上海交通大学医学院

Biodegradable zinc alloys have emerged as promising materials for orthopedics, owing to their favorable mechanical properties and controllable degradation rates. In this study, a series of novel Zn-2Cu-0.8Lix Mg ( x = 0.2, 0.4, and 0.8, in wt.%) alloys was developed to enhance both strength and biocompatibility. The effects of Mg addition on microstructure, mechanical properties, in vitro and in vivo degradation, and biocompatibility were systematically investigated. The microstructure of these as-extruded alloys consists of a matrix of beta-LiZn4 and the second phases of eta-Zn and Mg2 Zn11 . Compared to Zn-2Cu-0.8Li alloy, the addition of Mg improves the mechanical strength of Zn-2Cu-0.8Li- x Mg alloys. Notably, the as-extruded Zn-2Cu-0.8Li-0.4Mg alloy shows a well-balanced combination of mechanical properties, with an ultimate tensile strength of 556.5 f 1.6 MPa and a fracture elongation of 37.1% f 4.3%. Furthermore, the alloy exhibits a favorable degradation rate, with in vitro and in vivo rates of 201 f 6 }m/year and 86 f 6 }m/year, respectively. In vitro cell experiments demonstrated that this alloy extract possesses excellent biocompatibility, and a rat femoral shaft fracture model revealed that it has good bone repair capability. Therefore, these results indicate that Zn-2Cu-0.8Li-0.4Mg alloy is a highly promising candidate material for orthopedic applications. (c) 2026 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.

This review delves into brain imaging genomics, an interdisciplinary field merging brain imaging, genomics, and additional biomarkers with clinical data. The primary aim is to uncover new insights into the brain’s phenotypic, genetic, and molecular characteristics. We emphasize recent advances in genome-wide association studies and transcriptome-wide association studies, especially their integration with MRI-derived phenotypes in humans. These studies are crucial for understanding how various factors influence brain structure and function in normal and pathological states. Furthermore, this review highlights imaging transcriptomics progress in non-human primates, essential for elucidating brain organization and improving animal models evolutionarily to bridge gaps in understanding human disorders. We conclude that brain imaging genomics is set to transform research in neurological and psychiatric disorders, owing to its holistic approach that merges advanced genetic analysis with detailed imaging, will deepen our understanding of the brain, and usher in a new epoch in brain imaging research.

The neurobiological mechanisms underlying the therapeutic effects of meditation therapy in patients with schizophrenia remain poorly understood, therefore, we aim to investigate the relationship between clinical symptoms and the structural changes in brain gray matter in a meditation intervention trial. Han inpatients with schizophrenia admitted to the Shanghai First Civil Affairs Mental Health Centre in 2018 were recruited and randomly assigned to the meditation (Med) or the conventional (CON) treatment group in an eight-months trial. MRI data was collected at the outset, three months, and eight months of treatment. Fifty-eight male subjects completed all evaluations and MRI scans, including 30 in Med and 28 in CON. The findings indicate that meditation can significantly inhibit the extensive regional atrophy of gray matter volume (GMV). A positive correlation was observed between the GMV change rate in the right thalamus and the reduction rate of PANSS positive score (r = 0.3850, p = 0.0028), negative score (r = 0.3789, p = 0.0034) and total score (r = 0.3705, p = 0.0042); The GMV change rate in right insula was positively correlated with PANSS positive score (r = 0.2976, p = 0.0233), negative score (r = 0.3987, p = 0.0019), general score (r = 0.3216, p = 0.0138) and total score (r = 0.3765, p = 0.0035). Additionally, the GMV change rate in the right supplementary motor area was positively correlated with negative scores (r = 0.3633, p = 0.0051). while the right paracentral lobule positively correlated with PANSS general score (r = 0.2791, p = 0.0338). These findings demonstrate the effectiveness of meditation in chronic schizophrenia and suggest it as a promising method for the alleviation of clinical symptoms in schizophrenia.

