The Insulin-Like Growth Factor System and Nutritional Assessment

Over recent years there has been considerable interest in the role of the insulin-like growth factor (IGF) system in health and disease. It has long been known to be dysregulated in states of under- and overnutrition, serum IGF-I levels falling in malnourished patients and responding promptly to nutritional support. More recently, other proteins in this system have been observed to be dysregulated in both malnutrition and obesity. Currently no biochemical marker is sufficiently specific for use in screening for malnutrition, but levels may be valuable in providing information on nutritional status and in monitoring of nutritional support. All have limitations as nutritional markers in that their serum levels are influenced by factors other than nutritional status, most importantly the acute phase response (APR). Levels should be interpreted along with clinical findings and the results of other investigations such as C-reactive protein (CRP). This paper reviews data supporting the use of proteins of the IGF system as nutritional markers.


Introduction
e insulin-like growth factor (IGF) system consists of the peptide hormones IGF-I and -II, their cell surface receptors, and IGF-binding proteins (IGFBPs). IGF-I and -II are the principal components of the system. ey have biological roles in mediating the effects of growth hormone (GH) and during development [1]. eir actions can be divided into rapid metabolic effects and longer-term growth-promoting effects [2]. Compared to insulin both circulate in relatively high concentrations in serum. ey are also released into interstitial �uids where they act locally in a paracrine or autocrine manner. e IGFs generate their biological responses by activating a receptor tyrosine kinase (RTK) which initiates an intracellular signalling cascade [3]. eir function is modulated by the IGFBPs [4].

e IGFs and eir
Receptors. IGF-I exerts acute anabolic actions on protein and carbohydrate metabolism by increasing the cellular uptake of amino acids. As such, it is an important regulator of muscle mass. It has long-term effects on cell proliferation, differentiation, and apoptosis [4,5]. It is a potent mitogen increasing DNA synthesis and accelerating cell cycle progression. IGF-I is predominantly synthesised in liver, in response to GH, but is also synthesised to a lesser extent in various other tissues. Other factors which regulate its production include genetic factors, nutrition, and other hormones such as thyroxine, gonadal steroids, and cortisol. Its biological half-life in serum is only a few hours. Most serum IGF-I is bound to IGFBP-3 and other proteins of this family. A small proportion (<1%) which exists as free IGF-I is thought to be the bioactive component. Serum IGF-I levels are low at birth and increase through puberty to the age of 20 years, thereaer declining gradually with age [6]. Levels must therefore be interpreted using age-related reference ranges. Disease-related dysregulation of IGF-I may affect local or systemic actions of IGF-I. Localised growth disturbance may occur, for example, in bone disease where the pathology is con�ned to a speci�c region [7][8][9]. Disturbance of systemic IGF-I actions is implicated in metabolic disorders such as diabetes, malnutrition, obesity, acromegaly, and GH de�ciency. Its measurement in serum has an established place in the clinical management of the latter two conditions. IGF-II is a polypeptide synthesised mainly in liver which, in its mature form, is about 70% identical to IGF-I [10]. In comparison to IGF-I, its physiological role and regulation are relatively poorly understood but it is thought to be involved in foetal differentiation, metabolic regulation, cell growth, and cell division, being a more potent mitogen than IGF-I. e principal regulator of IGF-II gene transcription is unclear, GH having no effect on its expression in humans, but its expression is in�uenced by various other hormones and growth factors [5,11]. In serum, IGF-II occurs at levels around three times higher than those of IGF-I. e concentration of biologically active free IGF-II parallels that of total IGF-II. IGF-II has been observed to be dysregulated in a variety of disease states notably malnutrition, obesity, and non-islet cell tumour hypoglycaemia (NICTH), a rare condition in which tumours secrete excessive IGF-II [12][13][14]. IGF-II is less oen measured than IGF-I in clinical contexts, the only recognised indication for its measurement being investigation of suspected NICTH.
e IGFs bind principally to two cell surface transmembrane receptors, namely, the IGF receptor type 1 (IGF-1R) and IGF receptor type 2 (IGF-2R), but they also bind to the insulin receptor (I-R) [15]. e acute and long-term biological actions of the IGFs are mediated via binding to the IGF-1R and activation of its RTK activity [3,16]. e IGF-2R has no signalling activity. It is thought to regulate IGF-II availability to receptors by sequestering it and enabling its degradation [10]. (IGFBPs). e IGFBPs are a family of six proteins which bind the IGFs thereby regulating their availability to receptors. ey are synthesised in liver, connective tissue, and various other cell types [7]. e individual proteins have distinct binding preferences, IGFBP-2, for example, binding preferentially to IGF-II [17]. e functions of the IGFBPs in vivo are only partly understood, but suggested functions and other features of the IGFBPs are listed in Table 1.

