European Association for the Study of Diabetes (EASD) 50th Annual Meeting

September 15-19, 2014; Vienna Austria; Day #3 Highlights – Draft

Executive Highlights

We are coming back to you from Vienna, where the diaTribe Foundation symposium on “Solvable Problems in Diabetes” capped another learning-filled day at the European Association for the Study of Diabetes (EASD) 2014. We have reached the halfway point of the meeting (already!), though attendees’ energy and excitement shows no signs of dwindling. Below, we have included a streamlined look at the top ten highlights of the day.

1. The diaTribe Foundation hosted its inaugural EASD event, “Solvable Problems in Diabetes,” in which our own Ms. Kelly Close (The diaTribe Foundation, San Francisco, CA) moderated a wide-ranging, provocative discussion with Professors Philip Home (Newcastle University, Newcastle Upon Tyne, United Kingdom) and Jens Sandahl Christiansen (Aarhus University, Aarhus, Denmark).

2. Dr. Ira Gantz (Merck Sharp & Dohme Corp, Whitehouse Station, NJ) presented new phase 3 results of Merck’s omarigliptin, showing that the once-weekly DPP-4 inhibitor achieves statistically significantly A1c-lowering compared to placebo.

3. Dr. Matthew Riddle presented exciting new data from a JDRF-supported phase 1 dose-finding study comparing three fixed-dose combinations of pramlintide/regular human insulin.

4. A Medtronic poster presented additional data from the randomized, six-month Opt2mise trial, highlighting that patients in the worse control had the largest reductions in A1c. One patient experienced an impressive 6% decline in A1c from an 11.5% baseline.

5. A Novo Nordisk-sponsored study on hypoglycemia highlighted much higher-than-expected self-reported hypoglycemia in an impressively large 27,585 insulin-treated people with diabetes in 24 countries.

6. We heard new follow-up data on Novo Nordisk’s Saxenda (liraglutide 3.0 mg for obesity) from the SCALE Obesity and Prediabetes trial; the main takeaway was the need for continuous treatment in order to experience a benefit.

7. Similarly, in the SCALE Diabetes trial, Dr. Ralph DeFronzo investigated the effect of ceasing liraglutide treatment on safety and efficacy parameters in a 12-week off-treatment follow-up period after the original 56-week trial.

8. Full-year results on Lilly/BI’s empagliflozin/linagliptin FDC in treatment-naïve type 2 diabetes patients showed numerically but not statistically significantly greater A1c reductions from baseline with the high-dose combination.

9. Full results from the Sanofi-sponsored observational non-randomized INITIATOR trial found that patients that initiated Lantus (insulin glargine) treatment saw a lower increase in drug costs and greater adherence to treatment than those that began on Victoza (liraglutide), although those effects could be attributable to differences in each patient population.

10. A Janssen-sponsored study by Dr. Gwilym Thompson (Janssen UK, High Wycombe, United Kingdom) et al. concluded that J&J’s Invokana (canagliflozin) may be more cost-effective than Merck’s Januvia (sitagliptin) as a third-line treatment option for type 2 diabetes.

Top 10 Highlights

1. The diaTribe Foundation hosted its inaugural EASD event, “Solvable Problems in Diabetes,” in which our own Ms. Kelly Close (The diaTribe Foundation, San Francisco, CA) moderated a wide-ranging, provocative discussion with Professors Philip Home (Newcastle University, Newcastle Upon Tyne, United Kingdom) and Jens Sandahl Christiansen (Aarhus University, Aarhus, Denmark). The distinguished panel shared thoughts on a host of key issues, including differences in government regulation and reimbursement policy between the US and Europe, how to maximize the benefits of existing tools, and how they would invest $10 billion in diabetes. The discussion also pulled insightful questions from the audience and featured a “lightning round,” with rapid-fire questions such as “soda tax: yes or no?” and “biggest patient barrier: hypoglycemia or weight gain?” We so appreciate all of our sponsors and attendees for making this event a success – please look for our full report in the upcoming weeks for in-depth coverage of the event. 

