Prevalence of hypothalamic-pituitary-adrenal axis suppression in children treated for asthma with inhaled corticosteroid

Abstract

OBJECTIVE:

To determine the prevalence of hypothalamic-pituitary-adrenal (HPA) axis suppression in asthmatic children on inhaled corticosteroids (ICS).

METHODS:

Clinical and demographic variables were recorded on preconstructed, standardized forms. HPA axis suppression was measured by morning serum cortisol levels and confirmed by low-dose adrenocorticotropic hormone stimulation testing.

RESULTS:

In total, 214 children participated. Twenty children (9.3%, 95% CI 5.3% to 13.4%) had HPA axis suppression. Odds of HPA axis suppression increased with ICS dose (OR 1.005, 95% CI 1.003 to 1.009, P<0.001). All children with HPA axis suppression were on a medium or lower dose of ICS for their age (200 μg/day to 500 μg/day). HPA axis suppression was not predicted by drug type, dose duration, concomitant use of long-acting beta-agonist or nasal steroid, or clinical features.

CONCLUSION:

Laboratory evidence of HPA axis suppression exists in children taking ICS for asthma. Children should be regularly screened for the presence of HPA axis suppression when treated with high-dose ICS (>500 μg/day). Consideration should be given to screening children on medium-dose ICS.

Current clinical practice guidelines and consensus statements recommend daily inhaled corticosteroids (ICS) for treatment of nonintermittent asthma in children (1–4). ICS use is considered safe; however, high-dose use in children is met with some caution (5–7). Various drug factors affect the systemic availability of ICS, including particle size, per cent lung deposition, oral bioavailability and serum protein binding, and, therefore, the potential for adverse effects (8,9).

Hypothalamic-pituitary-adrenal (HPA) axis suppression is among the most important potential adverse effect of ICS, the mechanism of which is well defined. Through negative feedback, exogenous cortisol has inhibitory effects at the level of the hypothalamus, the pituitary gland and, to some extent, the adrenal gland itself (10,11). With prolonged suppression, endogenous cortisol release is reduced and the adrenal glands can atrophy, leading to a functional deficiency in the HPA axis. In cases where the HPA axis is unknowingly suppressed, abrupt withdrawal of steroid treatment could lead to adrenal crisis. Perhaps more worrisome is the child with unknown HPA axis suppression who undergoes significant stress (eg, illness, surgery, accident). Additionally, if the HPA axis response is severely limited there is risk of coma and death (12).

The effects of ICS on the HPA axis have been reviewed extensively (7,9,13). In one meta-analysis of 27 studies (7) HPA axis suppression was consistently found in children and adults on high doses of ICS (>750 μg/day fluticasone or dose equivalent). A more recent review including only long-term data from randomized controlled trials in children failed to find any evidence of HPA axis suppression (13). However, the majority of studies reviewed lacked dynamic tests of adrenal function, such as adrenocorticotropic hormone (ACTH) stimulation testing. Of the three studies using these tests, none used a low-dose ACTH stimulation test, which is possibly a more sensitive measure of adrenal function (7).

The occurrence of adrenal crisis in children is not uncommon. High doses of ICS had a 4.7% occurrence rate of adrenal crisis in the only published report (12). In that study, paediatricians and endocrinologists were surveyed for cases of adrenal crisis associated with treatment of asthma that were confirmed by abnormal HPA axis function tests. At the time of diagnosis of acute adrenal crisis, the mean dose of ICS was 980 μg/day fluticasone in children, with a mean duration of 1.7 years. The authors acknowledge that a key limitation of their study was the low response rate to their survey. Although this could indicate an overestimate of the proportion of children with adrenal crisis (due to the response bias of noncases), it is more likely that cases were under-reported or under-recognized, leading to a rather conservative measurement of prevalence. Certainly, it is possible that a larger percentage of those children on ICS had HPA axis suppression, because not all of those with HPA axis suppression develop acute adrenal crisis. Based on these reviews, clinical practice guidelines recommend monitoring of children receiving ≥500 μg/day of inhaled fluticasone (or equivalent doses of other ICS) (14). Additionally, parents should be informed about the potential need for additional steroid therapy during times of stress, such as illness and surgery (14).

