770P Prognostic and predictive value of ribonucleotide reductase subunit family M1/M2 in urothelial carcinoma

Abstract

Ribonucleotide reductase catalytic subunit M1 (RRM1) and ribonucleotide reductase regulatory subunit M2 (RRM2) are involved in DNA replication and DNA repair. Preclinical and clinical evidence has shown that high RRM1 expression is associated with resistance to gemcitabine and cisplatin in certain cancer types, but the predictive and prognostic value of RRM1 and RRM2 in urothelial cancer (UC) remains to be studied. Microarray datasets related to gene expression in UC were obtained from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases and supplementary data from published papers. Kaplan-Meier analyses were performed for mRNA expression and clinical outcome correlations. The associations between mRNA expression and clinicopathological parameters were analysed using Chi-square/Fisher exact tests and Mann-Whitney U test. In the UC cohort from the TCGA (n = 407), high expression of RRM1 (RRM1-H) and RRM2 (RRM2-H) were significantly associated with worse outcomes. The median OS in RRM1-H population was 28.80 m vs 98.47 m (HR=2, 95%CI 1.3-3.1, P=0.003), while the differences in RRM2 expression levels were also associated with distinct clinical outcomes (median OS of patient with RRM2-H, 27.30 m vs 47.43 m, HR=1.5, 95%CI 1.1-2, P=0.008). To further evaluate the prognostic value of RRM1 and RRM2, another 2 cohorts with adequate information on mRNA expression and survival were analysed (GSE13507, the CNUH cohort, 165 primary UC patients with surgical resection of a transitional cell carcinoma; and GSE32894; the SSH cohort, 224 patients undergoing transurethral resection) and results showed that RRM1-H or RRM2-H were associated with a poor OS in UC (HR, 2.7; 95% CI, 1.3-1.5; P =0.005 in CNUH cohort. HR, 6.6; 95% CI, 2.03-22.0; P < 0.001 in SSH cohort). Patients with high transcript abundance of RRM1 and RRM2 were validated to show worse survival benefits (P < 0.05), which indicate that RRM1 and RRM2 are potential prognostic biomarkers. We also conducted an analysis of an immunotherapy cohort (IMvigor 210, a single-arm, multicentre, phase 2 study, included 348 patients with atezolizumab treatment), in this cohort, patients with RRM1-H or RRM2-H were associated with higher Objective Response Rate (ORR) (24%), and associated with a better tendency for OS in UC(HR 0.68, 95%CI 0.45-1.00, P= 0.061). Further analysis indicated that RRM1 and RRM2 expression levels were positively correlated with PD-L1 expression, tumour-immune phenotypes (P < 0.05 for All). High RRM1 or RRM2 expression levels were a potential prognostic biomarker in UC patients across different disease courses. In addition, RRM1 or RRM2 expression levels showed some predictive value for patients with UC receiving immunotherapy.