Low Diagnostic Utility of Overnight High-Dose Dexamethasone Suppression Test in ACTH-Dependent Cushing’s Syndrome

Link to article at PubMed

Endocr Pract. 2021 Jul;27(7):723-727. doi: 10.1016/j.eprac.2020.12.011. Epub 2020 Dec 29.

ABSTRACT

OBJECTIVE: Overnight high-dose dexamethasone suppression test (ON-HDDST) is a simple test to localize the source of ACTH in patients with ACTH-dependent Cushing's syndrome (CS). However, previous studies have reported its varying accuracy. We studied the utility of ON-HDDST in diagnosing Cushing's disease (CD) in a series of patients with CD and ectopic ACTH syndrome (EAS).

METHODS: We conducted a retrospective study of 88 patients with ACTH-dependent CS (plasma ACTH > 20.0 pg/mL), who underwent an ON-HDDST. CD and EAS were diagnosed in 68 and 20 patients, respectively. Patients were investigated using MRI of the sellar region, CT of the thorax/abdomen, Gallium-68-DOTANOC PET scan, and bilateral inferior petrosal sinus sampling as required.

RESULTS: Patients with EAS had a significantly higher serum cortisol after ON-HDDST than patients with CD (median [IQR], 19.9 [12.4-31.1] μg/dL vs 9.9 [5.1-25.0] μg/dL, P <.01). A suppressed ON-HDDST (≥50% fall from baseline) was noted in 44 (65%) patients with CD and 3 (15%) patients with EAS (P <.0001). Among patients with CD, cortisol suppression >50% was noted in 35 (76%) of patients with microadenoma and 7 (44%) with macroadenoma. Among patients with EAS, ON-HDDST was suppressed in 1 of 6 patients (17%) with an occult tumor and 2 of 14 patients (14%) with a localized tumor. The ROC curve plotted for the percentage suppression of cortisol had an area under the curve (AUC) of 0.72 (P =.01). The best test parameters, with 65% sensitivity, 85% specificity, 94% positive predictive value, 42% negative predictive value, and 69% accuracy, were at 50% cutoff level.

CONCLUSION: The ON-HDDST had a poor diagnostic value in differentiating CD and EAS.

PMID:34250909 | DOI:10.1016/j.eprac.2020.12.011

Leave a Reply

Your email address will not be published.