The Effect of Growth Hormone on the Response of Total and Acylated Ghrelin to a Standardized Oral Glucose Load and Insulin Resistance in Children with Prader-Willi Syndrome

Abstract

Context: Fasting levels of plasma ghrelins are grossly elevated in children with Prader-Willi syndrome (PWS). The cause of this elevation and the regulation of ghrelins in PWS is largely unknown. The regulatory role of individual nutritional components and of GH is not well characterized.

Objective: We investigated the influence of GH on acylated (aGhr) and total ghrelin (tGhr) concentrations before and after an oral glucose load, and on insulin resistance in PWS children.

Design, Patients, and Interventions: In a clinical follow-up study, plasma ghrelins were measured during an oral glucose tolerance test, and parameters of insulin resistance were determined in 28 PWS children before and/or 1.18 (0.42–9.6) yr (median, range) after start of GH therapy (0.035 mg/kg body weight per day).

Main Outcome Measures: Fasting and postglucose concentrations of aGhr and tGhr and homeostasis model assessment 2 insulin resistance were the main outcome measures.

Setting: The study was conducted in a single center (University Children’s Hospital).

Results: High fasting [1060 ± 292 (sd) pg/ml; n = 12] and postglucose trough (801 ± 303 pg/ml; n = 10) tGhr concentrations in GH-untreated PWS children were found to be decreased in the GH-treated group (fasting 761 ± 247 pg/ml, n = 24, P = 0.006; postglucose 500 ± 176 pg/ml, n = 20; P = 0.006). In contrast, aGhr concentrations and insulin resistance were not changed by GH treatment. Both aGhr and tGhr concentrations were decreased by oral carbohydrate administration, independent of the GH treatment status.

Conclusions: Our results indicate that, in PWS children, aGhr and tGhr are differentially regulated by GH.

FASTING LEVELS OF plasma ghrelin, the orexigenic stomach-derived hormone, are grossly elevated in children with Prader-Willi syndrome (PWS) (1–3). This may contribute to the observed hyperphagia. The cause of the ghrelin elevation and the mechanisms of ghrelin regulation in PWS are largely unknown (4). However, the capacity of a mixed meal to suppress ghrelin is maintained in children with PWS (5–8). This effect appears to be lost in some, but not all PWS adults (2, 8). The regulatory role of individual nutritional components like carbohydrates is less well characterized. In a recent study, glucose administration was shown to decrease the concentration of ghrelins in young PWS children (9). Also, reduced insulin resistance has been implicated in the development of hyperghrelinemia in PWS adults (10).

The negative correlation between body mass index (BMI) and ghrelin seen in healthy obese children and lean and obese adults (11, 12) was not observed in some PWS cohorts (3, 13). However, in other groups of PWS individuals, this relationship was shown to be maintained at higher ghrelin levels (14). Because more than 85% of PWS children fulfill the auxological and laboratory criteria of GH deficiency, GH treatment is an established therapy in PWS children (15). It has normalizing effects on disturbed body composition, and contributes to weight loss in PWS (16–19). GH treatment appeared to have no significant effect on basal ghrelin concentrations in previous studies involving children and young adults with PWS (20, 21). With one exception (9), ghrelin concentrations in PWS have been reported as total ghrelin (tGhr), with no distinction being made between the GH secretagogue receptor-binding acylated form and the desacyl form.

To further elucidate the role of GH and carbohydrates in the regulation of the ghrelin isoforms in PWS, we aimed to characterize insulin resistance and the response of tGhr and acylated ghrelin (aGhr) to a standardized oral carbohydrate load. We report on the effect of GH therapy on these parameters and on BMI in PWS children and adolescents.

Patients and Methods

Patients

This study was carried out as a single center study in the tertiary care setting of the University Children’s Hospital in Essen. Twenty-eight consecutively admitted children and adolescents with molecularly proven PWS (13 female), age 2.5–16.1 yr [BMI sd score (SDS) range −2.04 to 4.31] were included in the study. They were undergoing an oral glucose tolerance test (oGTT) as routine evaluation before and during GH treatment (0.035 mg/kg body weight daily sc; maximum dose 2.7 mg). In addition to glucose and insulin, total and aGhr concentrations were measured in the samples obtained throughout the tests. Thirty-six tests were carried out. Test results were grouped in two different ways for analysis. First, in eight children who had an oGTT before and after the start of GH therapy, data were compared intraindividually. In this subset, the age difference between examinations pre- and post-GH therapy (n = 8) was 1.14 (0.29–1.8) yr (median, range). For further analysis, test data for all children were combined. They were grouped according to their GH treatment status (n = 12 before, n = 24 after the start of GH therapy). In this analysis, children who had only one oGTT (before or after the onset of GH treatment) were combined with those tested before and after treatment and the pretreatment and posttreatment levels were compared. For the 24 GH-treated children, 1.18 (0.42–9.56) yr (median, range) had elapsed between the start of GH treatment and the oGTT. BMI was transformed into BMI SDS using standards derived from a population of healthy German children (22).

