Abstract
BACKGROUND: Evidence remains insufficient to clarify whether hyperglycemia or hyperlipidemia exerts a greater influence on the development of liver steatosis and fibrosis. Furthermore, it remains uncertain whether distinct glucose-lipid metabolic profiles are associated with an increased risk of these hepatic conditions. METHODS: We used NHANES data to assess liver steatosis and fibrosis via Controlled Attenuation Parameter (CAP) and Liver Stiffness Measurement (LSM), then propensity-score weighted regression was used to access associations between seven glycolipid metabolic biomarkers and the two liver indices. Radar plots and network analysis were utilized to illustrate metabolic distribution patterns and delineate key metabolic signatures distinguishing four liver health phenotypes: Healthy Control, Steatotic Liver, Fibrotic Liver, and Combined Steatotic-Fibrotic Liver. Linear regression analyses were further conducted to compare the levels of steatosis and fibrosis across eight metabolic subgroups, categorized according to distinct combinations of dyslipidemia, dysglycemia, and insulin resistance. Lastly, a tree-based algorithm was employed to identify distinct glycolipid metabolic profiles associated with increased severity of hepatic steatosis and fibrosis. RESULTS: A total of 9698 individuals were included (mean age: 44.26[Formula: see text]20.80 years). Propensity score-weighted regression showed that TG, GLU, HbA1c, and HDL were significantly associated with both CAP and LSM, while TC and LDL were associated with CAP only. Radar plots and network analysis revealed that the steatosis-fibrosis group had the most adverse metabolic profile, with the lowest HDL, highest insulin, HbA1c, and GLU levels. TG and GLU consistently ranked as the top two central metabolic hubs across all liver groups, while HbA1c ranked third except in the steatotic-fibrotic group, where insulin resistance prevailed. Among subgroups with a single metabolic abnormality, CAP was highest in the insulin resistance (IR)-only group (B = 14.433, P < 0.001), followed by the dysglycemia-only group (B = 10.142, P < 0.001), and lowest in the dyslipidemia-only group. In IR-associated subgroups, CAP was significantly higher when IR co-occurred with dysglycemia (B = 57.393, P < 0.001) than with dyslipidemia (B = 53.635, P < 0.001). The triple abnormalities group exhibited the highest overall CAP (B = 79.811, P < 0.001) as well as the highest liver stiffness measurement (LSM) (B = 1.543, P < 0.001). Tree-based analysis further identified that CAP was highest in individuals with insulin ≥ 14.705 µU/mL and GLU ≥ 5.861 mmol/L, while LSM was highest in those with insulin between 17.1 and 28.205 µU/mL, HbA1c ≥ 6.15%, and GLU ≥ 7.861 mmol/L. CONCLUSIONS: Metabolic abnormalities related to hyperglycemia may be more closely associated with hepatic steatosis and fibrosis than those related to hyperlipidemia.