Endo 2019

March 23-26, 2019; New Orleans, LA; Full Report – Draft

Executive Highlights

  • Greetings from New Orleans, where ENDO 2019 recently concluded, featuring a very notable post hoc of the DEPICT trials for AZ’s SGLT-2 inhibitor Farxiga in type 1 which found meaningfully lower DKA risk in the subgroup of participants with BMI ≥27 kg/m2. We’re thrilled investigators got this data out so quickly – Forxiga for type 1 was JUST approved in the EU with an equivalent BMI restriction, following CHMP’s recommendation – and even more excited about what it shows: Among those ≥27 kg/m2 (~50% of all participants), 1.7% had DKA on Farxiga vs. 1.0% on placebo, compared to 4.0% and 1.1% in the whole cohort, indicating that higher BMI was associated with much lower DKA risk. That said, a study of Yale medical records found 21 episodes of SGLT-2 inhibitor associated DKA in 17 patients (one with type 1 diabetes) – of which 52% occurred with euglycemic blood glucose (<300 mg/dl). Importantly, the average hospital stay was seven days, and presenter Dr. George Stamatiades noted that many of these patients were in the hospital for “a couple of days” without knowing that they were in DKA – reflecting the importance of educating hospitalists on euglycemic DKA.

  • Also in therapy, full results from PIONEER 3 for just-submitted oral semaglutide build upon the topline data released last June (and were very notably published in JAMA with an editorial from Dr. Irl Hirsch). In particular, this poster offered more details on the intention-to-treat analysis of PIONEER 3: Compared to DPP-4 inhibitor sitagliptin (Merck’s Januvia), 14 mg oral semaglutide beat sitagliptin all the way out to 1.5 years, with a treatment difference of 0.4% at 78 weeks (1.1% vs. 0.7%, p<0.001). Dr. Paresh Dandona presented very topline results from a positive 26-week trial of liraglutide in type 1, but we were most fascinated by the announcement that JDRF has funded his group to conduct a study (n=114) of triple therapy in type 1, including insulin, semaglutide, and dapagliflozin. Results from the LiraAdd2SGLT-2i study (n=303, Victoza vs. placebo as an add-on to SGLT-2s) showed that at 26 weeks, Victoza group drove a ~0.7% greater reduction in A1c compared to placebo.

  • In tech, a real-world observational study (n=93) of the Medtronic MiniMed 670G in children and adolescents found high rates of discontinuation. 38% of participants withdrew from the study, most commonly due to the frequent alarms, forced exits, and calibration requirements of the system. Of those who remained on the 670G, significant A1c reductions were observed at six-months (-0.3%), but statistical significance was not maintained at 12 or 24 months. One of the main takeaways from the 670G observational study was the need for simpler, more user-friendly technology – we think the FreeStyle Libre pretty much hits the nail on the head here. Baseline (blinded) data from the WISDM trial evaluating real-time CGM in older adults (≥60 years) with type 1 diabetes found that participants (n=203) spent 57% time in range, 35% >180 mg/dl, and 5% <70 mg/dl. An interesting analysis of T1D Exchange data presented by Dr. Carol Wysham found that the combination of high mean A1c and high A1c variability was associated with worse outcomes.

Greetings from New Orleans, where ENDO 2019 just wrapped up. This meeting brought a lightning storm of learnings, which we’ve packed into 22 highlights + exhibit hall coverage below, by these categories: (i) SGLT inhibitors and GLP-1 agonists in type 1; (ii) Cardioprotective agents; (iii) diabetes technology; (iv) Glycemic targets and guidelines; (v) Epidemiological insights and post-hoc analyses; and (vi) Obesity. Read on!

Table of Contents 

SGLT Inhibitors and GLP-1 Agonists in Type 1

DEPICT Post Hoc of BMI ≥27 kg/m2 Shows Meaningfully Lower DKA Risk with Farxiga Compared to Overall Cohort: 1.7% on Farxiga vs. 1.0% on Placebo Experienced DKA

A risk/benefit analysis of DEPICT-1 and DEPICT-2 demonstrated that patients with a BMI ≥27 kg/m2 experienced less DKA than the overall, pooled population, while efficacy and CGM-based outcomes were similar between the ≥27 kg/m2 subgroup and the entire cohort. In the entire pooled population from both DEPICT studies over 52 weeks, 4.0% of participants on Farxiga and 1.1% on placebo experienced DKA (4.6 and 1.3 events per 100 patient years; definite DKA events only). However, in the subgroup of patients with BMI ≥27 kg/m2, only 1.7% on Farxiga and 1.0% on placebo experienced DKA (1.9 and 1.2 events per 100 patient years) – wow! Of note, this analysis includes 51% of the total Farxiga group (n=272) and 54% of the total placebo group (n=289), with similar baseline characteristics and demographics between the treatment groups. We find it particularly encouraging that >half of total DEPICT participants comprise this analysis, both (i) supporting the notion that this trend is real and (ii) reflecting that a truly substantial proportion of type 1s stand to qualify for use of the therapy if and when restricted to those ≥27 kg/m2.

Among the authors of this poster are Drs. Chantal Mathieu, Paresh Dandona, and Moshe Philip, and we’re thrilled that the DEPICT team got this analysis out so quickly following a February 1 opinion from EMA’s CHMP that recommended approval of the type 1 adjunct for adults with a BMI ≥27 kg/m2 with drug initiation and supervision by physicians experienced in type 1. CHMP has put forth an equivalent opinion for Sanofi/Lexicon’s sotagliflozin (this came a month later), and EMA is expected to grant final approval to both SGLTs for type 1 within 1H19. The BMI requirement came as a surprise to many given a lack of previous emphasis on BMI as a mediator of DKA risk, and it was alsopoint of critique at ATTD 2019. However, these data show that there is, indeed, substantial merit to the restriction recommended by CHMP – bravo to the agency for zeroing in on a salient and clinically useful selection criterion. Farxiga for type 1 remains under review at FDA, and we’ll be extremely curious to see how this data might impact discussion at a possible Advisory Committee meeting (and FDA’s final decision). FDA just issued a CRL for sotagliflozin for type 1.

  • Participants with BMI ≥27 kg/m2 saw efficacy with Farxiga similar to the entire pooled cohort. Placebo-adjusted A1c reduction with Farxiga was 0.43% at 24 weeks (95% CI: -0.55, -0.31) and 0.34% at 52 weeks (95% CI: -0.48, -0.19). Body weight loss was 3.1 kg (~6.8 lbs) at 24 weeks and 3.6 kg (~7.9 lbs) at 52 weeks with Farxiga (placebo adjusted). Placebo-adjusted difference in mean time in range was 9.7% at 24 weeks, translating to an additional 2.3 hours in range on Farxiga, with no increase in time <70 mg/dl. And on the composite endpoint of ≥0.5% A1c reduction without severe hypoglycemia, the odds ratio with Farxiga vs. placebo was 3.52 (95% CI: 2.39-5.21) at 24 weeks and 2.63 (95% CI: 1.78-3.88) at 52 weeks. There was also a trend toward less hypoglycemia with Farxiga in the ≥27 kg/m2 subgroup: The exposure-adjusted incidence rate for severe hypoglycemia was ~16 events per 100 patient years with Farxiga, compared to ~24 on placebo.

  • On average (mean), participants in the BMI ≥27 kg/m2 groups (Farxiga and placebo) were 45 years old, 42%-50% male, and had a BMI of ~32 kg/m2 – we’re curious how this relatively high BMI impacted results. Are those with a BMI of 27 at greater risk of DKA compared to those with a BMI of 32 kg/m2? We doubt DEPICT had enough DKA events to be sensitive to a continuum of risk, but we do think this is an important clinical question and we’ll be interested to see how else researchers can parse out DKA risk. Average total daily insulin dose was 71-72 units, duration of diabetes 21-22 years, and A1c 8.4%; 42%-44% were on pump therapy and 28%-32% used CGM. 

Dr. Paresh Dandona Presents New Topline 26-Week Data on Liraglutide for Type 1, Teases Triple Therapy Study of Insulin/Semaglutide/Dapagliflozin (n=114) Funded by JDRF

Following presentation at ADA 2018 of a promising year-long study of liraglutide in type 1, Dr. Paresh Dandona detailed similarly inspiring results from a separate 26-week study in type 1s with overweight and obesity. This JDRF-supported study randomized 64 type 1s with overweight and obesity to either liraglutide (n=37) or placebo (n=27). Significant improvements with liraglutide vs. placebo were seen on A1c (~0.4% treatment difference), time in range, time in hyperglycemia, weight loss, total fat percentage by body composition, systolic blood pressure, fasting and postprandial endogenous GLP-1 levels, and several biomarkers of inflammation, favoring liraglutide. Please note that hard numbers were not provided for much of this data.

  • A trend toward an increase in time in hypoglycemia was seen with liraglutide treatment, but this difference was not significant. We note that hypoglycemia events were balanced across all groups in data presented by Dr. Dandona at ADA 2018; however, this only included CGM data from four weeks compared to 26 weeks in this study. During Q&A, Dr. Cecilia Low-Wang pressed on this issue, and Dr. Dandona highlighted that “we have not had one patient leave this drug because of a hypoglycemia effect.” Dr. Low-Wang also asked about the insulin titration protocol that Dr. Dandona uses with these type 1s on liraglutide, and he detailed that while both basal and prandial insulin doses fall, he tries to stick to mainly minimizing prandial insulin doses, since this is where liraglutide exerts most of its effect (i.e., postprandial glucose lowering).  

  • Weight loss with liraglutide was entirely via body fat composition and not in lean body mass. Dr. Dandona noted that this was one of the most exciting results of the study. Total fat mass significantly declined in the liraglutide group compared to placebo, while lean body mass was unchanged in both groups. Visceral adipose tissue (VAT) also significantly fell in the treatment group. Dr. Dandona explained that this finding of VAT decline is actually the first clinical evidence of the phenomenon, as similar data has not been shown in type 2 studies yet. Mechanistically, it was also seen that levels of UCP2 – a protein associated with increasing the metabolic state of fat cells – was also increased in the treatment group. This increase was associated with an elevation in CPT1 level, which is an enzyme that transfers free fatty acids across the mitochondrial membrane to the mitochondrial matrix and causes increases in fat burn.

  • Dr. Dandona highlighted the increase in endogenous GLP-1 level with liraglutide treatment as a “very remarkable effect.” Significant increases in fasting and postprandial endogenous GLP-1 levels were associated with liraglutide treatment. Dr. Dandona hypothesized that there “must be some positive feedback regulator” responsible for this effect and pointed toward preliminary data from in vitro studies that show incubating islets with GLP-1 increases GLP-1 secretion. Interestingly, this data also implies that there may be local GLP-1 secretion from islets as well – this point got a murmur of surprise from the audience.

  • Very excitingly, Dr. Dandona divulged that JDRF has just funded his group to complete a triple therapy study of insulin/semaglutide/dapagliflozin in type 1 diabetes. He also noted that this is the largest grant the JDRF has ever given for a study – wow! The trial will be 52 weeks long and will enroll 114 people with type 1 diabetes. A 2:1 randomization protocol will be used, with a control group of 38 patients using standard of care insulin therapy. In the treatment arm, 76 patients will be assigned to an insulin therapy + semaglutide (Novo Nordisk’s Ozempic) arm for the first 26 weeks of the study, which is notable in and of itself. After 26 weeks, half of this arm will also have dapagliflozin added, while maintaining insulin and semaglutide. No further detail on endpoints or expected completion date were shared, but we’re extremely excited nonetheless. Dr. Dandona expressed the hope that “we can get 60-70% of type 1s under 7.0% A1c” with triple therapy – this would surely be a smashing success! On balance, however, we imagine that using triple therapy in type 1 takes a tremendous amount of care and expertise from both patient and provider (as does using just one adjunct therapy). Questions abound: How effective will semaglutide – a much longer-acting GLP-1 than liraglutide – be on glucose and weight in type 1? How will semaglutide and dapagliflozin interact in type 1, particularly on insulin requirements and DKA and hypoglycemia risk? Is there a specific reason semaglutide and dapagliflozin were chosen amongst the wide variety of GLP-1s and SGLT-2s available?