2. Dr. Ira Gantz (Merck, Whitehouse Station, NJ) presented the first phase 3 results on Merck’s omarigliptin, showing that the once-weekly DPP-4 inhibitor had a comparable safety and efficacy profile compared to the once-daily Januvia (sitagliptin) . The 24-week, non-inferiority trial was conducted in Japanese patients (n=414) with type 2 diabetes and compared the safety, efficacy, and tolerability of omarigliptin 25 mg with both placebo and sitagliptin 50 mg. Omarigliptin met the 0.3% A1c margin for non-inferiority, with an efficacy profile comparable to that of sitagliptin – both achieved placebo-adjusted A1c reductions of ~0.8%. Both agents also significantly reduced two-hour post-meal and fasting plasma glucose levels relative to placebo and showed no significant difference in safety profile. Omarigliptin was weight neutral, similar to sitagliptin. These results strengthen the case for omarigliptin as a safe, efficacious, and perhaps less burdensome alternative to sitagliptin. As a reminder, Merck intends to file for approval by the end of 2014. See Merck’s press release on this presentation here.

3. Dr. Matthew Riddle presented exciting new data from a JDRF-supported phase 1 dose-finding study comparing three fixed-dose combinations of pramlintide/regular human insulin (at 6, 9, and 12 mcg/unit of insulin) to regular human insulin alone. The randomized, single masked, four-way crossover study enrolled 19 type 1 patients who ate a standardized 600-calorie meal for breakfast; glucose and glucagon AUC, along with blood glucose, were followed for three hours post meal. Dr. Riddle called the results “quite dramatic” – every pramlintide/insulin fixed-dose ratio reduced incremental glucose AUC0-3 h by >50%, and all dose ratios reduced incremental glucagon AUC0-3 h by >50%. The results were highly statistically significant, and were even more notable given that the mealtime insulin dose was reduced by 30% in the pramlintide arms. Interestingly, there were no significant differences between any of the ratios. A plot of postprandial glucose showed the most dramatic improvement with the fixed-dose combination 30-120 minutes post-meal (e.g., ~155 mg/dl [insulin+pramlintide] vs. ~260 mg/dl [insulin alone] at 60 minutes), though by three hours, postprandial glucose was not statistically different between the arms of the study (~260 mg/dl). Further studies are planned in ongoing basal-bolus treatment – Q&A implied these will take place over 24 hours. It was not clear if the insulin and pramlintide were co-formulated, but we assume they were. We believe this fixed-dose approach has particular promise in a pumped formulation for the artificial pancreas, where meal excursions are still a challenge for control algorithms. More details below.

4. A Medtronic poster presented additional data from the randomized, six-month Opt2mise trial, comparing insulin pump therapy (n=168) to MDI (n=163) in type 2 patients in poor control (mean A1c: 9.0%); the primary data was published in the Lancet in July and shared in a late-breaking poster at ADA 2014. In the main analysis, A1c declined by 1.1% in those on an insulin pump compared to 0.4% in the MDI group (p<0.001) after 27 weeks. This poster presented a valuable sub-analysis of the relationship between different baseline variables and A1c reduction in the pump group. An illuminating chart plotting A1c reduction against baseline A1c – see a picture here – was a clear reminder of why super-responders matter, as well as the need to think beyond average changes. One patient experienced an astounding 6% drop (!) in A1c from a baseline of 11.5%, while 19 patients (by our count) experienced A1c reductions of 2% or greater. Twenty-one patients, by contrast, experienced an increase in A1c while on pump therapy – we wonder what can be learned from those patients about optimal initiation of pump therapy in type 2 diabetes. Overall, it was highly encouraging to see that efficacy was stronger in the patients in the worst control, since this would support strong cost-effectiveness data and bring so much help to so many patients that need it. Notably, the poster also found that older patients and those with lower cognitive state achieved comparable improvements in A1c, countering the view that pumps are too complicated for certain type 2 patients.

