Clin Invest Med 1996; 19 (4): 279-285
(Original manuscript submitted Sept. 19, 1995; received in revised form Mar. 14, 1996; accepted Apr. 25, 1996)
Copyright 1996, Canadian Medical Association
Design: Experimental use of DXA to assess human body composition before and after rapid saline infusion.
Participants: Six healthy men.
Outcome measures: Weight measurements, DXA scanning results and skinfold thicknesses taken on the first day of the experiment and on the second day, before and after rapid saline infusion.
Results: After the infusion, the subjects' weight increased by a mean 2.26 kg (standard deviation 0.199 kg). At each of the four readings, there was a strong correlation between weight and DXA-derived total mass (r = 0.999) and between skinfold-derived fat-free mass and DXA-derived lean mass (r = 0.941 to 0.957). Following infusion, no differences were found between the measured and theoretical (i.e., preinfusion value plus weight change) values for total mass (p = 0.22), lean soft-tissue mass (p = 0.10) and lean mass (p = 0.09). The bias was -0.669 (95% confidence interval [CI] 0.18 to -1.49) for total mass, -0.65 (95% CI 0.16 to -1.47) for lean soft-tissue mass, and
-0.14 (95% CI 0.11 to -0.38) for lean mass.
Conclusions: DXA is an improvement over previous dual-energy technologies and appears to provide sufficient accuracy to detect small (less than 2.5 kg) changes in mass in individual, healthy men, over a short period and under non-steady-state conditions. Therefore, DXA may also be of practical use for longitudinal assessment of weight change.
[Table of Contents]
Résumé
Objectif : Évaluer la capacité de l'absorptiométrie biénergique à rayons-x (ABX) à mesurer avec précision de faibles changements dans une masse de tissu mou maigre. On a suggéré récemment que l'ABX était un moyen exact et non effractif d'analyser la composition du corps.
Conception : Utilisation expérimentale de l'ABX pour évaluer la composition du corps humain avant et après une infusion rapide de solution physiologique.
Participants : Six hommes en bonne santé.
Mesures des résultats : Pesées, résultats de scanographie ABX et épaisseur du pli cutané le premier jour de l'expérience et le lendemain, avant et après une infusion rapide de solution physiologique.
Résultats : Après l'infusion, le poids des sujets a augmenté en moyenne de 2,26 kg (écart type, 0,199 kg). À chacune des quatre lectures, on a établi un lien solide entre le poids et la masse totale dérivée par ABX (r = 0,999) et entre la masse sans gras dérivée du pli cutané et la masse maigre dérivée de l'ABX (r = 0,941 à 0,957). Après l'infusion, on n'a constaté aucune différence entre les valeurs mesurées et les valeurs théoriques (c.-à-d. valeur préalable à l'infusion plus changement du poids) dans le cas de la masse totale (p = 0,22), dans celui de la masse de tissu mou maigre (p = 0,10) et dans celui de la masse maigre (p = 0,09). La déviation s'est établie à -0,669 (intervalle de confiance [IC] à 95 %, 0,18 à -1,49) dans le cas de la masse totale, à -0,65 (IC à 95 %, 0,16 à -1,47) dans celui de la masse de tissu mou maigre et à -0,14 (IC à 95 %, 0,11 à -0,38) dans celui de la masse maigre.
Conclusions : L'ABX est une amélioration par rapport aux techniques biénergétiques antérieures et semble assurer une exactitude suffisante pour repérer de faibles fluctuations (moins de 2,5 kg) de la masse chez des hommes en bonne santé pendant une courte période et dans des conditions non stables. L'ABX pourrait donc avoir une certaine utilité pratique dans l'évaluation longitudinale des changements de poids.