Ethical considerations

The study protocol was approved by the local ethics committee. Written informed consent was obtained from the parents of the patients and from those children old enough to follow the explanations.

oGTT

Carbohydrate load was carried out after an overnight fast by giving 1.75 g/kg body weight (maximum amount 75 g) of a commercially available glucose syrup (Dextro O.G.-T.; Roche, Grenzach-Wyhlen, Germany) within 1–5 min by mouth. Blood samples for glucose and insulin determination were obtained before and 30, 60, 90, and 120 min after carbohydrate ingestion. At the same time points, blood for measurement of total and aGhr was collected into EDTA-coated tubes and immediately transferred on ice water to the laboratory, where plasma was separated in a refrigerated centrifuge within the next 30 min. The plasma was then stored at −80 C for up to a maximum of 2 months before assay.

All oGTTs were performed between 0800–1000 h. All patients received their last GH injection on the day before the test between 1900–2200 h.

Assays

Blood glucose was determined by a standard glucose oxidase method. Plasma insulin was measured using a commercially available RIA (Insulin RIA; Biochem ImmunoSystems, Freiburg, Germany). Intraassay (interassay) variability was characterized by a coefficient of variation of 4.5–7.4% (4.5–8.0%) within a concentration range of 9.2–94.2 μU/ml (8.8–95.2 μU/ml). Sensitivity was 1.0 μU/ml.

tGhr was measured by a commercially available RIA (Phoenix Pharmaceuticals, Inc., Belmont, CA). The antibody has 100% cross-reactivity to human desacyl and acylated human ghrelin. Intraassay (interassay) variability was 4.3% (16.4%), sensitivity was 80 pg/ml.

aGhr was measured by a commercially available RIA (Linco Research, St. Charles, MO). The antibody has 100% cross-reactivity to human aGhr and <0.1% cross-reactivity to desacyl human ghrelin. Intraassay (interassay)variability was 13.9% (20.4%). Percent homeostasis model assessment 2 β-cell function (HOMA2B) and the degree of homeostasis model assessment 2 insulin resistance (HOMA2IR) were determined from fasting glucose and insulin concentrations by homeostasis model assessment 2 (HOMA2) modeling (23). HOMA2 parameters were calculated using the HOMA Calculator (The University of Oxford 2004; www.dtu.ox.ac.uk). Insulin resistance was assumed at HOMA2IR values greater than 4 (24).

Statistical evaluation

If not indicated otherwise, data are reported as mean ± sd. Group and intraindividual comparisons were made by the Wilcoxon rank sum test and the Wilcoxon signed rank test, respectively. Significance was taken as P < 0.05. Associations between parameters were assessed separately by Spearman rank correlation. Model-selection multiple stepwise regression analysis was used to evaluate the relative importance of GH treatment status, carbohydrate administration, BMI SDS, and HOMA2 parameters for the variation of the ghrelin plasma concentration in PWS. Parameters were allowed to enter into the model with F statistics significant at the 0.9 level, and allowed to stay in the model at the 0.5 significance level. Calculations were performed using the SAS software (release 8.2, edition 2002; SAS Institute, Cary, NC).

Results

Ghrelin response to GH treatment

First we consider the eight PWS children for whom data were collected both before and after the start of GH treatment. BMI SDS (P = 0.0078), basal tGhr (P = 0.0078), and postcarbohydrate trough tGhr (P = 0.0313) concentrations in plasma decreased significantly after the start of GH treatment (Fig. 1). In contrast, no GH-related decrease was observed for the aGhr concentrations, either before or after carbohydrate administration.

Fig. 1.

