Obstructive sleep apnea shares a common genetic liability with type 2 diabetes, hypertension, and adiposity: a multivariate genome-wide association study

Obstructive sleep apnea (OSA) frequently co-occurs with obesity, hypertension, and type 2 diabetes, but the contribution of shared genetic architecture to this clustering remains unclear. We characterized the common polygenic component linking OSA with cardiometabolic traits. We integrated European-ancestry genome-wide association studies (GWAS) of OSA, body mass index, type 2 diabetes, and hypertension using Genomic Structural Equation Modeling (Genomic SEM) to derive a latent multivariate obstructive sleep apnea–cardiometabolic disease factor (mvOSACMD). We then performed multivariate GWAS, locus discovery, transcriptome-wide association studies (TWAS) with fine-mapping and colocalization, enrichment and single-cell spatial analyses, heritability partitioning, and Polygenic Risk Score with Continuous Shrinkage (PRS-CS) posterior effect-size estimation. The common-factor model fitted well (comparative fit index = 0.968; standardized root mean square residual = 0.043). Multivariate GWAS tested 4,844,928 single nucleotide polymorphisms (SNPs), identifying 553 genomic loci and 808 independent lead variants. SNP-level heterogeneity testing with the Q SNP statistic showed that 799 lead variants (98.9%) had no genome-wide significant heterogeneity, and heterogeneous variants occurred within 27 loci (4.9%). Ninety-five variants reached genome-wide significance in mvOSACMD but not in any input GWAS. Fine-mapping prioritized 33 variants with mean posterior inclusion probability ≥ 0.95 across Sum of Single Effects (SuSiE) and FINEMAP, including six lead variants. TWAS with gene-level fine-mapping prioritized 234 genes, of which 101 colocalized with the mvOSACMD signal at posterior probability > 0.8. Enrichment analyses identified neuronal and neurodevelopmental regulatory programs, and single-cell spatial mapping showed the strongest enrichment in the central nervous system (CNS), peaking in brain ( P = 1.87 × 10⁻¹⁹) and spinal cord ( P = 8.51 × 10⁻¹⁷). Partitioned heritability was enriched in evolutionarily conserved elements and active regulatory annotations. OSA and cardiometabolic traits share a measurable common genetic component, with prioritized genes and enriched annotations pointing to neuronal and neurodevelopmental regulatory programs. The specificity of CNS enrichment to the shared factor remains uncertain because input-trait enrichment profiles were not directly compared. This variant- and gene-level resource provides hypotheses for functional follow-up; clinical polygenic prediction requires independent validation with individual-level data.