New guidelines on glucocorticoid-induced adrenal insufficiency: the end of short synacthen test in rheumatology?

Glucocorticoids (GCs) are widely employed in the management of inflammatory rheumatic diseases, particularly rheumatoid arthritis (RA), polymyalgia rheumatica, giant cell arthritis and systemic lupus erythematosus.1–3 The timing and approach for discontinuing long-term GC treatment remain a significant issue for rheumatologists in terms of disease control and also safety concerns related to the well-known risk of GC-induced adrenal insufficiency (GIAI).4–8 For instance, although the EULAR recommendations for RA management advocate for rapid GC tapering (aiming for discontinuation within about 3 months of initiation), recent data from the international TOCERRA and PANABA observational studies indicate that 40% of the 12 334 participants with RA continued oral GC therapy 1 year after initiating biological disease-modifying antirheumatic drugs, with a median time to GC discontinuation exceeding 2 years.2 9 In practice, Delteil et al reported two distinctive trajectories of GC discontinuation in patients with RA.10 One-quarter of them achieved GC discontinuation within the first year, while the remaining three-quarters had a more progressive GC tapering with a low-dose maintained at 2 years.10 These observations underscore the challenges associated with discontinuing GCs, which stem from inadequate disease control and safety concerns, including the risk of GIAI. Sagar et al reported that 43% of long-term GC-treated patients with rheumatic diseases exhibited evidence of possible GIAI.11 Comparable rates were observed in patients with RA treated with 5–7.5 mg/day of prednisolone for 3–6 months,12 13 and in those with polymyalgia rheumatica/giant cell arthritis.14 Furthermore, 13% of patients with systemic lupus erythematosus who had either discontinued prednisolone or were receiving <5 mg/day showed biological evidence of adrenal insufficiency.15 Finally, overt adrenal insufficiency was identified in 20% of patients with rheumatic and musculoskeletal diseases who experienced disease relapse during GC tapering.16

Prolonged administration of GCs at supraphysiological levels suppresses the hypothalamic-pituitary-adrenal (HPA) axis by inhibiting the production of corticotropin-releasing hormone by the hypothalamus and adrenocorticotropic hormone (ACTH) by corticotropic cells in the anterior pituitary, ultimately leading to adrenal cortex atrophy. Incomplete recovery of the HPA axis following GC discontinuation may result in adrenal insufficiency associated with severe clinical outcomes, such as adrenal crisis and increased mortality.17 18 Therefore, clinicians must maintain a high index of suspicion in patients at risk of GIAI to facilitate rapid diagnosis and appropriate management. Assessing this risk is particularly challenging due to significant interindividual variation in HPA axis recovery after GC tapering or withdrawal. Key risk factors to consider include GC potency, route of administration, dose, duration of treatment and patient age.18 19 Genetic polymorphisms also play a role, although to a lesser extent, as they are not routinely assessed in clinical practice.18 19

Guidelines on the diagnosis and management of GIAI were recently published for the first time by the European Society of Endocrinology (ESE) and the Endocrine Society (ES).20 21 The experts recommend that the risk for adrenal insufficiency should be considered for GC treatment durations exceeding 3–4 weeks and doses greater than the physiological levels (ie, equivalent of 15–25 mg/day hydrocortisone or 4–6 mg/day prednisone). Figure 1 summarises the two proposed approaches for managing the risk of adrenal insufficiency during the discontinuation of GC therapy. The initial step in both approaches involves tapering GC doses to physiological replacement levels, following a switch from long-acting GCs (eg, dexamethasone or betamethasone) to short-acting GCs (eg, prednisone or hydrocortisone) when applicable. Interestingly, the recent STAR study demonstrated that tapering prednisone by 1 mg/day each month or replacing prednisone with 20 mg/day of hydrocortisone for 3 months, followed by a reduction to 10 mg/day for an additional 3 months, resulted in comparable rates of successful GC withdrawal at 1 year in patients with RA.22 Once tapering of GC doses to physiological levels has been achieved, the ESE/ES experts propose two options: either gradually tapering GCs while monitoring for clinical signs of adrenal insufficiency (option 1), or, more cautiously, assessing morning serum cortisol levels at least 24 hours after the last GC dose (option 2). To date, no studies have shown the superiority of either approach. Physicians must keep in mind that, in addition to implementing an appropriate GC discontinuation scheme, preventing adrenal crises necessitates patient education on sick-day rules, emphasising the importance of temporarily increasing GC doses during periods of physiological stress or surgical procedures.23

We believe that option 1 proposed by the ESE/ES experts has significant limitations. The guidelines lack precise recommendations on how and when to monitor for clinical signs of adrenal insufficiency, and these signs—such as fatigue, nausea and muscle or joint pain—can overlap with symptoms of underlying rheumatic diseases, complicating the clinical assessment. In option 2, if the morning serum cortisol is >300 nmol/L (>10 µg/dL), GC therapy can be discontinued. Conversely, if the level is <150 nmol/L (<5 µg/dL), patients are likely to have persistent adrenal insufficiency, and GC replacement should be continued at a physiological dose, with reassessment after a few months (between 1 and 6 months).24 For serum cortisol levels between 150 and 300 nmol/L (5 and 10 µg/dL, respectively), HPA axis recovery is possible but uncertain. In such cases, the experts recommend continuing GC replacement at a physiological dose, with a repeat morning serum cortisol measurement after a few weeks until recovery occurs. The ESE/ES experts propose that waking salivary cortisone or cortisol may represent a promising alternative to serum cortisol, offering the advantage of home-based sampling.25 It is important to emphasise that caution should be exercised when applying such an algorithm based on cortisol levels in biological samples, due to the variability inherent in different assay methods.

