Prognostic and predictive genomic biomarkers in metastatic clear cell renal cell carcinoma (mccRCC): A large retrospective analysis of real-world data from a US-based clinico-genomic database (CGDB).

Abstract

473 Background: The absence of validated prognostic and predictive biomarkers constitutes a significant barrier to appropriate selection among immune checkpoint inhibitor combinations (ICI-C) for individual patients (pts) with mccRCC. Our study aimed to evaluate the prognostic and predictive value of single-gene alterations (GAs) and gene clusters in a large real-world case series of mccRCC. Methods: This study employed data from the US-based de-identified Flatiron Health-Foundation Medicine Inc. (FH-FMI) CGDB, comprising 858 mccRCC pts diagnosed between 2011 and 2022 in 280 US cancer clinics. The presence of GAs was determined using Foundation Medicine tests on tumor tissue specimens. The co-occurrence and mutual exclusivity of the most common GAs were assessed utilizing Fisher's exact test. The Louvain algorithm was applied to identify three co-occurring gene clusters (C) on the gene network graph: C1: VHL , SETD2 , PBRM1 , KDM5C , NFE2L2; C2: TP53 , TSC1 , TERT , DNMT3A ; C3: CDKN2A , CDKN2B , BAP1 , NF2 , MTAP . Pts were labelled as C+ if they had one or more mutations within a specific cluster and none in the others. Cox proportional hazard model was used to discern the prognostic and predictive value of GAs. Median overall survival (mOS) and progression-free survival (mPFS) were estimated using the Kaplan-Meier method. Results: The prognostic analysis encompassed 782 pts with mccRCC, while the predictive analysis involved 493 pts profiled with FMI genomic tests within 3 months of starting first-line systemic therapy (1L). GAs of seven genes ( CDKN2A , CDKN2B , TP53 , PTEN , NF2 , PIK3CA , MTAP ) correlated with a poorer prognosis, whereas PBRM1 mutants exhibited a favorable OS. C1+ pts had higher OS compared to C1- (mOS: 40 vs. 19 months; HR: 0.63, p <0.001), while C3+ was associated with worse OS (mOS: 18 vs. 27 months; HR: 1.36, p = 0.023). In 1L, 201 pts (40.8%) received antiangiogenic monotherapy (AAm) (mPFS: 7.0 months) and 172 pts (34.9%) received ICI-C (16.4% ICI+ICI, 18.5% ICI+AA) (mPFS: 6.8 months). 170 (34.5%), 43 (8.7%), and 71 (14.4%) pts were categorized as C1+, C2+, and C3+, respectively. TERT , TSC1 , and TET2 alterations were positive predictors of 1L-PFS for ICI-C vs. AAm (HR=0.44, 0.23, 0.20; p<0.05). C1+ favored AAm over ICI-C (HR=2.30, p=0.01) while C2+ favored ICI-C over AAm (HR=0.39, p=0.039). Among ICI-C, ICI+AA was more effective than ICI+ICI in pts with SETD2 alterations (HR=0.38, 0.020), and less effective in mutant TSC1 (HR=8.60, p=0.013). Conclusions: This study highlights the prognostic value of 8 GAs and 2 clusters of co-occurring GAs, and identifies 3 GAs and 1 cluster of co-occurring GAs to be positive predictors of 1L-PFS for ICI-C in a large mccRCC population. Prospective validation is required to confirm the prognostic and predictive role of these GAs and to tailor therapeutic strategies.