Myocardial triglycerides and systolic function in humans: In vivo evaluation by localized proton spectroscopy and cardiac imaging


  • Lidia S. Szczepaniak,

    Corresponding author
    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    2. Department of Radiology, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    • The University of Texas, UT Southwestern Medical Center, Department of Internal Medicine/Hypertension, 5323 Harry Hines Blvd., J4.134, Dallas, TX 75390-8586
    Search for more papers by this author
  • Robert L. Dobbins,

    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    Search for more papers by this author
  • Gregory J. Metzger,

    1. Philips Medical Systems, Dallas, Texas
    Search for more papers by this author
  • Greta Sartoni-D'Ambrosia,

    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    2. Department of Radiology, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    Search for more papers by this author
  • Debbie Arbique,

    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    Search for more papers by this author
  • Wanpen Vongpatanasin,

    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    Search for more papers by this author
  • Roger Unger,

    1. Touchstone Center for Diabetes Research, Dallas, Texas
    Search for more papers by this author
  • Ronald G. Victor

    1. Department of Internal Medicine, University of Texas, Southwestern Medical Center at Dallas, Dallas, Texas
    2. Donald W. Reynolds Cardiovascular Clinical Research Center, Dallas, Texas
    Search for more papers by this author


Recent experimental data suggest that adiposity directly damages the heart by promoting ectopic deposition of triglyceride, a process known as myocardial steatosis. The goal of this study was to develop and validate proton magnetic resonance spectroscopy (1H MRS) as an in vivo tool to measure myocardial lipid content. Complementary studies in rat tissue ex vivo and in 15 healthy humans in vivo provided evidence that 1H MRS constitutes a reproducible technique for the measurement of myocardial triglyceride. In myocardial tissue from Zucker rats, the 1H MRS measurement of triglyceride matched that obtained by biochemical measurement (P < 0.001). In human subjects triglyceride was evident in the hearts of even the very lean individuals and was elevated in overweight and obese subjects. Increased myocardial triglyceride content was accompanied by elevated LV mass and suppressed septal wall thickening as measured by cardiac imaging. Magn Reson Med 49:417–423, 2003. © 2003 Wiley-Liss, Inc.

The prevalence of overweight and obesity is increasing at an alarming rate in all industrialized countries, reaching epidemic proportions in the United States (1–5). The obesity epidemic is responsible for the dramatic escalation in the incidence of type 2 diabetes and is postulated to be a key factor in the recent leveling of the decline in coronary heart disease death rates. However, the underlying mechanisms by which adiposity contributes to heart disease are incompletely understood. In all populations studied, the risks of hypertension, hyperlipidemia, and insulin resistance increase linearly with increasing body mass index (BMI). Thus, the current view is that adiposity indirectly increases cardiovascular risk by enhancing established multiple risk factors.

In contrast to this rather traditional view, recent experimental data prompt the novel hypothesis that adiposity promotes ectopic deposition of triglyceride within cardiac myocytes that may directly damage the myocardium (6). The triglyceride content of nonadipocytes normally is tightly regulated by the oxidation of free fatty acids. When fatty acid oxidation is inadequate, as in the Zucker fatty rat (a genetic model of extreme obesity), cardiomyocytes accumulate large amounts of triglycerides. In the animal model, this cardiac steatosis was shown to trigger a pathological signaling cascade resulting in apoptosis and decreased left ventricular systolic function.

To determine if this basic research can be translated to the clinical setting, we first needed to develop a noninvasive technique to accurately measure myocardial triglycerides in vivo. Here we performed complementary studies in rats and humans to determine whether proton-localized magnetic resonance spectroscopy (1H MRS) can be used to noninvasively investigate cardiac steatosis. Initially, we validated the method in myocardial sections from Zucker rats. Then we performed a series of studies in healthy human subjects to: 1) document the intrasubject reproducibility of the myocardial triglyceride measurement in vivo; 2) examine the cross-sectional relation between myocardial triglyceride content and BMI over a wide range of values; and 3) determine if myocardial triglyceride deposition could be an independent determinant of left ventricular structure and function.


