Sleep-Related Breathing Disorders in Prepubertal Children with Prader-Willi Syndrome and Effects of Growth Hormone Treatment

Abstract

Context: Recently, several cases of sudden death in GH-treated and non-GH-treated, mainly young Prader-Willi syndrome (PWS), patients were reported. GH treatment in PWS results in a remarkable growth response and an improvement of body composition and muscle strength. Data concerning effects on respiratory parameters, are however, limited.

Objective: The objective of the study was to evaluate effects of GH on respiratory parameters in prepubertal PWS children.

Design: Polysomnography was performed before GH in 53 children and repeated after 6 months of GH treatment in 35 of them.

Patients: Fifty-three prepubertal PWS children (30 boys), with median (interquartile range) age of 5.4 (2.1–7.2) yr and body mass index of +1.0 sd score (−0.1–1.7).

Intervention: Intervention included treatment with GH 1 mg/m²·d.

Results: Apnea hypopnea index (AHI) was 5.1 per hour (2.8–8.7) (normal 0–1 per hour). Of these, 2.8 per hour (1.5–5.4) were central apneas and the rest mainly hypopneas. Duration of apneas was 15.0 sec (13.0–28.0). AHI did not correlate with age and body mass index, but central apneas decreased with age (r = −0.34, P = 0.01). During 6 months of GH treatment, AHI did not significantly change from 4.8 (2.6–7.9) at baseline to 4.0 (2.7–6.2; P = 0.36). One patient died unexpectedly during a mild upper respiratory tract infection, although he had a nearly normal polysomnography.

Conclusions: PWS children have a high AHI, mainly due to central apneas. Six months of GH treatment does not aggravate the sleep-related breathing disorders in young PWS children. Our study also shows that monitoring during upper respiratory tract infection in PWS children should be considered.

PRADER-WILLI SYNDROME (PWS) is characterized by muscular hypotonia, hypogonadism, psychomotor delay, obesity that may become extreme after the age of 2–4 yr, and short stature. Sleeping disorders and respiratory disorders such as hypoventilation, decreased pulmonary function, obstructive and central sleep apnea, and abnormal ventilatory and arousal response during hypercapnia may also occur (1–5). Hypothalamic dysfunction is thought to be responsible for many features of PWS (6).

The underlying cause of PWS is a paternal deletion or a uniparental maternal disomy of 15q11–13. In 1–5%, PWS is the result of an imprinting-center mutation, which causes genes in the paternally inherited chromosome 15q11–13 to be silenced (7, 8).

Several reports demonstrated that GH treatment results in a remarkable growth response but also in an impressive improvement of body composition, with decline in fat percentage and increment in lean body mass, muscle strength, and agility (9–11). Preliminary studies suggested that GH might improve psychosocial development in PWS (12). Data on effects of GH on respiratory parameters in young, prepubertal PWS children are, however, very limited. Haqq et al. (13) found after 6 months of GH a slight reduction in sleep apnea incidence in 12 PWS children, aged 4.5–14.5 yr. Lindgren et al. (14) found improved CO₂ responsiveness in 9 children with PWS after 6–9 months of GH, compared with baseline. Recently several reports have been published on sudden death in children with PWS during GH treatment (15, 16). Unexpected death, however, has also been described in non-GH-treated children with PWS (17, 18). In fact, Whittington et al. (19) reported an overall death rate of 3%/yr for PWS patients in one U.K. health region.

In our study we evaluated the occurrence of sleep-related breathing disorders (SRBDs) in 53 young, prepubertal children with PWS and the effects of 6 months of GH treatment in 35 of them.

Patients and Methods

Patients

In April 2002, a multicenter, randomized, controlled, prospective GH trial in PWS children was started investigating the effects of GH treatment vs. no GH on growth, body composition, activity level, and psychosocial development. Participants fulfilled the following inclusion criteria: 1) genetically confirmed diagnosis of PWS by positive methylation test; 2) age between 6 months and 16 yr; 3) bone age less than 14 yr (girls) or 16 yr (boys); 4) in children over 3 yr, height sd score (SDS) for age below zero; and 5) in children over 3 yr, if height is greater than 0 SDS, weight-for-height SDS must be more than +2 SDS, according to Dutch standards (20, 21). Patients with noncooperative behavior or patients receiving medication to reduce fat were excluded. All patients older than 3 yr started a diet and exercise program 3 months before start of the study. Children were enrolled in the study irrespective of their GH status. Patients received somatropin (Genotropin; Pfizer, New York, NY) in a dose of 1 mg/m²·d. The first 4 wk of treatment, they received only 0.5 mg/m²·d to prevent fluid retention.

