Biomarkers in RCC: Current Landscape and Potentially Actionable Developments

Rana McKay, MD, University of California, San Diego, speaks with David Braun, MD, PhD, Yale Cancer Center, on the current biomarker landscape for renal cell carcinoma and how biomarkers have shaped the treatment paradigm to date.

Dr. McKay: Why don’t we start by having you describe the current biomarker landscape in RCC and how these biomarkers have potentially shaped treatment, if at all.

Dr. Braun: The short answer is the majority of biomarkers probably haven’t shaped treatment just yet, but I think there is room for some optimism. There have been disappointments in the biomarker space in kidney cancer, but I think there are some shining bits of light that hold promise for the future. In my mind, there is sort of genomic biomarkers, or things intrinsic in the tumor itself. Total mutation burden. Neoantigens. These sorts of things. And while those have a certainly predictive effect in other tumor types, and we know melanoma, non-small cell lung cancer, and there’s a pan-histology approval for pembrolizumab with a high tumor mutation burden, that association doesn’t seem like it really holds water for kidney cancer.

Beyond these aggregate measures of mutations, we’ve looked at individual mutations that might have an impact. While there are maybe signals here or there of things that are not inert, there are some things that maybe have some impact, some influence on response. Look at things like PBR1 loss of function mutations or copy number variants in regions of chromosome 9p. There’s nothing that’s risen to the level of an actionable biomarker that’s really an impact for the patient in front of us should they get Treatment X or Treatment Y. I think genomic biomarkers have been really thoroughly explored, but still nothing that’s risen to the level of biomarker.

The second bucket is transcriptomics. Gene expression. Really looking at not individual genes, but signatures of genes that, together, mark some behavior of the tumor as a whole. For these, they’re not just the tumor cells themselves, but the tumor and the surrounding microenvironment. There have been lots of forays into this space, but the one that really stands out to me is work done by Doctors Motzer, Rini, and colleagues at Genentech, where they really broke down clear cell kidney cancers into different molecular clusters of subsets based on their transcriptomic patterns. There were ones that were more angiogenic. There were ones that were more infiltrated. There were ones that were more proliferative. There were also some ones that we still don’t understand the biology, like small nuclear RNA.

But what they had seen, and this was in the context of the IMmotion151 trial, the atezolizumab trial versus sunitinib, was there appeared to be certain molecular subtypes that might be treated effectively with sunitinib, TKI monotherapy, things like the angiogenic subtypes, and others that really seem to benefit from the addition of immune therapy. I think there’s caveats there, of course, that the trial did not lead to an FDA approval. These are not using FDA-approved drugs, and so we still have to really validate: are these clusters real and do they apply to have a predictive value for other therapies?

But for me, that was the first kind of interesting point, where it really did seem to distinguish outcome. There are actually trials that are already being designed based on these subtypes, like the OPTIC RCC trial, having immune therapy is still as a backbone, but whether you do an IO/TKI, angiogenic subtypes, or a pure IO-based approach in immune or immune subtypes.

So, that’s the transcriptomic landscape. There’s other things which are kind of earlier stage, especially endogenous retroviruses and these sorts of things, but that, to me, has the most sort of promise for the future. Though, again, nothing that’s ready for prime time yet.

The third bucket is maybe the oldest: histology and pathology. Looking under the microscope, some things like PD-L1, which we know has lots of troubles and is probably not the strongest predictor in kidney cancer by any means and has mixed results. But actually, the old looking-under-the-microscope of the pathologists looking at sarcomatoid histology, that’s maybe one of the best biomarkers  for choosing an IO-based therapy. We know from work from Dr. Tanir and others, looking at the CheckMate 214 trial, that those patients with sarcomatoid histology really stand out as having some exceptional responses. A nearly 20% complete response rate. Around a 60% overall response rate compared to the 40% for the overall population. Out of all of the ones that are probably actionable, it’s the oldest, the looking-out-of-a-microscope, and seeing the sarcomatoid histology, that probably is the most predictive at this point.

Dr. McKay: Thank you so much for that wonderful summary of where we are with biomarkers and the current landscape. You began to touch on the actionable biomarkers, and in the present day, the only largely actionable biomarker is patients with sarcomatoid histology who really should be receiving immunotherapy-based treatments and likely maybe nivo/ipi. Can you comment on what other actionable biomarkers are potentially out there, and maybe even share with us the story around PD-L1? There was a lot of excitement and enthusiasm around that biomarker upfront, and what has happened with that biomarker.

Dr. Braun: I think PD-L1 is something that has an intuitive place as a biomarker for anti-PD-L1-based therapy. If you don’t have to ligand for this inhibitory checkpoint, naively, you wouldn’t expect the tumor to necessarily be responsive to anti-PD-1-based therapy. There have been challenges that I would group into technical and biological. Technical challenges are just practical, right? There’s been a history of antibody variation that’s sort of impacted staining and what you can call this PD-L1-positive and negative. There has not been a pure consensus on. What is important is a PD-L1-positive tumor, so is it having it on the tumor cells? Is it having it on other immune cells, like myeloid cells? Some combination? Obviously, there’s scores that incorporate one or both of those. There are certainly technical elements that I would put in the realm “not perfectly solved problems.” Certainly advances in those, but not perfectly solved.

The others are biological, which is: PD-L1 is not a static marker and it’s not homogeneous. It’s not that you sample one bit, a tiny piece of tumor, at one moment in time and that is universally reflective of the PD-L1 status for that patient. We know it’s dynamic. It’s an interferon gamma responsive gene. As immuno-filtration differs, either over time or over space, in different spots of the same tumor or in different tumors within the body, there will be different levels of PD-L1 expression. So, capturing this one moment in time with the biopsies is a challenge. Those are the challenges.

In RCC, it’s my impression of an aggregate of these studies that, again, PD-L1 is not inert. It’s not that PD-L1 has zero impact. As we look across studies, especially the more modern checkpoint therapy trials in the frontline, I would say we see some enrichment in my eyes for PD-L1-positive patients, slightly higher response rates, better responses, better survival, these sorts of things, in that PD-L1-positive population, but it doesn’t rise to the level of a discriminant biomarker. When we talk about biomarkers, really what we want is EGFR, right? We want something where we get a test and it says positive or negative, and we know that if it’s positive, it has a high likelihood of working, and if it’s negative, it’s probably not worth giving. PD-L1 comes nowhere close to that level in kidney cancer. I think that’s been the challenge, and when we think about both the technical and the biological issues, I think it kind of makes sense in that context.

View their other comments on Novel Biomarkers, Strategies for Patient Selection, and Novel Therapies With the Most Promise.