  • Dr. Dandona noted that he has “the world’s largest number of type 1s on liraglutide” and that they like the drug so much that they refuse to leave it. Dr. Dandona joked that these patients “are addicted to liraglutide” because of the drug’s efficacy and impact on glycemic variability. We’re glad to see this sustained excitement and use of liraglutide in type 1, even after Novo Nordisk has declined to further pursue the therapy in this population following middling results in the ADJUNCT ONE and ADJUNCT TWO trials. Just last week at ENDO Fellows 2019, we heard similar enthusiasm from Dr. David Harlan on GLP-1s for type 1 – Dr. Harlan detailed that he uses this therapy in almost all of his patients with diabetes who show measurable C-peptide. Considering this enthusiasm from patients and providers, we really do hope to see larger scale studies of GLP-1s in type 1 diabetes. At the very least, we imagine that positive data like this from Dr. Dandona’s group will help liraglutide become a more popular adjunct therapy in type 1, especially with a generic version set to hit the market in 2023.

Medical Records Study (n=21 episodes) Finds Half of SGLT-2 Associated DKA Events are Euglycemic (<300 mg/dl), Leading to Delayed Diagnoses and Prolonged Hospital Stays

Dr. George Stamatiades (Yale New Haven Health) presented clinical and biochemical characteristics from 21 SGLT-2 inhibitor related DKA events, highlighting the prevalence of euglycemic DKA and its impact on delaying diagnosis of DKA in hospital settings. Seventeen total patients who were taking an SGLT-2 inhibitor and had a documented DKA episode were identified through medical record examination; of note, only one of these patients had type 1 diabetes. Eleven episodes were observed with canagliflozin, six with empagliflozin and four with dapagliflozin. At diagnosis, all patients had ketones in their urine, and on average had a beta-hydroxybutyrate reading of 6.4 mmol/l (well above the >3 mmol/l threshold for “probable DKA” established in recent consensus guidelines). Very notably, 52% of these cases were euglycemic DKA episodes with blood glucose levels <300 mg/dl (admittedly, a high threshold), and 47% were <250 mg/dl. On average, blood glucose levels at presentation were 329 mg/dl, with a range of 154-777 mg/dl. The average hospital stay was 7 days (!) for those admitted with DKA – much higher than the average stay of ~2 days normally associated with DKA events – and the median was 5 days (average duration of insulin infusion 3.7 days and median 3 days). On this point, Dr. Stamatiades noted that many of these patients were in the hospital for “a couple of days” without knowing that they were in DKA because of otherwise normal blood glucose levels at presentation. This led to delayed diagnoses and progression to more severe DKA, necessitating longer and more intensive hospital stays. Of course, data like these represent one of the greatest fears with SGLT-2 inhibitor use, as the paradoxical nature of euglycemic DKA (most patients, providers, and emergency room staff are used to only checking for DKA with elevated blood sugar levels) may cause many DKA events to be missed and progress to more severe episodes that endanger the patient and result in elevated costs for the entire healthcare system. It’s clear that as usage of SGLT-2 inhibitors continues to grow – especially among type 1s – increased education and awareness (especially in emergency rooms, where early diagnosis of euglycemic DKA will be key) is paramount in ensuring safe use of these nephro-and cardioprotective agents.

Cardioprotective Agents

Poster Displays Full PIONEER 3 Results: 14 mg Oral Semaglutide Shows Significant A1c and Weight Loss Benefits vs. DPP-4 Sitagliptin; ~12% of Patients on 14 mg Discontinue Therapy Following AEs

Novo Nordisk presented full results from PIONEER 3 (oral semaglutide vs. sitagliptin) on a poster, adding more granularity to the topline data released last June (JAMA publication and editorial from Dr. Irl Hirsch). The company just submitted its first-in-class oral GLP-1 agonist to FDA, and given the use of a priority review voucher, a decision is expected in six months, by September 2019. Notably, this poster shared specific numbers from the trial’s primary intent-to-treat (ITT) analysis, whereas the topline release focused more so on the secondary on-treatment analysis. FDA will use the ITT in its evaluations (this follows all participants, even if they discontinued medication or needed rescue therapy); key efficacy data is summarized in the table below. Baseline A1c was ~8.4%. The higher doses of oral semaglutide (7 mg and 14 mg) showed significantly superior A1c-lowering after 26 weeks, with estimated treatment differences vs. sitagliptin (Merck’s DPP-4 inhibitor Januvia) of 0.2% and 0.5%, respectively (both p<0.001). Only the 14 mg dose demonstrated significant A1c benefit after 78 weeks (1.5 years), with an estimated treatment difference of 0.4% (p<0.001). A1c reduction was significantly inferior with 3 mg oral sema vs. sitagliptin after 26 weeks, which highlights dose-dependent potency of Novo Nordisk’s newest GLP-1 candidate. Management has mentioned to us in the past that they’ll submit 14 mg oral semaglutide to FDA, but there was no confirmation in the company’s NDA announcement of which dose(s) were filed. Even weight loss proved to be dose-dependent in PIONEER 3 (and other PIONEER trials): After 26 weeks, the highest dose of oral semaglutide gave ~2 lbs more weight loss vs. the 7 mg dose, and after 78 weeks, it still gave ~1 lb additional weight loss. The injectable formulation of semaglutide is currently under investigation for obesity (the phase 3 STEP trials and SELECT CVOT), and we’ll be curious to see how oral semaglutide is leveraged for its pronounced weight loss benefits as well. Despite some concerns over the tolerability of oral semaglutide (more on this below), we are very confident that this treatment will be approved for type 2 diabetes, and we’re terribly excited.

PIONEER 3 Summary of Results (Intent-to-Treat Analysis)


Oral semaglutide 3 mg

Oral semaglutide 7 mg

Oral semaglutide 14 mg

Sitagliptin 100 mg

A1c drop at 26 weeks





A1c drop at 78 weeks





Weight loss at 26 weeks (lbs)





Weight loss at 78 weeks (lbs)





*p<0.001 favoring oral semaglutide. **p<0.05 favoring sitagliptin. ***p<0.05 favoring oral semaglutide.

  • Unsurprisingly, GI symptoms were more common with the GLP-1 vs. DPP-4 and were also dose-dependent, affecting 32%, 35%, and 42% of participants on 3 mg, 7 mg, and 14 mg oral semaglutide, respectively, vs. 32% of those on sitagliptin. Moreover, 12% of patients randomized to 14 mg oral sema discontinued the medication due to adverse side-effects, compared to only ~6% of patients taking the lower doses and only 5% of patients on Januvia. The main criticism of on-treatment analysis is that it excludes data from participants who dropped off therapy, thereby inflating efficacy results by ignoring adherence challenges. It was reassuring to see that the ITT findings from PIONEER 3 were largely similar to the on-treatment findings. That said, adherence will only be more difficult in the real world compared to an RCT setting, and to this end, we wonder if having multiple doses of oral semaglutide on the market would ease patients through the GI side-effects and allow for smooth up-titration. The PIONEER 3 poster emphasized that the most common GI symptom was mild/moderate nausea, and that this nausea was transient.

  • Overall adverse events were balanced across all treatment arms in PIONEER 3. Severe AEs occurred in 10% of patients on 3 mg oral sema, 8% of patients on 7 mg oral sema, 9% of patients on 14 mg oral sema, and 11% of patients on Januvia. Hypoglycemia was numerically more common with sitagliptin vs. any dose of semaglutide (8% vs. 5%-8%), though the difference was not statistically significant (and we’d add that both DPP-4s and GLP-1s come with low hypo risk due to their mechanism of action).

  • PIONEER 3 enrolled 1,864 patients with type 2 diabetes and ran for 83 weeks total. Week 78 marked the end of the treatment period, with a low, medium, or high dose of oral semaglutide, or with sitagliptin. All pills were taken once-daily in the morning.

KOLs Emphasize Role of SGLT-2s, GLP-1s for Cardiovascular and Renal Risk Reduction, Offering Clinical and Patient-Centered Tips and Insight

In a master clinician roundtable focused on therapy selection for CV outcomes “and beyond,” thought leaders – Drs. Dan Drucker, Hertzel Gerstein, Jane Reusch, and moderator Carol Wysham – offered salient and actionable insight into the management of cardiovascular and renal disease in diabetes, naturally highlighting a role for SGLT-2 inhibitors and GLP-1 agonists. We’ve summarized their acumen in four sections below – these are all well worth a read!

Type 1

  • “[On GLP-1s or SGLT-2s or type 1], now we’re entering the valley of the vacuum of data. Jane, you’d be sued, but Europe is going to start using SGLT-2 inhibitors in type 1 very soon. FDA just said not so fast. We don’t have data that CV or renal risk reduction extends to type 1. Some would say it’s the same pathophysiology but it’s actually slightly different in some instances. Same deal with GLP-1 agonists: Is the pathophysiology we’re intervening with in GLP-1 agonists identical in type 1? I’d discuss with the patient: Some of you will say there’s a possibility you might get benefit, it is off label, and we have to make that decision. Some of you say no, there’s no RCT benefit and it might do harm. I would respect that. I would also respect someone who says, ‘Let’s give it a shot.’ That’s clinical judgment in the absence of perfect data.” – Dr. Drucker

CV Risk in Diabetes

  • One of the things Carol and I learned recently working collaboratively with cardiologists at ACC is that there are many unrecognized, undiagnosed patients with symptomatic heart failure. This was not really on our radar until the somewhat unexpected findings with empagliflozin. We’re now seeing that early on in type 2 there’s already decreased adiponectin and exercise tolerance. Cardiac dysfunction may be one of the earliest complications in diabetes.” – Dr. Reusch

  • “I work right next door to Dr. Bernie Zinman who goes to bed wearing pajamas that say ‘empagliflozin.’ You’ve strongly suggested diastolic dysfunction or heart failure in addition to coronary disease in this patient, and I strongly feel an SGLT-2 inhibitor with proven CV benefit such as empagliflozin would be appropriate. If it’s strictly for glucose lowering, a GLP-1 agonist would be excellent therapy – it gives comparatively more glucose lowering and a little weight loss would help. Of course, we’d discuss pros and cons with the patient. And you’ll all have point of care ultrasounds in your offices soon: If that showed diastolic dysfunction and impaired ejection fraction, I’d wear the pajamas.” – Dr. Drucker

Renal Effects of SGLT-2s

  • “I’m a bit of a surgical personality: This person must be on an SGLT-2 inhibitor. He’s heading toward CKD, dialysis, and renal replacement therapy. The only medication we have that is known to protect against further eGFR decline, even when there’s zero glucose lowering, is an SGLT-2 inhibitor. GLP-1 agonists also exhibit cardioprotection even at lower levels of eGFR, but the worst thing that can happen to this patient is dialysis and the need for a kidney transplant. Sue me [for the lack of indication] – this person gets empagliflozin.” – Dr. Drucker

  • “SGLT-2 inhibitors in the lower GFR category are probably still very cardio and renal protective, but they are going to lower the eGFR. If you don’t think about it and all of a sudden GFR is 29, it might startle you, the patient, or their PCP. You need to have the discussion that is actually renal protective in the long term because it’s reducing GFR in the glomerulus.” – Dr. Gerstein

