Leptin a new biological marker for evaluating malnutrition in elderly patients

Transcription

1 (2007) 61, & 2007 Nature Publishing Group All rights reserved /07 $ ORIGINAL ARTICLE Leptin a new biological marker for evaluating malnutrition in elderly patients O Bouillanne 1, J-L Golmard 2, C Coussieu 3, M Noël 4, D Durand 5, F Piette 6 and V Nivet-Antoine 5 1 Service de Gérontologie 2, Hôpital Emile-Roux, Assistance Publique Hôpitaux de Paris (AP-HP), Limeil-Brévannes, France; 2 Service de Biostatistiques, Université Paris 6, Paris, France; 3 Service de Biochimie médicale, Hôpital Hôtel-Dieu, AP-HP, Paris, France; 4 Service de Biochimie Endocrinienne, Hôpital Robert-Debré, AP-HP, Paris, France; 5 Service de Biochimie, Hôpital Charles-Foix, AP-HP, Université Paris 5, Ivry sur Seine, France and 6 Service de Médecine Interne et Gérontologie, Hôpital Charles-Foix, AP-HP, Université Paris 6, Ivry sur Seine, France Background: There is no single universally accepted biochemical marker of nutritional status in the elderly. Many markers are affected by non-nutritional factors. Objective: The purpose of this study was to determine the biological parameters best related to anthropometric markers of malnutrition in an elderly polypathological population, and determine cutoff values for these potential parameters to diagnose malnutrition. Design: This prospective study enrolled 116 elderly hospitalized patients and 76 elderly outpatients. Nutritional status (albumin, transthyretin, body mass index (BMI), skinfold thickness) and biological parameters (leptin, insulin-like growth factor-1 (IGF-1), IGF binding protein-1 (IGFBP-1), IGFBP-3, C-reactive protein (CRP), orosomucoid) were assessed. We defined malnutrition according to the lowest quartile of BMI and skinfold thickness measured in a large healthy elderly French sample population. Results: In this sample of elderly patients (age: 8577 years old), leptin concentration was the only biological parameter significantly related to nutrition status. Independent correlations were found between leptin concentration and BMI, skinfold thickness and sex. The relationship between nutritional status and leptin concentration is significantly different in each sex: the more the patients are undernourished, the lower the leptin concentration in both sexes. The optimal leptin cutoff value for the diagnosis of malnutrition in this population was 4 mg/l in men (sensitivity 0.89, specificity 0.82) and 6.48 mg/l in women (sensitivity 0.90, specificity 0.83). Conclusion: Leptin concentration is highly correlated with anthropometric data whereas albumin or transthyretin are known to be also influenced by morbidity and inflammatory conditions. Serum leptin concentration could be used for nutritional assessment in elderly patients with acute diseases. (2007) 61, doi: /sj.ejcn ; published online 6 December 2006 Keywords: leptin; body mass index; skinfold thickness; elderly; malnutrition Introduction Malnutrition in elderly populations is considered a major public health problem. The Pubmed database includes 12 original articles from 2004 specifically focused on this topic. Correspondence: Dr O Bouillanne, Service de Gérontologie 2, CH Emile-Roux (Assistance Publique Hôpitaux de Paris), Limeil-Brévannes, France. olivier.bouillanne@erx.aphp.fr Contributors: OB, VN-A and FP contributed to the design of the experiment. OB, VN-A contributed to the collection of data. J-LG contributed to the analysis of data. OB, VN-A, FP contributed to the writing of the manuscript. CC, MN, DD provided significant advice. Received 19 January 2006; revised 9 August 2006; accepted 1 October 2006; published online 6 December 2006 The parameters measured always include weight and body mass index (BMI), a large majority include Mini Nutritional Assessment (MNA), albumin and transthyretin, whereas some articles focus on anthropometric data (calf circumference, skinfold thickness) or other biological data (insulin-like growth factor-1 (IGF-1), IGF binding proteins (IGFBPs)) or measurement of body composition. In many studies, parameters are combined in indexes because no single parameter alone is able to diagnose malnutrition in a very old polypathological population. Thus, there is no consensus on the best method for accurately assessing nutritional status in the elderly. Anthropometric measurements are modified by fluid imbalance and many of the classical biological parameters are affected by non-nutritional factors. Albumin and

