| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Articles |
Sir Quinton Hazel Molecular Medicine Research Center, Biological Sciences, University of Warwick, Coventry, United Kingdom CV4 7AL; Department of Clinical Endocrinology, Medizinische Hochschule (R.H., G.B.), D 30623 Hannover, Germany; and Department of Endocrinology, Christie Hospital (R.D.M., S.M.S.), Manchester, United Kingdom M20 4BX
Address all correspondence and requests for reprints to: Prof. E. W. Hillhouse, The Sir Quinton Hazel Molecular Medicine Research Center, Department of Biological Sciences, University of Warwick, Gibbet Hill Road, Coventry, United Kingdom CV4 7AL. E-mail: ehillhouse{at}bio.warwick.ac.uk
Abstract
GH therapy is associated with a reduction in fat mass and an increase in lean mass in subjects with GH deficiency (GHD). Leptin, like GH, plays an important role in the regulation of body composition. GH treatment has been shown to reduce serum leptin; however, the physiological interactions between the leptin system (free leptin, bound leptin, and soluble leptin receptor) and the GH/IGF-I system largely remain unknown.
Twenty-five patients with childhood (n = 10) and adult-onset (n = 15) GHD were studied. GH status had previously been determined using an insulin tolerance test and/or an arginine stimulation test. The following parameters were recorded at baseline (V1) and then after 3 months (V2) and 6 months (V3) on GH treatment: fat mass, body mass index (BMI), and waist/hip ratio (WHR); blood samples were taken after an overnight fast for free leptin, bound leptin, soluble leptin receptor, insulin, and IGF-I.
At V2 and V3, respectively, a fall in free leptin (P < 0.001 for each), and at V3 a fall in in percent fat mass (P < 0.001) were observed. There were no significant changes in BMI or WHR. Simultaneously, there was a rise in insulin (P = 0.068 and P < 0.001), IGF-I (P < 0.001 and P < 0.001), bound leptin (P = 0.005 and P < 0.001), and soluble leptin receptor (P = 0.61 and P < 0.001). A positive relationship was noted between free leptin and BMI (P < 0.001) and between free leptin and fat mass (P < 0.001), and a negative relationship was found between free leptin and IGF-I (P < 0.001) and, within patient, between free leptin and insulin (P < 0.001). There was no significant correlation between free leptin and WHR. Bound leptin had a positive association with IGF-I (P < 0.001) and insulin (P = 0.002) and a negative relationship with percent fat mass (P = 0.023). Soluble leptin receptor was also positively related to IGF-I (P < 0.001).
In conclusion, our data suggest that the reduction in serum leptin with GH treatment, as noted by others, is mediated through a fall in free leptin. The fall in free leptin and in part the rise in bound leptin are most likely through a reduction in percent fat mass. However, the observed changes in free leptin and bound leptin and, more importantly, the rise in soluble leptin receptor, are not explained entirely by modifications in body composition and may be a direct result of GH/IGF-I.
LEPTIN, A 16-kDa circulating peptide and product of the adipose tissue-specific ob gene, functions as a sensor of fat mass, regulating metabolism and body composition (1) through its receptor (2). In humans, circulating leptin levels correlate with body weight and body fat mass (3, 4), suggesting that obesity may be related in part to differences in leptin secretion and/or sensitivity (5). It is recognized that in humans leptin circulates in both bound and free forms; the amount of each is dependent on factors such as obesity, fasting, and refeeding (6). Together with leptin receptors, which belong to the class I cytokine receptor family (7), bound leptin and free leptin comprise the leptin system (6). Soluble receptors of several cytokines (e.g. IL-2, IL-4, and TNF
) modulate the availability of free hormones (cytokines) for their biological actions by acting as competitive inhibitors of the membrane receptors (8), a finding true for the soluble leptin receptor (9). The regulation of leptin production in humans is poorly understood. Glucocorticoids (10) and insulin (11, 12) are examples of factors that increase leptin production, whereas androgens decrease leptin production (13). The relationship between GH and leptin is interesting, particularly as GH, like leptin, plays an important role in the regulation of body composition (14).
