Impact of body-composition methodology on the composition of weight ...

2 downloads 0 Views 568KB Size Report
Feb 20, 2013 - BACKGROUND/OBJECTIVES: We intended to (i) to compare the composition of weight loss and weight gain using densitometry, deuterium ...
European Journal of Clinical Nutrition (2013) 67, 446–454 & 2013 Macmillan Publishers Limited All rights reserved 0954-3007/13 www.nature.com/ejcn

REVIEW

Impact of body-composition methodology on the composition of weight loss and weight gain M Pourhassan1, B Schautz1, W Braun1, C-C Gluer2, A Bosy-Westphal1,3 and MJ Mu¨ller1 BACKGROUND/OBJECTIVES: We intended to (i) to compare the composition of weight loss and weight gain using densitometry, deuterium dilution (D2O), dual-energy X-ray absorptiometry (DXA), magnetic resonance imaging (MRI) and the four-compartment (4C) model and (ii) to compare regional changes in fat mass (FM), fat-free mass (FFM) and skeletal muscle as assessed by DXA and MRI. SUBJECTS/METHODS: Eighty-three study participants aged between 21 and 58 years with a body mass index range of 20.2–46.8 kg/m2 had been assessed at two different occasions with a mean follow-up between 23.5 and 43.5 months. Body-weight changes within o 3% were considered as weight stable, a gain or a loss of 43% of initial weight was considered as a significant weight change. RESULTS: There was a considerable bias between the body-composition data obtained by the individual methods. When compared with the 4C model, mean bias of D2O and densitometry was explained by the erroneous assumption of a constant hydration of FFM, thus, changes in FM were underestimated by D2O but overestimated by densitometry. Because hydration does not normalize after weight loss, all two-component models have a systematic error in weight-reduced subjects. The bias between 4C model and DXA was mainly explained by FM% at baseline, whereas FFM hydration contributed to additional 5%. As to the regional changes in body composition, DXA data had a considerable bias and, thus, cannot replace MRI. CONCLUSIONS: To assess changes in body composition associated with weight changes, only the 4C model and MRI can be used with confidence. European Journal of Clinical Nutrition (2013) 67, 446–454; doi:10.1038/ejcn.2013.35; published online 20 February 2013 Keywords: body composition; deuterium dilution (D2O); air-displacement plethysmography (ADP); dual-energy X-ray absorptiometry (DXA); magnetic resonance imaging (MRI); four-compartment model (4C model)

INTRODUCTION Increased body fat is associated with high risk of metabolic disorders and cardiovascular disease, thus, weight loss is recommended in obese subjects.1 However, some studies suggest that weight loss itself is associated with increased mortality.2–4 This association appears to be independent of the underlying disease5 and maybe due to the loss of fat-free mass (FFM). Because dietinduced weight loss results in both losses in fat and lean tissue,6 the preservation of FFM is a specific concern of weight-loss therapy in the obese patient.7 In addition, weight maintenance is challenging to the majority of patients who had lost weight successfully, but undesirable weight regain and weight cycling is a common phenomenon.8 It is still unclear whether weight cycling carries an independent health risk.9 Adverse health consequences of weight cycling might be because of a different composition of weight regain when compared with weight loss, that is, weight cycling results in a higher ratio of fat mass (FM) to FFM.2 At present, there is limited information about the composition of body-weight changes with weight regain after weight loss.6 To our knowledge, body-composition analysis addressing longterm changes of body weight has rarely been performed using state-of-the-art methodologies in previous studies. A recent study

showed that among 2163 older men and women, significantly more lean mass (as measured by dual-energy X-ray absorptiometry (DXA)) was lost during weight loss than was regained over periods of 2–4 years.5 In contrast, data from a prospective study using a low calorie diet in 24 obese women (age 49–67 years) found that, although the mean body weight 4 years after weight-loss intervention was no longer significantly different from baseline, the mean lean mass significantly exceeded the baseline value (44.9±1.0 kg vs 43.6±1.1 kg, respectively).10 In another longitudinal population study on 2436 Danish adults aged 35, 45, 55 or 65 years, changes in body composition were measured by bioelectrical impedance analysis.2 In this study, FFM made up 41% and 35% of weight loss and 24% and 15% of weight regain in men and women, respectively. However, after adjustment for agerelated changes in body composition, gain and loss in FFM were not significantly different.2 Discrepant results among different studies may not only be explained by differences in age but may also partly be due to differences in the methods used for body-composition analysis. Dilution techniques, densitometry and DXA are among the most commonly used methods to assess changes in body composition. However, a major limitation to the application of these techniques

1 Institute of Human Nutrition and Food Science, Christian-Albrechts University, Kiel, Germany; 2Sektion Biomedizinische Bildgebung, Klinik fu¨r Diagnostische Radiologie, MOIN CC, Universita¨tsklinikum (University Medical Center) Schleswig Holstein (UKSH), Kiel, Germany and 3Institute of Nutritional Medicine, University of Hohenheim, Stuttgart, Germany. Correspondence: Professor Dr MJ Mu¨ller, Institut fu¨r Humanerna¨hrung und Lebensmittelkunde, Agrar-und Erna¨hrungswissenschaftliche Fakulta¨t, Christian-AlbrechtsUniversita¨t zu Kiel, Du¨sternbrooker Weg 17-19, Kiel D-24105, Germany. E-mail: [email protected] Received 22 January 2013; accepted 22 January 2013; published online 20 February 2013

Body composition during weight loss and weight gain M Pourhassan et al

447 in measuring changes is the inherent assumption of a constant density of FFM based on a fixed proportion of water, mineral and protein in this compartment.11 These assumptions may not be valid in obese patients mainly because of a higher water content of FFM, and especially during the dynamic state of weight changes,7 that is, because of a change in body water that is most exclusively from lean mass or FFM. Therefore, the variability in the density of FFM can affect the accuracy of these methods.12 To take into account the water and mineral content of FFM, a fourcompartment (4C) model has been recognized as the most accurate method (that is, the gold standard) to assess FFM.13 To our knowledge, only a few studies have assessed the validity of different methods to measure body-composition changes during weight loss or regain in overweight and obese subjects using the reference 4Cmodel.7,11,14 In these studies, FM and FFM measured by alternative techniques differed significantly from the reference method, thus raising questions about their value in assessing changes in body composition during weight changes. In addition, regional changes and distributions of fat and lean mass during weight loss and weight gain remain a controversy and may affect disease risk.15 Evaluation of regional body composition is of great importance because changes in FM and FFM may specifically and differentially affect the limbs and trunk.16 In addition to diet and lifestyle, age may also have an impact on the regional changes in FM and FFM. With advancing age, bone mineral and lean mass preferentially decrease,17 whereas body FM concomitantly increases and is more prone to be in the abdominal region.18 Magnetic resonance imaging (MRI) is the gold standard for assessment of regional body composition. In addition, DXA, anthropometric measures and bioelectrical impedance analysis have been applied to assess body composition in different parts of the body, but none of these techniques had been systematically used to address changes in body composition with weight changes. The aims of the present study were (i) to compare the composition of weight loss and weight gain using different methods including the gold standard, that is, the 4C model and (ii) to compare regional changes in FMDXA and adipose tissueMRI (ATMRI), as well as lean soft tissueDXA (LSTDXA) and skeletal muscleMRI (SMMRI), with weight loss and weight gain. We have used deuterium dilution (D2O), DXA and air-displacement plethysmography (ADP) for evaluating the composition of weight loss and weight gain in 83 healthy people with intentional diet-induced weight loss and participants with spontaneous weight gain, aged between 21 and 58 years. A 4C model was used as a reference. To understand how regional fat and lean mass in the limbs are altered during weight loss and weight gain, we measured the amount of regional fat and LST (LST of the trunk, arms and legs) by DXA with weight loss and weight gain using regional AT and SM volumes as assessed by whole-body MRI as the reference.

