Followers of Endocyte had pinned their hopes on overall survival data from vintafolide’s phase II Target study providing definitive proof of efficacy. With results presented at the Esmo congress on Saturday, however, defeat has been snatched from the jaws of victory.
Even a broad data dredge has barely managed to tease out a subgroup benefit – after one accepts Target’s advantageous statistical design, that is. Endocyte had risen 6% on Thursday in anticipation of the late-breaking Esmo data, but when the markets open again tomorrow investors will almost certainly be left nursing losses.
There were already question marks around vintafolide, Endocyte’s lead project. European approval was revoked after the Proceed study in ovarian cancer was halted for futility, prompting Merck & Co, which had paid $120m up front to access vintafolide, to pull out of the deal (Trial miss leaves Endocyte at the mercy of Merck, May 6, 2014).
Back in March, progression-free survival (PFS) results from the Target trial, which pitched vintafolide and docetaxel against docetaxel alone in a second-line setting for NSCLC, caused Endocyte to surge, and the company managed to raise $102m from investors. The fate of vintafolide, which comprises vinblastine linked to a compound that targets the folate receptor, thus rested on further analyses of Target – most crucially of patients’ overall survival (OS).
In fact, the OS data showed vintafolide scoring a miss. Risk of death was reduced by 12% – though with the upper confidence interval bound at 1.36 – with a p value of 0.2874.
One of the doubts around Target had been its statistical design, which determined that a result would be significant if its one-sided p value was under 0.1. In practice, trying to hit a one rather than two-sided p value reduces the rigour of the result, and cutting the threshold from the near-universally accepted 0.05 to 0.1 cuts this further still.
It was this way vintafolide managed to hit a median PFS benefit, its primary endpoint, with a 25% reduction in risk of progression and one-sided p value of 0.0696. The same methodology was applied to the OS calculation, and on these shortened odds the company still had to perform multiple secondary analyses before showing something it considered important.
Thus it looked at OS in patients stratified for numerous baseline factors – a miss – those with adenocarcinoma – not significant – and then stratified adenocarcinoma patients, which showed a 49% reduction in risk of death (one-sided p=0.0147). Endocyte stated that these analyses were pre-specified, but made no mention of adjusting the result for multiplicity.
|How vintafolide reaches a survival benefit by multiple subgroup analyses|
|Hazard ratio (95% CI)||One-sided p value|
|All comers in Target study||0.88 (0.58-1.36)||0.2874|
|All comers adjusted for baseline factors||0.75 (0.48-1.18)||0.1066|
|Adenocarcinoma patients||0.70 (0.40-1.22)||0.1018|
|Adenocarcinoma patients adjusted for baseline factors||0.51 (0.28-0.94)||0.0147|
Surely vintafolide will now be consigned to the scrapheap, but a bigger issue is whether the rest of the pipeline will head that way too, since like vintafolide all of Endocyte’s projects comprise a small-molecule targeting agent linked to a cytotoxic payload.
While Target’s lead investigator, Dr Nasser Hanna, from the Indiana University Department of Medicine, US, said the results were very interesting, the best he could say of vintafolide was that perhaps its cytotoxic element needed to be reconsidered.
But the Target data also cast doubt on the concept of directing a cytotoxic at the folate receptor. This mechanism should logically result in a targeted effect, and yet perversely the study reveals an increase in several “off-target” treatment-emergent adverse events, including peripheral neuropathy.
Since Endocyte is sitting on $121m of cash it will no doubt plough on with its pipeline, but the Target results show Merck’s decision to hit the eject button to have been the right one.