Category: Statistics

Bioinformatics. 2025 Nov 9:btaf617. doi: 10.1093/bioinformatics/btaf617. Online ahead of print.

ABSTRACT

MOTIVATION: Studies of microbial communities, represented by the relative abundances of taxa at various taxonomic levels, have underscored the significance of microbiota in numerous aspects of human health and disease. A pivotal challenge in microbiome research lies in pinpointing microbial taxa associated with disease outcomes, which could play crucial roles in prevention, detection, and treatment of various health conditions. Alongside these relative abundance data, taxonomic information sometimes offers a unique lens to explore the impact of shared evolutionary histories on patterns of microbial abundance.

RESULTS: In pursuit of this goal, we utilize the tree structure to more flexibly identify taxa associated with disease outcomes. To enhance the accuracy of our selection process, we introduce auxiliary knockoff copies of microbiome features designated as noise. This approach allows for the assessment of false positives in the selection process and aids in refining it towards more precise outcomes. Extensive numerical simulations demonstrate that our methodology outperforms several existing methods in terms of selection accuracy. Furthermore, we demonstrate the practicality of our approach by applying it to a widely used gut microbiome dataset, identifying microbial taxa linked to body mass index.

AVAILABILITY AND IMPLEMENTATION: TCVS R code is available at https://github.com/Yicong1225/TCVS.

PMID:41206954 | DOI:10.1093/bioinformatics/btaf617

Eur J Cardiothorac Surg. 2025 Nov 9:ezaf391. doi: 10.1093/ejcts/ezaf391. Online ahead of print.

ABSTRACT

OBJECTIVES: Left carotid-subclavian bypass (LCSB) is a classic strategy for left subclavian artery (LSA) revascularization in TEVAR patients. Its employment has been reduced in recent years due to the advent of endovascular solutions for LSA management.Data on outcomes of LCSB is lacking, especially for graft-related complications and patency at follow-up.

METHODS: All patients who underwent TEVAR with LCSB from November 2005 to January 2025, in an elective or urgent setting, were retrospectively analysed in terms of pre- and intraoperative characteristics, short- and mid-term outcomes.In-hospital outcomes were compared between the urgent and elective groups. LCSB patency at follow-up imaging was reported. A Kaplan-Meier analysis was performed on survival, freedom from reintervention and LCSB patency.

RESULTS: LCSB was performed in 161 patients, 36 of which (22.3%) were urgent procedures. In-hospital mortality was 3.7%, with no significant difference between the elective and the urgent group (3.2% vs 5.6% respectively, p = 0.491). There was a not statistically significantly higher incidence of stroke in the urgent patients (0.8% vs 5.6%, p = 0.057). LCSB-related complications occurred in 12 patients (7.4%). Overall LCSB patency at last available follow-up was 97.4%. LSA embolization was necessary in 7 cases (4.5%) due to type II endoleak. At 5 years, survival was 87.4%, freedom from reintervention was 88.5% and LCSB patency was 99%.

CONCLUSIONS: Left carotid-subclavian bypass is safe and effective as a LSA revascularization strategy. Even in urgent patients, LCSB was not linked to worse in-hospital outcomes.

PMID:41206953 | DOI:10.1093/ejcts/ezaf391

J Natl Cancer Inst. 2025 Nov 9:djaf308. doi: 10.1093/jnci/djaf308. Online ahead of print.

ABSTRACT

BACKGROUND: The associations between different types of diabetes, characterized by distinct pathophysiology and genetic architecture, and pancreatic ductal adenocarcinoma (PDAC) risk are not understood.

METHODS: We investigated associations of genetic susceptibility to type 2 diabetes (T2D), eight T2D mechanistic clusters, type 1 diabetes (T1D), and maturity-onset diabetes of the young (MODY) with PDAC risk. We used genome-wide association study (GWAS) summary-level statistics for T2D (242,283 cases, 1,569,734 controls), T1D (18,942 cases, 501,638 controls), and PDAC (10,244 cases and 360,535 controls) in individuals of European ancestry.