In rodents, leptin stimulates spontaneous pulsatile GH secretion and the GH response to GHRH (15). In fasted rats, the pattern of GH pulsatility is eliminated with a near absence of spontaneous peaks, but the administration of leptin by the intracerebroventricular route restores the altered pattern. When fed rats receive antileptin antibodies via the intracerebroventricular route, the normal pattern is reversed to an absence of pulses, reminiscent of the fasting state (16). GH alone does not stimulate leptin release from rodent adipocytes in vitro, but it significantly potentiates leptin release from both mouse and rat adipose tissue in the presence of dexamethasone or insulin, in the absence of lipolysis (17). In humans, the idea of an interaction between GH status and leptin is attractive, because 1) leptin levels are increased in subjects with GH deficiency (GHD) (18, 19) and are lower in acromegaly (20, 21); and 2) there is an inverse relationship between body fatness and GH secretion in both obese (22) and normal weight individuals (23). Although an inverse relationship between leptin and GH has been demonstrated in human subjects (24), there is no direct evidence of leptin inhibiting GH secretion. Furthermore, administration of GH in the acute setting has been shown to have no effect on leptin levels (25), although others have demonstrated an increase in leptin concentrations 24 h after exogenous GH administration (26). Chronic administration of GH is reported to be associated with a decrease in leptin levels by some researchers (27, 28, 29, 30, 31, 32), although not by others (18, 26, 33, 34). In addition, studies showing a positive correlation between body mass index (BMI) and leptin, have failed to show any correlation between leptin and IGF-I levels, both before (35) and after (36) GH therapy.
GHD individuals have abnormal body composition, with increased fat mass and decreased lean mass, which to an extent can be reversed by GH therapy (37). However, in these individuals, who have raised leptin levels (18, 19), the effects of GH on leptin remain controversial, as mentioned above, with the difference in serum leptin levels attributed to the alterations in body composition. To understand the physiological significance between GH and leptin, we examined the interaction between the GH/IGF-I system and components of the leptin system after chronic administration of GH in GHD individuals. We have attempted to determine whether leptin signaling can be potentially modified by the changes in the concentrations of leptin-binding proteins, in particular the soluble leptin receptor, through an interaction with the GH/IGF-I system, leading to an alteration in leptin levels, notably the bioavailable form-free leptin.
Subjects and Methods
Patient details
The study comprised 25 severely GH-deficient (GHD) adults (8 males and 17 females), of mixed adult and childhood onset, with an age range of 16–70 yr (mean, 38 yr). Severe GHD was defined as a peak GH response of less than 3 µg/liter to a provocation test. Two provocative tests were used in all patients with isolated GHD. The insulin tolerance test was used in 21 patients. The arginine stimulation test was used in 20, and the glucagon stimulation test was used in 6 patients in whom an insulin tolerance test was contraindicated or a second test was undertaken. The patients were recruited from the endocrine out-patient population at the Christie Hospital, Manchester. The study was approved by the South Manchester local regional ethics committee.
The underlying pathological diagnoses were nonfunctioning pituitary adenoma (n = 4), craniopharyngioma (n = 3), meningioma (n = 3), astrocytoma (n = 3), acute lymphoblastic leukemia (n = 3), prolactinoma (n = 2), idiopathic hypopituitarism (n = 2), histiocytosis X (n = 1), pinealoma (n = 1), dysgerminoma (n = 1), ependymoma (n = 1), and lymphosarcoma (n = 1). The patients consisted of those with a primary hypothalamo-pituitary tumor or infiltration (n = 12) and patients with GHD resulting from cranial irradiation received in the treatment of primary brain tumors or as prophylaxis for acute lymphoblastic leukemia (n = 13). None of the patients received GH in the 12 months before the study. Thirteen patients were gonadotropin deficient, 10 were cortisol deficient, 8 were T4 deficient, and 4 were antidiuretic hormone deficient. All were receiving stable replacement therapy throughout the study.
