Serum leptin and gonadotropin levels in patients with ... - Nature

Molecular Psychiatry (1999) 4, 71–75  1999 Stockton Press All rights reserved 1359–4184/99 $12.00

ORIGINAL RESEARCH ARTICLE

Serum leptin and gonadotropin levels in patients with anorexia nervosa during weight gain A Ballauff1, A Ziegler2, G Emons3, G Sturm3, WF Blum4,5, H Remschmidt1 and J Hebebrand1 1

Department of Child and Adolescent Psychiatry of the University of Marburg; 2Institute of Medical Biometry and Epidemiology of the University of Marburg; 3Department of Gynecology of the University of Marburg; 4University Children’s Hospital, Giessen; 5Lilly Deutschland GmbH, Bad Homburg, Germany

Keywords: luteinizing hormone; follicle stimulating hormone; fertility; amenorrhea; reproduction; eating disorder Leptin plays an important role in reproductive function. In patients with acute anorexia nervosa, serum leptin levels have repeatedly been shown to be lower than in age-matched controls. We have previously hypothesized that the amenorrhea characteristic of anorexia nervosa is related to this low leptin secretion. In an attempt to address this hypothesis, serum levels of leptin and follicle stimulating hormone (FSH) and luteinizing hormone (LH) of 16 female inpatients with anorexia nervosa or an eating disorder not otherwise specified (atypical anorexia nervosa) were measured on a biweekly basis during weight gain. We hypothesized that a serum leptin level of 1.85 ␮g L−1 would be associated with gonadotropin levels at or above the minimal level observed during the menstrual cycle in healthy adult fertile females. Our results revealed that increments of LH levels generally tracked increments of leptin levels during the first weeks of treatment. Similarly, in those patients with low referral leptin levels, FSH initially also tracked leptin levels. In contrast, a relationship between gonadotropin levels and leptin secretion was no longer discernible after LH and FSH levels had peaked. Those patients with exceedingly low leptin levels upon admission revealed a slow increase of gonadotropin levels. Our hypothesis of a threshold leptin level of 1.85 ␮g L−1 was supported for LH only.

The first hint of leptin’s important role in fertility became evident upon reproductive rescue of female ob/ob mice treated with exogenous leptin.1 Young wild-type mice treated with leptin reproduce earlier than non-treated animals.2 Leptin also blunts the response of the reproductive axis to starvation; both levels of luteinizing hormone (LH) and follicle stimulating hormone (FSH) do not drop to levels characteristic of semi-starvation, if food-deprived mice are injected with leptin.3 In in vitro studies leptin produced a dose-related increase in FSH and LH in hemi-anterior pituitaries. Low doses of leptin produced increases in secretion

of LH-releasing hormone (LH-RH) upon incubation of median eminence-arcuate nuclear explants from male rats.4 In ovariectomized female rats plasma LH, but not FSH levels, increased after intraventricular application of leptin.4 Leptin acts both at the hypothalamic and pituitary level to stimulate nitric oxide release which in turn mediates LH-RH or LH release, respectively.5 In humans leptin might be the mediator between the adipose tissue and the central events underlying menarche and maintenance of reproductive function. Indeed, a leptin threshold level for initiation of reproductive ability in women has been suggested.6 In girls leptin levels increase during puberty,7,8 leptin increments are followed by increments of FSH and later by LH and estradiol.8 Leptin possibly regulates the minute-to-minute oscillations in the levels of LH and estradiol.9 Mutations in both the leptin gene and the leptin receptor gene are associated with primary amenorrhea in females.10,11 In postmenarcheal females, secondary amenorrhea has been found to be associated with plasma leptin levels below 3 ␮g L−1 and an absent diurnal rhythm of leptin secretion.12,13 In our own studies performed with a sensitive radioimmunoassay,7 a leptin level of 1.85 ␮g L−1 predicted both past and present episodes of amenorrhea in a study group of underweight female students,14 whose leptin levels were predicted by both the degree of restrained eating and a history of weight loss.15 Leptin levels in female adolescents with acute AN were below this level of 1.85 ␮g L−1, whereas most patients with bulimia nervosa had higher leptin levels.14,16 In this study we explore the relationship between leptin and LH and FSH secretion, respectively, in 14 patients with AN and in two patients with an eating disorder not otherwise specified (atypical anorexia nervosa) at referral for inpatient treatment and during therapeutically induced weight gain. Based on our previous results,14 we hypothesized that a serum leptin level of 1.85 ␮g L−1 would be associated with LH and FSH levels at or above the minimal level observed during the menstrual cycle in healthy adult fertile females. Similar to our previous reports,14,16 leptin levels at referral were ⱕ1.92 ␮g L−1 (mean 0.69 ± 0.63; range 0.04–1.92). BMIs and % body fat initially ranged from 11.9 to 17.4 kg m−2 (mean 14.3 ± 1.5) and from 5.1 to 21.6 (mean 14.2 ± 5.9), respectively. At referral serum levels of LH and FSH ranged from 0.01 to 1.2 mIU ml−1 (mean 0.30 ± 0.31) and from 0.01 to 7.5 mIU ml−1 (mean 2.00 ± 2.25), respectively. Thus, LH levels were under the minimal level (0.8 mIU ml−1) of the reference range of the respective ELISA in 15 out of the 16 patients. In contrast, initial FSH levels of only eight patients were below the minimal normal level of 1.6 mIU ml−1. Correlations between initial log10 leptin levels and LH and FSH levels at referral were 0.54 and 0.59, respectively (P-values ⬍0.05). BMI and % body fat were not correlated (r = 0.04 and r = 0.07, respectively) with the initial LH level (P-values ⬎0.15). The FSH level at referral was correlated with BMI (r = 0.51; P⬍0.05), but not with % body fat (r = 0.24, P = 0.37). Therapeutically induced weight gain led to

