Body composition assessment in adults with cystic fibrosis: comparison of dual-energy X-ray absorptiometry with skinfolds and bioelectrical impedance analysis

Nutrition 21 (2005) 1087–1094 www.elsevier.com/locate/nut

Applied nutritional investigation

Body composition assessment in adults with cystic fibrosis: comparison of dual-energy X-ray absorptiometry with skinfolds and bioelectrical impedance analysis Susannah King, M.Nutr.Diet.a,b,*, John Wilson, Ph.D.a,c, Tom Kotsimbos, M.D.a,c, Michael Bailey, M.Sc. (Statistics)d, and Ibolya Nyulasi, M.Sc.b,e a

Department of Medicine, Monash University, Alfred Hospital, Commercial Road, Melbourne, Victoria, Australia b Nutrition Department, Alfred Hospital, Melbourne, Victoria, Australia c Department of Allergy, Immunology and Respiratory Medicine, Alfred Hospital, Melbourne, Victoria, Australia d Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Victoria, Australia e Department of Medicine, Monash University, Monash Medical Centre, Clayton, Victoria, Australia Manuscript received August 31, 2004; accepted April 4, 2005.

Abstract

Objective: We compared body composition measurement in adults with cystic fibrosis (CF) by using non-invasive methods (skinfold thicknesses and bioelectrical impedance analysis [BIA]) with dual-energy X-ray absorptiometry (DXA). Methods: Seventy-six adults with CF (mean age 29.9 ⫾ 7.9 y, mean body mass index 21.5 ⫾ 2.5 kg/m2) were studied. Body composition was measured to calculate fat-free mass (FFM) using DXA, the sum of four skinfold thicknesses, and BIA (predictive equations of Lukaski and of Segal). Results: Mean FFM values ⫾ standard deviation measured using DXA were 54.8 ⫾ 7.3 kg in men and 41.2 ⫾ 3.9 kg in women. Mean FFM values measured using BIA/Lukaski were 51.5 ⫾ 7.8 kg in men and 40.4 ⫾ 4.9 kg in women (P ⬍ 0.0005 for men, not significant for women for comparison with DXA). Mean FFM values measured using BIA/Segal were 54.2 ⫾ 7.5 kg for men and 44.1 ⫾ 5.9 kg for women (not significant for men, P ⬍ 0.0005 for women for comparison with DXA). Mean FFM values measured using skinfolds were significantly higher than those for FFM with DXA (57.2 ⫾ 7.2 kg in men, 43.3 ⫾ 4.3 kg in women, P ⬍ 0.0005 for comparison with DXA). The 95% limits of agreement with FFM using DXA were, for men and women, respectively, ⫺8.3 to 1.7 kg and ⫺6.4 to 4.8 kg for BIA/Lukaski, ⫺4.8 to 3.6 kg and ⫺3.1 to 8.9 kg for BIA/Segal, and ⫺2.8 to 7.3 kg and ⫺1.5 to 5.7 kg for skinfolds. Conclusion: This study suggests that skinfold thickness measurements and BIA will incorrectly estimate FFM in many adults with CF compared with DXA measurements of FFM. These methods have limited application in the assessment of body composition in individual adult patients with CF. © 2005 Elsevier Inc. All rights reserved.

Keywords:

Cystic fibrosis; Nutritional assessment; Body composition; Dual-energy X-ray absorptiometry; Bioelectrical impedance analysis

Introduction Cystic fibrosis (CF) is an autosomal recessive genetic disorder that affects approximately 1 in 2500 live births in Caucasians. The clinical manifestations of CF include proThis study was funded by the Australian Cystic Fibrosis Research Trust. Susannah King is supported by a National Health and Medical Research Council of Australia Postgraduate Scholarship. * Corresponding author. Tel.: ⫹61-3-9903-0152; fax: 61-3-9521-2124. E-mail address: [email protected] (S. King). 0899-9007/05/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.nut.2005.04.005

gressive lung disease; pancreatic insufficiency, resulting in intestinal maldigestion and malabsorption; and malnutrition, factors contributing to which include increased energy expenditure, reduced energy intake, and nutrient losses secondary to malabsorption [1]. Nutritional status in patients with CF is associated with lung function [2]. Poor nutrition has been associated with complications of CF, including reduced bone mineral density [3], and is independently associated with poor survival [4]. Recent studies have shown that a significant proportion