2 648 transthyretin concentrations are influenced by hemoconcentration and inflammation-dependent. Thus, these parameters are inappropriate as indicators of nutritional status in elderly populations, which often suffer from heart failure (edema) or chronic diseases with inflammatory or/and septic complications (Baxter, 1999). Other biological parameters such as IGF-1 associated with IGFBPs are sometimes used (Janssen et al., 1998; Campillo et al., 2000). Leptin, which has been strongly correlated with BMI and tissue adiposity, is a promising parameter (Friedman and Halaos, 1998). Chronic feeding control depends on leptin released from fat stores. Adipocyte leptin acts on the leptin receptors in the brain by controlling food intake and energy expenditure. Thus, we decided to study a cohort of elderly hospitalized patients and elderly outpatients and measure both anthropometric and biological parameters such as IGF-1, IGFBPs and leptin. The purpose of this study was to determine the biological parameters best related to anthropometric markers of malnutrition in elderly patients and determine cut-off values for these potential parameters to diagnose malnutrition. Methods Subjects One hundred and sixteen consecutive elderly patients were prospectively admitted in a hospital-based geriatric rehabilitation unit (designed as hospitalized patients). The main patient diagnoses for hospitalized patients on admission were: rehabilitation after falls (10%), stroke (8%), Alzheimer s disease (7%), bedsores (3%). Seventy six consecutive elderly patients were evaluated in a geriatric day hospital (designed as outpatients). The main outpatient diagnoses were Alzheimer s disease (37%), depression (10%) and other medical diseases. Patients were excluded if they did not give informed consent (themselves or their proxy), if they were less than 70 years old, if they underwent enteral or parenteral nutrition, if they had endocrine disease, especially diabetes and thyroid diseases, if they had edema or congestive cardiac insufficiency or advancing cancer, or if they needed palliative care (or if their life expectancy was evaluated at less than 45 days). Procedures Each patient underwent a clinical examination including measurement of weight, the sum of four subcutaneous skinfold thickness tests (triceps, biceps, suprascapular, suprailiac as measured with Harpenden calipers) and height evaluation (using knee height (KH) and Chumlea equations). The estimated height (H) was derived from KH and age using the following equations from Chumlea (Chumlea et al., 1985): For men H (cm) ¼ 2.02 KH (cm) 0.04 age (years) þ For women H (cm) ¼ 1.83 KH (cm) 0.24 age (years) þ BMI was calculated as body weight in kg divided by height squared in m 2. The subjects were fasted from hours the evening before testing. At hours, a venous blood sample was drawn. The serum was frozen and stored at 801C until assay. All samples were analyzed together. We defined malnutrition according to the lowest quartile of BMI and skinfold thickness measured in a large French population sample of healthy elderly subjects living independently at home and aged between 65 and 97 years, as described by Delarue et al. (1994). These authors stratified of BMI and skinfold thickness values by sex and four 5-year age classes starting at 65 years old. We divided our patients into three classes: Class 1 (probable malnutrition) had BMI and skinfold thickness values under the cutoff of the lowest quartile defined by Delarue. Class 2 (intermediate state of nutrition) covered all other subjects. Class 3 (normal nutrition or overweight) had BMI and/or skinfold thickness values over the cut-off of the highest quartile. The highest and lowest quartiles of Delarue s sample are presented after stratification by sex: cutoff values are given as an example for the over 80 years old bracket (Table 1). Assays Biological serum parameters including albumin, transthyretin, C-reactive protein, orosomucoid, glucose, thyrotropic hormone (TSH) and leptin were measured. In addition, hospitalized patients also had simultaneous serum measurements for IGF-1, IGFBP-1 and IGF-BP-3. Creatinine clearance was calculated using the Modification of Diet in Renal Disease (MDRD) formulas (National Kidney Foundation, 2002). Serum albumin, transthyretin, C-reactive protein and orosomucoid concentrations were measured by immunonephelometry on an Image analyzer (Beckman Coulter) apparatus. Serum glucose and creatinine were assessed by a glucose oxidase method and a modified Jaffé reaction, respectively, on a Hitachi analyzer (Roche diagnostics). Ultra-sensitive human serum TSH was determined by chemiluminescence immunoassay on an Access analyzer (Beckman Coulter). Serum leptin concentrations were measured by radioimmunoassay. For concentrations between 10 Table 1 Highest and lowest quartile of Delarue s sample presented after stratification by sex: example cut-off values are given for age over 80 years BMI Skinfold thickness in mm Lowest quartile Highest quartile Lowest quartile Highest quartile Men Women Abbreviation: BMI, body mass index.

