Recovery of Hypothalamic-Pituitary-Adrenal Axis in Pediatric Cushing Disease

Summary

Objectives:

The postoperative period of Cushing disease (CD) is complicated by a phase of adrenal insufficiency (AI). Factors that influence the duration of AI and its prognostic significance for CD recurrence in children have not been extensively studied. We investigated whether clinical or biochemical factors contribute to the duration of AI, and the correlation of the recovery process with the risk for recurrence.

Design:

Patients with pediatric-onset CD who were followed up for at least 3 months after transsphenoidal surgery (TSS) (n=130) were included in the study. Multivariable Cox proportional hazards analysis was used to assess the association of biochemical and clinical factors with duration of AI.

Results:

Overall, 102 patients recovered adrenal function during their follow-up. Median time to recovery was 12.7 months [95%Confidence Intervals (CI): 12.2-13.4]. On multivariable analysis, clinical (age, gender, disease duration, puberty stage, BMI z-score, tumor size, invasion of the cavernous sinus, year of surgery) and biochemical (midnight cortisol, morning ACTH) factors did not correlate with the time to recovery, except for increase of recovery time noted with increase of urinary free cortisol (UFC) [Hazard Ratio (HR):0.94, 95%CI:0.89-0.99]. Among patients who eventually recovered adrenal function, the risk for CD recurrence was associated with the time to recovery (HR:0.86, 95%CI:0.75-0.99).

Conclusions:

Recovery of adrenal function in patients with CD after TSS may not be associated with most clinical and biochemical factors in the preoperative period except for total cortisol excretion, as evidenced by the correlation with UFC. Earlier recovery is associated with higher risk for recurrence, which has implications for the patients’ follow up and counseling.

Introduction

Cushing disease (CD), caused by adrenocorticotropic hormone (ACTH) secreting pituitary adenomas, is the most common etiology of Cushing syndrome (CS) in children older than 7 years of age.¹ Successful management of CD is followed by a postoperative period of adrenal insufficiency (AI), which is the result of suppression of the normal hypothalamic-pituitary-adrenal (HPA) axis by endogenous hypercortisolemia.^2,3 The duration of postoperative AI and the mechanism of HPA axis recovery are not clearly understood, but it is thought that the rate-limiting step lies on the recovery of corticotrophin-releasing hormone (CRH) secretion by hypothalamic cells.⁴

Previous studies, mostly in adults, tried to identify factors that may affect HPA axis recovery. Although it is tempting to hypothesize that the higher or longer the exposure to glucocorticoids, the more severe the suppression, studies have not consistently found that clinical or biochemical factors significantly contribute to the duration of HPA suppression. ^5-9 Furthermore, the range of the reported median time to recovery varies between various cohorts (ranging from 6 months to more than 2 years). Of note, what has been more commonly reported is the association of early HPA axis recovery with higher risk for CD recurrence.^5,7 However, only one study, from our center, has previously described the HPA axis recovery in pediatric patients, who may behave differently than adults due to varying sensitivity or maturation of the HPA axis, and this prior study because of its lower numbers did not allow for a meaningful statistical analysis of risk for recurrence.¹⁰

In the present study, we investigated the association of the time to recovery of HPA axis after successful treatment in a large cohort of pediatric patients with CD with the age of diagnosis and other relevant clinical and biochemical factors. As a secondary aim, we also queried the possible association of recovery time with risk for recurrence that was suggested by our previous inconclusive reports.

