Institutional variability in the accuracy of urinary cytology for predicting recurrence of transitional cell carcinoma of the bladder

OBJECTIVE: To assess the contemporary inter-institutional accuracy of urinary cytology in predicting the recurrence of transitional cell carcinoma (TCC) of the bladder, in a large multi-institutional cohort from four continents, as cystoscopy and urinary cytology represent the 'gold standards' for surveillance of TCC recurrences, but the ability of cytology to predict recurrence varies. PATIENTS AND METHODS: Ten institutions contributed 2542 patients with a history of superficial TCC, of whom 898 had TCC recurrence. Age- and gender-adjusted logistic regression models were used to evaluate the association between urine cytology and TCC recurrence. The predictive accuracy derived from the logistic regression model was tested using the area under the receiver operating characteristic curve. The resulting predictive accuracy estimates were internally validated with 200 bootstrap re-samples. RESULTS: The mean (range across institutions) age of the patients was 65 (48-69) years and 75 (67-87)% were men. Cytology was positive in 19 (10-38)% of patients; recurrence was identified in 35 (27-54)% of patients. The sensitivity was 38-65% across institutions. Urinary cytology varied significantly in its ability to predict recurrence of bladder cancer. Institution-specific predictive accuracy adjusted for gender and age was 0.627-0.893. Stratifying by grade and stage only partly attenuated the discrepancies between centres. CONCLUSIONS: The variability of urinary cytology results was very appreciable among the 10 centres and ranged from poor (63%) to excellent (89%).