Associations between 12 insulin resistance surrogates with metabolic dysfunction-associated steatotic liver disease...

Metabolic dysfunction-associated steatotic liver disease (MASLD) has been shown to be intimately linked to the presence of insulin resistance. This study aimed to comprehensively evaluate 12 insulin resistance surrogates in relation to MASLD risk and all-cause mortality, utilizing nationwide data from the National Health and Nutrition Examination Survey (NHANES) III (1988-1994).

This study analyzed 10 303 adults aged greater than or equal to 20 years from the NHANES III (1988-1994) cohort, identifying 2199 individuals with MASLD. The mortality data for this cohort were collected from the National Death Index. Twelve surrogate markers of insulin resistance were evaluated, including triglyceride-glucose (TyG) index, TyG-BMI, TyG-waist circumference, TyG-waist-to-height ratio, C-reactive protein-triglyceride-glucose index, atherogenic index of plasma (AIP), AIP-BMI, AIP-waist circumference, AIP-waist-to-height ratio, estimated glucose disposal rate (eGDR), metabolic score for insulin resistance, and homeostatic model assessment of insulin resistance (HOMA-IR). Statistical analyses employed logistic regression and Cox proportional hazards models to assess associations. Additionally, restricted cubic splines (RCSs) and Kaplan-Meier curves were utilized alongside subgroup and sensitivity analyses. Predictive performance was examined using receiver operating characteristic analysis and machine learning (XGBoost).

Among 10 303 participants, MASLD prevalence was 21.3%. In models that have undergone full adjustment, eGDR was negatively correlated with the risk of MASLD [odds ratio = 0.827, 95% confidence interval (CI): 0.780-0.878], while all other indices showed positive associations. In 2199 MASLD patients with 1015 deaths during follow-up, TyG-related indices, C-reactive protein-triglyceride-glucose index, and HOMA-IR were significant correlates with higher all-cause mortality, whereas eGDR was inversely correlated (hazard ratio = 0.887, 95% CI: 0.844-0.933). Two sensitive analyses further supported the overall results in the overall models. The RCS curve exhibited nonlinear dose-response relationships for several indices. XGBoost analyses identified eGDR as the strongest predictor of mortality among insulin resistance surrogates, second only to age.

Most insulin resistance surrogates were significantly associated with both MASLD risk and mortality, while eGDR emerged as a robust protective factor with superior predictive performance. These results emphasize the pivotal role of insulin resistance in MASLD and highlight eGDR as a promising noninvasive tool for stratifying risk and predicting adverse outcomes in clinical and public health settings.