Effects of Growth Hormone on Pulmonary Function, Sleep Quality, Behavior, Cognition, Growth Velocity, Body Composition, and Resting Energy Expenditure in Prader-Willi Syndrome

The objective of this study was to investigate the effects of GH administration on pulmonary function, sleep, behavior, cognition, linear growth velocity, body composition, and resting energy expenditure (REE) in children with Prader-Willi syndrome. The study used a 12-month, balanced, randomized, double-blind, placebo-controlled, cross-over experimental design. Twelve subjects were randomized to GH (0.043 mg/kg·d) or placebo intervention for 6 months and then crossed over to the alternate intervention for 6 months. Differences in outcome variables were determined by paired t tests. Peak flow rate, percentage vital capacity, and forced expiratory flow rate improved and number of hypopnea and apnea events and duration of apnea events trended toward improvement after GH intervention. The only difference in cognition or behavior was an increase in hyperactivity scale on the Behavior Assessment System for Children after GH intervention. Linear growth velocity, REE, and lean mass were higher (67%, 19%, and 7.6%, respectively), and fat mass and percentage body fat were lower (10.3% and 8.1%, respectively) after GH intervention. GH administration did not change mean fasting ghrelin concentration. GH intervention improved body composition and REE and may contribute to better sleep quality and pulmonary function. GH administration did not impact fasting ghrelin concentration.

PRADER WILLI SYNDROME (PWS) is a genetic disorder associated with morbid obesity that occurs with a frequency of 1 in 15,000 live births (1). Hypothalamic dysfunction is thought to be the basis for many of the distinguishing features of PWS including deficient GH secretion, short stature, insatiable hunger, morbid obesity, hypogonadism, aberrant body temperature control, and sleep disturbances (2). Individuals with PWS have short stature, higher fat mass, and lower lean mass and bone mineral content (BMC) than age-matched controls, similar to individuals deficient in GH (3). The associated obesity alone does not account for the low GH secretion (4–8).

Individuals with PWS also have restrictive ventilatory impairment primarily because of respiratory muscle weakness (9) and abnormal ventilatory responses to hypoxia or hypercapnia (10, 11). In addition, children with PWS have abnormalities of sleep and arousal including excessive daytime sleepiness, sleep-disordered breathing with increased number of apnea events, decreased nadir of oxygen saturation, increased maximum heart rate, and abnormal arousal and cardiorespiratory responses to hypoxia and hypercapnia (12–16). Although diminished GH secretion has been demonstrated in other individuals with primary sleep abnormalities (17, 18), few studies to date have looked at the effect of GH administration on sleep quality or pulmonary function in children with PWS (10, 19).

Whitman and Accardo (20) administered the Survey Diagnostic Instrument to parents of 35 adolescents with PWS to assess behavioral problems. The following behavioral problems were identified: neurosis, somatization, conduct disorder, violent/antisocial disorder, and hyperactivity (20). In children with short stature, GH treatment may ameliorate some preexisting behavioral problems. Stabler et al. (21) tested 195 children [72 children with GH deficiency (GHD) and 59 children with idiopathic short stature] for intelligence, academic achievement, social competence, and behavior problems before and yearly after GH treatment for 3 yr. Intelligence quotient (IQ) and achievement scores did not change with GH therapy. However, after GH treatment, Child Behavior Checklist (CBCL) scores for total behavior problems were improved in patients with GHD (P < 0.001) and idiopathic short stature (P < 0.003) (21). Several studies demonstrated that GH administration affects behavior in children with PWS. Lindgren et al. (22) showed that GH treatment improved behavior in subjects with PWS, and Akefeldt and Gillberg (23) reported no improvement in behavior. Lindgren et al. (22) also noted that parents reported increased activity during GH treatment, but this finding was not proven in an open study. Most recently Whitman et al. (24) reported a reduction in depressive symptoms and no behavioral deterioration on the modified Oxford Survey Diagnostic Instrument with use of GH therapy in PWS. Because of these contradictory findings, we wanted to assess the effect of GH therapy on behavior problems in children with PWS.

Adults with PWS have elevated fasting plasma concentrations of ghrelin (25), a newly described gastroenteric hormone, which is high in concentration before meals and suppressed by food intake (26). Ghrelin was originally described as a potent GH secretagogue receptor agonist and was subsequently found in pharmacologic doses to stimulate hunger and food intake in rodents and humans (27, 28). It has been postulated that high fasting ghrelin concentrations found in individuals with PWS may contribute to the increased food intake and morbid obesity associated with this disorder. This elevation in ghrelin concentration, however, may simply reflect a compensatory increase in ghrelin secretion secondary to feedback from decreased GH levels. If this feedback mechanism exists, then GH treatment should decrease ghrelin concentrations. No publications to date have reported the effect of GH administration on serum ghrelin concentration in PWS.

