Determination of Right Ventricular Structure and Function in Normoxic and Hypoxic Mice
A Transesophageal Echocardiographic Study
Background—Noninvasive cardiac evaluation is of great importance in transgenic mice. Transthoracic echocardiography can visualize the left ventricle well but has not been as successful for the right ventricle (RV). We developed a method of transesophageal echocardiography (TEE) to evaluate murine RV size and function.
Methods and Results—Normoxic and chronically hypoxic mice (Fio2=0.11, 3 weeks) and agarose RV casts were scanned with a rotating 3.5F/30-MHz intravascular ultrasound probe. In vivo, the probe was inserted in the mouse esophagus and withdrawn to obtain contiguous horizontal planes at 1-mm intervals. In vitro, the probe was withdrawn along the left ventricular posterior wall of excised hearts. The borders of the RV were traced on each plane, allowing calculation of diastolic and systolic volumes, RV mass, RV ejection fraction, stroke volume, and cardiac output. RV wall thickness was measured. Echo volumes obtained in vitro were compared with cast volumes. Echo-derived cardiac output was compared with measurements of an ascending aortic Doppler flow probe. Echo-derived RV free wall mass was compared with true RV free wall weight. There was excellent agreement between cast and TEE volumes (y=0.82x+6.03, r=0.88, P<0.01) and flow-probe and echo cardiac output (y=1.00x+0.45, r=0.99, P<0.0001). Although echo-derived RV mass and wall thickness were well correlated with true RV weight, echo-derived RV mass underestimated true weight (y=0.53x+2.29, r=0.81, P<0.0001). RV mass and wall thickness were greater in hypoxic mice than in normoxic mice (0.78±0.19 versus 0.51±0.14 mg/g, P<0.03, 0.50±0.03 versus 0.38±0.03 mm, P<0.04).
Conclusions—TEE with an intravascular ultrasound catheter is a simple, accurate, and reproducible method to study RV size and function in mice.
Genetically altered mice are of great value to study the role of specific genes in cardiac development, structure, and function. Manipulation of the murine genome can produce ventricular hypertrophy and enhanced or decreased left ventricular function.1 The need for noninvasive methods to study the heart of the mouse is enhanced by the frailty of these animals and the possible variations in hemodynamic parameters produced by anesthesia2 and thoracotomy.3 Noninvasive methods allow serial studies of the heart and assessment of the natural history of cardiac responses to stress. Transthoracic echocardiography has recently been applied to study left ventricular structure and function in mice.4 5 6 7 However, because of its size, location, and unusual geometry, the murine right ventricle (RV) is not well visualized and measured with existing echocardiographic techniques. In mouse models of pulmonary hypertension, a noninvasive method for assessing RV size, weight, and function would be of great benefit.
We developed a transesophageal echocardiographic (TEE) technique to permit us to image the murine RV. The purpose of this study was to determine the feasibility of this method, to test its effect on hemodynamic parameters, and to evaluate its accuracy in assessing RV volumes, function, and myocardial weight in mice under baseline conditions and after chronic hypoxic exposure.
After institutional approval by the Massachusetts General Hospital Subcommittee on Research Animal Care, we studied adult mice of several strains and both sexes weighing 18 to 33 g.
Chronic Hypoxic Exposure
For the chronic hypoxic studies, mice were housed in chambers in which they breathed an inspired oxygen fraction (Fio2) of 0.11 for 21 days. The Fio2 was measured daily with a polarographic electrode.
Mice were anesthetized with intraperitoneal ketamine (100 mg/kg) and xylazine (5 mg/kg) and placed supine on a heated operating table, with a 16-gauge thermistor in the rectum to monitor body temperature. After tracheostomy, an endotracheal tube was inserted and sutured with a 4-0 silk ligature. Volume-controlled ventilation was initiated (respiratory rate of 110 to 120 breaths per minute, Fio2 of 1.0). Airway pressure was continuously monitored. Anesthesia was maintained by intraperitoneal injections of ketamine (100 mg/kg) and xylazine (1 mg/kg). An injection of pancuronium (2 mg/kg) was added for muscle relaxation.
A right carotid artery catheter was inserted (7 mice) to monitor blood pressure. For fluid replacement and RV pressure measurements, a silicone catheter was introduced into the jugular vein (5 mice). The ECG was monitored on the echocardiogram, with 2 electrodes (Red Dot, 3 M) on the upper limbs and 1 electrode on a lower limb.
