Plasma Steroid Profiles in Subclinical Compared With Overt Adrenal Cushing Syndrome

Jimmy Masjkur; Matthias Gruber; Mirko Peitzsch; Denise Kaden; Guido Di Dalmazi; Martin Bidlingmaier; Stephanie Zopp; Katharina Langton; Julia Fazel; Felix Beuschlein; Stefan Richard Bornstein; Martin Reincke; Graeme Eisenhofer


J Clin Endocrinol Metab. 2019;104(10):4331-4340. 

In This Article


The main findings of our study are twofold. First, we show that the measurement of multiple steroids from a single baseline plasma sample can identify patients with SC with high accuracy and distinguish them from normal controls and subjects in whom disease was excluded. Second, we demonstrate that patients with SC, although heterogeneous in terms of standard diagnostic tests and glucocorticoid output, have a distinct steroid fingerprint that separates them from patients with AC in discriminant analysis. Our data extend those of our previous publications addressing similar diagnostic questions in primary aldosteronism.[23–25] Together, these data suggest that cumbersome, multistep biochemical testing for diagnosis of adrenal pathologies could be radically simplified by a single test measuring plasma concentrations of multiple steroids.

Numerous previous studies have shown that no single test to establish hypercortisolism has 100% sensitivity or perfect accuracy, and so patients may be diagnosed incorrectly with SC because of false-positive results.[26,27] Lack of specificity of conventional tests can have far-reaching consequences, as for example in patients with pronounced metabolic syndrome who might become candidates for unilateral adrenalectomy when presenting with false-positive tests for autonomous cortisol secretion.[14] Inadequate diagnostic sensitivity also remains a pressing problem. Missing the correct diagnosis in a mildly symptomatic patient with a true cortisol hypersecreting tumor can delay adequate treatment, with potential for an adverse outcome.

Nocturnal salivary cortisol concentration, an early hallmark of hypercortisolism, has been extensively shown in some studies to be poorly sensitive for SC.[1,9,24] Our data confirm in a large data set poor discrimination of patients with SC compared with those with AC, using SFC measurements. Elevations in UFC have been widely used to predict the chronic manifestation of subtle cortisol excess despite the drawbacks due to common sampling errors and problems with cross-reactivity to other steroids in immunoassays. Of note, patients with SC in our study had similar mean levels of UFC as patients without Cushing syndrome. However, UFC provided reasonable separation of patients with AC from other groups. In summary, the ROC curves of the routine tests for diagnosis of SC in our study revealed lower diagnostic effectiveness of these methods compared with the selected steroid combination.

Among the various tests examined for differentiating patients with and without SC, all performed poorly except for the DST. This can be explained in part by the strong emphasis clinicians involved in our study likely placed on the DST, which, according to the recently published European guideline,[14] classifies autonomous cortisol secretion according to this test. There has been disagreement on the best cutoff for cortisol after DST. A recent study revealed that the optimal value could even be lower than the commonly used value of 1.8 μg/dL (50 nmol/L).[2,28] The response of cortisol after DST in patients who do not exhibit signs and symptoms specific for AC have to be analyzed together with other results from routine tests. Indeed, some patients with positive DST results were eventually categorized as EX, because there was no evidence of abnormal cortisol excess in other hormonal parameters, making this a source of uncertainty.[26] The more common threshold value of 1.8 μg/dL was used in our study to show that the level of cortisol after the DST remains the most sensitive assay alongside the combination of steroids to subgroup patients with subtle autonomous cortisol hypersecretion.

More recently, the discussion about cutoffs for correct classification of hormonal excess has moved toward prediction of long-term outcomes in SC. It is a well-established concept that hypercortisolism can lead to associated comorbidities, such as metabolic, musculoskeletal, and cardiovascular diseases,[2] similar to patients with overt Cushing syndrome. A retrospective study analyzed outcome, using the classification of patients with adrenal incidentalomas as nonsecreting or as stable SC, according to the cortisol levels after DST, and in another cohort with increasing cortisol levels on follow-up. This arbitrary classification was able to identify patients with increased cardiovascular events and mortality rates.[29] The data were confirmed by two quite similar studies.[30,31] Future studies will need to show whether a classification based on steroid fingerprinting will make it possible to predict long-term outcome in patients with subclinical hypercortisolism.

Several studies have reiterated the advantages of LC-MS/MS over immunoassays for measurements of steroids.[10,11,15] Here, we show that plasma 11-deoxycortisol, 11-deoxycorticosterone, DHEA, DHEA-S, and corticosterone levels can reliably distinguish patients with and without SC. Circulating levels of DHEA-S have been found in the high-normal range in patients with ACTH-dependent Cushing syndrome, whereas both AC and SC are characterized by decreased circulating basal concentrations of DHEA-S and DHEA.[32–34] The inverse association of DHEA-S and cortisol excess has been revealed in several studies to be exclusively associated with adrenocortical adenoma.[11,35] Indeed, our data also confirm lower DHEA and DHEA-S levels in patients with AC and SC, supporting previous suggestions that ACTH is a major determinant of their secretion.[36,37]

Low levels of progesterone in the SC and AC groups in our study can be explained by two mechanisms. Continuously elevated cortisol levels inhibit the pituitary-gonadal axis, leading to anovulatory cycles, prohibiting the action of progesterone in the uterus as well as the secretion of GnRH from the thalamus.[38] This effect likely contributes to low progesterone in our mainly female, premenopausal cohort with florid AC. In the sex-balanced, predominantly postmenopausal SC cohort, precursor substrate flow characteristics, enhanced by partially suppressed plasma ACTH, could result in the pattern of low plasma pregnenolone and progesterone concentrations. Indeed, chronic ACTH excess influences intra-adrenal use of δ5-pregnene, leading to elevation of pregnenolone levels,[39,40] which is very much in line with our findings.

Our study has several limitations, one of which involved the older age of the SC group than other cohorts and the female predominance of the AC cohort. Because the latter is an established observation in AC[41] and because adrenal incidentalomas that prompt consideration of hypercortisolism are predominantly found in the elderly,[2,42] neither of these issues is easily addressed by matched populations. Rather, we addressed these potential confounders by use of models and normalizations that took advantage of data from a larger reference population. It nevertheless remains possible that even with these corrections, age, sex, and menstrual phase differences may have contributed to differing patterns of steroids among the AC and SC groups. Another limitation was that the diagnosis and exclusion of hypercortisolism were based on routine tests that are not infallible. Findings that there were differences among steroids between the reference and EX groups and that the EX group had levels of many steroids intermediate between the reference and SC and AC groups raise the possibility of milder forms of adrenal cortical dysfunction within the EX group that were not identified by routine tests. Finally, it would have been useful to assess outcomes after adrenalectomy. However, the indication of adrenalectomy for patients with SC is often not achieved.

In conclusion, this study establishes a method to identify patients with SC using LC-MS/MS measurements of a panel of adrenal steroids. Performance of the steroid profile was more advanced than using SFC and UFC and had similar accuracy as routine test combinations that included the DST. Thus, the plasma steroid panel could serve as a single test alternative to screen for SC or to confirm the findings on DST. Versatility of a single steroid-profiling method for detecting other disorders of adrenal steroidogenesis provides another advantage for use of the method in the routine laboratory.