The size of adrenal incidentalomas correlates with insulin resistance. Is there a cause-effect relationship?

P>Context Adrenal incidentalomas (AI) have often been associated with a high prevalence of insulin resistance (IR) and cardiovascular risk factors, although direct measurement of insulin sensitivity (IS) has never been carried out. Objective We aimed to investigate whether the morphological and hormonal features of AI correlate with the presence and severity of IR, using the hyperinsulinaemic euglycaemic clamp (HEC). Design and Measurements Forty patients with AI (22 women) with a mean age of 58 center dot 5 +/- 11 center dot 1 years underwent hormonal and morphological evaluation. Nineteen patients with AI without known history of diabetes mellitus (DM) or impaired glucose tolerance (IGT) and 17 matched controls underwent oral glucose tolerance test (OGTT) and hyperinsulinaemic euglycaemic clamp (HEC). Results Diabetes mellitus was observed in 13 patients (33%), while three (8%) had IGT. Thirty-one of the AI were nonfunctioning (82 center dot 5%), whereas two (5%) secreted cortisol (Cushing's syndrome) and seven (12 center dot 5%) showed subclinical secretion of cortisol. The 19 patients with nonfunctioning AI were more insulin resistant than controls (glucose up-take: 4 center dot 58 +/- 1 center dot 80 vs 5 center dot 85 +/- 2 center dot 48 mg/kg/min respectively; P = 0 center dot 01); IS was inversely related to the mass size (r = -0 center dot 57; P = 0 center dot 04), free urinary cortisol (r = -0 center dot 68; P = 0 center dot 01), serum cortisol after 1-mg dexamethasone suppression (-0 center dot 65; P = 0 center dot 02) and percentage of trunk fat mass (-0 center dot 77; P = 0 center dot 02) and directly related to serum adreno cortico tropic hormone (ACTH) (r = 0 center dot 62; P = 0 center dot 03). After performing multivariate regression, the mass size was found to be the most powerful predictor of IR. Conclusion Our study showed a high prevalence of insulin resistance in patients with nonfunctioning AI and suggests its possible involvement in AI growth.