One of the most notable recommendations concerns dynamic tests to assess HPA function. The experts recommend against the routine use of dynamic testing to assess HPA axis recovery in patients tapering or discontinuing GC therapy. Dynamic testing includes the 250 µg ACTH1–24 stimulation test, also known as the short synacthen test (SST), and, less frequently, the more labour-intensive insulin-induced hypoglycaemia and metyrapone stimulation tests. The experts suggest that dynamic testing is reserved for patients with an intermediate morning serum cortisol between 150 and 300 nmol/L (5 and 10 µg/dL, respectively).

It should be noted that the quality of evidence supporting the ESE/ES recommendation against routinely performing dynamic tests to assess HPA axis recovery in patients tapering or discontinuing GC therapy was rated as very low by the experts.20 21 Indeed, the existing literature is limited, with only a few retrospective studies providing comparative data between baseline serum cortisol and SST, none of which accounted for clinical signs of adrenal insufficiency. Sagar et al found a strong correlation between 9 a.m. cortisol levels and the response to SST, defined as a 30 min cortisol level >450 nmol/L (>16.3 µg/dL), in 238 patients evaluated for GC-induced adrenal insufficiency risk (cortisol measured on the Siemens Advia Centaur analyser).11 Although all patients with 9 a.m. serum cortisol levels >350 nmol/L (>12.6 µg/dL) passed SST, 20% of patients with cortisol levels >250 nmol/L (>9.1 µg/dL) exhibited an insufficient SST response.11 In the study by Sbardella et al, a morning cortisol (assessed before 10 a.m. and measured on the Advia Centaur analyser) threshold of 185 nmol/L (6.7 µg/dL) predicted failure in SST response in 1019 patients with 95% sensitivity, while a cut-off of 358 nmol/L (13.0 µg/dL) had 100% specificity for predicting a normal SST response.26 Debono et al reported that a baseline serum cortisol level <152 nmol/L (<5.5 µg/dL), measured using the Roche Elecsys analyser, had a positive predictive value of 91% and a specificity of 95% for predicting failure in SST response (ie, 30 min cortisol <430 nmol/L or <15.6 µg/dL). A cut-off of 310 nmol/L (11.2 µg/dL) had a negative predictive value of 93% and a sensitivity of 96% for excluding GC-induced adrenal insufficiency .25 More recently, Eng et al reported that a baseline cortisol level >300 nmol/L (>10.9 µg/dL) predicted SST response with 93% sensitivity, while a baseline cortisol level <100 nmol/L (<3.6 µg/dL) confirmed adrenal insufficiency with 97.3% specificity.27 Combining the results of a previous SST with those of a new morning serum cortisol measurement has been shown to predict the results of a new SST with greater accuracy than relying on the new morning cortisol measurement alone.28

The ESE/ES panel highlights the risk of misinterpretation while strictly adhering to the proposed rule-in and rule-out cortisol thresholds. Instead, they recommend viewing cortisol levels as a continuum when assessing HPA axis recovery. This caution is especially warranted due to the substantial variability between assay methods. As a result, the diagnostic performance of recommended thresholds, which are based on studies using specific methods, may differ significantly when applied to other assay techniques. In this context, the SST offers a notable advantage by measuring the dynamic response in cortisol levels between baseline and post-synacthen administration, rather than relying on a static cortisol level, which can be difficult to compare to unstandardised thresholds. A difference in cortisol levels >100 nmol/L (>3.6 µg/dL) between baseline and 30 min post-synacthen administration has been shown to outperform baseline cortisol levels in predicting future HPA axis recovery in 110 GC-exposed patients with an initial failed SST.29

In addition, the SST may provide valuable insights into conditions where cortisol-binding globulin (CBG) levels are altered, resulting in serum total cortisol to not accurately reflect biologically active free cortisol. Pregnancy and oral oestrogen use are the most common conditions associated with elevated CBG and total serum cortisol, which may lead to an underestimation of GC-induced adrenal insufficiency risk.30 Although not specifically developed in the setting of GC tapering, previous guidelines from the Endocrine Society recommend SST as the test of choice for suspected adrenal insufficiency.31 However, cautions must be exercised by applying trimester-specific cut-offs for SST, which account for the physiological rise in cortisol levels during pregnancy .32 Conversely, liver failure, nephrotic syndrome and malnutrition are associated with reduced CBG and total serum cortisol, potentially leading to overestimation of adrenal insufficiency risk. Finally, the SST may be particularly useful in conditions that disrupt circadian rhythm, such as night shift work, jet lag and severe insomnia, where morning serum cortisol levels may appear falsely low.33 34

In conclusion, the retrospective design of the studies comparing morning cortisol with SST and the absence of direct head-to-head comparison studies between these tests and other dynamic tests such as the insulin tolerance test or metyrapone test represent significant limitations to draw definitive conclusions regarding the routine use of SST for assessing HPA axis recovery in patients tapering or discontinuing GC therapy, as acknowledged by the ESE/ES experts themselves. Furthermore, additional studies are required to evaluate the diagnostic performance of the SST by incorporating clinical signs of adrenal insufficiency as a criterion for judgement, rather than relying solely on biological endpoints. The ongoing placebo-controlled clinical trial Taper Or Abrupt Steroid Stop (ClinicalTrials.gov ID NCT03153527) is expected to provide valuable insights into the performance of both the SST and baseline cortisol measurements for predicting GC-induced adrenal insufficiency in patients initiating a systemic GC discontinuation, with the monitoring of clinical outcomes such as adrenal crisis, hospitalisations and mortality. Finally, the REPLACE study (EudraCT ID 2020-006121-65) will likely provide interesting findings on the association between response to SST and the development of adrenal insufficiency symptoms in patients with polymyalgia rheumatica and with biochemical evidence of mild-to-moderate adrenal insufficiency after cessation of GC treatment.