Association between TyG index trajectory and new-onset lean NAFLD: a longitudinal study

Area of Science:

• Metabolic Syndrome Research
• Liver Disease Epidemiology
• Biomarker Discovery

Background:

• Nonalcoholic fatty liver disease (NAFLD) is a growing global health concern.
• Lean NAFLD, occurring in individuals without obesity, presents unique diagnostic challenges.
• Identifying reliable biomarkers for early risk assessment is crucial for timely intervention.

Purpose of the Study:

• To identify longitudinal trajectories of the triglyceride glucose (TyG) index.
• To investigate the association between these TyG index trajectories and the risk of developing lean NAFLD.
• To develop a predictive model for lean NAFLD using machine learning.

Main Methods:

• Latent Class Growth Modeling (LCGM) was used to define TyG index trajectories in 1,109 participants.
• Cox proportional hazard models and restricted cubic splines (RCS) analyzed the association with incident lean NAFLD.
• Light Gradient Boosting Machine (LightGBM) was employed for predictive modeling and an online risk assessment tool was created.

Main Results:

• Three distinct TyG index trajectories were identified: low stable, moderate stable, and high stable.
• A high stable TyG index trajectory was significantly associated with an increased risk of lean NAFLD (HR: 2.668).
• A nonlinear dose-response relationship was observed between TyG index and lean NAFLD risk; LightGBM achieved high predictive accuracy (Test AUC 0.766).

Conclusions:

• The triglyceride glucose (TyG) index is a promising noninvasive biomarker for lean NAFLD.
• Specific longitudinal TyG index trajectories correlate with varying risks of lean NAFLD.
• The developed machine learning model and online tool can aid clinical risk assessment for lean NAFLD.