Jul 2026· Cancer Research· Vol 86, pp. A026-A026· 0 citations
TL;DR
An artificial intelligence (AI)-driven integrative framework that combines digital pathology with multi-omics profiling to enable cell-type–resolved characterization of tumor biology in rare and environmentally associated cancers and highlights the power of AI-driven digital pathology combined with transcriptomic and proteomic integration to uncover cell-type–specific mechanisms underlying rare cancers.
Abstract
Rare cancers present persistent challenges in biomarker discovery and clinical translation due to limited sample availability, histopathologic heterogeneity, and fragmented molecular data. To overcome these barriers, we developed an artificial intelligence (AI)-driven integrative framework that combines digital pathology with multi-omics profiling to enable cell-type–resolved characterization of tumor biology in rare and environmentally associated cancers. Our approach leverages whole-slide imaging and computational pathology algorithms to perform high-resolution cell typing and spatial characterization of the tumor microenvironment. These spatially informed features are integrated with matched transcriptomic and proteomic data using machine learning models to identify robust, biologically interpretable biomarkers. By linking cellular architecture with molecular signatures, our framework captures tumor heterogeneity on both structural and functional levels. As a proof-of-concept, we applied this platform to arsenic-associated bladder cancer, an exposure-driven malignancy with regionally rare incidence but significant global health impact. We identified distinct cell-type–specific gene and protein expression patterns associated with disease risk and progression. Integrative modeling revealed key pathways linking environmental exposure, tumor organization, and immune microenvironment dynamics. Biomarker candidates demonstrated reproducibility across independent cohorts and tissue-based validation datasets. Importantly, this framework is designed for translational scalability, incorporating predictive modeling for patient stratification and deployment through cloud-based analytical pipelines. By enabling the integration of histopathologic features with multi-omics data in low-sample settings, our approach addresses a critical gap in rare cancer research and supports the development of clinically actionable biomarkers. This study highlights the power of AI-driven digital pathology combined with transcriptomic and proteomic integration to uncover cell-type–specific mechanisms underlying rare cancers. Our platform provides a generalizable strategy to advance precision oncology, improve diagnostic accuracy, and facilitate equitable access to data-driven care for patients with rare and understudied malignancies.
Sandeep K. Singhal. AI-Enabled Digital Pathology and Multi-Omics Integration for Cell-Type–Resolved Biomarker Discovery in Environment-Associated Cancers [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Breaking Barriers in the Fight against Rare Cancers; 2026 Jul 18-20; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(14_Suppl):Abstract nr A026.
Precision phenomics is introduced as a unifying framework that links molecular, spatial, functional, and clinical characteristics of tumors to support personalized cancer management and has the potential to transform lung cancer research and improve patient outcomes.
Translational oncology has moved rapidly from histopathology and single-analyte biomarkers toward multi-dimensional molecular profiling. Yet many clinically deployed tests still use reductionist biomarker strategies that under-represent cancer complexity. This review examines whether a mechanistic, multi-layered, and specimen-aware approach can improve cancer detection, classification, prognosis, minimal residual disease (MRD) assessment, and therapeutic selection. Evidence across solid tumours shows that genomic alterations alone incompletely explain tumour state, metastatic behaviour, immune evasion, or therapeutic vulnerability. Integrated genome and transcriptome analyses, proteogenomics, single-cell atlases, fragmentomic, methylation based cell-free DNA assays, metabolomics and microbiome assessments reveal clinically relevant biology that single modality tests cannot determine. Minimally invasive collected specimens can extend access to screening, diagnosis and longitudinal monitoring, but the choice of specimen should be matched to disease biology and analytes that represent mechanisms of oncogenesis. However, translation remains constrained by pre-analytical variability, contamination, differences in tumour shedding behaviour, clonal haematopoiesis, translation of generated models, incomplete external validation and uncertain downstream clinical utility for emerging platforms. This review provides a commentary on the future of cancer diagnostics, the considerations and barriers to clinical translation, the relationship between utility and dimensionality of biomarkers assessed and the emerging rationale towards mechanistically grounded integrated models.
