Serum bilirubin improves the risk predictions of cardiovascular and total death in diabetic patients

Background: Bilirubin is a potential endogenous inhibitor of atherosclerosis. We investigated the association of bilirubin and cardiovascular (CV) and all-cause mortality including potential improvements in bilirubin risk reclassification in asymptomatic diabetic patients. Methods: We enrolled 2936 asymptomatic diabetic subjects. The serum bilirubin was measured, and future CV and all-cause death were the primary endpoints. Results: The follow-up period was 5.4 +/- 3.0 y. There were 218 deaths including 95 cardiovascular deaths. The occurrence of CV death and all-cause death were negatively correlated with increasing serum bilirubin quintiles and actual bilirubin values. Serum bilirubin was negatively associated with incident cardiovascular death (hazard ratio: 0.26, 95% CI, 0.11-0.61, p = .01) and all-cause death (hazard ratio: 0.30, 95% CI, 0.17-0.51, p <= .001). The addition of bilirubin for cardiovascular death increased the C-statistic from 0.713 (95% CI, 0.664-0.762) to 0.729 (95% CI, 0.681-0.776) (P = .008) and showed an integrated discrimination improvement (IDI) of 0.012 (P < .0171) with 8.57% improvement in net reclassification analysis (P = .0224). These results suggest additional predictive value is possible via total bilirubin levels for future CV deaths in diabetic patients. In terms of all-death, the addition of bilirubin significantly increased the C-statistic (from 0.769 to 0.78, P = .0064)-a 3.52% net reclassification improvement (P = .0307). It did not improve the IDI (p = .1505). Conclusions: Higher serum concentrations of bilirubin are associated with a decreased risk of developing CV and all-cause death in diabetic patients. Bilirubin improved the risk prediction of cardiovascular death but provided only a slightly better prediction of all-cause death than conventional risk factors.