Tumour infiltrating lymphocytes (TILs) are a key component of the tumour microenvironment. To establish a clinically relevant TILs cut-off for patients with oesophago-gastric (OG) cancer, it is essential to know whether TILs density varies by patient and/or disease characteristics. TILs were quantified as TILs/mm2 (TILs density) by a deep-learning algorithm applied to digitised Haematoxylin/Eosin (H E)-stained biopsies and resection specimens from 4628 patients from nine phase III trials. 4533 patients with TILs density and matched clinicopathological data were included in the final analyses. Associations between TILs density, disease stage, geographical region (UK versus Asia), sex, age, and treatment were analysed. Median TILs density was higher in pre-treatment biopsies from patients with early-stage versus late-stage disease (962 vs 479 TILs/mm2, p < 0.001). Within the same geographical region and disease stage, TILs density was similar across different chemotherapy regimens. In UK-led trials of early-stage disease, post-chemotherapy resections showed higher TILs density than chemotherapy-naïve resections (618 vs 571 TILs/mm2, p = 0.003). TILs density was higher in Asian tumours compared to UK tumours (1419 vs 571 TILs/mm2, p < 0.001). No significant associations were observed with age or sex. This is the largest study to date evaluating TILs density in OG cancer. TILs density varied with stage and geographical region but not by age or sex. These findings may explain enhanced response to immunotherapy observed in published studies of patients with early-stage disease and highlight the need to account for baseline TILs heterogeneity when interpreting TILs as a possible biomarker in future studies.

OBJECTIVE: Extranodal diffuse large B-cell lymphoma (DLBCL) of the oral cavity and maxillofacial region (OC-MR) is rare compared to Waldeyer’s ring (WR) DLBCL, characterized by marked aggressiveness and prognostic heterogeneity. This study aims to develop a special risk stratification model for extranodal DLBCL in OC-MR to overcome the limitations of the conventional International Prognostic Index (IPI) to improve outcome prediction in this high-risk population. METHOD: We conducted a retrospective analysis of 76 OC-MR DLBCL patients (43 extranodal and 33 WR-DLBCL) diagnosed between January 2015 and March 2022. Survival outcomes were assessed using Kaplan-Meier methodology with log-rank testing. Prognostic factors were identified through Cox regression and LASSO analysis, with model validation via concordance index (C-index). RESULT: Patients with extranodal OC-MR DLBCL demonstrated significantly greater multisite involvement (p < 0.001) and worse clinical outcomes compared to WR-DLBCL, including shorter median overall survival (17 vs. 26 months, p = 0.0329), shorter progression free survival (13 vs. 25 months, p = 0.0136) and lower objective response rates (67.4% vs. 84.4%, p = 0.003). Hypoalbuminemia (< 37 g/L), elevated C-reactive protein (> 5.83 mg/L), and high Ki67 index (> 80%) predicted poor prognosis. These clinicopathological biomarkers were integrated with IPI to develop the novel exploratory KACIPI model, which demonstrated a potentially improved discriminative ability than IPI (C-index 0.768 vs. 0.680, p < 0.001) and effectively stratified extranodal OC-MR DLBCL risk groups (p < 0.05). CONCLUSION: Extranodal OC-MR DLBCL exhibits distinct clinicopathological features and inferior survival outcomes. The exploratory KACIPI model may provide improved risk prediction for extranodal OC-MR DLBCL by incorporating tumor proliferation, inflammatory markers, nutritional status into conventional IPI score, enabling more accurate identification of high-risk patients.

Venetoclax plus hypomethylating agents (HMAs) is a standard therapy for older or unfit patients with acute myeloid leukemia (AML); however, some patients exhibit suboptimal responses, potentially associated with T-cell exhaustion. Our preclinical findings that statins enhance HMA efficacy by boosting anti-tumor T-cell responses prompted us to translate this strategy to the clinic. A multicenter phase II clinical trial (ChiCTR 2500111931) was conducted to evaluate the efficacy and safety of adding rosuvastatin to venetoclax and azacitidine (venetoclax-azacitidine) in older/unfit AML patients. After induction therapy with this triple combination, the cohort achieved a complete response (CR) rate of 55.5

Obsessive-compulsive disorder (OCD) affects 2–3 https://clinicaltrials.gov/study/NCT04086446 .

