34: Real-world performance of genomic, histologic, and radiographic features in predicting long-term outcomes in patients with clear cell renal cell carcinoma treated with first-line immunotherapy-based combination regimens.

Background: In clear cell renal cell carcinoma (ccRCC), clinicopathologic disease features including sarcomatoid dedifferentiation, somatic mutations, or depth of response have been commonly implemented to stratify patients to PD-1 based combinations ipilimumab/nivolumab (IO/IO) or targeted-therapy/immunotherapy (TKI/IO); however, data validating this utilization is limited.

Methods: Patients treated at MSKCC with IO-based combination therapy for ccRCC between 1/1/2014 and 12/31/2020 were identified. Cohorts were defined by first-line treatment: IO/IO or TKI/IO. Responses were assessed via Response Evaluation Criteria in Solid Tumors (RECIST) V1.1 criteria. Progression-free survival on next-line therapy (PFS-2) is defined as time from first-line therapy start to second progression or death. Deep response is defined as target lesion shrinkage ≥80%. Sarcomatoid features were ascertained using pathology reports. Tumor mutations were profiled via MSK-IMPACT. Restricted mean survival time (RMST) up to 48 months was calculated for PFS-2 in each treatment group. To test for heterogeneity of treatment effect among subgroups, sarcomatoid features and genomic alterations are included in a generalized linear model for RMST and interaction test performed. PFS-2 in deep responders was compared between treatment groups.

Results: One-hundred-seventy-three patients were included (90 received IO/IO; 83 TKI/IO); 141 patients underwent MSK-IMPACT. Among altered genes, only BAP1 was significantly associated with PFS-2: median PFS-2 for BAP1-altered vs -unaltered patients was 17 months (95% CI 11, 27) vs 44 (29, 53), p=0.006. In the RMST model for PFS2, the interaction between treatment group (TKI/IO vs. IO/IO) and presence/absence of sarcomatoid features or BAP1 alteration was not significant (respectively, p=0.21; p=0.65) (Table). Among deep responders, PFS-2 events occurred in 2/17 patients in the IO/IO group and 4/23 patients in the TKI/IO group (log-rank p=0.75).

Conclusions: The evaluated candidate biomarkers did not demonstrate predictive or prognostic potential for stratification to initial treatment modality. TKI/IO and IO/IO remain acceptable standards of care, where clinical stratification remains key for decision making.