Histological variants and lymphovascular invasion in upper tract urothelial carcinoma can stratify prognosis after radical nephroureterectomy.

Objectives: Patients with histological variants (HV) of bladder cancer have more advanced disease and poorer survival rates than those with pure urothelial carcinoma (UC). Moreover, lymphovascular invasion (LVI) is an important biomarker after RNU in systematic reviews and meta-analyses. Thus, here we investigated the clinical and prognostic impact of HV and LVI in patients with upper tract urothelial carcinoma (UTUC) treated with radical nephroureterectomy (RNU).

Methods: Data from 223 UTUC patients treated with RNU without neoadjuvant chemotherapy were retrospectively evaluated. We analyzed differences in clinicopathological features and survival rates between patients with pure UC and those with HV. Conditional survival (CS) analysis was performed to obtain prognostic information over time.

Results: A total of 32 patients (14.3%) had HV, with the most common variant being squamous differentiation, followed by glandular differentiation. UTUC with HV was significantly associated with advanced pathological T stage (pT ≥ 3), higher tumor grade (G3), and LVI, compared to pure UC (all P < 0.01). Progression-free survival (PFS), cancer-specific survival (CSS), and overall survival (OS), were all significantly worse in the HV group compared to the pure UC group (all, P < 0.001). In multivariable analysis, HV and LVI were independent predictors of CSS and OS. We classified the patients into three groups using these two predictors: low-risk (neither HV nor LVI), intermediate-risk (either HV or LVI), and high-risk (both HV and LVI). Significant differences in PFS, CSS, and OS rates were found among the 3 groups. In CS analysis, the conditional PFS, CSS, and OS rates at 1, 2, 3, 4, and 5 years improved with increased duration of event-free survival. CS analysis revealed that most progression events occurred within 2 years after RNU, and patients with risk factors had worse PFS at all time points.

Conclusions: A risk model using HV and LVI can stratify PFS, CSS, and OS of patients treated with RNU. In addition, CS analysis revealed that HV and LVI were poor prognostic factors over time after RNU.