Serum leptin and gonadotropin levels in AN patients A Ballauff et al

72

increments of BMI, % body fat, log10 leptin, LH and FSH (all P-values ⬍0.0005) between referral and discharge. As reported previously,16 the peak leptin values (mean 8.05 ± 4.05 ␮g L−1; range 2.79–14.6) were high in comparison to BMI-matched controls (data not shown). These peaks were reached after a median of 16.5 weeks. Mean peak concentrations for LH and FSH were 5.26 ± 5.22 mIU ml−1 (range: 0.6–23.1) and 5.85 ± 1.78 mIU ml−1 (range: 2.7–9.4), respectively. These peak concentrations were reached after a median of 16 and 11 weeks, respectively. Figures 1–5 illustrate the temporal relationship over a time period of ⱖ16 weeks for the increments of the five measured variables in five patients who serve to document the full range of patterns observed in all 16 patients during inpatient treatment and in single cases after discharge. The variability of the individual courses is readily apparent. Premenarcheal patients (n = 4; aged 12–14 years) did not show a pattern consistently different from patients with secondary amenorrhea. At a descriptive level increments of LH levels generally tracked increments of leptin levels during the first weeks of treatment (Figures 2–4). Similarly, in those patients with low referral leptin levels, FSH initially also tracked leptin levels. In contrast, a relationship between gonadotropin levels and leptin secretion was no longer discernible after LH and FSH levels had peaked (eg Figures 3 and 4). The three patients with the lowest leptin levels at referral (⬍0.01 ␮g L−1) serve to illustrate an extreme group of females in whom the delay between the increase in leptin secretion and the increments of both LH and FSH was pronounced. Thus, in these patients both LH and FSH levels were still low despite achievement of high leptin levels upon weight gain after several weeks of treatment. These patients all had referral BMIs ⬍13.2 kg m−2; their initial percent body fat was ⬍8.5. Only a single additional patient with a referral leptin level of 0.98 ␮g L−1 and without a pronounced reduction of gonadotropin levels had both BMI and percent body fat below these respective values. We conclude that normalization of gonadotropin secretion takes considerably longer in those patients who initially have very low leptin levels. In the light of the intraindividual variability observed, we attempted to investigate the relationships between leptin and FSH and LH secretion, respectively, by statistical approaches. This modelling should be helpful for future research in this area. However, upon attainment of higher case numbers, analyses using different subgroups appear more appropriate. For the purpose of this study we assumed that a critical leptin threshold for the reproductive axis exists. If leptin levels are below this hypothesized threshold, gonadotropin levels should be below the reference range. Leptin levels above the critical threshold would lead to gonadotropin levels within the reference range, thus representing the initial signal in the ensuing cascade that (re)establishes reproductive function. However, once leptin secretion exceeds the threshold level required to trigger this cascade, leptin levels should no

