Lilly announces first topline phase 3 data on its novel basal insulin peglispro (LY2605541) in type 2 diabetes patients – May 12, 2014

This morning, Lilly announced mixed topline results from three phase 3 trials of its novel basal insulin peglispro (formerly known as LY2605541; a PEGylated version of the rapid-acting insulin analog Humalog [insulin lispro]) in type 2 diabetes patients: IMAGINE-2 (peglispro vs. insulin glargine on top of OADs in insulin-naïve patients), IMAGINE-4 (peglispro and insulin lispro vs. insulin glargine and insulin lispro), and IMAGINE-5 (peglispro vs. insulin glargine alone or in combination with OADs).

Remarkably, in all three trials (IMAGINE-2, -4, and -5) peglispro led to statistically superior A1c reductions compared to the market-leading Lantus (insulin glargine), and it also yielded statistically significantly less nocturnal hypoglycemia and weight gain than Lantus (the alluring weight loss seen in phase 2, however, was not replicated). Although no specific values were provided, showing A1c-lowering superiority to Lantus is a pretty big milestone, one that no other basal insulin has achieved this clearly and consistently. Peglispro’s efficacy profile sets it apart from other “next-generation” basal insulins (such as Novo Nordisk’s Tresiba and Sanofi’s U300 glargine), which have only demonstrated A1c non-inferiority to Lantus with secondary benefits such as reduced nocturnal hypoglycemia.

However, it appears that most of the worrying hepatic and lipid issues seen in peglispro’s phase 2 results were also observed in phase 3. There were no cases of severe drug-induced liver injury (Hy’s Law), but peglispro did lead to a greater frequency of elevations in liver enzyme (ALT) levels (specifically, more patients on peglispro experienced an ALT elevation to three-times the upper limit of normal, although the number of these cases was relatively small). There was also an increase in liver fat observed on average in the studies of peglispro substituted for prior basal insulin. Investigators saw modest but statistically significant increases in plasma triglycerides in all three trials, with moderate reductions in HDL seen in two of the trials (IMAGINE-4 and -5).

Encouragingly, though, LDL levels were unchanged or reduced slightly, and cardiovascular adverse events were balanced between peglispro and insulin glargine, with the upper bound of the 95% confidence interval not crossing 1.8 (pooled analysis of all clinical trials in type 2 diabetes patients). Based on current data, Lilly does not expect to conduct a pre-approval CVOT.

 We expect a vigorous discussion of the full data (presentation expected in 2015), as peglispro appears to be a potentially high-risk, high-reward candidate. Liver safety can be a touchy topic for FDA Advisory Committees and providers alike, and a lot will depend on the magnitude of the changes observed – modest effects (akin to what has been seen in some statin trials) may not be a cause for derailment, especially given the drug’s remarkable efficacy and extra-glycemic benefits. Given the liver and lipid safety issues, it is possible that any hypothetical regulatory approval may come with some caveats, such as requirements for liver function tests. If it reaches the market, we imagine that the product could command a pricing premium over Lantus due to its constellation of benefits. In our view, prescriber opinions on safety may be the more important determinant of peglispro’s success – in a recent presentation at CODHy Latin America, Dr. Philip Home gave a rather pessimistic talk on hepato-specific insulin (read our coverage).  

The cloud of candidates ready to swoop in after Lantus loses patent exclusivity, including peglispro, other novel basal insulins (such as Novo Nordisk’s Tresiba [insulin degludec] and Sanofi’s U300 glargine), and a field of insulin glargine biosimilars (from Lilly/BI, Merck/Samsung Bioepis, and Biocon) continues to draw nearer (see table 2 below for the competitive landscape of new basal insulins).

Read on for our more detailed take on these results, which includes some learnings from our discussion with Lilly’s Distinguished Medical Fellow, Dr. David Kendall.