5. A Novo-Nordisk-sponsored study examined the prevalence of self-reported hypoglycemia in an impressive 27,585 insulin-treated type 1 (n=8,022) and type 2 patients (n=19,563) in 24 countries. The experimental design included a six-month retrospective period and a one-month prospective period; patient diaries were used to assist recall and record hypoglycemic events. Severe hypoglycemia used the ADA definition, “Requiring third party assistance.” The poster had three important findings: (i) estimated rates of hypoglycemia were higher than in previously reported studies; (ii) incidence rates rose in the prospective period, indicating significant underreporting of hypoglycemia; and (iii) there was no correlation between A1c and hypoglycemia (in line with data from the T1D Exchange, and running counter to the conventional wisdom from DCCT). The tables below summarize the hypoglycemia prevalence (% of patients) and incidence (events per patient-year) – we were particularly struck by the self-reported prevalence of severe hypoglycemia, which occurred in 27% of type 1 patients and 16% of type 2 patients in the six-month retrospective period, and 14% of type 1 patients and 9% of type 2 patients in the four-week prospective period. Overall, 83% of type 1s and ~49% of type 2s experienced one or more hypoglycemia events in a four-week period. We believe the data underscore a very critical point: hypoglycemia of all types still occurs far too frequently in type 1 and type 2 diabetes.

6. We heard new follow-up data on Novo Nordisk’s Saxenda  (liraglutide 3.0 mg for obesity) from the SCALE Obesity and Prediabetes trial; the main takeaway was the need for continuous treatment in order to experience a benefit. Dr. Xavier Pi-Sunyer (Columbia University, New York, NY) briefly discussed results from a follow-up period between 56 and 68 weeks, in which participants from the 56-week SCALE Obesity and Prediabetes trial were re-randomized from Saxenda to placebo. As a reminder, the trial’s primary results showed that Saxenda led to significant (69% vs. 33% with placebo) reversion from prediabetes to type 2 diabetes. During the follow-up period, participants initially on Saxenda re-randomized to the placebo group regained more weight than those who remained on the drug (2.9% vs. 0.7%); in addition, the percentage of patients with prediabetes rose from 8.0% to 22.4% in the placebo group, while it declined from 9.1% to 8.6% in the Saxenda group – it was striking that the incidence of prediabetes rose so rapidly after drug discontinuation. Dr. Pi-Sunyer did note that no participants developed type 2 diabetes during the re-randomized period. Such results are particularly notable given the concerns raised during last week’s FDA Advisory Committee meeting on Saxenda about the lack of data on long-term use of the drug.

7. Similarly, in the SCALE Diabetes trial, Dr. Ralph DeFronzo (University of Texas Health Science Center, San Antonio, TX) investigated the effect of ceasing liraglutide treatment on safety and efficacy parameters in a 12-week off-treatment follow-up period after the original 56-week trial. The results demonstrated that after halting treatment with Saxenda, patients regained weight (~2%), and the reductions in fasting plasma glucose and systolic blood pressure seen during the trial were all rapidly reversed. The increase in heart rate and lipase activity seen with liraglutide, however, were also reversed.

8. Following the positive 52-week results on Lilly/BI’s empagliflozin/linagliptin fixed-dose combination as add-on to metformin (presented yesterday), we were eager to see 52-week results on the combination in treatment-naïve patients. The 24-week results from that trial that we saw presented as a poster at ADA were somewhat surprising, as the high-dose combination (empagliflozin 25 mg/linagliptin 5 mg) achieved a numerically but not statistically significantly greater A1c reduction from baseline than empagliflozin 25 mg alone (-1.08% vs. -0.95%, p = 0.179). We thought that this might be due to a temporary lull in the high-dose combination’s performance that might equalize by the one year mark (the low-dose combination outperformed the high-dose combination at week 24), but the 52-week results were very much in line with the half-year results: from a mean baseline of ~8%, the high-dose combination achieved an A1c reduction of 1.17% compared with 1.01% with empagliflozin 25 mg (p = 0.176) and 0.51% with linagliptin 5 mg (p<0.001). The high-dose combination did perform significantly better than empagliflozin 25 mg in the subgroup of patients with baseline A1c greater than 8.5% (-1.99% vs. -1.53%; p = 0.036). These unexpected results might be due to the relatively small sample size: 677 patients were randomized between five groups. Overall, we would put more stock in the study presented yesterday, which tested the combination in a more clinically relevant scenario (in addition to metformin).