3 and 100 ng/ml, we used a human leptin radioimmunoassay HL-81K kit (Linco Research, St Louis, MO, USA). The limit of sensitivity for this human leptin assay is 0.5 ng/ml. For concentrations between 4.9 and 25.6 ng/ml, within-assay coefficient of variation was % and between-assay coefficient of variation was 3 6.2%. For concentrations below 10 ng/ml, we used a human leptin radioimmunoassay SHL-81K kit (Linco Research, St Louis, MO, USA). The lowest concentration of leptin that can be detected by this assay is 0.05 ng/ml. For concentrations between 0.4 and 4.24 ng/ml, within-assay coefficient of variation (CV) was % and between-assay coefficient of variation was %. Serum IGF-1 concentrations were measured by radioimmunoassay (CIS Bio International, Gif-sur-Yvette, France). Intra- and interassay precision were 2.0 and 6.0%, respectively. Serum IGFBP-1 was assayed with the disease susceptibility locus (DSL) ACTIVE IGFBP-1 coated-tube immunoradiometric assay (IRMA) kit (DSL, Cergy Pontoise, France). The detection limit was 0.33 ng/ml. Within-assay CVs ranged between 2.7 and 5.2%; between-assay CVs ranged between 3.5 and 6%. No cross-reactivity with IGFBP-2,-3, or -4 was detected. Serum IGFBP-3 was assayed with the DSL ACTIVE IGFBP3 IRMA. The detection limit was 0.05 ng/ml. Within-assay CVs ranged between 1.8 and 3.9%; between-assay CVs ranged between 0.5 and 1.9%. No cross-reactivity with IGFBP-1,-2, or -4 was detected. Samples were analyzed in duplicate. Statistics Mean values of clinical and biological patient characteristics according to sex and hospitalized/outpatient status were compared using two-way analyses of variance (ANOVAs) and the relationship between sexes, hospitalized/outpatient status and nutritional status were tested using Fisher exact tests. Mean clinical and biological parameter of the three classes of nutritional status were tested using a one-way ANOVA. When globally significant, the means were compared pairwise using Tukey tests. Correlations between serum leptin and clinical or biological variables were assessed using Spearman rank correlation tests. Finally, multiple linear regressions relating leptin serum to specific subset of parameters were also performed. Cut-off values were determined empirically by visual analysis of receiver operating characteristic (ROC) curves. Multiple comparisons were taken into account using a Bonferonni correction according to a number of 17 mean parameters: for the ANOVAs and correlation coefficient tests, P-values lower than 0.05/17 ¼ were considered significant. All tests were two-sided, and all computations were performed using the SAS statistical software package (version 8; SAS Institute France, Grégy-sur-Yerres, France). Results One hundred and sixteen hospitalized and 76 outpatients were included (age: 8577 years old). Table 2 presents the serum biological parameters and clinical characteristics of the hospitalized and outpatient populations by sex. Significant differences are observed concerning age, albumin, transthyretin, transferrin, CRP, orosomucoid, weight and skinfold thickness between hospitalized and outpatients. Hospitalized patients were older, slimmer, had lower albumin, transthyretin and transferrin and higher CRP and orosomucoid. Sex ratio distribution, classes of nutrition, glucose, leptin values and BMI did not reach significance. As shown in Table 2, IGFBP-3, leptin and weight were the only parameters significantly different between men and women. Table 3 reports the clinical and biological characteristics of patients classified according to their nutritional status. Albumin, transthyretin, IGF1, IGFBP-1 and IGFBP-3 values as well as acute-phase proteins (CRP and orosomucoid) were not significantly different between the three classes of nutrition (data not shown for orosomucoid). As expected, anthropometric data were significantly different between these classes. The only biological parameter significantly related to the 3 classes of nutrition defined from the two anthropometric parameters (BMI and skinfold thickness) was leptin concentration, with significant pairwise differences using Tukey tests. We decided to focus on leptin concentration and determine the best relation between leptin concentration, sex and anthropometric data. Firstly, using