NPJ Precision Oncology最新文献

Chemoresistance remains a critical challenge in breast cancer (BC) treatment. By integrating multi-omics (single-cell, spatial, and bulk transcriptomics) with clinical validation, we identified a specific COL3A^high CAF subset that drives BC chemoresistance. Mechanistically, these CAFs undergo lipid metabolic reprogramming, secreting excess oleic acid via SCD. This oleic acid binds to ENO1 on tumor cells, activating the PI3K/Akt pathway and inhibiting chemotherapy-induced apoptosis. Simultaneously, COL3A^high CAFs orchestrate an immunosuppressive niche by recruiting regulatory T cells and impairing cytotoxic CD8⁺ T cells. Our findings establish COL3A^high CAFs as key mediators of resistance through metabolic symbiosis and immune evasion. The strong correlation between COL3A^high CAF abundance and clinical poor response highlights their potential as both predictive biomarkers and therapeutic targets to overcome chemoresistance in BC patients.

The effective application of precision oncology in solid tumors remains challenging due to genetic heterogeneity and the absence of actionable alterations in some cancers. In this review, we discuss the integration of liquid biopsy and mutational signatures as a potential framework to address these limitations by enabling longitudinal detection of mutational processes that arise during tumor development and evolution. Together, these complementary approaches hold substantial promise for enhancing cancer screening, refining diagnosis, and guiding personalized therapeutic strategies, thereby advancing the field of precision oncology.

Genetic interactions (GIs) drive carcinogenesis and treatment resistance via non-additive phenotypic effects between genes. Traditional bulk-based methods fail to capture cell-type-specific interactions in heterogeneous tumors like lung adenocarcinoma (LUAD), limiting precision oncology. Resolving cell-type-specific GIs at single-cell resolution persists as a major hurdle, hindered by limited analytical methodologies. Here, we develop scPGI-finder, a computational framework that identifies gene pairs whose coordinated high expression is associated with higher proliferation-related fitness at single-cell resolution, which we refer to operationally as single-cell positive genetic interactions (scPGIs). Using scPGI-finder, we identify 49,808 and 15,896 scPGIs spanning epithelial cells and T cells in LUAD, respectively. The predicted scPGIs display tighter junctions in the protein interaction network compared to non-scPGIs. Furthermore, we demonstrate the predictive power of scPGIs for malignancy and immunotherapy response through multi-omics validation across diverse cohorts. Notably, with a mean area under the ROC curve (AUROC) of 0.974 in bulk tissue validation, the epithelial-derived scPGI classifier enables concordant malignancy identification across scales ranging from epithelial single cells and lung cancer cell lines, through spatial transcriptomic maps, to bulk LUAD tissue profiles. Additionally, a six-scPGI T cell signature reliably forecasts immunotherapy efficacy, with AUROC values exceeding 0.80 across multiple datasets. Together, our research advances the understanding of underlying cancer-positive GIs at the single-cell level. scPGIs of epithelial and T cells serve as robust biomarkers for malignancy evaluation and treatment response, offering a translational framework for precision oncology.

Sex differences in cancer susceptibility and prognosis are partially driven by sex chromosomes and sex hormones. However, the molecular mechanisms underlying the higher incidence and mortality of multiple myeloma (MM) in males remain poorly defined. In this study, we identify the Y-linked gene EIF1AY as a tumor-suppressive regulator in male MM. Clinical analysis reveals that partial deletions of EIF1AY in male MM patients are significantly associated with disease progression, reduced treatment responsiveness, and shorter overall survival. Functionally, loss of EIF1AY promotes M2 macrophage polarization and recruitment, thereby enhancing MM cell proliferation. Mechanistically, EIF1AY forms a protein complex with RPS4Y1 that directly binds to and stabilizes CD134 mRNA, thereby promoting CD134 expression in MM cells. The RPS4Y1-EIF1AY-CD134 axis suppresses IL-4 and IL-13 secretion from MM cells, which in turn downregulates the membrane receptor DDR1 on co-cultured macrophages, thereby inhibiting M2 macrophage polarization and recruitment, and ultimately restraining MM cell proliferation. These findings uncover a feed-forward loop in which the RPS4Y1-EIF1AY-CD134 axis suppresses IL-4/IL-13-DDR1 signaling, thereby suppressing M2 macrophage polarization and recruitment, and sustaining tumor growth through reciprocal crosstalk between tumor cells and macrophages. Collectively, our study elucidates a novel immune regulatory pathway driving sex differences in MM and highlights EIF1AY as a promising target for precision immunotherapy in male patients.

Compressive stresses are linked to the malignancy state of tumors. These stresses can drive cancer cells toward a malignant phenotype. The objective of this study is to investigate how patient-specific heterogeneity of a tumor tissue influences the stresses experienced by tissue components that are believed to play important roles in malignancy state. A unique image-based, physics-driven in silico modeling is developed, replicating a breast tumor tissue with the complexity and heterogeneity as observed in humans. This model employes images acquired by Fourier transform infrared (FTIR) microscopy which images and classifies breast tissues into six components including non-cancerous, malignant, others, dense, loose, and reactive stroma. We show that heterogeneous tissues having small and disconnected pieces of malignant components experience higher stresses, highlighting the dependency of stress magnitude on components' configuration, neighborhood, and initial surface area. Our in silico model predicts stresses on pre-cancerous lesions in the range that drive them to become lethal.

