Evaluation of multidetector computed tomography urography and ultrasonography for diagnosing bladder cancer

To evaluate and compare the diagnostic accuracy of multidetector computed tomography urography (CTU) and ultrasonography (US) for diagnosing bladder cancer.

A consecutive series of 143 patients over 40-years of age, presenting with macroscopic haematuria and without urinary tract infection underwent same-day CTU, US, and flexible cystoscopy. CTU and US were independently rated on a five-point scale for the presence of bladder cancer without knowledge of the reference standard of flexible or rigid cystoscopy and/or biopsy results. Diagnostic accuracy was assessed by receiver operating characteristic (ROC) analysis and likelihood ratios.

For CTU, a rating of 5 (definitely tumour) was highly specific for bladder cancer (96.5%, 95%CI: 91.3-99%), effectively confirming diagnosis (positive likelihood ratio 25.6, 95%CI: 9.7-67.4). For US, specificity was also high (94.7%, 95%CI: 88.9-98%) with a positive likelihood ratio of 13.1 (95%CI: 5.8-29.6). Sensitivity at this rating was substantially higher for CTU (89.7%, 95%CI: 72.7-97.8%) than US (69%, 95%CI: 49.2-84.7%). Standardized partial area (Az) under the ROC curve between 95-100% specificity, representing the average sensitivity in this range, was significantly greater (0.88 versus 0.61, p<0.05) for CTU than US.

The specificities of CTU and US for the diagnosis of bladder cancer were similar, but CTU was more sensitive. Although the sensitivity of CTU was not high enough to replace flexible cystoscopy in the diagnostic pathway, the high specificity enables direct referral to rigid cystoscopy, bypassing flexible cystoscopy and expediting diagnosis and treatment in those patients testing positive.