Paul K. Paik, MD:Since the initial bolus of studies taking a look at single-agent immunotherapy that began around 2015 and then 2016, the field quickly moved on to taking a look at combinations. If the single drug had efficacy, perhaps giving it together with some other medications would have even greater efficacy. And the 2 areas in which it has been explored are I-O/I-O combinations, and then the ones that have been of interest that have led to some shift in FDA approvals have been chemotherapy plus immunotherapy, particularly in the treatment-naïve first-line setting.
KEYNOTE-021 cohort G was a study that was also presented at the end of 2016 that led to accelerated approval of the regimen of carboplatin/pemetrexed/pembrolizumab given in the first-line setting for lung adenocarcinoma patients. It’s important to note that we have not yet had any trial data read out for squamous cell lung cancer patients. KEYNOTE-021 cohort G was a randomized phase II trial, so smaller numbers against standard-of-care carboplatin/pemetrexed followed by maintenance pemetrexed in this population. And the intention-to-treat results were positive. The overall response rate was higher at 55% versus around 30%. PFS was also improved significantly, again a hazard ratio of around 0.6. And overall survival, while not fully mature yetand so we have to be cautious about this—an early peak last year demonstrated that the survival curves were beginning to separate, which is an important thing to note about this regimen.
The trial on its surface then was encouraging, but the problem is in the details, and the details are important. We no longer think of the FDA approval of pembrolizumab in the first-line setting of the first-line subgroup as a unified group of patients; we think of them, at this point, as the ≥50% or the <50% subgroup of patients. And we have our standard-of-care therapies based on that. KEYNOTE-021 cohort G did not take that into consideration. It started before, of course, the FDA approval of pembrolizumab in the high PD-L1 expressers.
And so, if you take a look at the subgroups of patients by PD-L1 status, the PD-L10 patients’ response rates seemed to be better than with chemotherapy. And the high expressers, PD-L1 ˃50%, the response rate was also better in the high expresser versus chemotherapy. There was, however, a problematic subgroup, the 1% to 49% of patients, which is about one-third of the patients, where it appeared that chemotherapy actually was better by response rate than pembrolizumab. That difference was something like 38% in favor of chemotherapy versus 26% in favor of pembrolizumab.
And so, because that’s the case and because the control arms are different, depending on the degree of PD-L1 expression, while it is an FDA-approved regimen, I think we have to be careful about looking at the data and trying to figure out in whom to give it. We have no direct guidance, for example, that in the ≥50% subgroup that this triplet regimen is better than just pembrolizumab by itself. In fact, it may expose patients to more toxicity. You’re giving more drugs, and you’re giving chemotherapy to this subgroup of patients. We don’t know yet prospectively what is better, so we have to be a little cautious about that. And we also have to be cautious because of that problematic 1%-to-49% subgroup that the other subgroup may not be as great as might be evident if the numbers were higher.
There is a confirmatory randomized phase III trial where the top-line results have just been read out. This is KEYNOTE-189. This wasagain, in the intention-to-treat population—a positive study, but we will have to see on its presentation later this year what the different subgroups look like and whether or not there is equal efficacy among these subgroups before, for me, we roll these out in terms of treating our patients.
The clinical implications of KEYNOTE-021 cohort G we began to see shortly after its accelerated FDA approval. It’s important to mention, again, that some people have begun giving it in certain subgroups. Perhaps the most common subgroup that oncologists have identified for this regimen are patients who have a large burden of disease, who are very symptomatic of their disease, who have lung adenocarcinoma where we really need to shrink their tumors fairly quickly and by a lot in order to get them to feel better. Because, at the end of the day, that is the primary thing that we’re trying to achieve immediately at the beginning of our patients’ diagnoses. And so, it stands to reason then that you would want to select a regimen that seems to have the highest response rate, the highest chance of working in these patients. And, again, in that subgroup analysis for the 0% patients and the ≥50% patients, as well as in the intention-to-treat population, the numbers are better for the triplet combination of carboplatin/pemetrexed/pembrolizumab. So, in our very symptomatic large-burden patients, this is a reasonable indication or place to give the triple combination and something that would be approved.
I think at the end of the day, it is a discussion with the patient about what the data look like and about how they’re feeling, what their performance status is, and what their past medical history is in terms of recommending this and prescribing this. Clearly, the patients who have existing autoimmune diseases, they would not be eligible for this. Patients who are relatively frail, one would need to consider whether or not giving triplet chemotherapy in these patients or even doublet chemotherapy in these patients would be a good idea. So, what I’d say is that until we really see that confirmatory KEYNOTE-189 data, it is a case-by-case basis. And I think by-and-large, this is how practitioners in the academic world and in the community setting have been doing this, which I think is a very reasonable way to end up utilizing it in the accelerated approval phase.
One of the concerns that we have had in the academic circles with the accelerated approval of carboplatin/pemetrexed/pembrolizumab, the absence of mature overall survival data, is the question as to whether or not we are burning through our options too quickly for our patients. And this gets at the question as to whether or not sequencing, for example, chemotherapy followed by single-agent immunotherapy progression might be better or even similar as giving all 3 drugs up front. We don’t yet know what strategy is going to be the best, but of course, practically speaking for those of us who opt to give carboplatin/pemetrexed/pembrolizumab, we are then in a lurch a little bit in terms of what to do in the second-line setting, which used to be we give immunotherapy. So, here then, the standard of care becomes docetaxel with or without ramucirumab. That really is the gold standard. The trial data support docetaxel/ramucirumab as being better than docetaxel. I personally give docetaxel in a weekly regimen for better tolerability in this setting. But it does again consume through 2 lines of therapy at the beginning, forcing us to look for other options in the format of chemotherapy or consideration, of course, of a clinical trial in the second-line setting in these patients.
Transcript edited for clarity.