Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Int J Body Compos Res. Author manuscript; available in PMC 2011 May 3.
Published in final edited form as:
Int J Body Compos Res. 2004 January 1; 1(4): 155–160.
PMCID: PMC3086261

Measurement of Body and Liver Fat in Small Animals Using Peripheral Quantitative Computed Tomography


Peripheral quantitative computed tomography (pQCT) was used to determine percent body fat in mice, and relative liver fat in lemmings fasted for 0, 6, 12 or 18 hours to induce a wide range of liver fat content. Accuracy of the pQCT was determined by comparing pQCT-derived fat to that from chemical extraction using 30 male mice (whole body) and 26 female lemmings (liver only). To determine whether pQCT could measure changes in liver fat (%) in live animals, two groups of lemmings were scanned on 4 consecutive days under anesthesia. Controls (n = 3) had ad libitum access to food, whereas the fasted group (n = 5) was deprived of food for 18 hr before being measured on day 2 and then refed. The coefficient of variation (CV) for determining percent body fat in mice using the pQCT was 3.9% (±1.8 SD). Percent body fat determined by pQCT significantly overestimated percent fat as measured by chemical extraction (14.5 ± 3.2 vs 12.3 ± 2.9% respectively, P < 0.01, mean ± SD). However, percent body fat by pQCT was highly related to chemical extraction percent fat (r = 0.95, P < 0.001). The liver attenuation values from pQCT were highly related to percent liver fat (r=0.98, P<0.001) in lemmings. The technique showed excellent precision with a CV of 0.3 ± 0.1%. The two groups (control vs fasted) did not differ in their percent liver fat on day 1 (5.4% vs 5.8%). On day 2 the fasted group had a significantly higher percent liver fat than controls (5.9% vs 17.3%; p<0.05). Following refeeding, there were no significant group differences in percent liver fat on days 3 and 4. Our data indicate that pQCT has good accuracy and precision for determining percent body fat, and liver fat in small animals and can be used to track changes in liver fat over time.


Recent advances in imaging techniques have allowed for in vivo measures of body composition in small rodents. For instance, dual-energy X-ray absorptiometry (DXA) has been used to measure lean tissue mass, fat mass, and bone mineral content in rats (14), mice (58), and lemmings (9). This technology enables investigators to study body composition changes within an animal over time or after a treatment regimen.

Computed tomography (CT) has been used extensively for determining adiposity in human studies (10). The use of CT in animal studies has been much more limited. Ross et al., (11) utilized a human clinical CT instrument to measure adipose tissue in rats. Their results showed that the CT-derived data were not significantly different from that obtained by standard chemical analysis, suggesting that CT was a useful tool for determining body composition in rats. However, the cost of large clinical-CT instruments and issues concerning access may limit its usefulness for animal research. Smaller peripheral quantitative computed tomography instruments (pQCT) may be of greater use to animal researchers due to their lower cost and portability. pQCT has been used extensively for determining volumetric bone mineral content and density in rodent models (12;13). However, to our knowledge these instruments have not been used for determining relative body and liver fat in rodents.

Computed tomography allows for the measurement of attenuation (density) of individual tissues. Thus CT has been used in human studies to estimate the relative fat content of liver and muscle (1416). Since fat has a lower attenuation than lean tissue, a decrease in CT attenuation of an organ or muscle indicates an increase in fat infiltration. Most studies have examined only the attenuation value of the organ or muscle and have not attempted to convert the data into a percent fat value (1416). Ricci et al., (17) utilized phantoms that simulated normal to fatty livers to calibrate CT density to percentage hepatic fat content. The generated regression equation was then used to predict relative liver fat in 29 patients and the value was compared to histomorphometric data from a liver biopsy. A strong positive relationship (r = 0.83) between CT-predicted relative fat and histomorphometry was achieved (17). Thus, CT is a useful tool for measuring relative liver fat in vivo.

Computed tomography techniques have not yet been applied by researchers employing small rodents for the study of obesity and nonalcoholic fatty liver disease (NAFLD). Instead, animals are typically killed and liver fat content is determined histologically (18;19), or by chemical fat extraction techniques (19). Clearly these end-point studies do not allow for longitudinal measures of relative fat content. In this study, we validated the use of pQCT for measuring percent body fat in mice, and percent liver fat in lemmings and then longitudinally followed the progression and resolution of fatty liver in the same animal model. We chose the collared lemming (Dicrostonyx groenlandicus) as a model system since this animal develops fatty liver with fasting as has been shown with other Arvicoline (voles and lemmings) rodents (2022). Thus, the purpose of the present studies was to determine the validity and utility of pQCT for measuring relative body and liver composition of small animals.


