Growth Hormone (GH) Treatment Decreases Plasma Kisspeptin Levels in GH-Deficient Adults with Prader–Willi Syndrome

Obesity and growth hormone (GH)-deficiency are consistent features of Prader–Willi syndrome (PWS). Centrally, kisspeptin is involved in regulating reproductive function and can stimulate hypothalamic hormones such as GH. Peripherally, kisspeptin signaling influences energy and metabolic status. We evaluated the effect of 12-month GH treatment on plasma kisspeptin levels in 27 GH-deficient adult PWS patients and analyzed its relationship with metabolic and anthropometric changes. Twenty-seven matched obese subjects and 22 healthy subjects were also studied. Before treatment, plasma kisspeptin concentrations in PWS and obese subjects were similar (140.20 (23.5–156.8) pg/mL vs. 141.96 (113.9–165.6) pg/mL, respectively, p = 0.979)) and higher (p = 0.019) than in healthy subjects (124.58 (107.3–139.0) pg/mL); plasma leptin concentrations were similar in PWS and obese subjects (48.15 (28.80–67.10) ng/mL vs. 33.10 (20.50–67.30) ng/mL, respectively, p = 0.152) and higher (p < 0.001) than in healthy subjects (14.80 (11.37–67.30) ng/mL). After GH therapy, lean body mass increased 2.1% (p = 0.03), total fat mass decreased 1.6% (p = 0.005), and plasma kisspeptin decreased to levels observed in normal-weight subjects (125.1(106.2–153.4) pg/mL, p = 0.027). BMI and leptin levels remained unchanged. In conclusion, 12-month GH therapy improved body composition and decreased plasma kisspeptin in GH deficient adults with PWS. All data are expressed in median (interquartile range).


Introduction
Kisspeptin is a hormone that promotes the onset of puberty by stimulating the secretion of gonadotropin-stimulating hormone (GnRH). Kisspeptin can also stimulate the release of other pituitary hormones such as prolactin, growth hormone (GH), oxytocin, and vasopressin [1].
Kisspeptin is mainly synthesized by neurons in the arcuate nucleus of the hypothalamus. Its synthesis is modulated by energy balance, decreasing in situations of insufficient weight (e.g., anorexia nervosa) or excess weight (e.g., obesity) [2]. Low kisspeptin levels are considered the cause of the hypogonadotropic hypogonadism seen in some patients with metabolic syndrome and obesity [2,3].
Kisspeptin-producing neurons express receptors for leptin, a hormone produced in adipose tissue that informs these neurons about the status of energy reserves [4,5]. nar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.  recorded subjects' height, determined by a stadiometer (Harpenden, Holtain Ltd., Dyfed, UK); body weight, measured to the nearest 0.1 kg with standard equipment; body mass index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.

Methods
Blood was extracted from all participants at 8 AM after overnight fasting. Plasma samples for kisspeptin and leptin measurements were kept at −80 °C until analysis. We recorded subjects' height, determined by a stadiometer (Harpenden, Holtain Ltd., Dyfed, UK); body weight, measured to the nearest 0.1 kg with standard equipment; body mass index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.

Methods
Blood was extracted from all participants at 8 AM after overnight fasting. Plasma samples for kisspeptin and leptin measurements were kept at −80 °C until analysis. We recorded subjects' height, determined by a stadiometer (Harpenden, Holtain Ltd., Dyfed, UK); body weight, measured to the nearest 0.1 kg with standard equipment; body mass index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.

PWS Patients (n = 27) Obese Subjects (n = 27) Healthy Subjects (n = 22) p-Value
Glucose (mmol/L) 5.00 (4.55-6.99) 4.83 (4.77-5.49) 4.36 (4.16-5.02) * 0.006 weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.  Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.  was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. uous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age. Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. routine automated analyzer was used for other laboratory tests. The homeostatic model assessment for insulin resistance (HOMA-IR) index was calculated as fasting plasma glucose (mmol/L) x fasting insulin (µIU/mL))/22.5 [20]. Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. intraassay coefficient of variation (CV) <10%; interassay CV < 12%). To determine concentrations of leptin in plasma, we used the Human Leptin ELISA Kit (Biorbyt, Cambridge, UK) (lower limit of detection, 10 pg/ml; intraassay CV < 7.6%; interassay CV < 8.4%). A routine automated analyzer was used for other laboratory tests. The homeostatic model assessment for insulin resistance (HOMA-IR) index was calculated as fasting plasma glucose (mmol/L) x fasting insulin (µIU/mL))/22.5 [20]. Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA). Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. Blood was extracted from all participants at 8 AM after overnight fasting. Plasma samples for kisspeptin and leptin measurements were kept at −80 °C until analysis. We recorded subjects' height, determined by a stadiometer (Harpenden, Holtain Ltd., Dyfed, UK); body weight, measured to the nearest 0.1 kg with standard equipment; body mass index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group.