In April 2003 we started a polysomnography (PSG) study in addition to the original protocol. For the PSG study, we used the following inclusion criteria: 1) prepubertal at baseline and at repeated PSG; 2) no upper respiratory tract infection (URTI) during PSG; and 3) no previous GH treatment. On November 11, 2005, 83 patients had been included in the original study. Twenty-five were excluded from the PSG study because they received GH treatment before start of the PSG study. For one patient, parents refused PSG, three were pubertal at repeated PSG, and one was excluded because of treatment with nasal continuous positive airway pressure. As a result, 53 patients were eligible for analysis of baseline PSG. Thirty-nine children had a PSG repeated after 6 months of GH treatment. Fourteen patients were followed up in the control group of the original study. Their PSG will be repeated at 6 months after start of GH treatment. Because all patients were stratified for age and body mass index (BMI) before randomization in the original study, these patients were not different from those who had repeated PSG. Of the 39 patients with repeated PSG, four had URTI during the second PSG and were therefore excluded from group analysis.

The study protocol was approved of the Medical Ethical Committee of Erasmus Medical Center, Rotterdam, The Netherlands. Informed consent was obtained from the parents.

Anthropometry

Supine length was recorded below the age of 2.5 yr and thereafter standing height, measured with a Harpenden stadiometer. Weight was assessed on an accurate scale, and BMI (kilograms per square meter) was calculated. Height and BMI were converted into SDS according to Dutch references for age (20, 21). Calculations were performed with Growth Analyzer version 3.0 (www.growthanalyser.org).

Polysomnography

PSG was performed before and after 6.6 (6.1–7.3) months of GH treatment. All PSGs were performed in one specialized sleep center (A.W.d.W., sleep specialist). Children were admitted to the sleep center at 1700 h, accompanied by one parent. Patients underwent complete overnight PSG. Recordings included electroencephalogram, electrooculogram, one channel derivation of electrocardiogram, and surface electromyography of the submental muscle and both anterior tibial muscles. Nasal-oral airflow was monitored by nasal pressure prongs fixed in the nose, respiratory effort by thoracoabdominal strain gauges, and oxygen saturation (SaO₂) by pulse oximetry. All PSG studies were evaluated independently by two persons, both certified in PSG analysis. In case of major discrepancies between both assessments, a third expert opinion was asked. The polygraphic records were scored according to standard criteria of Rechtschaffen and Kales (22). A period of apnea or hypopnea was defined as more than 90% (apnea) or 50% (hypopnea) reduction of airflow for three breaths or longer. For hypopneas, the additional criterion was a reduction of SaO₂ of 4% or more. Periods of apnea and hypopnea were counted over the period of sleep during the night and calculated as mean per hour of sleep [apnea hypopnea index (AHI)]. An AHI above 1 per hour is considered pathological (23). Apneas were considered obstructive when absence of airflow occurred without a decrease in respiratory effort and central when thoracic movements were absent. Abnormal SaO₂ was defined as SaO₂ less than 92% or more than 4% below baseline values during three breaths or longer. Otorhinolaryngological examination consisted of tonsil inspection every 3 months according to the Brodsky staging system (24). Snoring was recorded in a structured interview with parents. When snoring or obstructive sleep apnea (OSA) was diagnosed, fiberoptic endoscopy was performed by an ear-nose-throat surgeon. If adenoid or tonsil hypertrophy was found, adenotonsillectomy was performed.

Data analysis

Statistical analysis was performed by the Statistical Package for Social Sciences (SPSS; version 11; Chicago, IL). Most of the data obtained in our patients were not Gaussian distributed. We therefore expressed our data as median and interquartile range (iqr). Nonparametric tests (Wilcoxon signed ranks test) were used to compare results before and after start of GH. Correlations were calculated using Spearman correlation coefficients. A χ² test was used to evaluate whether OSA is more common in children with BMI over +2 SDS, compared with normal-weight children. Associations among snoring, tonsillar size, and AHI were calculated with Kruskal-Wallis tests. P < 0.05 was considered statistically significant.

Results

Clinical characteristics at baseline

Fifty-three prepubertal PWS children (30 boys) participated in the PSG study. The median (iqr) age was 5.4 yr (2.1–7.2) and the median (iqr) BMI was 1.0 SDS (−0.1–1.7). Sixteen patients had paternal deletion, 21 had maternal disomy, and four had an imprinting center mutation. In 12 patients, diagnosis was confirmed by a positive methylation test for PWS but was not yet further specified (Table 1).

TABLE 1.