9. We saw full results from the Sanofi-sponsored INITIATOR trial, a real-world observational longitudinal cohort study comparing baseline characteristics, safety, efficacy, and cost for patients that started on either Lantus (insulin glargine) or Novo Nordisk’s GLP-1 agonist Victoza (liraglutide). As presenter Dr. Philip Levin (MODEL Clinical Research, Baltimore, MD) made very clear, the results did not necessarily serve as a direct safety or efficacy comparison for the two agents, but rather provide insights into the differences in the patient populations that opt for either drug. Patients that initiated injectable therapy with Victoza were around 10 kg heavier and had a lower baseline A1c (~8.1% vs. ~9.7%) than those that began on Lantus. Interestingly, Lantus showed an advantage over Victoza with regards to persistence to treatment (64% vs.50%) and increase in drug-related cost borne by payers ($300-$1000 per six months). However, with no data on what patients paid out of pocket, and with very significant differences between the patient populations that began on either therapy, the interpretation of these findings is far more limited than what could be interpreted if this was a randomized trial.

10. A Janssen-sponsored study by Dr. Gwilym Thompson (Janssen UK, High Wycombe, United Kingdom) et al. concluded that J&J’s Invokana (canagliflozin) may be more cost-effective than Merck’s Januvia (sitagliptin) as a third-line treatment option for type 2 diabetes. The study used the ECHO-T2DM model to simulate health outcomes and costs over 40 years for a total of 2 million hypothetical patients with poor glycemic control on metformin and a sulfonylurea. These patient characteristics were estimated based on information from one of Invokana’s phase 3 trials, which demonstrated greater efficacy with Invokana compared to Januvia. The simulation found that despite slightly higher absolute costs, treatment with Invokana was ultimately more cost-effective due to an increase in quality-adjusted life years; key contributors to the increase included improvements in BMI, greater overall survival, and less need for intensification of therapy. We wonder to what extent these types of simulated cost-effectiveness comparisons are persuasive to payers – we have heard in the past that payers are receptive to data from large databases and other non-randomized real-world sources, but simulations often involve a number of assumptions that are difficult to prove.

Honorable Mention

In a morning gathering with Novo Nordisk Chief Scientific Officer Dr. Mads Thomsen, we learned that Novo Nordisk plans to mount a large-scale foray into the field of obesity therapies. To start, of course, the company’s current presence in obesity was no secret, with a recent 14-to-1 vote in favor of approval at an FDA Advisory Committee meeting for Saxenda (liraglutide 3.0 mg for obesity). However, the company has previously characterized the Saxenda project as a way to take advantage of liraglutide’s effect on weight, rather than a large-scale push into obesity therapy. However, Dr. Thomsen stated that Novo Nordisk is now interested in creating a broader portfolio of biologic agents that more directly tackle the physiologic underpinnings of obesity, citing PYY and amylin as possible targets (in addition to GLP-1) and listing both satiety regulation and modification of energy expenditure as possible mechanistic avenues to explore. More broadly, Dr. Thomsen suggested that obesity is poised to see an explosion in therapy options along the lines of what the diabetes field has experienced over the past few decades. Resources to support this new obesity program may come from the recent discontinuation of the company’s anti-inflammatory program. Dr. Thomsen provided more details during an interview with another news source, suggesting that the new initiative could be housed in the company’s existing Seattle facility. Having Novo Nordisk committed to obesity as well as diabetes is a major win for patients, and Novo Nordisk’s entry might represent an opportunity for obesity, prediabetes, and diabetes to be framed more as multiple parts of a single continuum.