Patients with intracerebral hemorrhage (ICH) are at high risk of venous thromboembolism (VTE). Current risk assessment tools are limited and not tailored for neurocritical care populations. This study aimed to develop and validate machine learning-based models to predict VTE in ICH patients. Clinical data of 872 ICH patients admitted to the Neurosurgical ICU of Huashan Hospital from June 2018 to July 2023 were analysed. After univariate analysis, feature selection was performed using Random Forest Importance Ranking and LASSO regression. Three machine learning models (random forest, logistic regression, and LASSO logistic regression) were trained using 10-fold cross-validation. The dataset was randomly split into training (80%) and validation (20%) sets. Model performance was evaluated using accuracy, sensitivity, specificity, F1 score, AUC, calibration curves, and decision curve analysis. Among 421 patients included in the final analysis, 215 (51.1%) developed VTE. Five independent predictors (BMI, D-dimer, homocysteine, triglycerides, albumin) were identified. All three models showed strong discriminatory performance, with the random forest model achieving the highest AUC (0.98) and PR-AUC (0.98), followed by logistic regression (AUC 0.94, PR-AUC 0.91) and LASSO-LR (AUC 0.93, PR-AUC 0.91). Machine learning-based models incorporating metabolic and clinical predictors can accurately stratify VTE risk in ICH patients. The random forest model demonstrated superior performance and clinical applicability, highlighting potential for guiding early prophylactic interventions.

Abstract Objectives To investigate the role of super-resolution contrast-enhanced ultrasound (SR-CEUS) in evaluating inflammatory activity in Crohn’s disease (CD). Materials and methods In this prospective study, we consecutively enrolled CD patients confirmed by clinical and ileocolonoscopic findings. All patients underwent B-mode ultrasound (BMUS), color Doppler flow imaging (CDFI), CEUS, and SR-CEUS within 1 week of ileocolonoscopy. SR-CEUS quantitative parameters were recorded, with simple endoscopic score for Crohn’s disease (SES-CD) as the reference standard. Diagnostic performance was evaluated using receiver operating characteristic (ROC) curve analysis. Results 52 consecutive CD patients were categorized into active (SES-CD ≥ 3, n = 30) and inactive (SES-CD < 3, n = 22) groups. SR-CEUS clearly visualized the intramural microvascular architecture of the bowel wall. SR-CEUS yielded an AUC of 0.903 with 86.4% sensitivity (95% CI: 66.7–95.3%), and 86.7% specificity (95% CI: 70.3–94.5%) for assessing inflammatory activity, significantly outperforming both CDFI (p = 0.014) and CEUS (p = 0.045), while showing no statistically significant difference in comparison with BMUS (p = 0.988). Furthermore, the combination of BMUS and SR-CEUS achieved an AUC of 0.967 for diagnosing active CD, with 100% sensitivity (95% CI: 85.1–100%) and 86.7% specificity (95% CI: 70.3–94.7%), which was significantly superior to BMUS alone (p = 0.038). Conclusions SR-CEUS provides quantitative microvascular perfusion maps that display vascular density, flow velocity, and direction, offering a non-invasive tool for evaluating inflammatory activity in CD. Critical relevance statement This study demonstrates that super-resolution contrast-enhanced ultrasound (SR-CEUS) provides a novel, non-invasive approach for quantitative evaluation of inflammatory activity in Crohn’s disease (CD), which serves as a valuable supplement or alternative to endoscopy in routine monitoring. Key Points An unmet need remains for accurate, non-invasive tools to assess CD activity. SR-CEUS outperforms conventional CDFI and CEUS in distinguishing active from inactive CD. Combining SR-CEUS with standard BMUS yields excellent diagnostic accuracy, establishing this combined approach as a promising non-invasive alternative for monitoring inflammatory activity in CD patients. Graphical Abstract

This study aims to compare the efficacy and safety of catheter-directed mechanical thrombectomy using the Tendvia system versus systemic thrombolysis in patients with intermediate-to-high-risk acute pulmonary embolism. This is a 1:1 block-randomized, controlled, open-label parallel-group, multi-center, superiority trial. Eligible patients aged 18–75 years, diagnosed with intermediate-to-high-risk acute pulmonary embolism and hemodynamic deterioration, will be included. The main exclusion criteria are unsuitable target vessel criteria; sustained systolic hypotension; severe pulmonary hypertension; hematocrit < 28

