Multiscale image-molecular integration for interpretable immunotherapy biomarkers: from radiology to spatial omics

Cancer immunotherapy has changed the treatment landscape for many malignancies, yet reliable prediction of benefit, lack of benefit or resistance, and immune-related toxicity remains a major challenge. Current biomarkers, including PD-L1 expression, tumor mutational burden, microsatellite instability, immune-cell infiltration, and selected molecular signatures, provide useful information in specific settings but often fail to capture the full biological and clinical complexity that shapes response. There is a growing need for integrative approaches that connect information across complementary measurement scales, from whole-body imaging and whole-slide histopathology to multiplex tissue imaging, spatial omics, immune repertoire profiling, and molecular data.The Research Topic focuses on multiscale image-molecular integration for the development, interpretation, and validation of immunotherapy biomarkers. The collection aims to bring together studies that combine radiology, radiomics, digital pathology, pathomics, multiplex imaging, spatial transcriptomics, spatial proteomics, genomics, transcriptomics, proteomics, metabolomics, immune repertoire data, and clinically annotated information to support more accurate and biologically meaningful prediction of immunotherapy response. The emphasis is on methodological integration and translational value: how different data layers can be aligned, modeled, interpreted, and validated to generate biomarkers that are robust, reproducible, and relevant to cancer immunology and immunotherapy. A central goal is to move beyond single modality markers toward integrated biomarker frameworks that capture complementary information across scales. Radiological imaging can provide lesion-level and whole-body information on tumor burden, heterogeneity, and treatment dynamics. Whole-slide histopathology can capture tissue morphology, immune infiltration, stromal patterns, and spatial organization. Multiplex imaging and spatial omics can provide higher resolution information on cell states, cellular neighborhoods, antigen presentation programs, immune cell localization, molecular gradients, and therapy-associated remodeling. Molecular profiling can add genomic, transcriptomic, proteomic, metabolic, and receptor-level information that helps explain the biological basis of response or non-response. When integrated carefully, these data sources can support the discovery of interpretable immunotherapy biomarkers and improve the biological grounding of predictive models.The Research Topic places strong emphasis on validation, interpretability, reproducibility, and responsible translation. Computational and AI-based studies should not stop at model performance alone; submissions should aim to explain which biological or image-derived features drive predictions, how findings generalize across cohorts or platforms, and whether the proposed biomarkers add value beyond existing clinical or molecular markers. Studies based mainly on public datasets should include appropriate validation, such as e.g. independent cohorts, external benchmarking, orthogonal assays, spatial confirmation, cross-platform testing, functional support, or clinically meaningful immunotherapy endpoints. Purely descriptive analyses without validation or biological interpretation are outside the intended focus. We welcome the following article types: Brief Research Report, Hypothesis and Theory, Methods, Mini Review, Opinion, Original Research, Perspective, Review, Systematic Review, and Technology and Code. Potential themes include (but are not limited to):• Multiscale integration of radiology, whole-slide histopathology, multiplex imaging, spatial omics, molecular profiling, and clinically annotated data.• Radiomics, pathomics, and spatial omics approaches for immunotherapy biomarker discovery.• Interpretable AI models for response, non-response, and immune-related toxicity prediction.• Spatially resolved and image-linked biomarkers of immune activity, immune exclusion, antigen presentation, and therapy-associated remodeling.• Immune repertoire, neoantigen, genomic, transcriptomic, proteomic, metabolomic, and imaging-derived features in integrated biomarker models.• Benchmarking, external validation, model explain-ability, reproducibility, harmonization, and transparent reporting of computational immuno-oncology tools.• Translation of high dimensional image-molecular signatures into scalable, clinically interpretable biomarker assays.Overall, the Research Topic aims to advance methodologically rigorous and biologically grounded approaches for integrating imaging and molecular data across scales. By focusing on interpretability and validation, the collection seeks to support the development of immunotherapy biomarkers that are not only predictive, but also reproducible, explainable, and relevant for future translational cancer immunology research.Please note that manuscripts consisting solely of bioinformatics or computational analysis of public genomic or transcriptomic databases which are not accompanied by robust and relevant validation (clinical cohort or biological validation in vitro or in vivo) are out of scope for this Research Topic.Please note Dr Nektarios A Valous was granted a patent in the US (US11501444B2) Topic editors: Nektarios A. Valous, Meggy Suarez-Carmona Source: https://www.frontiersin.org/research-topics/83327/multiscale-image-molecular-integration-for-interpretable-immunotherapy-biomarkers-from-radiology-to-spatial-omics