Purpose: To test the hypothesis that voided urinary levels of nuclear matrix protein 22 (NMP22) would add to the predictive ability of urine cytology in the diagnosis, staging and grading of bladder transitional cell carcinoma (TCC), and to evaluate the diagnostic performance of different NMP22 cut-points.
Materials: NMP22 level and barbotage cytology were evaluated in voided urine specimens collected before cystoscopy from 302 subjects with a history of TCC, 32 subjects with benign urologic pathologies, and 10 healthy volunteers.
Results: 180 patients (52%) had bladder TCC. Higher levels of NMP22 and positive cytology were independently associated with an increased risk of TCC, invasive stage, and high grade. The NMP22 value with equal sensitivity and specificity for prediction of bladder cancer was 6.5U/ml; for prediction of grade 3 TCC it was 13.5U/ml; and for prediction of invasive tumor stage it was 17.4U/ml. The NMP22 cut-point of 6.5U/ml outperformed the 10U/ml cut-point in all pathologic stages and grades. The diagnostic sensitivity of the cytology and NMP22 combined was superior across all pathologic stages and grades to that of either marker alone. NMP22 and cytology stratified patients into groups with significantly different risk for TCC presence, invasive stage, and high grade.
Conclusions: 6.5U/ml is a robust NMP22 cut-point for bladder cancer surveillance. The diagnostic sensitivities of the combined NMP22 and cytology for TCC presence, stage, and grade were significantly higher than those of single marker alone. The combination of urine cytology and NMP22 could be used to tailor the frequency of cystoscopic follow up.