The rapid development of Vision Foundation Models (VFMs), particularly Vision Transformers (ViT) and Segment Anything Model (SAM), has sparked significant advances in the field of medical image analysis. These models have demonstrated exceptional capabilities in capturing long-range dependencies and achieving high generalization in segmentation tasks. However, adapting these large models to medical image analysis presents several challenges, including domain differences between medical and natural images, the need for efficient model adaptation strategies, and the limitations of small-scale medical datasets. This paper reviews the state-of-the-art research on the adaptation of VFMs to medical image segmentation, focusing on the challenges of domain adaptation, model compression, and federated learning. We discuss the latest developments in adapter-based improvements, knowledge distillation techniques, and multi-scale contextual feature modeling, and propose future directions to overcome these bottlenecks. Our analysis highlights the potential of VFMs, along with emerging methodologies such as federated learning and model compression, to revolutionize medical image analysis and enhance clinical applications. The goal of this work is to provide a comprehensive overview of current approaches and suggest key areas for future research that can drive the next wave of innovation in medical image segmentation.

Aim To develop effective prevention measures, a deep understanding of the evolution patterns and trends of pancreatitis burden is essential. Our study aims to quantify the burden related to pancreatitis in 204 countries and regions from 1990 to 2021. Methods Data related to pancreatitis were derived from the Global Burden of Disease Study in 2021. The burden of pancreatitis was assessed using incidence, disability-adjusted life years (DALYs), deaths and their corresponding age-standardized rates (ASRs), stratified by age, sex, Sociodemographic Index (SDI) and Human Development Index (HDI). The estimated annual percentage change was used to quantify the variation in ASRs. The analysis covered the period from 1990 to 2021. Results In 2021, there were 2,741,736 new cases of pancreatitis (95% UI 2,413,878-3,133,076), leading to 122,416 deaths (95% UI 109,848-141,362), accounting for 0.22% of global deaths, causing a loss of 4,101,154 DALYs (95% UI 3,647,631-4,684,283). The burden of pancreatitis in 2021 and its trends from 1990 to 2021 showed substantial differences based on sex, SDI quintiles and geographical regions. Conclusion Based on our results, the burden of pancreatitis is high and increasing among males and the elderly. Countries with high levels of SDI bear a greater disease burden. Forecast analysis predicts that by 2050, the number of deaths and DALYs related to pancreatitis will continue to rise. Understanding the disease burden and future burden trends associated with pancreatitis is crucial for implementing effective interventions to alleviate the global burden.

Vision-language foundation models have shown great promise in computational pathology but remain primarily data-driven, lacking explicit integration of medical knowledge. We introduce knowledge-enhanced pathology (KEEP), a foundation model that systematically incorporates disease knowledge into pretraining for cancer diagnosis. KEEP leverages a comprehensive disease knowledge graph encompassing 11,454 diseases and 139,143 attributes to reorganize millions of pathology image-text pairs into 143,000 semantically structured groups aligned with disease ontology hierarchies. This knowledge-enhanced pretraining aligns visual and textual representations within hierarchical semantic spaces, enabling a deeper understanding of disease relationships and morphological patterns. Across 18 public benchmarks (over 14,000 whole-slide images) and 4 institutional rare cancer datasets (926 cases), KEEP consistently outperformed existing foundation models, showing substantial gains for rare subtypes. These results establish knowledge-enhanced vision-language modeling as a powerful paradigm for advancing computational pathology.

