A multi-center comparison of dual energy X-ray absorptiometers: In vivo and in vitro soft tissue measurement

Transcription

1 European Journal of Clinical Nutrition (1997) 51, 312±317 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00 A multi-center comparison of dual energy X-ray absorptiometers: In vivo and in vitro soft tissue measurement CD Economos 1, ME Nelson 1, MA Fiatarone 1, GE Dallal 1, SB Heyms eld 2, J Wang 2, S Yasumara 3, RMa 3, AN Vaswani 4, M Russell-Aulet 5 and RN Pierson 2 1 Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, Boston, MA; 2 Body Composition Unit at St. Luke's- Roosevelt Hospital, New York, NY; 3 Brookhaven National Laboratory, Upton, NY; 4 Osteoporosis Research Clinic, Winthrop University Hospital, Mineola, NY and 5 The University of Michigan, Ann Arbor, MI, USA Objective: To assess intra- and inter-site soft tissue variability by dual energy X-ray absorptiometry (DXA). Design: Cross-sectional trial. Setting: Three medical research institutions. Subjects: Five humans (in vivo) and four phantoms (in vitro), con gured from two whole body phantoms with arti cial skeletons and thickness overlays. Interventions: Duplicate total-body DXA scans were performed on all subjects at each institution within a 15 d period. Results: All intra-site coef cients of variation (CV) were < 0.5% for total tissue mass, but in vitro and in vivo Cvs were 7.2% and 2.3% for fat mass (FM) and 2.5% and 0.9% for lean mass (LM), respectively. Several totalbody and regional FM and LM measurements were signi cantly different between sites (P <0.05), with percent differences between sites ranging from 2.6±13.3% for FM and from 1.6±13.6% for LM. Site 2 was consistently lower for FM and Site 3 was consistently lower for LM. Conclusions: These results stress the need for both rigorous and standardized cross-calibration procedures for soft tissue measurement by DXA. Sponsorship: This work has been supported in part by NIH Training Grant #T32AG00209, grant P01-DK42618 from the National Institutes of Health, federal funds from the US Department of Agriculture, and Agricultural Research Services contract 53-3K Dr Nelson is currently a Brookdale National Fellow. Descriptors: dual energy X-ray absorptiometry; DXA; densitometry; body composition; lean mass; fat mass. Introduction The nutritional assessment of an individual is enhanced tremendously by the accurate measurement of body composition. Since their derivation in the early 1980s, dual energy projection methods have been used to quantify bone mass and assess fracture risk. Now the prevailing technique, dual energy X-ray absorptiometry (DXA), also measures total-body soft tissue and quanti es the fat and lean components. [The acronym DXA is used throughout this text as recommended by Wilson (Wilson et al, 1990).] DXA is non-invasive and when performed properly it is safe and well tolerated, characterizing it as a promising technique for extensive use in body composition research. However, individual variability in the composition of soft tissue as well as DXAs dif culty in precisely partitioning the soft tissue into lean and fat compartments are recognized obstacles for DXAs routine use (Roubenoff et al, 1993). Therefore, further validation is required before DXA Correspondence: Dr ME Nelson, HNRC, 711 Washington Street, Tufts University, Boston, MA 02111, USA. The contents of this publication do not necessarily re ect the views or policies of the US Department of Agriculture, neither does the mention of trade names, commercial products, or organizations imply endorsement by the US government. Portions of this research were presented at the American Society for Clinical Nutrition Meeting May, 1995 in San Diego, CA. Received 13 September 1996; revised 18 December 1996; accepted 13 January should be accepted as a standard measuring technique for body composition. The reliability of DXA depends upon operator-, subject-, and apparatus-factors. The precision of individual DXA