Association of lean NAFLD with carotid plaque prevalence and incident diabetes: separate analyses in two Asian populations

Introduction The clinical heterogeneity of non-alcoholic fatty liver disease (NAFLD) across BMI strata warrants further investigation. This study examined the cross-sectional association between lean NAFLD and carotid plaque (CP) in a Chinese health check-up population, and the prospective association between lean NAFLD and incident diabetes in a Japanese cohort. Methods The study comprised a Chinese cross-sectional analysis ( n = 12,283) and a Japanese prospective cohort analysis ( n = 14,271). In both analyses, participants were classified into four groups: G1 (BMI ≤ 23 kg/m 2 , without NAFLD), G2 (BMI > 23 kg/m 2 , without NAFLD), G3 (BMI ≤ 23 kg/m 2 , with NAFLD; i.e., lean NAFLD), and G4 (BMI > 23 kg/m 2 , with NAFLD; i.e., non-lean NAFLD). Multivariable logistic regression was used to examine the prevalence of CP and of hypoechoic or heteroechoic carotid plaques, and Cox proportional hazards models were used to estimate incident diabetes risk. Additive interaction was assessed using the relative excess risk due to interaction (RERI), attributable proportion (AP), and synergy index (SI). Results In the cross-sectional analysis, G3 (OR = 1.596, 95% CI 1.282–1.983, p < 0.001) and G4 (OR = 1.461, 95% CI 1.276–1.672, p < 0.001) were significantly associated with higher odds of CP relative to G1. Among participants with CP, the proportion of hypoechoic or heteroechoic carotid plaques was highest in G4 and was significantly higher than that in G3 (adjusted p = 0.018). In the prospective analysis, both G3 (HR = 5.637, 95% CI 3.626–8.762) and G4 (HR = 8.133, 95% CI 5.970–11.078) showed significantly elevated diabetes risk versus G1 ( p < 0.001); risk did not differ significantly between G3 and G4 ( p > 0.05). Additive interaction analysis showed no significant interaction between BMI and NAFLD for CP ( p > 0.05). Conclusion No statistically significant difference was detected between lean NAFLD and non-lean NAFLD regarding their associations with carotid plaque prevalence and incident diabetes risk. Notably, this non-significant finding does not indicate equivalence of risk magnitudes between the two groups, and the relatively small number of diabetes events in the lean NAFLD group may have limited the statistical power to detect potential group differences.