Predictive value of phenotypic signatures of bladder cancer response to cisplatin-based neoadjuvant chemotherapy.

Background: Cisplatin-based neoadjuvant chemotherapy (NAC) for muscle-invasive urothelial carcinoma of the bladder confers only a modest survival advantage. Patients with pathologic progression on NAC have poor outcomes related to a delay in definitive surgical management.

Objective: To characterize the value of epithelial-mesenchymal transition (EMT) effectors and other novel biomarkers to predict response to NAC.

Methods: A tissue microarray was constructed from patients with clinical stage T2 urothelial carcinoma of the bladder using transurethral resection (TUR) specimens (N = 69) and case-matched post-NAC cystectomy specimens (N = 51). Patients were stratified based on pathologic response to NAC and cancer-specific survival. Biomarker expression in TUR specimens was correlated with pathologic response to NAC and clinical outcomes. Phenotypic changes in expression induced by cisplatin-based NAC were characterized in primary TUR and post-NAC cystectomy and lymph node metastasis specimens.

Results: Increased expression of mesenchymal markers and actin-cytoskeleton regulators, as well as low apoptosis index, in TUR specimens was associated with pathologic progression on cisplatin-based NAC. Overexpression of N-cadherin and decreased apoptosis was predictive of disease-specific mortality. NAC decreased cofilin phosphorylation and induced a mesenchymal-epithelial transition phenotype.

Conclusions: The epithelial-mesenchymal transition phenotype and actin-cytoskeleton remodeling are associated with pathologic progression on NAC. Chemotherapy induced an mesenchymal-epithelial transition phenotype and decreased cofilin phosphorylation, providing new insights into therapeutic resistance mechanisms. Primary tumors with high apoptosis rates exhibited favorable response to NAC and lower mortality rates, indicating that these tumors may be sensitized to therapy. Further validation will establish these novel signatures as predictors of therapeutic response.