Long-Term Enoxaparin Use and Osteoporosis Risk: A Real-World Cohort Study with Integrative Computational and Network Toxicology Approaches

Long-Term Enoxaparin Use and Osteoporosis Risk: A Real-World Cohort Study with Integrative Computational and Network Toxicology Approaches | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Long-Term Enoxaparin Use and Osteoporosis Risk: A Real-World Cohort Study with Integrative Computational and Network Toxicology Approaches Zhen-jiang Liu, Xiao-min Li, Fei Liu, Zhi-meng Zhang, Chao-lu Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8842742/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 07 Apr, 2026 Read the published version in Journal of Translational Medicine → Version 1 posted 4 You are reading this latest preprint version Abstract Background Enoxaparin sodium is a widely used low-molecular-weight heparin for thromboprophylaxis and anticoagulation. Although it is generally considered to have a lower skeletal toxicity profile than unfractionated heparin, emerging evidence suggests that prolonged exposure may adversely affect bone metabolism. However, robust real-world evidence and mechanistic insights linking long-term enoxaparin use to osteoporosis remain limited. Methods We conducted a large retrospective cohort study using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, including 23,852 adult patients with documented enoxaparin exposure and complete follow-up. The association between enoxaparin exposure duration and osteoporosis risk was evaluated using multivariable logistic regression, subgroup analyses, restricted cubic spline modeling, and causal mediation analysis. To explore potential molecular mechanisms, we integrated network toxicology, transcriptomic analysis, weighted gene co-expression network analysis (WGCNA), machine learning–based feature selection, and molecular docking. Results Prolonged enoxaparin exposure was significantly associated with an increased risk of osteoporosis in a duration-dependent and nonlinear manner. Patients exposed for more than 90 days had a much higher risk of osteoporosis compared to those exposed for less than 15 days. This risk remained strong even after demographic factors, clinical covariates, and dosing intensity were taken into account. Restricted cubic spline analysis confirmed a significant nonlinear exposure–response relationship. Mediation analyses indicated that dosing frequency and daily dose partially mediated this association, while exposure duration remained the predominant driver. Network toxicology and enrichment analyses implicated oxidative stress, inflammatory signaling, apoptosis, and osteoclast differentiation pathways. Integrative WGCNA and machine learning identified CDK16 and VHL as core regulatory genes. Molecular docking demonstrated stable binding affinities between enoxaparin and both targets, supporting their potential involvement in enoxaparin-associated bone dysregulation. Conclusion Long-term enoxaparin use is associated with an increased risk of osteoporosis, exhibiting clear duration-dependent and nonlinear characteristics. Integrating real-world epidemiologic evidence with systems-level network toxicology highlights CDK16- and VHL-centered pathways as potential mechanistic mediators. Enoxaparin Osteoporosis Network toxicology Real-world evidence MIMIC-IV database Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 1. Introduction Osteoporosis is a systemic skeletal disorder characterized by reduced bone mass and impaired bone microarchitecture, with its primary clinical outcomes being increased bone fragility and significantly elevated fracture risk [ 1 ]. The accelerating pace of global population aging has led to a persistent rise in the incidence of osteoporosis [ 2 , 3 ]. Statistics indicate that approximately 200 million people worldwide suffer from osteoporosis, affecting about 30% of postmenopausal women and 20% of men over 65 years of age [ 4 ]. Osteoporosis has become a serious public health issue, substantially increasing disability rates, mortality, and healthcare costs among the elderly and those with chronic diseases [ 5 ]. Osteoporotic fractures, particularly hip and spinal fractures, often lead to long-term functional impairment and reduced quality of life and are closely associated with increased all-cause mortality risk [ 6 , 7 ]. The development of osteoporosis results from the combined effects of multiple factors, including traditional risk factors such as aging, changes in sex hormone levels, genetic susceptibility, and metabolic abnormalities, as well as modifiable factors like drug exposure [ 8 , 9 ]. In recent years, drug-related bone loss has garnered increasing attention, particularly among hospitalized patients and those requiring long-term medication, as its potential impact on skeletal health cannot be overlooked [ 10 , 11 ]. Anticoagulant therapy is widely employed in clinical practice for the prevention and treatment of venous thromboembolism during the perioperative period of orthopedic, trauma, and major abdominal surgeries [ 12 , 13 ]. Enoxaparin sodium, as a low molecular weight heparin, has become one of the most commonly used anticoagulants in clinical practice due to its stable pharmacokinetic properties, reliable anticoagulant effects, and ease of use [ 14 ]. Compared with standard heparin, enoxaparin is generally considered to carry a lower risk of adverse skeletal effects [ 15 ]. However, a growing body of experimental and clinical research suggests that long-term exposure to heparin-like drugs may adversely affect bone metabolism by suppressing osteoblast activity, enhancing osteoclast-mediated bone resorption, and disrupting calcium homeostasis [ 16 , 17 ]. Despite enoxaparin's widespread and prolonged clinical use, systematic and reliable real-world evidence linking it to osteoporosis risk remains lacking. From a biological mechanism perspective, the development of osteoporosis stems not only from an imbalance between bone formation and resorption but is also closely associated with signaling pathways involving oxidative stress, chronic inflammation, hypoxia responses, and abnormal transcriptional regulation [ 18 , 19 ]. Long-term drug exposure may accelerate bone loss by activating inflammatory responses and oxidative stress, inhibiting osteoblast differentiation, promoting osteoclast generation, and disrupting the bone microenvironment [ 16 , 20 ]. However, the specific molecular mechanisms through which prolonged enoxaparin exposure mediates osteoporosis remain poorly understood. With the continuous refinement of large-scale electronic health record databases, real-world data has emerged as a crucial platform for evaluating long-term drug safety and its long-term outcomes [ 21 ]. Concurrently, advances in systems biology and computational biology methods have enabled the integration of clinical epidemiological research with transcriptomics and network analysis [ 22 ]. Weighted Gene Co-expression Network Analysis (WGCNA) identifies gene co-expression modules closely associated with disease, while machine learning methods aid in screening key risk features from high-dimensional data. Network toxicology and molecular docking analysis provide mechanistic support for elucidating potential connections between drugs, targets, and pathways [ 23 – 25 ]. Based on this, the present study aims to systematically evaluate the association between long-term enoxaparin use and osteoporosis risk and to explore its potential molecular mechanisms. Using the Medical Information Modeling and Informatics Consortium version IV (MIMIC-IV) database, we assessed the relationship between enoxaparin exposure duration and the risk of osteoporosis development through multivariate regression, subgroup analysis, nonlinear modeling, and mediation analysis. Subsequently, integrating network toxicology, transcriptomic data analysis, WGCNA, and machine learning methods, we screened key regulatory genes potentially mediating enoxaparin-associated bone loss and validated potential interactions between enoxaparin and core targets through molecular docking analysis. By integrating real-world clinical evidence with multi-level computational biology analysis, this study provides a systematic evaluation of skeletal safety during long-term enoxaparin therapy and offers novel theoretical insights into its potential molecular mechanisms. 2. Materials and methods 2.1 Data Availability Data are available from the corresponding author upon reasonable request. This study has been reported per the STROCSS 2025 guidelines [ 26 ]. All data used in this research were obtained from the publicly accessible MIMIC-IV database, a large, de-identified critical care dataset jointly developed by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center [ 27 ]. Access to the MIMIC-IV repository was granted to the lead author (Zhen-Jiang Liu, credential ID 12911307) after successful completion of the National Institutes of Health web-based human-subjects protection training. The Institutional Review Board of Beth Israel Deaconess Medical Center reviewed and approved the acquisition of patient-level information and the creation of the research dataset, subsequently authorizing data sharing and waiving the requirement for informed consent. All analytical procedures adhered strictly to regulations governing patient privacy and data confidentiality. No proprietary or restricted-access datasets were used. 2.2 Study Design and Population We conducted a retrospective cohort analysis using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, a large, freely accessible critical care dataset jointly maintained by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, spanning the years between 2008 and 2019. We included adult patients aged ≥ 18 years who had complete medication administration records documenting the use of enoxaparin and had complete follow-up data relevant to the assessment of osteoporosis, ensuring accurate identification of exposure duration and dosage patterns. Exposure was defined as long-term use of enoxaparin (≥ 30 days) and non-long-term use (< 30 days), while the primary outcome was the development or documented diagnosis of osteoporosis during or after the exposure period. As MIMIC-IV contains fully de-identified patient information, prospective registration with a clinical research registry was not required. After applying exclusion criteria—patients younger than 18 years, those lacking medication exposure records, those with incomplete diagnostic or covariate information, and individuals with pre-existing osteoporosis—the final analytic cohort consisted of 23852 eligible adult patients meeting all inclusion conditions. The MIMIC-IV project was approved by the Institutional Review Boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, with a waiver of informed consent due to the use of de-identified data. 2.3 Outcome Measurement The primary outcome was the development of osteoporosis following enoxaparin exposure, identified through International Classification of Diseases (ICD) diagnostic codes recorded in the MIMIC-IV database. Osteoporosis was defined using ICD-9 codes (73300, 73301, 73302, 73303, 73309) and ICD-10 codes (M80-M82), encompassing both age-related and secondary osteoporosis. To ensure temporal validity, only cases in which osteoporosis was diagnosed after the initiation of enoxaparin therapy were classified as outcome events. Patients with any documented osteoporosis diagnosis prior to exposure served as exclusions. All outcome classifications were based on clinician-entered diagnoses from inpatient hospital encounters, ensuring clinical verification. 2.4 Covariates Covariates were selected based on clinical relevance and established protocols in prior pharmacoepidemiologic studies using the MIMIC database. Demographic variables included age, sex, and race, categorized in MIMIC-IV as White, Black, Asian, Latino, and Other. Body mass index (BMI) was included as a continuous variable given its established association with bone metabolism. Clinical comorbidities included documented deep vein thrombosis (DVT) according to ICD-coded diagnoses in the medical record. To characterize enoxaparin exposure patterns, medication-related variables were also incorporated as covariates, including total cumulative dose, number of administrations, total duration of therapy (days), and average daily dose, each extracted from the complete medication administration records. These variables reflect both the intensity and duration of anticoagulant exposure and were selected to account for dose-dependent effects on bone health. All covariates were chosen a priori based on biological plausibility and literature-supported associations with osteoporosis risk, anticoagulation exposure, or both. All candidate covariates were screened using univariate logistic regression, and those with p < 0.05 (including age, sex, DVT status, total cumulative enoxaparin dose, number of administrations, total days of therapy, and average daily dose) were retained for multivariable adjustment. Participants with missing data for any covariate (< 5% of the total cohort) were excluded using complete-case deletion. 2.5 Acquisition of Osteoporosis Targets 2.5.1 Acquisition of differentially expressed gene of Osteoporosis through bioinformatics analysis We obtained microarray dataset GSE35958 from the GEO database ( https://www.ncbi.nlm.nih.gov/geo/ ). For data normalization, log2 transformation was applied, and individual batch effects were eliminated via the removeBatchEffect in the limma R package. Additionally, the limma R package was utilized to identify differentially expressed genes (DEGs) between osteoporosis samples and normal samples, which were then visualized through heat maps and volcano plots. The criteria were set at adjusted p 1.5. The Database for Annotation, Visualization and Integrated Discovery (DAVID) was used to collect data for conducting GO analysis and enrichment with the KEGG pathways. GO analysis categorizes enriched terms into biological processes (BP), cellular components (CC), and molecular functions (MF), which are used to provide insights into the roles of potential targets. KEGG was used to systematically analyze the gene functions and metabolic pathways involved in the target genes, which is helpful to fully understand the overall effect of the predicted targets in the organism. 2.5.2 Identify overlapping genes targets To identify more osteoporosis-related targets, three databases—GeneCards, OMIM, and TTD—were queried using “osteoporosis” as the keyword. The genes in the GeneCards database are screened based on the median gene relevance score to ensure validity. Integrate the obtained genes with the genes obtained from bioinformatics analysis and eliminate duplicate genes to obtain the final disease target genes. 2.6 Identification of enoxaparin-associated targets The chemical structure and SMILES notation of enoxaparin sodium were downloaded from PubChem ( https://pubchem.ncbi.nlm.nih.gov ). Potential enoxaparin target proteins were predicted using ChEMBL ( https://www.ebi.ac.uk/chembl/ ), STITCH ( https://stitch-db.org/ ), and Swiss Target Prediction ( https://swisstargetprediction.ch/ ), with species restricted to “Homo sapiens” and a prediction probability > 0. The intersecting genes between predicted enoxaparin targets and osteoporosis-associated targets were identified using Venn diagram analysis. These overlapping genes were considered potential mechanistic mediators of long-term enoxaparin exposure contributing to osteoporosis. 2.7 Construction of the protein–protein interaction (PPI) network and identification of core target genes Bioinformatic analysis of the overlapping genes was conducted via the STRING platform to map PPI networks. The STRING database analysis was conducted with taxonomic restriction to Homo sapiens, implementing an interaction confidence cut-off > 0.4, and medium false discovery rate (FDR) stringency. The resultant network data were then processed through Cytoscape 3.10.1 for topological mapping. Network topology evaluation was conducted using CytoNCA, a dedicated plugin for identifying densely connected functional clusters within protein interaction networks. Nodes with centrality values above the median were identified as core targets. 2.8 GO and KEGG pathway analysis Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted using the ClusterProfiler package (version 4.14.6). The intersecting targets from the Venn diagram were submitted to the ClusterProfiler Package, with “Homo sapiens” selected. GO terms were categorized by biological process (BP), molecular function (MF), and cellular component (CC), and pathways were sorted by p-value. Visualizations for the top GO terms and KEGG pathways were generated using the Microbiotics website ( http://www.bioinformatics.com.cn/ ), highlighting the most significant results. 2.9 Construction of the Weighted Gene Co-expression Network and Identification of Key Modules WGCNA was mainly employed to screen the key modules and obtain hub genes strongly correlated with disease traits. The “WGCNA” package in R software facilitated the identification and visualization of unsigned co-expression network modules within the GSE35958 dataset. The construction of the weighted gene network began with the selection of an optimal soft thresholding power (β). This parameter was chosen using the pickSoftThreshold function, which explored a range from 1 to 20. This soft threshold was then applied to form a co-expression similarity matrix, which was subsequently raised to calculate the adjacency matrix. The selection of the best power to establish this adjacency matrix was guided by scale-free topology criteria, ensuring strong gene-gene correlations. Following this, the adjacency matrix was transformed into a Topological Overlap Matrix (TOM) to measure dissimilarities. A hierarchical clustering function was then employed to group genes with similar expression profiles into distinct modules. The derived gene modules were subsequently associated with clinical features by assessing both gene significance (GS) and module membership (MM). Modules exhibiting a positive correlation with these clinical features were identified and designated as key co-expression modules. 2.10 Construction of diagnostic model by machine learning The Lasso regression model was employed to identify biomarkers most significantly associated with OP among 134 differentially expressed genes. The regularization parameter (λ) was optimized through cross-validation, and model performance was evaluated by tracking the binomial bias value. By visualizing biomarker coefficients at different λ values, markers maintaining statistical significance across λ variations were identified. Influence direction maps were constructed for these markers, where positive and negative coefficients represent their respective promoting or suppressing effects on the target outcome. The biomarker with the most significant coefficient was ultimately selected as a potential key predictor. The results were cross-referenced with biomarkers identified by the WGCNA method. A Venn diagram identified overlapping biomarkers between the two approaches, highlighting the most consistent biomarkers across different analytical pathways. These overlapping biomarkers were considered key indicators and potential mechanism-mediated factors in the study. 2.11 Molecular docking analysis of Enoxaparin Sodium with core OP-related targets A comprehensive examination of the interactions between Enoxaparin Sodium-OP and the key target proteins was identified through utilizing the molecular docking approach. The three-dimensional coordinates of core target proteins were acquired from the Protein Data Bank (PDB), while the Enoxaparin Sodium-OP’ stereochemical configuration was retrieved from PubChem (CID: 16667706). Structural optimization procedures in PyMOL 3.1.3.1 involved ligand excision and solvent molecule elimination. Following hydrogen atom addition via AutoDockTools’ structural refinement module (v1.5.6). Molecular docking simulations were executed through the AutoDock Vina (v1.1.x) command-line interface, with resultant binding poses rendered in PyMOL (v3.1.3.1). 2.12 Statistical Analysis Continuous variables were examined for normality using the Shapiro–Wilk test prior to further statistical analysis. Continuous variables were presented as median (interquartile range) and compared between groups using the Wilcoxon rank-sum test. Categorical variables were represented as frequencies and percentages and compared between groups using the chi-square test. Descriptive statistics were used to summarize baseline characteristics of the study cohort. Continuous variables were reported as mean ± standard deviation (SD) or median (interquartile range, IQR), as appropriate, while categorical variables were presented as counts and percentages. Between-group comparisons were performed using Student’s t-test or the Wilcoxon rank-sum test for continuous variables and the chi-square test (or Fisher’s exact test when appropriate) for categorical variables, consistent with commonly used reporting in epidemiologic analyses. For improved clarity in the representation of the odd ratio (OR), total days of administrations value was divided by 10. The association between long-term enoxaparin exposure and osteoporosis was evaluated using logistic regression models, and effect estimates were expressed as odds ratios with 95% confidence intervals (CIs). Covariates were selected a priori based on clinical relevance (age, sex, race, BMI, DVT status, and enoxaparin exposure pattern variables), and candidate covariates were screened using univariate logistic regression; variables meeting the prespecified threshold (P < 0.05) were included in multivariable adjustment. Model 1 was unadjusted; Model 2 was adjusted for age, sex, race, BMI, and DVT; and Model 3 was adjusted for age, sex, race, BMI, DVT, average number per day, and daily average dose. To assess potential effect modification, prespecified subgroup analyses were conducted across clinically relevant strata (sex, age, BMI, race, DVT, average number per day, daily average dose), and interaction was tested by adding multiplicative interaction terms to multivariable models with significance assessed via P for interaction. To characterize a potential nonlinear exposure–response relationship, restricted cubic spline (RCS) regression was applied to model the association between enoxaparin exposure duration (days) and osteoporosis risk; overall and nonlinearity P values were reported to evaluate whether the relationship deviated from linearity. The mediating roles of enoxaparin administration intensity (average administrations/day and daily average dose) in the exposure duration–osteoporosis association were examined using causal mediation analysis, reporting the average causal mediation effect (ACME), average direct effect (ADE), and proportion mediated. All analyses were conducted in R. A two-tailed p < 0.05 defined statistical significance. All statistical analyses were conducted with R 4.0.2, in conjunction with SPSS 25.0. 