Most recently, dual-energy absorptiometry has been proposed as a safe, convenient, noninvasive method of body-composition analysis.[1-3] This technology measures the relative attenuation of two distinct energy levels of radiation as they pass through body tissues. Dual-energy absorptiometry has the distinct advantage over other techniques of not requiring any assumptions about water content or density of lean-tissue mass, although it does assume constant attenuations for lean soft and fat masses. The lack of assumptions about density is particularly important when subjects other than young, healthy adults are investigated.[4,5] The original instruments for dual-energy absorptiometry used a radionuclide source of photons at two discrete energy levels (dual-photon absorptiometry or DPA). However, we have previously demonstrated that DPA does not accurately detect small changes in fat-free mass induced by transpiration or saline infusion.[6]
The Hologic alternating scanner, a newer instrument for dual-energy absorptiometry, employs an x-ray generator that produces an x-ray source of radiation at alternating energies of 70 µ 140 keV. Dual energy x-ray absorptiometry (DXA) produces a low amount of radiation; it has a steady radiation source, rapid scan times and a small pixel size. These advantages may make it an improvement over DPA.
Initial studies have suggested that DXA is better able to to distinguish fat mass from lean soft-tissue mass than DPA.[7] In an analysis of pig carcass, DXA accurately measured fat mass and lean soft-tissue mass.[8] Furthermore, it also accurately measured the changes in fat mass and lean soft-tissue mass induced by placing 8.8 kg of porcine lard on the trunks of human subjects, although it also showed an unexplained change in bone mineral mass. Going and associates[9] found that, although DXA accurately measured small changes (approximately 1.5 kg) in total body mass induced by fluid restriction, followed by oral rehydration with water, it had a limited capacity to resolve these small changes into changes in lean soft-tissue mass. In addition, a study of children and young adults found that DXA measurement of total mass underestimated scale weight.[10]
The infusion of normal saline solution is another way of rapidly inducing changes in lean soft-tissue mass. This method should not affect the ability of DXA scanning to compartmentalize the soft mass into its lean and fat components. The method of our study is similar to that of Going and associates;[9] however, unlike the oral-hydration protocol used by these authors, the saline-infusion protocol used in our study has been shown to increase both extracellular fluid and the ratio of extracellular to intracellular fluid.[11] In addition, with saline infusion, all of the extra fluid load is distributed to the lean soft-tissue mass, rather than being partially retained in the gastrointestinal tract, as may occur with oral hydration. The results may also be affected by differences in precision between DXA scanners.[12]
We conducted this study to evaluate whether small changes in lean soft-tissue mass induced by intravenous infusion of normal saline solution could be detected by the DXA scanner currently employed at our facility. If DXA scanning is accurate, then rapid saline infusion should increase only the lean soft-
tissue mass and masses that include this compartment.
[Table of Contents]
Methods
We studied six healthy male subjects 22 to 40 years of age. After signing informed consent, each subject was asked to void completely and to wear light clothing. The subject was then weighed on an electronic balance (Ancaster Scale Co., Brantford, Ont.) that is accurate to 0.1 kg. The subject was then asked to refrain from voiding for the remainder of the study. We also recorded each subject's height without shoes, accurate to 0.1 cm, on a stadiometer (Holtain Ltd., UK).
Each subject then underwent skinfold-thickness measurements at four sites for estimation of fat-free mass, according to the method developed by Durin and Womersley.[13] Following this, the subject underwent full-body DXA scanning. One hour later, the subject was reweighed and rescanned to determine the reproducibility of DXA scanning. The total fat-free mass was estimated the second time from the second weight determination and the previous skinfold thicknesses.
One to 2 days later, the subjects returned to the laboratory. They were again asked to void completely, were weighed and underwent skinfold-thickness measurements and DXA scanning. They were then asked to refrain from voiding for the rest of the study. Each subject was given an infusion over 1 hour with approximately 30 mL/kg of normal saline solution through an antecubital intravenous line. Immediately after the infusion, the subject was reweighed and rescanned. Postinfusion fat-free mass was determined from the subject's weight after the infusion and the skinfold thickness measured before the infusion.
