Shared Genetic Architecture and Causal Relationship Between Serum 25-Hydroxyvitamin D and Bone Mineral Density

Abstract

Context: Despite the well-established regulatory role of vitamin D in maintaining bone health, little is known about the shared genetics and causality of the association between serum 25-hydroxyvitamin D (25OHD) and bone mineral density (BMD).

Objective: We aimed to investigate the shared genetic architecture and causal relationship between serum 25OHD and BMD, providing insights into their underlying biological mechanisms.

Methods: Leveraging individual-level data from the UK Biobank (UKB) cohort and summary-level data from the genome-wide association studies (GWASs) conducted on European individuals for serum 25OHD (N = 417 580) and estimated heel BMD (eBMD, N = 426 824), we systematically elucidated the shared genetic architecture underlying serum 25OHD and eBMD through a comprehensive genome-wide cross-trait design.

Results: Despite a lack of global genetic correlation (rg=-0.001; P = .95), a statistically significant local signal was discovered at 5p11-5q11.9. Two-sample mendelian randomization (MR) indicated no causal association in the overall population (β=.003, 95% CI, -0.04 to 0.03; P = .93), while positive causal effects were observed in males (β=.005, 95% CI, 0.00 to 0.01; P = .03) and older individuals (β=.009, 95% CI, 0.00∼0.02; P = .01) according to one-sample MR. A total of 49 pleiotropic single-nucleotide variations (SNVs), with 4 novel SNVs (rs1077151, rs79873740, rs12150353, and rs4760401), were identified, and a total of 95 gene-tissue pairs exhibited overlap, predominantly enriched in the nervous, digestive, exocrine/endocrine, and cardiovascular systems. Protein-protein interaction analysis identified RPS9 and RPL7A as hub genes.

Conclusion: This study illuminates the potential health benefits of enhancing serum 25OHD levels to mitigate the risk of osteoporosis among men and individuals older than 65 years. It also unveils a shared genetic basis between serum 25OHD and eBMD, offering valuable insights into the intricate biological pathways.

Keywords: bone mineral density; causal inference; genetic correlation; pleiotropic loci; serum 25-hydroxyvitamin D.

Figure 1.

Flowchart of the overall study design. We first investigated the genetic correlation leveraging the GWASs of serum 25OHD and eBMD. We then conducted 2-sample MR and 1-sample MR to detect putative causal links between serum 25OHD and eBMD. Finally, we applied cross-trait meta-analysis and transcriptome-wide association study to identify pleiotropic loci and shared gene-tissue pairs, and further carried out a series of functional analyses to understand the potential biological mechanisms. Abbreviations: GWAS, genome-wide association study; LDSC, linkage disequilibrium score regression; MR, mendelian randomization; PLEIO, Pleiotropic Locus Exploration and Interpretation using Optimal test; STRING, Search Tool for the Retrieval of Interacting Genes/Proteins; SUPERGNOVA, SUPER GeNetic cOVariance Analyzer; UKB, UK Biobank.

Figure 2.

Local genetic correlation between serum 25-hydroxyvitamin D (25OHD) and estimated heel bone mineral density (eBMD). (A) Quantile-quantile (QQ) plot presents region-specific P values from the local genetic correlation by SUPERGNOVA. (B) Manhattan-style plot shows the estimates of local genetic covariance. Bold dots represent a genomic region contributing to a significant local genetic correlation with serum 25OHD and eBMD after multiple testing adjustments (P < .05/2353). $r_g$, genetic correlation.

Figure 3.

Causal relationships between serum 25-hydroxyvitamin D (25OHD) and bone mineral density. (A) Bidirectional 2-sample mendelian randomization (MR) analysis. We applied univariate MR and multivariate MR to explore bidirectional causality between serum 25OHD and estimated heel bone mineral density (eBMD). (B) One-sample MR analysis. Linear MR was conducted across the overall, sex-, and age-specific populations, followed by nonlinear MR categorizing participants into 4 subgroups based on quartiles of residual serum 25OHD levels. Abbreviations: FA-BMD, forearm bone mineral density; FN-BMD, femoral neck bone mineral density; LS-BMD, lumbar spine bone mineral density; Q, quartile; TB-BMD, total body bone mineral density.

Figure 4.

Protein-protein interaction network of shared genes between serum 25-hydroxyvitamin D (25OHD) and estimated heel bone mineral density (eBMD). We constructed protein-protein interaction network for the genes located at pleiotropic loci identified in Pleiotropic Locus Exploration and Interpretation using Optimal test (PLEIO) and transcriptome-wide shared genes obtained from transcriptome-wide association studies (TWAS). The circles represent genes, with deeper color indicating a higher connectivity degree. The lines indicate interactions among genes, with thicker lines representing higher combined scores.