Disclosures: MPR, YX WKW, and DLE are employees of General Electric Healthcare Lunar, or its affiliates. FPM and SR are employees of Nestle Research Center.
Precision of a new tool to measure visceral adipose tissue (VAT) using dual-energy X-Ray absorptiometry (DXA)†
Version of Record online: 12 MAR 2013
Copyright © 2012 The Obesity Society
Volume 21, Issue 1, pages E134–E136, January 2013
How to Cite
Rothney, M. P., Xia, Y., Wacker, W. K., Martin, F.-P., Beaumont, M., Rezzi, S., Giusti, V. and Ergun, D. L. (2013), Precision of a new tool to measure visceral adipose tissue (VAT) using dual-energy X-Ray absorptiometry (DXA). Obesity, 21: E134–E136. doi: 10.1002/oby.20140
- Issue online: 12 MAR 2013
- Version of Record online: 12 MAR 2013
- Accepted manuscript online: 6 NOV 2012 08:14AM EST
- Manuscript Accepted: 17 OCT 2012
- Manuscript Received: 10 FEB 2012
A new tool to quantify visceral adipose tissue (VAT) over the android region of a total body dual-energy x-ray absorptiometry (DXA) scan has recently been reported. The measurement, CoreScan, is currently available on Lunar iDXA densitometers. The purpose of the study was to determine the precision of the CoreScan VAT measurement, which is critical for understanding the utility of this measure in longitudinal trials.
Design and Methods:
VAT precision was characterized in both an anthropomorphic imaging phantom (measured on 10 Lunar iDXA systems) and a clinical population consisting of obese women (n = 32).
The intrascanner precision for the VAT phantom across 9 quantities of VAT mass (0–1,800 g) ranged from 28.4 to 38.0 g. The interscanner precision ranged from 24.7 to 38.4 g. There was no statistical dependence on the quantity of VAT for either the inter- or intrascanner precision result (p = 0.670). Combining inter- and intrascanner precision yielded a total phantom precision estimate of 47.6 g for VAT mass, which corresponds to a 4.8% coefficient of variance (CV) for a 1 kg VAT mass. Our clinical population, who completed replicate total body scans with repositioning between scans, showed a precision of 56.8 g on an average VAT mass of 1110.4 g. This corresponds to a 5.1% CV. Hence, the in vivo precision result was similar to the phantom precision result.
The study suggests that CoreScan has a relatively low precision error in both phantoms and obese women and therefore may be a useful addition to clinical trials where interventions are targeted towards changes in visceral adiposity.
The precision of a measurement tool is a critical consideration for inclusion in longitudinal and interventional studies (1). Precision has important implications on study design, such as sample size determination and optimization of monitoring intervals between measurements. Recently, a new tool to quantify visceral adipose tissue (VAT) mass and volume with dual-energy X-ray absorptiometry (DXA) has been described (2). This tool, which has been commercialized under the brand CoreScan (GE Healthcare, Madison, WI), was reported to have a high correlation with computed tomography (CT). The CoreScan algorithm computes the VAT within the android region of a total body DXA scan. This region is generally around 10 cm in height and extends from the iliac crest towards the chin. The algorithm was shown to provide accurate results on adult men and women with BMI 18.5-40 kg m−2. The initial description of the method, however, did not include information about repeat measurement precision. The purpose of this study was to demonstrate the precision of the CoreScan VAT measurement tool in phantoms and a convenience sample of obese women.
All data acquired as part of this study was acquired with Lunar iDXA systems (GE Healthcare, Madison, WI). The Lunar iDXA is a high weight limit (204 kg), narrow angle fan beam DXA system and is the system which currently has been cleared by the FDA for the measurement of VAT. All Lunar iDXA systems used in this study were used according to the manufacturer's instructions.
A custom phantom was designed to enable DXA measurements of specific masses of fat contained within an anthropomorphic abdominal phantom (Figure 1). The phantom is comprised of two materials: a simulated lean tissue based on a Computerized Imaging Reference Systems (CIRS, Norfolk, VA) standard polymer, which measures ∼−3% fat on DXA (equivalent to salinated water) and a simulated adipose tissue, which measures ∼75% fat on DXA. The physical construction of the phantom consists of an outer shell of adipose tissue, a fixed intra-abdominal region containing simulated lean tissue, and four holes that can accept removable rods. By inserting either lean or adipose rods into the holes, nine VAT masses, ranging from 0 to 1,800 g, can be produced. Intrascanner precision was characterized based on five replicate measurements of each phantom configuration. Interscanner precision was characterized based on measurements of each configuration on 10 Lunar iDXA systems.
An independent ethics committee located at the Hospital of Lausanne (Center hospitalier universitaire vaudois, CHUV, Switzerland) approved this study. A total of 32 obese women were recruited by the Nestle Research Center to participate in a nutrition research study. Informed written consent was obtained from all subjects. Exclusion criteria for the study included male gender, pregnancy, and BMI < 30 kg m−2. Total body scans were acquired on a Lunar iDXA system. All subjects were scanned while wearing a hospital gown and with all metal artifacts removed from their body. Patient positioning was performed according to the operator's manual. Each subject underwent two total body scans, with repositioning between scans.
