Joint association of estimated glucose disposal rate and systemic inflammation response index with mortality in metabolic-dysfunction associated steatotic liver disease: evidence from two cohort studies

BACKGROUND AND AIMS: Mortality in metabolic dysfunction-associated steatotic liver disease (MASLD) is primarily driven by the synergy between insulin resistance (IR) and systemic inflammation. However, practical tools for integrated risk assessment remain scarce. This study aimed to evaluate the individual and joint prognostic value of the estimated Glucose Disposal Rate (eGDR) and Systemic Inflammation Response Index (SIRI)-validated surrogates for IR and inflammatory status-to refine risk stratification and elucidate their reciprocal associations on MASLD survival. METHODS: This observational analysis included 7520 MASLD adults from the continuous National Health and Nutrition Examination Survey (NHANES) (1999-2018) and an independent external validation cohort of 1182 ultrasound-confirmed patients from NHANES III (1988-1994), with mortality linked through 2019. Multivariable Cox models and restricted cubic splines were employed to evaluate the associations between biomarkers and mortality. The combined model's predictive performance and clinical benefit were quantified via Receiver Operating Characteristic curves, Net Reclassification Improvement (NRI), Integrated Discrimination Improvement (IDI), and Decision Curve Analysis. Furthermore, exploratory bidirectional mediation analysis was conducted to assess the potential statistical interplay between the two markers. RESULTS: Over a median follow-up of 138 months, 1375 all-cause and 442 cardiovascular deaths occurred. Lower eGDR and higher SIRI were independently and linearly associated with increased mortality. A low eGDR/high SIRI phenotype was identified as the highest-risk category, exhibiting a 1.860-fold risk of all-cause mortality (95% CI 1.439-2.405) and a 2.395-fold risk of cardiovascular mortality (95% CI 1.379-4.159). The combined model demonstrated superior predictive accuracy (Area Under the Curve (AUC) 0.686-0.848) and significant reclassification improvement (NRI 0.161; IDI 0.017), and higher clinical net benefit than either indicator alone (p < 0.001). Furthermore, mediation analysis suggested that SIRI statistically accounted for 10.31% of the association between eGDR and all-cause mortality, highlighting a potential reciprocal statistical interplay. Crucially, this high-risk 'low eGDR/high SIRI' phenotype was successfully validated in the imaging-confirmed external cohort (ACM: HR = 2.354; CVM: HR = 3.153; both p < 0.001). CONCLUSIONS: Integrating eGDR and SIRI identifies a high-risk MASLD phenotype with the poorest prognosis, reflecting a synergistic metabolic-inflammatory burden. This joint assessment significantly improves predictive accuracy and offers superior net clinical benefits for long-term mortality prediction.