Therefore, the primary aims of this study were to determine whether GH administration in children with PWS improves pulmonary function, behavior, cognition, and sleep quality and lowers fasting serum ghrelin concentration. The secondary aims of the study were to confirm the effects of GH administration on growth velocity, biomarkers of weight regulation and metabolism, body composition, and resting energy expenditure. We hypothesized that with GH intervention, resting energy expenditure should improve concurrent with an increase in lean mass; pulmonary function tests may improve because of improved respiratory muscle strength; sleep would improve because of improved respiratory muscle strength and GH effects on central control of sleep; IQ would not change, but behavior would improve because of enhanced alertness and energy to tolerate life stressors.

Subjects and Methods

Subjects

Fourteen children (seven males and seven females) with PWS, naïve to GH treatment, were enrolled after informed consent was obtained from their parents or guardians. Ten subjects (71%) had a deletion of paternal chromosome 15q11–13. Four subjects (29%) had maternal disomy also commonly described in PWS. At baseline, participants were 4.5–14.5 yr of age (Table 1). All participants had normal free T₄ (FT4) and TSH concentrations with either endogenous (n = 13) or T₄ replacement therapy (n = 1). Individuals who had other chronic illnesses or were taking medications that impact long-term bone mineralization or body composition were excluded from the study. Two subjects withdrew from the study before completion, one because of relocation, and the other because of refusal of daily injections. The study was reviewed and approved by the Oregon Health and Science University (OHSU) Institutional Review Board and the General Clinical Research Center Scientific Advisory Committee.

Table 1.

Characteristics of study subjects at study entry

Subject no.	Age (yr)	Sex	BMI-SDS	Fat mass (kg)	Lean mass (kg)
1	4.5	F	3.8	14.6	10.0
2	7.0	F	2.7	26.2	19.7
3	7.3	F	3.0	33.4	20.2
4	8.5	F	2.0	16.9	15.1
5	9.7	F	2.3	29.8	20.4
6	13.9	F	2.9	66.8	51.8
7	5.9	M	2.3	11.5	11.2
8	6.0	M	2.8	12.6	12.5
9	8.8	M	2.8	42.3	25.9
10	12.2	M	1.7	32.0	24.7
11	12.6	M	1.2	20.8	24.8
12	14.6	M	2.8	58.9	35.8

Experimental design

The study used a 12-month, balanced, randomized, double-blind, placebo-controlled, cross-over experimental design. Subjects were randomized to receive GH (n = 6) (Genotropin, 0.043 mg/kg·d plus inactive ingredients; Pharmacia Corp., Peapack, NJ) or placebo (n = 6) (inactive ingredients; Pharmacia Corp.) for 6 months. GH and placebo were given once a day by sc injection using a Genotropin pen. Subjects crossed over to the alternate intervention for an additional 6 months. GH treatment was evenly distributed over the year. Subjects were admitted to the OHSU General Clinical Research Center at 0, 6, and 12 months for 36 h to complete the outcome measurements. Participants were seen at 3 and 9 months for anthropometric measurements and to monitor side effects and compliance with the intervention by documenting empty pen cartridges used to deliver the GH or placebo solutions.

Bone age and linear growth velocity

Bone age was determined at 0 and 12 months; x-ray films of the wrist were interpreted by a single pediatric endocrinologist according to the method of Greulich and Pyle (29). Height was measured at screening and at 0, 6, and 12 months using a wall-mounted stadiometer (Holtain Ltd., Crymych, UK). Linear growth velocity was calculated based on height measurements collected at screening, 0, 6, and 12 months as the difference in height in centimeters divided by the duration in days between measurements and expressed as centimeters per year.

Pulmonary function

Vital capacity (VC), percentage VC expelled in 1 sec, forced expiratory volume expired in 1 sec (FEV₁), forced expiratory flow rate (FEF _25–75) and peak flow rate (PFR) were assessed by spirometry (Koko PFT System, Pulmonary Data Service Instrumentation Inc., Louisville, CO) at 0, 6, and 12 months.

Sleep studies

Episodes of apnea and hypoxia and oxygen saturation were measured by polysomnography using channels for heart rate, thoracic impedance, nasal thermistor, pulse oximetry, and microphone for snoring annotation using an EdenTrace II+ digital recorder (Nellcor-Puritan Bennett, Eden Prairie, MN) and analysis software (Sandman, Melleville, Ottawa, Canada). Measurements were obtained for up to 10 h throughout each night while the subject slept at 0, 6, and 12 months.