SUBJECTS AND METHODS Eighty-three study participants (59 women and 24 men), aged between 21 and 58 years with a body mass index (BMI) range of 20.2–46.8 kg/m2 who participated in previous studies and had been assessed at two different occasions at the Institute for Human Nutrition at Christian-AlbrechtsUniversity were recruited from the local community by advertisement in newspapers and notice board postings. The present study focused on comparison of weight-loss- and weight-gain-associated changes in body composition. Body-weight changes withino3% were considered as weight stable, whereas a gain or a loss of 43% of initial weight was considered as a significant weight change.19 Net body-weight change was calculated, and subjects were grouped into three weight-change categories: 30 subjects with weight loss who participated in an intervention study on a low-calorie diet with a follow-up of 23.5±22.4 months; 33 subjects gained weight spontaneously with a mean period of 33.6±29.1 months; and 20 weight-stable subjects with a mean follow-up period of 43.5±25.0 months. All investigations have been performed between 2005 and & 2013 Macmillan Publishers Limited

2011. Exclusion criteria were any use of medication that has an influence on body composition, metallic implants, pregnancy, smoking and acute or chronic disease (for example, diabetes). The study was approved by the local ethical committee of the Christian-Albrechts-University zu Kiel, and all participants provided informed written consent. Subjects arrived between 0700 and 0900 hours in the morning after an overnight fast at the metabolic ward of the Institute for Human Nutrition.

ANTHROPOMETRIC MEASUREMENTS Body weight was measured to the nearest 0.01 kg by using an electronic scale coupled to the BOD POD device with participants wearing light clothes (Tanita, Tokyo, Japan). Height was measured to the nearest 0.5 cm by using a Seca stadiometer (Vogel & Halke, Hamburg, Germany), with subjects standing erect and without shoes. BMI was calculated as weight in kg/height in m2. Hip circumference was measured at the level of the symphysis and waist circumference was measured in a standing position using the mean of two measures obtained midway between the lowest rib and iliac crest at the end of normal exhalation. DEUTERIUM DILUTION The D2O procedure has been described in greater detail in elsewhere.13 Briefly, total body water (TBW) in liters was determined by D2O dilution (99.9%; Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). After obtaining 40 ml venous blood samples, each subject drank an oral dose of 0.4 g D2O per kg body weight with an amount of 100 ml tap water. Four hours later, a second blood sample was taken. Blood samples were centrifuged immediately after collection. Serum was stored at  40 1C. Before infrared analysis, samples were centrifuged for 3 h in ultrafiltration tubes (Vivaspin 4; VivaScience AG, Hannover, Germany). The concentration of D2O was measured in ultrafiltrate by fast-Fourier infrared spectroscopy. Infrared spectra of the samples were measured in the range of 2200–2800 cm  1 using a FTS 2000 Series spectrophotometer (Digilab, Marlborough, MA, USA) equipped with a CaF2 sample cell (omnicell Specac Ltd., Orpington, UK). A calibration curve with 0.5, 1.0 and 2.0 g D2O/l distilled water was used for quantification. Peak height was assessed by the manufacturer’s software (Version Merlin 3.4). The D2O concentration in the sample before ingestion of the dose was used as baseline value, which had to be subtracted from the result of the second sample 4 h after ingestion of D2O. For one result, four measurements each consisting of 16 scans were averaged. A hydration of 73.2% was assumed for calculation of FFM. FMD2O was calculated as body weight minus FFMD2O. AIR-DISPLACEMENT PLETHYSMOGRAPHY ADP was performed by the BOD POD device (Cosmed s.r.l., Rome, Italy). A two-step calibration was carried out before each measurement. In the first step, the volume of the empty chamber, and in the second step, the volume of a 50-l calibration cylinder, was measured. When entering the BOD POD device, all participants wore tight-fitting underwear (that is, brassiere and pants) and a swim cap. There were instructed to sit motionless during the 50-s body-volume measurement. Two repeated volume measurements were performed, averaged and corrected for predicted body surface area and measured thoracic gas volume using the BOD POD software (version 4.5.0). FM% was calculated from body mass and volume via body density.20 FFMADP was calculated as body weight minus FMADP. DUAL-ENERGY X-RAY ABSORPTIOMETRY DXA whole-body measurement was performed (QDR4500A Hologic Inc., Bedford, MA, USA). Subjects lay supine with arms European Journal of Clinical Nutrition (2013) 446 – 454

Body composition during weight loss and weight gain M Pourhassan et al

448 and legs at their sides during the 10-min scan. Scans were performed by a licensed radiological technician. Manufacturer’s software (version V8.26a:3) was used for analysis of bone mineral content (BMC) and FM, respectively. FFMDXA was calculated as body weight minus FMDXA. FOUR-COMPARTMENT MODEL A 4C model, which divides the body into lipids, water, mineral and protein, was used as the criterion method. Measurement of BMC (from DXA), body volume (from ADP) and TBW (from D2O) were combined to yield a criterion 4C model estimation of FM; the errors in each measurement are aggregated/propagated into the 4C model, proportionally modulated by the constants:21 FM4C (kg) ¼ 2.7474 body volumeADP (l)  0.7145 TBW (l) þ 1.4599 BMCDXA (kg)  2.0503 weight (kg) The 4C model is based on the assumption of a fixed ratio of osseus- to non-osseus-mineral content of the body. Total body mineral (density ¼ 3.0375 g/cm3) is acquired by BMC multiplied with 1.2741.22 The densities of water, protein and fat are assumed to be 0.99371, 1.34 and 0.9007 g/cm3, respectively. FFM4C was calculated as the difference of body weight and FM4C. REGIONAL BODY COMPOSITION WITH DXA AND MRI Regional body composition was determined in subpopulation of 64 subjects using DXA and MRI. The measured amounts of LST and fat tissue (using DXA) and muscle mass and AT (using MRI) in the regions of trunk, arms and legs were determined. MRI protocols have been described in greater detail elsewhere.23,24 Briefly, volume of SM (SMMRI), subcutaneous AT (SATMRI) and visceral AT (VATMRI) were obtained by a 1.5-T scanner (Magnetom Vision at baseline and 6-year follow-up or Avanto at 6-year follow-up; Siemens, Erlangen, Germany) using a T1-weighted gradient echo sequences (TR (time to repeat) 575 ms and TE (time to echo) 15 ms for Magnetom Vision and TR 157 ms and TE 4 ms for Siemens Avanto). The two MRI devices have been validated cross-sectional. Subjects were examined in a supine position with their arms extended above their heads. Continuous transversal images with 8-mm slice thickness and 2-mm interslice gaps were obtained (in subpopulation of 36 subjects at baseline, continuous transversal images with 10-mm slice thickness and 10-mm interslice gaps were obtained) and analyzed from wrist to ankle using the SliceOmatic software (version4.3; Tomovision, Montreal, Canada). Images in abdominal and thoracic regions were measured with subjects holding their breath. Arms and legs were segmented from wrist to humerus heads and from femur heads to ankle, respectively. Trunk was defined as the region between femur heads and humerus heads. VATMRI was segmented from the top of the liver to femur heads. SMMRI (kg) was calculated for muscle volume using a density of 1.04 kg/l.25 STATISTICAL ANALYSIS Statistical analysis was performed using SPSS statistical software (SPSS 17.0, Inc., Chicago, IL, USA). All data are given as means±s.d. Differences between parameters of body composition assessed by different methods and differences between baseline and followup within each of the three weight-change groups (weight lost, weight stable and weight gained) were analyzed by paired sample t-test. After multiple comparison adjustments, differences between men and women and between weight-change groups at baseline were analyzed by unpaired t-test. Bland–Altman analysis was performed to compare bodycomposition variables assessed by different methods and 4C model as a criterion method.26 Stepwise regression analysis was European Journal of Clinical Nutrition (2013) 446 – 454

