medRxiv : the preprint server for health sciences最新文献

Angioedema is a life-threatening adverse drug reaction associated with renin-angiotensin-aldosterone system (RAAS) inhibitors, characterized by localized swelling in the deep layers of the skin. Well-established evidence indicates an up to fivefold higher incidence of RAAS inhibitor-induced angioedema in self-identified Black patients compared to White patients. The mechanisms underlying this health disparity remain poorly understood and are often attributed to race, a poor proxy for interindividual genetic similarity and social stressors. Here, we investigate the genetic and social determinants of RAAS inhibitor-induced angioedema as well as the etiology of this racial difference. In particular, we (1) discovered OTULINL and CRABP1 as novel loci for RAAS inhibitor-induced angioedema, (2) confirmed the importance of bradykinin for this adverse drug reaction, (3) reported the first significant genome-wide association in self-identified Black participants, (4) identified alcohol use as an important social determinant, (5) demonstrated the strong role of variants enriched in 1000 Genomes African superpopulation-like genomes as the driver of racially differential angioedema risk, and (6) demonstrated the combined role of polygenic effect size and allele frequency differences in explaining these racial differences. Our results suggest that a clinical precision medicine tool may more precisely predict for whom RAAS inhibitors should be avoided (to prevent angioedema) compared to using race. These findings ultimately underscore the value of an evidence-based approach to removing race from treatment guidelines, which carries less potential harm than other removal strategies.

Background: Primary central nervous system (CNS) tumors affect patients' psychological well-being and quality of life. Individualized approaches, such as Managing Cancer and Living Meaningfully (CALM), have shown potential in advanced cancers for improving these outcomes.

Aims: This study assessed the effects and feasibility of CALM delivered remotely to a diverse cohort of patients with a primary CNS tumor.

Methods: Patients completed 3-6 remote CALM sessions focusing on 4 interrelated domains. Depression, death anxiety, attachment style, and quality of life were assessed at study enrollment, 3-months, and 6-months into the intervention.

Results: Of the 19 patients enrolled, 15 (79% retention rate) completed the study. Most patients had a high-grade (47%) tumor, mainly diagnosed in the brain (60%). The median age was 44 years (range, 24-70). Feasibility was demonstrated through adherence to completing outcome questionnaires and a high level of patient satisfaction (100% found it worthwhile). Although no statistically significant changes were seen in depression, death anxiety, attachment anxiety, or quality of life (p > 0.05; g = -0.09 to 0.78) at any measured time, a clinically meaningful decrease in depression was observed at the 6-month point (mean difference = -3.36, p = 0.13) among spine tumor patients.

Conclusions: This study demonstrated that delivering CALM via telehealth is feasible, as evidenced by high compliance, low attrition, and acceptability among patients diagnosed with CNS tumors. The findings indicated meaningful reductions in depressive symptoms among patients with spinal cord tumors. These preliminary positive findings justify further evaluation of the feasibility and effectiveness of CALM in a larger sample.

Trial registration: ClinicalTrials.gov ID NCT04852302.

Background: Gender disparities in academic medicine have been previously reported, but prior bibliometric studies have been limited by small sample sizes and reliance on manual gender annotation methods. These bottlenecks constrain previous analyses to only a small subset of clinical literature. To assess gender-based differences in authorship trends, research impact, and scholarly output over time in clinical research at scale, we hypothesized that large language models (LLMs) can be an effective tool to facilitate systematic bibliometric analysis of academic research trends.

Methods: We conducted a retrospective, cross-sectional bibliometric study evaluating manuscripts published between January 2015 and September 2025 across over 1,000 PubMed-indexed academic medical journals. Over 1 million manuscripts, written by more than 10 million authors across 13 medical specialties, were analyzed. To enable this large-scale study, the genders of manuscript authors were annotated using a scalable LLM-based pipeline compatible with consumer-grade hardware.

Results: We found that the proportion of female principal investigators has increased over time across different medical subspecialties. However, studies led by male authors tended to be published in higher-impact journals and cited more frequently than those led by female authors. We also observed that researchers of the same gender tended to work together when compared to colleagues of the opposite gender.