3. Results 3.1 Baseline characteristics of all participants Among the 23,852 participants receiving enoxaparin sodium who were included in the analysis, 902 were categorized into the long-term exposure group, while 22,950 were assigned to the non–long-term exposure group. As shown in Table 1 , participants in the long-term exposure group were more likely to be male and were significantly younger, with a lower median age, compared with those in the non–long-term exposure group (P < 0.001). No significant difference in body mass index was observed between the two groups (P = 0.968). Notably, the prevalence of osteoporosis was significantly higher among participants in the long-term exposure group than among those in the non–long-term exposure group (P < 0.001). In addition, the total cumulative dose, total number of administrations, total duration of use, and average daily dose of enoxaparin sodium were all significantly greater in the long-term exposure group (all P < 0.001). Table 1 Baseline characteristics of the osteoporosis and non-osteoporosis groups Variables Overall (n = 23852) Osteoporosis group (n = 2291) Non-osteoporosis group (n = 21561) P Sex, n(%) < 0.001 Male 10331(43.81) 336(14.67) 9995(46.36) Female 13521(56.19) 1955(85.33) 11566 (53.64) Age (years) 64.00(51.00,76.00) 78.00(67.00, 86.00) 62.00(49.00, 74.00) < .001 BMI 27.37(23.63, 32.02) 24.94(20.68, 28.62) 27.62 (23.82, 32.34) < .001 Race, n(%) < .001 ASIAN 634(2.66) 63(2.75) 571(2.65) BLACK 2938(12.32) 178 (7.77) 2760(12.80) LATINO 1007(4.22) 70 (3.06) 937(4.35) OTHER 1964(8.23) 129(5.63) 1835(8.51) WHITE 17309(72.57) 1851(80.79) 15458(71.69) DVT, n(%) 0.626 Yes 2655(11.13) 262 (11.44) 2393 (11.10) No 21197(88.87) 2029 (88.56) 19168 (88.90) Total dose(mg) 320.00(180.00,660.00) 320.00(200.00, 640.00) 320.00(180.00, 660.00) 0.995 Total number of administrations(times) 7.00(4.00,12.00) 8.00(5.00, 14.00) 6.00(4.00, 12.00) < .001 Total days of administrations(days) 5.00(3.00,9.00) 6.00(4.00, 10.00) 5.00(3.00, 9.00) < .001 Average number per day(times) 1.00(1.00, 2.00) 1.00(1.00, 2.00) 1.00(1.00, 2.00) < .001 Daily average dose(mg) 49.00(40.00,120.00) 40.00(40.00, 87.50) 50.00(40.00, 120.00) < .001 Among all participants, 2,291 were identified as having osteoporosis. Compared with participants without osteoporosis, those with osteoporosis were significantly older and predominantly female, and they exhibited a lower body mass index (P < 0.001). Regarding race/ethnicity, participants with osteoporosis were more likely to be non-Hispanic White, whereas the proportions of Black, Latino, and other racial groups were lower in the osteoporosis group (P < 0.001). The prevalence of deep vein thrombosis did not differ significantly between participants with and without osteoporosis (P = 0.626). With respect to enoxaparin sodium exposure, although no significant difference was observed in total cumulative dose between the two groups (P = 0.995), participants with osteoporosis had a significantly longer duration of treatment and higher exposure-related measures than those without osteoporosis (all P < 0.001). Detailed baseline characteristics are summarized in Tables 1 and 2 . Table 2 Baseline characteristics of the long-term medication and non-long-term medication groups Variables Overall(n = 23852) Long-term medication group(n = 902) Non-long-term medication group(n = 22950) P Sex, n(%) < .001 Male 10331(43.81) 467(51.77) 9864 (42.98) Female 13521(56.19) 435(48.23) 13086(57.02) Age(years) 64.00(51.00,76.00) 60.00(50.00,69.00) 64.00(51.00,76.00) < .001 BMI 27.37(23.63, 32.02) 27.33(23.47, 31.86) 27.37(23.63, 32.02) 0.968 Race, n(%) 0.694 ASIAN 634(2.66) 29 (3.22) 605 (2.64) BLACK 2938(12.32) 117 (12.97) 2821 (12.29) LATINO 1007(4.22) 36 (3.99) 971 (4.23) OTHER 1964(8.23) 67 (7.43) 1897 (8.27) WHITE 17309(72.57) 653 (72.39) 16656 (72.58) Osteoporosis, n(%) 0.002 Yes 2291(9.61) 114(12.64) 2177 (9.49) No 21561(90.39) 788(87.36) 20773(90.51) DVT, n(%) < .001 Yes 2655(11.13) 228(25.28) 2427 (10.58) No 21197(88.87) 674(74.72) 20523 (89.42) Total dose(mg) 320.00(180.00,660.00) 3960.00(2080.00,6480.00) 320.00(180.00,600.00) < .001 Total number of administrations(times) 7.00(4.00,12.00) 64.00(47.00,93.00) 6.00(4.00,12.00) < .001 Total days of administrations(days) 5.00(3.00,9.00) 42(34.00,57.00) 5.00(3.00,8.00) < .001 Average number per day(times) 1.00(1.00, 2.00) 1.54 (1.08, 1.91) 1.00 (1.00, 2.00) < .001 Daily average dose(mg) 49.00(40.00,120.00) 87.53(43.46,136.11) 47.69(40.00-120.00) < .001 3.2 Duration-dependent association between enoxaparin exposure and osteoporosis risk Multivariable regression models, with sequential adjustment for demographic characteristics, clinical covariates, and enoxaparin dosing parameters, were used to assess the association between long-term enoxaparin exposure and the risk of osteoporosis (Table 3 ). In the unadjusted model (Model 1), a significant association was observed between longer duration of enoxaparin exposure and an increased risk of osteoporosis. When exposure duration was modeled as a continuous variable, each unit increase in exposure duration was associated with an 8% higher risk of osteoporosis (OR = 1.08, 95% CI: 1.06–1.11; P < 0.001). When enoxaparin exposure duration was categorized into quartiles—Q1 ( 90 days)—a clear duration–response relationship was observed (P for trend = 0.035). Compared with patients exposed for less than 15 days (Q1), those receiving enoxaparin for 15–30 days (Q2) and 30–90 days (Q3) showed progressively higher risks of osteoporosis (OR = 1.19, 95% CI: 1.03–1.38; P = 0.019 and OR = 1.32, 95% CI: 1.07–1.64; P = 0.011, respectively). The highest risk was observed among patients with prolonged exposure exceeding 90 days (Q4), who exhibited more than a twofold increased risk of osteoporosis (OR = 2.31, 95% CI: 1.31–4.06; P = 0.004). After adjustment for age, sex, race, body mass index, and deep vein thrombosis status (Model 2), the association between enoxaparin exposure duration and osteoporosis risk remained robust. Each unit increase in exposure duration was associated with a 16% increase in osteoporosis risk (OR = 1.16, 95% CI: 1.08–1.24; P < 0.001). Relative to short-term exposure (< 15 days), patients exposed for more than 90 days continued to demonstrate a markedly elevated risk of osteoporosis (OR = 12.23, 95% CI: 4.10–36.51; P < 0.001), with a statistically significant increasing trend across exposure categories (P for trend = 0.023). In the fully adjusted model (Model 3), which additionally accounted for enoxaparin administration intensity, including average daily dose and average number of administrations per day, the association was modestly attenuated but remained statistically significant. Each unit increase in exposure duration was associated with a 15% higher risk of osteoporosis (OR = 1.15, 95% CI: 1.08–1.23; P < 0.001). Compared with patients treated for less than 15 days, those with prolonged exposure exceeding 90 days retained a substantially increased risk of osteoporosis (OR = 11.94, 95% CI: 3.98–35.79; P < 0.001). Importantly, the duration-dependent trend persisted across all models (P for trend = 0.015), supporting a dose–time–response relationship between long-term enoxaparin use and osteoporosis risk. Table 3 Logistic regression analysis of the associations between long-term enoxaparin exposure and the risk of osteoporosis in all participants Categories Model 1 Model 2 Model 3 OR(95%CI) P-value P for trend OR(95%CI) P-value P for trend OR(95%CI) P-value P for trend Continuous variable per unit 1.08 (1.06 ~ 1.11) < .001 1.16 (1.08 ~ 1.24) < 0.001 1.15 (1.08 ~ 1.23) < 0.001 Quartile 0.035 0.023 0.015 Q1(N = 20931) Ref Q2(N = 2018) 1.19 (1.03 ~ 1.38) 0.019 1.42 (1.02 ~ 1.98) 0.038 1.35 (0.97 ~ 1.88) 0.079 Q3(N = 825) 1.32 (1.07 ~ 1.64) 0.011 1.44 (0.83 ~ 2.50) 0.199 1.42 (0.81 ~ 2.47) 0.219 Q4(N = 78) 2.31 (1.31 ~ 4.06) 0.004 12.23 (4.10 ~ 36.51) < .001 11.94 (3.98 ~ 35.79) < .001 Model 1: unadjusted Model 2: adjusted for age, sex, race, BMI, DVT Model 3: adjusted for age, sex, race, BMI, DVT, average number per day, daily average dose 3.3 Subgroup analysis Subgroup analyses were performed to assess the consistency of the association between long-term enoxaparin exposure and osteoporosis risk across clinically relevant strata (Fig. 2 ). Overall, long-term enoxaparin exposure was significantly associated with an increased risk of osteoporosis in the total population (OR = 1.38, 95% CI: 1.13–1.69; P = 0.002). This association was consistently observed in both female (OR = 1.59, 95% CI: 1.26–2.02; P < 0.001) and male patients (OR = 1.58, 95% CI: 1.02–2.44; P = 0.039), with no significant interaction by sex (P for interaction = 0.976). Stratified analyses by age revealed a stronger association among patients younger than 65 years (OR = 2.68, 95% CI: 1.95–3.69; P < 0.001), whereas the association was not statistically significant among those aged 65 years or older (OR = 1.22, 95% CI: 0.93–1.60; P = 0.143), and a significant interaction with age was observed (P for interaction < 0.001). With respect to race, a significant association was detected only among White patients (OR = 1.50, 95% CI: 1.20–1.87; P < 0.001), while no significant associations were observed in other racial groups, and no significant interaction was identified (P for interaction = 0.214). The association between enoxaparin exposure and osteoporosis risk was also evident among patients without deep vein thrombosis (OR = 1.42, 95% CI: 1.13–1.78; P = 0.003), but not among those with deep vein thrombosis, with no significant interaction (P for interaction = 0.615). Furthermore, patients receiving more than one enoxaparin administration per day (OR = 1.66, 95% CI: 1.10–2.49; P = 0.015) and those with a daily average dose ≥ 49 mg (OR = 1.64, 95% CI: 1.28–2.10; P < 0.001) exhibited higher osteoporosis risk; however, no significant interactions were observed for administration frequency or daily dose (P for interaction = 0.552 and 0.139, respectively). 3.4 Nonlinear association between enoxaparin exposure duration and osteoporosis risk Restricted cubic spline analysis was performed to explore the potential nonlinear relationship between enoxaparin exposure duration and the risk of osteoporosis (Fig. 3 ). The analysis demonstrated a significant overall association between exposure duration and osteoporosis risk (P for overall < 0.001), along with a statistically significant nonlinear relationship (P for nonlinearity < 0.001). As shown in the spline curve, the odds of osteoporosis increased rapidly during the early phase of enoxaparin exposure, followed by a more gradual but sustained increase with longer exposure duration. The risk continued to rise as exposure days accumulated, particularly beyond prolonged treatment periods, indicating a clear time-dependent pattern. These findings suggest that longer durations of enoxaparin use are associated with progressively higher osteoporosis risk in a nonlinear manner. 3.5 Mediation effects of enoxaparin administration intensity on the association between exposure duration and osteoporosis risk Mediation analyses were conducted to examine whether enoxaparin administration intensity mediated the association between exposure duration and osteoporosis risk (Fig. 4 ). The results indicated that the average number of administrations per day partially mediated the relationship between days of enoxaparin exposure and osteoporosis, with a significant average causal mediation effect (ACME = 0.000321, P < 0.001). The average direct effect (ADE) of exposure duration on osteoporosis risk remained statistically significant (ADE = 0.007211, P < 0.001), and the proportion of the total effect mediated through administration frequency was estimated at 4.43% (P < 0.001). Similarly, daily average dose was identified as a significant mediator in the association between exposure duration and osteoporosis risk (ACME = 0.000422, P < 0.001), while the direct effect of exposure duration remained significant (ADE = 0.007622, P < 0.001). The proportion of the association mediated by daily average dose was 6.01% (P < 0.001). These findings suggest that enoxaparin administration intensity partially mediates the relationship between prolonged exposure duration and increased osteoporosis risk. 3.6 Screening candidate targets and analyzing the potential biological functions in osteoporosis To screen candidate targets associated with osteoporosis, transcriptomic data from bone mesenchymal stem cell samples were retrieved from the Gene Expression Omnibus (GEO) database (accession number: GSE35958), which includes gene expression profiles from 4 healthy controls and 5 patients with osteoporosis. Differentially expressed genes (DEGs) between osteoporosis and healthy states were identified using the limma package in R. Significant transcriptional alterations were visualized through hierarchical clustering heatmaps and volcano plots. The volcano plot revealed a clear separation in gene expression patterns between osteoporosis and control groups, based on the thresholds of P 1.5 (Fig. 5 A). Hierarchical clustering analysis based on the top DEGs further revealed distinct expression patterns between osteoporosis and control samples, indicating robust transcriptional heterogeneity associated with disease status (Fig. 5 B). To detect the biological characteristics of DEGs in osteoporosis, we performed GO and KEGG analysis via the DAVID database. Gene Ontology (GO) enrichment analysis of the identified DEGs showed significant enrichment in multiple biological processes, cellular components, and molecular functions (Fig. 5 C). Notably, enriched biological processes were primarily related to extracellular matrix organization, collagen fibril organization, cell–substrate adhesion, and regulation of protein targeting. Cellular component analysis highlighted endoplasmic reticulum lumen, Golgi apparatus, and extracellular matrix–associated structures, while molecular function analysis emphasized DNA-binding transcription factor activity, RNA polymerase II–specific binding, cadherin binding, and phosphoric ester hydrolase activity. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed that the DEGs were significantly involved in multiple signaling pathways relevant to bone metabolism and inflammatory regulation (Fig. 5 D). The most enriched pathways included PI3K–Akt signaling, focal adhesion, human cytomegalovirus infection, regulation of actin cytoskeleton, cellular senescence, insulin signaling, estrogen signaling, HIF-1 signaling, and phospholipase D signaling pathways. These pathways are closely associated with osteoblast differentiation, osteoclast activity, cellular stress responses, and immune-inflammatory processes, suggesting that dysregulation of these signaling networks may contribute to osteoporosis pathogenesis. To identify additional osteoporosis-related targets, disease-associated genes were further retrieved from the GeneCards, OMIM, and Therapeutic Target Database (TTD) databases. As illustrated in Fig. 5 E, a total of 2,340 genes were obtained from GeneCards, 24 genes from OMIM, and 21 genes from TTD. Intersection analysis revealed a limited number of overlapping genes shared among these databases, representing high-confidence osteoporosis-associated targets. After removing duplicate entries, a comprehensive set of disease-related candidate genes was established for subsequent analysis. Meanwhile, osteoporosis-related genes were independently identified from the GEO transcriptomic dataset, yielding 773 disease-associated genes (Fig. 5 F). By intersecting the GEO-derived genes with the curated disease-associated targets, overlapping genes were identified. After removing duplicate targets, 3233 targets associated with osteoporosis were obtained through using a Veen diagram. 3.7 Screening core target genes of enoxaparin-associated osteoporosis through a PPI network We first investigated the relationship between enoxaparin sodium and osteoporosis using a network toxicology approach. The chemical structure of enoxaparin sodium was first retrieved and used as the input compound for target prediction (Fig. 6 A). After removing duplicates, 530 targets of enoxaparin sodium were obtained via screening ChEMBL, STITCH, and Swiss Target Prediction databases (Fig. 6 B). As demonstrated in Fig. 6 C, 134 core target genes of enoxaparin sodium-MPs causing osteoporosis were obtained through overlapping 3233 target genes of osteoporosis and 530 target genes of enoxaparin sodium. To further reveal the interactions between common targets, a medium-confidence PPI network (confidence score ≥ 0.400) was constructed via the STRING database by inputting the 134 shared targets between enoxaparin sodium and osteoporosis. The results showed a total of 211 edges and 134 nodes, illustrating the complex interactions among the potential targets (Fig. 6 D). A topologically refined protein interaction network was generated to map inter-target relationships, with nodes symbolizing protein targets and edges denoting interaction events. Within the PPI network diagram, nodes exhibiting the highest degree are highlighted in dark red, with the color intensity diminishing as the degree reduces (Fig. 6 E). 3.8 Functional enrichment analysis of core target genes To elucidate the biological significance of the core target genes associated with long-term enoxaparin exposure and osteoporosis risk, Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed. GO analysis categorized the enriched terms into biological process (BP), cellular component (CC), and molecular function (MF) domains. As shown in Fig. 7 A, the core target genes exhibited distinct functional enrichment patterns across the three GO categories. In the BP category, the core target genes were predominantly enriched in processes related to response to oxidative stress, cellular response to chemical stress, response to peptide hormone stimulus, and regulation of cell development and differentiation. In the CC category, enriched terms were mainly associated with nuclear and chromosomal components, including chromosomal regions and transcription factor complexes. In the MF category, significant enrichment was observed for DNA-binding transcription factor binding, transcription coregulator activity, and transmembrane transporter activity. To further illustrate the distribution of enriched GO terms, a circular GO visualization was generated (Fig. 7 B). BP and MF categories showed higher enrichment intensity and gene ratios than CC terms, indicating functional convergence of the core genes on stress response, transcriptional regulation, and signaling integration. The GO dot plot demonstrated that several BP terms related to oxidative stress and inflammatory signaling showed high gene ratios and low adjusted p-values despite modest gene counts, suggesting that a limited number of core genes may exert substantial regulatory effects (Fig. 7 C). KEGG pathway analysis further revealed significant enrichment in pathways associated with bone remodeling and inflammation, including NF-κB, MAPK, PI3K–Akt, JAK–STAT signaling, apoptosis, and osteoclast differentiation (Fig. 7 D). The gene–pathway network indicated that core genes were involved in multiple pathways simultaneously, highlighting their roles as regulatory nodes coordinating signaling crosstalk. Overall, these results suggest that long-term enoxaparin exposure may influence osteoporosis risk through integrated regulatory networks governing oxidative stress, inflammation, and bone metabolism. 3.9 Weighted gene co-expression network analysis To identify gene co-expression modules associated with osteoporosis under long-term enoxaparin exposure, weighted gene co-expression network analysis (WGCNA) was performed using normalized gene expression data (Fig. 8 ). Hierarchical clustering was conducted to assess sample quality, and one potential outlier was identified and removed prior to network construction to improve robustness (Fig. 8 A). The remaining samples showed consistent clustering patterns and were retained for subsequent analysis. The sample dendrogram combined with the trait heatmap demonstrated a clear correspondence between expression-based clustering and clinical phenotypes, supporting the biological relevance of the dataset (Fig. 8 B). To construct a scale-free co-expression network, an appropriate soft-thresholding power was selected. As shown in Fig. 8 C and Fig. 8 D, the scale-free topology fit index (R²) increased and reached a plateau above 0.8 with increasing soft-threshold values, while mean connectivity gradually decreased. Based on these criteria, an optimal soft-threshold was chosen to balance scale-free topology and network connectivity. Using the selected soft-thresholding power, genes were hierarchically clustered into distinct co-expression modules by dynamic tree cutting. The gene dendrogram and corresponding module color assignments are shown in Fig. 8 E, revealing multiple modules with distinct expression patterns. Module–trait correlation analysis indicated that two modules, ME maroon and ME sienna3, were significantly associated with osteoporosis, while showing weak or opposite correlations with control samples (Fig. 8 F and Fig. 8 G), suggesting their involvement in osteoporosis-related transcriptional alterations. To further validate the relevance of OP-associated modules, relationships between module membership (MM) and gene significance (GS) for OP were examined. As shown in Fig. 8 F and Fig. 8 G, genes within these modules exhibited strong positive correlations between MM and GS (cor = 0.81 and cor = 0.96, both with extremely low p-values), indicating high intramodular connectivity and strong associations with the OP phenotype. Collectively, these results demonstrate that the identified OP-associated modules are robust, biologically meaningful, and closely linked to disease status, providing a solid foundation for subsequent integration with machine learning and downstream functional analyses. 3.10 Machine learning–based identification of key genes To further identify key genes associated with osteoporosis risk under long-term enoxaparin exposure, a machine learning–based feature selection strategy was applied. Least absolute shrinkage and selection operator (LASSO) regression was employed to reduce dimensionality and select the most informative predictors. Using ten-fold cross-validation, the optimal regularization parameter (λ) was determined based on the minimum binomial deviance criterion. As shown in Fig. 9 A, the binomial deviance decreased with increasing penalization, and a parsimonious model with a limited number of non-zero coefficients was obtained at the optimal λ value, indicating effective feature reduction and avoidance of overfitting. The coefficient profiles of candidate genes under different penalty strengths are illustrated in Fig. 9 B. As the regularization parameter decreased, only a small subset of genes retained non-zero coefficients, while the majority of coefficients shrank toward zero. This pattern indicates that these retained genes contributed independently and stably to the prediction of osteoporosis status. To further characterize the contribution of each selected gene, the direction and magnitude of their regression coefficients were examined. As shown in Fig. 9 C, several genes, including CDK16, SEC22B, ARFIP2, and VHL, exhibited positive coefficients, suggesting a potential association with increased osteoporosis risk. In contrast, AP1S1 and PTPN9 displayed negative coefficients, indicating a potential protective association. To enhance the robustness of gene selection, the machine learning–derived genes were intersected with genes from osteoporosis-associated co-expression modules identified by WGCNA. As shown in Fig. 9 D, two genes were shared between the WGCNA modules and the LASSO-selected gene set. These overlapping genes were therefore defined as core genes for subsequent functional enrichment analysis and molecular docking validation. 