Subjects and Methods

Subjects

Subjects enrolled under protocol 97-CH-0076 designed for genetic investigation and treatment of pituitary tumors, were screened for eligibility in the study. Patients were included based on the following criteria: children (< 18 years at presentation) with diagnosis of CD as defined by the Endocrine Society Guidelines and previously described [abnormal measurements in at least two of the following criteria: elevated 24 hour urinary free cortisol (UFC), elevated midnight serum cortisol [>121 nmol/L (4.4 mcg/dl) in children], and failure to suppress cortisol to 1mg (or weight-based equivalent dose) overnight dexamethasone suppression test (post-dexamethasone cortisol > 50 nmol/L (1.8 mcg/dL)],^11,12 successful surgical treatment [defined as postoperative cortisol nadir level < 138 nmol/L (5 mcg/dL)], and minimum postoperative follow up of 3 months. All patients underwent transsphenoidal surgery (TSS) for resection of the pituitary adenoma, either at our Institute or at another hospital. Only patients after first TSS were included in the study, to avoid dependent events in the survival analysis or effect of multiple surgical interventions on the outcome. Patients who received intracranial/pituitary radiation or medical therapy after surgery for inhibition of cortisol production, such as steroidogenesis inhibitors, were likewise excluded. Some of the patients included in the current manuscript (n= 44) were previously described by Lodish et al.¹⁰ After surgery, all patients were prescribed oral hydrocortisone at 8-12mg/m²/day and weaning of the dose was performed according to previously described recommendations and guided by their biochemical results and sympotms.¹⁰

Informed consent was obtained from parents and assent from patients if developmentally appropriate. All study procedures were approved by the Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD) Institutional Review Board (IRB).

Clinical and biochemical data

Clinical data and biochemical results were extracted from electronic medical records. Duration of disease was calculated as the time between the age at weight gain in excess of previous percentile, as noted on patient’s growth chart, and the age at first biochemical diagnosis. Height, weight and body mass index (BMI) z-scores were calculated based on age- and sex-matched references from the 2000 CDC growth charts. Serum cortisol and plasma ACTH levels were calculated as the average value of the corresponding levels performed at 23:30 h and 00:00 h (reported as midnight values) and 07:30 h and 08:00 h (reported as morning values). Twenty-four-hour urinary free cortisol (UFC) was calculated as the average of the first 2-3 samples reported in the electronic medical records. Given the possible differences in the assays and reported reference range for UFC, we calculated the increase of UFC based on the upper limit of normal (ULN) according to the following formula: UFCxULN= UFC/Upper limit of the reference range. Serum cortisol was measured with solid-phase, competitive chemiluminescent enzyme immunoassay on Siemens Immulite 2500 analyzer (Malvern, PA, USA). Plasma ACTH was measured with chemiluminescence immunoassay on Siemens Immulite 200 XPi analyzer (Malvern, PA, USA). UFC was measured with chemiluminescent enzyme immunoassay until 2011 and with High Performance Liquid Chromatography/ Tandem Mass Spectrometry since 2011(LC-MS/MS).

Primary and Secondary Endpoints

The primary outcome was time from first TSS to recovery as defined by a serum cortisol level of > 500 nmol/L (18 mcg/dL). All patients had a morning cortisol measurement (8:00 am). If the serum morning cortisol was > 500 nmol/L (18 mcg/dL), no further evaluation was performed, and it was considered that the patient had recovered. If morning cortisol was < 500 nmol/L (18 mcg/dL), a standard dose ACTH stimulation test (performed with IV administration of 250 mcg of cosyntropin) was performed and peak cortisol > 500 nmol/L (18 mcg/dL) was consistent with recovery. The secondary outcome was time to recurrence as defined by the period between TSS and first biochemical confirmation of recurrence.

