Peripheral artery disease (PAD) continues to increase worldwide, yet early risk stratification remains difficult because traditional obesity measures provide limited metabolic insight. A study published in Cardiovascular Diabetology evaluated whether the triglyceride-glucose-a body shape index (TyG-ABSI), which integrates the triglyceride-glucose (TyG) index and a body shape index (ABSI), predicts long-term PAD risk.
The analysis included 390,274 adults from the UK Biobank. Participants were categorized by TyG-ABSI quartiles and evaluated for incident PAD. Associations between TyG-related indices and PAD were assessed using multivariable-adjusted Cox regression, Kaplan-Meier survival analysis, and restricted cubic spline models. Robustness analyses included competing risk models, propensity score matching, subgroup analyses, and external validation using the National Health and Nutrition Examination Survey (NHANES). Consensus k-means clustering identified metabolic phenotypes, and feature selection methods, including least absolute shrinkage and selection operator, Boruta, and minimum redundancy maximum relevance, identified key predictors for PAD risk models.
Higher TyG-ABSI levels were associated with greater PAD incidence. The 15-year cumulative PAD incidence was 4.16% in the highest TyG-ABSI quartile compared with 0.98% in the lowest quartile. Each 1-standard deviation increase in TyG-ABSI was associated with higher PAD risk (hazard ratio 1.22; 95% confidence interval 1.17-1.27). Clustering analysis identified four metabolic phenotypes, with the highest PAD risk observed in the insulin resistance and glucose dysfunction subgroup (hazard ratio 7.48; 95% confidence interval 6.82-8.21 vs healthy phenotype). Logistic regression showed the most stable predictive performance with a validation area under the curve of 0.788 (95% confidence interval 0.778-0.798).
These findings indicate that TyG-ABSI independently predicts long-term PAD risk. Data-driven phenotyping and interpretable machine learning approaches may support more precise PAD risk stratification and individualized risk prediction.