instruments for the assessment of bone is excellent, with coef cients of variation (CV) < 1.0% (Johnson et al, 1991; LeBlanc et al, 1990; Mazess et al, 1989; Nuti et al, 1987; Fuller et al, 1992; Compston et al, 1992). On the other hand, Cvs for soft tissue measurements of up to 3.1% for lean tissue, 8.6% for fat tissue, and 7.9% for percent body fat have been observed (Table 1) Mazess et al, 1990; Johnson et al, 1991; Fuller et al, 1992; Compston et al, 1992; Heyms eld et al, 1989; Kelly et al, 1991; Slosman et al, 1992; Haarbo et al, 1991; Pritchard et al, 1993; Rico et al, 1994). The difference in precision between bone and soft tissue is not surprising since DXA was essentially designed to assess bone and soft tissue is heterogeneous. Nevertheless, assumptions that bone precision extends to soft tissue precision are often made erroneously. DXAs widespread availability and relatively inexpensive operation have made it an appealing technique for body composition assessment in multi-center research trials. Cross-validation studies of instruments at different sites are necessary in order to directly compare body composition data generated by different instruments. However, the few studies in the literature examining the agreement between DXA instruments from the same manufacturer, located at different clinical centers, have only

2 Table 1 Summary of intra-site DXA soft tissue precision A. Short-term 313 Author Instrument Duration Subjects (n) Scans/subject CV (%) Fat Lean % Fat Heyms eld 1989 Lunar DPX 5 days 4 5 Ð Ð 0.50 Mazess 1990 Lunar DPX 5±7 days Fuller 1992 Lunar DPX 1 day Ð Compston 1992 Lundar DPX 1 day Ð Pritchard 1993 Hologic QDR days Kelly 1991 Hologic QDR day 13 2± Rico 1994 Norland XR-26 1±16 weeks 10 2 Ð Economos (current study) B. Long-term Lundar DPX 1 day Lunar DPX 1 day Lunar DPX-L 1 day Author Instrument Duration Subjects (n) Scans/subject CV (%) Fat Lean % Fat Johnson 1991 Lunar DPX 1 and 9 months Ð Haarbo 1991 Lunar DPX 1 and 6 months Slosman 1992 Hologic QDR and 12 months Ð Rico 1994 Norland XR-26 1±24 weeks 8 6 Ð examined bone measurements (Kelly et al, 1989; Blake et al, 1991; Orwoll et al, 1991; Rencken et al, 1991; Trevisan et al, 1992; Economos et al, 1996; Wahner et al, 1994). To date, no studies have reported precision in soft tissue between sites, using the same manufacturer, with the exception of a brief report (Paton et al, 1995). Since it is not appropriate to assume that the level of agreement for bone, using instruments from the same manufacturer at different sites would be the same for soft tissue, we conducted a multi-center comparison of DXAs for in vivo and in vitro soft tissue. In a three-site research project that proposed to pool and analyze data cumulatively, the performance of instruments common to each of the sites was investigated to verify the agreement in measurements of total body weight, total soft tissue mass (TM), fat tissue mass (FM), and lean tissue mass (LM) within and between sites. Methods For a more detailed description of the methods please refer to a previously published paper (Economos et al, 1996). Research sites Within 15 d, all measurements were performed at three research sites. The centers were: (site 1) The Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, Boston, MA, (site 2) St. Lukes Roosevelt Hospital Center, New York, NY, and (site 3) Winthrop University Hospital, Mineola, NY. Instruments DXA measurements were made with Lunar DPX (sites 1 and 2), software version 3.6z, and DPX-L (site 3) software version 1.3z, which, according to the manufacturer, are analogous. The differential absorption of two X-ray energies allows two components, differing in electron density, to be determined simultaneously (Mazess et al, 1990). The soft tissue mass of the whole body is determined by measuring the R value, the ration of high energy to low energy attenuation, for each pixel that contains no signi cant bone, and extrapolating for soft tissue content in bone-containing