Behavior and cognition

Behavioral and cognitive assessments were performed at 0, 6, and 12 months by a single, blinded, medical psychologist using the Behavior Assessment System for Children (BASC), CBCL, Vineland, and Wechsler Intelligence Scale for Children III (30–33). The same parent responder was interviewed at each time point, with the exception of two occasions, to maintain consistency in responses.

Blood sample analysis

Stimulated GH concentration was measured after oral L-dopa (10 mg/kg) administration at the screening visit only. Fasting serum concentrations of IGF-I [Diagnostic Systems Laboratories (DSL), Webster, TX], IGF-binding protein (IGFBP)-3 (DSL), immunoreactive ghrelin (Phoenix Pharmaceuticals, Belmont, CA), FT4 (chemiluminescent assay), TSH (Quest Diagnostics, Inc. Nichols Institute, San Juan Capistrano, CA), glucose (glucose oxidase method), insulin (DSL), leptin (DSL), osteocalcin (Diagnostic Products, Los Angeles, CA), N-telopeptide (Ostex International, Inc., Seattle, WA), triglycerides (glyphosphate oxidase method), and cholesterol (timed end point method) were measured at 0, 6, and 12 months. The age-dependent sd score (SDS) for IGF-I and IGFBP-3 were calculated for each child and then averaged. Normative data were provided by DSL, based on multicenter studies performed with the DSL-6600 ACTIVE IGFBP-3 immunoradiometric assay (IRMA) kit and three separate pediatric studies using the DSL-5600 ACTIVE IGF-I IRMA kit (expected values, DSL-6600 ACTIVE IGFBP-3 IRMA kit and DSL-5600 ACTIVE IGF-I IRMA kit, Web site: http://www.dslabs.com).

Body composition

Weight was measured with a digital scale (Scale-Tronix, Inc., Wheaton, IL). Triceps and subscapular skin-fold thicknesses and midupper arm, waist, and hip circumference measurements were obtained using Lange skinfold calipers (Beta Technology Inc., Santa Cruz, CA) and a flexible, nonstretching, fiberglass measuring tape, respectively. Skinfold measurements were taken by a single, trained investigator on the right side of the body and circumference measurements were taken at the midpoint of the upper right arm, minimum waist diameter, umbilicus diameter, and maximum hip diameter while the subject stood relaxed after exhaling. Measurements were taken at 0, 3, 6, 9, and 12 months of intervention. Body composition was measured by dual-energy x-ray absorptiometry (QDR 4500W/CE; Hologic, Inc., Bedford, MA) at 0, 6, and 12 months. Dual-energy x-ray absorptiometry scan measurements were limited by a weight requirement of less than 300 pounds in one individual. Total lean mass (LM), fat mass (FM), percentage body fat, total and lumbar spine BMC, and bone mineral density (BMD) were calculated. Weight, height, and body mass index (BMI) SDS were obtained using Epi Info 2000 (www.cdc.gov/epiinfo/).

Resting energy expenditure

Resting energy expenditure (REE) was determined by open-circuit indirect calorimetry using a ventilated hood collection system (Vmax 29n; SensorMedics Corp., Yorba Linda, CA) at 0, 6, and 12 months. Measurements were conducted between 0600 and 0800 h after an 8-h overnight fast while the subject was awake but inactive and comfortably resting on a hospital bed. Each measurement lasted approximately 55 min or until the subject terminated the procedure. Estimates of predicted REE were derived from the Harris-Benedict, Schofield, and Cunningham equations (34–36). A variety of regression equations, derived from measured REE in healthy individuals, have been developed to predict REE. The Harris and Benedict equations, published in 1919, are among the most commonly used equations and are based on measured REE of healthy young adults (20- to 70-yr-olds). In healthy adults, these equations predict REE to within 14% of measured values (34). The Schofield prediction equations reported in 1985 were derived from measured REE in a much larger sample that included children. Inclusion of weight and height into the Schofield equations improved prediction of REE within each age and gender category, especially among children less than 3 yr of age. Although the Schofield equations are accepted as a means to estimate REE in healthy children, there is less confidence in their use among children with various chronic diseases or significantly aberrant body weight (35). The Cunningham equation is derived from the same data as the Harris and Benedict equations and takes into account only lean body mass in its calculations (36). The percentage of predicted REE was calculated as measured REE divided by predicted REE and multiplied by 100.

