Leptin and the skeleton


  • Tom Whipple,

    Corresponding author
    1. Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA
      Tom Whipple, MS, PT, Penn State Orthopaedic & Sports Medicine Center, 1850 East Parle Avenue, Suite 112, University Park, PA 16803. Tel: 814 235 4737; Fax: 814 235 2492. E-mail: tjw208@psu.edu
    Search for more papers by this author
  • Neil Sharkey,

    1. Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA
    Search for more papers by this author
  • Laurence Demers,

    1. Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA
    Search for more papers by this author
  • Nancy Williams

    1. Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA
    Search for more papers by this author

Tom Whipple, MS, PT, Penn State Orthopaedic & Sports Medicine Center, 1850 East Parle Avenue, Suite 112, University Park, PA 16803. Tel: 814 235 4737; Fax: 814 235 2492. E-mail: tjw208@psu.edu


Leptin is a 167 amino acid hormonal protein product of the obesity gene that has been widely studied since its discovery by Zang et al. (1994). Initially described as being involved with satiety and energy balance, leptin was proposed to be the anti-obesity factor involved in a feedback loop from the adipocyte to the hypothalamus (Zang et al., 1994). Increasing evidence continues to underscore the importance of leptin as a crucial hormone in the regulation of food intake as well as body weight in both animals and humans (Pekkymounter et al., 1995; Considine et al., 1996; Mercer et al., 1997; Trayhurn et al., 1999). Further investigation has elucidated a role for leptin in the regulation of metabolism (Emilsson et al., 1997; Kamohara et al., 1997), sexual development (Mantzoros et al., 1997; Matkovic et al., 1997) reproduction (Chehab et al., 1996), haematopoiesis (Gainsford et al., 1996), immunity (Lord et al., 1998), gastrointestinal function (Bado et al., 1998), sympathetic activation (Collins et al., 1996) and angiogenesis (Sierra-Honinmann et al., 1998). Given the inter-relationships between these processes and bone, recent research efforts have been directed at investigating leptin–skeletal interactions.

Several lines of evidence suggest that bone remodelling and thus skeletal homeostasis is governed by endocrine and/or humoral factors (Riggs & Melton, 1986; Simonet et al., 1997). Among anthropometric and metabolic factors, body weight is a major determinant of bone density (Mazess et al., 1987). The obese gain more bone during the formative years and have a slower rate of bone loss later in life (Tremollieres et al., 1993). These effects appear to be related more to fat mass than lean mass (Reid et al., 1992). Thus far, the protective effect of obesity has been elusive and is explained by either the peripheral aromatization of adrenal androgens to oestrone in adipose tissue and/or to mechanical loading factors. However, these two mechanistic models do not fully resolve the relationship between obesity and increased bone mineral density (BMD) as fat mass and bone density are still correlated after adjusting for oestrogen status (Reid et al., 1992), and this relationship continues to exist equally at both weight bearing and nonweight bearing sites (Hla et al., 1996). Because leptin is positively correlated with body fat and is involved in reproductive hormonal regulation, defining its role in skeletal physiology has received increasing attention. The purpose of this article is to provide an overview of leptin as it relates to the skeleton, as well as to highlight considerations for future research endeavors.

In-vitro evidence for leptin–skeletal interactions

Thomas and colleagues investigated the role of human recombinant leptin on a conditionally immortalized human marrow stromal cell line, hMS2-12 (Thomas et al., 1999). This cell line has the potential to differentiate to either the osteoblast or adipose phenotype. Administration of leptin to the stromal cells resulted in mRNA and protein expression for the leptin receptor as well as induced selective differentiation of the osteoblast vs. adipocyte phenotype. No significant proliferative effect of leptin on hMS2-12 cells was reported. Because of the absence of Cbfa1, an early response gene involved with osteoblast commitment (Komori et al., 1997), and the presence of alkaline phosphatase and osteocalcin, the conclusion was that the differentiation occurred at the maturation vs. the commitment stage. In abstract, Halloway and colleagues demonstrated that murine leptin was able to inhibit osteoclastogenesis of human mononuclear cells cultured on bone (Holloway et al., 2000).

