Background: Accurate measurement of serum cortisol is essential in the investigation of the HPA axis. It has been documented that routine immunoassays are liable to both under- and over-recovery leading to inaccurate results and subsequent inappropriate investigations. This study seeks to provide an up-to-date assessment of the accuracy of the major immunoassay platforms and compares results to a liquid chromatographytandem mass spectrometry (LCMS/MS) candidate reference measurement procedure (cRMP).
Design: Serum remaining from routine analysis was aliquoted and distributed to four different centres for cortisol analysis by their respective immunoassay. In addition, aliquots were analysed by a routine LCMS/MS assay and a LCMS/MS cRMP to provide metrologically traceable results for comparison. Cohort groups included: males (n=42), postmenopausal females (n=44), pregnant patients (n=68), patients taking prednisolone (n=42), and patients taking the 11β-hydroxylase inhibitor, metyrapone (n=27). Cortisol-binding globulin (CBG) was measured in the postmenopausal female and pregnant cohorts.
Results: Considerable bias was observed across the male (−15 to +15%) and postmenopausal female (−13 to +19%) cohorts. In both the prednisolone and metyrapone samples, all immunoassays over-recovered cortisol (prednisolone patients: 142707 nmol/l and metyrapone patients: 93118 nmol/l). The pregnancy cohort displayed a bias ranging from −45 to +8%. Higher concentrations of CBG were observed during pregnancy than in the post menopausal state. The routine LCMS/MS method showed no significant bias relative to the cRMP.
Conclusions: Accurate quantitation of cortisol by current immunoassays is compromised by the non-specificity of the antibody for cortisol and matrix effects. Users should be aware of the limitations of their current assay and consider these when interpreting results. This is especially pertinent in patients taking metyrapone where dosage may be titrated against cortisol concentration.