Xiao‐Meng Wang, Hao Yu Yan, Wen‐Fang Zhong, Jia‐Hao Xie, Huan Chen, Junjie Wang, Wei‐qi Song, Dong Shen, P. Zhang, Xi Zhang, Jiao‐Jiao Ren, Dan Liu, Zhi‐Hao Li, Chen Mao
AIMS: Longitudinal evidence is limited on how changes in insulin resistance (IR) indices-including estimated glucose disposal rate (eGDR), triglyceride-glucose index (TyG), metabolic score for insulin resistance (METS-IR), and lipid accumulation product (LAP)-relate to liver-related adverse outcomes. This study aims to assess their associations and the discriminative performance of IR indices. MATERIALS AND METHODS: IR indices were calculated from UK Biobank data at two surveys (2006-2010 and 2012-2013). Liver-related adverse outcomes, including liver disease, major adverse liver outcomes (MALO), and metabolic dysfunction-associated steatotic liver disease (MASLD), were identified via ICD-10 codes. K-means clustering defined four change patterns per index, and cumulative averages reflected long-term exposure. Cox regression estimated hazard ratios (HRs) and 95% confidence intervals (CIs). Discriminative performance was assessed using receiver operating characteristic (ROC) curves. RESULTS: The participants were followed for a mean of 9.7 years. Compared with persistently low eGDR levels, the persistently high group was associated with significantly lower risks of liver-related adverse outcomes, with HRs of 0.52 (95% Cl: 0.35-0.77) for liver disease, 0.30 (0.19-0.49) for MALO, and 0.31 (0.17-0.55) for MASLD. In contrast, persistently high TyG, METS-IR, and LAP were associated with increased risks of liver-related adverse outcomes, with METS-IR showing the strongest association with MASLD (HR = 10.50, 4.00-27.58). Cumulative eGDR was inversely associated with liver-related adverse outcomes (per 1 SD increase: HRs ranged from 0.52 to 0.68), whereas TyG, METS-IR, and LAP were positively associated, with METS-IR showing the strongest link to MASLD (HR = 1.70, 1.48-1.96). LAP demonstrated the highest discriminative performance in ROC analysis, particularly in females and those under 60 (AUC for MALO in females: up to 0.813). CONCLUSIONS: Dynamic changes in IR indices are independently associated with liver-related adverse outcomes. Among these indices, LAP showed relatively stronger discriminative performance in females. Collectively, these indices may have potential utility as non-invasive markers for liver disease risk stratification.