  •  “I like to [check creatinine with an SGLT-2 inhibitor] mainly because I want the PCP to understand that it can drop and I don’t want them to worry about it too much. It’s hard to know what to expect and what’s ‘acceptable.’ I think you’ll tell them that if GFR is normal at 70-80, you can see a 10 point drop. You may see more at lower GFRs – just a tiny stitch in GFR will increase creatinine more. I want the patient and PCP to see what it is so they’re not panicking six months from now. I reassure that if I stop the drug it’ll probably come right back up, but we started the drug for a reason.” – Dr. Reusch

    • “Nephrologists don’t worry if an ACE inhibitor causes a 30% change in GFR” – Dr. Wysham

Additional Insight

  •  “Patients with gestational diabetes need to be followed very closely for development of persistent hyperglycemia. More than one-third don’t have their routine postpartum checkup for glucose. And correct me if I’m wrong, but this is the strongest indication for metformin for diabetes prevention.” – Dr. Wysham

  • “Just telling a patient at risk to change diet and activity is not effective. We often move to medications where, if we invested similarly in lifestyle regimens, that would have long-term therapeutic benefit not only on the number of medications needed, but longevity.” – Dr. Reusch

    • “We think it’s not effective to push this on patients because we don’t think it’s easy to adhere to lifestyle interventions. Well, people don’t adhere to medications. You have to put everything together.” – Dr. Gerstein

  • “[The amputation signal in CANVAS] may very well have been a real signal, or it may have been a fluke. We always forget that even if a trial has lots of people, events that occur rarely can occur randomly imbalanced. There are three trials, only one saw something: Unless there’s a really good reason why canagliflozin different – there are always reasons, but a really good compelling one – then maybe it’s not real. That’s my theoretical answer. In practice, I’d say to the patient that with one drug there was a signal. I’ll suggest another, and all the evidence so far does not suggest a risk: Are you comfortable with that? I probably wouldn’t prescribe canagliflozin even if I don’t believe there’ a real risk.” – Dr. Gerstein

LIRA-ADD2SGLT2i (Victoza vs. Placebo on Top of SGLT-2s) Results Further Support Efficacy and Safety of GLP-1 + SGLT-2 Combo Therapy

Dr. Lawrence Blonde presented results from the Novo Nordisk-sponsored LiraAdd2SGLT-2i study (n=303), which tested liraglutide (Victoza) vs. placebo as an add-on to SGLT-2 inhibitors. At 26 weeks, the Victoza group saw a 0.98% A1c reduction compared to 0.30% in the placebo group, for a treatment difference of 0.68% (95% CI: -0.89, -0.48) with most of the difference achieved by eight weeks and maintained through six months. On body weight, there was actually no significant difference: The Victoza group lost 2.8 kg (~6.2 lbs) on average, but the placebo group also lost 2.0 kg (~4.4 lbs), for a treatment difference of only 0.8 kg (95% CI: -1.73, 0.09). The session moderator commented during Q&A that this magnitude of improvement is unusual for a placebo group, and Dr. Blonde concurred by commenting that (i) it’s possible that investigators naturally emphasized lifestyle changes more to those not seeing improvements (because they were on placebo) and (ii) AWARD-10 saw a similar effect (6.8 pounds lost with high-dose Trulicity vs. 4.6 pounds with placebo). As would be expected, significantly more participants on Victoza than on placebo achieved an A1c <7.0% (52% vs. 23%, OR=5.13, 95% CI: 2.67-9.87) and <6.5% (34% vs. 10%, OR=6.77, 95% CI: 2.94-15.57). The Victoza group also had more reach an A1c <7.0% without symptomatic hypoglycemia or weight gain (48% vs. 19%, OR=4.76, 95% CI: 2.46-9.20) and achieve ≥1.0% A1c reduction with weight loss of ≥3% (30% vs. 8%, OR=5.23, 95% CI: 2.24-12.17). This trial completed May 2018 and was referenced in the company’s 2Q18 press release, and it joins Lilly’s AWARD-10 and AZ’s DURATION-8 in supporting the glycemic benefits to GLP-1 + SGLT-2 combination therapy. Interestingly, weight loss benefit was nearly additive with Bydureon + Farxiga in DURATION-8, though our understanding is that this trial was designed differently. To be sure, the question of GLP-1 + SGLT-2 combination therapy is certainly an interesting one, and we’ve heard substantial enthusiasm among thought leaders for the potential these hold together – given their very different but similarly glucose-dependent mechanisms of glucose-lowering and, apparently, of cardioprotection, it’s possible that using these together would give striking CV risk reduction, weight loss, and glycemic outcomes. At the same time, it’s important to get more patients on at least one of these agents, and cost will undoubtedly keep this combo out of reach for many – this is too bad since there may be far more downstream costs that could be avoided.

  • Safety findings were as expected and support the safety of combining this GLP-1 agonist with an SGLT-2 inhibitor. The most common adverse events with Victoza were nausea (26% vs. 6% on placebo), vomiting (8% vs. 2%), diarrhea (9% vs. 3%), constipation (9% vs. 0%), and decreased appetite (9% vs. 0%). Dr. Blonde emphasized that there was little to no increased hypoglycemia risk with Victoza on top of an SGLT-2 inhibitor: 9% with Victoza vs. 8% with placebo experienced any hypoglycemia (33 vs. 11 episodes). There were no episodes of severe hypoglycemia, and severe or blood glucose confirmed symptomatic hypoglycemia occurred only three times in the placebo group and zero times with Victoza. Importantly, there were no reports of acute renal failure, DKA, foot ulcer, or amputation, bolstering the understood safety profile of GLP-1 + SGLT-2 combination therapy. 

  • By design, 303 people with type 2 on a stable SGLT-2 dose (with or without metformin) were randomized to either liraglutide 1.8 mg or placebo in a 2:1 ratio for 26 weeks. An impressive 98% completed the trial, with 92% remaining on liraglutide and 93% on placebo until the end. Liraglutide was titrated per the label and Dr. Blonde commented that only ~a dozen participants did not reach the 1.8 mg dose (and remained at 1.2 mg). At baseline, the entire enrollment had a mean age of 55 years, diabetes duration of 10 years, A1c of 8.0%, FPG 160 mg/dl, and BMI of 32.2 kg/m2. An unsurprising 94% were taking metformin and by SGLT-2 inhibitor, 49.5% were on AZ’s Farxiga (dapagliflozin), 25.7% Lilly/BI’s Jardiance (empagliflozin), and 24.8% J&J’s Invokana (canagliflozin).

Real World Study of GLP-1s Finds No Significant Difference Between Bydureon and Victoza on CV Outcomes

A poster from Northwestern Medicine and United Health detailed a retrospective (2011-2015) real world study of GLP-1s and CV outcomes, demonstrating no significant difference between AZ’s Bydureon (once-weekly exenatide) and Novo Nordisk’s Victoza (liraglutide). On the composite cardiovascular outcome (congestive heart failure, stroke, ischemic heart disease, and peripheral artery disease), Bydureon (n=1,451) was associated with a HR of 1.33 vs. Victoza (n=7,531), but the 95% confidence interval was wide and crossed unity (0.73-2.39). Of course, one issue with observational studies of GLP-1 agonists and CV outcomes is that these agents seem to exert cardioprotective effects over a long period: For example, one can observe very gradual but meaningful separation of the Kaplan-Meier curves in the LEADER CVOT for Victoza; number of events was also not provided, but we suspect it was on the low side. By year two, only 340 patients on Bydureon and 2,209 on Victoza are included in the primary outcome analysis. We do note that the population on Bydureon was comparatively “sicker” than the Victoza group: A higher proportion of those on Bydureon had an A1c >10% (7% vs. 4%), obesity (41% vs. 37%), hypertension (73% vs. 65%), and dyslipidemia (66% vs. 59%). Because this is a retrospective observational study, direct and definitive comparisons cannot be made between these two GLP-1s. Still, these data align with growing consensus in the field of a class wide cardioprotective effect with GLP-1s, and buttress recent data that further support the EXSCEL CVOT as a positive trial for Bydureon. As a reminder, EXSCEL very narrowly missed on CV superiority for Bydureon (HR=0.91, 95% CI: 0.83-1.00, p<0.001 for non-inferiority, p=0.061 for superiority).

Diabetes Technology

WISDM Baseline Blinded CGM Data in Older Adults (≥60 Years) Without Recent CGM Use: 57% Time-In-Range, 35% >180 mg/dl, 5% Time <70 mg/dl,

International Diabetes Center’s Dr. Anders Carlson shared baseline data from the Helmsley- and JDRF-funded WISDM study (n=203) evaluating the effect of CGM in older adults ≥60 years with type 1 diabetes. While Dr. Carlson expects full results from real-time wear to be released later this year (ClinicalTrials.gov slates completion for October 2019), he was able to present blinded Dexcom G4 CGM data collected for up to three weeks (minimum 10 days) prior to the real-time CGM intervention. Similar to the CITY young adult CGM study, participants could not have used real-time CGM in the past three months and had to have a baseline A1c <10%. During this blinded baseline period, participants:

  • Spent 57% time-in-range (~14 hours/day)

  • Spent 35% time >180 mg/dl (~8 hours/day) and 12% time >250 mg/dl (~3 hours/day)

  • Spent 5% time <70 mg/dl (72 min/day) and 1.6% time <54 mg/dl (24 min/day)

  • Had mean coefficient of variation (CV) of 42% (While consensus has not yet been reached on the optimal CV, many thought leaders, including Dr. Irl Hirsch, recommend ≤33%.)

Fairly high hyperglycemia figures may be do to participant age  ≥60; regimens may have been purposefully skewed to protect against hypoglycemia. Dr. Carlson added that the hypoglycemia metrics are close to the newly agreed upon targets aiming for <4% time <70 mg/dl and <1% time <54 mg/dl. The only measurement found to be associated with hypoglycemia was (not surprisingly) hypoglycemia unawareness – those with unawareness spent twice as much time <54 mg/dl (19 mins/day vs. 39 mins/day; p=0.03). There were no differences in overall daytime and nocturnal hypoglycemia. However, among those with hypoglycemia unawareness, time <70 mg/dl and <54 mg/dl was significantly greater during the day than at night.

Interestingly, retirement and lower total daily dose of insulin were associated with higher time-in-range. Household income between $35,000-$75,000 was shown to be associated with lower time-in-range. We’re eager to see the full results of the CGM intervention.

  • The contrast between these data and that seen in the CITY (ages 14-25) baseline phase is stark: In that group, median time-in-range was just 35%, with median time >180 mg/dl up to 60%. Which group will benefit from real-time CGM to a greater extent?

