Scolaris Content Display Scolaris Content Display

Hypothalamic‐pituitary‐adrenal (HPA) axis suppression after treatment with glucocorticoid therapy for childhood acute lymphoblastic leukaemia

Collapse all Expand all

Abstract

Background

Glucocorticoids play a major role in the treatment of acute lymphoblastic leukaemia (ALL). However, supraphysiological doses can suppress the hypothalamic‐pituitary‐adrenal (HPA) axis. HPA axis suppression resulting in reduced cortisol response may cause an impaired stress response and an inadequate host defence against infection, which remain a cause of morbidity and death. Suppression commonly occurs in the first days after cessation of glucocorticoid therapy, but the exact duration is unclear. This review is the second update of a previously published Cochrane review.

Objectives

To examine the occurrence and duration of HPA axis suppression after (each cycle of) glucocorticoid therapy for childhood ALL.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2016, Issue 11), MEDLINE/PubMed (from 1945 to December 2016), and Embase/Ovid (from 1980 to December 2016). In addition, we searched reference lists of relevant articles, conference proceedings (the International Society for Paediatric Oncology and the American Society of Clinical Oncology from 2005 up to and including 2016, and the American Society of Pediatric Hematology/Oncology from 2014 up to and including 2016), and ongoing trial databases (the International Standard Registered Clinical/Social Study Number (ISRCTN) register via http://www.controlled‐trials.com, the National Institutes of Health (NIH) register via www.clinicaltrials.gov, and the International Clinical Trials Registry Platform (ICTRP) of the World Health Organization (WHO) via apps.who.int/trialsearch) on 27 December 2016.

Selection criteria

All study designs, except case reports and patient series with fewer than 10 children, examining effects of glucocorticoid therapy for childhood ALL on HPA axis function.

Data collection and analysis

Two review authors independently performed study selection. One review author extracted data and assessed 'Risk of bias'; another review author checked this information.

Main results

We identified 10 studies (total of 298 children; we identified two studies for this update) including two randomised controlled trials (RCTs) that assessed adrenal function. None of the included studies assessed the HPA axis at the level of the hypothalamus, the pituitary, or both. Owing to substantial differences between studies, we could not pool results. All studies had risk of bias issues. Included studies demonstrated that adrenal insufficiency occurs in nearly all children during the first days after cessation of glucocorticoid treatment for childhood ALL. Most children recovered within a few weeks, but a small number of children had ongoing adrenal insufficiency lasting up to 34 weeks.

Included studies evaluated several risk factors for (prolonged) adrenal insufficiency. First, three studies including two RCTs investigated the difference between prednisone and dexamethasone in terms of occurrence and duration of adrenal insufficiency. The RCTs found no differences between prednisone and dexamethasone arms. In the other (observational) study, children who received prednisone recovered earlier than children who received dexamethasone. Second, treatment with fluconazole appeared to prolong the duration of adrenal insufficiency, which was evaluated in two studies. One of these studies reported that the effect was present only when children received fluconazole at a dose higher than 10 mg/kg/d. Finally, two studies evaluated the presence of infection, stress episodes, or both, as a risk factor for adrenal insufficiency. In one of these studies (an RCT), trial authors found no relationship between the presence of infection/stress and adrenal insufficiency. The other study found that increased infection was associated with prolonged duration of adrenal insufficiency.

Authors' conclusions

We concluded that adrenal insufficiency commonly occurs in the first days after cessation of glucocorticoid therapy for childhood ALL, but the exact duration is unclear. No data were available on the levels of the hypothalamus and the pituitary; therefore, we could draw no conclusions regarding these outcomes. Clinicians may consider prescribing glucocorticoid replacement therapy during periods of serious stress in the first weeks after cessation of glucocorticoid therapy for childhood ALL to reduce the risk of life‐threatening complications. However, additional high‐quality research is needed to inform evidence‐based guidelines for glucocorticoid replacement therapy.

Special attention should be paid to patients receiving fluconazole therapy, and perhaps similar antifungal drugs, as these treatments may prolong the duration of adrenal insufficiency, especially when administered at a dose higher than 10 mg/kg/d.

Finally, it would be relevant to investigate further the relationship between present infection/stress and adrenal insufficiency in a larger, separate study specially designed for this purpose.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

Suppression of the stress system in children who received synthetic stress hormones for acute lymphoblastic leukaemia

Review question

We reviewed the evidence for suppression of the stress system/hypothalamic‐pituitary‐adrenal (HPA) axis (how often does it happen? how long does the suppression persist?) after treatment with synthetic stress hormones/glucocorticoids in children with acute lymphoblastic leukaemia (ALL).

Background

ALL is the most frequent type of cancer among children. Glucocorticoids, such as prednisone and dexamethasone, play a very important role in the treatment of ALL. However, high‐dose glucocorticoids can cause suppression of the stress axis (in medical terms, the hypothalamic‐pituitary‐adrenal (HPA) axis). Suppression of the stress or HPA axis results in inadequate cortisol production. Cortisol is the natural stress hormone found in humans. When this hormone is produced insufficiently, response to stressors (e.g. trauma, surgery, inflammation) may be impaired and defence against infections may be inadequate. Therefore, insufficient production of cortisol remains a cause of morbidity and death in childhood. The occurrence and duration of HPA axis suppression after glucocorticoid therapy for childhood ALL are unclear.

Study characteristics

This systematic review included eight cohort studies and two randomised studies with a total number of 298 patients. All studies assessed adrenal function in paediatric patients treated with glucocorticoids for ALL. The evidence is current to December 2016. None of these studies assessed the HPA axis at the level of the hypothalamus, the pituitary, or both. We could not combine the results of different studies because of heterogeneity.

Key results

Adrenal insufficiency occurred in nearly all children during the first days after completion of glucocorticoid therapy. Most children recovered within a few weeks, but a small number had ongoing adrenal insufficiency lasting up to 34 weeks. Three studies looked into differences in duration of adrenal insufficiency between children who received prednisone and those who were given dexamethasone (two types of glucocorticoids). Two of these three studies found no differences. In the other study, children who received prednisone recovered earlier than those who received dexamethasone. Also, treatment with a certain antifungal drug (fluconazole) seemed to prolong the duration of adrenal insufficiency. Two studies investigated this. Finally, two studies evaluated the presence of infection/stress as a risk factor for adrenal insufficiency. One study found no relationship. The other study reported that increased infection was associated with a longer duration of adrenal insufficiency.

More high‐quality research is needed to define the exact occurrence and duration of HPA axis suppression. Then adequate guidelines for glucocorticoid replacement therapy can be formulated.

Quality of the evidence

All of the included studies had some risk of bias issues.

Authors' conclusions

Implications for practice

Upon review of currently available evidence, we can conclude that adrenal insufficiency routinely occurs during the first days after cessation of glucocorticoid therapy, but the exact duration of adrenal insufficiency remains unclear. Data on levels of the hypothalamus and the pituitary are not available; therefore, we can draw no conclusions regarding these outcomes. Most children in included studies seemed to recover at between three days and seven weeks. However, a small number had prolonged adrenal insufficiency, persisting up to several months. Clinicians may consider prescribing glucocorticoid replacement therapy (e.g. hydrocortisone) during periods of serious stress in the first weeks after cessation of glucocorticoid therapy for childhood acute lymphoblastic leukaemia (ALL), to reduce the risk of life‐threatening complications. If replacement therapy is indicated, its beneficial effects and side effects should be evaluated. Until results of future adequate studies on the incidence and duration of hypothalamic‐pituitary‐adrenal (HPA) axis suppression become available, clinicians may consider performing an HPA axis stimulation test, for example, two months after cessation of glucocorticoids, to determine whether the HPA axis has recovered, and whether replacement therapy provided during periods of stress can be discontinued. Exclusively morning cortisol levels are inappropriate for evaluation of HPA axis suppression because they reflect only basal cortisol prediction ‐ not the ability of the HPA axis to respond to stress (Agwu 1999).

Special attention should be paid to children receiving fluconazole therapy, and perhaps similar antifungal drugs, as such treatment may prolong the duration of adrenal insufficiency.

We can make no definitive conclusions regarding differences in occurrence and duration of adrenal insufficiency in terms of type (predniso(lo)ne vs dexamethasone); (cumulative) dose, duration, and method of cessation (abrupt or gradual) of glucocorticoid therapy; and other risk factors such as infection/stress.

Implications for research

Studies examining HPA axis suppression after high‐dose glucocorticoid therapy for childhood ALL are scarce, especially RCTs. High‐quality research regarding occurrence and duration of HPA axis suppression after glucocorticoid therapy for childhood ALL is needed to inform adequate evidence‐based guidelines for glucocorticoid replacement therapy. Studies evaluating long‐term effects of glucocorticoid therapy on the HPA axis are also needed.

Future studies should focus on identifying differences in effects of type, (cumulative) dose, repeated exposure, duration, and method of cessation of glucocorticoid therapy and other risk factors on occurrence and duration of HPA axis suppression. The number of included children should be sufficient to obtain the power needed for reliable results.

Furthermore, an interesting and relevant topic for future research would be the (genetic) susceptibility of individuals to HPA axis suppression after glucocorticoid treatment.

