RESUMO
BACKGROUND: UroScreen is a prospective study for early diagnosis of bladder cancer (BC) in chemical workers formerly exposed to aromatic amines, aimed to assess the performance of molecular tumor markers in comparison with urinary cytology. Here we evaluate the cancer-predictive values and potential effect modifiers of fluorescence-in-situ-hybridization (FISH). SUBJECTS AND METHODS: A FISH test was performed in 7,091 urine samples from 1,609 subjects between 2007 and 2010. Cystoscopy was recommended in case of positive or suspicious findings. Logistic regression models were applied to estimate the influence of potential test confounders like urinary creatinine and hematuria on detecting BC. Receiver operating characteristic (ROC) curves for FISH were adjusted for test confounders. Cancer-predictive values were calculated from test results in the last sample before diagnosis. RESULTS: Histopathology revealed 16 incidental BCs and 5 recurrent tumors in 20 study participants. FISH was positive in 9 BC cases of which 7 were high grade. Cytology detected 8 tumors. FISH overlapped with cytology in 7 cases. Sensitivity was 45.0% and PPV (positive predictive value) was 16.4% in all and 53.85% and 13.21% in high-grade tumors. Specificity and negative predictive value (NPV) were 96.97% and 99.26% in all bladder tumors. BC detected during UroScreen was associated with an odds ratio (OR) of 6.88 (95% CI 1.72-27.44) for positive FISH and with an OR of 8.81 (95% CI 1.41-54.96) for gross hematuria. The adjusted area under the curve was 0.77 (95% CI 0.62-0.92) for all and for high-grade lesions (0.85; 95% CI 0.69-1.00). CONCLUSIONS: FISH showed a performance in detecting bladder cancer comparable to cytology but a larger number or false-positive results. It remains to be investigated if chromosomal instability can be detected earlier than morphologic changes of exfoliated bladder cancer cells.