Reseland et al. (2001) reported that isolated human osteoblasts in culture were capable of expressing both leptin and the leptin receptor. This finding was limited to a single time-point observation and was proposed to indicate that leptin expression may be limited to either the mineralization and/or osteocyte transition period of osteoblast differentiation. Incubation of both osteoblasts and osteosarcoma cell lines with recombinant leptin was found to be associated with enhanced formation of mineralized nodules (Reseland et al., 2001).

Preliminary evidence obtained in vitro indicates that leptin is a hormone with the capacity to simultaneously stimulate osteoblast and inhibit osteoclast differentiation. Collectively, these findings suggest that leptin may have the potential to directly modulate both anabolic and protective skeletal functions. As obesity and BMD are correlated, it appears tenable that leptin may govern the balance between body composition and BMD.

Leptin and bone metabolism in animal models

Hoggard et al. (1997) previously demonstrated high levels of gene expression for both leptin and the leptin receptor in fetal murine mouse cartilage/bone. The results from this experiment reveal that the genetic expression of leptin is age, tissue and species specific.

Extensive experimentation by Ducy et al. (2000) on leptin deficient (ob/ob) and leptin receptor deficient (db/db) mice provides novel evidence that the skeletal effects of leptin may in part be governed through the central nervous system. Both mutant strands of mice are described as being obese, hypogonadal and have high levels of cortisol. Based on human correlational studies, the obese phenotype would favour increased BMD (Mazess et al., 1987; Reid et al., 1992; Tremollieres et al., 1993). Conversely, both hypogonadism and hypercortisolism should predictably result in decreased bone mass becuase sex hormone deficiency has been demonstrated to favour bone resorption (Riggs et al., 1986) and hypercortisolism has been shown to inhibit bone formation (Reid, 1997). In both mutants, a high bone mass phenotype that precedes the onset of obesity is expressed. Based upon histomorphological examination, mutant bone formation rate was increased by 60–70% compared to controls. No receptors for leptin were identified on the osteoblast. Neither osteoblast surface, nor osteoblast number were increased, indicating that enhanced osteoblastic function was responsible for the improvements in bone mineral content. Furtheremore, continuous central infusion of leptin to the intracerebroventricular region for 28 days resulted in significant bone loss. The conclusion based on these results was that leptin has an inhibitory effect on skeletal mass and its actions are mediated though an undefined hypothalamic/central nervous system pathway. In addition to providing evidence for the central regulation of bone turnover, these findings reveal an unexpected discrepancy between the mechanism of action of leptin in the control of body weight and of bone formation.

Ducy et al. (2000) proposed that leptin resistance may provide an explanation as to why their findings appear contradictory to the observation that obese humans, with elevated leptin levels, demonstrate increased BMD. It has been shown that mechanisms for leptin resistance may exist in the obese and that the ratio of cerebral spinal fluid leptin levels to serum leptin levels is significantly greater in lean individuals than in obese individuals (Caro et al., 1996; Schwartz et al., 1996; Couce et al., 2001). As such, if leptin resistance is operational and dictates skeletal homeostasis, accurately defining an individual's degree of leptin resistance, as well as detailing both central and peripheral leptin concentrations, would appear to be necessary requisites in future studies of leptin–skeletal interaction.

Baldock et al. (2002) recently contributed to our understanding of the modulation of bone formation by the central nervous system via neuropeptide Y (NPY), a downstream modulator of leptin, and the Y2 receptor. Genetically obese mice (ob/ob, db/db) that are leptin or leptin receptor deficient demonstrate increased activity of NPY circuits in the hypothalamus due to a lack of leptin-induced inhibition (Stephens et al., 1995). NPY mediates its effects through the activation of multiple different receptors in the mouse (Y1, Y2, Y4, Y5, Y6), all of which are expressed in the hypothalamus (Blomqvist & Herzog, 1997; Parker & Herzog, 1999). Both leptin and Y2 receptors are located on the NPY-expressing neurones of the arcuate nucleus and are thought to share common signalling pathways (Stephens et al., 1995; Broberger et al., 1997; Baskin et al., 1999; King et al., 2000). Y2 receptor deficient knockout models were generated by Baldock et al. (2002) and were found to have a twofold increase in trabecular bone volume of the distal femoral metaphysis compared to controls. The Y2 receptor deletion stimulated osteoblastic activity and increased bone mineralization rate independent of alterations in bone cell surface measurements, plasma total calcium or hormonal output. The result of hypothalamic Y2 receptor deletion appears to release a centrally mediated tonic inhibition of osteoblasts cells. These data suggest that Y2 receptor signalling is important in the central regulation of bone mass although the specific mechanisms for this influence are not well understood.