Real-World Observational Study (n=93) of MiniMed 670G in Children/Adolescents Finds 38% Discontinuation Rate; For Those Who Continued, A1c Reductions (~0.3%) at Six Months Not Maintained at 12 and 24 Months

Boston Children’s Hospital’s Dr. Gregory Goodwin presented results from a real-world observational study (n=93) of Medtronic’s MiniMed 670G in children and adolescents (average age: 17.2 years) showing high rates of discontinuation (nearly 40%). All participants were on pumps prior to the study and received dedicated 670G training, consisting of a two-hour initial clinic visit, a two-week follow-up visit, during which Auto Mode was initiated, a 4-6-week follow-up visit with a diabetes nurse educator, and quarterly follow-ups thereafter. Despite the large degree of support, ten patients (11%) never initiated Auto Mode, with the predominant issue being sensor difficulties. Of those who continued into Auto Mode, an additional 25 participants (27%) withdrew from the study, amounting to a total discontinuation rate of 38%. As Dr. Goodwin noted, this rate stands in significant contrast to clinical trials finding withdrawal rates closer to 5% – we wonder if the nearly 8x rate of discontinuation in this study had more to do with demographics, degree of follow-up/support in previous trials vs. this study, or something else. Average time to withdrawal from the study was 2.5 months, and participants mostly commonly cited frequent alarms, forced exits, and calibration requirements as reasons for discontinuation. Withdrawals were more common in older patients, as the mean age of those who stopped using the system was 18.4 years (just over the threshold of college age, unsurprisingly) vs. 16.6 years for those who stuck with it. Dr. Goodwin concluded that “more user-friendly technology” is needed for widespread, effective use of hybrid closed loop systems in type 1 diabetes. While we fully agree that more user-friendly tech will benefit greater swathes of the population, it’s hard to ignore the fact that 670G is a first-generation system that FDA approved in ~three months based on a single-arm study; perfect first-to-market systems are rarely expected to be perfect. Medtronic has a rich closed loop pipeline which promises to push its offerings in the direction of user-friendliness one year out and beyond (and has already enhanced the Guardian Sensor 3 transmitter so that it doesn’t require as much user interaction). Nearer term, Tandem’s Control-IQ, slated for a US launch between “summer and end of Q3 (September),” will provide the first FDA-approved hybrid closed loop in the US that leverages the factory-calibrated Dexcom G6 and TypeZero’s (Dexcom) algorithm with automatic boluses. Future systems promise to make meaningful gains along the adoption curve.

  • Patients who did remain on the 670G (n=58) saw modest outcomes. While A1c significantly declined by 0.27 percentage points (baseline: 8.0%) at six months, statistical significance was not maintained at 12 or 24 months. Still, Dr. Goodwin pointed out a significant association was shown between lower A1c and greater time in Auto Mode. Average 670G duration of use was 9.6 months, making this study the “longest longitudinal follow-up” of the system to Dr. Goodwin’s knowledge.

    • Dr. Goodwin also pointed out the “very wide range” for time in Auto Mode, which averaged 50% but spanned from 5%-90%.

Ms. Davida Kruger Presents Dr. Bergenstal’s Nine Steps to Interpreting CGM; Plans to Bring Direct-To-Consumer CGM Study in Primary Care to Large Health Systems

In a session sponsored by Abbott, Henry Ford Health System’s Ms. Davida Kruger described the nine steps she takes to instigate better patient communication surrounding the AGP CGM data display. [This procedure was first outlined by Dr. Rich Bergenstal at ADA 2018.] Ms. Kruger first recommends checking for adequate data (14-days of readings are ideal). Next, she marks up the AGP with her patients, noting factors that may affect a patient’s individual management plan, such as wakeup, meal, and insulin dosing times. Importantly, prior to offering her own analysis, Ms. Kruger asks patients what they see. She underscored that simply asking is not enough – to reap the full benefits of this step, physicians actually have to listen to the response! Following this brief discussion, Ms. Kruger identifies patterns of hypoglycemia with her patients, ensuring that she is “treating the cloud.” To this end, Ms. Kruger cautioned clinicians against only looking at the darker trace on the AGP, which reflects just 50% of a patient’s glucose readings. Hypoglycemia trends are identified next followed by areas depicting wide glucose variability. In working to mitigate this variability, Ms. Kruger explained that clinicians tend to jump directly to an insulin dose adjustment. However, she underscored that there are many factors that could contribute to such activity. We love Ms. Kruger’s suggestion to then compare the current AGP to past AGPs and reinforce successful strategies. Focusing on what is working, rather than only chastising patients for their mistakes, is a crucial component to Adam’s Bright Spots mentality. Lastly, Ms. Kruger agrees on an action plan with her patients, and then makes copies of the AGP for the patient and EMR. While we recognize that many clinicians new to CGM may find the data overwhelming at first, having a simple nine-step plan will hopefully prove helpful in standardizing their approach.

  • During Q&A, Ms. Kruger referenced an ongoing project to introduce CGMs in primary care. We assume she was referring to the Helmsley-funded study conducted by the Jaeb Center in collaboration with Cecilia Health and Wisconsin Research and Education Network, first described at ATTD. As Ms. Kruger explained, the study allows patients to choose their device (Dexcom G6 or Abbott FreeStyle Libre) and initiate the CGM outside the clinic with remote education provided by a CDE. She noted that when the pilot wraps up, the team “hopes to do a huge study.” Large systems including Geisinger and Blue Cross have already “been approached.” At ATTD, Dr. Roy Beck expected to also incorporate decision support tools and mental health support/coaching as the study expands. We’re very excited about the study, which we see as an important stepping stone towards acquiring CGM without a prescription.

Dr. Robert Califf on Harnessing Google Search History for Suicide, Diabetes Prevention; PCORnet “Open for Business”

In a tour de force presentation on the impact of big data on healthcare, Duke University School of Medicine/Verily’s Dr. Robert Califf used diabetes as an example of how people’s search history can be harnessed towards prevention. For context, he described how when people currently search for depression on Google, a pop-up appears asking if they’d like to take a questionnaire. If they choose to do so, they are directed to take the PHQ-9, a depression screening and diagnostic tool. According to Dr. Califf, “the very large number of Americans who have filled out the questionnaire and have been shown to be seriously and significantly depressed” are then referred to a patient advocacy group. He acknowledged that the ideal next step would be referring these respondents to a psychotherapist; however, it then becomes very complicated to determine which psychotherapists are worthy of referral and how to match patients appropriately. Dr. Califf pointed to diabetes as a disease state posing similar, yet complex promise. He posited that many individuals likely search for topics related to diabetes prior to seeking medical attention – some of the one billion Google health queries made every day. If Google can figure out how “to link up physicians” with patients’ search history accordingly, patients may be reached proactively. Dr. Califf acknowledged the obvious privacy concerns, though stipulated that “the majority” of people would be willing to let their search history be used to improve their health (we already do it to improve our entertainment and online purchasing experiences, after all).

While it may be too extreme for physicians or health systems to reach out to individuals based on their search history at the moment, we think the current middle ground in which patients searching prediabetes are offered the CDC’s “one-minute” prediabetes risk test is still immensely valuable. Perhaps eventually the test can pop up from other related searches beyond the most obvious terms.

  • Dr. Califf provided an updated on PCORnet, the “network of networks,” which aims to combined data across health systems to accelerate research. To date, the database accounts for one-third of the US population and comprises 30 health systems and two health plans. Combined, more than 100 million patients who have had a medical encounter in the past five years are represented. Dr. Califf announced that the system is “now open for business.” We’ll be very eager to see the kind of innovative research that arises from this massive undertaking.

Pathway Genomics’ Dr. Michael Nova on Leveraging AI and Genetics to Identify Optimal Diets for Diabetes and Obesity

Pathway Genomics’ Dr. Michael Nova described how his company is leveraging artificial intelligence and genetics to determine disease risk and optimal treatments. In one example, Dr. Nova described how diabetes is a “super-problem” in Mexico, with diabetes was identified as the country’s “number one killer.” Pathway Genomics sequenced genomes from 2,000 type 2 adults, creating a “very rich database” of genetic information while also incorporating clinical attributes such as diet, lifestyle, and comorbidities. Dr. Nova believes this work reflects “the largest sequencing database of type 2s with clinical information.” An artificial intelligence algorithm concluded that there was a significant difference in the ideal diets for adults with type 2 diabetes from Mexico. For example, he noted that many more Mexicans would benefit from switching to a high-fat diet – we’re not quite sure how this conclusion was reached definitively, considering that this was not a crossover study. Moreover, Dr. Nova’s algorithm was able to use the data to predict the onset of diabetes with ~90% accuracy, which he modestly characterized as “pretty good” – this is exactly the kind of data that would be useful in risk stratification and getting greater ROI from prevention activities. Dr. Nova sees a lot of promise for this kind of narrow application when it comes to AI in healthcare. While he still thinks AI is not quite ready to be used more broadly, he is excited to see how it can be applied in specific disease states. We frequently hear – most recently at SXSW – that AI is ready for prime time when it comes to image recognition, but broader applications in domains such as predictive analytics have a bit left to prove.

  • Dr. Nova provided another example of harnessing genetic information to identify ideal diets in patients with obesity. Pathway Genomics genotyped over 300 contestants from the Biggest Loser TV show, revealing that over 97% of these individuals exhibited high or above average genetic markers for obesity – resulting in an obesity risk far greater than would be expected in the general population. By applying a convolutional neural network, Dr. Nova’s team identified clusters from the 2.5 million+ data points. In this way, they were able to determine the individuals that would respond best to a certain diet such as low-fat vs. low-carb.

  • Applying genetics to obesity and diabetes is a burgeoning field, with 23andMe recently creating a new genetic report for type 2 diabetes risk and partnering with Lark Health to incorporate genetic information into Lark’s Wellness Program and CDC-recognized Diabetes Prevention Program. While there has certainly been some skepticism as to the validity of this approach, we can appreciate the appeal on the consumer end. According to Dr. Nova, if patients receive genetic information about themselves, they tend to demonstrate “much more” adherence to weight management regimens and are “more liable to change behavior.” We would be fascinated to see a study comparing weight loss outcomes between patients who received genetic information vs. those who do not. Indeed, we wonder if the genetic information would even have to be accurate or if it is simply the notion that their regimen is based around personalized, scientific data that patients find so compelling.  

  • We were very interested to hear an audience member ask Dr. Nova whether precision medicine should be focused on prevention or treatment. He responded that, while “prediction is great,” ultimately, “everyone wants to be told what to do.” He provided oncology as an example of a field in which precision medicine is already being applied in this way – a specific tumor is analyzed to determine the optimal therapy. While we appreciate from a business standpoint that it is much easier to sell a treatment (i.e., consumers are more likely to take action when they are already diagnosed with a condition), prevention presents a plethora of opportunities to cut down on healthcare costs, not to mention promote general wellness. Moreover, the actual science need not be nearly as complex as identifying a specific therapy – even risk stratification could play an enormous role in reaching out to patients at risk of diabetes or other chronic conditions. In our view, there’s no shortage of resources, so why ask which one we should focus on? A more interesting question, at this point, would be: “Which has more near-term potential, precision medicine in prevention or in disease treatment?”

Glycemic Targets and Guidelines

Drs. Buse and Wilt Debate ADA vs. ACP Guidelines for A1c: Is there a Benefit to Targeting <7%?; Considering RCT Evidence, Target vs. Achievement, Cost Effectiveness, and Practical Implications

Drs. John Buse and Timothy Wilt debated the resolution, “A reasonable A1c goal for many nonpregnant adults is <7%.” While Dr. Wilt sought to construct an argument that an A1c <7% has not been associated with any meaningful, long-term outcomes benefit, Dr. Buse aimed more to illustrate the practical function and implications of an A1c target – and why it’s beneficial for many to target <7%. Importantly, this debate was focused on type 2 diabetes. As Dr. Marie McDonnell pointed out in moderating the debate, virtually all professional guidelines leave room for patient-tailored targets. In his rebuttal, Dr. Buse said, “There’s no dichotomy between the major statements of both organizations. For many, <7% is reasonable. For most, 7%-8% is good.” Indeed, Dr. McDonnell closed by asking whether the question is one of semantics: “It occurs to me that when we set up a target, we often don’t arrive at it. If it’s <7% in a trial, we might get to 7.5%. Could you discuss target vs. achievement?” In response, Dr. Buse said that this was actually the root issue, quipping that “clearly I didn’t make that point strong enough” (for reference, an audience vote that began at an ~80/20 split in favor of <7% and ended at ~60/40). Below, we outline each presenter’s argument:

Dr. Buse concluded that the crux of achieving optimal diabetes outcomes is the attained A1c and how it is approached – an A1c goal of <7% is “fundamentally a tactic” aimed at achieving lower average A1c: “I do think a target of <7% is a tactic to get to 7%-8%, which I do think is a reasonable A1c. One of the slides I didn’t show is data from Scandinavia showing that if average A1c achieved over 20 years is 7.5% or less, there are almost no clinically meaningful microvascular complications. The fear is that the average A1c is going to be 7.8% or 7.9%, not the 7.4% it’s at now,” and that will lead to more complications and poorer quality of life for patients. This is precisely the reaction we had when these guidelines were released: The ACP statement gives little guidance for personalization (or use of CGM or newer therapy classes), making it easy to apply a 7%-8% target to all type 2s and allowing population-level rises in A1c. Evidence aside, we appreciated one of Dr. Buse’s closing remarks: “Would you as an individual aim for an A1c target of <7% if you had diabetes?” We suspect most in the room would have answered yes.

