Development and Validation of an Artificial Intelligence‐Driven Multigene Assay Decision‐Support System for Personalized Management of T1 High‐Grade Bladder Cancer
Year: 2026
Type: article
Abstract: T1 high‐grade bladder cancer presents a therapeutic dilemma, as current risk stratification fails to guide the choice between early cystectomy and bladder‐preserving therapy. In our recent work, we defined a high‐risk multiomic subtype driven by anoikis resistance and immune evasion and developed the T1HG‐UCBguider framework for risk stratification; however, its reliance on multiomic sequencing limits clinical implementation. Here, we translate this framework into Guider, an artificial intelligence‐driven reverse transcription‐quantitative polymerase chain reaction (RT‐qPCR) decision‐aid designed for cost‐conscious use in routine pathology workflows and integrating a 30‐gene TaqMan assay with clinicopathological variables to predict progression and recurrence, with exploratory treatment‐context modules for no instillation, Bacillus Calmette–Guérin, and intravesical chemotherapy. In a multicenter ambispective cohort of 1198 patients, Guider achieved > 10% higher absolute predictive performance than guideline‐based risk groups and the published 12‐gene progression score within this Chinese T1 high‐grade cohort. Reclassification identified 82.5% of guideline‐defined high‐risk patients who progressed and 45.1% of very‐high‐risk nonprogressors who might be considered for bladder preservation. Guider could potentially support treatment de‐escalation in 29.7% and escalation in 12.9% of patients. These findings establish a scalable translational framework bridging multiomic discovery and clinical decision support, warranting prospective interventional validation in geographically, ethnically, and demographically diverse populations.
Source: MedComm
DOI: 10.1002/mco2.71155
Language: English
Subfield: Surgery
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $3,500
OpenAlex ID: W7237364325