Intraindividual comparisons of BMI SDS (n = 8) (A), fasting plasma tGhr (picograms per milliliter; n = 8) (B), and postcarbohydrate (CH) tGhr trough concentration (picograms per milliliter; n = 7) (C), fasting plasma aGhr (picograms per milliliter; n = 8) (D), and postcarbohydrate (CH) aGhr trough concentration (picograms per milliliter; n = 7) (E) in the eight children and adolescents with PWS who underwent an oGTT before and after the start of GH therapy. P values indicate significant differences. n.s., Not significant.

Ghrelin response to an oral carbohydrate load

In response to a standardized oral carbohydrate load, both plasma tGhr and aGhr concentrations decreased significantly in PWS children irrespective of their GH treatment status. In the GH-untreated PWS children, basal fasting tGhr concentrations decreased from 1060 ± 292 pg/ml (n = 12) to postcarbohydrate trough concentrations of 801 ± 303 pg/ml (n = 10; P = 0.044) (Fig. 2, A and B; left columns). aGhr decreased from 187 ± 99 pg/ml (n = 12) to trough concentrations of 106 ± 36 pg/ml (n = 10; P = 0.035) (Fig. 2, C and D; left columns). Mean trough tGhr (aGhr) concentrations in GH-untreated children amounted to 66.1% (70.3%) of their respective basal fasting levels. In the GH-treated children, fasting tGhr concentrations decreased from 761 ± 247 pg/ml (n = 24) to postcarbohydrate trough concentrations of 500 ± 176 pg/ml (n = 20; P = 0.0007) (Fig. 2, A and B; right columns). aGhr decreased from 184 ± 95 pg/ml (n = 24) to trough concentrations of 118 ± 66 pg/ml (n = 20; P = 0.033) (Fig. 2, C and D; right columns). The relative mean decrease of tGhr (69.8%) and aGhr (65.8%) did not differ from the mean decrease recorded for the GH-untreated group.

Fig. 2.

Fasting plasma tGhr (picograms per milliliter) (A), plasma tGhr trough concentration after an oral carbohydrate load (B), fasting plasma aGhr (C), and plasma aGhr trough concentration after an oral carbohydrate load (D) in children and adolescents with PWS before (− GH Tx) and during (+ GH Tx) GH therapy. Significant differences between ghrelin values before and after carbohydrate administration are indicated by symbols (*, P = 0.044; **, P = 0.0007; †, P = 0.035; ‡, P = 0.033). Brackets with P values indicate significant differences with regard to GH treatment status. n.s., Not significant. Box-and whisker-plot: dot, mean; box, interquartile range; whiskers, range.

With regard to the effects of GH treatment on tGhr and aGhr concentrations, the results of the 36 oGTTs considered altogether confirmed the data of the eight patients for whom measures were available both before and after GH treatment. The observed tGhr decrease after start of GH therapy resulted in an increased mean aGhr to tGhr ratio postcarbohydrates in GH-treated PWS (untreated vs. treated: 0.18 ± 0.11 and 0.29 ± 0.14; P = 0.038).

Stepwise multiple regression analysis performed for model-building purposes revealed that, in a model including GH treatment status, carbohydrate administration, BMI SDS, chronological age, HOMA2B, and HOMA2IR, 48% of the variability of plasma tGhr concentration, but only 24% of aGhr variability could be explained by the model (Table 1). For tGhr variability, GH treatment status (21.3%) and carbohydrate administration (17.8%) played a major role; BMI SDS contributed only 6.9%. The influence of HOMA2 parameters and chronological age was low. With regard to aGhr variability, major explanatory factors were carbohydrate administration (14.4%) and HOMA2IR (5.3%). Chronological age (2.4%) and HOMA2B (1.9%) played only a minor role. There appeared to be no influence of GH treatment status and BMI SDS on aGhr.

TABLE 1.

Effects of GH treatment status, carbohydrate administration, BMI SDS, chronological age, HOMA2B, and HOMA2IR on total and acylated plasma ghrelin concentration

Model dependent variable	Step	Model variable	Partial r²a	Model r²b	F valuec	Pc
tGhr	1	GH STATUS	0.2130	0.2130	17.32	<0.0001
	2	CH ADMIN	0.1783	0.3912	18.45	<0.0001
	3	BMI SDS	0.0691	0.4603	7.93	0.0065
	4	CA	0.0186	0.4789	2.18	0.1447
	5	H2IR	0.0017	0.4806	0.19	0.6619
	6	H2B	Not includedd
aGhr	1	CH ADMIN	0.1436	0.1436	10.73	0.0017
	2	H2IR	0.0528	0.1964	4.20	0.0447
	3	CA	0.0243	0.2207	1.84	0.1795
	4	H2B	0.0193	0.2401	1.55	0.2176
	5	BMI SDS	Not includedd
	6	GH STATUS	Not includedd

GH STATUS, Before/after start of GH treatment; CH ADMIN, before/after carbohydrate administration; CA, chronological age; H2B, HOMA2B; H2IR, HOMA2IR.

