Intra-abdominal Fat Burden Discriminated In Vivo Using Proton Magnetic Resonance Spectroscopy

Authors


Molecular Pathology Unit, Center for Cancer Research, National Cancer Institute, 37 Convent Drive, Building 37, Room 2000, Bethesda, MD 20892. E-mail: ms43b@nih.gov

Abstract

Objective: To assess proton magnetic resonance spectroscopy (1H-MRS) as a means to distinguish among mice with disparate intra-abdominal body fat compositions, and to measure changes in intra-abdominal fat burden during weight loss and regain.

Research Methods and Procedures: Intra-abdominal fat burden was analyzed as a ratio of integrated areas under the curves of fat to water 1H-MRS signals collected from a region of interest standardized across B6.V-Lepob, C57BL/6, and A-ZIP/F mice that exhibited various genotypically related body fat compositions, ranging from obese (B6.V-Lepob) to minimal body fat (A-ZIP/F). 1H-MRS analysis of fat burden was compared with intra-abdominal fat volume and with a single cross-sectional intra-abdominal fat area calculated from segmented magnetic resonance images. Similar measurements were made from obese B6.V-Lepob mice before, during, and after they were induced to lose weight by leptin administration.

Results: Relative amounts of intra-abdominal fat analyzed by 1H-MRS differed significantly according to body composition and genotype of the three strains of mice (p < 0.05). Intra-abdominal fat assessed by 1H-MRS correlated with both intra-abdominal fat volume (r = 0.88, p < 0.001) and body weight (r = 0.82, p < 0.001) among, but not within, all three genotypes. During weight loss and regain, there was a significant overall pattern of changes in intra-abdominal fat quantity that occurred, which was reflected by 1H-MRS (p = 0.006).

Discussion: Results support the use of localized 1H-MRS for assessing differences in intra-abdominal fat. Refinements in 1H-MRS voxel region of interest size and location as well as instrument precision may result in improved correlations within certain body compositions.

Introduction

Anatomically distinct body fat depots have unique metabolic properties (1). Evidence linking intra-abdominal visceral obesity with diseases such as type 2 diabetes mellitus (2), heart disease (3), and cancer (4) has stimulated interest in developing methods to evaluate intra-abdominal fat in human and animal models. Controversy exists as to how well anthropometric measures of obesity reflect intraabdominal fat burden in animal studies (5). Precise methods, such as differential buoyancy (6), electromagnetic conductance (7), specific gravity (8), D2O dilution (9), and DXA (10), exist to study body composition in vivo, but these have not been used to measure intra-abdominal fat depots.

Accurate quantification of individual fat depots may be obtained in humans and animals using non-invasive imaging, such as computed tomography and magnetic resonance imaging (MRI)1 (11, 12, 13, 14). The extended time required for image segmentation, particularly whenever precision of volumetric data is sought, can be a limitation of MRI and computed tomography. Therefore, a precise technique for assessing intra-abdominal fat with improved throughput and safety is desirable.

Proton magnetic resonance spectroscopy (1H-MRS) is a rapid, non-invasive technique that has been used to assess total body fat in vivo (9, 15, 16, 17). Localized 1H-MRS, in which the proton chemical spectrum is generated from an operator-selected region of interest (ROI), has been used to evaluate lipid content, specifically in liver and muscle (15, 18, 19, 20, 21). Aims in this study included the use of localized 1H-MRS, focused on an intra-abdominal ROI, to examine whether various amounts of intra-abdominal fat could be distinguished among mice with different body fat compositions and to evaluate intra-abdominal fat for changes during weight loss and regain.

Research Methods and Procedures

Experimental Design

Images and spectroscopic measurements were made using 1H-MRS and MRI to compare amounts of intra-abdominal fat burden in three mouse genotypes that represent a wide range of body adipose tissue depositions. The mice included obese (B6.V-Lepob/J), lean (C57BL/6), and a third body conformation with minimal adipose tissue, designated A-ZIP/F mice. Measurements were also made from the obese (B6.V-Lepob/J) mice before, during, and after they were induced to lose weight. Relative abundances of intra-abdominal visceral fat were analyzed by comparing ratios of the areas under the curve (AUCs) of fat and water from localized 1H-MRS spectra with intra-abdominal visceral fat volume and cross-sectional fat area obtained by MRI.