  • Dr. Buse demonstrated that the statement is question is self-evidently true: There are many individuals for whom an A1c <7% is reasonable. He presented data from the Carolina Data Warehouse (~12 hospitals and 200 practices in NC, n>70,000). Overall, 50% were already achieving an A1c <7%. Among three key groups – (i) those who already have an A1c <7% without adverse events and >10 year life expectancy (~20% of total population), (ii) women of childbearing potential (~10% of total population), and (iii) people with CVD or CKD (~50% of total population) – 39%, 42%, and 50% were already achieving an A1c <7%, and many more could get there safely (not on insulin or SUs).

  • Dr. Buse also explained that changing the A1c target to <8% could dramatically reduce clinician ability to prescribe lifesaving drugs for patients. He showed that many current prior authorization forms require that a patient has an A1c ≥6.5%, based on the ≤7% target from ADA/EASD. If a shift toward higher recommended A1c took place, and that translated to prior authorizations, it would disqualify many patients already “doing well” from being able to use newer, better therapies – including those used for CV risk reduction in addition to glucose lowering (GLP-1s and SGLT-2s).

Dr. Timothy Wilt, who helped author the ACP’s contentious 2018 guidelines, argued that a target of 7%-8% for most people with type 2 diabetes incorporates both clinical and patient values, but he also left room for personalization based on patient discussion, preferences, health, and life expectancy plus costs of care and treatment burden. Interestingly, he commented, “My primary care colleagues and a large majority of the patients I treat were thankful and receptive of the recommendation.” One key issue emerged later in his presentation: With pay for performance measures, HCPs are currently asked to target an A1c of 7% in a 65 year old on three drugs with an A1c of 7.5% – and we certainty agree with Dr. Wilt that rigidity of these guidelines is a huge problem. But should we be more flexible with all type 2s? In evaluating the benefits vs. harms of specific A1c targets, Dr. Wilt ran through ACCORD, ADVANCE, UKPDS, and VADT: By his characterization, two of these showed no benefit (ADVANCE, VADT), one mixed or small benefit (UKPDS – insulin/SU and metformin arms), and the last showed harm (ACCORD). To be fair, as Dr. Buse pointed out, the early stopping of ACCORD for harm taught us not to blindly lower A1c with reckless ambition; we also feel that all of these trials would be designed very differently if conducted with modern medicine and technology, and many question whether they created a big enough glycemic divide to show any difference: Is there no benefit simply because it has not been demonstrated in type 2 diabetes? Nevertheless, Dr. Wilt did aptly point to a fairly large gap in the RCT evidence, and he further cited the harm that lower A1c targets can cause – namely, higher healthcare burden, costs, and increased adverse events.

  • Dr. Wilt strongly emphasized the importance of cost-effective and cost-conscious care, and this is where the two speakers seemed to diverge most strongly: Is it more cost-effective to lower glucose to <7% with a pricier GLP-1 agonist (potentially causing weight loss, preventing CV events, avoiding hypoglycemia and hospitalizations) or to target 7%-8% with NPH or premix insulin (a cheaper medication but one associated with more hypoglycemia – which is not necessarily A1c dependent – and weight gain)? To be sure, this hypothetical vastly over-simplifies a complex debate, but it seemed to us that Drs. Buse and Wilt were taking disparate stances on this question. We certainly agree with Dr. Wilt’s assertion that diabetes medications cost a lot, US healthcare is poor value, and high list prices are contributing to sky-high healthcare costs – he specifically questioned the near-ubiquitous use of insulin pens, “how much are we willing to pay for convenience?” But while Dr. Buse asserted using newer, pricier therapies like GLP-1s and SGLT-2s up-front prevents costly complications down the line, Dr. Wilt directly questioned the data supporting their cost-effectiveness and whether they would meet ICER’s $150,000/QALY threshold.

Endocrine Society Guidelines for Older Adults Recommend Individualized Glycemic Targets (Deprioritizing A1c) and Hypoglycemia Minimization

The Endocrine Society just published a Clinical Practice Guideline on Treatment of Diabetes in Older Adults (press release); following an overview of the guidelines from Dr. Derek LaRoith, Dr. Anne Peters moderated a series of case-based discussions between three members of the writing committee – Drs. LeRoith, Marie McDonnell, and Mark Molitch – designed to elucidate key takeaways from the new guideline. Epidemiologically, the guidelines emphasize that prediabetes is particularly prevalent in older people, but, positively, interventions aimed at diabetes prevention are very effective in this group. Similarly, type 2 prevalence increases with age, driving microvascular and macrovascular complications. As such, regular screening for prediabetes and diabetes is important. The guidelines emphasize shared decision-making and individualization as key, given heterogeneity in the health of older individuals. They also note the unique issues affecting older individuals with diabetes: sarcopenia, frailty, and cognitive dysfunction, all of which can contribute to lower treatment adherence, hypoglycemia risk, falls, and loss of independence. As the overall and diabetes populations age, the ability of endocrinologists and PCPs to effectively and safely treat older individuals with diabetes and prediabetes is only going to become more important, and we’re glad to see this focus from The Endocrine Society. Read on below for key takeaways from the five cases presented.

  • Screening: In what Dr. Peters characterized as the most surprising recommendation, the guidelines advocate for a 2-hour, 75 g oral glucose tolerance test (OGTT) as the best test of diabetes in older individuals. While the panelists were sympathetic to the challenge such a test poses to many patients – both in terms of comfort and time – Dr. McDonnell explained that the literature is very clear that the diagnostic specificity and sensitivity of A1c declines with age. Using A1c (or FPG) alone runs the risk of missing diabetes diagnoses in older adults, and this is most relevant to those at highest risk of diabetes. Dr. McDonnell explained that clinical judgement is also important – in an individual with metabolic syndrome, high triglycerides, low HDL, and an A1c of 6.2%, an OGTT is highly recommended. On balance, however, Dr. Molitch commented that both accuracy and convenience are important: ADA moved to include A1c in its diagnostic criteria because it’s easy. And while not all clinicians will move to use the OGTT commonly, he said it’s important to recognize that A1c is not as accurate in older adults. Dr. Peters added that how much information the patient wants is also important – in this case, a highly-functional and engaged 78 year old woman was presented, and Dr. Peters commented that a diabetes diagnosis would change how she would treat the patient. After all, as Dr. LaRoith explained, a precise diagnosis will have a meaningful impact on both treatment and what medications can be prescribed.

  • Glycemic Targets: Blood glucose targets are prioritized over A1c (particularly in light of lower accuracy, above), and glycemic targets should be chosen via shared-decision making based on assessment of the patient’s overall health. The panelists endorsed the established concept of “bucketing”  or “grouping” patients into good, intermediate, and poor health to help determine A1c targets. The idea of an “A1c floor” was keenly addressed, particularly for patients on hypoglycemia-causing medications – if an older adult has an A1c of 6.5% on nothing that can cause hypoglycemia (SUs or insulin), that’s great. Dr. LeRoith explained that targets can also move: Particularly when starting at a higher A1c, it’s best to lower gradually and proceed carefully.

  • Medication Selection and Hypoglycemia: In those over 65 years old, outpatient diabetes regimens should be specifically designed to minimize hypoglycemia risk. Glycemic targets should be tailored to overall health and management strategy, regardless of medication used. The panel considered an 80 year old woman (BMI 24.3 kg/m2, eGFR from 60 to 30 in past two years) with 20+ years of type 2 diabetes who is careful about food, medications, and exercise; she’s been on basal-bolus for 10 years but has decreased appetite, is alone most of the day, and is having mild hypoglycemia several times a week. While the audience poll suggested switching to basal insulin plus a DPP-4 inhibitor, Dr. Molitch did question whether a DPP-4 would be effective after 20 years of diabetes, explaining that a post-meal reduced dose of prandial insulin could also be an option. Due to decreased appetite and normal BMI, he warned against a GLP-1 (and did the same for metformin due to GFR) and emphasized that serious consideration should be given to raising glucose targets. Dr. McDonnell highlighted the value of CGM for figuring out exactly what’s happening. One danger, the panel said, is that patients can be seriously insulin deficient – and taking them off insulin can do more harm than good. Per Dr. McDonnell, insulin production in older adults is often not just low, but late and inefficient (C-peptide can be useful if “insulin therapy doesn’t necessarily match the natural history of the disease”). Dr. Peters noted value in caregivers for insulin dosing but also in simplified insulin regimens (e.g. 2x daily dosing).

  • Lipids: In those 65 years and older, statin therapy and an annual lipid profile are recommended for reducing CV and mortality risk. No specific LDL-C targets are recommended in this guideline, though the authors do recommend generally less strict LDL-C targets for those ≥80 years old with short life expectancy. See diabetes-specific cholesterol guidelines from AHA/ACC.

  • Blood pressure: In those 65 years and older, an ACE inhibitor or ARB should be used as first-line therapy. For those 65-85 years old, a target of 140/90 mmHg is recommended to decrease CV and renal risk. Higher-risk groups (previous stroke, progressing CKD, albuminuria) can be considered for a lower, 130/80 mmHg target, at which point monitoring and screening for orthostatic hypotension is important. Conversely, a higher target may be considered for those with high disease complexity and poor health (145-160/90 mmHg), and shared decision making is always important. Dr. McDonnell praised ADA and other organization for not applying the SPRINT trial and a 130/80 mmHg target to diabetes universally, noting that autonomic neuropathy common in patients with diabetes can lead to negative side effects when blood pressure is too low.

Endocrine Society Commissions Task Force to Develop Innovative Models of Care; Dr. Robert Gabbay Address HCP Burnout; 1.4% A1c Reduction Just From co-PCP/Pharmacist Diabetes Management?

In a session chaired by Joslin’s Dr. Robert Gabbay, we learned that the Endocrine Society has commissioned a task force to examine innovative models of care. As Dr. Gabbay asserted, such innovation can help to tackle the concerning amount of physician burnout in the US by “reconnecting to the joy and work that had us follow this career in the first place.” The impact of burnout, Dr. Gabbay explained, is widespread, affecting physicians both professionally and personally – he reiterated that, every day in the US, a physician commits suicide. Dr. Gabbay identified five major themes that can help to mitigate burnout: (i) finding meaning; (ii) having a sense of autonomy; (iii) comradery; (iv) equity; and (v) co-creating solutions. Interestingly, the degree to which these themes must be fulfilled is fairly small: evidence suggests that physicians who spend at least 20% of their professional effort focused on work they find “most meaningful” are at a “dramatically” lower risk for burnout. In fact, a ceiling effect at 20% is observed, meaning that physicians who spend even 50% of their time on meaningful pursuits are no less at risk of burnout than those who spends 30% of their time. How can physicians get to that 20%? Dr. Gabbay outlined four (easier-said-than-done) steps: (i) identify a passion; (ii) make a business case; (iii) speak to administration; and (iv) seek out funding if needed. The need to make a business case was particularly emphasized during the session. We were intrigued by a question from an audience member positing whether medical education should be revamped to help future physicians acquire business acumen. Dr. Gabbay responded: “Absolutely. If we’re going to take back our practice, we have to have some knowledge.” See below for examples of innovative models of care presented by members of the Endocrine Society taskforce.

