AI-driven Immuno-Radio-Multi-Omics in triple-negative breast cancer: mapping the immune landscape for precision immunotherapy
Abstract
In triple-negative breast cancer (TNBC), tumor immune microenvironment (TIME) is highly heterogeneous and constitutes a key determinant of response to immunotherapy. Nevertheless, current clinical assessment of biomarkers, notably programmed death-ligand 1 (PD-L1) expression and tumor-infiltrating lymphocytes (TILs), relies primarily on static and localized tissue sampling, constrained by spatial heterogeneity, sampling bias, and lack of longitudinal assessment, which limits its capacity to fully represent the evolving TIME and support adaptive treatment decision-making. Artificial intelligence (AI)-driven radiomics has emerged as a non-invasive framework enabling the conversion of routine imaging into high-throughput biomarkers, capturing TIME imaging phenotypes. When integrated with genomic data, this strategy extends into radiogenomics, allowing the association of radiomics features with underlying molecular features to better characterize the TNBC landscape. However, given the inter/intra-diversity of TNBC-TIME, integrating complementary multi-omics layers becomes necessary to fully reflect TIME landscape and represent the interplay between TIME components. Several findings from studies on solid tumors, including TNBC, suggest that AI-derived radiomics signatures are associated with immune and molecular patterns and may predict PD-L1 expression, TILs, TIME phenotypes and response to immune checkpoint blockade. Beyond static and isolated biomarker development, ‘Immuno-Radio-Multi-Omics’ could enable dynamic non-invasive whole-tumor immune profiling, enhanced patient stratification, and support the future development of longitudinal and adaptive therapeutic strategies. Meanwhile, its clinical translation is currently limited by the lack of large and prospective datasets, standardization, validation and interpretability. Through this mini-review, we aim to critically discuss current evidence, limitations and next frontiers of ‘Immuno-Radio-Multi-Omics’ in TNBC, emphasizing its potential to advance dynamic and personalized immunotherapy strategies in TNBC.
Related
- 860 mRNA-LNP-mediated immunomodulation for precision cancer vaccines and immunotherapy
- Precision targeting of tumor-associated macrophages via a 6-substituted pyrrolo[2,3-d]pyrimidine chimeric molecule for enhanced cancer immunotherapy
- Mapping tsetse fly connectivity in Uganda with machine learning landscape genetics
- Molecular Checkpoints of Nucleic Acid Sensing in Tumor Cells: Linking Innate Immune Evasion to Precision Cancer Immunotherapy
- Accuracy, Precision, and Reproducibility of Four T1 Mapping Sequences: A Head-to-Head Comparison of MOLLI, ShMOLLI, SASHA, and SAPPHIRE
BibTeX
@article{haloum2026aidriven,
title = {{AI-driven Immuno-Radio-Multi-Omics in triple-negative breast cancer: mapping the immune landscape for precision immunotherapy}},
author = {Ihsane Haloum and Abdallah Badou},
journal = {Frontiers in Immunology},
year = {2026},
month = nov,
doi = {10.3389/fimmu.2026.1968249},
url = {https://www.frontiersin.org/articles/10.3389/fimmu.2026.1968249},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - AI-driven Immuno-Radio-Multi-Omics in triple-negative breast cancer: mapping the immune landscape for precision immunotherapy
AU - Ihsane Haloum
AU - Abdallah Badou
PY - 2026
DA - 2026/11/04
JO - Frontiers in Immunology
DO - 10.3389/fimmu.2026.1968249
UR - https://www.frontiersin.org/articles/10.3389/fimmu.2026.1968249
AB - In triple-negative breast cancer (TNBC), tumor immune microenvironment (TIME) is highly heterogeneous and constitutes a key determinant of response to immunotherapy. Nevertheless, current clinical assessment of biomarkers, notably programmed death-ligand 1 (PD-L1) expression and tumor-infiltrating lymphocytes (TILs), relies primarily on static and localized tissue sampling, constrained by spatial heterogeneity, sampling bias, and lack of longitudinal assessment, which limits its capacity to fully represent the evolving TIME and support adaptive treatment decision-making. Artificial intelligence (AI)-driven radiomics has emerged as a non-invasive framework enabling the conversion of routine imaging into high-throughput biomarkers, capturing TIME imaging phenotypes. When integrated with genomic data, this strategy extends into radiogenomics, allowing the association of radiomics features with underlying molecular features to better characterize the TNBC landscape. However, given the inter/intra-diversity of TNBC-TIME, integrating complementary multi-omics layers becomes necessary to fully reflect TIME landscape and represent the interplay between TIME components. Several findings from studies on solid tumors, including TNBC, suggest that AI-derived radiomics signatures are associated with immune and molecular patterns and may predict PD-L1 expression, TILs, TIME phenotypes and response to immune checkpoint blockade. Beyond static and isolated biomarker development, ‘Immuno-Radio-Multi-Omics’ could enable dynamic non-invasive whole-tumor immune profiling, enhanced patient stratification, and support the future development of longitudinal and adaptive therapeutic strategies. Meanwhile, its clinical translation is currently limited by the lack of large and prospective datasets, standardization, validation and interpretability. Through this mini-review, we aim to critically discuss current evidence, limitations and next frontiers of ‘Immuno-Radio-Multi-Omics’ in TNBC, emphasizing its potential to advance dynamic and personalized immunotherapy strategies in TNBC.
SN - 1664-3224
ER -