Metabolic Syndrome as a Predictor of Adrenal Functional Status: A Discriminant Multivariate Analysis Versus Logistic Regression Analysis

Patients harboring adrenal tumors are characterized by higher prevalence of metabolic syndrome (MetS) components and a higher incidence of cardiovascular complications, especially in cases of subclinical or overt hormonal hypersecretion. Early detection and referral of those patients in tertiary centers could prevent unfavorable outcomes. In this cross-sectional, retrospective study, we evaluated 111 consecutive patients with adrenal incidentalomas and 14 patients with known hypersecretory adrenal lesions (autonomous cortisol secretion, primary aldosteronism, and pheochromocytoma), who were investigated in our clinic. Based on the different distribution of MetS components in patients with non-functional and functional adrenal lesions we introduced a predictive model of hormonal hypersecretion using those components. We performed multivariate discriminant analysis and compared predictive results with conventional multiple logistic regression analysis. Diabetes, impaired glucose tolerance, impaired fasting glucose, hypertension, body mass index, HDL-cholesterol levels, triglyceride levels, drug treatment for lipid disorder (statins, fenofibrate, and fish oils, alone or in combination), and maximal adrenal lesion diameter were used as discriminating covariates. Multivariate discriminant function exhibited a sensitivity of 77.27% and specificity of 73.08% in predicting adrenal hormonal hypersecretion. Receiver operating characteristic curve of discriminant predictive function had an area under the curve value of 0.785, S.E. 0.04. Logistic function delivered comparable results. MetS components exhibit a good predictive feature of hormonal hypersecretion in patients with adrenal tumors. Predictive functions may help in the search for an easy and generally available algorithm to validly predict the functional activity of adrenal masses.