Abstract

OBJECTIVE:

A recent meta-analysis of 13 prospective studies reported that higher levels of adiponectin were significantly associated with lower risk of type 2 diabetes. Most previous studies, however, were limited in their ability to adjust for appropriate confounding variables. Our objective, therefore, was to study this association after adjustment for directly measured adiposity and insulin sensitivity, expressed as the insulin sensitivity index (S(I)).

RESEARCH DESIGN AND METHODS:

The study included 1,096 Hispanic and African American participants free of diabetes at baseline (2000-2002) who returned for follow-up after 5 years. S(I) was determined from frequently sampled intravenous glucose tolerance tests with minimal model analysis. Visceral adipose tissue (VAT) area was determined by computed tomography. Diabetes and impaired fasting glucose (IFG) were defined using American Diabetes Association criteria. Multivariate generalized estimating equation logistic regression models were used to account for correlations within families.

RESULTS:

A total of 82 subjects met criteria for incident diabetes. After adjustment for age, sex, ethnicity, and smoking, adiponectin was significantly inversely associated with diabetes (odds ratio [OR] 0.54 per 1 SD difference [95% CI 0.38-0.76]). The association remained significant after additional adjustment in individual models for BMI, homeostasis model assessment of insulin resistance, or VAT (all P < 0.05). However, adiponectin was no longer associated in separate models adjusted for S(I) or IFG (OR 0.81 [0.56-1.16] and 0.75 [0.53-1.06], respectively).

CONCLUSIONS:

Adiponectin was inversely associated with incident diabetes after adjustment for conventional anthropometric and metabolic variables or VAT. Adjustment for detailed measures of S(I) attenuated this relationship, however, suggesting that the link between adiponectin and diabetes may operate at least in part through insulin resistance.