Clinical parameters of baseline group

	Median	iqr
No.	53
Gender (males/females)	30/23
Age (yr)	5.4	2.1–7.2
BMI (kg/m2)	17.7	15.9–19.4
BMI (SDS)	1.0	−0.1–1.7
AHI	5.1	2.8–8.7
Central apnea index	2.8	1.5–5.4
Obstructive apnea index	0.0	0.0–0.4
Hypopnea index	0.9	0.0–2.7
Duration longest apnea (sec)	15.0	13.0–28.0
Genetic defect
Paternal deletion (n)	16
Maternal disomy (n)	21
Imprinting center mutation (n)	4
Unknown (n)	12

Data are expressed as median (iqr).

Thirty-nine patients (23 boys) started GH at a dose of 1 mg/m²·d. The first month of GH, they received only 0.5 mg/m²·d, to avoid fluid retention.

Respiratory parameters at baseline

At baseline, the median (iqr) AHI was 5.1 (2.8–8.7) (Fig. 1). Of these, 2.8 per hour (1.5–5.4) were identified as central apneas, 0.0 per hour (0.0–0.3) as obstructive apneas, and 0.9 (0.0–2.7) as hypopneas. The longest median (iqr) duration was 15.0 sec (13.0–28.0). In all children, the AHI exceeded the normal range of 0–1 per hour, indicating that SRBDs do frequently occur, even in normal-weight prepubertal children with PWS. In the total patient group, no correlation was found between BMI SDS and AHI. Forty-five of our 53 patients were not obese. Of these, only 9% had OSA (four of 45), defined as obstructive apnea index over 1 per hour. In contrast, in our eight patients who were obese (i.e. BMI over +2 SDS), 50% had OSA (four of eight) (prevalence of OSA in normal weight vs. obese patients, P = 0.01). We found a negative correlation between both age and BMI and the number of central apneas (r = −0.34, P = 0.01 and r = −0.33, P = 0.017, respectively). There was no significant difference in AHI with regard to gender or genetic defect. Tonsil size as assessed by Brodsky staging system, was not associated with the AHI (data not shown).

Fig. 1.

AHI of 53 PWS children before start of GH. Line represents the upper limit of normal, i.e. AHI = 1 per hour.

Respiratory parameters after 6 months of GH

Thirty-five prepubertal children had PSG repeated after 6 months of GH treatment (Table 2). This group of 35 children had a median (iqr) age of 6.0 yr (2.4–8.6) and median (iqr) BMI of 0.8 SDS (−0.1 to 1.5) before GH. At baseline, median (iqr) AHI in this group was 4.8 per hour (2.6–7.9), of which 2.9 per hour (1.5–5.2) were indicated as central and 0.0 (0.0–0.3) as obstructive. After 6 months of GH (1 mg/m²·d), a nonsignificant decline in the AHI was found to 4.0 (2.7–6.2). This decline was mainly due to a reduction in central apneas to 2.2 per hour (0.8–4.1). In five, adenoidectomy and/or tonsillectomy was performed because adenoidal and/or tonsil hypertrophy developed during the follow-up period. There was no association between changes in AHI and changes in number of awakenings or rapid-eye movement sleep percentage (data not shown).

TABLE 2.

Clinical and respiratory parameters of GH-treated group

	Before start of GH		After 6 months of GH		P value
	Median	iqr	Median	iqr
No.	35		35
Gender	20 boys/15 girls		20 boys/15 girls
Age (yr)	6.0	2.3–8.6	6.8	3.1–9.9
BMI (kg/m2)	17.1	15.9–19.2	17.6	15.7–18.9	0.38
BMI (SDS)	0.8	−0.1 to 1.5	0.8	−0.1 to 1.2	0.19
AHI	4.8	2.6–7.9	4.0	2.7–6.2	0.36
Central apnea index	2.9	1.5–5.2	2.2	0.8–4.1	0.15
Obstructive apnea index	0.0	0.0–0.3	0.0	0.0–0.2	0.73
Hypopnea index	0.7	0.–1.9	1.0	0.7–2.0	0.26
Longest apnea (sec)	15.0	13.0–28.0	17.0	14.0–23.3	0.92

Data are expressed as median (iqr). Respiratory parameters did not change significantly during GH treatment (Wilcoxon signed rank test).

Breathing disorders during illness

Four patients were excluded from analysis because of URTI. The results of their PSGs during health and illness are listed in Table 3. In one of them, PSG was repeated after recovery and adenoidectomy. In this particular patient, the AHI before GH treatment was 7.9 per hour (100% central); during illness after 6 months of GH treatment, the AHI had impressively increased to 38.6 per hour (1.2 central apneas per hour, 12.4 obstructive apneas per hour, and 25.1 hypopneas per hour), whereas after recovery and adenoidectomy, AHI was 3.4 per hour (100% central).