In the 8 mice in which echocardiographic measurements of the cardiac output were validated, a small midsternal incision from intercostal spaces 2 through 5 was made. A 1-mm ultrasonic flow probe (1RB/T106, Transonic Systems Inc) was positioned around the ascending aorta.
Hemodynamic signals were amplified, transferred to an analog-to-digital converter, displayed on a computer screen, and recorded at 1500 Hz (DI 220, Dataq Instruments).
After placement of all catheters, TEE was performed. Thereafter, the mice were killed with an intravenous injection of 100 mg/kg body wt pentobarbital, the heart was removed, and the RV free wall was dissected free, blotted, and weighed.
Images were obtained with a 3.5F/30-MHz intravascular ultrasound catheter (Sonicath cv, Mansfield, Boston Scientific Corp) and standard echocardiographic system (HP Sonos Intravascular). This system allows 2-dimensional imaging with a maximal frame rate of 30 Hz and an axial resolution of 100 μm at 7 mm.
In Vitro Validation of RV Volume Measurement
Seven normoxic animals were used for the in vitro validation of RV volume. After euthanasia, the heart was removed and agarose was injected via the pulmonary artery into the RV. The heart was stabilized in a saline bath. The probe was positioned along the posterior wall of the heart, mimicking the location of the esophagus. It was then manually withdrawn in measured steps of 1 mm with a scanning duration of 5 seconds for each plane. Four to 6 horizontal planes were obtained. After imaging of the heart, the agarose cast was removed and immersed in water, and its volume was measured by water displacement.
In Vivo Assessment of RV Size and Function
Anesthetized mice were imaged in a supine position. The ultrasound catheter was introduced into the esophagus after the latter was filled with echocardiographic gel. The probe was carefully advanced until the liver was visualized. Instrument settings were optimized and usually consisted of a nonlinear postprocessing curve, maximal gain, and compression ranging between 45 and 55 dB. Image acquisition was initiated and the probe manually withdrawn in measured steps of 1 mm with an imaging duration of 5 seconds at each plane. Four to 6 horizontal planes were obtained.
Validation of Cardiac Output
In 9 anesthetized and ventilated mice (4 normoxic and 5 chronically hypoxic), cardiac output assessed by echocardiography was compared with cardiac output determined by the ultrasonic flow probe.
Validation of RV Mass
A total of 27 anesthetized and ventilated mice (15 normoxic and 12 hypoxic) were studied for assessment of RV weight. Echocardiographic RV free wall mass calculation and wall thickness measurements were compared with true RV free wall weight.
Comparison With MRI
After TEE imaging and euthanasia of the animal, MRI was performed in 1 normoxic mouse with a 2-dimensional multislice spin-echo pulse sequence. All MRIs were acquired in a SISCO (Varian Associates Inc) system equipped with a Nalorac 2.0-T (proton frequency at 84.74 MHz), 18-cm horizontal-bore superconducting magnet. The images were acquired with a 2-cm-diameter receiver coil located at the site of interest and inductively coupled to a single-loop 1-cm-diameter transmitter coil. After the location of the heart was determined, a multislice (n=9) set of images in transverse orientation was acquired in an attempt to mimic those obtained from the TEE. The following parameters were used: repetition time, TR, 2 seconds; echo time, TE, 40 ms; 4 signals acquired per phase-encoding step; receiver bandwidth, 10 kHz; slice thickness, 1 mm (gap, 0.1 mm); field of view, 3×3 cm, with an in-plane resolution of 256×256 pixels. Three measurements of the RV free wall thickness were taken in each plane, toward the base of the heart.
After validation of the method for assessing RV volumes, cardiac output, and RV free wall weight, 6 normoxic and 6 chronically hypoxic age-matched SV 129 mice underwent TEE. In 6 of these animals (3 normoxic and 3 chronically hypoxic), dobutamine (1 μg/g) was administered as an intravenous bolus, and imaging was repeated 2 minutes after the injection. The animals were killed and their RVs dissected free and weighed.
