Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2009 Aug 6.
Published in final edited form as: Int J Body Compos Res. 2009 Jan 1;7(1):21–26.

Precision and accuracy of bioimpedance spectroscopy for determination of in vivo body composition in rats

DL Smith Jr 1, MS Johnson 1, TR Nagy 1
PMCID: PMC2722071  NIHMSID: NIHMS117235  PMID: 19668348

Abstract

Objective

Bioelectrical impedance analysis (BIA) of hydration and body composition has made significant progress during the past 3 decades. With the development of Bioimpedance spectroscopy (BIS), bioimpedance has been expanded to reliably predict extracellular fluid (ECF) and total body water (TBW), allowing the calculation of fat-free mass (FFM) and fat mass (FM). In this study, a new BIS device (ImpediVet™), designed for body composition measurements in animals, was assessed for precision and accuracy in measuring TBW, FFM and FM in rats.

Methods

In a validation study, 25 rats were measured for body composition (TBW, FFM and FM) using BIS and chemical carcass analysis (CCA). BIS precision was assessed by the coefficient of variation using multiple BIS readings, while BIS accuracy was assessed by regression analysis of BIS and CCA values for each body compartment. In a cross-validation study, prediction equations generated from the validation group for TBW, FFM and FM were applied to an independent cohort of 25 rats that were measured by BIS and CCA. Linear regression analysis and paired t-tests were used to assess significance of relationships and measurement differences within groups.

Results

In the validation study, BIS was highly correlated with CCA for TBW (r2=0.988), FFM (r2=0.987) and FM (r2=0.966). Even so, BIS significantly underestimated TBW (mean: −31.07 g, −13.3%, p<0.001) and FFM (−50.69 g, −15.5%, p<0.001), while overestimating FM (+65.75 g, +63.5%, p<0.001). In the independent, cross-validation group of rats the prediction equations accurately predicted carcass values for TBW (−0.2%, p=0.350), FFM (−0.2%, p=0.457) and FM (+1.5%, p=0.508).

Conclusion

Based on these results, BIS provided a precise and accurate means to determine in vivo body composition in rats.

Keywords: Body Composition, Bioimpedance Spectroscopy, Rats, Fat, Lean

Introduction

The accurate determination of in vivo body composition is critical for the interpretation of many types of animal studies. Although quantitative magnetic resonance (QMR) (1–3) and dual-energy X-ray absorptiometry (DXA) (4;5) have proven useful for determining in vivo body composition, they require large and costly equipment. Bioelectrical impedance analysis (BIA) is a low cost alternative for the measurement of hydration status and body composition in humans (6). However, bioimpedance spectroscopy (BIS) has not been well developed for small animals. This study investigated the precision and accuracy of the handheld ImpediVET™ bioimpedance spectroscopy device for body composition measurements (total body water – TBW, fat-free mass – FFM, and fat mass – FM) in rats. Bioimpedance spectroscopy (BIS) utilizes the passage of a small electric current through an organism to determine the impedance of current flow (7). The impedance is proportional to the length, inversely proportional to cross-sectional area and directly related to the composition of the material through which current passes (8). Assuming a simple cylinder shape and using a range of frequencies (from 3 – 1000 kHz) combined with the body weight and resistivity coefficients derived for rats – TBW (extracellular (ECW) and intracellular water (ICW)), FFM and FM are calculated. The composition of compartments is determined by the values derived for each fraction of the calculated water, with the low frequency current passing primarily around and between cell membranes; while the high frequency current passes both around and through cell membranes (9). Based on the differential water composition of fat and lean tissues, an estimation of the total FFM and FM can be determined (10–14).

Using a group of Osborne-Mendel albino rats, BIS was assessed for precision and accuracy in the determination of TBW, FFM and FM compared with chemical carcass analysis (CCA). A separate cohort of rats was used to cross-validate the prediction equations generated from the regression of BIS and CCA.

