GH is a single chain protein with 191 amino acids and two disulfide bonds. The GH gene is located on chromosome 17. Approximately 75% is secreted in the 22kD form, while the remainder consists of a 20kD variant produced by alternate splicing. GH is secreted by the somatotrope cells located primarily in the lateral wings of the anterior pituitary. The morphological characteristics and number of these cells are remarkably constant throughout life, while secretion changes - as mentioned below. GH secretion occurs in a pulsatile fashion, and in a circadian rhythm with a maximal release in the second half of the night. Two hypothalamic hormones regulate GH secretion: Growth Hormone Releasing Hormone (GHRH) with a stimulatory action at the level of gene transcription and somatostatin (SST) with an inhibitory effect on the GH secretion from the pituitary gland. Various synthetically produced GH releasing compounds and the recently discovered natural hormone ghrelin probably have a dual effect in increasing the release of GHRH and inhibiting SST action, thereby obtaining a very powerful stimulation of GH secretion. Ghrelin in the systemic circulation derives from the stomach, but it remains to be convincingly demonstrated that gut-derived ghrelin is a regulator of GH secretion(2). Interestingly, exogenous ghrelin stimulates food intake and gastric emptying (2) In everyday life it is known, that stress, hypoglycaemia and ingestion of protein (high levels of circulating amino acids) stimulates GH secretion, while high levels of glucose and FFA inhibits secretion (Fig 1).

GH acts both directly through its own receptor and indirectly through the induced production of Insulin-like Growth Factor I (IGF-I). IGF-I is synthesised both in the liver and in the periphery, and is an important mediator of GH actions. It circulates bound to a number of different binding proteins of which IGFBP-3 is the most important.

With the introduction of dependable radioimmunological assays it was recognised that circulating GH was blunted in obese subjects (3), and that normal aging was accompanied by a gradual decline in GH levels (4). The latter observation led Rudman et al. (5) to the hypothesis that many of the senescent changes in body composition and organ function were related to or caused by hyposomatotropinemia. The term "somatopause" may be considered a paraphrase for Rudmans hypothesis although it remains uncertain who introduced this persuasive term.

More recent studies have uniformly documented that hypopituitary adults with severe GH-deficiency are characterised by increased fat mass and reduced lean body mass (LBM) (6). It is also known that normal GH levels can be restored in obese subjects following massive weight loss (7), and that GH substitution in GH-deficient adults normalises body composition (6).

What remains unknown is the cause-effect relationship between hyposomatotropenimia and senescent changes in body composition. Is the propensity for gaining fat and loosing LBM initiated or preceded by a primary age-dependent decline in GH secretion and action or vice versa?: accumulation of fat mass secondary to non-GH dependent factors (e.g. life style, dietary habits) results in a feedback inhibition of GH secretion

Moreover, little is known about possible age-associated changes in GH pharmacokinetics and bioactivity.

Assessment of GH status by means of standardized stimulation tests remains a cornerstone for the diagnosis of GH deficiency in children. The reason for this is that pituitary GH is released in a pulsatile and episodic manner separated by long intervals with low GH levels. A similar approach is used when evaluating hypopituitary adults, in whom it has been shown that stimulated GH release allows a better separation between patients and normal subjects as compared to 24-h spontaneous GH release (8). It is, however, noteworthy that stimulated GH peak levels are subject to a very pronounced inter- and intrasubject variability. A number of physiological variables such as body composition, nutritional status, physical fitness and sex steroids are known to influence GH release, but the degree to which each of these factors contributes to the individual variation is not clear . In adults it has been reported that the GH response to clonidine declines with age (9), whereas the response to arginine primarily appears to be determined by gender with higher levels in females (10). The association between body composition and stimulated GH release in healthy adults was assessed in a cross-sectional study in 42 clinically non-obese adults between 27-59 years (22 females/20 males) who underwent 2 stimulation tests (clonidine and arginine) in addition to in-depth measures of body composition and physical fitness (VO2-max) (11). Elderly people (mean age 50 years) had a lower peak GH response to both secretagogues, and females had a higher response to arginine when compared to males. Body mass index and intra-abdominal fat content (CT scan) was higher in "older" people and in males compared to "young" people and females, respectively and lean body mass was higher in males compared to females, whereas physical fitness was higher in young people compared to older people. Multiple regression analysis, however, revealed that intra-abdominal fat mass was the most important and negative predictor of peak GH levels (Fig.2), where as both age, gender and physical fitness were of minor importance. Lean body mass was not significantly associated with GH status in either males or females.