Animals and Treatments

Animals included five 10-week-old female B6.V-Lepob/J (Lepob) mice (Jackson Laboratories, Bar Harbor, ME), five 9-week old female C57BL/6 mice (NCI Frederick Cancer Research and Development Center, Frederick, MD), and three adult (4 to 5 months old) female FVB-Tg(AZIP/F)1Vsn/J (A-ZIP/F) mice (available to us by donation from Dr. Charles Vinson, National Cancer Institute/NIH, Bethesda, MD). All mice were fed the same diet ad libitum (NIH-31 mouse ration; Zeigler Bros., Inc., Gardners, PA). Mice were weighed and underwent 1H-MRS and MRI three times at weekly intervals. Throughout anesthesia and imaging, mice were monitored with the aid of a respiratory gaiting instrument (SA Instruments, Inc., Stony Brook, NY), which also helped to minimize motion artifact from respiratory excursions during data acquisition. During scans, body temperature was maintained at 37° C using ambient air warmed by a small-rodent heater system (SA Instruments, Inc.).

During the weight loss induction study, all five Lepob mice were weighed and underwent 1H-MRS and MRI on Study Days 1, 8, 15, 22, 28, 35, 43, and 71. For purposes of injection control, the Lepob mice were given 0.1 mL sterile 0.9% NaCl solution alone subcutaneously twice daily on Days 2 to 15 before the start of leptin administration. On Day 16, Lepob mice were induced to lose body weight by administration of 5 μg recombinant murine leptin/g body weight in 0.1 mL sterile 0.9% NaCl (Biomyx Technology, San Diego, CA) twice daily for a 13-day period (Study Days 16 to 28).

MRI and Spectroscopy Data Acquisition

1H-MRS and MRI were performed on a 7T/21 horizontal Avance scanner (Bruker BioSpin Corp., Billerica, MA). Before data acquisition from each animal, the radiofrequency coil was tuned and matched, and radiofrequency pulses were optimized. Coronal magnetic resonance (MR) images encompassing the whole abdomen were acquired using fast spin echo sequences [echo time (TE), 10 milliseconds; repetition time (TR), adjusted with respiratory synchronization to ∼1500 milliseconds; eight repetitive acquisitions were averaged; four echoes]. Coronal images were obtained using a 256 × 256 matrix with a 60-mm × 60-mm field of view (Lepob) or a 256 × 192 matrix with a 60-mm × 45-mm field of view (C57BL/6, A-ZIP/F), providing 0.23-mm pixel in-plane resolution for all three genotypes. Three 1.0-mm-thick axial images with similar resolution were made of each mouse, using a spin echo sequence (TE, 10 milliseconds; TR, ∼1500 milliseconds; four averages).

1H-MRS was performed using a 15-mm × 5-mm × 5-mm (375 μL) voxel within the mouse abdomens and the imaging phantoms (see below). Voxel volume and position (intra-abdominal, level of the fifth lumbar vertebra) were held constant across all genotypes (Figure 1). The cranial aspect of the voxel coincided with the axial MRI slice used to calculate the cross-sectional area of intra-abdominal fat. The consistency of voxel placement was ensured by reviewing the series of coronal abdominal MR images.

Figure 1.

Position of the 375-μL intra-abdominal 1H-MRS voxel ROI centered below the fifth lumbar vertebra [white rectangle depicted superimposed on single axial (transverse) section MR image] for (A) C57BL/6 (lean) mouse, (B) A-ZIP/F (minimal fat) mouse, and (C) Lepob (obese) mouse. Anatomic structures in white on image typically represent fat. Corresponding representative 1H-MRS spectra are depicted in the right column panel for each respective mouse strain. Each spectrum includes two perceptibly matched peaks representing water (H2O) and fat (−CH2−). Integrated AUC for each peak amplitude was used to represent intra-abdominal visceral water and fat, respectively (see “Research Methods and Procedures”). Complexity of spectral patterns varies uniquely by genotype.

