outcomes like achievement of glycemic targets (A1c), BMI, blood pressure, incidence of micro and macrovascular outcomes, incidence of hypoglycemia, patient-reported quality of life, healthcare resource use, and factors associated with treatment choices. One year in, this is already a very rich dataset, and it could potentially shed light on how different therapies affect micro and macrovascular outcomes in the real world (this isn’t exactly a head-to-head trial, but it could certainly illuminate advantages/disadvantages of different treatment options relative to one another, which is something we desperately need in the push for personalized medicine). Dr. Ji announced that 15,992 participants have been recruited for DISCOVER from 38 counties, mostly low- and middle-income. This study size is impressive; Dr. Ji described it as “ambitious,” especially considering the standardized data capture from very different healthcare systems around the world. More specifically, DISCOVER includes 2,002 type 2 diabetes patients from the Americas (but notably, not the US), 811 from the continent of Africa, 3,480 from Europe, 2,183 from the Eastern Mediterranean, 4,156 from the Western Pacific, and 3,360 from Southeast Asia – this is truly a global investigation, creating a realistic picture of what diabetes care looks like in different places around the world.

Methods and Baseline Patient Characteristics

Kamlesh Khunti, MD (University of Leicester, UK)

Dr. Kamlesh Khunti described study methods and baseline patient characteristics for the DISCOVER program (n=15,992 from 38 countries), pointing to considerable treatment inertia on a global scale. DISCOVER used a standardized case report form to collect information on demographics, physical exam and lab test results, first- and second-line treatment, treatment changes and reasons for change, hypoglycemia and vascular complication incidence, and, notably, patient-reported outcomes. Variables were all measured according to standard clinical practice – importantly, these standards may have differed from region to region, although the study was designed to maximize consistency in data capture. DISCOVER enrolled 4,156 participants from the Western Pacific, 3,480 from Europe, 3,360 from South East Asia, 2,002 from the Americas (no US), 2,183 from the Eastern Mediterranean, and 811 from Africa. Dr. Khunti pointed out that while >90% of patients had a reported systolic blood pressure and BMI, fewer had a recorded A1c (77%), LDL or triglycerides (60%), serum creatinine (55%), or albuminuria (15%) measurement, with wide variation across geographies. Mean diabetes duration was 5.6 years, mean A1c 8.4%, and mean BMI 29.6 kg/m2; these were reasonably constant across regions. Health insurance coverage was split between public/governmental (55%), private (16%), mixed (3%), and no insurance (26%); 48% were employed or self-employed, 23% retired, and 29% not working. Inclusion criteria were broad and included type 2 diabetes, age >18 years, and status of initiating a second-line therapy (adding or switching) after first-line oral treatment. Patients were excluded for type 1 diabetes, pregnancy, chemotherapy and steroid use, dialysis or renal transplants, if first-line treatment was injectable or herbal, interventional trial participation, and conditions that might cause early loss to follow up. Sites were selected to represent a country’s diversity of treatment (e.g. spanning urban and rural locations in the right proportion); 673 sites have been involved, including 41% PCP offices, 42% endocrinology practices, 15% internal medicine, 1% cardiology, and 1% other.

Baseline A1c Control and Prevalence of Complications

Mikhail Kosiborod, MD (University of Missouri, Kansas City, MO)