Measurement of percent body fat in mice


Three-week old male C57BL/6J mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice were maintained in 28 × 13 × 13-cm cages, provided with rodent diet (7012C; Harlan-Teklad, Madison, WI) and tap water ad libitum. Photoperiod and temperature in the room were maintained at 12:12-hour light/dark and 22 ± 2°C. Animals were either housed singly or in groups of 5 for a period of 9 weeks and were killed with CO2. We have previously reported on the effects of housing condition on phenotypic variance in these animals (23). The male mice (n = 10) used for the precision study were of mixed background (C57BL/6J × 129/SvJ). Experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham.


Accuracy of the pQCT was determined by comparing percent body fat of mice determined from 5 scans of the pQCT and percent body fat assessed by chemical extraction. Precision was determined by scanning 10 mice, five times each, with repositioning after each scan, and determining the coefficient of variation of percent body fat. Accuracy and precision were performed in both anesthetized and euthanized mice.

pQCT scans

Mice were placed on a plexiglass sheet and computed tomographic images were obtained using a Stratec XCT Research SA+ pQCT (Norland Medical Systems Inc.; Fort Atkinson, WI). Unless otherwise stated, the mice were killed prior to the measurement. A scout scan of the mouse was performed starting at the distal end of the skull. The scout image was then used to position five transverse scans. The anterior scan was taken just distal to the head of the humerus and the most posterior scan just distal to the head of the femur. The remaining three scans were evenly spaced between the proximal and distal scans. Voxel size was set at 0.30 mm, scan speed at 20 mm/sec, and collimation width at 0.46 mm. Total scan time (scout scan and 5 transverse scans) was approximately eight minutes. Transverse slices were analyzed using the software provided by the manufacturer (Version 5.40). Total cross-sectional area of each slice (mm2) was determined by setting the attenuation threshold at 0.2047 (1/cm) to exclude all air points. The cross-sectional area of fat (mm2) was determined by setting the attenuation threshold at 0.2773 (1/cm) and determined as the loss in area (cross-sectional area using 0.2047 minus the cross-sectional area using 0.2773). Percent fat content for each slice was determined as: (fat area/total area) × 100 and averaged over the five slices.

Chemical carcass analysis

Body composition was determined by Soxhlet extraction of the body of the mouse as previously described (8). The animals’ head was removed after the scans, thus the data are for the body only.

Measurement of percent liver fat in collared lemmings


Female lemmings (Dicrostonyx groenlandicus) 8–10 weeks old were obtained from our breeding colony at the University of Alabama at Birmingham.


To determine the accuracy of pQCT to measure relative liver fat, twenty-six female lemmings were euthanized after fasting for 0, 6, 12 or 18 hours, and placed in the pQCT scanner (Stratec XCT SA+). By fasting the lemmings for differing times we were able to obtain a wide range of percent liver fat. A validation equation was obtained from the comparison of pQCT attenuation values and percent liver fat from the chemical extraction. In addition, eight female lemmings were euthanized, and scanned five times each, repositioning between each scan to examine the precision of the technique. The attenuation values from each scan were used to calculate the coefficient of variation.

Longitudinal changes

To examine whether pQCT can detect longitudinal changes in percent liver fat, eight female lemmings were anesthetized (Isofluorane) and a pQCT scan taken of the liver on four consecutive days. On day 1 (pm) after scanning, lemmings were assigned to 2 groups: controls (n=3) and fasted (n=5). The control group had ad libitum access to food throughout the four days, whereas the fasted group were fasted after the first scan for 18 hours (till day 2: am), before being allowed ad libitum access to food for days 3 and 4. Percent liver fat was calculated from the pQCT attenuation values using the validation equation.

pQCT scans

Lemmings were placed on a plexiglass sheet and computed tomographic images were obtained using a Stratec XCT Research SA+ pQCT (Norland Medical Systems Inc.; Fort Atkinson, WI). A scout scan of the lemming was performed to locate the base of the lungs. Four consecutive images were taken 2mm apart starting from the base of the lungs (only data from the first image are presented here). The settings for the pQCT were the same as for the mice. Attenuation values were obtained for the liver in the first image.

Measurement of liver fat

Following the scans, the liver was removed, weighed, and dried to constant weight (60°C). Fat content of the dry liver was determined by chemical extraction (Soxhlet). Percent fat content was calculated using the wet liver weight and the fat content.


