Will diabetes mellitus induce aplastic anemia by immune factors?: A two-sample and mediation Mendelian randomization study

Diabetes is a prevalent global metabolic and endocrine disorder that is associated with a high incidence of complications and organ system dysfunction. Bone marrow (BM) as a previously neglected site of diabetic end-organ damage, characterized by microangiopathy, neuropathy, fat deposition, and inflammation. As a result, diabetes may lead to negative consequences for physiologic hematopoiesis, which may increase the risk of aplastic anemia (AA). Summary genetic data for diabetes mellitus (DM) were sourced from FinnGen; while the data for AA and immune cell traits were obtained from the IEU Open GWAS database. We performed a two-sample univariable Mendelian randomization (MR) analysis to investigate the causal effects of DM on AA. Simultaneously, multivariable MR was utilized to further estimate the direct effect of subgroup (distinct types) of diabetes on AA. Then, a two-step mediation MR analysis was conducted to examine 731 immune cell traits that may mediate these effects. Several methods were used to evaluate the robustness of the results, including sensitivity analyses with Cochran Q statistic, MR-Egger and MR-PRESSO, which also help mitigate potential bias from horizontal pleiotropy. The two-sample univariate MR analysis demonstrated DM was significantly and positively linked to the incidence of AA (IVW, OR = 1.12; 95% CI: 1.05-1.95; P = 5.11e-04), which was comparable to the direct effect estimated for type 2 diabetes on AA risk in multivariable MR (multivariable IVW, OR = 1.18; 95% CI: 1.03-1.35; P = 1.96e-02). In two-step mediation MR, we explored 54 immune cell traits associated with DM, among these, only the Resting CD4 + regulatory T cell absolute count emerged as a potential mediator influencing the risk of AA, accounting for 10.64% of the effect. We found robust genetic evidence for a causal association between DM and AA risk, and rest CD4+ Treg absolute count, might mediate this effect. However, the potential implications of our findings for AA prevention require validation through well-powered randomized clinical trials.