Dr. Mikhail Kosiborod presented key findings from the DISCOVER study so far, emphasizing a few key messages: (i) mean worldwide A1c for the type 2 diabetes population is far above goal, (ii) global complication rate is high, despite this population having a relatively short duration of diabetes (4-7 years, 5.6 years on average), and (iii) there’s evidence for broad treatment inertia, given that people starting a second diabetes drug displayed high A1c and frequent complications. First off, <80% of participants had a recorded A1c assessment, which Dr. Kosiborod described as an important finding in itself, revealing gaps in screening and monitoring. A1c screening rate was lowest in Africa at 57%, followed by Southeast Asia at 61%, ranging up to the Eastern Mediterranean with 94%. When screening was defined as A1c or fasting plasma glucose evaluations, the rate increased, but was still <90% globally. During later Q&A, Dr. Kosiborod pointed out that patients in some parts of the world have to pay out of pocket for screening, delivering this powerful quote: “This has been an eye-opening experience for someone who practices in the US. What we think of as basic measures, this is something that people pay out of pocket for in large parts of the world, so it’s not surprising that we see low rates of screening.” Mean A1c across the entire global study population was 8.4%, from a high of 8.7% in the Eastern Mediterranean to a low of 8.1% in Europe. Nearly one-third of the population had A1c ≥9%, while 55% had A1c ≥8%, and only 15% met an A1c target <7%. This is highly indicative of treatment inertia, Dr. Kosiborod explained, because major guidelines recommend second-line therapy initiation after ~three months not at glycemic goal, and it’s disheartening to see so many individuals waiting until they surpass an A1c of 9% before adding a second agent. The prevalence of microvascular complications (CKD, albuminuria, retinopathy, retinal laser photocoagulation, autonomic neuropathy, peripheral neuropathy, or erectile dysfunction) was 18% globally, ranging from 13% in the Americas to 22% in Southeast Asia. Macrovascular complications were defined broadly (coronary artery disease, stroke, transient ischemic attack, carotid artery stenting, carotid endarterectomy, peripheral artery disease, diabetic foot, amputation, defibrillator use, or heart failure) and appeared in 13% of the overall study population, ranging from 10% in Africa and the Americas to 18% in Europe. Dr. Kosiborod’s main takeaway from this outcomes data was that a large proportion of patients relatively early in the course of diabetes present with microvascular disease, macrovascular disease, or a combination of both. We already see meaningful implications of this finding: As one example, many thought leaders have argued that diabetes CVOTs are hardly generalizable to the real-world diabetes population because they enroll very high-risk patients, but DISCOVER suggests that the real-world diabetes patient is actually quite likely to have existing micro or macrovascular disease. Dr. Kosiborod listed four correlates for both micro and macrovascular complications: (i) male gender, (ii) older age, (iii) smoking, and (iv) a history of hypoglycemia (we find this last one particularly notable, in the movement for glycemic outcomes beyond A1c). He spoke to the “remarkable opportunity for aggressive risk factor screening and modification early in the disease process.” Indeed, while one objective of this program was to reveal the true global picture on diabetes, the researchers also anticipate many actionable insights, and the need for better screening (of glycemia, of micro and macrovascular risk factors) is one of them.

First- and Second-Line Treatment

Kamlesh Khunti, MD (University of Leicester, UK)

Dr. Khunti took the stage once again to discuss first- and second-line therapy choices in the DISCOVER study population. There was regional variation, but the most common first-line treatment across the board was monotherapy, ranging from 87% of regimens in Africa to 52% in Southeast Asia, where fixed-dose combinations are common according to Dr. Khunti. More specifically, 61% of all individuals enrolled in DISCOVER took metformin as first-line; 17% received a metformin/SU combination upfront, and 8% were prescribed SU monotherapy. Second-line therapies were more typically combinations, ranging from 71% in the Western Pacific to 85% in the Eastern Mediterranean. Insulin was used as second-line in 5% of Southeast Asian patients, up to 15% of patients in the Western Pacific, and 9% of cases overall. Second-line use of metformin/SU co-administration was seen in 24% of patients globally, while 23% of all participants took a metformin/DPP-4 inhibitor combination as second-line. Additionally, a multinomial regression model with region as a fixed effect examined individuals receiving metformin monotherapy as first-line treatment and found that region was an important predictor of second-line treatment. The Americas favor SUs over DPP-4 inhibitors and insulin, while Southeast Asia and the Western Pacific heavily favor DPP-4 inhibitors over SUs (this would be our preference as well, given the hypoglycemia, weight gain, beta cell burnout, and possible CV harm associated with sulfonylureas). DPP-4 inhibitors were a more likely addition in the elderly than SUs, and SUs were in turn more likely than insulin. At higher A1c, SUs were more likely to be added than DPP-4 inhibitors, but insulin was a more likely addition than SUs in those with A1c ≥9%. Follow-up analysis for DISCOVER is currently underway, and will reveal more about the clinical effects of various second-line treatment options over three years. This could be incredibly important insight for the diabetes field, in differentiating between drug classes and matching the right therapy to the right patient at different stages of disease.


Glasgow’s Dr. Naveed Sattar presented strengths and limitations of the DISCOVER study, finding more strengths in this analysis overall. He highlighted the value of real-world data from diverse parts of the world, including regions that have never before been represented in a diabetes study. He called out the standardized data capture, which allows for comparisons between geographies, and discussed the potential for one area to learn from another (what strategies are working elsewhere for population-level diabetes management?). Dr. Sattar also summarized the important insights gained on prevalence of diabetes complications, patterns of glycemia and risk factors, and quality of care around the world. As for limitations, he shared an earlier concern he had that DISCOVER would enroll patients who are more educated than average. However, after seeing the full presentation, he noted a high percentage of individuals who had no formal schooling. Lastly, Dr. Sattar mentioned the gaps in data: Not everyone in the study had an A1c reading, but this too was telling, revealing how so many people with diabetes around the world aren’t properly screened, let alone treated. Ultimately, he described the DISCOVER program as a “much-needed snapshot of diabetes care standards for people transitioning from first-line therapy to second-line therapy,” and we certainly appreciate the massive effort behind this global, real-world study, which has already provided new knowledge on diabetes care around the world and which will hopefully reveal strategies for improvement in the prospective follow-up.