Statistical methods

Data for the 12 subjects completing the protocol were summarized at baseline, after 6 months of GH, and after 6 months of placebo using means ± sd and percentages. Differences in means of outcome variables after 6 months of GH (n = 12) and placebo (n = 12) interventions were assessed by paired t tests using SigmaStat software (SPSS, Inc., Chicago, IL). When data followed a nonnormative distribution, Wilcoxin sign-rank tests were used to assess significant differences. Differences were considered significant when P was less than 0.05. Effect of treatment order was assessed using two-way repeated measures ANOVA (one-factor repetition). There was no statistically significant impact of treatment order on the outcome variables assessed.

Results

Subject characteristics

Twelve subjects (six males and six females) completed the protocol. At baseline, mean chronological and bone ages were 9.7 ± 3.3 yr and 10.0 ± 4.2 yr, respectively. Mean BMI SDS was 2.5 ± 0.7 (Table 1). Mean IGF-1 SDS and IGFBP-3 SDS were −1.10 ± 1.15 and −1.67 ± 1.10, respectively, representing GHD.

Pulmonary function

At baseline, subjects demonstrated modest impairment of pulmonary function as demonstrated by: PFR of 62% ± 13.6%, FEV₁ of 74% ± 9.0%, VC of 74% ± 9.1%, percentage VC exp in 1 sec of 79% ± 34%, and FEF_25–75 of 80% ± 24%, compared with standards established for age- and sex-matched children (37). After GH intervention, PFR, percentage VC, and FEF_25–75 improved significantly, compared with placebo (Table 2). No difference was seen in FEV₁ or VC after GH or placebo interventions.

Table 2.

Measures of pulmonary function after 6 months of placebo and GH interventions

Pulmonary function	Placebo intervention (n = 12)	GH intervention (n = 12)
PFR (liter/min)	2.5 ± 1.1	3.3 ± 1.3a
FEV1 (liter)	1.41 ± 0.52	1.49 ± 0.59
VC (liter)	1.74 ± 0.79	1.68 ± 0.73
VC (%)	84 ± 10	91 ± 10a
FEF25–75	1.32 ± 0.46	1.76 ± 0.74a

Results are expressed as mean ± sd.

^a

P < 0.05, difference in means after placebo and GH intervention.

Sleep quality

At baseline, subjects had more hypopnea and apnea events on polysomnography than normal children (38). After GH intervention the average number of apnea events (50.6 ± 69.0 vs. 27.3 ± 20.4, P = 0.26), duration of each apnea event (19.3 ± 4.1 sec vs.16.0 ± 5.7 sec, P = 0.15), and number of hypopnea episodes (146 ± 55 vs.114 ± 65, P = 0.18) trended lower than after placebo intervention but were not statistically significant. These values still exceeded the predicted number for normal children, even during the GH intervention period. Virtually 100% of study participants exhibited clinically significant frequency of apnea and hypopnea events. This finding was independent of BMI so that even children with PWS of normal body weight demonstrated sleep-disordered breathing.

Behavior

Mean T scores on the CBCL and BASC at baseline and after 6 months of placebo or GH interventions are presented in Tables 3³ and 4⁴. The sensitivity of the CBCL was better at discriminating problem areas for the children with PWS than the BASC. For each subscale of the BASC and CBCL, no significant differences were found after GH and placebo interventions, with the exception of a significant increase in the hyperactivity scale on the BASC with GH intervention, compared with placebo. There were also no differences in behavioral items specific to PWS (skin picking, obsession with food, overeating, overtiredness) after placebo and GH interventions.

Table 3.

Behavior items for CBCL

CBCL scales	Baseline mean T-scorea	Placebo intervention T-scorea	Treatment intervention T-scorea
Total score	63.6 ± 11.0b	61.3 ± 9.6	61.0 ± 11.7
Internalizing	59.2 ± 11.6	55.3 ± 7.8	53.8 ± 9.0
Externalizing	58.2 ± 9.6	51.2 ± 16.7	55.3 ± 12.9
Withdrawn	60.9 ± 9.7	57.7 ± 9.8	59.5 ± 10.0b
Somatic complaints	61.3 ± 8.5	58.1 ± 7.5	57.1 ± 4.1b
Anxious/depressed	56.8 ± 9.3	53.7 ± 3.9	53.0 ± 5.2b
Social problems	69.8 ± 7.5c	67.7 ± 5.8c	67.8 ± 9.9bc
Thought problems	66.8 ± 9.6	67.0 ± 8.8c	67.0 ± 10.0c
Attention problems	60.9 ± 7.3	57.9 ± 7.4	60.2 ± 8.1
Delinquent behavior	59.3 ± 9.8	58.3 ± 9.3	58.3 ± 10.4
Aggressive behavior	58.1 ± 8.0	55.8 ± 8.7	57.7 ± 9.3

Results are expressed as mean ± sd.

