This work is a publication of the USDA/ARS Children's Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine, and Texas Children's Hospital, Houston, TX, U.S.A. The contents of this publication do not necessarily reflect the views or policies of the USDA, nor does mention of trade names, commercial products, or organizations imply endorsement.
Bone Mineral and Body Composition Measurements: Cross-Calibration of Pencil-Beam and Fan-Beam Dual-Energy X-Ray Absorptiometers†
Article first published online: 27 OCT 2009
Copyright © 1998 ASBMR
Journal of Bone and Mineral Research
Volume 13, Issue 10, pages 1613–1618, October 1998
How to Cite
Ellis, K. J. and Shypailo, R. J. (1998), Bone Mineral and Body Composition Measurements: Cross-Calibration of Pencil-Beam and Fan-Beam Dual-Energy X-Ray Absorptiometers. J Bone Miner Res, 13: 1613–1618. doi: 10.1359/jbmr.19126.96.36.1993
- Issue published online: 27 OCT 2009
- Article first published online: 27 OCT 2009
- Manuscript Accepted: 10 JUN 1998
- Manuscript Revised: 19 MAY 1998
- Manuscript Received: 4 FEB 1998
Pencil-beam dual-energy X-ray absorptiometers (DXA) are being replaced with instruments that rely solely on fan-beam technology. However, information has been lacking regarding the translation of bone mineral and body composition data between the two devices. We have compared total body scans using pencil-beam (Hologic QDR-2000W) and fan-beam (Hologic QDR-4500A) instruments for 33 children (ages 3–18 years) and 14 adults. Bone mineral content (BMC), bone mineral density (BMD), fat, lean, and body fatness (%fat) values were highly correlated (r2 = 0.984–0.998) between the two DXA instruments. The mean differences between the paired measurements were: ΔBMC = 7.5 ± 73.6 g, ΔBMD = 0.0074 ± 0.0252 g/cm2, Δlean = 1.05 ± 1.8 kg, Δfat = −0.77 ± 1.7 kg, and Δ%fat = −0.94% ± 2.5%. The BMC and BMD values were not statistically different, whereas the differences for the body composition values were significant (p < 0.02–0.005). Regression equations are provided for conversion of bone and body composition data between pencil-beam and fan-beam values for the whole body. To test the performance of these equations for a second group (23 subjects), predicted values were compared with the measured data obtained using the fan-beam instrument. The mean differences were −1.0% to 1.4%, except for body fat mass, where the difference was 6.4%. For cross-sectional studies, the two DXA technologies can be considered equivalent after using the translational equations provided. For longitudinal studies in which small changes in body composition for the individual are to be detected, we recommend that the same DXA instrument be used whenever possible. For example, transition from a pencil-beam to a fan-beam instrument could, in extreme cases, result in differences as large as 19% for the estimate of body fat mass.
DUAL-ENERGY X-RAY ABSORPTIOMETRY (DXA) is a technique that is widely used for the assessment of skeletal mass.1–4 Interest has been increasing in the use of DXA for the measurement of soft tissue composition.5,6 In response to these demands, the DXA instruments have undergone continuing technological development. With each upgrade, issues arise in regard to compatibility with older-generation instruments and reference values established for the older DXA instruments.
Previous studies have provided comparisons of pencil-beam DXA instruments by different manufacturers,7–9 and comparisons of different generations of DXA instruments by the same manufacturer.10–12 In most cases, these comparisons have involved the assessment of bone mineral density (BMD), usually of the spine and femur. Comparisons of soft tissue measurements usually have not been performed. Furthermore, the age range of the subjects examined in these studies has been limited to adults and therefore indirectly limited in regard to body size. We wanted to examine a full range of body sizes representative not only of adults but also of children.