Figure 1 Serum leptin, luteinizing hormone (LH) and follicle stimulating hormone (FSH) levels in a 17-year-old postmenarcheal female with an eating disorder not otherwise specified (atypical anorexia nervosa) during therapeutically induced weight gain. A similarly late increase of LH and FSH levels was observed in two other patients who also had a BMI ⬍13.2 kg m−2, percent body fat ⬍8.5 and a leptin level ⬍0.1 ␮g L−1 upon referral. In this patient LH and FSH increments did not set in until after week 10 despite considerable weight gain, increase in percent body fat and an ensuing high leptin secretion. At week 16 the LH level was still below the reference range.

longer bear a systematic relationship to LH or FSH levels. Accordingly, because we expect a monotone instead of a linear relationship, classical pooled intraclass correlations are not adequate to analyze the assumed relationship. Therefore, we analyzed the relationship with a simple and readily testable hypothesis: once the gonadotropin levels are within the reference range, leptin levels have surpassed the critical level of 1.85 ␮g L−1. We equated the postulated threshold value with this leptin level because of our own aforementioned results obtained in independent study groups.14 Furthermore, we chose the minimal levels of the reference ranges for both LH (0.8 mIU ml−1) and FSH (1.6 mIU ml−1) in healthy eumenorrheic females to assess if upon attainment of these values leptin levels were at or above the critical level of 1.85 ␮g L−1. Of the 15 patients who reached a LH level of 0.8 mIU L−1 during inpatient treatment, 12 had a leptin level ⱖ1.85 ␮g L−1 at the time when this LH level was initially reached or surpassed (binomial test: nominal P = 0.018; corrected P = 0.036). However, the same leptin level showed no systematic relationship to the FSH level of 1.6 mIU ml−1 (nominal P = 0.75; corrected P = 1). Thus, 12 out of the 16 patients reached an FSH level ⱖ1.6 mIU ml−1 despite the fact that leptin levels were still below 1.85 ␮g L−1. This partially reflects the fact that FSH levels in eight patients were already


73

Figure 2 Serum leptin, LH and FSH levels in a 14-year-old postmenarcheal female with anorexia nervosa during therapeutically induced weight gain. This patient illustrates the pattern of secretion of leptin and LH and FSH for patients with low, but not extremely low leptin levels (⬎0.1 ␮g L−1) at referral. FSH secretion rises within the first weeks and is followed by an increase of LH secretion. The patient had a bleeding episode during week 21 of inpatient treatment.

Figure 3 Serum leptin, LH and FSH levels in a 14-year-old premenarcheal female with anorexia nervosa during therapeutically induced weight gain. The 30-week-long treatment led to a prolonged period of high leptin secretion (as of week 16 levels exceeded 6.5 ␮g L−1), which seemingly had no effect on FSH and LH secretion, which peaked at weeks 10 and 14, respectively.

greater than 1.6 mIU ml−1 at initiation of treatment. We conclude that the leptin level of 1.85 ␮g L−1 seemingly serves as a threshold for LH secretion only. Possibly, a lower leptin threshold exists for FSH secretion. Caveats of our study include the biweekly sampling procedure, the low case number and our inability to address diurnal and ultradian rhythmic secretion of the investigated hormones. Nevertheless, our longitudinal study design has proven useful for an analysis of the

relationship between leptin and gonadotropin secretion in patients with AN during weight gain. Further studies are clearly warranted to verify the threshold leptin value of 1.85 ␮g L−1 for LH, to determine if a threshold for FSH secretion can be calculated, to understand the role of leptin in the initiation of secondary amenorrhea in patients with AN and finally to understand the mechanisms relevant for the return of menstrual function in these patients once leptin levels


74

Figure 4 Serum leptin, LH and FSH levels in a 13-year-old postmenarcheal female with anorexia nervosa during therapeutically induced weight gain. This patient had a high leptin level (⬎1.5 ␮g L−1) upon referral. Both FSH and LH levels increase systematically during the first 4 weeks. Thereafter, no relationship to leptin secretion is evident except for the final levels, which stem from a blood sampling 5 weeks after discharge from inpatient treatment. The patient had lost a weight of 3 kg; both leptin and LH levels dropped. The patient reported a single bleeding episode 1 week after discharge.

