Predictive model for autonomous cortisol secretion development in non-functioning adrenal incidentalomas

We aimed to develop a predictive model able to stratify patients with non-functioning adrenal incidentalomas (AIs), according to their risk for developing autonomous cortisol secretion (ACS) during follow-up.

This was a retrospective study of patients with non-functioning AIs consecutively evaluated at a single institution between 2013 and 2019 in whom hormonal follow-up information was available for at least 1 year. Clinical, biochemical, and radiological features were used to build a multivariate Cox regression model using the estimation of all possible equations.

We included 331 patients with non-functioning AIs. ACS (post-dexamethasone suppression test (DST) serum cortisol > 1.8 µg/dL) developed in 73 patients during a median follow-up time of 35.7 months [range 12.8-165.4]. The best predictive model for ACS development during follow-up combined age, post-DST serum cortisol, and bilaterality at presentation and showed good diagnostic accuracy (AUC-ROC 0.70 [95% CI 0.65-0.75]). The lowest risk for ACS development was found among patients < 50 years old with cortisol post-DST values < 0.45 µg/dL and with unilateral tumors (risk 2.42%). Baseline post-DST serum cortisol levels at diagnosis were the most important factor for the development of ACS during follow-up (hazard ratio 3.56 for each µg/dL, p < 0.001). The rate of ACS development was associated with post-DST cortisol levels, being 19.2, 32.3, and 68.1 cases/10,000 person-years for patients with baseline post-DST cortisol < 0.9 µg/dL, 0.9-1.3 µg/dL, and > 1.3 µg/dL, respectively.

After ruling out malignancy, follow-up visits for patients < 50 years old with unilateral non-functioning AIs and post-DST serum cortisol < 0.45 µg/dL are considered unnecessary given the low risk of developing ACS during follow-up.