Novo Nordisk reports topline results from final phase 3a trials for faster-acting insulin aspart (Faster aspart) – March 26, 2015

Novo Nordisk recently announced topline results from the final two of four total phase 3a trials for its faster-acting formulation of insulin aspart (Faster aspart). Both trials met their primary endpoint of non-inferiority vs. NovoLog (insulin aspart), with one trial finding a modest but significant advantage with Faster aspart taken at mealtime (-0.32%) vs. NovoLog taken at mealtime (-0.17%) after 26 weeks (mean baseline 7.6%). Interestingly, this trial (Onset 1) had a third arm testing Faster aspart dosing post-mealtime; that group had a non-inferior A1c reduction from baseline (-0.13%) compared to NovoLog taken at mealtime, though it did not match Faster aspart mealtime dosing. In both trials, Faster aspart was associated with significantly reduced postprandial glucose excursions vs. NovoLog – the magnitude of the difference at one hour post-meal was 11 to 18 mg/dl in the two trials, which we see as meaningful.   

At this point in the development of rapid-acting insulin analogs, marginal improvements in A1c are harder and harder to achieve and we were not expecting to see a major improvement here. This is partially because faster-on/faster-off insulins tend to reduce hypoglycemia, which bumps up A1c. As an example, the most recent phase 3 data on Sanofi/MannKind’s inhaled insulin Afrezza in type 1 diabetes patients only barely found non-inferiority vs. injectable rapid-acting insulin analogs. The formulation improvements in Faster aspart appear to be more keyed towards hastening onset of action rather than expediting metabolism or deactivation. Perhaps as a result, there was no change in overall hypoglycemia seen with Faster aspart compared with NovoLog. In fact, the rate of hypoglycemia was statistically significantly higher with Faster aspart in the post-meal period as a result of redistribution due to the faster action. We will be looking closely at the hypoglycemia data when the full results from the trial are presented. We wonder if the Faster aspart dose was not adjusted downward to fully account for its faster onset of action – intuitively, using the same dose of a faster-acting insulin could lead to more early hypoglycemia. 

Taken as a whole, the topline results are positive but perhaps not mind-blowingly so. The improvement in postprandial glucose and possibility for dose flexibility represents a meaningful benefit. Whether those factors enough to win reimbursement, particularly in type 2, is a tougher question. The press release did not disclose when full results would be presented, but noted that Novo Nordisk plans to file Faster aspart with the US and EU around “the turn of the year.” This is quite a while given that all four pivotal trials are complete, although we imagine Novo Nordisk’s regulatory affairs team must be busy with the interim analysis for the DEVOTE CVOT for Tresiba (insulin degludec) and possible filings for Xultophy (insulin degludec/liraglutide), both of which are likely seen as higher-priority projects than Faster aspart.

• We do not expect Faster aspart to be a game-changing insulin, but even marginal improvements in onset of action could yield meaningful benefits for pumps and the closed loop in particular. The JDRF’s Dr. Aaron Kowalski argued during a workshop at ATTD that current insulin analogs do not work fast enough to allow for a fully automated closed-loop system. PK/PD data presented in a poster at last year’s ADA on Faster aspart showed a significant reduction in time to first appearance (4.85 minutes vs. 11.8 minutes with NovoLog) and time to 50% maximum concentration (20.7minutes vs. 31.6 minutes). The potential for post-meal dosing would be a major quality-of-life improvement for patients, and we were glad to see Novo Nordisk pursue a unique three-arm study design for Onset 1 that tested post-meal dosing.
• Other companies developing ultra-rapid-acting insulins include: Lilly/Adocia (BioChaperone Lispro in phase 2) and Biodel (BIOD-123 ready for phase 3, U400 BIOD-531 in phase 2). Sanofi/MannKind’s inhaled insulin Afrezza acts very rapidly, although this is not fully reflected on the current label. Halozyme is attempting to move forward with its Hylenex tool for expediting insulin absorption.
• Summary of topline results from the Onset 1 trial:
  
  • Study design: the trial randomized 1,143 type 1 diabetes patients successfully optimized on basal insulin therapy with Levemir (insulin detemir) during an eight-week run-in. Patients were randomized in a double-blind fashion to Faster aspart taken at mealtime, NovoLog taken at mealtime, and Faster aspart taken post-meal. The topline data announcement did not disclose the exact timing of dosing in the “post-meal” group, something we will be interested to learn more about when full results come out.
  • Efficacy: at the primary endpoint at 26 weeks, Faster aspart at mealtime reduced A1c by 0.32% from a common baseline of 7.6%; mealtime NovoLog reduced A1c by 0.17% from baseline, while post-meal Faster aspart reduced A1c by 0.13%. Faster aspart at mealtime and post-meal were both non-inferior to NovoLog at mealtime, and Faster aspart at mealtime was statistically significantly better than NovoLog. Patients on Faster aspart achieved significantly greater improvements in one and two-hour postprandial glucose increments during a meal test than patients on NovoLog – the largest difference, after one hour, was over 18 mg/dl.
  • Hypoglycemia: The overall rate of severe or confirmed hypoglycemia was similar between both insulins. However, the timing of the hypoglycemic events were redistributed as a result of the faster action.. It appears that post-meal Faster aspart dosing was not implicated in an increase in postprandial hypoglycemia, which could be the contribution of delayed dosing.
• Summary of topline results from the Onset 2 trial:
  
  • Study design: the trial randomized 689 type 2 diabetes patients inadequately controlled on a combination of basal insulin and oral drugs following an eight-week run-in period. Patients were randomized to Faster aspart or NovoLog (taken at mealtime) for 26 weeks.
  • Efficacy: A1c changes were comparable between groups, falling from 7.9% at baseline to around 6.6% in both groups (achieving non-inferiority). The Faster aspart group saw an ~11 mg/dl improvement in one-hour postprandial glucose; the improvement at two hours was numerically better but not statistically significant.
  • Hypoglycemia:  The overall rate of severe or confirmed hypoglycemia was similar between both insulins. However, the timing of the hypoglycemic events was redistributed as a result of the faster action..

-- by Manu Venkat and Kelly Close