DXA measurements were made with a QDR-2000 whole-body scanner (Hologic Inc., Waltham, Mass.), with the subject supine. The device produces a stable x-ray source of radiation at alternating energies of 70 µ 140 keV. Scans proceed diagonally from head to toe and take less than 7 min per scan. Each scan comprises approximately 854 000 pixels, each with an area of 1.46 mm[2]. For all scans, a calibration phantom is also scanned, resulting in a constant calibration. All scans were performed by the same technician (J.M.H.).
As previously described,[14] dual-energy absorptiometry considers the body to be composed of two compartments: a bone mineral mass, or bone tissue stripped of its organic and fluid components, and a soft-tissue mass, composed largely of water. When each pixel is scanned, the relative attenuation of the two energy signals is calculated. The attenuation ratio of pixels containing only soft tissue (about half of all pixels) is inversely related to the fat content of the pixels; pixels containing bone have a much higher ratio than those made up solely of lean tissue. A weighted mean attenuation ratio is calculated from the pixels without bone and then applied to the bone-containing pixels to calculate bone mass. After the bone mass of each pixel is calculated, reanalysis of the soft tissue and the attenuation ratio allows for compartmentalization of the soft mass into its lean and fat components. The values for each pixel are summed to derive bone, fat, lean soft-tissue and total mass, which is the sum of the previous masses.
The reproducibility of DXA-derived measurements of bone mass, fat mass, lean soft-tissue mass, lean mass (lean soft-tissue mass and bone mass) and total mass was assessed by Student's t-test, coefficients of variation and repeatability coefficients.[15] Values for weight, skinfold-derived fat-free mass and DXA-derived masses were compared over time by analysis of variance (ANOVA) for repeated measures. A p value of less than 0.05 was considered significant. When ANOVA suggested that there were significant differences, post-hoc Scheffé testing was carried out. For each of the four measurement times, weight and skinfold-derived fat-free mass were compared with DXA-derived total mass and lean mass, respectively, by regression analysis. The postinfusion measured values for total, lean soft-tissue and lean mass were compared, by Student's t-test, with the respective calculated theoretical values (i.e., preinfusion value plus weight change). Given our small sample size, the means of each pair of values (theoretical and measured) were plotted against the difference between each pair of values as another method of determining the sensitivity of the DXA scanner. The agreement between the theoretical and measured values for total, lean soft-tissue and lean mass were summarized by calculating the bias (the mean difference) and a 95% confidence interval.[15]
[Table of Contents]
Results
The subjects were 31.3 years of age on average (standard deviation [SD] 7.55 years) and had a mean height of 176.3 cm (SD 3.44 cm) and mean weight of 73.0 kg (SD 7.03 kg).
The DXA measurements taken on the first day were highly reproducible for bone and total mass but were less so for soft mass divided into fat and lean compartments (Table 1). Given that the SD of the differences in lean soft-tissue mass was so small (0.278 kg) in relation to the minimum increase in weight induced by saline infusion (1.91 kg), we are confident that detection of this minimal increase in weight is within the resolution of the DXA scanner.
The saline infusion resulted in a mean weight change of 2.26 kg (SD 0.199 kg, range 1.91 to 2.51 kg). Weight, DXA-derived lean soft-tissue, lean and total mass changed significantly only after the infusion (Table 2).
Weight was correlated with DXA-derived total mass at each of the four readings and for each time (r = 0.999). Except at the second measurement, the intercepts of the regression analysis did not differ from 0 and the slopes did not differ from 1. At the second measurement, the intercept was *0.46 kg (SD 0.22 kg) and the slope was 1.007 (SD 0.002) (Fig. 1). The standard errors for the four readings were 0.096, 0.193, 0.201 and 0.111 kg.
Skinfold thickness-derived fat-free mass was correlated with DXA-derived lean mass at each of the four readings (r = 0.941 to 0.957). Intercepts did not differ from 0 and slopes did not differ from 1. The standard errors for the four readings were 1.813, 1.913, 2.179 and 2.022 kg.
