Bone Abstracts

Background: Recent technological ‘omics’ advances have empowered us to identify associations between genetic markers and various traits. Knowledge of serum metabolites as intermediary phenotypes can help us achieve a better understanding of the causal pathways from genes to complex diseases like osteoporosis.

Methods: In the context of TwinsUK study, serum metabolomic profiles of 6055 participants were assessed using a non-targeted mass spectrometry platform (Metabolon, Inc., Durham, NC, USA). The concentrations of 510 serum metabolites (211 unknown and 299 known molecules including amino-acids, lipids, carbohydrates, vitamins, peptides, and xenobiotics) were measured. Out of 6055 participants, 5224 (86%) were female twins with at least one hip and spine DXA measurement (Hologic devices) and 5605 (92%) had genome-wide genotyping scans (Illumina platforms imputed to 2.5 million single nucleotide polymorphisms). Genome-wide association studies (GWAS) for all metabolites and direct associations between metabolites and bone phenotypes (femoral neck, total hip, and lumbar spine BMD) were measured using mixed-effects models adjusting for age, height, weight and family-relatedness. Causal associations between metabolites and bone phenotypes were assessed using genetic markers as instrumental variables.

Results: Several metabolites showed significant associations (corrected for multiple testing) with bone phenotypes including prolyl-hydroxy-proline (P=1.65×10⁻¹⁷), pipecolic acid (P=1.27×10⁻⁷) and several sulphated steroids. In Mendelian randomisation analysis, epiandrosterone sulphate (encoded by CYP3A5 on chromosome 7q22.1) and 4-androsten-3β,17β-diol disulphate (encoded by SULT2A1 on 19q13.33) were causally associated with femoral neck BMD. Total hip BMD changes were additionally caused by changes in butyryl-carnitine levels (encoded by ACADS on 12q24.31). Lumbar spine BMD was causally associated to an unknown metabolite encoded by ABCC4 on chromosome 13q32.1.

Discussion: To our knowledge, this is the first metabolome-genome-wide Mendelian randomisation study of human bone mineral density. Causal associations observed in this study can direct us to biological pathways involved in the pathogenesis of osteoporosis. Our results need replication in other studies.