Background

Of all malignancies in children, acute lymphoblastic leukaemia (ALL) occurs most frequently. In the Netherlands, ALL is newly diagnosed in approximately 120 children annually (DCOG 2014). Treatment and survival rates for childhood ALL have substantially improved over time, and morbidity and mortality due to treatment‐related side effects have become increasingly important. Unfortunately, up to 5% of children die as the result of toxic side effects of treatment, and this percentage is even greater in higher‐risk subgroups. The main cause of this treatment‐related mortality is infection associated with cytotoxic and immunosuppressive drugs (Christensen 2005; Pruckner 2009; Rubnitz 2004; Wheeler 1996). Glucocorticoid therapy is an important contributing factor to the occurrence and severity of infection (Te Poele 2007). Several studies have reported an increase in sepsis and lethal infections among children with ALL when prednisone was substituted for the more potent glucocorticoid dexamethasone (Hurwitz 2000; Igarashi 2005; Te Poele 2007).

Glucocorticoids play a major role in the treatment of ALL, inducing apoptosis of lymphoblastic cells (Planey 2000). Children with ALL receive cyclical courses of high‐dose glucocorticoids such as prednisone (or prednisolone) and dexamethasone. However, supraphysiological doses of glucocorticoids can suppress hypothalamic secretion of corticotrophin‐releasing hormone (CRH) and adrenocorticotropic hormone (ACTH) by the pituitary gland, resulting in secondary adrenal cortex atrophy with delayed recovery of hypothalamic‐pituitary‐adrenal (HPA) axis function. In states of profound or prolonged ACTH deficiency, the adrenal glands may be temporarily unable to generate sufficient cortisol (Henzen 2000; Krasner 1999). Supraphysiological glucocorticoid therapy is the most common cause of secondary adrenal insufficiency (Shulman 2007). The HPA axis plays a major role in the stress response and in host defence against infection. Stressors such as trauma, surgery, or inflammation stimulate the HPA axis, leading to an increase in cortisol production. In turn, increased cortisol levels cause anti‐inflammatory effects and inhibition of pro‐inflammatory cytokines (Hettmannsperger 1992; Nyhlén 2000; Waage 1988). Suppression of the HPA axis resulting in reduced adrenal cortisol production represents an impaired stress response and an inadequate host defence against infection and remains a cause of morbidity and death in childhood (Shulman 2007).

Various tests of HPA axis function have been well established. Morning serum cortisol value reflects basal adrenal function but gives no indication of capacity to respond to stress (Agwu 1999; Schlaghecke 1992). Stimulation tests are used to assess the response of the HPA axis to stress. The insulin tolerance test is considered the most reliable way to evaluate HPA axis function at the level of the pituitary and the adrenal glands, but it is associated with potentially dangerous side effects (Shah 1992). The CRH stimulation test is indicated only for individuals with a central disorder of the HPA axis (Maghnie 2005; Van Tijn 2008). The glucagon test has proved to be a safe and reliable method for testing HPA axis function at the levels of the pituitary and the adrenal glands, but it may induce mild inadvertent side effects (Böttner 2005; Rao 1987; Vanderschueren‐Lodeweyckx 1974). A well‐established alternative method for testing the HPA axis at the level of the adrenal gland without undesirable side effects is the low‐dose (1 µg) ACTH stimulation test, which can detect more subtle degrees of adrenal atrophy caused by central adrenal insufficiency than are detected by the 'normal' ACTH stimulation test (250 µg). Results of the low‐dose ACTH test closely correlate with those of the insulin tolerance test (Abdu 1999; Dickstein 1997; Tordjman 2000).

Several studies have prospectively assessed HPA axis function following treatment with high‐dose glucocorticoids for ALL. Adrenal stimulation tests have been performed and repeated until cortisol levels were normalised. Most children seemed to recover in a few weeks, but prolonged suppression may occur, lasting longer than several months in some cases (Einaudi 2008; Mahachoklertwattana 2004; Rix 2005; Salem 2015; Vestergaard 2011). Children with ALL plus HPA axis suppression with fever or other stressors may benefit from glucocorticoid replacement therapy (e.g. hydrocortisone); therefore, it is important to derive a consistent picture of HPA axis impairment after corticosteroid therapy.

This is the second update of the first Cochrane systematic review of HPA axis function after glucocorticoid therapy for childhood ALL (Gordijn 2012; Gordijn 2015). On the basis of information provided in this review, adequate guidelines for glucocorticoid substitution can be formulated and implemented to reduce risk of infection in childhood ALL.

Objectives

To examine the occurrence and duration of HPA axis suppression after (each cycle of) glucocorticoid therapy for childhood ALL.

Methods

Criteria for considering studies for this review

Types of studies

All study designs except case reports and patient series with fewer than 10 children, examining effects of glucocorticoid therapy for childhood ALL on HPA axis function. This effect can be evaluated both during treatment for ALL (after cessation of a glucocorticoid course) and after all ALL treatment is completed. Review authors resolved disagreements concerning the definition of a cohort study by consensus, with no third party arbitration needed.

Types of participants

Children between the ages of 0 and 18 years who were treated with glucocorticoids for ALL, irrespective of the duration of follow‐up after completion of glucocorticoid therapy. Exclusion criteria consisted of cranial radiotherapy because this treatment may damage the HPA axis, as well as assessment of HPA axis function by a CRH stimulation test only because this test is indicated only for patients with a central disorder of the HPA axis (Maghnie 2005; Van Tijn 2008).

Types of interventions

Glucocorticoid therapy (prednisone, prednisolone, dexamethasone) during treatment for ALL. The intervention was not compared with a control because this option was not available (except in the included randomised controlled trials (RCTs)).

Types of outcome measures

Outcomes reported here were not used as criteria for including studies but are the outcomes of interest within studies identified for inclusion.

Primary outcome

  • Adrenal insufficiency (occurrence and duration), measured by early‐morning plasma cortisol levels (between 8 and 10 a.m.) or by stimulation tests (e.g. low‐dose ACTH stimulation test, glucagon stimulation test). We used the cutoff limit as defined by the authors of original studies.

Secondary outcomes

To examine whether adrenal insufficiency after administration of glucocorticoids is dependent on:

  • moment of testing after cessation of glucocorticoid therapy;·

  • (cumulative) dose of glucocorticoids;

  • type of glucocorticoid: prednisone, prednisolone, or dexamethasone;

  • duration of glucocorticoid therapy;

  • method of cessation of glucocorticoid therapy: abrupt or gradual; or

  • other possible risk factors (such as fluconazole and presence of infection/stress).

Search methods for identification of studies

See Cochrane Childhood Cancer methods used in reviews (Module CCG).

The objective of the literature search was to identify all studies, except case reports and case series, reporting on HPA axis function after glucocorticoid therapy for childhood ALL. Cochrane Childhood Cancer ran searches in the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and Embase; the review authors ran all other searches.

Electronic searches

We searched the following electronic databases: CENTRAL (2016, Issue 11), MEDLINE/PubMed (from 1945 to 12 December 2016), and Embase/Ovid (from 1980 to 12 December 2016). We have provided in appendices the search strategies used for individual electronic databases (using a combination of controlled vocabulary and text words) (Appendix 1; Appendix 2; Appendix 3).

Searching other resources

We located information about trials not registered in CENTRAL, MEDLINE, or Embase, either published or unpublished, by searching the reference lists of relevant articles and review articles. We handsearched conference proceedings of the International Society for Paediatric Oncology (SIOP) and the American Society of Clinical Oncology (ASCO) (from 2005 up to and including 2016). Moreover, for this second review update, we handsearched conference proceedings of the American Society of Pediatric Hematology/Oncology (ASPHO) (from 2014 up to and including 2016). For the search strategy, see Appendix 4.

We scanned the following registers for ongoing trials: the International Standard Randomized Controlled Trial Number (ISRCTN) register (http://www.controlled‐trials.com), the National Institutes of Health (NIH) register (www.clinicaltrials.gov), and the International Clinical Trials Registry Platform (ICTRP) of the World Health Organization (apps.who.int/trialsearch) (all screened 27 December 2016; for the search strategy, see Appendix 5).

We imposed no language restrictions. We will update these searches every two years.

Data collection and analysis

Selection of studies

Two review authors independently selected studies meeting the inclusion criteria. Review authors resolved disagreements by consensus, with no third party arbitration needed. We obtained in full for closer inspection any study that seemed to meet the inclusion criteria upon review of titles or abstracts, or both. We clearly stated reasons for exclusion of any study considered for this review. We have included in the review a flow chart for selection of studies.

Data extraction and management

One review author performed data extraction using standardised forms; another review author checked the data recorded. Review authors were not blinded to journal, authors, or institution. We extracted data on the following categories: study characteristics, children, interventions, outcome measures, length of follow‐up, risk factors, and 'Risk of bias' assessment.

We resolved disagreements between review authors by consensus, with no third party arbitration needed.

Assessment of risk of bias in included studies

We based our assessment of risk of bias on previously described checklists for observational studies according to evidence‐based medicine criteria (Grimes 2002; Laupacis 1994). One review author performed the risk of bias assessment of included studies; another review author checked these assessments. We have described in Table 1 the 'Risk of bias' assessment criteria for observational studies. In assessing RCTs, we used 'Risk of bias' items as described in the module of Cochrane Childhood Cancer (Module CCG) and based on recommendations provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) (see Table 2). We resolved disagreements among review authors by consensus, with no third party arbitration needed.

