Transcript: Daniel J. George, MD: In the last few minutes, I wanted to switch gears 1 more time to biomarkers. As we see this portfolio expand to different treatment options, ideally we’d like to be more precise in our patient selection. Prognostic factors are 1 thing, predictive factors are another, and developing biomarkers that can help us predict for that clinical benefit is key. Sarcomatoid elements are 1, and that may not be a sexy new 1, but it does seem to be a negative prognostic factor though a positive predictive factor.

Nizar M. Tannir, MD, FACP: For IO [immuno-oncology] therapy.

Daniel J. George, MD: What about PD-L1 [programmed death-ligand 1] status? We talked a little about that in addition to sarcomatoid. Are there other settings where we’re using PD-L1 status in renal cell?

Chung-Han Lee, MD, PhD: PD-L1 status is still fairly controversial as a biomarker within RCC [renal cell carcinoma]. Certainly, in CheckMate 214, they shared the associations with higher PD-L1 status, like being associated with better responses, but for the TKI-IO combinations, it has not been as predictive. The jury is still out in that sort of setting and especially with that many different PD-1 [programmed cell death protein 1] assays that we could use, whether we’re looking at the tumor itself or at the immune cells, whether it’s a combined score, or regardless of what the cutoff should be. Those are the things that still need to be worked out. It doesn’t seem like the cutoffs that were used in other disease groups are directly translatable to RCC.

Robert S. Alter, MD: Right. If you look at the CheckMate 025, JAVELIN Renal 101, and even [A031203] CABOSUN [cabozantinib-sunitinib], their primary endpoints were met with progression-free survival and overall survival, irrespective of the PD-L1 status. When atezolizumab-bevacizumab came out, and we looked at that data that David McDermott presented 3 years ago, we felt as if our patients with PD-L1 would thrive on these targeted immunotherapies, and we’re just seeing it makes us a little more excited, but not as if that is our biomarker of choice.

Nizar M. Tannir, MD, FACP: I think it’s fair to say that in renal cell carcinoma, at least PD-L1 as just a single biomarker is not there to really select therapy. I think it is prognostic, definitely for patients who are treated with TKIs alone. We know there are the data that patients with tumors that are PD-L1—positive, don’t do well at all with TKI as monotherapy. I think there is definitely a higher response rate in patients who receive immune checkpoint inhibitors if their tumors are PD-L1–positive, but you do get benefit in patients with PD-L1–negative, and that’s the KEYNOTE-426, CheckMate 214. All those phase III trials have shown benefit for both groups, PD-L1–positive and PD-L1–negative.

The jury is still out for what is the best biomarker, if there is any biomarker to really help us predict. There was an abstract presented at ASCO [American Society of Clinical Oncology Annual Meeting] in the clinical science symposium on the JAVELIN Renal 101 phase III trial, looking at predictive biomarkers, and it had interesting data. It shows that you can do whole exome sequencing and RNA sequencing on 700 patients, but this data is specific to a regimen. If you do the same thing for IMmotion151, or for JAVELIN Renal 101, or for CheckMate 214, we’ll come up with a different signature of genes that predict response. I really don’t think we are there to use this data in the clinic for our patients. The biomarker that I’m really interested in, that I think we should have—and we talked about some of the challenging toxicities of these patients—the immune checkpoint inhibitor. Can we predict, before we give a patient an immune checkpoint inhibitor, who is going to get this bad severe toxicity? I think it is really important to spare that patient the horrendous toxicity if we knew beforehand that they have a polymorphism or something that predicts for really a bad outcome.

Daniel J. George, MD: Some small antibody load or anti-muscle antibodies or something like this—myasthenia gravis. We’ve seen some of these really severe cases and gone back and looked at blood and found a small level of antibody present. I think that’s a good point. I am glad to see that our clinical trials are including more of this prospective collection of tissue and blood to look at these for these predictive markers. I do think we need to get there. I agree with you: I don’t think they’re ready for prime time, but it’s high time that we started building this kind of prospectively into our trials.

Chung-Han Lee, MD, PhD: I completely agree. I think it’s very important that we incorporate these biomarkers and further this development, because with so many options, we can’t run a large phase III of every single permutation. We’re going to have to really let some of the science guide us in terms of what we do and how we stratify.

Daniel J. George, MD: One last thing, you know we’re here at ASCO—care for every patient, learn from every patient. There’s going to be need for some registry-type work in which we get in some real-world data on patients and maybe also some imaging and blood collections on patients.

Nizar M. Tannir, MD, FACP: Dan, you’re doing that at Duke Cancer Center in Durham, North Carolina.

Daniel J. George, MD: We’re planning something like this—a large, probably electronic health record—based prospective cohort study—to look at both academic and community centers, at outcomes in patients with metastatic renal cell carcinoma, collecting baseline blood samples, and imaging repository, and a tissue repository. It’s a big, long effort, but I think these are the kind of things that can really confirm and corroborate some of the markers that may evolve in the future, and maybe find that common marker across studies that we’re looking for and things like that. I think a lot more work needs to be done. Exciting times in the field. The plot only thickens.

Transcript Edited for Clarity