- Email: [email protected]
Comparison of different methods to assess body composition of weight loss in obese and diabetic patients
Diabetes Research and Clinical Practice 77 (2007) 405–411 www.elsevier.com/locate/diabres
Comparison of different methods to assess body composition of weight loss in obese and diabetic patients P. Ritz a,b,*, A. Salle´ b, M. Audran c, V. Rohmer b a
b
Inserm UMR 694, Angers, France Poˆle de maladies me´taboliques et me´decine interne, CHU, 49033 Angers, France c Poˆle oste´o-articulaire; Inserm EMI 0335, Angers, France Received 31 July 2006; accepted 8 January 2007 Available online 16 February 2007
Abstract Estimating body composition is important to understand the metabolic and cardiovascular effects of adiposity. Estimating changes in body compartments arising from weight loss strategies is equally important to evaluate their benefits and risks, particularly in frail populations (elderly or diabetic), and following bariatric surgery. Body compartments were evaluated in 50 obese subjects (25 diabetic, 25 non-diabetic) before and after a 7 kg weight loss obtained after 6 months of calorie restriction and orlistat. Fat and fat-free mass (FFM) were estimated by bioelectrical impedance analysis (BIA), dual X-ray absorptiometry (DXA), plethysmography (BodPod) and a combination of these in a 3- or 4-compartment model, the latter being considered the reference method. FFM hydration was the ratio of total body water (BIA) to FFM. FFM hydration was significantly higher than classical values (75.9 3.0%, P < 0.0001), and decreased with weight loss (74.2 3.3%). Compared to the 4-compartment, the 3-compartment model gave the most accurate fat and FFM estimation. A significant bias was observed with DXA, BodPod or BIA. Compartment changes induced by weight loss were accurately evaluated by DXA, being particularly precise in the 3-compartment analysis. No effect of diabetes per se was observed. A 3- or 4-compartmental analysis is necessary to accurately estimate body composition and its changes during weight loss. # 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Air displacement plethysmography; Obesity; Fat mass; Dual X-ray absorptiometry; Total body water
1. Introduction The incidence of obesity is increasing worldwide making this disease a pandemic [1]. Excess weight is associated with increased risk of metabolic disorders and cardiovascular diseases [2–5]. New pharmacological treatments are under investigation [6], and there is a renewed interest in surgical procedures with a particular emphasis on benefits in diabetic patients [7,8]. The * Corresponding author. Tel.: +33 241 354 499; fax: +33 241 354 969. E-mail address: [email protected] (P. Ritz).
evaluation of those treatment most of times targets body weight changes. Excess body fat is the main cause for health hazards induced by obesity (metabolic and cardiovascular disorders), and the preservation of fat-free mass (FFM) is a target in the treatment of obesity (especially in the elderly). Treatments of diabetes are known to increase weight and the clinical significance of the composition of weight gain (fat or fat-free) is important. Therefore, an assessment of body composition changes is desirable as an evaluation of the procedures. Data on the validity of body composition measurements in the obese persons are scarce [9]. It is now
0168-8227/$ – see front matter # 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.01.007
406
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
accepted that no one single method of measurement provides accurate and precise estimates [9,10]. Indeed, simple methods such as bioelectrical impedance analysis (BIA), and measurements derived from body density rely upon assumptions regarding constant relationships between gross components of fat-free mass. Those assumptions may not be valid in diseased states, during obesity and during weight changes [10]. In the healthy individuals as in diseased patients the 4compartmental approach can be regarded as the reference technique [10,11]. The 4-compartmental approach is based on measurements of body volume, total body water, bone mineral content (BMC) and weight. It obviates the need for those assumptions mentioned above. The 3-compartmental approach is simpler since it does not involve an estimate of bone mineral content. It has been challenged against one of the most accurate method (in vivo neutron activation analysis) and proved accurate [12]. Air displacement plethysmography (ADP) has attracted recent attention because it is an easy and reliable way of measuring body volume and density [13], which are key parameters in the 3- and 4-compartmental analysis of body composition. It is accurate for the measurement of adiposity up to 40% [14], and precise to about 2% [13,14]. There is only one evaluation of this technique as compared to 4-compartmental approach in obese subjects and in people with adiposity higher than 40% [9]. This evaluation did not consider diabetic patients. Furthermore, the accuracy and precision of the methods used to estimate body composition should be evaluated during weight changes. Therefore, the aim of the study was to investigate the validity of ADP estimates as compared to 4-compartmental approach in non-diabetic and diabetic obese patients both before and after weight loss. 2. Subjects and methods 2.1. Protocol Fifty volunteers were recruited to participate in a study on 6 months weight changes induced by mild calorie restriction and Orlistat. We present here the body composition data both at the beginning of the study and after weight loss, and compare results obtained with the different methods. Volunteers signed informed written consent for a protocol that was approved by the local medical school ethical committee. 2.2. Patients Twenty-five type 2 diabetic patients were recruited from the diabetes clinic in the university hospital in Angers, France.