Open in table viewer
Table 1. 'Risk of bias' assessment criteria for observational studies

Internal validity

External validity

Study group

Selection bias (representative: yes/no):

  • if it consisted of more than 90% of the original cohort of children with ALL treated with glucocorticoids; or

  • if it was a random sample with respect to treatment.

Reporting bias (well defined: yes/no):

  • if treatment protocol was mentioned; and

  • if (cumulative) dose of glucocorticoid treatment was mentioned; and

  • if type of glucocorticoid treatment was mentioned; and

  • if duration of glucocorticoid treatment was mentioned; and

  • if method of cessation of glucocorticoid treatment was mentioned.

Follow‐up

Attrition bias (adequate: yes/no):

  • if outcome was assessed at end date of the study for 60% to 90% of study group; or

  • if outcome was assessed for more than 90% of the study group but with an unknown end date.

Reporting bias (well defined: yes/no):

  • if length of follow‐up was mentioned; and

  • if frequency of measuring outcomes was mentioned.

Outcome

Detection bias (blinding: yes/no):

  • if outcome assessor was blinded to glucocorticoid treatment.

Reporting bias (well defined: yes/no):

  • if methods of detection were described; and

  • if outcome definition was objective and precise.

Risk estimation

Confounding (adjustment for other factors: yes/no):

  • if important prognostic factors (i.e. age, sex, cotreatment) or follow‐up was taken adequately into account.

Analysis (well defined: yes/no):

  • if risk ratio, odds ratio, attributable risk, linear or logistic regression model, mean difference, or Chi2 statistic was calculated.

ALL: acute lymphoblastic leukaemia.

Open in table viewer
Table 2. 'Risk of bias' assessment criteria for randomised controlled trials

Selection bias

Sequence generation (adequate: yes/no):

  • if the rule for allocating interventions to participants was based on some chance (random) process.

Allocation concealment (adequate: yes/no):

  • if the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Performance bias
Blinding of care providers (yes/no):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Blinding of participants (yes/no):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Detection bias
Blinding of outcome assessors (yes/no; assessed for each outcome separately):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Attrition bias
Incomplete outcome data (adequate: yes/no; assessed for each outcome separately):

  • if incomplete outcome (attrition and exclusions) data have been adequately addressed.

Reporting bias
Selective outcome reporting (yes/no):

  • if reports of the study were free of the suggestion of selective outcome reporting.

Other bias
Other bias (yes/no):

  • if the study was free of other problems (i.e. potential source of bias related to specific study design, premature termination of the study due to some data‐dependent process, extreme baseline imbalance) that could put it at high risk of bias.

Measures of treatment effect

Prevalence of HPA axis suppression at several follow‐up time points and time duration of HPA axis suppression.

Dealing with missing data

We contacted authors of individual studies to ask for clarification of unclear data or to request missing data regarding selection of studies, 'Risk of bias' assessment, or data extraction.

Assessment of heterogeneity

We planned to assess heterogeneity both by visually inspecting forest plots and by performing a formal statistical test for heterogeneity, that is, the I2 statistic. However, because we were not able to pool the results of included studies, this approach was not applicable.

Assessment of reporting biases

We planned to use a funnel plot to quantify the potential presence of publication bias. However, because we were not able to pool the results of included studies, this approach was not applicable.

Data synthesis

We planned to perform analyses using the statistical software Comprehensive Meta‐Analysis (Biostat, Inc., Englewood, NJ, USA) (Biostat, Inc, USA).

Across studies, we planned to conduct a multi‐variate linear meta‐regression analysis model using a backwards selection strategy (P < 0.10) to examine the relation between potential predictive factors and HPA axis suppression.

However, because we were not able to pool the results of included studies, this was not applicable, and we have described the results of individual studies separately.

Sensitivity analysis

We planned to perform a sensitivity analysis for 'Risk of bias' assessment criteria used. However, because we were not able to pool the results of included studies, this was not applicable.

Results

Description of studies

See Characteristics of included studies and Characteristics of excluded studies.

Results of the search

For the original version of this review (Gordijn 2012), searches run in the electronic databases of CENTRAL, MEDLINE/PubMed, and Embase/Ovid revealed a total of 1388 references. Initial screening of titles and abstracts led to exclusion of 1375 references that clearly did not fulfil all criteria for inclusion of studies in this review. For the remaining 13 references, we examined full‐text articles and deemed that 7 of the 13 were eligible for inclusion in this review. We have provided the reasons for exclusion under Characteristics of excluded studies.

Scanning the reference lists of relevant articles and reviews did not reveal additional eligible studies.

Scanning the conference proceedings of SIOP and ASCO did not reveal additional eligible studies, and scanning of ongoing trials databases revealed no ongoing studies.

Searches run for the first update (Gordijn 2015) using the electronic databases of CENTRAL, MEDLINE, and Embase, on 16 June 2014, revealed a total of 314 references. Initial screening of titles and abstracts led to exclusion of 310 references that clearly did not fulfil all criteria for inclusion of studies in this review. We examined the full‐text versions of three of the four articles (one full‐text article was not yet available and was placed in the studies awaiting classification list) and determined that one of these was eligible for inclusion in this review. We have provided the reasons for exclusion under Characteristics of excluded studies.

Scanning the reference lists of relevant articles and reviews did not reveal additional eligible studies.

Scanning the conference proceedings of SIOP and ASCO did not reveal eligible studies, and scanning of ongoing trials databases did not reveal any ongoing studies.

Searches run for the second update in the electronic databases of CENTRAL, MEDLINE, and Embase, on 12 December 2016, revealed a total of 236 references. Initial screening of these titles and abstracts led to exclusion of 235 references that clearly did not fulfil all criteria for inclusion of studies in this review. We examined the full‐text version of one article and determined that this article was eligible for inclusion in the review.

Scanning the reference list of the relevant article did not reveal additional eligible studies.

Scanning the conference proceedings of SIOP, ASCO, and ASPHO revealed one eligible study. The full‐text article for this study was not yet available (see Characteristics of studies awaiting classification) (Schlosser 2016).

Scanning the ongoing trials databases did not reveal any ongoing studies.

For the first update of the review, we placed one study in the list of studies awaiting classification. This study has since been published (Perdomo‐Ramírez 2016). We examined the full‐text article and determined that this study was eligible for inclusion in this review update.

In summary, we included 10 articles in this review. We attempted to contact trial authors to clarify aspects of study design and data analysis. We have summarised characteristics of the included studies in the Characteristics of included studies table. See Figure 1 for a flow diagram showing selection of studies for this systematic review.


Study flow diagram.

Study flow diagram.

Included studies

All included studies evaluated adrenal function after glucocorticoid therapy for childhood ALL. None of these studies assessed the HPA axis at the level of the hypothalamus, the pituitary, or both. Studies included a total of 298 children. The 10 included studies examined adrenal function after different types, doses, and durations of glucocorticoid therapy and after different methods of cessation of glucocorticoid therapy. Three studies examined effects of dexamethasone on adrenal function (Cunha 2004; Felner 2000; Kuperman 2001). One study examined effects of prednisone on adrenal function (Perdomo‐Ramírez 2016). Six studies evaluated effects of both dexamethasone and predniso(lo)ne on adrenal function (Einaudi 2008; Kuperman 2012; Mahachoklertwattana 2004; Petersen 2003; Rix 2005; Salem 2015). Some investigators measured adrenal function by early‐morning plasma cortisol levels (between 8 and 10 a.m.) (Cunha 2004; Kuperman 2001). Others used the (low‐dose) ACTH stimulation test (Einaudi 2008; Felner 2000; Kuperman 2012; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005; Salem 2015 ). Six studies performed follow‐up tests until normalisation of adrenal function (Einaudi 2008; Felner 2000; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Salem 2015). Durations of follow‐up for the other four studies were one month, two weeks, two months, and two to seven days, respectively (Cunha 2004; Kuperman 2001; Kuperman 2012; Rix 2005). Eight of the 10 included studies were observational studies (Cunha 2004; Felner 2000; Kuperman 2001; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005; Salem 2015). Two were RCTs evaluating prednisone versus dexamethasone (Einaudi 2008; Kuperman 2012). In one of the RCTs, both treatment groups received prednisone before randomisation (Einaudi 2008).

Excluded studies

We have provided in the Characteristics of excluded studies table information on the eight studies excluded during examination of full‐text articles. The most common reasons for exclusion were cranial irradiation therapy and lack of (adequate) HPA axis function tests.

Risk of bias in included studies

Cohort studies

For evaluation of 'internal validity' of the eight included cohort studies, we assessed risks of selection bias, attrition bias, detection bias, and confounding. Upon obtaining additional information from trial authors, we determined the following. Risk of selection bias (based on representativeness of the study group) was low in four of the eight studies, as the study group consisted of more than 90% of the original cohort (Felner 2000; Kuperman 2001; Petersen 2003; Rix 2005). One study selected an unrepresentative study group of about 30% of the original cohort (Cunha 2004). For the three other studies, neither published articles nor correspondence with trial authors yielded information on selection of children (Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Salem 2015). Risk of attrition bias (based on completeness of follow‐up) was low in seven of the eight studies, as investigators assessed outcomes for 60% to 90% of the study group at the end date of the study (Cunha 2004; Felner 2000; Kuperman 2001; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005). For the other study, neither the published article nor correspondence with trial authors yielded information on risk of attrition bias (Salem 2015). The authors of two studies reported that the outcome assessor was not blinded (Perdomo‐Ramírez 2016; Rix 2005). None of the other studies provided information on blinding of the outcome assessor to glucocorticoid treatment, so we could not rule out detection bias. Two of the eight included cohort studies addressed risk factors for development or persistence of adrenal insufficiency (Petersen 2003; Salem 2015). Confounding (based on important risk factors and follow‐up taken into account) was not present in these studies.

