body composition following stem cell transplant

Accepted Manuscript

Body composition following stem cell transplant: comparison of bioimpedance and air-displacement plethysmography

Yun-Chi Hung, BHlthSci Judith D. Bauer, PhD Pamela Horsely, GradDipDieteticsLeigh C. Ward, PhD John Bashford, RACP Elisabeth A. Isenring, PhD

PII: S0899-9007(14)00083-5

DOI: 10.1016/j.nut.2014.01.017

Reference: NUT 9217

To appear in: Nutrition

Received Date: 8 August 2013

Revised Date: 2 December 2013

Accepted Date: 29 January 2014

Please cite this article as: Hung Y-C, Bauer JD, Horsely P, Ward LC, Bashford J, Isenring EA,Body composition following stem cell transplant: comparison of bioimpedance and air-displacementplethysmography, Nutrition (2014), doi: 10.1016/j.nut.2014.01.017.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by University of Queensland eSpace

https://core.ac.uk/display/43342665?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1

http://dx.doi.org/10.1016/j.nut.2014.01.017

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPTStem cell transplant and impedance

1

Body composition following stem cell transplant: comparison of bioimpedance and air-

displacement plethysmography

1Yun-Chi Hung BHlthSci,

1,2Judith D Bauer PhD,

3Pamela Horsely GradDipDietetics,

4Leigh C

Ward PhD, 5John Bashford RACP and

1,6Elisabeth A Isenring PhD

1 Centre for Dietetics Research, School of Human Movement Studies, The University of

Queensland, Brisbane, Australia 2 The Wesley Research Institute, Brisbane, Queensland, Australia

3 Nutrition Services Department, The Wesley Hospital, Brisbane, Queensland, Australia

4 School of Chemistry and Molecular Biosciences, The University of Queensland, St Lucia,

Australia 5 Haematology & Oncology Clinics of Australia, The Wesley Medical Centre, Brisbane,

Australia 6 Department of Nutrition & Dietetics, Princess Alexandra Hospital, Brisbane, Queensland,

Australia.

Running title: Stem cell transplant and impedance

Address for correspondence: Yun-Chi Hung

Building 26B

School of Human Movement Studies,

The University of Queensland,

Brisbane, QLD,

Australia 4072

Telephone: +61 7 3365 6313

Fax: +61 7 3365 6877

E-mail: [email protected]

Word count: 5297

Number of Figures: 4

Number of Tables: 2

MANUSCRIP

T

ACCEPTED


2

Acknowledgements 1

We thank The Wesley Research Institute for funding the study, and the use of the body 2

composition laboratory. We are grateful for the assistance of Catherine Kirk and the 3

cooperation of participating patients. 4

Contribution of Authors 5

Yun-Chi Hung contributed to the conception and design of the study; generation, collection, 6

assembly, analysis and interpretation of data; drafting of the manuscript; and approval of 7

the final version of the manuscript. Judith Bauer contributed to the conception and design 8

of the study; revision of the manuscript; and the approval of the final version of the 9

manuscript. Pamela Horsley contributed to the conception and design of the study; and the 10

approval of the final version of the manuscript. Leigh Ward contributed to the analysis of 11

the data; revision of the manuscript; and the approval of the final version of the manuscript. 12

John Bashford contributed to the conception and design of the study; and the approval of 13

the final version of the manuscript. Elisabeth Isenring contributed to the conception and 14

design of the study; revision of the manuscript; and the approval of the final version of the 15

manuscript. 16

Conflicts of Interest 17

Author Ward consults to ImpediMed Ltd., a manufacturer of impedance devices. 18

ImpediMed Ltd had no involvement in the design, execution of this study or the preparation 19

of this manuscript. Other authors have no conflicts of interest to report. 20

21

22

23

MANUSCRIP

T

ACCEPTED


3

Abstract 24

Objective: To assess the agreement between detected changes in body composition 25

determined by bioimpedance spectroscopy (BIS) and air-displacement plethysmography 26

(ADP) amongst cancer patients undergoing peripheral blood stem cell transplantation 27

(PBSCT); to assess the agreement of absolute values of BIS with ADP and dual energy x-ray 28

(DXA). Methods: Forty-four adult haematological cancer patients undergoing PBSCT 29

completed both BIS and ADP assessment at pre-admission and at three months post-30

transplantation. Body composition was assessed using BIS (ImpSFB7, Impedimed, Brisbane, 31

Australia) and ADP (BOD POD, COSMED, Concord, CA, USA). A subsample (n=11) were 32

assessed by DXA (Hologic, QDR 4500A fan-beam scanner, MA, USA) at post-transplantation. 33

Results were examined for the BIS instrument’s default setting and three alternative 34

predictive equations from the literature. Agreement was assessed by the Bland-Altman 35

limits of agreement analysis while correlation was examined using the Lin’s concordance 36

correlation. Results: Changes in body composition parameters assessed by BIS were 37

comparable to those determined by ADP regardless of the predictive equations used. Bias of 38

change in fat free mass was clinically acceptable (all <1 kg), although limits of agreement 39

were wide (> ±6kg). Overall, the BIS predictive equation accounting for body mass index 40

performed the best. Absolute body composition parameters predicted by the alternative 41

predictive equations agreed with DXA and ADP better than the BIS instrument’s default 42

setting. Conclusion: Changes predicted by BIS were similar to those determined by ADP at a 43

group level, however, changes at an individual level should be interpreted with caution due 44

to wide limits of agreement. 45

Key Words: Body composition, stem cell transplantation, cancer, air-displacement 46

plethysmography, bioelectrical Impedance, spectroscopy 47

MANUSCRIP

T

ACCEPTED


5

Introduction 49

Unintentional weight loss and a decline in nutritional status are frequently reported 50

amongst cancer patients due to a combination of treatment side effects or the disease itself. 51