Table 1. Characteristics of the study population stratified by gender at baseline (mean±s.d.)

Age (y) Height (m) Weight (kg) BMI (kg/m2) WC (cm) Hip (cm) Prevalence of normal weight Prevalence of overweight Prevalence of obesity

All (n ¼ 83)

Females (n ¼ 59)

Males (n ¼ 24)

36.36±8.90 1.71±0.08 88.86±19.06 30.09±5.67 98.03±14.40 109.86±13.42 20.5%

34.50±7.69 1.68±0.07 87.09±19.91 30.55±6.03 97.47±14.82 112.24±14.23 20.3%

40.91±10.15** 1.79±0.04*** 93.21±16.34 28.95±4.60 99.40±13.53 104.00±9.00** 20.8%

30.1%

27.1%

37.5%

49.4%

52.5%

41.7%

Abbreviations: BMI, body mass index; Hip, hip circumference; WC, waist circumference. **Po0.01 and ***Po0.001 difference between gender.

used to determine the relationships between bias of changes in FM and FFM (assessed by comparison between 4C model and alternative methods) as dependent variable, and age, BMI, waist circumference, hip circumference, change in waist circumference and hip circumference, FM% at baseline, FM% at follow-up, % change in FM and change in FFM hydration as independent variables. Pearson’s correlation coefficient was calculated for relationships between variables. All tests were two-tailed, and a P-value o0.05 was accepted as the limit of significance.

RESULTS Basal characteristics of the study participants are shown in Table 1. Of 83 participants, 71% were female and 29% were male, with an age range between 21 and 58 years. The study population showed a wide BMI range with no sex differences. Prevalence of normal weight, overweight and obesity were 20%, 30% and 50%, respectively. In all, 27% of female and 37% of male subjects were overweight. Descriptive characteristics of the study population at baseline and follow-up are given in Table 2. In all, 36% of the study population were weight losers. Weight loss ranged between  3.3 and  25.4 kg. In contrast, 40% of the study participants gained weight. Weight gain ranged between 3.5 and 14.5 kg. The reminder of the participants (24%) maintained their weight within 3% of baseline values. At baseline, weight losers had a significantly higher body weight, BMI, waist circumference and hip circumference compared with weight-gain and weight-stable groups. There was no difference in age between the three weightchange categories. Results for FM and FFM at baseline and follow-up and respective changes are summarized in Table 3. The 4C model suggested significant losses or gains in FM as well as in FFM. With weight loss, 79 and 21% were explained by FM and FFM respectively. In contrast with weight gain, 90% was explained by FM, with only 10% left for FFM. In weight-loss and weight-gain groups, FM and FFM as measured by all methods significantly differed from baseline values, except for FFMADP that remained unchanged in the weight-gain group. In the weight-stable group, a significant gain in FM was observed using ADP and 4C model, whereas FFM increased according to the DXA measurements. In the weight-loss group, changes in FM and FFM determined by ADP, DXA and D2O did not differ from the reference 4C model (Table 4). In contrast, in the weight-gain group, DXA and D2O significantly underestimated the gain in FM4C (Po0.001, Po0.05) and overestimated the gain in FFM4C (Po0.001, Po0.05), whereas & 2013 Macmillan Publishers Limited

Body composition during weight loss and weight gain M Pourhassan et al

449 Table 2.

Characteristics of the study population stratified by weight change (mean±s.d.) Weight loss (n ¼ 30)

Age (y) Height (m) Weight (kg) BMI (kg/m2) WC (cm) Hip (cm)

Weight gain (n ¼ 33)

T0

T1

DT1  T0

T0

T1

36.93±8.44 1.72±0.08 99.84±18.89a 33.57±5.41a 106.92±12.77a 117.60±13.22a

38.93±9.78

2.00±2.21  11.19±4.92***  3.90±1.74***  10.26±6.64***  7.44±3.64***

36.18±10.00 1.71±0.08 86.52±16.63c 29.49±4.68d 95.67±11.81c 108.19±11.75c

39.03±10.42

88.65±17.23 29.66±4.64 96.65±10.97 110.16±12.11

93.01±18.14 31.56±5.20 99.65±11.10 114.41±11.81

Weight stable (n ¼ 20) DT1  T0 2.84±2.42 6.49±3.326*** 2.07±1.15*** 4.97±6.93** 6.22±7.00***

T0

T1

DT1  T0

35.80±8.01 1.71±0.07 76.23±13.75b 25.86±4.32b 88.60±13.55b 101.00±9.79b

39.45±8.74

3.65±2.10

77.09±14.45 26.01±4.60 88.56±14.06 103.15±10.19

0.86±1.41* 0.15±0.52 0.04±5.01 2.14±5.00

Abbreviations: BMI, body mass index; Hip, hip circumference; WC, waist circumference. *Po0.05, **Po0.01 and ***Po0.001 difference between T0 and T1 within group; aPo0.01 difference between weight loss and weight gain within time point; bPo0.001 difference between weight loss and weight stable within time point; cPo0.05 and dPo0.01 difference between weight gain and weight stable within time point.

Table 3.