^a

Partial r² is the partial coefficient of determination, indicating the proportion of total (acylated) ghrelin variability that is explained by a single model variable.

^b

Model r² is the cumulative sum of partial coefficients of determination, indicating the proportion of total (acylated) ghrelin variability that is explained by the group of model variables, the highest number showing the proportion explained by all included variables (the model).

^c

F value. P, Critical value and P value for keeping a model variable in the model.

^d

“Not included” indicates that a variable was chosen, but did not fulfill the criteria (P < 0.5) for staying in the model, meaning that this variable did not contribute to explain total (acylated) ghrelin variability in plasma.

There were no correlations between BMI SDS and basal aGhr concentration in GH-treated or GH-untreated patients. However, in GH-untreated PWS children, a significant negative correlation existed between BMI SDS and parameters reflecting the postcarbohydrate tGhr decrease (BMI SDS vs. postcarbohydrate tGhr trough: r = −0.71, P = 0.021; vs. postcarbohydrate tGhr trough expressed in percent of basal tGhr: r = −0.72, P = 0.03; vs. 2 h postcarbohydrate tGhr: r = −0.85, P = 0.004). These correlations were lost after the start of GH treatment (Fig. 3).

Fig. 3.

Correlation (Spearman rank) of BMI SDS with plasma tGhr (picograms per milliliter) 2 h after oral carbohydrate load (upper panels) and tGhr trough concentration after oral carbohydrate load, expressed in percent of fasting tGhr concentration (lower panels) in children and adolescents with PWS before GH treatment (left panels) and on GH treatment (right panels).

Glucose, insulin, and HOMA2 indices

Fasting blood glucose concentration was normal in the PWS patients at all time points in the study. Two patients had disturbed glucose tolerance, indicated by an increased blood glucose 2 h after oral carbohydrate load. In one girl, a 2-h blood glucose of 9.9 mmol/liter (normal: <7.77) was found before the start of GH therapy. This girl had the highest BMI SDS (4.31) of the PWS cohort. The other girl with a 2-h blood glucose of 8.3 mmol/liter shortly after the start of GH therapy had normal glucose tolerance on subsequent examinations during GH treatment. Two other PWS girls with normal glucose tolerance under GH therapy had HOMA2IR values of 5.6 and 7.3 that were considered indicative of insulin resistance. Basal fasting insulin was elevated in one patient before GH treatment (28.2 μU/ml). In this patient, insulin returned to normal with GH treatment. Two other patients showed elevated fasting insulin concentrations (45.4 and 59.3 μU/ml) while under GH treatment. Fasting and maximum insulin concentration after carbohydrate load, and HOMA2IR did not differ between the GH-treated and the GH-untreated group.

In GH untreated PWS children, there were no correlations between BMI SDS and HOMA2IR (and HOMA2B). After the start of GH treatment in the PWS group, a moderate correlation between BMI SDS and HOMA2IR developed (r = 0.47, P = 0.021). A moderate correlation of insulin resistance with the 2-h postcarbohydrate load aGhr concentration (r = −0.49, P = 0.034) was seen only in the GH-treated PWS group. Otherwise, no correlations for ghrelin data with HOMA2 parameters were found.

Discussion

PWS is the most prevalent form of syndromal obesity and results from a loss of function of the paternally expressed genes on chromosome 15q11–13. Disturbed satiety and hyperphagia (25–27) are the main factors contributing to the development of morbid obesity in PWS. Compared with simple obesity, obesity in PWS is characterized by an increased relative contribution of body fat to total body weight (28, 29), a decreased ratio of visceral to total adipose tissue (30), a lesser degree of insulin resistance (31), and by massively elevated total and aGhr concentrations (1–3, 6, 9, 10, 13, 14, 20, 21, 32, 33). Because ghrelin has orexigenic properties, the latter observation could explain at least partially the hyperphagia. Furthermore, the majority of PWS patients, but not simple obese individuals, fulfill the auxological and laboratory criteria of GH deficiency. Hence, GH is used for the treatment of PWS children in many countries (15). In this study, we have demonstrated for the first time that GH treatment is able to decrease both elevated basal and postcarbohydrate tGhr concentrations in PWS children and adolescents. This was associated with a significant intraindividual decrease of BMI SDS in patients. aGhr concentrations were not changed by GH treatment in our PWS cohort. However, both acylated and desacyl ghrelin concentrations decreased after carbohydrate administration in patients and controls, confirming a recent observation by Paik et al. (9).

