Novel predictive biomarkers of response to immune checkpoint blockade with nivolumab ± ipilimumab in the TITAN-RCC phase 2 trial.

367 Background: Despite promising therapeutic efficacy of immune checkpoint inhibitors in renal cell carcionoma (RCC) response varies significantly between individual patients. Therefore, predictors of response are urgently needed. To identify novel predictive biomarkers, we explored the characteristics of blood-circulating immune cell subsets within the population of the TITAN-RCC trial (NCT02917772) applying a tailored approach with nivolumab (nivo) and ipilimumab (ipi). Methods: In TITAN-RCC, patients with intermediate and poor risk advanced clear cell RCC started nivo Q2W induction. Upon early progressive disease (PD, week 8) or non-response (stable disease or PD) at week 16, patients received 2-4 nivo+ipi “boost” cycles. Responders (complete or partial response) to nivo induction continued with maintenance but could receive nivo+ipi for later PD. Blood samples for biomarker analyses were taken at baseline, during nivo induction and nivo+ipi “boost” cycles. Samples from 198 RCC patients (105 first line, 93 second line) were analyzed by multi-parametric flow cytometry for frequency and phenotype of T cell, monocyte, myeloid-derived suppressor cell (MDSC) and dendritic cell (DC) subsets. Baseline data were associated to response upon nivo induction and data from samples taken prior to first “boost” to response to nivo+ipi. We applied uni- and multivariable logistic regression modelling to investigate the association between treatment response and immune parameters. Here we report on single marker models (adjusted for age and gender). Results: Higher percentages of blood-circulating 4-1BB+ CD4+ T cells (adjusted odds ratio (ORadj) 1.05, 95% CI 1.02-1.08), 4-1BB+ CD8+ T cells (ORadj 1.03, 95% CI 1.01-1.07) and LAG3+ CD4+ T cells (ORadj 1.03, 95% CI 1.01-1.05) were found in responders to nivo induction compared to non-responders. Moreover, in patients receiving nivo+ipi “boosts”, a higher proportion of PD-L1+ CD14+ monocytes (ORadj 1.22, 95% CI 1.06-1.58), PD-L1+ early-stage MDSC (ORadj 1.14, 95% CI 1.02-1.41) and PD-L1+ plasmacytoid DC (ORadj 1.08, 95% CI 1.01-1.17) was observed in responders compared to non-responders. Conclusions: We identified various immune cell-related parameters that are associated with increased or poor therapeutic efficacy in RCC patients which are currently investigated in multi-marker models. These parameters may represent novel predictive biomarkers for response to nivo±ipi in clear cell RCC. Clinical trial information: NCT02917772.