Conclusions: While our findings revealed persistent gender-based differences in authorship trends, citation practices, and journal placement, we also observed ongoing, meaningful progress in female representation within academic medical research over time. Our results suggest that LLMs can be a powerful tool to scalably and periodically track this continued progress in future academic medical research.

Plain language summary: Academic research is important to advance the field and practice of medicine. To obtain an accurate picture of the differences in medical research and impact between male and female researchers, we leveraged large language models (LLMs) to identify author genders for over one million medical research papers published between 2015 and 2025. We found that the number of women serving as lead researchers has increased over time across many medical specialties. However, important gaps in achieving gender equality in medical research remain. Our study ultimately helps demonstrate that LLMs can help us monitor gender-based trends in academic research in the future.

Background: Functional neurological disorder (FND) is associated with alterations in functional brain networks, yet relationships between peripheral autonomic physiology and brain architecture remain poorly characterized. This pilot study examined associations between cardiac autonomic metrics and resting-state functional connectivity (rsFC) in FND.

Methods: Twenty females with FND and 23 age-matched female psychiatric controls (PCs) completed questionnaires, 10-min resting photoplethysmography recordings, and same-day resting-state fMRI. Interbeat interval (IBI) and heart rate variability (HRV) metrics were extracted. Whole-brain rsFC was quantified using weighted-degree [centrality]. Within-group analyses tested associations between cardiac autonomic metrics and weighted-degree rsFC separately in FND and PC cohorts, adjusting for age, head motion, and antidepressant/β-blocker use - while applying a cluster-wise correction.

Results: Cardiac (IBI and HRV) metrics did not differ between FND and PC cohorts, and these metrics did not correlate with FND symptom severity, somatic symptom burden, affective symptoms, or childhood trauma. In FND, shorter IBI (i.e., faster resting heart rate) correlated with increased weighted-degree rsFC in bilateral supplementary motor area (SMA) and right precentral/superior frontal regions, whereas higher HRV primarily correlated with decreased weighted-degree rsFC in the bilateral SMA, mid-cingulate cortex, and right amygdala, anterior insula, and lateral orbitofrontal cortex. In PCs, autonomic-rsFC associations were more spatially restricted to the anterior/mid-cingulate and SMA.

Conclusion: In FND, individual differences in resting autonomic physiology related to the centrality of brain areas that are part of the central autonomic, salience, and allostatic-interoceptive networks. These findings suggest that the relationship between autonomic physiology and network architecture may be important in FND.

Allocation out of sequence (AOOS) allows organ procurement organizations (OPOs) to bypass the standard match-run to expedite kidney placement and prevent nonuse. Inclusion of all AOOS attempts is vital when attempting to assess impact of AOOS on organ utility, including those attempts that do not lead to successful transplant. We assessed the frequency of AOOS documentation in discarded kidneys. Using Scientific Registry of Transplant Recipients (SRTR) Potential Transplant Recipient (PTR) offer data from 2021-2024, we identified match-runs with at least one discarded kidney. AOOS was defined according to Health Resources and Services Administration (HRSA) guidelines and match runs were stratified by kidney recovery and disposition patterns, focusing on 2024 when AOOS was well established. AOOS coding frequency was assessed within each group and across OPOs. In 2024, only 4.3% of all match-runs with at least one discarded kidney contained evidence of AOOS documentation. Across OPOs, AOOS-coded discards ranged from 0.0% to 17.1% (median 3.9%, IQR [2.7-7.6%]). AOOS documentation among discarded kidneys remains rare and inconsistent, suggesting major data-capture deficiencies when attempting to accurately assess AOOS efforts. Improved AOOS reporting is essential before future expedited allocation pathways can be effectively evaluated or implemented.

Introduction: Changes in mental health symptoms, and their timing in the preclinical period of dementia, are not well characterised.