3.11 Molecular docking validation The two core genes identified through integrative WGCNA and machine learning analyses were subjected to molecular docking simulations to systematically evaluate their potential binding affinities with enoxaparin sodium. In molecular docking analysis, lower binding energy indicates a more stable ligand–receptor complex and stronger interaction potential, whereas higher binding energy reflects weaker binding stability. As shown in Fig. 10 , the docking results revealed favorable binding affinities between enoxaparin sodium and both target proteins. Specifically, CDK16 exhibited the strongest binding affinity (− 8.8 kcal/mol; Fig. 10 A), followed by VHL (− 6.2 kcal/mol; Fig. 10 B). Notably, the binding energies of enoxaparin sodium with both core targets were lower than − 5 kcal/mol, suggesting a high level of binding stability. These results indicate a strong molecular affinity between enoxaparin sodium and the identified core proteins, supporting the potential involvement of these targets in mediating the biological effects of long-term enoxaparin exposure. The docking findings provide structural evidence that enoxaparin may directly interact with key regulatory proteins implicated in osteoporosis-related pathways, thereby contributing to altered bone metabolism and disease progression. 4. Discussion This large-scale retrospective cohort analysis based on the MIMIC-IV database systematically revealed a significant and time-dependent association between long-term enoxaparin exposure and increased osteoporosis risk, filling a gap in real-world evidence. Across multiple multivariable models, enoxaparin exposure days demonstrated a clear gradient increase in osteoporosis risk: compared to those exposed for 90 days exhibited a markedly elevated risk (OR = 2.31, 95% CI: 1.31–4.06; P = 0.004). This association remained robust after stepwise adjustment for demographic characteristics, BMI, deep vein thrombosis, and dosing intensity, suggesting a potential time-dependent cumulative effect. Although enoxaparin is generally considered to have a lower skeletal toxicity risk than unfractionated heparin [ 28 ], our findings suggest that prolonged exposure may still exert clinically significant adverse effects on bone health, consistent with prior clinical observations [ 15 ]. Given enoxaparin's widespread use in hospitalized patients and those requiring long-term anticoagulation therapy [ 29 , 30 ], even a moderate increase in risk may translate to a substantial absolute disease burden at the population level, particularly among high-risk patients requiring prolonged anticoagulation [ 31 , 32 ]. Furthermore, after further adjusting for cumulative dose, dosing frequency, and daily average dose, the association between enoxaparin exposure duration and osteoporosis risk remained significant. This suggests that exposure duration itself may be a key determinant of osteoporosis risk, rather than solely dependent on dose factors. These findings extend the evidence of osteotoxicity in heparin drugs—previously primarily based on experimental studies—to the real-world clinical setting, providing quantitative epidemiological support for the skeletal safety of long-term enoxaparin therapy. Beyond linear associations, this study further revealed a significant nonlinear relationship between enoxaparin exposure duration and osteoporosis risk through restricted cubic spline analysis (P < 0.001). Osteoporosis risk rose rapidly during the early exposure phase and continued to increase throughout long-term treatment, suggesting a cumulative toxicity profile with an unclear threshold[ 33 , 34 ]. In recent years, a growing body of research indicates that drug-induced bone metabolism abnormalities, such as those caused by nucleoside analogues and zoledronic acid, often exhibit such nonlinear exposure-response patterns. These patterns feature gradually increasing risk with prolonged exposure duration, accompanied by time-dependent tolerance or cumulative effects. Early imbalances in bone remodeling may lay the groundwork for subsequent accelerated bone loss [ 35 – 38 ]. Although the magnitude of risk varies across different exposure phases, the overall spline curve exhibits a sustained upward trend, supporting the biological plausibility of cumulative effects of long-term enoxaparin exposure on bone metabolism. Given the observational design of this study and the absence of dynamic bone density measurements, the findings should be regarded as hypothesis-generating evidence. Future prospective studies with continuous bone density monitoring are needed to determine whether clinically relevant exposure duration thresholds exist. Subgroup analysis revealed a significant association between long-term enoxaparin use and osteoporosis risk, particularly among women, younger patients, and Caucasians. Gender analysis showed that female patients had a significantly higher risk than males (OR = 1.59, 95% CI: 1.26–2.02, P < 0.001), potentially related to gender-specific differences in bone metabolism and hormonal changes [ 18 ]. Age analysis revealed that patients under 65 years old (OR = 2.68, 95% CI: 1.95–3.69, P < 0.001) exhibited greater sensitivity to enoxaparin exposure, whereas those aged 65 and above showed no significant increased risk. This suggests that enoxaparin-related bone loss may be more pronounced in populations with relatively active baseline bone metabolism. This finding aligns with evidence from animal models and clinical observations that heparin-induced inhibition of osteoblast activity is more readily apparent in younger individuals [ 15 , 39 ]. In contrast, no significant association was observed in elderly patients, potentially due to age-related bone loss mechanisms partially masking the drug effect [ 40 , 41 ]. Previous systematic reviews have also noted that multiple comorbidities in the elderly population may dilute drug-specific skeletal risks [ 5 , 42 ]. Furthermore, racial analysis revealed a higher osteoporosis risk among white patients (OR = 1.50, 95% CI: 1.20–1.87, P < 0.001), potentially attributable to interracial genetic differences or lifestyle factors [ 43 ]. In BMI and deep vein thrombosis (DVT) subgroups, although both were associated with risk, they did not significantly alter the relationship between enoxaparin use and osteoporosis, suggesting that osteoporosis development may be influenced by the duration and dose of drug exposure [ 44 ]. Notably, high-frequency administration (OR = 1.66, 95% CI: 1.10–2.49, P = 0.015) and higher doses (OR = 1.64, 95% CI: 1.28–2.10, P < 0.001) significantly increased osteoporosis risk, further underscoring the impact of drug dosage and treatment intensity on bone health. In summary, this study reveals a significant impact of long-term enoxaparin exposure on osteoporosis risk, particularly in specific populations, suggesting the need for individualized assessment and monitoring during prolonged therapy. Mediation analysis results indicate that enoxaparin dosing intensity (including average daily dosing frequency and daily dose) partially mediated the relationship between exposure duration and osteoporosis risk. However, the proportion of variance explained was relatively limited, and the direct effect of exposure duration remained predominant. The number of daily doses (ACME = 0.000321, P < 0.001) partially mediated the relationship between enoxaparin exposure and osteoporosis, accounting for 4.43% of the mediating effect (P < 0.001). Higher dosing frequency may amplify the drug's cumulative effects, thereby influencing bone metabolism. Furthermore, average daily dose (ACME = 0.000422, P < 0.001) also mediated the relationship between enoxaparin exposure and osteoporosis, accounting for 6.01% of the mediating effect (P < 0.001), underscoring the critical role of drug dosage in osteoporosis development. These findings suggest that while higher dosing intensity may exacerbate osteoporosis risk, prolonged exposure itself is the core driver of abnormal bone metabolism[ 15 , 20 ]. This aligns with evidence from animal models and clinical observations indicating that bone loss associated with heparin drugs is primarily time-dependent rather than purely dose-dependent. From a clinical practice perspective, this finding implies that reducing daily dosage or administration frequency alone may be insufficient to fully offset the bone health risks associated with long-term enoxaparin therapy. This aligns with multiple systematic reviews indicating that short-term (3–6 months) exposure to low molecular weight heparins does not significantly increase fracture risk, whereas longer exposure periods (over 12 months) may lead to bone density decline [ 32 ]. When evaluating anticoagulant benefits versus adverse effects, cumulative exposure duration should be considered a critical risk assessment metric, particularly in high-risk populations requiring long-term anticoagulation [ 45 ]. Regular bone density monitoring is recommended for early intervention against potential bone loss. To mitigate this risk, drug usage strategies should be carefully adjusted in high-risk patients. To explore the potential molecular mechanisms underlying enoxaparin-associated osteoporosis, this study integrated network toxicology, transcriptomics analysis, weighted gene co-expression network analysis (WGCNA), and machine learning methods. Functional enrichment analysis consistently revealed that candidate target genes were predominantly enriched in pathways including oxidative stress, inflammatory signaling, apoptosis, and osteoclast differentiation—all critical processes in bone remodeling and homeostasis. Cross-screening via WGCNA and LASSO regression identified CDK16 and VHL as potential core genes. Molecular docking analysis revealed stable binding potentials between enoxaparin and both CDK16 and VHL, with binding energies below − 5 kcal/mol, suggesting potential direct molecular interactions. This finding structurally supports enoxaparin's influence on key bone metabolism-related proteins. However, molecular docking is a computational prediction method whose results primarily reflect potential binding possibilities rather than actual biological effects. Future studies should validate the biological effects of these pathways using osteoblast and osteoclast models. CDK16 belongs to the non-classical cyclin-dependent kinase family and primarily participates in biological processes outside the cell cycle, including cell differentiation, vesicular transport, and the integrated regulation of multiple signaling pathways [ 46 ]. In the context of bone biology, a growing hypothesis suggests that CDK16 dysfunction may disrupt the differentiation of bone marrow mesenchymal stem cells toward the osteogenic lineage, leading to reduced osteoblast production and diminished bone matrix mineralization capacity [ 15 , 20 ]. In this study, functional enrichment analysis of enoxaparin-related osteoporosis targets revealed significant enrichment of multiple signaling pathways closely associated with osteogenic differentiation, including PI3K–Akt, MAPK, focal adhesion, actin cytoskeleton regulation, and cellular senescence pathways [ 47 – 49 ]. These signaling networks ultimately converge at the regulatory level of core osteogenic transcription factors such as RUNX2, SP7 (Osterix), and COL1A1 [ 50 ]. Long-term disruption of these pathways may cause fate deviation of bone marrow mesenchymal stem cells, suppressing the osteogenic process and thereby inducing insufficient bone formation. Enoxaparin, a highly sulfated, strongly negatively charged glycosaminoglycan-like molecule, may indirectly affect CDK16-related signaling networks through prolonged exposure by altering extracellular matrix interactions, cell-matrix adhesion, and mechanical signaling transduction within the bone marrow microenvironment [ 51 , 52 ]. Furthermore, the significant enrichment of oxidative stress and inflammation-related pathways in this study suggests that persistent inflammatory stimulation may further suppress osteogenic differentiation, with CDK16 potentially acting as a signaling integration node to amplify these adverse effects [ 18 ]. Collectively, these findings support the potential mechanism whereby enoxaparin inhibits osteoblast differentiation and reduces bone formation capacity by regulating CDK16-associated networks. VHL is a key regulator in cellular hypoxic adaptation, primarily acting by controlling the stability of hypoxia-inducible factors (HIFs) [ 53 , 54 ]. In bone tissue, the VHL–HIF axis plays a central role in coordinating angiogenesis, energy metabolism, and bone remodeling [ 55 , 56 ]. Physiological hypoxia and moderate HIF activation are essential for coupling osteogenesis with angiogenesis, whereas disruption of this axis may lead to pathological bone loss [ 57 , 58 ]. Functional enrichment analysis in this study revealed significant enrichment of multiple pathways, including HIF-1 signaling, PI3K–Akt, NF-κB, apoptosis, and osteoclast differentiation. These signaling networks exhibit substantial overlap with the VHL–HIF regulatory system. In the clinical population represented by MIMIC-IV, patients commonly exhibit heightened inflammatory responses, oxidative stress, and impaired tissue perfusion. Against this backdrop, dysregulation of the VHL–HIF axis may amplify effects on bone metabolism [ 18 ]. Molecular docking analysis indicates stable binding potential between enoxaparin and VHL, suggesting potential direct or indirect modulation of VHL function. Although molecular docking results cannot directly prove its biological effects in vivo, they provide structural support for enoxaparin's interference with HIF homeostasis. This process may impair VEGF-mediated angiogenesis, alter osteoblast metabolic programs, and promote osteoclast activation through inflammatory and oxidative stress pathways, ultimately causing bone-angiogenesis coupling imbalance and exacerbating bone resorption. Based on the above evidence, this study proposes a “dual pathway synergistic model” to explain the potential association between long-term enoxaparin exposure and increased osteoporosis risk. On one hand, enoxaparin may inhibit osteoblast differentiation and mineralization by regulating CDK16-associated cell cycle-independent differentiation gating and stress signaling integration. On the other hand, it may weaken endovascular support and promote enhanced osteoclast activity by disrupting VHL–HIF homeostasis and its interactions with inflammatory and oxidative stress pathways. Both pathways ultimately converge onto common downstream signaling networks involving PI3K–Akt, MAPK, NF-κB, apoptosis, and osteoclast differentiation, manifesting as cumulative, nonlinear exposure-response relationships at the population level. Additionally, this study has several limitations. First, the retrospective observational design restricts causal inference and cannot fully exclude the influence of unmeasured confounders such as vitamin D status, glucocorticoid use, and baseline bone mineral density. Second, osteoporosis diagnosis relied on ICD codes rather than direct bone density measurements, potentially introducing some risk of misclassification. Our findings primarily reflect clinically recognized osteoporosis rather than subclinical bone loss. Moreover, the MIMIC-IV database primarily comprises critically ill patients, necessitating further validation of the generalizability of these findings. Mechanistically, network toxicology and molecular docking analyses are inherently predictive methods, carrying inherent risks of circular reasoning. Therefore, independent experimental studies are urgently needed to validate these findings. 5. Conclusion In this retrospective cohort study based on the MIMIC-IV database, prolonged enoxaparin exposure was found to be associated with an increased risk of osteoporosis, exhibiting a clear duration-dependent and nonlinear pattern. By integrating real-world epidemiologic evidence with network toxicology, transcriptomic analysis, machine learning, and molecular docking, CDK16- and VHL-centered signaling networks were identified as potential molecular pathways linking long-term enoxaparin use to dysregulated bone remodeling. Nevertheless, the observational nature of the study, reliance on diagnosis codes rather than direct bone mineral density measurements, and the possibility of residual confounding limit definitive causal inference. Moreover, while network toxicology and molecular docking provide biologically plausible mechanistic hypotheses, they do not establish functional causality. Accordingly, these findings should be interpreted as hypothesis-generating and underscore the need for prospective, multicenter longitudinal studies with precise exposure characterization, serial bone density assessments, and adjudicated outcomes, together with targeted experimental validation, to confirm causality, clarify underlying mechanisms, and guide individualized risk assessment and prevention strategies for patients requiring long-term anticoagulation. Declarations Acknowledgements The authors would like to express their gratitude to the MIMIC database for providing data support. Author contributions LZJ completed the manuscript drafting. WCL were responsible for the study design and manuscript revision. LZJ, LXM, LF and ZZM performed data curation and statistical analysis. Funding This work was supported by the Clinical Evidence-Based Research Special Project for the Construction of High-Level Traditional Chinese Medicine Hospitals of Wangjing Hospital, China Academy of Chinese Medical Sciences (WJYY-XZKT-2023-14). Data availability The data that support the findings of this study are openly available in the Medical Information Mart for Intensive Care (MIMIC)-IV database at 10.13026/kpb9-mt58. Ethical Approval This study was based on publicly available anonymized MIMIC-IV data. Ethical approval was not required. Consent for publication Not applicable. Competing interests The authors report no competing interests related to the conduct of this study, the authorship, or the publication of this work. References Compston JE, McClung MR, Leslie WD, Osteoporosis. Lancet (London England). 2019;393(10169):364–76. GBD LBMD. The global, regional, and national burden attributable to low bone mineral density, 1990–2020: an analysis of a modifiable risk factor from the Global Burden of Disease Study 2021. Lancet Rheumatol. 2025;7(12):e873–94. Liang H, et al. Global epidemiology and burden of osteoporosis among postmenopausal women: insights from the Global Burden of Disease Study 2021. npj aging. 2025;11(1):78. Kanis JA et al. European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA, 2019. 30(1): pp. 3–44. Hoong CWS et al. Advances in the management of osteoporosis. BMJ (Clinical research ed.), 2025. 390: p. e081250. Tai T, et al. Refracture risk and all-cause mortality after vertebral fragility fractures: Anti-osteoporotic medications matter. J Formos Med Association = Taiwan yi zhi. 2023;122(Suppl 1):S65–73. Nasser MI, et al. Impact of hip fracture on survival, disability, pain, and health-related quality of life in Zimbabwe: a prospective cohort study. lancet Healthy Longev. 2025;6(9):100766. Wang L, Chen L, Chen K. Hormone-Related and Drug-Induced Osteoporosis: A Cellular and Molecular Overview. Int J Mol Sci. 2023;24(6):5814. Smit AE, Meijer OC, Winter EM. The multi-faceted nature of age-associated osteoporosis. Bone Rep. 2024;20:101750. Song X, et al. Advances in the study and treatment of glucocorticoid osteoporosis: A review. Medicine. 2025;104(22):e42668. Stegmayr B, Zuo L, Zadora W. Lipid and Bone Effects of Heparin Use During Hemodialysis. Semin Nephrol. 2023;43(6):151480. Tran HA et al. 2025 Guidelines for direct oral anticoagulants: a practical guidance on the prescription, laboratory testing, peri-operative and bleeding management. Internal medicine journal, 2025. 55(7): pp. 1174–1183. Stevens SM, et al. Antithrombotic Therapy for VTE Disease: Compendium and Review of CHEST Guidelines 2012–2021. Chest. 2024;166(2):388–404. Zhang Y, Guo S, Xu J. Multifunctional applications and research advances of low-molecular-weight heparin. Front Pharmacol. 2025;16:1585762. Li Y, et al. Impaired bone healing by enoxaparin via inhibiting the differentiation of bone marrow mesenchymal stem cells towards osteoblasts. J Bone Miner Metab. 2022;40(1):9–19. Yang S et al. Effect of long-term use of unfractionated or low-molecular-weight heparin on bone mineral density in maintenance hemodialysis patients. Hemodialysis international. International Symposium on Home Hemodialysis, 2020. 24(3): pp. 374–382. Yokoyama S, et al. Association between oral anticoagulants and osteoporosis: Real-world data mining using a multi-methodological approach. Int J Med Sci. 2020;17(4):471–9. Luo J, et al. Oxidative stress and inflammation: roles in osteoporosis. Front Immunol. 2025;16:1611932. Iantomasi T, et al. Oxidative Stress and Inflammation in Osteoporosis: Molecular Mechanisms Involved and the Relationship with microRNAs. Int J Mol Sci. 2023;24(4):3772. Mori Y, Aizawa T. Impacts of low molecular weight heparin on bone healing and osseointegration. J Bone Miner Metab. 2022;40(2):357. Parums DV. Med Sci monitor: Int Med J experimental Clin Res. 2025;31:e951118. A Review of the Importance and Relevance of Real-World Data and Real-World Evidence. Kumar R, Romano JD, Ritchie MD. Network-based analyses of multiomics data in biomedicine. BioData Min. 2025;18(1):37. Shen X, et al. Using weighted gene co-expression network analysis to identify key genes related to preeclampsia. Front Immunol. 2025;16:1569591. Kanapari A, et al. Current applications and future challenges of machine learning and artificial intelligence in clinical trials: A scoping review. Digit health. 2025;11:20552076251393272. Wang C, et al. Network toxicology combined with molecular docking technology to explore the molecular mechanism of amatoxin causing liver injury. Sci Rep. 2025;15(1):26068. Riaz A. A., Transparency in the Reporting of Artificial Intelligence – The TITAN Guideline. Premier J Sci, 2025. 2. Johnson AEW, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci data. 2023;10(1):1. Zhang Y, Sun T, Jiang C. Biomacromolecules as carriers in drug delivery and tissue engineering. Acta Pharm Sinica B. 2018;8(1):34–50. Piazza G, et al. Superficial Vein Thrombosis: A Review. JAMA. 2025;334(22):2020–30. Monreal M, et al. Comparative safety of enoxaparin versus other low-molecular-weight heparins in cancer-associated venous thromboembolism: a real-world cohort study from RIETE. Res Pract Thromb haemostasis. 2025;9(8):103244. Chen Y, et al. Association between enoxaparin administration and ICU 28-day mortality in sepsis patients: A comparative study. Respir Med. 2026;251:108543. Gajic-Veljanoski O, et al. Effects of Long-Term Low-Molecular-Weight Heparin on Fractures and Bone Density in Non-Pregnant Adults: A Systematic Review With Meta-Analysis. J Gen Intern Med. 2016;31(8):947–57. Jakonyte A, et al. Risk of Osteonecrosis of the Jaw in Patients Treated with Zoledronic or Alendronic Acid: A Systematic Review. Med (Kaunas Lithuania). 2025;61(7):1159. Chen L, et al. Bone loss in young adults with HIV following antiretroviral therapy containing tenofovir disoproxil fumarate regimen using machine learning. Front Pharmacol. 2025;16:1516013. Yip TC, et al. Long-term use of tenofovir disoproxil fumarate increases fracture risk in elderly patients with chronic hepatitis B. J Hepatol. 2024;80(4):553–63. Huang P, et al. A novel evidence of serial changes of bone mineral density in chronic hepatitis B patients treated with entecavir. Hep Intl. 2021;15(2):310–7. Soldado-Folgado J, et al. Changes in bone quality after switching from a TDF to a TAF based ART: A pilot randomized study. Front Endocrinol. 