One recently published study has evaluated the prevalence of HPA axis suppression in children receiving low to moderate doses of budesonide ICS (15). Metyrapone testing was performed on 26 children with asthma, and 35% were found to have evidence of suppression. Factors influencing HPA axis suppression included ICS dose, body size and use of nasal steroid. Although metyrapone testing is considered a gold standard test for HPA axis suppression (16), it is not readily available in most centres and is not practical for use as a screening test. Additionally, due to the small number of children in this trial and the fact that only one type of ICS was being studied, more trials are needed to confirm these findings and to explore additional factors involved in HPA axis suppression. Therefore, we sought to determine the prevalence of HPA axis suppression as measured by morning plasma cortisol levels and confirmed by low-dose ACTH stimulation testing, in children (eight months to 18 years of age) who were being treated with any ICS for longer than three months in a paediatric asthma clinic.

METHODS

Study design and population

A convenience sample of children, eight months to 18 years of age, being treated in the regional paediatric asthma centre (Orillia, Ontario) was selected. The centre treats approximately 700 children with asthma. In a recent audit of the centre (17), it was observed that of 267 new patients (mean age 5.5 years), the mode asthma severity was ‘moderate persistent’ (1) and the majority received ICS (96%).

Parents of children being treated in the clinic were approached for inclusion in the study over a nine-month period (September 9, 2010 to June 6, 2011). All patients were considered for cortisol testing based on previous concerns about HPA axis suppression. Children not on a stable dose of ICS for a period greater than three months and children requiring oral steroids in the previous three months were excluded from the study.

Ethics

The present study was granted ethical approval from the governing body at Orillia Soldiers’ Memorial Hospital (Orillia, Ontario), the Clinical Research Ethics Committee of the Cork Teaching Hospitals (Cork, Ireland) and passed independent ethical review by Institutional Review Board Services (Aurora, Ontario). A member of the research team obtained written and informed consent and assent from parents and patients before participation in the study.

Protocol

Eligibility was determined by the certified asthma educator at the clinic and confirmed by the paediatrician on duty. Baseline demographic and relevant clinical data were recorded in a preconstructed, standardized form. Collected information included: age, sex, weight, height, racial background, ICS (type, dose, frequency, route), other medication (long-acting beta-agonist or inhaled nasal corticosteroid) and symptoms compatible with HPA axis suppression according to clinical observation of the paediatrician. These features included presence or absence of anorexia, malaise, nausea, morning headache, poor weight gain, hypoglycemia, abdominal pain, fatigue, weakness, myalgia, growth suppression and subjective clinical impression of suspected HPA axis suppression. This last variable was included to evaluate the effect of physician intuition on detecting cases of HPA axis suppression. ICS dose was interpreted as low, medium or high, as defined by the Canadian Thoracic Society Guidelines (4). In children younger than six years of age, who are not defined in the guidelines, dose categorization was interpreted using the criteria from the youngest age group.

Cortisol testing

A standardized testing protocol was kindly provided by the Children’s Hospital of Eastern Ontario (18). Children were instructed to have their morning serum cortisol level measured at 08:00. If low cortisol levels were detected, a follow-up low-dose (1 μg by intravenous [IV] bolus) synthetic ACTH (Cortrosyn, Amphastar Pharmaceuticals Inc, USA) test was scheduled on an inpatient basis. The cut-off for low morning serum cortisol level is a laboratory-specific value, which is 170 nmol/L at the centre, and is equivalent to that used in the regional protocol provided. The low-dose ACTH stimulation test involves baseline measurements followed by plasma cortisol measurements at 15 min, 30 min and 60 min (18). The stock solution was prepared in the pharmacy 24 h before testing. A 1 μg/mL solution was prepared in a 25 mL syringe using exact methods. One millilitre was given to each patient by IV bolus at baseline. Before each inpatient test, children were asked to withhold exogenous glucocorticoids for 24 h, if possible. The test was nonfasting and conducted in the morning (between 08:00 and 10:00). Anesthetic cream was applied to the IV site before insertion. In all cases, normal functional status was defined as peak plasma cortisol >500 nmol/L.

