Patient Selection for Pembrolizumab in HNSCC

Transcript:Ezra Cohen, MD: It’s really amazing when you think about it, that they are changing the way, I think, that we view these patients and approach these patients. The data so far are very encouraging, as we’ve all said. But we’ve talked about, let’s say, 20% of patients responding, maybe a little bit less than that. Tanguy, tell us, are there biomarkers that can help us select these patients? Should we be using the biomarkers in patients with head and neck cancer? What’s going on there?

Tanguy Y. Seiwert, MD: It’s a great question. Obviously, we would like to know who are the patients who have benefit and who are the patients who have the most benefit. That would be ideal. The approval for pembrolizumab and, likely, the pending approval for nivolumab will be without patient selection, without a biomarker, and treat all-comers. I think that’s appropriate. We do, however, have data about biomarkers that suggest that we can enrich patients. So, PD-L1 has been tested. The ligands, PD-1, have been tested in immunohistochemistry assays across multiple tumor types. And across multiple tumor types, including head and neck cancer, we can see an enrichment of patients with benefit.

If you do PD-L1 selection, you could potentially get up to, if you use a 50% cutoff or something close to that, a response rate of 40%, maybe somewhat higher. We’ve actually seen this in lung cancer, and I would say that’s probably similar in head and neck cancer. However, even patients who are PD-L1—negative still have benefit, and we see responses in about 8% to 10% of patients. It is an assay that can maybe help you inform patients that are more likely. However, it is not an assay that can select patients. Some patients who are PD-L1–negative still have great benefit and durable benefit.

So, right now, I think we should use it without the biomarker. However, I think when we come to the first-line setting, we may actually have benefit by selecting patients. In fact, we have seen very exciting data here at the ESMO 2016 meeting using pembrolizumab in first-line treatment of non—small cell lung cancer. Incredible data, with a hazard ratio of 0.5 in the first-line treatment, randomizing the first-line setting in strongly PD-L1–positive patients, doublet chemotherapy versus pembrolizumab with PD-1 inhibition, and pembrolizumab clearly remarkably outperforming chemotherapy. I think we will see some of those data likely as part of a first-line setting in head and neck cancer—that would be the KEYNOTE-048 study. And so, I think the answer to the biomarker question right now is no selection, but I think the question remains open because it might play a role in the first-line setting.

The other thing I would like to mention is that because PD-L1 is a complex biomarker, there’s an interest in novel biomarkers. I’ve been involved in looking at a novel biomarker called the interferon gamma signature. This is an assay that can also be done actually quite easily with a rapid turnaround, and it seems to perform somewhat better than PD-L1 expression. It seems to actually have a high negative predictive value. When you have a negative assay for the signature, it may eventually allow us to exclude some patients who have no chance of having a benefit. I think it further needs to be validated. We need to have a real clear assay available. So, I think it will take some time. There are other things that are being explored. People are looking at mutational load or immunogenic mutations, and maybe dynamic biomarkers. Again, these are all experimental. But I think there’s a clear, large need to identify those patients that have more benefit. And I think we are working on that.

Ezra Cohen, MD: Kevin, Viktor, any other comments about the biomarkers?

Viktor Grünwald, MD: I would like to comment, actually, because I think that you raised a very good point. What we do is we look for a biomarker that delivers a predictivity of response. But, basically, what I’m more worried about is patients that do not respond to immunotherapy, because they might not have a chance to go on another round of chemotherapy whatsoever. So, I think the question that we have for patients not responding to PD-1 inhibitors is, what is the best treatment for these patients? Is it chemotherapy? Is it a combination? Is it something else?

I think this is something you should be more clear about. And one way to do it, hopefully, is a biomarker that would predict progressive disease to me would be much more usable than enriching patients, saying, “Oh, the chance of response is not 10%, it’s 30%.” And I see it more as a strategy for treatment algorithms, as you said, maybe in the first-line setting or maybe certifying for intense immunotherapy versus chemotherapy or something like this, but not so much right now at the point where we are.

Transcript Edited for Clarity