image pixels (Mazess et al, 1990; Peppler and Mazess, 1981). Theoretically, DXA measures three tissues: bone mineral, fat, and lean (fat-free mineral-free tissue). Human subjects (in vivo) and phantoms (in vitro) Five healthy subjects (3M, 2F) varying in size (53.6± 99.5 kg body weight) and age (28±61 y of age) and two sizes of whole body polyethylene Bottle Mannequin Absorption Phantoms (BOMAB), with overlays, from Atlan-Tech, Atlanta, GA. Were used in the study. The phantom sizes approximate a reference man (66.4 kg), and a small reference woman (49.9 kg) (Kramer et al, 1991). The mass attenuation coef cients (m/r) used in calculations of photon (X-ray) penetration in biological materials were similar for the materials used and the tissues that were simulated at photon energies of 40 and 70 kev (Hubbell 1982). When the overlays were lled with water and laid over the water- lled phantoms, they represented a large man of 97.3 kg and a large woman of 78.5 kg and were used to determine whether any changes in attenuation occurred in larger subjects, which may require correction. Protocol Within a two day period at each research site, two totalbody scans were performed in immediate succession, after repositioning of each subject and phantom con guration (with and without overlay). The protocol for obtaining and analyzing data was standardized. For each total-body scan, the total and regional (appendicular and trunk), fat mass (FM), lean mass (LM) (fat-free, mineral-free), total tissue mass (TM) (fat lean masses), and total body weight (fat lean bone masses) were

3 314 Table 2 Comparison of tissue weight by scale and by DXA for all sites A. In vivo Scale weight (kg) b DXA weight a (kg) c DXA vs scale correlation Mean d (kg) s.d. d RC Site 1 n ˆ r-0.09; P ˆ Site 2 n ˆ r ˆ 0.99 P ˆ Site 3 n ˆ r ˆ 0.99; P ˆ Values are presented as mean s.d. a Total body weight ˆ fat mass lean mass bone mineral content. b NS (between sites) by ANOVA.#15; c Site 1 > Site 2 and 3 by ANOVA. d ˆ [DXA weight (g) (fat mass lean mass bone mineral content) Ð Scale weight (g)]. Mean d ˆ [( Pd )/n]. s.d. d ˆ Standard deviation of the differences ˆ square root of [( Pd )/n]. RC ˆ 2s.d. d B. In vitro Scale weight (kg) b DXA weight a (kg) c DXA vs scale correlation Mean d (kg) s.d. d RC Site n ˆ r ˆ 0.99; P ˆ Site 2 n ˆ r ˆ 0.99; P ˆ Site 3 n ˆ r ˆ 0.99; P ˆ Values are presented as mean s.d. a Total body weight ˆ fat mass lean mass bone mineral content. b NS (between sites) by ANOVA. c Site 2 > Site 3 by ANOVA. d ˆ [DXA weight (g) (fat mass lean mass bone mineral content) - Scale weight (g)]. Mean d ˆ [( d )/n]. s.d. d ˆ Standard deviation of the differences ˆ square root of [( d )/n]. RC ˆ 2s.d. d. measured. We report lean mass, which does not include the bone mineral, although there were no differences in the analyses whether the bone mineral data were included or excluded. The bone results from the study have been previously reported (Economos et al, 1996). Statistical analysis Total body weight by DXA, calculated from the average of measurements 1 and 2, was compared with total body weight by scale within sites using the Bland and Atlman method (Bland and Atlman, 1986). Between site comparisons of total body weight by DXA and total body weight by scale was done using Analysis of variance (ANOVA) with Tukeys HSD (honesty signi cant difference) for both groups [humans (n ˆ 5) and phantoms (n ˆ 4)]. The signi cance level was set at P < Intra-site reproductiblity was calculated as the coef cient of variation (CV) from duplicate TM, FM, and LM measurements by the formula from Nilas (Nilas et al, 1988). The CV provides a measure that can be compared across studies of measurement precision (Table 1). For inter-site comparison, the site means and standard deviations of the total and regional (appendicular and trunk), and LM were calculated from the average of measurements 1 and 2. ANOVA was used to test for signi cant differences between sites and Tukeys HSD was used