Images were analyzed from the videotapes on an image processor (Hewlett-Packard Sonos 2500). For in vitro RV volume validation, the RV endocardial borders were traced on 1 frame of each plane, resulting in an RV area for each level. RV volume was calculated by Simpson’s method. For in vivo RV volume calculation, to identify end diastole and end systole, the RV endocardial borders were traced on 30 consecutive frames. End-diastolic area was defined as the largest RV area and end-systolic as the smallest. End-diastolic and end-systolic RV volumes (EDRVV, ESRVV) were calculated by Simpson’s method. Stroke volume was defined as EDRVV−ESRVV and RV ejection fraction (RVEF) as (EDRVV−ESRVV)/EDRVV. The echo-derived cardiac output was calculated as stroke volume×heart rate. For the in vivo weight calculation, the RV endocardial borders were traced in diastole (5 measures) and systole (5 measures) in 5 consecutive cardiac cycles in each plane, yielding 10 RV endocardial areas (RVendo). The epicardial borders were traced on the same frames, yielding 10 corresponding RV epicardial areas (RVepi) (see Figure 1⇓). Ten RV free wall areas were then calculated as RVepi−RVendo. The mean of the RV free wall area in each plane was used to calculate the total RV free wall volume by Simpson’s method. RV free wall mass was obtained by multiplying this volume by the specific density of the myocardium. RV free wall thickness was also directly measured; for each plane, 5 measurements were made in diastole and 5 in systole in 5 consecutive cardiac cycles, with the tricuspid valve used as a landmark (Figure 1⇓). The mean value was considered to be representative of RV wall thickness.
To assess the variability of volumes and cardiac output measurements, a total of 210 measurements were performed on 2 mice by 2 observers (M.S.-C. and P.R.H.). A single observer (M.S.-C.) repeated the measurements several weeks later. For the variability of mass measurements, a similar procedure was repeated for a total of 160 measurements in 2 animals. Three animals were selected for assessment of the interobserver variability for the measurements of wall thickness (90 measurements). Interobserver and intraobserver variabilities were calculated as the difference between the 2 observations divided by the means of the observations and expressed as both absolute numbers and percentages.6
All data are presented as mean±SD. Linear regression was used to correlate the echocardiographic findings with in vitro cast volumes, true RV free wall weight, and ultrasonic flow probe–derived cardiac output. The mean difference between echo-derived and true values was also calculated. The error (echo-derived minus actual cast volume, echo-derived minus flow-probe cardiac output, or echo-derived RV mass minus actual weight) was analyzed as a function of true values by the method of Bland and Altman.8 Paired t tests were used to compare baseline and dobutamine-stimulated states and unpaired t tests for baseline and chronic hypoxic states. A value of P<0.05 was considered significant.
In Vitro Validation of RV Volume Measurement
A photograph of an RV cast is shown in Figure 2⇓. The mean volume of 7 casts was 52±20 μL (range, 30 to 80 μL). Echocardiographic measurements were obtainable for all the casts and correlated closely with the volume of the cast (y=0.82x+6.03, r=0.88, P<0.01, Figure 3⇓). The mean difference between true and echo-derived volumes was −3.2±9.4 μL (P=NS). There was no significant relation between volume errors and actual volume by linear regression.
Hemodynamic Effects of the TEE Probe
There was no significant effect of inserting the TEE probe on the heart rate, mean arterial pressure, peak inspiratory pressure, RV pressures, or cardiac output of mice (Table 1⇓).
Feasibility and Interpretative Variability
Technically interpretable studies were obtained in 8 of 9 mice for the validation of cardiac output. The heart could be visualized on the echocardiogram, and a subjective assessment of RV size was made in all mice, but because of inadequate delineation of the epicardium, the RV mass could be measured in only 23 of the 25 surviving mice (92%). The end-diastolic and end-systolic frames of a mouse TEE are shown in Figure 4⇓. The results of interobserver and intraobserver variabilities are summarized in Table 2⇓. The maximum variability occurred for the determination of RV mass (10.4±3.6% for intraobserver and 18.0±25% for interobserver variability).
Validation of In Vivo RV Volumes and Cardiac Output
The mean cardiac output of the anesthetized mice was 7.9±2.7 mL/min as assessed with TEE and 7.4±2.6 mL/min measured with the flow probe (Table 3⇓). The correlation between these 2 measurements was highly significant whether the baseline, dobutamine-enhanced, or combined states were considered (y=1.00x+0.41, r=0.99, P<0.0001 for baseline; y=0.99x+0.67, P=NS for dobutamine; and y=1.00x+0.45, r=0.99, P<0.0001 for the combined state, Figure 5⇓). The mean difference between echo-derived and flow-probe cardiac output was 0.48±0.44 mL/min (P<0.005). There was no significant relation between cardiac output error and flow probe cardiac output by linear regression.
