Blog

Sep 14, 2023

Genetically predicted circulating serum homocysteine levels on osteoporosis: a two

Scientific Reports volume 13,

Scientific Reports volume 13, Article number: 9063 (2023) Cite this article

1 Altmetric

Metrics details

To investigate the causal relationship between circulating serum homocysteine (Hcy) levels and osteoporosis (OP). Using public datasets gathered from independently published genome-wide association studies (GWAS), Mendelian randomization (MR) analysis was done to investigate the causal influence of Hcy on OP. SNPs were selected from a meta-analysis of GWAS on Hcy concentrations in 44,147 individuals of European ancestry. Meanwhile, SNPs of individuals of European descent for OP were extracted from the Genetic Factors of Osteoporosis Consortium (GEFOS) UK Biobank. The odds ratio (OR) of inverse variance weighted (IVW) approaches was established as the primary outcome. Moreover, weighted median (WM) and MR-Egger regressions were included in the sensitivity analysis. There were no causal effects of Hcy on forearm bone mineral density and lumbar bone mineral density according to IVW, MR-Egger, and WM analyses (all p > 0.05). In the IVW, we discovered the causality between genetically predicted Hcy and heel bone mineral density (H-BMD) with an OR of 0.96 [95% confidence interval (CI) = 0.927–0.990, p = 0.011]. In the additional sensitivity analysis, WM regression (OR = 0.97, 95% CI = 0.995–1.076, p = 0.084) and MR-Egger regression (OR = 0.98, 95% CI = 0.918–1.049, p = 0.609) yielded values that were comparable in direction but less precise. The MR-Egger intercept, funnel plot, and IVW all indicate the absence of any discernible directional pleiotropy. The leave-one-out analysis revealed that a single SNP did not influence the results of the MR analysis. In conclusion, our MR investigation revealed evidence of a causal relationship between circulating serum Hcy levels and H-BMD, but not OP in the European population. However, larger sample sizes are needed in the future to get more reliable conclusions.

Osteoporosis (OP) is a worldwide health issue defined by bone microstructure degeneration and low bone mineral density (BMD), which may lower bone strength and increase bone brittleness1.

Recent studies have demonstrated a positive correlation between age and OP prevalence, with an incidence rate of 3.2% in individuals aged 40–49 and 32.5% in those aged 65 and above, highlighting the considerable risk OP poses to the physical and mental well-being of middle-aged and elderly individuals, substantially impacting their daily functioning2. Therefore, early detection and intervention, as well as effective control of the progression of osteoporosis, are essential.

Methionine metabolism results in the formation of a sulfur-containing amino acid, homocysteine (Hcy). It is affected by genetics, age, vitamin B12, folic acid, vitamin B6, and other factors and is mainly used in clinical risk prediction of cardiovascular and cerebrovascular diseases. Recent research has linked elevated Hcy levels to decreased bone density3. The relative risk of fracture increases with increased serum Hcy concentration4,5. According to some studies, high Hcy may affect bone matrix and diminish bone quality by inhibiting cross-linking between collagen fibers and hindering the formation of a collagen network structure6,7. Mild to moderately elevated Hcy levels have been shown to activate osteoclast formation and function by generating intracellular reactive oxygen species (ROS) in the bone marrow, resulting in increased bone resorption. Additionally, elevated Hcy levels have been linked to decreased osteocalcin expression and increased osteopontin expression, which disrupts the normal function of osteoblasts, resulting in decreased bone formation and ultimately contributing to osteoporosis8,9. Numerous studies have postulated that higher serum Hcy levels may be a risk factor for osteoporotic fractures in the elderly. However, the results from these studies have yielded conflicting findings regarding the relationship between Hcy, BMD, and fractures. After controlling for age and gender, a Dutch study of middle-aged and elderly persons over 55 revealed that Hcy was not associated with bone mineral density. After correcting for age, gender, body mass, and other variables, fracture's relative risk (RR) rose by 1.4 when the natural logarithm of Hcy increased by 1 standard deviation, with no significant gender difference4. Similar results were found in a 7-year Swedish study of 996 women (75 + years)5. High Hcy levels were not significantly associated with fractures, and the highest quartile of Hcy levels had lower femoral neck BMD and greater trochanteric BMD than the other groups and lower lumbar spine BMD (L2–L4)10. After adjusting for parathyroid hormone, smoking, and recent fractures, high Hcy levels were not correlated with low femoral neck BMD10.