1088

S. King et al. / Nutrition 21 (2005) 1087–1094

of adults with CF have low fat-free mass (FFM) [5]. This is significant because it has been suggested that depletion of FFM is associated with more severe CF lung disease [6]. The measurement of body composition in chronic diseases is an essential aspect of providing optimal nutritional management [7, 8]. The importance of measuring body composition in CF patients is highlighted by McNaughton et al.’s finding that weight-based indicators of nutritional status underestimated the prevalence of malnutrition in children and adolescents with CF compared with total body potassium measurement [9]. Body composition measurement should be undertaken using reliable and accurate methods. There are several methods that have been used in studies reporting body composition in CF, including dual-energy X-ray absorptiometric (DXA) scanning [5,6,10 –12]. The precision of DXA measurements is high, and the radiation dose is considered low enough to be safe for single and repeated measurements to be taken [13]. DXA is considered to be a useful and reliable method for body composition assessment [14]. Non-invasive methods of body composition assessment, including skinfold thicknesses and bioelectrical impedance analysis (BIA), are able to be readily performed by trained operators in most settings because equipment is easily transported. They also have the advantage of being relatively quick to perform and do not involve exposure to ionising radiation. BIA is based on the conduction of an electrical current through the body [15]. Several equations have been proposed for predicting total body water (TBW), FFM, and fat mass from resistance measurements obtained using BIA [16,17]. It has been suggested that underwater weighing is not suitable as a method for measuring body composition in people with CF [18], and total body potassium measurement is available only in specialized centers. Despite the number of studies reporting body composition measurements in CF by using various methods, there are few studies in adults comparing different methods to ascertain which are sufficiently reliable and accurate to be used in clinical practice. If non-invasive, quick methods for measuring body composition in adults with CF are to be useful in the assessment of nutritional status, it is important to ascertain how well they compare with a widely used reference method such as DXA. The aim of this study was to compare body composition in adults with CF using two non-invasive, quick methods (BIA and skinfolds) with DXA.

Material and methods Subjects The study population included 76 adults (46 men, age range 19 –59 y) in whom the diagnosis of CF had been previously made by a sweat chloride test, genotype analysis, and an appropriate CF phenotype. Subjects were part of a

study assessing bone mineral density and body composition and were randomly selected from the population of 174 non-transplanted adults with CF attending the regional Adult Cystic Fibrosis Service at the Alfred Hospital (Melbourne, Australia) during the study period (May 2000 to April 2001). Subjects were excluded from the study if they were pregnant, were unable to complete the study requirements, or were receiving treatment for reduced bone mineral density. The aim was to obtain a large study sample that was representative of an adult CF population. To ascertain how well this aim was met, genotypic and phenotypic characteristics (CF genotype, gender, age, forced expiratory volume in 1 s as a percentage of predicted [FEV1%], and pancreatic status) of all 174 patients attending the service were recorded and the study sample characteristics were compared with the non-studied population. The study was approved by the Alfred Hospital institutional ethics committee and written informed consent was obtained from each subject before participation. Assessments The following data were collected from each subject’s medical record and at interview: pancreatic status, as assessed by use of pancreatic enzyme replacement therapy, and CF genotype, identified by using a panel of probes for ten common mutations (⌬F508, ⌬F507, A445E, G542X, G551D, N1303K, R117H, R553X, V520F, 621 ⫹ 1G-T). Lung function was assessed by measuring FEV1 with a Lilly pneumotachometer attached to a Jaeger spirometer (Master screen pneumo, version 4.0, Jaeger, Wurzburg, Germany). The highest lung function in the 3 mo before body composition measurement was recorded. Height and weight were measured in light clothing without shoes and were used to calculate body mass index (kg/m2) and FEV1% using the predictive equations of Knudson et al. [19]. Subjects were not required to fast prior to body composition assessment, but were instructed to refrain from ingesting caffeine or alcohol for at least 2 hours prior to measurements being taken. Body composition measurements to determine FFM were made using each of the following three methods. 1. DXA scanning (Lunar DPX-IQ, version 4.7e, Lunar Radiation Corporation, Madison, WI, USA), according to standardized procedures recommended by the manufacturer. Calibration using the manufacturer’s phantom was performed on the day of each scan. Subjects reclined on a flat couch, were dressed in light clothing, and wore no metal objects. The whole body was scanned. Scan duration typically ranged from 10 to 30 min depending on the subject’s size, with radiation dose for a scan being approximately 0.5 ␮Sv. FFM was calculated by using the sum of the estimates of lean tissue mass and bone mineral content for each subject. DXA was the reference method.

