The Effect of Leptin Replacement on Parathyroid Hormone, RANKL-Osteoprotegerin Axis, and Wnt Inhibitors in Young Women With Hypothalamic Amenorrhea

Abstract

Context:

Recombinant leptin (metreleptin) treatment restores bone mineral density in women with hypothalamic amenorrhea (HA), a condition characterized by hypoleptinemia, which has adverse impact on bone health.

Objective:

The objective of the study was to investigate how metreleptin exerts its positive effect on bone metabolism in humans.

Design:

This was a randomized, double-blinded, placebo-controlled study.

Setting:

The study was conducted at Beth Israel Deaconess Medical Center (Boston, Massachusetts).

Patients and Interventions:

Women (n = 18) with HA and hypoleptinemia for at least 6 months were randomized to receive either metreleptin or placebo for 36 weeks. Serum samples were obtained at baseline and 12, 24, and 36 weeks of treatment.

Main Outcome Measures:

Circulating levels of leptin, intact PTH (iPTH), receptor activator of nuclear factor-κB ligand (RANKL), osteoprotegerin (OPG), sclerostin, dickkopf-1, and fibroblast growth factor-23.

Results:

Metreleptin administration significantly increased leptin levels throughout the treatment period (P = .001). iPTH decreased over the 36 weeks of treatment (P = .01). There was a trend toward a decrease in serum RANKL and increase in serum OPG in the metreleptin-treated group. The RANKL to OPG ratio was significantly decreased within the metreleptin (P = .04) but not the placebo group. Metreleptin had no effect on serum sclerostin, dickkopf-1, and fibroblast growth factor-23.

Conclusions:

Metreleptin treatment over 36 weeks decreases iPTH and RANKL to OPG ratio levels in hypoleptinemic women with HA.

Hypothalamic amenorrhea (HA) occurs in the context of chronically low energy availability either from excessive exercise and/or weight loss or psychological stress (1). The dysfunction of the hypothalamic-pituitary-gonadal axis leads to hypoestrogenemia (2), which unfavorably affects bone health (3). Low IGF-1 levels, relative hypercortisolemia, low protein intake, and secondary hyperparathyroidism (due to low calcium and/or vitamin D intake) may contribute to estrogen-independent bone loss (4). Furthermore, bone formation markers decrease and resorption markers increase during weight loss independently of menstrual status, eg, levels of estrogen (5). Consequently, bone mass accretion is compromised in these individuals, resulting in a lower peak bone mass with a considerable impact on future bone health and fracture risk (6). The limited treatment options available for the optimization of bone health in women with HA highlight an unmet need (7).

Circulating leptin levels reflect the availability of energy stores, and hypoleptinemia, a state evident in HA (8), may mediate the adverse effect of HA on the skeleton. Long-term (up to 2 y) administration of human recombinant N-methionyl-leptin (meterleptin) in patients with HA significantly increased bone mineral content and bone mineral density (BMD) at the lumbar spine (9). Although these promising results suggest a novel therapeutic potential of leptin replacement on bone health in HA, its mechanistic role on bone metabolism remains to be elucidated.

During the last decade, the identification of several new molecules involved in bone metabolism regulation has shed light in our understanding of the pathophysiology of bone diseases. These include the receptor activator of nuclear factor-κB ligand (RANKL) and its decoy receptor, osteoprotegerin (OPG), the inhibitors of the canonical Wingless (Wnt)/β-catenin pathway sclerostin and dickkopf-1 (DKK-1), and the osteocyte-derived fibroblast growth factor-23 (FGF23). In this randomized, double-blinded, placebo-controlled study, we aimed to investigate the effect of meterleptin treatment on PTH, the RANKL-osteoprotegerin axis, sclerostin, DKK-1, and FGF23 in patients with HA, thereby investigating potential mechanisms by which meterleptin could exert its beneficial effect on bone metabolism in chronic energy deprived, hypoleptinemic HA patients.

