We found that clear differences in AST, ALT, and AST/ALT ratio curves between age- and sex- different subgroups; the AST levels tended to decrease as age, but the ALT curves were U-shaped, which resulted in trends of the AST/ALT ratio to reduce according to age and skewed to the left. In the cardiometabolic risk assessment based on our novel references, components of MetS significantly increased in high levels of AST, ALT, and AST/ALT ratio; some of the components were increased even in borderline high levels of liver enzymes in the analysis of ANCOVA with correction of age, sex, BMI, etc. Similarly, the odds ratio of metabolic risk factors also increased in the high AST, ALT, and the low AST/ALT ratio; some components of MetS were significantly increased even in borderline high levels.
In our findings, AST and ALT were higher in boys than in girls and differ over time; AST tended to decrease with age but ALT levels were U-shaped and increased with age. Those tendencies were consistent with previous reports in Korean subjects 18 and those of other ethnicities, for instance, an obvious increase in ALT levels was observed in male after the age of 11 years 17, and a sex difference with a continuous decrease in concentration from ages 2 to 14 years was observed for AST 9,17,20,21. Percentile distributions of the AST/ALT ratio showed sex differences across all observed ages in our data; the ratio was significantly higher in girls than in boys. Despite conflicts with the data of the HELENA study 10, our results were similar to those of a recent Chinese study 22, which seems to be due to ethnic or geographic differences.
Regarding on MetS, an increase in the prevalence of MetS according to AST, ALT, and the AST/ALT ratio was more clearly observed as BMI increased (Figs. 3 and 4). This result suggested that although a strong correlation between high levels of AST, ALT, and the AST/ALT ratio and most cardiometabolic risk factors exist, the interpretation of liver markers in children and adolescents could be adapted differentially depending on the differential BMI values. Previously, AST or AST/ALT ratios were known to indicate progression to diabetes 23. Similarly, in our findings, high serum AST, ALT, and the AST/ALT ratio were all associated with a high OR of elevated glucose (Table 4). These results strongly supported that a decreased AST/ALT ratio might effectively predict increased cardiometabolic risk, especially insulin resistance when high ALT exists. Of interest, our results showed that the associations of AST, ALT, and the AST/ALT ratio with cardiometabolic risk factors were significant both in boys and girls, which partially conflicts with a previous European study 10. The differential distribution of overweight or obese children in the study population might affect the analysis. In addition, differential body adiposity might also influence screening tests of liver markers. Therefore, detailed comparative studies on this point will be needed in future studies.
Based on the percentile distribution, we suggested that the upper limits of normal of AST, ALT and the AST/ALT ratio were 23 IU/L and 20 IU/L (< 75th percentile), 19 IU/L and 14 IU/L (< 75th percentile), and 1.1 and 1.35 (< 25 percentile) in boys and girls, respectively. Similar studies with non-overweight adolescents proposed sex-specific thresholds for ALT levels < 25 IU/L in males and < 22 IU/L in females to detect pediatric chronic liver disease 8. However, Labayen et al. suggested upper limits of normal for ALT of 24–25 (75th percentile) and 22–24 IU/L (75th percentile) and thresholds of the AST/ALT ratio associated with high cardiometabolic risk of 1.0-0.74 and 0.86–0.87 (ranging from 13-35th percentile) for boys and girls, respectively 10. In those studies, the estimated upper limits of normal of ALT in adolescents were higher than those in our report. This might be due to differences in ethnicities between the studies; other possibilities are differences in the proportion of subjects with obesity and severity of central adiposity levels. Although direct comparison of the AST/ALT ratio between our data and previous reports is difficult, it seems that stricter levels of the AST and ALT should be applied for Korean adolescents, especially for the overweight or obese subgroup for the precise estimation of cardiometabolic prognosis.
The main limitation of this study is the cross-sectional analysis, which cannot identify the temporal association of MetS with AST, ALT, and the AST/ALT ratio. A large population-based, longitudinal cohort study could address this limitation in the future by serial measurements of liver enzymes and follow-up for the occurrence of cardiometabolic events. The other limitation is that our data were from subjects of one ethnicity in a single country. Thus, comparisons and meta-analysis with other ethnic groups will be needed for the broad application of pediatric reference intervals. Despite of information about a family history of premature coronary heart disease, we could not exclude familial hyperlipidemia during subject selection due to limitations on laboratory tests. Combined familial hyperlipidemia frequently accompanies NAFLD in approximately 49–76% of cases 24, which implies that possible effects of these comorbidities were not completely excluded in our data. Other liver markers such as gamma-GT have also been suggested to be strong predictors of cardiovascular disease and T2DM in adults 25, and metabolic risks in adolescents 10; however, we did not analyze other possible markers in our current study.
In conclusion, we newly established reference values for AST, ALT, and the AST/ALT ratio based on the risk assessment of MetS components. High levels of AST and ALT and a low AST/ALT ratio were closely associated with the prevalence of MetS and its components. In particular, overweight and obese children and adolescents have a considerably higher prevalence of MetS and its components when liver enzymes exceed the upper limits of normal than do normal subjects. Both ALT and the AST/ALT ratio were effective in screening for metabolic risk in both sexes in a Korean population. Therefore, the age- and sex-specific reference values provided in this study may contribute to the early diagnosis and treatment of MetS.