S. King et al. / Nutrition 21 (2005) 1087–1094

2. BIA was undertaken with a SEAC Multiple Frequency Bioimpedance Meter (Model SFB3, version 1.0, Uniquest Ltd, Brisbane, Queensland, Australia). Measurements were undertaken as previously described [15] by a single trained observer (S.K.). Subjects reclined on a flat couch, with limbs not touching each other. Electrodes were placed on the right side of the body between the distal prominences of the radius and ulna; the distal end of the third metacarpal; between the median and lateral malleoli at the ankle; and at the distal end of the third metatarsal. FFM was calculated from the measurements of resistance made at 50 kHz by using previously validated predictive equations of Lukaski [17], and Segal (fatness-specific equations) [16] as supplied by the manufacturer. These equations were selected because the age distribution of the populations used was similar to that in the present study. Two estimates of FFM were made for each subject, and the results were averaged. 3. Skinfold thicknesses at four sites (triceps, biceps, subscapular and suprailiac, on the right side unless contraindicated) using standardized Harpenden skinfold callipers (John Bull, British Indicators Ltd, United Kingdom) was measured by a single trained observer (S.K.). Two measurements were taken at each site, with a third measurement being taken if the two readings differed by more than 10%. The two closest readings were averaged. Percentage of fat was calculated from the sum of four skinfolds using the equations of Durnin and Wormersley [20]. FFM was then calculated from the estimate of percentage of fat by subtracting fat mass from total body weight. Statistical analysis Statistical analyses were performed using STATA 7.0 (STATA Corp., College Station, TX, USA). Data were expressed as mean ⫾ standard deviation. Unpaired t tests were performed to compare means of demographic and body composition variables between men and women. FFM measured using DXA was compared with that using other methods by using paired t tests, Pearson’s correlations, and Bland-Altman analysis [21]. Mean bias was calculated as the difference between the mean FFM for each method and the mean FFM for DXA, i.e., FFM/method ⫺ FFM/DXA. The 95% limits of agreement were calculated as bias ⫾ two standard deviations for each method to assess how well FFM estimates obtained using skinfolds and BIA were likely to agree with those obtained using DXA for the assessment of individual subjects. The standard deviation of the mean bias for each method was calculated and expressed as a percentage of the average of FFM/method and FFM/ DXA. This measurement represents the uncertainty of individual measurements of FFM made when using non-invasive methods (BIA and skinfolds) compared with DXA measurements and is reported in this study as the coefficient

1089

of variation for each method compared with DXA: coefficient of variation (%) ⫽ 100% ⫻ standard deviation/(mean FFM/method ⫹ FFM/DXA) [22]. Pearson’s correlation coefficients were calculated to assess the correlations between the difference in FFM between DXA and each method and the average of FFM/DXA and FFM/method to determine whether the bias for each method was consistent across the range of FFM values. Because the distribution of FFM differed between men and women, analyses were undertaken on men and women separately.

Results Table 1 lists the demographic and anthropometric characteristics of the subjects in the study. The study sample of 76 patients showed characteristics similar to those of the non-studied population, confirming that the study sample was representative of the total clinic population (data not shown). Sixty-eight subjects (89%) had pancreatic insufficiency and were taking pancreatic enzyme replacement therapy. Thirty-five subjects (46%) were homozygous for the ⌬F508 mutation; 31 (41%) were heterozygous for the ⌬F508 mutation; and 10 (13%) had no allele for ⌬F508. The gender ratio in this study (60.5% male) reflected that observed in most adult CF populations, which is explained by the poorer survival of females into adulthood [23]. FFM measured using all methods was significantly correlated with DXA FFM (BIA/Lukaski r ⫽ 0.95, BIA/Segal r ⫽ 0.94, skinfolds r ⫽ 0.97, P ⬍ 0.001 for all correlations). However, there were differences in mean FFM between methods (Table 2). Skinfolds in men and women and BIA/ Segal in women tended to overestimate FFM compared with DXA. In contrast, BIA/Lukaski in men tended to underestimate FFM. The mean bias for each method compared with DXA is presented in Table 3, in addition to the results of the BlandAltman analysis. This analysis shows that, when comparisons were performed to estimate the agreement between

Table 1 Demographic and anthropometric characteristics of the study population of 76 adults with cystic fibrosis* Parameter

Total (n ⫽ 76)

Men (n ⫽ 46)

Women (n ⫽ 30)

Age (y) Height (m) Weight (kg) BMI (kg/m2) FEV1%

29.9 (7.9) 1.68 (0.09) 61.0 (9.9) 21.5 (2.5) 60.5 (21.2)

29.2 (7.5) 1.73 (0.07) 64.4 (10.1) 21.5 (2.6) 56.5 (23.6)

31.1 (8.6) NS 1.61 (0.07)‡ 56.0 (7.3)‡ 21.6 (2.5) NS 66.6 (15.4)†

BMI, body mass index; FEV1, forced expiratory volume in 1 s as a percentage of predicted; NS, not significant * Data are means (standard deviations). Significance estimates are for differences between men and women using unpaired t tests: † P ⬍ 0.05, ‡ P ⬍ 0.0005.