Figure 5 Serum leptin, LH and FSH levels in a 17-year-old postmenarcheal female with anorexia nervosa of the binge eating/purging type during therapeutically induced weight gain. Both FSH and LH secretion intermediately drops to almost undetectable levels. Peaks in leptin secretion are accompanied by peaks in gonadotropin levels. This patient was known for her manipulative behavior. Unallowed drinking of water to manipulate body weight presumably underlies the considerable variation of percent body fat as measured by bioelectric impedance.

have passed the assumed threshold. Patients with exceedingly low leptin levels represent a distinct subgroup because their gonadotropin levels respond to the increased leptin secretion only after a considerable time lag.

Materials and methods Fourteen patients with a DSM-IV diagnosis of AN (13 of the restricting type and one of the binge eating/purging type) and two patients with a DSM-IV


diagnosis of an eating disorder not otherwise specified (EDNOS) who did not fulfill the second DSM-IV criterion for AN were included in this study. Diagnoses were based on the revised module for eating disorders (kindly provided by Professor Wittchen, Munich) of the Composite International Diagnostic Interview.17 The patients’ mean age at referral for inpatient treatment was 15 ± 1.8 (s.d.) years (range 12–18). The mean BMI at referral was 14.3 ± 1.5 kg m−2 (range 11.9–17.4). The mean duration of the eating disorder prior to referral was 13 ± 16.5 months (range 2–60). Four patients were premenarcheal. In postmenarcheal females amenorrhea had lasted for an average of 9.9 ± 13.8 months (range 3–48). Patients were treated as described previously.16 For the purpose of this study patients were longitudinally assessed for a minimum of 12 weeks during inpatient treatment (mean: 21.1 ± 8.1 weeks, range 12–39). Body height and weight were measured at referral; weight was subsequently measured at least weekly. Bioelectrical impedance measurements (Data-Input; 2000-S) and bloodsampling were performed at 8.00–8.30 am after an overnight fast upon referral and subsequently biweekly and concomitantly to the anthropometric measurements during inpatient treatment. At single time points patients were not available for a variety of reasons including short-term discharges during inpatient treatment. Seven of the 16 patients were followed up as outpatients after discharge. Serum leptin levels were determined as described previously.7,16 Serum LH and FSH levels were measured using the SR1-EIA-analyzer (Biochem Immuno Systems, Freiburg, Germany); the standard curves cover the range of 1–150 and 200 mIU ml−1, respectively, with intra- and interassay CV values less than 6%. Low levels (⬍1 mIU ml−1 for FSH and ⬍0.05 mIU ml−1 for LH) were linearly extrapolated. The lower limits of detection for FSH and LH are 0.2 and 0.05 mIU ml−1, respectively; the inter/intra assay CVvalues are 4% up to 90% (FSH) and 30% (LH) at decreasing levels. The references ranges for FSH and LH in healthy premenopausal females are 1.6– 12.2 mIU ml−1 and 0.8–15.3 mIU ml−1, respectively. This study was approved by the Ethics Committee of the University of Marburg. Written informed consent was obtained from the patients or in the case of minors, their parents. The exact two-sided one sample bionomial test was applied to investigate whether leptin levels had surpassed the hypothesized threshold of 1.85 ␮g L−1 at the time when LH and FSH levels reached the reference ranges of 0.8 mIU ml−1 and 1.6 mIU ml−1, respectively. These statistical tests are not affected by the higher intra/inter assay variations of FSH and LH for inpatients with very low gonadotropin levels (⬍0.5 mIU ml−1 for LH and ⬍1 mIU ml−1 for FSH). The t-test for dependent samples was used to assess increments of BMI, % body fat, log10 leptin, LH and FSH between referral and discharge, because the Lilliefors18 statistic did not reveal deviations from normality for the variables that were used for testing. Pearson cor-

relation coefficients and the corresponding test statistics were calculated to analyze associations at referral. Adjustments for the two statistical tests of primary interest performed in this paper were done according to Bonferroni.

