Corticosteroid-binding globulin affects the relationship between circulating adiponectin and cortisol in men and women.

Abstract

Inflammatory pathways are increasingly recognized to be tightly associated with insulin resistance in humans. The promoter region of the adiponectin gene--Apm1--encompasses consensus sequences for glucocorticosteroid receptor responsive element. Dexamethasone induced downregulation of adiponectin secretion in vitro, whereas prednisolone administration increased circulating adiponectin concentrations. As previous studies have demonstrated an inverse relationship between corticosteroid-binding globulin (CBG), body mass index, and insulin resistance, we studied whether CBG could explain cortisol-to-adiponectin relationship. One hundred twenty-two healthy subjects were enrolled in a cross-sectional study. Plasma CBG and serum cortisol concentration were measured by radioimmunoassay. The cortisol-to-CBG ratio was used to calculate free cortisol. An RIA kit (Linco Research, St Louis, MO) was used to measure adiponectin levels. Insulin resistance was calculated using the homeostatis model of assessment (HOMA) value. Circulating adiponectin was associated with serum CBG ( r = 0.38, P < .00001), both in men ( r = 0.26, P = .03, n = 79) and women ( r = 0.48, P = .003, n = 43), and with insulin resistance (HOMA index) ( r = -0.30, P < .0001) in both. Free cortisol correlated negatively with adiponectin only in women ( r = -0.32, P = .04), but not in men ( r = 0.01, P = .89). Serum CBG concentration was significantly lower among men in the lowest quartile of adiponectin when compared with the remaining subjects (37.3 +/- 5.7 vs 40.6 +/- 5.1, P = .016), whereas men in the highest quartile of adiponectin showed significantly increased free cortisol index (14.2 +/- 3.3 vs 12.2 +/- 3.1, P = .039). Women in the lowest quartile of adiponectin also displayed significantly lower CBG concentration than that present in the remaining subjects (38.6 +/- 6.9 vs 44.4 +/- 5.5, P = .016), whereas free cortisol index was not significantly different across adiponectin quartiles ( P = .1). In a stepwise regression analysis, body mass index ( P = .0011), CBG ( P = .0047), and sex ( P = .04) contributed to 15%, 8%, and 3%, respectively, of adiponectin variance. Using CBG as dependent variable, both adiponectin ( P = .0002) and fasting cortisol ( P = .019) contributed to 14% and 4%, respectively, of CBG variance. In summary, circulating adiponectin, CBG concentration, and fasting cortisol were significantly interrelated in healthy subjects. A significant sexual dimorphism exists in this association.