^a

T score ≥70 designates clinically significant behavior (behaviors at this level deviate from the norm such that a child needs further evaluation by a professional child psychologist or psychiatrist).

^b

Analysis performed using Wilcoxon sign-rank test due to non-normative distribution of data.

^c

T score of 67–70 designates borderline significant behavior.

Table 4.

Behavior items for BASC

BASC scales	Baseline mean T-scorea	Placebo intervention T-scorea	Treatment intervention T-scorea
Composite externalizing	49.8 ± 9.8b	47.7 ± 8.8	52.6 ± 12.1
Composite internalizing	50.9 ± 8.4	49.3 ± 8.6	51.8 ± 10.4
Composite behavioral symptom index	53.0 ± 11.0	52.8 ± 15.9	56.1 ± 11.2
Composite adaptive	43.8 ± 8.0	44.1 ± 12.9	41.5 ± 10.3
Hyperactivity	49.4 ± 11.0	47.4 ± 7.8	53.1 ± 10.1c
Aggression	49.5 ± 7.2	45.6 ± 6.3	48.7 ± 9.4
Anxiety	49.8 ± 8.1	48.3 ± 7.0	50.3 ± 6.5
Depression	52.8 ± 11.3	51.9 ± 11.0	52.5 ± 10.6
Somatization	49.7 ± 6.3	48.1 ± 10.3	51.1 ± 16.9
Atypicality	60.5 ± 17.6d	49.1 ± 10.0	54.4 ± 18.2b
Withdrawal	45.0 ± 8.2	45.3 ± 6.6	45.5 ± 6.0
Attention problems	50.9 ± 6.0	50.3 ± 7.4	55.8 ± 9.7
Adaptability	43.8 ± 8.0	44.1 ± 12.9	41.5 ± 10.3
Social skills	50.1 ± 6.5	50.8 ± 12.5	47.5 ± 10.9

Results are expressed as mean ± sd.

^a

T score ≥70 designates clinically significant behavior scale.

^b

Analysis performed using Wilcoxon sign-rank test due to non-normative distribution of data.

^c

P < 0.05, comparing difference in means after placebo and GH intervention.

^d

T score = 60–69 designates at-risk behavior scale.

Cognition

At baseline, all subjects demonstrated moderate mental retardation (mean full-scale IQ = 52 ± 4; mean verbal IQ = 56 ± 9; mean performance IQ = 55 ± 7). No significant differences in mean full-scale, verbal, or performance Wechsler Intelligence Scale for Children III scores were observed after GH or placebo interventions. At baseline, the mean standard score on the Daily Living Skills domain of the Vineland was 48 ± 10. No significant differences were seen after GH and placebo interventions in standard scores on the Daily Living Skills domain of the Vineland.

GH axis

L-Dopa stimulated peak GH concentrations were less than 10 ng/ml in all 12 subjects. At baseline, mean IGF-I and IGFBP-3 concentrations (SDS) were 169.3 ± 155.7 ng/ml (SDS −1.10 ± 1.15) and 2169 ± 1010 ng/ml (SDS −1.67 ± 1.10), respectively. Values obtained from each study participant suggested frank GH deficiency. After GH administration, mean IGF-I and IGFBP-3 concentrations (SDS) increased to 719.5 ± 378.6 ng/ml (SDS 3.9 ± 2.5) and 6029 ± 1311 ng/ml (SDS 1.1 ± 1.8), respectively (Table 5).

Table 5.

Fasting blood concentrations of biomarkers of weight regulation and metabolism after 6 months of placebo and GH interventions

Parameter	Placebo intervention (n = 12)	GH intervention (n = 12)
IGF-I (ng/ml)	232 ± 182	720 ± 379ab
IGFBP-3 (ng/ml)	4247 ± 1209	6029 ± 1311c
Leptin (ng/ml)	54.3 ± 46.2	49.7 ± 39.3
Ghrelin (pmol/liter)	361 ± 309	272 ± 204
FT4 (pmol/liter)	14.8 ± 1.4	12.9 ± 1.5d
TSH (mU/liter)	2.04 ± 1.13	1.81 ± 0.79
Insulin (pmol/liter)	64.2 ± 39	64.2 ± 42.6
Glucose (mmol/liter)	4.8 ± 0.5	5.0 ± 0.7
Osteocalcin (nmol/liter)	7.8 ± 5.9	10.5 ± 5.7a
Triglycerides (mmol/liter)	0.92 ± 0.42	0.80 ± 0.52
Total cholesterol (mmol/liter)	4.5 ± 1.7	4.7 ± 0.9a

Results are expressed as mean ± sd.