Studies in Rats

Nonsurvival studies on animal subjects were approved by the Institutional Animal Care and Research Advisory Committee of the University of Texas Southwestern Medical Center. The purpose of these studies was to validate 1H MRS measurement of myocardial triglyceride against standard biochemical assay. We tested fragments of myocardial tissue from biventricular septum. Thirteen Zucker rats, 8–14 weeks of age, were sacrificed to study myocardial triglyceride content over a wide range of values. A small piece of ventricular septum (∼15 mg) was trimmed under 10× magnification to remove visible adipose tissue, rinsed in normal saline, and blotted dry to limit contamination from blood or plasma lipids. The cleaned sample and a capillary tube containing the concentration standard (2 mmol of formic acid solution) were placed in a standard 5 mm tube filled with deuterated saline. High-resolution spectra were collected at physiological temperature using 300 MHz Varian Inova system. The single pulse sequence with 90° flip angle of 7 μs, an interpulse delay of 5 sec, acquisition time of 1.8 sec, 4096 data points, a spectral width of 6 kHz, and water presaturation was used for data acquisition. The T1 saturation factor for formic acid was calculated from spectra collected with an interpulse delay of 5 sec and 20 sec and with the same remaining parameters. Immediately upon completion of the magnetic resonance experiment, the tissue sample was frozen and stored in liquid nitrogen for subsequent biochemical measurement of the total myocardial triglyceride content (7). Tissue was powdered under liquid nitrogen and neutral lipids were extracted using chloroform:methanol as outlined by Folch et al. (8). Triglyceride content was measured using standard colorimetric assay (Sigma Diagnostics, St. Louis, MO, procedure no. 337) and was calibrated against a triolein triglyceride standard (9, 10). The assay reaction releases glycerol from triglycerides through the action of lipoprotein lipase and the glycerol was quantified by absorption. The interassay coefficient of variation was 4%.

The signal of methylene protons in the spectrum was selected for quantitative analysis of triglyceride (10). The area under the resonance was measured by line fitting and converted to the tissue triglyceride concentration using Eq [1]. The average number of protons (n = 68) present in methylene groups per molecule of triglycerol in heart tissue was determined from previously published results of chemical analyses (11–15). Concentration of triglycerides from magnetic resonance signal was calculated with the assumption that only triglycerides contribute to that lipid signal. Other lipids from membranes in myocardial cells may contain methylene protons, but they are not recognized by magnetic resonance due to restricted freedom of motion (15):

equation image(1)

where CTG = concentration of triglyceride measured by magnetic resonance spectroscopy [μmol/g]; Qstd = quantity of spectroscopic standard (mg); S = saturation factor; NTG = proton density of methylene groups (68); Nstd = proton density of standard (1); ATG = area under methylene resonance; Astd = area under resonance from standard; Wt = sample weight (g); MWtTg = molecular weight of the triolein triglyceride standard (0.885 mg/μmol).

Studies in Human Subjects

All experimental protocols were approved by the Institutional Review Board at the University of Texas Southwestern Medical Center and all human subjects gave their informed written consent to participate. Fifteen overtly healthy volunteers (seven men, eight women) were recruited with a wide range BMIs. None of the subjects had a history of hypertension, heart disease, or diabetes.

