Advancements in AI Transform Risk Assessment for Bladder Cancer

Dr. Roger Li, a genitourinary oncologist at Moffitt Cancer Center, recently addressed the risk of disease progression in patients with low-grade non–muscle-invasive bladder cancer (NMIBC). His insights focused on the emerging role of artificial intelligence (AI) in refining risk stratification for these patients. While low-grade NMIBC is often associated with positive oncological outcomes, Dr. Li highlighted that progression can occur along a spectrum, leading to various clinical implications.

Progression from low-grade NMIBC to high-grade, muscle-invasive, or metastatic bladder cancer is uncommon, typically seen in fewer than 5% of patients. In larger patient cohorts, Dr. Li noted, only a handful may experience a shift to muscle-invasive disease. Nevertheless, he emphasized that clinicians should not overlook the more prevalent transition from low-grade to high-grade disease, which presents significant management challenges.

Dr. Li estimates that between 10% to 20% of patients with low-grade NMIBC may advance to high-grade disease. He pointed out that variability in grading, often due to subjective interpretations among pathologists, complicates risk assessments. To address these discrepancies, there is an increasing interest in AI-driven pathology tools.

AI models trained on digitized hematoxylin and eosin (H&E) slides represent a practical solution. These tools utilize standard pathology images already prevalent in clinical settings, eliminating the need for specialized sequencing platforms often required for genomic assays. Dr. Li explained that AI can analyze nuclear and cellular features on a scale that surpasses human capability. By examining thousands of morphologic parameters, AI can identify patterns linked to clinically important outcomes, such as the risk of progression to high-grade disease.

The granular analysis provided by AI could enhance prognostication and facilitate earlier identification of risk. If validated through prospective studies, AI-assisted pathology could lead to more personalized surveillance and treatment strategies for patients with low-grade NMIBC. Those identified with higher-risk morphologic signatures might benefit from closer monitoring or earlier therapeutic interventions. In contrast, patients with lower-risk profiles could be spared unnecessary procedures or overtreatment.

Dr. Li concluded that AI-enabled pathology has the potential to significantly improve clinical decision-making. By offering objective, reproducible, and widely accessible risk stratification tools, this technology could enhance outcomes for patients with NMIBC, ultimately evolving the standard of care in this domain.