RESULTS: Two-sample Mendelian randomization (MR) using the Robust Adjusted Profile Score (MR-RAPS) method indicated that genetically predicted T2D was associated with PDAC risk (OR = 1.10; 95% CI 1.05-1.15), particularly the T2D obesity (OR = 1.28; 95% CI 1.15-1.42) and lipodystrophy (OR = 1.25; 95% CI 1.03-1.51) clusters. No association was observed for T1D with PDAC risk (OR = 1.01; 95% CI 0.99-1.02). Pathway/gene-set analysis using the summary-based Adaptive Rank Truncated Product (sARTP) method revealed a significant association between the MODY gene-sets and PDAC risk (P = 1.5 × 10-8), which remained after excluding 20 known PDAC GWAS loci (P = 7.6 × 10-4). HNF1A, FOXA3, and HNF4A were the top contributing genes after excluding the previously identified GWAS loci regions.

CONCLUSIONS: Our results from this genetic association study support that T2D, particularly the obesity and lipodystrophy mechanistic clusters, and MODY genomic susceptibility regions play a role in the etiology of PDAC.

PMID:41206949 | DOI:10.1093/jnci/djaf308

Bioinformatics. 2025 Nov 9:btaf605. doi: 10.1093/bioinformatics/btaf605. Online ahead of print.

ABSTRACT

SUMMARY: Holo-omics is an emerging research area that integrates multi-omic datasets from the host organism and its microbiome to study their interactions. Recently, curated and openly accessible holo-omic databases have been developed. The HoloFood database, for instance, provides nearly 10,000 holo-omic profiles for salmon and chicken under controlled treatments. However, bridging the gap between holo-omic data resources and algorithmic frameworks remains a challenge. Combining the latest advances in statistical programming with curated holo-omic data sets can facilitate the design of open and reproducible research workflows in the emerging field of holo-omics.

AVAILABILITY AND IMPLEMENTATION: HoloFoodR R/Bioconductor package and the source code are available under the open-source Artistic License 2.0 at the package homepage https://doi.org/10.18129/B9.bioc.HoloFoodR.

SUPPLEMENTARY INFORMATION: Available in the package vignette https://ebi-metagenomics.github.io/HoloFoodR/articles/case_study.html.

PMID:41206936 | DOI:10.1093/bioinformatics/btaf605

Bioinformatics. 2025 Nov 9:btaf616. doi: 10.1093/bioinformatics/btaf616. Online ahead of print.

ABSTRACT

MOTIVATION: The most informative genome-wide association studies (GWAS) are meta-analyses that have combined multiple studies to increase the GWAS sample size. Statistical fine-mapping is a key downstream analysis of GWAS to jointly evaluate the probability of causality of all variants in a genomic region of interest. Current fine-mapping methods are miscalibrated in the meta-analysis setting due to variation in sample size across the variants.

RESULTS: We introduce FINEMAP-miss, a new fine-mapping method that extends the FINEMAP model to account for variant-specific missingness. We show that FINEMAP-miss is well-calibrated in meta-analysis simulations where the standard fine-mapping fails. Compared to the summary statistics imputation approach, FINEMAP-miss provides clear improvement when the causal variants have low imputation information or when the sample size or complexity of the meta-analysis setting increase. We successfully apply FINEMAP-miss on a breast cancer GWAS meta-analysis where neither the standard fine-mapping nor the summary statistics imputation are applicable.

AVAILABILITY: An open source implementation of FINEMAP-miss as an R package (“finemapmiss”) is available at https://github.com/JoonasKartau/finemapmiss. The archived version of FINEMAP-miss used for this publication can be found on Zenodo at https://doi.org/10.5281/zenodo.17492622.

SUPPLEMENTARY INFORMATION: Supplementary Data is available at the journal’s web site.

PMID:41206934 | DOI:10.1093/bioinformatics/btaf616

J Natl Cancer Inst. 2025 Nov 9:djaf326. doi: 10.1093/jnci/djaf326. Online ahead of print.

ABSTRACT

INTRODUCTION: Although myeloma represents a key success story in oncology, some drugs have failed to meet primary endpoints in randomized controlled trials (RCTs), despite promising early-phase activity. This analysis aimed to understand factors that increase the likelihood of meeting primary endpoints in myeloma RCTs.

METHODS: Myeloma RCTs published through October 2023 were identified using MEDLINE, PubMed, Embase, and the Cochrane Registry. Studies were classified as head-to-head (substituting one regimen for another) or add-on (adding one drug to existing regimen). Trials were considered successful if they achieved statistical significance for primary outcomes. Logistic regression identified predictors of meeting trial endpoints.