  • Vanderbilt University Medical Center’s Dr. Michelle Griffith described how her health system is using direct-to-patient telemedicine to improve patient outcomes while also enhancing physicians’ experiences. A nationwide survey (n=4,345) on attitudes surrounding telemedicine in primary care indicated that 52% of respondents would like to see their own provider via telemedicine. 15% would consider leaving their current provider to see one who offered telemedicine. For endocrinologists, telemedicine improves space utilization, increased efficiency for certain types of visits, allowed for greater practice variety, and served as a practice differentiator. After initiating telemedicine at Vanderbilt, 91% of patients felt satisfied overall and 81% indicating feeling a connection with their provider equivalent to an in-person visit. We think telemedicine will play a major role in diabetes management given the need for high-touch care. Moreover, endocrinologists are at high-demand – we were shocked to hear that there are “zero or one” endocrinologists in the state of California between Sacramento and the Oregon border. For patients living in this area, having access to an endocrinologist via telemedicine could literally be life-changing. The big hurdle for telemedicine today – according to a Rock Health survey and JAMA letter – is moving adoption beyond young, urban-dwellers toward less mobile populations in more remote locations.

  • Parkview Health’s Dr. Jeffrey Boord advocated for co-management of diabetes between primary care physicians and pharmacists. This solution has been implemented at Parkview Health, and according to Dr. Boord, “patients really like having another person helping,” and “the primary care physicians love it.” In this model, clinical pharmacists focus on patients’ diabetes management, developing individual care plans, providing education, initiating therapy, and monitoring patient adherence. As Dr. Boord put it, pharmacists are “masters of barrier removal,” as they are uniquely equipped to navigate the complex web of formularies and prior-authorizations, as well as consider and respond to factors like reluctance and affordability. Outcomes at Parkview have been very strong: at just three months, patients saw an A1c reduction of 1.4% (baseline A1c: 9.8%). We think this is a great model – it allows primary care physicians to focus on the myriad other healthcare concerns that arise in a general doctor visit, while still providing patients with access to high-quality diabetes care.

  • Other examples of innovative models that were presented included team-based care, a transfer summary for pediatric to adult transitions in care, and electronic consults (eConsult). The eConsult essentially formalizes the work that many specialists already do for primary care physicians. PCPs routinely will ask for specialists’ advice via email. With eConsult, the process is recorded and, more importantly, can be billed. UCSF’s Dr. Elizabeth Murphy noted that at San Francisco General Hospital, there are over 50,000 eConsults annually, substantially saving time. 50% of endocrine referrals were handled by eConsults and never made it to the clinic. While the proportion is smaller for diabetes (20%-25%), it’s clear that this solution is able to cut down on unnecessary clinic visits.

Dr. Jack Leahy Reinforces Importance of Insulin in Type 2 Treatment in Light of Newest ADA/EASD Consensus Guidelines Promoting GLP-1 as First-Line Injectable

In reviewing the new ADA/EASD consensus recommendation of GLP-1 agonists as first-line injectable therapy, Dr. Jack Leahy remarked that insulin is “a little under respected” and emphasized that “insulin is an important part of what we do in type 2 diabetes.” Although he understands the reasons behind the recommendation, Dr. Leahy does feel a “general sense that insulin has been underrepresented on these guidelines.” This sentiment is surely understandable, as basal insulin is placed near the bottom of each specific part of the treatment algorithm. Nonetheless, the evidence supporting the recommendation of GLP-1s over basal insulin as the first line injectable in type 2 treatment (save specific circumstances of severe insulin deficiency) is strong; we particularly note the mass of evidence showing similar A1c lowering and superior effects on weight with GLP-1s compared to insulin. Generally, we remain thrilled to see GLP-1s promoted over basal insulin for most patients, given preferable effects on weight, the option of once-weekly dosing, little to no hypoglycemia risk, and long-term beta cell preservation; we’ve also seen this recommendation received with notable enthusiasm among thought leaders. Still, Dr. Leahy’s general point does stand, and the community should not disregard the role of insulin in treating type 2 diabetes.

  • Discussing next-gen basal insulins, Dr. Leahy emphasized that the defining clinical benefit of these agents is a lower risk of hypoglycemia, rather than superior A1c reductions. Dr. Leahy reviewed clinical trials for both Sanofi’s Toujeo (insulin glargine U300) and Novo Nordisk’s Tresiba (insulin degludec), showing that although significant A1c differences were not observed between these next-gen agents and their predecessors, significantly lower rates of hypoglycemia (especially nocturnal) were seen. On this front, Dr. Leahy argued that next-gen basals should be targeted to people “doing okay with A1c but having trouble with lows, especially at night.” In discussing which next-gen to pick for a patient, Dr. Leahy emphasized that there are more similarities than differences between Toujeo and Tresiba (we couldn’t agree more) and that as a provider “you choose what you want to choose or what your insurance company forces you to choose. There’s not much of a difference either way.” We agree with Dr. Leahy’s thinking here, and maintain that while comparisons between Toujeo and Tresiba are interesting and inevitable, the focus should be on getting as many patients as possible on one of these next-gen agents rather than older insulins.

Dr. Philip Zeitler Challenges Therapeutic Inertia in Youth-Onset Type 2 Diabetes: “We Shouldn’t Be Waiting Long to Treat This”

As part of a “Meet the Professor” session on youth onset type 2 diabetes, Dr. Philip Zeitler advocated for more aggressive treatment of this population. Dr. Zeitler emphasized that with youth onset type 2 diabetes, “the problem is that you [the provider] really can’t wait. A big problem we have is therapeutic inertia. The rise in A1c in these kids is not linear, but really exponential. You’ll see them for a few visits, and their A1c is in the fives, and then it will suddenly jump all the way up to nine. We shouldn’t wait long to be treating these kids.” We’re glad to see Dr. Zeitler shine a light on therapeutic inertia in the youth population, as this specific sub-population was not directly addressed in ADA’s recent Summit on Therapeutic Inertia.

During Q&A, some questioned whether such intense pharmacotherapy is appropriate in this population, especially considering the young age of many. Shouldn’t we first focus on improving lifestyle choices in these young children, the audience wondered? Dr. Zeitler provided the following response on balancing lifestyle intervention and pharmacotherapy or surgery: “This is a serious lifestyle disease, but you have to remember that these are also children with serious health risks that we cannot ignore just by thinking that this is a lifestyle disease. You have got to treat what is in front of you. It’s not just their lifestyle, and their lifestyle won’t change right away. You have to keep them from getting into trouble [by using other approaches].” Dr. Zeitler continued: “It may seem that they’re only kids and that we should therefore focus on lifestyle interventions. But the fact of the matter is that this is a more aggressive disease in kids than in adults [see RISE peds at ADA 2018]. They are losing beta cell function more quickly and developing complications more quickly than adults. [Treatment] may also include referrals for bariatric surgery. Surgical therapy is much more successful for almost all of these problems that this population faces.”

  • Dr. Zeitler explained that he views prescribing metformin to the youth population with type 2 onset as a “beta cell stress test.” He explained that “if you can’t get them to an A1c of 6.5% with metformin, their chance of going out of control is very much increased.” In terms of next-line therapy, insulin is prescribed (to our understanding, this is a function of label indication more than anything – Victoza has been submitted for youth type 2). However, Dr. Zeitler commented that A1c often doesn’t improve with insulin because “this is a very challenging population with a lot of different barriers affecting treatment” – we imagine he was referring to adherence.

Epidemiological Insights and Post-hoc Analyses

Dr. Carol Wysham presented results from a cross-sectional, observational study on A1c variability and diabetes complications using data from the T1D Exchange. To measure A1c variability, intrapersonal mean and standard deviation (SD) of available A1c values were calculated and the 50th percentiles of mean A1c and SD were used to categorize patients four groups (“Mean A1c” x “A1c variability” matrix). Those in the high mean/high SD group, compared to the other three groups, were significantly younger (30 vs. 37-44 years old) and had a shorter duration of type 1 diabetes (16 vs. 21-26 years); they also had a lower BMI and lower usage rate of CGM (24% vs. 27-44%). Most importantly, patients with a high mean/high SD were also significantly more likely to experience diabetes-related severe hypoglycemia, DKA, peripheral neuropathy, nephropathy, and renal disease. Rates of DKA were particularly elevated in this group when compared to those with a low mean/low SD (HR=13.32, 95% CI: 2.95-62.05). While these results are not necessarily surprising, they do further support the probable link of glycemic variability with severe hypoglycemia, DKA, and long-term outcomes and complications. Dr. Wysham noted that application of these results to clinical settings is possible, as high A1c variability may help identify patients at a particularly high risk for complications.

Significant Association Between Recurrent Hypoglycemia and Care Fragmentation Shown in Chicago

Northwestern University’s Dr. Clare O’Connor presented data demonstrating a significant association between recurrent hypoglycemia and fragmentation of care in Chicago-based people with diabetes. Recurrent hypoglycemia was defined as a patient seen in the emergency department and/or admitted to the hospital for hypoglycemia on more than one encounter, while fragmentation was defined as emergency department visits and/or hospitalization for hypoglycemia at more than one participating site. Data were collected from the Chicago HealthLNK Data Repository during 2006-2012, comprising six institutions across Chicago. To compile patient data, an algorithm sifts through patient records, removes personal health information, and links patient records between institutions into one database. Out of a total of 1,775,069 patients, 187,644 were determine to have diabetes (11%). 9,741 (5%) had at least one hypoglycemia event recorded, and 2,897 (1.5%) were shown to have recurring hypoglycemia. Fragmented care was identified in 10.5% of those with recurrent hypoglycemia – a significant association (IRR: 1.12, p=0.036). Dr. O’Connor concluded that there may be a causative association between the number of hypoglycemia events and fragmentation of care. Given that incidence of hypoglycemia was also shown to be highly associated with mortality as compared to those with diabetes but without hypoglycemia, Dr. O’Connor hopes to conduct further analysis examining the relationship between mortality and fragmentation of care. Hopefully these results generate greater discussion on how to provide continuous, wraparound care without patients resorting to visiting multiple health systems.

New DCCT/EDIC Analysis Finds that Early Glomerular Hyperfiltration is Not Associated with Development of Stage 3 CKD 

Northwestern’s Dr. Mark Molitch presented findings from a new DCCT/EDIC analysis indicating that early hyperfiltration in type 1 diabetes is not directly associated with an increased risk of developing stage 3 CKD. Of note, previous clinical studies have suggested a potential association between glomerular hyperfiltration and increased risk of later developing albuminuria in type 1 and type 2 diabetes; however, the totality of evidence is mixed and unclear. This specific analysis used long-term eGFR data from the DCCT and follow-up EDIC study to examine a potential link between the two variables. The group defined hyperfiltration as an eGFR of >140 ml/min/1.73 m2 (literature definitions vary between 120-150 ml/min/1.73 m2). Kaplan-Meier survival curves of the two groups (those with hyperfiltration and those without) did not indicate a significant difference in the rate of progression to an eGFR <60 ml/min/1.73 m2, the defined threshold for stage 3 CKD. Analysis using different definitions of hyperfiltration (>130 or >150 ml/min/1.73 m2) yielded similar results. The trend held when individuals were analyzed based on conventional vs. intensive treatment enrollment in the DCCT, representing encouraging consistency. That said, this is one piece of observational data, and we’re curious to learn more about the clinical value of hyperfiltration – if it were associated with progression to renal disease, would hyperfiltration serve as an independent marker of risk or simply identify patients whom are already known to be at high risk (i.e., poorer glycemic control, hypertension, higher BMI, etc).