to compare sites and adjust for multiple comparisons. Results were considered signi cant if P < The inter-site Cvs for FM and LM were calculated as: (CV ˆ s.d./mean 6 100), where the mean ˆ site 1 site 2 site 3/3. The mean percentage difference (MPD) between sites was calculated for FM and LM as MPD (%) ˆ [(site 1 7 site 2j)=(site 1 site 2/2)] All data were analyzed using the SAS statistical package, version 6.06 (SAS 1990). Results Total body weight comparisons Tables 2a and 2b show the in vivo and in vitro comparisons, both within and between sites, for total body weight by scale and by DXA. Within sites, total body weight by scale and by DXA were highly correlated (r ˆ 0.99, P < ), but the Bland and Altman method showed poor agreement between body weight by scale and by DXA for sites 1 and 3 in vivo and sites 1, 2, and 3 in vitro. There were no signi cant differences between sites for total body weight by scale, but there were signi cant differences in vivo and in vitro in the total body weight by DXA. Intra-site comparisons For total-body TM, Cvs were < 0.5% with ranges of 0.3± 0.6% for in vivo and 0.3±0.7% for in vitro. Total-body FM Cvs ranged from 1.0±3.5% in vivo and 3.1±10.7% in vitro and total body LM Cvs ranged from 0.7±1.1% in vivo and 1.0±3.5% in vitro. Inter-site comparisons Inter-site variability of total FM and LM, expressed as Cvs between the three sites, is depicted in Figure 1. The two human subjects that were at the extreme ends of the body size spectrum (subject 1 ˆ 1 largest and subject 5 ˆ smallest), had the highest inter-site Cvs for FM (1 ˆ 9.2%, 5 ˆ 11.0%) and LM (1 ˆ 2.9% and 5 ˆ 3.6%). There were no trends related to size for phantom precision. Figure 2 shows the in vivo and in vitro comparisons of the FM and LM between sites. The in vivo total FM from site 2 was signi cantly lower than sites 1 and 3 (P < 0.05) by 13.3% and 12.0%, respectively. The in vivo total LM from site 3 was signi cantly lower than sites 1 and 2 (P < 0.05) by 3.5% and 4.9%, respectively, and the in vitro LM from site 3 was signi cantly lower than site 2

4 315 Figure 1 Inter-site precision of DXA soft tissue measurements. CV ˆ coef cient of variation. The values represent the fat and lean mass Cvs [(s.d./mean) 6 100] calculated for each individual human (in vivo, n ˆ 5) and phantom (in vitro, n ˆ 4) across the three sites. The means are shown for comparison between in vivo and in vitro results and fat and lean results. The mean inter-site CVs were 7.5 and 7.9% for in vivo and in vitro fat mass, and 2.7% and 4.1% for in vivo and in vitro lean mass. Figure 3 In vivo and in vitro regional soft tissue by DXA. The values represent the means s.e.m. for appendicular (arms legs) and trunk fat and lean tissue in grams for each site, both in vivo and in vitro. ANOVA and Tukeys HSD were used to test for differences between sites (P < 0.05). Figure 2 In vivo and in vitro total body soft tissue by DXA. The values represent the means s.e.b. for total-body fat and lean tissue in grams for each site, both in vivo and in vitro. ANOVA and Tukeys HSD were used to test for differences between sites (P < 0.05). (P < 0.05) by 13.6%. To further examine the differences between sites, Figure 3 shows the regional (appendicular and trunk) comparisons of FM and LM from each site. Discussion An ample amount of work has been done to investigate the intra- and inter-site precision of DXA for bone (Mazess et al, 1990; Johnson et al, 1991; LeBlanc et al, 1990; Mazess et al, 1989; Nuti et al, 1987; Fuller et al, 1992; Compston et al, 1992; Kelly et al, 1991; Economos et al, 1996; Seto et al, 1991). Within the last ve years, quite a few studies examining soft tissue intra-site precision have been published (see Table 1). While most of the total-body lean tissue Cvs would be considered acceptable, fat tissue Cvs are not. Furthermore, regional Cvs are generally discouraging, with reports of regional lean tissue measurements ranging from 1.3±4.2% and regional fat tissue