All statistical analyses were conducted using SAS (ver. 8.0, SAS Institute, Inc., Cary, NC). Precision was taken as the mean intra-individual coefficients of variation (CV) for 10 (8 for lemmings) animals that were scanned 5 times each with repositioning between scans. Percent body fat determined by pQCT and chemical extraction were compared by paired t tests and Pearson-product moment correlations. Prediction equations were determined using linear regression analysis. Residual plots were used to determine whether there was an association between the difference between the methods and the chemically-derived data (considered the gold standard). Data were considered significant when P < 0.05.


Percent body fat in mice

Accuracy of the pQCT for determining percent body fat was determined by comparing the pQCT fat measures with those obtained from chemical carcass analysis. pQCT-determined fat significantly (P < 0.001) overestimated percent fat mass as determined by chemical extraction (14.5 ± 3.2 vs. 12.3 ± 2.9%). Although pQCT overestimated chemically-extracted percent fat content, the two were highly correlated (r = 0.95, P < 0.01). Regression analyses revealed that chemically-extracted percent fat content could be predicted by pQCT (Fig. 1; chemical %fat = 0.86*pQCT %fat − 0.2; r2 = 0.91, root mean standard error (RMSE) = 0.9%; P < 0.001). The extent of overestimation was significantly related to percent fat (Fig. 2). pQCT overestimated more at lower percent fat, and overestimated less at higher percent fat. The intra-individual CVs ranged from 0.9% to 7.4% with a mean of 3.9% and an SD of 1.8%.

Figure 1
Relationship between body fat of mice determined by chemical carcass analysis and by pQCT. The regression line is chemical: %fat = 0.86*pQCT %fat −0.2; r2 = 0.91, RMSE = 0.9%; P < 0.001. The dashed line represents the line of identity. ...
Figure 2
Residual plot of the difference between the pQCT and chemically-derived percent fat as a function of chemically-derived percent fat in mice. The extent of pQCT overestimation was significantly related to percent body fat (P<0.001).

Accuracy and precision were also measured for live mice with similar results. pQCT-derived fat was significantly higher than chemically extracted percent fat (P<0.001, mean difference = 6.5 ± 2.2 %), although the two were very highly correlated (r = 0.98, P<0.001). The CV’s of the live mice ranged from 1.2% to 6.2% with a mean of 3.8% and a SD of 1.7%.

Percent liver fat in lemmings

Liver fat, as measured by chemical extraction, ranged from 3.0% to 27.8%, with a mean of 11.5 % and SD of 6.7 %. By fasting for different durations we obtained a nine-fold variation in liver fat. Attenuation values of the liver (from pQCT) were highly correlated with chemically-extracted percent liver fat (r=−0.98, P<0.01; Fig. 3). The validation equation obtained from this relationship was: chemical %liver fat = 267 − (777 * liver attenuation). There was no significant association between the difference in percent liver fat by the two methods and the actual percent liver fat (P>0.05, Fig 4.). The CV’s for percent liver fat ranged from 0.2 to 0.5 %, with a mean of 0.3% and standard deviation of 0.1%.

Figure 3
Relationship between percent liver fat in lemmings determined by chemical analysis and pQCT attenuation. The regression line is: chemical % fat = 267 − (777 * attenuation); r2=0.96; RMSE = 1.4%; P < 0.001).
Figure 4
Residual plot of the difference between the pQCT and chemically-derived percent liver fat as a function of chemically-derived percent liver fat in lemmings. There was no significant bias in the difference between the methods at different levels of percent ...

Longitudinal measures of liver fat in lemmings

On day 1 prior to fasting, there was no significant difference in percent liver fat between the control and fasted group (P>0.05; 5.4 % vs 5.8%; Fig. 5). After an 18-hour fast, the fasted group had significantly greater percent liver fat than the control group (17.3% vs 5.9%; P<0.05). After refeeding, there was no significant difference in percent liver fat on days 3 and 4 (P>0.05).

Figure 5
Effect of fasting on percent liver fat in lemmings. *** P < 0.001 versus control value.


Recent advances in instrumentation are allowing for in vivo analysis of body composition in small rodents. Many of these instruments were originally designed for measuring bone mineral density and content in human appendages. For example peripheral DXA instruments originally designed for determining bone density in the human os calcis and the forearm have been validated and used to determine fat, lean, and bone in small rodents (8;9). Similarly, pQCT instruments that were designed for determining trabecular and cortical bone density in the human ulna have been adapted and used to measure bone mineral density in rodents (12;13). This transfer of technology to animal researchers will lead to exciting new research. In this paper we describe the use of pQCT for determining percent body and liver fat in small rodents, and the potential benefits to both obesity and hepatic steatosis animal models.

