Hot Versus Cold Tumors: Predictive Markers in Melanoma

Transcript:Keith Flaherty, MD: It’s been a great discussion. I think the final point we should touch on is this notion that we have this armamentarium of therapies. We have clinical parameters we’re using now to try to guide therapy. Obviously, it would be nice to be thinking in the future that we’d have some strategy for navigating patients to treatments. There’s very intriguing data in the immune space that supports the idea that we might be able to take the temperature of tumors, to some degree, and understand which ones are close, just maybe one shove away.

Jeff, can you give a few comments on your view of that emerging evidence? I think this audience will be aware of the concept that PD-L1 testing in other tumor types, not melanoma, has been a part of clinical trials, and they’re watching these drugs roll out. I’d be interested in your thoughts about PD-L1 testing and then also maybe what are some of these other signals that we’re seeing in terms of other markers.

Jeffrey Weber, MD, PhD: Well, there is a lot of discussion about PD-L1 as a predictive biomarker. So, why would you want a predictive biomarker? Either to tell you who you should treat, or tell you who you shouldn’t treat. As a marker to decide who you shouldn’t treat, PD-L1 staining falls short, certainly in melanoma. And I heard Chuck Drake, from Hopkins, give a terrific talk about the data using PD-L1 with immunohistochemical staining as a predictive biomarker. And it turns out, it doesn’t work in squamous lung cancer. It works very poorly in renal cell cancer. It works well in non-squamous. It works reasonably well in melanoma to enrich for those who will respond and will have long survival and benefit.

The problem is, it doesn’t go the other way. It doesn’t allow you to filter out the non-responders. BRAF mutation testing is a wonderful predictive marker. I assume you all would agree, I would never treat a BRAF wild-type patient with BRAF plus MEK or BRAF inhibitors, it makes no sense. It’s the perfect way of eliminating people from getting treatment. In fact, I think you’d agree, because of paradoxical MAP kinase activation, it would be detrimental.

Keith Flaherty, MD: It could be worse, right.

Jeffrey Weber, MD, PhD: It would make things worse. With PD-L1 staining, the data from the BMS-066 trial would suggest it’s more of a prognostic marker because, even in the chemotherapy treated patients, PD-L1-positive folks who got their dacarbazine did better than PD-L1-negative patients who got their dacarbazine. So I don’t see a future for it, I see it as a scientific marker of an emerging concept that Antoni Ribas and Paul Tumeh have proposed.

Tom Gajewski has taken it to another step to try to understand the idea of the hot tumor and the cold tumor. The cold tumor has no PD-L1 expression, no T-cell infiltrate. Those things, by the way, go together because the influx of T-cells making gamma interferon probably upregulates the PD-L1, versus a hot tumor, which is maybe 30%, 40% of all the melanoma patients where you have T-cells infiltrating. They’re in the tumor. They get out or into the tumor itself, and there’s PD-L1 positivity throughout the tumor, and those are the patients who are going to do well. Those are probably the patients who respond to pembrolizumab, nivolumab, and ipilimumab.

So the question is, as you’re implying, how do you get from a cold tumor to a hot tumor? That’s the new frontier and understanding how to impact on the tumor microenvironment with an HDAC inhibitor with TRAIL antibodies, with IDO inhibitors, with OX40 antibodies, with ipilimumab, radiation, toll-like receptor agonists. I think that’s the new frontier, and I think we’ll see real progress in melanoma when we can make a cold tumor into a hot one.

Keith Flaherty, MD: So, Jason, why don’t you close us out with the targeted therapy considerations? Do you have a sense that there’s an emerging understanding? Start with the BRAF-mutant population, because beyond BRAF testing, which gets you kind of started, we know there are patients who have quite divergent outcomes. Any sense that we’re going to be able to envision the idea of testing additional genetic factors to help try to understand who are those patients who could derive long-term benefit versus the ones who unfortunately would have short-term benefit?

Jason Luke, MD, FACP: Well, I think there’s a number of markers that are worthy of investigation that just need more work, and you’ve presented a great talk on this at this SMR meeting. One of them is phosphorylated-S6 looking at activation of parallel pathways and how that is influenced by MAPK signaling. There are a number of other molecules that are important to melanoma around MITF and oxidative phosphorylation that may also be able to predict. Now, the hard part is, how are we going to do those tests? I think the only real answer, and I think this is what you suggested as well, is we’re going to find a peripheral way to do this through cells in the blood or something else. I think that really needs to be an area of strong work moving forward for BRAF mutant melanoma.

And I will just say that I also think that we shouldn’t lose sight of the fact that BRAF/MEK is great. It’s possible that BRAF/MEK plus something else would be even better. And we’ve had some difficulty doing those clinical trials, because just so many things have happened in the field. But we should probably go back and reexamine some of those things as well. There’s biology there through upregulation of RTKs and other things. An Hsp90 inhibitor may be able to really do some damage there. Upregulation of MET seems to be a nongenomic mechanism of resistance.

So, again, multiple triplet combinations in the targeted therapy field. And then you move past BRAF and other subsets of melanoma, which was another big part of this meeting. And there’s activity with a MEK inhibitor perhaps with an addition of a CDK4 inhibitor in RAS mutant melanoma. We now know from the TCGA analysis, that the NF1 mutant population is another molecular subset of melanoma for which we really could be able to do some things, and investigating MEK inhibitors in that subset could be efficacious. So, I think we can take what we learned in the BRAF-mutant space, start to apply it to these other molecular subsets, but then simultaneously try to look for real-time diagnostics to tell us how our targeted therapy is working in this individual patient and how that can guide things.

Transcript Edited for Clarity