Assessing genotype−phenotype correlations in colorectal cancer with deep learning: a multicentre cohort study
Year: 2025
Type: article
Study design: Observational Study
Abstract: BACKGROUND: Deep learning-based models enable the prediction of molecular biomarkers from histopathology slides of colorectal cancer stained with haematoxylin and eosin; however, few studies have assessed prediction targets beyond microsatellite instability (MSI), BRAF, and KRAS systematically. We aimed to develop and validate a multi-target model based on deep learning for the simultaneous prediction of numerous genetic alterations and their associated phenotypes in colorectal cancer. METHODS: In this multicentre cohort study, tissue samples from patients with colorectal cancer were obtained by surgical resection and stained with haematoxylin and eosin. These samples were then digitised into whole-slide images and used to train and test a transformer-based deep learning algorithm for biomarker detection to simultaneously predict multiple genetic alterations and provide heatmap explanations. The primary dataset comprised 1376 patients from five cohorts who underwent comprehensive panel sequencing, with an additional 536 patients from two public datasets for validation. We compared the model's performance against conventional single-target models and examined the co-occurrence of alterations and shared morphology. FINDINGS: The multi-target model was able to predict numerous biomarkers from pathology slides, matching and partly exceeding single-target transformers. In the primary external validation cohorts, mean area under the receiver operating characteristic curve (AUROC) for the multi-target transformer was 0·78 (SD 0·01) for BRAF, 0·88 (0·01) for hypermutation, 0·93 (0·01) for MSI, and 0·86 (0·01) for RNF43; predictive performance was consistent across metrics and supported by co-occurrence analyses. However, biomarkers with high AUROCs largely correlated with MSI, with model predictions depending considerably on morphology associated with MSI at pathological examination. INTERPRETATION: By use of morphology associated with MSI and more subtle biomarker-specific patterns within a shared phenotype, the multi-target transformers efficiently predicted biomarker status for diverse genetic alterations in colorectal cancer from slides stained with haematoxylin and eosin. These results highlight the importance of considering mutational co-occurrence and common morphology in biomarker research based on deep learning. Our validated and scalable model could support extension to other cancers and large, diverse cohorts, potentially facilitating cost-effective pre-screening and streamlined diagnostics in precision oncology. FUNDING: German Federal Ministry of Health, Max-Eder-Programme of German Cancer Aid, German Federal Ministry of Education and Research, German Academic Exchange Service, and the EU.
Source: The Lancet Digital Health
Authors: Marco Gustav, Marko van Treeck, Nic G Reitsam, Zunamys I Carrero, Chiara M L Loeffler, Asier Rabasco Meneghetti, Bruno Märkl, Lisa A Boardman, Amy J French, Ellen L Goode, Andrea Gsur, Stefanie Brezina, Marc J Gunter, Neil Murphy, Pia Hönscheid, Christian Sperling, Sebastian Foersch, Robert Steinfelder, Tabitha Harrison, Ulrike Peters, Amanda Phipps, Jakob Nikolas Kather
Institutions: University Hospital Carl Gustav Carus, Else Kröner Fresenius Center for Digital Health, Technische Universität Dresden, University of Augsburg, Mayo Clinic, Medical University of Vienna, Centre international de recherche sur le cancer, Imperial College London, German Cancer Research Center, Nationales Centrum für Tumorerkrankungen Dresden, University Medical Center of the Johannes Gutenberg University Mainz, Fred Hutch Cancer Center, University of Washington, University of Leeds, Heidelberg University, University Hospital Heidelberg, National Center for Tumor Diseases
Language: English
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $6,680
Funders: U.S. Department of Health and Human Services, World Health Organization, Connecticut Department of Energy and Environmental Protection, Compagnia di San Paolo, Kræftens Bekæmpelse, Deutsches Krebsforschungszentrum, Centre International de Recherche sur le Cancer, University of Iowa, “la Caixa” Foundation, National Research Council, Cancer Research UK, National Institute for Health and Care Research, Department of Health and Social Care, Imperial College London, European Cooperation in Science and Technology, Deutscher Akademischer Austauschdienst, Deutsche Forschungsgemeinschaft, Institut National de la Santé et de la Recherche Médicale, Bundesministerium für Bildung und Forschung, Austrian Science Fund, Cancerfonden, Ministerie van Volksgezondheid, Welzijn en Sport, Bundesministerium für Gesundheit, Herzfelder'sche Familienstiftung, Ligue Contre le Cancer, Vetenskapsrådet, Österreichische Forschungsförderungsgesellschaft, Associazione Italiana per la Ricerca sul Cancro, Deutsche Krebshilfe, Institut Gustave-Roussy, Mutuelle Générale de l'Education Nationale, Medizinische Universität Graz, Junta de Andalucía, Gemeinsame Bundesausschuss, National Institutes of Health, Rollins School of Public Health, HORIZON EUROPE Framework Programme, Medical Research Council, World Cancer Research Fund, Instituto de Salud Carlos III, NIHR Imperial Biomedical Research Centre, Cancer Research UK Cambridge Institute, University of Cambridge, National Heart, Lung, and Blood Institute, National Cancer Institute, Office of Research Infrastructure Programs, National Institutes of Health, National Health Service Corps
Awards: G1041994376, G1075876736, G1211708387, ERA-NET Transcan - Tangerine - Künstliche Intelligenz-basierte End-to-End-Vorhersage des Ansprechens auf Krebsimmuntherapie, G1569319872, Verbund SWAG - SchWArmlernen und Generative Modelle zur Synthese und Nutzbarmachung hochqualitativer Daten in der Krebsmedizin - Standort Universitätsklinikum Würzburg, CAMINO - Programm zur Karriereentwicklung für forschende Ärztinnen und Ärzte in der multidimensionalen Tumorforschung, G2820917408, G2942021335, Translational Research Program in Colorectal Cancer Disparities, G3328679326, G3717981967, G3827278403, G3848825097, G4487933309, G4499735890, Health of vegetarians, G4681072085, G4712284355, G4881028819, G5220992053, G5514419823, G5788595627, G6037373127, G6894818919, Verbund SWAG - SchWArmlernen und Generative Modelle zur Synthese und Nutzbarmachung hochqualitativer Daten in der Krebsmedizin - Standort Universität Mainz, G7821233859, Verbund PEARL - Prävention von Darmkrebs im jungen und mittleren Erwachsenenalter - Standort Dresden, CompLS - Runde 5 - Verbundprojekt: TRANSFORM LIVER - Weiterentwicklung von Vision Transformern zur Entdeckung von Biomarkern bei Lebererkrankungen - Teilprojekt A, Molecular pathological epidemiology of colorectal cancer, G8698948422, Entitätsübergreifende, morphologiebasierte Vorhersage des Therapieansprechens auf Immuncheckpoint Inhibitoren mittels Deep Learning, Mayo Clinic Comprehensive Cancer Center Support Grant, Cancer Center Support Grant, G942809861
OpenAlex ID: W4413325002