Comparison of TEE With MRI
The RV of the mouse studied with both MRI and TEE was imaged on 4 consecutive horizontal planes with the pullback of the TEE probe and on 4 consecutive planes with the MRI. RV wall thickness measured on the TEE was 0.41±0.11 mm and on the MRI, 0.46±0.10 mm.
Effect of Dobutamine on the Hemodynamic Parameters
Dobutamine injection produced a significant increase in heart rate (597±78 versus 523±74 bpm at baseline, P<0.02), stroke volume (18±4 versus 15±4 μL at baseline, P<0.02), and echocardiographically determined cardiac output (11.1±2.8 versus 7.9±1.8 mL/min at baseline, P<0.0005) (Table 4⇓). There was a trend for RVEF to increase after dobutamine (0.62±0.08 versus 0.53±0.10, P=0.051).
Assessment of RV Weight by TEE
RV free wall mass obtained by echo varied between 10.5 and 28.1 mg (mean, 17.4±5.7 mg). True RV free wall weight varied between 15.8 and 46 mg (mean, 28.7±8.7 mg). The correlation between echo-derived RV free wall mass and actual weight was close (y=0.53x+2.29, r=0.81, P<0.0001, Figure 6⇓), although there was a consistent underestimation of the actual weight by TEE. The mean difference between echo-derived RV mass and RV true weight was 11.27±5.36 mg (P<0.0001). This difference was correlated to RV true weight (−0.47x+2.29, r=0.77, P<0.0001). Mean RV wall thickness measured by TEE was 0.48±0.11 mm (range, 0.34 to 0.72 mm) and correlated well with true weight (r=0.81, P<0.0001).
Effect of Hypoxia on the RV
The total body weight of animals breathing 11% O2 for 3 weeks was significantly less than that of normoxic animals (22.4±3.4 versus 28.8±3.9 g, P<0.05), and their anesthetized heart rate was slower (462±71 versus 570±43 bpm, P<0.05) (Table 5⇓). Mean blood pressure was unchanged in hypoxia (86±23 mm Hg in hypoxia versus 119±37 mm Hg in normoxia). Illustrations of the midventricular TEE planes of a normoxic and a hypoxic mouse are provided in Figure 7⇓. In hypoxia, RV chamber volumes were unchanged in absolute values but were increased when normalized by body weight. Similarly, stroke volume was unchanged in absolute value but was increased when normalized by body weight. RVEF and cardiac output were unchanged. There was a 59% increase in true RV free wall weight in hypoxic mice. Similarly, echocardiographic RV free wall mass was increased by 53% in hypoxic mice. Although not significant, there was a trend for nonnormalized echocardiographic RV free wall mass to be greater in hypoxic mice than in normoxic mice (P=0.054). The RV free wall was thicker in hypoxic animals.
Mice provide a unique laboratory tool to study the roles of various genes on cardiac development and function because of the extensive knowledge of their genome, the development of genetically engineered mice, and their short reproductive cycle. Although transthoracic echocardiography of the mouse has been useful to measure left ventricular dimensions, function,4 5 6 and mass,4 5 the RV is more difficult to image and evaluate. Assessment of the size and function of the murine RV is of particular interest in the study of the pathophysiology of pulmonary hypertension. Because hypoxia-induced pulmonary hypertension has been linked with angiotensin II and nitric oxide,9 10 the development of endothelial nitric oxide synthase (NOS3)–deficient mice11 and of ACE-knockout mice12 may illuminate the mechanisms underlying the development of pulmonary hypertension. Initial results have shown that NOS3-deficient mice have a higher pulmonary artery pressure than wild-type controls.3 These recent advances underline the importance of developing a method for minimally invasive RV function assessment.
In the present study, we report the novel use of catheter-based ultrasound to image and quantify the RV. The catheter was introduced into the esophagus to create TEE images of the right heart. Excellent correlations of RV volumes and function were found compared with cast volumes and directly measured cardiac output, respectively. Echocardiography-derived RV free wall mass and true RV free wall weight correlated closely, although echocardiographic measures underestimated true weight.
TEE in mice has been reported by others, with disappointing results.4 Early attempts used a 20-MHz multielement-array transducer with a maximal depth of field of 4 mm, which was insufficient to visualize both ventricles. In our method, the ultrasound catheter used a single, mechanically rotating 30-MHz crystal, providing better depth of field and resolution. In all but 1 animal, both ventricles were entirely visualized within a depth of 7 mm from the catheter.
