Detection and resection of carcinoma in situ of the bladder: Implications for clinical trial design

•carcinoma in situ (CIS) is under-detected by routine endoscopic evaluation of the bladder.•CIS may be surgically eradicated in a proportion of patients.•Single arm trials may over-estimate complete response rate since some responses are likely achieved by complete resection of CIS and others are likely due to under-detection of persistent CIS.•Random biopsy and use of fluorescent cystoscopy will improve CIS detection in clinical trials.•Randomized clinical trial design will overcome these limitations. The path to approval of novel therapeutics for patients with BCG-unresponsive non-muscle invasive bladder cancer (NMIBC) requires demonstration of efficacy in eradicating carcinoma in situ (CIS), as determined by cytology, white light cystoscopy and only sometimes mandatory re-biopsy. This paradigm is based on the premise that CIS, in contrast to papillary tumors, cannot be completely resected. We aimed to determine the accuracy of CIS by standard means and the rate at which CIS may be eradicated by transurethral bladder tumor resection (TURBT). We performed a retrospective analysis of consecutive patients who underwent radical cystectomy (RC) for high risk NMIBC or muscle invasive bladder cancer (MIBC) between 2005 and 2019 in a tertiary academic center. The concordance in the presence of CIS in matched TURBT and RC samples was calculated. Complete pathologic information was available for 816 patients with urothelial carcinoma. CIS was detected at TURBT in 354 (43.4%) patients (64.0% NMIBC, 32.3% MIBC) and at RC in 436 (53.4%) patients (64.7% NMIBC, 47.4% MIBC). CIS was missed by TURBT in 199 (45.6%) of those cases (NMIBC 25.4%, MIBC 60.6%). CIS detected on TURBT was not found in the RC specimen in 33.1% (117/354) of cases. Lack of prospective bladder mapping and central pathology review are limitations. Our results suggest that TURBT is inaccurate in detecting CIS. The absence of CIS in the RC specimen after detection in the matched TURBT specimen suggests that CIS may be completely resected by TURBT in a proportion of patients. These factors need to be considered in the design of clinical trials in patients with NMIBC. The use of random biopsies or enhanced cystoscopy could improve the accuracy of CIS detection, but the former is associated with patient morbidity and randomization would alleviate concern about these variables impacting clinical trial outcomes.