IGF Binding Proteins
In circulation they are present predominantly as 150 kDa ternary complexes consisting of IGF-I or -II, IGFBP-3, and acid-labile subunit (ALS), a protein synthesised mainly in liver [11]. e remainder of circulating IGFs are present mainly as binary complexes with IGFBPs or in the free form [2]. Ternary complexes are unable to cross endothelia and so extend the half-life of IGFs in the circulation. Various mechanisms regulate release of IGFs from these complexes allowing interaction with receptors [18]. Proteases have been identi�ed which can fragment IGFBPs enhancing IGF release [7]. A variety of catabolic states are known to be accompanied by increased proteolysis of IGFBP-3. IGFBPs are also degraded by matrix metalloproteinases. Release of the IGFs from their binding proteins is also regulated by IGFBP phosphorylation and cell surface association [19].

Malnutrition in Hospitalised Patients
Malnutrition is common amongst hospital patients, in whom it predisposes to disease, increases the risk of complications, and delays recovery. As well as being associated with an adverse outcome, these factors tend to increase the length of stay (LOS) in hospitals which has considerable cost implications. It is therefore vital to detect malnutrition as early as possible in order that nutritional support can be commenced. Nutritional support buys time for therapeutic interventions to aid recovery and reduces morbidity and mortality [20]. It can be delivered orally, enterally, or parenterally depending on the clinical situation. e �rst two routes are the more physiological, parenteral nutritional (PN) being reserved for patients with intestinal failure in whom the gut is either dysfunctional or inaccessible.
It is well recognised that malnutrition in hospitals is oen overlooked and consequently undertreated. Nutritional screening is therefore widely advocated by professional bodies, so that patients at risk of malnutrition can be identi�ed [21]. Whilst a number biochemical tests are sensitive for the detection of malnutrition, their diagnostic speci�city is too low for use in screening. Currently, nutritional screening is carried out using screening tools such as the malnutrition universal screening tool (MUST) [22].

Nutritional Assessment of Patients
Once the decision has been taken to provide nutritional support for an individual, it is essential that their nutritional requirements are met if the full bene�t is to be realised. When patients are referred for nutritional support, the dietitian should carry out a formal nutritional assessment, part of which should include an estimation of the patient's energy and protein requirements. Energy requirements can be calculated based on body weight and age using a method such as the Scho�eld equation to determine the approximate basal metabolic rate (BMR) [23,24]. Protein requirements can be calculated from body weight. ese methods only provide estimates which can be signi�cantly altered in, for example, critical illness, and it should be borne in mind that the �nal arbiters of the success of nutritional support are weight restoration and recovery. Following commencement of nutritional support, individual patients must therefore be closely monitored for evidence of nutritional repletion, irrespective of whether estimated requirements have been met. Where the patient is not recovering, consideration needs to be given to adjust the nutritional support provision. e National Institute for Health and Clinical Excellence (NICE) guidelines for nutritional support in adults (2006) provide guidance on the monitoring of patients [25]. Both anthropometric measurements and the results of biochemical tests contribute to this process.

Protein Markers in Nutritional Assessment
Biochemical markers can be useful adjuncts in the nutritional assessment and monitoring nutritional support, though their clinical utility is currently a topic of much debate and research. Markers currently available include transferrin, prealbumin, and retinol-binding protein, the serum levels of which may be subnormal in malnutrition. e current NICE guidelines do not advocate measurement of these as part of the laboratory monitoring of nutritional support [25], but their levels have been widely measured in research studies investigating malnutrition. Novel biochemical markers are likely, in time, to �nd a place in clinical management of patients. However, the assessment of novel nutritional markers is complicated in that there is no universally accepted "gold standard" method for nutritional assessment, making it difficult to de�ne the diagnostic performance of new markers. Instead, studies have compared novel markers to methods such as detailed nutritional assessment and subjective global assessment [26].