Change in the proportion of fat mass (FM) and fat-free mass (FFM) can be variable 52

depending on the type and stage of disease. Impaired FFM needs to be identified because 53

unfavourable body composition changes are associated with adverse clinical outcomes such 54

as mortality [1], reduced functioning, and poorer quality of life [2]. In a previous study 55

conducted by the authors [3], reduced nutritional status was accompanied by a concurrent 56

loss of LBM, and reduced quality of life after a group of cancer patients underwent PBSCT. 57

Bioimpedance technology such as bioimpedance spectroscopy (BIS) is a portable, 58

computerized technology for body composition assessment that is non-invasive and easy to 59

operate and relatively inexpensive. This technology provides accurate results amongst 60

healthy subjects [4, 5] but a limited number of studies have examined its validity amongst 61

the cancer population and the inter-changeability of BIS with other laboratory measures. 62

How portable devices such as BIS compare with laboratory methods such as dual-energy X-63

ray absorptiometry (DXA), and air-displacement plethysmography (ADP) is of interest 64

because compliance to routine assessments requiring complex procedures or low 65

convenience (i.e. need to travel) can be challenging amongst critically-ill patients. 66

In this study, the body composition of adult cancer patients treated with peripheral blood 67

stem cell transplantation (PBSCT) was examined. The aims of this study were to compare 68

the agreement of absolute values estimated by BIS relative to ADP, and DXA; and changes in 69

body composition detected by BIS relative to ADP at three months post-PBSCT. 70

Methods 71

MANUSCRIP

T

ACCEPTED


6

Subjects were haematological cancer patients treated with PBSCT at a single transplant 72

centre, the Haematology and Oncology Clinics of Australia, The Wesley Hospital, Brisbane, 73

Australia. Sixty-five subjects were scheduled to complete ADP and BIS assessment up to two 74

weeks before PBSCT (pre-admission), and at three months post-PBSCT; 44 subjects 75

completed assessments at both time points; 11 out of the 44 subjects completed a once-off 76

DXA scan. Ethical approval was granted by the Multidisciplinary Ethics Committee of the 77

hospital (Ref: 1107, and Ref: 1017) and Medical Research Ethics Committee of The 78

University of Queensland (HMS10/0306.r1 and HMS11/2405). 79

All subjects wore a tight fitting, one-piece Lycra® suit and a Lycra cap provided by the lab. 80

Height was measured to the nearest 0.1 cm using a wall-mounted stadiometer, and weight 81

was measured to the nearest 0.1 kg (TBF-300A, Tanita Inc, Tokyo, Japan). BIS, followed by 82

ADP measurements were completed within a 15-minute period. 83

Bioimpedance spectroscopy 84

Participants were assessed with whole body BIS (ImpSFB7, Impedimed, Brisbane, Australia). 85

The theories and principles of bioimpedance techniques have been detailed previously [6]. 86

In short, body composition is derived from impedance which measures the decrease in 87

voltage of an applied electric current due to resistance in the human body (i.e. non-88

conductive tissues such as fat). For each assessment, this BIS device obtains impedance data 89

across a spectrum of 256 frequencies between 3 to 1000 kHz. Prediction of body 90

composition was performed by fitting the impedance data to the Cole-Cole model to 91

determine resistance at zero and infinite frequencies using manufacturer’s software. These 92

resistance values were then applied to Hanai mixture theory equations to predict total body 93

water, FFM and FM [6, 7]. Hanai mixture theory equations require the input of values for 94

MANUSCRIP

T

ACCEPTED


7

resistivities of intra- and extracellular water, a body proportion correction factor (Kb), body 95

density (Db) and lean tissue hydration fraction to predict body fat-free mass from the 96

predicted TBW [8]. Body composition estimates from BIS data are therefore dependent 97

upon the values chosen for these parameters. For SFB7, the coefficient of variation for 98

repeated measures has been determined as <0.5% [9]. Estimates of body composition were 99

predicted using the default parameters provided with the instrument, those of: de Lorenzo 100

et al. [7] and Matthie et al. [10]; Moissl et al. [11]; and produced by the authors in an 101

independent study [12]. The electrode configuration for whole-body assessment has been 102

described previously [6]. 103

Air-displacement plethysmography 104

ADP was conducted using a BOD POD unit (COSMED, Concord, CA, USA). ADP is considered 105

as an alternative to underwater weighing; it is based on the two compartment model which 106

separates the body into two distinct chemical components composed of FFM and FM [13]. 107

The principles of ADP have been detailed elsewhere [14]. FFM and FM can be derived from 108

volume, density and weight using the equation for the general population by Siri [15]. 109