a multiple regression analysis, we checked independent correlations between leptin concentration and, respectively, BMI (Po0.0001), skinfold thickness (Po0.01) and sex (Po0.0001) (data not shown). Second, we confirmed that the relation between nutritional status and leptin concentration was significantly different in both sexes (Po0.001 for men and Po for women) (Figure 1): the more the patients were undernourished, the lower the leptin concentration, in both sexes. Third, correlations between leptin concentration and biological values and age were studied (Table 4). There was no correlation between leptin and the inflammatory markers CRP and orosomucoid. Leptin concentration was significantly correlated with IGFBP-1 in males only. In this study, IGFBP-1 showed a weak inverse relationship with BMI in women only (Po0.015) and a strong inverse relationship with the sum of four skinfold thickness in each sex (r ¼ in males, Po and r ¼ in females, Po0.0001) (data not shown) and was not correlated with the three classes of nutrition (Table 3). Comparatively, the correlations between IGF-1 or IGFBP-3 and BMI and between IGF-1 or IGFBP-3 and the sum of four skinfold thickness were not significant (data not shown). We then realized a stepwise multiple linear regression analysis including hospitalization status, sex, nutritional status and age. Age is known to be an independent contributor of leptin concentrations (Isidori et al., 2000) and was therefore included in the model, although it was obviously not significantly correlated to leptin concentration in our study. There was no effect of hospitalization 649

4 650 Table 2 Biological serum parameters and clinical characteristics of the hospitalized and the outpatient populations by sex Hospitalized patients Outpatients P a Women (n ¼ 91) Men (n ¼ 25) Women (n ¼ 54) Men (n ¼ 22) Hosp./Outp. Women/men Classes of nutrition (1/2/3) 38/33/20 9/8/8 15/17/22 9/5/8 NS NS Age b o NS Glucose (mmol/l) NS NS Albumin (g/l) o NS Transthyretin (mg/l) o NS Transferrin (g/l) o NS CRP (mg/l) NS Orosomucoid (g/l) o NS TSH (mu/l) NS NS IGF 1 (mg/l) NS IGF-BP 1 (mg/l) NS IGF-BP 3 (mg/l) o Leptin (mg/l) NS Weight (kg) o o BMI (kg/m 2 ) NS NS Sum of four skinfold thickness (mm) o NS Abbreviations: BMI, body mass index; CRP, C-reactive protein; IGF-BP1 insulin-like growth factor binding protein; NS, not significant after Bonferonni correction (see Methods); TSH, thyrotropic hormone. a Two-way analyses of variance (ANOVAs), Hosp./Outp., hospitalized patients versus outpatients by sex, Women/men, women versus men by hospitalized/outpatient status. b Means 7 s.d. (all values). Table 3 Clinical and biological characteristics of patients classified according to their nutritional status a Nutritional status P b Clinical parameters (n ¼ 71) (n ¼ 63) (n ¼ 58) Age c NS Sex ratio (Men/Women) 18/53 13/50 16/42 NS Body mass index (kg/m 2 ) o Sum of the four skinfold thickness (mm) o Classical biological parameters (n ¼ 71) (n ¼ 63) (n ¼ 58) Albumin (g/l) NS Transthyretin (mg/l) NS CRP (mg/l) NS Other biological parameters (n ¼ 67) (n ¼ 57) (n ¼ 53) Leptin (mg/l) o Other biological parameters (only measured in (n ¼ 44) (n ¼ 37) (n ¼ 26) hospitalized patients, respectively 25 men and 82 women) IGF-1 (mg/l) NS IGFBP-1 (mg/l) NS IGFBP-3 (mg/l) NS Abbreviations: CRP, C-reactive protein; IGF-1, insulin-like growth factor-1; IGFBP-1, IGF binding protein-1;ns, not significant after Bonferonni correction (see Methods). a CRP. b Chi-square and one-way ANOVAs according to nutritional status. c Means 7 s.d. (all such values). status, and the relationships between leptin concentration and other parameters are explained in the following formula. The equation written as follows explains 35% of variance (R 2 ). Age (in years) is expressed in relation to the mean age of our population, which was 85 years: In elderly women: Mean leptin value (mg/l) in class 1 ¼ 3.7 ((age 85) 0.10) Mean leptin value (mg/l) in class 2 ¼ 12.2 Mean leptin value (mg/l) in class 3 ¼ 20.2 In elderly men: Mean leptin value (mg/l) in class 1 ¼ 2.15 ((age 85) 0.21) Mean leptin value (mg/l) in class 2 ¼ 2.85 ((age 85) 0.11)