Methods

Animals

A total of 50 Osborne-Mendel albino rats were used in the study. Animals were multiply housed in standard rat cages at 22±1°C on a 12:12 light:dark cycle. Animals were allowed ad libitum access to standard chow (Harlan Teklad Rodent Diet #2016; Madison, WI, USA) and water until the time of measurements. Twenty-five Osborne-Mendel rats (validation group: n=14 male, 503.7±170.0 g, 16.7±1.9 cm; n=11 female, 372.5±54.7 g, 15.5±0.9 cm) were anesthetized and measured using the ImpediVET™ Bioimpedance Spectroscopy device (ImpediMed Limited, Brisbane, Australia). Chemical carcass analysis (CCA) was performed for all rats for the determination of body composition. A second group of 25 rats (cross validation: n=15 male, 495.6±131.2 g, 16.2±1.8 cm; n=10 female, 355.4±70.7 g, 14.8±.09 cm) was used for an independent cross validation of the prediction equations. All animal care and use procedures were approved by the UAB Institutional Animal Care and Use Committee.

BIS measurements

All animals were weighed to the nearest 0.1 g live weight and anesthetized with isoflurane (3–5% isoflurane in O2) for the duration of the BIS measurements. Anesthetized animals were placed flat on their abdomen on a non-conductive surface with fore and hind limbs beside the body and tail extended distally. The tips of four 25 gauge 1” needles (NN*2525R; Terumo Medical Corp, Elkton, MD, USA) were bent at a 90° angle, 5 mm from the tip. Needles were inserted subdermally along the dorsal midline of the animal at the anterior edge of the eye orbit (source 1), anterior edge of the pinna (detector 1), the sacral-caudal junction (detector 2) and fur line at the base of the tail (source 2). Electrodes and needles were attached per manufacturer’s instructions. The length between the 2 detector needle electrodes was measured along the dorsal midline with a standard measuring tape to the nearest 0.5 cm and used for the BIS device calculations. Body proportion = 1.0, body density = 1.05 g/cm3, hydration constant = 0.732 and resistivity coefficients of ρe = 325, ρi = 752 for males and ρe = 289, ρi = 669 for females were used for all calculations. Whole body resistance and reactance readings were acquired using a single spectrum acquisition from 4 khz to 1000 khz. Three consecutive measurements were performed for a single positioning of the 4 needle electrodes. Needle electrodes were subsequently repositioned 2 additional times for a total of 9 measurements (3 positions, 3 measurements each). Acquired data was downloaded and processed using the provided bioimpedance software (ImpediVet Vet BIS1 v. 1.0.0.4). Following the final BIS measurement, animals were euthanized by carbon dioxide asphyxiation and subjected to chemical carcass analysis including total water, fat free dry mass, fat mass and ash content.

Chemical carcass analysis

Carcass composition was determined for the validation and cross-validations groups using the method of Harris and Martin 1984 (15;16) with modifications. Briefly, euthanized animals were weighed to the nearest 0.1 g body weight, autoclaved for 99 minutes at 120 °C and allowed to cool overnight at 4°C. Carcasses were brought to room temperature and weighed to the nearest 0.1g. Deionized water was added to the carcass proportional to recorded body weight (1:1 – water:body weight). Rat carcasses were homogenized using an industrial stainless steel blender (Waring Commercial Model 38BL19 – CB10 and CB-2-10 Container; Waring Commercial, Torrington, CT, USA). Three aliquots of approximately 8 ml of the homogenate were added to 50 ml conical tubes and weighed to the nearest 0.0001 g. These samples were subjected to fat extraction by KCl-chloroform extraction. Ten ml of methanol and 5 ml of chloroform were added to each tube and briefly vortexed, followed by incubation on ice for 1 hr. An additional 5 ml of chloroform and 5 ml 1.0M KCl were added to each tube, vortexed and kept on ice for 20 minutes. These samples were then centrifuged at 810 × g in a swinging-bucket centrifuge at 4°C for 30 minutes. The liquid layer on top of the pellet was carefully aspirated and the chloroform layer (below the pellet) was gently poured into aluminum drying pans (pre-weighed to the nearest 0.0001 g) and allowed to evaporate in a fume hood overnight. The evaporated fat samples were placed in a 6 5–70°C oven for approximately 30 minutes to evaporate any residual moisture, and then cooled to room temperature in a dessicator before being weighed to the nearest 0.0001 g.