Predictive thoracic gas volume inbuilt in the BOD POD software was used. 110

The system is composed of a fibreglass chamber that measures body volume and an 111

external electronic scale that measures weight. The chamber volume is calibrated daily, 112

while the scale is calibrated fortnightly; calibration is performed using the manufacturer-113

provided calibration cylinder (50.099 L) and calibration weights (20 kg) respectively. Two to 114

three repeated measurements were conducted for each participant as instructed by the 115

computer. Participants remained still for each measurement which lasted less than a 116

minute. 117

MANUSCRIP

T

ACCEPTED


8

Dual energy X-ray (DXA) absorptiometry 118

Body composition was assessed with whole-body DXA scan (Hologic, QDR 4500A fan-beam 119

scanner, MA, USA) and adult software version 13.3. Daily calibration was performed with 120

phantom spine, and steps provided by the manufacturer. The theories and principles of DXA 121

have been detailed previously [16]. In summary, body composition is determined through 122

the measurement of mass attenuation coefficients, the ratio values, image processing, and 123

soft tissue distribution models. Body composition was calculated by algorithms provided by 124

the manufacturer (Bioimp, software version 5.3.1.1) [17]. 125

Participants wore light clothing; all metal objects were removed (i.e. jewellery, glasses, zips). 126

Participants were instructed to remain still for up to 7 minutes during the scan. 127

Statistical Analyses 128

Statistical analysis was performed using SPSS version 20.0.0 (IBM SPSS statistics, Chicago, IL, 129

USA). Normality was tested with Shapiro-Wilk’s test. Descriptive statistics were calculated 130

for baseline characteristics (age, and BMI), and body composition parameters at different 131

time points. Paired t-test was used to assess the mean differences between body 132

composition parameters measured by BIS relative to ADP or DXA. Bland-Altman approach 133

was used to assess the agreement between ADP and BIS; limits of agreement (LOA) were 134

calculated as ± two standard deviations of bias [18]. Correlation between results of the 135

methods was assessed with Lin’s concordance correlation [19]. Statistical significance was 136

reported at the conventional p < 0.05 level (two-tailed). A clinically acceptable difference for 137

FFM between the methods was defined a priori as ≤1 kg [20]. 138

Results 139

MANUSCRIP

T

ACCEPTED


9

Sixty-five subjects were included in this study; 44 subjects (52.3% male) who completed 140

both baseline and follow-up assessment at three-month post-transplantation were 141

analysed. 142

Non-completion of assessments was mainly due to inconvenience to travel to the hospital in 143

the desired time frame. 144

Median age was 56.5 years-old (range 22-75 y), and median BMI was 28.0 kg/m2 (range 145

16.4–47.6 kg/m2); 36.4% of the subjects were overweight (BMI 25 to <30 kg/m

2), 31.8% 146

were of normal weight (BMI 18.5 to <25 kg/m2), 29.5% were obese (BMI >30 kg/m

2), and 147

2.3% was underweight (BMI <18.5 kg/m2)[21]. Results of ADP showed FFM (kg) was higher 148

amongst males (p < 0.001). In the sub-group of participants (n=11/44) who underwent DXA 149

examination, both FFM and FM predicted by DXA relative to ADP was not significantly 150

different (FFMDXA-ADP = -0.91kg ± LOA 4.2, p = 0.186; FMADP-DXA 1.39 kg ± LOA 4.2, p = 0.055). 151

Mean weight loss amongst obese subjects (-7.1 ± 4.9 kg) was significantly higher than 152

normal or underweight subjects (-2.9 ± 3.4 kg) (p = 0.032) but similar to overweight subjects 153

(-4.3 ± 4.3 kg) (p = 0.235); there was no difference in weight loss between males and 154

females (p = 0.275), and younger or older group (age ≤60 and >60 y) (p = 0.272). 155

Different BIS prediction methods produced different absolute values for FFM and FM (Table 156

1). Results obtained using the alternative BIS predictive equations were generally in 157

agreement, e.g. predicted FFM at baseline varied by only 3.2 %, whereas the default 158

instrument predictions were in poorer agreement with ADP values and those obtained using 159

the alternative BIS predictive equations. Notably, however, the changes in body 160

composition from baseline to post-PBSCT measured by BIS were similar irrespective of the 161

prediction method used. This is explored further below. 162

MANUSCRIP

T

ACCEPTED


10

Agreement of absolute values 163

Since the alternative predictive equations performed similarly, results in Table 2 are shown 164

for default instrument and Moissl et al. [11] methods only. 165

Although the results predicted by Ward et al. [12] (i.e. difference FFM -2.2 kg (CI95% -3.6, -166

0.8), p = 0.003) and Moissl et al. [11] (i.e. difference FFM -1.2 kg, CI 95% -2.3, -0.2, p = 0.021) 167

at baseline, and Ward et al. [12] prediction at post-transplant (i.e. difference FFM -1.4 kg, CI 168

95% -2.8, 0.0, p = 0.045) were significantly different to ADP, the agreement relative to ADP 169

were better compared to the instrument’s default setting. Results predicted by De Lorenzo 170

et al. [7] method were not significantly different relative to ADP at both time points. 171

Results predicted by the BIS alternative predictive equations agreed well (all p > 0.05) with 172

DXA, whereas results predicted by the default setting was significantly different (all p ≤ 173

0.001). 174

Limits of agreement were wide for all comparisons (i.e. LOA ± 7 to LOA ±11 kg for both FFM 175

and FM). 176

Agreement of detected changes (Δ) before and after transplantation 177

Predicted ΔFFM and ΔFM (Table 2) by ADP and all BIS methods were similar but limits of 178

agreement were wide. The method by Moissl et al. [11] performed the best (mean bias 179

ΔFFM = 0.29 kg, LOA ± 6.4 kg, p = 0.547), followed by Ward et al. [12] (mean bias ΔFFM = 180