Materials and Methods

Clinical study

Apparently healthy HA women (n = 18) between the ages of 18 and 35 years, with low body weight and hypoleptinemia, were recruited through advertisements in the community in this randomized, double-blinded, placebo-controlled study. Fasting morning leptin levels at screening were less than 5 ng/mL. All subjects had stable body weight for at least 6 months before entering the study and were within 15% of their ideal body weight at the time of screening. Exclusion criteria were as follows: 1) coexisting medical conditions, including active eating disorders (screened for on the basis of questionnaires); 2) medications known to affect bone metabolism; 3) secondary amenorrhea due to hyperprolactinemia, hypothyroidism, Cushing's syndrome, congenital adrenal hyperplasia, polycystic ovarian syndrome, or primary ovarian failure; and 4) a positive pregnancy test at baseline. Pregnancy test results were performed at each follow-up visit and were negative throughout the study (9, 10).

Subjects were randomized to meterleptin (n = 10) or placebo (n = 8), administered sc daily between 7:00 pm and 11:00 pm for up to 36 weeks. All subjects began meterleptin administration at a dose of 0.08 mg/kg for 12 weeks. If menstruation occurred during this time, subjects continued on the 0.08-mg/kg dose. Otherwise, the dose was increased to 0.12 mg/kg. Placebo and meterleptin were both provided by Amylin Pharmaceuticals.

At baseline, a detailed history was obtained, a physical examination was performed, body mass index (BMI) was calculated and total fat and total BMD, as well as BMD at the lumbar spine (LS), and the nondominant hip and radius were measured. Morning fasting blood samples were obtained from all subjects at baseline and at 12, 24, and 36 weeks for the measurement of serum total calcium, phosphate, albumin, creatinine, intact PTH (iPTH), 25-hydroxyvitamin D, OPG, RANKL, sclerostin, DKK-1, FGF23, LH, FSH, estradiol (E2), and leptin levels. Corrected calcium was calculated by the following formula: corrected calcium (milligrams per deciliter) = Ca (milligrams per deciliter) + 0.8 × [4.0 − albumin (grams per deciliter)]. BMD was measured by dual-energy x-ray absorptiometry at baseline using a Discovery 4500A Densitometer (Hologic). The study protocol was approved by the Institutional Review Board at Beth Israel Deaconess Medical Center, and written informed consent was obtained from all subjects.

Assays

Biochemical analysis for serum total calcium, phosphate, albumin, and creatinine was assayed on the Roche Cobas c311 (Roche Diagnostics). iPTH was assayed with the Immulite 1000 immunoassay analyzer (Siemens Healthcare Diagnostics). RIAs were used for the measurement of 25-hydroxyvitamin D [DiaSorin; sensitivity 1.5 ng/mL; intraassay coefficient of variation (CV) 8.6%–12.5% interassay CV 8.2%–11.0%] and leptin (Millipore; sensitivity of 0.437 ng/mL, intraassay CV 3.4%–8.3%, interassay CV 3.0%–6.2%). An ELISA was used for the measurement of OPG (R&D Systems; sensitivity 62.5–4000 pg/mL), RANKL (Biovendor; sensitivity 0.4 pmol/L, intraassay CV 7.3%–11.5%, interassay CV 11.2%–12.8%), sclerostin (Biomedica; sensitivity 3.2 pmol/L, intraassay CV 5%–7%, interassay CV 3%–10%), DKK-1 (Biomedica; sensitivity 1.7 pmol/L intraassay CV: 3% interassay CV: 5%), FGF23 (Immutopics, Inc; sensitivity 1.5 RU/mL, interassay CV 2.4%–4.7%, intraassay CV 1.4%–2.4%), LH (DSLabs, Diagnostic Systems Laboratories, Inc; sensitivity 0.10 mIU/mL, intraassay CV 5.6%–6.8%, interassay CV 5.3%–7.8%), FSH (ALPCO; sensitivity 2.5–200 mIU/mL, intraassay CV 3.5%–5.8%, interassay CV 3.4%–7.7%), E2 (ALPCO; sensitivity 3–200 pg/mL, intraassay CV 5.5%–7.9%, interassay CV 6.8%–8.8%). All serum samples were stored at −80°C until analysis. All samples were analyzed in duplicate.

