Insulin Resistance Is Better Estimated by Using Fasting Glucose,Lipid Profile,and Body Fat Percent Than by HOMA-IR in Japanese Patients with Type 2 Diabetes and Impaired Glucose Tolerance: An Exploratory Study

Abstract

Aims:

The aim of the present study is to estimate insulin resistance (IR) using clinically available parameters except for serum insulin or C-peptide concentration to overcome the limitation of homeostasis model assessment of IR (HOMA-IR), which has been widely used in clinical practice.

Patients and Methods:

Fifty-two admitted patients with type 2 diabetes or impaired glucose tolerance were enrolled, and steady state plasma glucose (SSPG) method and cookie meal tolerance test were performed together with fasting blood sampling and anthropometric measurements. Insulin sensitivity measured by SSPG was estimated as glucose clearance corrected by the excretion of glucose into urine (C-GC).

Results:

Log-transformed (C-GC) was negatively correlated with fasting plasma glucose (FPG), log (Fasting triglyceride: TG), log (Fasting TG/Fasting high-density lipoprotein cholesterol: HDLC), and their area under the curves (AUCs). Fasting and AUC-HDLC was positively and fasting free fatty acid (FFA) was negatively correlated with log (C-GC). Body fat (%) was negatively correlated with log (C-GC). Multiple regression analysis on log (C-GC) as an outcome variable revealed that FPG, log (AUC-TG/AUC-HDLC), body fat (%), and fasting FFA were selected as significant predictive variables and contributed to log (C-GC) by 60% (adjusted R ²). Replacing log (AUC-TG/AUC-HDLC) with its fasting value, log (Fasting TG/Fasting HDLC), this model still showed a strong contribution to log (C-GC) by 57% (adjusted R ²). These contributions were stronger than those in log (HOMA-IR) (52.5%), log (Fasting C-peptide) (45.7%) to log (C-GC).

Conclusions:

It is plausible that our estimation for IR without the inclusion of plasma insulin concentration can be applied in Japanese patients whose HOMA-IR is not appropriately available. The model using fasting values is less complicated and could be the best way for the estimation of IR.