Growth Hormone Treatment of Adults with Prader-Willi Syndrome and Growth Hormone Deficiency Improves Lean Body Mass, Fractional Body Fat, and Serum Triiodothyronine without Glucose Impairment: Results from the United States Multicenter Trial

Abstract

Context: GH replacement in Prader-Willi syndrome (PWS) children has well-defined benefits and risks and is used extensively worldwide. Its use in PWS adults has been limited by documentation of benefits and risks, as determined by larger multisite studies.

Objectives: Our objective was to evaluate the effectiveness and safety of GH in GH-deficient genotype-positive PWS adults.

Design: We conducted a 12-month open-label multicenter trial with 6-month dose-optimization and 6-month stable treatment periods.

Setting: The study was conducted at outpatient treatment facilities at four U.S. academic medical centers.

Patients: Lean and obese PWS adults with diverse cognitive skills, behavioral traits, and living arrangements were recruited from clinical populations.

Intervention: Human recombinant GH (Genotropin) was initiated at 0.2 mg/d with monthly 0.2-mg increments to a maximum 1.0 mg/d, as tolerated.

Main Outcomes Measures: Lean body mass and percent fat were measured by dual-energy x-ray absorptiometry.

Results: Lean body mass increased from 42.65 ± 2.25 (se) to 45.47 ± 2.31 kg (P ≤ 0.0001), and percent fat decreased from 42.84 ± 1.12 to 39.95 ± 1.34% (P = 0.025) at a median final dose of 0.6 mg/d in 30 study subjects who completed 6–12 months of GH. Mean fasting glucose of 85.3 ± 3.4 mg/dl, hemoglobin A1c of 5.5 ± 0.2%, fasting insulin of 5.3 ± 0.6 μU/ml, area under the curve for insulin of 60.4 ± 7.5 μU/ml, and homeostasis model assessment of insulin resistance of 1.1 ± 0.2 were normal at baseline in 38 study initiators, including five diabetics, and remained in normal range. Total T₃ increased 26.7% from 127.0 ± 7.8 to 150.5 ± 7.8 ng/dl (P = 0.021) with normalization in all subjects, including six (20%) with baseline T₃ values at least 2 sd below the mean. Mildly progressive ankle edema was the most serious treatment-emergent adverse event (five patients).

Conclusions: This multicenter study demonstrates that GH improves body composition, normalizes T₃, and is well tolerated without glucose impairment in PWS genotype adults.

Prader-willi syndrome (PWS) is a complex genetic disorder involving failure of expression of paternal alleles in the PWS region of chromosome 15q11–13. The PWS phenotype is variable, but is usually characterized by hyperphagia, childhood growth failure with adult short stature, hypogonadotropic hypogonadism, and body composition abnormalities including markedly increased body fat and decreased lean mass and diminished bone mineral density. Several studies have documented the presence of GH and IGF-I deficiency.

GH replacement in children with PWS is approved worldwide and is well accepted and extensively used due to well-documented benefits and defined safety profile (1–10). Previous studies suggest that GH therapy may be beneficial in adults with PWS (11, 12). However, previous data are limited by small sample sizes and inclusion of subjects lacking genetic confirmation. GH use in PWS adults remains controversial due to pleiotropic effects of GH on glucose homeostasis and perceived potential risks for diabetes, the metabolic syndrome (MS), and other obesity- and age-related comorbidities in overweight and obese PWS adults. Previous studies have not specifically addressed adults with wide ranges of intellectual performance, behavioral abnormalities, GH exposure, metabolic characteristics, gonadal status, and living situations.

We conducted a 1-yr, multicenter open-label study to evaluate safety and effectiveness of GH therapy in genotype-positive, GH-deficient (GHD) PWS adults. This open-label study of somatropin (Genotropin) was designed to replicate clinical practice and provide guidelines for GH use in diverse PWS subjects of varying age, GH exposure, sex-steroid replacement, cognitive function, living situations, and comorbidities at four clinical sites.