5 Leptin (µg/l) Mean leptin value (mg/l) in class 3 ¼ ((age 85) 0.11) We made separate calculations of the sensitivity and specificity of leptin concentration to diagnose definite malnutrition (class 1) for each sex (Figures 2 and 3). Next, we determined cutoff values on the ROC curves, and found that the optimal cutoff was 4 mg/l in men (sensitivity 0.89, specificity 0.82) and 6.48 mg/l in women (sensitivity 0.90, specificity 0.83). Discussion class 1 (n=18) Men Women class 1 (n=49) class 2 (n=13) class 2 (n=44) The purpose of this study was to determine the biological parameters best related to anthropometric markers of 3.81 class 3 (n=15) class 3 (n=38) Figure 1 Leptin concentrations (means7s.d.) per sex and nutritional status. Relationships between nutritional status and leptin concentration were tested using one-way ANOVA for men and women. Significance levels are: Po0.001 for men and Po for women. Table 4 Univariate correlation between serum leptin and clinical and biological variables a Women Men r P b r P Age NS NS Albumin (g/l) NS NS Transthyretin (mg/l) NS NS CRP (mg/l) NS NS Orosomucoid (g/l) NS NS IGF-1 (mg/l) NS NS IGFBP-1 (mg/l) NS IGFBP-3 (mg/l) NS NS Abbreviations: CRP, C-reactive protein; IGF-1, insulin-like growth factor-1; IGFBP-1, IGF binding protein-1; NS, not significant after Bonferonni correction (see Methods). a CRP from all 192 patients of the whole population for the first five lines and from the 116 hospitalized patients for the last three lines. b Linear coefficient of correlation test. sensitivity Figure 2 sensitivity Figure specificity ROC curve of leptin levels (mgl) in women specificity ROC curve of leptin levels (mgl) in men. malnutrition in an elderly hospitalized or outpatient population, and determine cutoff values for these potential parameters to diagnose malnutrition. Given the absence of an international consensus definition of malnutrition in a very old polypathological population, we chose to define malnutrition according to the lowest quartile of BMI and skinfold thickness measured in a large French population sample of healthy elderly subjects living independently at home, as described by Delarue (Delarue et al., 1994). In the elderly over 70 years old, ESPEN guidelines (Beck and Ovesen, 1998) recommend the use of the MNA to detect the risk of developing malnutrition among in home-care programs, nursing homes and hospital settings (Hasselmann and Alix, 2003; Kondrup et al., 2003). The MNA is questionnaire-based and is more suitably adapted to the elderly at home. For adults, the ESPEN guidelines recommend using a combination of BMI and weight loss (Malnutrition Universal Screening Tool or MUST). Neither BMI nor weight loss should be used alone. The cutoff points used for screening are a BMI less than 20 kg/m 2 and weight loss of over 5% during the last 3 6 months (Beck and Ovesen, 1998). However, usual weight is frequently impossible to obtain in elderly patients 651

6 652 (Kuczmarski et al., 2001). Only half of the elderly can remember their habitual weight, and they are seldom weighed, even under professional care (Robbins, 1989). With increasing age, the optimal BMI for healthy elderly people increases from to kg/m 2. This is why, to identify the elderly at-risk patient, Beck suggests that a BMI of less than 24 kg/m 2 should be used in combination with other parameters (Beck and Ovesen, 1998). The use of other anthropometric parameters such as skinfold thickness needs experienced hands, as these relatively simple measurements can be compromised by a number of errors in technique as well as by intra-and inter-examiner errors. Furthermore, these anthropometric parameters are influenced by fluid imbalance. The first objective of this study was to focus on biological parameters, which are best related with anthropometric markers of malnutrition, to identify which parameter can be used to diagnose malnutrition in elderly pathological populations. As shown by our results, serum albumin and transthyretin values decrease dramatically in hospitalized patients. Concurrently, higher CRP and orosomucoid values were observed in this population, which often suffers from inflammation or/and infection. A reduction in the hepatic synthesis of albumin may be the consequence of inflammatory conditions rather than nutritional status; this amply reflects its proxy value as an indicator of active disease and inflammation. Our