Two previous studies (21, 34) have not been able to demonstrate an effect of GH on circulating tGhr in PWS. Hoybye (34) did not find an effect of GH on the fasting levels of tGhr in PWS adults. In a study of PWS children in an age range similar to our cohort, Haqq et al. (21) reported tGhr data that were compatible with a trend toward lower concentrations after GH, but the result did not reach significance. However, our children had a lower BMI SDS with a normal BMI in seven of 13 GH-untreated and 19 of 24 GH-treated children, which is in contrast to the finding of this feature in only two of 12 children in the study of Haqq et al. This suggests that, in PWS children with a higher BMI, GH-induced ghrelin suppression may be harder to achieve. Also, we observed that, with GH treatment, a strong and significant negative correlation of BMI SDS with ghrelin suppression by carbohydrates was abolished. This is consistent with the hypothesis that GH treatment alters the way in which BMI and carbohydrate-suppressed tGhr levels interact.

The mechanisms by which GH treatment decreases tGhr are unclear. Very low dose short-term (7 d) GH treatment of non-GH-deficient men with visceral obesity increased IGF-I and GH, but did not alter plasma ghrelin (35). In GH-deficient adults, short-term (8 d) administration of GH in a conventional replacement dose led to a significant decrease in ghrelin levels and an increase in circulating IGF-I. This is consistent with an inhibitory feedback of GH, IGF-I, or both on ghrelin in GH deficiency (36). After long-term GH treatment (6–12 months) in GH-deficient adults and children, ghrelin was found to be increased (36), unchanged (13, 37, 38), or decreased (39). In the latter study, the reduced ghrelin concentration was concurrent with a mean 27% decrease in fat mass. The lipolytic, glucose-promoting, and insulin-increasing activities of GH were assumed to contribute to the ghrelin decrease under the condition of GH deficiency.

One of the factors necessary for ghrelin reduction after a mixed meal is insulin. This is demonstrated by the fact that, in type 1 diabetes, a postprandial ghrelin decrease occurs only in the presence of insulin (40). PWS patients are unique in that they exhibit a diminished and delayed insulin secretion in response to carbohydrates and a lesser degree of insulin resistance than would be expected for the degree of their obesity (31). Looking at the time course of insulin resistance and the response of insulin and glucose to oral glucose administration during long-term GH treatment in PWS, l’Allemand et al. (41) described a transient increase in fasting insulin and insulin resistance over the first 3 yr of treatment that returned to normal after 36 months of GH therapy. Although overt insulin resistance develops in only a minority of PWS patients, it appears that, as long as fat mass is not reduced, GH is capable of inducing fasting insulin and insulin resistance. The majority of our GH-treated group was in that early phase of GH treatment: only five of 24 patients had been treated for more than 3 yr. However, fasting (postcarbohydrate) insulin concentrations and HOMA2IR did not differ between the GH-treated and GH-untreated group. Therefore, differences in the acute insulin response to carbohydrates between groups did not explain the observed tGhr suppression in the GH-treated state. Indeed, the explanation that, in PWS, GH-induced insulin resistance may play a role in modifying ghrelin concentration is compatible with some of our findings. Initially, in our untreated patients, an association of BMI with HOMA2 insulin resistance was lacking, and was only observed after the start of GH therapy. Furthermore, only in the GH-treated group did we find a moderate correlation of insulin resistance with the aGhr concentration 2 h after carbohydrate loading. This is consistent with the findings of Paik et al. (9), showing that the amount of aGhr suppression after carbohydrate loading increased with whole body insulin sensitivity in PWS. Also, when we tried to pinpoint factors influencing ghrelin concentrations by model-selection multiple stepwise regression analysis, HOMA2 insulin resistance was kept in the model, explaining some of the total and more of the aGhr variability.

