Real-world daily clinical use of liquid chromatography-tandem mass spectrometry (LC-MS/MS) for adrenal hormones measurement: comparison with chemiluminescent immunoassays in health and adrenal disorders

BACKGROUND: In clinical practice immunoassays (IA) represent the most used method for measuring adrenal steroids, despite the risk of cross-reactivity that may lead to falsely high results. The comparison with liquid chromatography–tandem mass spectrometry (LC-MS/MS), considered the reference method, has not yet been evaluated across the pathophysiological spectrum of steroid production in large real-world cohorts. AIM: To compare serum concentrations of cortisol, aldosterone (ALDO), 17α-hydroxyprogesterone (17-OHP), Δ4-androstenedione (Δ4) and DHEAS obtained by IA and LC-MS/MS, analyzing systematic differences and analytical performance of the methods across concentration ranges in a real-world setting in healthy and disease. METHODS: In this prospective real-world observational study, 953 patients (374 women, 579 men; median age 52.5 years) were included, undergoing hormonal evaluation at the Unit of Endocrinology the University of Modena and Reggio Emilia between 2022 and 2024. On the same fasting blood samples, morning cortisol levels (n=953) were measured by Immunoassay (CMIA DxI800,Beckman Coulter); 17-OHP (n=242) and DHEAS (n=229) by CLIA iSYS (Immunodiagnostic Systems); ALDO (n=125) and Δ4 (n=218) by CLIA Liaison XL (Diasorin); and all adrenal steroids by LC-MS/MS (MassChrom® Steroids, Chromsystems). Results are expressed as median and interquartile range. RESULTS: Cortisol concentrations measured by LC-MS/MS [14.75(11.23–18.56)µg/dL] were significantly (p<0.001) higher than those obtained by IA [13.60(10.50–17.00)µg/dL], with a good correlation between the methods (R²=0.769). Agreement was high at lower cortisol concentrations: the prevalence of values <5µg/dL was identical (5.9%), and that for values <6.7µg/dL was overlapping (7.7% by IA, 7.6% by LC-MS/MS). Furthermore, the correlation in the lower ranges was particularly high (R²=0.914 for values <5µg/dL; R²=0.925 for values <6.7µg/dL). However, increasing discordance emerged at higher values: the prevalence of cortisol <10µg/dL was greater by IA than to LC-MS/MS (21.8% vs 19.0%; R²=0.845). In contrast, LC-MS/MS identified more subjects with values >20µg/dL (17.8% vs 9.9%, R²=0.215), >22.6µg/dL (9.2% vs 5.1%, R²=0.214) and >25µg/dL (5.3% vs 2.9%, R²=0.226). Regression analysis showed a reduction in agreement at high cortisol concentrations. Regarding ALDO, 17-OHP, DHEAS and Δ4, LC-MS/MS measurement yielded significantly lower median concentrations (p<0.001) compared to IA, with values of 86.5(46.9–169.4) vs 145.0(77.3–243.0) pg/mL for ALDO(R²=0.709); 0.69(0.35–1.28) vs 1.20(0.70–1.80) ng/mL for 17-OHP(R²=0.715); 1.07(0.49–1.83)vs1.55(0.66–2.39) µg/mL for DHEAS(R²=0.738); and 1.00(0.57–1.51)vs1.25(0.59–2.13) ng/mL for Δ4(R²=0.746). Despite the overall good correlation, agreement was concentration-dependent: for ALDO, concordance decreased for values >100 pg/mL (R²=0.523), overestimating the prevalence of hyperaldosteronism compared to LC-MS/MS (66% vs 45%). For 17-OHP, agreement dropped (R²=0.539) at values near or above the screening cut-off for 21-hydroxylase deficiency (>2 ng/mL), where the IA generates a high rate of false positives, bringing the prevalence of values above this cut-off to 22% compared to the 12% by LC-MS/MS. CONCLUSION: Although the methods show a good overall correlation, differences emerge at specific diagnostic cut-offs. IA tends to underestimate cortisol concentrations compared to LC-MS/MS, with clinical implications in terms of overestimation of hypocortisolism and underestimation of hypercortisolism. IA overestimates ALDO, 17-OHP, DHEAS and Δ4, generating frequent false positives in the critical screening ranges for hyperaldosteronism and 21-hydroxylase deficiency. This limitation carries the risk of misclassification, reaffirming LC-MS/MS as the gold standard for steroid measurement also in endocrinological practice.