In the same population 24-h spontaneous GH levels were also analysed by means of deconvolution analysis of samples obtained every 20 minute. Mean GH levels, GH production rate and GH burst amplitude were higher in young people and in females as compared to older people and males, (12). Multiple regression analysis again suggested that intraabdominal fat mass was the single most important and negative determinant of GH status. Fasting levels of insulin, IGF-I and free fatty acids did not correlate with either estimates of GH status. Surprisingly, LBM exhibited a weak inverse correlation with mean 24-h GH release, but LBM was not associated with other attributes of GH status and was not an independent determinant by multiple regression analysis.

A detailed analysis of GH secretion in relation to body composition in elderly subjects has, to our knowledge, not been performed. Instead serum IGF-I has been used as a surrogate or proxy for GH status in several studies of elderly men (13-15). These studies comprise large populations of ambulatory, community-dwelling males aged between 50-90 years. Not unexpectedly serum IGF-I declined with age (Fig. 3), but IGF-I failed to show any significant association with body composition or physical performance (13-15). As also pointed out by some of the authors, however, the validity of IGF-I as an indicator of GH secretion is uncertain - in particular in adults. It is evident that serum IGF-I levels are low in GH deficient children and elevated in active acromegaly, but serum IGF-I levels correlate only weakly with GH status in healthy young and mid-life adults, and a large proportion of hypopituitary GH-deficient adults may have IGF-I levels within the normal range (16). The residual or non-GH-dependent determinants of IGF-I in adults remain elusive and merits future research.

Considering the great interest in the actions of GH in adults surprisingly few studies have addressed possible age-associated differences in the responsiveness or sensitivity to GH. In normal adults the senescent decline in GH levels is paralelled by a decline in serum IGF-I, suggesting a down-regulation of the GH-IGF-I axis. Administration of GH to elderly healthy adults has generally been associated with predictable albeit modest effects on body composition and a high incidence of side-effects (17). Whether this reflects an unfavourable balance between effects and side effects in older people or employment of excessive doses of GH is uncertain, but it is evident that older subjects are not resistant to GH. Studies in GH deficient adults with pituitary disease strongly suggest that the dose requirement declines with age. Short-term dose response studies clearly demonstrate that older patients require a lower GH dose to maintain a given serum IGF-I level (18-19), and it has been observed that serum IGF-I increases in individual patients on long-term therapy if the GH dosage remains constant (20). It has also recently been reported that hypopituitary patients above 60 years are highly responsive to even a small dose of GH (21). Interestingly, there appears to be a gender difference in GH deficient adults with men being more responsive in terms of IGF-I generation and fat loss during therapy (22).

The pharmacokinetics and short-term metabolic effects of a near physiological intravenous GH bolus (200 µg) were compared in a group of young (" 30 years) and older (" 50 years) healthy adults (23). The area under the GH curve was significantly lower in older subjects, whereas the elimination half-life was similar in the 2 groups, suggesting both an increased metabolic clearance rate (MCR) and apparent distribution volume (Vd) of GH in older subjects. Both MCR and Vd showed a strong positive correlation with fat mass, although multiple regression analysis revealed age to be an independent positive predictor. The short-term lipolytic response to the GH bolus was higher in "young" as compared to "older" subjects, respectively. Interestingly, the same study revealed that the GH binding protein (GHBP) correlated strongly and positively with abdominal fat mass (24).