• Peglispro demonstrated statistically superior A1c lowering over insulin glargine (Sanofi’s market-leading Lantus) in IMAGINE-2, -4, and -5 trials. This data makes peglispro the first insulin agent to demonstrate consistent A1c-lowering superiority against insulin glargine. The company did not disclose the specific non-inferiority/superiority margin used in the trials, but based on other non-inferiority/superiority studies for diabetes drugs, we imagine that the margin was in the neighborhood of 0.3-0.4%. The consistent demonstration of A1c superiority over insulin glargine is perhaps the biggest takeaway from the topline results, as the improvement in insulin analogs over the past decade has made it harder and harder to achieve a major jump in efficacy over established players – as a result, differentiation has largely depended on factors like weight or hypoglycemia benefits (peglispro demonstrated those too – see below). This A1c benefit has the potential to be a massively powerful differentiating factor, although one that will have to be weighed against a number of safety concerns (see below).
  
  • Two of the three trials (IMAGINE-2 and -4) included in the topline data release were double-blinded, with patients and providers using the same titration protocol for both peglispro and insulin glargine to treat to target. This makes the finding of A1c lowering superiority quite robust. Although the topline data release did not include any data on insulin dosage during the trial, we imagine that the nocturnal hypoglycemia and weight benefits seen with peglispro may have allowed patients to more safely achieve a higher average insulin dose, which could have driven the A1c benefit. Alternatively, peglisprop’s hepatoselectivity may have rendered it more potent than the same dose of insulin glargine – this is all speculative, and Lilly plans to present on peglispro’s mechanism at a conference in 2014, at which time we’ll learn more about how it works.
  • For background: in phase 2, peglispro was found to be superior to insulin glargine in terms of A1c lowering in type 1 diabetes patients, but only achieved non-inferiority in type 2 diabetes patients. The smaller number of patients in phase 2 may have had a role in that difference, but we wonder if we can expect even greater relative efficacy in the type 1 diabetes phase 3 trials (topline data expected in 3Q14). In any case, it was encouraging to see consistent demonstration of A1c lowering superiority in type 2 diabetes, and we see no reason why the type 1 diabetes trials should not achieve at least a comparable level of efficacy.  
• The reduction in nocturnal hypoglycemia and weight benefits seen with peglispro in phase 2 were also seen in phase 3 (however, the weight loss observed in phase 2 was not replicated in phase 3 where peglispro was only associated with lower weight gain than Lantus). The press release stated that the incidence of nocturnal hypoglycemia was significantly lower in the peglispro arms than the insulin glargine arms, and that patients on peglispro experienced significantly less weight gain. Given that weight and nocturnal hypoglycemia are increasingly being recognized as critical factors from a patient perspective, these benefits are powerful differentiating factors for peglispro. The lower weight gain is particularly noteworthy for a drug within a class that is plagued by adherence issues associated with weight gain. 
  
  • Mechanistically, hepato-selective action has been hypothesized to contribute to both the nocturnal hypoglycemia and weight benefits. Much of the weight gain associated with other insulins is due to action in the periphery, such as glucose uptake by muscle and adipocytes and promoting fat deposition in adipose. Targeting insulin action to the liver avoids this peripheral uptake, minimizing weight gain. Nocturnal hypoglycemia is largely driven by insulin action in the periphery as well, explaining the benefit seen on that front with an agent like peglispro that acts less in the periphery. However, it has been theorized that hepato-selectivity could be associated with an increase in daytime hypoglycemia, as hepatic glucose production is one of the body’s primary defenses against daytime hypoglycemia associated with increases in energy expenditure. Dr. Philip Home (Newcastle University, Newcastle Upon Tyne, UK) discussed this topic at a presentation on hepato-selective insulins at this year’s CODHy Latin America meeting (read our coverage) – in that talk, he was fairly negative on hepato-selective insulin. Lilly’s press release did not share any information about daytime or total hypoglycemia.
• The lipid and hepatic safety concerns seen in phase 2 were also reported in phase 3 – the drug’s overall safety profile is challenging to assess conclusively, and the full safety data from the complete phase 3 program will be under close scrutiny. In all three trials, significantly more patients on peglispro experienced an increase in the liver enzyme alanine aminotransferase (ALT) to greater than three times the upper limit of the normal range compared to patients on insulin glargine. Somewhat reassuringly, no cases of Hy’s Law (a way to measure severe drug-related liver injury) were seen in any of the trials. Patients across the three trials had modest but statistically significant increases in triglycerides with peglispro relative to insulin glargine, and in two of the trials (IMAGINE-4 and -5) this change was accompanied by a small but statistically significant increase in HDL cholesterol. Encouragingly, though, LDL cholesterol (one of the lipid parameters most tightly correlated with cardiovascular outcomes) was either unchanged or decreased slightly with peglispro relative to insulin glargine. Peglispro’s impact on lipid and hepatic parameters are increasingly well characterized, but that does not make them less worrisome – FDA Advisory Committees take hepatic safety quite seriously, as severe liver injury can be very difficult or even impossible to reverse. However, changes in lipids and liver enzymes are not always cause for worry, as statins demonstrate (the class has been linked with some slight changes in liver enzymes, but is still seen as relatively safe).
  
