A 5-gene Mutation Signature Predicts Response to Atezolizumab in Metastatic Bladder Urothelial Carcinoma

Background: Immune checkpoint inhibitors have improved outcomes in metastatic urothelial carcinoma, but predictive biomarkers beyond programmed death-ligand 1 (PD-L1) expression and tumor mutational burden (TMB) are urgently needed. In this study, we aimed to identify a novel mutation-based signature that can accurately predict response to atezolizumab and to evaluate its predictive performance independently and in combination with TMB.

Methods: We analyzed 298 patients from the IMvigor210 trial treated with atezolizumab. A 5-gene mutation signature comprising HOOK1, LGSN, PRR32, ADCY5, and ADAMTSL2 was derived via Fisher's exact test and least absolute shrinkage and selection operator (LASSO) regression. Patients were classified as mutation carriers (≥1 mutation in five genes) or non-carriers. We evaluated association with response, overall survival (OS), and pathway alterations.

Results: In the cohort, 32 patients (10.7%) were identified as mutation carriers. Mutation carriers had higher response rates than non-carriers (81.2% vs. 14.8%, P<0.0001). The 5-gene mutation signature independently predicted response, while TMB did not (P=0.0716). A combined TMB and 5 gene mutation signature model achieved AUC of 0.833, outperforming TMB (0.743) or signature (0.729) alone. Carriers had significantly prolonged OS (median not reached vs. 9.49 months; hazard ratio (HR)=0.105, P<0.0001). Gene Set Enrichment Analysis (GSEA) revealed cell cycle upregulation and downregulation of Epithelial-Mesenchymal Transition (EMT)/angiogenesis/KRAS pathways in carriers.

Conclusions: The 5-gene mutation signature is a robust independent predictor of response and survival in anti-PD-L1-treated metastatic urothelial cancer. Combining it with TMB enhances predictive accuracy, offering a practical strategy for patient selection in clinical practice.