measurements ranging from 2.4±12.6% (Mazess et al, 1990; Johnson et al, 1991; Fuller et al, 1992). The incongruous ndings from these studies do not support the use of this technique for body composition measurement as a gold standard. Compared to the precision of other body composition techniques ( 1% for muscle tissue from computed tomography, 1.5% for lean and fat tissue for hydrodensitometry, 7% for skinfolds, 2% for bioelectric impedance (Nelson et al, 1996)] DXA is comparable to the less sophisticated techniques. Nevertheless, DXA is currently being used as a standard technique for the assessment of body composition and to monitor the conventional change in body composition over time or as a result of an intervention. Unlike bone, standard monitoring of soft tissue precision using phantoms of known composition is relatively limited, and most often small, partial body phantoms are used (Kelly et al, 1991; Haarbo et al, 1991; Jensen et al, 1993; Laskey et al, 1991). The soft tissue CVs in these studies vary dramatically, from 0.3±7.0%, due to the various phantom sizes and thicknesses and the variability of the fat and lean equivalent materials. Using weight stable, whole-body phantoms with invariable body compositions we were able to show that intra-site reproducibility for total-body TM is highly acceptable and similar to results seen in humans. In fact, much of the data supporting DXAs ability to accurately measure soft tissue is based on evidence that total tissue mass by DXA correlates signi cantly to scaled weight (Gotfredsen et al, 1986; Ellis et al, 1994; Johansson et al, 1993; Dawson- Hughes et al, 1992; Svendsen et al, 1991; Going et al, 1993). However, simple correlation analyses can be misleading because they do not address the reproducibility, only the linear association. This is supported by our data, shown in Table 2, where correlationss between DXA weight and scale weight were very strong, nevertheless

5 316 we found signi cant differences between method, indicating disagreement in the results. Still, as made evident by a review of the literature, researchers often draw erroneous conclusions regarding DXAs ability to measure body composition merely based on evidence that total tissue weight reconstruction is satisfactory; a phenomenon which provides no direct evidence about its accuracy for measuring the individual masses of lean, fat, or bone. Our data supports the argument that DXAs ability to predict total tissue weight is constant, but its ability to compartmentalize FM and LM is variable. Over the last few years, a number of studies have begun to investigate soft tissue accuracy by DXA, (Pierson et al, 1991; Withers et al, 1992; Jensen et al, 1993; Ellis et al, 1994; Svendsen et al, 1993; Jebb et al, 1995). Nonetheless, until a number of comparisons using well-developed measurement techniques are repeated, accuracy cannot be con rmed. Based on our calculations using physical tables published by Hubbel (Hubbell, 1982), DXA principles (Mazess et al, 1990), and the experience of others (Kelly et al, 1991; Laskey et al, 1991), we expected the materials used to construct our phantoms to be detected as the same proportion of LM and FM with repeat scanning. However, compared to humans, the LM and FM of the phantoms were poorly reproduced; CVs up to 2.5 for LM and up to 10.7 for FM. Since our results were not consistent with our presumptions, we resolved that these phantoms do not have much potential for use in maintaining intra-site quality control for DXA scanning in the total-body mode. The CVs reported here are clearly un t for use in research and underscores the need for the development of a soft tissue phantom for total-body DXA studies. The question of whether similar DXA instruments at various centers demonstrate acceptable agreement for soft tissue has only once before been reported, which was a letter to the editor (Paton et al, 1995). Despite excellent agreement for TM between identical instruments, the