Statistical analysis

Data for continuous variables are presented as mean ± SEM. Data for categorical variables are presented as numbers and/or percentages. Kolmogorov-Smirnov test was used to check the normality of distributions of the continuous variables. A Mann-Whitney test was used for between group comparisons, in case of 2 groups of continuous variables. A χ² test or Fischer's exact test was used for between-group comparisons, in case of categorical variables. A repeated-measures, one-way ANOVA or a Friedman's test was used for within-group comparisons. In case of a statistically significant difference in an ANOVA or Friedman's test, a Bonferroni post hoc adjustment was used for multiple pairwise comparisons. A repeated-measures, one-way analysis of covariance (ANCOVA) was used to adjust within-group comparisons for potential cofounders. A logarithmic transformation was performed for the needs of the ANCOVA, when necessary. A repeated-measures multivariate ANOVA was used to check for between-variables interactions over time. The Spearman's coefficient (r_s) was used for bivariate correlations. The partial coefficient (r_p) was used for binary correlations adjusted for group. A two-sided value of P < .05 was considered statistically significant in all the above-mentioned tests. The statistical analysis was performed by SPSS version 21.0 for Macintosh (IBM Corp).

Results

Comparative baseline characteristics of both groups are presented in Table 1. There were no statistically significant differences in age, BMI, total fat, BMD, E2, LH, and FSH.

Table 1.

Comparative Baseline Data of the Study Groups

Variable	Metreleptin	Placebo	P Value for Between Groupsa
Age, y	26.1 ± 1.4	25.1 ± 1.3	.75
BMI, kg/m2	21.1 ± 0.6	20.1 ± 0.7	.21
Total fat, kg	13.8 ± 0.9	11.8 ± 1.0	.21
Total BMD, g/cm2	1.10 ± 0.03	1.16 ± 0.03	.09
LS BMD, g/cm2	0.92 ± 0.04	0.97 ± 0.03	.22
Total hip BMD, g/cm2	0.91 ± 0.03	0.93 ± 0.04	.79
FN BMD, g/cm2	0.82 ± 0.03	0.83 ± 0.04	.76
Radius BMD, g/cm2	0.57 ± 0.02	0.54 ± 0.01	.37
Es2, pg/mL	23.3 ± 9.6	12.9 ± 1.5	.48
LH, mIU/mL	8.8 ± 3.3	12.8 ± 5.5	.47
FSH, mIU/mL	5.3 ± 0.5	4.6 ± 0.5	.21

Data are presented as mean ± SEM.

^aMann-Whitney test.

Comparative biochemical data over time for both groups are presented in Table 2. As expected, in the meterleptin-treated group, but not in the control group, the serum leptin levels significantly increased at week 12 and remained increased up to week 36.

Table 2.