The variability in diagnostic criteria for homocysteinemia and osteoporosis among the studies and the limited number of individuals with hyperhomocysteinemia could be potential reasons for the inconsistent findings reported. Moreover, traditional observational studies can suffer from confounding bias, leading to discrepant results and concerns. Additionally, the temporal sequence between exposure and outcome is often misinterpreted, further complicating results interpretation. To address these issues, Katan proposed Mendelian Randomization (MR) research11.

MR is a method that uses instrumental variables (IVs) to examine exposure factors and simulates the influence of exposure factors on diseases by using the causal relationship effect between genotype and disease12. Given the random distribution of alleles during gamete formation, the association between genetic variations and diseases is less likely to be influenced by confounding bias and reverse causation compared to traditional epidemiological studies. Therefore, causal inference using genetic data has distinct advantages13. As a result, we postulated that this method could uncover causal relationships between Hcy and OP, thus offering further insights into the etiology of OP (Fig. 1) to facilitate early detection of osteoporosis and provide evidence-based intervention.

Diagram depicting two-sample Mendelian randomization analysis.

To minimize the potential impact of linkage disequilibrium (LD) on the analysis, we selected SNPs that were independent of each other. Specifically, the LD of SNPs associated with Hcy was constrained to r2 < 0.001 within a window size of 10,000 kb, po p = "1000G EUR". As instrumental variables for the MR analysis, 18 SNPs associated with circulating serum Hcy levels at genome-wide significance (p < 5 × 10–8) in the studies of the GWAS Catalog, including 44,147 individuals of European ancestry, were utilized14. The corresponding summary statistics for the associations between the Hcy-associated SNPs and F-BMD from a GWAS of 8143 individuals (ID: ieu-a-997), L-BMD from a GWAS of 28,498 individuals (ID: ieu-a-982), and H-BMD from a GWAS of 142,487 individuals(ID: ieu-a-GCST006288), of European ancestry in the GEFOS consortium UK Biobank (Neale Lab) were obtained. F-BMD, L-BMD, and H-BMD were derived from dual-energy X-ray absorptiometry. Relevant ethics committees approved all studies contributing data to these analyses.

Based on the selecting principle of IVs in MR analysis, these variants were in LD, so 5 SNPs were excluded (rs7422339, rs12134663, rs957140, rs12821383, and rs2851391). Hypertension15, serum Vit-B1215, high uric acid16, and lipid metabolism disorders17 are likely to be important confounders in the context of the Hcy-BMD relationship. PhenoScanner (www.phenoscanner.medschl.cam.ac.uk.) was used to assess the association of these SNPs with confounders such as hypertension, serum folic acid, high uric acid, and lipid metabolism disorders. 13 of the SNPs that were significantly associated with Hcy. SNP (rs154657) significantly correlated with hypertension, SNP (rs548987) significantly correlated with uric acid, SNP (rs1801222) significantly correlated with folic acid, and 2 SNPs (rs838133 and rs2251468) significantly correlated with lipid metabolism were excluded (Table S1). Eight SNPs strongly associated with homocysteine were included as final instrumental factors (Table 1). In order to standardize the effect alleles direction in the exposure dataset, data harmonization was performed between the exposure dataset and the outcome dataset.

Three MR approaches, including IVW, MR-Egger, and WM, were executed using the Two Sample MR packages in R version 4.2.2 (www.r-project.org). The results were reported as the mean effect per 1-SD genetically predicted rise in serum Hcy (log-transformed) and a 95% confidence interval (95% CI). In addition, a three-sided p < 0.017 = 0.05/3 was statistically significant.

The Cochran's Q test was used to identify heterogeneity among instrumental variables derived from IVW estimation. P < 0.05 was deemed statistically significant, indicating that heterogeneity may occur. The MR-Egger intercept was used to identify the directional pleiotropy of instrumental variables and was generated from the MR-Egger regression. When P < 0.05, it was deemed statistically significant, indicating that directional pleiotropy may occur. The leave-one-out test was employed to examine the strength of instrumental factors.

