Sex differences in the association between fatty liver index and biological aging: A mediation analysis of insulin resistance in a cross-sectional study

Aging challenges global healthcare systems, with fatty livers potentially relevant to the process. This study sought to investigate the sex-specific associations of fatty liver index (FLI) with biological aging (BA), assessed via phenotypic age (PhenoAge) and PhenoAge acceleration (PhenoAgeAccel), and explored the mediating role of the metabolic score for insulin resistance (METS-IR). Utilizing nationally representative data from 16,479 American adults in the National Health and Nutrition Examination Survey (NHANES 1999-2020), this cross-sectional study employed weighted multivariable regression, receiver operating characteristic curves, restricted cubic splines, threshold effect analyses, sensitivity analyses, and mediation analyses. FLI was positively correlated with BA overall (P < .001), with significant sex interaction effects observed (Pinteraction < .05). For every 1 - standard deviation increase in FLI, females exhibited stronger associations with PhenoAge (βfemale = 3.33, 95% confidence interval [CI]: 3.07-3.58; βmale = 2.54, 95% CI: 2.20-2.89) and PhenoAgeAccel risk (ORfemale = 1.89, 95% CI: 1.76-2.03; ORmale = 1.57, 95% CI: 1.46-1.69; all P < .001). The area under the curve of FLI for PhenoAgeAccel was higher in females (0.73, 95% CI: 0.72-0.75) than in males (0.67, 95% CI: 0.66-0.68). Restricted cubic splines revealed a biphasic upward trend in the FLI-BA relationship across both sexes (Pnonlinear < .001), with sex-divergent threshold effects (Plikelihood ratio < .001). METS-IR partially mediated the FLI-BA association, with a greater percentage of mediation effect in males (PhenoAge: 20.23%; PhenoAgeAccel: 36.84%) than females (8.73%; 15.93%). A robust positive association exists between FLI and BA, partially mediated by METS-IR. Females exhibited stronger association strength, while males demonstrated a greater proportion of the METS-IR's mediation effect.