0.75 kg LOA ± 6.6 kg, p = 0.138), De Lorenzo et al. [7] (mean bias ΔFFM = 0.91 kg, LOA ± 8.4 181

kg, p = 0.160), and lastly the default setting (mean bias ΔFFM = 0.92 kg, LOA ± 7.2 kg, p = 182

0.101). 183

MANUSCRIP

T

ACCEPTED


11

Correlation 184

The strength of correlation between results measured by BIS and ADP was based on the 185

criteria proposed by McBride et al. [22]. Absolute FFM, and FM measured by the default 186

setting correlated poorly (all strength < 0.90) with ADP; in contrast, absolute FFM and FM 187

measured by the alternative BIS predictive equations correlated moderately (strength 0.90-188

0.95) to substantially (strength 0.95-0.99), but poorly for percentage FM (strength <0.90). 189

Correlations for ΔFFM, ΔFM, and Δ%FM were poorer for all methods (strength 0.16-0.61). 190

Bland-Altman plots 191

Bland-Altman plots of absolute value at pre-admission and three months post-192

transplantation were similar. For ease of presentation, only the plots for pre-admission 193

comparison are presented for the BIS default setting and Moissl et al. [11] method. Figure 194

1a shows the default setting overestimated FFM relative to ADP and bias increases as FFM 195

increases. Figure 1b shows the Moissl et al. [11] method overestimated FFM slightly but the 196

trend of increase or decrease in bias towards extreme ends of FFM is less pronounced. 197

Similar trends were observed for FM (plots not shown). Figure 2a and Figure 2b are Bland-198

Altman plots showing ΔFFM detected by the default setting and Moissl et al. [11] method 199

compared to ADP. Regardless of the direction of change, disagreement between the default 200

setting and ADP tended to increase as the magnitude of change increased. Mean bias was 201

slightly improved when results were re-analysed with the Moissl et al. [11] method but a 202

similar trend of increase in bias was observed for increased changes in FFM. Bland-Altman 203

plots showed similar systematic variation for the bias ΔFM for both methods (plots not 204

shown). 205

MANUSCRIP

T

ACCEPTED


12

Discussion 206

This study is the first to examine the agreement of body composition methods amongst 207

cancer patients treated with PBSCT. This study found a portable device such as BIS was 208

sensitive enough to detect changes in body composition comparable to that assessed by 209

ADP amongst cancer patients who experienced moderate weight loss at three months after 210

PBSCT. Agreement at a group level (mean bias) for detected changes in FFM was clinically 211

acceptable (<1 kg) although limits of agreement were wide as observed in other studies [23-212

27]. 213

214

Relative to ADP and DXA, the agreement of absolute values predicted by the default setting 215

was poor. Similar findings were reported in other cancer studies that examined cross-216

sectional results [28, 29]. 217

218

No cancer studies which examined the agreement between predicted changes in 219

longitudinal studies could be identified in the literature. However, studies on overweight or 220

obese subjects (without cancer) demonstrated that portable methods such as bioimpedance 221

devices can accurately measure changes in the body composition parameters after weight 222

loss regardless of large discrepancies between cross-sectional results [24-27, 30, 31]. 223

224

Absolute results of FFM tend to be overestimated by the BIS default setting which was 225

similar to studies examining healthy subjects [32, 33], overweight or obese subjects, [24, 31, 226

34] and subjects with clinical conditions [23, 28]. When BIS results were re-analysed with 227

the methods by De Lorenzo et al. [7], Ward et al. [12], and Moissl et al. [11], the agreements 228

MANUSCRIP

T

ACCEPTED


13

of absolute results relative to ADP improved substantially. Bland-Altman plots showed the 229

systematic variation in the bias of FFM relative to ADP observed for the BIS default setting 230

(Figure 1a) was reduced when BIS results were re-analysed using the method by Mossil et al. 231

[11] (Figure 1b). A similar effect was observed for FM (plots not shown). 232

233

For the agreement in ΔFFM (Figures 2a and 2b) however, the method of Moissl et al. [11] 234

did not reduce the systematic variation in the bias of ΔFFM observed in the BIS default 235

setting. Changes in body hydration or altered composition of FFM may explain the increase 236

in discrepancy as weight loss or weight gain increases [35]. The BIS method, unlike 237

empirically-derived prediction equations, predicts body water volumes based on 238

fundamental equations incorporating the resistivity of body fluids (ECW and ICW) [7, 8]. 239

However, transformation of TBW to FFM assumes a hydration fraction and is therefore 240

prone to error when hydration state changes. At the time of study, the participants were 241

nutritionally stable and unlikely to be dehydrated. Other variables (i.e. disproportion loss of 242

FFM or FM) specific to our population may be present. 243

244

The limits of agreement for ΔFFM or ΔFM in this study were 2 to 4 kg wider than some 245

reported in the literature [24, 25, 27]. These differences may be due to sample 246

characteristics as subjects in these studies were younger, healthy participants undergoing 247

intentional weight loss program; three of these studies were composed of females only 248

while one analysed by gender separately. For each of the BIS methods, we examined bias 249

and limits of agreement separately by gender, age group (≥ 60 y and < 60 y) and BMI 250

MANUSCRIP

T

ACCEPTED


14

categories. Results were not significantly different for the Moissl et al. [11] method, 251

however, results predicted by the default setting, Ward et al. [12] and De Lorenzo et al. [7] 252

became significantly different to ADP for female subjects (p<0.05), and younger subjects ≤60 253

years old (p<0.05). Results showed Moissl et al. [11] method is more suitable for the 254

assessment of samples with mixed characteristics (i.e. gender, age, and BMI) than the other 255

