Abstract
Echocardiographic techniques are commonly utilized to describe the rodent cardiovascular phenotype. These approaches are contrasted with other in vivo methods and are positioned in the assay selection process by a review of studies from our laboratory and others. Although not conventionally considered a biomarker, the technique has the potential to be exploited as a marker of intentional or unanticipated toxic biological effects in the preclinical development of drugs and chemicals.
Introduction
For technical and economic considerations, rodent models are increasingly used to identify the mechanisms underlying cardiovascular development, function, and disease. However, phenotypic characterization of these small and rapidly beating hearts in vivo necessitate miniaturization and refinement of techniques that are used to study the cardiovascular system in larger animals (Table 1). Selecting the appropriate assay to evaluate the cardiovascular phenotype should not be equated with selecting the most available technique, although the importance of handiness, expediency and cost cannot be minimized.
Several issues should be considered when selecting an in vivo assay (Table 2). First, while accuracy, reproducibility and diagnostic discrimination of the assay are critical factors, they may not be similar among laboratories, particularly when only a few groups have investigated a technique. Second, the effects of anesthesia, mechanical ventilation, and surgical instrumentation may each independently and variably influence the specific physiologic parameter under study, rendering difficult interpretation and extrapolation. Conversely, interpretation of conscious studies not conducive to telemetric monitoring should allow for the effects of restraint and in some instances, recovery from anesthesia. Third, the suitability of a phenotypic assay is influenced by the developmental stage or age of the mouse. For example, technical considerations often preclude accurate echocardiographic studies in fetal and neonatal mice (in this instance, ultrasound biomicroscopy (UBM) with its 30-micron axial resolution would be more appropriate). Fourth, the anticipated phenotype may have an impact on the type of test. For example, the need to make multiple serial assessments often dictates the use of echocardiography; diastolic function is a particularly thorny problem that frequently mandates invasive catheterization. Fifth, the particular strain of mouse may significantly and selectively affect parameters of cardiovascular performance; for example, the C57BL/6J strain has eccentric LV hypertrophy, reduced LV fractional shortening, and increased maximal treadmill exercise times compared to the A/J strain (Hoit et al., 2002). Finally, local expertise, technical difficulty of the procedure, and cost are often paramount considerations. The remainder of this lecture examines in detail examples using echocardiography to assess the rodent cardiovascular phenotype.
Theoretical Background
Ultrasound is a mechanical radiation that can be directed in a beam, conforms to the laws of reflection and refraction, and is reflected by small object. (Feigenbaum, 1994). Ultrasound requires a medium for propagation and travels poorly through air and highly dense substances, such as lung and bone, respectively. When ultrasound traveling in one medium reaches a medium with different acoustic impedance (density), a small percent of ultrasound energy is reflected. The intensity of the reflected energy is directly related to the differences in acoustic impedances of the two media and is used by the scan converter on the ultrasonograph to form an image.
Reflection is also related to the thickness of the target and the angle of incidence between the target and the ultrasound beam. An important physical characteristic of ultrasound is that it is attenuated as it travels through a medium, and this loss of acoustic energy is directly related to the frequency of ultrasound. However, attenuation is generally not an issue in rodents because of the shallow imaging depths that are employed.
Either a single crystal or an array of elements that is either mechanically or electronically steered produces a 3-dimensional ultrasound beam. Focusing the beam, which is accomplished either acoustically or electronically, creates optimal beam intensity. Resolution is the ability to separate two targets as distinct echos; the most important factors influenc-ing resolution are the frequency of the ultrasound beam and the beam width. To the extent that system gain directly affects beam width, gain is a determinant of resolution. Therefore, imaging with high frequency transducers and appropriate use of the instrument’s gain controls are critically important for obtaining accurate ultrasound images.
The Doppler equation relates the velocity (V) of a moving target (e.g., red blood cells) to the shift in frequency (Δf) produced by the moving target, the speed of sound (c), the transducer frequency (ft), and the intercept angle (θ) between the ultrasound beam and the direction of the target:
Thus, the angle between the ultrasound beam and the direction of blood flow should be <20° (i.e., cos θ is near 1) to record accurate velocities, and transducer frequency should be low to record high velocities. Because the optimal recording conditions for Doppler (which resolves best with a parallel beam and low frequencies) are diametrically opposed to those for imaging (which resolves best with a perpendicular beam and high frequencies), optimal imaging may not provide optimal Doppler, and vice versa.
