Heterogeneity of diabetes as a risk factor for major adverse cardiovascular events in anticoagulated patients with atrial fibrillation: an analysis of the ARISTOTLE trial

Aims Whether diabetes without insulin therapy is an independent cardiovascular (CV) risk factor in atrial fibrillation (AF) has recently been questioned. We investigated the prognostic relevance of diabetes with or without insulin treatment in patients in the ARISTOTLE trial. Methods and results Patients with AF and increased stroke risk randomized to apixaban vs. warfarin were classified according to diabetes status: no diabetes; diabetes on no diabetes medications; diabetes on non-insulin antidiabetic drugs only; or insulin-treated. The associations between such patient subgroups and stroke/systemic embolism (SE), myocardial infarction (MI), and CV death were examined by Cox proportional hazard regression, both unadjusted and adjusted for other prognostic variables. Patients with diabetes were younger and had a higher body mass index. Median CHA(2)DS(2)VASc score was 4.0 in patients with diabetes and 3.0 in patients without diabetes. We found no significant difference in stroke/SE incidence across patient subgroups. Compared with no diabetes, only insulin-treated diabetes was significantly associated with higher risk. When adjusted for clinical variables, compared with no diabetes, the hazard ratios (HRs) for MI (95% confidence intervals) were for diabetes on no medication: 1.15 (0.62-2.14); for diabetes on non-insulin antidiabetic drugs: 1.32 (0.90-1.94); for insulin-treated diabetes: 2.34 (1.43-3.82); interaction P = 0.008. HRs for CV death were for diabetes on no medication: 1.19 (0.86-166); for diabetes on non-insulin antidiabetic drugs: 1.12 (0.88-1.42); for insulin-treated diabetes 1.85 (1.36-2.53), interaction P = 0.001. Conclusion In anticoagulated patients with AF, a higher risk of MI and CV death is largely confined to diabetes treated with insulin.