1H MRS Experimental Protocol for the Quantitative Measurement of Myocardial Triglycerides In Vivo

Imaging and localized proton spectroscopy of human myocardium were performed using a 1.5 T Gyroscan NT whole body clinical scanner (Philips Medical Systems, The Netherlands) equipped with a package for localized spectroscopy. A combination of a whole body coil for RF transmitting and a surface coil with a diameter of 17 cm for signal receiving were used. Cardiac and respiratory gating, at the end of systole and at expiration, eliminated artifacts from motion. A 6 cm3 volume was selected within the ventricular septum from cine dynamic images of heart. Double spin echo sequence (PRESS) with echo time of 25 ms was used for spectral localization and data collection (16, 17). Spectra with and without water suppression were collected and proton signal from water in cardiac tissue served as an internal reference for chemical shift offset and fat concentration (9, 18–20). NUTS software (ACORNNMR, Fremont, CA) was used to process data. The areas under signals from water and methylene groups of fatty acids in triglycerides were quantified using a line-fitting procedure and the values were corrected for spin–spin relaxation (9). The content of myocardial triglycerides can be directly expressed as a fat-to-water signals ratio but the values were converted to more familiar units of μmol/g incorporating the following assumptions: 1) Methylene resonance originates in average from 68 protons per molecule of triglyceride. The proton density of methylene groups in myocardial tissue triglycerides can vary from one individual to another and ideally would be measured separately for each subject. Given the risks associated with cardiac biopsy we used average value estimated for animal myocardium (12, 21). 2) Myocardial tissue water content is 73% by weight (22).

With the values listed above, the fat-to-water ratio converts to myocardial tissue triglyceride content in units of μmol/g wet tissue (conversion factor 11.81) using the following equation:

equation image(2)

where CTG = concentration of myocardial triglyceride (μmolTG/g); CW = concentration of water, 55 (mmol water/gwater); NTG = average proton density of methylene groups in myocardial triglyceride, 68 (mmolproton/mmolTG); NW = proton water density, 2 (mmolproton/mmolwater); Tw = myocardial tissue water, 0.73 (gwater/gtissue); F/W = ratio of areas under methylene and water magnetic resonance signals.

Multiple measurements of myocardial triglycerides were performed in a subgroup of subjects. In six subjects myocardial triglycerides were measured twice during the same magnet session. Three subjects returned for an additional, third measurement. The coefficient of variation was calculated as a ratio of SD and mean.

Myocardial Function

Dynamic cine MRI, a 3D technique, enables complete evaluation of heart anatomy and function (23–27). It provides high spatial resolution for the assessment of function and sufficient temporal resolution for isolation of cardiac events, such as end diastole and systole. Dimensions of the left ventricle (LV) were estimated from two cine images in long and in short heart axis views at the level of the mitral valve. LV volume and mass were calculated with the assumption that the LV has an ellipsoidal shape (23–26). LV ejection fraction was calculated as a ratio of stroke volume and LV volume in diastole with stroke volume defined as a difference of LV volume in diastole and systole (23). Regional LV systolic function was monitored through assessment of septal wall thickening during cardiac cycle (27):

equation image(3)

where Th = septal wall thickening; ThES = septal wall thickness at systole; ThED = septal wall thickness in diastole. The concentric remodeling of LV was evaluated by LV concentricity factor calculated as a ratio of LV mass and LV diastolic volume (25).


Ex Vivo Studies in Rats

The main findings from these validation studies are presented in Figs. 1 and 2. Spectra of myocardial triglyceride in sections of biventricular myocardium in fatty Zucker rats are shown in the lower panel of Fig. 1. We detected more intense triglyceride signal in the heart of fatty Zucker rat compared to lean. Electron microscope (EM) images were used for qualitative demonstration of the presence of lipid droplets within myocardial cells. In the upper panel of Fig. 1, EM images of sections of heart tissue from lean and fatty Zucker rats with triglyceride droplets close to mitochondria are clearly visible (6). Although EM images illustrate the presence of triglyceride droplets within myocytes, we preferred to validate spectroscopic measurements of myocardial triglyceride against values obtained by standard biochemical assay. Indeed, over a wide range of values the MRS measurement of myocardial triglyceride content correlated (r2 = 0.94) with the biochemical measurement and reached statistical significance (P < 0.001), as demonstrated in Fig. 2.

Figure 1.