One limitation of the technology transfer is that the hardware is seldom changed to specifically accommodate different size rodents. For instance, the GE-Lunar PIXImus DXA has an image area of only 60 × 80 mm. Although this may be ideal for measuring the human os calcis, it is too small to easily measure large mice (ob/ob, mice over 50g; (24)). Even scanning smaller mice (25–35 g) is problematic since the length of the mouse often exceeds 80 mm. Thus, the head of the mouse must be excluded from the analysis using an exclusion region of interest.

A second limitation of DXA instruments is that only areal (two-dimensional) measures are possible. Thus, DXA is not ideal for obtaining information on specific fat pads. One study has used small animal DXA to estimate body fat distribution (7), but little data were shown on the validity of these measures.

Computed tomography is capable of giving information on both absolute adiposity and on specific depots (11). pQCT may be especially useful to animal researchers given the relatively low cost and portability. The instrument used in the current study has a gantry opening that allows animals up to 90 mm in diameter to be scanned. Thus the machine may be able to scan a rat from birth through adulthood allowing detailed longitudinal studies to be performed.

In the present study, we focused on the ability of pQCT to measure percent body fat and liver fat in small rodents. The precision of the instrument (CV = 3.9%) for determining percent body fat is similar to that of small animal DXA (8;9). Although pQCT overestimated percent fat as determined by carcass analysis, the relationship between the two methods was good (r2 = 0.91), suggesting the pQCT may be a useful method for determining body composition of mice.

In a pioneering study by Ross et al., (11), it was found that CT accurately measured fat mass when compared to chemical carcass analysis and the two methods correlated extremely well (r = 0.99). In that study 12 transverse slices were taken for each rat as compared to only five transverse scans in the present study. A reduction in slice number might be expected to decrease precision and accuracy. We chose 5 slices for practical reasons, since increasing slice number increases total scan time. In our protocol, the entire scan time (scout scan and 5 transverse slices) was approximately 8 minutes. We felt that this was a reasonable period of time to keep animals anesthetized and to allow a fairly high throughput. In studies where greater precision and accuracy are needed, it may be necessary to increase slice numbers.

It is also possible that our selection of the fat threshold was too high. We based our thresholds on visual analysis of selected scans. The attenuation threshold of 0.2773 (1/cm) appeared to give a clear distinction between fat and lean tissue. Therefore we did not change the threshold values in an attempt to lower the percentage fat measured by the instrument. Instead we chose to use a regression technique to correct the pQCT data based on carcass analysis.

In addition to being able to discriminate different tissues (fat and lean), computed tomography should be able to give information on the relative fat infiltration of an organ or muscle. As the fat content of lean tissue increases, the attenuation value should decrease given that fat attenuates X-rays to a lesser extent than pure lean tissue. We therefore used this phenomenon to first, create a regression equation relating percentage liver fat from chemical analysis to the measured attenuation value of the liver. Our findings revealed that the two parameters were highly related and reproducible. With this data in hand, it was then possible to track changes in percent liver fat within the same animal over time by measuring the attenuation value of the liver and then converting this value into percentage liver fat. Although we chose lemmings as our model system, clearly the technique will be useful for other small animal models of hepatic steatosis.

Our results suggest that pQCT may be a useful method for measuring body composition in small rodents. The instrument is small and relatively inexpensive (compared to large clinical-CT instruments) making it feasible for laboratory animal investigations.


This work was supported by NIH R01-DK54918 (TRN), the University of Alabama at Birmingham Clinical Nutrition Research Center (DK56336), Center for Metabolic Bone Disease (AR46301), and Health Services Foundation General Endowment Fund.