The primary study endpoint was change in dual-energy x-ray absorptiometry (DXA)-determined lean body mass (LBM) and percent body fat. Secondary measures included glucose tolerance and insulin response curves, and MS components. Thyroid function tests, including T₃, were also measured to characterize the thyroid axis in PWS adults and GH-mediated thyroid effects on lean mass, as reported in non-PWS GHD adults (13–15).

Subjects and Methods

Study population

Adults with a molecular diagnosis of PWS were recruited from clinical populations at four U.S. sites (Fig. 1). Study inclusion criteria were age greater than 18 yr, naive to GH or off therapy for a minimum of 12 months, basal sex- and age-specific IGF-I sd score (SDS) less than or equal to −1, stable estrogen/progesterone or testosterone replacement (in subjects receiving gonadal replacement), and evidence of dietary control of weight before study enrollment. Patients were excluded from the study if they had a major systemic illness, severe mental retardation, active tumors, poorly controlled diabetes, or evidence of diabetic retinopathy. The study was approved by Institutional Review Boards at each of the four study sites, and informed consent was obtained from both study subjects and their guardians.

Fig. 1.

Flow chart of the study. AE, Adverse event; WMC, Westchester Medical Center.

GH stimulation tests

GH was measured at baseline and 0, 30, 60, 90, and 120 min after the administration of 500 mg l-dopa, prepared in oral unit-dose formulation from pharmaceutical grade l-dopa powder (Bryce Laboratories, Stamford, CT). Secretagogue selection was based on the presumed hypothalamic level of the GH secretory defect in PWS and the desirability of avoiding iv infusion in the study subjects. All subjects were tested in the morning in the fasted state. GHD was defined a priori as a peak response less than or equal to 10 μg/liter.

Study protocol

The study was conducted between 2002 and 2005. Patients were seen monthly for the first 3 months and at 6 months for a dose-optimization phase and at subsequent 3-month intervals to evaluate treatment response, record anthropometric and behavioral data, monitor side effects and obtain serial IGF-I determinations. GH dosing was initiated at 0.2 mg/d and increased monthly in 0.2-mg increments, as tolerated, unless the IGF-I was greater than or equal to 1 sd, in which case the GH dose was decreased to the previous level for the remainder of the study. The primary study endpoints were DXA-determined LBM and percent fat. Sample size calculations were based on the power to detect a 5% change at a significance of 0.05.

Outcome and safety measures

Body composition was determined at baseline and study completion by DXA (Hologic QDR instruments, Bedford, MA). IGF-I values were obtained at the GH stimulation test, treatment initiation and 1-month intervals for the first 3 months, and 6- and 12-month follow-up visits and were sent to a single standard reference laboratory.

Adverse events were assessed at each study visit. Baseline and follow-up electrocardiograms and cardiac ultrasounds were performed in all subjects. Baseline sleep studies were not included in the initial protocol but were added to determine study eligibility in patients with clinical symptoms suggestive of obstructive sleep apnea.

Anthropometric measures and behavioral assessments

Height, weight, waist and hip circumference, and blood pressures were measured by a single trained observer per site. Waist circumference was recorded at the narrowest diameter between the xiphoid process and the iliac crest; hip circumference was defined as the widest diameter below the umbilicus. Dietary review and a 20-item behavioral assessment form were completed at all study visits by the accompanying parent or caregiver.

GH, glucose tolerance testing, thyroid function tests, and other laboratory determinations

GH levels were measured by a solid-phase competitive chemiluminescent enzyme immunoassay (Immulite-1000 analyzer; Diagnostic Products Corp., Los Angeles, CA) with sensitivity of 0.01 μg/liter and intra- and interassay coefficients of variation (CV), respectively, of 2.9 and 4.2%. All laboratory examinations were performed in a single central laboratory (with the exception of glucose and lipid profiles, sent to affiliated clinical laboratories).

Standard 75-g oral glucose tolerance tests were performed in the morning after a 12-h fast in 33 nondiabetic patients. Insulin levels were determined with a DPC-Immulite assay with intra- and interassay CV of 5.7 and 5.9%, respectively, and no cross-reactivity to proinsulin.

