Causal relationship between aging-related biomarkers and osteoporosis: A bidirectional two-sample Mendelian randomization study Authors: Rui Xu, Yin Sun, Gui-guo Xu, Hai-bo Li, Hui Zhong, Zhi Peng, Sheng Lu Published: 2026 ABSTRACT ObjectiveThe potential causal role of molecular biomarkers of biological aging in osteoporosis remains uncertain. This study aimed to evaluate whether genetically predicted levels of aging-related traits are causally associated with the risk of osteoporosis.MethodsIn this study, we selected eight exposures: telomere length, mitochondrial DNA copy number, insulin-like growth factor 1 levels, DNA methylation PhenoAge acceleration, leucine, isoleucine, valine, and Dickkopf-related protein 1. A bidirectional two-sample Mendelian randomization analysis was conducted among the European population. The inverse-variance weighted method served as the primary analysis, complemented by Mendelian randomization–Egger, weighted median, mode-based methods, Cochran's Q test, the Mendelian randomization–Egger intercept, and leave-one-out analysis to assess robustness, heterogeneity, and pleiotropy.ResultsThe inverse-variance weighted analysis showed nominal associations (p < 0.05) between genetically predicted higher insulin-like growth factor 1 (odds ratio = 0.910, 95% confidence interval: 0.840 to 0.985, p = 0.019) and isoleucine (odds ratio = 0.780, 95% confidence interval: 0.618 to 0.984, p = 0.036) with a lower risk of osteoporosis and elevated Dickkopf-related protein 1 (odds ratio = 1.195, 95% confidence interval: 1.043 to 1.368, p = 0.010) with a higher risk of osteoporosis; however, none remained significant after Bonferroni correction (p > 0.00625). Sensitivity analyses suggested a statistically significant association for Dickkopf-related protein 1 (weighted median and mode, p < 0.00625), whereas no significant causal effects were observed for any of the other biomarkers. Reverse Mendelian randomization revealed no evidence that osteoporosis influenced any of the biomarkers.ConclusionsIn this study, no robust causal effects were identified between the eight aging biomarkers and osteoporosis in the European population. However, the nominal associations observed for insulin-like growth factor 1, isoleucine, and Dickkopf-related protein 1 warrant further investigation into the aging-skeleton axis. Introduction Osteoporosis is a systemic skeletal condition characterized by low bone mass and deteriorated bone structure. It is one of the most prevalent age-associated conditions worldwide, with a prevalence of 18.3%. 1 It significantly increases the risk of fragility fractures. Among individuals more than 50 years old, approximately half of women and one-fifth of men will experience at least one osteoporotic fracture. 2 Therefore, osteoporosis is a major public health issue worldwide, contributing substantially to disability, mortality, and an increased healthcare burden, especially among older adults. With the accelerating pace of population aging, this burden is expected to continue rising. The pathogenesis of osteoporosis is complex and is influenced by multiple genetic and environmental factors, including ethnicity, genetic background, age, hormonal status, immune factors, medication use, inflammation, nutrition, and physical activity.3,4 Although these factors are closely associated with the development of osteoporosis, the underlying mechanisms by which biological aging, the central driver of this condition, contributes to its pathogenesis have not yet been fully elucidated. Aging is a progressive biological process characterized by degenerative changes at the cellular, tissue, and organ levels over time, leading to functional decline and an increased susceptibility to various diseases, including osteoporosis, Alzheimer's disease, neurodegeneration, stroke, cancer, cardiovascular disease, and diabetes as well as increased mortality. 5 In recent years, the concept of biological aging has gained increasing attention and is considered a better predictor of health outcomes than chronological age because it reflects functional decline at the molecular and cellular levels. To date, a variety of multiomics aging biomarkers have been identified and actively investigated, including those derived from genomics, transcriptomics, proteomics, metabolomics, and microbiomics, with growing interest in their potential associations with osteoporosis. 