TABLE 3.

Respiratory parameters during health and URTI (n = 4)

	Health		URTI		P value
	Median	iqr	Median	iqr
AHI	5.7	3.1–9.5	36.5	18.1–39.5	0.07
Central apnea index	2.7	1.8–6.6	6.4	1.9–6.6	0.27
Obstructive apnea index	0.0	0.0–2.5	8.9	2.4–21.0	0.07
Hypopnea index	1.1	0.0–3.6	7.2	3.6–20.3	0.14
Longest apnea (sec)	24.5	18.5–33.5	35.5	27.3–49.8	0.27

Data are expressed as median (iqr). Respiratory parameters were as measured by PSG in health and during an episode of URTI. P values are depicted for the statistical analyses (Wilcoxon signed rank test) between measurements during health and URTI.

One patient in our study died unexpectedly. This 3-yr-old boy had GH treatment for 13 months. He responded very well in terms of growth and body composition. In this particular patient, PSG was performed before (AHI 1.7 per hour, 100% central) and after 6 months of GH (AHI 1.4 per hour, 67% central, 33% hypopnea). Six weeks before his death, BMI was 1.6 SDS and tonsils were assessed as Brodsky I-II. He had mild URTI and was clinically evaluated by his pediatrician the day before his death. At that time he had URTI but was in good condition, running around and not generally ill. During the night, he suddenly deteriorated and was found dead in the morning. Autopsy did not reveal the cause of death.

Discussion

We found an increased AHI in 53 young, prepubertal children with genetically confirmed diagnosis of PWS. The high AHI was mainly due to central apneas and hypopneas. In the total group of mainly nonobese PWS children, obstructive apneas were rare. In contrast, obstructive apneas were found in four of the eight overweight patients. After 6 months of GH treatment, a nonsignificant decrease of AHI was found, mainly due to a decrease in central apneas. No significant change in obstructive apneas was found during GH. Illness or adenoid/tonsil hypertrophy, however, did result in a marked increase in SRBDs, and particularly OSA. Our study also shows that a relatively normal PSG does not exclude the possibility of unexpected death during mild URTI.

The increased number of central apneas in our young PWS children suggests a central origin of SRBDs. A hypothalamic origin of SRBDs in PWS was already postulated 20 yr ago (25). A decreased number of oxytocin neurons in the hypothalamic paraventricular nucleus was reported, which might also be involved in reduced neural modulation of breathing (26, 27). Recently Ren et al. (28) proposed that neurally differentiated embryonal carcinoma-cell derived factor (Necdin) deficiency may contribute to the observed respiratory abnormalities in individuals with PWS because Necdin is one of the protein-coding genes that are deficient in PWS (29). Deficiency of Necdin in mice results in neonatal hypoventilation, which is usually fatal (30).

We found a negative association of both age and BMI, with number of central apneas. Because in PWS children age and BMI are highly correlated, we cannot distinguish whether this is an effect of age or BMI. From a pathophysiological point of view, we consider it more likely to be an effect of age. In fact, our data are in line with a previous report, indicating that central apneas are more common in younger, healthy children, although within the normal range (31). The mechanism is unclear and might be related to a relatively more immature respiratory control in younger children. However, we cannot exclude that underweight in young PWS infants might contribute to as well.

OSA was uncommon in normal-weight PWS patients. However, four of the eight overweight (defined as BMI over +2 SDS) patients (50%) had signs of OSA. Increased BMI has been associated with decreased SaO₂ and higher AHI in older PWS children and adults (32). Harris et al. reported an improvement of OSA and hypoventilation after weight loss in children and adults with PWS (33). Tonsil hypertrophy may also play a role in OSA. Children with PWS might have a smaller naso- and oropharynx, which could contribute to obstruction (3). Recently an improvement in AHI and SaO₂ was reported after adenotonsillectomy in five PWS children with OSA (34).

After 6 months of GH treatment, a nonsignificant decline in AHI was found, compared with baseline, mainly due to a lower number of central apneas. Thus, our study indicates that GH had no adverse effects on the respiration of PWS children. Several publications reported sudden death in infants and children with PWS during GH treatment (15, 16, 35). Several suggested a causal relationship between GH and sudden death in PWS.