In sharp contrast to the findings of Ducy et al. (2000), Steppan et al. (2000) demonstrated that peripheral leptin administration has anabolic rather than inhibitory effects on skeletal mass and bone metabolism in the ob/ob mouse. Daily intraperitonial leptin administration for 3 weeks resulted in significant increases in total bone mineral content (15·4%) and density (4·3%), as well as increases in total body bone area (12%) and femoral length (5·4%) compared to controls. Additionally, trabecular mineral content was increased by 84·2%; whereas cortical mineral content was increased by 16%. These findings were obtained despite decreased food intake and decreasing body weight (Steppan et al., 2000).

A bone sparing effect of leptin has also been demonstrated in the Sprague-Dawley rat (Burguera et al., 2001). The effect of leptin on bone mass was examined for 1 month in ovariectomized rats and was contrasted with that of 17α-ethinyl oestradiol (E2) as well as a combination of leptin and E2. Leptin was effective in reducing 69% of the trabecular bone loss experienced by control rats and demonstrated a protective effect on trabecular architecture and number. The combination of leptin and E2 further decreased bone turnover rate compared to isolated E2 treatment, suggesting the need to further elucidate the interactive effects of leptin and oestrogen.

Results from preliminary animal experiments provide evidence that leptin is a potent modulator of the rodent skeleton. These data suggest that leptin action is dependent upon age and species differences, as well as method and timing of administration. The need to examine both central and peripheral leptin–skeletal interactions has been identified by the dichotomous results of studies performing central vs. peripheral leptin administration. Leptin action appears to be anabolic and/or skeletal-protective when administered peripherally, yet produces negative effects when administered centrally. Therefore, the net result on bone may be secondary to the combination of positive peripheral and negative central effects, depending on serum leptin concentration as well as blood–brain barrier permeability. As such, the need to advance our understanding of leptin resistance and its associated consequences is self-evident. Finally, leptin action appears to be contingent upon the presence of other hormones, highlighting the need to detail leptin–hormone interactions as they relate specifically to the skeleton.

Leptin and bone metabolism during human development

There is evidence that leptin may be an important growth factor during human fetal development, with a positive correlation having been established between umbilical cord blood leptin concentration and newborn weight, body mass index (BMI) and arm fat (Hassink et al., 1997). The major determinant of fetal leptin concentration is the accumulation of fat mass (Clapp & Kiess, 1998). Ogueh et al. (2000) measured fetal blood leptin levels, carboxy-terminal pro-peptide of type I pro-collagen (PICP; a marker of bone formation) and cross-linked carboxy-terminal telopeptide of type I collagen (ICTP; a marker of bone resorption). Their results revealed a positive correlation between leptin concentration and gestational age (r = 0·240, P = 0·042) and a negative correlation with ICTP (r = −0·420, P = 0·001). The correlation between fetal leptin concentration and ICTP suggests that leptin may decrease bone resorption with a net effect of increasing bone mass.

A positive correlation between leptin concentrations and adiposity in children exists (Hassink et al., 1996), and these children achieve puberty earlier than children of normal weight (Jarurantanasirkul et al., 1997). Additionally, leptin concentrations were found to decrease with advancing Tanner stage of development in both normal-weight and obese children, independent of obesity (Hassink et al., 1996). Thus, higher concentrations of leptin in the prepubertal stage compared to the postpubertal stage may reflect a regulatory mechanism for times of rapid growth and development. Matkovic et al. (1997) evaluated the association between leptin and bone mass in 343 healthy young Caucasian females during peak growth. Their results showed a positive relationship between serum leptin and whole body BMD (r = 0·307, P < 0·0001). Leptin did not show any influence on bone mineral content when bone area was among the predictor variables. Instead, leptin was related to bone area, suggesting a modelling response on the periosteal envelope (Matkovic et al., 1997). Klein et al. (1998) also reported that there was no difference in the bone density of 18 obese children, with elevated leptin levels compared to 30 normal weight controls. However, obese children were discovered to have a greater bone age to chronological age ratio than nonobese children of similar pubertal stage (Klein et al., 1998). Because biochemical measures of bone turnover were not obtained in these studies, the specific mechanisms of skeletal modelling and/or remodelling leading to the expressed alterations in bone morphology or maturity remain undefined.

