Diagnostic, Staging, and Grading of Urothelial Carcinomas from Urine: Performance of BCA-1, a Mini-Array Comparative Genomic Hybridisation–Based Test

Abstract

BackgroundCytogenetic abnormalities occur at an early stage of bladder urothelial carcinomas (BUC), and their frequency increases as the cancer becomes more advanced. ObjectiveTo assess the diagnostic performance of a test based on cytogenetic abnormalities to diagnose, stage, and grade BUC from the urine. Design, setting, and participantsWe used a 341 bacterial artificial chromosome (BAC) comparative genomic hybridisation (CGH)–array chip (BCA-1) designed to include loci affected in BUC. The chip was first used on 32 frozen BUC biopsies to design staging (BN0) and grading (BN1 and BN2) prediction models based on Bayesian networks analysis. The models were then validated on external data obtained from 98 tumour samples using a 2464 BAC CGH-array chip. The performance of the test was finally assessed on 44 urine pellets collected, including 22 patients who had BUC and 22 controls. MeasurementsWe measured sensitivity and specificity to diagnose BUC stage and grade from urine pellets. Results and limitationsIn the urine, BCA-1 test sensitivity was 95%, specificity was 86%, and accuracy was 91%. The BN0 staging model identified T1–4 tumours in the urine with a sensitivity of 90%, a specificity of 83%, and an accuracy of 87%. The BN1 and BN2 grading models detected high-grade disease with a sensitivity, specificity, and accuracy of 86%, 88%, and 87%, respectively, using BN1 and 100%, 63%, and 82%, respectively, using BN2. BN models performed with similar sensitivity but reduced specificity using the external data. BCA-1 failed to produce results for eight additional samples (failure rate: 9%). The test needed high quantities and quality of DNA, and external validation in larger, prospective, and better-designed studies is necessary to confirm feasibility and performance. ConclusionsThe BCA-1 mini–CGH-array chip detected BUC in urine with a high diagnostic performance. It could also accurately discriminate low-grade from high-grade tumours and, to a lesser extent, lamina propria–invasive tumours from pTa tumours.