Methods
Blood was extracted from all participants at 8 AM after overnight fasting. Plasma samples for kisspeptin and leptin measurements were kept at −80 °C until analysis. We recorded subjects' height, determined by a stadiometer (Harpenden, Holtain Ltd., Dyfed, UK); body weight, measured to the nearest 0.1 kg with standard equipment; body mass index (BMI); and body composition determined by dual-energy x-ray absorptiometry (Lunar Prodigy−963, Chicago, IL, USA).
Patients with PWS were treated with recombinant GH (Genotonorm Miniquick ® , Pfizer, New York, NY, USA), starting with a dose of 0.2mg/day and adjusting the dose at 1, 3, 6, and 12 months to achieve high-normal insulin-like growth factor-1 (IGF-1) levels for the patient's age.
Only PWS patients received GH treatment. After 12 months' GH treatment, patients' analytic and anthropometric parameters were measured again with the same protocol.

Statistical Analyses
Continuous variables are reported as medians and interquartile ranges (IQR). Categorical variables are reported as frequencies and percentages. To compare all the continuous variables at baseline, between the 3 groups, we used Kruskal-Wallis test followed by the Mann-Whitney U test with Bonferroni correction. To compare all variables before and after GH treatment we used the Wilcoxon signed-rank test. To study the relationship between variables, we used Spearman's rank-order correlation test. Statistical significance was fixed at p < 0.05. All analyses were done with IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA).

Baseline Findings
3.1.1. Patient Characteristics: Table 1 reports the baseline characteristics of subjects in the PWS, obese, and normal weight groups. The PWS and obese groups did not differ in weight, BMI, waist, percentages of body fat and lean mass, glucose, HOMA-IR, follicle-stimulating hormone, or testosterone (only measured in men) levels. Females in the PWS group had lower levels of estradiol and luteinizing hormone than females in the obese group. .
In PWS subjects, baseline kisspeptin levels were similar to those in the obese group and higher than those in the healthy group (Table 1, Figure 1). There were no differences between diabetic and non-diabetic subjects (141.1 pg/mL (121.1-152.6) vs. 140.2 pg/mL (128.2-163.6), respectively, p = 0.798). Patients treated with sex steroids had higher levels of kisspeptin (149.4 pg/mL (138.3-165.3) vs. 133.6 pg/mL (117.8-142.2) than those not treated with sex steroids, p = 0.01), but these two groups did not differ in terms of BMI (p = 0.981), testosterone levels in men, or estradiol levels in women (data not shown). PWS = Prader-Willi syndrome; BMI = body mass index; HOMAR-IR = homeostatic model assessment for insul resistance; LH = luteinizing hormone; FSH = follicle-stimulating hormone. p-values are marked with * Kruskall-Wallis te for comparisons of PWS vs. Obese vs. Healthy. Mann-Whitney U test with Bonferroni correction for the followin comparisons: ⴕ PWS vs. Obese, λ PWS vs. Healthy, and ** Obese vs. Healthy. All variables except for sex are reported median (interquartile range).
In PWS subjects, baseline kisspeptin levels were similar to those in the obese and higher than those in the healthy group (Table 1, Figure 1). There were no diffe between diabetic and non-diabetic subjects (141.1 pg/mL (121.1-152.6) vs. 140.2 (128.2-163.6), respectively, p = 0.798). Patients treated with sex steroids had higher of kisspeptin (149.4 pg/mL (138.3-165.3) vs. 133.6 pg/mL (117.8-142.2) than tho treated with sex steroids, p = 0.01), but these two groups did not differ in terms of = 0.981), testosterone levels in men, or estradiol levels in women (data not shown).
In PWS subjects, baseline leptin levels were similar to those in the obese grou higher than those in the healthy group (Table 1, Figure 2). Leptin levels did not between diabetic and non-diabetic subjects (45.30 ng/mL (19.60-61.80) vs. 51.0 (29.50-82.40), respectively, p = 0.866) or between those treated with sex steroids and
In PWS subjects, baseline leptin levels were similar to those in the obese group and higher than those in the healthy group (Table 1, Figure 2). Leptin levels did not differ between diabetic and non-diabetic subjects (45.
By contrast, leptin levels were not significantly different after treatment (Table 2, Figure 2) and no significant changes in bone mineral density were observed; Z-scores remained lower than expected for age ( Table 2).   By contrast, leptin levels were not significantly different after treatment (Table 2, Figure 2) and no significant changes in bone mineral density were observed; Z-scores remained lower than expected for age (Table 2).

