Transcript:Robert L. Ferris, MD, PhD: Human papillomavirus (HPV) is responsible for an increasing subset of head and neck cancers. It turns out that head and neck cancers in the oropharynx, the tonsil and the base of the tongue, are the primary site where HPV causes head and neck cancers. So, when we talk about sub-sites and risk stratification, we need to recognize that HPV-positive head and neck cancers are essentially exclusively located in the tonsil and the base of the tongue, otherwise known as the oropharynx. Risk stratification based on HPV status is crucial because survival and clinical outcome is dramatically improved in HPV-positive cancers. Therefore, we think about oropharynx cancer, which is 80% to 90% HPV-positive, as a good prognosis subsite and a low risk stratification, a group of patients who are destined to do very well.

P53 is the most commonly altered gene in all of human cancers, and head and neck cancer is no exception. It’s genetically altered 50% to 60% of the time, and it is a poor prognostic factor when it is altered versus when p53 is wild-type. Another alteration is upregulation of the epidermal growth factor receptor, or EGFR. The higher the EGFR expression, the worse the prognosis. So, using these molecular markers or drivers, we can stratify head and neck cancers into the poor prognostic ones with high EGFR expression and altered p53, whereas the p53 normal, or wild-type, and the EGFR-low patients tend to do better.

Ezra Cohen, MD: When we think about the typical head and neck cancer patient, the first thing that we think about with respect to risk stratification is whether their cancer is related to HPV or not. Now, of course, we always take into account staging, the TNM system, performance status, the age of the patient, and their comorbidities. All of those form our gestalt with respect to how that patient is going to do, and what we can tell them about their prognosis. But, in terms of molecular markers, by far the most important one is HPV-status. And, of course, we now recognize that HPV-positive patients, when we equalize, all those other parameters do much better than HPV-negative.

But, there are some emerging factors that we’re now beginning to pay attention to. And, of course, many of them are related to immunotherapy, and either prognostic or predictive markers are related to that. What we’ve come to recognize for head and neck cancer, as for many other cancers, is a tumor that has a microenvironment that appears to have responded to cancer. Let’s call it an inflamed microenvironment or CD8-positive infiltrated. It is a situation histologically where we can say that the immune system has somehow recognized and responded to this cancer. We’ve come to recognize that those patients tend to do better. And so, that is a part of the prognostic gestalt for our patients.

With respect to PD-L1, what we’ve come to notice is that although it may not have prognostic implications, that’s really still getting worked out. What we have come to recognize is that it does have a relationship with response rate to PD-1 or PD-L1 antibodies. That’s a complicated story because what we’ve come to recognize is that there are different antibodies that can measure PD-L1. There are different ways to interpret the assay. There are different cutoffs. To make a sweeping statement like, “PD-L1 expression has a relationship with response rate,” is always going to be fraught with some danger. But, that relationship in general is probably true. Now when we get into more of the specifics of that, what we come to recognize is that, first of all, the expression on stromal cells in head and neck cancer, as opposed to some other malignancies like lung cancer, is critically important. In fact, most of the PD-L1 is expressed on stromal cells—like macrophages or other white cells in the environment, even some T cells in the environment—as much as, or more than, the actual tumor cells.

The other thing that we have to take note of is that despite the fact that there is some relationship between response rate and PD-L1 status—when we apply agents like pembrolizumab and nivolumab—right now, even in PD-L1-negative patients, those drugs appear to do better than standard of care. So, even in a PD-L1-negative patient, we can still see a response rate—it’s usually about 10%—and it still appears that, for instance for a drug like nivolumab, it is going to improve overall survival compared to a cytotoxic chemotherapy. And so, for head and neck cancer, we don’t use PD-L1 as a determinant for therapy. But, that story, of course, is going to evolve probably tremendously over the next couple of years.

Transcript Edited for Clarity