1090

S. King et al. / Nutrition 21 (2005) 1087–1094

individual FFM measurements obtained using different methods, the limits of agreement were wide. The coefficients of variation (CV) for FFM measurements using each method compared with DXA were for males and females respectively: 4.7% and 6.9% for BIA (Lukaski), 3.9% and 7.0% for BIA (Segal), and 4.5% and 4.3% for skinfolds. The 95% limits of agreement with DXA FFM were for males ⫺8.3 to 1.7 kg for BIA (Lukaski); ⫺4.8 to 3.6 kg for BIA (Segal); and ⫺2.8 to 7.3 kg for skinfolds. For females, the 95% limits of agreement were ⫺6.4 to 4.8 kg for BIA (Lukaski); ⫺3.1 to 8.9 kg for BIA (Segal); and ⫺1.5 to 5.7 kg for skinfolds. These differences are shown in Fig. 1, which represents the tendency for BIA/Lukaski to underestimate FFM in men and the tendency for BIA/Segal in women and skinfold thickness measurements in men and women to overestimate FFM compared with DXA. Bland-Altman plots of the difference between FFM measurements made in CF subjects using DXA and each method plotted against the average of the FFM measurements are shown in Fig. 2. There was a significant and positive correlation between the means and the differences for women for BIA/Lukaski (r ⫽ 0.40, P ⫽ 0.03) and BIA/Segal (r ⫽ 0.70, P ⬍ 0.0001), suggesting that the bias was not consistent across the range of FFM measurements. No significant correlations were found for skinfolds in men or women or in men when using BIA/Lukaski or BIA/Segal.

Discussion In this study of 76 adults with CF, body composition was measured using quick, non-invasive methods (BIA using the Lukaski equation, BIA using the Segal equation, and skinfold thicknesses), and results were compared with those obtained using DXA as the reference method. Skinfold thicknesses in men and women and BIA/Segal in women overestimated mean FFM compared with DXA, whereas BIA/Lukaski in men underestimated mean FFM. Although FFM measured using all methods was highly correlated with FFM measured using DXA, there was a large discrepancy in comparisons on an individual basis for the non-

Table 3 Mean bias, SD of the bias, CV, and 95% limits of agreement of FFM for each method compared with DXA in 46 male and 30 female adults with cystic fibrosis Comparison

Men (n ⫽ 46) BIA (Lukaski) versus DXA BIA (Segal) versus DXA Skinfolds versus DXA Women (n ⫽ 30) BIA (Lukaski) versus DXA BIA (Segal) versus DXA Skinfolds versus DXA

Mean bias (kg)

SD of bias (kg)

CV (%)

95% limits of agreement (kg)

⫺3.3 ⫺0.6 2.3

2.5 2.1 2.5

4.7% 3.9% 4.5%

⫺8.3 to 1.7 ⫺4.8 to 3.6 ⫺2.8 to 7.3

⫺0.8 2.9 2.1

2.8 3.0 1.8

6.9% 7.0% 4.3%

⫺6.4 to 4.8 ⫺3.1 to 8.9 ⫺1.5 to 5.7

BIA, bioelectrical impedance analysis; CV, coefficient of variation; DXA, dual-energy X-ray absorptiometry; FFM, fat-free mass; SD, standard deviation * Mean FFM/method ⫺ mean FFM/DXA. † 100% ⫻ SD/(mean FFM/method ⫹ FFM/DXA). ‡ Mean bias ⫾ 2 SD.

invasive methods. This is reflected in the results of the Bland-Altman analysis, in which the 95% limits of agreement with FFM measured using DXA were wide for all three methods. This suggests that BIA using the Lukaski or the Segal equations and skinfold thickness measurements will incorrectly estimate FFM in many adults with CF compared with DXA measurements. This study is the first to report comparisons of body composition measurements made using non-invasive bedside methods with those made using DXA in a large group of adults with CF. Our findings confirm those of previous studies in CF patients using other reference methods, which have found good correlations between methods on a group basis, but that agreement between methods for individuals varied widely. In a study of children with CF, measurements of lean body mass were made using dual-photon absorptiometry (a method similar to, but now superseded by, DXA), skinfold thicknesses, and BIA [24]. Although results ob-

Table 2 Mean fat-free mass (kg) measured by different body composition methods* Method

Total

Men

Women

DXA BIA/Lukaski BIA/Segal Skinfolds

49.5 (9.1) 47.1 (8.8)‡ 50.2 (8.5)† 51.7 (9.2)‡

54.8 (7.3) 51.5 (7.8)‡ 54.2 (7.5) NS 57.2 (7.2)‡

41.2 (3.9) 40.4 (4.9) NS 44.1 (5.9)‡ 43.3 (4.3)‡

BIA, bioelectrical impedance analysis; NS, not significant * Data are means (standard deviations). Significance estimates are for differences between DXA FFM and each other method, using paired t tests: † P ⬍ 0.05, ‡ P ⬍ 0.0005.