It is obvious that the mechanism underlying the so-called somatopause involves other and perhaps more complex mechanisms than the female menopause, which predominantly is caused by gonadal resistance to gonadotropins. A prospective long-term study of normal adults with serial concomitant estimations of GH status and adiposity would provide useful information. Evaluation of GH sensitivity as a function of age, sex and body composition would also be worthwhile. In the mean time the following hypothesis may be proposed (Fig. 4): 1. Changes in life-style and genetic predispositions promote accumulation of body fat with aging 2. The increased fat mass increases FFA availability, inducing insulin resistance and hyperinsulinemia 3. High insulin levels suppress IGFBP-1 resulting in a relative increase in free IGF-I levels 4. Systemic elevations in FFA, insulin and free IGF-I suppresses pituitary GH release, which further increases fat mass 5. Endogenous GH is cleared more rapidly in subjects with high amount of fat tissue. The very strong positive correlation between fat mass and GHBP could suggest that GH is cleared in adipose tissue by a receptor mediated mechanism. Clearly, future studies are needed to substantiate or refute this simplified model. At present it is equally premature and unwarranted to recommend GH treatment to reverse the age-associated deterioration in body composition and physical performance.

The involvement of the pituitary gland in the regulation of substrate metabolism was originally detailed in the classic dog studies by Houssay (25). Fasting hypoglycaemia and pronounced sensitivity to insulin were described as salient features of hypophysectomised animals. These symptoms were readily corrected by administration of anterior pituitary extracts. It was also noted that pancreatic diabetes was alleviated by hypophysectomy. Finally, excess of anterior pituitary lobe extracts aggravated or induced diabetes in hypophysectomised dogs.

Luft et al. (26) clearly demonstrated the glycaemic control to deteriorate following exposure to a single supraphysiological dose of human GH in hypophysectomised adults with type 1 diabetes mellitus. Somewhat surprisingly, only modest effects of GH on glucose metabolism were recorded in the first metabolic balance studies involving adult hypopituitary patients (27, 28).

More recent studies on glucose homeostasis in GH deficient adults have generated results, which at first glance may appear contradictory. Insulin resistance may be more prevalent in untreated GH deficient adults (29, 30), whereas the impact of GH replacement on this feature seems to depend on the duration and the dose. Below, some of the metabolic effects of GH in human subjects, with special reference to the interaction between glucose and lipid metabolism, will be reviewed.

Almost forty years ago it was shown that infusion of high dose GH into the brachial artery of healthy adults reduced forearm glucose uptake in both muscle and adipose tissue (31). This was parallelled by a drop in RQ and an increase in muscle uptake of FFA, both of which suggest oxidation of FFA by the muscle. This pattern was opposite that of insulin, and co-administration of insulin and GH resulted in only minimal changes in net fluxes of glucose and FFA across the forearm bed. These studies clearly indicated direct insulin antagonistic effects of GH on muscle and adipose tissue.

The introduction of reliable radioimmunoassays for GH revealed the pulsatile and episodic nature of GH release (32) now known to be generated by alternating secretion of GHRH and SST. A GH pulse is released roughly every second hour with a mean daily secretion of 0.5 mg (33). Apart from a well-known circadian variation in terms of elevated nocturnal GH levelsduring the early hours of sleep, GH secretion is amplified during fasting and stress, whereas meals suppress GH release. We studied the metabolic effect of a physiological GH bolus in the postabsorptive state, and demonstrated stimulation of lipolysis following a lag time of 2-3 hours to be the most consistent effect (34). Plasma glucose, on the other hand exhibited only minimal fluctuations, and serum insulin and C-peptide levels remained completely stable. This was associated with subtle reductions in muscular glucose uptake and oxidation, which could reflect substrate competition between glucose and fatty acids (i.e. the glucose/fatty acid cycle). In line with this, sustained exposure to high GH levels induces both hepatic and peripheral (muscular) resistance to the actions of insulin on glucose metabolism together with increased (or inadequately suppressed) lipid oxidation. Apart from enhanced glucose/fatty acid cycling, it has been shown that GH induced insulin resistance is accompanied by reduced muscle glycogen synthase activity (35) and diminished glucose dependent glucose disposal (36). Bak et al. (35) also demonstrated insulin binding and insulin receptor kinase activity from muscle biopsies to be unaffected by GH.

