BMC Medicine最新文献

Background: Despite rapid advances in medical artificial intelligence (AI), robust evidence for real-world clinical application-particularly in low-resource settings (LRS)-remains limited. To address this gap, we conducted a multicenter evaluation of an AI-based quality control (AI-QC) system for fetal ultrasound images across hospitals in Guizhou Province, China.

Methods: We implemented an independent, post-examination AI-QC system in Guizhou. After image uploaded, the system assigns a 0-100 score and classifies images as standard (≥ 80), basic-standard (60-79), or non-standard (< 60). From September 2020 to May 2025, we prospectively collected ultrasound examinations uploaded by sonographers. Examinations were categorized into four types according to national guideline: first-trimester scan (2 planes), basic biometry scan (3 planes), limited anomaly scan (11 planes), and standard anomaly scan (23 planes). First-trimester and standard anomaly scans represent the highest technical demands. Quality was assessed at two levels: examination level (proportion of required images per examination classified as standard; 100% defined as full-standard); and plane level (proportion of images for a given view classified as standard). Primary outcomes were temporal trends in these two measures.

Results: We analyzed 61,959 examinations (551,144 images) from 186 sonographers at 34 hospitals. Over 36 months, the combined proportion of first-trimester and standard anomaly scans increased from 33.1% to 66.8% (p < 0.0001). The proportion of full-standard examinations increased significantly across all categories: first-trimester scans from 39.5% to 82.1%, basic biometry from 46.3% to 65.5%, limited anomaly from 29.2% to 58.8%, and standard anomaly scans from 16.1% to 53.3% (all p < 0.0001). By 18-24 months post-deployment, most counties surpassed a 60% examination-level standardization threshold; for example, for first-trimester scans, the proportion of counties with mean rates ≥ 60% increased from 31.6% to 68.4% (p for trend < 0.0001). At the plane level, representative views showed improvement; for example, standard transthalamic plane images increased from 91 to 97% (p for trend < 0.0001), accompanied by marked reductions in common deficiencies.

Conclusions: AI-based quality control was associated with improved image quality in LRS, with sustained improvements over time. Future studies linking image quality to diagnostic performance and perinatal outcomes are needed to establish clinical benefit.

Background: Early life exposure to fine particulate matter (PM_2.5) is linked to child development, but limited studies have focused on the specific constituents of PM_2.5, and the potential moderating role of residential greenness remains unknown.

Objectives: The aim of this study is to investigate the effects of constituent-specific PM_2.5 on child development and explore the moderating role of greenness.

Methods: This study included 8327 children aged 1-6 years from a nationwide cross-sectional survey in China. We assessed PM_2.5 and its five compositions across six exposure windows, covering the preconception, prenatal, and the year before the developmental assessment. Developmental quotients (DQ) across five domains were examined using the Children's Developmental Scale of China to assess the child development. Greenness was estimated using the Normalized Difference Vegetation Index within a 1000-m buffer (NDVI_1000m). Multivariable linear regression models were used to evaluate the associations of PM_2.5, its constituents, and child DQ, as well as the effect modification by NDVI_1000m.

Results: PM_2.5 and its constituents were negatively associated with child DQ in motor, adaptability, and social skills. These effects varied by DQ domains and constituents, fine and gross motor domains showed more consistent negative associations, and stronger associations were observed for black carbon (BC), organic matter, and sulfates. Especially, PM_2.5 and all five compositions across six windows were consistently associated with lower fine motor DQ [e.g., β = - 4.66, 95% CI: - 5.45, - 3.87, for a 1-ln μg/m³ increase in PM_2.5 in the year preceding the test]. Additionally, significant interactions were found between PM_2.5, its constituents, and greenness. For example, BC exposure during the year before the test was related to greater decrements in fine motor DQ among children with low NDVI_1000m (β = - 5.01, 95% CI: - 6.01, - 4.00), compared to those with high NDVI_1000m (β = - 1.74, 95% CI: - 3.12, - 0.36).

Conclusions: PM_2.5 and its constituents were inversely linked to child development across motor, adaptability, social domains, with greenness potentially serving as a protective factor.

