Abstract
BACKGROUND: Heterogeneity of muscle-invasive bladder cancer (MIBC) has been characterized using whole-genome mRNA expression data, showing distinct molecular and clinicopathological characteristics by subtypes. However, associations between risk factors and molecular subtypes have not been reported. METHODS: Four previously published schemes were used to categorize molecular subtypes in 372 MIBC patients from the Cancer Genome Atlas (TCGA). Data on gene expression (RNA-seq), demographic, and clinicopathological characteristics were retrieved through TCGA data portal. Polytomous logistic regression was used to estimate the associations of subtypes by different schemes with age at diagnosis, obesity, and smoking. RESULTS: While some quantitative variation was evident, distinct molecular subtype schemes showed considerable consistency in the association with the risk factors. Generally, compared to patients with luminal-like tumors, patients with basal-like subtypes were more likely to be older (OR(75 + yrs vs. <60 years) range = 1.32-2.89), obese (OR(obese vs. normal) range = 1.30-3.05), and to start smoking at early age (OR(<18 years vs. 25+ years) range = 1.11-4.57). CONCLUSIONS: Different molecular subtypes of MIBC may have distinct risk profiles. Large population-based studies with detailed information on bladder cancer risk factors are needed to further define etiologic heterogeneity for bladder cancer.