6 For example, observational studies have reported conflicting findings regarding the association between telomere length and osteoporosis. Some studies support an association between shorter leukocyte telomere length and lower bone mineral density or a higher fracture risk,7,8 whereas another cohort study found no significant relationship. 9 This inconsistency underscores the lack of consensus regarding the causal or correlative role of telomere shortening in the pathogenesis of osteoporosis.10,11 More recently, multiple cross-sectional and cohort studies have reported associations between serum insulin-like growth factor 1 (IGF-1) levels and bone health in specific clinical populations. Lower IGF-1 levels have been associated with an increased risk of osteoporosis in patients with Parkinson's disease, 12 have been negatively associated with osteodystrophy in individuals with cirrhosis, 13 and have been positively correlated with bone formation markers in physically active adults. 14 In addition, among patients with type 2 diabetes, higher IGF-1 levels have been associated with greater bone mineral density (BMD).15,16 Despite these associations, observational studies cannot definitively establish causality because of potential confounding, reverse causation, and measurement bias. To address these limitations, Mendelian randomization (MR) provides a powerful epidemiological approach that uses genetic variants as instrumental variables (IVs) to infer causal relationships between exposures and outcomes. Because germline genetic variants are randomly assigned at conception and remain fixed throughout life, MR minimizes confounding and reverse causation and avoids the ethical and logistical challenges of long-term randomized trials in aging research. To address these uncertainties, this study leveraged large-scale genome-wide association study (GWAS) summary statistics and employed a bidirectional two-sample MR design. Although several MR studies have investigated the relationship between aging and osteoporosis, these studies have primarily focused on individual or specific aging metrics (e.g. epigenetic aging and mitochondrial DNA copy number (mtDNA-CN)).17,18 In contrast, this study aims to broaden the scope by systematically evaluating a more comprehensive array of aging-related biomarkers. We initially screened more than 60 aging-related biomarkers highlighted by Zhang et al., 19 and ultimately selected 8 representative biomarkers for which high-quality GWAS data and valid IVs were available. These included telomere length, mtDNA-CN, IGF-1 levels, DNA methylation PhenoAge acceleration, branched-chain amino acids (leucine, isoleucine, and valine), and Dickkopf-related protein 1 (DKK1). The study aimed to address two key questions: (a) whether genetically predicted levels of these aging biomarkers causally influence the risk of osteoporosis and (b) whether genetic predisposition to osteoporosis exerts a reverse causal effect on these biomarkers. Methods Study design In this study, we screened a panel of multiomics-derived aging biomarkers reported by Zhang et al.19 as potential exposures. We then systematically searched publicly available GWAS summary statistics to identify datasets with sufficient single nucleotide polymorphisms (SNP) sizes and high-quality IVs. Specifically, exposure factors were preliminarily selected based on the availability of large-scale GWAS summary statistics from individuals of European ancestry with sufficient SNP count (>one million) and more than five high-quality IVs (SNPs reaching genome-wide significance (p < 5 × 10−8), independent (r2 < 0.001 within a 10,000 kb window), and with an F-statistic >10). The final set of exposures included eight continuous traits: telomere length, mtDNA-CN, IGF-1 levels, DNA methylation PhenoAge acceleration, and three branched-chain amino acids (leucine, isoleucine, and valine) as well as DKK1. In the study, osteoporosis was designated as the outcome of interest. Following the identification of both exposures and outcome, we implemented a bidirectional two-sample MR to explore the relationships between these aging-related biomarkers and the risk of osteoporosis. The study adhered strictly to the three key assumptions: (a) relevance, whereby the selected genetic variants are robustly associated with the exposure; (b) independence, whereby the IVs are not associated with confounders of the exposure–outcome relationship; and (c) exclusion restriction, whereby the IVs influence the outcome exclusively through the exposure, without horizontal pleiotropic pathways. 