Inverse linear associations between liver aminotransferases and incident cardiovascular disease risk: The PREVEND study

- ALT and AST are each approximately log-linearly associated with CVD risk.
- The associations are inverse and independent of conventional CVD risk factors.
- The associations hold true across the normal ranges of the aminotransferases.
- ALT or AST is unlikely to improve CVD risk assessment in the general population.

BackgroundAlanine aminotransferase (ALT) and aspartate aminotransferase (AST) have been linked with an increased risk of type 2 diabetes, but their relationships with cardiovascular disease (CVD) are uncertain. We aimed to assess the associations of ALT and AST with CVD risk and determine their potential utility for CVD risk prediction.MethodsALT and AST measurements were made at baseline in the PREVEND prospective cohort involving 6899 participants aged 28-75 years without pre-existing CVD.ResultsDuring 10.5 years of follow-up, 729 CVD events were recorded. Serum aminotransferases were strongly correlated with each other and each weakly correlated with several cardiovascular risk markers. ALT and AST were each approximately log-linearly associated with CVD risk. In analyses adjusted for conventional risk factors, the hazard ratios (95% CIs) for CVD per 1 standard deviation increase in loge ALT and loge AST were 0.87 (0.79-0.94; P = 0.001) and 0.91 (0.84-0.98; P = 0.017) respectively. The associations remained consistent after additional adjustment for several potential confounders including alcohol consumption, fasting glucose, and C-reactive protein, with corresponding hazard ratios of 0.88 (0.80-0.96; P = 0.003) and 0.92 (0.84-0.99; P = 0.029). The inverse associations persisted within normal ranges of the aminotransferases. Adding ALT or AST to a CVD risk prediction model containing established risk factors did not improve the C-index or net reclassification.ConclusionsAvailable data suggest the liver aminotransferases are each inversely, independently, and approximately log-linearly associated with CVD risk. Nonetheless, they provide no significant improvement in CVD risk assessment beyond conventional CVD risk factors.