The Biochemical Diagnosis of Adrenal Insufficiency With Modern Cortisol Assays

Reappraisal in the Setting of Opioid Exposure and Hospitalization

Caitlin Colling; Lisa Nachtigall; Beverly M. K. Biller; Karen K. Miller


Clin Endocrinol. 2022;96(1):21-29. 

In This Article

Materials and Methods

Study Population

This was a retrospective study of adult patients (age ≥ 18) who underwent cosyntropin stimulation testing for adrenal insufficiency as part of clinical care within Mass General Brigham (MGB) health system between May 31, 2015 and October 30, 2020. Patients were identified using the Research Patient Data Registry (RPDR), a centralized clinical data registry at MGB containing data on 6.5 million patients and 2 billion facts. Diagnostic codes and demographic and laboratory data were obtained from RPDR, which includes outpatient, inpatient and emergency visits within MGB. Detailed prescription data were obtained from the MGB Enterprise Data Warehouse (EDW). EDW is an integrated data mart with clinical data sets including diagnoses, labs, medications and procedures incorporated from multiple sources, including electronic medical records, clinical and financial systems.

Cohort 1: Presence of opioid-induced adrenal insufficiency was evaluated in inpatients or outpatients with an active outpatient opioid prescription(s) for greater than 30 days before cosyntropin stimulation testing (cohort one, Figure 1A). Opioids were defined as long-acting or short-acting according to the CDC Opioid National Drug Code and Oral Morphine Milligram Equivalent (MME) Conversion File.[21] Patients exposed to both long-acting and short-acting opioids were included in the long-acting group.

Figure 1.

Inclusion/exclusion flow diagrams. (A) Cohort 1: Flow diagram for patients with chronic opioid exposure undergoing cosyntropin stimulation test (CST). (B) Cohort 2: Flow diagram for patients undergoing cosyntropin stimulation test with morning cortisol collected. aOutliers defined as three times the interquartile range above the third quartile of peak cortisol

Cohort 2: The validity of low morning cortisol for the diagnosis of adrenal insufficiency was evaluated in inpatients and outpatients who underwent cosyntropin stimulation tests and had a morning cortisol level collected between 6 and 10 AM (Cohort 2, Figure 1B). Opioid-exposed was defined as having an inpatient or outpatient opioid prescription for hospitalized patients or outpatients, respectively, at the time of testing. Controls were opioid-unexposed, defined as having no inpatient or outpatient opioid prescriptions for at least 14 days before testing. Cosyntropin stimulation testing was categorized as inpatient or outpatient, due to the unknown effect of hospitalization on cortisol circadian rhythm.

For all analyses, patients with a history of cirrhosis, pituitary disease, or treatment with immune checkpoint inhibitors or with elevated ACTH (>100 pg/ml) were excluded. Diagnoses were identified by International Classification of Disease Ninth and Tenth Revision codes within 6 months of testing. Cosyntropin stimulation tests performed in ICU were excluded due to changes in cortisol metabolism and volume of distribution during critical illness.[22] Patients who received oral, intravenous or intraarticular glucocorticoids within 30 days or oral estrogen at the time of testing were excluded. An adequate cosyntropin stimulation test, defined as a serum cortisol level drawn between 25 and 70 min after administration of 250 μg of cosyntropin, or maximum cortisol ≥405 nmol/L (14.7 μg/dl) on the day of cosyntropin administration was required for inclusion. Only the first cosyntropin stimulation test was included for patients with more than one qualifying test. Adrenal insufficiency was diagnosed by maximum cortisol <405 nmol/L (14.7 μg/dl), with an analysis applying the standard cutoff of <500 nmol/L (18.1 μg/dl) performed. Only cosyntropin stimulation tests performed with modern specific cortisol assays, Architect (Abbott Laboratories) immunoassay or Elecsys Cortisol II (Roche Diagnostics), were included. The cutoff of 405 nmol/L was selected to reflect the improved specificity of monoclonal antibody immunoassays compared to older assays resulting in a lower diagnostic cutoff for abnormal HPA axis function.[16,23] As with all disorders that utilize biochemical testing in the diagnosis, abnormal or normal tests are interpreted incorporating the clinical circumstances and the pre-test probability of the disease. Patients who failed to achieve a cortisol of 405 nmol/L (14.7 μg/dl) after stimulation were classified as having adrenal insufficiency, recognizing that in clinical settings information including risk factors and symptoms would ultimately factor strongly into whether adrenal insufficiency is diagnosed. Low morning cortisol was defined as serum morning cortisol of <83 nmol/L (3 μg/dl) collected between 6 and 10 AM. Daily morphine milligram equivalent was calculated using the CDC Oral MME Conversion File.[21] Only outpatient opioid prescriptions were included and intravenous, cough, and cold formulations were excluded. Duration of exposure was calculated as the continuous time in months of opioid prescriptions before cosyntropin stimulation testing. One-year cumulative opioid exposure was calculated as sum of MME for all opioid prescriptions with start date within one year of date of cosyntropin stimulation test.

ACTH and cortisol tests were run by each hospital laboratory within MGB as part of clinical care. Assays used for measuring cortisol varied over time. The majority of cosyntropin stimulation tests were performed at Massachusetts General Hospital where cortisol was measured with Architect (Abbott Laboratories) immunoassay from the start of study through January 2019, after which the Elecsys Cortisol II (Roche Diagnostics), an electrochemiluminescence competitive immunoassay, was used. The Architect and Elecsys Cortisol assays were compared with LC-MS/MS and GC-MS, respectively, with slopes of 1.0 ± 0.1 and a correlation coefficient of ≥0.95.[24,25] Brigham Health, the site of 20% of cosyntropin stimulation tests analysed, also measured cortisol with Elecsys Cortisol II from December 2016 through study end. ACTH was measured with the Elecsys ACTH assay (Roche Diagnostics), an electrochemiluminescence immunoassay, at all locations. The MGB Institutional Review Board approved this study with waiver of signed consent.

Statistical Analysis

Categorical data are presented as number (percentage of subjects), and continuous data are presented as the median (interquartile range). The statistical significance of differences between categorical data was assessed with Pearson's χ 2 tests or Fisher's Exact Test if fewer than 10 observations per category; unpaired t-tests were used for normally distributed continuous data, and Wilcoxon's rank sum test were used for nonnormal continuous data. For the evaluation of morning serum cortisol values for the diagnosis of adrenal insufficiency, the sensitivity and specificity of a morning serum cortisol was calculated. For all analyses, a two-tailed p value <.05 was considered statistically significant. JMP (version 15; SAS Institute Inc.) was used for statistical analyses.