Pulse.

a daily field guide to health research that matters

◆ Console

‹ Thu · 13 Aug 2026
Near-term implementable finding

A stromal-immune computational pathology signature for prognosis and immune checkpoint inhibitor response in localized, locally advanced and metastatic urothelial carcinoma: A multicenter retrospective study.

A deep-learning biomarker combining immune and stromal features better predicts immunotherapy response in bladder cancer than existing tests, potentially guiding treatment selection.

This multicenter retrospective study developed and externally validated a combined TLS-TSR risk score in 884 UC patients across five cohorts (FAHZU, Emory, TCGA, QDPH, TRRC) using deep learning-based nuclei classification and automated stromal segmentation. The integrated score outperformed individual biomarkers (C-index 0.65–0.69), remained an independent PFS predictor in multivariable analysis, and was strongly associated with ICI response—establishing it as a clinically actionable computational biomarker for UC management.

What the study was

Study design
retrospective_cohort_multicenter
Population
Urothelial carcinoma patients across five independent cohorts
Sample size
884
Category
Diagnostics
Maturity
Validated
Journal
Eur J Cancer

Why it surfaced

Multicenter external validation across 5 cohorts in 884 patients is exceptional for computational pathology; ICI response prediction AUC 0.745 is clinically meaningful; directly bridges AI diagnostics and precision immunotherapy selection.

A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.