Reference List

1. Bertin E, Ruiz J-C, Mourot J, Peiniau P, Portha B. Evaluation of dual-energy X-ray absorptiometry for body-composition assessment in rats. Journal of Nutrition. 1998;128:1550–1554. [PubMed]
2. Rose BS, Flatt WP, Martin RJ, Lewis RD. Whole body composition of rats determined by dual energy X-ray absorptiometry. Journal of Nutrition. 1998;128:246–250. [PubMed]
3. Jebb SA, Garland SW, Jennings G, Elia M. Dual-energy X-ray absorptiometry for the measurement of gross body composition in rats. British Journal of Nutrition. 1996;75:803–809. [PubMed]
4. Nagy TR, Prince CW, Li J. Validation of peripheral dual-energy X-ray absorptiometry for the measurement of bone mineral in intact and excised long bones of rats. Journal of Bone and Mineral Research. 2001;16:1682–1687. [PubMed]
5. Brommage R. Validation and calibration of DEXA body composition in mice. Am.J.Physiol Endocrinol.Metab. 2003;285:E454–E459. [PubMed]
6. Iida-Klein A, Lu SS, Yokayama K, Dempster DW, Nieves JW, Lindsay R. Precision, accuracy and reproducibility of dual X-ray absorptiometry measurements of mice in vivo. Journal of Clinical Densitometry. 2003;6:25–33. [PubMed]
7. Masuzaki H, Paterson J, Shinyama H, et al. A transgenic model of visceral obesity and the metabolic syndrome. Science. 2002;294:2166–2170. [PubMed]
8. Nagy TR, Clair AL. Precision and accuracy of dual-energy X-ray absorptiometry for determining in vivo body composition of mice. Obesity Research. 2000;8:392–398. [PubMed]
9. Hunter HL, Nagy TR. Body composition in a seasonal model of obesity: longitudinal measures and validation of dual-energy X-ray absorptiometry. Obesity Research. 2002;10 [PubMed]
10. Després J-P, Ross R, Lemieux S. Imaging techniques applied to the measurement of human body composition. In: Roche AF, Heymsfield SB, Lohman TG, editors. Human Body Composition. Champaign, IL: Human Kinetics; 1996. pp. 149–166.
11. Ross R, Léger L, Guardo R, De Guise J, Pike BG. Adipose tissue volume measured by magnetic resonance imaging and computerized tomography in rats. Journal of Applied Physiology. 1991;70:2164–2172. [PubMed]
12. Andersson N, Lindberg MK, Ohlsson C, Andersson K, Ryberg B. Repeated in vivo determinations of bone mineral density during parathyroid hormone treatment in ovariectomized mice. Journal of Endocrinology. 2001;170:529–537. [PubMed]
13. Rosen HN, Tollin S, Balena R, et al. Differentiating between orcheictomized rats and controls using measurements of trabecular bone density: a comparison among DXA, histomorphometry, and peripheral quantitative computerized tomography. Calcified Tissue International. 1995;57:35–39. [PubMed]
14. Simoneau J-A, Colberg SR, Thaete FL, Kelley DE. Skeletal muscle glycolytic and oxidative enzyme capacities are determinants of insulin sensitivity and muscle composition in obese women. FASEB Journal. 1995;9:273–278. [PubMed]
15. Goto T, Onuma T, Takebe K, Kral JG. The influence of fatty liver on insulin clearance and insulin resistance in non-diabetic Japanese subjects. International Journal of Obesity. 1995;19:841–845. [PubMed]
16. Katoh S, Hata S, Matsushima M, et al. Troglitazone prevents the rise in visceral adiposity and improves fatty liver associated with sulfonylurea therapy - a randomized controlled trial. Metabolism. 2001;50:414–417. [PubMed]
17. Ricci C, Longo R, Gioulis E, et al. Noninvasive in vivo quantitative assessment of fat content in human liver. Journal of Hepatology. 1997;27:108–113. [PubMed]
18. Baffy G, Zhang C-Y, Glickman JN, Lowell BB. Obesity-related fatty liver is unchanged in mice deficient for mitochondrial uncoupling protein 2. Hepatology. 2002;35:753–761. [PubMed]
19. Brix AE, Elgavish A, Nagy TR, Gower BA, Rhead WJ, Wood PA. Evaluation of liver fatty acid oxidation in the leptin-deficient obese mouse. Molecular Genetics and Metabolism. 2002;75:219–226. [PubMed]
20. Mosin AF. Some physiological and biochemical features of starvation and refeeding in small wild rodents (Microtinae) Comparative Biochemistry and Physiology. 1982;71A:461–464. [PubMed]
21. Mosin AF. On the energy fuel of voles during their starvation. Comparative Biochemistry and Physiology. 1984;77A:563–565. [PubMed]
22. Nagy TR, Pistole DH. The effects of fasting on some physiological parameters in the meadow vole, Microtus pennsylvanicus. Comparative Biochemistry and Physiology. 1988;91:679–684. [PubMed]
23. Nagy TR, Krzywanski D, Li J, Meleth S, Desmond R. Effect of group vs. single housing on phenotypic variance in C57BL/6J mice. Obesity Research. 2002;10:412–415. [PubMed]
24. Nagy TR. The use of dual-energy X-ray absorptiometry for the measurement of body composition. In: Speakman JR, editor. Body Composition Analysis of Animals: A Handbook of Non-Destructive Methods. Cambridge: Cambridge University Press; 2001. pp. 211–229.