Serum TSH, total and free T₄, and total T₃ were measured in the fasting state at baseline, 6 months, and study completion using a DPC-chemiluminescent assay (respective CV of 10.0 and 6.2%, 6.7 and 6.7%, 5.5 and 5.0%, and 10.1 and 7.5%). Serial IGF-I measurements were assessed at every study visit (Esoterix Endocrine Laboratories, Calabas Hills, CA), with CV of 8.3 and 5.4%.

Total fasting adiponectin was collected during the baseline oral glucose tolerance tests in 32 nondiabetic subjects and assayed with an ELISA kit from LINCO (Bellerica, MA) with CV 6.2 and 9.3%.

Components of the MS and relevant covariates were assessed at baseline in 38 GHD patients (Table 1) using National Cholesterol Education Program guidelines (16). Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated by the formula: fasting insulin (micro-units/milliliter) times fasting glucose (millimoles/liter) divided by 22.5 (17).

TABLE 1.

Metabolic characteristics of 38 GH-deficient PWS adults at study enrollment

Parameters	Mean ± se	Median	Range
Age (yr)	30.5 ± 1.5	30.0	32 (17–49)
Height (cm)	152.3 ± 1.7	150.6	49.3 (133.7–183.0)
Weight (kg)	79.6 ± 4.1	74.2	103.2 (42.3–145.5)
BMI (kg/m2)	34.7 ±1.7	32.5	41.6 (22.0–63.6)
Waist (cm)	99.5 ± 4.0	95.9	125.8 (44.3–170.0)
Hip (cm)	115.1 ± 4.3	113.0	134.5 (45.5–180.0)
WHR	0.85 ± 0.03	0.86	0.67 (0.32–0.99)
Systolic BP (mm Hg)	116.3 ± 2.6	116.0	70 (90–160)
Diastolic BP (mm Hg)	73.7 ± 1.8	70.0	44 (48–92)
Fasting glucose (mg/dl)	85.3 ± 3.4	84.5	125 (50–175)
2-h glucose (mg/dl)	100.6 ± 7.2	93.0	117 (64–181)
AUC-insulin (μU/ml)	60.4 ± 7.5	49.5	171.8 (13.4–185.2)
Fasting insulin (μU/ml)	5.3 ± 0.6	4.9	17.2 (2.0–19.2)
HOMA-IR	1.1 ± 0.2	0.9	4.5 (0.3–4.8)
HbA1c (%)	5.5 ± 0.2	5.2	5.0 (3.9–8.9)
Total T3 (ng/dl)	130.9 ± 7.6	129.0	204.7 (46.3–251.0)
Total T4 (μg/dl)	8.6 ± 0.3	8.4	10.2 (4.2–14.4)
Free T4 (ng/dl)	1.1 ± 0.04	1.1	1.1 (0.6–1.7)
TSH (μU/liter)	1.5 ± 0.2	1.2	7.7 (0.01–7.7)
Total cholesterol (mg/dl)	191.5 ± 7.5	184.5	181 (99–280)
HDL cholesterol (mg/dl)	51.0 ± 2.6	49.5	56 (29–85)
LDL cholesterol (mg/dl)	118.7 ± 6.5	110	131 (70–201)
Triglycerides (mg/dl)	122.6 ± 10.5	107.0	239 (49–288)
Adiponectin (mg/dl)	20.8 ± 2.8	16.0	0.0 (0.0–50.0)
Lean body mass (kg)	42.6 ± 2.1	38.7	48.2 (22.0–70.2)
Percent fat	43.0 ± 1.1	44.4	21.3 (31.5–52.8)
IGF-I (μg/ml)	119.1 ± 11.1	102.0	313.0 (37.0–350.0)
Base IGF SDS	−1.9 ± 0.1	−2.0	3.07 (−3.22 to −0.15)

BP, Blood pressure; LDL, low-density lipoprotein; WHR, waist to hip ratio.