BIS methods in this study. Owing to wide limits of agreement, ADP and BIS are not 256

interchangeable; routine assessments on the same individual should be conducted with the 257

same method consistently. 258

259

There is currently no population-specific resistivity coefficient or recommendation on the 260

type of BIS equation suitable for the assessment of body composition of patients treated 261

with PBSCT. Results of this study showed the default setting is not optimal for assessing the 262

PBSCT population since alternative predictive equations produced better agreement with 263

the reference method. Absolute (i.e. cross-sectional) results should be interpreted with 264

caution as results vary greatly with the choice of predictive equations and the characteristics 265

of the subjects being assessed [36]. Notably, for example in the present study, BIS 266

prediction, when analysed using the method of Moissl et al. [11] performed significantly 267

better than when the BIS default settings were used. The Moissl et al. [11] approach 268

specifically takes account of the variation of impedance-based predictions of body 269

composition with BMI; our population was skewed toward subjects who were either 270

overweight (36.4%) or obese (29.5%). 271

272

MANUSCRIP

T

ACCEPTED


15

There are no similar studies amongst the cancer population; most studies compared cross-273

sectional results only. This study has highlighted that the accuracy of BIS for prediction of 274

absolute body composition is dependent upon the choice of appropriate values for 275

coefficients in the predictive equations; the default settings of BIS cannot be relied upon. It 276

should be noted, however, that these settings may be changed by the user. Considering 277

changes in body composition parameters have more clinical meaning than results collected 278

at a single time point, it would be of great interest for future studies to compare more 279

thoroughly the validity of detected changes in body composition rather than the agreement 280

of cross-sectional results alone. Different predictive equations should be explored. In this 281

regard, all BIS methods were observed in the present study to predict similarly change in 282

body composition both in direction and magnitude; these predictions were also similar to 283

those determined by ADP. Further studies are needed to confirm the sensitivity of BIS in 284

tracking body composition changes relative to other laboratory methods, as well as amongst 285

other clinical populations with varying degrees of weight loss. 286

287

Strengths and limitations 288

This study provides new information as it is the first to examine the agreement of body 289

composition methods amongst PBSCT patients. Our study had a moderate sample size 290

considering studies examining detected body composition changes had sample sizes of less 291

than 60 [25-27, 30, 31]; these studies included healthy overweight subjects participating in 292

weight loss programs whereas our subjects were cancer patients; it is more challenging to 293

obtain large sample in prospective studies particularly amongst the cancer populations. 294

MANUSCRIP

T

ACCEPTED


16

295

A limitation of this study was that DXA could only be performed at one time point amongst a 296

small number of participants because the instrument was located at a different site; the 297

distance of travel had an effect on compliance. The 32% (n=21/65) of patients who were not 298

included in this study due to incomplete body composition data may bias the results if they 299

had different degree of FFM and FM change compared to patients included in this study. 300

301

Conclusion 302

Amongst patients treated with PBSCT, the BIS approach to prediction of FFM and FM, 303

particularly the Moissl method [11], was highly correlated with body composition 304

determined by ADP although the bias and limits of agreement were large for the SFB7 305

default method. The Moissl method [11] was in closer agreement with ADP but the 306

relatively large limits of agreement may preclude its use in individuals. With respect to 307

changes in FFM and FM, changes predicted by BIS were similar to those determined by ADP 308

at a group level, however, agreement at an individual level should again be interpreted with 309

caution due to wide limits of agreement. 310

311

Conflict of interest 312

Author Ward has consulted to ImpediMed Ltd., a manufacturer of impedance devices. 313

ImpediMed Ltd had no involvement in the design, execution of this study or the preparation 314

of this manuscript. Other authors declare no conflict of interest. 315

316

MANUSCRIP

T

ACCEPTED


17

References 317

1. Gale CR, Martyn CN, Cooper C, Sayer AA. Grip strength, body composition, and 318

mortality. Int J Epidemiol 2007;36:228-235. 319

2. Thibault R, Genton L, Pichard C. Body composition: Why, when and for who? Clin 320

Nutr 2012;31:435-447. 321

3. Hung Y-C, Bauer J, Horsley P, Waterhouse M, Bashford J, Isenring E. Changes in 322

nutritional status, body composition, quality of life, and physical activity levels of 323

cancer patients undergoing autologous peripheral blood stem cell transplantation. 324

Support care cancer 2013;21:1579-1586. 325

4. Korth O, Bosy-Westphal A, Zschoche P, Gluer CC, Heller M, Muller MJ. Influence of 326

methods used in body composition analysis on the prediction of resting energy 327

expenditure. Eur J Clin Nutr 2006;61:582-589. 328

5. Jebb SA, Cole TJ, Doman D, Murgatroyd PR, Prentice AM. Evaluation of the novel 329

Tanita body-fat analyser to measure body composition by comparison with a four-330

compartment model. Brit J Nutr 2000;83:115-122. 331

6. Kyle UG, Bosaeus I, De Lorenzo AD, Deurenberg P, Elia M, Gómez JM, et al. 332

Bioelectrical impedance analysis—part I: review of principles and methods. Clin Nutr 333

2004;23:1226-1243. 334

7. De Lorenzo A, Andreoli A, Matthie J, Withers P. Predicting body cell mass with 335

bioimpedance by using theoretical methods: a technological review. J Appl Physiol 336