Limitations of Protein Markers.
All nutritional marker proteins in current use suffer from the limitation that factors other than nutritional status in�uence their serum concentrations, for instance, hydration status, disease, surgery, and abnormal liver function [27]. In practice it is difficult to separate the in�uence of these factors from that of malnutrition. e most problematic confounder is the acute phase response (APR) which causes decreased levels of these proteins [28]. During the APR, cytokines such as interleukin-6 (IL-6) precipitate a decline in the protein concentrations over a characteristic time course [29]. ey do this by increasing capillary permeability, increasing protein catabolism, and reducing the hepatic synthesis of proteins [30]. e magnitude of the reduction in levels correlates with the severity of the injury [31,32]. ese changes occur independently of total body protein status and irrespective of whether the patient is receiving adequate nutritional support [29]. Levels of nutritional marker proteins tend to be inversely related to the levels of APR markers such as C-reactive protein (CRP). Measurement of CRP is therefore of value in interpretation. e APR is particularly problematic in hospitalised patients where most patients in need of nutritional support are also acutely ill. Nutritional markers may therefore �nd greater use in outpatients or in the context of epidemiological research where subjects are relatively well. eir responsiveness to changes in nutritional status may make them of more value in monitoring than in baseline assessment of the patient. Although the available markers may be poor at detecting nutritional status during acute illness, they are recognised to have prognostic value and as such may be useful parameters T 2: Features of the ideal protein-energy marker.
Short biological half life �xists primarily within an accessible body �uid Limited homeostatic regulation Constant rate of catabolism �nin�uenced by vitamin or mineral status �nin�uenced by pathology other than malnutrition Measurement simple, cheap, and available locally to include in a biochemical pro�le in patients referred for nutritional support. is prognostic value relates to their place as negative APR markers. It is worthwhile to consider the characteristics of the ideal nutritional marker which would not be subject to any of the limitations discussed ( Table  2). Whilst the ideal marker does not exist, it is useful to keep this list in mind when considering the merits of current or novel markers. e remainder of this paper focuses on proteins of the IGF axis as potential markers of nutritional status.

The Role of IGF Axis Proteins in Managing Malnutrition and Nutritional Assessment
Serum levels of proteins of the IGF system have been widely reported to be altered in malnutrition. In view of this, there is increasing interest in their potential clinical utility in nutritional assessment of patients and in monitoring nutritional support provision. Table 3 summarises the effects of malnutrition and obesity on levels of the IGFs and IGFBPs.
To date IGF-I is the most extensively researched member of this family [33][34][35].

IGF-I.
Currently, the main clinical use of IGF-I measurement is in the assessment of pituitary GH status, that is, to detect de�ciency in the case of hypopituitarism and excess in acromegaly. However, its serum level has also long been recognised to re�ect nutritional status, declining during fasting and starvation [46,61]. Following one week of starvation, levels are comparable to those observed in hypopituitarism. It correlates with anthropometric indices such as BMI and the levels of other protein nutritional marker proteins [62]. e decline in its levels appears to be greater in those with protein energy malnutrition as compared to protein malnutrition alone [63]. e precise mechanism for this decline is not known but likely re�ects the decrease in hepatic protein synthesis which accompanies malnutrition. Its physiological purpose may be to divert substrates away from growth and towards metabolic homeostasis, a process assisted by the relatively insulin-resistant state which exists in starvation. It may also contribute to the catabolic reaction observed in starvation and sepsis. Conversely, IGF-I levels return to normal in response to nutritional support during which they have been observed to climb incrementally in relation to weight gain [46,64,65]. e short serum half life of IGF-I renders it more sensitive than previous markers to short-term changes in nutritional status, an attribute which increases its value in monitoring [66]. Moreover, serum IGF-I appears to be sensitive to both the amount and type of fat provided in nutritional support. Fish oil and low-fat solutions were signi�cantly correlated to serum IGF-I, appearing to promote a more rapid recovery of its levels [67]. As well as being used as a marker of repletion, monitoring the IGF-I level may help to avoid overfeeding of patients and the adverse consequences thereof. Although it may have a place in monitoring, IGF-I like the older nutritional markers performs poorly as a screening test for malnutrition [30]. IGF-I is considered in more detail below in various clinical situations associated with malnutrition.