An additional three aliquots of the homogenate were partitioned to 17 ml porcelain crucibles and immediately weighed to the nearest 0.0001 g. These samples were subsequently dried for 7 days at 65–70°C in an oven, brought to room temperature in a dessicator and weighed to the nearest 0.0001 g. Samples were then ashed overnight at 600°C in a muffle furnace. After cooling to room temperature in a dessicator, the inorganic ash content of the porcelain crucibles was measured to the nearest 0.0001 g.

Statistics

Data were analyzed with SAS 9.1 statistical software package (SAS Institute, Cary, NC, USA). BIS and CCA were compared for both the validation and cross-validation groups using linear regression analysis and paired t-tests. Results were considered significant when p≤0.05. The coefficient of variation (CV) was calculated as the standard deviation divided by the mean (× 100) for indicated BIS readings. Bias in TBW, FFM and FM was determined by comparing the residual difference between the measured BIS – CCA value with the CCA value. Prediction equations were generated from the validation group (for TBW, FFM and FM) based on the linear regressions.

Results

Validation - Precision

To assess the precision of the BIS device, 3 consecutive readings were acquired from a single needle electrode placement and the coefficient of variation (CV) was calculated. As described in Table 1, the CV was approximately 2% or less for TBW, FFM and FM. Additionally, 3 needle electrode placements, with 1 reading acquired at each (readings #1, 4, 7), were assessed for precision with TBW and FFM exhibiting 2.7% and FM 5.4% (Table 1). Finally, all 9 BIS measurements were utilized (3 needle electrode positions with 3 reads each) for precision determination. TBW and FFM CV’s were 2.5% while FM was 4.8% (Table 1).

Table 1.

Coefficient of Variation (mean±s.d.) for TBW, FFM and FM for BIS using combinations of BIS readings based on needle electrode replacement in the validation group.

BIS Reading TBW FFM FM
1 – 3 0.9% ± 1.3% 0.9% ± 1.3% 2.1% ± 4.6%
1, 4, 7 2.7% ± 1.5% 2.7% ± 1.5% 5.4% ± 4.8%
All 2.5% ± 1.0% 2.5% ± 1.0% 4.8% ± 4.3%
Open in a new tab

Validation - Accuracy

To assess the accuracy of the BIS device, the measured TBW, FFM and FM were compared with CCA using the same data acquired for the precision measures described above. By using only the first acquired BIS reading, BIS was highly correlated with CCA for all measures (Table 2). Acquiring two additional readings while maintaining the same needle electrode placement resulted in similar r2 values (Table 2, reading #1–3). The relationship between BIS and CCA was improved by removing and reinserting the needle electrodes, and using the mean of the first BIS reading at each needle electrode reposition (reading #1, 4, 7) for each animal (Table 2). A final comparison utilizing the mean of all 9 readings of BIS for each animal was similarly highly correlated with CCA (Fig. 1.A,B,C; Table 2). Despite the high correlations, BIS significantly underestimated TBW (mean= −31.07 g, −13.3%, p<0.001) (Fig. 2.A) and FFM (mean= −50.69 g, −15.5 %; p<0.001) (Fig.2.B) while significantly overestimating FM (mean= 65.75 g, 63.5 %; p<0.001) (Fig. 2.C). There was also a significant bias in the measurement of TBW, FFM and FM with the underestimation increasing with increasing TBW and FFM, and the overestimation increasing with increasing FM (p<0.05, Fig. 2.A–C). Based on the linear regressions derived for each body composition measure using all 9 BIS readings, prediction equations were generated to correct for BIS - CCA difference in TBW, FFM and FM (Table 3).

Table 2.

Relationship (r2) between BIS and CCA measures of TBW, FFM and FM.

