Causal relationship between Type 1 diabetes and osteoporosis and fracture occurrence: a two-sample Mendelian randomization analysis.

Abstract

Based on the random assignment of gametes at conception, Mendelian randomization (MR) analysis simulates randomized controlled trials in an observational setting. Therefore, we used MR to assess the association causality of type 1 diabetes (T1D) with fractures and osteoporosis. From a genome-wide association meta-analysis, independent single nucleotide polymorphisms closely associated with T1D were selected as instrumental variables. Data on fracture and osteoporosis were obtained from the FinnGen Consortium. We performed a two-sample MR analysis, using inverse-variance weighted (IVW) as the primary analysis method, to assess possible causal associations between T1D and bone risk. The results were verified by MR-Egger regression and median weighted method (WME). MR-PRESSO and MR-Egger intercepts were used to evaluate the horizontal pleiotropy of instrumental variables, and the Q-test and "leave-one-out" methods were used to test the heterogeneity of MR results. IVW (OR=1.040, 95% CI=0.974-1.109, P=0.238), MR-Egger regression (OR=1.077, 95% CI=0.921-1.260, P=0.372) and WME (OR=1.021, 95% CI=0.935-1.114, P=0.643) all showed that there was no causal relationship between T1D and osteoporosis, but the direction was consistent. The indicative significance of IVW results in T1D and forearm fractures (OR=1.062, 95% CI=1.010-1.117, P=0.020), but the results are not robust enough. There was no causal effect in femur, lumbarand pelvis, or shoulder andupper arm fractures. After MR analysis, although T1D may be a risk factor for bone health, we do not have sufficient evidence to support a causal effect of T1D on osteoporosis and fractures at a genetically predicted level. More cases need to be included for analysis.