Tumor necrosis factor mediates the impact of PM2.5 on bone mineral density: Inflammatory proteome Mendelian randomization and colocalization analyses

Abstract

To assess the causal effect of particulate matter 2.5 (PM_2.5) on human bone mineral density (BMD) and to explore the possible mechanism and proportion mediated by inflammation-related protein. The genetic correlation between PM_2.5 and BMD was assessed using the Linkage Disequilibrium Score (LDSC), and the causal effect between PM_2.5 and BMD was assessed by two-sample Mendelian randomization (TSMR). A 2-step Mendelian randomization (MR) approach was employed to evaluate the potential role of inflammation-associated protein as the mediator in the causal association between PM_2.5 and BMD. The multivariate Mendelian randomization (MVMR) study was designed to perform mediation analyses, exclude possible confounders and calculate the proportion of mediation. Subsequently, we used Bayesian colocalization analysis to consolidate the MR results. Finally, using drug-target MR design, we evaluated the potential repurposing of tumor necrosis factor (TNF) inhibitors for the treatment of osteoporosis (OP). The results of the analyses show that BMD is negatively influenced by PM_2.5 (Inverse variance weighted [IVW] beta [β] = -0.288, 95% confidence interval [CI]: -0.534 - -0.042, P < 0.05). PM_2.5 has a positive causal association with TNF (IVW β = 1.564, 95% CI: 0.155 - 2.973, P < 0.05) and a negative causal association with protachykinin-1 (TAC-1) (IVW β = -1.654, 95% CI: -3.063 - -0.244, P < 0.05). TNF has a negative causal association with BMD (Wald ratio β = -0.082, 95% CI: -0.165 - 0.000, P < 0.05) and TAC-1 has a positive causal association with BMD (IVW β = 0.042, 95% CI: 0.007 - 0.077, P < 0.05). After adjusting TNF and TAC-1, PM_2.5 has no causal association with BMD (IVW β = -0.200, 95% CI: -0.579 - 0.179, P > 0.05). After adjusting PM_2.5 and TAC-1, there was still a negative causal association between TNF and BMD (IVW β = -0.089, 95% CI: -0.166 - -0.012, P < 0.05). In the final drug-target MR study, the protective effect of TNF/TNF receptor 1 (TNFR1) inhibition on BMD was observed. For every 10% decrease of circulating C-reactive protein (CRP) achieved by TNF/TNF receptor 1 (TNFR1) blockade, β was 0.540 (95% CI: 0.040-1.040) for BMD. We found a negative causal association between PM_2.5 and BMD and that causal association was mediated by TNF. The results of drug-target MR do support TNFR1 as a promising target for OP prevention among the general population.

Keywords: Bone mineral density; Drug target; Mendelian randomization; Osteoporosis; PM(2.5); Tumor necrosis factor.