For evaluation of 'external validity' of the included cohort studies, we assessed risk of reporting bias. Five studies did not define the study group well in terms of treatment protocol and (cumulative) dose, type, duration, and form of cessation of glucocorticoid treatment (Cunha 2004; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005). Two studies did not mention the treatment protocol (Felner 2000; Kuperman 2001). One study did not mention the duration of tapering of glucocorticoid treatment, thus cumulative dose and total duration of glucocorticoid treatment were unclear (Salem 2015). All eight studies defined follow‐up well, as trial authors mentioned both length of follow‐up and frequency of measurement (Cunha 2004; Felner 2000; Kuperman 2001; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005; Salem 2015). For these studies, outcomes were well defined, methods of detection were described, and outcome definitions were objective and precise (Cunha 2004; Felner 2000; Kuperman 2001; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Petersen 2003; Rix 2005; Salem 2015). One of the two studies addressing risk factors defined risk estimation (based on calculation of risk ratio, odds ratio, attributable risk, linear or logistic regression model, mean difference, or Chi2 statistic) well (Salem 2015). The other study did not define risk estimation well (Petersen 2003).

See also Table 3.

Open in table viewer
Table 3. Risk of bias in included observational studies

Study

Representative study group

Complete follow‐up assessment

Blinded outcome assessor

Adjustment for important confounders

Well‐defined study group

Well‐defined follow‐up

Well‐defined outcome

Well‐defined risk estimation

Cunha 2004

No, based on additional information provided by trial authors, the study group described did not consist of more than 90% of the original cohort and was not a random sample.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, treatment protocol and (cumulative) dose, type, duration, and form of cessation of glucocorticoid treatment were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Felner 2000

Yes, based on additional information provided by trial authors, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

No, treatment protocol was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Kuperman 2001

Yes, based on additional information provided by trial authors, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

No, treatment protocol was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Mahachoklertwattana 2004

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, treatment protocol and (cumulative) dose, type, duration, and form of cessation of glucocorticoid treatment were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Perdomo‐Ramírez 2016

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

No, outcome assessor was not blinded to glucocorticoid treatment. This information was based on additional information provided by trial authors.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid therapy were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Yes, mean difference was calculated.

Petersen 2003

Yes, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment.

Yes, important prognostic factors or follow‐up was taken into account.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid treatment were mentioned. Information on the method of cessation of glucocorticoid treatment was based on additional information provided by trial authors.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

No, risk ratio, odds ratio, attributable risk, linear or logistic regression model, mean difference, or Chi2 statistic was not calculated.

Rix 2005

Yes, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

No, outcome assessor was not blinded to glucocorticoid treatment.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid treatment were mentioned. Information on the method of cessation of glucocorticoid treatment was based on additional information provided by trial authors.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Salem 2015

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Unclear whether outcome was assessed for 60% to 90% at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, important prognostic factors or follow‐up was taken into account.

No, duration of tapering of glucocorticoid treatment was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Yes, mean difference was calculated.

RCTs

For included RCTs, we evaluated risks of selection bias, performance bias, detection bias, attrition bias, reporting bias, and other biases. Upon receiving additional information from trial authors, we made the following determinations. We found no selection bias (based on sequence generation and concealment of allocation) in two studies (Einaudi 2008; Kuperman 2012). We could not rule out performance bias (based on blinding of care providers and children) or detection bias (based on blinding of outcome assessors) in one RCT (Einaudi 2008). This same RCT was susceptible to reporting bias, as trial authors did not report all of the study's prespecified primary outcomes (Einaudi 2008). (For assessment of selective outcome reporting bias, we compared the methods and results sections of included RCTs.) We found no performance bias, detection bias, or reporting bias in the other RCT (Kuperman 2012). For both RCTs, we judged that risks of attrition bias (based on completeness of outcome data) and other bias (i.e. based on potential sources of bias related to specific study design, premature termination of the study due to some data‐dependent process, or extreme baseline imbalance) were low (Einaudi 2008; Kuperman 2012).

See also Table 4.

Open in table viewer
Table 4. Risk of bias in included randomised controlled trials

Study

Adequate sequence generation?

Adequate allocation concealment?

Blinding?

Incomplete outcome data addressed?

Free of selective reporting?

Free of other bias?

Einaudi 2008

Yes, according to additional information provided by trial authors, the rule for allocating interventions to children was based on some chance (random) process.

Yes, according to additional information provided by trial authors, the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Based on additional information provided by trial authors, care providers, children, and outcome assessors were not blinded.

Yes, no outcome data were missing.

No, "adrenal function completely recovered in the 12 children evaluated with subsequent low‐dose ACTH test (in 4, 3, and 5 patients after 4, 8, and 10 weeks, respectively)". However, it was not reported which of these children received prednisone and which received dexamethasone. Therefore, not all of the study's prespecified primary outcomes were reported.

Yes

Kuperman 2012

Yes, the rule for allocating interventions to children was based on some chance (random) process.

Yes, according to additional information provided by trial authors, the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Yes, based on additional information provided by trial authors, care providers, children, and outcome assessors were all blinded.

Outcomes were assessed for 83% to 93% of the study population.

Yes

Yes

ACTH: adrenocorticotropic hormone.

Effects of interventions

Adrenal insufficiency (occurrence and duration)

We could extract from all included studies data on the prevalence and duration of adrenal insufficiency after treatment with glucocorticoid therapy for childhood ALL. However, it should be noted that individual studies used different types and (cumulative) doses of glucocorticoids, and we detected differences in the duration and method of cessation of glucocorticoid therapy. Methods of testing adrenal function varied as well. Owing to this heterogeneity, pooling of results was not possible. For more information, see Characteristics of included studies.

ACTH stimulation test

Two studies used the ACTH stimulation test with comparable cutoff limits (stimulated cortisol 18 µg/dL (500 nmol/L)) in measuring adrenal function (Felner 2000; Petersen 2003). Before glucocorticoid therapy, adrenal function was normal in all children in one study (Felner 2000). However, all 10 children (100%) had insufficient cortisol levels one day after abrupt cessation of 28 days of dexamethasone at 6 mg/m2/d. Three out of 10 children (30%) had ongoing adrenal insufficiency after four weeks, but all recovered after eight weeks. Another study assessed two types of glucocorticoid therapy: induction therapy comprising 35 days of prednisolone 60 mg/m2/d with tapering over nine days, and reinduction therapy comprising 21 days of dexamethasone 10 mg/m2/d with tapering over nine days (Petersen 2003). Reinduction therapy followed induction therapy, thus children in the dexamethasone group also received induction therapy with prednisolone. Investigators did not assess HPA axis function before glucocorticoid therapy. After induction therapy (n = 10), 7 out of 10 children (70%) had adrenal insufficiency within the first week. Six children (60%) had ongoing adrenal insufficiency after three weeks, and four children (40%) after seven weeks. These four children remained insufficient at the end of follow‐up, that is, after 10, 11, 11, and 19 weeks, respectively. The last child, who showed no recovery by 19 weeks, received, in addition to induction therapy, two one‐week‐long reinduction courses including prednisolone 60 mg/m2/d during adrenal insufficiency. After completing reinduction therapy (n = 7), five out of seven children (71%) had adrenal insufficiency within the first week. Four children (57%) had ongoing adrenal insufficiency after three weeks, and three children (43%) after seven weeks. These three children remained insufficient at the end of follow‐up, that is, after 16, 33, and 34 weeks, respectively. One of these children, who showed no recovery after 16 weeks, received an additional one‐week‐long reinduction course of prednisolone 60 mg/m2/d during the period of adrenal insufficiency. The other two children, who showed no recovery after 33 and 34 weeks, received three additional one‐week‐long reinduction courses, including prednisolone 60 mg/m2/d, during adrenal insufficiency. See Table 5 for an overview of the prevalence and duration of adrenal insufficiency in studies that used an ACTH suppression test.

Open in table viewer
Table 5. Prevalence and duration of adrenal insufficiency evaluated by an ACTH stimulation test

Felner et al

Therapy: dexamethasone (cumulative dose 168 mg/m2)

Time after cessation

Before

1 day

4 weeks

8 weeks

n insufficient/n total

0/10

10/10

3/10

0/10

Petersen et al (1)

Therapy: prednisolone (cumulative dose 2257.5 mg/m2)a

Time after cessation

1 week

3 weeks

7 weeks

End of follow‐up: 10, 11, 11, and 19 weeks, respectively

n insufficient/n total

7/10

6/10

4/10

4/10

Petersen et al (2)

Therapy: dexamethasone (cumulative dose 236.25 mg/m2)b

Time after cessation

1 week

3 weeks

7 weeks

End of follow‐up: 16, 33, and 34 weeks, respectively

n insufficient/n total

5/7

4/7

3/7

3/7

ACTH: adrenocorticotropic hormone.
a One child received additional 840 mg/m2 prednisolone during the period of adrenal insufficiency.

b These children received prednisolone 2257.5 mg/m2 as induction therapy before. Three high‐risk patients received an additional 560 mg/m2 prednisolone in advance. Furthermore, owing to persistent adrenal insufficiency, one of these high‐risk children received an additional 420 mg/m2 prednisolone during the period of insufficiency, and the other two high‐risk children received an additional 1260 mg/m2 prednisolone during that period. 

