Glucocorticoid induced adrenal insufficiencyBMJ 2021; 374 doi: https://doi.org/10.1136/bmj.n1380 (Published 12 July 2021) Cite this as: BMJ 2021;374:n1380
- 1Institute of Metabolism and Systems Research, University of Birmingham, Birmingham, UK
- 2Division of Endocrinology, Metabolism and Nutrition, Department of Internal Medicine, Mayo Clinic, Rochester, MN 55905, USA
- Correspondence to: I Bancos
Synthetic glucocorticoids are widely used for their anti-inflammatory and immunosuppressive actions. A possible unwanted effect of glucocorticoid treatment is suppression of the hypothalamic-pituitary-adrenal axis, which can lead to adrenal insufficiency. Factors affecting the risk of glucocorticoid induced adrenal insufficiency (GI-AI) include the duration of glucocorticoid therapy, mode of administration, glucocorticoid dose and potency, concomitant drugs that interfere with glucocorticoid metabolism, and individual susceptibility. Patients with exogenous glucocorticoid use may develop features of Cushing’s syndrome and, subsequently, glucocorticoid withdrawal syndrome when the treatment is tapered down. Symptoms of glucocorticoid withdrawal can overlap with those of the underlying disorder, as well as of GI-AI. A careful approach to the glucocorticoid taper and appropriate patient counseling are needed to assure a successful taper. Glucocorticoid therapy should not be completely stopped until recovery of adrenal function is achieved. In this review, we discuss the factors affecting the risk of GI-AI, propose a regimen for the glucocorticoid taper, and make suggestions for assessment of adrenal function recovery. We also describe current gaps in the management of patients with GI-AI and make suggestions for an approach to the glucocorticoid withdrawal syndrome, chronic management of glucocorticoid therapy, and education on GI-AI for patients and providers.
Exogenous glucocorticoids can cause adrenal insufficiency (glucocorticoid induced adrenal insufficiency; GI-AI), which is among the most common causes of cortisol deficiency.1 Its true prevalence is difficult to ascertain because of the heterogeneity of studies. Moreover, a mismatch exists between the high prevalence of “biochemical” GI-AI and the reporting of its “clinical” consequences, suggesting potential under-diagnosis and lack of awareness among clinicians.2 GI-AI is one of the most dangerous adverse effects of glucocorticoid therapy and puts patients at risk of life threatening adrenal crises. Discontinuation of long term glucocorticoid therapy can be difficult; to stop treatment safely, clinicians must consider both the recovery of normal cortisol secretion and potential withdrawal symptoms. Despite widespread use of glucocorticoids, the evidence guiding the management of GI-AI originates from very heterogeneous studies in terms of design, patient population, sample size, types and regimens of glucocorticoids used, and assessment of adrenal function. Therefore, the level of evidence is low, which results in variation in management across different centers and clinicians. The aim of this review is to provide an up-to-date overview of the main aspects of GI-AI and discontinuation of glucocorticoids.
Epidemiology of glucocorticoid induced adrenal insufficiency
Systemic glucocorticoids are used by 1-3% of the general population.3456 Focusing on the long term use of oral glucocorticoids, the prevalence is 0.5-1.8%.578910 Oral glucocorticoids are more commonly used by women and older people; chronic use is reported in 1.4% of patients older than 55 years and in 3% of women over the age of 80.37810 A systematic review and meta-analysis found that the absolute risk of GI-AI induced by oral glucocorticoids was 48.7%, with the highest risk in people with hematologic malignancies (60%), followed by patients with a history of renal transplant (56.2%), inflammatory bowel disease (52.2%), and rheumatologic disorders (39.4%).11
The use of inhaled glucocorticoids has also increased over time in both children and adults,1213 and the overall risk of GI-AI in these patients is 7.8%.11 Adult patients treated with topical or nasal glucocorticoids were reported to have a lower risk of GI-AI of 4.2-4.7%.11 Patients treated with intra-articular glucocorticoids had biochemical evidence of GI-AI in 52.2% of cases; however, this estimate was based on pooled evidence from only four studies including a total of 69 patients.11
Sources and selection criteria
We searched PubMed for studies using the terms “glucocorticoid-induced adrenal insufficiency”, “steroid-induced adrenal insufficiency”, “tertiary adrenal insufficiency”, “glucocorticoid withdrawal syndrome”, and “glucocorticoid taper.” We considered all studies published in the English language between 1 January 2010 and 15 November 2020. Relevant publications outside this timeline were selected on the basis of review of the bibliography. Unless stated, we report data from studies that assessed adrenal function by dynamic testing (adrenocorticotropic hormone (ACTH) stimulation tests, corticotropin releasing hormone stimulation test, overnight metyrapone test, insulin tolerance test). We predefined the priority of study selection for this review according to the level of the evidence (systematic reviews and meta-analyses of literature and randomized controlled trials, if any), on the basis of the population of interest (studies reporting on GI-AI, with clear description of the type, dose, and duration of glucocorticoid use), the sample size (studies with larger sample size were prioritized), and the time of publication (more recent studies were prioritized).
Pathophysiology of hypothalamic-pituitary-adrenal axis suppression and recovery
Glucocorticoids that directly enter the systemic circulation and those surviving the first pass metabolism after gastrointestinal absorption exert negative feedback on the corticotropin releasing hormone producing neurons and pituitary corticotroph cells. This leads to reduced adrenal cortisol production and, after prolonged exposure, adrenal cortical hypoplasia and atrophy (fig 1, top panel).141516 After the effect of glucocorticoid therapy on the hypothalamic-pituitary-adrenal axis (HPAA) has worn off, production of ACTH and corticotropin releasing hormone typically recovers first, followed by that of cortisol, which can remain suppressed in the long term if adrenal atrophy has ensued (fig 1, bottom panel).171819
Factors affecting likelihood of glucocorticoid induced adrenal insufficiency
The risk of developing GI-AI is difficult to predict on an individual basis. The glucocorticoid formulation used, dose, and treatment duration can affect this risk, but a great degree of overlap exists.112021 Nevertheless, multiple studies have assessed the various factors that contribute to development of GI-AI (box 1), which can help clinicians to establish the risk individualized to each patient (table 1). This classification, together with clinical judgment, can be useful in deciding the management of patients exposed to exogenous glucocorticoids (fig 2 and fig 3); in particular, we focused on systemic and inhaled glucocorticoids, as their effects on the HPAA have been examined more extensively. However, it is important to note that high quality evidence is limited.