Questions and Answers

Dr. Sattar: Gentlemen, were any of you surprised by any of the findings?

Dr. Kosiborod: I’m not sure if surprised is the right word – more like “unexpected” – but participants in this study were relatively young, with relatively short duration of diabetes, and still we saw complication rates that are quite high. We’re used to seeing high event rates in diabetes CVOTs, which specifically select for high-risk patients. DISCOVER enrolled a very, very different patient population. There was no requirement for individuals to have vascular disease at baseline, and overall, these patients would be considered low-risk. So, are we missing the boat? Are we thinking of these patients as low-risk when in fact they aren’t? What can we do to minimize the risk of complications? Very shortly, we’ll have one-year data from DISCOVER, and this will be one of the most insightful takeaways.

Dr. Khunti: My surprise was that we go to all these diabetes meetings, and we look at these fantastic RCTs showing how we can improve care, but this global population is not receiving the basic care they need. Look at lipids, albuminuria. These are basic standards of care that every patient in the world should get.

Comment: Congratulations on a really wonderful session. This is the type of study that is much-needed in diabetes. I do want to point out that Sub-Saharan Africa was not represented, so that may be an issue. It’s important to encourage those regions with the lowest resources to participate in these types of studies going forward. But this is a great start.

Q: I was struck by the map of macrovascular complication rates around the world. Prevalence was high in Russia, for example. Are there other risk factors besides diabetes that are more common there and explaining this?

Dr. Kosiborod: You make a keen observation. It’s important to keep in mind that there are many reasons why there could be higher macrovascular prevalence in any given region. Screening strategies are different in different parts of the world. So, you can imagine a scenario where someone screens aggressively for complications, and complications are thus reported as high, but that doesn’t mean the quality of care is suboptimal for patients there. It also depends what kind of healthcare setting patients come from (rural, urban, etc.).

Dr. Ji: This program will most likely be adapted by IDF, and it could be expanded into other regions for a more global picture on diabetes management.

Q: I’d also like to highlight that only 2.3% of participants were Black, which shows the under-representation of Sub-Saharan Africa. Also, the US wasn’t represented. This is a big country with a very heterogeneous patient population. That would have helped the final result.

Dr. Kosiborod: There’s a very good reason for this exclusion. There’s a parallel registry program called the Diabetes Collaborative Registry – now with nearly 2 million patients enrolled – between the ACC, ADA, AACE, Joslin Diabetes Center, and American College of Physicians (APA) that’s collecting this data in the US. This US data will complement the findings from DISCOVER.

Q: Drawing on Steno-2, when multifactorial intervention led to better outcomes, what is most important? Is it to control A1c with any drug, or to focus on the pathophysiology of diabetes?

Dr. Khunti: This study was not looking at the therapies in terms of how they’re being used. One of the primary reasons for doing this research is access to care – the basic level of care that some patients are getting is quite low. Certainly, we’ll look to see newer therapies coming onto the market in our prospective follow-up, but it’s more about the quality of life that these patients are getting. And, this is the first time we’re getting data from some countries that have never been represented in any study previously.

Q: I was surprised that the influence of cost on choosing a second-line drug was actually quite low in this study. How can you explain this, even though the study was conducted in low- to middle-income countries?

Dr. Khunti: Some places have a minimum number of drugs available – maybe only metformin, sulfonylureas, and insulin. So these professionals won’t say that cost comes into the equation, because it doesn’t.

Dr. Sattar: Two final questions for the panel. Why was albuminuria collection so low, only 16%? And what are the specific aims of three-year follow-up?

Dr. Khunti: These are basic quality of life measures that patients should be getting, and it’s just not happening, especially in lower-income countries.

Dr. Kosiborod: Let me double up on what Kamlesh just said – this has been an eye-opening experience for someone who practices in the US. What we think of as basic measures, this is something that people pay out of pocket for in large parts of the world, so it’s not surprising that we see low rates of screening (for albuminuria, and for other complications). It’s not because clinicians aren’t thinking of it. Patients may choose not to do it because of other priorities and limited resources. For me, the most interesting question to look at in follow-up will be: What’s the incidence of diabetes complications in people who didn’t have complications before? What are the predictors of that? Having this rich dataset will allow us to better determine who develops a complication vs. who doesn’t. And we’ll be able to see what treatment patterns make an impact.


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