Purinergic P2 receptors have long been attractive therapeutic targets, as evidenced by the clinical success of P2Y12 and P2X3 receptor antagonists. However, P2Y2 receptor remains underexplored in oncology, despite its emerging role in cancer biology and its established clinical utility as the target of the agonist diquafosol for dry eye disease. P2Y2, a G protein-coupled receptor that is activated by extracellular ATP and UTP, integrates multiple oncogenic signaling pathways to drive proliferation, epithelial–mesenchymal transition, invasion, metastasis, and drug resistance. Mechanistically, these effects are mediated through crosstalk with key signaling axes, including EGFR, MAPK, PI3K–AKT, β-catenin, integrins, and HIF-1α. Beyond cancer cell-intrinsic programs, P2Y2 also remodels the tumor microenvironment by promoting angiogenesis, inflammation, and immune evasion, thereby accelerating disease progression. Conversely, P2Y2 activation can induce apoptosis in certain contexts, highlighting its context-dependent, bidirectional role. This review synthesizes current knowledge of P2Y2 structure, signaling mechanisms, and cancer-relevant functions, and surveys advances in pharmacological modulation, including agonists, antagonists, gene-directed strategies, and emerging antibody-based approaches. This review also discusses bioinformatic evidence supporting P2Y2 as a potential biomarker and outlines future directions integrating structural biology and digital medicine to advance P2Y2-targeted precision oncology.
Promoters are central to regulating gene transcription by orchestrating cell-type- and developmental-stage-specific expression programs. However, the intrinsic heterogeneity of transcription factor binding sites, characterized by variable lengths, complex combinatorial patterns, and substantial sequence diversity across cell types, poses significant challenges to the robustness and generalizability of models relying on single-feature representations. To address these limitations, we propose MuSE-Promoter, a deep ensemble framework that integrates multi-scale feature fusion with weighted ensemble learning for accurate promoter identification across diverse cell lines. MuSE-Promoter constructs parallel feature extraction channels that combine contextual sequence embeddings from the DNABERT model and Word2Vec embeddings with handcrafted descriptors, including tri-nucleotide physicochemical properties (TPCP) and reverse-complement k-mer frequencies (RCKmer). A multi-scale convolutional neural network augmented with squeeze-and-excitation (SE) attention captures hierarchical motif patterns while effectively suppressing noise, followed by a Transformer module to model long-range dependencies. Furthermore, the deep learning branch is integrated with a Random Forest classifier through a learnable weighted ensemble strategy, thereby enhancing cross-domain robustness and prediction stability. Comprehensive evaluations across human cell lines from multiple tissues and Arabidopsis thaliana datasets demonstrate that MuSE-Promoter consistently outperforms state-of-the-art methods. Notably, it achieves superior generalization performance in challenging scenarios, including cross-cell-line transfer and enhancer–promoter discrimination. Collectively, MuSE-Promoter provides a powerful computational framework for large-scale promoter annotation, offering new insights into the regulatory mechanisms underlying complex transcriptional regulation.
Breast cancer remains the most prevalent malignancy among women worldwide, characterized by profound heterogeneity and therapeutic resistance that pose significant challenges to both research and clinical management. Breast cancer organoids (BCOs), as three-dimensional in vitro models derived from patient tissues, closely recapitulate the key structural, molecular, and heterogeneous characteristics of primary tumors, surpassing conventional models in fidelity and clinical relevance. This review systematically synthesizes the key advancements in BCO technology, encompassing standardized sample processing, optimized culture systems, and the integration of dynamic microfluidic platforms that better mimic the tumor microenvironment. Due to the high degree of genomic and phenotypic concordance between BCOs and their parental tumors, BCOs have proven indispensable in subtype-specific research, such as illuminating mechanisms of endocrine resistance in luminal tumors, uncovering metastatic pathways in HER2+ cancers, and identifying novel therapeutic vulnerabilities in triple-negative breast cancer. Furthermore, organoid-based metabolic and signaling pathway analyses have revealed new regulators of tumor progression. In the realm of clinical translation, BCOs are increasingly used for high-throughput drug screening and personalized drug sensitivity testing, guiding therapeutic decisions. The establishment of organoid biobanks and the implementation of standardized protocols are facilitating large-scale studies and multicenter collaborations. Despite ongoing challenges in microenvironment simulation and protocol standardization, the convergence of BCOs with emerging technologies, such as single-cell multi-omics, CRISPR screening, and engineered microenvironments, holds immense potential. With their unparalleled capacity to model tumor evolution and therapeutic responses, BCOs represent a paradigm shift in precision oncology, ultimately bridging the gap between basic research and individualized patient care.