Factors affecting the risk of glucocorticoid induced adrenal insufficiency (GI-AI)
Glucocorticoid mode of administration
Systemic administration (including oral, intravenous, intramuscular)
Factors increasing the risk of GI-AI:
Daily administration for >2-4 weeks
Multiple daily split doses
Factors reducing the risk of GI-AI:
Alternate day administration
Pulse systemic therapy (intermittent intravenous administration of very high doses of glucocorticoids over a few days or weeks)
Factors increasing the risk of GI-AI:
High daily doses given for >6-12 months
Treatment with fluticasone propionate
Concomitant use of oral glucocorticoids (including intermittent use—eg, in chronic obstructive pulmonary disease)
Lower body mass index (children)
Higher compliance with treatment (children)
Factors increasing the risk of GI-AI:
Repeated injections with high doses of glucocorticoids
Factors increasing the risk of GI-AI:
Long term and frequent use of large amounts of high potency glucocorticoids
Prolonged use on inflamed skin or with impaired barrier function
Use on mucous membranes, eyelids, and scrotum
Larger body surface area to body weight ratio (children)
Glucocorticoid half life and potency
The use of long acting glucocorticoids (ie, with a longer biological half life) and formulations with higher glucocorticoid potency (ie, stronger binding to the glucocorticoid receptor) predispose to more pronounced adrenal suppression because of the continuous effect on the HPAA. This is particularly relevant for systemic administration. See table 2 for a list of widely used systemic glucocorticoids.
Higher doses of glucocorticoids correlate with increased inhibition of corticotropin releasing hormone secretion, especially when given daily over a prolonged period. The evidence for the effect of the dose of systemic glucocorticoids on manifestations of GI-AI is, however, limited. A stronger relation between the risk of GI-AI and administered doses has been described for inhaled glucocorticoids and can inform the risk of adrenal suppression (table 1 and table 3).
CYP3A4 is the major pathway for the inactivation of most prescribed glucocorticoids. CYP3A4 inhibitors are expected to increase the systemic exposure to synthetic glucocorticoids (ie, higher risk of GI-AI):
Strong inhibitors include boceprevir, ceritinib, clarithromycin, cobicistat, darunavir, idelalisib, indinavir, itraconazole, ketoconazole, lopinavir, mifepristone, nefazodone, nelfinavir, posaconazole, ritonavir, saquinavir, telaprevir, telithromycin, and voriconazole
Moderate inhibitors include amiodarone, aprepitant, cimetidine, conivaptan, crizotinib, ciclosporin, diltiazem, dronedarone, erythromycin, fluconazole, fosamprenavir, fosaprepitant, grapefruit juice, imatinib, isavuconazole, netupitant, nilotinib, ribociclib, schisandra, and verapamil
CYP3A4 inducers are expected to decrease the systemic exposure to synthetic glucocorticoids. Therefore, patients who have underlying GI-AI may develop signs and symptoms of cortisol deficiency:
Strong inducers include apalutamide, carbamazepine, enzalutamide, fosphenytoin, lumacaftor, lumacaftor-ivacaftor, mitotane, phenobarbital, phenytoin, primidone, and rifampicin
Moderate inducers include bexarotene, bosentan, cenobamate, dabrafenib, efavirenz, elagolix/estradiol/norethindrone acetate combination, eslicarbazepine, etravirine, lorlatinib, modafinil, nafcillin, pexidartinib, rifabutin, rifapentine, and St John’s wort (Hypericum perforatum)
Concomitant use of other drugs
Megestrol acetate potentially increases the risk of adrenal suppression because of its glucocorticoid activity.
High dose medroxyprogesterone acetate can inhibit release of adrenocorticotropic hormone and exerts weak glucocorticoid activity
CYP3A4=cytochrome P450 3A4; HPAA=hypothalamic-pituitary-adrenal axis
Systemic glucocorticoid therapy
Systemic glucocorticoids, especially those with longer half life and higher potency, are more likely to cause GI-AI compared with other administration routes because of the direct negative feedback on the HPAA (box 1 and table 2).22 Treatment with bedtime glucocorticoid administration and multiple split doses is more likely to affect circadian ACTH release and cause GI-AI.232425 On the other hand, alternate day administration, as well as pulse therapy with high dose systemic glucocorticoids, is associated with a lower risk of GI-AI because the HPAA has more time to recover (box 1).232627282930
Several studies have suggested a direct relation between the dose and duration of systemic glucocorticoid therapy and the likelihood of GI-AI, although the evidence is limited.223313233 A recent analysis of medical records from primary and secondary care in England found that the incidence of GI-AI and iatrogenic Cushing’s syndrome among chronic users of oral glucocorticoids (mainly prednisolone) increased with higher daily and cumulative doses.2 Although these data provide a good population based estimate of the risk of GI-AI, establishing this risk at the individual level is difficult.113233343536
Short courses (<2 weeks) of high dose glucocorticoids are usually associated with a swift recovery of adrenal function,183738 although cases of GI-AI lasting more than seven to 14 days have been described.31394041424344 The possible exception to this is in patients who receive frequent short glucocorticoid courses, as in the treatment of asthma or chronic obstructive pulmonary disease and during emetogenic chemotherapy. Of 37 healthy people given prednisone at doses ranging from 2.5 mg to 60 mg over seven days on three occasions, separated by 14 day washout periods, two developed GI-AI, lasting as long as four months in one of them.43
In a prospective study of 311 patients with exacerbation of chronic obstructive pulmonary disease treated with prednisolone for ≤14 days, 9%, 3%, and 2% had biochemical evidence of GI-AI when assessed at one, three, and six months after discontinuing glucocorticoids.45 Of note, a substantial number of these patients were exposed to short courses of glucocorticoid therapy for exacerbations of chronic obstructive pulmonary disease before and during the study.46 In a prospective study of patients with cancer treated with at least three short courses of high dose dexamethasone (three to four days) every two to four weeks to counteract chemotherapy induced nausea, 15% of patients developed GI-AI within three to six months.47 Patients who also received megestrol acetate had a threefold higher risk of GI-AI, possibly because of its glucocorticoid activity (box 1).4748
Children also have a significant risk of GI-AI when exposed to systemic glucocorticoids, which may be higher in those aged under 5 years.49 A retrospective study of 103 children receiving supraphysiological doses of glucocorticoids for prolonged periods of time (median >1 year) did not find an association between the duration of treatment, the cumulative glucocorticoid dose, and the risk of GI-AI.50 A systematic review of 10 studies that included 298 children with acute lymphoblastic leukemia showed that GI-AI occurred in nearly all cases treated with high dose systemic glucocorticoids for less than two months.51 Most children recovered adrenal function within a few weeks, but GI-AI persisted in 11% of patients retested at 12-34 weeks.51
