A Large-Scale Multi-omics Polygenic Risk Score Analysis Identified Candidate Biomarkers Associated with Heel Bone Mineral Density

Bone mineral density (BMD) is a critical indicator of osteoporosis (OP). Utilizing the latest multi-omics quantitative trait loci (QTLs) data, we aim to identify novel candidates associated with heel BMD (hBMD).

We collected QTLs data from the INTERVAL cohort (independent of the UK Biobank (UKB)) that is a randomised trial including approximately 50,000 healthy blood donors enrolled from 25 centres of England's National Health Service Blood and Transplant. We then calculated individual polygenic risk score (PRS) for 13,646 RNAs, 308 proteins (Olink), 2379 proteins (SomaScan), 726 metabolites (Metabolon) and 141 metabolites (Nightingale) in UKB. hBMD was measured by quantitative ultrasound. Generalized linear model was used to evaluate the associations between multi-omics PRS and hBMD in 96,165 subjects. Integrated analysis of multi-omics was performed on the MetaboAnalyst 6.0 platform. Replication analysis was performed using the internal Olink proteomics and Nightingale metabolomics data of UKB. We subsequently explored the causal effects of the targets on hBMD using mendelian randomization (MR) analysis. Significant associations were determined using a false discovery rate (FDR)-adjusted P-value (Padj < 0.05).

We identified 195 hBMD-associated genes, such as WNT16 (Padj = 6.417 × 10- 14); 180 proteins such as COL1A1 (Padj = 3.132 × 10- 24); 21 metabolites, such as total cholesterol (Padj = 0.008). Those associated proteins/metabolites were replicated in UKB, such as COL1A1 (Padj = 7.480 × 10- 4) and total cholesterol (Padj = 7.773 × 10- 8). The integrated analysis of multi-omics identified several genes/metabolites (DGKZ, PLPP3, GPD1L, CHKB and glycerol 3-phosphate) enriched in the overlapping pathway-glycerophospholipid metabolism. Moreover, MR detected several novel biomarkers among the top/enriched targets, which were causally associated with hBMD, such as NAP1L2 (β= - 0.019, P = 0.003) and PLPP3 (β = - 0.026, P = 0.041).

Our study not only identified several novel candidate biomarkers such as PLPP3, RGMB, and RNF128 for hBMD but also provided genetic evidence supporting their potential causal roles in bone mineral regulation. These findings could shed light on the molecular underpinnings of OP.