2023;14:1076739. Black DM, et al. Bisphosphonates and the risk of atypical femur fractures. Bone. 2022;156:116297. Yang H, et al. Safety and Efficacy of Low Molecular Weight Heparin for Thromboprophylaxis in the Elderly: A Network Meta-Analysis of Randomized Clinical Trials. Front Pharmacol. 2021;12:783104. Ensrud KE, Crandall CJ. Osteoporos Annals Intern Med. 2024;177(1):ITC1–16. Department of Error. Lancet (London England). 2022;400(10354):732. Ayers C, et al. Effectiveness and Safety of Treatments to Prevent Fractures in People With Low Bone Mass or Primary Osteoporosis: A Living Systematic Review and Network Meta-analysis for the American College of Physicians. Ann Intern Med. 2023;176(2):182–95. Noel SE, Santos MP, Wright NC. Racial and Ethnic Disparities in Bone Health and Outcomes in the United States. J bone mineral research: official J Am Soc Bone Mineral Res. 2021;36(10):1881–905. Chandran M et al. Impact of osteoporosis and osteoporosis medications on fracture healing: a narrative review. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA, 2024. 35(8): pp. 1337–58. Karcutskie CA, et al. Association of Anti-Factor Xa-Guided Dosing of Enoxaparin With Venous Thromboembolism After Trauma. JAMA Surg. 2018;153(2):144–9. Dixon-Clarke SE, et al. Structure and inhibitor specificity of the PCTAIRE-family kinase CDK16. Biochem J. 2017;474(5):699–713. Xu H et al. A disproportionality analysis of low molecular weight heparin in the overall population and in pregnancy women using the FDA adverse event reporting system (FAERS) database. Frontiers in pharmacology, 2024. 15: p. 1442002. Li D, et al. 7,8-DHF inhibits BMSC oxidative stress via the TRKB/PI3K/AKT/NRF2 pathway to improve symptoms of postmenopausal osteoporosis. Volume 223. Free radical biology & medicine; 2024. pp. 413–29. Liu C, et al. Research on the role and mechanism of the PI3K/Akt/mTOR signalling pathway in osteoporosis. Front Endocrinol. 2025;16:1541714. Artigas N, et al. Mitogen-activated protein kinase (MAPK)-regulated interactions between Osterix and Runx2 are critical for the transcriptional osteogenic program. J Biol Chem. 2014;289(39):27105–17. Signorelli SS, et al. Anticoagulants and Osteoporosis. Int J Mol Sci. 2019;20(21):5275. Shen X, Yao Q, Ma L. Mechanisms responsible for the ability of enoxaparin sodium to inhibit inflammatory responses in the immune microenvironment of bone repair: A transcriptomic sequencing study. PLoS ONE. 2025;20(9):e0332041. Mendoza SV, et al. Osteocytic oxygen sensing: Distinct impacts of VHL and HIF-2alpha on bone integrity. Bone. 2025;192:117339. Chen K, et al. Osteocytic HIF-1α Pathway Manipulates Bone Micro-structure and Remodeling via Regulating Osteocyte Terminal Differentiation. Front cell Dev biology. 2022;9:721561. Janeczko K, Agoro R. Oxygen Sensing in Osteocytes: From Physiology to Age-related Osteoporosis. Curr Osteoporos Rep. 2025;23(1):28. Chen W, et al. HIF-1α Regulates Bone Homeostasis and Angiogenesis, Participating in the Occurrence of Bone Metabolic Diseases. Cells. 2022;11(22):3552. Wang Y, et al. The hypoxia-inducible factor alpha pathway couples angiogenesis to osteogenesis during skeletal development. J Clin Investig. 2007;117(6):1616–26. Knowles HJ. Hypoxic regulation of osteoclast differentiation and bone resorption activity. Hypoxia (Auckland, N.Z.), 2015. 3: pp. 73–82. Li L, et al. Suppression of Hypoxia-Inducible Factor 1α by Low-Molecular-Weight Heparin Mitigates Ventilation-Induced Diaphragm Dysfunction in a Murine Endotoxemia Model. Int J Mol Sci. 2021;22(4):1702. Supplementary Files strocss2025GuidelineChecklist.docx Cite Share Download PDF Status: Published Journal Publication published 07 Apr, 2026 Read the published version in Journal of Translational Medicine → Version 1 posted Reviewers agreed at journal 16 Feb, 2026 Reviewers invited by journal 16 Feb, 2026 Editor assigned by journal 15 Feb, 2026 First submitted to journal 10 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8842742","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":592190704,"identity":"7fb9754f-3b2f-4ec3-bb67-655b83de5960","order_by":0,"name":"Zhen-jiang Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6klEQVRIiWNgGAWjYBACNvnHBx984LGpb2xvSHyQUFFDWAsfQ1qy4QyZNMbmngOPDR6cOUZYixxDjpk0j81hxvYZjs8kH7YwE+EwhmPJxjw5acy8M5jTKhIb2Bj427sT8GthbD74cM4ZGzbJ2W1pNxJ3yDBInDm7Ab8WZrZkg7c9aTyGc84AtZxhYzCQyCWghY3HTIL332EJ+xv53woS25iJ0MLDYybJw3PYgHFGQhoDcVok2ICBzJOWwNhzIFki4cwxHoJ+kZ/BDI7KBEZgVH78UVEjx9/ei18LBuAhTfkoGAWjYBSMAqwAAOTcTOAOyjxzAAAAAElFTkSuQmCC","orcid":"","institution":"China Academy of Traditional Chinese Medicine: China Academy of Chinese Medical Sciences","correspondingAuthor":true,"prefix":"","firstName":"Zhen-jiang","middleName":"","lastName":"Liu","suffix":""},{"id":592190705,"identity":"a3557453-61c1-44ba-8d7f-a27faf4cf7c6","order_by":1,"name":"Xiao-min Li","email":"","orcid":"","institution":"China Academy of Traditional Chinese Medicine: China Academy of Chinese Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Xiao-min","middleName":"","lastName":"Li","suffix":""},{"id":592190706,"identity":"4007d675-af36-42eb-9d88-1ff6db0f0819","order_by":2,"name":"Fei Liu","email":"","orcid":"","institution":"China Academy of Traditional Chinese Medicine: China Academy of Chinese Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Liu","suffix":""},{"id":592190707,"identity":"53c2c2ab-f4a4-4c22-928f-50d4e8386d3a","order_by":3,"name":"Zhi-meng Zhang","email":"","orcid":"","institution":"China Academy of Traditional Chinese Medicine: China Academy of Chinese Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Zhi-meng","middleName":"","lastName":"Zhang","suffix":""},{"id":592190708,"identity":"0f123e8c-f71f-415e-9512-62fa887b6af2","order_by":4,"name":"Chao-lu Wang","email":"","orcid":"","institution":"China Academy of Chinese Medical Sciences","correspondingAuthor":false,"prefix":"","firstName":"Chao-lu","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2026-02-10 15:38:48","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8842742/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8842742/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12967-026-08093-2","type":"published","date":"2026-04-07T15:58:55+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":102993259,"identity":"e0adeb53-b0d2-4f44-b512-5d0e9b28ad03","added_by":"auto","created_at":"2026-02-19 11:46:18","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":156428,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the patient selection process in the trial.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/0578f3f42b451da6fe03efaa.png"},{"id":103056471,"identity":"24bcb634-8d80-4783-8a0f-82ff3a155689","added_by":"auto","created_at":"2026-02-20 09:11:29","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":732767,"visible":true,"origin":"","legend":"\u003cp\u003eSubgroup analysis of the association between long-term enoxaparin exposure and osteoporosis risk across clinically relevant subgroups. BMI, body mass index; DVT, deep venous thrombosis; CI, confidence interval; OR, odds ratio.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/2a70901c5169fafa70306c8c.png"},{"id":102993256,"identity":"37512ad7-e6c1-457a-9765-971e6f0fa060","added_by":"auto","created_at":"2026-02-19 11:46:18","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":139828,"visible":true,"origin":"","legend":"\u003cp\u003eAssociation between enoxaparin exposure duration and osteoporosis risk assessed using restricted cubic spline analysis. Solid lines represent adjusted ORs, and shaded areas indicate 95% confidence intervals. CI, confidence interval; OR, odds ratio.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/3496ae3a142c22200559fc1e.png"},{"id":102993262,"identity":"b06c00f1-05e3-4742-a0b4-181b61fbfa6e","added_by":"auto","created_at":"2026-02-19 11:46:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":347047,"visible":true,"origin":"","legend":"\u003cp\u003eMediator analysis of the association between enoxaparin exposure duration and osteoporosis risk based on dosing intensity. A: Mean daily dosing frequency; B: Mean daily dose. ACME: Average causal mediating effect; ADE: Average direct effect; OP: Osteoporosis.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/e3c42ffd970003f7badc78cd.png"},{"id":103049721,"identity":"9ddce309-273f-484d-8c7e-859a5a4b14e0","added_by":"auto","created_at":"2026-02-20 07:45:11","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1003419,"visible":true,"origin":"","legend":"\u003cp\u003eScreening candidate targets and potential biological functions in osteoporosis. \u003cstrong\u003ePanel A:\u003c/strong\u003e Volcano plot of differentially expressed genes (DEGs) identified from the GEO dataset \u0026nbsp;(GEO accession: GSE35958). Red dots represent significantly upregulated genes, blue dots indicate significantly downregulated genes, and gray dots denote genes without significant differential expression. \u003cstrong\u003ePanel B:\u003c/strong\u003e Hierarchical clustering heatmap of DEGs between osteoporosis and control samples. Rows represent genes, and columns represent individual samples, with red and blue color gradients indicating relative upregulation and downregulation, respectively. \u003cstrong\u003ePanel C:\u003c/strong\u003eGene Ontology (GO) enrichment analysis of DEGs. The top enriched terms are shown for Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) categories, ranked by adjusted P values. \u003cstrong\u003ePanel D:\u003c/strong\u003e Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. The top enriched pathways are displayed, with circle size corresponding to the number of enriched genes and color intensity indicating −log10 (adjusted P value). \u003cstrong\u003ePanel E:\u003c/strong\u003e Venn diagram illustrating the overlap of osteoporosis-related targets obtained from the GeneCards, OMIM, and Therapeutic Target Database (TTD) databases. \u003cstrong\u003ePanel F:\u003c/strong\u003e Venn diagram showing the intersection between osteoporosis-related targets derived from public disease databases and DEGs identified from the GEO dataset.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/c28b3979d1f0f9a59b951cd8.png"},{"id":103050069,"identity":"ffaf91ce-6efc-43bb-b9c7-02119e5babe6","added_by":"auto","created_at":"2026-02-20 07:47:57","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1780534,"visible":true,"origin":"","legend":"\u003cp\u003eNetwork toxicology analysis identifying potential targets of enoxaparin associated with osteoporosis. \u003cstrong\u003ePanel A:\u003c/strong\u003e Two-dimensional chemical structure of enoxaparin sodium obtained from the PubChem database. \u003cstrong\u003ePanel B:\u003c/strong\u003eVenn diagram showing the overlap of predicted enoxaparin-related targets derived from the ChEMBL, STITCH, and SwissTargetPrediction databases. \u003cstrong\u003ePanel C: \u003c/strong\u003eVenn diagram illustrating the intersection between enoxaparin- associated targets and osteoporosis-related genes obtained from the GEO and disease-related databases, identifying shared candidate targets. \u003cstrong\u003ePanel D:\u003c/strong\u003eProtein–protein interaction (PPI) network of the overlapping targets constructed using the STRING database, where nodes represent proteins and edges indicate predicted or known protein–protein interactions. \u003cstrong\u003ePanel E:\u003c/strong\u003e Topological analysis of the PPI network highlighting hub genes based on network parameters, with node size and color intensity reflecting relative importance within the interaction network.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/db007966195b366b2ac69503.png"},{"id":102993257,"identity":"b9dd5302-087d-49e0-b158-be57949e7b11","added_by":"auto","created_at":"2026-02-19 11:46:18","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":1858394,"visible":true,"origin":"","legend":"\u003cp\u003eFunctional enrichment analysis of core target genes associated with osteoporosis under long-term enoxaparin exposure. \u003cstrong\u003ePanel A:\u003c/strong\u003e Gene Ontology (GO) enrichment bar plot showing the top enriched terms across the biological process (BP), cellular component (CC), and molecular function (MF) categories. The bar height represents the number of enriched genes for each term. \u003cstrong\u003ePanel B:\u003c/strong\u003e Circular GO enrichment visualization illustrating the distribution and relative contribution of enriched BP, CC, and MF terms, with gene counts, enrichment ratios, and statistical significance integrated across functional categories. \u003cstrong\u003ePanel C:\u003c/strong\u003e GO enrichment dot plot displaying the relationship between gene ratio and adjusted p-value for significantly enriched BP, CC, and MF terms. Dot size indicates the number of genes, and color intensity represents the level of statistical significance. \u003cstrong\u003ePanel D:\u003c/strong\u003e KEGG pathway enrichment analysis visualized by a gene–pathway network and corresponding dot plot, showing the association between core genes and significantly enriched pathways. The size and color of the dots reflect gene counts and −log10(p-value), respectively.\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/7649ac8ad8feac3ac04a61b0.png"},{"id":102993263,"identity":"ff12a365-8516-444a-b1f9-f97ed8c672ff","added_by":"auto","created_at":"2026-02-19 11:46:18","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1874342,"visible":true,"origin":"","legend":"\u003cp\u003eWeighted gene co-expression network analysis (WGCNA) identifying osteoporosis-associated gene modules under long-term enoxaparin exposure. \u003cstrong\u003ePanel A:\u003c/strong\u003e Sample clustering dendrogram used to detect potential outliers based on hierarchical clustering. One sample above the cutoff line was identified as an outlier and removed prior to network construction. \u003cstrong\u003ePanel B:\u003c/strong\u003e Sample dendrogram with corresponding trait heatmap showing the distribution of clinical phenotypes (control and OP) across samples, indicating consistency between expression-based clustering and clinical traits. \u003cstrong\u003ePanel C:\u003c/strong\u003e Scale-free topology fit index (R²) as a function of soft-thresholding power, showing that the network approaches scale-free topology at the selected power. \u003cstrong\u003ePanel D:\u003c/strong\u003e Mean connectivity analysis corresponding to different soft-thresholding powers, demonstrating a gradual decrease in connectivity with increasing power. \u003cstrong\u003ePanel E:\u003c/strong\u003e Gene dendrogram and module color assignment generated by hierarchical clustering and dynamic tree cutting, illustrating the identification of distinct gene co-expression modules. \u003cstrong\u003ePanel F:\u003c/strong\u003e Scatter plot showing the relationship between module membership (MM) and gene significance (GS) for osteoporosis in the maroon module, indicating a strong positive correlation between intramodular connectivity and phenotype relevance. \u003cstrong\u003ePanel G:\u003c/strong\u003e Scatter plot showing the relationship between module membership (MM) and gene significance (GS) for osteoporosis in the sienna3 module, further confirming the strong association between module genes and the OP phenotype.\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/25fdd2407e5fb72c0c33f425.png"},{"id":103050100,"identity":"8d99ef13-b789-421d-83f3-ffa7dcbdc867","added_by":"auto","created_at":"2026-02-20 07:48:12","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":640621,"visible":true,"origin":"","legend":"\u003cp\u003eMachine learning–based identification of core genes associated with osteoporosis risk under long-term enoxaparin exposure. \u003cstrong\u003ePanel A:\u003c/strong\u003e Ten-fold cross-validation curve of the LASSO regression model showing the relationship between binomial deviance and log-transformed regularization parameter (λ). The dotted vertical lines indicate the optimal λ values, and the numbers above the plot represent the number of non-zero coefficients. \u003cstrong\u003ePanel B:\u003c/strong\u003e LASSO coefficient profiles of candidate genes as a function of log(λ), illustrating the shrinkage of regression coefficients with increasing penalization and the selection of key features. \u003cstrong\u003ePanel C:\u003c/strong\u003e Direction and magnitude of LASSO regression coefficients for the selected genes. Positive coefficients (red bars) indicate genes positively associated with osteoporosis risk, whereas negative coefficients (blue bars) indicate genes potentially associated with reduced risk. \u003cstrong\u003ePanel D: \u003c/strong\u003eVenn diagram showing the intersection between genes identified by WGCNA and those selected by LASSO regression. Two overlapping genes, CDK16 and VHL, were identified as core genes and selected for subsequent functional enrichment and molecular docking analyses.\u003c/p\u003e","description":"","filename":"image9.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/411cd3a732a2336f1dfeb60e.png"},{"id":102993266,"identity":"af2cf21b-8cbe-4c63-9ceb-9d20581f0a24","added_by":"auto","created_at":"2026-02-19 11:46:19","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":2529249,"visible":true,"origin":"","legend":"\u003cp\u003eMolecular docking results of enoxaparin sodium with target proteins. Molecular docking results of enoxaparin sodium with \u003cstrong\u003ePanel A:\u003c/strong\u003e CDK16(−8.8 kcal/mol), \u003cstrong\u003ePanel B:\u003c/strong\u003e VHL (−6.2kcal/mol). The results showed the binding positions of the enoxaparin sodium molecule within the surface cavities of each protein and the interactions with the residues.\u003c/p\u003e","description":"","filename":"image10.png","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/521df7d2a5172ab2e676f8ad.png"},{"id":106809627,"identity":"6085b8dc-ae27-4889-8e4b-5e71fcfe3744","added_by":"auto","created_at":"2026-04-13 16:11:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":12442684,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/92699964-f150-4932-bd6c-6b73d99b4c59.pdf"},{"id":103049569,"identity":"6288380f-0714-44d8-84a9-55396c2adeb4","added_by":"auto","created_at":"2026-02-20 07:42:44","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":38952,"visible":true,"origin":"","legend":"","description":"","filename":"strocss2025GuidelineChecklist.docx","url":"https://assets-eu.researchsquare.com/files/rs-8842742/v1/8a10f80d1f70684c6ab88dc5.docx"}],"financialInterests":"","formattedTitle":"Long-Term Enoxaparin Use and Osteoporosis Risk: A Real-World Cohort Study with Integrative Computational and Network Toxicology Approaches","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOsteoporosis is a systemic skeletal disorder characterized by reduced bone mass and impaired bone microarchitecture, with its primary clinical outcomes being increased bone fragility and significantly elevated fracture risk [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. The accelerating pace of global population aging has led to a persistent rise in the incidence of osteoporosis [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Statistics indicate that approximately 200\u0026nbsp;million people worldwide suffer from osteoporosis, affecting about 30% of postmenopausal women and 20% of men over 65 years of age [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Osteoporosis has become a serious public health issue, substantially increasing disability rates, mortality, and healthcare costs among the elderly and those with chronic diseases [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Osteoporotic fractures, particularly hip and spinal fractures, often lead to long-term functional impairment and reduced quality of life and are closely associated with increased all-cause mortality risk [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe development of osteoporosis results from the combined effects of multiple factors, including traditional risk factors such as aging, changes in sex hormone levels, genetic susceptibility, and metabolic abnormalities, as well as modifiable factors like drug exposure [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. In recent years, drug-related bone loss has garnered increasing attention, particularly among hospitalized patients and those requiring long-term medication, as its potential impact on skeletal health cannot be overlooked [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAnticoagulant therapy is widely employed in clinical practice for the prevention and treatment of venous thromboembolism during the perioperative period of orthopedic, trauma, and major abdominal surgeries [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Enoxaparin sodium, as a low molecular weight heparin, has become one of the most commonly used anticoagulants in clinical practice due to its stable pharmacokinetic properties, reliable anticoagulant effects, and ease of use [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Compared with standard heparin, enoxaparin is generally considered to carry a lower risk of adverse skeletal effects [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. However, a growing body of experimental and clinical research suggests that long-term exposure to heparin-like drugs may adversely affect bone metabolism by suppressing osteoblast activity, enhancing osteoclast-mediated bone resorption, and disrupting calcium homeostasis [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Despite enoxaparin's widespread and prolonged clinical use, systematic and reliable real-world evidence linking it to osteoporosis risk remains lacking.\u003c/p\u003e \u003cp\u003eFrom a biological mechanism perspective, the development of osteoporosis stems not only from an imbalance between bone formation and resorption but is also closely associated with signaling pathways involving oxidative stress, chronic inflammation, hypoxia responses, and abnormal transcriptional regulation [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Long-term drug exposure may accelerate bone loss by activating inflammatory responses and oxidative stress, inhibiting osteoblast differentiation, promoting osteoclast generation, and disrupting the bone microenvironment [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. However, the specific molecular mechanisms through which prolonged enoxaparin exposure mediates osteoporosis remain poorly understood.\u003c/p\u003e \u003cp\u003eWith the continuous refinement of large-scale electronic health record databases, real-world data has emerged as a crucial platform for evaluating long-term drug safety and its long-term outcomes [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Concurrently, advances in systems biology and computational biology methods have enabled the integration of clinical epidemiological research with transcriptomics and network analysis [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Weighted Gene Co-expression Network Analysis (WGCNA) identifies gene co-expression modules closely associated with disease, while machine learning methods aid in screening key risk features from high-dimensional data. Network toxicology and molecular docking analysis provide mechanistic support for elucidating potential connections between drugs, targets, and pathways [\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eBased on this, the present study aims to systematically evaluate the association between long-term enoxaparin use and osteoporosis risk and to explore its potential molecular mechanisms. Using the Medical Information Modeling and Informatics Consortium version IV (MIMIC-IV) database, we assessed the relationship between enoxaparin exposure duration and the risk of osteoporosis development through multivariate regression, subgroup analysis, nonlinear modeling, and mediation analysis. Subsequently, integrating network toxicology, transcriptomic data analysis, WGCNA, and machine learning methods, we screened key regulatory genes potentially mediating enoxaparin-associated bone loss and validated potential interactions between enoxaparin and core targets through molecular docking analysis. By integrating real-world clinical evidence with multi-level computational biology analysis, this study provides a systematic evaluation of skeletal safety during long-term enoxaparin therapy and offers novel theoretical insights into its potential molecular mechanisms.