Statistical Analysis

Categorical data are presented as counts and proportions and compared between subjects with vs. without recurrence using χ² test or Fisher exact test, as appropriate. Continuous data were checked for normality based on histogram distribution and the Shapiro-Wilk statistical test. Continuous data with normal distribution are described as mean (standard deviation) and were compared between groups using student’s t-test. Continuous data without normal distribution are presented as median (interquartile range) and were compared between groups using Wilcoxon rank-sum test. Median time to recovery and 95% confidence intervals (CI) were calculated using the Kaplan Meier method. To investigate potential collinearity of the covariates of interest, the variance inflation factor (VIF) was calculated and covariates with VIF > 4 were considered as potentially collinear with others; the covariates morning cortisol vs. midnight cortisol, year of surgery vs. surgeon met this criterion. After excluding morning cortisol and year of surgery from the model, all remaining VIFs were <4. The proportional hazards (PH) assumption of the Cox PH test was tested by calculating the Schoenfeld residual using the cox.zph function. Covariates with p-values < 0.05 were further investigated through direct inspection of the zph plot. No evidence of non-proportionality was found. For continuous covariates, the linearity assumption was assessed by performing a lack-of-fit test comparing a linear fit with a non-linear fit based on a restricted cubic spline with 3 knots, and no evidence of a non-linear relationship was found. For the 11 model covariates considered in the regression analyses, 19.2% subjects were missing a value for at least one of the 11 covariates. However, only 7 of the 11 covariates were missing values and the percent of missing values ranged between 2.3% and 7.7%, well below the recommended maximum of 20% to 25%. Multiple imputation was implemented on the model covariates. Firstly, the missing data were filled in five times to generate five complete data sets as per the Full Conditional Specification method. Secondly, the five complete data sets were analyzed using standard statistical analyses; here, Cox proportional hazards regression. Lastly, the results from the five complete datasets were combined using Rubin’s rules to produce the final inferential results to account for the additional variability from the imputation. Results were considered significant if p-value was < 0.05. Statistical analyses were performed in R. Plots were created in R using the ggplot2 package.

Results

Characteristics of the patient cohort

Of the 231 pediatric patients with CD evaluated under protocol 97-CH-0076 from 1997 until 2019, 130 patients fulfilled the inclusion criteria. Patients were excluded because of persistent disease (n=32), insufficient follow up time or data (n=67) or lost to follow up before a TSS was performed (n=2) (Figure 1). Of patients with persistent disease, 10 patients underwent second TSS within one month after the first one, as per our institute’s protocol, and achieved remission. However, these patients were excluded from the analysis since they had two interventions. The baseline clinical and biochemical data of the included patients at the time of their diagnosis are presented in Table 1.

Figure 1.

Flowchart of included and excluded patients of the study.

Table 1.

Clinical and biochemical characteristics of the included patients at the time of diagnosis and based on the status of recurrence. BMI: Body mass index, IQR: interquartile range, UFCxULN: (Urinary free cortisol) x (Upper Limit of normal)

	No Recurrence (n=115)	Recurrence (n=15)	Overall (n=130)	p-value (Recurrence vs No recurrence)
Age at diagnosis (years)	13.1 (3.1)	13.2 (2.7)	13.1 (3.0)	.92
Gender				.32
Female	61 (53.0%)	10 (66.7%)	71 (54.6%)
Male	54 (47.0%)	5 (33.3%)	59 (45.4%)
Disease duration (years) *	2.45 (1.5)	3.00 (1.5)	2.50 (1.5)	.45
Puberty Tanner stage, n (%)				.14
Stage 1	31 (27.0%)	2 (13.3%)	33 (25.4%)
Stage 2	23 (20.0%)	4 (26.7%)	27 (20.8%)
Stage 3	11 (9.6%)	4 (26.7%)	15 (11.5%)
Stage 4	16 (13.9%)	1 (6.7%)	17 (13.1%)
Stage 5	27 (23.5%)	1 (6.7%)	28 (21.5%)
Height z-score at diagnosis	−1.14 (1.3)	−0.948 (1.4)	−1.12 (1.3)	.63
Weight z-score at diagnosis	1.41 (1.1)	1.38 (1.2)	1.41 (1.1)	.93
BMI z-score at diagnosis *	2.17 (0.6)	2.01 (0.9)	2.15 (0.9)	.7
Morning cortisol (nmol/L) *	491.1 (91.1)	513.2 (343.5)	491.1 (325.6)	.24
Midnight cortisol (nmol/L) *	416.6 (190.4)	507.7 (317.3)	430.4 (303.5)	.68
UFCxULN *	3.90 (5.4)	6.49 (4.5)	4.0 (4.8)	.45
Morning ACTH (pmol/L) *	8.3 (6.1)	9.9 (5.3)	8.4 (5.4)	.24
Tumor size (mm) *	4.00 (3.1)	5.50 (5)	4.0 (5.4)	.04
Cavernous sinus invasion, n (%)				.075
No	97 (84.3%)	8 (53.3%)	105 (80.8%)
Yes	14 (12.2%)	4 (26.7%)	18 (13.8%)