The purpose of this paper was to determine whether direct comparisons can be made between whole-body data for bone mineral and body composition measurements obtained with the older-generation pencil-beam and the newer-generation fan-beam DXA instruments. In particular, we wanted to determine whether the two different instruments would provide the same values for children, affording direct interchange of data between the instruments. Also, we wanted to know whether adjustments will be required for reference values for children originally obtained using the older-generation pencil-beam measurements.
MATERIALS AND METHODS
For the cross-comparison of the two DXA instruments, total-body measurements were obtained for 47 subjects (21 males, 26 females), including 33 children (age range 3–18 years). For each subject, the pencil-beam measurement was obtained first, immediately followed by the fan-beam measurement. A second group of 23 subjects (ages 5–46 years) was later examined to test the validity of the cross-calibration equations developed for the initial group. Body weight was determined using a calibrated electronic scale in the laboratory, and weights were recorded to the nearest 0.1 kg. This protocol was approved by the Institutional Review Board for human studies, and informed consent was obtained for each subject.
The pencil-beam measurements were obtained using a Hologic QDR-2000 W instrument and analyzed with “Enhanced Whole Body V5.71” software. The subjects were scanned across the body's width, starting at the head. A whole-body scan with the pencil-beam instrument took about 12 minutes for children and 15 minutes for adults. Based on duplicate measurements over a 10-day period in nine subjects (weight range 40–105 kg), the mean precision for the total body values were: 0.9% (BMD), 1.2% (bone mineral content [BMC]), 2.5% (lean), and 3.9% (fat).
The fan-beam measurements were obtained using a Hologic QDR-4500A absorptiometer (Hologic, Inc., Waltham, MA, U.S.A.), with the results analyzed using Whole Body Fan Beam V8.21a:3 software. The body was scanned from the head to the feet in three sections, starting with the right side of the body. A whole-body measurement typically took about 3 minutes for all subjects. For each subject's set of scans, the bone and body composition analyses, including the setting of boundaries for the various body regions, were performed by a single operator. Based on duplicate total body measurements with the fan-beam instrument in seven subjects (weight range 50–120 kg), the precision for the respective body composition values were: 0.8% (BMD), 1.1% (BMC), 2.3% (lean), and 3.3% (fat).
Tabulated data are reported as the mean ± SD. To determine the degree of interchangeability of values between the two DXA instruments, the method proposed by Bland and Altman was used.13 For this analysis, the differences between the two DXA measurements were calculated and examined as a function of the average value for the two measurements. Translational equations for bone and body composition values between the two DXA instruments were based on least-squares linear regression analyses; the regression equation, correlation coefficient (r), and standard error of the estimate (SEE) are provided. The SEE is mathematically the same as the root mean square error. To test the validity of these translational equations, the percentage difference (defined as 100 × [measured − predicted]/measured) for the retest group was used. If the mean value for the percentage difference exceeded the precision of the measure, the equation used for the predicted value would be considered inadequate for translation between the two DXA measurements. All statistical procedures were performed using the statistical package MINITAB (Minitab, Inc., State College, PA, U.S.A.), and test results with a p value <0.05 were considered significant.
The mean, SD, minimum, and maximum values for the BMD, BMC, fat, lean, and percentage fat (%fat) for the pencil-beam and fan-beam DXA measurements are given in Table 1. In addition, the DXA-based values for body weight (WtDXA) are also provided. It is evident that for this group of subjects, there was little difference in the mean values for the bone or body composition measurements. The range of values for the BMD, BMC, fat, and %fat measurements, as represented by the SDs, was slightly lower for the fan-beam instrument than for the pencil-beam instrument. There was better agreement with values for body weight based on the fan-beam instrument than for those based on the pencil-beam instrument.