Type 2 diabetes was defined according to WHO [15]. They were only treated with oral therapy, 23 patients took metformine, which was started 37 months (range 1–156 months) before the protocol was initiated. One patient was excluded from the study for mild kidney failure. Twenty-five obese patients were recruited at the same time as diabetic subjects. They were considered non-diabetic on the basis of a normal plasma glucose concentration (<110 mg/dl) at the first visit before any diet was implemented. Neither obese nor type 2 diabetic patients were taking drugs known to influence body weight. They received their scheduled treatment for hypertension and lipid abnormalities. 2.3. Weight reduction program After the initial measurement at M0 patients entered a weight reduction program, similar for both groups. The dietician set diet reduced by 20–25% energy from the usual energy intake (3 days food recall), which followed the French national recommendations for macronutrient content (Programme National Nutrition Sante´, www.sante.gouv.fr). Diet was evaluated at each visit, and further advice was given to the patients to optimise adhesion to initial recommendations. The dietician was blind to weight changes and diagnosis. Furthermore, patients received 120 mg Orlistat (Roche pharmaceuticals, Neuilly, France), thrice a day before each main meal. Orlistat was continued for 6 months until the week before the end-point measurement when it was discontinued. Patients were encouraged to increase their physical activity. 2.4. Body composition measurements Height was measured bare feet and in undergarments, with heels together and touching the base of a vertical scale, which was set to the nearest 0.1 cm (SECA, Hamburg, Germany). Weight was measured after breakfast, in undergarments, on a plethysmographic weighing scale, which was calibrated to the nearest 0.01 kg. The scale was calibrated prior to each test. Weight was then measured at the monthly visit on the same scale. Air displacement plethysmographic measurements were carried out using the BOD-POD body composition system (Bod-Pod, DS MEDICA, USA) according to the manufacturers’ instructions and recommendations [13]. The test was carried while the patients where in their undergarments and were wearing a swimming cap. The patient was initially weighed on an electronic scale. The procedure involved the calibration of the system when empty and then when a 50-l metal cylinder was placed inside. The patient was then seated within a 450-l chamber inside the machine, and asked to remain still with hands positioned on their thighs and breathing normally. A minimum of two consecutive body volume measurements (Vb), each lasting 50 s, were carried out. If the two volumes differed from each other by more than 150 ml, a third measurement was conducted. Body density (Db) was calculated by the machines’ integrated computer using the equation: Db = weight/Vb. Predicted respiratory volumes were
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
considered in the calculations. It was shown that predicted respiratory volumes were as accurate as measured volumes even in obese subjects, provided they do not suffer from respiratory disease [14]. Bone mineral content was measured in the dual X-ray absorptiometry (DXA) facility, being scanned using a Hologic 4500A device, a test taking little over 3 min to perform. Total body fat, fat-free mass and bone mineral content were computed by the systems integrated software (Version 11.2.3). Body water was measured by multifrequency bioelectrical impedance analysis using a BIA analyser (Analycor 4, Spengler, France), a test performed following 30 min rest in a temperature controlled room. Four surface electrodes were placed onto clean degreased skin at limb ends in a standardised manner, the measurement being performed on the right side of the body. Reproducibility of this technique is to within 2 V. Total body water (TBW) and extracellular water (ECW) was calculated using equations shown to be valid in diabetic people [16]. These equations were set in comparison with deuterated water and bromide dilution. It has been shown that in obese individuals and provided that the BIA estimates of body water rely on validated equations, 3-compartment estimates of fat did not differ whether BIA of isotope dilution was used [9]. 2.5. Calculations and statistics Calculations of body compartments. The 4-compartment approach was used as the reference technique. It computes fatness from body density, total body water, weight and bone mineral content as described in Packianathan et al. [17]. fat mass ¼ 2:747 weight=density 0:71 TBW þ 1:46 BMC 2:05 weight volume ¼ weight=density where TBW is in kg, BMC is bone mineral content in kg, weight is in kg and density in kg/l. The 3-compartment model was used as a surrogate for the 4-compartment approach [17]. It computes fatness from body density, total body water, and weight (% fat = 2.1176/density 0.78 TBW/weight 1.351). It does not require DXA measurement of bone mineral content. Fatness was also taken from the DXA scanning as provided by the machine (with software Version 11.2.3). Fatness was computed from density as acquired by plethysmography according to Siri [18]: %fat ¼ 4:95=density 4:55 Finally fatness was calculated from body weight and fatfree mass, which was calculated from TBW considering a 72% hydration coefficient. This 72% coefficient may not be accurate in diabetic persons but is also in Siri’s calculation from density [18].
407
Furthermore, hydration of fat-free mass was computed from fat-free mass taken from the 4-compartment approach and total body water [19]. 2.5.1. Statistics Results are expressed as means S.D. Initial values at M0 were compared between groups with a factorial analysis of variance. A comparison between baseline values (M0) and at 6 months (M6) among those obese patients and those with diabetes was carried out using an analysis of variance (ANOVA) for repeated measurements with the interaction between time and group (obese versus diabetics). A P-value of <0.05 was considered significant in all statistical comparisons.
3. Results Table 1 displays the physical characteristics of the patients. The obese diabetic and non-diabetic patients did not differ with respect to body composition, although the diabetic patients were slightly older. Patients were well matched for fat-free mass and fat mass (estimated with the 4-compartment model). Table 1 also displays the metabolic characteristics of the patients. As expected, HbA1c was higher in the diabetic group. Subjects lost a mean 7.2 3.17 kg weight over the 6 months. Weight loss was similar between diabetic and obese patients (P = 0.55). Body composition changes were made of 97% fat and 3% fat-free mass (4compartment evaluation). This was similar whether the patients were diabetic or not (P = 0.29). Hydration of fat-free mass was similar at M0 between the two groups at 75.9 3.0% (P = 0.63). It changed with weight loss to be 74.2 3.3% (P = 0.0001) at M6 similar in diabetics and non-diabetics (P for interaction 0.99). Both at M0 and M6 hydration of fat-free mass was significantly different from 72 or 73% (all P-values <0.0001). Table 2 displays the body composition data acquired with the different techniques. Data at M0 and M6 were Table 1 Physical characteristics and biological parameters at baseline Diabetic
Age (year) Weight (kg) BMI (kg/m2) HbA1c (%) Fat-free mass (kg) Fat mass (kg)
Obese
P-value
Mean
S.D.
Mean
S.D.