authors reported an average 3.0 kg and 3.9 kg difference for fat and lean tissue, respectively. Our nding that the inter-site precision of FM was less favorable than that of LM is consistent with intra-site ndings by the current study and others (Table 1), which all report better precision for LM than for FM. Signi cant disagreements among instruments, as reported in this cross-sectional study as well as Paton et al, 1995, would make it dif cult to compare the results obtained in different laboratories and would interfere seriously with the use of DXA for analysis of soft tissue in both research and clinical medicine. Furthermore, our results indicate that single-and multicenter intervention studies should be extremely concerned about this variability since the expected changes in FM and LM may be less than the variability within and between sites, which could confound the results from longitudinal studies. It is also important to note that the present day study was performed under very stringent conditions. We used one analyzer, weight-stable humans and whole-body phantoms, internally well-controlled research facilities, and DXA instruments of the same manufacturer, yet there were still discrepancies, indicating that differences in the machines calibration was the likely source of variability. Therefore, the development and availability of appropriate quality control soft tissue phantoms that would allow the con dent interpretation of DXA from serial patient measurements cannot be overstated, especially for situations where soft tissue thicknesses change over time, as in growth, weight loss or gain, illness and disease. Because of the proprietary nature of the DXA it is not possible to fully speculate the reasons for the imprecision of the technique. However, several factors are most likely responsible: The technique was not developed for extreme body thicknesses; partial body phantoms varying in size, thickness and soft tissue equivalent materials have been used to determine and monitor precision; human soft tissue is heterogeneous in nature; and the technique must extrapolate for soft tissue content in 40% of the body area where there are bone containing image pixels. It is imperative for scientists to work closely with the DXA manufacturers to develop this promising technique to its maximum potential. Conclusions Our results show that very similar or identical DXA instruments may produce signi cant inter- and intra-site differences in soft tissue measurements with human subjects and whole-body phantoms. In addition, whole-body phantoms can be used for multi-center cross-comparison studies when the identi cation of differences between sites is essential and when the transfer of human subjects between sites is not feasible. Nevertheless, for intra- and inter-site measurements to be truly comparable, measurements with a group of common in vivo and in vitro phantoms are suggested. These results stress the need for both rigorous and standardized cross-calibration procedures for soft tissue measurement by DXA. AcknowledgementsÐWe would like to thank the members of the Calcium and Bone Metabolism Laboratory at The Jean Mayer USDA HNRC at Tufts University, The Bone Density and Body Composition Laboratories at St. Luke-Roosevelt Hospital and the Bone Laboratory at Winthrop University Hospital. References Blake GM, Tong CM & Fogelman I (1991): Intersite comparison of the Hologic QDR-1000 dual energy x-ray bone densitometer. Br. J. Radiol. 64, 440±446. Bland JM & Altman DGM (1986): Statistical methods for assessing agreement between two methods of clinical measurement. The. Lancet. 2, 307±310. Compston JE, Bhambhani M, Laskey MA, Murphy S & Khaw KT (1992): Body composition and bone mass in post-menopausal women. Clin. Endocrinol. 