Changes in fat mass (FM) and fat-free mass (FFM) over time as measured with D2O, ADP, DXA and 4C within groups (mean±s.d.) Weight loss (n ¼ 30)

Weight gain (n ¼ 33)

Weight stable (n ¼ 20)

T0

T1

DT1  T0

T0

T1

DT1  T0

T0

T1

FM (kg) D2O ADP DXA 4C

37.76±13.71 41.45±15.31 37.55±12.95 38.74±14.52

28.86±10.80 31.96±12.65 29.68±11.34 29.86±11.55

 8.90±6.28***  9.49±5.46***  7.86±3.84***  8.87±5.89***

28.54±10.01 30.79±12.89 29.28±10.33 29.24±11.29

33.56±11.99 37.15±14.46 33.20±11.63 34.89±13.03

5.02±4.10*** 6.36±3.29*** 3.91±2.51*** 5.64±3.36***

21.97±10.18 22.85±12.30 22.08±10.74 22.04±11.10

22.49±10.27 24.02±12.33 22.42±10.58 22.89±11.22

0.51±3.03 1.17±1.67** 0.34±1.47 0.85±1.79*

FFM (kg) D2O ADP DXA 4C

62.07±12.95 58.36±11.11 62.28±12.63 61.10±12.44

59.78±11.13 56.69±9.93 59.44±11.80 58.78±10.72

 2.28±4.06**  1.67±2.08***  2.84±3.10***  2.31±3.02***

57.98±13.03 55.69±12.52 57.24±12.79 57.27±12.92

59.44±12.03 55.83±12.09 60.47±13.52 58.12±12.18

1.46±2.86** 0.13±1.60 3.23±2.25*** 0.84±1.82*

54.26±8.61 53.39±10.24 54.03±9.89 54.18±9.31

54.60±9.39 53.07±10.25 55.48±9.90 54.19±9.83

0.34±2.94  0.31±1.44 1.45±1.23*** 0.01±1.63

74.32±2.20

74.43±1.70

0.10±2.10

74.10±2.45

74.97±2.22

0.86±2.04*

73.49±2.98

73.87±1.94

0.38±2.49

FFMhydration (%)

DT1  T0

Abbreviations: ADP, air-displacement plethysmography; 4C, four-compartment model; D2O, deuterium dilution; DXA, dual-energy X-ray absorptiometry; MRI, magnetic resonance imaging. FFMhydration ¼ TBWD2O /FFM4C. *Po0.05, **Po0.01 and ***Po0.001 difference between T0 and T1 within group.

ADP overestimated the gain in FM4C (Po0.05) and underestimated the gain in FFM4C (Po0.05). We also estimated the FFM hydration by dividing TBWD2O by FFM4C. At baseline, FFM hydration did not differ between weightloss and weight-gain groups. At follow-up, FFM hydration significantly increased (Po0.05) with weight gain. In contrast, there was no significant difference in FFM hydration with weight loss (P ¼ 0.781). Mean result, bias and 95% limit of agreement for changes in fat mass (DFM) and fat-free mass (DFFM) are shown in Table 4. In the total study population, DFFMDXA was overestimated when compared with DFFM4C (Po0.01). Figures 1 and 3 show good absolute agreement between all methods and the 4C-model for assessment of either DFM or DFFM. Limits of agreement (mean bias and 95% confidence interval) were narrow for D2O and wider for the other methods. Systematic errors were observed for the assessment of DFMDXA, DFFMADP, DFFMDXA and DFFMD2O. DXA overestimated the loss and underestimated the gain in DFM (Table 4, Figure 1). By contrast, DXA and D2O underestimated the loss and overestimated the gain in DFFM (Table 4, Figure 3). In addition, ADP overestimated the loss and underestimated the gain in DFFM (Table 4, Figure 3). The bias between DFM4C and DFM assessed by other methods was correlated with BMI (r ¼ 0.23, Po0.05 for ADP, r ¼  0.23, Po0.05 for D2O), change in FFM hydration (r ¼  0.85, Po0.001 for ADP, r ¼  0.26, Po0.05 for DXA and r ¼ 0.98, Po0.001 for D2O) change in waist or hip circumference (r ¼ 0.45, r ¼ 0.38, both Po0.001 for DXA), DFM% and FM% at baseline (r ¼ 0.74, r ¼  0.34, both Po0.001 for DXA). In a stepwise multiple regression analysis with the bias between DFM4C and DFM assessed by alternative methods as dependent variable and age, WC, Hip, FM% at baseline, FM% at follow-up and DFM% and change in FFM hydration as independent variables, BMI and & 2013 Macmillan Publishers Limited

change in FFM hydration explained 5.1 and 73.1% of the variance in bias between DFM4C and DFMADP, respectively. BMI and change in FFM hydration explained 5.1 and 96.3% of the variance in bias between DFM4C and DFMD2O, respectively. DFM% and change in FFM hydration explained 54.3 and 6.8% of the variance in bias between DFM4C and DFMDXA, respectively. Other variables did not contribute to the variance in bias between DFM4C and DFM assessed by alternative methods. The bias between DFFM4C and DFFM assessed by other methods was also correlated with BMI (r ¼ 0.25, Po0.05 for DXA, r ¼ 0.23, Po0.05 for D2O, r ¼  0.24, Po0.05 for ADP), change in FFM hydration (r ¼ 0.85, Po0.001 for ADP, r ¼ 0.22, Po0.05 for DXA and r ¼  0.98, Po0.001 for D2O), hip circumference (r ¼ 0.28, Po0.01 for DXA), FM% at baseline (r ¼ 0.47, Po0.001 for DXA) and change in waist or hip circumference (r ¼  0.45, r ¼  0.35, both Po0.001 for DXA). In a stepwise multiple regression analysis using the bias between DFFM4C and DFFM assessed by alternative methods as dependent variable and age, WC, Hip, FM% at baseline, FM% at follow-up and DFM% and change in FFM hydration as independent variables, baseline FM% and change in FFM hydration explained 11.9 and 73.5% of the variance in bias between DFFM4C and DFFMADP, respectively. Baseline FM% and change in FFM hydration explained 54.3 and 5.0% of the variance in bias between DFFM4C and DFFMDXA, respectively. BMI and change in FFM hydration explained 5.1 and 96.3% of the variance in bias between DFFM4C and DFFMD2O. Other independent variables did not contribute to the variance in bias between DFFM4C and DFFM assessed by alternative methods. Table 5 compares changes in regional FM and LST using DXA with changes in AT and SM using MRI. In the weight-loss group, total weight loss was estimated to consist of 25.3% LST (mainly European Journal of Clinical Nutrition (2013) 446 – 454

Body composition during weight loss and weight gain M Pourhassan et al

450 Table 4.

Results of the limit of agreement analysis: mean result (±s.d.), bias and 95% limit of agreement for changes in fat mass (DFM) and fat-free mass (DFFM) measured by ADP, DXA and D2O and compared with results from the 4C model. ADP

DXA

D2O

All (n ¼ 83) DFM (kg) Bias vs DFM4C kga Correlation,b r DFFM (kg) Bias vs DFFM4C kga Correlation,b r

 0.62±8.04  0.14±1.94  0.19  0.62±1.91 0.13±1.94 0.42**

 1.20±5.96 0.45±2.87 0.59** 0.60±3.61  1.11±3.25ww  0.33**

 1.09±7.79 0.34±1.67  0.08  0.15±3.72  0.33±1.66  0.64**

Weight loss (n ¼ 30) DFM (kg) Bias vs DFM4C kga Correlation,b r DFFM (kg) Bias vs DFFM4C kga Correlation,b r

 9.49±5.46 0.62±2.19 0.19  1.67±2.08  0.65±2.20 0.46**

 7.86±3.84  1.01±3.16 0.66**  2.84±3.10 0.53±3.77  0.02

 8.90±6.28 0.03±1.76  0.22  2.28±4.06  0.03±1.76  0.60**

Weight gain (n ¼ 33) DFM (kg) Bias vs DFM4C kga Correlation,b r DFFM (kg) Bias vs DFFM4C kga Correlation,b r

