INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

LEFT VENTRICULAR HYPERTROPHY is both an adaptive mechanism and a pathological result of many cardiac diseases. Animal models of pressure overload have been developed to study this process. These include surgical manipulation (constriction of the aorta), renal ablation (by excision or drugs), and intravascular volume expansion with high-salt diets (6). These approaches suffer from an inherent dissimilarity to human hypertension that is chronic, usually gradual in onset, and appears to have a large genetic component. Genetic models of hypertension in rodents are available; however, the specific genetic abnormalities have rarely been identified (9).

In contrast, transgenic models of hypertension provide the capacity to analyze the effects of specific genes of interest in a model that may more closely resemble the human disease. One limitation of transgenic analysis is the need to analyze hemodynamics and end-organ pathophysiology in a 30-g animal with a 200-mg heart. The assessment of hypertrophy in mice traditionally has relied on necropsy, increasing the numbers of animals required and preventing serial or functional studies. More recently, echocardiography using high-frequency transducers has proven to be useful in estimating LV mass in mice (4, 12). Magnetic resonance imaging (MRI) as a three-dimensional approach, independent of geometric assumptions, has proven in numerous previous studies to be an accurate technique to quantify left ventricular (LV) mass and function (14). We have extended previous work on cardiac MRI in small animals to the evaluation of LV mass and function in the mouse. We have applied this method to transgenic mice with a deletion of the guanylyl cyclase A (GCA) gene, which produces sustained hypertension, and evaluated the development of cardiac hypertrophy in this model.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animal Protocol

MRI Technique

Multislice, multiphase cine MRI was performed. Each study included a scout, coronal plane long axis of the left ventricle and a set of short-axis acquisitions. Multiframe, short-axis gradient-echo sequences were used to measure LV end-systolic volume (LVESV) and end-diastolic volume (LVEDV) and estimate LV mass. Four or five slices perpendicular to the long axis were obtained for each heart spanning apex to base. The slice thickness was 1.6 mm with a 0.2-mm gap between slices. The matrix was 256 × 256, with a field of view of 50 mm (yielding voxel sizes of 0.19 × 0.19 × 1.6 mm), flip angle of 30°, repetition time of 39 ms, and echo time of 14 ms.

The pulse sequence was set for a heart rate of 250/min with five cardiac phases and a temporal resolution of 39 ms. Depending on the heart rate and R-R interval length, a trigger delay was introduced. Thus end-diastolic and end-systolic images were captured over two contiguous cardiac cycles. The frame with the largest chamber dimensions was used as end diastole for mass and volume measurements, and the image with the smallest chamber volume was taken as the systolic image.

Data Analysis

Image analysis. For LV mass determination, the epicardial and endocardial border of each slice was identified during diastole and traced by hand. The LV wall volume was calculated as follows where the distance between slice center equals slice thickness plus slice gap.

Necropsy analysis. The animals were killed with ether and cervical dislocation, and the ventricles were dissected free of atria, fat, and large blood vessels. During initial experiments, the weight of the right ventricular free wall was found to be <10% of the total ventricular weight. Because variation in right ventricle mass contributed only marginally to total ventricular mass and the dissection of the free wall was prone to error, we included the right ventricle in all necropsy mass determinations.

Statistics

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Each MRI study had an average duration of 60 min. In one early study, images were inadequate for analysis because of motion artifact. One animal died from presumed anesthetic complications during the scan. Figure 1 shows short-axis views in diastole and systole in control and transgenic mice. Correlation between LV mass evaluated by MRI and at necropsy was excellent, with LV_necropsy = 1.04 × LV_MRI + 4.69 mg (r² = 0.95) (see Fig. 2). Bland-Altman analysis (Fig. 3) demonstrated good agreement between MRI and necropsy determinations.

View larger version (158K): [in this window] [in a new window]   Fig. 1.   Magnetic resonance images obtained in control and transgenic mice of similar age (8 mo) and body weight (42 g). Diastolic and systolic images from a guanylyl cyclase A (GCA) +/+ mouse are shown in A and B, respectively. Diastolic and systolic images from a GCA / mouse are shown similarly in C and D, respectively. There is evidence of increased wall thickness in GCA / images.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Linear regression analysis of left ventricular (LV) mass estimated by magnetic resonance imaging (MRI) and LV mass measured at necropsy.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Scatter plots showing mean LV mass by necropsy and MRI (horizontal axes) and difference between necropsy and MRI measurements (vertical axes). Mean difference (solid line) ± 2SD (dashed lines) is shown. There is a good agreement between MRI and necropsy measurements. MRI measurements are slightly and consistently lower than necropsy measurements.

The most likely explanation for the slightly greater mass of hearts at necropsy is our inclusion of right ventricular free wall, which was not dissected away from the LV. There was excellent reproducibility of magnetic resonance measurements with low intraobserver (3 ± 1%) and interobserver (7 ± 6%) variability. The interstudy variability for quantification of LV mass, LVEDV, and LVESV was 1 ± 0.5, 7 ± 4.3, and 3 ± 1.7%, respectively.

LV mass obtained with MRI was significantly greater in transgenic (GCA /) mice than in isogenic controls [GCA /, 226 ± 43 mg (n = 14) vs. controls, 156 ± 14 mg (n = 10); P < 0.0001]. The LV mass index (LVMI) was determined in each animal [LVMI (mg/g) = LV mass determined by MRI (mg)/body weight (g)]. Figure 4 demonstrates the time course for development of cardiac hypertrophy in GCA / mice as determined by MRI. The regression equation for the 14 GCA / mice is Thus, for each month of age, the LVMI increases by 0.137 mg · g body wt¹ · mo¹. The regression equation for the 10 GCA +/+ mice is For this group, the increase in LVMI is much less, only 0.0032 mg · g body wt¹ · mo¹. The t-test performed to compare the two coefficients was not significant (P = 0.252). However, there is a trend present that could reach significance with a larger sample size.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   LV mass index by MRI vs. age for all mice examined in this study. Regression lines for LV mass index vs. age for each group are indicated by solid lines and are further described in text.

