Real-world patterns of first-line treatment selection after the introduction of enfortumab vedotin plus pembrolizumab for advanced urothelial carcinoma

Abstract Background The approval of enfortumab vedotin plus pembrolizumab (EVP) has substantially changed the first-line treatment landscape for locally advanced or metastatic urothelial carcinoma. However, little is known about how first-line treatment is selected in routine clinical practice following the introduction of EVP. Methods This multicenter retrospective study included 175 consecutive patients with untreated locally advanced or metastatic urothelial carcinoma who initiated first-line management after EVP approval between September 2024 and January 2026. Patients received EVP, gemcitabine plus cisplatin plus nivolumab (GCN), or non-EVP management, including platinum-based chemotherapy, pembrolizumab monotherapy, supportive care only, or treatment refusal. Clinical factors associated with non-EVP selection were evaluated using receiver operating characteristic analyses, including area under the curve (AUC), multivariable logistic regression, decision curve analysis, and a nomogram. The EV-ineligible criteria (EVITA) were explored as a secondary assessment. Results Among 175 patients, 136 (78%) received EVP, 35 (20%) received non-EVP management, and four (2%) received GCN. Patients managed without EVP were substantially older and had poorer ECOG performance status than those receiving EVP. Age demonstrated the highest discriminative ability for non-EVP selection (AUC 0.85), followed by ECOG performance status (AUC 0.77), both outperforming EVITA (AUC 0.71). In multivariable analysis, older age and worse ECOG performance status were independently associated with non-EVP selection, whereas EVITA was not. Decision curve analysis showed that a model incorporating age and ECOG performance status achieved the greatest net benefit, and the addition of EVITA did not improve model performance. A nomogram based on age and ECOG performance status demonstrated excellent discrimination (AUC 0.87) and good internal calibration. Conclusions Following the introduction of EVP, first-line treatment selection in clinical practice was primarily associated with age and performance status. Simple clinical variables outperformed the composite EVITA criteria in describing real-world treatment selection, highlighting the gap between structured eligibility tools and contemporary clinical practice.