Qualitative illustration of the myocardial triglyceride in sections of tissue from rat heart: Lower panel, high-resolution proton magnetic resonance spectra (1H MRS) from ∼15 mg of myocardial biventricular septum. Upper panel, Corresponding EM with clearly visible triglyceride droplets close to mitochondria within myocardial cell (6). EM images of myocardial tissue from one lean and one fatty Zucker rat were used for qualitative demonstration of the presence of lipid droplets present within myocardial cells. The images of sections of heart were obtained using Philips CM 10 electron microscope. The samples were prepared as described previously (6). Resonances from fatty acids of triglyceride are significantly elevated in concert with increased number of triglyceride droplets in tissue.

Figure 2.

The concentration of myocardial triglyceride was evaluated independently by biochemical assay (y-axis) and by proton MRS (x-axis) in sections of septum (∼15 mg) from Zucker rats. High correlation of both measurements is in agreement with the similar results obtained previously in various tissues (9, 10, 12, 14). Rats were allowed food prior to the experiment.

In Vivo Studies in Humans

Table 1 presents clinical characteristics of the human subjects. By incorporating cardiac and respiratory gating and focusing the spectroscopic volume within the ventricular septum, high-quality spectra were obtained revealing intramyocardial triglyceride resonance at the characteristic chemical shift of 1.4 ppm (Fig. 3) relative to water at 4.8 ppm. When a sample volume within the anterior LV wall was studied, an additional resonance corresponding to epicardial fat at 1.6 ppm was identified (Fig. 4). Similar fat bicompartmentation is well documented in skeletal muscle (9, 18–20). To simplify the analysis and enhance reproducibility, all subsequent experiments were confined to measurement of intramyocardial triglyceride within ventricular septum.

Table 1. Clinical Characteristics of the Study Population
GenderBMI kg/m2Height mWeight kgAge yearsSBP mm HgDBP mm HgHR beats/minEF %PP mm HgCo I/min
  1. BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; EF, ejection fraction; PP, pulse pressure; CO, cardiac output; m, male; f, female.

Figure 3.

Selection of volume of interest (VOI = 6 cc) in the ventricular septum of the myocardium throughout the experiment is indicated by the white box on the cine image (long axis view, end of systole). Next to the image the corresponding magnetic resonance spectrum from VOI is displayed along with enlarged signal from myocardial triglyceride.

Figure 4.

Selection of VOI in the ventricular septum and in the LV wall with corresponding spectra. Spectrum from septum displays single resonance from myocardial triglyceride. Spectrum from left ventricular wall contains additional resonance from epicardial fat.

For the subset of subjects in whom repeated spectroscopy measurements were performed, the coefficient of variation in the measurement of myocardial triglyceride content was 17% (Fig. 5). For the group as a whole, although myocardial triglyceride content increased linearly with BMI (r2 = 0.55), the relation between BMI and myocardial triglyceride was not significant (P = 0.07, data not included).

Figure 5.

a: Illustration of the reproducibility of 1H MRS measurement of intramyocardial triglyceride in vivo. Intrasubject reproducibility is demonstrated by spectra of myocardial triglyceride from three separate measurements in a single subject collected 2 hr apart (gray solid and black dashed lines) and 5 days later (black solid line). b: Intersubject reproducibility is demonstrated by plot of multiple measurements of myocardial triglyceride in several subjects. Y-axis represents myocardial triglyceride measured once and X-axis represents values obtained in second measurement for the same subjects.

As expected, LV mass for each of these clinically healthy individuals fell within established norms. The normal values of LV mass according to the literature (24, 25) are on average 145 ± 65 g, as indicated by gray box in Fig. 6. LV mass of our subjects ranged from 60–209 g, median 127 g, and the average value of 141 ± 51 g. LV mass positively correlated with adiposity measured as BMI (r2 = 0.27, P = 0.045; Fig. 6). The strength of the linear correlation of LV mass, however, was tighter with myocardial triglyceride content (r2 = 0.61, P < 0.001; Fig. 6). When myocardial function was assessed by cardiac imaging, LV ejection fraction was unrelated to myocardial triglyceride content in our subjects (data not shown). Septal thickening, a measure of regional systolic function, was inversely related to septal triglyceride content (r2 = 0.37, P = 0.016; Fig. 7). Note that the only two subjects with septal thickening below the normal range (60 ± 36; (26)) were found to have the highest myocardial triglyceride content. The increased myocardial lipids were also associated with the concentric remodeling of LV, as demonstrated in Fig. 8 (r2 = 0.61, P = 0.00065).