A volume-selected single free induction decay was acquired using point-resolved spectroscopy sequences (PRESS TE, 16 milliseconds; TR, 2 seconds; spectral width, 4 KHz; 64 averages) for a total time of 2 minutes, after optimizing the magnetic field within each selected voxel individually, to be as uniform as possible (automatic and manual shimming). This was considered to represent signals from the mid-abdomen for the mice. Signals were acquired separately from each phantom.

Image and Spectroscopic Data Analysis

The intra-abdominal fat volume was calculated from the series of coronal images using Amira image analysis software (version 3.0; Mercury Computer Systems, Inc., Chelmsford, MA). To accomplish this, manual delineation of the abdomen was followed by automated segmentation of intra-abdominal fat based on pixel intensity. Visual examination was used to correct for artifacts introduced during automated segmentation. Intra-abdominal adipose tissue cross-sectional area was similarly calculated from an axial image at the level of the fourth and fifth lumbar intervertebral space.

The 1H-MRS signal acquired in the time domain was subjected to Fourier transformation followed by manual phase correction to produce frequency domain spectra for examination. The strongest resonance signals from 1H-MRS of normal biological tissues acquired in this manner are generated by water and lipid, resulting in two primary signal amplitudes (peaks) separated by ∼3.4 to 3.5 ppm (9). Spectral peaks interpreted as representing water were calibrated to 4.7 ppm (15, 16, 18). Alternatively, some spectra were calibrated by setting the lipid peak amplitude at 1.2 ppm. After calibration, spectra were deconvoluted, and a Lorentzian fit was applied. Spectral data analyses were applied to total MR signal affecting an entire spectroscopic histogram with the use of XWIN-NMR software (Bruker BioSpin Corp.). The AUC values for each peak of interest in the spectrum were calculated.

Quality Control

Imaging and spectroscopy phantoms (7.35-mm internal diameter × 54-mm long), composed of water (Ultra-Pure; KD Medical, Columbia, MD) and 20% lipid emulsion in water (Liposyn II 20%; Hospira, Inc., Lake Forest, IL), were positioned directly beneath the mouse abdomen during each data acquisition session. These served as spectroscopy assay standards, with the water phantom generating a single peak and the Liposyn phantom generating both water and methylene proton (lipid) peaks. Phantom spectral data from each acquisition episode were analyzed as described above.

Individual phantom measurements of water and lipid AUC values were compared with the mean AUC measurements from the phantoms over the entire study. Any data point in which an individual phantom measurement was >2 standard deviations from the mean resulted in exclusion of that animal's data from further study. As a result, first imaging session data from Lepob mice nos. 3 and 4 and third imaging session data from A-ZIP/F mouse no. 2 were not studied further.

Owing to variations over the course of the study in 1H-MRS readings from phantom standards, within-session-day phantom readings were averaged and used to normalize each in vivo water and fat reading from each animal according to a formula: mouse AUC × [mean all phantom AUCs (by day)]/(mouse-matched phantom AUC), using the phantom measurements for either water or fat (Liposyn) acquired along with a given mouse's readings, as appropriate. Comparisons of 1H-MRS analyses among animals were made using ratios of the normalized fat and water AUC values from each acquisition.

Statistical Analysis

One-way ANOVA was used to assess the effects of genotype on diverse body composition variables measured by means of 1H-MRS and MRI. Results of ANOVA for genotype, as well as a posteriori comparison of the means using Tukey's honestly significant difference procedure (JMP 5.0; SAS Institute, Inc., Cary, NC), are reported. To account for large differences in variance between genotypes, data were natural log-transformed throughout for the ANOVA. Note that back-transformed data are reported in Table 1. Spearman correlations were calculated between body composition variables overall and within genotypes, using untransformed data. To incorporate information from multiple scans, individual mice were treated as the unit of observation, and scans were weighed accordingly. For example, if a mouse was scanned three times, then each scan received a weight of 0.33. Thus, p values reported for correlation analyses are based on the number of mice scanned, not the number of scans. Finally, in the weight loss induction study, ANOVA was performed using age (days on study) as the independent variable; modest sample sizes precluded use of repeated measures ANOVA.