No need for ethical approval as used of anonymous open data.

Using IVW, MR-Egger, and WM regressions, the causal relationship between serum Hcy levels and osteoporosis indicators (F-BMD, L-BMD, and H-BMD) was analyzed. Upon examining the causal relationship between genetically predicted Hcy and F-BMD, and L-BMD using the IVW approach, no significant causal association was observed (all p value > 0.017) (Table 2). Furthermore, the regression slopes were observed to be in the same direction (Fig. 2). Conversely, utilizing the IVW technique, a causal and inverse relationship was found between genetically predicted Hcy and the H-BMD risk using the IVW technique [OR = 0.96, 95% CI = 0.927–0.990, p = 0.011] (Table 2). Similar trends were seen in the WM regression estimations [OR = 0.97, 95% CI = 0.995–1.076, p = 0.084]. Statistical significance was not found using the MR-Egger technique (OR = 0.98, 95% CI = 0.918–1.049, p = 0.609) (Fig. 3). Nonetheless, the consistency in the direction of the estimates generated by the three approaches lends credence to the findings' veracity. Evidently, from Figs. 2 and 4, each genetic variant contributes to the development of osteoporosis.

A scatter plot illustrating the causal relationship between Hcy and the risk of osteoporosis. The slope of the straight line represents the magnitude of the causal relationship. IVW, inverse-variance weighted; MR, Mendelian randomization; Hcy, homocysteine; F-BMD, forearm bone mineral density; L-BMD, lumbar bone mineral density; H-BMD, heel bone mineral density.

Forest plot to visualize the causal effect of Hcy on the risk of osteoporosis. IVW, inverse-variance weighted; MR, Mendelian randomization; Hcy, homocysteine; F-BMD, forearm bone mineral density; L-BMD, lumbar bone mineral density; H-BMD, heel bone mineral density.

Forest plot to visualize the causality of every single SNP on osteoporosis risk. (A) Hcy on F-BMD risk, (B) Hcy on L-BMD risk, (C) Hcy on H-BMD risk.

The IVW approach revealed no heterogeneity among the eight SNPs significantly associated with Hcy (p value > 0.017) (Table S2). The funnel plot's impact on causation was roughly symmetrical (Fig. 5). The MR-Egger-regression showed no directional pleiotropy (Hcy to F-BMD: Egger intercept = − 0.011, SE = 0.018, p = 0570; Hcy to L-BMD: Egger intercept = 0.017, SE = 0.120, p = 0.425; Hcy to H-BMD: Egger intercept = − 0.002, SE = 0.002, p = 0.4449). The leave-one-out analysis showed that a single SNP did not drive the MR analysis results (Fig. 6).

Funnel plots to visualize the overall heterogeneity of MR estimates for the effect of Hcy on osteoporosis. (A) Hcy on F-BMD risk, (B) Hcy on L-BMD risk, (C) Hcy on H-BMD risk; MR, Mendelian randomization; Hcy, homocysteine; F-BMD, forearm bone mineral density; L-BMD, lumbar bone mineral density; H-BMD, heel bone mineral density.

Leave-one-out plot to visualize the causal effect of Hcy on the risk of osteoporosis when leaving one SNP out. MR, Mendelian randomization; Hcy, homocysteine; F-BMD, forearm bone mineral density; L-BMD, lumbar bone mineral density; H-BMD, heel bone mineral density.

Mendelian randomization was employed for the first time in this investigation to examine the association between Hcy and OP in the European population. The findings revealed a causal relationship between serum Hcy and H-BMD, but but no causal relationship existed between Hcy and F-BMD and L-BMD.