BIS Measurement TBW FFM FM
1st 0.983 0.981 0.961
Single placement (1–3) 0.984 0.984 0.959
3 placements (1, 4, 7) 0.986 0.987 0.968
All (1–9) 0.988 0.987 0.966
Open in a new tab

Table 4.

Cross validation group BIS values (mean±s.d., in grams) for TBW, FFM and FM before (uncorrected) and after (predicted) prediction equation correction with CCA TBW, FFM and FM. Paired t-test significance (p values) for predicted BIS vs. CCA.

Mean Uncorrected BIS Mean Predicted BIS Mean CCA p-value
TBW 200.2±59.7 231.1±62.7 232.7±58.9 0.350
FFM 273.4±81.5 323.1±91.3 325.1±86.4 0.457
FM 166.1±57.5 101.1±43.7 99.6±47.6 0.508
Open in a new tab

Figure 1.

Figure 1

Figure 1

Open in a new tab

Relationships between BIS and CCA for a) TBW b) FFM and c) FM for the validation group (n=25). Dotted line of identity shown for reference. (Mean of all 9 BIS measures for each rat).

Figure 2.

Figure 2

Figure 2

Open in a new tab

Bland-Altman plots of residual (BIS – CCA) for the validation group (n=25) for a) TBW b) FFM and c) FM. Group mean (solid line) ±2 s.d. (dotted lines) shown for each body composition compartment. Regression line (dashed line) demonstrating significant bias in TBW, FFM and FM.

Table 3.

Prediction equations for TBW, FFM and FM (in grams) generated from BIS values (9 BIS readings for each rat) for the validation group.

Dep. Variable Prediction Equations Model r2 P value
CCA TBW TBW = 1.051*BIS TBW + 20.7 0.988 <0.001
CCA FFM FFM = 1.119*BIS FFM + 17.0 0.987 <0.001
CCA FM FM = 0.761*BIS FM − 25.2 0.966 <0.001
Open in a new tab

Cross-validation

A cross validation group of 25 rats was subsequently measured for TBW, FFM and FM by BIS and CCA using the same techniques as the validation group. This cross validation group was similar to the validation group in all aspects regarding sex, body weight, and observed CCA measures of TBW, FFM and FM (Fig. 3). As with the validation group, BIS significantly underestimated TBW and FFM, while BIS significantly overestimated FM (Table 4 – mean uncorrected difference). However, by applying the prediction equations, corrected BIS measures of TBW, FFM and FM, were not significantly different from CCA (Table 4; Fig. 4.A–C).

Figure 3.

Figure 3

Open in a new tab

Box plots of CCA TBW, FFM and FM for the validation (n=25) and cross-validation groups (n=25) showing the 25th and 75th percentiles as the upper and lower half of each box along with the 10th and 90th percentiles as upper and lower error bars plus individual outliers.

Figure 4.

Figure 4

Figure 4

Open in a new tab

Bland-Altman plots of residuals (BIS – CCA) for the cross validation group (n=25) before and after prediction equation correction for a) TBW b) FFM and c) FM. Uncorrected (closed circle) group mean (solid line) and prediction corrected (open circles) group mean (dashed line) ±2 s.d. (dotted lines) shown for each body compartment.

Discussion

With the increase in the number of studies that require accurate body composition measures for animals, additional equipment is needed to provide accessibility to the broad range of investigators. In this study the precision and accuracy of the ImpediVet™ BIS device for measuring body composition in Osborne-Mendel albino rats weighing 200 to 800 grams was determined. The precision of the BIS was good, ranging from 0.9% to 5.4% and averaging 3.2% across all reading for TBW, FM and FFM (Table 1). Although the measured values for BIS TBW, FFM and FM were significantly different from CCA, they were still highly related in each case (Table 2, Fig. 1). However, there was a significant bias with increasing mass for TBW, FFM and FM (Fig 2.A–C), which may be related to the use of resistivity coefficients derived in Wistar rats using tritium dilution, compared with the chemical carcass techniques applied to Osborne Mendel rats in the present study (17). Even so, the derived prediction equations were able to correct for the error and bias of the measured body composition values in the cross-validation study group (Table 4, Fig. 4), indicating that the derived prediction equations work well and can be used to convert BIS measures to carcass equivalents for rats.