Background: Parental socioeconomic status is associated with health outcomes across age. However, the specific age-, gender-, and disease-related patterns linking parental income during early childhood to healthcare utilization across age remain poorly characterized. Enhanced understanding of these associations is essential to inform targeted interventions and improve health equity.

Methods: This cross-sectional study analyzed primary care consultations for all Norwegian residents aged 10-59 in 2018 (N = 2,882,669), merged with parental income records from 1958 up until 2017. The analysis was restricted to GP consultations for diseases or disorders, excluding consultations for symptoms and complaints. Analyses were stratified by sex, age, and type of health problem. We also examined how adult income and education mitigate socioeconomic disparities in healthcare utilization.

Results: Individuals from low parental income backgrounds had higher primary care utilization. In the lowest parental income quintile, females averaged 1.89 (SD 3.30) and males 1.24 (SD 2.65) consultations per year, compared to 1.60 (SD 2.94) and 1.00 (SD 2.21), respectively, in the highest quintile. Socioeconomic differences varied by age and disease type. Among females, the largest inequality occurred at age 24, with 1.97 (95% CI 1.89-2.05) consultations in the lowest quintile versus 1.17 (95% CI 1.11-1.23) in the highest. For males, inequality was largest at age 31, with 1.22 (95% CI 1.15-1.30) versus 0.78 (95% CI 0.74-0.83) consultations. Disease-specific differences showed the largest odds ratios for psychological (females: OR 1.54, 95% CI 1.51-1.57; males: OR 1.64, 95% CI 1.60-1.68) and endocrine/nutritional issues (females: OR 1.34, 95% CI 1.32-1.37; males: OR 1.35, 95% CI 1.31-1.38). Adjusting for adult education and income eliminated most disparities, except for musculoskeletal and endocrine/nutritional health problems.

Conclusions: Lower parental income in childhood was associated with higher adult primary care use, particularly for psychological, musculoskeletal, and endocrine conditions, with heterogeneity by age and sex. These associations were attenuated when accounting for individuals' own education and income, suggesting that policies that enhance educational attainment and economic opportunity can help reduce health inequalities.

Background: Evidence on the individual risks of type 1 and type 2 diabetes following gestational diabetes mellitus (GDM) exposure, particularly in large Asian populations, remains limited. We examined whether maternal GDM increases the risk of type 1 and type 2 diabetes in offspring in a Korean nationwide cohort.

Methods: After excluding mothers with preexisting diabetes, we analyzed a nationwide Korean birth cohort of 3,491,680 mother-child pairs from 2009 to 2018, followed for up to 14 years. Type 1 diabetes was defined as International Classification of Diseases, 10th Revision (ICD-10) E10 with insulin prescription; type 2 diabetes as ICD-10 E11-E14 with antidiabetic medication use. Cox proportional hazards models were used to estimate hazard ratios (HRs) across three groups, offspring born to mothers without GDM (reference), with GDM but not treated with insulin, and with GDM treated with insulin, adjusting for maternal and neonatal covariates.

Results: Of the total cohort, 424,185 (12.1%) pregnancies were complicated by GDM, among which 30,003 (7.1%) required insulin therapy during pregnancy. GDM without insulin therapy was not associated with type 1 diabetes in offspring (HR, 0.857 [95% confidence interval, 0.696-1.054]). However, offspring of GDM mothers requiring insulin during pregnancy had a higher risk of type 1 diabetes (HR, 1.936 [1.228-3.052]). For offspring type 2 diabetes, GDM without insulin during pregnancy was significantly associated with increased risk (HR, 1.281 [1.146-1.433]), with a greater risk among insulin-treated GDM pregnancies (HR, 4.329 [3.555-5.270]).

Conclusions: Maternal GDM without insulin therapy is associated with an increased risk of type 2 diabetes, but not type 1 diabetes, in offspring, whereas insulin treatment for GDM during pregnancy is associated with increased risk for both type 1 and type 2 diabetes. These findings underscore the need for individualized postnatal monitoring for offspring of mothers with GDM, with heightened attention for those whose mothers required insulin therapy during pregnancy.

Background: Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease with limited therapeutic options. The circulating metabolome comprises small molecules present in plasma/serum which are the intermediates and end-products of cellular metabolism, and is linked to ALS pathogenesis.