^*Corresponds to data described as median (Interquartile range)

Time to recovery of the complete cohort

Overall, 102 patients showed evidence of recovery of the HPA axis during their follow-up. Using a survival analysis, there was a 50% probability of recovery by a median time of 12.7 months postoperatively (95% CI: 12.2-13.4) (Figure 2). For those patients who recovered, the duration of postoperative AI ranged from 3 to 35 months, with median value of 12.3 months. Only 3 patients recovered within less than 3 months after TSS, while 3 patients recovered more than 2 years after surgery. Twenty-eight patients did not have recovery of the HPA axis by a median follow up time of 12.7 (18.3) months.

Figure 2.

Kaplan-Meier curve of time to recovery for the complete cohort. Dashed line corresponds to the median time to recovery (12.7 months). Light blue area corresponds to 95% confidence interval (CI) of the probability of recovery. Risk table shows all patients at risk at each time point.

Factors associated with time to recovery

Univariable analysis of the association of time to recovery with each variable of interest did not identify any factor significantly associated with the outcome (Table 2). On multivariable analysis, a negative correlation was identified with UFCxULN [beta coefficient= −0.06, Hazard Ratio (HR) = 0.94, 95% CI: 0.89-0.999], suggesting that for every increase of UFC by 1-fold the ULN, there was an increase of the time to recovery by 6%. Correlations with other variables such as duration of disease (HR= 0.87, 95% CI: 0.75-1.00) could be of interest, but did not reach statistical significance.

Table 2.

Results from the univariable model for the association of various factors with the time to recovery of the HPA axis. Results from the multivariable model for the association of various factors with the time to recovery of the HPA axis.

	Univariate analysis		Multivariate analysis
	HR	95% Confidence Intervals	HR	95% Confidence Intervals
Age at diagnosis	1.03	0.97 - 1.1	0.99	0.87 – 1.12
Gender (Male vs Female)	1.00	0.68 - 1.48	1.08	0.65 – 1.81
Duration of disease	0.98	0.87 – 1.1	0.87	0.75 – 1.00
Puberty Stage vs Tanner 1
Stage 2	1.35	0.76– 2.4	1.44	0.76 – 2.73
Stage 3	1.41	0.71 – 2.81	0.85	0.37 – 1.96
Stage 4	1.25	0.64 – 2.44	1.08	0.42 – 2.76
Stage 5	1.24	0.70 – 2.2	1.55	0.58 – 4.15
BMI at diagnosis Z-score	0.82	0.65 – 1.05	0.85	0.62 – 1.17
Midnight cortisol	1.00	0.99– 1.02	1.02	0.99 – 1.05
Morning ACTH	1.00	0.99 – 1.00	1.00	0.99 – 1.01
UFCxULN	0.97	0.93 – 1.01	0.94	0.89– 0.999
Tumor size	1.03	0.97 – 1.1	1.05	0.96 – 1.15
Cavernous sinus invasion (Yes vs No)	1.15	0.65 – 2.03	1.17	0.57 – 2.36
Surgeon vs. Surgeon 1
Surgeon 2	0.48	0.22– 1.06	1.35	0.25 – 7.3
Surgeon 3	0.29	0.13 – 0.69	0.78	0.15 – 4.2
Surgeon 4	0.78	0.3 – 2.01	2.21	0.39 – 12.66