A few technical points should be highlighted concerning the use of BIS in small mammals. Following a set protocol for animal handling, needle electrode placement and anesthesia reduced the variation in measured BIS values observed during the initial equipment testing. Thus, as with any device that relies on needle electrode placement, the potential for variation from less than meticulous care is present; however, with care and experience, reproducible placement of needles is achievable. Additionally, the distance measurement between the 2 interior needle electrodes, which directly contributes to the measured BIS values, should be standardized to reduce potential error. As discussed in the literature (6), hydration status can have a significant impact on the accuracy of the measured values. The rats used in this study were allowed ad libitum access to food and water until 30 minutes before the start of BIS measurement. As such, it is assumed that individual and group hydration status did not significantly affect the results. However, performing measures on animals with varying hydration status should be avoided; particularly since the inclusion of a single defined hydration coefficient assumes a similar hydration status between animals and across groups.

Based on the observed body composition measurements, it appears that taking a single measurement of body composition could provide a reliable assessment of body composition. A minimum of 9 readings per animal were acquired (3 per each needle electrode placement with 3 replacements). Taking additional readings, while maintaining the same needle electrode placement, provided little benefit in accuracy beyond a single measurement (Table 2). However, taking readings after multiple needle electrode placements provided a small improvement to the overall accuracy of the BIS measure (Table 2). This may be related to the averaging effect of reducing the influence of any one erroneous measurement that is generated from needle electrode placement differences. Although the variance increases with multiple needle electrode placements, the accuracy increases as well, providing a reasonable trade off with increased accuracy.

The validations of other instruments used to determine in vivo body composition have been previously reported. Dual-energy X-ray absorptiometry (DXA) determination of body composition in Sprague-Dawley rats had similar precision to BIS (CV: DXA – lean=0.83%, fat=3.65% vs. BIS – lean=0.9–2.5%, fat =2.1–4.8%; Table 1), but lower accuracy, particularly for fat determination (r2: DXA – lean=0.906, fat=0.352 vs. BIS – lean=0.987, fat=0.966, Table 2) (5). Likewise, total body electrical conductivity (TOBEC) measurements made in female Wistar rats were less well correlated with chemical carcass values than BIS (r2: TOBEC – lean=0.785, fat=0.952 vs. BIS – lean=0.987, fat=0.966; CV’s not reported) (18).

With proper vigilance in following a standardized protocol and careful attention to needle electrode placement, the BIS measures provide a precise and accurate means to determine in vivo body composition in rats.

Acknowledgements

This work was supported by a grant from ImpediMed, Inc. DLS was supported by T32DK062710. Chemical carcass analysis was provided by the UAB Small Animal Phenotyping Core (P30DK56336 and P60DK079626).