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Data Availability\u003c/h2\u003e \u003cp\u003eData are available from the corresponding author upon reasonable request. This study has been reported per the STROCSS 2025 guidelines [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. All data used in this research were obtained from the publicly accessible MIMIC-IV database, a large, de-identified critical care dataset jointly developed by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Access to the MIMIC-IV repository was granted to the lead author (Zhen-Jiang Liu, credential ID 12911307) after successful completion of the National Institutes of Health web-based human-subjects protection training. The Institutional Review Board of Beth Israel Deaconess Medical Center reviewed and approved the acquisition of patient-level information and the creation of the research dataset, subsequently authorizing data sharing and waiving the requirement for informed consent. All analytical procedures adhered strictly to regulations governing patient privacy and data confidentiality. No proprietary or restricted-access datasets were used.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Study Design and Population\u003c/h2\u003e \u003cp\u003eWe conducted a retrospective cohort analysis using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, a large, freely accessible critical care dataset jointly maintained by the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, spanning the years between 2008 and 2019. We included adult patients aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years who had complete medication administration records documenting the use of enoxaparin and had complete follow-up data relevant to the assessment of osteoporosis, ensuring accurate identification of exposure duration and dosage patterns. Exposure was defined as long-term use of enoxaparin (\u0026ge;\u0026thinsp;30 days) and non-long-term use (\u0026lt;\u0026thinsp;30 days), while the primary outcome was the development or documented diagnosis of osteoporosis during or after the exposure period. As MIMIC-IV contains fully de-identified patient information, prospective registration with a clinical research registry was not required. After applying exclusion criteria\u0026mdash;patients younger than 18 years, those lacking medication exposure records, those with incomplete diagnostic or covariate information, and individuals with pre-existing osteoporosis\u0026mdash;the final analytic cohort consisted of 23852 eligible adult patients meeting all inclusion conditions. The MIMIC-IV project was approved by the Institutional Review Boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, with a waiver of informed consent due to the use of de-identified data.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Outcome Measurement\u003c/h2\u003e \u003cp\u003eThe primary outcome was the development of osteoporosis following enoxaparin exposure, identified through International Classification of Diseases (ICD) diagnostic codes recorded in the MIMIC-IV database. Osteoporosis was defined using ICD-9 codes (73300, 73301, 73302, 73303, 73309) and ICD-10 codes (M80-M82), encompassing both age-related and secondary osteoporosis. To ensure temporal validity, only cases in which osteoporosis was diagnosed after the initiation of enoxaparin therapy were classified as outcome events. Patients with any documented osteoporosis diagnosis prior to exposure served as exclusions. All outcome classifications were based on clinician-entered diagnoses from inpatient hospital encounters, ensuring clinical verification.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Covariates\u003c/h2\u003e \u003cp\u003eCovariates were selected based on clinical relevance and established protocols in prior pharmacoepidemiologic studies using the MIMIC database. Demographic variables included age, sex, and race, categorized in MIMIC-IV as White, Black, Asian, Latino, and Other. Body mass index (BMI) was included as a continuous variable given its established association with bone metabolism. Clinical comorbidities included documented deep vein thrombosis (DVT) according to ICD-coded diagnoses in the medical record. To characterize enoxaparin exposure patterns, medication-related variables were also incorporated as covariates, including total cumulative dose, number of administrations, total duration of therapy (days), and average daily dose, each extracted from the complete medication administration records. These variables reflect both the intensity and duration of anticoagulant exposure and were selected to account for dose-dependent effects on bone health. All covariates were chosen a priori based on biological plausibility and literature-supported associations with osteoporosis risk, anticoagulation exposure, or both. All candidate covariates were screened using univariate logistic regression, and those with p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 (including age, sex, DVT status, total cumulative enoxaparin dose, number of administrations, total days of therapy, and average daily dose) were retained for multivariable adjustment. Participants with missing data for any covariate (\u0026lt;\u0026thinsp;5% of the total cohort) were excluded using complete-case deletion.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Acquisition of Osteoporosis Targets\u003c/h2\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e2.5.1 Acquisition of differentially expressed gene of Osteoporosis through bioinformatics analysis\u003c/h2\u003e \u003cp\u003eWe obtained microarray dataset GSE35958 from the GEO database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). For data normalization, log2 transformation was applied, and individual batch effects were eliminated via the removeBatchEffect in the limma R package. Additionally, the limma R package was utilized to identify differentially expressed genes (DEGs) between osteoporosis samples and normal samples, which were then visualized through heat maps and volcano plots. The criteria were set at adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log2 Fold Change| \u0026gt; 1.5. The Database for Annotation, Visualization and Integrated Discovery (DAVID) was used to collect data for conducting GO analysis and enrichment with the KEGG pathways. GO analysis categorizes enriched terms into biological processes (BP), cellular components (CC), and molecular functions (MF), which are used to provide insights into the roles of potential targets. KEGG was used to systematically analyze the gene functions and metabolic pathways involved in the target genes, which is helpful to fully understand the overall effect of the predicted targets in the organism.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e2.5.2 Identify overlapping genes targets\u003c/h2\u003e \u003cp\u003eTo identify more osteoporosis-related targets, three databases\u0026mdash;GeneCards, OMIM, and TTD\u0026mdash;were queried using \u0026ldquo;osteoporosis\u0026rdquo; as the keyword. The genes in the GeneCards database are screened based on the median gene relevance score to ensure validity. Integrate the obtained genes with the genes obtained from bioinformatics analysis and eliminate duplicate genes to obtain the final disease target genes.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.6 Identification of enoxaparin-associated targets\u003c/h2\u003e \u003cp\u003eThe chemical structure and SMILES notation of enoxaparin sodium were downloaded from PubChem (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubchem.ncbi.nlm.nih.gov\u003c/span\u003e\u003cspan address=\"https://pubchem.ncbi.nlm.nih.gov\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Potential enoxaparin target proteins were predicted using ChEMBL (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ebi.ac.uk/chembl/\u003c/span\u003e\u003cspan address=\"https://www.ebi.ac.uk/chembl/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), STITCH (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://stitch-db.org/\u003c/span\u003e\u003cspan address=\"https://stitch-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and Swiss Target Prediction (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://swisstargetprediction.ch/\u003c/span\u003e\u003cspan address=\"https://swisstargetprediction.ch/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), with species restricted to \u0026ldquo;Homo sapiens\u0026rdquo; and a prediction probability\u0026thinsp;\u0026gt;\u0026thinsp;0. The intersecting genes between predicted enoxaparin targets and osteoporosis-associated targets were identified using Venn diagram analysis. These overlapping genes were considered potential mechanistic mediators of long-term enoxaparin exposure contributing to osteoporosis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.7 Construction of the protein\u0026ndash;protein interaction (PPI) network and identification of core target genes\u003c/h2\u003e \u003cp\u003eBioinformatic analysis of the overlapping genes was conducted via the STRING platform to map PPI networks. The STRING database analysis was conducted with taxonomic restriction to Homo sapiens, implementing an interaction confidence cut-off \u0026gt;\u0026thinsp;0.4, and medium false discovery rate (FDR) stringency. The resultant network data were then processed through Cytoscape 3.10.1 for topological mapping. Network topology evaluation was conducted using CytoNCA, a dedicated plugin for identifying densely connected functional clusters within protein interaction networks. Nodes with centrality values above the median were identified as core targets.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.8 GO and KEGG pathway analysis\u003c/h2\u003e \u003cp\u003eGene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted using the ClusterProfiler package (version 4.14.6). The intersecting targets from the Venn diagram were submitted to the ClusterProfiler Package, with \u0026ldquo;Homo sapiens\u0026rdquo; selected. GO terms were categorized by biological process (BP), molecular function (MF), and cellular component (CC), and pathways were sorted by p-value. Visualizations for the top GO terms and KEGG pathways were generated using the Microbiotics website (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.bioinformatics.com.cn/\u003c/span\u003e\u003cspan address=\"http://www.bioinformatics.com.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), highlighting the most significant results.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e2.9 Construction of the Weighted Gene Co-expression Network and Identification of Key Modules\u003c/h2\u003e \u003cp\u003eWGCNA was mainly employed to screen the key modules and obtain hub genes strongly correlated with disease traits. The \u0026ldquo;WGCNA\u0026rdquo; package in R software facilitated the identification and visualization of unsigned co-expression network modules within the GSE35958 dataset. The construction of the weighted gene network began with the selection of an optimal soft thresholding power (β). This parameter was chosen using the pickSoftThreshold function, which explored a range from 1 to 20. This soft threshold was then applied to form a co-expression similarity matrix, which was subsequently raised to calculate the adjacency matrix. The selection of the best power to establish this adjacency matrix was guided by scale-free topology criteria, ensuring strong gene-gene correlations. Following this, the adjacency matrix was transformed into a Topological Overlap Matrix (TOM) to measure dissimilarities. A hierarchical clustering function was then employed to group genes with similar expression profiles into distinct modules. The derived gene modules were subsequently associated with clinical features by assessing both gene significance (GS) and module membership (MM). Modules exhibiting a positive correlation with these clinical features were identified and designated as key co-expression modules.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e2.10 Construction of diagnostic model by machine learning\u003c/h2\u003e \u003cp\u003eThe Lasso regression model was employed to identify biomarkers most significantly associated with OP among 134 differentially expressed genes. The regularization parameter (λ) was optimized through cross-validation, and model performance was evaluated by tracking the binomial bias value. By visualizing biomarker coefficients at different λ values, markers maintaining statistical significance across λ variations were identified. Influence direction maps were constructed for these markers, where positive and negative coefficients represent their respective promoting or suppressing effects on the target outcome. The biomarker with the most significant coefficient was ultimately selected as a potential key predictor. The results were cross-referenced with biomarkers identified by the WGCNA method. A Venn diagram identified overlapping biomarkers between the two approaches, highlighting the most consistent biomarkers across different analytical pathways. These overlapping biomarkers were considered key indicators and potential mechanism-mediated factors in the study.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e2.11 Molecular docking analysis of Enoxaparin Sodium with core OP-related targets\u003c/h2\u003e \u003cp\u003eA comprehensive examination of the interactions between Enoxaparin Sodium-OP and the key target proteins was identified through utilizing the molecular docking approach. The three-dimensional coordinates of core target proteins were acquired from the Protein Data Bank (PDB), while the Enoxaparin Sodium-OP\u0026rsquo; stereochemical configuration was retrieved from PubChem (CID: 16667706). Structural optimization procedures in PyMOL 3.1.3.1 involved ligand excision and solvent molecule elimination. Following hydrogen atom addition via AutoDockTools\u0026rsquo; structural refinement module (v1.5.6). Molecular docking simulations were executed through the AutoDock Vina (v1.1.x) command-line interface, with resultant binding poses rendered in PyMOL (v3.1.3.1).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e2.12 Statistical Analysis\u003c/h2\u003e \u003cp\u003eContinuous variables were examined for normality using the Shapiro\u0026ndash;Wilk test prior to further statistical analysis. Continuous variables were presented as median (interquartile range) and compared between groups using the Wilcoxon rank-sum test. Categorical variables were represented as frequencies and percentages and compared between groups using the chi-square test. Descriptive statistics were used to summarize baseline characteristics of the study cohort. Continuous variables were reported as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD) or median (interquartile range, IQR), as appropriate, while categorical variables were presented as counts and percentages. Between-group comparisons were performed using Student\u0026rsquo;s t-test or the Wilcoxon rank-sum test for continuous variables and the chi-square test (or Fisher\u0026rsquo;s exact test when appropriate) for categorical variables, consistent with commonly used reporting in epidemiologic analyses. For improved clarity in the representation of the odd ratio (OR), total days of administrations value was divided by 10. The association between long-term enoxaparin exposure and osteoporosis was evaluated using logistic regression models, and effect estimates were expressed as odds ratios with 95% confidence intervals (CIs). Covariates were selected a priori based on clinical relevance (age, sex, race, BMI, DVT status, and enoxaparin exposure pattern variables), and candidate covariates were screened using univariate logistic regression; variables meeting the prespecified threshold (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) were included in multivariable adjustment. Model 1 was unadjusted; Model 2 was adjusted for age, sex, race, BMI, and DVT; and Model 3 was adjusted for age, sex, race, BMI, DVT, average number per day, and daily average dose. To assess potential effect modification, prespecified subgroup analyses were conducted across clinically relevant strata (sex, age, BMI, race, DVT, average number per day, daily average dose), and interaction was tested by adding multiplicative interaction terms to multivariable models with significance assessed via P for interaction. To characterize a potential nonlinear exposure\u0026ndash;response relationship, restricted cubic spline (RCS) regression was applied to model the association between enoxaparin exposure duration (days) and osteoporosis risk; overall and nonlinearity P values were reported to evaluate whether the relationship deviated from linearity. The mediating roles of enoxaparin administration intensity (average administrations/day and daily average dose) in the exposure duration\u0026ndash;osteoporosis association were examined using causal mediation analysis, reporting the average causal mediation effect (ACME), average direct effect (ADE), and proportion mediated. All analyses were conducted in R. A two-tailed p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 defined statistical significance. All statistical analyses were conducted with R 4.0.2, in conjunction with SPSS 25.0.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Baseline characteristics of all participants\u003c/h2\u003e \u003cp\u003eAmong the 23,852 participants receiving enoxaparin sodium who were included in the analysis, 902 were categorized into the long-term exposure group, while 22,950 were assigned to the non\u0026ndash;long-term exposure group. As shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, participants in the long-term exposure group were more likely to be male and were significantly younger, with a lower median age, compared with those in the non\u0026ndash;long-term exposure group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). No significant difference in body mass index was observed between the two groups (P\u0026thinsp;=\u0026thinsp;0.968). Notably, the prevalence of osteoporosis was significantly higher among participants in the long-term exposure group than among those in the non\u0026ndash;long-term exposure group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In addition, the total cumulative dose, total number of administrations, total duration of use, and average daily dose of enoxaparin sodium were all significantly greater in the long-term exposure group (all P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of the osteoporosis and non-osteoporosis groups\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;23852)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOsteoporosis group\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;2291)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNon-osteoporosis group\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;21561)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eSex, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10331(43.81)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e336(14.67)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9995(46.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13521(56.19)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1955(85.33)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e11566 (53.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e64.00(51.00,76.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e78.00(67.00, 86.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e62.00(49.00, 74.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e27.37(23.63, 32.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24.94(20.68, 28.62)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27.62 (23.82, 32.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eRace, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASIAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e634(2.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e63(2.75)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e571(2.65)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBLACK\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2938(12.32)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e178 (7.77)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2760(12.80)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLATINO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1007(4.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e70 (3.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e937(4.35)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOTHER\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1964(8.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e129(5.63)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1835(8.51)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWHITE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17309(72.57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1851(80.79)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15458(71.69)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eDVT, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.626\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2655(11.13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e262 (11.44)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2393 (11.10)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21197(88.87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2029 (88.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e19168 (88.90)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal dose(mg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e320.00(180.00,660.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e320.00(200.00, 640.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e320.00(180.00, 660.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.995\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal number of administrations(times)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7.00(4.00,12.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8.00(5.00, 14.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.00(4.00, 12.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal days of administrations(days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.00(3.00,9.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.00(4.00, 10.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.00(3.00, 9.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage number per day(times)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.00(1.00, 2.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.00(1.00, 2.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.00(1.00, 2.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDaily average dose(mg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e49.00(40.00,120.