Inhaled glucocorticoid therapy
Inhaled glucocorticoids can be absorbed directly from the lungs or from the gastrointestinal tract. The drug that does not reach the lungs is deposited into the mouth and pharynx, where it can be swallowed and absorbed. The portion of the drug that survives the first pass metabolism in the liver exerts systemic effects (box 1, table 3, and fig 1, top panel).525354
A systematic review and meta-analysis found that patients with asthma or chronic obstructive pulmonary disease who used only inhaled glucocorticoids had a 6.8% risk of GI-AI.11 This risk had a positive correlation with both the glucocorticoid dose (up to 18.5% risk in those using high doses) and the treatment duration (risk up to 20.3% if used for longer than one year).11 This dose-response relation was also observed in other large cohorts.5556 Several studies have also shown that patients treated with inhaled glucocorticoids who also use oral glucocorticoids are much more likely to develop GI-AI (up to 43.7%).1156 Moreover, the cumulative dose of inhaled glucocorticoids, the concomitant use of nasal glucocorticoids, higher adherence to treatment, and low body mass index in children increase the likelihood of GI-AI.57585960
Most cases of GI-AI in patients using inhaled glucocorticoids have been linked to fluticasone propionate.61626364 This drug has a long half life (14.4 hours) and a strong binding affinity for the glucocorticoid receptor (18 times that of dexamethasone), which may lead to pronounced negative feedback on the HPAA.22 Fluticasone furoate has the same steroidal backbone as fluticasone propionate but a different ester substituent and has a longer retention time in respiratory tissues.65 This may explain the lower levels of systemic absorption and the low impact on the HPAA observed in both children and adults (box 1),66676869 although East Asian people may have a higher risk of systemic exposure to the drug.69 Ciclesonide and its active metabolite desisobutyryl-ciclesonide have high concentrations of plasma protein binding (that is, reduced biological activity) and high clearance rate (that is, reduced systemic exposure)5270; this might explain the low degree of HPAA suppression observed with this compound.717273 Therefore, inhaled ciclesonide may be a safer alternative than fluticasone propionate,5274 although cases of GI-AI have been associated with its use.557576
In children and adults with asthma, most studies did not find GI-AI when inhaled glucocorticoids were used at fluticasone equivalent doses below 400 µg/day and 500 µg/day, respectively.5277787980 However, long term users of inhaled glucocorticoids (>6-12 months) can develop GI-AI even at intermediate and low doses.687581 A study in 70 children with asthma treated with inhaled glucocorticoids for at least six months found GI-AI in 24% of cases, with most being treated for more than 24 months at fluticasone equivalent doses above 400 µg/day.82
Patients with cystic fibrosis often need inhaled glucocorticoids if they develop symptoms of asthma or in the setting of allergic bronchopulmonary aspergillosis. Limited evidence suggests that they may be at particularly high risk of GI-AI.1183 Patients who need prolonged courses of systemic glucocorticoids and antifungal drugs (strong inhibitors of CYP3A4) for acute allergic bronchopulmonary aspergillosis are more susceptible to HPAA suppression.83
Intra-articular and epidural glucocorticoid therapy
Intra-articular glucocorticoid injections are commonly used for symptom relief in patients with osteoarthritis and inflammatory arthropathies. They are favored over oral glucocorticoids because of the much lower systemic absorption and risk of adverse effects. However, a proportion of injected glucocorticoid reaches the systemic circulation and can affect the HPAA (box 1). Exogenous glucocorticoids can be detected in the urine for months after the injection, pointing to possible prolonged systemic absorption.8485 Repeated injections and higher doses of glucocorticoids increase the risk of GI-AI.8687888990 Moreover, patients with inflammatory arthropathies have a higher risk of systemic glucocorticoid absorption because of its injection into a highly vascularized area (box 1).89 This is particularly relevant in the pediatric population, as inflammatory (rather than degenerative) arthropathies are more common.86
In patients treated for osteoarthritis of the knee, biochemical evidence of GI-AI was observed two to four weeks after the injection of intra-articular methylprednisolone acetate and betamethasone acetate/betamethasone sodium phosphate in 25% and 5% of people, respectively.9192 Asymptomatic GI-AI was observed in 60% of patients receiving simultaneous bilateral knee injections of methylprednisolone acetate and persisted for eight weeks in 10% of cases.93 Interestingly, a minority of these patients had intermittent abnormalities in cortisol secretion over time, suggesting possible variation in systemic absorption of the glucocorticoid from the knee joint.93
The effects of epidural glucocorticoids on the HPAA have been assessed in a few small studies. Eight people receiving a single epidural injection of triamcinolone had a significant reduction in 24 hour urinary cortisol after 12 weeks, as well as a blunted response of ACTH and cortisol to corticotropin releasing hormone stimulation.94 Another study showed that 35% of patients receiving multiple epidural injections of triamcinolone acetate had GI-AI, which resolved within three months.95 This was also observed after injections of methylprednisolone acetate,9697 with patients receiving higher doses (80 mg) being at higher risk of GI-AI.97
Topical and rectal glucocorticoid therapy
The evidence of an effect of topical glucocorticoids on the HPAA is mostly anecdotal. Cases of GI-AI have been reported after exposure to nasal, intradermal, and cutaneous glucocorticoids.629899100101102 The pharmacokinetic and pharmacodynamic characteristics of the glucocorticoid used can influence its effect on adrenal function; however, short term administration at the recommended dosages is safe.103104 Skin inflammation, impaired barrier function due to damage, use of occlusive dressings, and the site of application affect the rate of systemic absorption in case of topical glucocorticoid use.101105 For example, percutaneous absorption is higher through mucous membranes, eyelids, and scrotum.101 The risk of GI-AI may be higher in infants because of a larger body surface area to body weight ratio (box 1).22
Rectal glucocorticoids are often used to limit systemic exposure in patients with inflammatory bowel diseases affecting the lower part of the colon and rectum. Although short term use is safe, systemic glucocorticoid absorption is possible.106 GI-AI has been described in users of prednisolone enemas.107 Rectal budesonide and beclometasone dipropionate may be safer than other glucocorticoids108109110111; however, most of the evidence is based on single serum cortisol measurements.