^a

Analysis was performed using Wilcoxon sign-rank test due to non-normative distribution of data.

^b

P < 0.001, difference in means after placebo and GH intervention.

^c

P < 0.01, difference in means after placebo and GH intervention.

^d

P < 0.05, difference in means after placebo and GH intervention.

Bone age, height, and linear growth velocity

Mean bone age at baseline was 10.2 ± 4.1 yr, compared with a mean chronological age of 9.7 ± 3.3 yr. At the end of the 12-month study, mean bone age was 11.3 ± 3.7 yr, compared with mean chronological age of 10.7 ± 3.3 yr. Mean height and height SDS at baseline were 128.9 ± 19.7 cm and −1.3 ± 1.2, respectively. Mean height and height SDS at the end of the study were 134.6 ± 19.3 cm and −1.2 ± 1.2, respectively. Mean growth velocity between the screening and baseline visits was 4.2 ± 2.3 cm/yr. Growth velocity was significantly higher after 6 months of GH, compared with placebo intervention (7.5 ± 3.5 vs. 4.5 ± 2.7 cm/yr; P = 0.03) (Table 6).

Table 6.

Anthropometric, growth velocity, and body composition measurements after 6 months of placebo and GH intervention

	Baseline (n = 12)	Placebo intervention (n = 12)	GH intervention (n = 12)
BMI (kg/m2)	30.8 ± 8.3	32.8 ± 9.7	31.2 ± 8.9a
BMI (SDS)	2.5 ± 0.7	2.5 ± 0.6	2.4 ± 0.5
Height (SDS)	−1.3 ± 1.2	−1.3 ± 1.3	−1.2 ± 1.1
Growth velocity (cm/yr)	4.2 ± 2.3	4.5 ± 2.7	7.5 ± 3.5a
Body fat (%)	54 ± 5.3	54.1 ± 5.6	49.7 ± 5.8a
Fat mass (kg)	29.6 ± 16.7	29.1 ± 14.1	26.1 ± 12.8a
Lean mass (LM) (kg)	22.5 ± 10.9	22.4 ± 8.5	24.1 ± 8.8a
Lumbar spine BMD (SDS)	−0.51 ± 0.30	−0.40 ± 1.4	−0.33 ± 1.4
Total BMC (g)	1263 ± 451	1342 ± 453	1337 ± 453
REE (kcal/d)	1288 ± 290	1285 ± 404	1533 ± 455a
REE (kcal/cm)	10.0 ± 1.3	9.6 ± 1.9	11.3 ± 2.2a
REE (kcal/kg LM)	67.1 ± 15.1	57.1 ± 13.1	64.5 ± 15.5b

Results are expressed as mean ± sd.

^a

P < 0.05, difference in means after placebo and GH intervention.

^b

P = 0.07, difference in means after placebo and GH intervention.

Biomarkers of weight regulation and metabolism

Mean fasting serum concentrations of leptin and ghrelin trended lower after 6 months of GH intervention than placebo, although the differences were not significant (P = 0.06 and P = 0.11, respectively) (Table 5). FT4 concentrations were significantly lower yet remained within the normal range after GH intervention, compared with placebo, but TSH concentration remained unchanged. Only one subject required thyroid hormone replacement while receiving GH treatment. No differences were observed in mean fasting concentrations of glucose, insulin, triglyceride, or cholesterol after GH and placebo interventions (Table 5).

Anthropometric and body composition measurements

Mean BMI was significantly lower (P = 0.04) after GH than placebo intervention. (Table 6). Likewise, triceps and subscapular skinfold and midarm circumference measurements were significantly lower after GH than placebo intervention (data not shown).

At baseline, mean LM was 22.5 ± 10.9 kg and mean FM was 29.6 ± 16.7 kg resulting in a mean percentage body fat of 54.0% ± 5.3%. After 6 months of GH intervention, mean FM (26.1 ± 12.8 kg) and mean percentage body fat (49.7% ± 5.8%) were significantly lower and mean LM (24.1 ± 8.8 kg) was significantly higher than after placebo intervention (Table 6). Lumbar spine BMD SDS and total body BMC SDS were not different after 6 months of GH or placebo intervention (Table 6). However, osteocalcin concentration, a measure of bone formation, trended higher after GH than placebo intervention, although it did not reach statistical significance (10.5 ± 5.7 vs. 7.8 ± 5.9 nmol/liter; P = 0.06) (Table 5).