51.9 105.4 37.4 7.2 62.7 42.7
10 15.6 4.1 1.1 11.2 8.0
44.6 106.5 37.8 5.3 63.3 43.9
15 20.3 5.4 0.4 15.2 10.6
0.05 0.87 0.83 <0.0001 0.88 0.64
408
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
Table 2 Body composition data in the patients with the different techniques C4 Fat mass (n = 88) Fat-free mass (n = 88) Fat mass changes (n = 44) Fat-free mass changes (n = 44) * ** ***
40.0 10.2 63.0 12.9 6.14 3.61 0.90 2.57
C3
BodPod *
40.5 10.0 62.3 12.3* 6.04 3.42 1.10 0.26***
DXA *
38.0 11.1 64.3 13.0* 7.69 4.85** 0.51 3.93*
BIA *
37.9 9.9 65.8 13.1* 5.55 3.14 1.57 1.87
38.1 9.0* 64.8 13.0* 4.52 3.43** 2.69 2.65*
P < 0.0001 vs. C4. P < 0.0002 vs. C4. P < 0.05 vs. C4.
pooled (n = 88 complete data sets). As compared to the 4-compartment estimate of fat mass each other estimate was significantly different. Fig. 1 displays the distribution of the differences between the 4-compartment estimate of fat mass and that from other techniques. It can be seen that those differences are normally distributed and of a very small magnitude when the 3-compartment technique is considered. This was not the case with DXA or BIA or plethysmography. As compared to the 4-compartment estimate of fatfree mass each other estimate (DXA, BIA, ADP) was significantly different (Table 2). Table 2 also shows the magnitude of the changes in fat mass and fat-free mass (n = 44 complete data sets) over the 6 months of weight reduction. As compared to the 4-compartment estimate, changes in fat mass were
accurately estimated by 3-compartment and DXA. Estimates of fat mass changes with BodPod and BIA were significantly different. The magnitude of changes in fat-free mass over the 6 months of weight reduction were small and significantly overestimated by BIA and the 3-compartment estimate. Fig. 2 displays the distribution of the differences between the 4-compartment estimate of fat mass changes over 6 months and that from other techniques. It can be seen that those differences are normally distributed and of a very small magnitude when the 3-compartment technique is considered. The differences between estimates obtained with each method and the 4-compartment estimate of fat mass, fatness and fat-free mass was not influenced by whether the patients were diabetic or not (all P values
Fig. 1. Accuracy of the determination of fat mass. Histograms represent the differences between fat mass determined by the reference 4compartment and alternative methods. BOD, plethysmography; DXA, dual X-ray absorptiometry, BIA, bioelectrical impedance analysis and C3, 3compartment analysis.
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
409
Fig. 2. Accuracy of the determination of fat mass changes induced by weight loss. Histograms represent the differences between fat mass determined by the reference 4-compartment and alternative methods. BOD, plethysmography; DXA, dual X-ray absorptiometry; BIA, bioelectrical impedance analysis and C3, 3-compartment analysis.
>0.25). This was also the case for the changes in body composition after 6 months. 4. Discussion This study shows that in obese persons, whether diabetic or not, different methods for measuring body composition provide different results. If the 4-compartment is preferred and considered the reference, only the 3-compartment method provides values that differ very little. Furthermore, differences between the 3-compartment and the reference estimates were distributed over a narrow range. For the estimation of the changes in fat mass or fat-free mass, DXA and the 3-compartment method performed well, contrary to BIA and plethysmography. Hydration of fat-free mass did change during weight changes. Diabetes status did not influence the difference between the methods. Excess body fat is the main cause for health hazards induced by obesity (metabolic and cardiovascular disorders), and the preservation of fat-free mass is an objective in the treatment of obesity. This is especially the case in the elderly were fat-free mass is reduced with ageing and its preservation is critical. Furthermore, elderly people make a significant proportion of subjects suffering from diabetes. Treatments of diabetes are
known to increase weight and the clinical significance of the composition of weight gain (fat or fat-free) is important. Therefore, an assessment of body composition changes is desirable, and an evaluation of the procedures is required since many methods can be used. Simple methods to address body composition are attractive in that they take little time to be performed and are not aggressive. For example, BIA takes a few seconds, plethysmography and DXA a few minutes when the patients are installed. Software’s often propose results that appear accurate. In fact, those methods are in essence 2-compartment analyses of body mass (fat and fat-free mass), one component is measured while the other is calculated as the difference with weight. For example, BIA estimates fat-free mass from total body water (divided by the hydration coefficient), and fat mass is the difference between fat-free mass and weight. The 2-compartment methods rely on assumptions about the composition of fat-free mass. BIA assumes that hydration of fat-free mass is constant between individuals, and with time during interventions. Plethysmography assumes that fat-free mass is made of 72% water; 21% protein and 7% minerals [18]. DXA although less sensitive to hydration changes relies on fixed attenuation coefficient of fat and fat-free tissues to X-rays [11]. This may or may not be
410
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
true is certain individuals and the weight changes may alter those proportions. For example, during diabetes, corrections of glycaemia and insulin treatment are known to induce changes in water retention and distribution, bone mass and protein and fat mass (through the anabolic effect of insulin) [20,21]. The 4-compartment approach obviates the need for assumptions about the gross composition of fat-free mass [10]. This is why it was chosen as the reference method for body composition assessment in this study. However, there are very little data in obese subjects and in diabetics and no evaluation of those methods during weight changes especially in diabetics [9,10]. The present data shows that hydration of fat-free mass decreases when weight decreases. Therefore, it invalidates the results obtained with plethysmography and BIA alone, which rely on a fixed hydration of fat-free mass. There was a 2 kg fat mass difference with 4-compartment model (Table 2 two first rows). This is also the case when changes in body composition are studied (Table 2 two last rows). Since DXA measurements are less sensitive to hydration, mean changes in fat or fat-free mass observed with DXA did not differ from those obtained with the 4compartment model. Classical values for the hydration of fat-free mass are 72 or 73% [18]. Here, the