Outcomes

The primary outcome measure of the present study was prevalence of HPA axis suppression, as assessed by morning serum cortisol level and confirmed by low-dose ACTH stimulation test. A secondary outcome measure was the clinical and drug factors that predict HPA axis suppression at time of diagnosis, as assessed by logistic regression analysis. Clinical factors were predetermined and included presence or absence of anorexia, malaise, nausea, headache, poor weight gain, hypoglycemia, abdominal pain, fatigue, weakness, myalgia, growth suppression or clinical suspicion of HPA axis suppression.

Planned analysis

Data were analyzed using the statistical package SPSS Statistics 19.0 (IBM Corporation, USA). Demographic data pertaining to cases and withdrawals were compared by Student’s t test or χ² test, as appropriate. Prevalence values were calculated as a number and percentage. A secondary analysis was planned using logistic regression to determine which clinical factors, if any, predicted HPA axis suppression while controlling for demographic variables. Preselected variables included height, weight and sex, and drug characteristics, such as ICS type, dose, duration of current dose and concomitant use of inhaled nasal steroid or long-acting beta-agonist. Variables entered into the prediction equation were not consistent with observed constraints, such as having fewer than 10 subjects per variable investigated (19) and, therefore, a comparison was made using Fischer’s exact test to preselect variables for entry into the model. Equivalent doses of ICS were calculated based on previously determined efficacy data (4). In all cases, a P≤0.05 was considered statistically significant.

RESULTS

Of the 321 children eligible during the study period, 311 agreed to participate and 214 had morning serum cortisol levels measured (Figure 1). Of those not tested, the most frequent reason was failure to visit the laboratory for a blood draw. In 21 patients (21.4%), this was due to fear of needles or citing that the procedure was ‘too invasive’. Due to the large number of withdrawals before testing, a comparison was made between the cases and withdrawals (Table 1). On comparison of the two groups, no statistically significant differences were found (P>0.05).

Figure 1).

Study flow diagram. * ≤170 nmol/L. ACTH adrenocorticotropic hormone

TABLE 1.

Characteristics of children who consented to morning serum cortisol testing and children who withdrew before testing

	Consented to morning serum cortisol testing (n=214)	Did not have morning serum cortisol tested (n=97)
Age, years	8.0±4.5	7.0±4.1
Male, n (%)	127 (59.3)	54 (56)
Caucasian, n (%)	204 (95.3)	94 (97.0)
Weight, kg	32.6±21.2	29.2±21.2
Height, cm	123.9±27.7	119.0±26.6
Dose duration, months	9.3±8.4	8.2±4.8
Nasal steroid, n (%)	13 (6.1)	2 (2.0)
Clinical suspicion of suppression, n (%)	1 (0.5)	0 (0)

All data presented as mean ± SD unless otherwise indicated. P>0.05 for all

A total of 43 patients (20.0%) had low morning serum cortisol levels and were required to attend for low-dose inpatient ACTH stimulation testing. One patient had an ACTH stimulation test despite having a normal morning cortisol level (178 nmol/L) and did not have evidence of suppression. A total of 20 children (9.3%, 95% CI 5.3% to 13.4%) had confirmed HPA axis suppression (46.5% of those with low morning serum cortisol levels). Table 2 summarizes the characteristics of these children, along with drug dose and peak cortisol response during ACTH stimulation testing.

TABLE 2.