The mean differences, defined as the fan-beam value minus the pencil-beam value, are presented in Table 2 for the bone and body composition values. Again, it is clearly evident that the BMD and BMC values, on average, were very similar between the two DXA instruments. For the lean compartment, the mean fan-beam estimate averaged about 1 kg higher (p < 0.005) than the mean estimate for the pencil-beam measurements. For the fat and %fat values, the mean estimates for the fan-beam measurements were lower by 0.77 kg (p < 0.005) and 0.94% (p < 0.02), respectively. For body weight, the mean difference between the fan-beam estimate and scale weight was small (−0.26 ± 0.59 kg) yet statistically significant (p < 0.005).
Figure 1 shows the relationships between the differences in the bone and body composition values for the fan-beam and pencil-beam measurements as a function of the average values for the two DXA measurements. The regression equations, correlation coefficient (r), and SEE for each of these graphs are provided in Table 2. The differences for the BMD, BMC, lean, fat, and %fat values were significantly dependent on the average value for that compartment. For the bone measurements, fan-beam data are higher for smaller (usually younger) subjects and lower for the larger (usually older) subjects, relative to the pencil-beam values. The crossover in values occurred at about midrange for BMD (1.023 g/cm2) and BMC (1659 g). For the lean and fat compartments, the crossovers did not occur at the midrange but at the lower range of values: 18.5 kg for lean mass and 8.5 kg for fat. For %fat, the crossover in values between the pencil-beam and fan-beam measurements occurred at 22.5%. That is, the subjects originally identified as lean (low %fat) remained lean with the fan-beam measurements, but each subject's %fat was increased by a few percent. Alternately, the %fat values of the fatter subjects decreased by 4–7% for the fan-beam measurement. The differences between the DXA-derived weights (sum of BMC, lean, and fat) and the scale weights were independent of the scale weights. We also examined the effects of body size (weight, height, and body mass index [BMI]) on the relationships seen in Fig. 1. The residuals for the regression analysis for the difference versus average relationships for BMD, fat, and %fat were independent of any of the body size parameters, whereas BMC and lean were influenced by BMI (p < 0.03).
For the translation of values between the two DXA instruments, a set of equations is provided in Table 3. The values for r2 are all nearly 1.0, indicating that the individual values for the pencil-beam and fan-beam measurements are highly correlated. This, however, does not necessarily ensure that they are directly interchangeable. If the two sets of measurements were identical (within the measurement errors associated with each instrument), then the differences shown in Fig. 1 would be independent of the average values for each compartment. Except for body weight, this was not the case. Thus, we examined the relationship between the residual (defined as the observed − predicted) for the difference values in relationship to the subject's body size (weight, height, and BMI). The variations in the BMD, fat, and %fat values were independent of weight, height, and BMI, whereas those for BMC and lean were influenced by BMI (p < 0.03).
To test the accuracy of the translational equations (provided in Table 3), pencil-beam and fan-beam data were obtained for a second group of subjects (cross-validation group). The mean values for the fan-beam measurements are presented in Table 4. The values for the pencil-beam measurements were substituted into the translational equations in order to predict the corresponding values for the fan-beam instrument. The mean, SD, and range of values for the percentage difference are included in Table 4. The mean percentage differences were 0.1% for BMD and −1.0% for BMC, respectively. These differences were within the precision of the DXA measurements for BMD (<1%) and BMC (<1.5%). For the soft-tissue compartments, the mean percentage difference values were 1.4% for the lean mass, 6.4% for fat mass, and −0.7% for %fat. The mean percentage differences for the lean mass and %fat values were also within the DXA measurement precision for these compartments, whereas that for the fat mass was somewhat higher.
Many investigators will be faced, at some point, with the data concerns raised when replacing or upgrading DXA instruments. These concerns may relate to longitudinal studies, continued recruitment of subjects in ongoing protocols, and the usefulness of reference ranges based on the older-generation DXA instruments. For the data from the newer instruments to be highly correlated with the data from the older instruments it is necessary, but not sufficient, for comparison between instruments. Although cross-calibrations based on phantoms and in vitro studies are helpful, these types of comparisons may not be representative of in vivo changes.8,9 It has been our experience that although skeletal values can appear to be consistent across generations of DXA instruments, this does not ensure that the body composition values for the lean and fat compartments are also correct. For example, in the present study, it took us some time to convince the manufacturer that the initial values for our fan-beam instrument was not accurate before they determined that it was a hardware misalignment. This problem was not detectable with the instrument's standard quality-control programs. Without an independent measure of the lean tissue mass, for example, these subtle differences might have been missed.