Paired t-tests revealed no differences between the measured and theoretical values for total, lean soft-tissue or lean mass (Table 3). The individual data values were also examined by plotting the means of each pair of values (theoretical and measured) against the difference between each pair of values. Fig. 2 is an example of such a plot, performed for the lean mass values. The bias for total mass was -0.669 (95% confidence interval [CI] 0.18 to -1.49), for lean soft-tissue mass was -0.65 (95% CI 0.16 to -1.47) and for lean mass was -0.14 (95% CI 0.11 to -0.38).
[Table of Contents]
Discussion
We have demonstrated that DXA scanning with the use of the Hologic QDR-2000 accurately measures small (2.0 to 2.5 kg) changes in lean soft-tissue mass induced by normal saline infusion, in healthy men. Measurements made at any particular time were strongly correlated with measured weights and derived values for fat-free mass. No differences were found between the postinfusion measured values and the theoretical values for total, lean soft-tissue or lean mass. DXA scanning was also highly reproducible with respect to total mass and bone mass. These results suggest that DXA scanning is a practical method for assessing small changes in body composition, occurring under non-steady-state conditions, in healthy men.
Although, as we have just noted, measurements made with the Hologic DXA scanner used in this study were highly reproducible for bone and total mass, they were less so when the soft mass was divided into its lean and fat components. These findings are consistent with the within-day precision results of Svendsen and associates,[8] who made double measurements of the body composition of pigs and adult humans with a Lunar total-body DXA scanner. Their coefficients of variation for bone mass were 0.9% for humans and 2.2% for pigs, which are slightly higher than, but comparable to, our results. They found, as did we, that the coefficients of variation for lean mass (1.5 for humans and 0.6 for pigs) and fat mass (4.6 for humans and 2.5 for pigs) are progressively higher than those for bone mass.
Our results are also consistent with those of Going and associates,[9] who evaluated the ability of DXA scanning to detect changes in soft-tissue mass induced by transpiration and oral rehydration with water. They found that estimates of bone mass were unaffected by changes in hydration. In addition, their study also demonstrated that excellent agreement between scale body weight and DXA total mass and correct measurement of small changes in total mass with DXA. As in our study, Going and associates found reasonable agreement between mean changes in weight and changes in DXA-derived lean soft-tissue mass. However, they found that DXA had some difficulty resolving small changes in soft-
tissue mass into changes in lean soft-tissue mass in individual subjects. Our results agree with these findings despite the fact that, compared with normal saline infusion, the protocol used in the study by Going and associates should not have affected the ratio of extracellular to intracellular fluid.
Although, as outlined above, the major findings of our study agree with those of Going and associates[9] and Svendsen and associates,[8] there are minor differences between these studies and ours. These differences may be explained by the findings of Pritchard and associates,[12] who demonstrated differences in body-composition analysis of the same subject by two densitometers, the Hologic QDR 1000W (QDR) and the Lunar DPX densitometer (DPX). The DPX was used for both of the previously mentioned studies, whereas we used the Hologic QDR 2000 (the newer model of the QDR). Pritchard and associates found a 3.1% systematic difference between the QDR and DPX in measurements of fat, with the QDR measurements 2% higher and those of the DPX 5% higher than underwater weighing. This variation was reflected in precision levels as well: the coefficient of variation was 1.2% for the QDR, whereas it was 3% to 4% for the DPX. These investigators concluded that it is important to specify the make of DXA instrument used for soft-tissue analysis in any given study.
Our results support the reproducibility of total, lean soft-tissue and lean mass measurements by DXA, reported in previously published articles.[8,9] DXA scanning is a significant improvement over DPA with respect to speed, accuracy and radiation exposure. It also provides sufficient accuracy to track the changes in the soft-tissue compartments induced by rapid saline infusion. Although it remains to be tested, DXA's ability to track slower, naturally occurring changes over longer periods (days or months) appears promising.
[Table of Contents]
Acknowledgements
Dr. L.C. Lands is a Chercheur-clinicien of the Fonds de la recherche en santé du Québec.
[Table of Contents]
References