REE

At baseline, mean REE was 1288 ± 290 kcal/d, which is 96% of mean REE predicted by the Harris-Benedict equation (1353 ± 372 kcal/d). Measured REE was 94.5% and 133% of REE predicted by the Schofield equation (1434 ± 387 kcal/d) and Cunningham equation (1008 ± 252 kcal/d), respectively. Therefore, before GH intervention, the Harris-Benedict equation most closely approximated measured REE in the children with PWS. Also, as expected, measured REE was strongly correlated with LM (r = 0.86). After GH intervention, mean REE was 1533 ± 455 kcal/d and was significantly higher than mean REE after placebo intervention (1285 ± 404 kcal/d; P = 0.02). This difference was maintained when REE was expressed as a function of height (11.3 ± 2.2 vs. 9.6 ± 1.9 kcal/cm; P = 0.02) and trended toward significance when REE was indexed to LM (64.5 ± 15.5 vs. 57.1 ± 13.1 kcal/kg LM; P = 0.07).

Adverse effects

Children treated with GH are at increased risk of scoliosis and glucose intolerance (39). Although four participants did have a mild degree of scoliosis (<15 degrees) at baseline, no subject developed a significant degree of scoliosis (>20 degrees) during treatment with GH. We found no evidence that GH intervention contributed to impaired fasting glucose concentrations. Fasting glucose measurements were less than 6.7 mmol/liter at all study time points. GH intervention (0.043 mg/kg·d) resulted in supranormal IGF-I and normal IGFBP-3 concentrations. The consequence of this elevation in IGF-I is not known at this time.

Discussion

Low GH secretion in children with PWS is well documented (4–8). Unlike other obese individuals who may have low levels of measured GH and high levels of IGF-I and IGFBP-3, children with PWS also demonstrate low levels of IGF-I and IGFBP-3, reflecting a true GH-deficient state (40). In addition, abnormal body composition, including increased FM and reduced LM, resembles individuals with GHD (3). Favorable changes in body composition with GH intervention, including decreased FM and increased LM, are also well described (5, 41–45).

PWS is also characterized by pulmonary abnormalities, poor sleep quality, maladaptive behaviors, and decreased cognition. Pulmonary abnormalities in PWS consist of a restrictive ventilatory impairment primarily as a result of respiratory muscle weakness as well as abnormal responses to both hypoxia and hypercapnia (9–11). Poor sleep quality manifests as increased number of apnea events per hour of sleep, decreased nadir of oxygen saturation, increased maximum of instantaneous heart rate, and decreased respiratory responses to hypercapnia during quiet sleep (12–16). In elderly individuals, low GH secretion has been shown to cause poor sleep quality (46). Maladaptive behaviors include dysphoria, anxiety, somatization, conduct disorder, and hyperactivity; cognitive delay is variable (mild to severe).

Previous studies evaluated the effectiveness of GH therapy in children with PWS to reduce FM and increase LM. Our results confirm that GH decreases FM and percentage body fat by 10.3% and 8.1%, respectively, and increases LM by 7.6% after 6 months of GH intervention. These changes in body composition have been associated with improvements in physical capability (19, 43). Children with PWS have significantly lower energy requirements for weight maintenance and are less physically active, compared with unaffected age-matched peers (47). The mechanism for these differences leading to the development of obesity is not clearly defined. Several hypotheses suggest decreased physical activity (47), decreased basal metabolic rate secondary to decreased LM (48), or increased energy intake as the primary cause of obesity in PWS (49). REE increased an average of 19% after GH intervention, which occurred concurrently with an average increase in LM of 7.6%.

After GH intervention, IGF-I and IGFBP-3 concentrations increased to 3.9 ± 2.5 SDS and 1.1 ± 1.8 SDS, respectively. On the current dosage of 0.043 mg/kg·d of GH, supranormal concentrations of IGF-I were achieved. Therefore, the GH doses used in this study were clearly in the pharmacologic range rather than physiologic range. These high IGF-I values may be due, in part, to the fact that the GH dosage used in this study was calculated based on actual body weight rather than an estimated ideal body weight. These IGF-I concentrations are similar to those seen in normal children on comparable doses of GH (50). However, it is not yet known what the concentration of free serum IGF-I is in these patients or the biological consequences, including cancer risk, if any, of this supranormal IGF-I concentration.