The rising incidence and mortality of malignant tumors worldwide call for more effective early cancer screening strategies. Liquid biopsy, particularly the analysis of plasma cell-free DNA (cfDNA), has emerged as a promising non-invasive approach with high sensitivity. cfDNA carries a wealth of tumor-derived molecular information, enabling comprehensive profiling of cancer-associated alterations. In this review, we first summarize seven major cfDNA detection technologies: polymerase chain reaction (PCR)-based assays, bisulfite sequencing, metagenomic sequencing, nanopore sequencing, whole-exome sequencing (WES), whole-genome sequencing (WGS), and CRISPR-based methods. We then describe eleven representative cfDNA features, including fragmentomics, transcription start site (TSS) nucleosome coverage, promoter fragmentation entropy (PFE), DNA methylation, transcription factor (TF) footprints, histone modification patterns, gene mutations, copy number variations (CNVs), T-cell receptor (TCR) repertoires, microbiome signatures, and end-motif patterns. Collectively, these features offer new opportunities for early cancer detection, recurrence monitoring, and therapeutic response assessment. We focus on the technologies and analytical methodologies underlying cfDNA detection, discuss their current and emerging applications in oncology, and highlight recent advances in cfDNA-based approaches. We also critically evaluate the advantages and challenges of clinical implementation and propose future directions for optimizing cfDNA-driven cancer screening strategies.
Elucidating the pathological mechanisms and enabling early, precise diagnosis of Alzheimer's disease (AD) are critical goals in clinical practice. Neuroimaging genetics provides a powerful framework for unraveling the complex genetic architecture and neural substrates of AD across macroscale brain structures and molecular pathways. This review systematically outlines state-of-the-art methodologies for computer-aided AD diagnosis using multimodal data. Specifically, we examine neuroimaging modalities including structural and functional magnetic resonance imaging (MRI), diffusion tensor imaging (DTI), and positron emission tomography (PET), alongside genetic data such as single-nucleotide polymorphisms (SNPs) and gene expression profiles. Correlation analyses are discussed for identifying preliminary associations between genetic variants and imaging phenotypes. Traditional machine learning approaches enable linear cross-modal data fusion, while deep learning methods capture complex nonlinear interactions between genetic and neuroimaging features. Accumulating evidence indicates that multimodal integration significantly improves the accuracy of both AD diagnosis and prognostic prediction. Future research should leverage large-scale longitudinal cohorts and foster interdisciplinary collaboration to enhance the interpretability and clinical applicability of deep learning models, thereby advancing AD research from associative observations toward mechanism-driven precision medicine.
Despite advances in treatment, cancer remains a leading cause of death worldwide. Immunotherapy has revolutionized clinical oncology, extending survival for many patients with advanced disease; however, low response rates in certain cancers remain a major challenge. Histamine, traditionally known as an inflammatory and allergic mediator, has recently been implicated in key aspects of cancer biology. Through binding to its receptors, histamine influences cancer cell proliferation, migration, invasion, apoptosis, and drug resistance. Furthermore, emerging evidence suggests that combining antihistamines with immunotherapeutic agents may improve patient survival. This review focuses on the roles of histamine and its receptors in modulating cancer progression and explores how antihistamines might interfere with these processes. We also discuss the significance of histamine within the tumor microenvironment and propose how modulating histamine signaling could enhance the efficacy of immunotherapy. These insights aim to guide the rational design of future chemo-immunotherapy combinations.
