Toni K. Choueiri, MD, on Renal Cell Carcinoma: Potential Predictive Biomarkers of Treatment Efficacy

What we presented was the biomarker analysis. So, we had data on RNA-Seq, on immunohistochemistry. We asked a lot of questions during this study. The first question we asked is if there are any gene signatures that were previously described in RCC with VEGF inhibitor, but PD-L1 inhibitor with nivolumab that's a PD-L1 inhibitor. So are they different? And actually we couldn't validate any of the signature that were described with PD-L1 inhibitors plus VEGF inhibitors such as the bevacizumab atezolizumab from emotion 151 or from JAVELIN Renal 101 with axitinib and avelumab. But of course this is a PD-L1 inhibitor, so we weren't able to validate that. But we actually found some hallmark gene sets and some individual genes associated with progression-free survival. And that started us to characterize the tumor microenvironment of the responders to PD-L1 plus VEGF inhibitor better. We also looked at immunohistochemistry, specifically at CDAT cell, at PD-L1 and at c-MET looking at different cutoff and definitions. And also we couldn't find anything that is predictive for the combination versus single agent sunitinib. The only thing that came up, and this was not surprising, is that some prognostic markers, so for example, in sunitinib treated patient PD-L1 positivity is associated with worse outcome. We knew that. This is not new at all and nivolumab, cabozantinib resulted in better PFS regardless. Same thing with MET, the membrane CT, c-Met expression was not associated with much, but the cytoplasmic c-Met was associated with worse prognosis. So this was not predictive. Overall in term of next step and what that means that we continue to try to identify predictive biomarkers, probably the next step using a composite model for nivolumab, cabozantinib efficacy. But it seems all these biomarker we identify, they may be specific to a drug or a combination rather than class of agents such as VEGF and PD-1, PD-L1 inhibitor.