6.36±3.29  0.72±1.60c 0.04 0.13±1.60 0.71±1.58c 0.16

3.91±2.51 1.73±2.52www 0.37* 3.23±2.25  2.39±2.78www  0.19

5.02±4.10 0.62±1.56c  0.48** 1.46±2.86  0.63±1.56c  0.69**

Abbreviations: ADP, air-displacement plethysmography; 4C, four-compartment model; D2O, deuterium dilution; DXA, dual-energy X-ray absorptiometry. *Po0.05 and **Po0.01. aBias was calculated as result obtained from reference method (4C) minus ADP, DXA and D2O measurement. 95% limits of agreement was calculated as ±2s.d. bCorrelation was calculated as Pearson correlation coefficient for the relationship between (resultreference c method þ resultother methods)/2 and the bias. Significant difference between reference method (4C) and results from ADP, DXA and D2O by paired samples t-test (wPo0.05, wwPo0.01 and wwwPo0.001).

explained by a loss in LSTlegs) and 10.9% SM (mainly explained by a loss in SMtrunk), as well as 74.7% FM (mainly explained by a loss in FMtrunk) and 89.1% AT (mainly explained by a loss in SATtrunk). In the weight gainer group, total weight gain consisted of 51.1% LST (mainly explained by a gain in LSTtrunk) and 26.1% SM (mainly explained by a gain in SMlegs), as well as 48.9% FM (mainly explained by a gain in FMtrunk) and 73.9% AT (mainly explained by a gain in SATtrunk). Using DXA, subjects lost more FM than LST (Po0.001). With weight gain, subjects gained approximately similar amounts of LST and FM (P ¼ 0.787). More LST was gained in the weight gainer group than was lost in the weight loser group. Using MRI, subjects lost more absolute AT than SM (Po0.001). Vice versa with weight gain, subjects gained more AT compared to SM (Po0.01). Approximately similar amounts of SM were gained in the weight gainer group and was lost in the weight loser group. In both weight change groups, the changes in FM were proportionally greater than changes in FFM (Table 3). In the weight loser group, the losses in FM as a percentage of initial FM were similar between methods (  22.3% (ADP),  20.9% (DXA) and  21.8% (D2O)) and all not significantly different from 4Cmodel (  21.5%) (Figure 2). Losses in FFM as a percentage of initial FFM were also similar between methods (  2.6% (ADP),  4.3% (DXA) and  3.0% (D2O)) and all not significantly different from 4C-model (  3.3%) (Figure 2). By contrast, in the weight gainer group, the gain in FM as a percentage of initial FM significantly differed between ADP and DXA (23.8 vs 14.1%) and between 4C and DXA (21.1 vs 14.1%). In addition, ADP estimated lower (0.4%) and DXA estimated higher (5.7%) gains in FFM as a percentage of initial FFM compared to 4C-model (1.9%). When compared with the 4C-refrence method, all other methods showed a different composition of weight gain (as a percentage of initial body composition) (Figure 2). European Journal of Clinical Nutrition (2013) 446 – 454

DISCUSSION Comparison of changes in body composition as assessed with different methods The primary aim of the present study was to compare the composition of weight loss and weight gain between methods using a 4C model as a reference. No significant differences were observed in changes in FM and FFM as measured by ADP, DXA or D2O when compared with the 4C-model (Table 4) with the exception of changes in FFMDXA. DXA systematically underestimated the loss and overestimated the gain in FFM with weight changes (Po0.01, Bland–Altman analysis in Figure 3b). This resulted in a significant overestimation of FFM gain in weight gainers (Table 4). Our findings are in line with Schoeller et al.27 who reported that QDR 4500 DXA overestimated FFM when compared with a 4-C model. Accordingly, Minderico et al.14 have shown that DXA overestimated FM loss as measured with QDR-1500 (pencilbeam mode) compared with a 4-C model in overweight and obese women, the bias increased with the degree of overweight and obesity. This might be due to tissue thickness that increases with increasing body weight. Overestimation of %FM by DXA was observed at a higher tissue thickness.28,29 In case of a fan beam technology, this leads to a magnification error. Williams et al.30 found a higher error in obese subjects using fan beam technology. In the present study, the bias between 4C-model and DXA was mainly explained by FM% at baseline whereas the change in FFM hydration contributed to additional 5% of the bias only (see results). With increasing adiposity, FFM hydration increases due to a higher water content of the lean compartment of AT.31 However, due to similar densities of fat and water, the attenuation coefficient of X-rays for both compartments is similar. That would imply an underestimation of lean mass with increasing FFM hydration in obesity or with weight gain. Since the bias we found was in opposite direction (for example, overestimation of lean mass with weight gain), the erroneous assumption of a constant hydration of FFM is unlikely to be the cause of the bias between DXA and 4C-model. In addition, different instrument manufacturers and software versions may also affect the accuracy of DXA results.28,32 Accordingly, the limits of agreement were wider for DXA when compared with the other methods; 5.39 to  7.60 kg DFFM for DXA compared with 4.01 to  3.76 kg DFFM for ADP and 2.99 to  3.67 kg DFFM for D2O). Roemmich et al.33 also found that DXA produced larger limits of agreement than other 2C age-adjusted models. In our total population, changes in body composition measured by D2O were not significantly different from the 4C-model (Table 4). However, similar to DXA, Bland–Altman analysis revealed a systematic bias (Figure 3c) and D2O significantly overestimated the gain in DFFM4C in the subgroup of weight gainers (Po0.05; Table 4). This may also be explained by the erroneous assumption of a constant hydration of FFM.34 In line with this hypothesis, the bias between 4C-model and D2O correlated with BMI, as well as with changes in FFM hydration (see results). About 15–30% of TBW is present in AT, which increases with increasing adiposity.35 Because FM is higher in women and obese individuals, the higher hydration of FFM causes an underestimation of FFM and overestimation of FM.36 The present study population primarily consists of women (71%) and overweight and obese subjects (79.5%). Therefore, mean FFM hydration was higher (74%) at baseline when compared with the hydration status assumed for normal weight individuals (73.2%). By contrast, FFM hydration did not normalize after weight loss (Table 3). These results are in line with the study of Das et al.37 who found no significant difference in FFM hydration of extremely obese patients after massive weight loss (  44 kg) caused by gastric bypass surgery. Hence we postulate that 2-C models (FMs and FFMs) that assume a constant hydration of FFM have a systematic error in weight-reduced subjects. & 2013 Macmillan Publishers Limited

Body composition during weight loss and weight gain M Pourhassan et al

451 15

15

4C - ADP

R2 = 0.3592

10

10 ΔFM4C - ΔFMDXA (kg)

ΔFM4C - ΔFMADP (kg)

4C - DXA

5 0 -5 -10 -15 -30 -25 -20 -15 -10 -5

5 0 -5 -10 -15

0

5

10 15 20

-30 -25 -20 -15 -10 -5

(ΔFM4C + ΔFMADP) / 2 (kg)

0

5

10

15

20

(ΔFM4C + ΔFMDXA) / 2 (kg) 4C - D2O

15

Female

Male

ΔFM4C - ΔFMD2O (kg)