I. Daniels, Andrew J. Page, Daniel Wise· Cancers· 0 citations
Triple-Negative Breast Cancer (TNBC) is characterized by high heterogeneity, poor prognosis, and limited targeted treatment options. Bridging the gap between molecular alterations and histopathological morphology remains a major challenge in precision oncology. We propose an interpretable, multi-modal framework that integrates histopathological image analysis with multi-omics profiling (somatic mutations, DNA methylation, copy number alterations), leveraging U-Net-based nuclei segmentation, vision-language models (BLIP), biomedical language models (BioGPT), and explainable AI (SHAP, LIME). Our framework achieves strong predictive performance (AUC = 0.989) and provides transparent, biologically grounded interpretations by integrating morphological features with genomically prioritized biomarkers. Cross-modal analysis confirms established TNBC drivers and generates novel, testable hypotheses associating specific epigenetic alterations with distinct morphological phenotypes. While causal validation requires future wet-lab experiments, our framework accelerates hypothesis-driven biomarker discovery by integrating complementary data modalities with language-based reasoning, providing a transparent foundation for hypothesis generation and clinical translation.
AL Imran, Khandokar Md. Rahat Hossain, SM Rafiqul Islam et al.· bioRxiv· 0 citations
Metastatic disease remains the leading cause of cancer-related death, yet most precision oncology strategies still emphasize profiling primary tumors and tracking cell-free tumor DNA (ctDNA). Although ctDNA has transformed genomic profiling, molecular residual disease monitoring, and early cancer detection, it cannot directly capture viable tumor cell states, phenotypic plasticity, or functional adaptations that drive metastatic spread. We propose that the next phase of precision oncology should integrate the cellular dimension of metastasis through systematic circulating tumor cell (CTC) profiling. The SCRUM-MONSTAR platform, one of the largest pan-cancer molecular profiling initiatives in Japan, offers an exceptional foundation for this transition through its nationwide infrastructure for multi-omics analysis, longitudinal biospecimen collection, and artificial intelligence-enabled clinical interpretation. By combining matched tissue profiling, serial ctDNA analysis, single-cell CTC transcriptomics, metabolomics, and organoid- and mouse-based functional modeling, SCRUM-MONSTAR-CTC could evolve into a translational ecosystem for anti-metastatic drug discovery. Within this framework, we highlight adherent-to-suspension transition (AST) as one representative, experimentally tractable plasticity program that enables tumor cells to survive in circulation and subsequently colonize distant organs. We envision that identifying and therapeutically targeting AST-related and other metastatic plasticity programs across tumor types will provide a path toward clinically actionable anti-metastatic therapies. More broadly, this framework could enable the identification of metastatic vulnerabilities, the development of biomarker-guided anti-metastatic trials, and the reverse translation of patient-derived discoveries into early-phase clinical testing. Precision oncology must move beyond cataloging tumor genomes and begin targeting metastasis as a dynamic biological process. UMIN000056873, approved by the Institutional Review Board of the National Cancer Center Hospital East.
T. Hashimoto, T. Shibuki, T. Fujisawa et al.· International Journal of Cli...· 0 citations
Clear cell renal cell carcinoma (ccRCC) is a highly heterogeneous cancer with complex tumor and immune microenvironment interactions influencing progression and therapy response. Advances in spatial technologies now allow for simultaneous spatial mapping of transcriptomic and proteomic features from patient tissue and phenotyping of blood cells, enabling detailed assessment of disease biology. This study highlights a multi-omic approach to characterizing systemic immune responses, tumor microenvironment niches, immune states and tumor heterogeneity in ccRCC patient samples.