RESULTS: A total of 145 comparisons from 123 RCTs were included. Only two factors were independently associated with meeting primary endpoints in multivariate analysis. Higher median participant age was associated with lower odds of meeting the primary endpoint (OR per one-year increase, 0.90, 95% CI: 0.83-0.98). Overall survival (OS) was the primary endpoint in 20/145 comparisons, of which 3/20 met their endpoint. Selecting OS as primary endpoint was associated with reduced likelihood of success compared with progression-free survival by 94% (OR: 0.06, 95% CI: 0.01-0.23). Head-to-head design was not associated with lower success rates than add-on design (OR: 0.59; 95% CI: 0.22-1.62).

CONCLUSION: Two key factors predicted higher likelihood of meeting endpoints: younger patient age and primary endpoints other than overall survival. Although head-to-head design is considered riskier, it was not associated with decreased success. This analysis aims to better inform clinicians, industry, and regulators in myeloma drug development.

PMID:41206930 | DOI:10.1093/jnci/djaf326

Oncologist. 2025 Nov 9:oyaf375. doi: 10.1093/oncolo/oyaf375. Online ahead of print.

ABSTRACT

BACKGROUND: Blinded independent central review (BICR) mitigates assessment bias in oncology trials but imposes significant operational burdens. Its value in hematologic malignancies-where multimodal response criteria reduce reliance on subjective imaging assessments compared to solid tumors-remains unestablished. This meta-analysis evaluates BICR-investigator concordance specifically in hematology trials.

METHODS: We systematically identified Phase II/III hematology trials (2014-2024) reporting progression-free survival (PFS) and/or objective response rate (ORR) assessments by both investigators and BICR from PubMed. Agreement was quantified using Pearson/Spearman correlation, pooled hazard ratio ratio (HRR, HRINV/HRBICR) for PFS, and odds ratio ratio for ORR (OddsRR, ORINV/ORBICR). We also analyzed the odds ratio for ORR for single arms (OddsINV/OddsBICR). Subgroup analyses assessed the impact of masking, cancer type based on imaging dependence, and sample size.

RESULTS: Data from 70 studies (37 PFS comparisons; 23 ORR comparisons; 29 single-arm ORR) were analyzed. For PFS, the pooled HRR was 0.96 (95% CI: 0.89,1.03), with perfect agreement in statistical significance (Cohen’s kappa = 1). For ORR, the pooled OddsRR was 0.99 (95% CI: 0.85, 1.14). Single-arm trials showed minimal odds difference between assessors (OR = 1.02, 95% CI: 0.90, 1.17). Subgroup analyses (masking, cancer type, sample size) consistently showed high agreement.

CONCLUSIONS: Investigator and BICR assessments demonstrated substantial concordance in hematology trials. The common applications of BICR in registration trials provide minimal added value for primary endpoint validation in this setting. We recommend prioritizing investigator training and standardized criteria to optimize resource allocation.

PMID:41206920 | DOI:10.1093/oncolo/oyaf375

J Am Med Inform Assoc. 2025 Nov 9:ocaf193. doi: 10.1093/jamia/ocaf193. Online ahead of print.

ABSTRACT

OBJECTIVE: Stigmatizing language (SL) in Electronic Health Records (EHRs) can perpetuate biases and negatively impact patient care. This study introduces a novel method for automatically detecting such language to improve healthcare documentation practices.

MATERIALS AND METHODS: We developed a multi-stage transfer learning framework integrating semantic, syntactic, and task adaptation using three datasets: hate speech, clinical phenotypes, and stigmatizing language. Experiments were conducted on stigmatizing language dataset which consists of 4,129 de-identified EHR notes (72.7% stigmatizing, 27.3% non-stigmatizing), split 80/20 for training and testing. Longformer, BERT, and ClinicalBERT models were evaluated, and model performance was assessed on 35 randomized subsets of the test set (each comprising 70% of test data). The Wilcoxon-Mann-Whitney test was used to evaluate statistical significance, with Bonferroni correction applied to control for multiple hypothesis testing. Baseline models included zero-shot and few-shot GPT-4o, Support Vector Machine, Random Forest, Logistic Regression, and Multinomial Naive Bayes.

RESULTS: The proposed framework achieved the highest accuracy, with fully adapted Longformer reaching 89.83%. Performance improvements remained statistically significant after Bonferroni correction compared to all baselines (p < .05). The framework demonstrated robust gains across different stigmatizing language types.

DISCUSSION: This study underscores the value of domain-adaptive NLP for detecting stigmatizing language in EHRs. The multi-stage transfer learning framework effectively captures subtle biases often missed by conventional models, enabling more objective and respectful clinical documentation.