Echocardiography to Assess the Phenotype
Echocardiography is unquestionably the foremost method for imaging the cardiovascular system in small animals. It is noninvasive, versatile, widely available, inexpensive, and well suited for serial studies over a wide range of ages. Using the techniques described in this review, the impact of a drug or toxin on the endocardium, myocardium, and pericardium can be determined in the in vivo context. Thus, valves are assayed for abnormal structure, thickness, and valve dysfunction (regurgitation, stenosis); cavity size, indices of global systolic function (either derived from the isovolumic or ejection phase), regional systolic function (wall motion score), diastolic function (E/A ratios, deceleration time), or combined systolic and diastolic function (myocardial performance index, MPI) are determined; pericardial effusions are readily recognized.
Two-dimensionally directed M-mode echocardiography has been used to phenotype the cardiovascular system in a host of genetic mouse models by many laboratories over the past 10 years (Hoit, 2001a). Broadband transducers operating at 12–15 MHz enhance spatial, contrast, and temporal resolution and near field definition, obviate the need for acoustic standoffs, and facilitate imaging the left atrium and right ventricle (Hoit, 2001). Left ventricular dimensions, function, and mass are reliably measured (Sakata et al., 1998; Kiriazis et al., 2002); however, assessment of the vascular system (like fetal echocardiography) cannot compare with UBM operating at 30–40 MHz. UBM limitations, which are largely related to inadequate temporal resolution, are rapidly being overcome. Importantly, in contrast to UBM, tissue Doppler, color M-mode, radial strain rate analysis, and velocity-vector imaging may be used to analyze cardiovascular function in a relatively load independent manner in rodents. For example, in one study, endocardial and epicardial tissue velocities and strain rate from the posterior LV wall on short axis views of the LV were obtained with low interpretative variability in mice (Sebag et al., 2005). These relatively load independent indices of LV contraction were significantly and linearly related to LV dP/dtmax and decreases were detected earlier than LV shortening fraction in a model of endotoxin-induced LV dysfunction.
Newer generation scanners with high frame rates permit high-resolution real time tomographic imaging in multiple planes, making it possible to calculate two-dimensional LV volume and mass using standard geometric formulae (Kiatchoosakun et al., 2002). Accurate, serial assessment of LV mass is possible with 2-D area-length methods. (Fard et al., 2000) and 3-dimensional echocardiography holds promise (Dawson et al., 2004). These methods eliminate the M-mode echo requirements for uniform ventricular geometry and wall motion; 2-D and 3-D echo accurately quantitate global and regional changes in ventricular volume in a post-infarction model (Scherrer-Crosbie et al., 1999; Dawson et al., 2004). In vivo assessment of ventricular volume permits quantitation of ejection phase indices of cardiac performance such as the ejection fraction and velocity of circumferential fiber shortening. Although noninvasive screening of LV volume and performance by echocardiographic interrogation represents an efficient tool for an initial cardiovascular description, loading conditions and heart rate in addition to myocardial contractility influences these parameters. Thus, it is important to remember that normal resting LV performance can occur despite depressed myocardial contractility by the operation of acute and chronic compensatory mechanisms, and vice versa. For example, overexpression of PKC ɛ produces mild concentric LVH and normal LV systolic shortening, but depressed isolated cardiomyocyte performance (Takeishi et al., 2000).
Wall Motion Score and Myocardial Performance Index in Rats
The rat infarct model has been used extensively because it simulates much of the pathophysiology of LV dysfunction after myocardial infarction in humans (Pfeffer and Braunwald, 1990) and is useful in the evaluation of experimental therapies for heart failure (Pfeffer et al., 1987). A reliable method to assess LV dysfunction in this model is essential because of the high variability in LV remodeling and contractile abnormalities following coronary artery ligation. For this purpose, we validated a 13-segment LV wall motion score index (WMSI) and the myocardial performance index (MPI) in Wistar rats that underwent either left coronary artery ligation or sham operation using 2-D and Doppler flow echocardiography against invasive indices obtained using a high fidelity catheter.
