AUA 2017: Development and external validation of a novel 12-gene signature on muscle-invasive bladder cancer for prediction of overall survival: linkage to data from The Cancer Genome Atlas

Boston, MA (UroToday.com) Molecular gene-expression profiling of bladder cancer by several groups (UNC, MDA, TCGA, Lund) has led to a better understanding that histologically similar bladder cancer can have distinct subgroups of biological behavior. These subgroups have been found to have different prognoses at baseline and even different responses to cisplatin-based neoadjuvant chemotherapy. However, the clinical utility of genome-wide expression profiling has called into question the applicability and generalizability in the clinic. Prior authors have attempted to pare down the number of genes that could be used to predict molecular subclass. Damrauer and Kim described a classifer using 47 genes called "BASE47" (PNAS 2014). Dadhania and Czerniak suggested that GATA3 and KRT5/6 immunohistochemistry could alone predict subclass with high accuracy (Ebiomedicine 2016). The authors used a reduced multivariate Cox regression model to pare down a classifier to 12 genes using the TCGA dataset.

In a validation dataset, their 12-gene classifier predicted overall survival with an AUC of 0.741 (compared to clinical data with an AUC of 0.667). Combining the clinical data and 12-gene classifier together increased the AUC to 0.768.
This relatively modest improvement in outcome prediction reminds us that gene-expression profiles in the tumors, while certainly important, are only one of many factors influencing patient outcomes.

Presented by: Fangning Wan, Shanghai China

Contributed by: Jed Ferguson, MD/PhD and Ashish Kamat, MD. MD Anderson Cancer Center, Department of Urology.

at the 2017 AUA Annual Meeting - May 12 - 16, 2017 – Boston, Massachusetts, USA

Login