1997;82:1542-1558. 337

8. Matthie JR. Second generation mixture theory equation for estimating intracellular 338

water using bioimpedance spectroscopy. J Appl Physiol 2005;99:780-781. 339

MANUSCRIP

T

ACCEPTED


18

9. Ward L. Inter-instrument comparison of bioimpedance spectroscopic analysers. The 340

Open Medical Device J 2009;1:3-10. 341

10. Matthie J, Zarowitz B, De Lorenzo A, Andreoli A, Katzarski K, Pan G, et al. Analytic 342

assessment of the various bioimpedance methods used to estimate body water. J 343

Appl Physiol 1998;84:1801-1816. 344

11. Moissl UM, Wabel P, Chamney PW, Bosaeus I, Levin NW, Bosy-Westphal A, et al. 345

Body fluid volume determination via body composition spectroscopy in health and 346

disease. Physiol Meas 2006;27:921-933. 347

12. Ward L, Isenring E, Dyer J, Battersby K, Bengston J, Kagawa M. Constancy of 348

resistivity coefficients for bioimpedance spectroscopy (BIS). Int J Body Comp Res 349

2008;6:1479-1456. 350

13. McCrory MA, Gomez TD, Bernauer EM, Mole PA. Evaluation of a new air 351

displacement plethysmograph for measuring human body composition. Med Sci 352

Sport Exer 1995;27:1686-1691. 353

14. Dempster P, Aitkens S. A new air displacement method for the determination of 354

human body composition. Med Sci Sport Exer 1995;27:1692-1697.. 355

15. Siri WE. Body composition from fluid spaces and density: Analysis of method. In: 356

Brozek J, Henschel A, Eds. Techniques for measuring body composition. National 357

Academy of Sciences: Washington DC 1961, p. 223-244. 358

16. Pietrobelli A, Formica C, Wang Z, Heymsfield SB. Dual-energy X-ray absorptiometry 359

body composition model: review of physical concepts. Am J Physiol-Endoc M 360

1996;271:E941-E951. 361

17. Laskey MA. Dual-energy X-ray absorptiometry and body composition. Nutrition 362

1996;12:45-51. 363

MANUSCRIP

T

ACCEPTED


19

18. Bland MJ, Altman DG. Statistical methods for assessing agreement between two 364

methods of clinical measurement. The Lancet 1986;327:307-310. 365

19. Lin LIK. A Concordance Correlation Coefficient to Evaluate Reproducibility. 366

Biometrics 1989; 45:255-268. 367

20. Barber MD, Ross JA, Voss AC, Tisdale MJ, Fearon KC. The effect of an oral nutritional 368

supplement enriched with fish oil on weight-loss in patients with pancreatic cancer. 369

Br J Cancer 1999;81:80-86. 370

21. Watterson C, Fraser A, Banks M, Isenring E, Miller M, Silvester C, et al. Evidence 371

based practice guidelines for the nutritional management of malnutrition in adult 372

patients across the continuum of care. Nutr Diet 2009;66:S1-S34. 373

22. McBride G. A proposal for strength-of-agreement criteria for Lin’s concordance 374

correlation coefficient. National Institute of Water and Atmospheric Research Ltd, 375

Hamilton, New Zealand. Client Report: HAM2005-062. Report to Ministry of Health 376

2005. 377

23. van Venrooij LMW, Verberne HJ, de Vos R, Borgmeijer-Hoelen MMMJ, van Leeuwen 378

PAM, de Mol BAJM. Preoperative and postoperative agreement in fat free mass 379

(FFM) between bioelectrical impedance spectroscopy (BIS) and dual-energy X-ray 380

absorptiometry (DXA) in patients undergoing cardiac surgery. Clin Nutr 2010;29:789-381

794. 382

24. Verdich C, Barbe P, Petersen M, Grau K, Ward L, Macdonald I, et al. Changes in body 383

composition during weight loss in obese subjects in the NUGENOB study: 384

Comparison of bioelectrical impedance vs. dual-energy X-ray absorptiometry. 385

Diabetes Metab. 2011;37:222-229. 386

MANUSCRIP

T

ACCEPTED


20

25. Minderico CS, Silva AM, Keller K, Branco TL, Martins SS, Palmeira AL, et al. Usefulness 387

of different techniques for measuring body composition changes during weight loss 388

in overweight and obese women. Br J Nutr 2008;99:432-441. 389

26. Mahon AK, Flynn MG, Iglay HB, Stewart LK, Johnson CA, McFarlin BK, et al. 390

Measurement of body composition changes with weight loss in postmenopausal 391

women: comparison of methods. J Nutr Health Aging 2007;11:203-213. 392

27. Thomson R, Brinkworth GD, Buckley JD, Noakes M, Clifton PM. Good agreement 393

between bioelectrical impedance and dual-energy X-ray absorptiometry for 394

estimating changes in body composition during weight loss in overweight young 395

women. Clin Nutr 2007;26:771-777. 396

28. Trutschnigg B, Kilgour RD, Reinglas J, Rosenthall L, Hornby L, Morais JA, et al. 397

Precision and reliability of strength (Jamar vs. Biodex handgrip) and body 398

composition (dual-energy X-ray absorptiometry vs. bioimpedance analysis) 399

measurements in advanced cancer patients. Appl Physiol Nutr Me 2008;33:1232-400

1239. 401

29. Ellegard LH, Ahlen M, Korner U, Lundholm KG, Plank LD, Bosaeus IG. Bioelectric 402

impedance spectroscopy underestimates fat-free mass compared to dual energy X-403

ray absorptiometry in incurable cancer patients. Eur J Clin Nutr 2008;63:794-801. 404

