Figure 3.

The effects of exemplar miRNAs on time to onset of ESKD were mediated by circulating levels of EPHA2. Shown are the effects of each exemplar miRNA (exposure) from each cluster on risk of ESKD during 10-year follow-up (outcome) mediated by EPHA2 (mediator). Combined T1D discovery and T2D replication cohorts were used in these analyses. The analyses were based on Cox proportional hazards models for time to onset of ESKD, adjusted for sex, duration of diabetes, systolic BP, baseline HbA1c, and eGFR, with variable stratification by study cohort. The effect of an miRNA on the time to ESKD (total effect) is split into a natural indirect effect (red arrow), measured as percent mediated (PM) by EPHA2, which exerts its role on the outcome through modulation of a protein level (mediator) and natural direct effect (blue arrow), which acts on the outcome independently from the mediator. Results of the mediation analysis for EFNA4 are provided in Supplemental Table 9.