An important observation in this study was that both total and aGhr levels decreased in response to carbohydrate loading, irrespective of whether the patients were treated with GH or not. In contrast, only tGhr levels decreased after GH treatment. To explain this, we propose that, in PWS, a chronically increased production of aGhr is accompanied by accelerated rate of deacylation. This would result in elevated tGhr concentrations in the face of normal levels of the acylated form. If this process was partly blocked by GH then tGhr levels would decrease. Among a number of ghrelin deacylation enzymes that have been identified [e.g. lysophospholipase I (42), esterases (43)], several are potential GH targets. In adult-onset GH deficiency patients, stable on long-term GH treatment, Gauna et al. (44) have indeed shown that administration of a small amount of aGhr increased aGhr levels. But, tGhr concentrations increased even more, and this resulted in an increased ratio of total to aGhr. Yet, omission of the evening GH dose, and its administration with aGhr the next morning did not blunt the rise of tGhr levels otherwise seen after treatment with the acylated form alone. However, this may not militate against the effect of GH we report with our patients, who were studied separately in a GH-naïve state and under chronic GH treatment.

Desacyl ghrelin is involved in many biological activities independent of the GH secretagogue receptor Ia. These include glucose homeostasis, lipolysis, cell proliferation and apoptosis (44–52). Ghrelin action on the hypothalamic level appears to be limited to aGhr acting via the GH secretagogue Ia receptor (53). Therefore, attempts to explain obesity in PWS by changes in desacyl ghrelin must address direct effects of this form on peripheral tissues. Until now, most studies have focused on short-term changes in ghrelin. These studies failed to establish a link between ghrelin levels and hunger or satiety in PWS. Acute somatostatin infusion lowered ghrelin concentrations in PWS adults, but did not affect appetite (32). Perhaps it is rather the long-term fluctuations in blood ghrelin levels corresponding to the chronic feeding state of the organism that affect weight in PWS.

The hypothesis of acylated or desacyl ghrelin playing a major role for obesity in PWS could be tested by a new class of pharmacological agents, the spiegelmers. These are L-enantiomer oligonucleotides, that can be constructed to specifically bind and inactivate aGhr. In diet-induced obese mice, administration of the spiegelmer NOX-B11 for 1 wk decreased body weight and body fat (54), an effect not seen in obese ghrelin-deficient animals. This supports a role of the acylated form of ghrelin for obesity, at least in this species. Further studies of the role of ghrelins in PWS-related obesity will also have to address the possibility that ghrelins and obestatin, another peptide encoded by the ghrelin gene, but with opposing effects on weight regulation (55), are differentially regulated by glucose or GH.

A limitation of our study is the high interassay variability of the aGhr measurements of 20.4%, exceeding most of the published data that vary less than 13% (9, 44). This could weaken the interpretation of our aGhr data.

In conclusion, our results showed that aGhr and tGhr levels are differentially regulated by GH in PWS children and adolescents. Chronic GH treatment decreased basal and postcarbohydrate plasma tGhr concentrations below those of GH-untreated PWS children, whereas aGhr concentrations remained unaffected. In contrast, both aGhr and tGhr concentrations were decreased by oral carbohydrate administration, independent of the GH treatment status. In intraindividual comparisons, the GH-induced tGhr decrease was associated with a decrease of BMI SDS. Changes in insulin resistance appeared to play only a minor role for tGhr decrease.

Acknowledgments

We thank H. Tourné for his technical assistance with the ghrelin measurements and N. Lehmann (Institute for Medical Informatics, Biometry and Epidemiology, University of Essen, Essen, Germany) for helpful discussions regarding statistical issues. We also thank R. D. Oades in the Department of Child and Adolescent Psychiatry, University of Essen, for a helpful review of the text.

Part of this work was presented at the 87th Annual Meeting of The Endocrine Society, San Diego, CA, June 4–7, 2005.

Disclosure Summary: B.P.H. serves on the Advisory Board of the German Pfizer Growth Database and has received lecture fees from Pfizer and Novo Nordisk. K.H., I.M.R., and N.U. have nothing to declare. K.M. has received consulting and lecture fees from Pfizer and Novo Nordisk. S.P. has received lecture fees from Novo Nordisk and Eli Lilly.

Abbreviations:

• aGhr,

  Acylated ghrelin;

• BMI,

  body mass index;

• HOMA2,

  homeostasis model assessment 2;

• HOMA2B,

  homeostasis model assessment 2 β-cell function;

• HOMA2IR,

  homeostasis model assessment 2 insulin resistance;

• oGTT,

  oral glucose tolerance test;

• PWS,

  Prader-Willi syndrome;

• SDS,

  sd score;

• tGhr,

  total ghrelin.