Li Min

Background: Gastric cancer (GC) is a molecularly heterogeneous disease. This study aims to explore whether endocytosis pathway-associated genes and immune checkpoint-related features were linked to GC subtypes and prognosis. Methods: Public transcriptomic datasets from TCGA and GEO were analyzed. ssGSEA was applied to predefined endocytosis-related gene sets, and NMF was used to identify molecular subtypes. Survival, tumor microenvironment features, tumor mutation burden, microsatellite instability, and differentially expressed genes were evaluated. A three-gene prognostic model was developed and assessed through retrospective internal and external validation across public cohorts for prognostic association only. Results: Two endocytosis-related subtypes were identified. The Group 1 subtype showed higher pathway activity, more immunosuppressive microenvironmental features, higher expression of selected immune checkpoint-related markers, and poorer survival. The Group 2 subtype showed lower pathway activity and more favorable outcomes. A three-gene signature (NRP1, SERPINE1, and MISP3) stratified patients into higher-risk and lower-risk groups across retrospective cohorts. Higher-risk cases were associated with worse overall survival and enrichment of epithelial-mesenchymal transition- and immune evasion-related pathways. This retrospective cross-cohort validation suggests that the prognostic signal was not restricted to a single dataset, but it does not demonstrate prospective clinical validity, analytical validity, or readiness for clinical implementation. Conclusions: Endocytosis-related molecular patterns may be associated with distinct biological and prognostic features in GC. The identified subtypes and prognostic signature should be interpreted as exploratory, hypothesis-generating findings that warrant further prospective, functional, and assay-level validation.

Extracellular vesicles (EVs) have emerged as superior tools compared with plasma tumor biomarkers for gastric cancer detection. However, clinical practice of EV detection for early stage gastric cancer (EGC) discrimination remains limited, primarily hindered by the low sensitivity of current detection technologies and the intrinsic complexity of EVs. Here, we propose a label-free microfluidic microsensor for the ultrasensitive and rapid detection of EVs in mixed environments. An estimated detection limit of 23.87 EVs/mL for EV detection is achieved by using the ultrahigh-Q whispering-gallery-mode microbubble resonator. Moreover, by incorporating an interpretable machine learning algorithm, the complex spectral features of EVs can be effectively captured for discrimination analysis. In clinical validation, the system achieves a discrimination performance of 90.91% in distinguishing EGC from healthy individuals, significantly outperforming that of commonly used tumor biomarkers. These findings provide robust evidence supporting the feasibility of machine learning-assisted, microcavity-enhanced EV analysis for distinguishing EGC from other clinical groups.

The optimal bowel preparation for acute lower gastrointestinal bleeding (ALGIB) in the emergency remains uncertain. This study aimed to compare the efficacy and safety of oral versus non-oral laxative approaches for emergency colonoscopy in ALGIB patients. We retrospectively analyzed 157 patients undergoing emergency colonoscopy for ALGIB, stratified into PEG (n = 67) and non-PEG (n = 90) groups. Inverse probability of treatment weighting (IPTW) was applied to adjust for baseline confounders, including pre-procedural suspected etiology. Primary outcomes were endoscopic diagnosis (including definitive and presumptive) and hemostasis rates. Secondary outcomes included bowel preparation quality, procedural completion, and medical resource utilization. Post hoc analyses were performed by categorizing etiologies into focal/subtle and diffuse/obvious types. After IPTW adjustment, overall core outcomes including endoscopic diagnosis (86.5% vs. 80.3%), stigmata of recent hemorrhage (SRH) detection (42.7% vs. 43.9%) and hemostasis (12.5% vs. 16.9%) were comparable between groups (all adjusted P > 0.05). The PEG group demonstrated significantly higher rates of adequate bowel preparation (BBPS ≥ 6: 70.6% vs. 31.8%, P < 0.001) and cecal intubation (86.2% vs. 55.9%, P = 0.001), with no significant differences between groups in procedural efficiency or resource utilization (all P > 0.05). However, outcomes were significantly etiology-dependent. Diffuse/obvious lesions were associated with significantly higher definite diagnosis rates (64.3% vs. 38.6%, P = 0.004) but lower cecal intubation success (66.1% vs. 84.3%, P = 0.017) compared to focal/subtle lesions. Clinical outcomes in ALGIB are fundamentally etiology-dependent. While PEG and non-PEG strategies achieve comparable overall diagnostic and therapeutic yields, their distinct clinical utilities vary across lesion types. These findings advocate for a stratified, etiology-driven approach to bowel preparation. Future efforts should prioritize the development of pre-procedural predictive models to refine individualized management strategies.