observed values in obese subjects were higher, as already published (75.6% in [9]). Therefore, an estimate of hydration is required for the measurements of body composition in obese persons. BIA has been validated against isotope dilution in this population including diabetic patients and can therefore be used [9,18]. The 3-compartment method integrates hydration in the calculations and is a simpler method that was shown to be accurate in young, obese and elderly individuals [9,12,22]. It does not require the determination of bone mineral content. This study shows that 3-compartment analyses perform as well as the 4-compartment. Individual differences with the reference 4C model were very small. In the next paragraph, a discussion of the arithmetic in the equations shows that the difference in the 3- and 4-compartments analyses could only be different when bone mineral content changes dramatically. This was not the case here (data not shown). Furthermore, DXA measurements are limited to patients below 130 kg with actual tables. Therefore, for those patients a 3-compartment approach is well suited. Close scrutiny of the 4-compartment equation (see methods, 18) shows that fatness is grossly influenced by body volume (which is almost equal to body weight since density ranges between 0.9 and 1.1) and body weight. Those two variables are arithmetically affected by large coefficients (2.75 and 2.05). Quite the opposite,
TBW (about 50% of body weight) is affected by a smaller coefficient (0.7) while bone mineral content is small. Since body weight can be measured with high accuracy and precision, most of the value given by the 4-compartement model depends on body volume. Body volume is measured by plethysmography, and is accurate even in obese patients as shown by Fields et al. [13] and Demerath et al. [14]. Body volume can be measured with a precision of about 2% [13,14]. However, this 2% imprecision propagates to a 6% imprecision of fatness. With the same reasoning an error by 5% in body water propagates to an imprecision of 3% fatness when the 4-compartment approach is used. In conclusion, the present study shows that obesity, diabetes and weight changes do affect hydration of fatfree mass. Therefore, simple methods relying on a fixed hydration (BIA, ADP) are not valid enough to estimate body composition. The 4- and 3-compartment approaches, which requires estimates of body volume and body water are precise and accurate and should be used in these circumstances. The 3-compartment is a suitable alternative. Acknowledgments This work was made possible through grants from ‘‘Alfediam’’, ‘‘Contrat de Plan Etat Re´gion’’ and ‘‘Programme Hospitalier de Recherche Clinique’’. References [1] C.L. Ogden, M.D. Carroll, L.R. Curtin, M.A. McDowell, C.J. Tabak, K.M. Flegal, Prevalence of overweight and obesity in the United States, 1999–2004, JAMA 295 (2006) 1549–1555. [2] K.M. Flegal, B.I. Graubard, D.F. Williamson, M.H. Gail, Excess deaths associated with underweight, overweight, and obesity, JAMA 293 (2005) 1861–1867. [3] T. Thum, S.D. Anker, Obesity and risk of myocardial infarction: the INTERHEART study, Lancet 367 (2006) 1051–1052, author reply 1054. [4] L.H. Kuller, Nutrition, lipids, and cardiovascular disease, Nutr. Rev. 64 (2006) S15–S26. [5] C. Thomas, E. Hypponen, C. Power, Type 2 diabetes mellitus in midlife estimated from the Cambridge Risk Score and body mass index, Arch. Intern. Med. 166 (2006) 682–688. [6] S.Z. Yanovski, Pharmacotherapy for obesity—promise and uncertainty, N. Engl. J. Med. 353 (2005) 2187–2189. [7] H. Buchwald, Y. Avidor, E. Braunwald, M.D. Jensen, W. Pories, K. Fahrbach, et al., Bariatric surgery: a systematic review and meta-analysis, JAMA 292 (2004) 1724–1737. [8] L. Sjostrom, A.K. Lindroos, M. Peltonen, J. Torgerson, C. Bouchard, B. Carlsson, et al., Lifestyle, diabetes, and cardiovascular risk factors 10 years after bariatric surgery, N. Engl. J. Med. 351 (2004) 2683–2693. [9] S.K. Das, S.B. Roberts, J.J. Kehayias, J. Wang, L.K. Hsu, S.A. Shikora, et al., Body composition assessment in extreme obesity
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
[10]
[11]
[12]
[13]
[14]
[15]
and after massive weight loss induced by gastric bypass surgery, Am. J. Physiol. Endocrinol. Metab. 284 (2003) E1080–E1088. I.C. Packianathan, D.B. Peterson, A. Wright, W.A. Coward, N. Finer, Use of a reference four-compartment model to evaluate the ability of alternative methods and prediction techniques to estimate body composition in type 2 diabetes and its changes following insulin treatment, Int. J. Body Compos. Res. 2 (2004) 1–8. A. Pietrobelli, S.B. Heymsfield, Z.M. Wang, D. Gallagher, Multi-component body composition models: recent advances and future directions, Eur. J. Clin. Nutr. 55 (2001) 69–75. A.M. Silva, W. Shen, Z.M. Wang, J.F. Aloia, M.E. Nelson, S.B. Heymsfield, et al., Three-compartment model: critical evaluation based on neutron activation analysis, Am. J. Physiol. 287 (2004) 962–969. D.A. Fields, M.I. Goran, M.A. McCrory, Body-composition assessment via air-displacement plethysmography in adults and children: a review, Am. J. Clin. Nutr. 75 (2002) 453–467. E.W. Demerath, S.S. Guo, W.C. Chumlea, B. Towne, A.F. Roche, R.M. Siervogel, Comparison of percent body fat estimates using air displacement plethysmography and hydrodensitometry in adults and children, Int. J. Obes. Relat. Metab. Disord. 26 (2002) 389–397. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, Diabetes Care, vol. 20, 1997, pp. 1183–1197.
411
[16] A. Salle, G. Guilloteau, M. Ryan, B. Bouhanick, P. Ritz, Effect of insulin treatment on the body composition of Type 2 diabetic patients, Diabet. Med. 21 (2004) 1298–1303. [17] I.C. Packianathan, N.J. Fuller, D.B. Peterson, A. Wright, W.A. Coward, N. Finer, Use of a reference four-component model to define the effects of insulin treatment on body composition in type 2 diabetes: the ’Darwin study’, Diabetologia 48 (2005) 222– 229. [18] W. Siri, Body composition from fluid spaces and density: analysis of methods, in: JBaA. Henschel (Ed.), Techniques for Measuring Body Composition, National Academy of Science, Washington, 1961. [19] P. Ritz, Body water spaces and cellular hydration during healthy aging, Ann. N.Y. Acad. Sci. 904 (2000) 474–483. [20] A. Brizzolara, M.P. Barbieri, L. Adezati, G.L. Viviani, Water distribution in insulin-dependent diabetes mellitus in various states of metabolic control, Eur. J. Endocrinol. 135 (1996) 609– 615. [21] E.Y. Tsui, X.J. Gao, B. Zinman, Bioelectrical impedance analysis (BIA) using bipolar foot electrodes in the assessment of body composition in Type 2 diabetes mellitus, Diabet. Med. 15 (1998) 125–128. [22] J.L. Clasey, J.A. Kanaley, L. Wideman, S.B. Heymsfield, C.D. Teates, M.E. Gutgesell, et al., Validity of methods of body composition assessment in young and older men and women, J. Appl. Physiol. 86 (1999) 1728–1738.