Oral budesonide, frequently used to treat inflammatory bowel disease, undergoes extensive inactivation during first pass metabolism. However, the drug can be absorbed systemically and cases of adrenal crisis and Cushing’s syndrome have been reported following prolonged use.112113 Patients who are taking higher doses of oral budesonide (eg, >6 mg) for more than eight weeks and those on long term treatment (regardless of the dose used) should be considered at higher risk of GI-AI.114115116117 This is particularly relevant in patients with recent exposure to oral glucocorticoids.
Dexamethasone and, to a lesser extent, predniso(lo)ne are metabolized by the hepatic cytochrome P450 3A4 (CYP3A4). Therefore, CYP3A4 inhibitors are expected to increase systemic exposure to glucocorticoid (box 1).51118 For example, coadministration of ritonavir and prednisone increased the circulating concentrations of the active metabolite prednisolone by up to 37%.118 The newer protease inhibitor dolutegravir does not inhibit CYP3A4 and could be an alternative in patients treated with glucocorticoids.119
Concomitant treatment with strong CYP3A4 inhibitors and inhaled, intranasal, and injectable fluticasone, budesonide, and triamcinolone is not recommended because of the preferential metabolism of these drugs by CYP3A4.90118120121122123 Beclometasone and flunisolide may be safer alternatives because of their different metabolism.124125 Ritonavir has been linked to many cases of iatrogenic Cushing’s syndrome and symptomatic GI-AI when used in combination with inhaled and intranasal glucocorticoids (mostly fluticasone propionate).52103122126 This is particularly relevant because asthma and chronic obstructive pulmonary disease are more prevalent in HIV infected populations.127
Other factors affecting risk of glucocorticoid induced adrenal insufficiency
Inter-individual variability in the sensitivity of the HPAA to exogenous glucocorticoids has been observed.44128129 For example, higher cortisol after administration of dexamethasone was associated with an inadequate response to the low dose ACTH stimulation test after a 14 day course of prednisolone in healthy volunteers.44 This variation is possibly linked—among other factors—to polymorphisms of the genes encoding the glucocorticoid and mineralocorticoid receptors.14129130131132 Glucocorticoid receptor polymorphisms were reported to be associated with increased sensitivity to glucocorticoids, leading to a higher prevalence of features associated with hypercortisolism and efficacy of glucocorticoids or, by contrast, a relative glucocorticoid resistance with lower rates of development of features suggestive of iatrogenic Cushing’s syndrome.133134135136137138139140141142 Time of glucocorticoid administration during the day is another important factor that likely contributes to the risk of GI-AI. Clock related genes in circulating blood cells from patients taking exogenous glucocorticoid multiple times a day were found to be abnormal, and they partially normalized after a switch to once a day administration.143 Administration of glucocorticoid in the evening is more likely to lead to GI-AI and glucocorticoid induced side effects as glucocorticoid sensitivity is higher at that time, parallel to the clock gene down-regulation.144 The metabolism of exogenous glucocorticoids (and hence exposure of the HPAA to negative feedback) can also be affected by sex, age, body mass index, liver and kidney failure, proinflammatory cytokines, and polymorphisms of CYP enzymes and the cytochrome P450 oxidoreductase.145
Clinical relevance of glucocorticoid induced adrenal insufficiency
We showed in the previous sections that a biochemical diagnosis of GI-AI is common in patients treated with glucocorticoids. However, a fundamental question needs to be answered—is this just a biochemical abnormality, or is it clinically relevant? A crucial point to consider is whether GI-AI without apparent signs and symptoms of cortisol deficiency is clinically relevant. A review of the literature found that very few studies assessing the risk of GI-AI in patients treated with glucocorticoids also focused on symptoms of adrenal insufficiency.11 In the pooled analysis, only 2% of patients reported symptoms consistent with adrenal insufficiency but, on dynamic testing, 19% of patients failed to achieve a normal cortisol response.11 Notably, although symptom assessment in this study was not standardized, these data suggest that patients with symptoms are only the tip of the iceberg. Moreover, patients with GI-AI can often present with non-specific signs and symptoms that can be attributed to other causes. Importantly, events such as infections and surgery in a patient with undiagnosed and untreated GI-AI can trigger a life threatening adrenal crisis because of the inability to mount an adequate stress response.146
Adrenal crisis and symptomatic glucocorticoid induced adrenal insufficiency
Several cases of adrenal crisis and hospital admission for GI-AI have been described, including a few deaths.1435147148149150 Of note, many patients presenting with adrenal crisis were treated with inhaled glucocorticoids only (mostly high dose fluticasone propionate), emphasizing the potential risks associated with their systemic absorption.1455150151152153 Symptomatic GI-AI and adrenal crises have also been described in patients exposed to oral glucocorticoids for less than two months, as well as in patients receiving topical, intra-articular, epidural, and rectal glucocorticoids.148587150154155156157158 Two studies found that patients with GI-AI had a higher incidence of adrenal crisis than did patients with other causes of adrenal insufficiency.159160 The most common triggering factor was infections; of note, several patients developed an adrenal crisis after reducing or discontinuing the dose of exogenous glucocorticoids, which is a preventable cause of morbidity.160
