Proteome-wide Mendelian randomization implicates TIMP2 as a putative causal protein for bone mineral density and fracture risk

Abstract

Osteoporosis is a highly prevalent cause of fractures in older adults and remains a critical source of morbidity that requires continued efforts to develop new therapeutics. Circulating proteins play a critical role in the pathophysiology of osteoporosis and offer opportunities to identify new putatively causal determinants of bone health. We therefore performed a large-scale proteome-wide Mendelian randomization (MR) analysis to systematically estimate the effects of genetically determined circulating proteins levels on bone mineral density (BMD) and fracture risk. Circulating proteins play critical role in osteoporosis pathophysiology and offer opportunities to identify new putative causal proteins of bone health, we performed large-scale proteome-wide Mendelian randomization (MR) to systematically estimate the effect of genetically determined circulating proteins levels on bone mineral density (BMD) and fracture risk. Genetic instruments were derived from cis-protein quantitative trait loci (cis-pQTLs) for 2,110 plasma proteins across four European ancestry cohorts and applied to genome-wide association studies (GWAS) of heel estimated BMD, femoral neck BMD, lumbar spine BMD, any fracture, and forearm fracture in up to 426,824 individuals of European ancestry. Across proteins and outcomes, 192 protein-outcome associations showed MR evidence of association and passed checks for weak instruments, heterogeneity, and horizontal pleiotropy, and 128 of these also showed strong colocalization with osteoporosis-related loci. We then prioritized proteins that replicated across cohorts, exhibited concordant effect directions, and were likely to be active in circulation, yielding 18 high-confidence causal proteins for BMD and fracture risk. These included established skeletal regulators such as sclerostin (SOST) and R-spondin-3 (RSPO3), which showed opposing effects consistent with their known biology, along with less well-characterized proteins. Among the latter, tissue inhibitor of metalloproteinases 2 (TIMP2) emerged as a particularly compelling candidate: higher genetically predicted TIMP2 levels were associated with lower BMD and increased forearm fracture risk. Gene-level and variant-level phenome-wide association analyses converged on musculoskeletal traits, and rare predicted damaging or loss-of-function variants in TIMP2 were associated with higher BMD T-scores at the heel, spine and hip. Our findings implicate several circulating proteins as putatively causal factors for osteoporosis and, among them, provide multiple layers of evidence supporting TIMP2 as a potential therapeutic target.