Memorandum

Topline results for EXSCEL cardiovascular outcome trial of AZ’s Bydureon (exenatide once-weekly) demonstrate CV non-inferiority, but not superiority – May 23, 2017

Executive Highlights

  • The EXSCEL cardiovascular outcome trial (CVOT) for AZ’s GLP-1 agonist Bydureon (exenatide once-weekly) met its primary safety endpoint by demonstrating non-inferiority to placebo for CV events but did not meet its co-primary efficacy endpoint.
  • CV events (as defined by three-point MACE [non-fatal MI, non-fatal stroke, and CV death]) were numerically lower in the Bydureon arm but did not reach statistical significance.
  • Full results will be presented at EASD 2017 in Lisbon, Portugal.

Early this morning, AZ announced that the EXSCEL cardiovascular outcome trial (CVOT) for GLP-1 agonist Bydureon (exenatide once-weekly) met its primary safety endpoint by demonstrating non-inferiority to placebo for CV events but did not meet its co-primary efficacy endpoint. As mandated by the 2008 FDA guidelines, EXSCEL demonstrated that Bydureon does not significantly increase the CV risk, as measured by three-point MACE (non-fatal MI, non-fatal stroke, and CV death), in patients with type 2 diabetes. That said, EXSCEL was also powered to demonstrate CV superiority – with over 14,000 patients enrolled – and the trial did not meet this endpoint, though AZ shared that numerically fewer CV events were observed in the Bydureon arm of the trial. The company also characterized the general safety and tolerability profile of Bydureon in the trial as consistent with previous trials – it’s a relief to know there are no major surprises on this front. Full results will be presented at EASD 2017 in Lisbon, Portugal and we’re eager to dig into the study design and findings more at that point.

Bydureon is the fifth CVOT to report topline results for a GLP-1 agonist (and the third trial powered to potentially demonstrate CV superiority). Previously, both Novo Nordisk’s Victoza (liraglutide) and once-weekly semaglutide demonstrated impressive cardioprotective benefits in the LEADER and SUSTAIN 6 trials, respectively. The SUSTAIN 6 results were particularly surprising and impressive, given the small size of the study and the fact that it was designed to meet a pre-approval inferiority threshold. On the other hand, results from the ELIXA trial of Sanofi’s Lyxumia/Adlyxin (lixisenatide) were resoundingly neutral. The smaller, pre-approval FREEDOM-CVO trial of Intarcia’s implantable ITCA 650 (exenatide mini-pump) also demonstrated non-inferiority but not superiority (it was not sufficiently powered to demonstrate superiority) – only topline results for this last trial have been announced and we eagerly await full results.

In our view, the EXSCEL results further suggest that there may be some heterogeneity in the CV effect of GLP-1 agonists. While it’s certainly possible that the findings of EXSCEL may be due to the study design (it’s been speculated that the Novo Nordisk trials enrolled a population that is better able to benefit from the CV effects of GLP-1 – we have no information on this), the very large size of EXSCEL and the fact that AZ is certainly no stranger to designing outcomes trials for their many cardiovascular drugs lead us to suspect that the non-inferiority finding is “real.” We think it might be quite possible that there may be drug-specific aspects driving CV superiority vs. non-inferiority. Thus far, we’ve seen CV superiority demonstrated for both of the human GLP-1-based drugs (liraglutide and semaglutide) and non-inferiority for the exendin-4-based drugs (lixisenatide and two formulations of exenatide). There isn’t a clear mechanism for this to our knowledge, though in general the human-based GLP-1 agonists have demonstrated more potent results for other diabetes endpoints as well (in a head-to-head study, semaglutide was superior to Bydureon in terms of A1c and weight loss). We look forward to the REWIND CVOT results for Lilly’s human GLP-1-based Trulicity (dulaglutide) to potentially shed more light on this matter – the trial is expected to complete in July 2018. Clearly there are lots of big picture questions in light of these topline results and we’re looking forward to hearing from key opinion leaders.

  • The non-inferiority finding of EXSCEL is a disappointment to AZ, which bet heavily on Bydureon demonstrating CV superiority given the size, length, and cost of this trial. Indeed, AZ recently created a combined Cardiovascular and Metabolic Disease unit, positioning the cardiorenal comorbidities of diabetes front and center for its diabetes portfolio. The company is certainly positioning its SGLT-2 inhibitor Farxiga (dapagliflozin) as the future headliner of this unit (and has recently initiated heart failure and renal outcome trials for the drug, even though DECLARE CVOT results are not yet available). That said, AZ management had looked to EXSCEL with some excitement and a positive readout in EXSCEL would’ve been much-welcomed. The third major leg of AZ’s diabetes portfolio – DPP-4 inhibitor Onglyza (saxagliptin) – also demonstrated CVOT results with non-inferiority overall but a statistically significant increase in heart failure. We imagine that given these neutral findings from EXSCEL and the unwieldy dosing procedure for Bydureon (though an auto-injector is under FDA review), alternate options like Victoza or once-weekly Trulicity will become increasingly popular. We also expect that there will be quite a demand for semaglutide within the GLP-1 agonist market, assuming its approved at the end of 2017, given the combination of potent A1c lowering and weight loss with cardioprotection in a once-weekly formulation. In light of these dynamics, we expect AZ will continue to emphasize Farxiga even more in future communications and our fingers are crossed for positive CV findings for that product.

Close Concerns Questions:

Q: How much of a numerical difference was there in CV events between the Bydureon and placebo arms? Was the p-value “close” to statistical significance?

Q: What was the end-of-trial A1c difference between the Bydureon and placebo arms? Some have attributed a portion of liraglutide and semaglutide’s CV benefit to potent glucose-lowering – was the A1c difference for EXSCEL comparable?

Q: How did use of other medications throughout the trial in each arm of EXSCEL compare to usage in LEADER or SUSTAIN 6? Some have suggested that substantially increased use of SUs in the placebo arm of LEADER could be driving some of the results – what did SU usage look like in EXSCEL?

Q: Are the EXSCEL results likely to be driven by a specific aspect of Bydureon or something related to the trial design, “luck,” etc.?

Q: What was the impact of Bydureon on hypoglycemia, renal outcomes, and retinopathy? What was the impact on heart failure?

Q: How will these results affect already-sluggish sales of Bydureon?

Q: What will the REWIND CVOT results for Lilly’s human GLP-1-based Trulicity (dulaglutide) look like?

Q: Will the Bydureon results lead some to question whether ITCA 650 may be associated with a CV benefit (given the exenatide molecule in both products)?

 

-- by Helen Gao and Kelly Close