10 5 0 -5 -10 -15 -30 -25 -20 -15 -10

-5

0

5

10

15

20

(ΔFM4C + ΔFMD2O) / 2 (kg)

Figure 1. Bland–Altman plots of limits of agreement for changes in FM (kg) between 4C and (a) ADP, (b) DXA and (c) D2O. Solid line indicates the mean difference and dashed lines indicate ±2 s.d. Open symbols for females; closed symbols for males.

weight gainer group ** weight loser group

***

40 30

Fat mass Fat free mass

Change in FM & FFM (as percentage of initial FM & FFM)

Change in FM & FFM (as percentage of initial FM & FFM)

**

20 10 0 -10 -20 -30 -40 4C

ADP

DXA

D2O

***

*

40 30 20 10 0 -10 -20 -30

Fat mass Fat free mass

-40 4C

ADP

DXA

D2O

Figure 2. Mean (±s.d) changes in FM and FFM as a percentage of initial FM and FFM shown for (a) weight losers (n ¼ 30) and (b) weight gainers (n ¼ 33). Changes in FM and FFM as a percentage of initial FM and FFM were significantly different in weight-gain group, *Po0.05, **P o0.01 and ***Po0.001 (paired t-test). No significant differences were observed in changes in FM and FFM as a percentage of initial FM and FFM in weight-loss group.

In contrast to D2O and DXA, densitometry (ADP) systematically overestimated the loss and underestimated the gain in FFM (Table 4). This is in line with data of Fields et al.38 who suggested that ADP underestimates FFM and overestimates FM in overweight adults. Comparable to DXA and D2O, hydration of FFM is likely the main source of the bias between ADP and a 4-C model. However, the bias is in the opposite direction because densities of fat and water are very similar and thus an increasing water fraction of FFM in obese people or after weight gain is misinterpreted as an increase in FM. In the present study, the bias between 4C-model and ADP with weight loss or weight gain was related to BMI and the change in FFM hydration (see results). Comparison of the composition of weight loss and weight gain Comparing FFM and FM, proportionally more FFM4C (3.3% as a percentage of initial FFM4C) was lost during weight loss than was & 2013 Macmillan Publishers Limited

gained (1.89% as a percentage of initial FFM4C) during weight gain (Figure 2). A 1 kg weight loss consisted of 0.20 kg FFM whereas a 1 kg weight gain is explained by 0.12 kg FFM only; this was not related to the time of follow up. Our data compare changes in two different populations. In contrast to our protocol, Beavers et al.6 investigated intra-individual changes in body composition (as measured by DXA) with weight loss and regain in a 6 months trial involving obese women. In that study, each 1 kg fat loss was associated with a loss of 0.26 kg FFM whereas the regain of 1 kg fat was associated with a gain of 0.12 kg lean tissue.6 Thus, these data are similar to our results. The implication of these findings is that in the long term weight cycling can promote sarcopenia which may be even worse in elderly patients. In fact, the results of the Health Aging and Body Composition Study have shown that for every kg weight loss there was a 0.42 kg and 0.06 kg loss of FFM in men and women, respectively, whereas for every kg of weight gain 0.37 and 0.32 kg of FFM increased.17 However, other European Journal of Clinical Nutrition (2013) 446 – 454

Body composition during weight loss and weight gain M Pourhassan et al

452 Table 5.

Changes in regional LST and FM using DXA and SM and AT using MRI within groups (mean±s.d). Weight loser (n ¼ 17)

Weight gainer (n ¼ 29)

T0

T1

DT1  T0

T0

T1

DXA LSTtrunk (kg) LSTarms (kg) LSTlegs (kg)

29.98±6.14 7.61±2.27 22.16±5.34

30.63±5.67 6.71±1.86 21.18±4.78

0.64±2.38  0.89±1.18**  0.98±1.45**

26.09±5.32 6.08±2.03 18.93±4.73

29.43±6.22 6.34±1.99 19.86±4.91

MRI SMtrunk (kg) SMarms (kg) SMlegs (kg)

11.01±2.83 4.39±1.34 15.47±4.20

10.30±2.32 4.42±1.28 15.21±3.75

 0.71±1.15* 0.02±0.42  0.26±1.22

9.71±3.12 3.64±1.09 13.45±3.61

DXA FMtrunk (Kg) FMarms (Kg) FMlegs (Kg)

15.90±5.37 3.94±1.58 11.92±6.03

13.23±4.61 3.02±1.15 9.88±5.45

 2.66±1.95***  0.91±0.54***  2.04±1.33***

MRI ATtrunk (l) SATarms (l) SATlegs (l)

19.88±6.13 3.61±1.23 13.42±6.34

14.52±5.92 3.05±1.05 11.23±6.13

 5.36±2.51***  0.56±0.32***  2.19±1.41***

Weight stable (n ¼ 18) DT1  T0

T0

T1

DT1  T0

3.33±2.85*** 0.26±0.43* 0.92±0.94***

23.91±4.51 5.68±1.57 17.35±3.06

26.30±4.64 5.51±1.59 17.54±3.13

2.39±1.54***  0.16±0.51 0.18±0.76

9.50±2.87 3.93±1.38 14.28±3.89

 0.21±0.95 0.29±0.52** 0.83±0.70***

8.57±2.13 3.18±0.95 11.83±2.21

8.32±1.92 3.56±1.03 12.64±2.37

 0.25±0.81 0.38±0.58* 0.81±0.98**

12.54±4.50 3.40±1.32 11.00±4.91

15.26±5.26 3.77±1.53 12.25±5.61

2.72±1.82*** 0.36±0.39*** 1.24±1.05***

9.58±5.89 2.76±1.47 9.06±3.92

10.08±5.76 2.68±1.44 9.08±3.88

0.53±1.01*  0.07±0.26 0.02±0.71

16.19±5.64 3.12±1.14 12.44±5.18

17.69±6.90 3.56±1.39 13.82±6.23

1.49±2.52** 0.43±0.56*** 1.37±1.21***

12.86±6.90 2.52±1.00 9.90±3.71

11.53±7.61 2.57±1.12 10.08±4.54

 1.32±1.58** 0.04±0.24 0.18±1.72

Abbreviations: AT, adipose tissue; ATtrunk, AT of the trunk (subcutaneous AT of the trunk þ visceral AT); DXA, dual X-ray absorptiometry; FM, fat mass; FMarms, FM of the arms; FMlegs, FM of the legs; FMtrunk, FM of the trunk; LST, lean soft tissue; LSTarms, LST of the arms; LSTlegs, LST of the legs; LSTtrunk, LST of the trunk; MRI, magnetic resonance imaging; SATarms, subcutaneous AT of the arms; SATlegs, subcutaneous AT of the legs; SM, skeletal muscle; SMarms, SM of the arms; SMlegs, SM of the legs; SMtrunk, SM of the trunk. *Po0.05, **Po0.01 and ***Po0.001 difference between T0 and T1 within group.