Anorexia Nervosa.
Malnourished patients with anorexia nervosa (AN) oen have multiple endocrine abnormalities, including amenorrhoea, hypothyroidism, hypercortisolism, and elevated G�, with suppressed IGF-I re�ecting G� resistance [34]. ese abnormalities are believed to be secondary to nutritional deprivation as they also occur in starvation states secondary to other causes and return to normal upon weight restoration [68]. In patients with eating disorders, it has been demonstrated that serum IGF-I levels are closely related to BMI, body fat, and body muscle mass, re�ecting the severity of nutritional depletion [36]. When treating AN, the aim is nutritional rehabilitation. Given that the serum IGF-I level responds to nutritional support, it has been investigated as an indicator of nutrition during treatment of adolescents with eating disorders [37]. Where there was net weight gain during treatment, IGF-I increased in parallel with the BMI standard deviation score, a measure of leanness. e authors concluded that IGF-I can be considered an indicator of nutritional status in these patients. It is sensitive to short-term weight changes and can be used at assessment and to monitor nutritional rehabilitation.
Assays are now available for free IGF-I enabling it to be measured in clinical studies. In common with total IGF-I, its levels fall in undernutrition [50,51]. Free IGF-I has been reported to be more sensitive to short-term nutritional status than total IGF-I [69]. It is usually estimated by measurement of immunoreactivity of IGF-I in serum following extraction of IGFBPs. Although this method is valid in many cases, there are conditions such as chronic renal disease, obesity, and during fasting in which directly measured IGF-I bioactivity may more closely re�ect the endogenous IGF-I bioactivity.
A study was carried out in which IGF-I bioactivity was measured by assaying IGF-1R RTK activation. is was compared to total IGF-I and free IGF-I. IGF-I bioactivity was signi�cantly reduced in patients with AN along with the two other parameters [1], all three parameters being closely correlated. is suggests that total IGF-I may considered as a surrogate marker for the IGF-I bioactivity. Whether these �ndings apply to malnourished states other than AN remains to be established. Studies of patients with AN have shown no change in IGFBP-3 proteolytic activity [50,51]. More research needs to be carried out into the utility of free IGF-I as a nutritional marker.

Paediatrics.
Numerous studies have investigated the utility of IGF system proteins as nutritional markers in children. In malnourished children, serum IGF-I correlates strongly with height SD score suggesting that it is a useful indicator of growth and nutritional status [57]. It has also been reported to correlate with economic status independently of overt malnutrition [70].
Studies have investigated IGF-I as a marker of lean body mass (LBM) in malnourished children with various medical disorders. In cystic �brosis (CF), the serum IGF-I level correlated with LBM, as evaluated by DEXA scanning, independently of weight [71]. e signi�cance of this is that reduction in LBM can impair respiratory function thereby worsening the clinical outcome for these patients. IGF-I could therefore be used as be a tool for identifying patients with CF at risk of deterioration in respiratory function. In a separate study of children with CF, decreased IGF-I levels were shown to re�ect growth retardation [58]. IGF-I was also used as a marker of LBM in a recent study of children commencing treatment for HIV infection [72]. During treatment, improved muscle mass, but not linear growth, was associated with normalised IGF-I concentrations. In children with congenital heart disease (CHD), the most important factor in�uencing serum IGF-I levels was the presence of cyanosis, levels being signi�cantly lower in cyanotic compared to acyanotic patients [73]. e authors suggested that chronic hypoxia plays a signi�cant role in the pathogenesis of malnutrition, re�ected by serum IGF-I levels. Moreover, IGF-I de�ciency in these patients may be responsible for the decrease in le ventricular mass.
Serum IGF-I has been used for monitoring children with short bowel syndrome (SBS) receiving PN, its level having been shown to improve in line with nitrogen balance [74]. It may therefore be of value along with other measurements in assessing protein nutritional status of these children. Not all children with SBS respond equally to standard nutritional approaches, and it is important to identify the subgroup who may respond poorly. IGF-I has been reported as an index of intestinal failure in children with SBS who merit more aggressive therapeutic intervention [75]. Intestinal transplantation can potentially liberate children with intestinal failure from the need for long-term PN, but it is difficult to achieve sufficient growth post-transplant in most patients. Negative linear growth velocity is predicted by low levels of IGF-I pre-transplant [76]. IGF-I levels may therefore be useful in identifying those patients requiring an intensive nutritional regimen aer transplant. Recombinant growth hormone (rhGH) is another treatment that may reduce the amount of PN required in patients with SBS. is has recently been trialled in PN-dependent children [77]. Although they remained on PN, there was a signi�cant increase in the amount of nutrition that could be delivered enterally rather than intravenously. is paralleled an increase in their energy balance over a period of 12 months following the treatment and serum IGF-I levels increased signi�cantly.