Fasting glucose and insulin concentrations were not different after GH intervention, and no patient developed glucose intolerance, a potential adverse effect of GH treatment. Fasting serum leptin concentrations were lower concurrent with a decline in total body FM after GH administration. Finally, GH administration did not significantly alter fasting serum concentrations of ghrelin in children with PWS. This latter finding is supported by a recent report that systemic GH administration to 23 patients with GHD (18 adult-onset GHD; 5 childhood-onset GHD), who achieved normal concentrations of serum IGF-I, did not alter ghrelin concentrations (51). Therefore, the elevated fasting serum ghrelin concentrations seen in children with PWS are not due to GHD.

After 6 months of GH treatment, various components of pulmonary function improved. It is possible that continued therapy would show even more improvement in pulmonary function. Because cardiorespiratory disease contributes the highest burden of morbidity and mortality in PWS, improvements in pulmonary function may benefit patient longevity. The prevalence of obstructive sleep apnea in this study was 100%, which agrees with some previous reports of a 50–100% prevalence of obstructive sleep apnea in PWS (52, 53). Following 6 months of GH intervention, the number of apnea events, duration of each apnea event, and number of hypopnea episodes decreased but not significantly. Therefore, a study with a greater number of subjects and longer duration of treatment may be warranted to draw stronger conclusions about the effect of GH therapy on sleep quality in children with PWS. It seems likely that early identification and treatment of these children’s sleep-disordered breathing may delay or prevent development of cor pulmonale, the primary cause of death in PWS.

No influence of GH intervention on cognitive function was seen in the children participating in this study. As well, no behavioral improvement or deterioration was noted with GH administration, which agrees with findings of others (23, 24).

In summary, although the current Food and Drug Administration indication for GH use in children with PWS is only for documented growth failure, GH intervention is likely to provide benefits for reasons other than improved adult height. The European Union has now approved GH for “improvement of growth and body composition in children with PWS.” For example, GH intervention clearly improves body composition (decreases FM and increases LM) and increases REE. Because REE contributes approximately 40% to total daily energy expenditure, this increase in REE may translate into improved ability of children with PWS to maintain a stable weight over time. In addition, improvements in muscle mass leading to improved pulmonary function with GH therapy may reduce the development of cardiorespiratory disease that is the most frequent cause of morbidity and mortality in this condition. Further studies need to be done to demonstrate any conclusive changes in sleep abnormalities with GH intervention. No adverse effects of GH administration were seen on glucose intolerance, scoliosis, or maladaptive behaviors during this short-term study, suggesting that the benefits of therapy outweigh any harm during 6 months of treatment. Additional studies need to be done to determine the optimal dosage of GH replacement therapy in children with PWS. Finally, just as the use of GH therapy is being reexamined in children with GHD during their transition to adulthood and beyond, further studies need to be done to evaluate the benefit of long-term GH therapy in individuals with PWS into adulthood on body composition, pulmonary function, sleep, and behavior.

Acknowledgements

We investigators thank the staff of the General Clinical Research Center for their dedicated patient care; Katherine Pratt for IGF-I and IGFBP-3 measurement; Sara Rae for measurements of pulmonary function; Breanna Nadeau for measurements of body composition; Barbara Gauthier, Dr. Robert Sack, Stacy Nordstrom, and Lisa DeJongh for sleep study analysis; and Vickie Nichols for randomization and education of the parents and subjects. We thank the families and children for their enthusiastic participation in this study. Also, we thank Dr. Ellen Magenis for help with recruitment of subjects and Dr. Gary Sexton for biostatistical advice.

This work was supported by grants from the General Clinical Research Center NIH/NCRR (M01-RR-00334), Pharmacia Corp. (to A.M.H., S.H.L., and R.G.R.), and NIH (K-23-DK-02689, to J.Q.P.).

Abbreviations:

• BASC,

  Behavior Assessment System for Children;

• BMC,

  bone mineral content;

• BMD,

  bone mineral density;

• BMI,

  body mass index;

• CBCL,

  Child Behavior Checklist;

• FEF_25–75,

  forced expiratory flow rate;

• FEV₁,

  forced expiratory volume expired in 1 sec;

• FM,

  fat mass;

• FT4,

  free T₄;

• GHD,

  GH deficiency;

• IGFBP,

  IGF-binding protein;

• IRMA,

  immunoradiometric assay;

• IQ,

  Intelligence quotient;

• LM,

  lean mass;

• PFR,

  peak flow rate;

• PWS,

  Prader Willi syndrome;

• REE,

  resting energy expenditure;

• SDS,

  sd score;

• VC,

  vital capacity.