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40.00(40.00, 87.50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e50.00(40.00, 120.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAmong all participants, 2,291 were identified as having osteoporosis. Compared with participants without osteoporosis, those with osteoporosis were significantly older and predominantly female, and they exhibited a lower body mass index (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Regarding race/ethnicity, participants with osteoporosis were more likely to be non-Hispanic White, whereas the proportions of Black, Latino, and other racial groups were lower in the osteoporosis group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The prevalence of deep vein thrombosis did not differ significantly between participants with and without osteoporosis (P\u0026thinsp;=\u0026thinsp;0.626). With respect to enoxaparin sodium exposure, although no significant difference was observed in total cumulative dose between the two groups (P\u0026thinsp;=\u0026thinsp;0.995), participants with osteoporosis had a significantly longer duration of treatment and higher exposure-related measures than those without osteoporosis (all P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Detailed baseline characteristics are summarized in Tables\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of the long-term medication and non-long-term medication groups\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariables\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall(n\u0026thinsp;=\u0026thinsp;23852)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLong-term medication group(n\u0026thinsp;=\u0026thinsp;902)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNon-long-term medication group(n\u0026thinsp;=\u0026thinsp;22950)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eSex, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10331(43.81)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e467(51.77)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9864 (42.98)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13521(56.19)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e435(48.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e13086(57.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge(years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e64.00(51.00,76.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e60.00(50.00,69.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e64.00(51.00,76.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27.37(23.63, 32.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e27.33(23.47, 31.86)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27.37(23.63, 32.02)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.968\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eRace, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.694\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eASIAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e634(2.66)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e29 (3.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e605 (2.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBLACK\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2938(12.32)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e117 (12.97)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2821 (12.29)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLATINO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1007(4.22)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e36 (3.99)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e971 (4.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOTHER\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1964(8.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e67 (7.43)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1897 (8.27)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWHITE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17309(72.57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e653 (72.39)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e16656 (72.58)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eOsteoporosis, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e0.002\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2291(9.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e114(12.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2177 (9.49)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21561(90.39)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e788(87.36)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20773(90.51)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDVT, n(%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2655(11.13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e228(25.28)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2427 (10.58)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21197(88.87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e674(74.72)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20523 (89.42)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal dose(mg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e320.00(180.00,660.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3960.00(2080.00,6480.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e320.00(180.00,600.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal number of administrations(times)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.00(4.00,12.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e64.00(47.00,93.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.00(4.00,12.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal days of administrations(days)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.00(3.00,9.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e42(34.00,57.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.00(3.00,8.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage number per day(times)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.00(1.00, 2.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.54 (1.08, 1.91)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.00 (1.00, 2.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDaily average dose(mg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49.00(40.00,120.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e87.53(43.46,136.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e47.69(40.00-120.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Duration-dependent association between enoxaparin exposure and osteoporosis risk\u003c/h2\u003e \u003cp\u003eMultivariable regression models, with sequential adjustment for demographic characteristics, clinical covariates, and enoxaparin dosing parameters, were used to assess the association between long-term enoxaparin exposure and the risk of osteoporosis (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In the unadjusted model (Model 1), a significant association was observed between longer duration of enoxaparin exposure and an increased risk of osteoporosis. When exposure duration was modeled as a continuous variable, each unit increase in exposure duration was associated with an 8% higher risk of osteoporosis (OR\u0026thinsp;=\u0026thinsp;1.08, 95% CI: 1.06\u0026ndash;1.11; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). When enoxaparin exposure duration was categorized into quartiles\u0026mdash;Q1 (\u0026lt;\u0026thinsp;15 days), Q2 (15\u0026ndash;30 days), Q3 (30\u0026ndash;90 days), and Q4 (\u0026gt;\u0026thinsp;90 days)\u0026mdash;a clear duration\u0026ndash;response relationship was observed (P for trend\u0026thinsp;=\u0026thinsp;0.035). Compared with patients exposed for less than 15 days (Q1), those receiving enoxaparin for 15\u0026ndash;30 days (Q2) and 30\u0026ndash;90 days (Q3) showed progressively higher risks of osteoporosis (OR\u0026thinsp;=\u0026thinsp;1.19, 95% CI: 1.03\u0026ndash;1.38; P\u0026thinsp;=\u0026thinsp;0.019 and OR\u0026thinsp;=\u0026thinsp;1.32, 95% CI: 1.07\u0026ndash;1.64; P\u0026thinsp;=\u0026thinsp;0.011, respectively). The highest risk was observed among patients with prolonged exposure exceeding 90 days (Q4), who exhibited more than a twofold increased risk of osteoporosis (OR\u0026thinsp;=\u0026thinsp;2.31, 95% CI: 1.31\u0026ndash;4.06; P\u0026thinsp;=\u0026thinsp;0.004). After adjustment for age, sex, race, body mass index, and deep vein thrombosis status (Model 2), the association between enoxaparin exposure duration and osteoporosis risk remained robust. Each unit increase in exposure duration was associated with a 16% increase in osteoporosis risk (OR\u0026thinsp;=\u0026thinsp;1.16, 95% CI: 1.08\u0026ndash;1.24; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Relative to short-term exposure (\u0026lt;\u0026thinsp;15 days), patients exposed for more than 90 days continued to demonstrate a markedly elevated risk of osteoporosis (OR\u0026thinsp;=\u0026thinsp;12.23, 95% CI: 4.10\u0026ndash;36.51; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with a statistically significant increasing trend across exposure categories (P for trend\u0026thinsp;=\u0026thinsp;0.023). In the fully adjusted model (Model 3), which additionally accounted for enoxaparin administration intensity, including average daily dose and average number of administrations per day, the association was modestly attenuated but remained statistically significant. Each unit increase in exposure duration was associated with a 15% higher risk of osteoporosis (OR\u0026thinsp;=\u0026thinsp;1.15, 95% CI: 1.08\u0026ndash;1.23; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Compared with patients treated for less than 15 days, those with prolonged exposure exceeding 90 days retained a substantially increased risk of osteoporosis (OR\u0026thinsp;=\u0026thinsp;11.94, 95% CI: 3.98\u0026ndash;35.79; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Importantly, the duration-dependent trend persisted across all models (P for trend\u0026thinsp;=\u0026thinsp;0.015), supporting a dose\u0026ndash;time\u0026ndash;response relationship between long-term enoxaparin use and osteoporosis risk.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLogistic regression analysis of the associations between long-term enoxaparin exposure and the risk of osteoporosis in all participants\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eCategories\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eModel 1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e \u003cp\u003eModel 2\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c10\" namest=\"c8\"\u003e \u003cp\u003eModel 3\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eP for trend\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eP for trend\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eOR(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eP for trend\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eContinuous variable per unit\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.08 (1.06\u0026thinsp;~\u0026thinsp;1.11)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.16 (1.08\u0026thinsp;~\u0026thinsp;1.24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1.15 (1.08\u0026thinsp;~\u0026thinsp;1.23)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQuartile\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.035\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003e0.023\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u003cb\u003e0.015\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ1(N\u0026thinsp;=\u0026thinsp;20931)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"9\" nameend=\"c10\" namest=\"c2\"\u003e \u003cp\u003eRef\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ2(N\u0026thinsp;=\u0026thinsp;2018)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.19 (1.03\u0026thinsp;~\u0026thinsp;1.38)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e0.019\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.42 (1.02\u0026thinsp;~\u0026thinsp;1.98)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e0.038\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1.35 (0.97\u0026thinsp;~\u0026thinsp;1.88)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.079\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ3(N\u0026thinsp;=\u0026thinsp;825)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.32 (1.07\u0026thinsp;~\u0026thinsp;1.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e0.011\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.44 (0.83\u0026thinsp;~\u0026thinsp;2.50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.199\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1.42 (0.81\u0026thinsp;~\u0026thinsp;2.47)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.219\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eQ4(N\u0026thinsp;=\u0026thinsp;78)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.31 (1.31\u0026thinsp;~\u0026thinsp;4.06)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e0.004\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e12.23 (4.10\u0026thinsp;~\u0026thinsp;36.51)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e11.94 (3.98\u0026thinsp;~\u0026thinsp;35.79)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"10\"\u003eModel 1: unadjusted\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"10\"\u003eModel 2: adjusted for age, sex, race, BMI, DVT\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"10\"\u003eModel 3: adjusted for age, sex, race, BMI, DVT, average number per day, daily average dose\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Subgroup analysis\u003c/h2\u003e \u003cp\u003eSubgroup analyses were performed to assess the consistency of the association between long-term enoxaparin exposure and osteoporosis risk across clinically relevant strata (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Overall, long-term enoxaparin exposure was significantly associated with an increased risk of osteoporosis in the total population (OR\u0026thinsp;=\u0026thinsp;1.38, 95% CI: 1.13\u0026ndash;1.69; P\u0026thinsp;=\u0026thinsp;0.002). This association was consistently observed in both female (OR\u0026thinsp;=\u0026thinsp;1.59, 95% CI: 1.26\u0026ndash;2.02; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and male patients (OR\u0026thinsp;=\u0026thinsp;1.58, 95% CI: 1.02\u0026ndash;2.44; P\u0026thinsp;=\u0026thinsp;0.039), with no significant interaction by sex (P for interaction\u0026thinsp;=\u0026thinsp;0.976). Stratified analyses by age revealed a stronger association among patients younger than 65 years (OR\u0026thinsp;=\u0026thinsp;2.68, 95% CI: 1.95\u0026ndash;3.69; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), whereas the association was not statistically significant among those aged 65 years or older (OR\u0026thinsp;=\u0026thinsp;1.22, 95% CI: 0.93\u0026ndash;1.60; P\u0026thinsp;=\u0026thinsp;0.143), and a significant interaction with age was observed (P for interaction\u0026thinsp;\u0026lt;\u0026thinsp;0.001). With respect to race, a significant association was detected only among White patients (OR\u0026thinsp;=\u0026thinsp;1.50, 95% CI: 1.20\u0026ndash;1.87; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while no significant associations were observed in other racial groups, and no significant interaction was identified (P for interaction\u0026thinsp;=\u0026thinsp;0.214). The association between enoxaparin exposure and osteoporosis risk was also evident among patients without deep vein thrombosis (OR\u0026thinsp;=\u0026thinsp;1.42, 95% CI: 1.13\u0026ndash;1.78; P\u0026thinsp;=\u0026thinsp;0.003), but not among those with deep vein thrombosis, with no significant interaction (P for interaction\u0026thinsp;=\u0026thinsp;0.615). Furthermore, patients receiving more than one enoxaparin administration per day (OR\u0026thinsp;=\u0026thinsp;1.66, 95% CI: 1.10\u0026ndash;2.49; P\u0026thinsp;=\u0026thinsp;0.015) and those with a daily average dose\u0026thinsp;\u0026ge;\u0026thinsp;49 mg (OR\u0026thinsp;=\u0026thinsp;1.64, 95% CI: 1.28\u0026ndash;2.10; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) exhibited higher osteoporosis risk; however, no significant interactions were observed for administration frequency or daily dose (P for interaction\u0026thinsp;=\u0026thinsp;0.552 and 0.139, respectively).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Nonlinear association between enoxaparin exposure duration and osteoporosis risk\u003c/h2\u003e \u003cp\u003eRestricted cubic spline analysis was performed to explore the potential nonlinear relationship between enoxaparin exposure duration and the risk of osteoporosis (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The analysis demonstrated a significant overall association between exposure duration and osteoporosis risk (P for overall\u0026thinsp;\u0026lt;\u0026thinsp;0.001), along with a statistically significant nonlinear relationship (P for nonlinearity\u0026thinsp;\u0026lt;\u0026thinsp;0.001). As shown in the spline curve, the odds of osteoporosis increased rapidly during the early phase of enoxaparin exposure, followed by a more gradual but sustained increase with longer exposure duration. The risk continued to rise as exposure days accumulated, particularly beyond prolonged treatment periods, indicating a clear time-dependent pattern. These findings suggest that longer durations of enoxaparin use are associated with progressively higher osteoporosis risk in a nonlinear manner.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Mediation effects of enoxaparin administration intensity on the association between exposure duration and osteoporosis risk\u003c/h2\u003e \u003cp\u003eMediation analyses were conducted to examine whether enoxaparin administration intensity mediated the association between exposure duration and osteoporosis risk (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). The results indicated that the average number of administrations per day partially mediated the relationship between days of enoxaparin exposure and osteoporosis, with a significant average causal mediation effect (ACME\u0026thinsp;=\u0026thinsp;0.000321, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The average direct effect (ADE) of exposure duration on osteoporosis risk remained statistically significant (ADE\u0026thinsp;=\u0026thinsp;0.007211, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and the proportion of the total effect mediated through administration frequency was estimated at 4.43% (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Similarly, daily average dose was identified as a significant mediator in the association between exposure duration and osteoporosis risk (ACME\u0026thinsp;=\u0026thinsp;0.000422, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while the direct effect of exposure duration remained significant (ADE\u0026thinsp;=\u0026thinsp;0.007622, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The proportion of the association mediated by daily average dose was 6.01% (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). These findings suggest that enoxaparin administration intensity partially mediates the relationship between prolonged exposure duration and increased osteoporosis risk.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Screening candidate targets and analyzing the potential biological functions in osteoporosis\u003c/h2\u003e \u003cp\u003eTo screen candidate targets associated with osteoporosis, transcriptomic data from bone mesenchymal stem cell samples were retrieved from the Gene Expression Omnibus (GEO) database (accession number: GSE35958), which includes gene expression profiles from 4 healthy controls and 5 patients with osteoporosis. Differentially expressed genes (DEGs) between osteoporosis and healthy states were identified using the limma package in R. Significant transcriptional alterations were visualized through hierarchical clustering heatmaps and volcano plots. The volcano plot revealed a clear separation in gene expression patterns between osteoporosis and control groups, based on the thresholds of P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log₂ fold change| \u0026gt; 1.5 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). Hierarchical clustering analysis based on the top DEGs further revealed distinct expression patterns between osteoporosis and control samples, indicating robust transcriptional heterogeneity associated with disease status (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003eTo detect the biological characteristics of DEGs in osteoporosis, we performed GO and KEGG analysis via the DAVID database. Gene Ontology (GO) enrichment analysis of the identified DEGs showed significant enrichment in multiple biological processes, cellular components, and molecular functions (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). Notably, enriched biological processes were primarily related to extracellular matrix organization, collagen fibril organization, cell\u0026ndash;substrate adhesion, and regulation of protein targeting. Cellular component analysis highlighted endoplasmic reticulum lumen, Golgi apparatus, and extracellular matrix\u0026ndash;associated structures, while molecular function analysis emphasized DNA-binding transcription factor activity, RNA polymerase II\u0026ndash;specific binding, cadherin binding, and phosphoric ester hydrolase activity. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed that the DEGs were significantly involved in multiple signaling pathways relevant to bone metabolism and inflammatory regulation (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). The most enriched pathways included PI3K\u0026ndash;Akt signaling, focal adhesion, human cytomegalovirus infection, regulation of actin cytoskeleton, cellular senescence, insulin signaling, estrogen signaling, HIF-1 signaling, and phospholipase D signaling pathways. These pathways are closely associated with osteoblast differentiation, osteoclast activity, cellular stress responses, and immune-inflammatory processes, suggesting that dysregulation of these signaling networks may contribute to osteoporosis pathogenesis. To identify additional osteoporosis-related targets, disease-associated genes were further retrieved from the GeneCards, OMIM, and Therapeutic Target Database (TTD) databases. As illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE, a total of 2,340 genes were obtained from GeneCards, 24 genes from OMIM, and 21 genes from TTD. Intersection analysis revealed a limited number of overlapping genes shared among these databases, representing high-confidence osteoporosis-associated targets. After removing duplicate entries, a comprehensive set of disease-related candidate genes was established for subsequent analysis.\u003c/p\u003e \u003cp\u003eMeanwhile, osteoporosis-related genes were independently identified from the GEO transcriptomic dataset, yielding 773 disease-associated genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF). By intersecting the GEO-derived genes with the curated disease-associated targets, overlapping genes were identified. After removing duplicate targets, 3233 targets associated with osteoporosis were obtained through using a Veen diagram.