OR18-04 Urine Steroid Metabolomics and Timed Urine Steroid Profiling: A Novel Test for the Diagnosis and Differential Diagnosis of Cushing’s Syndrome

Abstract Disclosure: A. Prete: HRA Pharmaceuticals, Lundbeck, Diurnal. L. Adbi: None. J.T. Bain: None. F. Shaheen: None. O. Suntornlohanakul: None. E. Deflorenne: None. A. Blanchard: None. M. Nestola: None. C. Pamporaki: None. I. Tizianel: None. E. Nowak: None. J. Bertherat: None. D.A. Vassiliadi: None. I. Bancos: None. C.M. Dennedy: None. D. Kastelan: None. M. Fassnacht: None. P. Mulatero: None. U. Ambroziak: None. A. Tabarin: None. M. Quinkler: None. S. Puglisi: None. J. Deinum: None. G.Å. Ueland: None. L. Lenzini: None. F. Beuschlein: None. M. Reincke: None. F. Ceccato: None. G. Eisenhofer: None. S.M. Corsello: None. L. Amar: None. M. Zennaro: None. N. Karavitaki: None. A.E. Taylor: None. W. Arlt: None. Introduction: Twenty-four-hour urinary free cortisol is used to diagnose Cushing’s syndrome (CS); amongst its limitations is failure to differentiate ACTH-dependent CS (AD-CS) from ACTH-independent CS (AI-CS). We tested the performance of urine steroid metabolomics (USM), the computational analysis of 24-hour urine steroid metabolome data by machine learning, for the diagnosis and differential diagnosis of CS. Given the physiological diurnal rhythm of cortisol secretion and its loss in CS, we also hypothesised that the difference in glucocorticoid excretion between CS and controls should be higher during nighttime, thereby facilitating more sensitive detection of CS. Methods: 264 subjects completed a 24-hour urine collection (40 AD-CS, 103 AI-CS due to adrenocortical adenoma or hyperplasia, 121 healthy subjects). A subset of 52 individuals (13 CS, 39 healthy) provided a nighttime urine collection and a daytime urine collection. Mass spectrometry-based multi-steroid profiling was used to quantify the urinary excretion of 27 steroid metabolites. Data were analysed by generalised matrix learning vector quantisation (a prototype-based supervised machine learning approach), multiple linear regression adjusted for age, sex, and BMI, and multiple logistic regression adjusted for age, sex, and BMI. Results: Twenty-four-hour USM demonstrated very high accuracy in differentiating CS from healthy individuals (AUC-ROC 0.99, 95%CI 0.99-0.99), reflected by higher urinary excretion of glucocorticoid and glucocorticoid precursor metabolites in CS. USM yielded high accuracy in differentiating AD-CS from AI-CS (AUC-ROC 0.91, 95%CI 0.89-0.94), with androgen metabolites being the most discriminatory. We observed significant sex differences in steroid excretion in both healthy individuals and those with CS. Women exhibited markedly lower excretion of both glucocorticoids and mineralocorticoids compared to men: average glucocorticoid excretion was reduced by 43% in healthy women and 30% in CS women, while average mineralocorticoid excretion was 29% lower in healthy women and 51% lower in CS women (results adjusted for age and BMI). Timed steroid excretion in healthy individuals reflected the diurnal pattern of adrenal steroid secretion, with lower nighttime than daytime excretion of glucocorticoid metabolites. Nighttime glucocorticoid metabolite excretion (AUC-ROC 1.00, 95%CI 0.99-1.00) performed better than daytime (AUC-ROC 0.91, 95%CI 0.77-1.00) and 24-hour excretion (AUC-ROC 0.94, 95%CI 0.86-1.00) in separating CS cases from healthy individuals. Conclusions: USM is a non-invasive, one-step candidate test for the accurate diagnosis and differential diagnosis of CS. Timed nighttime urine collection leverages cortisol circadian rhythmicity and improves the diagnostic accuracy of the current reference standard 24-hour collection. Presentation: Sunday, July 13, 2025