Methods: We conducted hypothesis-free two-sample Mendelian randomisation (MR) analysis of the concentration of 575 plasma/serum metabolites, to determine which are causally linked to risk of ALS. Significant metabolites were validated in an independent GWAS of plasma/serum metabolite concentrations and evaluated for sex-specific effects. Correlations between directly measured patient biofluid metabolite concentrations and ALS risk/severity were examined in 94 ALS patients and 40 controls. We experimentally assessed metabolic function in a murine neurons and human astrocytes carrying an ALS-associated G4C2-repeat expansion within C9orf72.

Results: MR causally associated five metabolites with ALS risk after multiple-testing correction. Higher serum concentration of glycoprotein acetyls (P = 9.7e - 9, β = 0.21) and the peptide DSGEGDFXAEGGGVR (P = 8.0e - 6, β = 0.22) was associated with increased ALS risk, whereas higher plasma concentration of phenylalanylserine, isobutyrylcarnitine, and acetylcarnitine was protective (P < 5e - 5, β = - 0.29 to - 0.72). DSGEGDFXAEGGGVR has been linked to glucose metabolism but we have used genetic fine-mapping to link DSGEGDFXAEGGGVR, neuronal glucose uptake through GLUT3, and ALS risk. Direct measurement of metabolite concentrations in patient biofluids revealed elevated acetylcarnitine levels in patients with ALS, which were associated with delayed symptom onset (Cox regression, P = 0.02, HR = 0.4). Similarly, lactate is elevated in ALS patient CSF (ANOVA, P = 1.3e - 3) and in patients with longer survival time (Cox regression, P = 0.03, HR = 0.3). Plasma fructose is elevated in ALS patients with shorter survival time (Cox regression, P = 0.02, HR = 1.1). In vitro, neurons and astrocytes carrying an ALS-associated G4C2-repeat expansion within C9orf72 demonstrated reduced metabolic flexibility.

Conclusions: We provide evidence that impaired energy substrate availability contributes to ALS risk and severity. CNS cell types differ in their use of energy substrates and therefore we postulate the relative importance of different cell types for different stages of disease. Our findings support further investigation of metabolic interventions to treat or prevent ALS.

Background: Despite significant progress in identifying risk factors for colorectal cancer (CRC), factors influencing survival in people with CRC remain less understood. Pro-inflammatory cytokines like interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) have been implicated in cancer progression and may influence CRC outcomes. We investigated associations between genetically predicted levels of IL-6 and TNF-α signaling pathways and mortality in people with CRC.

Methods: We conducted a two-sample Mendelian randomization (MR) analysis using cis-acting single nucleotide polymorphisms (SNPs) associated with soluble IL-6 receptor alpha (sIL6-RA) and IL-6 signal transducer gp130 (IL6ST), representing IL-6 signaling, and with TNF-α, and its soluble receptors (sTNF-R1, sTNF-R2). SNPs were obtained separately from two large genome-wide association studies (GWAS): deCODE and UK Biobank (UKB). The outcome was CRC-specific mortality among 16,964 CRC cases (4010 deaths) in the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO). Analyses were stratified by tumor site and stage. The inverse variance weighted (IVW) method, incorporating a correlation matrix for dependent SNPs, was used for primary analyses. Because literature links TNF-α to CRC incidence, we additionally performed a simulation study to evaluate the potential impact of collider bias resulting from restricting analyses to CRC cases.

Results: Genetically predicted sIL6-RA was weakly positively associated with CRC-specific mortality (deCODE-SNPs (n = 13) HR per 1 SD increase: 1.06; 95% CI: 1.00-1.12; UKB-SNPs (n = 11) HR: 1.09; 95% CI: 1.02-1.17). Genetically proxied IL6ST levels showed no association with CRC-specific mortality in the overall sample (deCODE-SNPs (n = 19) HR: 1.04; 95% CI: 0.90-1.21; UKB-SNPs (n = 9) HR: 1.11; 95% CI: 0.87-2.42), while higher IL6ST levels were associated with increased mortality among patients with stage 2/3 disease (deCODE-SNPs (n = 19) HR: 1.45; 95% CI: 1.10-1.91; UKB-SNPs (n = 9) HR: 1.87; 95% CI: 1.22-2.89). No associations were observed for TNF-α, sTNF-R1, or sTNF-R2. Findings for all exposures were consistent across both GWAS datasets. Simulation analyses for TNF-α indicated collider bias was present but limited in magnitude.

