diction of Stage, Grade, and Survival in Bladder Cancer

Download pose: To evaluate performances of published gene signatures for the assessment of urothelial oma. erimental Design: We evaluated 28 published gene signatures designed for diagnostic and progpurposes of urothelial cancer. The investigated signatures include eight signatures for stage, five for four for progression, and six for survival. We used two algorithms for classification, nearest centroid ication and support vector machine, and Cox regression to evaluate signature performance in four endent data sets. ults: The overlap of genes among the signatures was low, ranging from 11% among stage signatures % among survival signatures. The published signatures predicted muscle-invasive and high-grade rs with accuracies in the range of 70% to 90%. The performance for a given signature varied erably with the validation data set used, and interestingly, some of the best performing signatures not designed for the tested classification problem. In addition, several nonbladder-derived gene ures performed equally well. Large randomly selected gene signatures performed better than the shed signatures, and by systematically increasing signature size, we show that signatures with genes are needed to obtain robust performance in independent validation data sets. None of the hed survival signatures performed better than random assignments when applied to independent tion data. clusion: We conclude that gene expression signatures with >150 genes predict muscle-invasive Con growth and high-grade tumors with robust accuracies. Special considerations have to be taken when designing gene signatures for outcome in bladder cancer. Clin Cancer Res; 16(17); 4421–33. ©2010 AACR.