Genomic structural equation modeling uncovers novel risk loci for bone metabolic disorders: cross-tissue genetic mechanisms and bone-brain axis regulation.

OBJECTIVE

The genetic architecture underlying bone metabolic disorders (BMetDs) remains poorly characterized. Utilizing genomic structural equation modeling (Genomic-SEM) and multiple post-GWAS approaches, we estimated causal single nucleotide polymorphisms (SNPs) associated with BMetDs-independent metabolic variations, identifying 26 novel risk loci (including 14 genome-wide significant loci). Multi-omics analyses-including tissue-specific, cellular-level, and genomic element annotations-were applied to prioritize susceptibility loci and regulatory elements. Additionally, polygenic risk scores (PRS) derived from summary statistics were used to evaluate chromosomal contributions to BMetDs risk. For the first time, we present a comprehensive genetic landscape of BMetDs through a GWAS targeting an unmeasured latent phenotype.

METHODS

We conducted a multivariate GWAS of BMetDs using genomic structural equation modeling (Genomic-SEM) on six BMetDs-related traits: osteoporosis, type 2 diabetes (T2D), body mass index (BMI), osteoarthritis (OA), ankylosing spondylitis (AS), and rheumatoid arthritis (RA) (total sample size: N = 462,933-461,194 Europeans). Genetic covariance matrices were constructed via LD score regression, followed by latent factor modeling to dissect shared genetic architecture. Novel loci were prioritized using FUMA-based functional annotation, fine-mapping (SuSIE/FINEMAP), transcriptome-wide association studies (TWAS/FUSION), and cell-type enrichment analyses.

RESULTS

Genomic-SEM identified 37 novel risk loci (P < 5 × 10⁻⁸), including 14 genome-wide significant loci (P < 5 × 10⁻¹⁶). Fine-mapping revealed causal variants (posterior probability > 0.95) on chromosome 1 (rs1415145), chromosome 2 (rs35802221), and chromosome 7 (rs2367906). TWAS implicated FERMT3 (β = +1.85, P = 2.3 × 10⁻⁶), CCDC88B (β = +2.01, P = 1.7 × 10⁻⁵), and USP53 (β = -1.72, P = 4.1 × 10⁻⁴) in BMetDs through cerebellar-hypothalamic gene expression. Cell-type enrichment highlighted pancreatic B cells (FDR = 0.03) and brain non-myeloid neurons (FDR = 0.04) as key mediators. Partitioned heritability analysis demonstrated significant enrichment in enhancer (H3K27ac, P = 1.1 × 10⁻⁷) and chromatin boundary (CTCF, P = 3.4 × 10⁻⁵) regions, supporting epigenetic dysregulation in BMetDs pathogenesis.

CONCLUSION

This study uncovers novel genetic loci and cross-tissue regulatory mechanisms driving BMetDs, emphasizing the pivotal roles of the bone-brain axis and non-coding regulatory elements. Our findings bridge genomic insights with translational applications, offering actionable targets for precision interventions in metabolic bone diseases.