15

4C - ADP

4C - DXA R2 = 0.1097

R2 = 0.1798 10 ΔFFM4C - ΔFFMDXA (kg)

ΔFFM 4C - ΔFFM ADP (kg)

15 10 5 0 -5 -10

-15 -10

-5

0

5

10

5 0 -5 -10 -15 -10

15

(ΔFFM 4C + ΔFFM ADP) / 2 (kg)

ΔFFM 4C - ΔFFM D2O (kg)

0

5 DXA)

10

15

/ 2 (kg)

4C - D2O

15 10

-5

(ΔFFM 4C + ΔFFM

R2 = 0.4092

Female Male

5 0 -5 -10 -15 -10

-5

0

5

10

15

(ΔFFM 4C + ΔFFM D2O) / 2 (kg)

Figure 3. Bland–Altman plots of limits of agreement for changes in FFM (kg) between 4C and (a) ADP, (b) DXA and (c) D2O. Solid line indicates the mean difference and dashed lines indicate ±2 s.d. Open symbols for females, closed symbols for males.

longitudinal studies of body composition in older adults show a gain of FM and a loss of lean mass with time in weight stable individuals.39,40 In one of our recent studies we intra-individually followed weight loss and re-gain over 6 months of follow up in 103 overweight and obese subjects.41 The intra-individual comparison between the different body components lost and re-gained revealed that the regain was in proportion to weight loss except for a higher regain in AT of the extremities in women and a lower regain in extremity and visceral AT in man. There was also a lag of SM regain in the trunk behind the extremities. These data argued against the idea that after weight loss weight regain adversely affects fat distribution. However, in that study weight European Journal of Clinical Nutrition (2013) 446 – 454

regain was not complete and body weight as well as body composition differed between baseline and after the 6 months observation period. Comparison of regional changes between weight loser and weight gainer groups We have also compared the regional changes in FMDXA and ATMRI as well as LSTDXA and SMMRI with weight loss and weight gain. Using DXA, lean mass is mainly lost at the arms and legs and gained at the trunk whereas; by contrast, using MRI lean mass is mainly lost at the trunk and gained at the extremities (Table 5). & 2013 Macmillan Publishers Limited

Body composition during weight loss and weight gain M Pourhassan et al

453 Using DXA, weight change consisted of a loss (  47.4%), and a gain in FMtrunk ( þ 63.0%). When using MRI, weight change consisted of a loss (  66.0%) and a gain in ATtrunk ( þ 45.3%). These results suggest that there may be a regional redistribution of fat and lean mass with weight loss and weight regain. However, future research needs to be done using an intra-individual study design. Because AT does not resemble FM and LST is higher than SM (due to connective tissue and organ mass) absolute differences in the composition of weight loss and weight gain between both methods are obvious. However, this does not explain the differences in fat and lean redistribution with weight loss and weight gain that may partly be due to inherent assumptions of DXA software. Taken together, data on DXA- and MRI-derived changes in regional body composition cannot be directly compared with each other. The limitations of the DXA approach (and presumably the 2 component models) has to be taken into account. Study strengths and limitations Some limitations to the present study should be discussed. The number of men was small (n ¼ 24), therefore sex differences cannot be addressed with confidence. In addition, physical activity and fitness have not been addressed which are known to have an impact on the composition of weight change.42 Since in this study we did not assess intra-individual weight cycles, we cannot directly compare the composition of weight loss and weight gain. The strengths of this study is the concomitant use of a variety of highly standardized body composition techniques including a 4C-model as a gold standard that avoids the assumptions of different 2C-methods that may be violated during unstable conditions of weight loss and weight gain. It should be mentioned that 4C is balancing of the three measurements, with amended measurement errors from all of them. In addition, imaging technology allowed the evaluation of regional changes in body composition with weight loss and weight gain. CONCLUSIONS When compared with the 4C model ( ¼ gold standard), mean bias of D2O and densitometry methods is explained by the erroneous assumption of a constant hydration of FFM. This assumption leads to an underestimation of FM change measured by D2O and an overestimation of FM change measured by densitometry. Because hydration does not normalize after weight loss we can deduce that all 2C-models that are based on the assumption of a constant hydration of FFM have a systematic error in weight reduced subjects. The bias between 4C-model and DXA was mainly explained by FM% at baseline whereas the change in FFM hydration only contributed to additional 5% of the bias. As to the regional changes in body composition, MRI data cannot be replaced by DXA measurements. CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS The authors wish to thank Britta Jux, Klinik fu¨r Radiologische Diagnostik, UKSH Kiel, for their help in MRI scanning. The study was funded by Deutsche Forschungsgemeinschaft (DFG Mu¨ 714/ 8-3) BMBF Kompetenznetz Adipositas, Core domain ‘‘Body composition’’ (Ko¨rperzusammensetzung; FKZ 01GI1125)

DISCLOSURE The sponsor of the study (DFG, BMBF) had no role in study design, the collection, analysis and the interpretation of the data, writing the text or in the decision to submit the manuscript.

& 2013 Macmillan Publishers Limited

AUTHOR CONTRIBUTIONS ABW and MJM designed and supervised the study, ABW, MP and MJM wrote the final version of the manuscript, MJM and ABW had primary responsibility for the final content of the manuscript. ABW and WL performed all the investigations. BS, WL, MP organized the study, collected the data, did the segmentations of whole body MRI data and performed the statistical analyses. C-CG was responsible for DXA and MRI examinations.

