首都医科大学

BACKGROUND:The prognosis for esophageal squamous cell carcinoma (ESCC) patients with lymph node metastasis (LNM) is still dismal. Elucidation of the LNM associated genomic alteration and underlying molecular mechanisms may provide clinical therapeutic strategies for ESCC treatment. METHODS:Joint analysis of ESCC sequencing data were conducted to comprehensively survey SCNAs and identify driver genes which significantly associated with LNM. The roles of miR-548k in lymphangiogensis and lymphatic metastasis were validated both in vitro and in vivo. ESCC tissue and blood samples were analyzed for association between miR-548k expression and patient clinicopathological features and prognosis and diagnosis. RESULTS:In the pooled cohort of 314 ESCC patients, we found 76 significant focused regions including 43 amplifications and 33 deletions. Clinical implication analysis revealed a panel of genes associated with LNM with the most frequently amplified gene being MIR548K harbored in the 11q13.3 amplicon. Overexpression of miR-548k remarkably promotes lymphangiogenesis and lymphatic metastasis in vitro and in vivo. Furthermore, we demonstrated that miR-548k modulating the tumor microenvironment by promoting VEGFC secretion and stimulating lymphangiogenesis through ADAMTS1/VEGFC/VEGFR3 pathways, while promoting metastasis by regulating KLF10/EGFR axis. Importantly, we found that serum miR-548k and VEGFC of early stage ESCC patients were significantly higher than that in healthy donators, suggesting a promising application of miR-548k and VEGFC as biomarkers in early diagnosis of ESCC. CONCLUSIONS:Our study comprehensively characterized SCNAs in ESCC and highlighted the crucial role of miR-548k in promoting lymphatic metastasis, which might be employed as a new diagnostic and prognostic marker for ESCC.

Rationale: Excessive stress (distending pressure), strain (volume deformation), and drop in inspiratory alveolar pressure are proposed mechanisms for patient self-inflicted lung injury. Objectives: To dissect the influence of inspiratory effort, respiratory mechanics, and ventilation mode on lung stress, strain, and drop in inspiratory alveolar pressure; and explore their impact on oxygenation and lung compliance. Methods: International cohort study analyzing respiratory recordings (esophageal pressure) of patients with acute hypoxemic respiratory failure. Association between muscular pressure (Pmus), surrogates of stress (driving trans-alveolar pressure), strain (tidal volume), and inspiratory alveolar pressure relative to PEEP were explored with mixed-models, including interactions for ventilation mode, respiratory system elastance, and synchrony. Association between these and changes in oxygenation and lung compliance were explored. Measurements and main results: 60 patients from 15 centers represented 528 recordings (339,796 breaths). For each cmH(2)O Pmus increase there was an increase in driving trans-alveolar pressure (median[CI 95%] 0.28[0.27-0.29]cmH(2)O) and tidal volume (0.16[0.16-0.17]ml/kg of predicted body weight) and decrease in alveolar pressure (-0.25[0.24-0.6]cmH(2)O, p<0.001). Volume-control ventilation showed less increase in stress and strain surrogates than pressure-targeted modes, but more drop in alveolar pressure (p<0.001, Pmus:mode interaction). Breath-stacking was infrequent and associated with higher stress. Lower inspiratory alveolar pressure relative to PEEP was associated with subsequent worsening oxygenation (p=0.04) and higher stress with worsening lung compliance (p=0.023). Conclusion: Strong efforts are associated with high surrogates for lung stress, strain, and lower inspiratory alveolar pressure relative to PEEP, differently according to the mode of ventilation, being associated with subsequent worsening oxygenation and lung compliance.

Background Anemia can lead to secondary brain damage by reducing arterial oxygen content and brain oxygen supply. Patients with acute brain injury have impaired self-regulation. Brain hypoxia may also occur even in mild anemia. Red blood cell (RBC) transfusion is associated with increased postoperative complications, poor neurological recovery, and mortality in critically ill neurologic patients. Balancing the risks of anemia and red blood cell transfusion-associated adverse effects is challenging in neurocritical settings. Methods We searched the Cochrane Central Register of Controlled Trials (CENTRAL), Embase, and MEDLINE (PubMed) from inception to January 31, 2024. We included all randomized controlled trials (RCTs) assessing liberal versus restrictive RBC transfusion strategies in neurocritical patients. We included all relevant studies published in English. The primary outcome was mortality at intensive care unit (ICU), discharge, and six months. Results Of 5195 records retrieved, 84 full-text articles were reviewed, and five eligible studies were included. There was no significant difference between the restrictive and liberal transfusion groups in ICU mortality (RR: 2.53, 95% CI: 0.53 to 12.13), in-hospital mortality (RR: 2.34, 95% CI: 0.50 to 11.00), mortality at six months (RR: 1.42, 95% CI: 0.42 to 4.78) and long-term mortality (RR: 1.22, 95% CI: 0.64 to 2.33). The occurrence of neurological adverse events and most major non-neurological complications was similar in the two groups. The incidence of deep venous thrombosis was lower in the restrictive strategy group (RR: 0.41, 95% CI: 0.18 to 0.91). Conclusions Due to the small sample size of current studies, the evidence is insufficiently robust to confirm definitive conclusions for neurocritical patients. Therefore, further investigation is encouraged to define appropriate RBC transfusion thresholds in the neurocritical setting.

Wearable silent speech systems hold significant potential for restoring communication in patients with speech impairments. However, seamless, coherent speech remains elusive, and clinical efficacy is still unproven. Here, we present an AI-driven intelligent throat (IT) system that integrates throat muscle vibrations and carotid pulse signal sensors with large language model (LLM) processing to enable fluent, emotionally expressive communication. The system utilizes ultrasensitive textile strain sensors to capture high-quality signals from the neck area and supports token-level processing for real-time, continuous speech decoding, enabling seamless, delay-free communication. In tests with five stroke patients with dysarthria, IT's LLM agents intelligently corrected token errors and enriched sentence-level emotional and logical coherence, achieving low error rates (4.2% word error rate, 2.9% sentence error rate) and a 55% increase in user satisfaction. This work establishes a portable, intuitive communication platform for patients with dysarthria with the potential to be applied broadly across different neurological conditions and in multi-language support systems.

Existing deep learning models for defining pathology from clinical imaging data rely on expert annotations and lack generalization capabilities in open clinical environments. Here we present a generalizable vision–language model for Annotation-Free pathology Localization (AFLoc). The core strength of AFLoc is extensive multilevel semantic structure-based contrastive learning, which comprehensively aligns multigranularity medical concepts with abundant image features to adapt to the diverse expressions of pathologies without the reliance on expert image annotations. We conducted primary experiments on a dataset of 220,000 pairs of image–report chest X-ray images and performed validation across 8 external datasets encompassing 34 types of chest pathology. The results demonstrate that AFLoc outperforms state-of-the-art methods in both annotation-free localization and classification tasks. In addition, we assessed the generalizability of AFLoc on other modalities, including histopathology and retinal fundus images. We show that AFLoc exhibits robust generalization capabilities, even surpassing human benchmarks in localizing five different types of pathological image. These results highlight the potential of AFLoc in reducing annotation requirements and its applicability in complex clinical environments. An annotation-free vision–language model based on contrastive learning performs pathology localization and clinical diagnosis on a wide variety of medical image datasets.