20 The reporting of this study followed the Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR) guidelines. 21 This study used publicly available GWAS summary statistics; therefore, no additional ethical approval or written informed consent was required. A schematic overview of the analytical workflow is provided in Figure 1. Design of the study. Data resources The summary-level GWAS data for telomere length were obtained from the study conducted by Codd et al.,22 which included 472,174 UK Biobank individuals of European ancestry and encompassed 20,134,421 SNPs. The mtDNA-CN data were derived from the large-scale GWAS by Chong et al.,23 comprising 383,476 participants of European ancestry (British = 359,689; Irish = 10,598; other White = 13,189) and 11,173,383 SNPs. The IGF-1 levels data were sourced from Barton et al.,24 involving 435,516 UK Biobank individuals of European descent and 4,231,359 SNPs. Epigenetic aging, quantified as DNA methylation PhenoAge acceleration, was assessed using GWAS results from McCartney et al.,25 which included 34,463 European individuals from 28 cohorts (approximately 57% women) and 7,545,555 SNPs. Genetic instruments for plasma branched-chain amino acids, including leucine, isoleucine, and valine, were drawn from three separate GWASs reported by Richardson et al.,26 each based on approximately 115,000 UK Biobank participants of European ancestry and 11,590,399 SNPs. Additionally, summary statistics for DKK1 were obtained from Folkersen et al.,27 which analyzed 21,758 individuals of European ancestry from 13 cohorts and included 13,102,756 SNPs. The osteoporosis data were obtained from the FinnGen consortium (Release 12; https://r12.finngen.fi/). 28 This dataset included 10,461 clinically diagnosed cases and 473,264 controls of European ancestry, with genotype information covering 21,326,870 SNPs. Osteoporosis cases were identified based on the International Classification of Diseases (ICD) diagnostic codes: ICD-8 codes 7230 and 72391; ICD-9 codes within the range 7330–7331; and ICD-10 codes M80, M81, and M82 (Table 1).22–27 The GWAS datasets for the exposures and outcome were derived from independent populations with no sample overlap. Table 1.Overview of data source.GWAS datasetAccession numberPhenotypeEthnicitySample sizeNumber of SNPsDefinition of casesCodd et al.22 (PMID: 37117760)ieu-b-4879Telomere length (continuous)European472,17420,134,421/Chong et al.23 (PMID: 35023831)ebi-a-GCST90026372Mitochondrial DNA copy number (continuous)European383,47611,173,383/Barton et al.24 (PMID: 34226706)ebi-a-GCST90025989Insulin-like growth factor 1 levels (continuous)European435,5164,231,359/McCartney et al.25 (PMID: 34187551)ebi-a-GCST90014292DNA methylation PhenoAge acceleration (continuous)European34,4637,545,555/Richardson et al.26 (PMID: 35213538)ebi-a-GCST90092891Leucine (continuous)European115,07811,590,399/Richardson et al.26 (PMID: 35213538)ebi-a-GCST90092843Isoleucine (continuous)European115,07911,590,399/Richardson et al.26 (PMID: 35213538)ebi-a-GCST90092995Valine (continuous)European115,05211,590,399/Folkersen et al.27 (PMID: 33067605)ebi-a-GCST90012060Dickkopf-related protein 1 (continuous)European21,75813,102,756/FinnGenfinngen_R12_M13_OSTEOPOROSISOsteoporosis (yes/no)European10,461 cases and 473,264 controls21,326,870ICD-8: 7230|72391; ICD-9: 733(0–1); ICD-10: M80, M81, M82GWAS: genome-wide association study; ICD: International Classification of Diseases; SNPs: single nucleotide polymorphisms. Overview of data source. GWAS: genome-wide association study; ICD: International Classification of Diseases; SNPs: single nucleotide polymorphisms. MR analysis Selection of genetic variants To ensure the selection of strong and valid IVs, we first identified SNPs that were robustly associated with each exposure at the conventional genome-wide significance threshold (p < 5 × 10−8). 29 To minimize linkage disequilibrium (LD) among the selected variants, we performed LD-based clumping using a stringent threshold of r2 <0.001 within a 10,000-kb genomic window. 30 Only SNPs yielding an F-statistic exceeding 10 were retained for downstream analyses to mitigate potential bias arising from weak instrument strength. 