Critical Illness.
Critical illness, in common with malnutrition, sepsis, and aging, is associated with a dramatic reduction in the mass and functional status of skeletal muscle which results from protein catabolism. is leads to muscle weakness and prolonged mechanical ventilation, factors which tend to increase LOS in ITU. ere are a number of hormonal changes which occur in catabolic states including an increase in serum GH along with GH resistance leading to a reduction in both circulating IGF-I and its expression in skeletal muscle. ere is also an increase in insulin resistance. In common with other protein markers, IGF-I is negatively in�uenced by in�ammation, injury, and burns [28,[78][79][80]. Although the mechanism of this decrease is unclear, there is evidence that cytokines locally inhibit IGF-I expression, thereby contributing to catabolism [81].
In view of the role of IGF-I in in�uencing muscle mass and the pathophysiological role of its de�ciency in critical illness, there has been interest in use of treatment with rhGH to increase IGF-I levels therapeutically. is treatment has been demonstrated to result in increased IGF-I levels with nitrogen retention and decreased LOS. However, two large randomised trials in 1999 reported increased mortality associated with infection and organ dysfunction [82]. Small clinical trials have reexamined rhGH supplementation in prolonged critical illness, demonstrating increased serum levels of IGF-I and IGFBP-3 in patients receiving adequate nutritional support. In a recent trial of 30 multiple trauma patients with prolonged critical illness, rhGH treatment delivered intravenously in pulses normalised IGF-I levels [83]. It is possible that rhGH supplementation may be safer and more efficacious in certain critically ill patients, but the subgroup likely to bene�t needs to be de�ned. A clearer understanding of the GH resistance should lead to the development of new therapeutic strategies aimed at restoring the bene�cial effects of anabolic agents such as GH and IGF-I. If in time the balance of evidence supports GH treatment in some critically ill patients, there may be a place for monitoring IGF-I levels. Current guidelines continue to recommend against therapeutic use of rhGH in critical illness [84].

Obesity.
Obesity is well recognised to in�uence the IGF system though the effects are relatively poorly understood [85]. An increase in BMI is associated with a reduction in GH secretion, but IGF-I does not appear to fall to the same extent. ere is disagreement regarding the effect of obesity on IGF-I levels. Studies have reported IGF-I levels to be normal [38], high [39,40], or low [41,42]. Similarly there is variation between studies as to what happens to free IGF-I levels. ese have been reported as normal [43], high [44,45,48,49], or low [52], though the majority of studies suggest high levels. When IGF levels have been investigated in large populations, serum IGF-I levels appear to follow an inverted U-shaped association with BMI. Maximal IGF-I has been reported at BMI of 22-24 kg/m 2 [48], 22.5-29.5 kg/m 2 , [60] and 30-35 kg/m 2 [86] though the changes in IGF-I levels across the normal-to-high BMI range are relatively small. e elevation of free IGF-I in the obese at high BMI is a phenomenon that may be con�ned to nondiabetic subjects. In obese subjects with type 2 diabetes, free IGF-I was not signi�cantly elevated when compared to lean controls [44]. Subjects with type 1 diabetes had a marked reduction (>50%) in free IGF-I [44]. More work is required to con�rm these �ndings and to assess the mechanistic basis and consequences of overnutrition on free IGF-I levels.
Numerous studies have investigated how therapeutic interventions in obesity in�uence IGF-I levels. ese are more consistent in showing a reduction in levels in response to a variety of approaches, namely, caloric restriction [45], gastric banding [87], physical activity [88], and a combination of dietary restriction and exercise training [89]. Whilst these �ndings are interesting, measurement of IGF-I in the context of obesity is currently con�ned to research studies.

Scienti�ca
At present its measurement would not appear to be of value in guiding clinical management of obese patients. Allied to the study of IGF levels in obese subjects is an increasing interest in the link between obesity and cancer. It has been suggested that dysregulation of the IGF system plays a major role in this link [90]. Research will no doubt continue in this area, seeking to determine the underlying mechanisms, but it remains to be established whether measurement of IGF-I is of clinical value in predicting or monitoring mortality risk.

