RESUMO
OBJECTIVE: To determine the accuracy of voided urinary cytology (VUC) in predicting of non-muscle-invasive bladder cancer (NMIBC) risk stratification before surgery. METHODS: We prospectively collected data from all patients diagnosed with bladder cancer in our institution over 2 years. We have analyzed VUC accuracy of positive and suspicious VUC in the detection of high-risk tumors and negative and atypical VUC in the detection of low-risk tumors. To test this accuracy, we assessed sensitivity, specificity, positive (PPV) and negative predictive values (NPV), diagnostic odds ratio (DOR), and generated ROC curves (receiver operating characteristic curve). RESULTS: With 224 patients included, the positive VUC subcategory showed a specificity of 92.4% (95%CI: 83.2%-97.5%) and a PPV of 91.4 (95%CI: 81%-97.1%). DOR in this subgroup was 6.81. In the suspicious VUC, specificity was 90.9% (95%CI: 81.3%-96.6%), PPV was 88% (95%CI: 75.7%-95.5%) and DOR was 4.23. Combined analysis of positive and suspicious cytologies for detecting high-risk NMIBC showed a sensitivity of 65% (95%CI: 57.3%-73.2%) and a DOR of 9.51. Negative VUC showed high specificity in detecting low-risk (93.2% [95%CI: 87.9%-96.7%]) and a DOR of 6.90 (95%CI: 3.07-15.46). Atypical VUC was the least accurate and had rather low specificity and predictive values. CONCLUSIONS: VUC appears to be a good, inexpensive and easily available method to determine risk stratification before surgery. This can be useful in daily practice to determine which patients should receive a single instillation of MMC and to prioritize patients more likely to have a high- risk tumor.