31 Statistical analysis and sensitivity analysis The primary analysis was based on the inverse-variance weighted (IVW) method. 32 To evaluate the robustness of our findings, we complemented the main analysis with several alternative MR estimators, including MR-Egger, 33 weighted median, 34 simple mode, and weighted mode approaches. 35 We also conducted a series of sensitivity analyses to assess the validity of our results. Heterogeneity among SNP-specific causal estimates was quantified using Cochran's Q statistic, with a significant p-value (<0.05) indicated the presence of substantial heterogeneity. 36 Directional pleiotropy was evaluated via the MR-Egger intercept test, where a nonzero intercept suggests systematic pleiotropic effects. 33 In addition, a leave-one-out analysis was performed iteratively to determine whether any single variant disproportionately influenced the overall causal estimate. 37 And p <0.05 was considered to indicate a nominal association. To account for multiple comparisons, Bonferroni correction was applied, with statistical significance defined as p <0.00625 (i.e. 0.05 divided by the number of 8 tested exposures). 38 The statistical power of the MR estimates was evaluated using the mRnd web tool (https://shiny.cnsgenomics.com/mRnd/). All MR analyses were implemented in R version 4.4.3 using the ‘TwoSampleMR’ package. 39 Results Causal effects of aging-related biomarkers on osteoporosis After strict selection, we finally retained 134, 53, 384, 11, 16, 10, 18, and 9 independent IVs for telomere length, mtDNA-CN, IGF-1 levels, DNA methylation PhenoAge acceleration, leucine, isoleucine, valine, and DKK1, respectively. All IVs had F-statistics >10, indicating strong instrument strength. Details are provided in Supplementary Table S1 and S2. Using the IVW method as the primary analytical approach, we observed evidence of nominal associations (p < 0.05) for four biomarkers. Genetically predicted higher IGF-1 levels (odds ratio (OR) = 0.910, 95% confidence interval (CI): 0.840 to 0.985, p = 0.019) and greater isoleucine levels (OR = 0.780, 95% CI: 0.618 to 0.984, p = 0.036) were nominally associated with a reduced risk of osteoporosis, whereas elevated DNA methylation PhenoAge acceleration showed a negative and nominal association (OR = 0.964, 95% CI: 0.930 to 1.000, p = 0.048). In contrast, higher genetically predicted DKK1 levels were nominally associated with an increased risk of osteoporosis (OR = 1.195, 95% CI: 1.043 to 1.368, p = 0.010). However, after Bonferroni correction for multiple testing, none of the examined exposures remained significantly associated with osteoporosis (p > 0.00625). Specifically, genetically predicted telomere length (OR = 0.908, 95% CI: 0.806 to 1.022, p = 0.111), leucine (OR = 0.910, 95% CI: 0.700 to 1.184, p = 0.481), valine (OR = 0.921, 95% CI: 0.791 to 1.073, p = 0.292), and mtDNA-CN (OR = 1.083, 95% CI: 0.860 to 1.365, p = 0.498) were not significantly associated with osteoporosis. The weighted median and weighted mode analyses revealed statistically significant associations between DKK1 levels and osteoporosis (p < 0.00625). In contrast, the MR-Egger, weighted median, simple mode, and weighted mode methods consistently showed no statistically significant associations between the remaining exposures and osteoporosis. Details are presented in Figure 2 and Supplementary Table S3. Forest plot showing the causal effects of aging-related biomarkers on osteoporosis. MR-Egger regression analyses showed no pleiotropy for all exposures. The results of Cochran's Q tests indicated significant heterogeneity in the association between leucine and osteoporosis (Q = 26.44, Cochran's Q test p = 0.034). Although Cochran's Q tests indicated no heterogeneity in the association between the rest exposures and osteoporosis (Cochran's Q test p > 0.05). Leave-one-out analyses confirmed that no single SNP disproportionately influenced the overall causal estimates, supporting result robustness. Notably, the statistical power for the analyses of IGF-1 and DKK1 was ≥80%. Details are provided in Supplementary Table S3 and Supplementary Figures S1 to S8. Causal effects of osteoporosis on aging-related biomarkers Using 5–12 genome-wide significant SNPs associated with osteoporosis (after harmonization and clumping), we derived valid IVs (F-statistics >10) for each outcome trait. Details are