30. Pødenphant J, Gotfredsen A, Engelhart M, Andersen V, Heitmann BL, Kondrup J. 405

Comparison of body composition by dual energy X-ray absorptiometry to other 406

estimates of body composition during weight loss in obese patients with rheumatoid 407

arthritis. Scand J Clin Lab Inv 1996;56:615-625. 408

31. Frisard MI, Greenway FL, DeLany JP. Comparison of Methods to Assess Body 409

Composition Changes during a Period of Weight Loss. Obesity 2005;13:845-854. 410

MANUSCRIP

T

ACCEPTED


21

32. Volgyi E, Tylavsky FA, Lyytikainen A, Suominen H, Alen M, Cheng S. Assessing Body 411

Composition With DXA and Bioimpedance: Effects of Obesity, Physical Activity, and 412

Age. Obesity 2008;16:700-705. 413

33. Duz S, Kocak M, Korkusuz F. Evaluation of body composition using three different 414

methods compared to dual-energy X-ray absorptiometry. Eur J Sport Sci 2009;9:181-415

190. 416

34. Minderico CS, Silva AM, Teixeira PJ, Sardinha LB, Hull HR, Fields DA. Validity of air-417

displacement plethysmography in the assessment of body composition changes in a 418

16-month weight loss program. Nutr Metabolism 2006;3:32-39. 419

35. Das SK, Roberts SB, Kehayias JJ, Wang J, Hsu LKG, Shikora SA, et al. Body composition 420

assessment in extreme obesity and after massive weight loss induced by gastric 421

bypass surgery. Am J Physio-Endo M 2003;284:E1080-E1088. 422

36. Pichard C, Kyle UG, Slosman DO. Fat-free mass in chronic illness: comparison of 423

bioelectrical impedance and dual-energy x-ray absorptiometry in 480 chronically ill 424

and healthy subjects. Nutrition 1999;15:668-676.425

MANUSCRIP

T

ACCEPTED


22

Table 1 Body composition parameters (mean ±SD) measured at pre-admission and at 426

three months after peripheral blood stem cell transplantation N=44. 427

Variable

Pre-admission

Post-PBSCT

Detected change

Weight (kg) 81.5 ± 19.7 76.8 ± 18.1 -4.6 ± 4.5

FFM(kg)ADP 49.4 ± 10.9 48.4 ± 10.9

-1.0 ± 1.9

FM (kg) ADP 32.1 ± 14.4 28.5 ± 12.7 -3.6 ± 3.9

%FM ADP 38.3 ± 9.9 36.1 ± 9.4 -2.2 ± 3.3

FFM(kg) BIS SFB7 default 55.9 ± 13.46 54.0 ± 13.2

6 -1.9 ± 4.6

FM (kg) BIS 25.6 ± 11.06 22.8 ± 10.5

6 -2.8 ± 3.2

%FM BIS 38.3 ± 9.96 29.2 ± 8.7

6 -1.7 ± 3.8

FFM(kg) BIS De Lorenzo et al.1

50.2 ± 13.8 48.3 ± 13.9 -1.9 ± 5.2

FM (kg) BIS 31.3 ± 11.8 28.5 ± 11.2 -2.8 ± 3.7

%FM BIS 38.3 ± 8.9 37.1 ± 9.7 -1.2 ± 4.2

FFM(kg) BIS Ward et al.2

51.6 ± 13.45 49.9 ± 13.4

5 -1.7 ± 4.3

FM (kg) BIS 29.9 ± 11.75 27.0 ± 10.9

5 -2.9 ± 2.9

%FM BIS 36.4 ± 8.55 35.0 ± 9.0

-1.5 ± 3.2

FFM(kg) BIS Moissl et al.3

50.6 ± 11.75 49.4 ± 11.9 -1.3 ± 4.1

FM (kg) BIS 30.9 ± 13.25 27.5 ± 12.2 -3.4 ± 3.3

%FM BIS 37.0 ± 9.15 35.0 ± 9.6 -2.02 ± 3.5

MANUSCRIP

T

ACCEPTED


23

FFM(kg) DXA 4

NM 50.2 ± 9.5 NM

FM (kg) DXA 4

NM 25.7 ± 9.1 NM

%FM DXA 4

NM 33.0 ± 6.1 NM

1De Lorenzo et al. [7] 428

2Ward et al. [9] 429

3Moissl et al. [11] 430

4N=11 431

5p < 0.05, difference compared to ADP, paired t-test 432

6p < 0.001, difference compared to ADP, paired t-test 433

Abbreviations: ADP, air displacement plethysmography; BIS, bioimpedance spectroscopy; DXA, dual-434

energy X-ray absorptiometry; FFM, fat free mass; FM, fat free mass, NM, not measured; PBSCT, 435

peripheral blood stem cell transplantation. 436

437

MANUSCRIP

T

ACCEPTED


24

Table 2 Mean bias (95%CI) and limits of agreement for fat free mass (kg), fat mass (kg), and 438

percentage (%) fat mass, at pre-admission, and at three months after peripheral blood stem cell 439

transplantation as measured by bioimpedance spectroscopy relative to air-displacement 440

plethysmography, and dual-energy X-ray absorptiometry N=44. 441

Comparison Bias (95%CI)

Limits of

Agreement

5Correlation

r

Baseline

FFM (kg) ADP – BIS -6.5 (-8.0, -5.0)

4 ± 9.9 0.80

FM (kg) ADP – BIS SFB7 default 6.5 (5.0, 7.9)

4 ± 9.8 0.82

FM (%) ADP – BIS 7.3 (5.7, 8.8)

4 ± 10.2 0.63

FFM (kg) ADP – BIS -1.2 (-2.3, -0.2)

3 ± 6.8 0.95

FM (kg) ADP – BIS Moissl et al.