IGF-II.
Relatively few studies have investigated IGF-II in nutritional contexts, but, in common with IGF-I, it has generally been observed to be dysregulated in both under-and overnutrition, its serum level falling in chronic malnutrition and climbing during refeeding [34,46]. However, one study investigating short-term fasting observed no change in serum IGF-II [47]. IGF-II levels are reported to be elevated in obesity [49]. Following weight reduction in these subjects, both serum total and free IGF-II levels decreased signi�cantly. More work is required to examine the physiological regulation of IGF-II and its dysregulation in disease. e utility, if any, of its measurement in nutritional contexts remains to be established.

IGFBP-1.
IGFBP-1 is expressed predominantly by liver and is abundant in blood [91], its expression being regulated by insulin which inhibits transcription of the IGFBP-1 gene. Its mRNA has a half life of only 2 hours which results in the level of IGFBP-1 in blood varying reciprocally with insulin levels [92]. However, IGFBP-1 levels lag behind insulin levels re�ecting the longer half life of IGFBP-1 compared to insulin. is means that measurement of IGFBP-1 levels can be used to indirectly assess insulin secretion [91]. ese �ndings have led to interest in IGFBP-1 as a marker of insulin resistance. In support of its use in this context, serum IGFBP-1 levels have been shown to correlate with the gold standard method for assessing insulin resistance, namely, the hyperinsulinaemic, euglycaemic clamp [93]. Low fasting serum IGFBP-1 levels correlate with hyperinsulinaemia and a variety of other cardiovascular risk factors, including increased BMI, fasting hypertriglyceridaemia, and low HDLcholesterol concentrations all of which are features of the metabolic syndrome [94]. Indeed, its serum level has been suggested as a marker of metabolic syndrome in adults [53] and to be predictive of the development of insulin resistance in children [54]. Moreover, its levels increase in response to lifestyle interventions indicative of the reversible nature of insulin resistance upon life-style change [95]. In a study comparing it with other simple indices of insulin resistance in normal glucose-tolerant subjects, IGFBP-1 was suggested to be the most reliable marker [96]. It has the advantage of being a relatively stable analyte compared to insulin, thus having the potential to assess insulin resistance by means of a single fasting blood test. It may therefore have clinical utility in the assessment of cardiovascular risk in patients with metabolic syndrome and in monitoring the effect of therapeutic interventions.
IGFBP-1 has been less extensively studied in the context of undernutrition, but levels have been observed to be markedly elevated in patients with AN [51]. is is to be expected since low insulin levels during prolonged fasting have a permissive effect on IGFBP-1 gene transcription. It may be that the increase in IGFBP-1 during fasting acts as a protective mechanism against hypoglycaemia by binding IGF-I and thereby preventing its action. IGFBP-1 levels have been observed to be elevated and predictive of growth failure in patients with chronic renal failure [97].

IGFBP-2.
Relatively few studies have investigated IGFBP-2 in the context of malnutrition, but it appears that its serum levels rise in undernutrition and fall following recovery [34,59,102]. When compared with other proteins of the IGF system in a study of patients with AN, it was reported to be the best predictor of BMI [102]. In a large population cross-sectional study, IGFBP-2 levels decreased steadily with increasing BMI; the principle dietary factor determining its levels believed to be dairy protein [103]. IGFBP-2 is also of interest in metabolic syndrome and obesity, in which its serum levels fall [55]. Indeed, in a large population study, the reduction in IGFBP-2 levels associated with adiposity was reported to be the largest effect of adiposity observed on the IGF system [86]. Moreover, it has been suggested as an independent predictor of insulin resistance in a study comparing it to the hyperinsulinaemic euglycaemic clamp [56]. Although less well studied than IGFBP-1 in this context, it is potentially a more useful marker as its levels do not �uctuate in an insulin-dependent manner. Aside from nutritional contexts, IGFBP-2 levels are known to climb in NICTH and in other situations where the free IGF-II level is expected to be high [100]. Investigation of NICTH is currently the only established indication for its measurement.

IGFBP-3.
IGFBP-3 is the most abundant member of this family of proteins in serum. Its concentration generally parallels that of IGF-I both physiologically and in other clinical conditions and it is commonly measured along with the latter in the assessment of poor growth in children [104]. In studies of malnourished patients in which IGF-I levels have been observed to fall, the IGFBP-3 level also falls and recovers upon refeeding [57][58][59]. ere is a consistent relationship between serum IGFBP-3 concentration and caloric intake. One study suggested that it was a more sensitive marker of nutritional status than IGF-I [58]. As such, its levels have potential utility in monitoring the adequacy of nutritional support regimens. In obesity the serum IGFBP-3 level appears to climb being positively associated with central adiposity though, unlike in malnutrition, its level does not necessarily parallel that of IGF-I [60]. Derangement of thyroid status is a potential confounding factor in the interpretation of IGFBP-3 levels. Serum IGFBP-3 and IGF-I levels correlate positively with free thyroxine levels, likely because thyroid hormones have a role in regulating the synthesis of these proteins [57].

