CDKN2A alterations as markers of immune checkpoint blockade (ICB) resistance in urothelial carcinoma (UC).

Abstract

475 Background: ICB has shown clinical benefit across several metastatic carcinomas; however, predictive biomarkers are still lacking. CDKN2A is one of the most commonly altered genes across human cancers. With prior studies giving conflicting evidence regarding the association between CDKN2A alterations and ICBs, we examined the impact of CDKN2A alterations on clinical outcomes in UC patients treated with ICBs. Methods: Of 809 patients at the Dana Farber Cancer institute (DFCI) treated with ICBs only and with relevant cancer types and targeted exome sequencing data (Oncopanel), 235 (29%) had loss-of-function (LOF) mutations or homozygous deletions in CDKN2A. Overall survival (OS) was compared by Cox logistic regression between CDKN2A altered and CDKN2A wild type (WT) patients. Hazard ratio (HR) was derived using multivariable analysis (MVA), adjusted for prior lines of therapy and tumor mutational burden (TMB). A validation cohort from Memorial Sloan Kettering Cancer Center (MSKCC) (Samstein et al., Nature Genetics, 2019) of 811 cancer patients treated with ICBs was analyzed in a similar manner, adjusted for TMB. As a control, the association between CDKN2A alterations and OS was examined in a cohort of platinum-treated UC patients (N = 56) to determine whether CDKN2A alterations were predictive of response to ICIs. Results: For the DFCI and MSKCC cohorts, median follow-up was 26.9 and 24 months (m), respectively. In the DFCI and MSKCC cohorts, CDKN2A alterations were found in 32/90 (35%) and 22/104 (21.2%) of UC, respectively; 4/55 (7.3%) and 3/131 (2.3%) of renal cell carcinoma, respectively; 73/178 (41%) and 45/194 (23.2%) of melanoma tumors, respectively; 86/370 (23.2%) and 26/260 (10%) of non-small cell lung cancer (NSCLC) tumors, respectively; 18/66 (27.2%) and 4/53 (7.5%) of esophagogastric tumors, respectively; and 22/50 (44%) and 11/69 (15.9%) of head and neck, respectively. CDKN2A alterations were significantly associated with shorter OS and TTF in the DFCI UC and melanoma cohorts by MVA, and showed a trend towards significance in the MSKCC UC cohort (Table). There was no significant association between CDKN2A alterations and OS for the other cancer types in both cohorts; and no association with OS or TTF was seen in the DFCI cisplatin-treated UC cohort. Conclusions: CDKN2A alteration status may serve as a predictive biomarker in patients with UC treated with ICBs. Further studies are needed to examine the mechanism of this clinical effect. [Table: see text]