Gene expression test for the non-invasive diagnosis of bladder cancer: A prospective, blinded, international and multicenter validation study.

ABSTRACT

OBJECTIVE: This study aimed to validate, in a prospective, blinded, international and multicenter cohort, our previously reported four non-invasive tests for bladder cancer (BC) diagnosis based on the gene expression patterns of urine. METHODS: Consecutive voided urine samples from BC patients and controls were prospectively collected in five European centres (n=789). Finally, 525 samples were successfully analysed. Gene expression values were quantified using TaqMan Arrays and previously reported diagnostic algorithms were applied to gene expression data. Results from the most accurate gene signature for BC diagnosis were associated with clinical parameters using analysis of variance test. RESULTS: High diagnostic accuracy for the four gene signatures was found in the independent validation set (area under curve [AUC]=0.903-0.918), with the signature composed of two genes (GS_D2) having the best performance (sensitivity 81.48%; specificity 91.26%; AUC 0.918). The diagnostic accuracy of GS_D2 was not affected by the number of tumours (p=0.58) but was statistically associated with tumour size (p=0.008). Also, GS_D2 diagnostic accuracy increases with increasing BC tumour risk. We found no differences in the performance of the GS_D2 test among the populations and centres in detecting tumours (p=0.7) and controls (p=0.2). CONCLUSIONS: Our GS_D2 test is non-invasive, non-observer dependent and non-labour-intensive, and has demonstrated diagnostic accuracy in an independent, international and multicenter study, equal or superior to the current gold standard (cystoscopy combined with cytology). Additionally, it has higher sensitivity than cytology while maintaining its specificity. Consequently, it meets the requirements for consideration as a molecular test applicable to clinical practice in the management of BC.