Precocious puberty is a condition of rapid physical maturation characterized by an accelerated rate of development of secondary sex characteristics. It has been correlated with low birth weight, rapid catch-up growth and a variety of metabolic and hormonal abnormalities (Ibanez et al., 1999). Two of the hormonal disturbances are hyperinsulinaemia and hyperandrogenism, both of which have been associated with increases in BMD later in life (Barret-Connor & Ktitz-Silverstein et al., 1996; Good et al., 1999). Ibanez et al. (2000) reported that a sample of 52 normal weight girls with precocious puberty (age range 6·9–14·9 years) had higher leptin levels for their degree of BMI. These subjects had increased lumbar spine BMD (r = 0·42, P < 0·05) and advanced bone age for chronological age compared to a control population (Ibanez et al., 2000). It appears from these results that leptin levels may be elevated in certain circumstances independent of obesity and possibly be related to the interplay between other endocrine factors. However, because BMI vs. body composition was used to compare these subjects to controls, no firm conclusion can be made regarding these findings.

A detailed case study by Farroqi et al. (1999) of a 9-year-old leptin-deficient girl provides additional insight on the role of leptin in developmental skeleton physiology. In brief, the patient had a normal birth weight but had marked hyperphagia, and began gaining excessive weight at approximately 4 months of age. As a result of her excessive obesity, valgus deformities of the legs developed, for which she underwent lateral tibial osteotomies. Treatment with daily leptin injections was started when the patient was 9 years old and weighed 94·4 kg (> 99·9th percentile for age). Leptin dosage was equivalent to 10% of the child's predicted normal serum leptin concentration. The patient lost weight within 2 weeks of starting treatment and continued to do so over the 12-month treatment period, during which she lost a total of 16·4 kg, of which 95% was attributed to adiposity. Despite the loss of body weight, bone mineral mass was increased by 0·15 kg. Although multiple factors may have governed the increase in bone mass, it suggests that subcutaneous administration of leptin may at least preserve and/or potentially increase bone mass despite diminishing body weight, adiposity and reduced food consumption.

Collectively, studies that have examined the role of leptin in human skeletal development reveal that the the effects of the hormone may be expressed either by an anabolic action or by a permissive antiresorptive effect, or both. In addition, bone growth may be augmented by leptin-induced mechanisms that increase the rate at which maturation takes place, rather than result in definitive morphological or structural consequences per se. Clarification of the role of leptin during the developmental years would be enhanced by longitudinal studies that correlate the chronological, reproductive and bone age of a sample with specific architectural parameters as well as indices of skeletal remodelling.

Leptin and bone metabolism in the adult skeleton

Several correlational studies have demonstrated a positive effect of serum leptin on adult bone (Pasco et al., 2001; Thomas et al., 2001; Yoneda et al., 2001). In the largest study to date, Thomas and colleagues tested the hypothesis that serum leptin, insulin or oestrogen levels, alone or in concert, mediated the relationship between adiposity and BMD. Their study population included 343 male subjects and 349 female subjects. BMD, body composition, and biochemical markers of skeletal turnover were assessed and correlated with fasting serum hormone levels. Their findings revealed that serum leptin levels were associated with BMD in women, but not men. Overall, leptin explained 0·3%, 1% and 0·01% of the variance in BMD in the men at the total hip, mid-lateral spine and mid-distal radius, respectively. The comparable figures were 10%, 0·3% and 5% in premenopausal women, and 19%, 6% and 10%, in postmenopausal women. Among female subjects, leptin tended to be inversely associated with several markers of bone turnover [bone alkaline phophatase, osteocalcin, PICP, N-crosslinked telopeptides (NTx)]. The most consistent of these associations were noted with urine NTx (r =−0·24, P < 0·01) for premenopausal women), suggesting that the hormone has an antiresorptive effect.