The tapering down and discontinuation of glucocorticoids is a crucial time, as an underlying GI-AI can become clinically apparent. A Danish population based study showed that discontinuation of long term oral glucocorticoids increased the risk of developing hypotension, gastrointestinal symptoms, hypoglycemia, and hyponatremia.161 These signs, which are suggestive of untreated GI-AI, peaked up to three months after the glucocorticoids were stopped and persisted for several months; older patients taking higher daily doses of glucocorticoids for longer periods of time, as well as those exposed to infections, had the highest risk of developing symptoms.161 Intriguingly, mortality after cessation of chronic oral glucocorticoid therapy markedly increased in the first months after discontinuation, raising the suspicion of possible undiagnosed adrenal crises in this population.2 Another retrospective cohort study showed that 36 adrenal crises (7.1% of all cases) occurred within 30 days of discontinuation of glucocorticoids.152 Adrenal crises have also been reported in patients tapering down inhaled glucocorticoids or switching between different glucocorticoid preparations.153
Risks associated with inhaled glucocorticoid therapy
Cases of symptomatic GI-AI and iatrogenic Cushing’s syndrome have been linked to inhaled glucocorticoids.5262162 Most cases involved patients treated with fluticasone propionate ≥500 μg/day. A retrospective cohort study from Canada identified 392 hospital admissions due to GI-AI over a 15 year period among adults treated with inhaled glucocorticoids.59 Patients using higher daily doses (≥1000 µg/day) and cumulative doses (>157 000 µg/year) of glucocorticoids had an almost twofold higher risk of hospital admission than those with lower exposure.59 A study focusing on the medical records of general practices in the UK identified only 31 cases of established GI-AI linked to inhaled glucocorticoids in a cohort of 2.4 million people.56 Considering the widespread use of inhaled glucocorticoids, the low rates of GI-AI observed in these studies would suggest that this problem is largely unrecognized or under-reported.62
Clinical presentation of patients with hypothalamic-pituitary-adrenal axis suppression
A high degree of clinical suspicion is paramount when a history of glucocorticoid exposure exists. Patients with GI-AI may have no symptoms, present with varying degrees of symptoms of cortisol deficiency, or present with life threatening adrenal crisis (fig 4). Clinical diagnosis of GI-AI seems to be difficult, with most patients reporting multiple symptoms at the time of diagnosis and 66% of patients needing to see more than one physician to make the diagnosis of GI-AI.159 GI-AI generally has a less dramatic presentation than primary adrenal insufficiency; acute circulatory collapse and electrolyte abnormalities are infrequent because the secretion of mineralocorticoids is preserved (fig 1, top panel).114 Symptoms such as fatigue and abdominal pain are non-specific, and their onset may be insidious and mistakenly attributed to other causes or to the disease for which glucocorticoids were prescribed.14162163164 In children, stunted growth, reduced final height, and unexplained hypoglycemia can be signs of GI-AI and should prompt an assessment of adrenal function.76153165 Patients with untreated GI-AI may be more susceptible to infections and recover more slowly than those with normal adrenal function; this could be another warning sign in patients with chronic obstructive pulmonary disease and asthma treated with glucocorticoids.153
Patients presenting with signs and symptoms of cortisol excess due to glucocorticoid use (iatrogenic Cushing’s syndrome) have had marked systemic exposure and must be assumed to have a fully suppressed HPAA (fig 4). Iatrogenic Cushing’s syndrome can occur with any formulation of glucocorticoids and can take several months to resolve after the dose of glucocorticoid is decreased.87158166
Approaches to glucocorticoid taper
The goal of glucocorticoid therapy is the shortest treatment with the lowest effective dose (box 2). Glucocorticoid taper and eventual discontinuation in patients treated with exogenous glucocorticoids require consideration of several aspects of the care of patients. The first one relates to the original reason for the exogenous glucocorticoid therapy. If prescribed for a chronic condition (for example, rheumatoid arthritis, asthma, inflammatory bowel disease), this may relapse after discontinuation of glucocorticoids. A close collaboration with the medical team is needed to ensure that the patient is able to discontinue glucocorticoids and the underlying disorder is in remission or well controlled with an alternative, non-glucocorticoid therapy. Secondly, symptoms related to the glucocorticoid withdrawal should be anticipated and necessitate a careful approach. Thirdly, GI-AI should be expected and glucocorticoids should not be discontinued until recovery of the HPAA is documented.
Recommendations to reduce the risk of glucocorticoid induced adrenal insufficiency
Systemic glucocorticoid administration (including oral)
Whenever possible, use the lowest effective dose of glucocorticoids for the shortest period of time
Whenever possible, favor once daily dosing when using intermediate and long acting glucocorticoids (eg, prednisone, prednisolone, dexamethasone)
Whenever possible, avoid bedtime administration of glucocorticoids*
Do not taper down glucocorticoids if the treatment course is <2 weeks. The risk of HPAA axis suppression in such cases is low, and glucocorticoids can be discontinued abruptly. If treatment is prolonged beyond 2 weeks, the risk of HPAA suppression increases
Whenever possible, use the lowest effective dose of glucocorticoids for the shortest period of time
Use spacers and mouth rinsing to reduce systemic absorption through the gastrointestinal tract
Whenever possible, reduce the number of injections and space out the administration of intra-articular glucocorticoids†
Whenever possible, avoid simultaneous injections of multiple joints
Use the lowest effective dose of glucocorticoids
Favor triamcinolone hexacetonide over triamcinolone acetonide because of its higher residence time in the joint (lower risk of systemic absorption)
*Excluding modified release formulations (eg, modified release prednisone).
†Based on very low quality evidence and personal experience, the risk of sustained glucocorticoid induced adrenal insufficiency is higher in patients receiving multiple glucocorticoid injections, especially if over a short period of time. Patients should be considered to be at moderate to high risk of sustained glucocorticoid induced adrenal insufficiency if they receive >3 injections over the course of a year. Nevertheless, transient adrenal suppression is observed in >50% of patients for up to 2 months after intra-articular glucocorticoid injection. Health professionals should educate patients to report signs and symptoms of adrenal insufficiency, especially during the first 2 months after the injection.