Some investigators have provided comparisons of different manufacturers' pencil-beam instruments, while others have reported on the differences between the pencil-beam and fan-beam scan modes within the same DXA device.8,9,14–17 These comparisons have mainly been for regional BMD measurements of the spine and femur. Faulkner et al.14 compared the Hologic QDR-1000 and QDR-2000 measurements of the BMD of the spine and femur of older adult women, finding a statistical, but clinically insignificant, difference of about 0.8%. Abrahamsen et al.15 found similar differences and attributed them to small calibration drifts between the two instruments. Several investigators have reported a difference for bone measurements between the pencil-beam and fan-beam mode using the same DXA instrument.16,17 Abrahamsen et al.,15 for example, observed higher BMD values for the femur using the fan-beam mode, whereas the opposite was observed for the spine and whole body. For the BMC measurements, the fan-beam mode gave substantially lower values for the spine, femoral neck, and whole body. Estimates for total body fat mass were also reported to be highly dependent on the choice of scan modes. These investigators concluded that using only the spine phantom was not adequate for cross-calibration between the pencil-beam and fan-beam options. Tothill et al.,8,9 using whole-body phantoms, determined that interchangeability of results among various manufacturers' pencil-beam instruments was not always possible. More recently, these investigators18 have reported measurement anomalies for BMC and BMD associated with weight change and have questioned the accuracy of bone measurements with current pencil-beam DXA technology.
The mean and SD values for the total body BMC and BMD parameters, as given in Table 1, indicate that, on a group basis, there is adequate interchangeability between the two DXA technologies. That is, on average, either instrument should equally identify conditions of osteopenia. For clinical conditions, such as sarcopenia, malnutrition, or obesity, the identification of some individuals may be instrument dependent. Furthermore, the nonrandom pattern (see Fig. 1) for the difference between pencil-beam and fan-beam measurements suggests that when monitoring changes within an individual, the data for one of the two instruments will need to be adjusted. On a population basis, the mean adjustment for BMD would be 0.0074 g/cm2, while that for BMC would be 7.5 g. Although these adjustments are well within the precision errors (see Materials and Methods) for these parameters, they may not be adequate for the individual. For the bone mineral data, it is evident that for the smaller and/or younger subjects (generally, those with the lower values), the fan-beam instrument gives higher estimates than those obtained with the pencil-beam instrument, whereas for larger and/or older subjects (higher values) the reverse situation occurs.
Adjustments based on the mean difference in the present study would be 1.05 kg for the lean mass and −0.77 kg for body fat mass, respectively. From Fig. 1, it is evident that for most of the subjects in this study, the fan-beam measurement gave higher lean values and lower fat values, when compared with the pencil-beam values. The reason for these differences is unknown. One explanation may be that the manufacturer has changed the reference values or cutoff points for the attenuation coefficients used to define the relative lean and fat fractions of the total soft tissue mass. Alternately, these differences may reflect variations in body thickness which may produce magnification errors.19,20 The Bland-Altman analysis does not a priori assume that one technique is more accurate than the other but instead provides a measure of the interchangeability between the two methods.