37, 426±431. Dawson-Hughes B & Harris S. (1992): Regional changes in body composition by time of year in healthy postmenopausal women. Am. J. Clin. Nutr 56, 307±313. Economos CD, Nelson ME, Fiatarone MA, Dallal GE, Heyms eld SB, Russell-Aulet M, Yasumura S, Ma R, Vaswani AN, Pierson RN & Wang J (1996): A multi-center comparison of dual energy X-ray absorptiometers: In vivo and in vitro measurements of bone mineral content and density. J. Bone. Min. Res. 11, 275±285. Ellis KJ, Shypailo RJ, Pratt J & Pond WG (1994): Accuracy of dual-energy x-ray absorptiometry for body composition measurements in children. Am. J. Clin. Nutr. 60, 660±665. Fuller NJ, Jebb SA, Laskey MA, Coward WA & Elia M (1992): Fourcomponent 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. 82, 687±693. Fuller NJ, Laskey MA & Elia M (1992): Assessment of the composition of major body regions by dual-energy x-ray absorptiometry (DEXA) with special reference to limb muscle. Clin. Physiol. 12, 253±256. Going SB, Massett MP, Hall MC, Bare LA, Root PA, Williams DP & Lohman TG (1993): Detection of small changes in body composition by dual-energy x-ray absorptiometry. Am. J. Clin. Nutr. 57, 845±850. Gotfredsen A, Jensen J, Borg J & Christiansen C (1986): Measurement of lean body mass and total fat using dual photon absorptiometry. Metabolism 35, 88±93. Haarbo J, Gotfredsen A, Hassager C, Christiansen C (1991): Validation of body composition by dual energy X-ray absorptiometry (DEXA). Metabolism. 11, 331±341. Hangartner TN & Johnston CC (1990): In uence of fat on bone measure-

6 ments with dual-energy absorptiometry. Bone. and. Mineral. 9, 71±81. Heyms eld SB, Wang J, Heshka S, Keyayias JJ & Pierson RN (1989): Dual-photon absorptiometry: comparison of bone mineral and soft tissue mass measurements in vivo with established methods. Am. J. Clin. Nutr. 49, 1283±1289. Heyms eld SB, Wang J, Kehayias J, Heshka S, Lichtman SB & Pierson RN (1989): Chemical determination of human body density in vivo: relevance to hydrodensitometry. Am. J. Clin. Nutr. 50, 1282±1290. Hubbell JH (1982): Photon mass attenuation and energy-absorption coef cients from 1 kev to 20 MeV. Int. J. Appl. Rad. Isot. 33, 1269± Jebb AS, Goldberg GR, Jennings G & Elia M (1995): Dual-energy X-ray absorptiometry measurements of body composition: effects of depth and tissue thickness, including comparisons with direct analysis. Clin. Sci. 88, 319±324. Jensen MD, Kanaley JA, Roust LR, OBrien PC, Braun JS, Dunn WL & Wahner HW (1993): Assessment of body composition with use of Dual- Energy X-ray Absorptiometry: Evaluation and comparison with other methods. Mayo. Clin. Proc. 68, 867±873. Johansson AG, Forslund A, Sjodin A, Mallmin H, Hambraeus L & Ljunghall S (1993): Determination of body composition ± a comparison of dual-energy x-ray absorptiometry and hydrodensitometry. Am. J. Clin. Nutr. 57, 323±326. Johnson J & Dawson-Hughes B (1991): Precision and stability of Dual- Energy X-ray Absorptiometry Measurements. Calc. Tiss. Inter. 49, 174±178. Kramer GH, Burns L & Noel L (1991): The BRMD BOMAB phantom family. Health. Phys. 61, 895±902. Kelly TL, Richardson TL, Berger N, Steiger P & Stein, JA (1991): Assessment of body composition by dual x-ray absorptiometry using a simultaneously scanned calibration standard. Med. Sci. Sports. Exerc. 23, Suppl. (4), S90 (abstr.). Kelly TL, Slovik DM, Shoenfeld DA & Neer RM (1980): Calibration and standardization of bone mineral densitometers. J. Bone. Miner. Res. 4, 663±669. Laskey MA, Lyttle KD, Flaxman ME & Barber RW (1992): The in uence of tissue depth and composition on the performance of the Lunar dualenergy X-ray absorptiometer whole-body scanning mode. Eur. J. Clin. Nutr 46, 39±45. Laskey MA, Flaxman ME, Barber RW, Trafford S, Hayball MP, Lyttle KD, Crisp AJ & Compston JE (1991): Comparative performance in vitro and in vivo of lunar DPX and Hologic QDR-100 dual energy x-ray absorptiometers. Br. J. Radiol. 64, 1023±1029. LeBlanc AD, Schneider VS, Engelbretson DA & Evans HJ (1990): Precision of regional bone mineral measurements obtained from totalbody scans. J. Nucl. Med. 31, 43±45. Mazess R, Barden HS, Bisek JP & Hanson J (1990): Dual-energy x-ray absorptiometry for total-body and regional bone-mineral and soft-tissue composition. Am. J. Clin. Nutr. 51, 1106±1112. Mazess R, Collick B, Trempe J, Barden H & Hanson J (1989): Performance evaluation of a dual-energy X-ray Bone Densitometer. Calc. Tiss. Inter. 