Study protocol
The protocol had an open treatment design. Before being entered into the study all patients underwent a general physical examination and were taught how to self-inject using an automated pen device (Genotropin pen, Pharmacia Biotech, Milton-Keynes, UK). With the patient fasted, blood was drawn between 1000 and 1200 h for measurement of IGF-I; additionally, serum was saved at -80 C for analysis of insulin and the leptin subfractions. The subjects were then commenced on GH therapy at a daily dose of 0.26 mg to be injected at 2200 h. The GH dose was subsequently adjusted by 0.13 mg/d, at intervals of at least 4 wk to optimize the IGF-I level within the range of -2 to +2 SD of the age-related mean. The 25 patients in the study represent all patients seen consecutively who managed to achieve their maintenance GH dose at their 6-month review. IGF-I was measured at each visit, and serum was saved for analysis of fasting insulin and leptin subfractions. The data presented in the study are based on fasting blood samples taken for insulin, IGF-I, free leptin, bound leptin, and soluble leptin receptor at baseline (V1) and after 3 months (V2) and 6 months (V3) of GH treatment. Weight, height, waist, and hip circumference measures were taken by a single investigator at each visit. Body composition was measured by bioelectrical impedance (TBF-305, Tanita, Uxbridge, UK).
Assays
GH assays were performed by an in-house two-site immunoradiometric assay measured against the reference standard National Institute of Biological Standards and Control 80/505. IGF-I assays were performed at Sahlgrens Hospital by RIA, using synthetic IGF-I for labeling (Nichols Institute Diagnostics, San Juan Capistrano, CA). The samples were pretreated with an acid-ethanol solution to avoid binding protein interference. The reference range of the laboratory was used to calculate the IGF-I SD score. Serum insulin was measured by an immunometric assay (insulin RIA, Pharmacia Biotech). Free leptin, bound leptin, and soluble leptin receptor were measured using an in-house RIA, details of which have been published previously (38, 39).
Soluble leptin receptor assay: rabbits were immunized with a fragment (amino acids 354–368, Saxon Biochemical, Hannover, Germany) of the extracellular domain of the human leptin receptor to detect circulating forms of the short soluble receptor. Again, no sequence homology to any other protein in the Swiss-Prot databank was detected, and a naturally occurring Tyr in the sequence allowed for easy radiolabeling. Intra- and interassay variations for the soluble leptin receptor assay were 4.5% and 7.3%, respectively.
Statistical snalysis
The design of this observational study, in which multiple measurements were made within a patient, created a hierarchical data structure. Although the patient was the actual sampling unit, the unit of interest and hence the unit of analysis were, in fact, the components of the leptin system within patients. Hence, statistical analysis of the data were performed by means of generalized linear mixed (fixed and random effect) hierarchical (patient and observation level) statistical models, by which total observed variation could be apportioned to that occurring between patients vs. within a patient. Models were generated using MlwiN software (Institute of Education, London, UK), and the significance of model parameters was assessed by maximum likelihood methods. Model fit and underlying distributional assumptions were assessed by examination of standardized residuals. Two-tailed tests of statistical significance were derived from Wald 
2 values of appropriate linear contrasts. In all analyses, statistical significance was considered to be achieved for P < 0.05.
Results
The results of the study are presented in Table 1
and Figs. 1–5. For insulin measures, 41% of the total observed variation occurred between patients, with the remainder (59%) accounted for by within-patient variation. Compared with the baseline there was a marginally nonsignificant rise in insulin concentration at the 3-month visit (P = 0.068), although values at the 6-month visit were significantly higher than observed at 3 months (P < 0.001) or at baseline (P < 0.001; Fig. 1A).
|
|
|
|
|
|
).
After initiation of GH there was no significant change in BMI (Fig. 2A) and waist to hip ratio (WHR) from baseline values at either 3 months (P = 0.68 and P = 1.0, respectively) or 6 months (P = 0.74 and P = 0.36, respectively). However, a significant decline in the percentage of fat mass was noted between baseline and the 6-month visit (P < 0.001; Fig. 2B). These findings serve to illustrate that percent fat mass is a better measure than BMI in GHD adults.