provided in Supplementary Table S4 and S5. The IVW method found no significant causal effect of osteoporosis on any outcomes: telomere length (β = −0.006, 95% CI: −0.023 to 0.010, p = 0.438), mtDNA-CN (β = 0.014, 95% CI: −0.008 to 0.037, p = 0.218), IGF-1 levels (β = −0.015, 95% CI: −0.0068 to 0.038, p = 0.570), DNA methylation PhenoAge acceleration (β = −0.066, 95% CI: −0.522 to 0.390, p = 0.777), leucine (β = −0.013, 95% CI: −0.043 to 0.016, p = 0.375), isoleucine (β = −0.016, 95% CI: −0.046 to 0.014, p = 0.305), valine (β = −0.028, 95% CI: −0.058 to 0.001, p = 0.062), or DKK1 levels (β = −0.028, 95% CI: −0.110 to 0.054, p = 0.506). The MR-Egger, weighted median, simple mode, and weighted mode methods consistently showed no statistically significant associations between osteoporosis and any of the eight outcomes (p > 0.05). Details are presented in Figure 3 and Supplementary Table S6. Forest plot showing the causal effects of osteoporosis on aging-related biomarkers. No evidence of heterogeneity or horizontal pleiotropy was observed. Leave-one-out analyses further confirmed that the causal estimates were not driven by any individual SNP. Details are provided in Supplementary Table S6 and Supplementary Figure S9 to S16. Discussion This study explored the causal relationships between a suite of molecular biomarkers of biological aging, including telomere length, mtDNA-CN, IGF-1 levels, DNA methylation PhenoAge acceleration, leucine, isoleucine, valine, and DKK1, and the risk of osteoporosis. Although the IVW analyses did not identify any statistically significant associations between the eight aging-related biomarkers and osteoporosis after Bonferroni correction for multiple testing, several nominal associations warrant exploratory consideration. Specifically, genetically predicted higher IGF-1 and isoleucine levels were nominally associated with a reduced risk of osteoporosis, whereas elevated DKK1 levels were nominally associated with an increased risk. Conversely, we found no evidence of a reverse causal effect of osteoporosis on any of these biomarkers. IGF-1 is a key polypeptide hormone that is essential for bone development, remodeling, and the lifelong maintenance of bone mass, acting through both endocrine and autocrine/paracrine mechanisms. It promotes bone formation primarily by stimulating osteoblast activity and also modulates osteoclast function, thereby maintaining skeletal homeostasis throughout life.40,41 This study revealed a nominal association between genetically higher IGF-1 levels and reduced osteoporosis risk aligns with previous observational evidence.42,43 Notably, a previous MR study in European populations found that genetically proxied higher serum IGF-1 levels were associated with a decreased fracture risk and possibly greater estimated BMD (eBMD). 44 Although the MR estimate did not survive multiple-testing correction, the concordant direction of effect across biological, epidemiological, and genetic evidence strengthens the plausibility of IGF-1 as a modulator, rather than a sole causal driver, of osteoporosis pathogenesis. Branched-chain amino acids (leucine, isoleucine, and valine) have emerged as potential regulators of bone metabolism by promoting osteoblast activity through the mechanistic target of rapamycin (mTOR) and AMP-activated protein kinase (AMPK) signaling pathways while potentially suppressing osteoclast-mediated bone resorption via antioxidant and anti-inflammatory effects. 45 Our finding of a negative and nominal association between isoleucine and osteoporosis aligns with some observational studies. For example, Carbone et al. 46 and Urano et al. 47 reported positive and significant association between branched-chain amino acid levels and BMD (level 3 and level 2) in observational studies. The lack of significant causal effects for leucine and valine on osteoporosis may reflect limited power or context-dependent roles, indicating that further investigation would be valuable. Although initial advances have been made, research into the role of branched-chain amino acids in bone metabolism and osteoporosis is still nascent, with substantial knowledge gaps and methodological challenges that currently constrain its clinical translation. 48 DKK1, a potent Wnt/β-catenin pathway inhibitor, suppresses osteoblast activity and bone formation. 