Low‐dose ACTH stimulation test with comparable cutoff limits

Five studies used the low‐dose ACTH stimulation test with comparable cutoff limits (stimulated cortisol 18 µg/dL (500 nmol/L)) in measuring adrenal function (Einaudi 2008; Mahachoklertwattana 2004; Perdomo‐Ramírez 2016; Rix 2005; Salem 2015 ).

In one study, all children received induction therapy with prednisolone 40 mg/m2/d (Mahachoklertwattana 2004). This was followed four weeks after completion by maintenance therapy consisting of seven days of dexamethasone 8 mg/m2/d every four weeks. Baseline cortisol levels before induction therapy and two weeks afterwards were not significantly different between adrenal‐suppressed and adrenal‐unsuppressed groups. Eleven out of 24 children (46%) had adrenal insufficiency two weeks after abrupt cessation of 28 days of induction therapy. Nine children (38%) had ongoing adrenal insufficiency after four weeks, seven children (29%) after eight weeks, and three children (13%) after 12 weeks. The last three children remained insufficient at the end of follow‐up at 20 weeks.

Another study (Rix 2005) evaluated three types of glucocorticoid treatment. All children (at standard, intermediate, and high risk) received induction therapy (22 children in total were tested afterwards; for two children, no information was available) comprising 35 days of prednisolone 60 mg/m2/d with tapering over nine days. All children also received seven‐day courses of prednisolone 60 mg/m2/d without tapering (13 children in total were tested afterwards); intermediate‐ and high‐risk children also received 21 days of dexamethasone 10 mg/m2/d with tapering over nine days (seven children in total were tested afterwards). The seven‐day course of prednisolone and the dexamethasone course followed induction therapy. The intermediate‐risk group received the dexamethasone course before the seven‐day course of prednisolone, and the high‐risk group received the dexamethasone course after the seven‐day course of prednisolone. Trial authors provided information showing that 13 children were tested before induction therapy, and all had normal adrenal function. Sixteen out of 22 children (73%) had adrenal insufficiency one day after cessation of induction therapy; one was lost to follow‐up thereafter. Five out of 22 children (23%) were not tested at this time point. Eight children (36%) had ongoing adrenal insufficiency after three days; two were lost to follow‐up afterwards. Seven children (32%) with no confirmed adrenal recovery were not tested at this time point (including the child lost to follow‐up after the first test moment subsequent to cessation of induction therapy). Eight children (36%) had ongoing adrenal insufficiency at the end of five days of follow‐up. Three children (14%) (i.e. the three children lost to follow‐up) with no confirmed adrenal recovery were not tested at this time point. After a seven‐day course of prednisolone, all 13 children (100%) remained insufficient at the end of the follow‐up period of two days. However, two out of 13 children (15%) had insufficient adrenal function before receiving prednisolone therapy. Five out of seven children (71%) underwent a low‐dose ACTH test before receiving dexamethasone therapy; all had sufficient cortisol levels. One day after the dexamethasone course, two out of seven children (29%) had adrenal insufficiency; one was lost to follow‐up thereafter. Five out of seven children (71%) were not tested at this time point. Three children (43%) had ongoing adrenal insufficiency after three days; one was lost to follow‐up afterwards. Two children (29%) (including the child lost to follow‐up) with no confirmed adrenal recovery were not tested at this time point. One child (14%) had ongoing adrenal insufficiency at the end of the follow‐up period of seven days. Two children (29%) (i.e. the two children lost to follow‐up) with no confirmed adrenal recovery were not tested at this time point.

Another study examined two randomised arms of glucocorticoid therapy: 22 days of prednisone 60 mg/m2/d with tapering over nine days (n = 40), and 22 days of dexamethasone 10 mg/m2/d with tapering over nine days (n = 24) (Einaudi 2008). Both groups of children received seven days of prednisone 60 mg/m2/d in advance. At diagnosis, basal cortisol values were within the normal range for all children. In the prednisone arm, 32 out of 40 children (80%) had adrenal insufficiency one day after cessation of glucocorticoid therapy, eight children (20%) had ongoing adrenal insufficiency after 7 to 14 days, and five children (13%) after 28 days. All children (100%) recovered in 10 weeks. In the dexamethasone arm, 20 out of 24 children (83%) had adrenal insufficiency one day after cessation of glucocorticoid therapy. Four children (17%) had ongoing adrenal insufficiency after 7 to 14 days, and three children (13%) after 28 days. All children (100%) recovered in 10 weeks.

In another study, researchers evaluated two types of glucocorticoid treatment during both induction and reinduction phases of treatment (Salem 2015). During the induction phase, children whose condition was diagnosed before 2006 (dexamethasone group) received dexamethasone 6 mg/m2/d for 28 consecutive days with an unknown duration of tapering (n = 20), and children whose condition was diagnosed during and after 2006 (prednisone group) received prednisone 60 mg/m2/d for 28 consecutive days with an unknown duration of tapering (n = 20). During the reinduction phase, the dexamethasone group received dexamethasone 6 mg/m2/d for 21 days with an unknown duration of tapering, and the prednisone group received prednisone 60 mg/m2/d for 21 days with an unknown duration of tapering. Time that passed between induction and reinduction was not mentioned by trial authors. Adrenal function was tested at diagnosis (before start of therapy) and during both induction and reinduction at the following time points: immediately after the last steroid course (week 0), two weeks after completion of glucocorticoid treatment (week 2), four weeks after completion of glucocorticoid treatment (week 4), and every two weeks thereafter until HPA axis recovery was reached. Moreover, adrenal function was tested during infectious events. This study found that children who received dexamethasone had a statistically significantly longer duration of adrenal insufficiency than those who received prednisone (P < 0.005). At diagnosis, adrenal insufficiency was present in 11 out of 40 children (27.5%) ‐ 5 out of 20 (25%) in the dexamethasone group and 6 out of 20 (30%) in the prednisone group. In the dexamethasone group, 50% to 55% of participants had adrenal insufficiency at week 4, and 25% to 35% had ongoing insufficiency at week 6 during both induction and reinduction. All children recovered in 20 weeks. Mean duration to adrenal recovery was 7.20 weeks for the dexamethasone group (during both induction and reinduction). In the prednisone group, 25% to 35% showed adrenal insufficiency at week 2, and 5% to 30% had ongoing adrenal insufficiency at weeks 4 and 6, during both induction and reinduction. All children recovered in 20 weeks. Mean time until adrenal recovery was 3.00 and 4.80 weeks (respectively, during induction and reinduction) in the prednisone group. Trial authors did not provide specific data on participant levels or for each time point.

Another study evaluated 28 days of prednisone 60 mg/m2/d with nine days of tapering (n = 40) (Perdomo‐Ramírez 2016). Researchers included in the study only children without adrenal insufficiency before the start of treatment; therefore, basal cortisol and ACTH levels were normal in all children at the start of therapy. Three days after cessation of glucocorticoid therapy, adrenal insufficiency was present in 29 out of 40 children (72.5%). One of these children died before the second evaluation on day 7 after cessation of glucocorticoid therapy. Of the remaining 39 participants, 11 had adrenal insufficiency on day 7 after cessation (28.2%), and three (7.7%) had ongoing adrenal insufficiency 14 days after cessation of glucocorticoid treatment. All children (100%) recovered 30 days after cessation of glucocorticoid treatment.

Low‐dose ACTH stimulation test with lower cutoff limit

One study used the low‐dose ACTH stimulation test with a lower cutoff limit of 14.2 µg/dL (392 nmol/L) (Kuperman 2012). This cutoff value was determined by a control group of 16 children who underwent the low‐dose ACTH test because they were suspected of having any endocrinopathy other than adrenal insufficiency. Investigators examined two randomised arms of glucocorticoid therapy: 28 days of prednisone 40 mg/m2/d without tapering (n = 16), and 28 days of dexamethasone 6 mg/m2/d without tapering (n = 13). Before the start of glucocorticoid therapy, basal cortisol values were abnormal in three children (one in the prednisone arm and two in the dexamethasone arm). Children in both treatment arms (prednisone and dexamethasone) displayed similar mean peak cortisol levels before treatment. Thereafter, a low‐dose ACTH test was performed weekly for eight weeks. However, not all children underwent the stimulation test every week. Information on individual testing schedules and on adrenal recovery per child was not provided. In the prednisone arm, 7 out of 14 children (50%) had insufficient cortisol levels one week after the start of glucocorticoid therapy. Five children had insufficient cortisol levels in the second, third, and fourth weeks of therapy, out of 14 children (36%), 15 children (33%), and 14 children (36%), respectively. During week 5 ‐ the first week after cessation of glucocorticoid therapy ‐ 3 out of 13 children (23%) had adrenal insufficiency. During the second week after cessation (week 6), 4 out of 14 children (29%) had adrenal insufficiency. During the third week after cessation (week 7), 5 out of 14 children (36%) had adrenal insufficiency. At the end of follow‐up (week 8) ‐ four weeks after cessation of glucocorticoid therapy, 4 out of 15 children (27%) remained insufficient. In the dexamethasone arm, 4 out of 10 children (40%) had insufficient cortisol levels one week after the start of glucocorticoid therapy. During the second week of therapy, two out of seven children (29%) had adrenal insufficiency; in the third week, 5 out of 13 children (38%) were insufficient; and in the last week (week 4), only 1 out of 12 children (8%) had abnormal adrenal function. However, in subsequent weeks after cessation of glucocorticoid therapy, 3 out of 13 children (13%) and 5 out of 11 children (45%) were insufficient during week 5 and week 6, respectively. In week 7, 3 out of 10 (30%) children had adrenal insufficiency. At the end of follow‐up (week 8) ‐ four weeks after cessation of glucocorticoid therapy ‐ 3 out of 12 children (25%) remained insufficient.