With respect to free leptin measures, 89% of the total observed variation occurred between patients, with 11% accounted for by within-patient variation. A significant fall in free leptin concentrations from baseline was noted at each follow-up visit (3 months, P < 0.001; 6 months, P < 0.001) and also between the 3-month and 6-month visits (P = 0.015; Fig. 3A). Conversely, compared with baseline, bound leptin increased at each follow-up visit (3 months, P = 0.005; 6 months, P < 0.001) and between the 3-month and 6-month visits (P = 0.030; Fig. 3B); however, a significant rise in soluble leptin receptor was only achieved at the 6-month visit (P < 0.001; Fig. 3C).
Although free leptin was positively associated with insulin at each visit, the tendency for a simultaneous decline in free leptin and a rise in insulin over time is manifest as an increase in the magnitude (i.e. slope) of the relationship over time, but also as a significant negative relationship between the two measures within each individual (P < 0.001; Fig. 4). An interesting observation was that within patients the rising insulin and falling free leptin levels yielded a negative relationship, whereas between the patients (i.e. looking at the correlation at each visit) there was a positive relationship. This suggests that the two parameters may be simply changing independently in response to GH administration, thereby making it difficult to evaluate any causal link. Interestingly, it is worth noting that once the effect of IGF-I on free leptin levels had been accounted for, insulin did not add significantly (P = 0.13) to the multivariate analysis of relationships with free leptin in two-level generalized linear mixed model despite a correlation between insulin and IGF-I (r2 = 0.37).
In other univariate models free leptin was positively associated with percent fat mass at each visit (P < 0.001; Fig. 5A), and a significant positive relationship between the two measures was noted within each individual (P < 0.001; Fig. 5B). In addition, free leptin correlated positively with BMI (P < 0.001) and negatively with IGF-I (P < 0.001). Perhaps more interestingly, however, in multivariate analyses the magnitude of the negative relationship between IGF-I and free leptin was dependent on the BMI of the patient; the strongest relationship between the two measures was observed in individuals with the highest BMI, although the interaction term just failed to reach statistical significance (P = 0.09). Further, once this relationship among free leptin, body mass index, and IGF-I was accounted for, no further variables contributed significantly to improved model fit.
With respect to bound leptin, statistical models considering only single variables demonstrated significant positive associations with IGF-I (P < 0.001) and insulin (P = 0.002) and a negative association with percent fat mass (P = 0.023), but not with BMI (P = 0.83). However, once IGF-I was included in the model no other variables achieved significance. Similarly, soluble leptin receptor was only significantly positively associated with IGF-I (P < 0.001).
Discussion
In this paper we have demonstrated the impact of GH on the different components of the leptin system. Free leptin concentrations significantly decreased after GH administration, with a strong and significant correlation between free leptin and BMI and percent body fat mass, both before and after the GH treatment. Earlier reports also observed a fall in total leptin levels after GH therapy. However, in some studies (31) the fall in leptin levels seemed to antedate the change in body composition (reflected in a fall in fat mass), whereas others (27, 32) observed a decrease in leptin concentrations that was proportional to the decrease in body adiposity. Our data show that reduction in percentage body fat after GH treatment in GHD individuals, without a significant change in BMI or WHR, results in a reduction in free leptin concentrations. However, despite the strong correlation between change in percent body fat mass and change in free leptin concentrations, it is possible that the observed fall in free leptin concentrations might have at least in part reflected a direct interaction between the GH/IGF-I axis and leptin, rather than a reduction in fat cell mass. This direct interaction between GH/IGF-I and leptin has been demonstrated in the rodent model in which adipocytes, which express IGF-I receptors (40), have suppressed leptin mRNA expression after IGF-I treatment (41).