As mentioned previously, the three assumptions of the MR Model must be satisfied for causal estimates to be accurate. Thirteen highly-associated and independent SNPs closely linked to Hcy were identified in this study, satisfying the first assumption. Furthermore, the exclusion of five SNPs (rs838133, rs154657, rs548987, rs1801222, and rs2251468) connected with confounding variables, such as lipid metabolism, hypertension, uric acid, and vitamin B, as documented in previous studies, mitigated the bias caused by a polymorphism in MR trials, satisfying the second assumption. Additionally, the MR-Egger intercepts were near 0 in this investigation (p > 0.05), implying that unknown variables did not lead to pleiotropy, satisfying the third assumption. Therefore, the SNPs used and the outcomes of the research are credible.

BMD is a crucial indicator for the diagnosis and evaluation of osteoporotic fractures. Studying 188 postmenopausal Moroccan women, Quzzif et al. found that plasma Hcy was significantly higher in the osteoporosis group than in the non-osteoporosis group and that Hcy was inversely linked with lumbar and total hip BMD18. Significant negative correlations were discovered between Hcy and femoral neck BMD (beta = 0.032, p = 0.010) and lumbar spine (beta = − 0.098, p = 0.021) in a large cross-sectional study (n = 6107). Bone mineral density was found to be affected by serum folate and body mass index but not by Hcy, as shown by a stepwise regression analysis19,20. Although the association between serum Hcy and BMD remains contentious, increasing Hcy levels increase the OP risk, and it is commonly acknowledged that Hcy is a risk factor for OP and osteoporotic fractures.

There are limitations in this MR. First, although the causal association between Hcy and H-BMD was verified in this investigation, it was not established between Hcy, F-BMD, and L-BMD mainly because the sample size was insufficient to provide meaningful data. Second, we used aggregated GWAS data, and the lack of specific information on sex and age prevented us from conducting subgroup analysis. Third, since the research was based on genetic information from European groups, it needs to be seen whether the findings apply to Asian people. Fourth, due to the absence of GWAS data, this research gathered only H-BMD, F-BMD, and L-BMD data. Other BMD indices, such as femoral neck bone density and hip bone density, should be included in future research to get more trustworthy findings. Finally, although we obtained a small OR, focusing on the causal relationship between Hcy and H-BMD is warranted because of the large prevalence.

In conclusion, our MR investigation revealed evidence of a causal relationship between circulating serum Hcy levels and H-BMD, but not OP. This finding needs to be further verified in a study with a larger sample size.

The data and material that support the findings of this study are available from public datasets that could be found in GWAS Catalog.

Park, D. H. et al. Long-term proton pump inhibitor use and risk of osteoporosis and hip fractures: A nationwide population-based and multicenter cohort study using a common data model. J. Gastroenterol. Hepatol. 37(8), 1534–1543 (2022).

Article CAS PubMed Google Scholar

Lee, D. Y. & Chung, Y. S. Quality of life and patient satisfaction with raloxifene/cholecalciferol combination therapy in postmenopausal women. Sci. Rep. 12(1), 7142 (2022).

Article ADS CAS PubMed PubMed Central Google Scholar

Gjesdal, C. G. et al. Plasma total homocysteine level and bone mineral density: The Hordaland Homocysteine Study. Arch. Intern. Med. 166(1), 88–94 (2006).

Article CAS PubMed Google Scholar

McLean, R. R. et al. Homocysteine as a predictive factor for hip fracture in older persons. N. Engl. J. Med. 350(20), 2042–2049 (2004).

Article CAS PubMed Google Scholar

Enneman, A. W. et al. The association between plasma homocysteine levels, methylation capacity and incident osteoporotic fractures. Bone 50(6), 1401–1405 (2012).

Article CAS PubMed Google Scholar

Blouin, S. et al. Bone matrix quality and plasma homocysteine levels. Bone 44(5), 959–964 (2009).

Article CAS PubMed Google Scholar

Lubec, B. et al. Evidence for McKusick's hypothesis of deficient collagen cross-linking in patients with homocystinuria. Biochim. Biophys. Acta 1315(3), 159–162 (1996).

Article PubMed Google Scholar

Koh, J. M. et al. Homocysteine enhances bone resorption by stimulation of osteoclast formation and activity through increased intracellular ROS generation. J. Bone Miner. Res. 21(7), 1003–1011 (2006).