Methodological issues were posed by the rapid resting heart rate of our mice. Heart rate varied between 360 and 600 bpm, similar to that of conscious mice.11 Because the RV occupied ≈45° to 60° of the entire 360° sector and the frame rate was 30 Hz, the RV was imaged in 4 to 6 ms. This represents 6% of a cardiac cycle at a heart rate of 600 bpm. Thus, it is likely that during the time it takes to image the RV, there is minimal change in its size. However, the number of frames per cardiac cycle (3 to 5) is insufficient to display the entire cardiac cycle consecutively and to allow clear identification of both end diastole and end systole. We noted a lack of relationship between the mouse heart rate and the frame rate. Thus, during imaging, frames occurred at different points in each cardiac cycle. On the basis of this assumption, we postulated that if 30 consecutive frames were analyzed, 1 of the frames would occur at end diastole and 1 at end systole. The close correlation between echocardiographically estimated cardiac output and flow probe–measured cardiac output supports our measurement convention.
To calculate RV volume and weight, we attempted to mimic 3-dimensional methods, which have shown the greatest accuracy for these calculations in large animals.13 14 Data from every horizontal plane were combined by use of a method of disks. This method is less dependent on defining the specific dimensions of the heart or an assumption of the RV cavitary shape, as would be required for other established 2-dimensional methods.15 In our study, this was particularly important because the orientation of the heart to the esophagus varied considerably among mice. Use of a steerable catheter may lead to an improved ability to obtain reproducible views between mice and further improve the accuracy of RV measurements.16
Despite use of a different ultrasound technique, intraobserver and interobserver variabilities were similar to those reported in studies of the mouse left ventricle measured by transthoracic M-mode echocardiography.5 6
At present, few techniques to measure RV size and function exist. RVEF has been measured by quantitative digital microangiography after intrajugular contrast injection.17 Our measurements of EDRVV were significantly smaller than those measured by angiography.17 Two aspects of the angiography could have resulted in volume expansion. First, the mouse heart rate was slowed to 180 to 200 bpm. Second, RV volumes were measured immediately after injection of a volume load of 0.12 mL of contrast media. With the angiographic contrast technique, quantifiable images were present in 60% of the animals, whereas for our echocardiographic method, images could be analyzed in 89%. Assessment of RV hemodynamics and RV dP/dt is also possible by direct subxyphoidal puncture,9 but this technique does not allow for serial examinations or anatomic imaging.
We identified RV changes induced by chronic hypoxia with the TEE method. Hypoxia increased the echo-derived normalized RV mass and wall thickness. True RV weight increased by 59%, similar to the values reported by others after comparable hypoxic exposure in mice.18 Similarly, echocardiographic measurements of RV mass increased by a mean of 53%. EDRVV and ESRVV were both increased in hypoxic animals. Although not previously reported in mice, RV dilatation has been shown in healthy human volunteers living for 40 days at a simulated altitude of 8840 meters.19 RVEF was unchanged after 3 weeks of hypoxia in these wild-type animals. This is in contrast to the decrease found with radionuclide equilibrium angiography in rats submitted to 2 weeks of comparable hypoxic conditions.20 However, the magnitude of increase of RV systolic pressure reported in those rats (81±4 mm Hg in hypoxic rats versus 34±2 mm Hg in normoxic rats) is much higher than that observed in mice after exposure to 3 weeks of hypoxia (28±1 versus 19±1 mm Hg),21 and this might explain the discrepancy. In our study, cardiac output was unchanged, similar to that reported in rats submitted to comparable hypoxic conditions.22
Although the correlation between true weight and echo-derived mass was close, there was a consistent underestimation of true weight by the echocardiographic measures. Imprecision of pullback steps of the ultrasound catheter could contribute to this underestimation (ie, if a step had been >1 mm but was counted as 1 mm). However, when precise MRI slices of 1-mm thickness were obtained, a number of planes through the RV identical to that on the TEE was found. This suggests that the TEE pullback steps were accurate and equal to 1 mm. Another potential explanation for the weight underestimation by TEE would be underestimation of the wall thickness. However, measurements of the RV wall thickness at the base of the RV by TEE and MRI were comparable. Other factors contributing to the underestimation of RV weight might involve errors in tracing. The pulmonary outflow tract was imaged in all of the mice, but precise tracing of the epicardium in this region was often difficult, as was tracing of the epicardium in the apical region. Indeed, the attenuation of sound waves as they pass through the mitral apparatus often creates a region of acoustic shadowing in the RV apical region. Despite these limitations, relative changes in weight could be accurately tracked by TEE.