The beneficial effects of GH therapy are well recognized, with an increase in lean mass and a decrease in fat mass (37), as seen in our GHD individuals. The reduction in percent fat mass after GH treatment was associated with a significant increase in bound leptin concentrations, highlighting a negative correlation between the two parameters; no correlation was noted between bound leptin and BMI or WHR. These findings are in keeping with the observations that in lean subjects the majority of circulating leptin is in the bound form, whereas in individuals with increased fat mass the majority of leptin circulates in the free form (6). Given our findings, that GH decreased free leptin concentrations but increased bound leptin, it is possible that the change in total leptin concentrations during GH treatment may have been too small to detect in some of the studies reported in the literature.
Moreover, the results of our study suggest that the relationship between GH and leptin may extend beyond simply a decrease in free leptin and a rise in bound leptin concentrations, and that factors other than body fat mass may be involved. In particular, we have demonstrated that GH treatment may alter the concentrations of leptin-binding proteins, as it significantly increases the concentration of soluble leptin receptor. The soluble leptin receptor, a leptin-binding protein (6), binds leptin at a 1:1 ratio (42). Liu et al. (9) have shown inhibition of leptin binding to membrane leptin receptors by the soluble leptin receptor, which exerts a dominant negative effect on leptin signaling through the long isoform of the leptin receptor. More recently, Brabant et al. (43) demonstrated that leptin-binding proteins may indeed play a role in the regulation of the leptin system in human subjects. Interestingly, they found a significant positive correlation between serum and cerebrospinal fluid for both bound and free leptin concentrations, but more importantly they noted a positive correlation between the resting energy expenditure and bound leptin concentrations. These observations suggest that the balance between free and bound leptin may thus be a potential regulator of leptin bioavailability, influencing leptin signaling both peripherally and in the central nervous system. Hence, an increase in the soluble leptin receptor concentrations, as seen with GH therapy in our GHD individuals, might potentially modify leptin signaling via increased free leptin binding with a concomitant increase in bound leptin concentrations, consequently leading to decreased leptin sensitivity through a fall in the bioavailable form (free) of leptin. Therefore, the changes in free and bound leptin concentrations observed after GH treatment, which correlate with changes in percent fat mass, may also be explained by the rise in soluble leptin receptor concentration.
In addition to the described changes in serum concentrations of the components of the leptin system, we have demonstrated a significant correlation between these components and IGF-I. To date, the relationship between leptin and IGF-I has not been clearly understood. Houseknecht et al. (44) suggest the presence of a positive correlation between the concentrations of leptin mRNA and IGF-I mRNA in bovine adipose tissue. However, in several other studies (22, 26, 35) no correlation between leptin and IGF-I has been shown, although such a correlation might have been obscured by the fact that only total leptin levels were measured. Furthermore, Isozaki et al. (45) and Hardie et al. (46) have shown no direct effect of IGF-I on leptin gene expression, whereas others (47, 48) suggest a negative effect of IGF-I on leptin gene expression. In their study Nystrom et al. (13) demonstrated a negative correlation between leptin and IGF-I in men, although with borderline significance (P = 0.048). Therefore, in the present study we have shown for the first time a highly significant negative correlation between free leptin and IGF-I, with a concomitant significant positive correlation between bound leptin and IGF-I and between soluble leptin receptor concentrations and IGF-I. It should be noted that IGF-I was the only variable significantly associated with soluble leptin receptor concentrations. In addition, not only have we demonstrated a negative relationship between free leptin and IGF-I within the patients over the course of the study, but importantly our findings suggest that the magnitude of this relationship is greatest for patients with the highest BMI. In addition, the opposite correlations noted between IGF-I and the free and bound forms of leptin may help explain the conflicting data and poor correlation between IGF-I and total leptin reported by previous researchers. The presence of such a correlation between free leptin and IGF-I and between bound leptin and IGF-I, even when BMI and percent body fat had been accounted for confirms differential effects of GH administration on the leptin system. Further studies are required however to elucidate whether the effects of GH on the leptin system represent a direct GH effect or are mediated through IGF-I.
