Elevations in markers of liver injury and risk of type 2 diabetes: the insulin resistance atherosclerosis study

Abstract

A limited number of studies have reported associations of markers of liver injury, including elevated concentrations of aspartate aminotransferase (AST) and alanine aminotransferase (ALT), with prospective risk of type 2 diabetes. However, only one study has adjusted for a detailed measure of insulin sensitivity (insulin sensitivity index [S(i)]), which is important given associations of obesity and S(i) with nonalcoholic fatty liver disease (NAFLD). Our objective was to investigate the associations of elevated AST and ALT with incident type 2 diabetes among 906 participants in the Insulin Resistance Atherosclerosis Study who were nondiabetic at baseline. S(i) and acute insulin response (AIR) were measured directly from the frequently sampled intravenous glucose tolerance test among black, Hispanic, and non-Hispanic white participants aged 40-69 years. After 5.2 years, 148 individuals had developed type 2 diabetes. Baseline AST and ALT were positively correlated with fasting insulin (r = 0.22 and r = 0.35, respectively), waist circumference (r = 0.18 and r = 0.34), and fasting glucose (r = 0.13 and r = 0.29) and inversely with S(i) (r = -0.18 and r = -0.30; all P < 0.0001). In separate logistic regression models adjusting for age, sex, ethnicity, clinical center, and alcohol consumption, participants in the highest quartiles (Q4) of AST and ALT were at significantly increased risk of incident type 2 diabetes compared with those in the lowest three quartiles (Q1-Q3): AST: odds ratio (OR) 1.73 (95% CI 1.17-2.57); ALT: OR 2.32 (1.36-3.75). After further adjustment for smoking, waist circumference, triglyceride, HDL, impaired glucose tolerance, S(i), and AIR, both AST and ALT remained significantly associated with incident type 2 diabetes: AST, Q4 vs. Q1-Q3: OR 1.98 (1.23-3.17); ALT, Q4 vs. Q1-Q3: OR 2.00 (1.22-3.28). There were no interactions of sex, ethnicity, obesity, impaired glucose tolerance, or S(i) with AST or ALT in the prediction of type 2 diabetes. When entered into the same model with adjustment for demographic variables, both C-reactive protein and ALT independently predicted type 2 diabetes. In addition, AST and ALT were positively associated with incident type 2 diabetes after excluding former and moderate to heavy drinkers. In conclusion, AST and ALT independently predict type 2 diabetes. Baseline elevations of these markers may reflect NAFLD or related pathologies.