Identification of a novel urothelial carcinoma (UC) biomarker of lethality.

Abstract

4569 Background: Limited information exists regarding the changes that occur in UC tumors that ultimately metastasize, the lethal phenotype of UC. Recurrent copy number alterations (CNAs) have been previously identified in UC, but correlation with clinical outcomes has not been reported in metastatic UC. Methods: We identified a cohort of 94 UC patients uniformly-treated for distant metastases (platinum-based combination chemotherapy), with clinical annotatation and FFPE primary tumor tissue available. Overall survival (OS) was defined as the time from start of chemotherapy for metastatic disease to death or last follow-up. Tumor bearing areas were evaluated by a single pathologist, and genomic DNA was extracted and analyzed for copy number alterations using Agilent 4x180k high resolution comparative genomic hybridization (CGH) arrays (average resolution of 13kb). Gain and Loss Analysis of DNA (GLAD) was used to segment the array output, and Genomic Identification of Significant Targets in Cancer (GISTIC) v2 was used to assess the significance of gains/losses (log base 2 ratio > 0.9 and < -1.3, respectively). False discovery rates < 0.2 were considered significant to correct for multiple comparisons. A Cox model for survival controlling for known prognostic variables (PS and visceral metastasis) was used for analysis. Bootstrapping (500 re-samples) assessed the variability and bias of parameters as internal validation. Results: In the 94 patients (46 deaths) median OS was 15 months. 15 patients were noted to have copy number gain (CNG) of a 150 kb region of 1q23.3. These patients had a median OS of 7.3 months, compared to patients without CNG with a median OS of 18 months. Gain of 1q23.3 is significantly associated with lethal UC (adjusted HR 2.94, 95% CI 1.35-6.44, p=0.007, FDR=0.12) (Table). This finding was internally validated (HR=3.2, 95% CI 1.25-7.2). Conclusions: Genomic gain of a short segment of chromosome 1q23.3 is associated with a lethal phenotype of UC in patients with metastatic disease. External validation in an additional cohort is underway. HR (95% CI) P value 1q23.3 gain (yes vs. no) 2.94 (1.35, 6.44) 0.007 ECOG PS (1,2 vs. 0) 1.93 (1.00, 3.79) 0.05 Visceral disease (yes vs. no) 2.40 (1.25, 4.47) 0.008