IGF-1 values are consistent with those reported by (Campillo et al. (2000) in 40 elderly patients with hip fracture. The IGF-1 values reported by Ponzer et al. (1999) in elderly women with hip fracture are lower, but were obtained by another dosage method after extraction. The IGFBP-1 values we obtained for men are similar to those reported by Kalme et al. (2004) in elderly men without diabetes. The leptin concentrations of patients in Class 1 (probable malnutrition) of this study are comparable to those observed by Cederholm et al. (1997) in malnourished elderly patients and Haluzik et al. (1999) in malnourished adults. The leptin concentrations of patients in class 3 (normal nutrition or overweight) are in the same range as those observed by Isidori, Smirnoff and Van den Saffele in healthy elderly patients (Van den Saffele et al., 1999; Isidori et al., 2000; Smirnoff et al., 2001). It should be noted that in our study, leptin was not correlated with transthyretin, albumin, CRP or orosomucoid. Leptin concentration was correlated with BMI independently of CRP (r ¼ 0.72, Po0.0001, partial correlation coefficient test). Our analysis of the relation between biological and anthropometric markers of malnutrition in elderly patients revealed a very strong correlation between leptin concentration and BMI, and between leptin concentration and the sum of four skinfold thickness measurements. Leptin concentration was the only parameter significantly related to all three classes of nutrition defined from these two anthropometric parameters. However, this finding cannot be generalized to patients excluded from the study, such as diabetic patients. Our results could potentially be limited in initially obese patients or those with greater loss of lean tissue than fatty tissue. However, the number of patients potentially involved is probably relatively low in the age bracket used in our study. Furthermore, studies on moderately obese elderly patients living at home and following a diet highlight that the weight loss involved applied more to fatty tissue (especially subcutaneous fatty tissue) than lean tissue, regardless of other factors including diabetes or regular physical exercise (Gallagher et al., 2000; Giannopoulo et al., 2005; Villareal et al., 2006). This difference in leptinemia between the three nutritional classes cannot be explained by renal insufficiency. Creatinine clearance was calculated using the Modification of Diet in Renal Disease (MDRD) study formulas. Average creatinine clearance in malnourished patients (group 1) was ml/min compared to ml/min and ml/min in groups 2 and 3 patients, respectively. There was no correlation between leptin concentration and creatinine clearance (r ¼ 0.122, P ¼ 0.1). Comparatively, the correlation between IGF-1 and BMI and between IGF-1 and the sum of four skinfold thicknesses was not significant. Some studies have attempted to simultaneously correlate BMI or nutritional status with leptin and IGF-1 concentrations. Although some authors (Sullivan et al., 1990; Ponzer et al., 1999; Campillo et al., 2000) found a positive correlation between IGF-1 and BMI, this correlation appeared to be weak and was not reported by other authors. Lo et al. (2003) in male volunteers and patients with tumor, as well as Van den Saffele et al. (1999) in 372 male subjects (23 78 years), found similar results to ours: IGF1 concentration was not correlated to BMI and leptin concentration was correlated to BMI. Anssen et al. (1998). showed that variations in IGFBP-1, which are inversely correlated with IGF-1, are better related to BMI and nutritional status than IGF-1. This same pattern was also reported by Laughlin et al. (2004) in a study of 633 men and 552 women (51 98 years). IGFBP-1 in our study showed a weak inverse correlation with BMI in women only, a strong inverse correlation with the sum of four skinfold thicknesses in both sexes, and showed no relationship with our three classes of nutrition. Soderberg et al. (2001) reported a strong inverse correlation between BMI and IGFBP-1 in subjects between 40 and 60 years old. Soderberg s team also showed an inverse correlation between leptin concentration and IGFBP-1 (Po0.05) in men only, which is in agreement with the strong inverse correlation found in our study between leptin concentration and IGFBP-1 in men (P ¼ ). In our study, edema and fluid imbalance were strictly excluded, but both disorders are frequent in the elderly, and both influence weight and other anthropometric data (Baxter, 1999). In the elderly, subclinical edema influences weight, which in this condition is not strictly correlated with nutritional status. Leptin concentration could serve as a better nutritional marker than weight and BMI in these contexts. However, patients with advanced chronic heart

