Joint effects of phenotypic age acceleration and novel insulin resistance indicators on mortality in adults with metabolic dysfunction-associated steatotic liver disease: a prospective cohort study

MASLD is a heterogeneous metabolic disorder with substantial risks of all-cause mortality (ACM) and cardiovascular mortality (CVM). Although insulin resistance (IR) and accelerated biological aging are both linked to adverse outcomes, their joint prognostic value in MASLD remains unclear. This study evaluated the independent, joint, and incremental prognostic value of PhenoAgeAccel and three IR-related indicators, including TyG-ABSI, TyG-WHtR, and eGDR, for ACM and CVM. This prospective cohort study included 6,718 adults with FLI-defined presumed MASLD from NHANES 1999–2018, using FLI ≥ 60 as a non-invasive surrogate for hepatic steatosis rather than imaging, elastography, or histology. ACM was prespecified as the primary outcome and CVM as the secondary outcome. Independent and joint associations of PhenoAgeAccel, TyG-ABSI, TyG-WHtR, and eGDR with mortality were estimated in adjusted Cox proportional hazards models and further characterized using restricted cubic splines, with Fine-Gray competing-risk models accounting for non-cardiovascular deaths in CVM analyses. Incremental prognostic performance was assessed by time-dependent ROC curves, C-index, NRI, IDI, and decision curve analysis. Exploratory bidirectional indirect association analysis evaluated potential metabolic-aging links, while additive and multiplicative interaction analyses examined whether joint exposures conferred excess risk beyond individual associations. During a median follow-up of 148 months, 1,245 ACM and 401 CVM events occurred. In the fully adjusted model, higher PhenoAgeAccel, TyG-ABSI, and TyG-WHtR were associated with increased mortality risk, whereas higher eGDR was associated with lower ACM and CVM. Joint phenotype analyses showed that participants with both PhenoAgeAccel and high-risk IR-related profiles consistently had the greatest mortality risk. In particular, the PhenoAgeAccel > 0 and high TyG-ABSI phenotype showed the highest risk of ACM (HR = 2.234, 95% CI 1.740–2.869) and CVM (HR = 3.141, 95% CI 1.961–5.030). Joint models showed incremental gains in time-dependent AUC over the clinical baseline model, with supportive reclassification and decision-curve findings. Exploratory indirect association analyses suggested reciprocal links between PhenoAgeAccel and IR-related indicators. Exploratory interaction and absolute-risk analyses further supported elevated joint risk among dual high-risk phenotypes for both outcomes. PhenoAgeAccel and IR-related indicators were individually and jointly associated with ACM and CVM in adults with MASLD. Their combined assessment identified a dual high-risk phenotype and provided additional prognostic information beyond individual indicators. These findings provide hypothesis-generating support for external evaluation of a metabolic-aging approach to mortality risk stratification in MASLD.