Triglyceride glucose index as a surrogate measure of insulin sensitivity in obese adolescents with normoglycemia, prediabetes, and type 2 diabetes mellitus: comparison with the hyperinsulinemic–euglycemic clamp

    loading  Checking for direct PDF access through Ovid



There is a need for simple surrogate estimates of insulin sensitivity in epidemiological studies of obese youth because the hyperinsulinemic–euglycemic clamp is not feasible on a large scale.


(i) To examine the triglyceride glucose (TyG) index (Ln[fasting triglycerides (mg/dL) × fasting glucose (mg/dL)/2]) and its relationship to in vivo insulin sensitivity in obese adolescents (OB) along the spectrum of glucose tolerance and (ii) to compare TyG index with triglyceride/high-density lipoprotein TG/HDL and 1/fasting insulin (1/IF), other surrogates of insulin sensitivity.

Patients and design:

Cross-sectional data in 225 OB with normal glucose tolerance (NGT), prediabetes (preDM), and type 2 diabetes (T2DM) who had a 3-h hyperinsulinemic–euglycemic clamp and fasting lipid measurement.


Insulin-stimulated glucose disposal (Rd) declined significantly across the glycemic groups from OB-NGT to OB-preDM to OB-T2DM with a corresponding increase in TyG index (8.3 ± 0.5, 8.6 ± 0.5, 8.9 ± 0.6, p < 0.0001). The correlation of TyG index to Rd was −0.419 (p < 0.0001). The optimal TyG index for diagnosis of insulin resistance was 8.52 [receiver operating characteristic-area under the ROC curves (ROC-AUC) 0.750, p < 0.0001]. The ROC-AUC for 1/IF was 0.836. In multiple regression analysis, 64.8% of the variance in Rd was explained by TyG index, 1/IF, body mass index (BMI) z-score, glycemic group, and sex.


The TyG index affords an easily and widely available simple laboratory method as a surrogate estimate of insulin sensitivity that could be used repeatedly in large-scale observational and/or interventional cohorts of OB. Although not superior to 1/IF, TyG index offers the advantage of having a standardized method of measuring triglyceride and glucose, which is not the case for insulin assays.

Related Topics

    loading  Loading Related Articles