Chest X-ray (CXR) imaging is one of the most widely used diagnostic modalities in clinical practice, encompassing a broad spectrum of diagnostic tasks. Recent advancements have seen the extensive application of reasoning-based multimodal large language models (MLLMs) in medical imaging to enhance diagnostic efficiency and interpretability. However, existing multimodal models predominantly rely on ”one-time” diagnostic approaches, lacking verifiable supervision of the reasoning process. This leads to challenges in multi-task CXR diagnosis, including lengthy reasoning, sparse rewards, and frequent hallucinations. To address these issues, we propose CX-Mind, the first generative model to achieve interleaved ”think-answer” reasoning for CXR tasks, driven by curriculum-based reinforcement learning and verifiable process rewards (CuRL-VPR). Specifically, we constructed an instruction-tuning dataset, CX-Set, comprising 708,473 images and 2,619,148 samples, and generated 42,828 high-quality interleaved reasoning data points supervised by clinical reports. Optimization was conducted in two stages under the Group Relative Policy Optimization framework: initially stabilizing basic reasoning with closed-domain tasks, followed by transfer to open-domain diagnostics, incorporating rule-based conditional process rewards to bypass the need for pretrained reward models. Extensive experimental results demonstrate that CX-Mind significantly outperforms existing medical and general-domain MLLMs in visual understanding, text generation, and spatiotemporal alignment, achieving an average performance improvement of 25.1% over comparable CXR-specific models. On a real-world clinical dataset (Rui-CXR), CX-Mind achieves a mean recall@1 across 14 diseases that substantially surpasses the second-best results, with multi-center expert evaluations further confirming its clinical utility across multiple dimensions. CX-Mind establishes a new paradigm for constructing interpretable, and high-performing medical MLLMs.

Miyoshi myopathy is an autosomal recessive distal myopathy resulting from pathogenic variants in the DYSF gene encoding dysferlin. This study aimed to characterize a novel biallelic DYSF gene mutation as a cause of Miyoshi myopathy through comprehensive clinical, genetic and pathological analysis. We evaluated serum muscle enzymes and electromyography for myopathic features and performed muscle magnetic resonance imaging to characterize the pattern of involvement. We then conducted whole-exome sequencing with Sanger validation to identify potential causative variants and obtained a muscle biopsy for pathological confirmation. A 19-year-old male presented with progressive bilateral lower limb weakness and symmetric posterior calf muscles atrophy. Markedly elevated creatine kinase (18053 U/L) and myopathic electromyography supported a diagnosis of distal myopathy. Muscle imaging demonstrated severe gastrocnemius atrophy without inflammation. Whole-exome sequencing identified a novel homozygous nonsense variant in exon 20 of the DYSF gene (c.1851 C > G, p.Tyr617Ter), producing a premature stop codon. Muscle biopsy confirmed absent sarcolemmal dysferlin expression, supporting a loss-of-function mechanism. We reported a novel pathogenic variant in the DYSF gene, expanding the mutational spectrum of Miyoshi myopathy and reinforcing the role of biallelic DYSF mutations in disease inheritance.

Long-form clinical videos are central to visual evidence-based decision-making, with growing importance for applications such as surgical robotics and related settings. However, current multimodal large language models typically process videos with passive sampling or weakly grounded inspection, which limits their ability to iteratively locate, verify, and justify predictions with temporally targeted evidence. To close this gap, we propose , a tool-using clinical video reasoning model that performs coarse-to-fine evidence seeking over long-form procedures. By interleaving intermediate reasoning with targeted tool calls and verification on retrieved observations, MedScope produces more accurate and trustworthy predictions that are explicitly grounded in temporally localized visual evidence. To address the lack of high-fidelity supervision, we build , an evidence-centric, fine-grained clinical video suite. We then optimize with rounding-ware roup elative olicy ptimization (), which directly reinforces tool use with grounding-aligned rewards and evidence-weighted advantages. On full and fine-grained video understanding benchmarks, achieves state-of-the-art performance in both in-domain and out-of-domain evaluations. Our approach illuminates a path toward medical AI agents that can genuinely “think with videos” through tool-integrated reasoning. We will release our code, models, and data.

Recent breakthroughs in artificial intelligence through foundation models and agents have accelerated the evolution of computational pathology. Demonstrated performance gains reported across academia in benchmarking datasets in predictive tasks such as diagnosis, prognosis, and treatment response have ignited substantial enthusiasm for clinical application. Despite this development momentum, real world adoption has lagged, as implementation faces economic, technical, and administrative challenges. Beyond existing discussions of technical architectures and comparative performance, this review considers how these emerging AI systems can be responsibly integrated into medical practice by connecting deployable clinical relevance with downstream analytical capabilities and their technical maturity, operational readiness, and economic and regulatory context. Drawing on perspectives from an international group, we provide a practical assessment of current capabilities and barriers to adoption in patient care settings.