Methods: We followed 5,495 Whitehall II participants (median age 68.5; 72.1% male) from their mental health symptoms assessment using the Clinical Interview Schedule-Revised (CIS-R) starting in 2012/13 to dementia diagnosis, death, or 2024. Linear mixed effects regression assessed CIS-R score changes preceding dementia. Flexible parametric models estimated associations of mental health symptoms with dementia.

Results: Total CIS-R score increased (2.56 points [0.85-4.27]) in the 12 years preceding dementia. Having any mental health condition was associated with dementia in the short-term (HR at 3 years=4.04 [2.53-6.50]) but not the long-term (HR at 6 years=1.26 [0.63-2.49]). This pattern held for severe mental health conditions, concentration problems, depression, irritability, fatigue, anxiety, and worry.

Discussion: Awareness of mental health symptoms as preclinical indicators of dementia in the short-term may support timely diagnosis of dementia.

Background: Diagnosis of tuberculosis (TB) in people with HIV (PWH) remains difficult. Since the first pathogen-host interaction in TB occurs in the upper airway, host transcriptomic analysis on nasal specimens may identify novel diagnostic biomarkers. We aimed to demonstrate differences in nasal gene expression in PWH and TB disease versus PWH without TB, evaluate the performance of nasal signatures in predicting TB and compare nasal gene profiles with blood.

Methods: We enrolled adults in Uganda with newly diagnosed HIV and symptoms of pulmonary TB disease. We collected nasal cells and blood for RNA sequencing to identify differentially expressed genes (DEGs) and enriched pathways between PWH and TB disease and PWH without TB. Supervised machine-learning of gene expression data was used to predict TB.

Results: 40 PWH were enrolled (median age: 34 years, median CD4 count: 182), including 20 with TB disease and 20 without. We identified 44 nasal DEGs and 238 blood DEGs, with three overlapping DEGs between samples. Models trained using all 44 nasal DEGs had a cross-validated area under the curve between 0.87-0.90 for predicting TB disease. A simplified signature ( SPIB, SHISA2, TESPA1 and CD1B ) met WHO criteria for a TB triage test. Among adults with TB, pathways related to the inflammatory response were downregulated in nasal samples and upregulated in blood.

Conclusion: There are distinct nasal gene expression patterns associated with TB, not seen in blood. Differences in nasal gene expression in PWH who have TB disease, versus those without TB, highlight their potential as diagnostic biomarkers.

Background and aims: Molecular subtyping of pancreatic ductal adenocarcinoma (PDAC) into basal-like and classical has established prognostic and predictive value. However, its use in clinical practice is limited by cost, turnaround time, and tissue requirements, thereby restricting its application in the management of PDAC. We introduce PanSubNet (PANcreatic SUBtyping NETwork), an interpretable deep learning framework that predicts therapy-relevant molecular subtypes directly from standard hematoxylin and eosin (H&E)-stained whole-slide images.

Methods: PanSubNet was developed using data from 1 , 055 patients across two multi-institutional cohorts (PANCAN, n=846; TCGA, n=209) with paired histology and RNA sequencing data. Ground-truth labels were derived using the validated Moffitt 50-gene signature refined by GATA6 expression. The model employs dual-scale architecture that fuses cellular-level morphology with tissue-level architecture, leveraging attention mechanisms for multi-scale representation learning and transparent feature attribution.

Results: On internal validation within PANCAN using five-fold cross-validation, PanSubNet achieved mean area under the receiver operating characteristic curve (AUC) of 88.5% in high-confidence cases, with balanced sensitivity and specificity. External validation on the independent TCGA cohort without fine-tuning demonstrated robust generalizability (AUC 84.0% ). PanSubNet preserved and, in metastatic disease, strengthened prognostic stratification compared to RNA-seq-based labels. Prediction uncertainty linked to intermediate transcriptional states, not classification noise. Model predictions are aligned with established transcriptomic programs, differentiation markers, and DNA damage repair signatures.

Conclusions: By enabling rapid, cost-effective molecular stratification from routine H&E-stained slides, PanSubNet offers a clinically deployable and interpretable tool for genetic subtyping. We are gathering data from two institutions to validate and assess real-world performance, supporting integration into digital pathology workflows and advancing precision oncology for PDAC.