LV volumes and EF were calculated for each mouse (Table 1). There were no significant differences between the two groups in LVEDV, LVESV, SV, or EF.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The results of this study indicate that 1) MRI provides an accurate means for the noninvasive estimate of myocardial mass in the intact mouse; 2) MRI can noninvasively detect differences in myocardial mass between wild-type mice and siblings lacking the GCA gene; 3) GCA / mice have increased myocardial mass as compared with GCA +/+ mice, with no significant differences in LV volumes or ejection fraction; and 4) GCA / mice demonstrate a trend toward increasing myocardial mass with age.

Over the last decade, transgenic manipulation of the mouse has allowed dramatic progress in our understanding of cellular and molecular mechanisms of disease. Unfortunately, the application of these techniques to cardiovascular biology has been limited by the difficulty of evaluating cardiac phenotype in an animal with a heart rate of 450 beats/min and LV mass of <200 mg.

Echocardiography has been used extensively in studies of cardiac mass in humans (3) and large animal models (10, 13, 16). In the mouse, with the use of high-frequency transducers (7.5 and 9.0 MHz) (4, 12, 19), there was a good correlation between echocardiographic estimation of LV mass and LV mass at autopsy. Also, echocardiography has been used to evaluate ventricular function in transgenic mice (5). Previous studies performed in rodents to assess LV mass used geometric assumptions to estimate mass (4, 12, 13), and such assumptions may not be valid in hearts with segmental wall motion abnormalities or nonhomogeneous distributions of LV hypertrophy (4).

MRI has also been used extensively in the evaluation of ventricular mass in human (7, 8) and large animal models (16, 20, 23). An important feature of MRI in these studies has been the lack of assumptions regarding LV shape. Application of MRI methods to small animal models has been limited. Rose et al. (15) imaged mouse hearts using high-field magnets (7 and 9.4 T); however, no quantification or functional studies were performed. Recently, Siri et al. (18) performed gated MRI and M-mode echocardiography in normal and hypertrophied murine hearts. Again, imaging was performed at 9.4 T, and LV mass was estimated using a truncated ellipsoid model. LV mass was estimated more accurately by MRI than echocardiography in this study. Our study is the first to perform imaging using a conventional imaging system at 1.5 T, which would facilitate broad application of this method. In addition, we used Simpson's rule to calculate LV mass and volumes, which is not based on geometric assumptions and has been shown to be the most accurate method in other animal models (21).

The MRI assessment of the mouse heart has several limitations. First, the typical anesthetized mouse heart rate is ~450 beats/min, with a typical duration of systole of ~50 ms. Given the temporal resolution of 39 ms used in this study, a trigger delay must be used to ensure that images are obtained at both end diastole and end systole for the assessment of function. If higher temporal resolution becomes available, this will remove the need for multiple scans to determine the appropriate trigger delays. Second, the scan protocol was designed to obtain high spatial resolution and adequate temporal resolution for a moving object with an average epicardial volume of ~300 µl. The LV long axis, measuring 7-10 mm, was spanned with five 1.6-mm slices with a 0.2-mm gap to minimize total scan time. For comparison, determination of myocardial mass in humans using MRI typically uses eight to ten 1-cm-thick slices to span an LV long axis of 8-10 cm. Thus a relatively limited number of slices was used in our study, raising the possibility of partial volume effects. However, the excellent correlation with necropsy results suggests that the effect on the assessment of LV mass in the mouse is limited. Third, the right ventricle was not routinely dissected when necropsy measurements were done. However, the free wall of the RV constitutes substantially <10% of the total ventricular weight, and even a large variation in its mass would have had a negligible effect on the total measurement. MRI cardiac mass measurements were consistently smaller than necropsy, which may be a result of the dissection technique used. Fourth, our determinations of intracavitary volumes and derived stroke volume were slightly larger than a previous invasive study (1), and an invasive validation of LV volumes was not performed. The average body weight of the mice used in our study was higher (35 g) than in the invasive study (25-30 g), which may explain the larger LV volumes. Finally, the requirement for anesthesia limits the evaluation of cardiac function in many animal models; peripheral vasodilation, effects on central sympathetic output, and the direct chronotropic effect of some anesthetics may all modify ventricular function. In addition, no reference measure of function was made using another method, which would be required to validate the assessment of LV function.

The principal value of this study is the demonstration of an accurate, noninvasive, and widely available approach to the assessment of LV mass in transgenic mouse models. In addition, MRI permits the acquisition of functional data only available in vivo; however, further work needs to be done to validate the MRI method against an independent method for assessing function. The ability to assess hypertrophy and function will be valuable in defining the effects of genetic and pharmacological modifications on the heart, especially in longitudinal studies. Also, transgenic technology allows the combination of multiple interacting genetic modifications through simple breeding experiments. For example, mutations conferring an atherosclerotic or diabetic phenotype could be incorporated into the GCA / hypertensive mice, allowing study of a combination of cardiovascular insults commonly found in human patients. Moreover, as inducible and tissue-specific transgenic technologies progress, one can envisage sequential addition of cardiac manipulations, even more closely mimicking the evolution of heart disease in the human population. As cardiac evaluation of the mouse becomes easier, the general utility of this genetically modifiable, inexpensive model system should facilitate rapid progress in our understanding of mechanisms of disease and response to therapy.

Conclusion