  • We note that the “hyperfiltration theory” of diabetic nephropathy progression was cited by J&J researchers when beginning the CREDENCE renal outcomes trial for Invokana. In 2014, Dr. Norman Rosenthal (Canagliflozin Compound Development Team Leader at Janssen) explained to us that the hyperfiltration model is certainly a plausible one, given that one of the hallmarks of the diabetic kidney is an increase in intraglomerular pressure” and that SGLT-2 inhibition may reduce hyperfiltration in the kidney to improve renal outcomes. Of course, CREDENCE was stopped early in July 2018 due to significantly positive outcomes, and full CREDENCE results will be presented in Melbourne on April 15, at the World Congress of Nephrology. Recently at ATTD 2019, we heard Dr. Per-Henrick Groop point to reducing renal hyperfiltration as a putative mechanism for SGLT inhibitor mediated renal protection. This “tubular hypothesis” – see this key 2016 Circulation paper from Drs. Hiddo Heerspink, David Cherney, and others – appears to be a popular one, although much uncertainty surely remains on the exact mechanism that may be driving renal protection with SGLT inhibition. What insight can ongoing renal outcomes studies lend to the potential link between hyperfiltration and renal outcomes?


“Your Gut Might Just Be an Obesity-Promoting Gut” – Dr. Eckert-Norton Elucidates Pathophysiology of Obesity, Discusses Challenges to Weight Loss via Lifestyle Change

Dr. Margaret Eckert-Norton, former Secretary and Committee Chair for the Endocrine Nurses Society, summarized (i) new concepts in the pathophysiology of obesity and (ii) empirically-tested lifestyle interventions for weight loss. Throughout the first half of her talk, we were intrigued by the idea that a more accurate understanding of what causes obesity might reduce weight-based stigma. For example, Dr. Eckert-Norton distinguished between homeostatic (fulfilling biological need) vs. hedonic (for pleasure) eating. She suggested that food companies have leveraged food science to increase the palatability of what they sell, deliberately targeting the hedonic, reward-sensitive pathway. This gets at the environmental contribution to the obesity epidemic, which needs to be addressed first and foremost, in our view (e.g. by making healthy food more affordable/accessible, or perhaps taxing junk food). Dr. Eckert-Norton explained that ~70% of excess energy consumed is stored as fat, and that greater adiposity lowers the efficiency of energy expenditure (i.e. the same amount of exercise may burn fewer calories). Moreover, obesity is associated with both leptin and insulin resistance. The former phenomenon interferes with satiety, and the latter, of course, becomes type 2 diabetes. After weight gain, any energy deficit – such as the start of a new diet – leads to a hyper-insulinemic state (the body is accustomed to over-producing insulin, and now it’s not being used), which exacerbates insulin resistance and adiposity. With all this pathophysiology in mind, it shouldn’t come as a surprise that dietary modifications don’t always lead to immediate weight loss, Dr. Eckert-Norton argued. Given how frustrating weight loss can be, she urged nurses and other HCPs to go over these concepts with patients, eliminating any blame and setting reasonable expectations for body weight targets. She also shifted blame from individuals to their microbiome: “Your gut might just be an obesity-promoting gut.” Dr. Eckert-Norton discussed one case in which a fecal microbiome transplant led to new-onset obesity in the recipient (the donor was overweight, but otherwise healthy). The obesity that developed in the recipient was unresponsive to all traditional treatments. Dr. Eckert-Norton highlighted how over-prescription of antibiotics has “perturbed” the gut microbiome and may be implicated in the obesity epidemic, lowering the concentration of bacteria thought to protect against obesity.

  • In the second half of her presentation, Dr. Eckert-Norton echoed a refrain from obesity expert Dr. Donna Ryan – that the best diet for weight loss is the one a person can adhere to. She displayed weight loss data for four diets: (i) DASH; (ii) low-carb; (iii) low-fat; and (iv) Mediterranean. While the last of these has shown greatest adherence out to six years, Dr. Eckert-Norton emphasized that there’s no significant difference in weight loss achieved with each of these methods. Where there is a significant association is between diet adherence and weight loss. Personalizing a diet plan is thus paramount for successful obesity management.

  • Turning to physical activity, Dr. Eckert-Norton exclaimed that “sitting is the new smoking.” She noted that any exercise is better than none; for people with severe obesity who may not be able to walk long distances, even chair exercises can make a positive difference. Dr. Eckert-Norton framed physical activity as a critical prong in weight maintenance, though she showed how diet is the dominant factor for weight loss.

Dr. Lee Kaplan Touts Potential for Synergistic Weight Loss with Drugs + Surgery; Offers General Algorithm for Pharmacotherapy, Focused on Avoiding Inertia

Harvard’s Dr. Lee Kaplan shared valuable insight into the obesity pharmacotherapy and metabolic surgery landscapes, highlighting the additive and even synergistic benefits of using these together. Dr. Kaplan began his talk with an interesting hypothetical: It wouldn’t make sense to combine surgical weight loss procedures (~30% weight loss on average) with obesity pharmacotherapy (~3%-7% weight loss on average) if surgery was simply physically restricting food intake. After all, the weight loss seen with pharmacotherapy pales in comparison to the dramatic weight loss with surgical procedures and would not be worth the investment for most patients. However, Dr. Kaplan argued, it’s important to remember that surgical weight loss procedures do affect the fundamental biology of weight and do change the physiology of energy and fat balance regulation. As a result, he underscored that there is a chance for the combination of pharmacotherapy and surgical weight loss to be synergistic in driving down weight. Therefore, it’s important to fully consider pharmacotherapy in patients who are also considering weight loss surgery or have already undergone such a procedure. Indeed, we often hear pharmacotherapy promoted as a means of enhancing or maintaining weight loss post-surgery.

  • Dr. Kaplan outlined the anti-obesity medication algorithm that he uses in clinical practice. Patients are initiated on a drug therapy (Dr. Kaplan named the four branded therapies: Saxenda, Qsymia, Belviq, and Contrave) and treated for one to three months. After an initial treatment period, patients response is evaluated. If >10% weight loss is achieved, they continue on the therapy indefinitely. If 5-10% weight loss is achieved, therapy is continued but another drug is added. If <5% weight loss is met, the drug is stopped and a new drug is initiated. We appreciate the emphasis on timely progression, something sorely needed in both diabetes and obesity.

    • Dr. Kaplan noted a unique challenge associated with the current landscape: “With obesity pharmacotherapy, we have a lot of opportunity and a lot of mechanisms to learn about and to choose from. Learning how to use one or two or three of these drugs is inadequate for the task. We don’t have an anchor drug or class like we do for diabetes or hypertension.”

  • Additionally, Dr. Kaplan predicted that the next big advance will be identifying clinically useful predictors of response to individual therapies. For most treatment options, weight loss greatly varies among patients – this includes lifestyle interventions like low-carb diets, pharmacotherapies such as liraglutide, devices such as duodenal liners, and metabolic surgery. Not all patients react the same way to these interventions, and spectra of weight loss follow standard curves from those who respond with significant weight loss (>50 pounds) to those who do not respond at all or even gain weight. Predicting which patients will respond best to certain treatments is surely no small task, but the potential yield of is tremendous – it would save time, money, energy, and frustration for both patients and HCPs. To our knowledge, both regression models and genetic based prediction models for weight loss in those with obesity remain in their infancy, with much work ahead before such prediction techniques can reliably be used in clinical practice.

Technology Exhibit Hall


Abbott’s booth advertised the 14-day FreeStyle Libre, approved by the FDA last July. A screen boasted Abbott’s impressive real-world database, which was featured in an awesome poster at ATTD (see our report). With four years of data from over 470,000 readers, Abbott has amassed 4.8 billion glucose measurements. As the display noted, FreeStyle Libre 14-day users have averaged 12 scans/day, well above the ADA recommendation of 6-10 glucose checks per day for patients on MDI or insulin pump therapy. Affordability was emphasized both at the booth and during the company’s product theater. Indeed, the 14-day FreeStyle Libre still has the lowest startup cost on the US market, as users can purchase one sensor in-pharmacy for ~$60 cash and use the free FreeStyle LibreLink smartphone app, meaning that purchasing a reader is not required. As Dr. Ernest Asamoah (Marian University Osteopathic School of Medicine, Indianapolis, IN) claimed during the product theater, “I’ve used all the CGMs and they’re all great. The difference is the FreeStyle Libre is affordable and a lot of patients are paying out of pocket. This is why the FreeStyle Libre is the gamechanger in diabetes going forward.”

Companion Medical

At the Companion Medical booth, we learned that the InPen has officially launched in retail pharmacies as of March 8. Those with commercial insurance can expect to pay “no more” than $99, and Medicare copays “vary.” Overall, median copays are <$50 – similar to the estimate at JPM, where we heard that this figure was $50. While the representative could not provide a user base update from the 2,000 patients shared at JPM, he mentioned that it is “significantly growing” – we expect that with the added convenience of the recent pharmacy launch, these number will only continue to rise. Companion Medical has also “significantly” expanded its sales team to match the greater anticipated demand. As for the company’s plans for a study of InPen vs. traditional MDI using FreeStyle Libre, the representative shared that it has “already begun enrollment.” The trial will assess time-in-range as the primary endpoint at three and six months.


At the Dexcom booth, we learned that G6 Medicare shipping is estimated to begin this September. This is potentially a delay from timing shared on the company’s 4Q18 call in February, which estimated shipping in the “near-term.” As the booth representative pointed out, it is a responsible decision to delay shipping until supplies are sufficiently robust to support the Medicare population over time. In fact, the representative shared that Dexcom is doubling its manufacturing capacity to meet overall demand. No updates were provided on the lower-cost G6 transmitter, slated to launch in 2H19, or the G7 (Dexcom/Verily Gen 2), for which an initial launch is expected in late 2020.


Glooko representatives were excited to discuss the company’s recent decision to make its mobile app free for anyone with diabetes. Previously, there was an annual $59.95 subscription fee for individuals not sponsored through their provider, health plan, employer, or device company. This is an obvious and long-needed strategic move for Glooko. According to the representatives, providers have been “enthusiastic” to speak with their patients about downloading the app. Glooko is rated a strong 4.0/5 star rating on Google Play (913 ratings) and 3.1/5 stars on the App Store (64 ratings). Google Play notes 50,000+ installs.


An Insulet booth representative shared that the new Bluetooth-enabled Dash PDM is set to roll out in a matter of weeks, likely aligning with 4Q18 expectations to launch Dash at no upfront cost by the end of 1Q19. Later, management shared that Dash was launched in the pharmacy channel in select areas as part of the ongoing limited market release. Separately, brochure at the booth highlighted the impressive amount of data collected from the Omnipod Horizon hybrid closed loop trials. To date, the clinical program has involved 7,720 cumulative hours of hybrid closed loop from 132 participants. The booth representative confirmed that the pivotal trial is expected to be complete by the end of this year. As of the company’s 4Q18 call, Omnipod Horizon is expected to launch in 2H20 with direct smartphone control on Samsung Galaxy phones. We were also excited to see the brochure describe Insulet’s partnership with Tidepool to develop an automated insulin delivery system using the DIY Loop algorithm. Insulet remains the first and only pump company to partner with Tidepool, although we would not be surprised to see others join. Per the brochure, the Loop algorithm will run on the user’s smartphone and will “activate, control, and deactivate” the pod.


