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
Year: 2026
Type: article
Study design: Observational Study
Abstract: BACKGROUND: Tumor microenvironment (TME) including stromal composition and antitumor immune responses, plays an important role in cancer progression. Tumor-stroma ratio (TSR) and tertiary lymphoid structures (TLS) both shows prognostic value, but their combined significance across urothelial carcinoma (UC) remains unclear. METHODS: We retrospectively analyzed H&E-stained whole-slide images from 884 UC patients from five cohorts (FAHZU, n = 76; Emory, n = 94; TCGA, n = 291; QDPH, n = 330; TRRC, n = 93), encompassing both localized and advanced disease settings. TLS were identified using deep learning-based nuclei classification followed by identifying clusters of aggregated immune cells, while TSR was quantified using automated stromal segmentation. An integrated TLS-TSR risk score was trained using a Cox proportional hazards model in the FAHZU cohort and externally validated in four independent test cohorts. In the TRRC cohort, associations between these biomarkers and response to immune checkpoint inhibitor (ICI) were assessed. FINDINGS: Automated TLS density and TSR individually showed good concordance with pathologist assessments (TLS: ICC=0.852; TSR: Spearman ρ=0.826; both p < 0.001). High TLS density and low TSR were each associated with improved progression-free survival (PFS) across cohorts. The TLS-TSR score outperformed either biomarker alone (C-index: 0.65-0.69) and remained an independent predictor of PFS in multivariable analysis. TLS density showed strong association with ICI response (AUC 0.745, 95% CI 0.607-0.864). Transcriptomic analysis further revealed enrichment of extracellular matrix organization and stromal-related pathways in the high-risk group, accompanied by selective alterations in immune cell composition, indicating coordinated remodeling of both stromal and immune components of TME. INTERPRETATIONS: Integrating stromal composition and immune-related morphology may improve prognostic stratification across UC patients and identify histopathologic features associated with response to ICI.
Source: European Journal of Cancer
Institutions: Georgia Institute of Technology, Emory University, The Wallace H. Coulter Department of Biomedical Engineering, Zhongda Hospital Southeast University, Southeast University, Mansoura University, Atlanta VA Medical Center, Nanjing University of Information Science and Technology, Augusta University, First Affiliated Hospital Zhejiang University, Zhejiang University, Winship Cancer Institute, Cleveland Clinic, Guangdong Academy of Medical Sciences, Guangdong Provincial People's Hospital, Southern Medical University, Affiliated Hospital of Qingdao University
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $4,270
Funders: Institute for Technology in Health Care, U.S. Department of Veterans Affairs, Breast Cancer Research Foundation, Prevent Cancer Foundation, Bristol-Myers Squibb, AstraZeneca, Office of Research and Development, Biomedical Laboratory Research and Development, VA Office of Research and Development, Scott R. MacKenzie Foundation, Fund for Innovation in Cancer Informatics, National Institutes of Health, Advanced Research Projects Agency, Winship Cancer Institute, National Institute on Aging, National Heart, Lung, and Blood Institute, National Cancer Institute, National Institute of Allergy and Infectious Diseases, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Biomedical Imaging and Bioengineering, U.S. National Library of Medicine
Awards: G1046651757, G1365000891, Multimodal, Multi-scale Framework for Ethical AI Model Development, G1909130050, G2711701398, G3371128657, RADIOMIC APPROACHES TO IMPROVE TARGETING FOR ATRIAL FIBRILLATION CATHETER ABLATION, G6079499416, G6142425281, G7105568102, Computer-Assisted Histologic Evaluation of Cardiac Allograft Rejection, G8861662720, G8958113727, Histotools: scaling digital pathology curation tools for quality control, annotation, labeling, and dataset identification, Winship Cancer Institute Cancer Center Support Grant, Novel Radiomics for Predicting Response to Immunotherapy for Lung Cancer, Multi-Consortia Coordinating Center (MC2 Center) for Cancer Biology: Building Interdisciplinary Scientific Communities, Coordinating Impactful Resource Sharing, and Advancing Cancer Research, Oral Cavity Quantitative Histomorphometric Risk Classifier (OHbIC) in Oral Cavity Squamous Cell Carcinoma (OC-SCC), Quantitative imaging phenotypic classifier for distinguishing radiation effects from tumor recurrence in Glioblastoma, Prostate Cancer Biomarker and Imaging Validation Alliance: Emory University, University of Alabama Birmingham, and University of Texas Southwestern, Imaging signatures of early tuberculosis, P-CARRS-BRAIN: Multi-domain (genetic, socio-behavioral, vascular) risk factors and prediction of Alzheimer’s Disease continuum in South Asians in India, Prognostic and Predictive Digital Tissue Image Assay for Prostate Cancer, An AI-enabled Digital Pathology Platform for Multi-Cancer Diagnosis, Prognosis and Prediction of Therapeutic Benefit
OpenAlex ID: W7201866518