Considerations for glucocorticoid taper
The best choice of glucocorticoid to use during glucocorticoid taper has not been explored. However, differences in the pharmacokinetics and potency of various glucocorticoids likely influence both glucocorticoid withdrawal syndrome (GWS) and the rapidity of HPAA recovery.167 Glucocorticoid dose and taper are not standardized, leading to significant differences in practice. Reported tapering regimens are based on very low quality evidence.168 In several studies investigating the effect of different types and dosage of glucocorticoid therapy on the duration of HPAA recovery and/or GWS, evidence was inconclusive.169170171
Approaches to the glucocorticoid taper should include consideration of the duration of supraphysiological glucocorticoid use, the average dose used over this period, clinical manifestations of iatrogenic Cushing’s syndrome, overall functional status, and the risk of relapse of the underlying disorder that was treated with glucocorticoids (box 3). Rapid taper protocols were found to be less effective than a standard taper in maintaining remission.172 The patient’s previous experience with glucocorticoid taper and GWS can also guide the aggressiveness of taper. In the absence of well designed studies evaluating approaches to glucocorticoid taper, an individualized approach is needed for each patient. As most patients are treated with either predniso(lo)ne or dexamethasone, we suggest conversion to predniso(lo)ne therapy taken once a day in the morning as soon as possible. Initial rapid glucocorticoid taper of 5-10 mg weekly increments down to 20 mg daily is usually possible for patients taking predniso(lo)ne at >20-40 mg per day. A much slower taper is subsequently needed for most patients, which includes a weekly-monthly predniso(lo)ne decrease of 1-2.5 mg to avoid severe GWS (box 3). All patients should be extensively counseled on the symptoms of GWS and provided with a plan on management of severe GWS. The management plan includes supportive measures and a recommendation to incrementally increase glucocorticoid to the most recent dose associated with symptoms that were tolerable.
Approach to glucocorticoid taper and management of glucocorticoid induced adrenal insufficiency
Patient education (see fig 2)
Education on symptoms and management of GWS
Education on the symptoms and management of GI-AI
Provider education (see fig 2)
Written recommendations in patient’s medical record and letter on the management of glucocorticoid and adrenal insufficiency
Considerations for glucocorticoid taper
Initial rapid glucocorticoid taper
For patients on an average daily glucocorticoid dose of:
o PED >40 mg daily—decrease daily glucocorticoid dose by 5-10 mg weekly increments until PED 20 mg daily
o PED 20-40 mg daily—decrease daily glucocorticoid dose by 5 mg weekly increments until PED 20 mg daily
o PED 10-20 mg daily—proceed to the slow glucocorticoid taper recommendations
For patients taking twice a day predniso(lo)ne, or evening doses, switch predniso(lo)ne to once a day, in the morning
For patients taking dexamethasone, switch to predniso(lo)ne equivalent dose
Subsequent slow glucocorticoid taper
(Note that the decrement and the frequency of glucocorticoid decrease can be adapted according to the severity of the GWS.)
For patients transitioning after the initial rapid glucocorticoid taper or patients taking PED ≤20 mg daily:
o PED 10-20 mg daily—decrease daily glucocorticoid dose by 1-2.5 mg weekly decrements until 10 mg daily. If severe GWS, consider bi-monthly decrements
o PED 5-10 mg daily—decrease by 1 mg weekly decrements until 5 mg daily. If severe GWS, consider bi-monthly or even monthly decrements
o PED 5 mg daily*—do not stop or decrease glucocorticoid until HPAA recovery has been documented. Consider switching to hydrocortisone 20 mg (15 on waking, 5 mg at noon) if HPAA recovery is delayed
Assessment of HPAA recovery (see fig 3)
Timing of reassessment—should be done only after reaching PED 5 mg or a replacement dose of hydrocortisone (eg, 15 mg in the morning, 5 mg in the afternoon) for 1-4 weeks.
Frequency of reassessment—every 2-3 months. If HPAA has not recovered for 1-2 years, consider reassessment every 3-6 months.
Testing—assessment should be done in the morning, 24 hours off glucocorticoid therapy with the following tests:
Morning serum cortisol
Additional tests to consider (only if indeterminate morning cortisol results)
o Synthetic ACTH stimulation test
o Overnight metyrapone test (where available)
o Insulin tolerance test (very rarely)
Management of patients with established GC-AI (see fig 2)
Periodic clinical assessment for symptoms and signs of glucocorticoid over-replacement and under-replacement
Periodic biochemical assessment of HPAA recovery
Assure patient has optimal supplies (oral glucocorticoid, injectables)
Patient counseling on glucocorticoid therapy, special situations (shift work, pregnancy, concomitant diabetes mellitus, excessive physical activity), sick day rules
Chronic glucocorticoid replacement therapy:
o Hydrocortisone (eg, 15 mg on waking, 5 mg 6-8 hours later), or
o Predniso(lo)ne 4-5 mg on waking
o See “Emergent treatments” section for novel preparations
Sick day rules:
o Sick day rule 1—double or triple daily oral glucocorticoid dose during illness that requires bed rest and/or antibiotics or is associated with high fever (>38°C) until recovery. Double or triple the dose on the day of any minor procedure/surgery that does not require fasting (eg, procedures requiring local anesthesia and tooth extraction)
o Sick day rule 2†—self-inject glucocorticoid when unable to take oral glucocorticoid (eg, acute gastroenteritis), during severe illness, after major trauma, and if the patient is confused, drowsy, or unconscious. Also, self-inject glucocorticoid in preparation for surgery and procedures requiring general anesthesia, during prolonged fasting (eg, preparation for colonoscopy), and at the onset of adrenal crisis. Examples of recommended injectable doses in adults: hydrocortisone 50 mg every 6 hours, dexamethasone 4 mg/24 hours, methylprednisolone 40 mg/12 hours. If the patient is admitted to hospital, continuous hydrocortisone infusion should be favored over intermittent bolus administration
Wear medical alert bracelet/necklace
Carry steroid card/letter outlining GC-AI management
Become comfortable with GC-AI management
Practice glucocorticoid injections
Communicate with other members of the medical team with regard to GC-AI
ACTH=adrenocorticotropic hormone; HPAA=hypothalamic-pituitary-adrenal axis; GC-AI=glucocorticoid induced adrenal insufficiency; GWS=glucocorticoid withdrawal syndrome; PED=prednisolone equivalent dose.