Table 1.  Mean body weight, intra-abdominal fat volume, and intra-abdominal fat area in three mouse strains differing in body composition
 Mouse strain 
VariableA-ZIP/F (minimal fat)C57BL/6 (lean)Lepob (obese)p
  • *

    Replicate evaluations were performed on each mouse group member. Number indicates total measures made from all mice in the case of each parameter: mass, volume, area.

  • Letters appear in rank order by row, with lowest value indicated by a, highest value by c. Superscript letters indicate means that are significantly different, based on Tukey's honestly significant difference test on natural log-transformed values for all measurements taken from body weight or images.

  • Values in parentheses represent 1 standard error, untransformed.

  • §

    Intra-abdominal fat area at the level of the fourth to fifth lumbar vertebrae.

Mice (N)355 
Replicate measures (N)*81513 
Mass (g)25.1b (0.8)20.2a (0.6)55.7c (0.7)<0.0001
Fat volume (μL)176a (202)723b (148)12,822c (159)<0.0001
Fat area (mm2)§7.0a (13)38b (9)642c (10)<0.0001

Results

Measurements taken from three different mouse strains representing a wide range of body fat compositions included body weight, MRI volume and cross-sectional area of intra-abdominal fat, as well as 1H-MRS spectra from an intra-abdominal ROI. The mice studied had distinctly different body compositions (Table 1). Characteristic 1H-MRS spectra, including individual water and methylene proton (fat) peaks that had peak amplitudes that varied by genotype, were obtained (Figure 1). Additionally, water and lipid emulsion phantoms were incorporated as within-study measurement control standards during each data acquisition episode. Analysis of 1H-MRS spectra from the phantoms by study day revealed coefficients of variation of 12% for water AUC from both the water and lipid emulsion phantoms, and 13% coefficients of variation for lipid AUC from the lipid emulsion phantom. When evaluated over the entire study, the phantom AUCs displayed a trend in readings suggesting temporal variation in phantom values. This variation trend over time led to use of daily phantom standards data, acquired from within the same study day, for normalizing readings from mice.

Segmentation of a series of coronal plane MR images made during each acquisition episode provided three-dimensional models from which to calculate visceral fat volume in mice. This was accepted as the quantity of intra-abdominal visceral fat for purposes of this study. The mean volumes of intra-abdominal fat were unique and stratified according to visible amounts of fat present in the three phenotypically distinct Lepob, C57BL/6, and A-ZIP/F strains of mice (Table 1).

Intra-abdominal fat volume was compared with data derived by other MR methods of assessing body mass or amount of fat within the abdomen. The mean 1H-MRS fat/water ratios obtained from the intra-abdominal ROI for A-ZIP/F (0.05), C57BL/6 (0.34), and Lepob (1.73) mice revealed significant distinctions among the three mouse genotypes (p < 0.05), which were comparable to differences obtained by MR image analyses (Figure 2). Intra-abdominal fat as assessed by 1H-MRS correlated with both intra-abdominal fat volume (r = 0.88, p < 0.001) and body weight (r = 0.83, p < 0.001) across all three genotypes (Table 2). It is noteworthy, however, that there was less strong correlation between intra-abdominal fat quantity and fat assessed by 1H-MRS within certain specific mouse body compositions (Table 2).

Figure 2.

1H-MRS analyses of intra-abdominal visceral fat expressed as mean ratios of the AUCs of fat (methylene proton) to water (hydrogen proton) signals from an intra-abdominal ROI for three mouse strains with differing amounts of body fat. A-ZIP/F mice have minimal body fat, C57BL/6 (trim; i.e., lean) mice have ample fat in the mesentery, whereas Lepob (LEP/OB, obese) mice have grossly abundant intra-abdominal visceral fat depots. The three group means differed significantly from each other. ANOVA on natural log-transformed (Ln) values of AUC ratios was performed using Tukey-Kramer honestly significant difference test (n = 13, p < 0.05).