Comparative Biochemical Data of Both Groups at Baseline and Over Time

Variable	Group	Baseline	Week 12	Week 24	Week 36	P Value Within Groupsa
Leptin, ng/mL	Metreleptin	4.4 ± 0.8	23.8 ± 5.1b,c	30.1 ± 6.7b,c	28.0 ± 6.3b,c	.001
	Placebo	3.4 ± 0.6	3.2 ± 0.7	8.0 ± 2.9	4.8 ± 1.8	.23
RANKL, pmol/L	Metreleptin	303 ± 55	239 ± 36	268 ± 39	242 ± 25	.18
	Placebo	248 ± 34	236 ± 23	233 ± 24	251 ± 29	.69
OPG, pmol/L	Metreleptin	51.1 ± 3.6	53.2 ± 4.0	56.6 ± 3.8	56.3 ± 2.6	.23
	Placebo	49.8 ± 4.6	48.6 ± 7.2	48.7 ± 5.9	53.6 ± 4.1	.61
RANKL to OPG ratio	Metreleptin	6.6 ± 1.3	4.9 ± 0.9	5.1 ± 0.8	4.5 ± 0.6	.04
	Placebo	5.2 ± 0.9	5.6 ± 1.0	5.3 ± 0.9	4.8 ± 0.5	.82
DKK-1, ng/mL	Metreleptin	19.3 ± 2.8	20.5 ± 2.6	24.9 ± 2.2c	18.8 ± 2.3	.34
	Placebo	14.6 ± 2.9	15.8 ± 2.2	17.0 ± 2.9	16.2 ± 2.2	.76
Sclerostin, ng/mL	Metreleptin	0.433 ± 0.049	0.333 ± 0.031	0.355 ± 0.043	0.450 ± 0.059	.07
	Placebo	0.396 ± 0.052	0.352 ± 0.028	0.308 ± 0.038	0.432 ± 0.065	.23
FGF23, U/mL	Metreleptin	56.4 ± 12.7	64.1 ± 11.8	45.6 ± 8.6	42.1 ± 7.8	.32
	Placebo	54.8 ± 9.5	61.3 ± 15.5	55.9 ± 11.0	75.8 ± 17.3	.55
iPTH, pg/mL	Metreleptin	41.8 ± 6.3	42.8 ± 4.5	34.4 ± 3.6	26.7 ± 2.6b,d	.01
	Placebo	29.1 ± 3.1	32.2 ± 2.7	35.7 ± 5.4	30.4 ± 3.6	.59
25-Hydroxyitamin D, μg/dL	Metreleptin	20.9 ± 2.7	25.7 ± 2.4	24.0 ± 2.8	20.8 ± 2.4	.40
	Placebo	26.4 ± 3.6	23.1 ± 3.5	21.0 ± 3.3	24.2 ± 2.3	.35
Corrected calcium, mg/dL	Metreleptin	7.8 ± 0.4	8.8 ± 0.2	8.6 ± 0.1	8.3 ± 0.3	.12
	Placebo	8.4 ± 0.9	9.0 ± 1.0	7.6 ± 0.7	7.9 ± 0.7	.73
Phosphate, mg/dL	Metreleptin	4.2 ± 0.1	4.4 ± 0.2	4.3 ± 0.2	4.5 ± 0.2	.69
	Placebo	4.3 ± 0.2	4.0 ± 0.1	4.2 ± 0.2	4.1 ± 0.1	.49
Creatinine, mg/dL	Metreleptin	0.84 ± 0.03	0.79 ± 0.04	0.71 ± 0.06	0.80 ± 0.03	.13
	Placebo	0.86 ± 0.05	0.93 ± 0.10	0.74 ± 0.05	0.81 ± 0.02	.13

Data are presented as mean ± SEM.

^aWithin-group comparisons are repeated-measures ANOVA or Friedman's test; pairwise comparisons are Bonferroni post hoc test.

^bP < .05 compared with baseline.

^cP < .05 for between-group comparisons (Mann-Whitney test).

^dP < .05 compared with week 12.

^eP < .05 compared with week 24.

Baseline iPTH levels were not significantly different between groups (P = .093)

iPTH was significantly and progressively decreased within the meterleptin but not the placebo group. The between-group interaction over time for iPTH remained significant after adjustment for BMI (P = .02); BMI and age (P = .02); BMI, age, and total BMD (P = .01); or age, BMI, total BMD, and log(estradiol) (P = .01). The significance remained when BMI was replaced by total fat, and/or total BMD was replaced by LS, total hip, femoral neck (FN), or radius BMD in the above models or repeated-measures ANCOVA.

Serum RANKL and OPG did not change significantly within either group; however, there was a nonsignificant trend toward a decrease in serum RANKL and an increase in serum OPG in meterleptin-treated group. As a consequence, the RANKL to OPG ratio was significantly and progressively decreased within the meterleptin but not the placebo group. However, the between-group interaction over time did not remain significant for the RANKL to OPG ratio after adjustment for BMI (P = .23); BMI and age (P = .17); BMI, age, and total BMD (P = .24); or BMI, age, total BMD, and log(estradiol) (P = .40). These results remained essentially unchanged when BMI was replaced by total fat, or total BMD was replaced by LS, total hip, FN, or radius BMD in the above models or repeated-measures ANCOVA. There was no interaction between PTH and the RANKL to OPG ratio over time (P = .40).

Serum sclerostin did not significantly change within either group; however, there was a nonsignificant trend toward a decrease in serum sclerostin during weeks 12 and 24 (U shaped) only in the meterleptin group. The between-group interaction over time for sclerostin remained nonsignificant after adjustment for BMI (P = .27); BMI and age (P = .79); BMI, age, and total BMD (P = .68); or BMI, age, total BMD, and log(estradiol) (P = .61). These results remained unchanged when BMI was replaced by total fat, or total BMD was replaced by LS, total hip, FN, or radius BMD in the above models or repeated-measures ANCOVA.