Clinical Utility of IGF-I and Other Markers
e attraction of biochemical markers to the user is their objectivity. As such, they can complement clinical nutritional assessment, which is largely subjective in nature. is has prompted researchers to try to identify novel markers. To date a considerable evidence base supports the utility of IGF-I as a nutritional marker, its short half life in serum rendering it a more sensitive indicator of nutritional depletion than earlier markers. e main role for IGF-I measurement would appear to be in monitoring nutritional support in malnourished patients especially those in whom acute illness and liver disease can be excluded. It is unclear whether the levels of IGF proteins can be used to help de�ne recovery in these patients and what should be considered the target level. A premorbid value could be used as the target but is unlikely to be available. Although age-related reference ranges are available, normal values vary signi�cantly between individuals. A value toward the lower end of the reference range, for example, may be normal for one individual but not for another. Clearly when used in monitoring, IGF-I levels should be considered in clinical context along with anthropometric measurements and results of other biochemical tests. It has been suggested that relative changes in IGF-I may be more useful than absolute values [36]. Another potential use for IGF is in the prediction of relapse in AN. In this situation, body weight alone may be misleading as it re�ects �uid as well as body cell mass. Investigation of this possible utility would require the IGF-I level to be studied prospectively to assess whether its decline precedes that of signi�cant weight loss. Refeeding syndrome (RS) can occur as a complication following commencement of nutrition support in malnourished patients, a major component of which is hypophosphataemia [105,106]. Although guidelines make recommendations on how to recognise patients at risk of RS [25], prediction is difficult in practice and there is a need for better objective markers of risk. A recently published study on patients receiving PN examined three parameters as potential predictors of RS, namely, serum IGF-I measured before commencement of PN, the leptin level, and a "refeeding index" (RI) derived from the other two values [98]. Although RI predicted a fall in the serum phosphate level more sensitively and speci�cally than the other two parameters, IGF-I was a better predictor of mortality.
Serum IGF-I has signi�cant limitations as a nutritional marker, notably that its level is likely to be reduced during the APR regardless of the nutritional state. e underlying disease must therefore be taken into account as possibly in�uencing its levels. Its level is also lowered in disease of the liver, the main site of IGF-I synthesis, and can be elevated or lowered in renal disease due to changes in IGFBP levels [107]. IGF-I levels have been reported to be low in patients with untreated severe hypothyroidism but increased signi�cantly following thyroxine treatment [108]. is re�ects the regulatory in�uence of thyroxine on IGF-I levels. Its serum level is also in�uenced by GH status as discussed above. Whilst its measurement may be useful in selected cases, these confounding factors mitigate against its use generally in hospital patients or in nutritional screening.
e clinical utility of measurement of free IGF-I and other proteins of the IGF system in nutritional contexts is less clear at present. If they are to be of value, the result should lead to clinical action which could potentially improve the outcome for the patient. eir utility needs to be de�ned by future clinical validation studies. Future research on these markers also needs to evaluate interferences and confounding factors and address the question of target values. e prognostic value of the markers needs to be determined, that is, whether they can indicate the likelihood of an adverse outcome in the event that nutritional support is withheld. Potential clinical utility for the measurement of IGF axis proteins is summarised in Table 4.
Validation studies alone are insufficient to ensure that novel tests enter the laboratory repertoire. ey are unlikely to be offered widely unless recommended by clinical guidelines. Even then, logistical factors such as cost and assay availability may preclude their use. Where a test is to be of value in monitoring nutritional support, it is essential that the result is available the same day as the request so that the nutritional regimen can be modi�ed appropriately. is necessitates the assay being sufficiently rapid and available on site. e latter may not be feasible in, for example, a district general hospital where the demand for the test is low. In the case of markers of overnutrition or insulin resistance, turnaround time for results is likely to be less critical as the target population will be mainly outpatients. e logistical problems are likely to be overcome in time as assays become cheaper and more automatable, possibly resulting in their availability as near patient tests.