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e3.7 Screening core target genes of enoxaparin-associated osteoporosis through a PPI network\u003c/h2\u003e \u003cp\u003eWe first investigated the relationship between enoxaparin sodium and osteoporosis using a network toxicology approach. The chemical structure of enoxaparin sodium was first retrieved and used as the input compound for target prediction (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). After removing duplicates, 530 targets of enoxaparin sodium were obtained via screening ChEMBL, STITCH, and Swiss Target Prediction databases (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). As demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC, 134 core target genes of enoxaparin sodium-MPs causing osteoporosis were obtained through overlapping 3233 target genes of osteoporosis and 530 target genes of enoxaparin sodium. To further reveal the interactions between common targets, a medium-confidence PPI network (confidence score\u0026thinsp;\u0026ge;\u0026thinsp;0.400) was constructed via the STRING database by inputting the 134 shared targets between enoxaparin sodium and osteoporosis. The results showed a total of 211 edges and 134 nodes, illustrating the complex interactions among the potential targets (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). A topologically refined protein interaction network was generated to map inter-target relationships, with nodes symbolizing protein targets and edges denoting interaction events. Within the PPI network diagram, nodes exhibiting the highest degree are highlighted in dark red, with the color intensity diminishing as the degree reduces (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eE).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e3.8 Functional enrichment analysis of core target genes\u003c/h2\u003e \u003cp\u003eTo elucidate the biological significance of the core target genes associated with long-term enoxaparin exposure and osteoporosis risk, Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed. GO analysis categorized the enriched terms into biological process (BP), cellular component (CC), and molecular function (MF) domains. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA, the core target genes exhibited distinct functional enrichment patterns across the three GO categories. In the BP category, the core target genes were predominantly enriched in processes related to response to oxidative stress, cellular response to chemical stress, response to peptide hormone stimulus, and regulation of cell development and differentiation. In the CC category, enriched terms were mainly associated with nuclear and chromosomal components, including chromosomal regions and transcription factor complexes. In the MF category, significant enrichment was observed for DNA-binding transcription factor binding, transcription coregulator activity, and transmembrane transporter activity. To further illustrate the distribution of enriched GO terms, a circular GO visualization was generated (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). BP and MF categories showed higher enrichment intensity and gene ratios than CC terms, indicating functional convergence of the core genes on stress response, transcriptional regulation, and signaling integration. The GO dot plot demonstrated that several BP terms related to oxidative stress and inflammatory signaling showed high gene ratios and low adjusted p-values despite modest gene counts, suggesting that a limited number of core genes may exert substantial regulatory effects (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC). KEGG pathway analysis further revealed significant enrichment in pathways associated with bone remodeling and inflammation, including NF-κB, MAPK, PI3K\u0026ndash;Akt, JAK\u0026ndash;STAT signaling, apoptosis, and osteoclast differentiation (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD). The gene\u0026ndash;pathway network indicated that core genes were involved in multiple pathways simultaneously, highlighting their roles as regulatory nodes coordinating signaling crosstalk. Overall, these results suggest that long-term enoxaparin exposure may influence osteoporosis risk through integrated regulatory networks governing oxidative stress, inflammation, and bone metabolism.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e3.9 Weighted gene co-expression network analysis\u003c/h2\u003e \u003cp\u003eTo identify gene co-expression modules associated with osteoporosis under long-term enoxaparin exposure, weighted gene co-expression network analysis (WGCNA) was performed using normalized gene expression data (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e). Hierarchical clustering was conducted to assess sample quality, and one potential outlier was identified and removed prior to network construction to improve robustness (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). The remaining samples showed consistent clustering patterns and were retained for subsequent analysis. The sample dendrogram combined with the trait heatmap demonstrated a clear correspondence between expression-based clustering and clinical phenotypes, supporting the biological relevance of the dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). To construct a scale-free co-expression network, an appropriate soft-thresholding power was selected. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eC and Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eD, the scale-free topology fit index (R\u0026sup2;) increased and reached a plateau above 0.8 with increasing soft-threshold values, while mean connectivity gradually decreased. Based on these criteria, an optimal soft-threshold was chosen to balance scale-free topology and network connectivity. Using the selected soft-thresholding power, genes were hierarchically clustered into distinct co-expression modules by dynamic tree cutting. The gene dendrogram and corresponding module color assignments are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eE, revealing multiple modules with distinct expression patterns. Module\u0026ndash;trait correlation analysis indicated that two modules, ME maroon and ME sienna3, were significantly associated with osteoporosis, while showing weak or opposite correlations with control samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eF and Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eG), suggesting their involvement in osteoporosis-related transcriptional alterations. To further validate the relevance of OP-associated modules, relationships between module membership (MM) and gene significance (GS) for OP were examined. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eF and Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eG, genes within these modules exhibited strong positive correlations between MM and GS (cor\u0026thinsp;=\u0026thinsp;0.81 and cor\u0026thinsp;=\u0026thinsp;0.96, both with extremely low p-values), indicating high intramodular connectivity and strong associations with the OP phenotype. Collectively, these results demonstrate that the identified OP-associated modules are robust, biologically meaningful, and closely linked to disease status, providing a solid foundation for subsequent integration with machine learning and downstream functional analyses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e3.10 Machine learning\u0026ndash;based identification of key genes\u003c/h2\u003e \u003cp\u003eTo further identify key genes associated with osteoporosis risk under long-term enoxaparin exposure, a machine learning\u0026ndash;based feature selection strategy was applied. Least absolute shrinkage and selection operator (LASSO) regression was employed to reduce dimensionality and select the most informative predictors. Using ten-fold cross-validation, the optimal regularization parameter (λ) was determined based on the minimum binomial deviance criterion. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eA, the binomial deviance decreased with increasing penalization, and a parsimonious model with a limited number of non-zero coefficients was obtained at the optimal λ value, indicating effective feature reduction and avoidance of overfitting. The coefficient profiles of candidate genes under different penalty strengths are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eB. As the regularization parameter decreased, only a small subset of genes retained non-zero coefficients, while the majority of coefficients shrank toward zero. This pattern indicates that these retained genes contributed independently and stably to the prediction of osteoporosis status. To further characterize the contribution of each selected gene, the direction and magnitude of their regression coefficients were examined. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eC, several genes, including CDK16, SEC22B, ARFIP2, and VHL, exhibited positive coefficients, suggesting a potential association with increased osteoporosis risk. In contrast, AP1S1 and PTPN9 displayed negative coefficients, indicating a potential protective association. To enhance the robustness of gene selection, the machine learning\u0026ndash;derived genes were intersected with genes from osteoporosis-associated co-expression modules identified by WGCNA. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003eD, two genes were shared between the WGCNA modules and the LASSO-selected gene set. These overlapping genes were therefore defined as core genes for subsequent functional enrichment analysis and molecular docking validation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e3.11 Molecular docking validation\u003c/h2\u003e \u003cp\u003eThe two core genes identified through integrative WGCNA and machine learning analyses were subjected to molecular docking simulations to systematically evaluate their potential binding affinities with enoxaparin sodium. In molecular docking analysis, lower binding energy indicates a more stable ligand\u0026ndash;receptor complex and stronger interaction potential, whereas higher binding energy reflects weaker binding stability. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e, the docking results revealed favorable binding affinities between enoxaparin sodium and both target proteins. Specifically, CDK16 exhibited the strongest binding affinity (\u0026minus;\u0026thinsp;8.8 kcal/mol; Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eA), followed by VHL (\u0026minus;\u0026thinsp;6.2 kcal/mol; Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003eB). Notably, the binding energies of enoxaparin sodium with both core targets were lower than \u0026minus;\u0026thinsp;5 kcal/mol, suggesting a high level of binding stability. These results indicate a strong molecular affinity between enoxaparin sodium and the identified core proteins, supporting the potential involvement of these targets in mediating the biological effects of long-term enoxaparin exposure. The docking findings provide structural evidence that enoxaparin may directly interact with key regulatory proteins implicated in osteoporosis-related pathways, thereby contributing to altered bone metabolism and disease progression.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis large-scale retrospective cohort analysis based on the MIMIC-IV database systematically revealed a significant and time-dependent association between long-term enoxaparin exposure and increased osteoporosis risk, filling a gap in real-world evidence. Across multiple multivariable models, enoxaparin exposure days demonstrated a clear gradient increase in osteoporosis risk: compared to those exposed for \u0026lt;\u0026thinsp;15 days, patients exposed for 15\u0026ndash;30 days and 30\u0026ndash;90 days showed progressively higher risks, while those exposed for \u0026gt;\u0026thinsp;90 days exhibited a markedly elevated risk (OR\u0026thinsp;=\u0026thinsp;2.31, 95% CI: 1.31\u0026ndash;4.06; P\u0026thinsp;=\u0026thinsp;0.004). This association remained robust after stepwise adjustment for demographic characteristics, BMI, deep vein thrombosis, and dosing intensity, suggesting a potential time-dependent cumulative effect. Although enoxaparin is generally considered to have a lower skeletal toxicity risk than unfractionated heparin [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], our findings suggest that prolonged exposure may still exert clinically significant adverse effects on bone health, consistent with prior clinical observations [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Given enoxaparin's widespread use in hospitalized patients and those requiring long-term anticoagulation therapy [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], even a moderate increase in risk may translate to a substantial absolute disease burden at the population level, particularly among high-risk patients requiring prolonged anticoagulation [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Furthermore, after further adjusting for cumulative dose, dosing frequency, and daily average dose, the association between enoxaparin exposure duration and osteoporosis risk remained significant. This suggests that exposure duration itself may be a key determinant of osteoporosis risk, rather than solely dependent on dose factors. These findings extend the evidence of osteotoxicity in heparin drugs\u0026mdash;previously primarily based on experimental studies\u0026mdash;to the real-world clinical setting, providing quantitative epidemiological support for the skeletal safety of long-term enoxaparin therapy.\u003c/p\u003e \u003cp\u003eBeyond linear associations, this study further revealed a significant nonlinear relationship between enoxaparin exposure duration and osteoporosis risk through restricted cubic spline analysis (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Osteoporosis risk rose rapidly during the early exposure phase and continued to increase throughout long-term treatment, suggesting a cumulative toxicity profile with an unclear threshold[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. In recent years, a growing body of research indicates that drug-induced bone metabolism abnormalities, such as those caused by nucleoside analogues and zoledronic acid, often exhibit such nonlinear exposure-response patterns. These patterns feature gradually increasing risk with prolonged exposure duration, accompanied by time-dependent tolerance or cumulative effects. Early imbalances in bone remodeling may lay the groundwork for subsequent accelerated bone loss [\u003cspan additionalcitationids=\"CR36 CR37\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Although the magnitude of risk varies across different exposure phases, the overall spline curve exhibits a sustained upward trend, supporting the biological plausibility of cumulative effects of long-term enoxaparin exposure on bone metabolism. Given the observational design of this study and the absence of dynamic bone density measurements, the findings should be regarded as hypothesis-generating evidence. Future prospective studies with continuous bone density monitoring are needed to determine whether clinically relevant exposure duration thresholds exist.\u003c/p\u003e \u003cp\u003eSubgroup analysis revealed a significant association between long-term enoxaparin use and osteoporosis risk, particularly among women, younger patients, and Caucasians. Gender analysis showed that female patients had a significantly higher risk than males (OR\u0026thinsp;=\u0026thinsp;1.59, 95% CI: 1.26\u0026ndash;2.02, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), potentially related to gender-specific differences in bone metabolism and hormonal changes [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Age analysis revealed that patients under 65 years old (OR\u0026thinsp;=\u0026thinsp;2.68, 95% CI: 1.95\u0026ndash;3.69, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) exhibited greater sensitivity to enoxaparin exposure, whereas those aged 65 and above showed no significant increased risk. This suggests that enoxaparin-related bone loss may be more pronounced in populations with relatively active baseline bone metabolism. This finding aligns with evidence from animal models and clinical observations that heparin-induced inhibition of osteoblast activity is more readily apparent in younger individuals [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. In contrast, no significant association was observed in elderly patients, potentially due to age-related bone loss mechanisms partially masking the drug effect [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Previous systematic reviews have also noted that multiple comorbidities in the elderly population may dilute drug-specific skeletal risks [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Furthermore, racial analysis revealed a higher osteoporosis risk among white patients (OR\u0026thinsp;=\u0026thinsp;1.50, 95% CI: 1.20\u0026ndash;1.87, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), potentially attributable to interracial genetic differences or lifestyle factors [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. In BMI and deep vein thrombosis (DVT) subgroups, although both were associated with risk, they did not significantly alter the relationship between enoxaparin use and osteoporosis, suggesting that osteoporosis development may be influenced by the duration and dose of drug exposure [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. Notably, high-frequency administration (OR\u0026thinsp;=\u0026thinsp;1.66, 95% CI: 1.10\u0026ndash;2.49, P\u0026thinsp;=\u0026thinsp;0.015) and higher doses (OR\u0026thinsp;=\u0026thinsp;1.64, 95% CI: 1.28\u0026ndash;2.10, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) significantly increased osteoporosis risk, further underscoring the impact of drug dosage and treatment intensity on bone health. In summary, this study reveals a significant impact of long-term enoxaparin exposure on osteoporosis risk, particularly in specific populations, suggesting the need for individualized assessment and monitoring during prolonged therapy.\u003c/p\u003e \u003cp\u003eMediation analysis results indicate that enoxaparin dosing intensity (including average daily dosing frequency and daily dose) partially mediated the relationship between exposure duration and osteoporosis risk. However, the proportion of variance explained was relatively limited, and the direct effect of exposure duration remained predominant. The number of daily doses (ACME\u0026thinsp;=\u0026thinsp;0.000321, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) partially mediated the relationship between enoxaparin exposure and osteoporosis, accounting for 4.43% of the mediating effect (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Higher dosing frequency may amplify the drug's cumulative effects, thereby influencing bone metabolism. Furthermore, average daily dose (ACME\u0026thinsp;=\u0026thinsp;0.000422, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) also mediated the relationship between enoxaparin exposure and osteoporosis, accounting for 6.01% of the mediating effect (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), underscoring the critical role of drug dosage in osteoporosis development. These findings suggest that while higher dosing intensity may exacerbate osteoporosis risk, prolonged exposure itself is the core driver of abnormal bone metabolism[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. This aligns with evidence from animal models and clinical observations indicating that bone loss associated with heparin drugs is primarily time-dependent rather than purely dose-dependent. From a clinical practice perspective, this finding implies that reducing daily dosage or administration frequency alone may be insufficient to fully offset the bone health risks associated with long-term enoxaparin therapy. This aligns with multiple systematic reviews indicating that short-term (3\u0026ndash;6 months) exposure to low molecular weight heparins does not significantly increase fracture risk, whereas longer exposure periods (over 12 months) may lead to bone density decline [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. When evaluating anticoagulant benefits versus adverse effects, cumulative exposure duration should be considered a critical risk assessment metric, particularly in high-risk populations requiring long-term anticoagulation [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. Regular bone density monitoring is recommended for early intervention against potential bone loss. To mitigate this risk, drug usage strategies should be carefully adjusted in high-risk patients.\u003c/p\u003e \u003cp\u003eTo explore the potential molecular mechanisms underlying enoxaparin-associated osteoporosis, this study integrated network toxicology, transcriptomics analysis, weighted gene co-expression network analysis (WGCNA), and machine learning methods. Functional enrichment analysis consistently revealed that candidate target genes were predominantly enriched in pathways including oxidative stress, inflammatory signaling, apoptosis, and osteoclast differentiation\u0026mdash;all critical processes in bone remodeling and homeostasis. Cross-screening via WGCNA and LASSO regression identified CDK16 and VHL as potential core genes. Molecular docking analysis revealed stable binding potentials between enoxaparin and both CDK16 and VHL, with binding energies below \u0026minus;\u0026thinsp;5 kcal/mol, suggesting potential direct molecular interactions. This finding structurally supports enoxaparin's influence on key bone metabolism-related proteins. However, molecular docking is a computational prediction method whose results primarily reflect potential binding possibilities rather than actual biological effects. Future studies should validate the biological effects of these pathways using osteoblast and osteoclast models.\u003c/p\u003e \u003cp\u003eCDK16 belongs to the non-classical cyclin-dependent kinase family and primarily participates in biological processes outside the cell cycle, including cell differentiation, vesicular transport, and the integrated regulation of multiple signaling pathways [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. In the context of bone biology, a growing hypothesis suggests that CDK16 dysfunction may disrupt the differentiation of bone marrow mesenchymal stem cells toward the osteogenic lineage, leading to reduced osteoblast production and diminished bone matrix mineralization capacity [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. In this study, functional enrichment analysis of enoxaparin-related osteoporosis targets revealed significant enrichment of multiple signaling pathways closely associated with osteogenic differentiation, including PI3K\u0026ndash;Akt, MAPK, focal adhesion, actin cytoskeleton regulation, and cellular senescence pathways [\u003cspan additionalcitationids=\"CR48\" citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]. These signaling networks ultimately converge at the regulatory level of core osteogenic transcription factors such as RUNX2, SP7 (Osterix), and COL1A1 [\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. Long-term disruption of these pathways may cause fate deviation of bone marrow mesenchymal stem cells, suppressing the osteogenic process and thereby inducing insufficient bone formation. Enoxaparin, a highly sulfated, strongly negatively charged glycosaminoglycan-like molecule, may indirectly affect CDK16-related signaling networks through prolonged exposure by altering extracellular matrix interactions, cell-matrix adhesion, and mechanical signaling transduction within the bone marrow microenvironment [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. Furthermore, the significant enrichment of oxidative stress and inflammation-related pathways in this study suggests that persistent inflammatory stimulation may further suppress osteogenic differentiation, with CDK16 potentially acting as a signaling integration node to amplify these adverse effects [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Collectively, these findings support the potential mechanism whereby enoxaparin inhibits osteoblast differentiation and reduces bone formation capacity by regulating CDK16-associated networks.