As GH administration resulted in a rise in fasting insulin levels, this raised the question of whether and to what extent the observed changes in the leptin system could be mediated through a change in insulin levels. In humans, insulin has been shown to stimulate leptin production both in vitro (11) and in vivo (12). Recently, in healthy human subjects, we have demonstrated that free leptin levels significantly increased during a 26-h hyperinsulinemic euglycemic clamp (49). In the present study, however, despite a significant increase in insulin concentrations during GH treatment, there was a significant fall in free leptin levels at each visit. It would therefore seem likely that the effects of GH both directly and indirectly, via a change in body composition with concomitant fall in free leptin levels, outweigh any direct effects of insulin on free leptin concentrations. The question then arises of whether the increased insulin levels lead to the rise in bound leptin and soluble leptin receptor concentrations during GH treatment? To date there is no published evidence that insulin may increase the concentrations of bound leptin or the soluble leptin receptor. Recently, we failed to show any significant rise in bound leptin and soluble leptin receptor concentrations during a hyperinsulinemic euglycemic clamp (49), a finding true in our previous study of normal pregnancy (38). In the present study despite a significant positive correlation between bound leptin and insulin (P = 0.002), insulin was not a significant variable (P = 0.11) once IGF-I was included in the multiple variable analysis; IGF-I was the variable statistically related to bound leptin. Similarly, insulin was not significantly associated with soluble leptin receptor concentrations, either on its own (univariate analysis, P = 0.09) or in a model containing IGF-I. We can therefore postulate that the increase in bound leptin and soluble leptin receptor concentrations during GH treatment is likely to represent in part a direct GH effect on the leptin system, probably mediated through IGF-I. At the physiological level, GH, which is known to increase resting metabolic rate (50), may decrease leptin sensitivity by increasing soluble leptin receptor levels and consequently increase free leptin binding. This, in turn, may lead to an increased appetite and preservation of the body adipose reserves in the setting of an increased metabolic rate.
In conclusion, in GHD individuals we have demonstrated that GH treatment decreases free leptin levels, but increases the concentrations of bound leptin and the soluble leptin receptor. Insulin per se was not the factor responsible for these observed effects of GH. It is likely that the ratio of free/total leptin may not be constant, and that a dynamic equilibrium probably exists between the free form of leptin and the circulating binding proteins, where this balance may be affected by GH/IGF-I. Further studies are needed to elucidate the precise nature of such an interaction, and whether these (our) findings are a direct effect of GH and/or an effect of GH via IGF-I.
Footnotes
Abbreviations: BMI, Body mass index; GHD, GH deficiency; WHR, waist/hip ratio.
Received February 12, 2001.
Accepted November 5, 2001.
References
This article has been cited by other articles:
 |
|
 |
|
  D. E. Edwards, R. P. Bohm Jr, J. Purcell, M. S. Ratterree, K. F. Swan, V. D. Castracane, and M. C. Henson Two Isoforms of the Leptin Receptor Are Enhanced in Pregnancy-Specific Tissues and Soluble Leptin Receptor Is Enhanced in Maternal Serum with Advancing Gestation in the Baboon Biol Reprod, November 1, 2004; 71(5): 1746 - 1752. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. S. Randeva, K. C. Lewandowski, J. Komorowski, R. D. Murray, C. J. O'Callaghan, E. W. Hillhouse, H. Stepien, and S. M. Shalet Growth Hormone Replacement Decreases Plasma Levels of Matrix Metalloproteinases (2 and 9) and Vascular Endothelial Growth Factor in Growth Hormone-Deficient Individuals Circulation, May 25, 2004; 109(20): 2405 - 2410. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Saaresranta and O. Polo Sleep-disordered breathing and hormones Eur. Respir. J., July 1, 2003; 22(1): 161 - 172. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C U Nwokolo, D A Freshwater, P O'Hare, and H S Randeva Plasma ghrelin following cure of Helicobacter pylori Gut, May 1, 2003; 52(5): 637 - 640. [Abstract] [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