7 failure may present elevated serum concentrations of leptin (Schulze et al., 2003). We can conclude that, among the biological parameters available, leptin concentration, in this study and in agreement with the literature, appears sufficiently strongly correlated with anthropometric data to probably be the best marker available. This correlation is in accordance with the fact that leptin concentration reflects the metabolic reserve constituted by fat, especially peripheral subcutaneous fat. The independent correlations between leptin concentration and sex, leptin concentration and BMI and, to a lesser extent, between leptin concentration and the sum of four skinfold thicknesses observed in this study suggest that leptin is a useful diagnostic tool. Physiologically, leptin is secreted by large adipocytes of subcutaneous peripheral fat (Van Harmelen et al., 1998). BMI is a composite index in which weight is mainly composed of peripheral fat, trunk fat and lean mass. According to Gower et al. (2000), trunk fat is presumed to secrete less leptin that peripheral fat, whereas the same authors found a negative correlation between lean mass and leptin concentration. Our results suggest that sex (with more peripheral subcutaneous fat in women) and BMI provide a good assessment of peripheral subcutaneous fat and that the sum of skinfold thicknesses only slightly enhances the assessment. The second objective of this study was to determine cutoff values for leptin concentrations in order to diagnose malnutrition in elderly, hospitalized and outpatients. We found that the optimal cutoff was 4 mg/l in men and 6.48 mg/l in women, with good sensitivity and specificity. In conclusion, leptin concentration is highly correlated with anthropometric data used to define nutritional status, and is the only biological parameter significantly related to classes of nutrition, in contrast with albumin or transthyretin which are influenced by morbidity and inflammatory conditions. Leptin concentration decreases as undernourishment becomes more pronounced. Thus, leptin concentration may be particularly useful for patients who cannot be weighed owing to reduced mobility or postural instability. Serum leptin concentration could serve as a nutritional assessment indicator in elderly patients, even against a background of acute disease, which is frequent in this population. Acknowledgements This study was supported by a DRRC AP-HP grant, number Conflict of interest statement All authors have disclosed any financial and personal relationships with other people or organizations that could inappropriately influence their work. The corresponding author states that he had full access to all the data in the study and had final responsibility for the decision to submit for publication. References Baxter JP (1999). Problems of nutritional assessment in the acute setting. Proc Nutr Soc 58, Beck AM, Ovesen L (1998). At which body mass index and degree of weight loss should hospitalised elderly patients be considered at nutritional risk? Clin Nutr 17, Campillo B, Paillaud E, Bories PN, Noel M, Porquet D, Le Parco JC (2000). Serum levels of insulin-like growth factor-1 in the three months following surgery for a hip fracture in elderly: relationship with nutritional status and inflammatory reaction. Clin Nutr 19, Cederholm T, Arner P, Palmblad J (1997). Low circulating leptin levels in protein-energy malnourished chronically ill elderly patients. J Intern Med 242, Chumlea WC, Roche AF, Steinbaugh ML (1985). Estimating stature from knee height for persons 60 to 90 years of age. J Am Geriatr Soc 33, Delarue J, Constans T, Malvy D, Pradignac A, Couet C, Lamisse F (1994). Anthropometric values in an elderly French population. Br J Nutr 71, Friedman