See Table 6 for an overview of the prevalence and duration of adrenal insufficiency in studies that used a low‐dose ACTH test.

Open in table viewer
Table 6. Prevalence and duration of adrenal insufficiency evaluated by a low‐dose ACTH stimulation test

Mahachoklertwattana et al

Therapy: prednisolone (cumulative dose 1120 mg/m2)a

Time after cessation

2 weeks

4 weeks

8 weeks

12 weeks

20 weeks

n insufficient/n totalb

11/24

9/24

7/24

3/24

3/24

Rix et al (1)

Therapy: prednisolone (cumulative dose 2257.5 mg/m2)

Time after cessation

Before

1 day

3 days

5 days

n insufficient/n totalb

0/13

16/17

8/15

8/17

Rix et al (2)

Therapy: prednisolone (cumulative dose 420 mg/m2)c

Time after cessation

Before

2 days

n insufficient/n totalb

2/13

13/13

Rix et al (3)

Therapy: dexamethasone (cumulative dose 236.25 mg/m2)c

Time after cessation

Before

1 day

3 days

7 days

n insufficient/n totalb

0/5

2/2

3/5

1/5

Einaudi et al (1)

Therapy: prednisone (cumulative dose 1477.5 mg/m2)d

Time after cessation

1 day

7 to 14 days

28 days

42 days

10 weeks

n insufficient/n totalb

32/40

8/32

5/8

5/5

0/5

Einaudi et al (2)

Therapy: dexamethasone (cumulative dose 246.25 mg/m2)d

Time after cessation

1 day

7 to 14 days

28 days

42 days

10 weeks

n insufficient/n totalb

20/24

4/20

3/4

3/3

0/3

Kuperman et al 2012 (1)

Therapy: prednisone (cumulative dose 1120 mg/m²)

Time after cessation

Before (4 weeks after start of glucocorticoid treatment)

1 week

2 weeks

3 weeks

4 weeks

n insufficient/n totalb

5/14

3/13

4/14

5/14

4/15

Kuperman et al 2012 (2)

Therapy: dexamethasone (cumulative dose 168 mg/m²)

Time after cessation

Before (4 weeks after start of glucocorticoid treatment)

1 week

2 weeks

3 weeks

4 weeks

n insufficient/n totalb

1/12

3/13

5/11

3/10

3/12

Perdomo‐Ramírez et al 2015

Therapy: prednisone (cumulative dose 1837.5 mg/m2)

Time after cessation

Before

3 days

1 week

2 weeks

4 weeks

n insufficient/n totalb

0/40

29/40

11/28

3/11

0/3

Salem et al 2015 (1)

(both induction and reinduction phases)

Therapy: dexamethasone (cumulative dose unknown)

Time after cessation

Before (at diagnosis)

1 day to 18 weeks

20 weeks

n insufficient/n totalb

5/20

No data on patient level were available.

All children recovered.e

Salem et al 2015 (2)

(both induction and reinduction phases)

Therapy: prednisone (cumulative dose unknown)

Time after cessation

Before (at diagnosis)

1 day to 18 weeks

20 weeks

n insufficient/n totalb

6/20

No data on patient level were available.

All children recovered.e

ACTH, adrenocortiotropic hormone.
a Four weeks after completion of induction therapy, children received maintenance therapy consisting of a 7‐day course of high‐dose dexamethasone 8 mg/m2/d every 4 weeks. Cumulative dose depended on how long the child had been followed up.

b If not all children were tested at all time points, then "n total" = n tested.

c All children first received prednisolone (cumulative dose 2257.5 mg/m2). 

d After 7 days of prednisone (60 mg/m2/d, cumulative dose 420 mg/m2).

eFrom 4 weeks after cessation of glucocorticoid therapy, only children who were adrenal insufficient at that time point underwent further low‐dose ACTH testing every 2 weeks until adrenal recovery. Therefore it is unknown how many children were tested at 20 weeks (in both induction and reinduction phases).

Basal morning cortisol values

Two studies used basal morning cortisol values in measuring adrenal function. One study, which included 35 children, found that median basal cortisol levels were inhibited on the eighth day (1.2 µg/dL, range 0.9 to 132.7 µg/dL) and on the 28th day (0.9 µg/dL, range 0.9 to 6.6 µg/dL) of 28 days of dexamethasone 6 mg/m2/d and 48 hours after cessation (over 10 days) of dexamethasone treatment (2.4 µg/dL, range 0.9 to 11.2 µg/dL) compared with pre‐glucocorticoid therapy levels (17.5 µg/dL, range 7.6 to 40.9 µg/dL) (P = 0.01 for the three tests vs pre‐glucocorticoid levels) (Cunha 2004). Median basal cortisol levels one month after cessation of dexamethasone treatment (12.4 µg/dL, range 1.8 to 29.0 µg/dL), although slightly lower, did not show a significant difference compared with pre‐glucocorticoid therapy levels. No data on participant levels were provided.

In another study, which included 15 children, mean basal cortisol levels (± standard error of the mean) were significantly (P < 0.05) lower on day 7 (10.8 ± 1.0 µg/dL) and day 14 (11.5 ± 2.0 µg/dL) after abrupt cessation of 42 days of dexamethasone 6 mg/m2/d than at pretreatment (17.8 ± 1.3 µg/dL) (Kuperman 2001). Levels at day 7 and day 14 did not differ significantly. Additional information provided by trial authors revealed that all children (100%) had sufficient basal cortisol levels at diagnosis (> 7 µg/dL), whereas 4 out of 15 children (27%) had insufficient basal cortisol levels seven days after cessation of dexamethasone therapy. One child was lost to follow‐up thereafter. Fourteen days after cessation of dexamethasone therapy, 4 out of 14 children (29%) had insufficient basal cortisol levels. It should be noted that one of them had a sufficient basal cortisol level seven days earlier.

See Table 7 for an overview of the prevalence and duration of adrenal insufficiency in studies that used basal morning cortisol values.

Open in table viewer
Table 7. Prevalence and duration of adrenal insufficiency evaluated by basal morning cortisol values

Cunha et al

No data on patient levels were available.

Time after cessation

n insufficient/n total

Kuperman et al 2001

Therapy: dexamethasone (cumulative dose 252 mg/m2)a

Time after cessation

Before

1 week

2 weeks

n insufficient/n total

0/15

4/15

4/14

a Results based on basal cortisol levels; cutoff level 7 µg/dL = 194 nmol/L.

Risk factors for adrenal insufficiency

Four of the included studies addressed risk factors for development or persistence of adrenal insufficiency (Einaudi 2008; Kuperman 2012; Petersen 2003; Salem 2015).

Type of glucocorticoid

Three studies, including two RCTs, investigated differences between prednisone and dexamethasone in occurrence and/or duration of adrenal insufficiency Einaudi 2008; Kuperman 2012; Salem 2015). In one RCT, all children received prednisone before randomisation (Einaudi 2008). See results above. In summary, investigators in the two RCTs found no differences between prednisone and dexamethasone arms with regard to occurrence and duration of adrenal insufficiency (Einaudi 2008; Kuperman 2012). In the other (observational) study, children who received prednisone recovered earlier than those who received dexamethasone (Salem 2015).

Fluconazole

Two cohort studies evaluated fluconazole therapy as a risk factor for persistence of adrenal insufficiency. Three children in one study received fluconazole (Petersen 2003). Two of these children had ongoing adrenal insufficiency eight months after cessation of glucocorticoid therapy (dexamethasone). The third child recovered after three weeks. In the other study, up to 80% of all children received fluconazole at some point during induction or reinduction therapy (Salem 2015). Half of these children received a dosage above 10 mg/kg/d. These children had a longer duration of adrenal insufficiency than those who received lower doses of fluconazole (< 10 mg/kg/d). Investigators did not further specify this information and did not provide data on participant levels.

Infection/stress

Two studies investigated the presence of infection/stress as a risk factor for adrenal insufficiency (Kuperman 2012; Salem 2015). One of these trialsrandomly assigned children to receive prednisone or dexamethasone (Kuperman 2012). Investigators defined episodes of infection/stress by hospitalisation due to fever with or without neutropenia. In the prednisone group, six episodes of infection/stress occurred in children with adrenal insufficiency, and eight episodes in children with adequate cortisol levels (14 episodes of infection/stress in total). In the dexamethasone group, one episode of infection/stress occurred in a child with an insufficient cortisol level, and 10 episodes in children with adequate cortisol levels (11 episodes of infection/stress in total). Trial authors found no relationship between the presence of infection/stress and adrenal insufficiency. In the other cohort study, children received prednisone or dexamethasone according to their treatment protocol (which differed between children whose condition was diagnosed before 2006 and those whose condition was diagnosed during and after 2006) (Salem 2015). For both dexamethasone and prednisone groups, longer duration of adrenal insufficiency was associated with increased occurrence of infection (P = 0.002 and 0.005, respectively). Investigators did not further specify this information and provided no data on participant levels.