REFERENCES 1 Van Gaal LF, Mertens IL, De Block CE. Mechanisms linking obesity with cardiovascular disease. Nature 2006; 444: 875–880. 2 Heitmann BL, Garby L. Composition (lean and fat tissue) of weight changes in adult Danes. Am J Clin Nutr 2002; 75(5): 840–847. 3 Mikkelsen KL, Heitmann BL, Keiding N, Sorensen TI. Independent effects of stable and changing body weight on total mortality. Epidemiology 1999; 10(6): 671–678. 4 Newman AB, Yanez D, Harris T, Duxbury A, Enright PL, Fried LP et al. Weight change in old age and its association with mortality. J Am Geriatr Soc 2001; 49(10): 1309–1318. 5 Newman AB, Lee JS, Visser M, Goodpaster BH, Kritchevsky SB, Tylavsky FA et al. Weight change and the conservation of lean mass in old age: the Health, Aging and Body Composition Study. Am J Clin Nutr 2005; 82: 872–878quiz 915-6. 6 Beavers KM, Lyles MF, Davis CC, Wang X, Beavers DP, Nicklas BJ. Is lost lean mass from intentional weight loss recovered during weight regain in postmenopausal women? Am J Clin Nutr 2011; 94: 767–774. 7 Ritz P, Salle A, Audran M, Rohmer V. Comparison of different methods to assess body composition of weight loss in obese and diabetic patients. Diabetes Res Clin Pract 2007; 77: 405–411. 8 Van Wye G, Dubin JA, Blair SN, Di Pietro L. Weight cycling and 6-year weight change in healthy adults: the Aerobics Center Longitudinal Study. Obesity 2007; 15: 731–739. 9 Mason C, Foster-Schubert KE, Imayama I, Xiao L, Kong A, Campbell KL et al. History of weight cycling does not impede future weight loss or metabolic improvements in postmenopausal women. Metabolism 2012; 62: 127–136. 10 Hensrud DD, Weinsier RL, Darnell BE, Hunter GR. A prospective-study of weight maintenance in obese subjects reduced to normal body-weight without weightloss training. Am J Clin Nutr 1994; 60: 688–694. 11 Gately PJ, Radley D, Cooke CB, Carroll S, Oldroyd B, Truscott JG et al. Comparison of body composition methods in overweight and obese children. J Appl Physiol 2003; 95: 2039–2046. 12 Baumgartner RN, Heymsfield SB, Lichtman S, Wang J, Pierson RN Jr. Body composition in elderly people: effect of criterion estimates on predictive equations. Am J Clin Nutr 1991; 53: 1345–1353. 13 Korth O, Bosy-Westphal A, Zschoche P, Gluer CC, Heller M, Muller MJ. Influence of methods used in body composition analysis on the prediction of resting energy expenditure. Eur J Clin Nutr 2007; 61: 582–589. 14 Minderico CS, Silva AM, Keller K, Branco TL, Martins SS, Palmeira AL et al. Usefulness of different techniques for measuring body composition changes during weight loss in overweight and obese women. Br J Nutr 2008; 99: 432–441. 15 Tchoukalova YD, Koutsari C, Karpyak MV, Votruba SB, Wendland E, Jensen MD. Subcutaneous adipocyte size and body fat distribution. Am J Clin Nutr 2008; 87: 56–63. 16 Coin A, Giannini S, Minicuci N, Rinaldi G, Pedrazzoni M, Minisola S et al. Limb fatfree mass and fat mass reference values by dual-energy X-ray absorptiometry (DEXA) in a 20–80 year-old Italian population. Clin Nutr 2012; 31: 506–511. 17 Lee JS, Visser M, Tylavsky FA, Kritchevsky SB, Schwartz AV, Sahyoun N et al. Weight loss and regain and effects on body composition: the Health, Aging, and Body Composition Study. J Gerontol A Biol Sci Med Sci 2010; 65: 78–83. 18 St-Onge MP, Gallagher D. Body composition changes with aging: the cause or the result of alterations in metabolic rate and macronutrient oxidation? Nutrition 2010; 26: 152–155. 19 Stevens J, Truesdale KP, McClain JE, Cai J. The definition of weight maintenance. Int J Obes (Lond) 2006; 30: 391–399. 20 Siri WE. Body composition from fluid spaces and density: analysis of methods. 1961. Nutrition 1993; 9(5): 480–491, discussion, 92. 21 Fuller NJ, Jebb SA, Laskey MA, Coward WA, Elia M. Four-component model for the assessment of body composition in humans: comparison with alternative methods, and evaluation of the density and hydration of fat-free mass. Clin Sci 1992; 82: 687–693. 22 Brozek J, Grande F, Anderson JT, Keys A. Densitometric analysis of body composition: revision of some quantitative assumptions. Ann N Y Acad Sci 1963; 110: 113–140.

European Journal of Clinical Nutrition (2013) 446 – 454

Body composition during weight loss and weight gain M Pourhassan et al

454 23 Bosy-Westphal A, Kossel E, Goele K, Blocker T, Lagerpusch M, Later W et al. Association of pericardial fat with liver fat and insulin sensitivity after diet-induced weight loss in overweight women. Obesity 2010; 18: 2111–2117. 24 Later W, Bosy-Westphal A, Kossel E, Gluer CC, Heller M, Muller MJ. Is the 1975 Reference Man still a suitable reference? Eur J Clin Nutr 2010; 64: 1035–1042. 25 Snyder WS CM, Nasset ES, Karhausen LR, Howells GP, Tipton IH. Report of the Task Group on Reference Man. Pergamon Press: Oxford, New York, USA, 1975. 26 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Int J Nurs Stud 2010; 47: 931–936. 27 Schoeller DA, Tylavsky FA, Baer DJ, Chumlea WC, Earthman CP, Fuerst T et al. QDR 4500A dual-energy X-ray absorptiometer underestimates fat mass in comparison with criterion methods in adults. Am J Clin Nutr 2005; 81: 1018–1025. 28 Nakata Y, Tanaka K, Mizuki T, Yoshida T. Body composition measurements by dual-energy X-ray absorptiometry differ between two analysis modes. J Clin Densitom 2004; 7: 443–447. 29 Jebb SA. Measurement of soft tissue composition by dual energy X-ray absorptiometry. Br J Nutr 1997; 77: 151–163. 30 Williams JE, Wells JC, Wilson CM, Haroun D, Lucas A, Fewtrell MS. Evaluation of Lunar Prodigy dual-energy X-ray absorptiometry for assessing body composition in healthy persons and patients by comparison with the criterion 4-component model. Am J Clin Nutr 2006; 83: 1047–1054. 31 Wang Z, Deurenberg P, Wang W, Pietrobelli A, Baumgartner RN, Heymsfield SB. Hydration of fat-free body mass: review and critique of a classic bodycomposition constant. Am J Clin Nutr 1999; 69: 833–841. 32 Lohman T. Human Body Composition. In: Heymsfield SB LST, Wang Z (eds). Human Body Composition. 2nd edn, pp 63–78 (Human Kinetics: Springfield, IL, USA, 2005.

European Journal of Clinical Nutrition (2013) 446 – 454

33 Roemmich JN, Clark PA, Weltman A, Rogol AD. Alterations in growth and body composition during puberty I. Comparing multicompartment body composition models. J Appl Physiol 1997; 83: 927–935. 34 Fields DA, Goran MI. Body composition techniques and the four-compartment model in children. J Appl Physiol 2000; 89: 613–620. 35 Chumlea WC, Schubert CM, Sun SS, Demerath E, Towne B, Siervogel RM. A review of body water status and the effects of age and body fatness in children and adults. J Nutr Health Aging 2007; 11: 111–118. 36 Chumlea WC. Body composition assessment of obesity. In: RD Bray GA (eds). Overweight and the metabolic syndrome. from bench to bedside New York. Springer: New York, 2006 23–35. 37 Das SK, Roberts SB, Kehayias JJ, Wang J, Hsu LK, Shikora SA et al. Body composition assessment in extreme obesity and after massive weight loss induced by gastric bypass surgery. Am J Physiol Endocrinol Metab 2003; 284: E1080–E1088. 38 Fields DA, Wilson GD, Gladden LB, Hunter GR, Pascoe DD, Goran MI. Comparison of the BOD POD with the four-compartment model in adult females. Med Sci Sports Exerc 2001; 33: 1605–1610. 39 Gallagher D, Ruts E, Visser M, Heshka S, Baumgartner RN, Wang J et al. Weight stability masks sarcopenia in elderly men and women. Am J Physiol Endocrinol Metab 2000; 279: E366–E375. 40 Forbes GB. Longitudinal changes in adult fat-free mass: influence of body weight. Am J Clin Nutr 1999; 70: 1025–1031. 41 Bosy-Westphal A, Schautz B, Lagerpusch M, Pourhassan M, Braun W, Goele K. Effect of weight loss and regain on adipose tissue distribution, composition of lean mass, and resting energy expenditure in young overweight and obese adults. Int J Obesity. 2013, e-pub ahead of print, 5 February 2013, doi:10.1038/ijo.2013.1. 42 Kyle UG, Zhang FF, Morabia A, Pichard C.. Longitudinal study of body composition changes associated with weight change and physical activity. Nutrition 2006; 22: 1103–1111.

& 2013 Macmillan Publishers Limited