Conclusions: Our findings suggest that IL-6 signaling may play a role in CRC progression although of limited magnitude, whereas TNF-related pathways appear less relevant for prognosis.

Background: Social frailty, a state of vulnerability to social risk factors, has been linked to adverse physical, psychological, and cognitive outcomes. However, evidence on its role in the progression of multimorbidity across these domains is limited. This study examined social frailty in relation to transitions from a healthy state to physical, psychological, and cognitive conditions and to multimorbidity.

Methods: In this multi-cohort study across 24 countries in the USA, Europe, and Asia, middle-aged and older individuals with no preexisting physical, psychological, or cognitive conditions were included. Social frailty, assessed through baseline questionnaires, was defined across four domains: general resources, social behaviors, social resources, and basic social needs. Physical, psychological, and cognitive conditions were ascertained at baseline and reassessed four times during a mean follow-up of 7 years. Multi-state models and the group-based multi-trajectory model were used to examine the role of social frailty in the development of physical, psychological, and cognitive conditions and their progression to multimorbidity.

Results: Among the 7119 participants (mean age, 59.6 years [SD 8.5]; 3575 [50.2%] women), 778 (10.9%) were socially frail at baseline. During 49,648 person-years at risk, 2981 (41.9%) individuals progressed from a healthy state to one physical, psychological, or cognitive condition, and 1592 (22.4%) progressed to multimorbidity. Compared with socially robust participants, those with social frailty had a higher risk of progressing to any of these conditions (hazard ratio, 1.30; 95% CI, 1.17-1.44) and to multimorbidity (2.10, 1.67-2.65). Four outcome trajectories were identified, including "stably no physical, psychological or cognitive conditions" (48.0%), "increased physical conditions" (16.4%), "increased physical and psychological conditions" (8.7%), and "increased physical and cognitive conditions" (26.9%). Social frailty was associated with more than a twofold increase in risk for the latter three trajectories (odds ratios 2.34, 1.76-3.09, 2.27, 1.63-3.15 and 2.69, 2.11-3.44, respectively). The strongest associations were observed for frailty in the domains of general resources and basic social needs.

Conclusions: Social frailty is associated with faster transitions from a healthy state to physical, psychological, and cognitive conditions and further progression to multimorbidity. Early identification and interventions to address social frailty may help improve health in middle-aged and older adults.

Background: A quantitative biomarker for clinical staging is essential for amyotrophic lateral sclerosis (ALS) stratification. This study evaluated microstructural impairment in brain short association fibers (SAFs) across ALS stages via neurite orientation dispersion and density imaging (NODDI) and assessed correlations with disease severity.

Methods: Diffusion-weighted imaging data were collected from 87 ALS patients (categorized into four groups King's stages) and 37 healthy controls. Whole-brain SAF mapping was performed via a spherical deconvolution-driven probabilistic tractography approach. Diffusion tensor imaging (DTI) and NODDI parameters (neurite density index, NDI; orientation dispersion index, ODI; isotropic volume fraction, ISO) were estimated for each SAF.

Results: Seven SAFs connecting the left postcentral-precentral gyrus, left precentral-precentral gyrus, right postcentral-precentral gyrus, right paracentral-posterior cingulate gyrus, left paracentral-posterior cingulate gyrus, left precentral-superior parietal gyrus, and left precentral-superior frontal gyrus exhibited significant NDI differences across the five groups. Additionally, one fiber connecting the left medial orbitofrontal-rostral anterior cingulate gyrus demonstrated an ISO difference [false discovery rate (FDR)-corrected p < 0.05]. Progressive trends of NDI reduction and ISO increase were observed at higher ALS stages. No intergroup differences were found in the ODI or DTI parameters. The NDI values of these seven SAFs were positively correlated with disease severity scores (FDR-corrected p < 0.05). Combining NDI and ISO revealed moderate classification potential for ALS (area under the curve = 0.780).