CONCLUSION: This framework offers a statistically validated, high-performing framework for detecting stigmatizing language in EHRs, supporting responsible AI and promoting equity in clinical care.

PMID:41206907 | DOI:10.1093/jamia/ocaf193

JAMA. 2025 Nov 9. doi: 10.1001/jama.2025.21077. Online ahead of print.

ABSTRACT

IMPORTANCE: Risk stratification strategies in primary prevention of coronary events lack precision.

OBJECTIVE: To determine whether prediction of first coronary events is improved by adding information on coronary atherosclerosis from coronary computed tomography angiography (CCTA) to a model using the pooled cohort equation (PCE) risk score tool and the coronary artery calcification score (CACS).

DESIGN, SETTING, AND PARTICIPANTS: Observational cohort study including individuals aged 50 to 64 years randomly recruited from the general population and examined at 6 university hospitals in Sweden from 2013 to 2018, with a median follow-up of 7.8 years. A sample of 30 154 individuals underwent cardiopulmonary imaging, physical examinations, routine laboratory tests, questionnaires, and/or functional tests. This study included 24 791 individuals without previous cardiovascular disease for whom high-quality CCTA images were available. Events were followed up via registers until September 2024.

EXPOSURES: The information used from the CCTA images was the extent of coronary atherosclerosis (segment involvement score), presence of noncalcified atherosclerosis, and presence of coronary obstructive disease (stenosis ≥50%).

MAIN OUTCOMES AND MEASURES: The outcome was a composite of first occurrence of nonfatal myocardial infarction or death from coronary heart disease.

RESULTS: During follow-up, 304 coronary events occurred. Segment involvement scores of 3 to 4 and greater than 4 and presence of noncalcified atherosclerosis were associated with hazard ratios of 2.71 (95% CI, 1.34-5.44), 5.27 (95% CI, 2.50-11.07), and 1.66 (95% CI, 1.23-2.22), respectively. In a model based on the PCE and CACS, CCTA-derived data improved risk discrimination (C statistic improved from 0.764 to 0.779; P = .004) and risk reclassification (net reclassification improvement of 0.133 [95% CI, 0.031-0.165]), conferred a net correct upward reclassification of 14.2% in those with events and incorrectly classified 1.6% of participants not experiencing an event into a higher-risk category. Because of the low event rate in the cohort, reclassification mainly occurred in the group classified as at low risk (<5%) according to the PCE.

CONCLUSIONS AND RELEVANCE: Information on coronary atherosclerosis from CCTA modestly improved risk prediction beyond traditional risk factors and CACS in identifying individuals at risk of coronary events and in need of primary prevention.

PMID:41206900 | DOI:10.1001/jama.2025.21077

Int J Speech Lang Pathol. 2025 Nov 9:1-20. doi: 10.1080/17549507.2025.2582522. Online ahead of print.

ABSTRACT

PURPOSE: This systematic review examined the effectiveness and instructional characteristics of morphological awareness interventions for children aged 3 to 12 diagnosed with Developmental Language Disorder, in English- or Spanish-speaking contexts. It aimed to determine the outcomes of MA interventions and identify the instructional components most supported by evidence.

METHOD: Following PRISMA guidelines, a systematic search was conducted across five major databases (PsycINFO, PubMed, SCOPUS, Web of Science, and ERIC) between September and November 2024. Ten studies published between 1980 and 2024 met the inclusion criteria. The review synthesised results concerning both intervention outcomes and instructional features, using Frizelle and McKean’s (2022) Dose Form Framework.

RESULT: Only four studies reported statistically significant improvements in morphological awareness, with large effect sizes, indicating potential benefits of morphological awareness interventions for children with Developmental Language Disorder. However, many studies lacked rigorous methodology or failed to disaggregate outcomes for children with Developmental Language Disorder, limiting generalizability. Interventions with the strongest evidence base were those using explicit, clinician-directed instruction targeting affix identification and word construction.

CONCLUSION: The findings suggest promising but limited evidence supporting the effectiveness of morphological awareness interventions for children with Developmental Language Disorder. Further high-quality, controlled studies with detailed statistical reporting and consistent diagnostic labelling are needed. Structured, explicit instruction appears to be a critical feature of effective programs and should be prioritised in clinical and educational practice.

PMID:41206861 | DOI:10.1080/17549507.2025.2582522