References 1 Chehab FF, Lim ME, Lu R. Correction of the sterility defect in homozygous obese female mice by treatment with the human recombinant leptin. Nature Genet 1996; 12: 318–320. 2 Ahima RS, Dushay J, Flier SN, Prabakaran D, Flier JS. Leptin accelerates the onset of puberty in normal female mice. J Clin Invest 1997; 99: 391–395. 3 Ahima RS, Prabakaran D, Mantzoros C, Qu D, Lowell B, MaratosFlier E et al. Role of leptin in the neuroendocrine response to fasting. Nature 1996; 18: 250–252. 4 Yu WH, Kimura M, Walczewska A, Karanth S, McCann SM. Role of leptin in hypothalamic-pituitary function. Proc Natl Acad Sci USA 1997; 94: 1023–1028. 5 Yu WH, Walczewska A, Karanth S, McCann SM. Nitric oxide mediates leptin-induced luteinizing hormone-releasing hormone (LHRH) and LHRH and leptin-induced LH release from the pituitary gland. Endocrinology 1997; 138: 5055–5058. 6 Matkovic V, Ilich JZ, Skugor M, Badenhop NE, Goel P, Clairmont A et al. Leptin is inversely related to age at menarche in human females. J Clin Endocrinol Metab 1997; 82: 3239–3245. 7 Blum WF, Englaro P, Hanitsch S, Juul A, Hertel NT, Mu¨ller J et al. Plasma leptin levels in healthy children and adolescents: dependence on body mass index, body fat mass, gender, developmental stage and testosterone. J Clin Endocrinol Metab 1997; 82: 2904–2910. 8 Garcia-Mayor RV, Andrade MA, Rios M, Lage M, Dieguez C, Casanueva FF. Serum leptin levels in normal children: relationship to age, gender, body mass index, pituitary-gonadol hormones, and pubertal stage. J Clin Endocrinol Metab 1997; 82: 2849–2855. 9 Licinio J, Negra¨o AB, Mantzoros C, Kaklamani V, Wong ML, Bongiorno PB et al. Synchronicity of frequently sampled, 24-h concentrations of circulating leptin, luteinizing hormone, and estradiol in healthy women. Proc Natl Acad Sci USA 1998; 95: 2541–2546. 10 Cle´ment K, Vaisse C, Lahlou N, Cabrol S, Pelloux V, Cassuto D et al. A mutation in the human leptin receptor gene causes obesity and pituitary dysfunction. Nature 1998; 392: 398–401. 11 Strobel A, Issad T, Camoin L, Ozata M, Strosberg AD. A leptin missense mutation associated with hypogonadism and morbid obesity. Nature Genet 1998; 18: 213–215. 12 Laughlin GA, Yen SSC. Hypoleptinemia in women athletes: absence of a diurnal rhythm with amenorrhea. J Clin Endocrinol Metab 1997; 82: 318–321. 13 Tataranni PA, Monroe MB, Dueck CA, Traub SA, Nicolson M, Manore MM et al. Adiposity, plasma leptin concentration and reproductive function in active and sedentary females. Int J Obes 1997; 21: 818–821. 14 Ko¨pp W, Blum WF, von Prittwitz S, Ziegler A, Lu¨bbert H, Emons G et al. Low leptin levels predict amenorrhea in underweight and eating disordered females. Mol Psychiatry 1997; 2: 335–340. 15 von Prittwitz S, Blum WF, Ziegler A, Scharmann S, Remschmidt H, Hebebrand J. Restrained eating is associated with a low leptin synthesis in underweight females. Mol Psychiatry 1997; 2: 420–422. 16 Hebebrand J, Blum WF, Barth N, Coners H, Englaro P, Juul A et al. Leptin levels in patients with anorexia nervosa are reduced in the acute stage and elevated upon short-term weight restoration. Mol Psychiatry 1997; 2: 330–334. 17 WHO. Composite International Interview (CIDI): (a) CIDI Interview (version 1.0), (b) CIDI-user manual, (c) CIDI-training manual, (d) CIDI computer programs. WHO, Geneva. 18 Lilliefors HW. On the Kolmogorov–Smirnov test for the exponential distribution with mean unknown. J Am Stat Assoc 1969; 64: 387–389.

Correspondence: J Hebebrand, Clinical Research Group, Department of Child and Adolescent Psychiatry, Hans-Sachs-Str 6, 35033 Marburg, Germany. E-mail: Hebebran얀post.med.uni-marburg.de Received 12 June 1998; revised and accepted 25 August 1998

75