*Further decrease of glucocorticoid dose <5 mg PED can be considered in a subset of patients to accelerate the recovery of adrenal function (ie, 4 mg, or even 3 mg, PED). By contrast, a subset of patients may achieve recovery of adrenal function at higher PED (ie, 6-7.5 mg PED).
†All patients with GI-AI and their family members/partners should receive appropriate education and a demonstration of glucocorticoid injection.
Glucocorticoid withdrawal syndrome
GWS is a withdrawal reaction that may occur during the glucocorticoid taper owing to developed dependence on supraphysiological glucocorticoid concentrations.173 The pathophysiology of GWS is multifactorial and is likely mediated by suppressed corticotropin releasing hormone, central noradrenergic and dopaminergic system, decrease in pro-opiomelanocortin related peptides due to chronic suppression of HPAA, and the increase in cytokines (interleukin 6, tumor necrosis factor α) and prostaglandins.173174175 Patients who taper glucocorticoids often feel unwell even when still taking doses that are supraphysiological but lower than their usual glucocorticoid doses. Symptoms may include anorexia, nausea, emesis, lethargy, somnolence, arthralgia, myalgia, low grade fever, postural hypotension, and psychiatric symptoms (mainly depression, anxiety, and panic attacks).173176 GWS may last months, and patients report poor quality of life and wellbeing.170176 GWS is under-recognized as a separate entity, partly because of overlap of GWS symptoms with those of adrenal insufficiency and possibly the underlying disorder glucocorticoids were initially prescribed to treat. When physicians and patients do not recognize GWS as a separate entity, the recovery process can be difficult and prolonged, especially if glucocorticoid doses are increased unnecessarily. In the published evidence so far, very little emphasis has been placed on assessment of the severity of GWS during glucocorticoid taper. In a retrospective study of patients undergoing curative surgery for endogenous hypercortisolism, GWS was reported in 67% of patients and was associated with the severity of cortisol excess before surgery.170
Assessment of hypothalamic-pituitary-adrenal axis
The recovery of the HPAA should be assessed only in patients taking a prednisolone equivalent dose of no higher than 5 mg a day. HPAA recovery is likely to occur more rapidly if the patient is switched to hydrocortisone (short acting glucocorticoid) as opposed to predniso(lo)ne (box 1); however, head to head studies have not been performed. HPAA recovery is unlikely in patients treated with long acting glucocorticoids, such as dexamethasone. In a patient who is undergoing glucocorticoid taper, our approach is to assess HPAA recovery once the patient is treated with prednisolone equivalent doses of 5 mg for at least one to four weeks and reassess every three months until recovery occurs. In patients with GI-AI lasting for more than one to two years, reassessment can be spanned out to every six to 12 months. Several approaches to HPAA assessment can be used, all done after 24 hours without any glucocorticoid therapy other than dexamethasone. These most commonly include measurement of morning cortisol concentrations, synthetic ACTH stimulation tests, and, less commonly, the overnight metyrapone test and the insulin tolerance test (fig 3).
In a recent study of patients recovering from endogenous hypercortisolism, a morning cortisol ≥270 nmol/L (10 µg/dL) was associated with no reported symptoms of GWS or instances of adrenal crisis and thus could serve as a simple approach to HPAA assessment.170 Another large study of patients who had synthetic ACTH stimulation testing to assess for GI-AI also suggested that morning cortisol measurements can replace the dynamic testing, with cortisol cut-offs of 221 and 347 nmol/L showing 86% and 95% sensitivity, respectively.55
Notably, accurate interpretation of cortisol concentrations can be challenging. Morning cortisol concentrations vary according to the individual circadian rhythm and assays used for measurements. In addition, total cortisol concentrations can be elevated during pregnancy and in patients using oral estrogens, owing to higher concentrations of cortisol binding globulin.177 By contrast, total cortisol concentrations can be decreased in patients with low albumin and cortisol binding globulin, as in cirrhosis or critical illness.178179 In addition, cortisol concentrations vary depending on the assays used. For example, in a large study of patients undergoing synthetic ACTH stimulation testing, basal cortisol that excluded adrenal insufficiency varied between 336 and 506 nmol/L when measured by three different immunoassays.180 These special circumstances and assay variability need to be considered when interpreting cortisol cut-offs (fig 3).
The insulin tolerance test is traditionally considered the reference standard dynamic test for secondary adrenal insufficiency but is rarely needed for the diagnosis of GI-AI.181182183 Both high dose and low dose ACTH stimulation testing can be used in making the diagnosis of adrenal insufficiency.184 However, the performance of both tests is suboptimal in secondary adrenal insufficiency, with a high rate of false negative results (sensitivity of 64-83%).185 The overnight metyrapone stimulation test can also be used to assess the HPAA. Metyrapone inhibits the conversion of 11-deoxycortisol to cortisol, with subsequent stimulation of ACTH and accumulation of 11-deoxycortisol. The overnight metyrapone stimulation test was reported to have a similar performance to the insulin tolerance test and a superior performance to the ACTH stimulation tests.186187188 Use of the overnight metyrapone is limited by the availability of metyrapone and 11-deoxycortisol assays.
Management of glucocorticoid induced adrenal insufficiency
HPAA recovery in patients with GI-AI can take months to years.20 Patients taking higher glucocorticoid doses for a longer duration, those with a lower body mass index, and those with features of iatrogenic Cushing’s syndrome are likely to take longer to recover.170 A proportion of patients may develop chronic or permanent GI-AI. As reported in a systematic review of GI-AI, 15% of patients may have persistent GI-AI when reassessed three years after discontinuation of supraphysiological glucocorticoid therapy, and 2-7% of patients may never recover adrenal function.20189190
In a recent survey study that included 179 patients with GI-AI, patients reported a median age at diagnosis of 54 years and a median duration of GI-AI of three years.159 One in five patients with GI-AI reported not being aware of the diagnosis. Notably, compared with other causes of adrenal insufficiency, patients with GI-AI were more likely to report difficulty with management and less likely to have available injectable glucocorticoid at home or to wear medical alert gear. Despite taking a higher daily glucocorticoid dose, more patients with GI-AI reported symptoms due to under-replacement (fatigue, nausea). A higher number of patients with GI-AI needed stress dose steroids. When seeking care in the emergency setting, patients with GI-AI were around two times less likely to receive proper care than those with other causes of adrenal insufficiency. Overall, 59% of patients with GI-AI reported poor general health.159 These data reflect the fact that patients with GI-AI are less likely to receive adequate education on adrenal insufficiency than other patients, possibly owing to lack of awareness and the belief that the problem is temporary.