44, 228±232. Nelson ME, Fiatarone MA, Layne JE, Trice I, Economos CD, Fielding RA, Ma R, Pierson, RN & Evans WJ (1996): Analysis of bodycomposition techniques and models for detecting change in soft tissue with strength training. Am. J. Clin. Nutr. 63, 678±686. Nilas L, Hassager C & Christiansen C (1988): Long term precision of dual photon absorptiometry in the lumbar spine in clinical settings. Bone. and. Mineral. 3, 305±315. Nuti R, Righi G, Martini G, Turchetti V, Lepore C & Canaggia A (1987): Methods and clinical applications of total body absorptiometry. J. Nucl. Med. 31, 213±221. Orwoll E, Oviatt S and the Nafarelin bone study group (1991): Longitudinal precision of Dual-Energy X-ray Absorptiometry in a multicenter study. J. Bone. Miner. Res. 6, 191±197. Paton NIJ, Macalian DC, Jebb SA, Pazianes M, Grif n GE (1995): Dualenergy X-ray absorptiometry results differ between machines. Lancet. 346, 899. Peppler WW & Mazess RB (1981): Total body bone mineral and lean body mass by dual-photon absorptiometry. Theory and measurement procedure. Calc. Tiss. Inter. 33, 353±359. Pierson RN, Wang J, Heyms eld SB, Russell-Aulet M & Mazariegos M (1991): Measuring body fat: calibrating the rulers. Intermethod comparisons in 389 normal Caucasian subjects. Am. J. Physiol. 261, (Endocrinol. Metab). 24, E103±E108. Pritchard JE, Nowson CA, Strauss BJ, Carlson JS, Kaymakci B & Wark JD (1993): Evaluation of dual energy X-ray absorptiometry as a method of measurement of body fat. Eur. J. Clin. Nutr. 47, 216±228. Rencken ML, Murano R, Drinkwater BL & Chestnut III, CH (1991): In Vitro comparability of dual energy X-ray Absorptiometry (DEXA) bone densitometers. Calc. Tiss. Inte. 48, 245±248. Rico H, Revilla M, Villa L, Ruiz-Contreras D, Hernandez ER & Alverez de Buergo M (1994): The four-compartment models in body composition data from a study with dual-energy x-ray absorptiometry and nearinfrared interactance on 815 normal subjects. Metabolism. 43, 417±422. Roubenoff R, Kehayias JJ, Dawson-Hughes B & Heyms eld SB (1993): Use of dual-energy x-ray absorptiometry in body composition studies: not yet a gold standard. Am. J. Clin. Nutr. 58, 589±591. SAS (1990): SAS users guide: statistics, SAS Institute Inc.: Cary, NC Seto H, Kageyama M, Nomura K, Kakishita M & Tonami S (1991): Precision of Total-Body and Regional Bone Mineral Measurement by Dual-Energy X-ray Absorptiometry. Radiat. Med. 9, 110±113. Slosman DO, Casez J, Pichard C, Rochat T, Fery F, Rizzoli R, Bonjour JP, Morabia A & Donath A (1992): Assessment of whole-body composition with dual energy x-ray absorptiometry. Radiology. 185, 593±598. Svendsen OL, Haarbo J, Hassager C & Christiansen C (1993): Accuracy of measurements of body composition by dual-energy x-ray absorptiometry in vivo. Am. J. Clin. Nutr. 57, 605±608. Svendsen OL, Haarbo J, Heitmann BL, Gotfredsen A & Christiansen C (1991): Measurement of body fat in elderly subjects by dual-energy x- ray absorptiometry, bioelectrical impedance, and anthropometry. Am. J. Clin. Nutr. 53, 1117±1123. Trevisan C, Gandolini GG, Sibilla P, Penotti M, Caraceni M & Ortolani S (1992): Bone mass measurement by DXA: In uence of analysis procedures and interunit variation. J. Bone. Miner. Res. 7, 1373± Wahner HW, Looker A, Dunn WL, Walters LC, Hauser MF & Novak C (1994): Quality control of bone densitometry in a national healthy survey (NHANES III) using three mobile examination centers. J. Bone. Miner. Res. 9, 951±960. Wilson CR, Collier DB, Carrera GF & Jacobson DR (1990): Acronym for dual-energy x-ray absorptiometry. Radiology. 176, 875±876. Withers RT, Smith DA, Chatterton BE, Schultz CG & Gaffney RD (1992): A comparison of four methods of estimating body composition of male endurance athletes. Eur. J. Clin. Nutr. 46, 773±