Terahertz (THz) spectroscopy has recently gained significant attention as a powerful tool for biomacromolecule detection due to its exceptional sensitivity in capturing molecular fingerprints, attributed to its unique wavelength range. THz biosensing platforms offer distinct advantages in identifying molecular rotational and vibrational states, making them effective for high-sensitivity, high-precision biomacromolecule analysis. This review highlights recent advancements in THz biosensing platforms, focusing on their spectroscopy interpretation and practical applications. We begin by introducing commonly used THz equipment and various metamaterials, followed by an overview of the key spectroscopy interpretation strategies for THz spectral analysis, including refractive index sensing and molecular fingerprint recognition. We also summarize the machine learning (ML) methods applied to enhance spectral analysis in THz biosensing platforms. Lastly, we provide an integrated perspective on the spectroscopy interpretation, clinical applications, and recent technological advancements in THz biosensing platforms, with an emphasis on how to construct a standardized THz-based biosensing workflow. We believe THz biosensing platforms hold immense potential to revolutionize molecular detection by enabling precise, label-free, and highly sensitive analysis of biomacromolecule interactions, paving the way for breakthroughs in medical diagnostics, environmental monitoring, and biochemical research.

Plasma metabolomics offers significant potential for non-invasive biomarker discovery in gastric cancer (GC), yet conventional analytical workflows face challenges in absolute quantification and biological interpretability, hindering clinical translation. Here we present an innovative multi-phase hybrid framework integrating untargeted metabolomics with relative- and absolute-quantitative targeted metabolomics, coupled with a custom interpretability-driven algorithm for de novo biomarker identification. We perform metabolic profiling on 1,706 plasma samples from multicenter cohorts, identifying 84 key metabolites significantly enriched in caffeine metabolism and primary bile acid biosynthesis during the relative quantitation phase. By applying the custom algorithm to absolute quantitation data, we establish a 12-metabolite panel covering multiple functional metabolic modules. Machine learning-based diagnostic models using this signature achieve an area under the curve of 0.951 in validation cohort. Together, our study provides a robust and interpretable framework for translational metabolomics and establishes a GC detection biomarker panel, laying the foundation for future mechanistic research and clinical application.

Large‐scale population cohorts and biobanks are cornerstones of precision medicine, providing extensive multimodal data that afford unprecedented opportunities to elucidate the mechanisms of complex diseases. However, traditional biobanks have largely functioned as static data repositories, and prevailing analytical frameworks face critical challenges, including data heterogeneity, fragmented knowledge extraction, and a limited pace of clinical translation. This article advances a new paradigm of the “intelligent biobank,” repositioning these passive repositories as active “predictive and discovery engines.” We argue that the emergence of large language models (LLMs) offers a transformative approach to addressing these challenges. As a unified computational interface across data modalities, LLMs can harness advanced natural language understanding and generation to integrate genomic, phenomic, imaging, and other multidimensional data. This integration enables interpretable biomarker discovery, dynamic risk prediction, and mechanistic hypothesis generation. Such a transformation could substantially accelerate the translation of precision medicine from research into clinical practice.

Thoracic aortic aneurysm (TAA) is a life-threatening condition characterized by aortic dilation, inflammation, and extracellular matrix degradation. Despite advances in surgical management, effective pharmacological therapies are still lacking, largely due to an incomplete understanding of the cellular mechanisms driving disease progression. Although recent single-cell RNA sequencing (scRNA-seq) studies have revealed diverse cell types in TAA, the intercellular communication driving pathological remodeling is still poorly defined. Here, we performed integrated scRNA-seq analysis of human TAA (n = 8) and healthy aorta (n = 8) to construct a comprehensive cellular landscape. We identified a disease-associated crosstalk between SPP1⁺ myeloid cells and POSTN⁺ fibroblasts, mediated by SPP1 and MK signaling. These two cell subsets were enriched in TAA and co-activated TNF-α signaling via NF-κB and epithelial mesenchymal transition (EMT) pathways, thereby promoting inflammation and ECM remodeling. Cell-cell communication analysis further uncovered upregulated interactions involving SPP1-integrin (e.g., ITGAV/ITGA8/ITGA5+ITGB1) and MDK receptors (SDC4/SDC2/NCL/LRP1/ITGA4/ITGA6+ITGB1) in TAA. These computational findings were further supported by multiplex immunofluorescence and spatial transcriptomics analyses. By integrating key genes and signaling pathways, we identified hub genes and their associated transcription factors, whose regulatory activity was further supported by transcription factor regulon analysis. Our findings highlight the crucial role of myeloid-fibroblast interactions in driving TAA pathogenesis and identify potential therapeutic targets.