Statistical analyses

Statistical analyses were conducted using SPSS version 13.0. Nonnormally distributed variables were log-transformed or evaluated with nonparametric procedures. Data are expressed as the mean ± se and the median.

Data analysis included one-way ANOVA to compare group means for continuous variables across sites, paired t tests to determine mean group change from baseline at indicated time points, parametric and nonparametric correlations, and linear and logistic regression for covariate adjustment of main treatment effects. All reported P values are exact and represent two-tailed significance tests. Last observation carried forward was used to record final observations in patients who did not complete the 12-month final visit. No other corrections were made for missing data.

Results

Baseline characteristics of the initial study population

Thirty-eight subjects [25 women and 13 men; mean age, 30.5 ± 1.5 yr; and body mass index (BMI), 34.7 ± 1.7 kg/m²] met GHD inclusion criteria and were included in the study (Fig. 1). Mean and median GH peak responses in the study qualifiers were, respectively, 2.06 ± 0.32 and 1.35 μg/liter (Fig. 2). Thirty-four subjects had GH peaks less than or equal to 4 μg/liter; 19 had flat response curves with peak less than or equal to 1 μg/liter. Baseline mean ± se and median IGF-I and IGF-I SDS in the 38 study initiators were, respectively, 119.1 ± 11.1 and 102.0 μg/ml and −1.9 ± 0.1. Baseline characteristics of this study population are summarized in Table 1. Previous GH use was reported in eight (21.1%) subjects; 13 (34.4%) were on standard-dose sex steroid replacement or had regular menses at the onset of the study.

Fig. 2.

GH stimulation tests with l-dopa in 38 PWS adults who qualified for the study.

Metabolic parameters of the initial 38 study participants, including five with diabetes, include fasting glucose of 85.3 ± 3.4 mg/dl, hemoglobin A1c (HbA1c) of 5.5 ± 0.2%, fasting insulin and area under the curve (AUC)-insulin of 5.3 ± 0.6 μU/ml and 60.4 ± 7.5 μU/ml, and HOMA-IR of 1.1 ± 0.2. Glucose tolerance tests performed on 33 nondiabetic patients identified one patient with both fasting hyperglycemia and impaired glucose tolerance (IGT). Including the five patients with diabetes, only six of 33 (18.1%) with complete data met National Cholesterol Education Program consensus criteria for MS. Abnormal sex-specific waist circumferences (men, >102 cm; women, >88 cm), present in 24 (63.2%) of 38 subjects, was the most common MS component. Sixteen (of 35 with complete data) or 45.7% had abnormal sex-specific high-density lipoprotein (HDL) levels, and eight (25%) of 32 with available data had abnormal baseline triglycerides, whereas only four of 38 (10.5%) met MS blood pressure criteria.

Mean adiponectin was 20.8 ± 2.8 mg/liter, 23.4 ± 3.5 mg/liter in women and 15.0 ± 4.1 mg/liter in men (P = 0.163 for ANOVA by sex). Significant direct and inverse correlates of adiponectin included age (ρ = 0.621; P = 0.000), BMI (ρ = −0.553; P = 0.002), and fasting insulin (ρ = −0.511; P = 0.005).

GH dose optimization, body composition, and metabolic changes

IGF-I SDS improved from −1.8 ± 0.2 to −0.2 ± 0.2 at a median final dose of 0.6 mg/d with a range of 0.4–1.0 mg/d (Table 2). Significant improvements were observed in both LBM and percent fat. LBM increased from 42.66 ± 2.25 to 45.47 ± 2.31 kg (P ≤ 0.0001), and percent fat decreased from 42.84 ± 1.12 to 39.95 ± 1.34% (P = 0.025). Regression models indicated that these changes were independent of age, initial BMI, sex steroid use, and social setting and were noted in a subset of patients who were evaluated at 6 months. The percent increase in LBM was significantly greater in the male than the female subjects, 6.66 vs. 8.39%, whereas the decrease in percent fat mass, −4.35 vs. −6.28% was not (Fig. 3). (This is reported as a post hoc analysis, because the study was not powered to permit stratification by sex for the main treatment effect.)

