Brian I. Rini, MD, on Renal Cell Carcinoma: Exploratory Biomarker Results

KEYNOTE-426 was a randomized phase III trial in advanced frontline clear cell kidney cancer that randomized patients to pembrolizumab plus axitinib or sunitinib monotherapy. It was a registration trial that led to FDA approval, and the clinical data has been presented many times showing advantages to the combination arm over monotherapy and response rate, PFS and overall survival. Presented at ASCO as the first translational data from this trial. About 80 or 90% of the patients had baseline tissue that was analyzable. What was analyzed were three main things. One is RNA sequencing, looking at what's called a T-cell inflamed gene expression profile. This is a Merck-derived gene expression profile that basically defines inflammatory or hot tumors. An angiogenesis signature looking at expression of angiogenesis-related genes, and then some other various RNA-seq signatures. DNA alterations were looked at by whole exome sequencing, and what was looked at was the common alterations in kidney cancer like Von Hippel-Lindau and PBRM1 and the like. Then PDL-I staining, which has been reported before, but re-reported here was looked at in terms of the combined positive score. The translational data was linked to the five-year follow-up clinical data presented at ASCO last year and associated with outcome. The main findings were that within the pembo axitinib arm, patients who had higher expression of that T-cell-inflamed GEP, so more inflamed tumors, had better clinical outcomes, not surprisingly. In the sunitinib arm, patients who had higher expression of the angiogenesis-related genes had better clinical outcome. Again, not surprisingly. PDL-I unfortunately was not associated with outcome and it's been a fairly disappointing biomarker in kidney cancer. In terms of the non-GEP, RNA-seq signatures, not much was found. There was some association with a myeloid-derived suppressor cell signature, but it lost significance after adjusting for the T-cell-inflamed GEP, and some of the proliferative signatures were associated with worse outcome, especially in the sunitinib arm. Perhaps not surprisingly; just worse proliferative tumors that don't do as well. In terms of DNA alterations, what was seen is that there really wasn't any consistent association, and this has been seen before with DNA alterations in kidney cancer where there hasn't been good DNA alterations associated with outcome. In this analysis, within the pembro/axi arm, patients who had PBRM1 gene mutant versus wild type had about a 20% improvement in response rate, so there was some difference there. PBRM1 has been looked at in kidney cancer and had sort of variable results, so I think these contribute to those variable results. Then as mentioned, PD-L-I staining, unfortunately, was not really associated with any clinical outcome. In sum, what we see is that patients who have more inflamed tumors do better with the combination. We see higher response rates for the combination across cohorts, but again, it's sort of enriching for response within that patient population. The hard work now is really taking these results and using them prospectively. These results teach us about biology. They teach us about drug mechanism, a lot of which we knew, but I think what we really need to do is design biomarker-based trials that use, I think, baseline RNA-seq results, which I think are the most robust, assign patient's treatment based on that expression or not, and then associate that with outcome. That's how we're really going to move the field forward.