Table 2.  Correlations between body weight and intra-abdominal body fat measured by MR image analysis and 1H-MRS
 All 3 genotypes (n = 13 mice, 36 scans)*A-ZIP/F (minimal fat) (n = 3 mice, 8 scans)C57BL/6 (lean) (n = 5 mice, 15 scans)Lepob (obese) (n = 5 mice, 13 scans)
Body compositionrprprprp
  • FV, intra-abdominal fat volume (μL); WT, body weight (g); FA, intra-abdominal fat cross-sectional area at the fourth and fifth lumbar vertebrae (mm2); F/W, fat/water ratio calculated from localized intra-abdominal 1H-MRS spectral AUCs (see ′Research Methods and Procedures′).

  • *

    Statistical tests based on weighted analysis with sample size equal to number of mice (see ′Research Methods and Procedures′).

FV vs. WT0.98<0.0010.350.400.78<0.0010.620.023
FV vs. FA0.98<0.0010.300.470.650.080.560.045
F/W vs. WT0.83<0.0010.490.490.590.02−0.300.32
F/W vs. FV0.88<0.0010.280.510.670.0060.250.41
F/W vs. FA0.88<0.0010.780.020.670.0060.220.47

A second MRI method, using a fat area segmented from a single axial MR image, also revealed an intra-abdominal visceral fat quantity that differed significantly according to body composition and genotype of the three strains of mice (Table 1). The intra-abdominal cross-sectional fat area correlated well with the intra-abdominal fat volume across all three strains of mice (r = 0.98). The intra-abdominal fat area and 1H-MRS of visceral fat also correlated overall (r = 0.88), but these correlations were weaker in general within genotypes, particularly within the Lepob and A-ZIP/F mice (Table 2).

The Lepob obese strain of mice was induced to lose and regain body weight over time by administration and withdrawal of recombinant murine leptin. Lepob mice lack production of endogenous leptin, which contributes to their accumulation of adipose connective tissue. Administration of leptin to Lepob mice induced ∼22% reduction of mean body weight (Figure 3A). After the cessation of leptin administration, two of the Lepob mice were removed from the study because of marked appetite suppression (Study Days 28 to 32). Surviving Lepob mice regained lost body weight after leptin was withdrawn and gained additional body weight up to the study conclusion at Day 71 (Figure 3A). Correspondingly, there were significant changes in the amount of abdominal visceral fat lost and regained in response to leptin administration (Figure 3B). These significant differences in abdominal visceral fat volume occurring over time in Lepob mice during the weight loss study were reflected in analyses of the overall means of the fat/water ratios from 1H-MRS spectra (p = 0.006) (Figure 3C).

Figure 3.

Variations in (A) body weight, (B) intra-abdominal fat volume, and (C) intra-abdominal fat measured by localized 1H-MRS in obese (Lebob) mice that were induced to lose and regain body weight after parenteral administration and withdrawal of recombinant murine leptin. Mice were initially given twice daily saline injections beginning Day 1 (white arrowhead). Imaging assessments and measurements were made on Days 1, 8, 15, 22, 28, 35, 43, and 71. Significant changes in body weight (A) (p < 0.001) and intra-abdominal visceral fat volume (B) (p < 0.001) occurred in concert over time after leptin administration, which began on Day 16 (black arrow) and continued twice daily through Day 28 (white arrow). Overall changes in intra-abdominal visceral fat observed (C) were also significant as measured by 1H-MRS (p = 0.006). 1H-MRS measurements of intra-abdominal visceral fat are depicted as the ratio of the AUCs of fat (methylene proton) to water (hydrogen proton) signals from an intra-abdominal ROI (arbitrary units, see “Research Methods and Procedures”). All data are illustrated as group means ± standard deviation by study day. Data for five mice are shown through Study Day 28; thereafter data for three mice are included (see “Results”).