Correlations
Overall, baseline kisspeptin did not correlate significantly with weight, BMI, percentage of body fat or total body fat, percentage of lean mass or total lean mass, waist circumference, glucose, insulin, HOMA-IR, glycated hemoglobin, gonadotropins, or sex steroids. Baseline kisspeptin and leptin levels correlated only in the healthy group (r = 0.414, p = 0.05).
In the PWS group, baseline kisspeptin did not correlate significantly with any of the above variables or with IGF-1, bone mineral density (total femur Z-score, total spine Z-score), extremities/trunk body fat index, or appendicular skeletal muscle mass index. Moreover, the percentage of change in kisspeptin did not correlate with the percentage of change in body fat (whether measured in % or kg), lean body mass (whether measured in % or kg), HOMA-IR, insulin, glucose, or IGF-1.
In the PWS group, baseline leptin levels correlated with HOMA-IR (r = 0.317, p = 0.006), but after GH treatment, leptin levels did not correlate with HOMA-IR (r = 0.110, p = 0.602). Baseline leptin levels also correlated with percentage of body fat (r = 0.705, p < 0.001) and after treatment, with percentage of body fat (r = 0.692, p < 0.001), as well as with total body fat (r = 0.498, p = 0.01).
The percentage of change in IGF-1 did not correlate with the percentage of change in total body fat or in lean body mass.