Other outcome measures

Owing to heterogeneity, it was not possible to identify whether adrenal function after administration of glucocorticoids was dependent on the moment of testing; the (cumulative) dose, type, or duration of glucocorticoid therapy; or the method of cessation of glucocorticoid therapy.

Discussion

With improvement in survival among individuals with childhood acute lymphoblastic leukaemia (ALL), treatment‐related side effects have become increasingly relevant. Glucocorticoids play an important role in the treatment of those with childhood ALL, but supraphysiological doses may suppress the hypothalamic‐pituitary‐adrenal (HPA) axis, resulting in an impaired stress response and inadequate defence against infection (Henzen 2000; Krasner 1999). Children with HPA axis suppression may benefit from glucocorticoid replacement therapy (e.g. hydrocortisone) to reduce the risk of life‐threatening complications. HPA axis suppression commonly occurs in the first days after cessation of glucocorticoid therapy, but its exact duration is unclear. Adequate guidelines for glucocorticoid substitution are lacking. This is the second update of the first systematic review conducted to evaluate HPA axis function after treatment with glucocorticoid therapy for childhood ALL (Gordijn 2012; Gordijn 2015).

We identified two new studies published since the last review update (Gordijn 2015). In total, we identified 10 studies evaluating adrenal function after treatment with glucocorticoid therapy for childhood ALL, including two randomised controlled trials (RCTs). None of these studies evaluated the HPA axis at the level of the hypothalamus, the pituitary, or both. Owing to substantial differences in types and (cumulative) doses of glucocorticoids used, in duration and method of cessation of glucocorticoid therapy, and in methods of testing adrenal function, pooling of results was not possible. This systematic review used a very broad search strategy in identifying eligible studies. However, although it is unlikely that eligible studies were missed, it is never possible to rule out reporting bias.

Included studies showed that adrenal insufficiency occurs in almost all patients during the first days after cessation of glucocorticoid therapy for childhood ALL. Most children recovered from adrenal insufficiency within seven weeks. However, a small number of children had ongoing adrenal insufficiency lasting up to 34 weeks. At first impression, no significant differences in the occurrence and duration of adrenal insufficiency between different types, (cumulative) doses, durations, and methods of cessation of glucocorticoid therapy are evident, but because of heterogeneity between studies, we were not able to assess this further. Owing to this limitation and to the small numbers of children enrolled in the included studies (i.e. low power), we can provide no definitive conclusions. The two studies designed as RCTs enabled comparison between two different types of glucocorticoid therapy: prednisone versus dexamethasone (Einaudi 2008; Kuperman 2012). In one of these RCTs, both treatment groups received prednisone before randomisation (Einaudi 2008). The occurrence and duration of adrenal insufficiency did not differ between prednisone and dexamethasone (± prednisone) treatment arms. However, because of the low power of included RCTs, we cannot provide definitive conclusions on this topic. One observational study compared prednisone and dexamethasone as well. This study found that children who received dexamethasone had a longer duration of adrenal insufficiency than those given prednisone (mean duration to recovery 7.20 weeks in the dexamethasone group (during both induction and reinduction) vs 3.00 and 4.80 weeks (respectively, during induction and reinduction) in the prednisone group; P < 0,005) (Salem 2015). However, this comparison involved two (non‐random) groups of children on different ALL treatment protocols. Children whose condition was diagnosed before 2006 received dexamethasone according to the CCG‐1991 protocol, and those whose condition was diagnosed during and after 2006 received prednisone according to the modified BFM‐1990 protocol. Thus, results of comparisons of glucocorticoid therapy reported in this study should be interpreted with caution.

Previous studies demonstrated adrenal suppression after high‐dose fluconazole therapy (Albert 2001; Shibata 2001). Only two of the studies included in this review reported on fluconazole therapy (Petersen 2003; Salem 2015). In one of these studies, two of the three children receiving fluconazole had ongoing adrenal insufficiency eight months after cessation of dexamethasone therapy, whereas the third child recovered after three weeks (Petersen 2003). It should be considered that fluconazole therapy may have influenced the duration of adrenal insufficiency in these children. In the other study, up to 80% of children received fluconazole somewhere during induction or reinduction (Salem 2015). Half of these children received a dose higher than 10 mg/kg/d. Compared with children who received lower doses of fluconazole (< 10 mg/kg/d), these children had a prolonged duration of adrenal insufficiency (not specified). Investigators did not provide data on participant levels.

Two of the studies included in this review evaluated the presence of infection and/or stress episodes as a risk factor for adrenal insufficiency (Kuperman 2012; Salem 2015). One study randomly assigned children to receive prednisone or dexamethasone (Kuperman 2012). In the prednisone group, 14 episodes of infection/stress occurred (defined by hospitalisation due to fever with or without neutropenia). Six of these episodes occurred in children with insufficient cortisol levels; the other eight episodes concerned children with adequate cortisol levels. In the dexamethasone group, 11 episodes of infection/stress occurred: one in a child with an insufficient cortisol level, and the other 10 in children with adequate cortisol levels. In conclusion, trial authors found no relationship between the presence of infection/stress and adrenal insufficiency. In the other study, longer duration of adrenal insufficiency was associated with increased infection in both dexamethasone and prednisone groups (P = 0.002 and 0.005, respectively) (Salem 2015). Investigators did not further specify this information. In conclusion, it would be relevant to investigate this relationship further in a larger, separate study specifically designed for this purpose.

Owing to the paucity of RCTs on HPA axis suppression after glucocorticoid therapy in childhood ALL, most of the studies included in this systematic review were uncontrolled studies. We identified only two RCTs. The lack of control groups made it impossible for review authors to evaluate possible causes of HPA axis suppression other than glucocorticoid therapy. Moreover, all of the included studies used biochemical markers to evaluate adrenal insufficiency; only two studies additionally discussed the clinical consequences of adrenal insufficiency (Kuperman 2012; Salem 2015). All included studies had risk of bias issues, but currently they provide the best available evidence on occurrence and duration of adrenal insufficiency after glucocorticoid therapy in childhood ALL.

Study flow diagram.
Figures and Tables -
Figure 1

Study flow diagram.

Table 1. 'Risk of bias' assessment criteria for observational studies

Internal validity

External validity

Study group

Selection bias (representative: yes/no):

  • if it consisted of more than 90% of the original cohort of children with ALL treated with glucocorticoids; or

  • if it was a random sample with respect to treatment.

Reporting bias (well defined: yes/no):

  • if treatment protocol was mentioned; and

  • if (cumulative) dose of glucocorticoid treatment was mentioned; and

  • if type of glucocorticoid treatment was mentioned; and

  • if duration of glucocorticoid treatment was mentioned; and

  • if method of cessation of glucocorticoid treatment was mentioned.

Follow‐up

Attrition bias (adequate: yes/no):

  • if outcome was assessed at end date of the study for 60% to 90% of study group; or

  • if outcome was assessed for more than 90% of the study group but with an unknown end date.

Reporting bias (well defined: yes/no):

  • if length of follow‐up was mentioned; and

  • if frequency of measuring outcomes was mentioned.

Outcome

Detection bias (blinding: yes/no):

  • if outcome assessor was blinded to glucocorticoid treatment.

Reporting bias (well defined: yes/no):

  • if methods of detection were described; and

  • if outcome definition was objective and precise.

Risk estimation

Confounding (adjustment for other factors: yes/no):

  • if important prognostic factors (i.e. age, sex, cotreatment) or follow‐up was taken adequately into account.

Analysis (well defined: yes/no):

  • if risk ratio, odds ratio, attributable risk, linear or logistic regression model, mean difference, or Chi2 statistic was calculated.

ALL: acute lymphoblastic leukaemia.

Figures and Tables -
Table 1. 'Risk of bias' assessment criteria for observational studies
Table 2. 'Risk of bias' assessment criteria for randomised controlled trials

Selection bias

Sequence generation (adequate: yes/no):

  • if the rule for allocating interventions to participants was based on some chance (random) process.

Allocation concealment (adequate: yes/no):

  • if the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Performance bias
Blinding of care providers (yes/no):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Blinding of participants (yes/no):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Detection bias
Blinding of outcome assessors (yes/no; assessed for each outcome separately):

  • if knowledge of the allocated intervention was adequately prevented during the study.

Attrition bias
Incomplete outcome data (adequate: yes/no; assessed for each outcome separately):

  • if incomplete outcome (attrition and exclusions) data have been adequately addressed.

Reporting bias
Selective outcome reporting (yes/no):

  • if reports of the study were free of the suggestion of selective outcome reporting.

Other bias
Other bias (yes/no):

  • if the study was free of other problems (i.e. potential source of bias related to specific study design, premature termination of the study due to some data‐dependent process, extreme baseline imbalance) that could put it at high risk of bias.