Until now only limited data were available on the effects of GH on PSG. Miller et al. (36) recently reported an improvement of AHI after 6 wk of GH in most of her PWS patients. They performed PSG in children and adults of which 12 were children under the age of 12 yr. A subset of patients, however, had an increased AHI after 6 wk of GH. Most of these patients had URTI during the second evaluation (36). Haqq et al. (13) reported in a crossover study a decrease in AHI after 6 months of GH in 12 PWS children, aged 4.5–14.5 yr, although not statistically significant. Myers et al. (11) demonstrated that inspiratory and expiratory muscle strength improved in 20 children with PWS, aged 4–16 yr after 12 months of GH, compared with 10 controls. Lindgren et al. (14) found improved CO₂ responsiveness in 9 children with PWS after 6–9 months of GH, compared with baseline. A number of hormones, including GH and IGF-I, are involved in the physiological regulation of breathing (37). IGF-I receptors are located around the central chemoreceptors in the brain stem and also in the cerebellum in which the inputs from chemoreceptors are integrated (38). GH may therefore theoretically improve breathing via a direct mechanism.

In our study we found only a small number of obstructive apneas both before and during GH treatment. There was no increase in obstructive apneas during GH treatment. Five children had adenotonsillectomy before the second PSG was performed because of adenoid and/or tonsillar hypertrophy. Unfortunately, this might confound our results, but for obvious safety reasons, we could not avoid this. The AHI of these patients during both PSGs was not different, compared with the rest of the study group. We found no significant association between tonsil size or snoring and the AHI. Sleep apnea, both obstructive and central, occurs more frequently in adults with GH excess (acromegaly) (39) and is associated with thickening of the pharyngeal wall in the acromegalic patients (40). We cannot rule out that GH might have resulted in some adenoidhypertrophy because we performed fiberoptic endoscopy only when indicated by snoring or OSA during PSG. It has been suggested that GH treatment might increase tonsil size; however, to our knowledge, no controlled, prospective study has been performed.

One of our patients died unexpectedly during an episode of URTI. One of the most alarming findings is that this patient had near-normal sleep-related breathing during PSG, both before and during GH treatment. This points out that a near-normal PSG in a healthy PWS child does not guarantee he/she will not die during mild URTI. It might be related to a rise of apneas (both central and obstructive) during illness as shown in four of our patients who had a PSG during an episode of mild URTI. Unexpected deaths have been described in PWS children both without and during GH and have been attributed to several possible causes, such as respiratory dysfunction, cardiomyopathy, temperature instability, and adrenal insufficiency or combinations of these.

We recommend monitoring of SRBDs by PSG and regular ear-nose-throat evaluation in all PWS children, both before and during GH treatment. If adenoidhypertrophy or tonsillar hypertrophy occurs, adenotonsillectomy should be considered. It is important to mention that a relatively normal PSG does not exclude the possibility of unexpected death during mild URTI. Based on our results cardiorespiratory monitoring during URTI in children with PWS before and during GH treatment should be considered. Future studies are required for evaluating SRBDs in PWS during URTI to give recommendations with regard to monitoring during URTI.

In conclusion, our study shows that many prepubertal children with PWS have SRBDs, mainly due to central apneas. BMI or age cannot explain the variability in the severity of the SRBD, although OSA was more prevalent in children with obesity than in normal-weight children. After 6 months of GH, a nonsignificant decrease in AHI was found. Thus, our data are reassuring with respect to the effects of GH on SRBDs. Our study also shows that a normal PSG does not exclude the possibility of unexpected death during mild URTIs. During URTI, AHI may rise and obstructive apneas may occur.

Acknowledgments

The authors thank all the participating parents and children for their enthusiastic cooperation. The assistance of Mrs. M. van Eekelen is gratefully appreciated.

This work was supported by a grant from Pfizer.

Disclosure statement: The multicenter study in children with PWS was an investigator-responsible study. Only the investigators decided on study design, data collection, data analysis, and data interpretation; writing of the report; or the decision to submit the paper for publication. D.A.M.F., A.W.d.W., R.A.S.v.d.B., K.J., and H.H. have nothing to declare. A.C.S.H.-K. has delivered lectures and received reimbursement of travel/accommodation expenses at meetings sponsored by Pfizer.

Abbreviations:

• AHI,

  Apnea hypopnea index;

• BMI,

  body mass index;

• iqr,

  interquartile range;

• Necdin,

  neurally differentiated embryonal carcinoma-cell derived factor;

• OSA,

  obstructive sleep apnea;

• PSG,

  polysomnography;

• PWS,

  Prader-Willi syndrome;

• SaO₂,

  oxygen saturation

• SDS,

  sd score;

• SRBD,

  sleep-related breathing disorder;

• URTI,

  upper respiratory tract infection.