Reference List

  • 1.Taicher GZ, Tinsley FC, Reiderman A, Heiman ML. Quantitative magnetic resonance (QMR) method for bone and whole-body-composition analysis. Anal Bioanal Chem. 2003;377:990–1002. doi: 10.1007/s00216-003-2224-3. [DOI] [PubMed] [Google Scholar]
  • 2.Tinsley FC, Taicher GZ, Heiman ML. Evaluation of a quantitative magnetic resonance method for mouse whole body composition analysis. Obes Res. 2004;12:150–160. doi: 10.1038/oby.2004.20. [DOI] [PubMed] [Google Scholar]
  • 3.Napolitano A, Miller SR, Murgatroyd PR, et al. Validation of a quantitative magnetic resonance method for measuring human body composition. Obesity (Silver Spring) 2008;16:191–198. doi: 10.1038/oby.2007.29. [DOI] [PubMed] [Google Scholar]
  • 4.Nagy TR, Clair AL. Precision and accuracy of dual-energy X-ray absorptiometry for determining in vivo body composition of mice. Obes Res. 2000;8:392–398. doi: 10.1038/oby.2000.47. [DOI] [PubMed] [Google Scholar]
  • 5.Lukaski HC, Hall CB, Marchello MJ, Siders WA. Validation of dual x-ray absorptiometry for body-composition assessment of rats exposed to dietary stressors. Nutrition. 2001;17:607–613. doi: 10.1016/s0899-9007(01)00577-9. [DOI] [PubMed] [Google Scholar]
  • 6.Thomas BJ, Ward LC, Cornish BH. Bioimpedance spectrometry in the determination of body water compartments: accuracy and clinical significance. Appl Radiat Isot. 1998;49:447–455. doi: 10.1016/s0969-8043(97)00052-3. [DOI] [PubMed] [Google Scholar]
  • 7.Foster KR, Lukaski HC. Whole-body impedance--what does it measure? Am J Clin Nutr. 1996;64:388S–396S. doi: 10.1093/ajcn/64.3.388S. [DOI] [PubMed] [Google Scholar]
  • 8.Hoffer EC, Meador CK, Simpson DC. Correlation of whole-body impedance with total body water volume. J Appl Physiol. 1969;27:531–534. doi: 10.1152/jappl.1969.27.4.531. [DOI] [PubMed] [Google Scholar]
  • 9.Jenin P, Lenoir J, Roullet C, Thomasset AL, Ducrot H. Determination of body fluid compartments by electrical impedance measurements. Aviat Space Environ Med. 1975;46:152–155. [PubMed] [Google Scholar]
  • 10.Narath E, Skalicky M, Viidik A. Voluntary and forced exercise influence the survival and body composition of ageing male rats differently. Exp Gerontol. 2001;36:1699–1711. doi: 10.1016/s0531-5565(01)00145-0. [DOI] [PubMed] [Google Scholar]
  • 11.Skalicky M, Narath E, Viidik A. Housing conditions influence the survival and body composition of ageing rats. Exp Gerontol. 2001;36:159–170. doi: 10.1016/s0531-5565(00)00174-1. [DOI] [PubMed] [Google Scholar]
  • 12.Skalicky M, Narath E, Viidik A. Ad libitum fed rats can keep lean or grow fat when they age. Exp Gerontol. 2000;35:1093–1094. doi: 10.1016/s0531-5565(00)00124-8. [DOI] [PubMed] [Google Scholar]
  • 13.Staudinger FB, Rorie RP, Anthony NB. Evaluation of a noninvasive technique for measuring fat-free mass in poultry. Poult Sci. 1995;74:271–278. doi: 10.3382/ps.0740271. [DOI] [PubMed] [Google Scholar]
  • 14.Ilagan J, Bhutani V, Archer P, Lin PK, Jen KL. Estimation of body composition changes during weight cycling by bioelectrical impedance analysis in rats. J Appl Physiol. 1993;74:2092–2098. doi: 10.1152/jappl.1993.74.5.2092. [DOI] [PubMed] [Google Scholar]
  • 15.Harris RB, Martin RJ. Recovery of body weight from below "set point" in mature female rats. J Nutr. 1984;114:1143–1150. doi: 10.1093/jn/114.6.1143. [DOI] [PubMed] [Google Scholar]
  • 16.Hartsook EW, Hershberger TV. A SIMPLIFIED METHOD FOR SAMPLING SMALL ANIMAL CARCASSES FOR ANALYSES. Proc Soc Exp Biol Med. 1963;113:973–977. doi: 10.3181/00379727-113-28548. [DOI] [PubMed] [Google Scholar]
  • 17.Ward LC, Mason S, Battersby KJ. Bioimpedance spectroscopy for the measurement of body composition in laboratory rat in vivo. Int J Body Comp Res. submitted accompanying paper. Ref Type: Generic. [Google Scholar]
  • 18.Dickinson K, North TJ, Telford G, et al. Determination of body composition in conscious adult female Wistar utilising total body electrical conductivity. Physiol Behav. 2001;74:425–433. doi: 10.1016/s0031-9384(01)00535-2. [DOI] [PubMed] [Google Scholar]

RESOURCES