The neural phenotype in invasive urothelial carcinoma patients: Alternative score detection and prognostic implication.

Abstract

e17037 Background: Recent molecular subtyping studies (NGS) identified a subset (5-15%) of muscle invasive urothelial carcinoma (MIBC) with transcriptomic patterns consistent with neuroendocrine (NE) differentiation in the absence of NE histology (NE-like), representing a potentially high risk subgroup of carcinoma which may require a different treatment strategy. We recently set an alternative immuno-phenotypical score (Piescore), to discriminate Luminal from Basal from Neural carcinoma. Aim of this study was to test the ability of Piescore in identifying NE-like cases in a mono-institutional cohort of patients treated with trans-urethral resection and radical cystectomy (RC) and to correlate them with clinical outcomes. Methods: Transurethral resection specimens harbored foci of HG pT2 (MIBC) UC from 116 pts who subsequently underwent RC have been submitted for immunohistochemical analysis, using relevant gene-expression-based markers for Basal type (CD44, CK5/6) and Luminal type (CK20 and pPARg). Piescore divided Basal and Luminal types when at least 3 of the 4 markers were consistent with a specific phenotype; Mixed if two luminal and two basal markers were present simultaneously; NE-like when all four markers were negative. Results: Overall, the Piescore identified Basal phenotypes in 49 patients (42,2%), Luminal in 38 (32,7%), and Mixed in 9 (7,8%). No expression was identified in 20 patients (17,2%): 7 cases with morphological NE differentiation and 13 cases with classical urothelial histology, all consistent with NE-like phenotype. Interestingly, in 10/13 patients the NE-like phenotype was only documented in the muscle invasive component of the tumor whereas in the non-invasive component they retained Luminal phenotype in 9 cases and Basal in one. With a median follow up of 188 months, the pathological stage of disease (pT2 versus ≥pT3 and/or N+) and the tumor vascular invasion (absent versus present) resulted prognostic (Stage: 5-years DFS rate 65% versus 30%, p = 0.038; 5-years OS rate 69% versus 32%, p = 0.017) (vascular invasion: 5-years DFS rate 47% versus 24%, p = 0.020; 5-years OS rate 54% versus 21%, p < 0.001) in all population. No statistically significant differences in terms of pathological stage of disease, vascular invasion, DFS, and OS were observed in NE-like cases compared with non-NE-like cases. Conclusions: The NE-like urothelial carcinoma identified by Piescore immunophenotyping (CD44, CK5/6, CK20 and pPARg) did not show any statistically significant association with worse prognosis.