Recurrence risk and correlation with recovery time

To investigate whether recovery time correlates with the risk for recurrence, we computed the HR for the time to recurrence or persistent remission, based on the duration of postoperative AI for patients who recovered (n=102). An increased risk for recurrence was noted with earlier recovery (p= 0.0342). Specifically, there was 14% decrease of the risk for recurrence (HR: 0.86, 95% CI: 0.75-0.99) with every additional month of duration of AI. On review of the data, all patients who recurred (n= 15), had experienced recovery within 15 months after TSS (Figure 3), whereas none of the patients who recovered after 15 months or had not yet recovered at their latest follow up of more than 15 months postoperatively (n=22) had recurred.

Figure 3.

Kaplan-Meier curve presentation of time to recovery for patients who later recurred (grey) versus those with persistent remission (blue). Risk table shows all patients at risk at each time point. Hazard ratio (HR) corresponds to the cox model of time to recurrence or persistent remission (in months) based on recovery time.

Characteristics of patients with and without recurrence

Overall, 15 patients (11.5%) recurred at a median of 44 months (range: 6.3- 97 months) after TSS. We further investigated whether baseline characteristics of patients who later recurred differed from those of patients without recurrence during their follow up. Patients were similar at various baseline characteristics, including age, gender, anthropometric measurements and biochemical data (Table 1). Patients with recurrence had larger tumors at baseline as measured at preoperative MRI (5.5 mm vs 4 mm, p= 0.039). As one would expect, patients with cavernous sinus invasion were more represented in the cohort of patients with recurrence but this did not reach statistical significance and may be associated with the larger tumor size.

Discussion

The postoperative period after TSS for CD, is complicated by an expected phase of AI, which is of variable duration. Our data suggest that the period of AI is not dependent on several clinical and biochemical markers of CD in children, including the age of the patient at diagnosis, and several markers of hypercortisolemia, except UFC which may predict a small increase of the duration of suppression of HPA axis. However, we identified a significant association of earlier recovery with increased risk for recurrence in the pediatric population, which provides a proof-of-concept for what was previously reported in adults.

The lack of association of postoperative AI with baseline markers of CD is not unexpected. Previous studies in adults have described several potentially interesting factors; however, the results have not been consistent. Specifically, Serban et al described that the duration of AI was positively correlated with the duration of disease, but not with any other baseline factors, including cortisol levels (urine or serum).⁵ Klose et al, showed a potential correlation of age at diagnosis and duration of disease with the duration of AI, but not with other biochemical factors.⁸ In contrast, Hurtado et al described significant association of post-TSS AI with age and BMI at diagnosis, and duration of disease, as well as with post-dexamethasone suppression test cortisol levels;⁶ however, their study included only patients with ACTH-independent CS, which represents a fundamentally different process. The observation that duration of AI depends on the type of CS is further supported by other studies, that showed that patients with ectopic CS or CD recover HPA function earlier than patients with ACTH-independent CS.^7,13 Furthermore, a study from our group in a cohort of pediatric patients (n= 57) failed to identify any significant correlation with clinical and biochemical factors preoperatively. We confirmed these results in a larger cohort of patients (n= 130) and in more extensive set of variables included in the analysis.¹⁰

Recovery of the HPA axis following TSS in CD is thought to depend on the function of the CRH-secreting cells in the paraventricular nuclei of the hypothalamus.^4,14 The fact that we did not identify factors that contribute to this process with a more significant effect, may imply that other factors that were not included in this study play an important role in the recovery process. For example, neuropeptides (such as GABA and opioids) are affected by hypercortisolemia and have been shown to play significant role on CRH-secretion.^15,16 Those peptides need to be measured in the cerebrospinal fluid (CSF) which was not available from our patients. Another potential factor could be genetic background. Genetic variations of the genes implicated in the HPA axis have been extensively studied for correlation with disease susceptibility and treatment response.^17-19 Preliminary results from a pilot study from our group showed variants in potential genes of interest that may be associated with the recovery time, including TBX19, MC5R, CABLES1 and HSD11B1.²⁰ Further studies are underway to confirm functionally the effect of genetic factors in the recovery and overall function of the HPA axis.