JM, Halaos JL (1998). Leptin and the regulation of body weight in mammals. Nature 395, Gallagher D, Kovera AJ, Clay-Williams G, Agin D, Leone P, Albu J et al. (2000). Weight loss in postmenopausal obesity: no adverse alterations in body composition and protein metabolism. Am J Physiol Endocrinol Metab 279, E124 E131. Giannopoulou I, Ploutz-Snyder LL, Carhart R, Weinstock RS, Fernhall B, Goulopoulou S et al. (2005). Exercise is required for visceral fat loss in postmenopausal women with type 2 diabetes. J Clin Endocrinol Metab 90, Gower BA, Nagy TR, Goran MI, Smith A, Kent E (2000). Leptin in postmenopausal women: influence of hormone therapy, insulin and fat distribution. J Clin Endocr Soc 85, Haluzik M, Kabrt J, Nedvidkova J, Svobodova J, Kotrlikova E, Papezova H (1999). Relationship of serum leptin levels and selected nutritional parameters in patients with protein caloric malnutrition. Nutrition 15, Hasselmann M, Alix E (2003). Tools and procedures for screening for malnutrition and its associated in risks in hospital. Nutr Clin Metab 17, Isidori AM, Strollo F, More M, Caprio M, Aversa A, Moretti C et al. (2000). Leptin and aging: correlation with endocrine changes in male and female healthy adult populations of different body weights. J Clin Endocrinol Metab 85, Janssen JAMJL, Stolk RP, Pols HAP, Grobbee DE, Lamberts SWJ (1998). Serum total IGF-1, free IGF-1, and IGFBP-1 levels in an elderly population. Relation to cardiovascular risk factors and disease. Arteriolscler Thromb Vasc Biol 18, Kalme T, Seppälä M, Qiao Q, Koistinen R, Nissinen A, Harrela M et al. (2004). Sex hormone binding globulin and insulin-like growth factor binding protein-1 as indicators of metabolic syndrome, cardiovascular risk and mortality in elderly men. J Clin Endocrinol Metab 90, Kondrup J, Allison SP, Elia M, Vellas B, Plauth M (2003). ESPEN guidelines for nutrition screening Clin Nutr 22, Kuczmarski MF, Kuczmarski RJ, Najjar M (2001). Effects of age on validity of self-reported height, weight, and body mass index: findings from the Third National Health and Nutrition Examination Survey, J Am Diet Assoc 101, Laughlin GA, Barrettt-Connor E, Criqui MH, Kritz-Silverstein D (2004). The prospective association of serum insulin-like growth factor I (IGF-I) and IGF-binding protein-1 levels with all cause and cardiovascular disease mortality in older adults: the Rancho Bernardo study. J Clin Endocrinol Metab 89, Lo HC, Yang CS, Tsai LJ (2003). Simultaneous measurements of serum insulin-like growth factor-1 and leptin reflect the postoperative nutrition status of oral tumor patients. Nutrition 19,

8 654 National Kidney Foundation (2002). K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 39, S Ponzer S, Tidermark J, Brismar K, Söderqvist A, Cederholm T (1999). Nutritional status, insulin-like growth factor-1 and quality of life in elderly women with hip fractures. Clin Nutr 18, Robbins LJ (1989). Evaluation of weight loss in the elderly. Geriatrics 44, Schulze PC, Kratzsch J, Linke A, Schoene N, Adams V, Gielen S et al. (2003). Elevated serum levels of leptin and soluble leptin receptor in patients with advanced chronic heart failure. Eur J Heart Fail 5, Smirnoff P, Almiral-Seliger D, Schwartz B (2001). Serum leptin levels in the elderly: relationship with gender and nutritional status. J Nutr Health Aging 5, Soderberg S, Ahren B, Eliasson M, Dinesen B, Brismar K, Olsson T (2001). Circulating IGF binding protein-1 is inversely associated with leptin in non-obese men and obese postmenopausal women. Eur J Endocrinol 144, Sullivan DH, Patch GA, Walis RC, Lipschitz DA (1990). Impact of nutritional status on morbidity and mortality in a select population of geriatric patients. Am J Clin Nutr 51, Van den Saffele J, Goemaere S, de Bacquer D, Kaufman JM (1999). Serum leptin levels in healthy ageing men: are decreased serum testosterone and increased adiposity in elderly men the consequence of leptin deficiency? Clin Endocrinol 5, Van Harmelen V, Reynisdottir S, Eriksson P, Hoffstedt J, Lönnqvist F, Arner P (1998). Leptin secretion from subcutaneous and visceral adipose tissue in women. Diabetes 47, Villareal DT, Banks M, Sinacore DR, Siener C, Klein S (2006). Effect of weight loss and exercise on frailty in obese older adults. Arch Intern Med 166,