TABLE 2.

Metabolic and hormonal results in 30 PWS adults at baseline and after GH treatment

Parameters	Baseline		After GH treatment		P value
	Mean ± se	Median	Mean ± se	Median
Height (cm)	151.7 ± 2.2	150.6	147.6 ± 5.6	150.0	0.431
Weight (kg)	78.7 ± 4.7	71.8	78.8 ± 5.1	71.6	0.707
BMI (kg/m2)	33.7 ± 1.8	32.4	32.6 ± 1.6	31.2	0.194
Waist (cm)	101.4 ± 4.1	98.3	98.4 ± 3.3	96.5	0.299
Systolic BP (mm Hg)	116.0 ± 2.7	116.0	116.4 ± 2.1	113.5	0.866
Diastolic BP (mm Hg)	72.4 ± 2.1	70.0	70.6 ± 1.8	71.0	0.364
Fasting glucose (mg/dl)	81.1 ± 2.6	82.0	87.6 ± 3.1	86.0	0.010
2-h glucose (mg/dl)	90.0 ± 5.8	90.5	129.9 ± 11.6	123.5	0.015
AUC-insulin (μU/ml)	53.4 ± 6.2	50.6	80.3 ± 10.5	73.0	0.006
Fasting insulin (μU/ml)	4.9 ± 0.7	4.7	7.7 ± 1.2	5.7	0.003
HOMA-IR	1.1 ± 0.2	0.9	1.9 ± 0.3	1.1	0.002
HbA1c (%)	5.4 ± 0.2	5.0	5.5 ± 0.2	5.3	0.174
Total T3 (ng/dl)	127.0 ± 7.8	128.0	150.5 ± 7.8	142.0	0.021
Total T4 (μg/dl)	8.6 ± 0.4	8.5	7.9 ± 0.3	7.8	0.053
Free T4 (ng/dl)	1.1 ± 0.05	1.1	1.1 ± 0.03	1.0	0.104
TSH (μU/ml)	1.6 ± 0.3	1.3	1.6 ± 0.2	1.4	0.985
Total cholesterol (mg/dl)	194.1 ± 7.8	190.0	189.1 ± 7.3	180.0	0.423
HDL cholesterol (mg/dl)	51.0 ±3.2	49.5	51.1 ± 2.9	49.5	0.949
LDL cholesterol (mg/dl)	119.4 ± 7.2	113.0	117.4 ± 6.6	111.5	0.779
Triglycerides (mg/dl)	118.1 ± 11.0	97.0	179.3 ± 48.4	98.0	0.200
Lean body mass (kg)	42.7 ± 2.3	38.6	45.5 ± 2.3	42.3	0.000
Percent fat	42.8 ± 1.1	43.7	40.0 ± 1.3	40.3	0.025
IGF-I (μg/ml)	121.8 ± 17.3	107.0	290.8 ± 37.2	299.5	0.001
IGF SDS	−1.8 ± 0.2	−1.9	−0.2 ± 0.2	−0.4	0.000
No. of metabolic syndrome variables	1.4 ± 0.2	1.0	1.45 ± 0.2	1.0	0.646

BP, Blood pressure.

Fig. 3.

Changes in LBM and percent fat in male and female study subjects after GH treatment.

HbA1c did not vary significantly during the study: 5.4 ± 0.2 to 5.5 ± 0.2%. Fasting glucose increased from 81.1 ± 2.6 to 87.6 ± 3.0 mg/dl (6.6%) but remained in the normal range in 27 patients (including three with diabetes) with available data (P = 0.01). Fasting insulin of 4.9 ± 0.7 μU/ml and AUC-insulin of 53.4 ± 6.2 μU/ml increased to 7.7 ± 1.2 μU/ml and 80.3 ± 10.5 μU/ml (P = 0.006 and P = 0.003) in 22 tested (nondiabetic) study participants. HOMA-IR increased from 1.1 ± 0.1 to 1.9 ± 0.3 (P = 0.007). Management of diabetes did not change during the study. Of 29 patients with available data, the prevalence of MS increased after GH to 31.0% (nine patients).

