Yesterday, news that the Checkmate 026 trial that compared the PD1 checkpoint inhibitor Opdivo (nivolumab) to standard chemotherapy as first line treatment for advanced non-small cell lung cancer (NSCLC) failed to demonstrate an improvement in the primary endpoint of progression-free survival. This led to Bristol-Myers Squibb stock being spanked, with a 16% drop over the day, while the stock price for Merck , the makers of the remarkably similar agent Keytruda (pembrolizumab), jumped more than 10% on the same news, since Keytruda had been reported in a press release to beat standard chemotherapy in a similar trial, Keynote 024. 

This website isn't focused on analyzing what investors should do, but it's clear that these dramatic changes in stock price represent a major misinterpretation of the take-home conclusions from the press releases that have emerged thus far.  The fact that Keytruda beat chemo in a first line trial vs. chemotherapy but Opdivo failed to beat chemo in a first line trial doesn't mean that these agents are meaningfully different and that Keytruda is superior (disclosure: I've been paid to serve on advisory boards for both Merck and BMS in the past and am not pulling for either).  To understand why Keytruda beat first line chemotherapy in Merck's Keynote 024 trial but Opdivo didn't beat first line chemo in BMS's Checkmate 026 trial, I refer you to one of the major points I made back in mid-June, when I offered my commentary about the positive results with Keytruda (bold included when originally written on 6/16/16):

In contrast to Keynote 024, which focused on the cherry-picked population who we would have predicted to have the most immuno-sensitive disease, Checkmate 026 enrolled a much broader population of patients, including far more patients who were relatively less likely to respond well to immunotherapy. Specifically, though the method in which tumor cell staining with PD-L1, the marker we use to predict probability of benefit from immune checkpoint inhibitors targeting PD1 or PD-L1, is a little different for the different studies and agents, the trend is very consistent that the patients with cancers that have a lot of PD-L1 expression have the highest probability of responding well to immune checkpoint inhibitors and tend to have the longest benefit with them. The Keynote 024 trial enrolled only pateints with PD-L1 expression on at least 50% of their tumor cells, about 28% of the patients who would be tested for potential participation for the trial. In contrast, the Checkmate 026 trial was more liberal, though not extremely so, using a 5% cutoff that would have limited enrollment to 41% of all patients tested based on earlier work.  I believe that the most likely explanation for the failure of Opdivo in Checkmate 026 is that the dilution of the study population beyond what seems to be probably not more than a third of patients who are most likely to benefit from immunotherapy undermined what would have very likely been positive results with Opdivo if the Checkmate 026 study had enrolled the same 28% tested and found to be PD-L1 positive at a higher 50% expression cutoff.  

Ironically, a more liberal definition for inclusion has really helped BMS win tremendous market dominance in the second line setting, since Opdivo was approved without requiring PD-L1 testing and showed a benefit in an unselected population, while Merck focused on a narrower population selected by PD-L1 expression that led to their approval being only after patients had been tested and found to be positive for PD-L1. Both Opdivo and Keytruda beat Taxotere as second line treatment, but that is a relatively low bar that was achievable even without patient selection for the Checkmate trials with Opdivo in second line. And with two similar agents and no requirement for the extra time and money of PD-L1 testing for Opdivo, it has been the vastly preferred agent for second line. In contrast, though, first line platinum doublet chemotherapy is a more effective comparator arm, so a less restrictive enrollment less likely to favor immunotherapy recipients seems to have backfired for BMS  in this different setting.

We will soon have an avalanche of trial data on first line immunotherapy that will make it easier to interpret the landscape, and we don't even have the actual data beyond the press releases for these two trials. But at this moment, it would be a mistake to conclude that there is a clinically significant difference between Keytruda and Opdivo (except for the more convenient every three week schedule for Keytruda vs. every two weeks for Opdivo).  What we have clearly learned is that immunotherapy is not the clear best choice for all or even most patients. It appears that the benefit of single agent immune checkpoint inhibitors may well be limited to no more than about 1/4 to 1/3 of advanced NSCLC patients, even as first line therapy.