Although leptin is intimately related to adiposity, Pasco et al. (2001) demonstrated an association between serum leptin and BMD independent of body weight and body fat mass in a study of 214 healthy, nonobese women. Specifically, with bone density as the dependent variable, adjusting for age, body weight and body fat mass, the association with the natural logarithm of leptin reached a significant level at the lateral spine (partial r2 = 0·030, P = 0·011) and was of borderline significance at Ward's triangle and the trochanter (partial r2 = 0·012–0·017, P = 0·058–0·120). Positive correlations between leptin and bone mineral content were also reported at all sites measured. Additionally, in a sample of 51 dialysis patients, serum leptin was positively correlated to distal radius BMD for females (r = 0·469, P < 0·02) but not males (Yoneda et al., 2001).

In contrast to the previous findings, others have reported little or no interaction between leptin and bone (Martini et al., 2001; Odabasi et al., 2001). Rauch et al. (1998) examined the interactions between serum leptin and bone density, bone cortex geometry, and bone metabolism in 94 healthy women, aged 40–60 years. Total, trabecular and cortical bone density, as well as bone area, were measured by quantitative computed tomography (QCT) at the distal radius and bone turnover was assessed by measuring biochemical remodelling markers. Their findings revealed no interactions existed between leptin and any of the CT scan indices or markers of skeletal metabolism. However, PICP was associated with serum leptin in the postmenopausal group, after adjustment for BMI (r = −0·40, P = 0·009). In a second report, Iwamoto et al. (2000) investigated leptin–bone interactions in 76 healthy females, and found no correlation for whole body BMD and only weak correlations between serum leptin and BMD of the pelvis (r = 0·39, P < 0·05) and left leg (r = 0·41, P < 0·01) in premenopausal women. These correlations decreased after adjustment for BMI. However, leptin was found to be positively correlated with the serum bone formation marker alkaline phosphatase (r = 0·56, P < 0·001) and inversely correlated with the urinary resorption marker deoxypyridinoline (r = −0·32, P < 0·05) in premenopausal subjects, even after adjustment for BMI.

Four studies to date have examined leptin–bone interactions in samples composed of exclusively postmenopausal women (Goulding & Taylor, 1998; Odabasi et al., 2000; Martini et al., 2001; Blain et al., 2002). The conclusion of three independent investigations was that no significant or direct correlation exists between serum leptin and bone mass in the postmenopausal female. However, independent of the final conclusion of these studies, inspection of the data reveals that leptin–bone interactions may exist. Specifically, Odabasi et al. (2000) examined 50 subjects with osteoporosis and 30 age and BMI-matched healthy controls. Their findings revealed no correlation between plasma leptin and spinal BMD as determined by dual-energy X-ray absorptiometry in healthy subjects, but a weak correlation was found to exist in the subjects with osteoporosis (rs = 0·285, P = 0·045). Goulding & Taylor (1998) concluded that leptin had no direct effect controlling bone cell activity because no correlation could be identified between leptin and biochemical markers of bone turnover. However, the researchers did report that plasma leptin was correlated with both bone mineral content (r = 0·480, P < 0·001) and BMD (r = 0·551, P < 0·001).

In a recent and thorough investigation of 107 postmenopausal women aged 50–90 years, Blain et al. (2002) investigated whether leptin is an independent predictor of BMD. This study examined the relationships of BMD and bone turnover to serum leptin, 25(OH)D, 1,25(OH)2D, PTH, E2, dehydroepiandrosterone sulphate, GH, IGF-I, calcium intake and body composition. In a stepwise multiple regression statistical analysis, leptin explained 7·2% and 3·7% of the whole body and femoral neck BMD variance, respectively. Additionally, there was a significant inverse correlation between leptin and urinary C-telopeptide of type I collagen (r = −0·27, P < 0·01), indicating leptin action may be exerted via an antiresorptive mechanism (Blain et al., 2002). Blain and colleagues underscored the importance of adjusting for fat mass vs. BMI when examining leptin–skeletal interactions and proposed that failure to do so might explain some of the contradictory findings in previous studies.