Fig. 1. Scatterplot showing differences in FFM from DXA in 46 men and 30 women with cystic fibrosis. Difference from FFM as measured by DXA is presented as FFM/method ⫺ FFM/DXA (kg). BIA, bioelectrical impedance analysis; DXA, dual-energy X-ray absorptiometry; FFM, fat-free mass.

S. King et al. / Nutrition 21 (2005) 1087–1094

Fig. 2. Difference between FFM measured using each method and DXA plotted against the average of the two measurements. (a) BIA/Lukaski). (b) BIA/Segal. (c) Skinfolds. Difference from FFM as measure by DXA is presented as FFM/method ⫺ FFM/DXA (kg). The r values represent Pearson’s correlation coefficients for correlations between difference from FFM/DXA and FFM/mean (DXA ⫹ method). BIA, bioelectrical impedance analysis; DXA, dual-energy X-ray absorptiometry; FFM, fat-free mass.

tained using the three methods were highly correlated for the group, values for individual subjects differed widely. Other studies have compared TBW measurements made using BIA with those made using deuterium dilution in CF patients. TBW measurements made using BIA in CF patients were highly correlated with those made using deuterium dilution [25–27], but BIA underestimated TBW compared with deuterium dilution in one study [25] and overestimated TBW in another [27]. The latter study reported wide limits of agreements for TBW estimated using BIA compared with deuterium dilution. The development of

1091

separate predictive equations for use in CF patients was recommended [25,27]. These results, together with our findings, suggest that BIA predictive equations developed for healthy individuals are not suitable for use in CF. Holt et al. compared FFM measurements obtained using BIA and skinfold thicknesses with total body potassium (TBK) in healthy volunteers and subjects who had CF or end-stage liver disease [28]. As in our study, which compared BIA and skinfolds with DXA, they found that BIA and skinfold measurements were highly correlated with TBK in CF subjects. However, the level of agreement between methods using Bland-Altman analysis was unsatisfactory, with the coefficients of variation being 5.0% for BIA and 6.6% for skinfold thicknesses compared with the reference method (TBK), which were similar to our findings of 3.9% to 7.0% for BIA and 4.3% to 4.5% for skinfolds using DXA as the reference method. Another study from the same center reported that BIA measurements were highly correlated with TBK measurements in children with CF [29]. Although the limits of agreement with TBK were narrower for BIA than for anthropometric measurements, it was concluded that BIA measurements could not be considered accurate or reliable predictors of body cell mass or TBK in this population. There are no studies in adults with CF comparing skinfold thickness measurements of FFM with those obtained using DXA. In children with CF, skinfold thickness measurements overestimated mean FFM compared with those estimated using deuterium dilution [22], which is consistent with our finding in adults with CF using DXA as the reference method. The coefficient of variation for measurements of FFM made using skinfolds compared with deuterium dilution was 8.0%, which was higher than our finding of 4.5% for men and 4.3% for women. Age groups and reference methods differed among the reported studies, which may explain the range of results reported for coefficient of variation. There are a number of possible explanations for the findings we report in relation to CF. First, the predictive equations used to calculate body composition in this study were developed and validated in healthy individuals. There are no predictive equations for skinfold thickness measurements or BIA measurements that have been developed and validated specifically for use in subjects with CF. Other investigators concluded that disease-specific predictive equations need to be developed for BIA to be useful for assessing body composition in CF patients [25,27,30]. Second, it has been reported that serum sodium level can alter the resistance measurements observed using BIA [31]. The reasons for this are not clear. It is well known that CF patients have elevated concentrations of sodium and chloride in their sweat and are at increased risk of sodium depletion [32]. It has been suggested that alterations in skin electrolyte composition may influence measurement of whole-body electrical potentials [25], resulting in invalid results for the measurement of body composition in patients

