Validity of PD-L1 Testing in Melanoma

Transcript:Keith T. Flaherty, MD: I think this raises the point. When you think about the adoption of many approved drugs—particularly the large number of available in the United States—in terms of referral to major centers, uveal probably comes top of the list where you’d say that’s a group where it’s essentially clinical trials all the time. Mucosal, acryl, intermediate I suppose are next because they can respond to available therapies. Jeff, before we jump into the advanced disease contacts in therapy, I just want to get your quick thoughts on PD-L1 testing. So, before we leave this molecular testing paradigm, is that a feature in your practice?

Jeffrey S. Weber, MD: PD-L1 testing is assessment of surface staining, usually by immunohistochemistry, of the ligand for a PD-1, which is a break on the immune system present on T-cells. The problem with PD-L1 staining is it’s a 1940s assay, immunohistochemistry that was developed probably 70 years ago. It probably hasn’t advanced that much since. It’s an inducible marker, meaning it can go up or down depending on the level of T-cell infiltrate, level in inflammation within a tumor. Within a tumor in the body, you can actually have heterogeneous staining. So, depending on where you stick your needle, you can get one result or another. Within the body of the patient from one tumor to the other, it also may vary.

When you think about it, putting that together, it makes it one of the world’s worst potential predictive markers, because a predictive marker by its nature should be something that’s easily measurable, has a robust easy-to-manage test, and is essentially invariant like BRAF mutations. It’s the perfect predictive marker. The problem with PD-L1, as I’ve mentioned, is all of the pitfalls. And if it would distinguish someone who would get no benefit from PD-1 blockade or CTLA-4 blockade from those who would, it would be great. But unfortunately even though we all agree, if you’re a PD-L1 patient, you probably have a higher chance of responding to a PD-1 blocking or a PD-L1 blocking antibody, the PD-L1 negatives still can benefit. So, unfortunately, it hasn’t had that much traction in spite of its early promise. I personally don’t ever get it on my patients.

Jason J. Luke, MD: And I just put it in the context of other diseases that are being treated in the community. In lung cancer, the role of PD-L1 staining is debatable. People have opinions in melanoma at the current time that there really isn’t a role in the clinical space to do the testing, because if you actually look at the data, the response rate in the PD-L1-negative patients to a monotherapy PD-1 inhibitor is still better than what we see with ipilimumab or chemotherapy. So, even with the biomarker, the PD-1 inhibitor is still the best drug. Just give the drug. But that’s probably an evolving area as we get more agents, as we’ll consider what is the most appropriate.

Georgina Long, BSc, MBBS: And not just response rate, overall survival. The nivolumab versus DTIC trial clearly showed those who were PD-L1 negative had a much higher 1-year and 2-year overall survival than either DTIC arms, which had PD-L1 positive or negative. So, it’s clearly beneficial, I agree.

Jason J. Luke, MD: One could hypothesize that in a future environment in which we had prospective randomized data comparing targeted therapy and immunotherapy, there might be subset analyses where you could pull out patients. But we’re 5 years from that, that’s still out there. So, I don’t think in clinical practice that that’s an approach that really should be done.

Georgina Long, BSc, MBBS: PD-L1 is a blunt tool for inflammatory tumor. It’s a very blunt tool. And we’re getting better at trying to assess that. And although you made the comment that immunohistochemistry is old, it doesn’t mean it’s not good. One area where we’re really pursuing is in multiplex, which is basically immunohistochemistry but being able to visually, spatially see how cells are related. And that might be something that’s more informative about these when you get a whole tissue section of these inflamed tumors. Heterogeneity, though, is still a problem.

Jeffrey S. Weber, MD: Interesting factoid is that I heard a talk last night where a new molecular technique is being pursued on spectroscopy looking at protein and protein fragments within tumors to assess on a more molecular quantitative basis the expression of something like PD-L1. And the assessment showed that in a moderate-sized series of tumors, the false-positive and negative rates were huge, about 70%, comparing a more molecular rigorous quantitative technique to immunochemistry.

Transcript Edited for Clarity