Characteristics of children with hypothalamic-pituitary-adrenal axis (HPA) suppression confirmed by low-dose adrenocorticotropic hormone (ACTH) stimulation testing, according to type of inhaled corticosteroid (ICS)

Age, years	Sex	ICS dose, μg/day	Duration, months	Peak cortisol, nmol/L*
Fluticasone and salmeterol
7.0	M	500	33	<28
4.9	F	500	4	193
8.8	F	500	3	260
11.3	F	500	10	291
8.4	M	250	24	300
4	M	500	23	330
9	M	500	6	333
6.8	F	500	17	344
7.3	M	500	6	387
13.8	F	500	5	423
6.8	M	500	12	457
Fluticasone
11.8	F	500	3	<28
10.3	M	500	8	245
7.5	M	500	3	432
2.4	M	500	3	444
2.9	M	500	29	482
Beclomethasone
13.2	F	400	12	462
6.8	F	400	16	470
Ciclesonide
12.8	M	200	3	371
13.8	F	400	3	436

^*During low-dose ACTH stimulation test. F Female; M Male

All children with HPA axis suppression were on a medium or lower dose of ICS for their age (4). A statistically significant higher daily dose of ICS was found in children with HPA axis suppression compared with children with a normal HPA axis (median difference 250 μg/day, P<0.001). Fischer’s exact test did not reveal any other factors predicting HPA axis suppression, including clinical factors chosen apriori, drug type, duration, concomitant use of nasal steroid or long-acting beta-agonist (Table 3). Logistic regression revealed that weight-adjusted daily ICS dose was predictive of HPA axis suppression (OR 1.005, 95% CI 1.003 to 1.009, P<0.001). In other words, a 250 μg/day increase in dose is associated with an approximately 50% increase in the odds of HPA axis suppression.

TABLE 3.

Comparison of children with hypothalamic-pituitary-adrenal axis (HPA) axis suppression confirmed by low-dose adrenocorticotropic hormone (ACTH) stimulation testing

	HPA axis suppression (n=20)	Normal HPA axis (n=194)	P*
Age, years	8.5±3.5	8.0±4.6	0.56
Male, n (%)	14 (70.0)	113 (58.2)	0.35
Weight, kg	33.4±18.6	32.5±21.5	0.86
Height, cm	125.8±22.3	123.7±28.2	0.74
Dose duration, months	11.2±9.5	9.1±8.3	0.31
Oral steroids previous year, n (%)	4 (20.0)	25 (12.9)	0.49
Time since exposure, months	4.4±2.8	6.4±3.3	0.26
ICS type, n (%)†
Fluticasone	16 (80.0)	108 (57.1)	0.06
Beclomethasone	2 (10.0)	54 (28.5)	0.11
Ciclesonide	2 (10.0)	23 (12.1)	1
Budesonide	0 (0)	4 (2.1)	1
MDI with spacer, n (%)‡	20 (100)	171 (88.1)	0.14
ICS dose, μg/day, median (range)	500 (200, 500)	250 (100, 1000)	<0.001
Long-acting beta-agonist, n (%)	11 (55.0)	66 (34.0)	0.09
Nasal steroid, n (%)	2 (10.0)	11 (5.6)	0.35
Clinical suspicion of suppression, n (%)	0 (0)	1 (0.5)	1
Any clinical feature of suppression, n (%)	3 (15.0)	39 (20.1)	0.77

Data presented as mean ± SD unless otherwise indicated.

^*By 2-sided Fischer’s exact test, Student’s t test or Mann-Whitney U test, as appropriate;

^†Four children were taking fluticasone and ciclesonide, one child was taking fluticasone and beclomethasone and were not included in the denominator for this comparison;

^‡All children with metered dose inhalers (MDI) used spacers and otherwise used a dry-powder inhaler. ICS Inhaled cotricosteroid

DISCUSSION

We identified a significant prevalence of HPA axis suppression in children taking ICS for treatment of asthma (9.3%). In all cases, children were receiving medium doses or less of ICS (range: 200 μg/day to 500 μg/day fluticasone or equivalent). Additionally, daily ICS dose was predictive of HPA axis suppression, with a 250 μg/day increase equating to, approximately, a 50% increase in the odds of HPA axis suppression. Importantly, no other clinical factors were able to predict suppression in the identified cases compared with children with a normal HPA axis, and there was no statistically significant effect of drug type or duration of dose. Estimated prevalence of asthma in Canadian children is between 8% to 15%, indicating this is a significant clinical issue for physicians treating childhood asthma (20).