1092

S. King et al. / Nutrition 21 (2005) 1087–1094

with CF using BIA. This might explain the large variability in the BIA results and the wide 95% limits of agreement found between measurements of FFM using DXA and those using BIA for the two predictive equations we evaluated. In other conditions such as chronic liver disease, significant bias and wide limits of agreement have been reported for body composition measurements made using BIA compared with reference methods [33]. In the present study, in women using the BIA/Lukaski and BIA/Segal equations, the bias between FFM values by these equations and FFM by DXA was not consistent across the range of FFM measurements. The reasons for these findings in women but not in men are not entirely clear. No other studies in CF patients have reported these data separated by gender. In a study in adults with CF comparing TBW measurements made using deuterium dilution with those made using BIA, Richards et al. found a significant positive correlation between mean TBW and the difference in TBW between the two methods (male and female results combined) [27]. Another study in underweight elderly adults found that BIA formulas had a low reliability for predicting FFM in female subjects compared with DXA [34]. It is possible that body composition and muscle and fat distributions differ in elderly women and/or women with chronic illnesses such as CF compared with healthy subjects in whom the BIA equations were developed and validated. In a study of healthy adults that developed regression models to predict body composition using BIA measurements, it was found that the best models for women included more variables from the impedance measurements than did the models for men [35]. This observation may explain the greater variability in results obtained in women in our study using predictive equations with fewer variables and at single frequency. These findings suggest that caution needs to be exercised when interpreting results of BIA measurements in disease states, including CF. Third, skinfold thicknesses will be influenced by regional body fat distribution. The rationale for the conversion of skinfold thickness measurements to fat mass and FFM is based on key assumptions that subcutaneous adipose tissue represents a constant proportion of total body fat, and that the sites selected for measurement are representative of the average thickness of subcutaneous adipose tissue across the whole body [15]. These assumptions may not be valid under all conditions because the degree to which subcutaneous adipose tissue reflects total body fatness may change with age, gender, race, and disease [36]. There are no predictive equations developed and validated for the estimation of body composition from skinfold thickness measurements in disease states, including CF. Other limitations of this technique include the potential for inter- and intraobserver variabilities [36]. In our study, skinfold thicknesses tended to overestimate FFM in adults with CF, suggesting that the extent of FFM depletion could be underestimated if skinfold thicknesses are used to assess body composition. Interpretation of the results of body composition assessments made using skinfold thickness measurements and generic predic-

tive equations should take into account these limiting factors. Fourth, it is acknowledged that there are other non– disease-specific predictive equations for estimating body composition from skinfold thickness and BIA measurements. Different results may have been obtained if different predictive equations were used. The equations used in this study were selected because they had been developed and validated in relatively large populations with an age range similar to that of our CF subjects. Other studies in CF [22,24,27–29] and other populations [34,37] have reported conclusions similar to those of our study, suggesting that the choice of equation is not the crucial determinant of our conclusions. DXA was selected as the reference method in this study. The potential advantages of DXA for assessing body composition include high precision [13], good agreement with results obtained using “gold standard” multicompartment models [38], good short-term reproducibility [36], and scanning of the entire body. This means that subcutaneous and visceral fats are included in the estimates of fat mass (unlike skinfold thicknesses) and that the resulting estimates of FFM and fat mass have encompassed regional differences (which may not apply to site-specific measurements such as skinfold thicknesses). Currently, there are no studies in adults with CF comparing body composition measurements made using DXA with those made using TBK counting or in vivo neutron activation analysis (IVNAA). DXA has been validated against IVNAA, TBK counting, and TBW using deuterium dilution in other populations [36,39]. In a study comparing various body composition methods in adults with chronic liver disease, it was concluded that DXA was superior to anthropometry and BIA for measuring FFM and fat mass when compared with a gold standard multicompartment method [33]. Limitations of DXA include the lack of portability of the equipment, meaning that subjects can be assessed only at centers that have a DXA scanner; radiation exposure associated with the scan (albeit at a very small dose); and the requirement for subjects to lie still, which may limit its application in patients who have CF and severe lung disease if they cannot avoid coughing during the scan. Another potential limitation of DXA is that the estimation of FFM assumes a constant hydration of FFM, which may not hold true for all ages or in some disease states [13]. Alterations in the hydration of FFM have been shown to occur in different races, during pregnancy, and in association with edema [40]. Reports have suggested that a 5% change in the hydration of FFM will affect the estimate of body fat using DXA by 1% to 2.5% [38]. Although there are no reports documenting the hydration of FFM in adults with CF, in the present study, all subjects were Caucasian, and no subject was pregnant or had clinical signs of edema. This suggests that these factors are unlikely to affect the accuracy of DXA measurements in our adult patients with CF.

S. King et al. / Nutrition 21 (2005) 1087–1094

DXA as a body composition method has some other important limitations. Body composition estimates may differ between DXA systems produced by the various commercial manufacturers [42], meaning that repeat measurements on individual subjects should ideally be undertaken using the same system. DXA assumes a constant distribution of fat over bone tissue and bone-free tissue, which means that estimates of soft tissue may not be as accurate over regions with a high bone content (e.g., arms, thorax) as the regions with low bone content [43]. It is possible that this or other limitations of DXA may have contributed to our findings of differences between FFM measurements using DXA, BIA, and skinfold measurements. Nevertheless, despite the limitations of DXA, it has been argued that it is one of the best reference methods available for body composition assessment [14], particularly because of its wider availability compared with other techniques such as TBK counting and IVNAA. The consistency between our findings and those of other studies in CF patients comparing non-invasive methods with other reference methods suggest that the body composition measurements made using BIA and skinfolds be interpreted with caution until CF-specific predictive equations are developed and validated. The results of this study also emphasize the importance of gender-specific analysis of body composition data in studies of adults with CF because of the different distribution of FFM and body fat between men and women. This may have particular implications for women who have CF and poorer survival than men who have CF, for reasons that are not yet fully understood [44]. It is possible that nutritional factors may contribute to this survival disadvantage. Our finding that the bias in FFM measurements obtained using BIA was not consistent across the range of FFM values for women suggests the potential for misclassification of an individual patient’s nutritional status and for inappropriate institution or withholding of nutritional interventions. It is important to use accurate methods to assess and monitor nutritional status and body composition to select the most appropriate candidates for intervention.