Leptin and body composition analysis

BMI is often used as an index of obesity in correlational or cross-sectional research. However, BMI is calculated from weight and height measurements (kg/m2) and does not discriminate between fat and lean mass. Because leptin is produced by the adipocyte and is correlated to fat mass vs. lean or total mass, studies utilizing BMI rather than direct measures of body composition must be interpreted with caution. Not only can there be significant differences in body composition for a given BMI among individuals, but age-related height loss could potentially affect the interpretation of results (Sorkin et al., 1999). Accordingly, if body weight and adiposity remain constant, BMI will increase artifactually, as height is always lost with age. A 7%‘error’ in BMI is predicted to occur in the average female who experiences normal age-related height loss between early and late adulthood (Sorkin et al., 1999). Therefore, controlling for the association between leptin and BMD with BMI is likely to be less accurate than controlling for fat mass. Results from studies that have utilized direct measurements of body composition to examine leptin–skeletal interactions should be considered superior to those that have not.

The results from correlational and cross sectional human studies presented to date would appear to substantiate the need for additional, carefully controlled and standardized hypothesis-driven research. As such, it is recommended that future investigations of leptin and the skeleton not only consider the issues related to body composition, but also remember that leptin action may be specific to an the age, gender and/or hormonal status of an individual.

Gender-related differences in leptin secretion

Leptin concentrations in women are significantly higher than in men, despite subjects of opposite sex being matched for adiposity (Havel et al., 1996; Rosenbaum et al., 1996; Sad et al., 1997). This sexual dimorphism is identified in umbilical cord blood, suggesting that gender-based differences are well established during the intrauterine period (Matsuda et al., 1997). While Matsuda et al. (1997) concluded that gender-related differences in leptin are due to genetic factors, others have demonstrated that leptin is strongly influenced by the presence of other humoral factors. Casabiell et al. (1998) studied human omental adipose tissue cultures from both genders and reported that spontaneous leptin secretion was higher in females than males, and that both dexamethasone and oestradiol stimulated leptin secretion in women but not men, while neither progesterone nor oestradiol influenced leptin secretion in either sex. Shimizu et al. (1997) have demonstrated that oestrogen increases in-vivo leptin production in both rats and humans. A rise in testosterone was also shown to correspond with a reduction in leptin levels in developing boys (Garcia-Mayor et al., 1997), while testosterone replacement therapy causes a reduction in leptin levels in hypogonadal men (Jockenhovel et al., 1997).

In addition to a potential genetic bias for females to secrete more leptin per measure of adiposity than males at baseline, the respective reproductive hormones appear to strongly influence subsequent leptin release and action. Realizing that significant gender-related differences in leptin physiology exist is an important consideration for accurately defining leptin–skeletal interactions in future studies. Additionally, these data underscore the importance of examining leptin–skeletal interactions in concert with the existing hormonal milieu, rather than in isolation.

Although there is mounting evidence that leptin is a potent direct or indirect governor of the skeleton, leptin was initially introduced as an afferent metabolic signal of dietary energy and/or energy reserve. In this context, consideration of the diurnal rhythm of leptin, as well as its sensitivity to energy availability as it may relate to the skeleton, appears to be especially warranted.

Leptin diurnal rhythm and sensitivity to energy availability

A diurnal variation of plasma leptin has been observed in humans in which the zenith occurs at night and the nadir occurs during the day (Sinha et al., 1996). The average circadian amplitude between acrophase and nadir was 51·7% in obese, and 75·6% in lean, respectively. Findings by Schoeller et al. (1997) indicate that the diurnal rhythm of leptin is related to human feeding habits. Leptin levels are found to slowly rise after breakfast, become significant after lunch and to continue to increase after dinner. This diurnal pattern appears to be a very sensitive indicator of energy availability (Ahima et al., 1996; Chin-Chance et al., 2000).