All Lunar iDXA scan files were analyzed with enCORE version 13.6 to generate the CoreScan values. Microsoft Excel 2007 and Minitab version 12.23 were used for statistical analysis. For the phantom analysis, both inter- and intrascanner precisions of CoreScan VAT measurements were determined. Descriptions of subjects in the clinical experiments are reported as group mean, standard deviation (SD), and range. All precision results are reported as the root mean square SD (RMS-SD) and coefficient of variance (%CV).
The intrascanner precision for the VAT phantom across the nine VAT masses ranged from 28.4 to 38.0 g. The interscanner precision ranged from 24.7 to 38.4 g. There was no statistical dependence of the quantity of VAT on either in inter- or intrascanner precision result (P = 0.670). The combined impacts of the intra- and interscanner precision can be computed as the square root of the sum of the intra- and interscanner variance components. This analysis yielded a precision of 47.6 g for VAT mass, which corresponds to 4.8% CV on a 1 kg VAT mass.
The clinical study consisted of 32 obese women. The subjects had an average age of 35.3 ± 5.2 years (range: 25.2-45.2 years), height of 162.6 ± 7.1 cm (range: 150.0-178.5 cm), weight of 92.9 ± 11.0 kg (range: 73.8-112.5 kg), and BMI of 35.1 ± 3.1 kg m−2 (range: 30.0-40.0 kg m−2). The average VAT mass in this sample was 1110.4 ± 435.0 g (range: 404.2-2296.8 g). The RMS-SD precision was 56.8 g with a CV of 5.1%. VAT mass can be converted to VAT volume using a standard coefficient of 0.94 g cm−3. This analysis yields a precision for VAT volume of 60.2 cm3. The CV is the same for VAT mass and VAT volume. Bland-Altman analysis of the data found a non-significant (α = 0.05) bias of 15.4 g with a 95% confidence interval of −172.5 to +141.7 g.
Results of this study demonstrate the repeat measurement precision of the DXA-based CoreScan VAT estimation tool. Both phantom experiments and clinical experiments show that the precision of the measurement is around 5% CV or 50 g on a 1 kg VAT mass. Phantom experiments demonstrated that there was no statistically significant effect of the quantity of VAT on the precision of the measurement.
Previous reports of total body and regional precision for DXA measurements have shown %CV between 0.5 and 5% (3-8). Initial results suggest that total body and some larger subregions within the Lunar iDXA total body scan files have precision on the order of 1-3% (4, 9). The CoreScan value, while slightly higher than some other regions of interest is reasonable considering the small physical size of the region, and the mathematical modeling required to separate the subcutaneous from the visceral adipose tissue in the android region. Both of these factors would be expected to lead to small increases in precision error of CoreScan VAT.
Phantom experiments are an effective tool to test interinstrument performance. Many large trials in nutrition, weight management, and cardiometabolic disease require data to be collected at multiple centers. This study shows that by following standard DXA calibration procedures, CoreScan data can be collected at multiple sites with minimal effect of the instrument on VAT results.
Another encouraging finding of this study was that the phantom precision was similar to that observed in the clinical study. This is particularly significant since the clinical study was conducted on obese women. Obese subjects may be expected to have larger absolute precision error over replicate examinations due to the potential for increased quantum noise in the DXA image acquisition in these subjects. Women are also an important group because they have been previously shown to have the strongest associations between VAT accumulation and cardiometabolic disease. Results of this study suggest that CoreScan may be a useful tool for the evaluation of changes in VAT in obese women.
This study has several limitations. First, the phantom has a single subcutaneous fat geometry. Future phantom studies with varying subcutaneous fat geometry may improve our understanding of CoreScan technical performance. The clinical study only includes obese women. Future studies should quantify precision in obese men and subjects of both genders with normal and overweight BMI. Further, precision data should be collected in longitudinal studies in addition to short term measurements such as those presented here. Finally, intervention trials may benefit from a crossover study, where both a reference standard method, such as CT, and DXA VAT values are collected. This will increase confidence that differences in VAT that can be detected with volumetric imaging modalities can be replicated with DXA methods.
CoreScan VAT represents an automated, low cost, low dose procedure for quantifying VAT. The precision of the measurement is a critical component for consideration in longitudinal trials. Results of this study suggest that CoreScan has a low precision error in both phantoms and obese women and therefore may be a useful addition to trials where interventions are targeted toward changes in visceral adiposity.
Excel is a trademark of Microsoft Corporation in the United States and other countries. Minitab is a trademark of Minitab Inc.
- 3Precision of total body BMD and body composition using the GE Lunar densitometer. J Bone Miner Res 2005; 20: S220., , .