Comparison of different methods to assess body composition of weight loss in obese and diabetic patients P. Ritz a,b,*, A. Salle´ b, M. Audran c, V. Rohmer b a
b
Inserm UMR 694, Angers, France Poˆle de maladies me´taboliques et me´decine interne, CHU, 49033 Angers, France c Poˆle oste´o-articulaire; Inserm EMI 0335, Angers, France Received 31 July 2006; accepted 8 January 2007 Available online 16 February 2007
Abstract Estimating body composition is important to understand the metabolic and cardiovascular effects of adiposity. Estimating changes in body compartments arising from weight loss strategies is equally important to evaluate their benefits and risks, particularly in frail populations (elderly or diabetic), and following bariatric surgery. Body compartments were evaluated in 50 obese subjects (25 diabetic, 25 non-diabetic) before and after a 7 kg weight loss obtained after 6 months of calorie restriction and orlistat. Fat and fat-free mass (FFM) were estimated by bioelectrical impedance analysis (BIA), dual X-ray absorptiometry (DXA), plethysmography (BodPod) and a combination of these in a 3- or 4-compartment model, the latter being considered the reference method. FFM hydration was the ratio of total body water (BIA) to FFM. FFM hydration was significantly higher than classical values (75.9 3.0%, P < 0.0001), and decreased with weight loss (74.2 3.3%). Compared to the 4-compartment, the 3-compartment model gave the most accurate fat and FFM estimation. A significant bias was observed with DXA, BodPod or BIA. Compartment changes induced by weight loss were accurately evaluated by DXA, being particularly precise in the 3-compartment analysis. No effect of diabetes per se was observed. A 3- or 4-compartmental analysis is necessary to accurately estimate body composition and its changes during weight loss. # 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Air displacement plethysmography; Obesity; Fat mass; Dual X-ray absorptiometry; Total body water
1. Introduction The incidence of obesity is increasing worldwide making this disease a pandemic [1]. Excess weight is associated with increased risk of metabolic disorders and cardiovascular diseases [2–5]. New pharmacological treatments are under investigation [6], and there is a renewed interest in surgical procedures with a particular emphasis on benefits in diabetic patients [7,8]. The * Corresponding author. Tel.: +33 241 354 499; fax: +33 241 354 969. E-mail address: [email protected] (P. Ritz).
evaluation of those treatment most of times targets body weight changes. Excess body fat is the main cause for health hazards induced by obesity (metabolic and cardiovascular disorders), and the preservation of fat-free mass (FFM) is a target in the treatment of obesity (especially in the elderly). Treatments of diabetes are known to increase weight and the clinical significance of the composition of weight gain (fat or fat-free) is important. Therefore, an assessment of body composition changes is desirable as an evaluation of the procedures. Data on the validity of body composition measurements in the obese persons are scarce [9]. It is now
0168-8227/$ – see front matter # 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.diabres.2007.01.007
406
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
accepted that no one single method of measurement provides accurate and precise estimates [9,10]. Indeed, simple methods such as bioelectrical impedance analysis (BIA), and measurements derived from body density rely upon assumptions regarding constant relationships between gross components of fat-free mass. Those assumptions may not be valid in diseased states, during obesity and during weight changes [10]. In the healthy individuals as in diseased patients the 4compartmental approach can be regarded as the reference technique [10,11]. The 4-compartmental approach is based on measurements of body volume, total body water, bone mineral content (BMC) and weight. It obviates the need for those assumptions mentioned above. The 3-compartmental approach is simpler since it does not involve an estimate of bone mineral content. It has been challenged against one of the most accurate method (in vivo neutron activation analysis) and proved accurate [12]. Air displacement plethysmography (ADP) has attracted recent attention because it is an easy and reliable way of measuring body volume and density [13], which are key parameters in the 3- and 4-compartmental analysis of body composition. It is accurate for the measurement of adiposity up to 40% [14], and precise to about 2% [13,14]. There is only one evaluation of this technique as compared to 4-compartmental approach in obese subjects and in people with adiposity higher than 40% [9]. This evaluation did not consider diabetic patients. Furthermore, the accuracy and precision of the methods used to estimate body composition should be evaluated during weight changes. Therefore, the aim of the study was to investigate the validity of ADP estimates as compared to 4-compartmental approach in non-diabetic and diabetic obese patients both before and after weight loss. 2. Subjects and methods 2.1. Protocol Fifty volunteers were recruited to participate in a study on 6 months weight changes induced by mild calorie restriction and Orlistat. We present here the body composition data both at the beginning of the study and after weight loss, and compare results obtained with the different methods. Volunteers signed informed written consent for a protocol that was approved by the local medical school ethical committee. 2.2. Patients Twenty-five type 2 diabetic patients were recruited from the diabetes clinic in the university hospital in Angers, France.