49 In this study, genetically predicted higher DKK1 levels showed a nominal association with an increased risk of osteoporosis. However, findings from observational studies on the relationship between circulating DKK1 levels and BMD in postmenopausal women have yielded inconsistent findings. Some studies have reported that elevated serum DKK1 levels are associated with more severe osteoporosis, 50 whereas others have paradoxically observed a positive correlation between DKK1 levels and BMD in women with postmenopausal osteoporosis. 51 Given these conflicting results and the limited evidence currently available, the utility of DKK1 as a reliable biomarker for osteoporosis or fracture risk prediction warrants further investigation in well-designed prospective and genetic studies. DNA methylation PhenoAge acceleration is a measure of biological aging derived from an epigenetic clock that integrates methylation levels at 513 cytosine-phosphate-guaninedinucleotide (CpG) sites with 9 clinical biomarkers linked to mortality, yielding a predicted age that often diverges from chronological age.52,53 Greater PhenoAge acceleration, indicating an older biological than chronological age, has been associated with an increased risk of age-related morbidity and all-cause mortality. In this study, genetically predicted DNA methylation PhenoAge acceleration showed a negative and nominal association with osteoporosis risk. This finding contrasts with some observational studies reporting positive associations. For example, cross-sectional analyses of National Health and Nutrition Examination Survey (NHANES) and UK Biobank data have linked higher PhenoAge acceleration to lower BMD and a higher prevalence of osteoporosis.54,55 However, observational associations are often susceptible to residual confounding, particularly given the strong link between PhenoAge acceleration and behavioral or lifestyle factors. Supporting the complexity of this relationship, a twin cohort study found that although PhenoAge acceleration was associated with a higher risk of fracture or osteoporosis in individual-level analyses, the association was no longer significant in within-pair comparisons. 56 Similarly, another observational study suggested that primary osteoporosis is not reflected by disease-specific DNA methylation or accelerated epigenetic age in blood. 57 Furthermore, our findings are consistent with previous MR evidence showing no clear causal relationship between PhenoAge acceleration and BMD at the femoral neck, lumbar spine, or forearm. 17 The discrepancies among these studies may partly attributed to tissue specificity, as epigenetic clocks were mainly based on blood samples, which means that the clocks reflect the overall biology of an individual, and not the biology specific to bone tissue. Additionally, the discrepancy between our MR results and previous MR evidence may stem from differences in the outcomes assessed as well as limitations such as the relatively small number of available IVs for PhenoAge acceleration, which may have reduced statistical power and biased the estimates toward the null. Therefore, these findings should be interpreted with caution and validated in future studies using stronger genetic instruments and larger sample sizes. Telomeres are repetitive TTAGGG DNA sequences located at chromosome ends that are stabilized by the shelterin protein complex, play a critical role in maintaining genomic integrity and progressively shorten with aging.11,58 In this study, we found no evidence of a causal association between genetically predicted telomere length and osteoporosis risk. However, observational studies have reported conflicting associations, some showing no association between telomere length and osteoporosis risk, 9 whereas others suggesting a negative association.7,8 Similarly, mtDNA-CN, a readily measurable blood-based indicator of mitochondrial function that correlates with oxidative stress, cellular energy status, and mitochondrial membrane potential, showed no significant causal link to osteoporosis in our study, aligning with prior observational data.59,60 Collectively, these results cast doubt on the utility of telomere length or mtDNA-CN as standalone causal biomarkers of osteoporosis, although their potential contribution within a multimodal biomarker panel warrants further investigation.11,60 The present findings suggest that none of the individual eight aging biomarkers