The significance of this study lies initially on the counseling and guidance we provide to our patients. Based on this and previous studies, it seems that follow up to document the recovery of the axis is required within the first 1-2 years after TSS since most patients recover within the first 12-18 months. In our institute, we often perform the first postoperative ACTH stimulation test 6 months after TSS, since most patients do not recover adrenal function earlier. Based on the peak cortisol level we continue or wean the hydrocortisone dose (starting from the afternoon dose) until the 12-month ACTH-stimulation test. Any further adjustments of the hydrocortisone dose and the follow-up intervals (ranging from 3-6 months until recovery) are guided by the peak cortisol value. This will allow patients to avoid exposure to unnecessary glucocorticoid therapy. All patients should be counseled on the signs and symptoms of AI, since adrenal crisis may be a life-threatening situation, and can occur during the period of AI and potentially the months following recovery.²¹ All patients should be educated on management of illnesses, carry hydrocortisone injection and a medical alert for the duration of AI and, as we recommend in our institute, for at least six more months after documented recovery. Other than the possibility of adrenal crisis, patients should be aware that AI is associated with reduced quality of life and the transition from hypercortisolemia to hypocortisolemia (although under replacement therapy) may have an impact on their function, school performance and activity level.^22-24

Another significant implication of this study is the report of the prognostic importance of earlier recovery for future risk of recurrence. Previous studies have shown increased risk for recurrence with earlier recovery in adult patients, but this has not been reported previously in the pediatric population.^5,25 In adults with recurrence, recovery of the axis was reported by 22-36 months. We here describe that all children with recurrence had recovery of the HPA axis function by 15 months and a significantly increased risk for recurrence with every month of earlier recovery is noted. Closer surveillance and appropriate counseling of patients who recover early should be performed as appropriately.

Certain limitations exist for the current study. First, this was a retrospective study and thus data collection was not as complete as it could be. To overcome this limitation, we applied multiple imputation analysis to impute the missing data. Patients were followed up at certain intervals after surgery. As per protocol, most of the patients are seen around 6 months and 12 months after TSS, and then as needed every 3-6 months if they have not recovered adrenal function. Thus, we did not have assessment of the HPA axis at shorter time intervals. Furthermore, our data come from a center of referral for other institutes in the US and internationally, and thus, our patient population may not represent the typical cases of pediatric CD. However, this constitutes the largest pediatric cohort to our knowledge of a rare disease. Regarding the evaluation of recovery of HPA axis, we used the standard dose cosyntropin stimulation test. The gold standard test for diagnosis of AI is considered the insulin tolerance test.²² There are several studies, including three metanalyses, comparing the diagnostic accuracy between low (1 mcg) and standard (250 mcg) dose cosyntropin stimulation tests in patients with AI.^26-28 The most recent metanalysis included 30 studies of patients with secondary AI and showed no statistical difference in the diagnostic accuracy of the two tests, although both test showed low sensitivity.²⁶ Given the technical difficulties in performance of a low dose test, in our Institute, we have been performing only standard dose testing for our patients.^29,30 Additionally, all patients have been at least 3 months postoperatively and had been exposed to low ACTH secretion for sufficient time to induce adrenal atrophy.

In summary, we report that in pediatric patients with CD, postoperative duration of AI does not depend on age at diagnosis or several other clinical and biochemical factors (except for UFC). This implies that the process of recovery depends on other factors, not measured in the current study. However, earlier recovery is associated with higher risk for recurrence, which has implications for the patients’ follow up and counseling.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.