The impetus for this literature review and subsequent study was the presentation of a child, from our asthma clinic, to the emergency room during an intercurrent illness with significant hypoglycemia while on a medium dose of ICS. The 2004 Canadian Consensus Guidelines for Pediatric Asthma defined low-, medium- and high-dose ICS in terms of chlorofluorocarbon beclomethasone diproprionate, a medication that has not been available for use in Canada for some time (2). More recent consensus guidelines in Canada define dose categories based on hydrofluroalkane (HFA) beclomethasone dipropionate (ultra fine particle) (4). Specifically, a medium dose for children six to 11 years of age is 201 μg/day to 400 μg/day of HFA beclomethasone diproprionate (201 μg/day to 500 μg/day fluticasone). A medium dose for children older than 12 years of age is 251 μg/day to 500 μg/day of HFA belcomethasone dipropionate (equivalent for fluticasone). International guidelines reflect this classification system (1).

A review of the literature identified a single published study that measured the prevalence of HPA axis suppression in children being treated with ICS (15). Nine of 26 children (35%) were found to have suppression as identified by gold standard metyrapone testing. Unlike the current study, only the use of budesonide was studied and no effect of current ICS dose was found. Similarly, cumulative ICS dose was not predictive of HPA axis suppression, but was found to be statistically significant when combined with nasal steroid dose/m². Although we failed to find an effect of concomitant nasal steroid use, it is an important factor when considering total steroid dose in children with asthma. Additionally, we identified a slightly lower prevalence (9.3%) over a range of ICS types, and larger studies are needed to gain a more precise estimate. It is perhaps not surprising that we failed to find suppression in children using budesonide, because only four children in our sample were taking it.

The available literature on the use of basal adrenal function tests and dynamic HPA axis function tests has been reviewed extensively by Zöllner (16). Basal tests of adrenal function (including plasma, salivary and urinary cortisol) are unable to identify children who cannot adequately respond to stress but can be used in screening. Of the available dynamic tests of HPA axis function, the metyrapone test and the insulin tolerance test can be considered gold standards (16). However, the former is not available in most locations, and the latter is rarely used due to reports of death resulting from testing (21). Moreover, the insulin tolerance test is contraindicated in patients with epilepsy and heart disease, and in children younger than five years of age (22). The ACTH stimulation test is a good alternative test of adrenal functional capacity. Importantly, the conventional dose of 250 μg is considered supraphysiological (7) and could result in false-negative results. Therefore, the low-dose test was chosen, which may be more sensitive in detecting mild or evolving HPA axis suppression (23).

After the study began, a small (n=26) pilot study identified morning serum ACTH testing as a potential screening test for HPA axis suppression (24). However, ACTH has a short half-life that requires a strict protocol for testing. The authors concede that it would only be useful at large centres with onsite facilities familiar with the sampling and processing protocols. Although morning serum cortisol level was not recommended for screening purposes, it is weakly correlated to HPA axis suppression as detected by gold standard metyrapone testing (r=0.33, P=0.10)(24). The morning serum cortisol test is readily available to most clinicians in Canada. The specificity of the test is high (approaching 100% when a cutoff of 83 nmol/L is used); however, the sensitivity is poor (approximately 60%) and results are subject to both physiological and pathological variation (14,16). We used morning serum cortisol level to identify children with potential HPA axis suppression and for this reason may have missed children with suppression. The low-dose ACTH stimulation test is a suitable follow-up test to screening tests for suppression. It is also available at most paediatric asthma centres in Canada and has a high sensitivity for detecting HPA axis suppression (approximately 90%) (14). Importantly, the specificity of this test varies in the literature; 58% in one recent study (25) to approximately 90% (14), and the possibility of false-positives cannot be ruled out. Regardless, measuring morning serum cortisol levels raises the pretest probability in patients, and combined, both tests represent a practical approach.