Type 2 diabetes was defined according to WHO [15]. They were only treated with oral therapy, 23 patients took metformine, which was started 37 months (range 1–156 months) before the protocol was initiated. One patient was excluded from the study for mild kidney failure. Twenty-five obese patients were recruited at the same time as diabetic subjects. They were considered non-diabetic on the basis of a normal plasma glucose concentration (<110 mg/dl) at the first visit before any diet was implemented. Neither obese nor type 2 diabetic patients were taking drugs known to influence body weight. They received their scheduled treatment for hypertension and lipid abnormalities. 2.3. Weight reduction program After the initial measurement at M0 patients entered a weight reduction program, similar for both groups. The dietician set diet reduced by 20–25% energy from the usual energy intake (3 days food recall), which followed the French national recommendations for macronutrient content (Programme National Nutrition Sante´, www.sante.gouv.fr). Diet was evaluated at each visit, and further advice was given to the patients to optimise adhesion to initial recommendations. The dietician was blind to weight changes and diagnosis. Furthermore, patients received 120 mg Orlistat (Roche pharmaceuticals, Neuilly, France), thrice a day before each main meal. Orlistat was continued for 6 months until the week before the end-point measurement when it was discontinued. Patients were encouraged to increase their physical activity. 2.4. Body composition measurements Height was measured bare feet and in undergarments, with heels together and touching the base of a vertical scale, which was set to the nearest 0.1 cm (SECA, Hamburg, Germany). Weight was measured after breakfast, in undergarments, on a plethysmographic weighing scale, which was calibrated to the nearest 0.01 kg. The scale was calibrated prior to each test. Weight was then measured at the monthly visit on the same scale. Air displacement plethysmographic measurements were carried out using the BOD-POD body composition system (Bod-Pod, DS MEDICA, USA) according to the manufacturers’ instructions and recommendations [13]. The test was carried while the patients where in their undergarments and were wearing a swimming cap. The patient was initially weighed on an electronic scale. The procedure involved the calibration of the system when empty and then when a 50-l metal cylinder was placed inside. The patient was then seated within a 450-l chamber inside the machine, and asked to remain still with hands positioned on their thighs and breathing normally. A minimum of two consecutive body volume measurements (Vb), each lasting 50 s, were carried out. If the two volumes differed from each other by more than 150 ml, a third measurement was conducted. Body density (Db) was calculated by the machines’ integrated computer using the equation: Db = weight/Vb. Predicted respiratory volumes were
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
considered in the calculations. It was shown that predicted respiratory volumes were as accurate as measured volumes even in obese subjects, provided they do not suffer from respiratory disease [14]. Bone mineral content was measured in the dual X-ray absorptiometry (DXA) facility, being scanned using a Hologic 4500A device, a test taking little over 3 min to perform. Total body fat, fat-free mass and bone mineral content were computed by the systems integrated software (Version 11.2.3). Body water was measured by multifrequency bioelectrical impedance analysis using a BIA analyser (Analycor 4, Spengler, France), a test performed following 30 min rest in a temperature controlled room. Four surface electrodes were placed onto clean degreased skin at limb ends in a standardised manner, the measurement being performed on the right side of the body. Reproducibility of this technique is to within 2 V. Total body water (TBW) and extracellular water (ECW) was calculated using equations shown to be valid in diabetic people [16]. These equations were set in comparison with deuterated water and bromide dilution. It has been shown that in obese individuals and provided that the BIA estimates of body water rely on validated equations, 3-compartment estimates of fat did not differ whether BIA of isotope dilution was used [9]. 2.5. Calculations and statistics Calculations of body compartments. The 4-compartment approach was used as the reference technique. It computes fatness from body density, total body water, weight and bone mineral content as described in Packianathan et al. [17]. fat mass ¼ 2:747 weight=density 0:71 TBW þ 1:46 BMC 2:05 weight volume ¼ weight=density where TBW is in kg, BMC is bone mineral content in kg, weight is in kg and density in kg/l. The 3-compartment model was used as a surrogate for the 4-compartment approach [17]. It computes fatness from body density, total body water, and weight (% fat = 2.1176/density 0.78 TBW/weight 1.351). It does not require DXA measurement of bone mineral content. Fatness was also taken from the DXA scanning as provided by the machine (with software Version 11.2.3). Fatness was computed from density as acquired by plethysmography according to Siri [18]: %fat ¼ 4:95=density 4:55 Finally fatness was calculated from body weight and fatfree mass, which was calculated from TBW considering a 72% hydration coefficient. This 72% coefficient may not be accurate in diabetic persons but is also in Siri’s calculation from density [18].
407
Furthermore, hydration of fat-free mass was computed from fat-free mass taken from the 4-compartment approach and total body water [19]. 2.5.1. Statistics Results are expressed as means S.D. Initial values at M0 were compared between groups with a factorial analysis of variance. A comparison between baseline values (M0) and at 6 months (M6) among those obese patients and those with diabetes was carried out using an analysis of variance (ANOVA) for repeated measurements with the interaction between time and group (obese versus diabetics). A P-value of <0.05 was considered significant in all statistical comparisons.
3. Results Table 1 displays the physical characteristics of the patients. The obese diabetic and non-diabetic patients did not differ with respect to body composition, although the diabetic patients were slightly older. Patients were well matched for fat-free mass and fat mass (estimated with the 4-compartment model). Table 1 also displays the metabolic characteristics of the patients. As expected, HbA1c was higher in the diabetic group. Subjects lost a mean 7.2 3.17 kg weight over the 6 months. Weight loss was similar between diabetic and obese patients (P = 0.55). Body composition changes were made of 97% fat and 3% fat-free mass (4compartment evaluation). This was similar whether the patients were diabetic or not (P = 0.29). Hydration of fat-free mass was similar at M0 between the two groups at 75.9 3.0% (P = 0.63). It changed with weight loss to be 74.2 3.3% (P = 0.0001) at M6 similar in diabetics and non-diabetics (P for interaction 0.99). Both at M0 and M6 hydration of fat-free mass was significantly different from 72 or 73% (all P-values <0.0001). Table 2 displays the body composition data acquired with the different techniques. Data at M0 and M6 were Table 1 Physical characteristics and biological parameters at baseline Diabetic
Age (year) Weight (kg) BMI (kg/m2) HbA1c (%) Fat-free mass (kg) Fat mass (kg)
Obese
P-value
Mean
S.D.
Mean
S.D.
51.9 105.4 37.4 7.2 62.7 42.7
10 15.6 4.1 1.1 11.2 8.0
44.6 106.5 37.8 5.3 63.3 43.9
15 20.3 5.4 0.4 15.2 10.6
0.05 0.87 0.83 <0.0001 0.88 0.64
408
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
Table 2 Body composition data in the patients with the different techniques C4 Fat mass (n = 88) Fat-free mass (n = 88) Fat mass changes (n = 44) Fat-free mass changes (n = 44) * ** ***
40.0 10.2 63.0 12.9 6.14 3.61 0.90 2.57
C3
BodPod *
40.5 10.0 62.3 12.3* 6.04 3.42 1.10 0.26***
DXA *
38.0 11.1 64.3 13.0* 7.69 4.85** 0.51 3.93*
BIA *
37.9 9.9 65.8 13.1* 5.55 3.14 1.57 1.87
38.1 9.0* 64.8 13.0* 4.52 3.43** 2.69 2.65*
P < 0.0001 vs. C4. P < 0.0002 vs. C4. P < 0.05 vs. C4.