Discussion

Localized 1H-MRS was evaluated as a non-invasive method for analyzing intra-abdominal visceral adipose connective tissue, which is expressly distinct from other body fat depots. Furthermore, an approach was sought that would permit longitudinal estimates of intra-abdominal fat with improved throughput and without the necessity to remove mice from study for analyses. Relative estimates of visceral fat from analyses of localized 1H-MRS spectra generated from the abdominal cavity of mice were compared with intra-abdominal fat volume, and separately with the area of intra-abdominal fat. By making comparisons with intra-abdominal fat volume acquired contemporaneously, it was determined that localized 1H-MRS was particularly useful in specifically distinguishing the amount of intra-abdominal fat when comparing animals with obese, lean, or minimal body fat composition. Correlation coefficients between fat estimated by 1H-MRS and MRI fat volume reflected this.

Total intra-abdominal fat volume, derived using three-dimensional models reconstructed from segmented MR images, was used to represent fat for these comparisons. This was based on the previous acquisition of accurate fat volume from mice and the ability to track changes in body fat depots over time using MRI (13, 22). In previous studies, 1H-MRS measurements of fat within the general abdominal region or whole body correlated well with biochemical analyses of carcass fat (16) or percentage body fat measured by D2O dilution (9). Direct analysis of triglycerides from dog, rabbit, rat, and mouse livers and from muscle of rats and humans similarly correlated with analyses of fat by 1H-MRS (18, 19, 20, 21). Such findings support using localized 1H-MRS from an intra-abdominal ROI to evaluate abdominal visceral fat burden.

The cross-sectional area of intra-abdominal fat from the fourth to fifth lumbar intervertebral axial slice location correlated highly with intra-abdominal fat volume across the range of mouse body fat compositions. A correlation between intra-abdominal fat area and 1H-MRS analyses of intra-abdominal fat across all genotypes was also observed; however, it was somewhat less powerful than the correlation between fat volume and 1H-MRS. Use of the fourth to fifth lumbar intervertebral axial slice location corresponded with the location of the 1H-MRS intra-abdominal voxel ROI in this study. The fat area from a single abdominal axial MR image at the level of the fourth and fifth lumbar vertebrae typically correlates well with total intra-abdominal fat volume in humans (14, 23). It is possible that axial slice locations other than the fourth or fifth lumbar vertebrae may provide preferable comparisons with greater correlations, particularly within specific body conformation types; however, studies comparing more optimal locations for MRI axial slice fat area measurements have not been reported for mice. Measurement comparisons within specific genotypes, in contrast to comparisons across all three genotypes, did not correlate as well in every case. Correlations between fat analyzed by 1H-MRS and by MRI volume within genotypes were lower in Lepob and A-ZIP/F mice than in C57BL/6 mice. Correlations between the area of intra-abdominal fat and intra-abdominal fat volume were weaker within some, but not all, individual mouse genotypes, most notably in the A-ZIP/F mice, which have minimal body fat. Reasonable correlations were observed between area and 1H-MRS for lean mice and for mice with minimal body fat, but not for obese mice.

Lack of a stronger correlation between intra-abdominal fat analyzed by 1H-MRS and the amount of either fat volume or fat area may indicate some limitations in the 1H-MRS measurements made in this study. Limitations may arise owing to differences in density of fat per unit volume conceivably present in the various genotypes. With further optimization in mice, spectroscopic imaging, a method designed to yield both image and spectroscopic data from within the same voxel, may extend correlations between fat content and volume made in this study using localized 1H-MRS. Some compromise in throughput may be necessary, however.