Chronic inflammation in inflammatory bowel disease (IBD) is a recognized risk factor for colorectal cancer (CRC), but the early cellular and molecular events that drive this transition remain poorly understood. To address this, we analyzed over 600,000 single-cell transcriptomes from healthy colon, IBD lesions, and CRC tumors. Focusing on epithelial stem cells (ESCs), we identified a distinct IBD-associated subset that harbors copy number variations (CNVs) resembling those in mismatch repair-deficient (MMRd) tumors, including amplifications of chromosomes 4 and 6 and deletion of chromosome 16. Pseudotime and diffusion map analyses revealed that a fraction of IBD-derived ESCs acquire transcriptional features characteristic of cancer stem cells and follow a trajectory toward cancer-like states. Chromosome 6 amplification was associated with upregulation of MHC class II genes in both IBD and MMRd tumors; however, increased CD8+ T-cell infiltration was observed only in MMRd tumors. Comparative pathway analysis of cells with chromosome 4 or 6 amplification showed activation of interferon-response programs, including IL-27 signaling and CXCL10/CXCL11 expression. DNA-level validation in independent cohorts confirmed the presence of these amplifications in both CRC and IBD patients. In ESC-enriched colorectal tumors, a gene signature derived from these amplifications correlated with improved prognosis. Collectively, these findings define a pre-malignant ESC state in IBD that shares genomic and immunological features with MMRd CRC and suggest that inflammation-driven CNVs in ESCs may serve as early biomarkers for CRC risk stratification.
Cutaneous melanoma is an aggressive type of skin cancer that is often triggered by DNA damage induced by exposure to ultraviolet (UV) radiation. Despite initial high response rates to mutant BRAF- and MEK-targeting therapies, advanced melanomas inevitably develop resistance to these agents. Increasing evidence suggests that a complex network of interacting and intersecting signaling mechanisms directs melanoma progression and therapy responsiveness, among which the canonical and noncanonical Wnt signaling pathways play a commanding role. This review examines Wnt signaling crosstalk with BRAF/MEK and PI3K/AKT signaling networks and its role as a hub for coordinating phenotypic switching, metabolic rewiring, and immune evasion. It also explores and evaluates current therapeutic approaches developed for targeting Wnt signaling and its associated ubiquitin machinery, and proposes strategic considerations for overcoming toxicities associated with targeting the Wnt pathway.
Breast cancer continues to be a leading cause of cancer-related morbidity and mortality among women globally. Existing clinical management strategies are often constrained by limitations in precision, personalization, and operational efficiency. In recent years, artificial intelligence (AI) has emerged as a transformative force in breast oncology, revolutionizing diagnostic accuracy, refining treatment protocols, and enhancing prognostic assessments. By leveraging advanced computational techniques, AI improves the interpretation of ultrasound, radiological and histopathological images, facilitates the development of personalized treatment regimens, and enables more accurate risk stratification through the integration of multi-omics data. A deeper understanding of these AI-driven innovations is crucial for promoting the transition toward a more data-informed and individualized paradigm of breast cancer care. This review provides a comprehensive analysis of the expanding role of AI in breast oncology, highlighting its potential to redefine standards of care and shape the future trajectory of precision oncology.
Precision oncology for breast cancer is currently constrained by substantial biological heterogeneity, which renders traditional anatomical metrics—such as the Response Evaluation Criteria in Solid Tumors (RECIST)—inadequate for capturing distinct molecular subtypes and early therapeutic responses. Integrating artificial intelligence (AI) with multimodal molecular imaging establishes a transformative paradigm for non-invasive, quantitative assessment of tumor biology. This review synthesizes the pivotal role of AI in optimizing positron emission tomography (PET), functional and molecular magnetic resonance imaging (MRI), and optical/photoacoustic imaging, emphasizing technical milestones in reconstruction, automated segmentation, and deep feature fusion. We critically evaluate the clinical evidence for AI-enhanced molecular imaging across the continuum of care, focusing on diagnostic accuracy, non-invasive axillary staging, neoadjuvant therapy monitoring, and prognostic stratification. Despite this promise, the field faces substantial hurdles, including data standardization, algorithmic robustness, and translational barriers. Future progress hinges on personalized, adaptive imaging pathways and privacy-preserving collaborative frameworks, such as federated learning. Ultimately, accelerating clinical translation will require sustained interdisciplinary collaboration to shift the diagnostic paradigm from static anatomical observation toward intelligent, dynamic molecular assessment.