In conclusion, we recommend that in longitudinal studies it is advisable to perform all measurements for a given subject on one instrument whenever possible. For longitudinal studies where significant changes (>2× precision error) for the individual could be expected, use of the translational equations will significantly reduce biases introduced by changing instruments. At the time of an upgrade from pencil-beam to fan-beam technology for total body measurements, we recommend that translational equations be used, such as those reported in this study. Furthermore, if the fan-beam values for a subject are being compared with a reference range, then the reference values, if based on pencil-beam data, will need to be adjusted accordingly. This can be especially critical for the measurement of changes in bone and body composition during growth of children. Further analyses will be needed to determine if any of the differences observed in this study can be attributed to magnification effects associated with the fan-beam technology.
We thank T. Kelly at Hologic, Inc. for his interest in this research and for information related to the fan-beam instrument, and L. Loddeke for her editorial assistance with preparation of the manuscript. This work is supported by the U.S. Department of Agriculture, Agricultural Research Service under Cooperative Agreement #58-6250-6-001 with Baylor College of Medicine.
- 11990 Dual energy X-ray absorptiometry for total body and regional bone mineral and soft tissue composition Am J Clin Nutr 51:1106–1112., , ,
- 21989 Dual-photon absorptiometry: Comparison of bone mineral and soft tissue measurements in vivo with established methods Am J Clin Nutr 49:1283–1289., , , ,
- 31995 Bone mineral and body fat measurements by two absorptiometry systems: Comparisons with neutron activation analysis Calcif Tissue Int 56:93–98., , , , , ,
- 41996 Total body calcium and bone mineral content: Comparison of dual-energy X-ray absorptiometry with neutron activation analysis J Bone Miner Res 11:843–848., , , ,
- 51996 Dual-energy X-ray absorptiometry. In: RocheAF, HeymsfieldSB, LohmanT (eds.) Human Body Composition. Human Kinetics, Champaign, IL, U.S.A., pp. 63–78.
- 61993 Assessment of body composition with use of dual-energy X-ray absorptiometry: Evaluation and comparison with other methods Mayo Clin Proc 68:867–873., , , , , ,
- 71996 Comparison of body composition and bone mineral measurements from two DXA instruments in young men Am J Clin Nutr 64:669–676., , , , , ,
- 81994 Precision and accuracy of measurements of whole-body bone mineral: Comparisons between Hologic, Lunar, and Norland dual-energy X-ray absorptiometers Br J Radiol 67:1210–1217., ,
- 91994 Comparisons between Hologic, Lunar, and Norland dual-energy X-ray absorptiometers and other techniques used for whole-body soft tissue measurements Eur J Clin Nutr 48:781–794., , ,
- 101996 Dual energy X-ray absorptiometry: Inter-machine variability Int J Obes 20:1048–1050., ,
- 111995 Dual-energy X-ray absorptiometry results differ between machines Lancet 346:899–900., , , ,
- 121997 A preliminary evaluation of the Lunar Expert-XL for bone densitometry and vertebral morphometry J Bone Miner Res 12:136–143., , , , , , , ,
- 131986 Statistical methods for assessing agreement between two methods of clinical measurement Lancet 1:307–310.,
- 141993 Cross-calibration of DXA equipment: Upgrading from a Hologic QDR-1000W to a QDR-200 Calcif Tissue Int 52:79–84., , ,
- 151995 Cross calibration of QDR-2000 and QDR-1000 dual-energy X-ray densitometers for bone mineral and soft-tissue measurements Bone 16:385–390., , ,
- 161997 body composition by DEXA in older adults: Accuracy and influence of scan mode Med Sci Sports Exerc 29:560–567., , , , , ,
- 171993 Dual X-ray absorptiometry: a comparison between fan beam and pencil beam scans Br J Radiol 66:902–906., , ,
- 181997 Anomalies in the measurement of changes in total-body bone mineral by dual-energy X-ray absorptiometry during weight change J Bone Miner Res 12:1908–1921., , ,
- 191997 Correcting the magnification error of fan beam densitometers J Bone Miner Res 12:119–123., ,
- 201997 Comparison of two Hologic DXA systems (QDR 1000 and QDR 4500A) Br J Radiol 70:728–739., , ,