Thyroid measurements

None of the subjects were known to be hypothyroid or were on thyroid hormone replacement at enrollment. However, among the 30 study completers, six subjects (20%) had total T₃ values of at least 2 sd below the assay mean (127.0 ng/dl) at baseline. Baseline T₄, free T₄, and TSH were within the normal range for all subjects. Statistically significant GH-induced changes in total T₃ and total T₄ were observed (Fig. 4). Total T₃ increased by 26.7% (P = 0.021) from 127.0 ± 7.8 to 150.5 ± 7.8 ng/dl, with normalization in all subjects. Total T₄ decreased by 7.8% (P = 0.051) from 8.6 ± 0.4 to 7.9 ± 0.4 ng/dl. TSH, 1.56 ± 0.25 vs. 1.49 ± 0.17 μU/dl, and free T₄, 1.1 ± 0.1 vs. 1.0 ± 0.0 ng/dl, were essentially unchanged. Correlation of T₃ and LBM changes just failed to reach significance (P = 0.06).

Fig. 4.

Thyroid function tests at baseline and after GH treatment.

Behavioral assessment scores

Symptom scores completed by family or health aides present at the study visits demonstrated a reduction in total symptom score from a mean 8.3 ± 4.7 to 6.13 ± 4.5 (P = 0.04) in 15 subjects with evaluable symptom checklist scores.

Adverse events

Mild progression of preexisting ankle edema (five patients) was the most serious treatment-emergent adverse event; one patient withdrew from the study because of myalgias associated with lower leg swelling. There was no other evidence of fluid retention for these or any other subjects at baseline or during the study. Baseline cardiac ultrasounds revealed previously unidentified, clinically insignificant valvular defects in two of 38 study subjects but were otherwise essentially normal and were unchanged in follow-up studies.

Discussion

GH treatment effects

GH-mediated improvements in growth, body composition, and respiratory function have been documented in studies of children with PWS (1–6). Accordingly, GH was approved for “long-term treatment of pediatric patients with growth failure due to Prader-Willi Syndrome (PWS)” in the United States in 1999 and is currently the only medication specifically approved for use in PWS. Although most adults with PWS have deficiencies of GH or the GH/IGF axis and suffer similar body composition abnormalities compared with both children with PWS and adults with GH deficiency (18), data on GH treatment of adults with PWS is limited. In a study of 13 adults with genetic confirmation of PWS, a 12-month course of GH increased LBM (3.7 ± 0.9 kg) and decreased percent fat (−3.8 ± 1.3%), (12) with persistence of these effects in six subjects who elected to continue treatment for a median of 5.1 yr (19).

Our results from a multicenter study provide additional information regarding GH status and treatment effectiveness in a larger population of adults with a confirmed diagnosis of PWS. For our investigations, l-dopa was selected as the GH secretagogue based on the presumption that a GH secretory defect in PWS would likely be at the hypothalamic level. This testing confirmed GHD in 38 of 40 subjects (95%), in association with low IGF-I levels.

During the initial 6 months of the protocol, the GH doses were progressively increased as described in Subjects and Methods. During this period, five patients experienced ankle-limited swelling, and one subject had associated myalgias leading to discontinuation of GH treatment, whereas others remained in the study with adjustment of GH dose. The mean GH dose, 0.6 mg/d (∼0.008 mg/kg/d for our study population) at dose optimization is similar to doses used in non-PWS adult GHD (20) and is substantially lower than the 0.03–0.05 mg/kg/d used in pediatric PWS and GHD.

Over the subsequent 6 months of the study, the GH doses were maintained at a constant level. Our results provide confirmation that GH treatment, even over a relatively short period of time, leads to significant improvements in body composition in adults with PWS, including a decrease in percent fat and increased LBM and, typically, are independent of changes in BMI, a measurement that is not sensitive to shifts in body compartments. These improvements in LBM may be important in PWS, a condition in which LBM is severely deficient (21, 22).