Acknowledgment The authors acknowledge the assistance of Libby Francis with the conduct of the study.

References [1] Dowsett J. An overview of nutritional issues for the adult with cystic fibrosis. Nutrition 2000;16:566 –70. [2] Sinaasappel M, Stern M, Littlewood J, et al. Nutrition in patients with cystic fibrosis: a European Consensus. J Cystic Fibrosis 2002;1:51– 75. [3] King SJ, Topliss DJ, Kotsimbos TC, et al. Reduced bone density in cystic fibrosis: deltaF508 mutation is an independent risk factor. Eur Respir J 2005;25:54 – 61.

1093

[4] Sharma R, Florea VG, Bolger AP, et al. Wasting as an independent predictor of mortality in patients with cystic fibrosis. Thorax 2001; 56:746 –50. [5] Ionescu AA, Nixon LS, Evans WD, et al. Bone density, body composition, and inflammatory status in cystic fibrosis. Am J Respir Crit Care Med 2000;162:789 –94. [6] Ionescu AA, Nixon LS, Luzio S, et al. Pulmonary function, body composition, and protein catabolism in adults with cystic fibrosis. Am J Respir Crit Care Med 2002;165:495–500. [7] Pietrobelli A, Wang Z, Heymsfield SB. Techniques used in measuring human body composition. Curr Opin Clin Nutr Metab Care 1998;1: 439 – 48. [8] Plourde G. The role of radiologic methods in assessing body composition and related metabolic parameters. Nutr Rev 1997;55:289 – 96. [9] McNaughton SA, Shepherd RW, Greer RG, Cleghorn GJ, Thomas BJ. Nutritional status of children with cystic fibrosis measured by total body potassium as a marker of body cell mass: lack of sensitivity of anthropometric measures. J Pediatr 2000;136:188 –94. [10] Haslam RHM, Borovnicar DJ, Stroud DB, Strauss BJG, Bines JE. Correlates of prepubertal bone mineral density in cystic fibrosis. Arch Dis Child 2001;85:166 –71. [11] Rochat T, Slosman DO, Pichard C, Belli DC. Body composition analysis by dual-energy x-ray absorptiometry in adults with cystic fibrosis. Chest 1994;106:800 –5. [12] Salamoni F, Roulet M, Gudinchet F, et al. Bone mineral content in cystic fibrosis patients: correlation with fat-free mass. Arch Dis Child 1996;74:314 – 8. [13] Laskey MA. Dual-energy X-ray absorptiometry and body composition. Nutrition 1996;12:45–51. [14] Van Loan M. Is dual-energy X-ray absorptiometry ready for prime time in the clinical evaluation of body composition? Am J Clin Nutr 1998;68:1155– 6. [15] Lukaski HC. Methods for the assessment of body composition: traditional and new. Am J Clin Nutr 1987;46:537–56. [16] Segal KR, Van Loan M, Fitzgerald PI, Hodgdon JA, Van Itallie TB. Lean body mass estimation by bioelectrical impedance analysis: a four-site cross-validation study. Am J Clin Nutr 1988;47:7–14. [17] Lukaski HC, Bolonchuk WW. Theory and validation of the tetrapolar bioelectrical impedance method to assess human body composition. In: Ellis KJ, Yasumura S, Morgan WD, editors. Proceedings of an international symposium of in vivo body composition studies. London: Institute of Medicine and Physics; 1987. p. 410 – 4. [18] Newby MJ, Keim NL, Brown DL. Body composition of adult cystic fibrosis patients and control subjects as determined by densitometry, bioelectrical impedance, total-body electrical conductivity, skinfold measurements, and deuterium oxide dilution. Am J Clin Nutr 1990; 52:209 –13. [19] Knudson RJ, Lebowitz MD, Holberg CJ, Burrows B. Changes in the normal maximal expiratory flow-volume curve with growth and aging. Am Rev Respir Dis 1983;127:725–34. [20] Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16 –72 years. Br J Nutr 1974;32:77– 97. [21] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;I:307– 10. [22] de Meer K, Gulmans VA, Westerterp KR, Houwen RH, Berger R. Skinfold measurements in children with cystic fibrosis: monitoring fat-free mass and exercise effects. Eur J Pediatr 1999;158:800 – 6. [23] Corey M, Farewell V. Determinants of mortality from cystic fibrosis in Canada, 1970 –1989. Am J Epidemiol 1996;143:1007–17. [24] Lands LC, Gordon C, Bar-Or O, et al. Comparison of three techniques for body composition analysis in cystic fibrosis. J Appl Physiol 1993;75:162– 6.