Another study in the literature surveyed United Kingdom paediatricians and endocrinologists recording cases of adrenal crisis (12). The authors reported a 4.7% occurrence rate of adrenal crisis confirmed by dynamic testing of the HPA axis. Unlike the current study, cases were associated with high doses of ICS (from 500 μg/day to 2000 μg/day fluticasone or dose equivalent). Current European guidelines recommend screening children receiving ≥400 μg/day fluticasone or equivalent (26). A more recent North American guideline recommends screening be considered in children receiving ≥500 μg/day of fluticasone, with any symptoms regardless of ICS dose, on oral corticosteroid longer than two consecutive weeks, children who have received multiple courses of oral steroids for more than three weeks in the previous six months and children on high-dose ICS (14). The results of the present study support these guidelines and add the consideration of screening those on medium doses regardless of symptoms.

Every effort should be made to reduce the dose of ICS to the lowest effective dose that can achieve asthma symptom control. The use of a combined long-acting beta-agonist with ICS was not associated with the occurrence of HPA axis suppression in the present study, and it is possible that a lower effective dose could be achieved with its concomitant use. Low-dose ICS should be used as starting doses, and current Canadian guidelines support the use of add-on therapy rather than increasing the dose of ICS above 250 μg/day fluticasone or equivalent and certainly above 500 μg/day (4). Additionally, we found a nonsignificant trend with fluticasone toward HPA axis suppression. This finding is worrying in light of the high percentage of fluticasone use in practice and may be due to the drug’s increased lipophilicity and large particle size compared with other available compounds (27,28). This finding needs to be investigated in future, larger studies.

A previous case report identified four children with adrenal suppression secondary to fluticasone ICS, which improved upon switching to ciclesonide (29). For this reason, ciclesonide has been recommended in this situation (14,29). Although reports of side effects from ciclesonide are absent from the current literature, two children in our study were found to have HPA axis suppression with ciclesonide, representing the first described cases. Therefore, although ciclesonide may be beneficial in response to HPA axis suppression, more research is likely needed on its exact role in treatment.

The present study had several limitations. First, although the dose duration of ICS was recorded, the total length of ICS treatment since diagnosis was not known. Therefore, a cumulative effect of ICS usage could be present in our sample despite our inclusion criteria (at least three months of stable dosing and the absence of oral steroid use in the previous three months). Although oral steroid use may have lasting effects on the HPA axis (up to one year) (30), this exclusion criterion was not applied because it would limit the sample size and would have extended the recruitment period of the present study. However, this factor was investigated upon study completion (Table 3). Second, the asthma severity was not measured in each patient. However, a previous audit of our centre identified that the majority of children were of ‘moderate persistent’ asthma severity (1,17). Moreover, 96% of patients in the audit were treated as per the established guidelines published by the Global Initiative for Asthma (GINA) (1), indicating good overall adherence at the centre. Therefore, the results are most reflective of a secondary referral-only centre including patients with asthma previously unmanageable or difficult to manage by family physicians. Third, the morning serum cortisol level test has poor sensitivity and normal values do not rule out HPA axis suppression. Therefore, the prevalence may be a conservative estimate because a greater number of children could have abnormal ACTH stimulation tests, yet have morning serum cortisol levels within the normal range. Future, larger studies should attempt dynamic testing of the HPA axis on all study participants.

CONCLUSIONS

ICS are the best first-line medication for the management of non-intermittent asthma in children. However, as found in the present study, a significant number of children being treated for asthma with ICS have HPA axis suppression and are, therefore, at risk of adrenal crisis. Importantly, all children in the present study were on what is considered a medium or lower dose of ICS and no clinical factors (including symptoms, type of drug, duration or clinical suspicion) were able to predict HPA axis suppression. Our findings support previous recommendations to regularly screen children for the presence of HPA axis suppression when treated with high-dose ICS (>500 μg/day fluticasone). In addition, consideration should be given to screening children on medium-dose ICS. A finding of low morning serum cortisol levels should be followed by low-dose ACTH stimulation testing to confirm the presence of HPA axis suppression. Future research should focus on designing larger studies to further delineate the degree of suppression with multiple doses and types of ICS.