Research

Could Biomarkers and AI Reshape Bladder Cancer Surveillance?

Cystoscopy remains central to bladder cancer diagnosis and surveillance, but the need for repeated invasive procedures has driven an expanding search for reliable alternatives and complementary tools. Advances in urinary biomarkers, liquid biopsy, molecular profiling, and artificial intelligence are beginning to offer new ways to detect disease, estimate recurrence risk, and monitor tumor evolution. A recent review suggests that combining these technologies may eventually move bladder cancer surveillance toward a more individualized approach.

Bladder cancer is particularly suited to noninvasive molecular monitoring because urothelial tumor cells are in direct contact with urine. Tumor-derived DNA, RNA, proteins, extracellular vesicles, and other molecular signals can therefore potentially be detected without obtaining tissue. Researchers are investigating these signals for several clinical applications, including early detection, recurrence surveillance, treatment-response assessment, and identification of minimal residual disease. Some assays evaluate individual mutations or proteins, while newer approaches combine DNA methylation, genomic alterations, RNA signatures, and other molecular information.

A growing number of urine-based tests illustrate how quickly the field is evolving. Molecular assays evaluating alterations such as FGFR3 and TERT, DNA methylation signatures, messenger RNA, and protein biomarkers have demonstrated encouraging diagnostic performance in published studies. The review also describes several urinary assays already cleared for clinical use, alongside newer platforms being investigated for NMIBC surveillance. However, performance varies considerably by assay and clinical setting, and these technologies have not eliminated the need for cystoscopic evaluation.

Artificial intelligence could add another layer by analyzing information that would be difficult to interpret using conventional approaches alone. Machine-learning models can integrate imaging, pathology, molecular profiles, and clinical characteristics to identify patterns associated with diagnosis, recurrence, progression, or treatment response. AI-assisted cystoscopy is another emerging application. One system trained using more than 69,000 cystoscopic images achieved 93.9% accuracy and 95.4% sensitivity for lesion detection, while other models have demonstrated similarly encouraging results. AI-assisted urine cytology is also being investigated as a potential way to identify patients who may safely require fewer invasive evaluations.

The challenge now is determining how these technologies should fit into actual clinical care. Differences in specimen collection, assay methodology, patient populations, algorithms, and validation strategies make it difficult to translate promising performance into universal practice. Large prospective multicenter studies will be needed to establish whether biomarker- and AI-assisted approaches can safely change surveillance schedules or treatment decisions. For urologists, the emerging opportunity may therefore be less about replacing cystoscopy and more about combining molecular and computational information with established clinical tools to determine which patients require closer evaluation and which may eventually be candidates for less invasive monitoring.