There are several limitations to this echocardiographic method. Although the mice do not need to be killed after the procedure, esophageal intubation with the ultrasound catheter is required. This is well tolerated hemodynamically but necessitates anesthesia and tracheal intubation to prevent tracheal collapse. In this feasibility and validation study, the ultrasound probe was introduced after the trachea had been protected by intubation through a tracheostomy. However, tracheal intubation is possible in mice17 and will allow serial TEEs. Further refinements of the technique are thus required. Development of smaller catheters may obviate the need for airway protection. To ensure analysis of end diastole and end systole with the frame rate available on current catheters, 30 frames were analyzed in each plane, which is a somewhat lengthy procedure. Strategies that use newer imaging devices with higher frame rates should provide enhanced temporal resolution.
In conclusion, we have developed a novel echocardiographic method that allows visualization and quantification of RV structure and function in mice. This technique is nontraumatic and accurate. This approach allowed us to study the natural history of RV remodeling in a murine model of chronic hypoxic pulmonary hypertension.
These studies were supported by USPHS grant HL-42397. Dr Scherrer-Crosbie is a Research Fellow supported by the Harold M. English Fellowship (Harvard Medical School). Dr Steudel is a Research Fellow supported by the Deutsche Forschungsgemeinschaft (German Research Association, STE 835/1-2). We are grateful to John Newell for his assistance and advice in the statistical analysis of the results.
- Received January 26, 1998.
- Revision received March 26, 1998.
- Accepted April 1, 1998.
- Copyright © 1998 by American Heart Association
Steudel W, Ichinose F, Huang PL, Hurford WE, Jones RC, Bevan JA, Fishman MC, Zapol WM. Pulmonary vasoconstriction and hypertension in mice with targeted disruption of the endothelial nitric oxide synthase (NOS 3) gene. Circ Res. 1997;81:34–41.
Gardin JM, Siri FM, Kitsis RN, Edwards JG, Leinwand LA. Echocardiographic assessment of left ventricular mass and systolic function in mice. Circ Res. 1995;76:907–914.
Hoit BD, Khoury SF, Kranias EG, Ball N, Walsh RA. In vivo echocardiographic detection of enhanced left ventricular function in gene-targeted mice with phospholamban deficiency. Circ Res. 1995;77:632–637.
Tanaka N, Dalton N, Mao L, Rockman HA, Peterson KL, Gottshall KR, Hunter JJ, Chien KR, Ross J Jr. Transthoracic echocardiography in models of cardiac disease in the mouse. Circulation. 1996;94:1109–1117.
Hoit BD, Khan ZU, Pawloski-Dahm CM, Walsh RA. In vivo determination of left ventricular wall stress-shortening relationship in normal mice. Am J Physiol. 1997;272:H1047–H1052.
Zhu YJ, Kradin R, Brandstetter RD, Staton G, Moss J, Hales CA. Hypoxic pulmonary hypertension in the mast cell-deficient mouse. J Appl Physiol. 1983;54:680–686.
Nong Z, Stassen JM, Moons L, Collen D, Janssens S. Inhibition of tissue angiotensin-converting enzyme with quinapril reduces hypoxic pulmonary hypertension and pulmonary vascular remodeling. Circulation. 1996;94:1941–1947.
Jiang L, Siu SC, Handschumacher MD, Luis Guererro J, Vazquez de Prada JA, King ME, Picard MH, Weyman AE, Levine RA. Three-dimensional echocardiography: in vivo validation for right ventricular volume and function. Circulation. 1994;89:2342–2350.
Levine RA, Gibson TC, Aretz T, Gillam LD, Guyer DE, King ME, Weyman AE. Echocardiographic measurement of right ventricular volume. Circulation. 1984;69:497–505.
Rockman HA, Ono S, Ross RS, Jones LR, Karimi M, Bhargava V, Ross J Jr, Chien KR. Molecular and physiological alterations in murine ventricular dysfunction. Proc Natl Acad Sci U S A. 1994;91:2694–2698.
Klinger JR, Petit RD, Curtin LA, Warburton RR, Wrenn DS, Steinhelper ME, Field LJ, Hill NS. Cardiopulmonary responses to chronic hypoxia in transgenic mice that overexpress ANP. J Appl Physiol. 1993;75:198–205.
Hurford WE, Crosby G, Strauss HW, Jones R, Lowenstein E. Ventricular performance and glucose uptake in rats during chronic hypobaric hypoxia. J Nucl Med. 1990;31:1344–1351.