Detailed Discussion and Commentary

Novel Compounds on the Horizon

Pramlintide-Insulin Fixed-Dose Combination: A Phase 1 Dose Ratio-Finding Study In Patients With Type 1 Diabetes Mellitus

Matthew Riddle, MD (Oregon Health and Science University, Portland, OR)

Dr. Matthew Riddle presented exciting new data from a JDRF-supported phase 1 dose-finding study comparing three fixed-dose combinations of pramlintide/regular human insulin (at 6, 9, and 12 mcg/unit of insulin) to regular human insulin alone. The randomized, single masked, four-way crossover study enrolled 19 type 1 patients (mean A1c 7.8%, mean BMI 26 kg/m2) who ate a standardized 600-calorie meal for breakfast (insulin glargine the night before); glucose and glucagon AUC, along with blood glucose, were followed for three hours post meal. Dr. Riddle called the results “quite dramatic” – every pramlintide/insulin fixed-dose combination reduced incremental glucose AUC0-3 h by >50%, and all dose ratios reduced incremental glucagon AUC0-3 h by >50%. The results were highly statistically significant, and were even more notable given that the mealtime insulin dose was reduced by 30% in the pramlintide arms. Interestingly, there were no significant differences between any of the ratios. A plot of postprandial glucose showed the most dramatic improvement with the fixed-dose combination 30-120 minutes post-meal (e.g., ~155 mg/dl [insulin+pramlintide] vs. ~260 mg/dl [insulin alone] at 60 minutes), though by three hours, postprandial glucose was not statistically different between the arms of the study (~260 mg/dl). Dr. Riddle did not say it specifically, but that could imply that slowed gastric emptying was largely driving the improved glycemia with the insulin/pramlintide combination (which had worn off by three hours). It was not 100% clear if the insulin and pramlintide were co-formulated, but we assume they were, given the JDRF/Amylin partnership on this front from 2011. Further studies are planned in ongoing basal-bolus treatment – Q&A implied these will take place over 24 hours. We believe this fixed-dose approach has particular promise in a pumped formulation for the artificial pancreas, where meal excursions are still a challenge for control algorithms, and where bolus dosing can be more creative. 

  • Dr. Riddle covered the rationale for a fixed-dose combination of pramlintide and insulin. He explained that amylin is a second beta cell hormone that is co-secreted with insulin, implying that a fixed ratio would mimic physiology. Amylin has three modes of action: suppressing elevation of glucagon (especially following meals); slowing gastric emptying; and reducing food intake and enhancing satiety.
    • Pramlintide (Symlin) is a stable injectable analog of amylin currently sold by AstraZeneca. It reaches a peak after subcutaneous injection at 20 minutes, with a total duration of approximately three hours. Pramlintide is approved in the US at a fixed dose of 60 mcg in type 1 or 120 mcg in type 2, irrespective of the insulin dose. Dr. Riddle noted that pramlintide is not extensively used in real-world practice, in part due to the challenges of dosing it correctly.
  • Based on prior work (Wayer et al., Diabetes Care 2003), this study elected to use regular human insulin instead of a rapid-acting analog. The earlier study found that plasma glucose response was flatter when pramlintide was combined with regular human insulin vs. insulin lispro.
  • All dose ratios were well tolerated, and no treatment related hypoglycemia was reported. One participant reported nausea in all three pramlintide/insulin dose ratios, and one reported abdominal pain and diarrhea. The positive safety data, while of short duration, was encouraging to see at this early stage – real-world patient experience dosing pramlintide is often hampered by hypoglycemia.
  • This study was supported by JDRF (AstraZeneca provided pramlintide). As a reminder, JDRF and Amylin partnered back in 2011 to co-formulate pramlintide and insulin. BMS/AZ purchased In June 2012 for ~$7.0 billion, and AstraZeneca acquired BMS’ diabetes business in December 2013 for $2.7 billion in cash plus milestones. The acquisition included Symlin (pramlintide).

Questions and Answers

Dr. Tim Heise (Profil, Neuss, Germany): Nice presentation. Do you have data for more than three hours? It seemed that blood glucose was still rising at the end of three hours with the fixed-dose combination.