There was no pattern over time in the serum DKK-1, FGF23, calcium, phosphate, 25-hydroxyvitamin D, creatinine, and albumin levels in either group (Table 2). DKK-1 was different between groups only at week 24 (P = .043) but not at week 12 (P = .20) or week 36 (P = .42).

Bivariate correlations between the study's parameters at baseline are presented in Table 3. Serum leptin levels at baseline were not correlated with BMD at any site. iPTH was correlated with serum RANKL (r_s = 0.52; P = .03) but not OPG or their ratio. Partial correlations (adjustment for group) between δ (D) values (D: 36 month − baseline) are presented in Table 4. Interestingly, there was a nonstatistical positive trend between D(PTH) and D(FGF23) (r_p = 0.44; P = .08) but not between D(PTH) and D(corrected calcium) and between D(sclerostin) and D(RANKL) (r_p = 0.41; P = .10) and between D(sclerostin) and D(RANKL to OPG ratio) (r_p = 0.43; P = .08). There is also a nonsignificant inverse trend between D(PTH) and D(OPG) (r_p = −0.43; P = .08).

Table 3.

Bivariate Correlations at Baseline Between Circulating Study's Parameters in the Sum of Patients With Hypothalamic Amenorrhea

Variable	iPTH, pg/mL	RANKL, pmol/L	OPG, pmol/L	RANKL to OPG Ratio	Sclerostin, ng/mL	DKK-1, ng/mL	FGF23, U/mL	25OH VitD, μg/dL
Leptin, ng/mL	0.09 (0.73)	−0.12 (0.63)	0.12 (0.63)	−0.19 (0.44)	−0.43 (0.07)	0.06 (0.83)	0.02 (0.95)	0.23 (0.36)
iPTH, pg/mL		0.52 (0.03)	0.02 (0.93)	0.40 (0.10)	−0.27 (0.28)	−0.22 (0.38)	0.38 (0.12)	−0.07 (0.77)
RANKL, pmol/L			−0.46 (0.05)	0.94 (<0.001)	0.01 (0.97)	−0.27 (0.28)	0.26 (0.29)	−0.14 (0.58)
OPG, pmol/L				−0.71 (0.001)	−0.40 (0.10)	0.26 (0.30)	−0.38 (0.12)	−0.01 (0.98)
RANKL to OPG ratio					0.23 (0.36)	−0.27 (0.28)	0.39 (0.11)	−0.16 (0.53)
Sclerostin, ng/mL						0.19 (0.44)	0.22 (0.37)	−0.32 (0.19)
DKK-1, ng/mL							−0.11 (0.68)	−0.10 (0.69)
FGF23, U/mL								0.15 (0.56)

Data are presented as Spearman's coefficient of correlation; r_s (P value).

Table 4.

Partial Correlations (Adjustment for Group) Between δ-Values (D: 36 Month − Baseline) of Circulating Study's Parameters in Patients With Hypothalamic Amenorrhea

Variable	D(iPTH), pg/mL	D(RANKL), pmol/L	D(OPG), pmol/L	D(RANKL to OPG) Ratio	D(Sclerostin), ng/mL	D(DKK-1), ng/mL	D(FGF23), U/mL	D(25OH VitD), μg/dL
Leptin, ng/mL	0.27 (0.30)	0.33 (0.19)	−0.03 (0.91)	0.19 (0.48)	−0.28 (0.27)	−0.15 (0.57)	0.26 (0.32)	−0.36 (0.15)
D(iPTH), pg/mL		0.23 (0.37)	−0.43 (0.08)	0.31 (0.23)	−0.11 (0.68)	−0.09 (0.73)	0.43 (0.08)	0.03 (0.91)
D(RANKL), pmol/L			−0.21 (0.42)	0.91 (<0.001)	0.41 (0.10)	0.04 (0.89)	−0.17 (0.51)	−0.01 (0.98)
D(OPG), pmol/L				−0.51 (0.04)	−0.07 (0.78)	−0.02 (0.93)	−0.32 (0.21)	0.38 (0.14)
D(RANKL to OPG) ratio					0.43 (0.08)	0.04 (0.88)	−0.01 (0.98)	−0.10 (0.69)
D(Sclerostin), ng/mL						0.08 (0.76)	−0.21 (0.43)	0.15 (0.57)
D(DKK-1), ng/mL							−0.25 (0.34)	0.17 (0.52)
D(FGF23), U/mL								−0.25 (0.33)