Figure 6.

Myocardial triglyceride correlate positively with both BMI and LV mass. a: Correlation of LV mass with BMI. b: Correlation of LV mass with myocardial triglyceride. Normal levels of LV mass (145 ± 65; (25, 26)) are indicated in gray.

Figure 7.

Myocardial triglyceride is inversely correlated with the regional systolic function measured as septal thickening. Normal range of septal thickening (60 ± 36; (27)) is indicated in gray Interestingly, in two subjects with the highest myocardial triglyceride deposition the septal contractility was challenged the most.

Figure 8.

Concentric remodeling of LV with increased deposits of myocardial triglycerides documented by the relationship of myocardial lipids and concentricity index measured as a ratio of LV mass and volume. Normal levels of concentricity index (1 ÷ 1.3 (25)) are indicated in gray.


The major findings from our experiments are threefold. First, localized proton spectroscopy with cardiac and respiratory gating constitutes a valid and reproducible technique for the noninvasive measurement of myocardial triglyceride in humans. Second, triglyceride is detectable in the myocardium of healthy human subjects, even in those who are very lean. Third, in overweight subjects elevated myocardial triglyceride content is accompanied by increased LV mass and a subtle reduction of septal wall thickening. Thus, 1HMRS is a promising research tool to study the functional effects of myocardial triglyceride accumulation in humans.

The current work builds upon earlier studies that utilized 1H NMR to study myocardial triglyceride in laboratory animals (27, 28). Madden et al. (12) demonstrated that triglyceride content in isolated perfused rat hearts estimated by proton spectroscopy directly correlates with triglyceride content determined by chemical analysis. These initial proton NMR spectroscopy studies of myocardial triglyceride in animal hearts constituted a very important step toward the clinical application of in vivo monitoring of myocardial triglyceride. Extending the technique for noninvasive observation of triglyceride in working human heart is further complicated by artifact arising from cardiac and respiratory motion. den Hollander et al. (17) first demonstrated that utilization of cardiac gating with localized proton spectroscopy allows detection of myocardial triglyceride in vivo. The application of the combination of respiratory gating and cardiac triggering tremendously improved the quality and reproducibility of myocardial triglyceride signal, as demonstrated by Felblinger et al. (16). The current studies combined NMR spectroscopy and cardiac imaging to demonstrate that deposits of myocardial triglyceride were associated with important parameters of cardiac function.

Postmortem studies of human myocardial lipids performed on morbidly obese individuals (30–33) have identified two distinctly different patterns of myocardial fat deposition that could be detected by proton spectroscopy. The first is infiltration of adipocytes from the visceral epicardium to areas between the myocardial fibers, which is similar to the “marbling” present in skeletal muscle. However, while marbling is present even in normal skeletal muscle, fatty infiltration of the myocardium seems to be a pathological process restricted to morbid obesity and involving mainly the right ventricle (33). The second pattern, termed “fatty degeneration” in the old literature, is cardiac steatosis, deposition of triglyceride droplets within the cytosol of the cardiac myocytes (32). Prior to NMR this pathologic condition could only be assessed at autopsy or by intraventricular biopsy. Localized proton spectroscopy permits assessment of cardiac steatosis in vivo. Cardiac imaging combined with myocardial spectroscopy permits examination of associations between myocardial triglyceride levels and cardiac structure and function.

