S1-S10 Figs are included in the file.

Background
Observational studies have suggested a relationship between type-1 diabetes mellitus (T1DM) and systemic lupus erythematosus (SLE). In both autoimmunities, 25-hydroxyvitamin D (25-OHD) deficiency is common. However, the causality between T1DM, 25-OHD level and SLE remains largely unknown.
Methods
Independent genetic variants associated with T1DM, 25-OHD level, and SLE from the largest genome-wide association studies were used to conduct two-sample bidirectional Mendelian randomization (BIMR) and two-step Mendelian randomization (MR) analysis to estimate causal relationship between T1DM, 25-OHD level and SLE, and further multivariable Mendelian randomization (MVMR) was used to verify direct causality of T1DM and 25-OHD level on SLE. A series of sensitivity analysis as validation of primary MR results were performed.
Results
Consistent with the results of BIMR, there was strong evidence for a direct causal effect of T1DM on the risk of SLE (ORMVMR-IVW = 1.249, 95% CI = 1.148–1.360, PMVMR-IVW = 1.25×10−5), and 25-OHD level was negatively associated with the risk of SLE (ORMVMR-IVW = 0.305, 95% CI = 0.109–0.857, PMVMR-IVW = 0.031). We also observed a negative causal effect of T1DM on 25-OHD level (ORBIMR-IVW = 0.995, 95% CI = 0.991–0.999, PBIMR-IVW = 0.030) while the causal effect of 25-OHD level on the risk of T1DM did not exist (PBIMR-IVW = 0.106). In BIMR analysis, there was no evidence for causal effects of SLE on the risk of T1DM and 25-OHD level (PBIMR-IVW > 0.05, respectively).
Conclusion
Our MR analysis suggested that there was a network causal relationship between T1DM, 25-OHD level and SLE. T1DM and 25-OHD level both have causal associations with the risk of SLE, and 25-OHD level could be a mediator in the causality of T1DM and SLE.

研究背景
观察性研究提示1型糖尿病（type-1 diabetes mellitus, T1DM）与系统性红斑狼疮（systemic lupus erythematosus, SLE）之间存在关联。在这两种自身免疫性疾病中，25-羟维生素D（25-hydroxyvitamin D, 25-OHD）缺乏均较为常见。然而，T1DM、25-OHD水平与SLE之间的因果关联目前仍未完全明确。
研究方法
本研究采用最大规模全基因组关联研究（genome-wide association studies, GWAS）中与T1DM、25-OHD水平及SLE相关的独立遗传变异，开展两样本双向孟德尔随机化（two-sample bidirectional Mendelian randomization, BIMR）与两步孟德尔随机化（Mendelian randomization, MR）分析，以评估T1DM、25-OHD水平与SLE之间的因果关联；进一步采用多变量孟德尔随机化（multivariable Mendelian randomization, MVMR）验证T1DM与25-OHD水平对SLE的直接因果效应。同时开展一系列敏感性分析以验证核心孟德尔随机化分析结果的可靠性。
研究结果
与BIMR分析结果一致，本研究发现T1DM对SLE发病风险存在显著直接因果效应（多变量孟德尔随机化-IVW法：ORMVMR-IVW=1.249，95%置信区间（confidence interval, CI）=1.148~1.360，PMVMR-IVW=1.25×10⁻⁵）；25-OHD水平与SLE发病风险呈负相关（ORMVMR-IVW=0.305，95%置信区间=0.109~0.857，PMVMR-IVW=0.031）。此外，本研究观察到T1DM对25-OHD水平存在负向因果效应（BIMR-IVW法：ORBIMR-IVW=0.995，95%置信区间=0.991~0.999，PBIMR-IVW=0.030）；但25-OHD水平对T1DM发病风险无显著因果效应（PBIMR-IVW=0.106）。在BIMR分析中，未发现SLE对T1DM发病风险及25-OHD水平存在因果效应（两组BIMR-IVW检验P值均>0.05）。
研究结论
本研究的孟德尔随机化分析表明，T1DM、25-OHD水平与SLE之间存在网络化因果关联。T1DM与25-OHD水平均与SLE发病风险存在因果关联，且25-OHD水平可能是介导T1DM与SLE之间因果通路的中介变量。