Differences between local and review urinary cytology in diagnosis of bladder cancer. An interobserver multicenter analysis.

Objectives: The objective of the study is to evaluate the agreement of local and review urinary cytology in patients with newly diagnosed bladder cancer and in those being followed for their disease. In addition, the effect of the type of institution on agreement was determined.

Methods: A total of 652 consecutive patients with bladder cancer from 19 institutions were evaluated; 575 (88.2%) of the patients had cytopathological sample available for central review and were eligible for analysis. One hundred and twenty nine (22.4%) of the patients had newly diagnosed bladder cancer, whereas the remaining 446 (77.6%) patients were under follow-up. A voided urine sample was obtained prior to transurethral resection of the bladder (TURB) or routine follow-up cystoscopy and split for culture and cytology. The cytopathological samples were first analysed by a local pathologist, and then re-analysed by a central reviewer. The agreement of cytological readings was determined by Kappa coefficient.

Results: The sensitivities of local and review cytology in detection of primary bladder cancer were 38.8 and 31.0%, respectively. Recurrence was observed in 119 of the 446 (26.7%) patients under follow-up, of which both local and review cytology detected 21 (17.6%) cases. Specificities of local and review cytology were 97.6 and 96.6%, respectively. Overall agreement of urine cytology was good in patients with primary bladder cancer and moderate in those being followed for their disease as Kappa coefficients were 0.70 and 0.60, respectively. However, some disagreement was found when results were analysed according to type of institution, to type of primary tumour, and to result of follow-up cystoscopy. In patients with primary bladder cancer the Kappa coefficient was 0.86 (very good) in university hospitals and 0.36 (fair) in city hospitals. Accordingly, in patients under follow-up the Kappa coefficient was 0.65 (good) in university hospitals and 0.39 (fair) in district hospitals. Although the stage of primary tumour had no effect on agreement, agreement was moderate (Kappa coefficient 0.45) in those with low grade tumour and good (Kappa coefficient 0.67) in those with high grade tumour. In addition, Kappa coefficients were 0.65 (good) and 0.40 (fair) in those with and without recurrence at follow-up cystoscopy.

Conclusions: Although overall agreement of routine cytology was from moderate to good in both diagnosis and monitoring of bladder cancer, there is some variation in agreement according to the type of institution. Accordingly, grade of primary tumour and the result of follow-up cystoscopy had effect on agreement reflecting subjectiveness and weak reproducibility of this test. This not only emphasises the need for continuing education and quality control for urine cytologic analysis, but also inspires the development of more objective tests.