  • Liver fat was measured via MRI in a subset of patients in IMAGINE-2 and IMAGINE-5. Increased liver fat is associated with a range of metabolic disorders (either as a cause or an effect), and in extreme cases can lead to irreversible cirrhosis. In IMAGINE-2, patients on peglispro did not experience a change in liver fat, while patients on insulin glargine saw a reduction in liver fat. In IMAGINE-5, patients on peglispro saw an increase in liver fat that stabilized by the end of the trial, while patients on insulin glargine saw no change.   
  • Given the high background level of fatty liver-related disorders in the type 2 diabetes patient population (and the associated higher risk), the risk/benefit assessment for peglispro could differ between type 1 diabetes and type 2 diabetes. We wonder if the balance could differ to the point that regulatory agencies might consider approving the drug for type 1 diabetes but not type 2 diabetes, or if the type 2 diabetes approval could come with more caveats. Such a decision would be clouded by the fact that the type 1 diabetes population (in line with the general population) is seeing an increased incidence of obesity and metabolic dysfunction. Even without any formal contraindications or restrictions, we imagine that providers would use a high level of discretion in prescribing peglispro, taking care to avoid patients at the highest risk of liver disease (perhaps due to excessive alcohol intake or findings from liver function tests).  
• Lilly expects to file peglispro in the US and EU by the end of 1Q15. This represents a slight delay from previous company guidance of a “possible” submission in late 2014, and would likely mean a regulatory decision in early 2016. The company expects to release topline results from two phase 3 studies of peglispro in type 1 diabetes by the end of 3Q14, and plans to present full results from those trials (as well as the three type 2 diabetes trials) at scientific meetings in 2015. We are disappointed that we will have to wait possibly a year or more to see full detailed results on the three studies outlined in these topline results, although we understand that the study completion timeline may not have delivered results in time for submission to ADA 2014. See table 1 below for an overview of Lilly’s IMAGINE phase 3 program for peglispro. The press release did state that new data regarding peglispro’s mechanism of action will be presented at an “upcoming medical meeting in 2014” – we imagine ADA or EASD are possibilities.

Table 1: The IMAGINE phase 3 program for peglispro (LY2605541)

• At the moment, there is no definite reason why the FDA might request a dedicated cardiovascular outcomes trial, although there is always room for an agency surprise on that front. The triglyceride signal is the closest thing to a cardiovascular safety red flag, but the connection between triglyceride levels and cardiovascular outcomes is still somewhat cloudy (at least, relative to the connection between metrics like LDL and CV events). An analysis of all the clinical trials completed to date in the peglispro program found that the incidence of adverse CV events was comparable between insulin glargine and peglispro, with the upper bound of the 95% confidence interval below 1.8 (the FDA’s pre-specified threshold for a pre-approval outcomes trial).
  
  • The FDA required pre-approval CV data for Novo Nordisk’s ultra-long-acting basal insulin Tresiba (insulin degludec) – read our report on the FDA’s Complete Response Letter. However, in that case, the decision was based on an imbalance in the incidence of adjudicated CV events in pooled clinical registrational trials. It is unclear whether the change in a marker such as triglycerides or HDL in the absence of a trend towards an imbalance in outcomes would constitute sufficient reason for the FDA to request a CV trial pre-approval. We hope it does not, as such a paradigm would make the diabetes drug regulatory environment even more treacherous, as given the multitude of parameters investigated in a clinical trial, often times imbalances can emerge that are not highly clinically meaningful.

Table 2: Competitive landscape for new basal insulins

--by Manu Venkat, Jessica Dong, and Kelly Close