Biomarker Development in mRCC

Sumanta Kumar Pal, MD: Dave, I’m going to bring up that cancer cell paper from IMmotion151. I read it with a lot of interest. I sat down with the heat map that compartmentalized renal cell carcinoma into 7 groups. At the end of looking at that, I thought to myself, “How can we move this forward?” Is 7 groups too many for us to work with in this disease space?

David F. McDermott, MD: Not all groups will hold up. Group No. 7, which had an awesome response to I/O [immuno-oncology] therapy, was really small. So we’ll see. We obviously need a lot more data there. But those different subgroups were probably less—if you collapse them down, 2 of them were angio-high, at least 2 were I/O, and the other 2 were completely different. If we could break kidney cancer down into 3 subgroups, that would be helpful.

More important, the group at Genentech that drove this research deserves a lot of credit for wanting to study this because we did learn a lot about mechanisms of resistance. For example, there are other papers related to this work that look at myeloid infiltration of the tumor, not just in kidney cancer, but in other tumor types that have high levels of IL-6 and IL-8. Those are patients more likely to be resistant to PD-1, not just in kidney but in bladder, for example. Maybe that causes us to develop therapies that target the myeloid component, or lower IL-6 or IL-8 levels. We’re going to get much more out of this work when it comes to novel therapies than we are from selection, particularly in kidney cancer, because kidney cancer is so heterogeneous. When we get samples from patients, it’s often not the tumors we’re treating. It’s the primary tumor, which is often very different from the metastases. The individual metastases are different from one another.

If I was a junior faculty person wanting to do biomarker development, I would go as far away from kidney cancer as I possibly could, just because there are so many other tumor types that are much more homogeneous where tissue is much more accessible and where funding levels are higher. I would say to go somewhere else. Don’t do what I did because you end up with a lot of sadness.

Sumanta Kumar Pal, MD: I love it. That’s a really positive, inspirational message.

David F. McDermott, MD: Don’t end up like me? That’s good. Give it to them straight.

Tian Zhang, MD: But Dave, we might eventually get there in our careers. It’s really challenging to do prospectively designed studies with biomarker selection when it’s on a trans-chromosomic level. It also takes time and cost to subset out these patients. I have to be hopeful. I’m still of the young crowd in this group. Maybe we will get to a point where we’ll have biomarker-directed studies and be able to select those patient populations. It’s really important, as you say, to find the right patients to treat with the ipilimumab-nivolumab combination, the patients who will have complete responses and really durable responses, and to avoid the patients who will have resistance to ipilimumab-nivolumab up front. How do we go about that? How do we understand those resistance mechanisms? I fully agree with all your points. I wouldn’t give up on biomarkers yet.

David F. McDermott, MD: I was being a little over the top. For pure therapy, I/O [immuno-oncology]–I/O, you could come up with a marker. We should push to do that. There are just aspects of kidney cancer that make it more difficult. To be honest, industry doesn’t really like to fund science that narrows markets, so we have to push hard as a group and insist on it being done. For I/O–I/O, we can come up with a marker that narrows that population of patients getting it. I completely agree with you.

Nowadays, when you order a foundation assay, you get things you never could have imagined getting 5 or 10 years ago on the DNA side. I predict that 5 years from now, we’re going to get RNA sequencing data from these commercial platforms and you can make decisions. Using it to make clinical decisions is different from a companion diagnostic. A companion diagnostic, what you have in melanoma and lung cancer, needs a robust, likely tumor-based marker with a tumor-based therapy. With some of these novel therapies, like an HIF inhibitor, maybe you could get that—I want to believe you can, and we should work at it—but when you’re targeting the tumor micro-environment and it’s very heterogeneous and you’re combining therapies that target 2 different biologies, the clinical experience is outstripping the predictive world. But I was just kidding. You should keep doing what you’re doing, and I will clap from the audience when you’re up there.

TRANSCRIPT EDITED FOR CLARITY