Abbreviation: 25OH VitD, 25-hydroxyvitamin D. Data are presented as partial coefficient of correlation; r_p (P value).

Discussion

The adverse effect of chronic energy deprivation on bone health has been an area of concern in women with HA (11). We have previously reported that in such patients meterleptin treatment for up to 2 years significantly increased bone mineral content by 5.1% and LS-BMD by 8.6% (9). However, the exact mechanism through which leptin exerts its favorable effect on the skeleton is largely unknown. In this study, we tested the effect of meterleptin on several molecules involved in bone physiology and pathophysiology. We found that meterleptin treatment over 36 weeks decreased the RANKL to OPG ratio and iPTH levels in hypoleptinemic women with HA. We observed no effect on serum sclerostin, DKK-1, or FGF23.

Restoration of normal leptin serum levels per se may not be the determinant of bone mass restoration because its levels were not correlated with BMD at any site in our population and could not predict BMD in Caucasian (12, 13) or Chinese (14) women. It is more likely that meterleptin acts indirectly through a positive effect on the abnormalities in the gonadal, GH, and adrenal axes and more specifically an increase in estradiol and IGF-1 to IGF-binding protein 3 ratio while decreasing cortisol levels (10). It has also been proposed that leptin may centrally regulate both bone formation (15) and bone resorption (16) via the sympathetic nervous system, which favors bone resorption by increasing expression of RANKL and cocaine and amphetamine regulated transcript, which inhibits bone resorption by modulating RANKL expression. However, these effects of leptin on the sympathetic nervous system in mice have not yet been verified in human studies (17).

Regardless if direct or indirect, it is not clear whether the overall effect of meterleptin on bone is an antiresorptive or an osteoanabolic one. We have previously reported in this cohort that osteocalcin, a marker of bone formation, increases and urinary amino-terminal telopeptide of type 1 collagen adjusted for creatinine, a marker of bone resorption, stabilizes with meterleptin treatment compared with placebo (10). However, in the smaller extension trial totaling 2 years of meterleptin treatment, only carboxy-terminal telopeptide of type 1 collagen, a marker of bone resorption, decreased significantly (9), indicating that changes may occur over time. In this context, investigating molecules produced by or affecting bone cells could help us better understand the mechanisms involved.

In our study, iPTH was significantly and independently decreased with meterleptin treatment. It could be hypothesized that meterleptin, through restoring E2 levels, increases calcium absorption from the intestine (18) and reduces calcium renal excretion (19), and this leads to a suppression of iPTH secretion from the parathyroid glands. However, the decrease of iPTH with meterleptin in our population was independent from E2 levels. A direct effect of leptin on PTH could also be speculated. Serum leptin levels were significantly lower in hemodialysis females with higher iPTH (20, 21), but this could reflect decreased feeding in patients with more advanced disease and therefore higher PTH levels. On the other hand, leptin levels were positively correlated to iPTH in obese subjects (22) and were higher in kidney transplant recipients with higher iPTH (23) and in patients with primary hyperparathyroidism (24) and decreased after parathyroidectomy, suggesting that PTH may also regulate leptin production (25).

The clinical significance of this decrease in iPTH within the normal range in women with hypothalamic amenorrhea treated with meterleptin remains largely unknown, whereas the design of this study does not allow the extraction of secure conclusions. Based on PTH physiology and the inverse relationship between calcium and PTH described within the normal PTH range (26), we could speculate that iPTH decrease would diminish the osteoclastic activity, thereby possibly being beneficial for bone metabolism. However, this remains to be shown by further studies of different design.

