Membranous Nectin-4 expression to predict enfortumab vedotin response in metastatic urothelial carcinoma.

Abstract

854 Background: Enfortumab vedotin (EV) is a standard of care for metastatic urothelial carcinoma (mUC), yet robust predictive biomarkers are lacking. Although its target, Nectin-4, is widely expressed, reported levels vary and the relative impact of membranous versus cytoplasmic localization remains unclear. While NECTIN4 amplification strongly predicts EV benefit, it occurs in only ~25% of cases, underscoring the need for refined biomarkers among the remaining ~75% non-amplified patients. We validated NECTIN4 amplification and systematically characterized subcellular Nectin-4 expression, developing a membranous scoring algorithm to enhance patient stratification. Methods: We retrospectively analyzed 183 pretreated mUC patients receiving EV (Nov 2019–Apr 2025, multicenter). NECTIN4 amplification was assessed by FISH and correlated with Nectin-4 immunohistochemistry (IHC). A four-tier membranous scoring system (0–3+), adapted from CAP HER2 gastric criteria, was benchmarked against H-scores. We further built an integrated model combining amplification and membranous staining categories to predict response. Results were compared with biomarker data from the EV-301 trial. Results: Membranous Nectin-4 expression (median H-score 160, IQR 70–240) was significantly lower than cytoplasmic (200, IQR 125–250; P < 0.001). Combined membranous + cytoplasmic scoring yielded higher values (median 250; 79.6% ≥150), consistent with EV-301 (median 250; 82.6% ≥150). A ≥150 cutoff enriched for EV response (our cohort: 52.6% vs. 29.2%; EV-301: 45.8% vs. 20%, P = 0.0014). High membranous expression (2+/3+) predicted superior ORR (59% vs. 21%, P < 0.001), PFS (8.8 vs. 2.6 mo; HR = 0.36, P < 0.001), and OS (13.0 vs. 8.3 mo; HR = 0.50, P = 0.005), whereas cytoplasmic staining was non-predictive. NECTIN4 amplification conferred excellent outcomes (ORR 77.5%; mPFS 17.0 mo; mOS 30.0 mo; all P < 0.001). The integrated model delineated three prognostic groups: amplified, non-amplified/high membranous (2+/3+), and non-amplified/low (0/1+) with ORRs of 77.5%, 45.6%, and 19.4%; mPFS 17.0, 5.5, and 2.6 months; mOS 30.0, 9.5, and 6.9 months (all P < 0.0001). Conclusions: Membranous, but not cytoplasmic, Nectin-4 independently predicts EV response in mUC. NECTIN4 amplification identifies long-term responders, while our novel membranous scoring system refines stratification among non-amplified patients. Comparison with EV-301 suggests that previously reported high expression levels likely reflected combined membranous and cytoplasmic staining, diluting the predictive power of membranous expression alone. These data establish NECTIN4 amplification and membranous Nectin-4 expression as complementary biomarkers to optimize EV patient selection.