Dr. Riddle: We do not have data after three hours. It’s an important question. As expected, there was a rather strong suppression of postprandial glucose, with all three doses tested. There were essentially no nausea and hypoglycemia concerns. We feel safe over preceding over a longer 24-hour period. Another reason for the failure over time to continue to suppress glucose and glucagon levels was a dosage reduction in the pramlintide arm. Also, basal insulin was provided by glargine, and in these insulin sensitive people, blood levels of glargine were probably waning at that time. We don’t know the answer.

Dr. Heise: As you pointed out, these were insulin sensitive subjects with relatively low doses. Would you develop another dose ratio for type 2 diabetes patients? Or is this intended for type 1 only?

Dr. Riddle: We’re starting with type 1 diabetes. There is the greatest potential benefit in type 1, particularly in a nice closed-loop system. I am very interested myself in type 2 diabetes. We have done previously, a head to head comparison of a rapid-acting insulin vs. pramlintide on top of glargine. It was in this same dosage, 120 mcg, with meals. It turns out to about the same ratio. The results were interchangeable – non-inferiority. So in type 2 diabetes, you can get a quite considerable prandial effect with pramlintide.

Self-Reported hypoglycemia: A Global Study of 24 Countries with 27,585 Insulin-Treated Patients with Diabetes: The HAT Study

K. Khunti, S. Alsifri, R. Aronson, M. Cigrovski Berković, C. Enters-Weijnen, T. Forsén, G. Galstyan, P. Geelhoed-Duijvestijn, M. Goldfracht, R. Kapur10, N. Lalic, B. Ludvik, E. Moberg,U. Pedersen-Bjergaard, A. Ramachandran

This large study examined the prevalence of hypoglycemia in an impressive 27,585 insulin-treated type 1 (n=8,022; baseline A1c: ) and type 2 patients (n=19,563) in 24 countries. The experimental design included a six-month retrospective period and a one-month prospective period; patient diaries were used to assist recall and record hypoglycemic events. Severe hypoglycemia used the ADA definition, “Requiring third party assistance.” The poster had three important findings: (i) estimated rates of hypoglycemia were higher than previously reported; (ii) incidence rates rose in the prospective period, indicating significant underreporting of hypoglycemia; and (iii) there was no correlation between A1c and hypoglycemia (in line with data from the T1D Exchange, and running counter to the conventional wisdom from DCCT). The tables below summarize the hypoglycemia prevalence (% of patients) and incidence (events per patient-year) – we were particularly struck by the self-reported prevalence of severe hypoglycemia, which occurred in 27% of type 1 patients and 16% of type 2 patients in the six-month retrospective period, and 14% of type 1 patients and 9% of type 2 patients in the four-week prospective period. This translated to 2-5 severe hypoglycemia events per patient-year in type 1 and 0.9-2.5 events per patient-year in type 2. Overall, 83% of type 1s and ~49% of type 2s experienced one or more hypoglycemia events in a four-week period. Of course, the self-reported nature of the study is a potential limitation, but we still believe the data underscore a very critical point: hypoglycemia – both moderate and severe – still occurs far, far too often in type 1 and type 2 diabetes.

Table 1: Hypoglycemia prevalence (% of patients) in retrospective and prospective periods

 

Retrospective

Prospective

 

T1D

T2D

T1D

T2D

Any (4 weeks)

83%

51%

83%

47%

Nocturnal (4 weeks)

47%

22%

41%

16%

Severe

27%*

16%*

14%**

9%**

*six-month period; **four-week period

Table 2: Hypoglycemia incidence (events per patient-year) in retrospective and prospective periods

 

Retrospective

Prospective

 

T1D

T2D

T1D

T2D

Any (4 weeks)

52

17

73

19

Nocturnal (4 weeks)

16

5

11

4

Severe

2*

0.9*

5**

2.5**

*six-month period; **four-week period

 

-- by Melissa An, Adam Brown, Varun Iyengar, Emily Regier, Manu Venkat, and Kelly Close