Active acromegaly clearly unmasks the diabetogenic properties of GH. In the basal state plasma glucose is elevated despite compensatory hyperinsulinemia. In the basal and insulin-stimulated state (euglycemic glucose clamp) hepatic and peripheral insulin resistance is associated with enhanced lipid oxidation and energy expenditure (37). There is evidence to suggest that this hypermetabolic state ultimately leads to beta cell exhaustion' and overt diabetes mellitus (38), but a more recent study have demonstrated that the abnormalities are completely reversed after successful surgery (37). Conversely, it has been shown that only two weeks administration of GH in supraphysiological doses (8 IU/day) induces comparable acromegaloid - and reversible - abnormalities in substrate metabolism and insulin sensitivity (39).

Regardless of the exact mechanisms, the insulin antagonistic effects may cause concern when replacing adult GH deficient patients with GH, since some of these patients are insulin resistant in the untreated state. There is evidence to suggest that the direct metabolic effects on GH may be balanced by long-term beneficial effects on body composition and physical fitness, but some studies report impaired insulin sensitivity in spite of favourable changes in body composition. There is little doubt that these effects of GH are dose-dependent and may be minimised or avoided if an appropriately low replacement dose is used. Still, the pharmacokinetics of s.c. GH administration is unable to mimick the endogenous GH pattern with suppressed levels after meals and elevations only during postabsorptive periods, such as during the night. This may be considered the natural domain of GH action which coincides with minimal beta-cell challenge. This reciprocal association between insulin and GH and its potential implications for normal substrate metabolism was initially recognised by Rabinowitz & Zierler (47) . The problems arise when GH levels are elevated during repeated prandial periods. The classic example is active acromegaly, but prolonged high dose s.c. GH administration may cause similar effects. Subcutaneous administration of GH in the evening probably remains the best compromise between effects and side effects (48), but it is far from physiological. We know and understand that hypoglycaemia is a serious and challenging side effect of insulin therapy as a consequence of inappropriately high insulin levels (during fasting). As a corollary, we must realise that hyperglycaemia may result from GH therapy. It is therefore important to carefully monitor glucose metabolism and to use the lowest effective dose when replacing adults with GH.

The clinical picture of acromegaly and gigantism includes increased lean body mass of which skeletal muscle mass accounts for approximately 50 %. Moreover, retention of nitrogen was one of the earliest observed and most reproducible effects of GH administration in humans (1). Thoroughly conducted studies with GH administration in GH deficient children using a variety of classic anthropometric techniques strongly suggested that skeletal muscle mass increased significantly during treatment (49, 50). Indirect evidence of an increase in muscle cell number following GH treatment was also presented (49).

These early clinical studies were paralleled by equally impressive experimental studies in rodent models. GH administration in hypophysectomised rats increased not only muscle mass, but also muscle cell number (i.e. muscle DNA content) (49). Interestingly, the same series of experiments revealed that work-induced muscle hypertrophy could occur in the absence of GH. The ability of GH to stimulate RNA synthesis and amino acid incorporation into protein of isolated rat diaphragm suggested direct mechanisms of actions, whereas direct effects of GH on protein synthesis could not be induced in liver cell cultures (51). Another important observation of that period was made by Goldberg, who studied protein turnover in skeletal muscle of hypophysectomised rats with 3H-leucine tracer techniques. In these studies it was convincingly demonstrated that GH directly increased the synthesis of both sarcoplasmic and myofibrillar protein without affecting proteolysis (52).

The most substantial recent contributions within the field derive from human in vivo studies of the effects of systemic and local GH and IGF-I administration on total and regional protein metabolism by means of amino acid isotope dilution techniques. Horber and Haymond demonstrated that systemic GH administration for 7 days in normal adults increased whole body protein synthesis without affecting proteolysis (53), and similar data were subsequently obtained in GH deficient adults (54). Fryburg and Barret (55) infused GH (systemically for 8 hours) in normal adults and reported an acute stimulation of forearm (muscle) protein synthesis without any effects on whole body protein synthesis. By contrast Copeland and Nair (56) observed an acute stimulatory effect of GH on whole body protein synthesis, but no stimulatory effect on leg protein synthesis, in a design that also included co-administration of somatostatin to suppress insulin. Finally, Fryburg et al. (57) infused GH into the brachial artery, which was accompanied by a local increase in forearm muscle protein synthesis.