pooled (n = 88 complete data sets). As compared to the 4-compartment estimate of fat mass each other estimate was significantly different. Fig. 1 displays the distribution of the differences between the 4-compartment estimate of fat mass and that from other techniques. It can be seen that those differences are normally distributed and of a very small magnitude when the 3-compartment technique is considered. This was not the case with DXA or BIA or plethysmography. As compared to the 4-compartment estimate of fatfree mass each other estimate (DXA, BIA, ADP) was significantly different (Table 2). Table 2 also shows the magnitude of the changes in fat mass and fat-free mass (n = 44 complete data sets) over the 6 months of weight reduction. As compared to the 4-compartment estimate, changes in fat mass were
accurately estimated by 3-compartment and DXA. Estimates of fat mass changes with BodPod and BIA were significantly different. The magnitude of changes in fat-free mass over the 6 months of weight reduction were small and significantly overestimated by BIA and the 3-compartment estimate. Fig. 2 displays the distribution of the differences between the 4-compartment estimate of fat mass changes over 6 months and that from other techniques. It can be seen that those differences are normally distributed and of a very small magnitude when the 3-compartment technique is considered. The differences between estimates obtained with each method and the 4-compartment estimate of fat mass, fatness and fat-free mass was not influenced by whether the patients were diabetic or not (all P values
Fig. 1. Accuracy of the determination of fat mass. Histograms represent the differences between fat mass determined by the reference 4compartment and alternative methods. BOD, plethysmography; DXA, dual X-ray absorptiometry, BIA, bioelectrical impedance analysis and C3, 3compartment analysis.
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
409
Fig. 2. Accuracy of the determination of fat mass changes induced by weight loss. Histograms represent the differences between fat mass determined by the reference 4-compartment and alternative methods. BOD, plethysmography; DXA, dual X-ray absorptiometry; BIA, bioelectrical impedance analysis and C3, 3-compartment analysis.
>0.25). This was also the case for the changes in body composition after 6 months. 4. Discussion This study shows that in obese persons, whether diabetic or not, different methods for measuring body composition provide different results. If the 4-compartment is preferred and considered the reference, only the 3-compartment method provides values that differ very little. Furthermore, differences between the 3-compartment and the reference estimates were distributed over a narrow range. For the estimation of the changes in fat mass or fat-free mass, DXA and the 3-compartment method performed well, contrary to BIA and plethysmography. Hydration of fat-free mass did change during weight changes. Diabetes status did not influence the difference between the methods. Excess body fat is the main cause for health hazards induced by obesity (metabolic and cardiovascular disorders), and the preservation of fat-free mass is an objective in the treatment of obesity. This is especially the case in the elderly were fat-free mass is reduced with ageing and its preservation is critical. Furthermore, elderly people make a significant proportion of subjects suffering from diabetes. Treatments of diabetes are
known to increase weight and the clinical significance of the composition of weight gain (fat or fat-free) is important. Therefore, an assessment of body composition changes is desirable, and an evaluation of the procedures is required since many methods can be used. Simple methods to address body composition are attractive in that they take little time to be performed and are not aggressive. For example, BIA takes a few seconds, plethysmography and DXA a few minutes when the patients are installed. Software’s often propose results that appear accurate. In fact, those methods are in essence 2-compartment analyses of body mass (fat and fat-free mass), one component is measured while the other is calculated as the difference with weight. For example, BIA estimates fat-free mass from total body water (divided by the hydration coefficient), and fat mass is the difference between fat-free mass and weight. The 2-compartment methods rely on assumptions about the composition of fat-free mass. BIA assumes that hydration of fat-free mass is constant between individuals, and with time during interventions. Plethysmography assumes that fat-free mass is made of 72% water; 21% protein and 7% minerals [18]. DXA although less sensitive to hydration changes relies on fixed attenuation coefficient of fat and fat-free tissues to X-rays [11]. This may or may not be
410
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
true is certain individuals and the weight changes may alter those proportions. For example, during diabetes, corrections of glycaemia and insulin treatment are known to induce changes in water retention and distribution, bone mass and protein and fat mass (through the anabolic effect of insulin) [20,21]. The 4-compartment approach obviates the need for assumptions about the gross composition of fat-free mass [10]. This is why it was chosen as the reference method for body composition assessment in this study. However, there are very little data in obese subjects and in diabetics and no evaluation of those methods during weight changes especially in diabetics [9,10]. The present data shows that hydration of fat-free mass decreases when weight decreases. Therefore, it invalidates the results obtained with plethysmography and BIA alone, which rely on a fixed hydration of fat-free mass. There was a 2 kg fat mass difference with 4-compartment model (Table 2 two first rows). This is also the case when changes in body composition are studied (Table 2 two last rows). Since DXA measurements are less sensitive to hydration, mean changes in fat or fat-free mass observed with DXA did not differ from those obtained with the 4compartment model. Classical values for the hydration of fat-free mass are 72 or 73% [18]. Here, the observed values in obese subjects were higher, as already published (75.6% in [9]). Therefore, an estimate of hydration is required for the measurements of body composition in obese persons. BIA has been validated against isotope dilution in this population including diabetic patients and can therefore be used [9,18]. The 3-compartment