Big data determination and validation of reference range for 24-h urine cortisol by liquid chromatography-mass spectrometry.

Abstract

Objective: Twenty-four-hour urine-free cortisol (UFC) is a first-line test for Cushing syndrome (CS). A new mass spectrometry assay for UFC requires a validated, relevant reference range appropriate to a screening population.

Design: Combined retrospective and prospective cohort study in a government health system and tertiary endocrinology clinic, Canada. Participants were patients with potential features of CS.

Methods: The refineR reference interval algorithm was used to derive a middle 95%ile reference interval from 4830 UFC results in non-CS patients, compared with 120 prospective patients where evaluation excluded CS.

Results: Urine-free cortisol and 24-h urine volume were correlated (r = 0.28, P < .0001). There was no significant difference between the volume-corrected UFC distributions in the prospective vs retrospective populations (P = .09). Urine-free cortisol distribution was highly skewed (P < .0001) and showed strong sex interaction. The refineR-generated adult male UFC upper reference limit was 238 nmol/day (86.3 μg/day) and for females was 147 nmol/day (53.3 μg/day); urine volume-corrected, the upper limits were 89 nmol/L (32.3 μg/L) and 91 nmol/L (32.9 μg/L), respectively. Applied to both populations, between 3% and 8% of all results would be flagged high; most are expected to represent nonneoplastic (pseudo)Cushing's.

Conclusions: We used mass population data, where the prevalence of CS was likely very rare, plus a carefully phenotyped sample where CS was considered but excluded, to derive a validated reference interval for 24-h UFC by mass spectrometry in populations that reflect real-world use of the test. Given the highly skewed upper tail of the population distribution, it is probable that high test specificity for CS will require multimodality diagnostic confirmation.