Kolaczynski et al. (1996) investigated the response of leptin to short-term fasting and overfeeding in humans. Fasting resulted in a steady decline from baseline values and reached a nadir at 36 h, while restoration of normal food intake was associated with a rapid rise in leptin concentration and a return to baseline values within 24 h. Hilton & Loucks (2000) detailed the 24-h response of leptin to an energy deficient state created by either diet or exercise. Their findings demonstrated that, irrespective of whether the energy deficit was created by diet or exercise, both the 24-h mean and peak amplitude of leptin were significantly reduced. van Aggel-Leijssen et al. (1999) studied the effects of exercise and diet on 24-h leptin profiles and found that exercise under conditions of energy surfeit had no effect on the average 24-h leptin level, but increased the peak amplitude of leptin by two-fold compared with an exercise-energy-balanced trial. A noteworthy finding from this study is that, despite the two-fold increase in the nocturnal peak amplitude of leptin, 09·00 h baseline levels of plasma leptin remained unchanged. It has been proposed that the findings from a single morning blood sampling for leptin may be misleading, and may not accurately reflect the mean 24-h and/or peak concentration of the hormone (Kanabrocki et al., 2001). To date, all of the previously referenced correlational studies of leptin have analyzed only a single fasting sample to base their conclusions. The skeletal effects resulting from the dynamic profile, peak amplitude or total 24-h levels of of leptin remain largely undefined.

Accordingly, and perhaps most noteworthy, Laughlin & Yen et al. (1997) examined 24-h leptin profiles with controlled nutrient intake in highly trained female athletes with and without menstrual cyclicity and in BMI-matched cycling sedentary controls. Twenty-four hour leptin levels were decreased in both groups of athletes relative to controls and correlated with reduced body fat. The diurnal pattern of leptin was characterized by an approximate 50% rise from nadir to peak in both the cycling athletes and controls, but was completely absent in amenorrhoeic athletes. No diagnostic imaging studies or biochemical markers of skeletal metabolism were performed in these subjects. However, the correlation between decreased BMD and amenorrhoeic athletes has been well established by others (Loucks & Horvath et al., 1985; Yeager et al., 1993; Voss et al., 1998). These data underscore the dynamic nature of the 24-h profile of leptin and its sensitivity to energy availability, as well as indicating the need to further elucidate the significance of the peak amplitude of leptin on the skeleton.

In addition to controlling for the effects of energy and investigating the significance of the 24-h profile of leptin, it is also recommended that future experiments consider that leptin action may be involved with either acute or long-term regulation of skeletal metabolism, or both. Therefore, biochemical markers of skeletal turnover may be useful short-term end points in studies attempting to define the role of leptin as it relates to acute changes in skeletal metabolism, whereas diagnostic imaging studies and/or scans may be better suited to describe the long-term skeletal consequences of leptin.

The diurnal rhythm of biochemical markers of bone turnover

Biochemical markers of bone formation are direct or indirect products of the osteoblast and are reflective of different phases of bone formation whereas the majority of the markers used to identify bone resorption activity are degradation products of bone (type I) collagen (Seibel, 2000; Woitge & Seibel et al., 2001). Although no studies to date have directly compared the 24-h profiles of leptin with either type of marker, it is relevant to note that bone cell activity has a recognized circadian variation and is also acutely sensitive to energy availability. A nocturnal zenith in bone formation activity, similar to that for leptin, has been described (Gundberg et al., 1985; Nielsen et al., 1990a,b). In addition, a well-known 24-h profile for bone resorption exists, with the peak amplitude occurring during the late night/early morning and the nadir late in the afternoon (Eastell et al., 1992; Schlemmer et al., 1992). The circadian fluctuation in bone formation has been shown to be largely dependent on the circadian alteration in serum cortisol (Nielsen et al., 1988; Nielsen et al., 1992; Heshmati et al., 1998), whereas the aetiology of the circadian variation in bone resorption remains largely undefined.

The influence of dietary energy on bone remodelling

The circadian rhythm of bone resorption has been found to be independent of cortisol (Schlemmer et al., 1997), age, menopausal status and posture (Schlemmer et al., 1994). The effect of calcium homeostasis on bone resorption is contradictory. Two studies have concluded that either short- or long-term calcium supplementation can be effective at reducing the nocturnal rise in parathyroid hormone and bone resorption (Blumsohn et al., 1994; McKane et al., 1996), whereas other findings indicate that supplemental oral calcium, constant calcium infusion, or nasal calcitonin have no influence on the 24-h profile of bone resorption (Sairanen et al., 1994; Ledger et al., 1995). The circadian increase in bone resorption has been proposed to be secondary to an absence of food availability. As such, two studies have demonstrated in rats that when food intake is divided into multiple portions vs. consumed at one meal, the peak in bone resorption is blunted (Muhlbauer & Fleissch, 1995; Li & Muhlbauer, 1999). Muhlbauer & Fleissch (1995) demonstrated an increase in rat bone mass after only 30 days of food fractionation. It remains to be determined if human bone remodelling may be influenced by such dietary practices. It is also unknown whether an alteration in serum leptin concentration, commensurate with energy availability, is a key central metabolic signal and/or a direct governor of skeletal resorption activity.