Research on the intratumoral microbiota is shifting from descriptive analyses of presence and abundance toward understanding its spatial heterogeneity and local function. Here we propose the concept of the microbiota-residing spatial niche (MRSN), defined as a functional unit formed by interactions between microbiota and neighboring tumor, immune, and stromal cells within a defined spatial context, and characterized by four key features: spatial discernibility, functional consistency, microbial dependency, and clinical relevance. In this review, we discuss the limitations of single-omics approaches and present a framework centered on spatial multi-omics. This framework integrates spatial transcriptomics, multiplex immunoimaging, spatial metabolomics, and graph neural networks, enabling a stepwise analysis that first localizes microbiota, then characterizes their surrounding cellular neighborhoods, and finally validates their functional roles. We classify MRSNs into immunosuppressive, immunostimulatory, protumorigenic, and antitumorigenic types, and demonstrate how they modulate therapy responses through local metabolic remodeling and immune regulation. We also contrast the systemic immunomodulatory effects of gut microbiota with the localized influence of intratumoral microbiota. Finally, we discuss challenges in technical validation, temporal dynamics, and model translation, and propose strategies for precision interventions targeting niche vulnerabilities and clinical stratification. This framework provides a unified conceptual basis for understanding intratumoral microbiota functions and informs next-generation spatially guided therapeutic strategies.
Aging populations, climate change, and emerging infectious disease threats are increasingly straining traditional public health models. Digital twins (DTs), virtual replicas continuously updated with real-world data, have recently expanded from engineering into healthcare, with growing interest in population-level applications. In practice, DTs can integrate diverse data streams, including epidemiological, environmental, and emerging molecular surveillance data, to support dynamic population-level health modeling. Existing DT pilot projects demonstrate technical feasibility in controlled settings, but most applications in public health remain at an early developmental stage, with limited scale, validation, and real-world implementation. Progress will depend on robust data ecosystems, interoperable platforms, and transparent governance frameworks, while integration with advanced artificial intelligence may further expand future predictive capabilities. We review the potential of DTs in public health, with emphasis on infectious disease surveillance, emergency preparedness, environmental monitoring, chronic disease management, and policy evaluation. In addition, this review proposes a preliminary multi-dimensional framework that evaluates axes such as data integration capacity, predictive validity, operational usability, and governance, thereby providing a basis for assessing the maturity and translational readiness of DT systems in public health. DTs hold potential to support a shift in public health from reactive approaches toward more proactive and predictive practice. However, realizing this vision requires overcoming substantial foundational and translational challenges.
Achieving surgical precision in colorectal cancer remains challenging due to subjective interpretation of visual and tactile cues, which frequently results in positive resection margins or overlooked microscopic lesions. Although intraoperative fluorescence molecular imaging improves real-time visualization, its broader clinical adoption is hindered by interobserver variability, false-positive signals, and ambiguous boundary delineation. The integration of artificial intelligence (AI) is shifting fluorescence imaging from purely visual assistance toward quantitative, data-driven surgical guidance. Here, we systematically examine the synergistic evolution of three core pillars—targeted molecular probes, advanced imaging devices, and AI algorithms—focusing on recent progress in automated tumor segmentation, discrimination between malignant and inflammatory tissues, vascular mapping, sentinel lymph node detection, and anastomotic perfusion assessment. We also discuss the clinical utility of AI-enhanced fluorescence imaging in resecting primary and metastatic tumors, alongside its emerging role in surgical education, and critically appraise current obstacles concerning data availability, regulatory approval, and clinical validation. Finally, we outline prospective directions aimed at fostering more predictive and personalized surgical strategies, while recognizing that most of these applications remain in early-stage validation.