1 1.2 (0.2, 2.3)

3 ± 6.8 0.97

FM (%) ADP – BIS 1.2 (0.0, 2.4)

3 ± 7.9 0.91

Post-transplantation

FFM (kg) ADP – BIS -5.6 (-7.0, -4.2)

4 ± 9.1 0.84

FM (kg) ADP – BIS SFB7 default 5.7 (4.3, 7.1)

4 ± 9.2 0.82

FM (%) ADP – BIS 6.9 (5.4, 8.4)

4 ± 10.0 0.65

FFM (kg) ADP – BIS -0.9 (-2.1, 0.2) ± 7.3 0.95

FM (kg) ADP – BIS Moissl et al.

1 1.0 (-0.1, 2.1) ± 7.5 0.95

FM (%) ADP – BIS 1.1 (-0.4, 2.5) ± 9.6 0.87

FFM (kg) DXA - BIS 2

-5.9 (-8.7, -3.0) 4

± 8.4 0.81

FM (kg) DXA - BIS 2

SFB7 default 5.1 (2.4, 7.8)3 ± 8.0 0.69

FM (%) DXA - BIS 2

6.3 (3.1, 9.5) 4

± 9.5 0.33

FFM (kg) DXA – BIS 2

-1.5 (-3.2, 0.1) ± 4.9 0.96

FM (kg) DXA – BIS 2

Moissl et al. 1

0.7 (-0.7, 2.2) ± 4.3 0.96

MANUSCRIP

T

ACCEPTED


25

FM (%) DXA – BIS 2

0.9 (-1.3, 3.1) ± 6.7 0.79

Detected changes

ΔFFM (kg) ADP – BIS 0.92 (-0.19, 2.02) ± 7.2 0.45

ΔFM (kg) ADP – BIS SFB7 default -0. 76 (-1.87, 0.35) ± 7.3 0.47

ΔFM (%) ADP – BIS 0.40 (-1.68, 0.87) ± 8.4 0.31

ΔFFM (kg) ADP – BIS 0.29 (-0.68, 1.27) ± 6.4 0.50

ΔFM (kg) ADP – BIS Moissl et al. 1

-0.22 (-0.68, 1.27) ± 6.3 0.61

ΔFM (%) ADP – BIS -0.16 (-1.32, 0.99) ± 7.6 0.37

1Moissl et al. [11] 442

2n = 11 443

3p < 0.05, difference between methods, paired t-test 444

4p < 0.001, difference between methods, paired t-test 445

5Lin’s concordance correlation coefficient 446

Abbreviations: ADP, air displacement plethysmography; BIS, bioimpedance spectroscopy; DXA, dual-447

energy X-ray absorptiometry; FFM, fat free mass; FM, fat free mass; Δ, change between baseline and 448

post-transplantation. 449

450

451

452

453

MANUSCRIP

T

ACCEPTED


26

Figure 1. Bland-Altman plot of absolute fat free mass in kg (FFM) at pre-admission as predicted by (a) 454

bioimpedance spectroscopy default setting (BIS default), and (b) Moissl et al. method, relative to air 455

displacement plethysmography (ADP). Mean bias (solid line) and ±2SD (dashed lines). 456

457

Figure 2. Bland-Altman plot of change in fat free mass in kg (ΔFFM) between pre-admission and 458

three months after stem cell transplantation as predicted by (a) bioimpedance spectroscopy default 459

setting (BIS default), and (b) by Moissl et al. method, relative to air displacement plethysmography 460

(ADP). Mean bias (solid line) and ±2SD (dashed lines). 461

462

463

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

-6.5

-16.4

3.4

-25.0

-20.0

-15.0

-10.0

-5.0

0.0

5.0

10.0

30.0 40.0 50.0 60.0 70.0 80.0

FF

M A

DP

-F

FM

BIS

de

fau

lt

(FFM ADP + FFM BIS default ) /2

Female Male

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

-1.2

-8.1

5.6

-25.0

-20.0

-15.0

-10.0

-5.0

0.0

5.0

10.0

30.0 40.0 50.0 60.0 70.0 80.0

FF

M A

DP

-F

FM

Mo

issl

(FFM ADP + FFM Moissl) /2

Female Male

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

0.9

-6.3

8.2

-10.0

-5.0

0.0

5.0

10.0

15.0

-15.0 -10.0 -5.0 0.0 5.0 10.0

∆F

FM

AD

P -∆

FF

M B

IS d

efau

lt

(ΔFFM ADP + ΔFFM BIS default) /2

Female Male

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

0.3

-6.1

6.7

-10.0

-5.0

0.0

5.0

10.0

15.0

-15.0 -10.0 -5.0 0.0 5.0 10.0

ΔF

FM

AD

P -

ΔF

FM

Mo

issl

(ΔFFM ADP + ΔFFM Moissl) /2

Female Male

body composition following stem cell transplant

Documents