Exploring causal associations between the plasma proteome on bone mineral density: a Mendelian randomization study

Osteoporosis is characterized by reduced bone mineral density (BMD) and an increased risk of fractures, but the relationships between plasma proteins and site-specific BMD phenotypes remain unclear. We aimed to investigate the potential causal associations of plasma proteins on BMD using a Mendelian randomization (MR).

We conducted two-sample MR analysis as the primary analysis, using cis-protein quantitative trait loci (pQTLs) instruments from nine proteomic genome-wide association studies and summary statistics for total-body, forearm, femoral-neck, and lumbar-spine BMD. Prioritized associations were evaluated using colocalization, Steiger directionality testing, protein-altering variant (PAV) annotation, cis-expression quantitative trait locus (cis-eQTL) overlap, and targeted replication. Protein-protein interaction (PPI), pathway enrichment, and druggability analyses were exploratory.

23 study-specific associations met the significance thresholds and involved 12 proteins: F2, TGFBI, PLEKHA1, HTRA1, PKDCC, CFHR4, LRP4, RSPO3, DKK1, TNFRSF11A, SPP1, and IBSP. Strong colocalisation supported associations of F2 with total-body BMD, TGFBI with femoral-neck BMD, and PLEKHA1 with total-body and lumbar-spine BMD. Steiger testing supported directionality for 22 of 23 associations, except PKDCC-forearm BMD. Targeted replication provided no statistically significant, directionally concordant support. LRP4 relied on the only direct target-protein PAV, while 8 cis-pQTLs for 6 proteins overlapped with cis-eQTLs. Exploratory analyses implicated skeletal and extracellular-matrix processes and targets of prior pharmacological interest.

Using cis-pQTL-based MR, we identified circulating proteins associated with BMD phenotypes. Strong colocalization was observed for F2, TGFBI, and PLEKHA1, whereas the remaining associations require additional validation. These findings provide genetic support for established bone-related pathways and highlight candidate proteins.