The transition of PD1 and PD-L1 inhibitors into the first line setting for at least some patients with advanced non-small cell lung cancer (NSCLC) is poised to be the biggest sea change in the management of lung cancer over the next 1-2 years.  Based on the preliminary report that the Keynote 024 trial of the PD1 inhibitor Keytruda (pembrolizumab) vs. chemotherapy as first line treatment for patients with advanced NSCLC and PD-L1 expression of 50% or greater in their tumors (the subset of approximately 25-30% of patients who consistently demonstrate the best probability of major tumor shrinkage and prolonged survival from Keytruda and other PD1/PD-L1 checkpoint inhibitors), we already knew that the ground is shifting under our feet. But with literally dozens of first line trials with several agents and subtly different patient populations and comparisons, we have many open questions:

1) Will immunotherapy agents provide extremely similar results in comparable populations?

2) Which patients will truly benefit from immunotherapy more than chemotherapy? All patients, most, or only a rather selected subset with the highest predicted probability of immuno-sensitivity?

3) Will checkpoint inhibitors be most helpful when administered  as single agents in first line, combined with chemotherapy, or potentially combined with other immunotherapy agents?

4) Do immune checkpoint inhibitors need to really be administered indefinitely, or can patients do just as well with a limited duration of therapy followed by sustained benefit?

5) Will the lower bar of improved progression-free survival in first line treatment translate to the more valuable prize of prolonged overall survival, particularly after patients have the benefit of crossing over from one treatment to another? In other words, does the timing of treatment matter significantly, so that it's important to administer immunotherapy in the first line setting rather than later?

 The huge collection of trials testing first line immunotherapy in advanced NSCLC will test all of these issues. There are so many that will be reported in the next few years, it's essentially impossible to keep track of all of them. I think of them as a "Chinese menu" where companies can just order "family style" options they choose from the different columns to create their own new trial with their PD1/PD-L1 checkpoint inhibitor of choice.

Today, Bristol-Myers Squibb, who make the PD1 inhibitor Opdivo (nivolumab), offered a press release that the CheckMate 026 trial comparing Opdivo to any of various chemotherapy regimens as first line treatment of advanced NSCLC did NOT show any improvement in the primary endpoint of progression-free survival (PFS).  A total of 541 patients with either squamous or non-squamous NSCLC and with at least 5% of their tumor cells positive for PD-L1 (a rather liberal cutoff point achieved in about 40-45%  of cases) were randomized to receive Opdivo at the current standard dose of 2 mg/kg IV every 2 weeks or one of several platinum-based doublet chemotherapy options, with the choices differing a bit depending on the tumor subtype. That's all the information we have so far.

What does this mean? First, as I mentioned in the wake of the positive press on the Keynote 024 trial with Keytruda, the benefit seen there in first line was seen in the narrowest subset most likely to benefit from immunotherapy (those with >50% PD-L1 expression).  The CheckMate 026 trial studies a more diluted population that includes far more patients not as likely to benefit greatly. Though we have yet to see any progression-free survival curves, I think there's a great chance we'll see a criss-crossing shape such as that seen in the IPASS trial of EGFR inhibitor Iressa in Asian never-smokers, a shape that suggests two distinct populations within the study. Specifically, I think the problem with CheckMate 026 may be that the study population was too broad and that the patients with the most PD-L1 expression actually do the same as the Keytruda recipients on Keynote 024, but the rest of the patients on Checkmate 026 do better with first line chemotherapy. When you mix those two populations together, you end up with no difference overall. 

It's notable (and a bit ironic) that the more liberal enrollment criteria for Opdivo vs. Keytruda in the second line NSCLC space has led to this setting being completed dominated by Opdivo thus far. In the second line studies that led to FDA approval of both agents, the Opdivo trials enrolled patients regardless of PD-L1 staining levels (including even patients with no PD-L1 expression, least likely to benefit from PD1/PD-L1 checkpoint inhibitors); in contrast, the studies in previously treated pateints receiving Keytruda focused on patients with higher PD-L1 expression levels and therefore led to an approval that is conditional on PD-L1 testing and is only available for patients who test positive (with best results seen in patients with higher level PD-L1 expression overall). BMS got away with this more liberal enrollment and accrual in the second line setting because second line chemotherapy with docetaxel is a relatively low bar to exceed (proven survival benefit, but relatively underwhelming).  But first line doublet chemotherapy has more meaningful activity, so it's not at all surprising that a greater benefit from immunotherapy, at least as a single agent strategy, is only going to be apparent in a narrower subset most likely to benefit from immunotherapy. So BMS is now feeling the sting of the double-edged sword of less selective use of immunotherapy in a different setting. 

It's also possible that Ketruda is simply more effective than Opdivo, but the patterns in similar populations of patients have been so similar that I would consider that far less likely than the differences in results being due to differences in the study populations.

The press release on CheckMate 026 doesn't say anything about overall survival. It's harder to see changes in overall survival than progression-free survival (which is part of why the preliminarily reported survival benefit from Keynote 024 is particularly impressive to me), but overall survival could trend in the wrong direction for first line immunotherapy if many immunotherapy zealots (patients and physicians alike) misinterpret worse-looking scans as pseudoprogression even as patients experience a clinical decline. This could mean that patients who aren't benefiting from first line immunotherapy are actually being harmed by missing the opportunity to benefit from chemotherapy in a timely way.  If patients become too debilitated to receive doublet chemotherapy while progressing through immunotherapy, it could realistically prove to be net harmful. We have a precedent for this in the form of the European TORCH trial from several years ago that took a patient population unselected forn EGFR mutations (and in whom we could expect an activating EGFR mutation in 10% or less of the population) to receive either first line chemotherapy or Tarceva (erlotinib) and then switch to the other treatment upon progression: that trial was stopped early because of an observed far worse progression-free survival and overall survival in patients getting first line Tarceva. Though they crossed over to chemo upon progression, they benefited far less from the same chemo as second line treatment. That may well prove to be the case after first line immunotherapy for patients with less immuno-sensitive disease.

There is obviously a great deal still to learn. We don't have any of the actual data from either of these first line trials yet. But we now have our first hint that it's time to pause and conclude that immunotherapy, at least first line single agent therapy, isn't going to be the best treatment for all patients.  We need to enter into an era of more nuanced discussion of the benefits of immunotherapy and recognize that it's not the right tool for every job in every patient with cancer.