\u003c/p\u003e \u003cp\u003eVHL is a key regulator in cellular hypoxic adaptation, primarily acting by controlling the stability of hypoxia-inducible factors (HIFs) [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. In bone tissue, the VHL\u0026ndash;HIF axis plays a central role in coordinating angiogenesis, energy metabolism, and bone remodeling [\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e]. Physiological hypoxia and moderate HIF activation are essential for coupling osteogenesis with angiogenesis, whereas disruption of this axis may lead to pathological bone loss [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. Functional enrichment analysis in this study revealed significant enrichment of multiple pathways, including HIF-1 signaling, PI3K\u0026ndash;Akt, NF-κB, apoptosis, and osteoclast differentiation. These signaling networks exhibit substantial overlap with the VHL\u0026ndash;HIF regulatory system. In the clinical population represented by MIMIC-IV, patients commonly exhibit heightened inflammatory responses, oxidative stress, and impaired tissue perfusion. Against this backdrop, dysregulation of the VHL\u0026ndash;HIF axis may amplify effects on bone metabolism [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Molecular docking analysis indicates stable binding potential between enoxaparin and VHL, suggesting potential direct or indirect modulation of VHL function. Although molecular docking results cannot directly prove its biological effects in vivo, they provide structural support for enoxaparin's interference with HIF homeostasis. This process may impair VEGF-mediated angiogenesis, alter osteoblast metabolic programs, and promote osteoclast activation through inflammatory and oxidative stress pathways, ultimately causing bone-angiogenesis coupling imbalance and exacerbating bone resorption.\u003c/p\u003e \u003cp\u003eBased on the above evidence, this study proposes a \u0026ldquo;dual pathway synergistic model\u0026rdquo; to explain the potential association between long-term enoxaparin exposure and increased osteoporosis risk. On one hand, enoxaparin may inhibit osteoblast differentiation and mineralization by regulating CDK16-associated cell cycle-independent differentiation gating and stress signaling integration. On the other hand, it may weaken endovascular support and promote enhanced osteoclast activity by disrupting VHL\u0026ndash;HIF homeostasis and its interactions with inflammatory and oxidative stress pathways. Both pathways ultimately converge onto common downstream signaling networks involving PI3K\u0026ndash;Akt, MAPK, NF-κB, apoptosis, and osteoclast differentiation, manifesting as cumulative, nonlinear exposure-response relationships at the population level. Additionally, this study has several limitations. First, the retrospective observational design restricts causal inference and cannot fully exclude the influence of unmeasured confounders such as vitamin D status, glucocorticoid use, and baseline bone mineral density. Second, osteoporosis diagnosis relied on ICD codes rather than direct bone density measurements, potentially introducing some risk of misclassification. Our findings primarily reflect clinically recognized osteoporosis rather than subclinical bone loss. Moreover, the MIMIC-IV database primarily comprises critically ill patients, necessitating further validation of the generalizability of these findings. Mechanistically, network toxicology and molecular docking analyses are inherently predictive methods, carrying inherent risks of circular reasoning. Therefore, independent experimental studies are urgently needed to validate these findings.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn this retrospective cohort study based on the MIMIC-IV database, prolonged enoxaparin exposure was found to be associated with an increased risk of osteoporosis, exhibiting a clear duration-dependent and nonlinear pattern. By integrating real-world epidemiologic evidence with network toxicology, transcriptomic analysis, machine learning, and molecular docking, CDK16- and VHL-centered signaling networks were identified as potential molecular pathways linking long-term enoxaparin use to dysregulated bone remodeling. Nevertheless, the observational nature of the study, reliance on diagnosis codes rather than direct bone mineral density measurements, and the possibility of residual confounding limit definitive causal inference. Moreover, while network toxicology and molecular docking provide biologically plausible mechanistic hypotheses, they do not establish functional causality. Accordingly, these findings should be interpreted as hypothesis-generating and underscore the need for prospective, multicenter longitudinal studies with precise exposure characterization, serial bone density assessments, and adjudicated outcomes, together with targeted experimental validation, to confirm causality, clarify underlying mechanisms, and guide individualized risk assessment and prevention strategies for patients requiring long-term anticoagulation.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to express their gratitude to the MIMIC database for providing data support.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLZJ completed the manuscript drafting. WCL were responsible for the study design and manuscript revision. LZJ, LXM, LF and ZZM performed data curation and statistical analysis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Clinical Evidence-Based Research Special Project for the Construction of High-Level Traditional Chinese Medicine Hospitals of Wangjing Hospital, China Academy of Chinese Medical Sciences (WJYY-XZKT-2023-14).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are openly available in the Medical Information Mart for Intensive Care (MIMIC)-IV database at 10.13026/kpb9-mt58.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical Approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was based on publicly available anonymized MIMIC-IV data. Ethical\u003c/p\u003e\n\u003cp\u003eapproval was not required.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors report no competing interests related to the conduct of this study, the authorship, or the publication of this work.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eCompston JE, McClung MR, Leslie WD, Osteoporosis. Lancet (London England). 2019;393(10169):364\u0026ndash;76.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGBD LBMD. The global, regional, and national burden attributable to low bone mineral density, 1990\u0026ndash;2020: an analysis of a modifiable risk factor from the Global Burden of Disease Study 2021. Lancet Rheumatol. 2025;7(12):e873\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiang H, et al. Global epidemiology and burden of osteoporosis among postmenopausal women: insights from the Global Burden of Disease Study 2021. npj aging. 2025;11(1):78.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKanis JA et al. European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA, 2019. 30(1): pp. 3\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoong CWS et al. Advances in the management of osteoporosis. BMJ (Clinical research ed.), 2025. 390: p. e081250.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTai T, et al. Refracture risk and all-cause mortality after vertebral fragility fractures: Anti-osteoporotic medications matter. J Formos Med Association = Taiwan yi zhi. 2023;122(Suppl 1):S65\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNasser MI, et al. Impact of hip fracture on survival, disability, pain, and health-related quality of life in Zimbabwe: a prospective cohort study. lancet Healthy Longev. 2025;6(9):100766.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang L, Chen L, Chen K. Hormone-Related and Drug-Induced Osteoporosis: A Cellular and Molecular Overview. Int J Mol Sci. 2023;24(6):5814.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmit AE, Meijer OC, Winter EM. The multi-faceted nature of age-associated osteoporosis. Bone Rep. 2024;20:101750.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong X, et al. Advances in the study and treatment of glucocorticoid osteoporosis: A review. Medicine. 2025;104(22):e42668.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStegmayr B, Zuo L, Zadora W. Lipid and Bone Effects of Heparin Use During Hemodialysis. Semin Nephrol. 2023;43(6):151480.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTran HA et al. 2025 Guidelines for direct oral anticoagulants: a practical guidance on the prescription, laboratory testing, peri-operative and bleeding management. Internal medicine journal, 2025. 55(7): pp. 1174\u0026ndash;1183.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStevens SM, et al. Antithrombotic Therapy for VTE Disease: Compendium and Review of CHEST Guidelines 2012\u0026ndash;2021. Chest. 2024;166(2):388\u0026ndash;404.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Y, Guo S, Xu J. Multifunctional applications and research advances of low-molecular-weight heparin. Front Pharmacol. 2025;16:1585762.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, et al. Impaired bone healing by enoxaparin via inhibiting the differentiation of bone marrow mesenchymal stem cells towards osteoblasts. J Bone Miner Metab. 2022;40(1):9\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang S et al. Effect of long-term use of unfractionated or low-molecular-weight heparin on bone mineral density in maintenance hemodialysis patients. Hemodialysis international. International Symposium on Home Hemodialysis, 2020. 24(3): pp. 374\u0026ndash;382.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYokoyama S, et al. Association between oral anticoagulants and osteoporosis: Real-world data mining using a multi-methodological approach. Int J Med Sci. 2020;17(4):471\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo J, et al. Oxidative stress and inflammation: roles in osteoporosis. Front Immunol. 2025;16:1611932.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIantomasi T, et al. Oxidative Stress and Inflammation in Osteoporosis: Molecular Mechanisms Involved and the Relationship with microRNAs. Int J Mol Sci. 2023;24(4):3772.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMori Y, Aizawa T. Impacts of low molecular weight heparin on bone healing and osseointegration. J Bone Miner Metab. 2022;40(2):357.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParums DV. Med Sci monitor: Int Med J experimental Clin Res. 2025;31:e951118. A Review of the Importance and Relevance of Real-World Data and Real-World Evidence.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar R, Romano JD, Ritchie MD. Network-based analyses of multiomics data in biomedicine. BioData Min. 2025;18(1):37.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShen X, et al. Using weighted gene co-expression network analysis to identify key genes related to preeclampsia. Front Immunol. 2025;16:1569591.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKanapari A, et al. Current applications and future challenges of machine learning and artificial intelligence in clinical trials: A scoping review. Digit health. 2025;11:20552076251393272.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang C, et al. Network toxicology combined with molecular docking technology to explore the molecular mechanism of amatoxin causing liver injury. Sci Rep. 2025;15(1):26068.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiaz A. A., Transparency in the Reporting of Artificial Intelligence \u0026ndash; The TITAN Guideline. Premier J Sci, 2025. 2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnson AEW, et al. MIMIC-IV, a freely accessible electronic health record dataset. Sci data. 2023;10(1):1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Y, Sun T, Jiang C. Biomacromolecules as carriers in drug delivery and tissue engineering. Acta Pharm Sinica B. 2018;8(1):34\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePiazza G, et al. Superficial Vein Thrombosis: A Review. JAMA. 2025;334(22):2020\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMonreal M, et al. Comparative safety of enoxaparin versus other low-molecular-weight heparins in cancer-associated venous thromboembolism: a real-world cohort study from RIETE. Res Pract Thromb haemostasis. 2025;9(8):103244.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Y, et al. Association between enoxaparin administration and ICU 28-day mortality in sepsis patients: A comparative study. Respir Med. 2026;251:108543.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGajic-Veljanoski O, et al. Effects of Long-Term Low-Molecular-Weight Heparin on Fractures and Bone Density in Non-Pregnant Adults: A Systematic Review With Meta-Analysis. J Gen Intern Med. 2016;31(8):947\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJakonyte A, et al. Risk of Osteonecrosis of the Jaw in Patients Treated with Zoledronic or Alendronic Acid: A Systematic Review. Med (Kaunas Lithuania). 2025;61(7):1159.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen L, et al. Bone loss in young adults with HIV following antiretroviral therapy containing tenofovir disoproxil fumarate regimen using machine learning. Front Pharmacol. 2025;16:1516013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYip TC, et al. Long-term use of tenofovir disoproxil fumarate increases fracture risk in elderly patients with chronic hepatitis B. J Hepatol. 2024;80(4):553\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang P, et al. A novel evidence of serial changes of bone mineral density in chronic hepatitis B patients treated with entecavir. Hep Intl. 2021;15(2):310\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoldado-Folgado J, et al. Changes in bone quality after switching from a TDF to a TAF based ART: A pilot randomized study. Front Endocrinol. 2023;14:1076739.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlack DM, et al. Bisphosphonates and the risk of atypical femur fractures. Bone. 2022;156:116297.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang H, et al. Safety and Efficacy of Low Molecular Weight Heparin for Thromboprophylaxis in the Elderly: A Network Meta-Analysis of Randomized Clinical Trials. Front Pharmacol. 2021;12:783104.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEnsrud KE, Crandall CJ. Osteoporos Annals Intern Med. 2024;177(1):ITC1\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDepartment of Error. Lancet (London England). 2022;400(10354):732.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAyers C, et al. Effectiveness and Safety of Treatments to Prevent Fractures in People With Low Bone Mass or Primary Osteoporosis: A Living Systematic Review and Network Meta-analysis for the American College of Physicians. Ann Intern Med. 2023;176(2):182\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNoel SE, Santos MP, Wright NC. Racial and Ethnic Disparities in Bone Health and Outcomes in the United States. J bone mineral research: official J Am Soc Bone Mineral Res. 2021;36(10):1881\u0026ndash;905.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChandran M et al. Impact of osteoporosis and osteoporosis medications on fracture healing: a narrative review. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA, 2024. 35(8): pp. 1337\u0026ndash;58.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarcutskie CA, et al. Association of Anti-Factor Xa-Guided Dosing of Enoxaparin With Venous Thromboembolism After Trauma. JAMA Surg. 2018;153(2):144\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDixon-Clarke SE, et al. Structure and inhibitor specificity of the PCTAIRE-family kinase CDK16. Biochem J. 2017;474(5):699\u0026ndash;713.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu H et al. A disproportionality analysis of low molecular weight heparin in the overall population and in pregnancy women using the FDA adverse event reporting system (FAERS) database. Frontiers in pharmacology, 2024. 15: p. 1442002.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi D, et al. 7,8-DHF inhibits BMSC oxidative stress via the TRKB/PI3K/AKT/NRF2 pathway to improve symptoms of postmenopausal osteoporosis. Volume 223. Free radical biology \u0026amp; medicine; 2024. pp. 413\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu C, et al. Research on the role and mechanism of the PI3K/Akt/mTOR signalling pathway in osteoporosis. Front Endocrinol. 2025;16:1541714.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArtigas N, et al. Mitogen-activated protein kinase (MAPK)-regulated interactions between Osterix and Runx2 are critical for the transcriptional osteogenic program. J Biol Chem. 2014;289(39):27105\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSignorelli SS, et al. Anticoagulants and Osteoporosis. Int J Mol Sci. 2019;20(21):5275.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShen X, Yao Q, Ma L. Mechanisms responsible for the ability of enoxaparin sodium to inhibit inflammatory responses in the immune microenvironment of bone repair: A transcriptomic sequencing study. PLoS ONE. 2025;20(9):e0332041.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMendoza SV, et al. Osteocytic oxygen sensing: Distinct impacts of VHL and HIF-2alpha on bone integrity. Bone. 2025;192:117339.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen K, et al. Osteocytic HIF-1α Pathway Manipulates Bone Micro-structure and Remodeling via Regulating Osteocyte Terminal Differentiation. Front cell Dev biology. 2022;9:721561.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJaneczko K, Agoro R. Oxygen Sensing in Osteocytes: From Physiology to Age-related Osteoporosis. Curr Osteoporos Rep. 2025;23(1):28.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen W, et al. HIF-1α Regulates Bone Homeostasis and Angiogenesis, Participating in the Occurrence of Bone Metabolic Diseases. Cells. 2022;11(22):3552.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Y, et al. The hypoxia-inducible factor alpha pathway couples angiogenesis to osteogenesis during skeletal development. J Clin Investig. 2007;117(6):1616\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKnowles HJ. Hypoxic regulation of osteoclast differentiation and bone resorption activity. Hypoxia (Auckland, N.Z.), 2015. 3: pp. 73\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi L, et al. Suppression of Hypoxia-Inducible Factor 1α by Low-Molecular-Weight Heparin Mitigates Ventilation-Induced Diaphragm Dysfunction in a Murine Endotoxemia Model. Int J Mol Sci. 2021;22(4):1702.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Enoxaparin, Osteoporosis, Network toxicology, Real-world evidence, MIMIC-IV database","lastPublishedDoi":"10.21203/rs.3.rs-8842742/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8842742/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eEnoxaparin sodium is a widely used low-molecular-weight heparin for thromboprophylaxis and anticoagulation. Although it is generally considered to have a lower skeletal toxicity profile than unfractionated heparin, emerging evidence suggests that prolonged exposure may adversely affect bone metabolism. However, robust real-world evidence and mechanistic insights linking long-term enoxaparin use to osteoporosis remain limited.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe conducted a large retrospective cohort study using the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, including 23,852 adult patients with documented enoxaparin exposure and complete follow-up. The association between enoxaparin exposure duration and osteoporosis risk was evaluated using multivariable logistic regression, subgroup analyses, restricted cubic spline modeling, and causal mediation analysis. To explore potential molecular mechanisms, we integrated network toxicology, transcriptomic analysis, weighted gene co-expression network analysis (WGCNA), machine learning\u0026ndash;based feature selection, and molecular docking.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eProlonged enoxaparin exposure was significantly associated with an increased risk of osteoporosis in a duration-dependent and nonlinear manner. Patients exposed for more than 90 days had a much higher risk of osteoporosis compared to those exposed for less than 15 days. This risk remained strong even after demographic factors, clinical covariates, and dosing intensity were taken into account. Restricted cubic spline analysis confirmed a significant nonlinear exposure\u0026ndash;response relationship. Mediation analyses indicated that dosing frequency and daily dose partially mediated this association, while exposure duration remained the predominant driver. Network toxicology and enrichment analyses implicated oxidative stress, inflammatory signaling, apoptosis, and osteoclast differentiation pathways. Integrative WGCNA and machine learning identified CDK16 and VHL as core regulatory genes. Molecular docking demonstrated stable binding affinities between enoxaparin and both targets, supporting their potential involvement in enoxaparin-associated bone dysregulation.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eLong-term enoxaparin use is associated with an increased risk of osteoporosis, exhibiting clear duration-dependent and nonlinear characteristics. Integrating real-world epidemiologic evidence with systems-level network toxicology highlights CDK16- and VHL-centered pathways as potential mechanistic mediators.\u003c/p\u003e","manuscriptTitle":"Long-Term Enoxaparin Use and Osteoporosis Risk: A Real-World Cohort Study with Integrative Computational and Network Toxicology Approaches","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-19 11:46:07","doi":"10.21203/rs.3.rs-8842742/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2026-02-16T22:42:48+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-16T13:51:34+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-16T03:41:51+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Translational Medicine","date":"2026-02-10T10:34:35+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-translational-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jtrm","sideBox":"Learn more about [Journal of Translational Medicine](http://translational-medicine.biomedcentral.com)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/jtrm/default.aspx","title":"Journal of Translational Medicine","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"914b73fe-a61b-4631-aeab-7da5e5783d7b","owner":[],"postedDate":"February 19th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-13T16:07:41+00:00","versionOfRecord":{"articleIdentity":"rs-8842742","link":"https://doi.org/10.1186/s12967-026-08093-2","journal":{"identity":"journal-of-translational-medicine","isVorOnly":false,"title":"Journal of Translational Medicine"},"publishedOn":"2026-04-07 15:58:55","publishedOnDateReadable":"April 7th, 2026"},"versionCreatedAt":"2026-02-19 11:46:07","video":"","vorDoi":"10.1186/s12967-026-08093-2","vorDoiUrl":"https://doi.org/10.1186/s12967-026-08093-2","workflowStages":[]},"version":"v1","identity":"rs-8842742","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8842742","identity":"rs-8842742","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}