Objective: To examine whether baseline levels and short‐term changes of asymmetric dimethylarginine (ADMA) are associated with subclinical carotid atherosclerosis (SCA) and its progression in adults with metabolic dysfunction–associated steatotic liver disease (MASLD). Methods: A total of 600 adults with MASLD were prospectively recruited and followed annually for 2 years. Serum ADMA was measured at baseline and 1‐year follow‐up. SCA, defined as increased carotid intima‐media thickness or carotid plaque, was assessed at baseline, 1‐ and 2‐year follow‐up. Multivariable regression and propensity score‐based analyses were performed to assess the associations between baseline ADMA or change in ADMA and SCA risk. Results: Increased baseline ADMA level was significantly associated with SCA. Each 1‐standard deviation (SD) increase in baseline ADMA was associated with higher odds of SCA (odds ratio 2.24, 95% confidence interval [CI] 1.66–3.01), with a dose–response relationship across ADMA quartiles. Baseline ADMA was not associated with SCA progression; however, dynamic increase in ADMA over time was strongly associated with SCA progression at both 1 and 2 years. Each 1‐SD increase in absolute and relative changes of ADMA was associated with higher risks of SCA progression at 1 year (relative risk [RR] 1.82, 95% CI 1.60–2.07; RR 1.44, 95% CI 1.29–1.61) and 2 years (RR 1.52, 95% CI 1.36–1.70; RR 1.31, 95% CI 1.20–1.42). Associations were more pronounced among participants with pre‐existing SCA. Conclusion: ADMA is associated with the prevalence of SCA, and short‐term dynamic changes in ADMA may represent early indicators for SCA progression in adults with MASLD.

Autophagy is integral to the rapid proliferation of esophageal squamous cell carcinoma (ESCC), and its regulation presents a promising avenue for therapeutic intervention. Recent studies have elucidated the interplay between autophagy and glucose metabolism, while there is a paucity of anticancer drugs that concurrently target these 2 biological processes. In this study, we identified a natural compound, p -hydroxylcinnamaldehyde (CMSP), originally isolated from Cochinchina momordica seed (CMS) by our research team, which exhibits substantial anticancer activity against ESCC in both in vitro and in vivo models. The study demonstrates that CMSP induces apoptosis in ESCC cell lines and patient-derived organoid (PDO) models by disrupting autophagic flux. Mechanistically, CMSP specifically binds to the glycolytic enzyme LDHA in the cytoplasm, hindering its phosphorylation by blocking its membrane translocation and thereby disrupting its interaction with FGFR1. This inhibition results in decreased lactate production from glycolysis, reduced lysosomal acidity, and suppression of the AMPK/mTOR pathway, ultimately resulting in the blockade of autophagy and the induction of apoptosis. Furthermore, in vivo studies underscore the potential clinical application of CMSP in ESCC by disrupting autophagy. In summary, we propose a novel therapeutic strategy for the precision treatment of ESCC by simultaneously targeting glycolysis-mediated autophagy.

Venous thromboembolism (VTE) is the second‐leading cause of cancer‐associated mortality. Neutrophil extracellular trap formation (i.e., NETosis) is a crucial process in forming VTE in cancer patients. Nevertheless, how cancer cells contribute to NETosis remains unclear. This study investigated the potential activation effects of cancer cell‐derived extracellular vesicles (CC‐EVs) on neutrophils. Both small and large EVs (sEVs and lEVs) released from cancer cells are found to significantly induce NETosis in neutrophil‐like HL‐60 (dHL‐60) cells. Following an in‐depth exploration of EV‐induced NETosis, the specific molecular pathways involved in this biological process are elucidated. CYBA enriched in CC‐lEVs is delivered to dHL‐60, leading to a rapid increase in intracellular ROS levels and upregulation of citH3 expression. This cascade resulted in decondensed chromatin release and subsequent NETosis along with elevated MPO‐DNA levels. Injection of CC‐lEVs into mice caused more pronounced VTE, which is accompanied by increased peripheral blood levels of the MPO‐DNA and thrombin‐antithrombin complex. Inhibiting CYBA expression or ROS generation prevented NETosis in vitro and significantly reduced VTE in vivo. In conclusion, CC‐lEVs induce NETosis through the CYBA‐ROS‐citH3 pathway and increase VTE risk. Targeting CYBA expression or ROS production can provide novel strategies for preventing and treating VTE in high‐risk cancer patients.