Human studies have confirmed that alterations in dietary energy induce abnormalities in bone remodelling (Ricci et al., 2001) and that reduction in body weight leads to a loss of BMD (Ramsdale et al., 1994; van Loan et al., 1998; Salamone et al., 1999). Acute fasting has been demonstrated to cause both a reduction in bone formation (Grinspoon et al., 1995) and an increase in bone resorption (Schlemmer & Hassager, 1999), whereas studies of chronic energy deficit have revealed an uncoupling of bone remodelling in which bone resorption is significantly increased (Ricci et al., 2001). Conversely, despite hypoestrogenism and amenorrhoea, refeeding has only been shown to result in restoration and recovery of normal circadian bone resorption profiles in recovering anorexics (Caillot-Augusseau et al., 2000). Although multiple nutritionally regulated hormones are known to have significant consequences on bone (Laughlin & Yen, 1996), including the growth hormone-IGF-1 axis (Clemmons et al., 1981; Merimee et al., 1982; Isley et al., 1983), a complete discussion of nutritionally regulated skeletal endocrinology is beyond the scope of this article.

Regulation of the growth hormone axis by leptin

GH exerts an anabolic effect on the human skeleton (Dieguez et al., 1988) and its action is influenced by alterations in dietary energy availability and body composition. Obese humans have decreased GH levels (Diezguez & Casaneueva, 1995; Vahl et al., 1997), whereas energy deficit and anorexia nervosa are characterized by increased GH secretion (Frankel & Jenkins, 1975; Maes et al., 1991). IGF-1 level is usually considered normal in the obese (Cordido et al., 1991) and deficient in the undernourished (Clemmonds et al., 1981). Although direct evidence is lacking, inferential support for the role of leptin in the regulation of GH is accumulating (Nagatini et al., 2000). Preliminary research in animals and correlational findings in humans suggest that adipose tissue can influence GH secretion either by the release of free fatty acids or leptin (Casanueva et al., 1998; Dieguez et al., 2000). In normally fed laboratory rats, central administration of leptin antiserum leads to a reduction in GH, as well as preventing the inhibitory effect of fasting on GH levels (Carro et al., 1997; Vuagnat et al., 1998). Leptin-induced modulation of the GH axis has been demonstrated to occur at the hypothalamic GHRH- and somatostatin producing neurones, possibly through a hypothalamic NPY axis (Vuagnat et al., 1998; Caro et al., 1999). Because GH is reduced in the rat and increased in humans during fasting (Ho et al., 1988), translation of these data to the human condition remains unclear at present.

Currently, there is no direct evidence linking leptin with modulation of GH secretion in humans. However, obesity is simultaneously associated with elevated leptin and reduced growth hormone levels. Weight loss lowers plasma leptin and normalizes growth hormone (Rasmussen et al., 1995). The recovery of normal GH secretion following weight loss points to an acquired rather than a causal relationship between GH and obesity. Whether the relationship between leptin and GH is related to a state of leptin-resistance as previously discussed, or by a leptin-induced inhibition of GH secretion, remains to be determined. Because the GH axis is known to have a profound effect on bone, and leptin is potentially involved either directly or indirectly with its regulation, it is recommended that future studies carefully examine leptin–GH interactions while controlling for differences in body composition and nutritional status.

Conclusions and acknowledgement

The initial role of leptin as the human adipostat has been significantly expanded. Leptin is now recognized as a pleiotropic hormone responsible for the regulation of multiple systems and capable of exerting sophisticated effects on numerous targets. To date, a preliminary understanding of leptin–skeletal interactions has emerged from the findings of pioneering investigators. The complexity of the leptin-skeletal axis is self-evident, indicating the need for continued multidisciplinary collaborative research efforts. We wish to gratefully acknowledge the many investigators and clinicians whose valuable contributions have been highlighted in this manuscript.