Discussion
To our knowledge, this is the first study to evaluate plasma kisspeptin levels in PWS patients. We found that plasma kisspeptin levels in PWS patients with GH deficiency were similar to those in obese subjects matched by age, sex, and BMI and were higher than those in healthy controls. After 12 months of treatment with GH, plasma kisspeptin levels in PWS patients decreased to levels similar to those observed in healthy subjects.
The relationship between obesity and kisspeptin levels is unclear. Hestiantoro et al. [21] found kisspeptin levels were lower in obese than in normal-weight menopausal women. Pita et al. [22] found kisspeptin levels were higher in obese than in normal-weight prepubertal girls, but similar in obese and normal-weight prepubertal boys. Sitticharoon et al. [23] observed higher kisspeptin levels in obese than in non-obese men. In other studies, this group found no differences in kisspeptin levels between obese and normalweight women [24] or between obese and normal-weight girls with central precocious puberty [25]. Taken together, these findings suggest that kisspeptin levels may be influenced more by hormonal status than by obesity per se. Moreover, kisspeptin levels in women differ across the menstrual cycle, being highest in the luteal phase, followed by the preovulation phase and the follicular phase, suggesting that kisspeptin might play a role in peripheral reproductive regulation [26]. In the present study, kisspeptin levels were higher in obese patients (with or without PWS), but did not differ between sexes. However, all our PWS patients had some degree of hypogonadism, and not all of them were being treated with sex hormones. Furthermore, we did not take women's menstrual phase into account because most of these patients do not ovulate. Nevertheless, kisspeptin levels were higher in those with hormone replacement therapy than in those without, suggesting sex hormones might affect peripheral kisspeptin. These findings are in line with those reported in vitro experiments where estrogens stimulated Kiss-1 gene expression in cultured kisspeptin neurons, suggesting a central interaction between sex hormones and kisspeptin-producing cells [27].
The relationship between kisspeptin and BMI is also unclear. Some studies have found positive correlations. One study in men found that kisspeptin levels correlated positively with BMI and weight [23], and another in children found they correlated positively with BMI, weight, and waist circumference [28]. However, other studies have found negative correlations between kisspeptin levels and BMI and waist circumference in non-diabetic men and women [29] and between kisspeptin and BMI in anorectic women [30]. In the present study, we found no correlations between kisspeptin and BMI or waist circumference in the whole group or in the PWS group. Discrepancies between studies might be due to the inaccuracy of BMI and waist circumference as measures of body fat and/or to the heterogeneity and the small size of the cohorts studied.
In the PWS group, GH treatment resulted in a decrease in total body fat and an increase in lean body mass, as reported by other authors [31,32]. Because adipocytes appear to be a source of circulating kisspeptin, we expected decreases in body fat mass to result in decreases in kisspeptin levels. Moreover, adipocytes express the kisspeptin-1 receptor (KISS1R), indicating that kisspeptin secreted by adipose tissue could act as an adipokine or as autocrine/paracrine regulator of adipocyte function [33]. However, we found no correlations between the change in kisspeptin and the change in fat mass, lean mass, extremities/trunk body fat index, or appendicular lean mass index after treatment, suggesting there is no clear relationship between kisspeptin and body composition parameters. Furthermore, kisspeptin appears to regulate GH, but the role of kisspeptin in GH release remains unclear, especially due to discrepancies between in vivo and in vitro findings [1], although the data suggest that kiss1/KISS1R could play a role in a short or ultra-short feedback loop that regulates the function of somatotrophs [1]. Such a role would help explain why kisspeptin levels decreased in GH-deficient patients with PWS after one year of GH therapy. Nevertheless, we found no correlation between the changes in kisspeptin and IGF-1 levels. Further studies are necessary to test this hypothesis.
Leptin is mainly produced by adipose tissue [34], and its concentration in serum correlates with body energy reserves [35]. Leptin may signal peripheral energy status to the hypothalamus, thus influencing food intake and reproductive function [36]. Kisspeptin neurons in the hypothalamus have leptin receptors, and leptin increases Kiss1 mRNA expression in those neurons, thus supporting the existence of a leptin-kisspeptin-GnRH pathway in which kisspeptin would mediate between energy reserves and the maturation of the hypothalamic-pituitary-gonadal axis [2].
In our study, plasma leptin levels correlated with body fat mass in all groups at baseline, and no significant changes in leptin levels occurred in the PWS group after GH treatment. These findings are in line with those reported by Höybe et al. [37] in 2003 in a study with 17 adult patients with PWS, but discrepant with those reported by Myers et al. [38], who reported decreased leptin levels after GH treatment in children with PWS. Moreover, we found no correlation between plasma leptin and kisspeptin levels, indicating that these two peptides could be less interrelated peripherally than in the central nervous system.
Inactivation of kiss1r in mice results in obesity and diabetic phenotype [39], and kisspeptin and its receptor are expressed in metabolic tissues (e.g., fat, liver and pancreatic tissues) and likely plays a role in regulating insulin secretion. [6]. However, kisspeptin stimulates insulin secretion only when glucose levels are elevated, suggesting kisspeptin's role involves correcting hyperglycemia [40]. Glucagon secreted from pancreatic α-cells provokes kiss1 expression in the liver, increasing secretion of kisspeptin from the liver and thereby suppressing glucose-stimulated insulin secretion from pancreatic β-cells [33].
For all these reasons, we also analyzed the relationship between kisspeptin and glucose metabolism in our cohort of patients with PWS, 6 (22.2%) of whom had diabetes with good glycemic control (HbA1c < 7.5%). We found no difference in kisspeptin levels between diabetic and non-diabetic PWS patients, and kisspeptin levels did not correlate with plasma glucose, insulin, glycated hemoglobin, or HOMA-IR index. After GH treatment, no changes in these parameters or correlations between these parameters and kisspeptin were observed. These results corroborate those of a previous study that showed no effect of GH treatment on glycemic control in adults with PWS [32].