Screening for Cushing’s syndrome: The ‘CS-Screening’ Shiny app

Cushing’s syndrome (CS) results from supraphysiological exposure to glucocorticoids, due to administration of corticosteroids (exogenous CS) or autonomous cortisol overproduction, whether or not ACTH-dependent (endogenous CS). ¹ As CS is associated with significant morbidity and mortality, an early diagnosis of CS is warranted though it is very challenging partly because of similarity with other common conditions (i.e. pseudo-CS). ² In a recent article on diagnosing CS published in the British Journal of General Practice, the catalyst for the article was ‘the experience of one of the authors whose diagnosis of Cushing’s was not immediately recognised despite the presence of numerous suggestive clinical features and comorbidities’. ³ At the same time, clinicians are faced with an increasing number of patients requiring exclusion of CS, in the setting of increasing prevalence of hypertension and metabolic syndrome, and adrenal incidentalomas due to greater use of medical imaging. ³ Given the high sensitivity and specificity, the recommended initial tests for suspected CS are late-night salivary cortisol (LNSC), 24-hour urinary free cortisol (UFC) and 1-mg overnight dexamethasone suppression test (ONDST).^2,4 Disappointingly, in a recent study determining the relative frequency of the recommended tests for CS in primary care, it was found that the initial testing for CS was done by measurement of serum cortisol alone in 78.7% of cases. ⁴ The relatively infrequent requesting of the recommended tests for CS poses a risk of delayed or missed diagnosis. ⁴

We have created a web-based application (https://pathologyapps.shinyapps.io/CS-Screening/) to assist clinicians in selecting the most suitable tests when screening patients for CS. When the tests are performed, the Shiny app assists in result interpretation by reminding users of potential confounding factors.

The app incorporates several key messages from a recent guideline update. ² Users are reminded that synthetic glucocorticoids have the potential to cause similar symptoms as seen in endogenous CS; of note, glucocorticoid use is reportedly the most prevalent cause of CS. ¹ If exogenous glucocorticoid use is excluded, then the recommended initial tests for suspected CS include UFC, LNSC or, if the first two are not suitable, the ONDST. ²

If the LNSC test is performed, the app reminds users of the need for two LNSC tests to account for within-individual variability.^2,3,5 LNSC should not be done in patients with disruption of the normal day and night cycle, such as night-shift workers. It is noted in the web-based tool that LNSC is not recommended in patients with an incidental adrenal nodule as LNSC has lower specificity in these patients. ³

For the ONDST, serum cortisol concentration of less than 50 nmol/L at 0800 h after 1-mg dexamethasone given between 2300 h and midnight strongly predicts absence of CS. ² The CS-Screening Shiny app reminds users not to overlook several important confounding factors with DST; false-positive results might be seen with rapid absorption or malabsorption of dexamethasone due to chronic diarrhoea or coeliac disease; from concomitant treatment with CYP3A4 inducers (e.g. phenobarbital and carbamazepine) and from increased corticosteroid-binding globulin (CBG) concentrations caused by oral oestrogens or pregnancy, which can increase total cortisol concentrations.^1,2 While false-negative results are less common (due to inhibition of dexamethasone metabolism by concomitant medications such as fluoxetine or diltiazem, leading to a higher biologically available dose), these limitations are noted in the app. Decreased CBG and albumin concentrations, which can be noted in patients with concurrent nephrotic syndrome, also might produce a falsely low value.^1,2

For 24-hour UFC, at least two 24-hour urine collections are advised to account for within-patient variability, like the LNSC test.^1,2 UFC relies on accurate collection by the patient.^1,2 As urine volume and glomerular filtration rate strongly predict UFC, other screening tests such as LNSC might be preferred for patients with renal impairment (creatinine clearance <60 mL/min) or clinically significant polyuria (>5 L/24 h). ² These limitations are incorporated into the algorithm in the Shiny app. Prospective studies involving clinicians are required to determine if the web-based CS-Screening decision support tool increases detection rate and lead to better health outcome for patients subsequently treated for CS.