Fusobacterium nucleatum tumor infiltration is a hallmark of colorectal cancer (CRC) progression. However, whether and how other microbes influence F. nucleatum in CRC remains unclear. Here, we discover that the commensal fungus Candida albicans synergizes with F. nucleatum to promote CRC pathogenesis and contributes to poor patient outcomes. Co-inoculation of C. albicans and F. nucleatum accelerates CRC progression compared to either microbe alone in animal models. C. albicans strains isolated from CRC patients exhibiting enhanced hyphal morphogenesis and elevated FLO9 expression promote F. nucleatum-mediated tumor growth. Mechanistically, through the Flo9-RadD interaction, C. albicans facilitates F. nucleatum localization to the colonic mucosa and promotes F. nucleatum-driven malignant transformation. Translationally, we demonstrate that L-arginine disrupts the interaction between C. albicans and F. nucleatum, reducing their synergistic pro-tumor activity in CRC models in vivo. These findings highlight fungal-bacterial interaction as a critical etiologic mechanism in CRC and a potential therapeutic target.

Surgical scene understanding demands not only accurate predictions but also interpretable reasoning that surgeons can verify against clinical expertise. However, existing surgical vision-language models generate predictions without reasoning chains, and general-purpose reasoning models fail on compositional surgical tasks without domain-specific knowledge. We present Surg-R1, a surgical Vision-Language Model that addresses this gap through hierarchical reasoning trained via a four-stage pipeline. Our approach introduces three key contributions: (1) a three-level reasoning hierarchy decomposing surgical interpretation into perceptual grounding, relational understanding, and contextual reasoning; (2) the largest surgical chain-of-thought dataset with 320,000 reasoning pairs; and (3) a four-stage training pipeline progressing from supervised fine-tuning to group relative policy optimization and iterative self-improvement. Evaluation on SurgBench, comprising six public benchmarks and six multi-center external validation datasets from five institutions, demonstrates that Surg-R1 achieves the highest Arena Score (64.9

IntroductionHepatocellular carcinoma (HCC) is a leading cause of cancer-related death with limited treatment options. Dihydrodiol dehydrogenase (DHDH), an enzyme involved in D-xylose metabolism, has unclear roles in tumorigenesis and immune regulation. This study aims to investigate the clinical significance, biological functions, and immunomodulatory mechanisms of DHDH in HCC, and to explore the therapeutic potential of targeting its metabolic activity.MethodsDHDH expression and its clinical correlation were analyzed using TCGA-LIHC data and validated in HCC tissue microarrays. In vitro functional assays were performed using DHDH-overexpressing and knockdown Hepa1–6 cells. Immune interactions were assessed via co-culture with CD8+ T cells and flow cytometry. Subcutaneous tumor models in immunodeficient and immunocompetent mice, alongside HCC organoid models, were used to evaluate tumor growth and immune microenvironment changes. The therapeutic effect of D-xylose, alone or combined with anti-PD-L1, was examined in vivo.ResultsDHDH was highly expressed in HCC tissues and significantly associated with poor prognosis. Functional studies demonstrated that DHDH overexpression promoted HCC cell proliferation and invasion while suppressing CD8+ T cell activity, potentially through the upregulation of PD-L1 and downregulation of β2-microglobulin (B2M). Further investigations revealed that D-xylose, a metabolic substrate of DHDH, significantly enhanced intracellular NADPH production and reduced ROS levels, thereby alleviating oxidative stress-induced T cell dysfunction. In patient-derived organoid-PBMC co-culture models, D-xylose treatment markedly enhanced T cell immune activity.DiscussionDHDH drives HCC progression and immune evasion by promoting an immunosuppressive microenvironment. Targeting DHDH with D-xylose restores CD8+ T cell function and synergizes with immunotherapy.