Based on these recent studies it seems that the nitrogen retaining properties of GH predominantly involve stimulation of protein synthesis without affecting (lowering) proteolysis and clues are also provided about the underlying mechanisms. Theroretically, the protein anabolic effects of GH could be either direct, or mediated through IGF-I, insulin or lipid intermediates. GH receptors are present in skeletal muscle (58), which combined with Fryburgs intra-arterial GH studies, makes a direct GH effect conceivable. An alternative interpretation of Fryburgs data could be that GH stimulates local muscle IGF-I release, which subsequently acts in an autocrine/paracrine manner. The effects of systemic IGF-I administration on whole body protein metabolism seem to depend on ambient amino acid levels in the sense that IGF-I administered alone suppresses proteolysis (59) whereas IGF-I in combination with an amino acid infusion increase protein synthesis (60). Moreover, intra-arterial IGF-I in combination with systemic amino acid infusion increased protein synthesis (61). It is therefore likely that the muscle anabolic effects of GH at least to some extent are mediated by IGF-I. By contrast, it is repeatedly shown that insulin predominantly acts through suppression of proteolysis and this effect(s) appears to be blunted by co-administration of GH (62). The degree to which mobilisation of lipids contributes to the muscle anabolic actions of GH has so far not been specifically investigated.

In conclusion several experimental lines of evidence strongly suggest that GH stimulates muscle protein synthesis. This effect is presumably in part mediated through binding of GH to GH receptors in skeletal muscle. This does not rule out a significant role of IGF-I being produced either systematically or locally.

A interesting recent discovery has been that infusion of GH and IGF-I into the brachial artery increase forearm blood flow several fold (57, 63). This effect appears to be mediated through stimulation of endothelial nitric oxide release leading to local vasodilatation (64, 65). Moreover, co-infusion of a nitric oxide inhibitor with IGF-I appeared to blunt the stimulatory effect of IGF-I on forearm protein synthesis (64). It thus appears that an IGF-I mediated increase in muscle nitric oxide release accounts for some of the effects of GH on skeletal muscle protein synthesis. These intriguing observations may have many other implications. It is, for instance, tempting to speculate that this increase in skeletal muscle blood flow contributes to the GH induced increase in resting energy expenditure, since skeletal muscle metabolism is a major determinant of REE (66). Moreover, it is plausible that the reduction in total peripheral resistance seen after GH administration in GHDA is mediated by nitric oxide (65).

As previously mentioned the ability of acute and more prolonged GH administration to retain nitrogen in normal adults has been known for decades and more recent studies have documented a stimulatory effect on whole body and forearm protein synthesis.

Rudman et al. was the first to suggest that the senescent changes in body composition were causally linked to the concomitant decline in circulation GH and IGF-I levels (66). This concept, which is known by some as the somatopause, has recently been reviewed (67), and a number of studies with GH and other anabolic agents for treating the sarcopenia of ageing are currently in progress.

Placebo-controlled GH administration in young healthy adults (21-34 years) undergoing a resistance exercise programme for 12 weeks showed a GH induced increase in LBM, whole body protein balance and whole body protein synthesis, whereas quadriceps muscle protein synthesis rate and muscle strength increased to the same degree in both groups during training (68). In a similar study in older men (67 years) GH also increased LBM and whole body protein synthesis, without significantly amplifying the effects of exercise on muscle protein synthesis or muscle strength (69). An increase in LBM but unaltered muscle strength following 10 weeks of GH administration plus resistance exercise training was also recorded by Taafe et al. (70). A more recent study of 52 older men (70-85 years) treated with either GH or placebo for 6 months, without concomitant exercise, observed a significant increase (4.4 %) in LBM with GH, but no significant effects on muscle strength (71). Thus no significant clinical benefit from administrating GH to non-GH-deficient senescent patients has been documented yet.

Numerous studies have evaluated the effects of GH administration in chronic and acute catabolic illness. A comprehensive survey of the prolific literature within this field is beyond the scope of this review, but it is noteworthy, that HIV-associated body wasting is a licensed indication for GH treatment in the USA. In this patient category GH treatment for 12 weeks has been associated with significant increments in LBM and physical fitness (72, 73).