The wall segments were visualized from 2-D images taken from the parasternal long axis and from the basal and mid-papillary short axes. Regional wall motion was graded in each segment as: 1 = normal, 2 = hypokinetic, 3 = akinetic, 4 = dyskinetic and 5 = aneurysmal. Motion in the anteroseptal and posterior wall segments was scored from the clearer of either the parasternal long or short axis, but not both. WMSI was defined as the total of the wall motion scores divided by the number of segments scored. MPI was calculated using color-flow directed Doppler pulsed-wave tracings of mitral and aortic flow measured at the levelof the LV outflow tract (LVOT) from the apical four-chamber. The sample volume was placed between the aortic outflow and mitral inflow, and the waveforms were recorded when the mitral and aortic flows were distinct and both aortic and mitral valve clicks were clearly visible. Three consecutive beats were averaged to calculate MPI, which was defined as the sum of the isovo-lumic contraction and relaxation times divided by the ejection time.
The WMSI and MPI significantly correlated directly with end diastolic pressure (r = 0.72, 0.42; WMSI and MPI, respectively), the time constant of isovolumic relaxation, tau, (r = 0.68, 0.48), and inversely with peak +dP/dt (r = −0.68, −0.50), peak − dP/dt (r = −0.57, −0.44), LV developed pressure (r = −0.58, −0.42), area fractional shortening (r = −0.85, −0.53), and cardiac index (r = −0.74, −0.74). Stepwise linear regression analyses revealed that LV end-diastolic pressure, +dP/dt, area fractional shortening and cardiac index were independent determinants of WMSI (r = .994) and that CI and +dP/dt were independent determinants of MPI (r = .781). We concluded that the 13 segment WMSI and MPI were reproducible and correlated strongly with established echocardiographic and invasive indices of systolic and diastolic function, and could be used as indices of global LV function in the rat infarction model of heart failure
Transesophageal Echocardiography (TEE) in Rats
The ability to serially, accurately, reproducibly, and safely image valvular structures with high spatial and temporal resolution and to assess valvular insufficiency semiquantitatively with spectral and color flow Doppler is a major reason why echocardiography has emerged as the most frequently employed diagnostic modality in cardiology. Although imaging small animals with rapid heart rates is a challenge that has been met by the development of high frequency transthoracic transducers operating at high frame rates, a complete assessment of valvular function and integrity is often impossible with transthoracic transducers.
Transesophageal echocardiography (TEE) is frequently used to assess the structure and function of the cardiac valves in larger animals and man. Recently developed ultrasound tipped catheters have been developed for intracardiac monitoring in man (Packer et al., 2002); the feasibility of these catheters to perform TEE in small animals and infants has recently been demonstrated (Bruce et al., 1999; Kohl et al., 2001). Our experience with one of these catheters (Siemens AcuNav has indicated that rat TEE is feasible and safe, and yields reproducible views of all 4 cardiac valves, transvalvular and venous (caval and pulmonary venous) flows, and long axis views of the left ventricle otherwise difficult to obtain from transthoracic studies. Another group has successfully used this approach safely, and has applied this technique to assist intraoperative coronary ligation and to create a model of chronic mitral regurgitation (Gao et al., 2005).
The technique of TEE in rodents in our laboratory is as follows: rats are anesthetized with isoflurane and placed on a heating pad maintained at 37 °C. The trachea is orally intubated with a 24-gauge catheter, and the animal is ventilated with a small animal respirator. Needle electrodes are inserted in the limbs for an electrocardiogram. A 10 F variable frequency (5.5–10 MHz) phased-array intracardiac ultrasound catheter with full Doppler capability (Siemens AcuNav) is advanced from the esophagus to a position immediately posterior to the left atrium. From this position, small catheter manipulations are made under echo visualization and data are acquired on the ultrasonograph (Siemens Sequoia).
M-mode and 2D images of the mitral and aortic valve, diastolic transmitral, pulmonary venous, and systolic aortic pulsed wave Doppler are routinely recorded and if present, mitral and aortic regurgitation will be assessed using spectral and color flow Doppler (Figures 1–6). However, the use of high frequency Doppler imaging (4–7 MHz) underestimates the color flow jet area and therefore, the severity of valvular regurgitation, which, using jet area alone is semiquantitative at best.
Echocardiographic Evaluation of Diastolic Ventricular Function
The Doppler-determined temporal distribution of left ventricular filling is commonly used to assess LV diastolic function in species larger than mice. The ratio of transmitral filling (i.e., E/A ratios) identifies abnormalities of LV diastolic function related to impaired LV relaxation and reduced ventricular compliance, but alone is insensitive, largely because these ratios are dependent on loading conditions and heart rate. Several indices of diastolic function, each technically challenging, may be useful. For example, in larger animals, the pattern of pulmonary venous flow is coupled with trans-mitral profiles to more accurately diagnose diastolic dysfunction, and the isovolumic relaxation time is a surrogate of LV pressure decay. However, the former varies with changes in atrial pressure and function, and the latter varies not only with LV relaxation rate, but also with the level of aortic diastolic and left atrial pressures, which are often unknown.