Conclusions: Neurite injury in SAFs involving primary motor and extramotor areas worsened alongside clinical staging and motor disability in ALS. NODDI provides quantitative SAF-related biomarkers for assessing ALS disease severity.

Background: Patients with mild stroke (National Institutes of Health Stroke Scale [NIHSS] ≤ 5) and severe intracranial stenosis or occlusion represent a high-risk subgroup for early neurological deterioration (END) and are strong predictors of poor 3-month functional outcomes. The effectiveness of intravenous thrombolysis (IVT) in patients with minor stroke and large vessel severe stenosis or occlusion (LVSSO) remains uncertain. This study aims to assess the comparative efficacy and safety of IVT versus standard medical therapy (SMT) through a systematic review and meta-analysis of both randomized and observational studies.

Methods: A systematic search of PubMed, Embase, Cochrane Library, and Web of Science was conducted to identify studies evaluating IVT in minor stroke with LVSSO from database inception through March 13, 2025. The primary outcome was defined as an excellent functional outcome (modified Rankin Scale [mRS] 0-1) at 90 days. Pooled analyses were performed using odds ratios (ORs) with 95% confidence intervals (CIs) for the overall population and prespecified subgroups. The study protocol was prospectively registered in the PROSPERO international registry (CRD42025637459).

Results: A total of five studies were included, comprising 1090 patients with mild ischemic stroke and LVSSO treated with IVT and 3501 patients treated with SMT. Compared with SMT, IVT significantly improved the 90-day excellent functional outcome (mRS 0-1) (OR, 1.58 [95% CI, 1.16-2.15]). In the large vessel occlusion subgroup, IVT was also significantly associated with an excellent 90-day functional outcome (OR, 1.52 [95% CI, 1.03-2.23]). No significant differences were observed between groups in 90-day favorable functional outcome (mRS 0-2), symptomatic intracranial hemorrhage (sICH), 90-day mortality, or 90-day stroke recurrence.

Conclusions: These findings suggest that IVT may be associated with improved 90-day excellent functional outcomes in patients with mild acute ischemic stroke (AIS) and LVSSO, without increasing the risk of sICH.

Background: No randomized trials or large-scale observational studies directly compared thermal ablation (TA) and hepatectomy (HT) for solitary colorectal liver metastases (SCLM) ≤ 5 cm with long-term follow-up.

Methods: In this multicenter target trial emulation (2009-2024), 1,334 patients with SCLM ≤ 5 cm from 21 Chinese hospitals were enrolled. Propensity score matching (1:1) balanced baseline characteristics, yielding 437 matched pairs. Primary and secondary endpoints were progression-free survival (PFS) and overall survival (OS), respectively. Complications, hospital stay, and costs were analyzed. Sensitivity analyses were performed to evaluate the robustness of the results.

Results: After matching, median PFS was 1.81 years (95% CI 1.60-2.04) for TA vs 1.95 years (1.67-2.34) for HT (HR 0.94, 95% CI 0.82-1.09; P = 0.41). Median OS was 7.22 years (6.43-9.05) for TA vs 8.09 years (6.68-10.58) for HT (HR 0.89, 0.71-1.11; P = 0.30). The Five-year PFS and OS rates were comparable between groups. The PFS rates were 30.05% (95% CI: 25.32%-35.67%) for TA and 32.16% (95% CI: 27.49%-37.63%) for HT, while OS rates were 62.36% (95% CI: 56.24%-69.14%) for TA and 67.79% (95% CI: 62.38%-73.66%) for HT. TA reduced Clavien-Dindo III-IV complications (2.1% vs 5.0%; P = 0.017), hospital stay (median 3 vs 10 days; P < 0.001), and costs (median $4,820 vs $10,239; P < 0.001). Across all predefined subgroups (e.g., tumor size, CRS), treatment with TA resulted in PFS and OS comparable to those achieved with HT.

Conclusions: For SCLM ≤ 5 cm, TA provides comparable oncologic outcomes to HT, alongside superior safety and cost-effectiveness. This supports its role as a first-line therapeutic alternative for selected patients or when surgery is contraindicated.