Article CAS PubMed Google Scholar

Sakamoto, W. et al. Homocysteine attenuates the expression of osteocalcin but enhances osteopontin in MC3T3-E1 preosteoblastic cells. Biochim. Biophys. Acta 1740(1), 12–16 (2005).

Article CAS PubMed Google Scholar

Gerdhem, P. et al. Associations between homocysteine, bone turnover, BMD, mortality, and fracture risk in elderly women. J. Bone. Miner. Res. 22(1), 127–134 (2007).

Article CAS PubMed Google Scholar

Katan, M. B. Apolipoprotein E isoforms, serum cholesterol, and cancer. Lancet 1(8479), 507–508 (1986).

Article CAS PubMed Google Scholar

Bochud, M. & Rousson, V. Usefulness of Mendelian randomization in observational epidemiology. Int. J. Environ. Res. Public Health 7(3), 711–728 (2010).

Article CAS PubMed PubMed Central Google Scholar

Haycock, P. C. et al. Best (but oft-forgotten) practices: The design, analysis, and interpretation of Mendelian randomization studies. Am. J. Clin. Nutr. 103(4), 965–978 (2016).

Article CAS PubMed PubMed Central Google Scholar

van Meurs, J. B. et al. Common genetic loci influencing plasma homocysteine concentrations and their effect on risk of coronary artery disease. Am. J. Clin. Nutr. 98(3), 668–676 (2013).

Article PubMed PubMed Central Google Scholar

He, F. J. et al. Salt reduction to prevent hypertension and cardiovascular disease: JACC state-of-the-art review. J. Am. Coll. Cardiol. 75(6), 632–647 (2020).

Article CAS PubMed Google Scholar

Yao, X. et al. The association between serum uric acid and bone mineral density in older adults. Int. J. Endocrinol. 2020, 3082318 (2020).

Article PubMed PubMed Central Google Scholar

Kan, B. et al. Association between lipid biomarkers and osteoporosis: A cross-sectional study. BMC Musculoskelet. Disord. 22(1), 759 (2021).

Article CAS PubMed PubMed Central Google Scholar

Ouzzif, Z. et al. Relation of plasma total homocysteine, folate and vitamin B12 levels to bone mineral density in Moroccan healthy postmenopausal women. Rheumatol. Int. 32(1), 123–128 (2012).

Article CAS PubMed Google Scholar

Enneman, A. W. et al. The association between plasma homocysteine levels and bone quality and bone mineral density parameters in older persons. Bone 63, 141–146 (2014).

Article CAS PubMed Google Scholar

Anagnostis, P. et al. Bone health in patients with dyslipidemias: An underestimated aspect. Int. J. Mol. Sci. 23(3), 1639 (2022).

Article CAS PubMed PubMed Central Google Scholar

Download references

We thank the Genetic Factors of Osteoporosis Consortium (GEFOS)UK Biobank for the BMD GWAS datasets. We also thank the Women's Genome Health Study (WGHS) for other GWAS datasets. The coordinates system for genome positions is GRCH38/hg38.

This research was supported by the National Natural Science Foundation of China (81071494) and the Science and Technology Department of Hubei Province (2015BCA316). The funding agencies had no role in study design, data collection, analysis, publication decisions, or manuscript preparation.

Department of Orthopedics, Renmin Hospital of Wuhan University, Wuhan, China

ChenYu Wang & Bo Qiu

Department of Geriatrics, Renmin Hospital of Wuhan University, Wuhan, China

Xiang Zhang

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

W.Y. and B.Q. conceived of the presented idea. W.Y. and X.Z. developed the theory and performed the computations. W.Y. verified the analytical methods. All authors discussed the results and contributed to the final manuscript. The authors declare that there are no conflicts of interest regarding the publication of this paper.

Correspondence to Bo Qiu.

The authors declare no competing interests.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

Wang, C., Zhang, X. & Qiu, B. Genetically predicted circulating serum homocysteine levels on osteoporosis: a two-sample mendelian randomization study. Sci Rep 13, 9063 (2023). https://doi.org/10.1038/s41598-023-35472-2

Download citation

Received: 25 November 2022

Accepted: 18 May 2023

Published: 04 June 2023

DOI: https://doi.org/10.1038/s41598-023-35472-2

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.