method integrates hydration in the calculations and is a simpler method that was shown to be accurate in young, obese and elderly individuals [9,12,22]. It does not require the determination of bone mineral content. This study shows that 3-compartment analyses perform as well as the 4-compartment. Individual differences with the reference 4C model were very small. In the next paragraph, a discussion of the arithmetic in the equations shows that the difference in the 3- and 4-compartments analyses could only be different when bone mineral content changes dramatically. This was not the case here (data not shown). Furthermore, DXA measurements are limited to patients below 130 kg with actual tables. Therefore, for those patients a 3-compartment approach is well suited. Close scrutiny of the 4-compartment equation (see methods, 18) shows that fatness is grossly influenced by body volume (which is almost equal to body weight since density ranges between 0.9 and 1.1) and body weight. Those two variables are arithmetically affected by large coefficients (2.75 and 2.05). Quite the opposite,
TBW (about 50% of body weight) is affected by a smaller coefficient (0.7) while bone mineral content is small. Since body weight can be measured with high accuracy and precision, most of the value given by the 4-compartement model depends on body volume. Body volume is measured by plethysmography, and is accurate even in obese patients as shown by Fields et al. [13] and Demerath et al. [14]. Body volume can be measured with a precision of about 2% [13,14]. However, this 2% imprecision propagates to a 6% imprecision of fatness. With the same reasoning an error by 5% in body water propagates to an imprecision of 3% fatness when the 4-compartment approach is used. In conclusion, the present study shows that obesity, diabetes and weight changes do affect hydration of fatfree mass. Therefore, simple methods relying on a fixed hydration (BIA, ADP) are not valid enough to estimate body composition. The 4- and 3-compartment approaches, which requires estimates of body volume and body water are precise and accurate and should be used in these circumstances. The 3-compartment is a suitable alternative. Acknowledgments This work was made possible through grants from ‘‘Alfediam’’, ‘‘Contrat de Plan Etat Re´gion’’ and ‘‘Programme Hospitalier de Recherche Clinique’’. References [1] C.L. Ogden, M.D. Carroll, L.R. Curtin, M.A. McDowell, C.J. Tabak, K.M. Flegal, Prevalence of overweight and obesity in the United States, 1999–2004, JAMA 295 (2006) 1549–1555. [2] K.M. Flegal, B.I. Graubard, D.F. Williamson, M.H. Gail, Excess deaths associated with underweight, overweight, and obesity, JAMA 293 (2005) 1861–1867. [3] T. Thum, S.D. Anker, Obesity and risk of myocardial infarction: the INTERHEART study, Lancet 367 (2006) 1051–1052, author reply 1054. [4] L.H. Kuller, Nutrition, lipids, and cardiovascular disease, Nutr. Rev. 64 (2006) S15–S26. [5] C. Thomas, E. Hypponen, C. Power, Type 2 diabetes mellitus in midlife estimated from the Cambridge Risk Score and body mass index, Arch. Intern. Med. 166 (2006) 682–688. [6] S.Z. Yanovski, Pharmacotherapy for obesity—promise and uncertainty, N. Engl. J. Med. 353 (2005) 2187–2189. [7] H. Buchwald, Y. Avidor, E. Braunwald, M.D. Jensen, W. Pories, K. Fahrbach, et al., Bariatric surgery: a systematic review and meta-analysis, JAMA 292 (2004) 1724–1737. [8] L. Sjostrom, A.K. Lindroos, M. Peltonen, J. Torgerson, C. Bouchard, B. Carlsson, et al., Lifestyle, diabetes, and cardiovascular risk factors 10 years after bariatric surgery, N. Engl. J. Med. 351 (2004) 2683–2693. [9] S.K. Das, S.B. Roberts, J.J. Kehayias, J. Wang, L.K. Hsu, S.A. Shikora, et al., Body composition assessment in extreme obesity
P. Ritz et al. / Diabetes Research and Clinical Practice 77 (2007) 405–411
[10]
[11]
[12]
[13]
[14]
[15]
and after massive weight loss induced by gastric bypass surgery, Am. J. Physiol. Endocrinol. Metab. 284 (2003) E1080–E1088. I.C. Packianathan, D.B. Peterson, A. Wright, W.A. Coward, N. Finer, Use of a reference four-compartment model to evaluate the ability of alternative methods and prediction techniques to estimate body composition in type 2 diabetes and its changes following insulin treatment, Int. J. Body Compos. Res. 2 (2004) 1–8. A. Pietrobelli, S.B. Heymsfield, Z.M. Wang, D. Gallagher, Multi-component body composition models: recent advances and future directions, Eur. J. Clin. Nutr. 55 (2001) 69–75. A.M. Silva, W. Shen, Z.M. Wang, J.F. Aloia, M.E. Nelson, S.B. Heymsfield, et al., Three-compartment model: critical evaluation based on neutron activation analysis, Am. J. Physiol. 287 (2004) 962–969. D.A. Fields, M.I. Goran, M.A. McCrory, Body-composition assessment via air-displacement plethysmography in adults and children: a review, Am. J. Clin. Nutr. 75 (2002) 453–467. E.W. Demerath, S.S. Guo, W.C. Chumlea, B. Towne, A.F. Roche, R.M. Siervogel, Comparison of percent body fat estimates using air displacement plethysmography and hydrodensitometry in adults and children, Int. J. Obes. Relat. Metab. Disord. 26 (2002) 389–397. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, Diabetes Care, vol. 20, 1997, pp. 1183–1197.
411
[16] A. Salle, G. Guilloteau, M. Ryan, B. Bouhanick, P. Ritz, Effect of insulin treatment on the body composition of Type 2 diabetic patients, Diabet. Med. 21 (2004) 1298–1303. [17] I.C. Packianathan, N.J. Fuller, D.B. Peterson, A. Wright, W.A. Coward, N. Finer, Use of a reference four-component model to define the effects of insulin treatment on body composition in type 2 diabetes: the ’Darwin study’, Diabetologia 48 (2005) 222– 229. [18] W. Siri, Body composition from fluid spaces and density: analysis of methods, in: JBaA. Henschel (Ed.), Techniques for Measuring Body Composition, National Academy of Science, Washington, 1961. [19] P. Ritz, Body water spaces and cellular hydration during healthy aging, Ann. N.Y. Acad. Sci. 904 (2000) 474–483. [20] A. Brizzolara, M.P. Barbieri, L. Adezati, G.L. Viviani, Water distribution in insulin-dependent diabetes mellitus in various states of metabolic control, Eur. J. Endocrinol. 135 (1996) 609– 615. [21] E.Y. Tsui, X.J. Gao, B. Zinman, Bioelectrical impedance analysis (BIA) using bipolar foot electrodes in the assessment of body composition in Type 2 diabetes mellitus, Diabet. Med. 15 (1998) 125–128. [22] J.L. Clasey, J.A. Kanaley, L. Wideman, S.B. Heymsfield, C.D. Teates, M.E. Gutgesell, et al., Validity of methods of body composition assessment in young and older men and women, J. Appl. Physiol. 86 (1999) 1728–1738.