Endocrine Abstracts

Purpose: A combined model of clinical, biochemical and radiological variables could help to predict autonomous cortisol secretion (ACS) in adrenal incidentalomas (AI).

Methods: We analyzed retrospectively 100 patients diagnosed of AI between 2011 and 2015. AI was defined as an adrenal mass>1 cm, accidentally discovered by radiologic examination. ACS was ruled out (ACS-) by serum cortisol post-dexamethasone suppression test (Nugent) <3 μg/dl, and was confirmed by levels ≥3 μg/dl, normal cortisoluria and no typical data of Cushing’s syndrome. The statistical analysis was performed with STATA 13.0. For multivariate analysis variables were selected by a pvalue <0.1 on univariate analysis and previous literature findings.

Results: Ninety-three patients were included in the statistical analysis. Mean age was 62.9 years and 54% were women. Fourteen patients (15%) had ACS. In the univariate analysis, the variables associated with higher risk of ACS (expressed in odds ratio (OR) and/or the proportion/mean of the variable in ACS and ACS-) were: Nugent test (3.6 vs 1.5 μg/dl, P<0.00) and maximum adenoma diameter (MAD) (26.8 vs 17.2 mm, P=0.02). Higher risk of ACS was not related with age (64 vs 63 years, P=0.57), sex (ratio of masculinity 1.4 vs 0.8, P=0.4), HTA (OR 1.7, 71.4 vs 44.9%, P=0.08), diabetes (OR=1.4, 36 vs 25.6%, P=0.4), obesity (OR=0.6, 25 vs 37.6%, P=0.4), osteoporosis (OR=1.3, 9.1 vs 7.2%, P=0.8), glucose (108.3 vs 105.5 mg/dl, P=0.7), cortisoluria (70 vs 59.7 μg/24 h, P=0.5), DHEAS (70.2 vs 46.2 μg/dl, P=0.1); ACTH (20.1 vs 18.2 pg/ml, P=0.7) or bilaterality (OR=1.4, 20 vs 15%, P=0.7). No differences were found between ACS and radiological characteristics in the CT scan (calcification, necrosis, lipid content). In the logistic regression analysis, the variables male sex, age, HTA, diabetes, Nugent, DHEAS, MAD and bilaterality were included to elaborate the ACS predictor score. It was found that the model with the best predictive power for the ACS diagnosis included age, Nugent test and DHEAS levels, with sensitivity of 89% and specificity of 100%.

Conclusions: 15% of AI in our series had ACS. We identified Nugent test and MAD as predictors of ACS. The combined model with the best ACS diagnostic accuracy combined age, Nugent test and DHEAS levels, with a specificity of 100%. This combined score could be a very useful tool to identify ACS with a higher diagnostic value than the Nugent test alone regardless of the cutoff point used. Its high specificity makes it especially indicated in the screening of AI.