All patients at risk of or with established GI-AI should receive adequate education (fig 2). Special attention should be paid to increasing patient awareness (including over-the-counter drugs containing glucocorticoids and intra-articular glucocorticoid injections), measures targeted toward prevention of adrenal crisis, and the sick day rules (box 3). This includes counseling on glucocorticoid therapy, special situations that may require adaptation of glucocorticoid therapy (box 3), use of injectable glucocorticoid therapy, and signs and symptoms of over-replacement and under-replacement, including those of adrenal crisis (table 2). Patients with GI-AI should receive adequate education regardless of the projected duration of adrenal insufficiency.
Chronic glucocorticoid therapy most commonly consists of either twice daily hydrocortisone or predniso(lo)ne once in the morning. In the absence of a reliable biomarker, periodic clinical assessment will ensure that the dose is individualized to the patient’s needs. Guidelines suggest twice or thrice daily hydrocortisone as a valid replacement scheme in patients with adrenal insufficiency to mimic the physiological cortisol rhythm.191 This is also an appropriate option for GI-AI and should especially be considered when patients complain of glucocorticoid withdrawal symptoms; however, we postulate that avoiding late afternoon exposure to hydrocortisone by twice daily administration might accelerate the recovery of ACTH secretion by negative feedback.
In adult patients with adrenal crisis in a hospital setting, hydrocortisone 100 mg bolus injection followed by the intravenous infusion of 100 mg hydrocortisone per 24 hours is the preferable mode of treatment.192 Children with adrenal crisis should be treated with a lower hydrocortisone dose depending on their weight, usually 25-50 mg bolus injection followed by infusion of hydrocortisone 50-100 mg per 24 hours. When clinical improvement is seen, hydrocortisone should be converted to oral administration at an increased dose (usually double the usual daily dose) and gradually tapered to the usual regimen.191193194195
The HPAA and the peripheral sensitivity to glucocorticoids follow a circadian rhythm, which can affect the negative feedback response.196197 Novel modified release glucocorticoids have been designed as either immunosuppressive therapy to mirror the morning cortisol rise and improve clinical outcomes (for example, modified release prednisone in patients with rheumatoid arthritis) or replacement therapy to reproduce the circadian cortisol rhythm in patients with established adrenal insufficiency (for example, modified release hydrocortisone).197198 Potentially these modified release formulations could reduce the risk of adrenal insufficiency and promote recovery of adrenal function in patients exposed to glucocorticoids, but further research in this area is needed. No interventional trials in patients with GI-AI are ongoing.
Clear guidance exists about the prevention and emergency management of patients with confirmed or suspected adrenal insufficiency, including patient education, sick day rules, and management of adrenal crisis.21194 To the best of our knowledge, however, guidelines specifically focusing on GI-AI are lacking. Considering the widespread use of glucocorticoids in different doses and formulations for a vast number of conditions across multiple specialties, the fact that the recommendations of several medical societies do not fully cover or differ in terms of who is at risk of GI-AI, when glucocorticoid therapy can be discontinued, and how to do so comes as no surprise.191194 Although the decision to stop glucocorticoids is clearly linked to the underlying disease, the lack of a multidisciplinary consensus on the stratification of patients according to their risk of developing GI-AI and the management of glucocorticoid taper and GI-AI is likely associated with suboptimal outcomes.
Patients exposed to exogenous glucocorticoids are at risk of developing GI-AI, which needs to be promptly recognized, anticipated, and optimally managed. Patients treated with high doses of glucocorticoids for a longer duration may develop severe GWS when the treatment is tapered down. Individualized glucocorticoid taper can help with symptoms of glucocorticoid withdrawal and a quicker eventual recovery of adrenal function. Every patient with GI-AI should receive optimal education on management of adrenal insufficiency, including management of glucocorticoid therapy during an illness. Guidelines need to be developed to cover gaps in the management of patients treated with chronic glucocorticoids to raise awareness of GI-AI and GWS and provide clear recommendations that can be applied in the interdisciplinary setting.
What are the best predictors of glucocorticoid induced adrenal insufficiency and its duration?
What is the best approach to glucocorticoid taper to alleviate glucocorticoid withdrawal syndrome while achieving the quickest recovery of adrenal function?
What is the best approach to assessing the hypothalamic-pituitary-adrenal axis?
What strategies can be developed to increase awareness of glucocorticoid induced adrenal insufficiency among patients and health professionals?
We thank Wiebke Arlt and William Young for their critical review and useful suggestions on tables and figures.
Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors
Contributors: IB and AP conceptualized and performed the literature review for this manuscript. AP and IB drafted the manuscript. IB and AP contributed to the critical revision of the article. AP and IB gave final approval of the version to be published. IB is the guarantor.
Funding: AP is a Diabetes UK Sir George Alberti research training fellow (grant reference number 18/0005782). This research was supported by the Catalyst Award for Advancing in Academics from the Mayo Clinic (to IB) and the National Institute of Diabetes and Digestive and Kidney Diseases of the NIH under award K23DK121888 (to IB). The views expressed are those of the authors and not necessarily those of the NIH. The funders of the study had no role in the conceptualization, writing, and decision to publish the manuscript. All researchers were independent from funders.
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: IB is a consultant to CinCor, Corcept, Sparrow Pharmaceutics, Strongbridge, and HRA Pharma, outside this work. IB has received research support from HRA Pharma for an investigator initiated study outside this work.
Patient and public involvement: Patients or the public were not involved in the design, conduct, reporting, or dissemination plans for this work.
Provenance and peer review: Commissioned; externally peer reviewed.