Figures and Tables -
Table 2. 'Risk of bias' assessment criteria for randomised controlled trials
Table 3. Risk of bias in included observational studies

Study

Representative study group

Complete follow‐up assessment

Blinded outcome assessor

Adjustment for important confounders

Well‐defined study group

Well‐defined follow‐up

Well‐defined outcome

Well‐defined risk estimation

Cunha 2004

No, based on additional information provided by trial authors, the study group described did not consist of more than 90% of the original cohort and was not a random sample.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, treatment protocol and (cumulative) dose, type, duration, and form of cessation of glucocorticoid treatment were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Felner 2000

Yes, based on additional information provided by trial authors, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

No, treatment protocol was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Kuperman 2001

Yes, based on additional information provided by trial authors, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

No, treatment protocol was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Mahachoklertwattana 2004

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, treatment protocol and (cumulative) dose, type, duration, and form of cessation of glucocorticoid treatment were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Perdomo‐Ramírez 2016

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

No, outcome assessor was not blinded to glucocorticoid treatment. This information was based on additional information provided by trial authors.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid therapy were mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Yes, mean difference was calculated.

Petersen 2003

Yes, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment.

Yes, important prognostic factors or follow‐up was taken into account.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid treatment were mentioned. Information on the method of cessation of glucocorticoid treatment was based on additional information provided by trial authors.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

No, risk ratio, odds ratio, attributable risk, linear or logistic regression model, mean difference, or Chi2 statistic was not calculated.

Rix 2005

Yes, the study group described consisted of more than 90% of the original cohort.

Yes, outcome was assessed for 60% to 90% of the study group at the end date of the study.

No, outcome assessor was not blinded to glucocorticoid treatment.

Yes, treatment protocol and (cumulative) dose, type, and duration of glucocorticoid treatment were mentioned. Information on the method of cessation of glucocorticoid treatment was based on additional information provided by trial authors.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Salem 2015

Unclear whether the study group consisted of more than 90% of the original cohort, or if it was a random sample

Unclear whether outcome was assessed for 60% to 90% at the end date of the study.

Unclear whether outcome assessor was blinded to glucocorticoid treatment

Yes, important prognostic factors or follow‐up was taken into account.

No, duration of tapering of glucocorticoid treatment was not mentioned.

Yes, length of follow‐up and frequency of measuring were mentioned.

Yes, methods of detection were described, and outcome definition was objective and precise.

Yes, mean difference was calculated.

Figures and Tables -
Table 3. Risk of bias in included observational studies
Table 4. Risk of bias in included randomised controlled trials

Study

Adequate sequence generation?

Adequate allocation concealment?

Blinding?

Incomplete outcome data addressed?

Free of selective reporting?

Free of other bias?

Einaudi 2008

Yes, according to additional information provided by trial authors, the rule for allocating interventions to children was based on some chance (random) process.

Yes, according to additional information provided by trial authors, the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Based on additional information provided by trial authors, care providers, children, and outcome assessors were not blinded.

Yes, no outcome data were missing.

No, "adrenal function completely recovered in the 12 children evaluated with subsequent low‐dose ACTH test (in 4, 3, and 5 patients after 4, 8, and 10 weeks, respectively)". However, it was not reported which of these children received prednisone and which received dexamethasone. Therefore, not all of the study's prespecified primary outcomes were reported.

Yes

Kuperman 2012

Yes, the rule for allocating interventions to children was based on some chance (random) process.

Yes, according to additional information provided by trial authors, the randomisation method did not allow investigator and child to know or influence allocation of treatment before eligible children entered the study.

Yes, based on additional information provided by trial authors, care providers, children, and outcome assessors were all blinded.

Outcomes were assessed for 83% to 93% of the study population.

Yes

Yes

ACTH: adrenocorticotropic hormone.

Figures and Tables -
Table 4. Risk of bias in included randomised controlled trials
Table 5. Prevalence and duration of adrenal insufficiency evaluated by an ACTH stimulation test

Felner et al

Therapy: dexamethasone (cumulative dose 168 mg/m2)

Time after cessation

Before

1 day

4 weeks

8 weeks

n insufficient/n total

0/10

10/10

3/10

0/10

Petersen et al (1)

Therapy: prednisolone (cumulative dose 2257.5 mg/m2)a

Time after cessation

1 week

3 weeks

7 weeks

End of follow‐up: 10, 11, 11, and 19 weeks, respectively

n insufficient/n total

7/10

6/10

4/10

4/10

Petersen et al (2)

Therapy: dexamethasone (cumulative dose 236.25 mg/m2)b

Time after cessation

1 week

3 weeks

7 weeks

End of follow‐up: 16, 33, and 34 weeks, respectively

n insufficient/n total

5/7

4/7

3/7

3/7

ACTH: adrenocorticotropic hormone.
a One child received additional 840 mg/m2 prednisolone during the period of adrenal insufficiency.

b These children received prednisolone 2257.5 mg/m2 as induction therapy before. Three high‐risk patients received an additional 560 mg/m2 prednisolone in advance. Furthermore, owing to persistent adrenal insufficiency, one of these high‐risk children received an additional 420 mg/m2 prednisolone during the period of insufficiency, and the other two high‐risk children received an additional 1260 mg/m2 prednisolone during that period. 

Figures and Tables -
Table 5. Prevalence and duration of adrenal insufficiency evaluated by an ACTH stimulation test
Table 6. Prevalence and duration of adrenal insufficiency evaluated by a low‐dose ACTH stimulation test

Mahachoklertwattana et al

Therapy: prednisolone (cumulative dose 1120 mg/m2)a

Time after cessation

2 weeks

4 weeks

8 weeks

12 weeks

20 weeks

n insufficient/n totalb

11/24

9/24

7/24

3/24

3/24

Rix et al (1)

Therapy: prednisolone (cumulative dose 2257.5 mg/m2)

Time after cessation

Before

1 day

3 days

5 days

n insufficient/n totalb

0/13

16/17

8/15

8/17

Rix et al (2)

Therapy: prednisolone (cumulative dose 420 mg/m2)c

Time after cessation

Before

2 days

n insufficient/n totalb

2/13

13/13

Rix et al (3)

Therapy: dexamethasone (cumulative dose 236.25 mg/m2)c

Time after cessation

Before

1 day

3 days

7 days

n insufficient/n totalb

0/5

2/2

3/5

1/5

Einaudi et al (1)

Therapy: prednisone (cumulative dose 1477.5 mg/m2)d

Time after cessation

1 day

7 to 14 days

28 days

42 days

10 weeks

n insufficient/n totalb

32/40

8/32

5/8

5/5

0/5

Einaudi et al (2)

Therapy: dexamethasone (cumulative dose 246.25 mg/m2)d

Time after cessation

1 day

7 to 14 days

28 days

42 days

10 weeks

n insufficient/n totalb

20/24

4/20

3/4

3/3

0/3

Kuperman et al 2012 (1)

Therapy: prednisone (cumulative dose 1120 mg/m²)

Time after cessation

Before (4 weeks after start of glucocorticoid treatment)

1 week

2 weeks

3 weeks

4 weeks

n insufficient/n totalb

5/14

3/13

4/14

5/14

4/15

Kuperman et al 2012 (2)

Therapy: dexamethasone (cumulative dose 168 mg/m²)

Time after cessation

Before (4 weeks after start of glucocorticoid treatment)

1 week

2 weeks

3 weeks

4 weeks

n insufficient/n totalb

1/12

3/13

5/11

3/10

3/12

Perdomo‐Ramírez et al 2015

Therapy: prednisone (cumulative dose 1837.5 mg/m2)

Time after cessation

Before

3 days

1 week

2 weeks

4 weeks

n insufficient/n totalb

0/40

29/40

11/28

3/11

0/3

Salem et al 2015 (1)

(both induction and reinduction phases)

Therapy: dexamethasone (cumulative dose unknown)

Time after cessation

Before (at diagnosis)

1 day to 18 weeks

20 weeks

n insufficient/n totalb

5/20

No data on patient level were available.

All children recovered.e

Salem et al 2015 (2)

(both induction and reinduction phases)

Therapy: prednisone (cumulative dose unknown)

Time after cessation

Before (at diagnosis)

1 day to 18 weeks

20 weeks

n insufficient/n totalb

6/20

No data on patient level were available.

All children recovered.e

ACTH, adrenocortiotropic hormone.
a Four weeks after completion of induction therapy, children received maintenance therapy consisting of a 7‐day course of high‐dose dexamethasone 8 mg/m2/d every 4 weeks. Cumulative dose depended on how long the child had been followed up.

b If not all children were tested at all time points, then "n total" = n tested.

c All children first received prednisolone (cumulative dose 2257.5 mg/m2). 

d After 7 days of prednisone (60 mg/m2/d, cumulative dose 420 mg/m2).

eFrom 4 weeks after cessation of glucocorticoid therapy, only children who were adrenal insufficient at that time point underwent further low‐dose ACTH testing every 2 weeks until adrenal recovery. Therefore it is unknown how many children were tested at 20 weeks (in both induction and reinduction phases).

Figures and Tables -
Table 6. Prevalence and duration of adrenal insufficiency evaluated by a low‐dose ACTH stimulation test
Table 7. Prevalence and duration of adrenal insufficiency evaluated by basal morning cortisol values

Cunha et al

No data on patient levels were available.

Time after cessation

n insufficient/n total

Kuperman et al 2001

Therapy: dexamethasone (cumulative dose 252 mg/m2)a

Time after cessation

Before

1 week

2 weeks

n insufficient/n total

0/15

4/15

4/14

a Results based on basal cortisol levels; cutoff level 7 µg/dL = 194 nmol/L.

Figures and Tables -
Table 7. Prevalence and duration of adrenal insufficiency evaluated by basal morning cortisol values