Endocrine Abstracts

Background: The dexamethasone suppression test (DST) is among essential tests for hypercortisolism diagnosis. Due to high sensitivity and specificity, liquid chromatography-tandem mass spectrometry (LC-MS/MS) is replacing obsolete immunoassays (IA) for steroid measurement. However, exhaustive data about IA and LC-MS/MS concordance in the frame of DST are still not available.

Aim: To compare cortisol measurements by two routine IAs and LC-MS/MS in basal (F0) and post-DST (Fdst) sera. To assess between-method concordance in diagnosing hypercortisolism.

Methods: We enrolled 330 patients with hypercortisolism suspicion or adrenal mass. Of these, 287 patients had their paired F0 and Fdst measured by ElecsysE170 (Roche), and 43 by DxI800 (Beckman). All samples were also measured by a validated LC-MS/MS method. Methods were compared by Wilcoxon test and Passing and Bablok regression. We evaluated between-methods concordance in classifying patients in “non-secreting” (NS) (Fdst<18 ng/ml), “possible autonomous cortisol secretion” (PACS) (Fdst:18-50 ng/ml) and “autonomous cortisol secretion” (ACS) (Fdst≥50 ng/ml).

Results: F0 levels assayed by ElecsysE170 (median (min-max): 151.0 (47.0-380.0) ng/ml) were significantly higher than by LC-MS/MS (116.8 (38.1-249.5) ng/ml))(P<0.001), with significant proportional overestimation (slope(95CI): 1.366(1.309-1.423)) and constant underestimation (intercept(95CI): −7.9(-14.1--1.8)) of the former and R=0.916. Comparing DxI800 vs LC-MS/MS resulted in higher values (134.0 (52.0-269.0) ng/ml vs 115.1 (55.3-207.8) ng/ml, respectively) (P<0.001), significant slope (1.337(1.226-1.451)) and intercept (-18.6(-33,3--7.3)) and R=0.955. Regarding Fdst levels, no significant differences were found between ElecsysE170 (14.0 (3.00-290.0) ng/ml) and LC-MS/MS (13.70 (2.72-168.7) ng/ml) (P=0.060), with no significant proportional or constant errors and R=0.907. No difference was obtained for DxI800 (14.0 (6.0-258.0) ng/ml) vs LC-MS/MS (13.3 (6.9-199.0) ng/ml) (P=0.160), with no proportional error, R=0.910, and a small significant constant underestimation (intercept: −2.8; 95CI: −4.9--0.7). According to ElecsysE170, we classified 186 patients (64.8%) as NS, 88 (30.7%) as PACS and 13 (4.5%) as ACS. When using LC-MS/MS, 13 (7%, false negatives) of NS, 19 (21.6%, false positives) of PACS and 3 (23.1%, false positives) of ACS were not confirmed, with total 35 (12.2%) misdiagnoses. When using DxI800 Fdst, 26 (60.5%) were NS, 14 (32.5%) were PACS and 3 (7.0%) were ACS. Of these, LC-MS/MS did not confirm 1 (3.8%, false negative) within NS and 1 case (33.3%, false positive) within ACS classes, with total 2 (4.7%) misdiagnoses.

Conclusions: Unexpectedly, concordance between LC-MS/MS and IAs is poorer at basal than DST-suppressed cortisol levels. Nonetheless, a non-negligible amount of patients may receive a different diagnosis when using the more specific and accurate LC-MS/MS measurement.