examined are robust, standalone causal drivers of osteoporosis. Although these indicate a lack of direct causality, these “negative findings” hold clinical value by helping to preclude potentially futile therapeutic targets, thereby preventing the misallocation of medical resources. Furthermore, the results clarify that previously observed associations in epidemiological studies are likely driven by confounding factors rather than a direct biological effect, cautioning clinicians against interpreting fluctuations in these markers as direct indicators of bone deterioration. Consequently, our study highlights the potential value of considering approaches beyond single markers in clinical risk assessment. Instead, multimodal panel integrating genetic, epigenetic, metabolic, and proteomic markers—including, but not limited to, DKK1, IGF-1, and branched-chain amino acids—may provide complementary predictive value for identifying individuals at high risk of osteoporosis. Future research could explore the potential value of such combinatorial biomarkers alongside lifestyle modifications to better understand their impact on skeletal outcomes. There are several strengths to this study. We adhered strictly to the STROBE-MR reporting guidelines, used large-scale publicly available GWAS datasets with stringent quality control, and selected only strong IVs to minimize weak instrument bias. The bidirectional MR design helped rule out reverse causation, whereas multiple sensitivity methods collectively confirmed result robustness. Notably, the absence of significant directional pleiotropy across all exposures further strengthens the validity of our conclusions. Nevertheless, important limitations should be acknowledged. First, our analyses were restricted to participants of European ancestry; therefore, the findings may not be generalizable to other populations with different genetic architectures, lifestyle factors, or aging trajectories. Future MR studies including diverse ancestral populations are needed to evaluate the consistency of these findings across populations. Second, residual horizontal pleiotropy cannot be entirely excluded, particularly if genetic variants influence osteoporosis through pathways unrelated to the biomarker of interest. In addition, many aging-related biomarkers were not included because of the limited availability of GWAS data and suitable IVs during the initial screening phase. As GWAS resources expand and more robust genetic instruments become available, future studies will be better positioned to explore these additional candidates. Furthermore, as this study is an in silico bioinformatics analysis, it primarily provides hypothesis-generating evidence that warrants further validation through in vivo experiments and clinical studies. Conclusion In conclusion, this MR study found no robust causal evidence linking the eight aging-related biomarkers to osteoporosis. Nevertheless, the nominal associations observed for IGF-1, isoleucine, and DKK1 provide preliminary evidence that may inform future investigations into the mechanisms underlying age-related bone loss and its early diagnosis. These findings help steer efforts away from spurious associations and toward plausible biological pathways, underscoring that more extensive investigations are warranted to further explore the aging-skeleton axis. Supplemental Material sj-docx-1-imr-10.1177_03000605261469014 - Supplemental material for Causal relationship between aging-related biomarkers and osteoporosis: A bidirectional two-sample Mendelian randomization study Supplemental material, sj-docx-1-imr-10.1177_03000605261469014 for Causal relationship between aging-related biomarkers and osteoporosis: A bidirectional two-sample Mendelian randomization study by Rui Xu, Yin Sun, Gui-guo Xu, Hai-bo Li, Hui Zhong, Zhi Peng and Sheng Lu in Journal of International Medical Research sj-xlsx-2-imr-10.1177_03000605261469014 - Supplemental material for Causal relationship between aging-related biomarkers and osteoporosis: A bidirectional two-sample Mendelian randomization study Supplemental material, sj-xlsx-2-imr-10.1177_03000605261469014 for Causal relationship between aging-related biomarkers and osteoporosis: A bidirectional two-sample Mendelian randomization study by Rui Xu, Yin Sun, Gui-guo Xu, Hai-bo Li, Hui Zhong, Zhi Peng and Sheng Lu in Journal of International Medical Research