1094

S. King et al. / Nutrition 21 (2005) 1087–1094

[25] Azcue M, Fried M, Pencharz PB. Use of bioelectrical impedance analysis to measure total body water in patients with cystic fibrosis. J Pediatr Gastroenterol Nutr 1993;16:440 –5. [26] Borowitz D, Conboy K. Are bioelectric impedance measurements valid in patients with cystic fibrosis? J Pediatr Gastroenterol Nutr 1994;18:453– 6. [27] Richards ML, Bell SC, Edminston KA, Davies PSW. Assessment of bioelectrical impedance analysis for the prediction of total body water in cystic fibrosis. Asia Pac J Clin Nutr 2003;12:161–5. [28] Holt TL, Cui C, Thomas BJ, et al. Clinical applicability of bioelectric impedance to measure body composition in health and disease. Nutrition 1994;10:221– 4. [29] Quirk PC, Ward LC, Thomas BJ, et al. Evaluation of bioelectrical impedance for prospective nutritional assessment in cystic fibrosis. Nutrition 1997;13:412– 6. [30] Pichard C, Kyle UG, Slosman DO. Fat-free mass in chronic illness: comparison of bioelectrical impedance and dual-energy x-ray absorptiometry in 480 chronically ill and healthy subjects. Nutrition 1999; 15:668 –76. [31] Forbes GB. Perspectives on body composition. Curr Opin Clin Nutr Metab Care 2002;5:25–30. [32] Shepherd RW, Cleghorn G, Ward LC, Wall CR, Holt TL. Nutrition in cystic fibrosis. Nutr Res Rev 1991;4:51– 67. [33] Strauss BJ, Gibson PR, Stroud DB, et al. Total body dual X-ray absorptiometry is a good measure of both fat mass and fat-free mass in liver cirrhosis compared to “gold-standard” techniques. Melbourne Liver Group. Ann NY Acad Sci 2000;904:55– 62. [34] Lupoli L, Sergi G, Coin A, et al. Body composition in underweight elderly subjects: reliability of bioelectrical impedance analysis. Clin Nutr 2004;23:1371– 80.

[35] Chumlea WC, Guo SS, Cockram DB, Siervogel RM. Mechanical and physiologic modifiers and bioelectrical impedance spectrum determinants of body composition. Am J Clin Nutr 1996;64. [36] Brodie D, Moscrip V, Hutcheon R. Body composition measurement: a review of hydrodensitometry, anthropometry, and impedance methods. Nutrition 1998;14:296 –310. [37] Piers LS, Soares MJ, Frandsen SL, O’Dea K. Indirect estimates of body composition are useful for groups but unreliable in individuals. Int J Obes 2000;24:1145–52. [38] Lohman TG, Harris M, Teixeira PJ, Weiss L. Assessing body composition and changes in body composition. Another look at dualenergy X-ray absorptiometry. Ann NY Acad Sci 2000;904:45–54. [39] Ellis KJ. Human body composition: in vivo methods. Physiol Rev 2000;80:649 – 80. [40] Jebb SA, Elia M. Multi-compartment models for the assessment of body composition in health and disease. In: Davies PSW, Cole TJ, editors. Body composition techniques in health and disease. Cambridge: Cambridge University Press; 1995. p. 240 –54. [41] Hannan WJ, Cowen SJ, Freeman CP, Wrate RM, Barton J. Evaluation of dual-energy X-ray absortiometry for the assessment of body composition in anorexic females. In: Ellis KJ, Eastman JD, editors. Human body composition. New York: Plenum Press; 1993. p. 169 –72. [42] Tothill P, Hannan WJ, Wilkinson S. Comparisons between a pencil beam and two fan beam dual energy X-ray absorptiometers used for measuring total body bone and soft tissue. Br J Radiol 2001;74:166 –76. [43] Genton L, Hans D, Kyle UG, Pichard C. Dual-energy X-ray absorptiometry and body composition: differences between devices and comparison with reference methods. Nutrition 2002;18:66 –70. [44] Rosenfeld M, Davis R, FitzSimmons S, Pepe M, Ramsey B. Gender gap in cystic fibrosis mortality. Am J Epidemiol 1997;145:794 – 803.