Monitoring of glucose homeostasis was an important objective of the multicenter study, although baseline data indicated low prevalence of glucose abnormalities and insulin resistance, as reported in some (23–25) but not all previous studies (26). Despite modest increases in glucose and insulin, HbA1c and all measures of insulin and glucose homeostasis remained within normal limits in all study subjects, including five with diabetes and one with IGT. Thus, although GH may worsen IGT or diabetes, this was not demonstrated in this limited study population.

The prevalence of MS, unexpectedly low at baseline (six patients, 18.1%), increased after GH (among 29 patients with available data) to 31.0% (nine patients). In three patients, this increase was secondary to increased triglycerides and corresponded to weight gain, directly attributable to documented dietary lapses, presumably unrelated to GH treatment.

The thyroid axis in PWS

The thyroid axis is generally considered to be normal in PWS subjects because previous studies have not revealed any abnormalities in standard determinations of TSH, total or free T₄, or total and free T₃ (23, 27–29). However, Bray et al. (23) reported TRH-stimulated TSH values in nine PWS subjects that were 2 sd above those of six obese controls despite normal basal TSH, T₃, and T₄ levels in their cohort of 40 very obese subjects (mean BMI = 47.0 kg/m²), suggesting mild impairment in the hypothalamic-pituitary control system for the thyroid, and Tauber et al. (29) reported hypothyroidism demarcated by low T₄ and/or elevated TSH levels in 32% of 28 children with PWS, 23 of whom were genotype positive. In a recent study of 47 children and adults with PWS, Butler et al. (28) reported an incidence of hypothyroidism comparable to the general population based on TSH and T₄ levels. However, despite a normal reported mean total T₃ level, 137 ± 37 (sd) ng/dl, the level appeared low (<80 ng/dl) on inspection of individual bar graphs in five (12%) of 41 subjects with available data.

To our knowledge, the observed baseline T₃ abnormalities in 20% of GHD PWS subjects and their normalization with GH therapy are unique. However, they are consistent with reports in children with PWS (30) and in non-PWS GHD adults, documenting dose-dependent enhancement of extrathyroidal T₄ to T₃ conversion (13–15, 31) and suppression of circadian TSH levels after the administration of GH (15, 32) as well as with other widely reported subtle, but notable complexities in the interaction of GH with the hypothalamic-pituitary-thyroid axis (32). The concomitant group mean reduction in total T₄ levels after GH (8.6 to 7.9 ng/dl, P = 0.05) provides additional evidence of increased deiodination of T₄ in our PWS subjects and supports the hypothesis that peripheral conversion of T₄ to T₃ may contribute to GH-mediated enhancement of resting energy expenditure and LBM (15).

Type 2 iodothyronine deiodinase, predominantly found in skeletal muscle, has been identified as the major source of plasma T₃ in euthyroid subjects (33). Theoretically, activity of this deiodinase may be low in PWS, a condition characterized by a remarkably low level of LBM, and the resultant low T₃ levels may contribute to pathology in some patients. Augmentation of this deiodinase activity by increasing LBM, e.g. with GH therapy, could be particularly important in patients with PWS. Although our data correlating changes in LBM and T₃ did not reach statistical significance (P = 0.06), the observed trend and response to GH treatment are consistent with this hypothesis. In summary, our findings suggest that, as in other GHD subjects, PWS adults may have subtle thyroid abnormalities, especially T₃ alterations, which are reversed with GH replacement and that this may contribute to amelioration of the uniquely abnormal body composition associated with PWS.

Other qualitative findings

Profound baseline differences in cognitive and social skills, living situations, behavioral symptoms, and GH exposure limited acquisition and interpretability of GH-mediated changes in quality of life (e.g. the study cohort included two subjects who attended college; eight young adults who received GH throughout childhood, several of whom attended regular educational programs; and institutionalized, morbidly obese older adults with significant comorbidities). Behavioral assessment scores based on rating scales of typical PWS-associated symptoms demonstrated improvement in number and degree of many behavioral abnormalities, consistent with other studies (34).