In an update on the specifics of the OUS 670G launch, we learned that the 670G has rolled out to England and France, with Australia to follow “soon.” These launches are still very recent, with the booth representative estimating availability to have begun just within the past 90-120 days. The update at EASD called for an OUS launch in October (and a Canada launch was expected in Fall 2018) – we’re checking with the company on the status of 670G in these geographies. At the time of EASD, Belgium, Denmark, Finland, Italy, Netherlands, Slovenia, Spain, Sweden, Switzerland, and UK/Ireland were among the countries expected to be included in the initial launch. According to the booth representative, other areas in South America are slated for a launch “soon,” but as of now no calendar date has been shared. The booth representative noted that Colombia already has approval for the 670G. There were no updates on Medtronic’s exciting slew of pipeline efforts, which include four product launches expected in the April 2019-April 2020 (FY20) window as of JPM: (i) the next-gen “advanced hybrid closed-loop system,” now called the MiniMed “780G,” with automated correction boluses, Bluetooth, and remote software updating; (ii) non-adjunctive CGM claim and iCGM indication for what looked like Guardian Sensor 3 (enabling Medicare coverage); (iii) Guardian Connect on Android (previously expected by this coming April); and (iv) Sugar.IQ “Gen 2” with meal handling and carb counting advice (Nutrino acquisition) and predictive insights going beyond 60 minutes (partly launched last week for hypoglycemia).


The Senseonics booth included a section dedicated to the newly launched Eversense Bridge Program, offering the 90-day Eversense for $99 to eligible patients. A brochure highlighted the easy enrollment process: physicians need only add their interested patients to the Eversense Ordering System portal and check a box indicating that the patient wants to apply for the program. After that, Senseonics takes care of the rest, including confirming available insurance coverage via pre-authorization and appeal services and working with clinics to schedule sensor placements. One clinician visiting the booth commented that while she has done a few insertions, she believes that the 180-day Eversense XL will be the real gamechanger in adoption. While we are inclined to agree, the 90-day version should be sufficiently attractive to garner reasonable uptake. With cash dwindling (Senseonics does not believe it has enough remaining to fund operations through March 2020) establishing a strong user base in the US is more essential than ever. Still, we’re pleased to see Senseonics forging ahead on its rigorous pipeline efforts. A booth representative confirmed that the 180-day US clinical trial for Eversense XL is still enrolling, aligning with 4Q18 comments anticipating enrollment to wrap up in 3Q19. The booth representative expects to see FDA approval and launch for the Eversense XL in the next 18-24 months – the far end being a bit later than we might have anticipated, given the study is estimated for completion in December 2019. The booth representative also touted Eversense’s recently launched integration with Glooko, announced earlier this month.


At the Tandem booth, we learned that the t:slim X2 is now available in 11 countries outside the US plus Canada (i.e., 13 countries total). Per the company’s 4Q18 call in February, the Basal-IQ with G6 integration will begin launching internationally this year, but timing will vary country-to-country. The booth representative explained that a key step in this process is acquiring regulatory approval for the Tandem Device Updater (TDU). Booth representatives confirmed expectations to launch a t:slim X2 mobile app for Apple and Android in 2Q19 – towards the end of 4Q18 timing anticipating a broader 1H19 launch. The app will initially offer wireless data uploads from the t:slim X2 pump to t:connect, but wireless smartphone control of the pump remains the ultimate goal for the app. The booth representatives confirmed that the app will be free and will be enabled with a remote software upgrade. While representatives could not comment on the potential for Tandem to launch the Control-IQ/G6 hybrid closed loop with a pediatric indication (6+ years), they confirmed that the pivotal trial in 14+ years is fully enrolled (n=168). Per the 4Q18 call, the study is expected to complete in April and will report data at ADA in June.  As a reminder, Tandem now expects Control-IQ to launch in the US between “summer and end of Q3 (September),” as a brand-new study in 6+ years could enable a pediatric launch out of the gate.

Therapy Exhibit Hall


Amarin’s booth was comparatively tiny when stacked up to its exhibits at ACC 2019 and AHA 2018 (the first conferences where it’s had an outsized presence, following the emergence of Vascepa and positive REDUCE-IT results last September). Of course, this is understandable considering the different physician populations at each of these meetings (cardiologists vs. endocrinologists). The booth at ENDO emphasized REDUCE-IT and recent follow-up analyses showing that Vascepa reduced first, subsequent, and total CV events when compared to placebo in the REDUCE-IT trial. We’re curious to see how endos react to these results and how much they incorporate Vascepa into their prescribing practices. Of note, during their 3Q18 earnings call, Amarin management noted that its sales approach in reaching out to physicians will include 85% general practitioners, 7% cardiologists, and 5% endocrinologists. Nonetheless, seeing how large the diabetes cohort was in the REDUCE-IT trial and the ever-growing importance of CV risk reduction in diabetes care, we imagine that interest from endocrinologists will grow.


Amgen featured familiar promotional materials for its PCSK9 inhibitor Repatha. The booth emphasized the following message to endos: “For patients with established CVD, take action with Repatha; You can prevent another MI or stroke.” The booth focused on the unmet need in lowering LDL-C in many patients already on statins and the recently updated 2018 AHA/ACC guidelines that recommend a LDL-C threshold of 70 mg/dl for treatment in those with CVD. Similar to its booth at ACC 2019, signs also emphasized a theme of “lower” – lower LDL-C, lower CV risk, lower hassle, and lower cost. These last two points are notable in light of Amgen’s recent 60% price slash for Repatha, along with improved access to Repatha in light of fewer authorization criteria now required.


AZ brought a sizable booth focusing on GLP-1 receptor agonist Bydureon (exenatide) and SGLT-2 inhibitor Farxiga (dapagliflozin). Building on a strong ACC the week before, Farxiga stayed in the news at ENDO with (i) a key DEPICT post-hoc analysis presented on Saturday and (ii) approval for type 1 in Europe announced Monday. Notably, AZ’s booth generated a lot of buzz in the hall, as it hosted a boxing-themed professional photo booth where individuals could pose and have a motion blur photo taken of them – check out a photo of two of our associates in the booth here!


Eisai’s one-woman booth exclusively featured Lenvima, an oral treatment for thyroid cancer. The rep on hand was only familiar with the company’s oncology portfolio, and couldn’t speak to obesity drug Belviq (lorcaserin).


Lilly/BI have scaled back their exhibit hall presence at diabetes conferences in the past year or so, and this meeting was no exception. The companies shared a small booth in the back, offering informational brochures on GLP-1 agonist Trulicity (dulaglutide) and SGLT-2 inhibitor Jardiance (empagliflozin). Two glass cases displayed sample insulin pens as well as the IDEO-designed autoinjector for Trulicity (which is a large reason for the product’s popularity among patients). Lilly reps underscored the range of type 2 diabetes therapies in their portfolio, which covers basal and rapid-acting insulin, GLP-1, SGLT-2, and DPP-4. They seemed to emphasize that endocrinologists could whole-heartedly trust Lilly when it comes to diabetes care – we would’ve assumed that was not in question but more a matter of individualization/personalization, prioritization, and access.


We were impressed by Merck’s substantial investment in Pfizer-partnered SGLT-2 inhibitor Steglatro at ENDO, which included a full-screen ad that appeared every time the conference app was opened and materials throughout the convention center. In Merck’s medium-sized exhibit hall booth, Steglatro (ertugliflozin) and DPP-4 inhibitor Januvia (sitagliptin) were promoted in equal turn near a main entrance to the exhibit hall: Both Steglatro and fixed-dose combination Steglujan (ertugliflozin/sitagliptin) were prominently featured on informational panels. In contrast, partner Pfizer did not promote Steglatro in its booth, opting instead to promote acromegaly drug Somavert. Merck reps were enthusiastic about Steglujan (which we were glad to see advertised), given the strength and familiarity of Januvia, but they also continued to recognize that fixed-dose combinations are difficult to afford for many patients and taking an SGLT-2 inhibitor and DPP-4 inhibitor separately is typically more affordable. Of note, we’ve recently heard chatter that Steglatro and J&J’s Invokana are now neck-and-neck in terms of new-to-brand prescriptions, meaning as many patients new to SGLT-2s are being prescribed Invokana as are receiving Steglatro – we view this as hugely positive for the latter franchise.

Novo Nordisk

Novo Nordisk hosted a large booth – though certainly smaller than we’d see at ADA or EASD – to promote its extensive and diverse suite of diabetes products, focusing specifically on Ozempic, Victoza, Tresiba, Xultophy, and Fiasp – essentially all of the company’s current growth drivers. Undoubtedly, new GLP-1 agonist Ozempic was the primary emphasis of the booth: In each of the three corners with informational panels, one side of each was dedicated to Ozempic, with the others split between Xultophy, Fiasp, and Tresiba. While Victoza promotion was present, it was clearly a smaller focus, reflecting the company promotional shift away from Victoza and toward Ozempic. Reps highlighted particularly strong interest in Ozempic, and we noted that the stands near escalators that held prescribing information were frequently empty.


Sanofi’s booth told compelling patient stories to highlight the benefits of Lantus, Toujeo, Soliqua, Admelog, and Apidra. Overhead banners advertised Lantus (insulin glargine) with the slogan “no substitute needed,” which reps said should serve as a reminder to patients and providers that they don’t have to switch to Basaglar (Lilly/BI’s biosimilar glargine). They emphasized that patients like Lantus and providers are familiar and comfortable with it. In one rep’s words, “Basaglar may be glargine, but it’s not Lantus,” and we agree that these products are about more than just the molecule. For instance, some patients may prefer Sanofi’s SoloStar pens to Lilly/BI’s KwikPens, and there are certainly advantages to continuity of care over switching to a new basal insulin. Sanofi reps claimed that Lantus is still favored within some managed care plans, though our sense is that Basaglar has better payer coverage overall in the US (it’s preferred by major PBMs CVS Health and UnitedHealthcare, not to mention it’s on a lower tier within Medicare Part D). Indeed, Lantus sales have been on a consistent decline since 2015; product revenue fell 23% YOY in 2018 to $4.2 billion (from $5.2 billion in 2017). And while we support maximum therapeutic choice (i.e. patients/HCPs should be able to choose between Lantus and Basaglar, and for that matter, Toujeo, Levemir, and Tresiba), we only see an expanding role for biosimilar insulins in diabetes care going forward.

Regarding Toujeo, reps were particularly excited to talk about the SoloStar Max pen, which was FDA approved about this time last year. SoloStar Max holds 900 units of insulin glargine U300, can deliver up to 160 units/injection, and can be titrated in units of two. While the main target population for this higher-capacity pen is patients with large basal insulin requirements, reps suggested that it could be prescribed to anyone and might even reduce out-of-pocket cost for some (though this varies and depends on an individual’s specific insurance plan). Understandably, no one staffing the Sanofi booth could comment on last week’s Complete Response Letter for sotagliflozin in type 1 diabetes.


Xeris reps confirmed that an FDA decision on the Gvoke HypoPen (a liquid-stable glucagon autoinjector for hypoglycemia recovery) is expected by June 10. Following approval – and reps were quite confident about this – the next-generation glucagon therapy should hit US pharmacy shelves in August of this year. They explained that “HypoPen” is meant to evoke “EpiPen.” Xeris wants people with diabetes at risk for hypoglycemia to carry this product with them all the time, and they want bystanders to understand how to use it. Since it’s more or less common knowledge what EpiPens are for, management thought HypoPen would get the message across clearly that this autoinjector should be used during episodes of severe hypoglycemia. We like the idea of next-generation glucagon facilitating more open discussion about hypoglycemia and how to help someone with diabetes when they are experiencing a severe low. We’ll be curious to see if/how Xeris incorporates this education into its promotional strategy for the Gvoke HypoPen. If all goes according to plan, Xeris will be first-to-market with a next-gen glucagon; Lilly and Zealand will follow. See our glucagon competitive landscape for the complete picture.


--by Ann Carracher, Martin Kurian, Brian Levine, Payal Marathe, Maeve Serino, and Kelly Close