Relatively smaller, but significant, differences in degree of obesity were tracked over time using the localized 1H-MRS methods in the Lepob mice administered leptin. In response to leptin administration and withdrawal, Lepob mice lost, and then regained, significant body weight, similar to findings in a previous report on the effect of leptin (24). These changes were quantifiable by assessment of decreased, followed by increased, visceral adipose tissue volume over the course of the present study. Furthermore, these differences in mean intra-abdominal fat volume within Lepob mice during the study were also statistically significant, as detected by 1H-MRS. In this focus on whether 1H-MRS and MRI could detect changes in intra-abdominal visceral fat during weight loss and regain, analyses centered on the overall pattern of response to leptin treatment. A decrease in fat volume in response to leptin was associated with a transiently increased fat/water ratio occurring on Days 15 to 22. This observation remains unexplained. It is possible that it occurred because of differences in leptin response within subregions of intra-abdominal fat. Alternatively, the possibility remains that there was transient spurious elevation of the ratio, reflecting the propensity of leptin administration to decrease water intake. Studies designed specifically to analyze all possible means comparisons must take into consideration potential influences introduced by mouse strain or numbers used, the character of the ROI selected to sample the intra-abdominal fat, and variation in instrument readings.

Because 1H-MRS can be used to estimate the quantity of body fat (9, 15, 16, 17), it seems likely that refinement of the 1H-MRS method may yield improved predictability when attempting to assess fat burden within heterogeneous abdominal connective tissue and organs. This possibility is particularly relevant when the limitations of image analysis are considered. Segmentation of a series of MR images to calculate the volume of intra-abdominal fat required ∼45 minutes per mouse, whereas analysis of spectroscopic data was accomplished in <2 minutes. Strategies that shorten image acquisition parameters to improve throughput, such as reducing the number of images obtained, decrease the accuracy of fat volume estimation (25). Thus, as in previous studies (15, 17), 1H-MRS provided throughput advantages in acquiring and analyzing fat burden data from these study mice.

The ability to focus MR signal gradients within a tissue ROI presents opportunities to extrapolate from a representative specimen sample to an entire tissue. There are challenges, however, analogous to attempting organ disease diagnosis from a limited tissue biopsy. Another challenge is to address the fact that intra-abdominal fat is not a homogeneous tissue but is integrated throughout the viscera. Selection of the standardized ROI size and location may have affected the results, particularly within a given strain, in part, because the position of abdominal viscera may change and intestinal contents are in flux. Differences in fat topography were apparent in the three mouse strains. These challenges notwithstanding, an ROI was chosen for this study that would permit use of the same voxel size uniformly, regardless of body conformation, across the three mouse strains. Voxel size was dictated by the necessity that it fit within the smallest abdomen studied. Selection of the intra-abdominal voxel location was guided using static anatomic landmarks to focus MR pulse sequences reproducibly on heterogeneous intra-abdominal viscera and fat, to the exclusion of fat or water signal originating from the anatomically identifiable stomach, liver, urinary bladder, and subcutaneous tissues. Further method enhancements may enable improved within-strain correlations between 1H-MRS measurements and the quantity of visceral fat. Such method refinements may include experiments aimed at evaluating alternative voxel positions within the abdomen.

Localized 1H-MRS provides a means to assess intra-abdominal visceral fat non-invasively. Benefits derived include serial measurements, increased throughput compared with image analysis, and less reliance on carcass fat extraction methods or anthropometric assessments. Improvements in nuclear MR instrumentation and technique will likely provide further rationale for the use of 1H-MRS in longitudinal analyses of intra-abdominal fat in individuals. Such 1H-MRS approaches permit comparisons of abdominal visceral adipose tissue burden across experimental models, leading to applications in humans and, ultimately, to elucidation of the relevance of fat burden in diseases.

Acknowledgments

We gratefully acknowledge the critical contributions of Mary Angstadt, Michael Kolf, and Martin Lizak. This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. B.E.W. is an employee of the SAIC-Frederick, Inc., on contract to the National Cancer Institute Intramural Research Program (Award N01-CA-12,400).

Footnotes

  • 1

    Nonstandard abbreviations: MRI, magnetic resonance imaging; 1H-MRS, proton magnetic resonance spectroscopy; ROI, region of interest; AUC, area under the curve; Lepob, B6.V-Lepob/J; A-ZIP/F, FVB-Tg(AZIP/F)1Vsn/J; MR, magnetic resonance; TE, echo time; TR, repetition time.

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Ancillary