New Diagnostic Criteria for Subclinical Hypercortisolism Using Postsurgical Hypocortisolism

The Co-work of Adrenal Research Study

Seung Hun Lee; Kee-Ho Song; Jinju Kim; Sooyoun Park; Seong Hee Ahn; Hyeonmok Kim; Yoon Young Cho; Sunghwan Suh; Beom-Jun Kim; Jae Hyeon Kim; Jung-Min Koh


Clin Endocrinol. 2017;86(1):10-18. 

In This Article

Abstract and Introduction


Objective There is no consensus on the biochemical diagnostic criteria for subclinical hypercortisolism (SH). Using parameters related to the hypothalamic–pituitary–adrenal axis, we aimed to develop a diagnostic model of SH for predicting postsurgical hypocortisolism and metabolic complications.

Design Prospective and cross-sectional, observational, multicentre study in Korea.

Methods After exclusion of overt Cushing's syndrome, adrenal incidentaloma (AI) patients who underwent unilateral adrenalectomy (n = 99) and AI patients (n = 843) were included. Primary outcome was defined as the presence of postsurgical hypocortisolism; secondary outcome was the presence of ≥4 complications (components of the metabolic syndrome and low bone mass). Postsurgical hypocortisolism was determined on the fifth postsurgery day using the ACTH stimulation test.

Results Thirty-three of the 99 patients developed postsurgical hypocortisolism. Analysis of the presurgery overnight 1-mg dexamethasone suppression test (1-mg DST) showed that all patients with cortisol levels of >138 nmol/l experienced postsurgical hypocortisolism, whereas those with levels of ≤61 nmol/l did not. The models of (i) 1-mg DST >138 nmol/l or (ii) >61 nmol/l with the presence of one among low levels of ACTH and dehydroepiandrosterone–sulphate had the highest accuracy (89·9%, P < 0·001) and odds ratio [OR 111·62, 95% confidence interval (CI) 21·98–566·74, P < 0·001] for predicting postsurgical hypocortisolism. Finally, patients with the same criteria in the 843 AI patients showed the highest risk for having ≥4 complications (OR 3·51, 95% CI 1·84–6·69, P < 0·001), regardless of gender, age, body mass index and bilaterality.

Conclusions Our proposed model is able to accurately predict subtle cortisol excess and its chronic manifestations in AI patients.


Subclinical hypercortisolism (SH) is defined as a condition with excessive biochemical cortisol levels without classical signs or symptoms of overt hypercortisolism, for example purple striae, easy bruising, proximal muscle weakness and plethora.[1,2] Although emerging evidence suggests that SH should be corrected to prevent chronic sequelae,[3–8] there is no consensus regarding the biochemical criteria for diagnosing SH.[1] At present, SH diagnosis relies on altered biochemical parameters related to the hypothalamic–pituitary–adrenal (HPA) axis with different cut-off levels, for example (i) lack of cortisol suppression after a 1-mg overnight dexamethasone suppression test (1-mg DST); (ii) elevated 24-h urinary-free cortisol (UFC) levels; (iii) loss of cortisol diurnal rhythm; (iv) low ACTH levels; and/or (v) low dehydroepiandrosterone–sulphate (DHEA-S) levels.[1,2,9–15] The absence of gold standard criteria, mainly originating from a lack of large-scale prospective studies regarding its long-term sequelae,[1,6,14,16] makes it challenging for clinicians to diagnose SH. The chronic sequelae of SH, for example metabolic syndrome (MetS) components and low bone mass, occur more frequently even in healthy adults,[17] compared with the occurrence of SH (prevalence 0·2–2·0%).[1] A large-scale, long-term prospective study to demonstrate a statistically clear cut-off point has therefore proven difficult to perform. In this context, a recommendation for adrenalectomy in patients with SH largely depends on expert opinion.[1,2,9–12,14,15]

In our study, we hypothesized that the occurrence of immediate, postsurgical hypocortisolism might provide an alternative standard for diagnosing SH, because it indicates a presurgical inappropriate cortisol elevation.[1,18] However, only one study has previously evaluated the diagnostic criteria of SH using postsurgical hypocortisolism occurrence.[18] In that study, the presence of >2 alterations among 1-mg DST > 138 nmol/l, ACTH < 2·2 pmol/l, elevated UFC and midnight serum cortisol had the highest odds ratio (OR) for predicting postsurgical hypocortisolism [OR 10·45 #bib95% confidence interval (CI) 2·54–42·95]. However, that study suggested that postsurgical hypocortisolism could not completely be ruled out due to the <71·7% accuracy of any parameter combinations; in addition, the clinical relevance of the diagnostic criteria was not evaluated. In our multi-centre, prospective study on a relatively larger number of patients, we developed diagnostic criteria with improved prediction (OR 111·62, 95% CI 21·98–566·74) and accuracy (89·9%) for SH. Furthermore, we confirmed that the same diagnostic criteria were associated with metabolic complications in a large-scale, cross-sectional design.