Formal assessment of physical fitness was not included in our multicenter trial because of the wide variation in baseline exercise capacity precluding the use of a single study instrument. However, in qualitative data elicited from open-ended questions, study subjects previously engaged in regular exercise reported increased aerobic conditioning (increased duration on treadmill, biking, or lap swimming) and physical strength (load bearing in weight training) in supervised activities. Both study subjects and their guardians also reported increased energy levels, improved attention spans, and general sense of well-being. Future studies should include pre- and post-treatment measurements of these important domains in suitable patient subsets.

Strengths and limitations of the study

Our study design did not include a placebo arm because the study investigators could not justify the use of injections without discernible benefit to our study population. At the study initiation, 2002, GH-mediated body composition changes had been well documented in multiple studies of GHD adults and GH had been approved for growth failure in children with PWS, although its safety profile had not been established in adults with PWS. The preservation of normal fasting blood sugar and HbA1c, even among five diabetics enrolled in the study, supports the relative safety of GH in this population, but clearly, additional follow-up studies will be necessary to determine long-term consequences of GH treatment in this population.

We believe the diverse nature of the study population—the wide range of ages, comorbidities, and cognitive and social skills of our subjects—contributed to the reliability and validity of the study findings.

Although the low prevalence of PWS in the general population precluded restriction of the study cohort to a homogeneous sample, control for the influence of potentially important covariates, including sex steroid status, were addressed with multivariate analyses. The absence of effect modification of the GH-mediated improvements in body composition despite extreme ranges in various baseline characteristics, including age (17–49 yr), weight (42.3–145.5 kg), and BMI (22.0–63.6 kg/m²) as well as sex steroid status and GH exposure, suggests that the study findings may be extrapolated to additional heterogeneous adult PWS cohorts. The notable exception, effect modification of LBM by sex, with greater treatment response in men than women, has been previously reported in GH replacement studies of GHD adults with (12) and without PWS (20).

In conclusion, our 12-month multicenter trial in 38 diverse GHD genotype-positive PWS adults demonstrates that GH therapy improves LBM and percent fat and normalizes IGF-I levels without glucose impairment. The incidence of MS was low at baseline and, with the exception of triglyceride elevations attributable to dietary lapses in three study subjects, did not change appreciably during GH treatment. In addition, our findings suggest that total T₃ levels may be low in a significant proportion of this population and are normalized with GH therapy. As with treatment of non-PWS adult GHD subjects, some individuals may have significant GH-mediated water retention during dosage optimization, mandating judicious GH initiation, dosage adjustment, and careful monitoring. Long-term studies will be necessary to delineate further the risks and benefits of GH in this population.

Acknowledgments

We thank our patients with PWS and their families for their commitment to the study and Janalee Heineman and the PWSUSA for their support of the trial.

This study was supported by an unrestricted, investigator-initiated, clinical research grant from Pfizer.

Current address for P.D.K.L.: EMD Serono Inc., Rockland, Massachusetts 02370.

Disclosure Statement: H.R.M. is a consultant to Merck and receives grant support from Lilly and Glaxo-Smith-Kline. P.D.K.L. is currently employed by Serono. B.Y.W. and S.M. received grant support from Pfizer (2003–2005). Lilly, Serono, and Pfizer make rhGH, but these products are not approved for PWS. W.B.Z., M.F., M.C., B.P., and A.L.S. have nothing to report.

Abbreviations

• AUC,

  Area under the curve;

• BMI,

  body mass index;

• CV,

  coefficients of variation;

• DXA,

  dual-energy x-ray absorptiometry;

• GHD,

  GH-deficient;

• HbA1c,

  hemoglobin A1c;

• HDL,

  high-density lipoprotein;

• HOMA-IR,

  homeostasis model assessment of insulin resistance;

• IGT,

  impaired glucose tolerance;

• LBM,

  lean body mass;

• MS,

  metabolic syndrome;

• PWS,

  Prader-Willi syndrome;

• SDS,

  sd score.