Subtype-specific benefit of adjuvant chemotherapy in non-metastatic muscle-invasive bladder cancer with histologic variants: a four-endpoint, subtype-stratified analysis

Purpose Histologic variants (HV) of urothelial carcinoma portend adverse biology, yet subtype-specific guidance for adjuvant chemotherapy (ACT) after radical cystectomy (RC) remains limited. We aimed to disaggregate the prognostic effect of histology from the therapeutic effect of ACT across four endpoints and identify subtype-specific drivers of risk and response. Methods We assembled a neoadjuvant-naïve, pT2 + RC cohort operated between January 2013 and December 2024 at a single tertiary center (n = 555: 356 pure urothelial carcinoma [Pure UC], 199 HV). Overall (OS), cancer-specific (CSS), metastasis-free (MFS), and recurrence-free survival (RFS) were modeled in five-covariate Cox regression. The HV signal was disaggregated using three-way subtype coding, with within-subtype ACT effects and target × ACT interaction tests. In 110 patients with quantified variant component percentage (HV%), dose–response analyses were performed. Results ACT was independently favorable across all four endpoints (OS hazard ratio [HR] 0.41; CSS 0.42; MFS 0.63; RFS 0.49; all p ≤ 0.011). The HV-associated deficit localized specifically to the metastatic axis (MFS HR 1.29, p = 0.086). Three-way modeling identified sarcomatoid differentiation as the principal adverse driver (MFS HR 2.41, p = 0.038). Within-subtype analyses revealed chemoresistance in squamous, directionally favorable signals in micropapillary and sarcomatoid, and a marked RFS benefit in plasmacytoid (HR 0.07, p = 0.005). HV% showed no dose–response association or interaction with ACT. Conclusion ACT improved survival across all four endpoints; histology-associated risk concentrated along the metastatic axis and was driven by sarcomatoid and squamous differentiation, while the binary presence rather than quantitative extent of variant histology appeared therapeutically relevant.