The RANKL/OPG system is a major regulator of osteoclast differentiation and activity (27). Conditions favoring increased osteoclastic activity and bone resorption are associated with an increased RANKL to OPG ratio, whereas decreased RANKL to OPG ratios are related to reduced bone resorption (28, 29). Given that chronically energy deprived women with anorexia nervosa or HA and hypoleptinemia exhibit an increased RANKL to OPG ratio (30), the opposite would be expected after meterleptin initiation. Indeed, we observed a significant decrease in RANKL to OPG ratio resulting from a trend towards decrease in RANKL and an increase in the OPG serum levels. This finding is in alignment with previous in vitro data (31) and the above-mentioned decrease in serum carboxy-terminal telopeptide of type 1 collagen (9) [a bone resorption marker considerably more reliable than the ones in the urine (32)] and implies that meterleptin treatment reduces osteoclast activity.

The above-mentioned decrease of iPTH with meterleptin could contribute to the decrease in RANKL production and the subsequent decrease of osteoclastic activation (33); however, this seems unlikely, considering the lack of statistically significant interaction between PTH and the RANKL to OPG ratio over time in our patients. On the other hand, sclerostin showed a U-shaped trend between 12 and 24 weeks of treatment, whereas DKK-1 was not affected by meterleptin treatment. Sclerostin and DKK-1 are inhibitors of the Wnt/β-catenin pathway, which is considered a major promoter of bone formation (34). Given that osteocytes are the main source of RANKL (35) and sclerostin (36), but not DKK-1 (37), we could speculate that meterleptin acts on the osteocytes by decreasing the production of both RANKL (resulting in decreased bone resorption) and sclerostin (resulting in increased bone formation), and this could contribute to the increase in bone mass in HA patients. The U-shaped change of sclerostin over time might represent an escape phenomenon after the restoration of bone mass to the anticipated for age levels. There is no solid explanation for the between group difference in DKK-1 only at week 24, but this finding is probably of limited value, considering that there was no significant change in DKK-1 over time within either the control or meterleptin group.

This is the first human study to investigate the effect of meterleptin treatment on molecules regulating bone metabolism in hypoleptinemic women with HA. The study was well randomized, double blinded, and placebo controlled with stringent selection criteria of the study subjects. All assays had been rigorously evaluated to ensure results are reproducible and consistent.

Some limitations need to be acknowledged. The sample size was small and thus, the study's power is low, but the data reached statistical significance. The duration of intervention was only up to 36 weeks; the effect of a longer duration of leptin replacement on the OPG to RANKL ratio, sclerostin, DKK-1, and iPTH needs to be further elucidated before definite conclusions are drawn. It also remains to be shown whether the effect of leptin on these molecules might be dose dependent. Finally, the meterleptin group started at a higher (although statistically nonsignificant) iPTH level compared with placebo; however, this does not affect within-group comparisons. However, if the meterleptin and placebo groups had started both at approximately the same baseline level and had followed similar patterns over time, this would have possibly affected the between-group comparisons, ie, possibly the meterleptin group would have significantly lower iPTH as compared with the control group, especially at week 36.

Larger prospective interventional studies using varying doses of meterleptin are needed to clarify the effect of leptin on bone metabolism and specifically to affirm the effect on the RANKL/OPG dipole and explore the mechanism of PTH reduction as well as to investigate the effect of meterleptin on the osteocytes in in vitro studies. The effect of meterleptin treatment on PTH, RANKL, OPG, and Wnt inhibitors in the setting of bone microenvironment would be also of interest; however, this study would require repeated bone biopsies. Finally, it would be interesting to study the reversibility of the effects of meterleptin on bone metabolism and BMD after treatment discontinuation.

In summary, our data suggest that meterleptin treatment over 36 weeks decreases iPTH levels and the RANKL to OPG ratio in hypoleptinemic women with HA in favor of reduced osteoclast differentiation and activation. These results add to our understanding on the role of hypoleptinemia in the bone pathophysiology of chronically energy-deprived individuals. This study warrants larger clinical studies, comparing the effect of various doses of meterleptin and duration of treatment on bone parameters both in serum and bone microenvironment.

Funding Statement

This work was supported by the National Institutes of Health National Center for Research Resources Grant M01-RR-01032 (Harvard Clinical and Translational Science Center).