Increment of serum bilirubin as an independent marker predicting new-onset type 2 diabetes mellitus in a Korean population

Background and aims Several cross-sectional studies reported that serum bilirubin concentrations had an inverse association with type 2 diabetes mellitus (T2DM) prevalence. The aim of the current study was to investigate the relationship between percentage change in bilirubin levels (PCB) and incident risk of T2DM using a longitudinal model. Methods and results 22,084 participants who received regular health check-ups between 2006 and 2012 were enrolled. Multivariable-adjusted Cox regression models were used to determine the hazard ratio (HR) of incident T2DM based on PCB. PCB was determined by subtracting baseline serum bilirubin level (BB) from the bilirubin level at the end of follow-up or a year before the last date of diagnosis, dividing by BB and multiplying by 100. Compared to non-diabetics, BB was lower in the diabetic group at the initial visit. There were 20,098 participants without T2DM at the initial visit; 1253 new cases occurred during follow-up. As PCB increased, T2DM incidence also increased (P < 0.001). After adjusting for confounders, the HR of incident T2DM in the highest PCB quartile was 2.08 (95% confidence interval [CI] 1.76–2.46). This trend remained significant when PCB was analyzed as a continuous variable (HR for 1-SD increment, 1.25; 95% CI 1.19–1.31). Additional analysis comparing the rate of PCB during the follow-up period revealed that the serum bilirubin level of the Incident T2DM group increased before T2DM development and decreased rapidly thereafter compared to others (P < 0.001). Conclusions Bilirubin level increment over time is associated with T2DM development.