Evidence suggests that the color M-mode-determined velocity of early diastolic transmitral flow that propagates apically (Vp) is a relatively load-independent index of left ventricular filling (Garcia et al., 1999) in humans and larger animal models. We recently tested the ability of color M-mode Doppler propagation velocity and the myocardial performance index to detect differences in LV diastolic function in two genetically engineered mouse models that influence calcium (Ca2+) homeostasis (Schmidt et al., 2002).
Recall that Ca2+-sequestration by the sarcoplasmic reticulum (SR) Ca2+-ATPase (SERCA) is a key determinant of myocardial relaxation (Bers, 2001). Transgenic overexpression of the phosphoprotein, phospholamban (PLB) inhibits, and ablation (or knockout, KO) of PLB enhances SR Ca2+ sequestration, resulting in decreased and increased rates relengthening (relaxation) respectively (Kadambi and Kranias, 1997). Transgenic expression of a mutant form of PLB (PLB/N27A, a mutation of amino acid asparagine 27 to alanine) exhibits “superinhibition” of SERCA and results in markedly slowed myocardial relaxation (Zhai et al., 2000).
Ten-to 14 week-old mice expressing the superinhibitory mutant form of PLB (PLB/N27A; n = 12) were examined in parallel with age-matched PLB knockout mice (PLB/KO; n = 12) and controls expressing wild-type (WT) PLB (n = 12). A color Doppler sector map of transmitral inflow was displayed from the most apical window possible, and bisected by the M-mode cursor. The sweep rate was maximized and the baseline was shifted to alias (change colors) at 50% to 75% of the peak early transmitral pulsed Doppler velocity. The propagation velocity, Vp, was measured off-line with commercially available software as the slope of the aliasing velocity.
PLB/N27A mice displayed the characteristic pattern of impaired LV relaxation (i.e., decreased peak E velocity and E/A ratio compared to WT and PLB/KO), and prolonged isovolumic relaxation time (IVRT) compared to WT. In contrast, improved SR Ca2+ reuptake by PLB ablation produced opposite effects (i.e., peak E velocity was increased, IVRT was significantly shorter, and there was a strong tendency for E/A to be higher in PLB/KO compared to WT). Color M-mode Vp was 49.1 ± 1.2 cm/s in WT mice, 22.8 ± 1.7 cm/s in PLB/N27A, and 70.8 ± 10.3 cm/s (both p < 0.05) in PLB/KO. Both isovolumic contraction and relaxation times were significantly prolonged in PLB/N27A compared to WT and PLB/KO mice. Ejection time was also significantly longer in the mutant PLB/N27A mice, due in part to a slower heart rate. Since the increase in ejection time was not as pronounced as the increases in the isovolumic time intervals, the calculated myocardial performance index was significantly higher in PLB/N27A mice than in their WT or PLB/KO littermates. Thus, Doppler filling patterns and intervals provide precise, repeatable measurements of in vivo LV diastolic function with sufficient sensitivity to detect changes in LV diastolic filling in mice with genetically altered SR Ca2+ uptake capacity.
Another relatively load-independent index of myocardial relaxation, early transmitral Doppler tissue velocity, was decreased (as was the ratio of early to late diastolic transmitral tissue velocities) in an aortic banding model of LV hypertrophy and isolated diastolic dysfunction in mice (Schaefer et al., 2003) and in rats (Derumeaux et al., 2002).
Conclusions
Although there are many methods that assess the in vivo murine phenotype, none have the versatility, availability, cost, and ease of echocardiography to evaluate ventricular systolic function, and LV mass and geometry (remodeling). In addition, as an assay, echocardiography may be useful to measure diastolic ventricular function and quantify valvular injury. These methods have important implications for pathologists with an interest in the assessment of chemical and pharmaceutical compounds. For example, Gustafsson and coworkers recently studied the cardiac effects of a bioactive substance administered chronically to rats. Rats were injected subcutaneously with either serotonin or saline control daily for three months. Sixty percent (6/10) of rats receiving serotonin developed pathological echocardiograms; specifically, either isolated or combined aortic and pulmonic insufficiency. Histopathology confirmed thickened and retracted aortic cusps with deposits of collagen tissue and subendocardial carcinoidlike plaques (Gustafsson et al., 2005).