We profiled ccRCC patient samples using peripheral blood mononuclear cells (PBMC) and tumor-derived cells (TDCs) by CyTOF™ technology with a 50-plus-antibody panel. Matched formalin-fixed tumor tissue was examined by a spatial transcriptomic platform, hematoxylin and eosin (H&E) assessment, and Imaging Mass Cytometry™ (IMC™) technology with a 43-marker panel sequentially on the same tissue section. These datasets were analyzed to explore immune and tumor signatures linked to disease states.
We demonstrate compatibility of IMC workflows with H&E and spatial transcriptomic modalities. IMC technology identified spatial distribution and activation states of immune and tumor cells in ccRCC. Combining spatial transcriptomics and proteomics enabled detailed phenotyping of tumor metabolic and signaling states and immune cytokine and transcription factor expression. Comparison of PBMC and TDC composition provided insights into systemic and localized immune responses.
This study highlights the power of spatial multi-omic profiling to unravel the immune and oncologic landscape of ccRCC with high resolution. Identification of spatially defined immune cell states and niches offered insights into tumor immune evasion and resistance mechanisms. These findings pave the way for spatially informed biomarkers and potential precision therapies for ccRCC. For Research Use Only. Not for use in diagnostic procedures.
n/a
Tumor Immunology: Cellular Responses and Tumor Microevironment (TIME)
Michael J. Cohen, Q. Raza, N. Zabinyakov et al.· Journal of Immunology· 0 citations
Rare cancers have challenged conventional drug development because patient populations are too small for traditional disease-specific trials. Yet rare cancers may provide the paradigm for the future of oncology. Cancer classification has evolved from organ-of-origin taxonomy to molecular stratification and then to tumor-agnostic therapies targeting single biomarkers across histologies. However, comprehensive molecular profiling reveals that most tumors harbor not a single driver, but complex, unique molecular architectures. Thus, every cancer may be considered ultra-rare—a biologically distinct N-of-1.
Rare cancers expose a fundamental limitation of traditional treatment models. While tissue-specific and single-biomarker approaches have produced important advances, they do not capture the molecular complexity and heterogeneity observed within and across cancer types. Even tumors sharing histology or genomic alterations often differ substantially in co-occurring abnormalities, immune contexture, and evolutionary trajectories. Rare cancers therefore exemplify a broader principle: the individual patient, rather than the disease category, is the relevant unit of analysis. Precision oncology must move beyond matching one alteration to one drug and instead address the totality of a tumor/patient molecular profile. Because multiple dysregulated pathways commonly coexist, optimal therapy may require customized combinations of gene- and immune-targeted agents, cytotoxics, endocrine therapies, and repurposed drugs selected to address the unique molecular ecosystem of each tumor. The I-PREDICT study provided a blueprint for this transformative model. Patients received individualized therapies matched to their molecular alterations, and a Matching Score quantified the degree to which treatment addressed the tumor's molecular abnormalities. Clinical outcomes correlated linearly/significantly/independently with Matching Score, demonstrating that greater molecular matching was associated with greater benefit. Therefore, matching complexity matters, and individualized combination strategies can be safely implemented and evaluated. The N-of-1 (r)evolution carries major implications for clinical research. If each patient's cancer is molecularly unique, the question becomes whether molecular matching algorithms can effectively translate complex biological data into personalized treatment regimens. In this model, the algorithm—not the tumor type or individual drug—becomes the principal object of investigation.
Rare cancers foreshadow the future of precision oncology. As molecular characterization becomes increasingly comprehensive, classification based on tissue of origin or single biomarkers gives way to recognition that each tumor possesses a unique and complex molecular identity. The N-of-1 paradigm reframes cancer as requiring customized therapeutic solutions. Future advances will depend not only on new drugs but also on validating sophisticated matching algorithms capable of delivering truly personalized cancer care.
Razelle Kurzrock. Precision oncology and the N-of-1 revolution: Rare cancers as a paradigm [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Breaking Barriers in the Fight against Rare Cancers; 2026 Jul 18-20; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(14_Suppl):Abstract nr IA007.
R. Kurzrock· Cancer Research· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