In our experiment, before applying the method to humans, we first performed validation studies in Zucker rats. In tissue sections triglyceride droplets were identified within the cardiac myocytes positioned adjacent to the mitochondria, in close proximity to the enzymes involved in lipid oxidation. By comparing triglyceride values obtained by standard biochemical assays and 1H MRS on myocardial sections excised from fatty and lean Zucker rats of varying ages, we were able to validate the spectroscopic measurement of these intramyocellular triglycerides over a wide range of values.

To establish the spatial resolution and intrasubject reproducibility of in vivo cardiac-and-respiratory-gated localized proton spectroscopy in humans, we studied healthy subjects with BMI ranging from lean to moderately obese. We demonstrated that the technique is sufficiently sensitive to resolve triglyceride accumulated within cardiac myocytes from the large adipocyte pool in the pericardium. When the voxel was placed in the portion of free wall that abuts the visceral pericardium, two triglyceride resonances separated by 0.2 ppm were detected: one characteristic of adipocytes, and a second resonance characteristic of intramyocardial triglyceride. The identity of these peaks was previously established and validated in skeletal muscle tissue (9, 18, 19). When the sample volume was moved to the intraventricular septum, far away from epicardial fat, only a single resonance corresponding to intramyocardial triglyceride was detected. Because the deposition of triglyceride may not be completely homogeneous throughout the myocardium, we used a large sample volume (6 cc) encompassing a substantial portion of the intraventricular septum. This procedure provides a reproducible estimate of the triglyceride content of the septum because the coefficient of variation in the measurement was 17% when individual subjects were studied on multiple occasions. This within-subject coefficient of variation is small relative to the overall range of values for myocardial triglyceride between subjects. Most of the intrasubject variability in the measurement is related to inadvertent subject movement, as the variability was highest in individuals who had difficulty lying still and the lowest in those who stayed motionless (16).

That triglyceride was evident in the hearts of even the leanest subjects suggests that cardiac steatosis may not be restricted to extreme obesity. In our healthy subjects, myocardial triglyceride tended to be higher in the obese individuals, although this tendency did not achieve statistical significance. Thus, the measurement of myocardial triglyceride is likely to produce phenotyping information above and beyond that produced simply by measuring BMI. For example, the traditional notion is that obesity leads to a high cardiac output state to perfuse the excess adipose tissue, with the hyperdynamic circulation leading over time to eccentric LV hypertrophy. In this scenario, depressed myocardial contractility is a very late phenomenon seen with massive obesity. In contrast, in our study the increased LV mass seen with increased myocardial triglyceride was accompanied by both an increased concentricity of LV and a subtle decrease in regional systolic performance. From these initial cross-sectional studies, we clearly cannot determine whether the observed correlations between myocardial triglyceride and LV mass and function are causal. However, these human data are consistent with the lipotoxicity hypothesis of Unger and colleagues (6, 34). According to this theory, the triglyceride content of nonadipocyte cells is tightly regulated by mitochondrial oxidative enzymes. When triglyceride accumulation in myocardial cells surpasses oxidative capacity, the excessive triglyceride is converted to ceramide that activates iNOS, which in turn has been implicated in both hypertrophic signaling (35) and depressed myocardial contractility (36), the latter caused by apoptosis (34, 37).

The present experiments provide proof of concept of 1H MRS as a valid and reproducible clinical research tool to study the determinants and functional importance of cardiac steatosis in humans. The method offers a strategy for early diagnosis of myocardial complications that may occur due to lipotoxic disorders (6, 34, 38). One example is the rapidly growing population of diabetic patients who have increased morbidity and mortality from cardiac disease that is often attributed to greater prevalence of atherosclerotic risk factors that lead to myocardial ischemia. Diabetic cardiac disease, however, cannot be completely explained by traditional risk factors and may be related to derangements of myocardial lipid oxidation (39). Future studies are needed to establish the impact on cardiac steatosis of cardiovascular risks such as age, gender, ethnicity, physical activity, diet, and serum lipids. These initial data set the stage for future prospective studies to determine if interventions that reduce myocardial triglyceride favorably impact LV mass and function.