Artificial intelligence (AI) is being used in oncological drug development to address the high costs, low success rates, and long timelines that characterize traditional drug development pipelines. The use of machine learning (ML) and deep learning (DL) models in computer-aided drug design is constantly growing owing to their capacity to analyze large, heterogeneous datasets, their ability to capture nonlinear biological trends, and their integration of various molecular and clinical characteristics. AI applications accelerate target discovery by predicting protein structures, ranking disease-relevant genes, and assessing target drugability. AI can be used to conduct rapid searches of multiplexed chemical libraries, predict drug-target interactions, and optimize the pharmacological and physicochemical properties of drugs in virtual screening. Advanced neural network designs also aid in de novo drug design, which involves developing new molecular structures with therapeutic properties of interest. This review outlines how AI has been used for target identification, virtual screening, de novo molecular design, and, specifically, in cancer applications. It further discusses the major issues in AI-based drug development, such as data quality, model interpretation, computational constraints, and ethical and regulatory considerations, which remain essential obstacles to broader clinical translation.

Hepatocellular carcinoma (HCC) remains a major global health challenge due to its molecular heterogeneity, late diagnosis, and limited therapeutic options. Recent studies have identified isonicotinylation (K_inic), a novel lysine acylation, as a regulatory modification influencing carcinogenic protein activity and liver cancer progression. In this study, we established the K_inic Index (K_inicI), an artificial intelligence (AI)-driven predictive model that integrates multi-omics data and consensus clustering to classify HCC patients into two distinct K_inic subgroups. Patients in the high-K_inic subgroup exhibited significantly worse overall survival, demonstrating the value of K_inicI for risk stratification and outcome prediction. Machine learning approaches (LASSO, RSF) coupled with Shapley additive explanation (SHAP) analysis identified CYP2C9 and G6PD as the most influential prognostic variables associated with HCC progression. Single-cell and spatial transcriptomic analyses confirmed that CYP2C9 and G6PD are primarily localized in malignant hepatocytes with high metastatic potential, underscoring their clinical relevance. Importantly, using the GraphBAN deep learning framework and ADMET-AI screening, we prioritized candidate compounds targeting CYP2C9 and G6PD, followed by molecular docking that validated strong binding affinities, suggesting their potential as novel therapeutics. Together, our study demonstrates that K_inicI is a powerful AI-enabled platform for prognostic modeling, molecular stratification, and multitarget drug discovery, providing a foundation for precision oncology and resistance-aware treatment strategies in HCC patients.

Anaplastic large cell lymphoma (ALCL) is a rare form of mature T cell lymphoma in children, particularly the anaplastic large cell kinase (ALK) negative subtype. Despite frontline treatment advances, there is no standard approach to treat relapsed disease and prognosis remains poor. Recently, JAK/STAT activating mutations have been implicated in the pathogenesis of ALK-negative ALCL in adults, but the oncogenic drivers of this disease in children are not well characterized. Herein, we describe a case of a 13 year-old boy with early systemic relapse of ALK-negative ALCL harboring a rare ATXN2L::JAK2 fusion, who achieved complete remission with ruxolitinib monotherapy. Consolidative allogeneic hematopoietic stem cell transplant HSCT then lead to long-term remission. This case underscores the critical role of comprehensive genomic profiling for rare histologies and supports the potential utility of JAK/STAT pathway inhibitors in select patients with ALK-negative ALCL.

Immunotherapy has significantly improved the treatment of metastatic solid tumors; however, detecting early signs of response to enable timely intervention for resistant tumors remains challenging. A blood-only circulating tumor DNA (ctDNA) test may provide a rapid assessment of tumor response without reliance on matched tumor tissue. We applied a tissue-agnostic, genome-wide methylation enrichment assay, based on cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq), to plasma samples from patients in a phase 2 trial evaluating pembrolizumab across multiple solid tumors (NCT02644369). A decrease in ctDNA from baseline to pre-cycle 3 was significantly associated with higher objective response and clinical benefit rates and longer progression-free and overall survival in univariate analyses, with these associations remaining significant in multivariable models except for overall survival. These results validate a commercial-grade, tissue-agnostic plasma cfDNA methylation platform for immunotherapy response monitoring, which may facilitate earlier, more informed treatment decisions and improve patient outcomes.

Proteins are ultimately responsible for cellular phenotypes and are targeted by most anticancer drugs. However, beyond immunohistochemistry, proteins are not typically measured in precision oncology, meaning transcriptomics is used as a proxy. To determine how informative mRNA is for guiding personalised treatments, mRNA-protein correlations were analysed in three large pan-cancer datasets and made available in a web portal (https://oncorr.aws.procan.org.au/). OnCorr can be integrated into precision medicine programs to augment transcriptomics.