Correspondence of cystoscopy with pathology findings in the evaluation of bladder cancer

A systematic review and meta-analysis

Abstract

Introduction: The diagnosis and treatment non-muscle-invasive bladder cancer relies on cystoscopy to detect bladder tumors. Few studies, even less in the modern era of cystoscopes, have assessed their diagnostic performance. We aimed to evaluate the literature on cystoscopy as a diagnostic test for pathology-confirmed bladder cancer. We further aimed to ascertain the utility of recorded images and video at cystoscopy to correctly predict the pathologic diagnosis among patients investigated for bladder cancer.

Methods: This systematic review followed the PRISMA guideline and was registered in PROSPERO. We reviewed MEDLINE/Pubmed, Cochrane, and EMBASE databases from January 1995 to December 2024 for English- or French-language studies to address the research question: what is the utility of recorded images and video at cystoscopy to correctly predict the pathologic diagnosis among patients investigated for bladder cancer? Study selection was performed in duplicate, with consensus from a third reviewer to resolve differences, and the QUADAS-2 scale was used to assess the risk of bias in the study. After selecting the studies, the following variables were extracted: study characteristics, patient demographics, and diagnostic outcomes. The results were summarized in an Excel sheet and analyzed through a meta-analysis performed using RevMan software, version 5.4.

Results: Of 1165 extracted studies, we included 13 studies that reported on the correspondence of cystoscopy findings with pathology findings. Among studies with consecutive cystoscopy cohorts, six evaluated patients under investigation for bladder cancer, and five evaluated both patients with prior bladder cancer and those under investigation for bladder cancer. The sensitivity and specificity of white light cystoscopy from studies with per-lesion comparison to pathology are, respectively, 83.1% (95% confidence interval [CI] 71.9−90.5%) and 79.9% (95% CI 62.4−90.5%). Qualitatively, diagnostic results were lower for flat lesions compared to papillary lesions, with the accuracy of grade prediction variable among studies assessing this. Five studies reported data evaluating conserved photos or videos, with no significant differences in reported diagnostic test performance. Overall, most studies were at low risk of bias in the areas assessed. Nonetheless, limitations include the heterogeneity of clinical contexts and the detail of available information, with many studies indirectly related to our study question.

Conclusions: While relatively few studies directly address this topic, the current literature indicates that cystoscopy has high diagnostic value to detect pathology-confirmed bladder cancer; however, it may lead to both overdiagnosis or overtreatment of non-aggressive lesions and miss more aggressive ones. Lesion features may help indicate the level of diagnostic accuracy, while current data does not support the reliable diagnosis of grade at cystoscopy. Our findings support the use of recorded evaluations in clinical practice and for future research, which could complement real-time assessments.