Asynchronous Electrical Activation Induces Asymmetrical Hypertrophy of the Left Ventricular Wall
Background—Asynchronous electrical activation, induced by ventricular pacing, causes regional differences in workload, which is lower in early- than in late-activated regions. Because the myocardium usually adapts its mass and structure to altered workload, we investigated whether ventricular pacing leads to inhomogeneous hypertrophy and whether such adaptation, if any, affects global left ventricular (LV) pump function.
Methods and Results—Eight dogs were paced at physiological heart rate for 6 months (AV sequential, AV interval 25 ms, ventricular electrode at the base of the LV free wall). Five dogs were sham operated and served as controls. Ventricular pacing increased QRS duration from 47.2±10.6 to 113±16.5 ms acutely and to 133.8±25.2 ms after 6 months. Two-dimensional echocardiographic measurements showed that LV cavity and wall volume increased significantly by 27±15% and 15±17%, respectively. The early-activated LV free wall became significantly (17±17%) thinner, whereas the late-activated septum thickened significantly (23±12%). Calculated sector volume did not change in the LV free wall but increased significantly in the septum by 39±13%. In paced animals, cardiomyocyte diameter was significantly (18±7%) larger in septum than in LV free wall, whereas myocardial collagen fraction was unchanged in both areas. LV pressure-volume analysis showed that ventricular pacing reduced LV function to a similar extent after 15 minutes and 6 months of pacing.
Conclusions—Asynchronous activation induces asymmetrical hypertrophy and LV dilatation. Cardiac pump function is not affected by the adaptational processes. These data indicate that local cardiac load regulates local cardiac mass of both myocytes and collagen.
Ventricular pacing causes asynchronous electrical activation of the ventricles.1 In previous canine studies,2 3 we have shown that ventricular pacing decreases fiber shortening, contractile work, myocardial blood flow, and oxygen consumption in early-activated regions and increases these parameters in late-activated regions. The ventricular wall is known to adapt to changes in workload by changing cardiomyocyte size and extracellular matrix composition. These processes are supposed to be regulated by neurohumoral factors4 5 and cardiac load.6 7 8 Although studies on unloaded papillary muscles6 and isolated myocytes7 8 and in mathematical simulations9 support the role of load-regulated growth, it is unknown whether local differences in workload, as in cardiac pacing, result in regional differences in myocardial mass. Neither is it known whether such an asymmetrical hypertrophy, if any, results in changes in left ventricular (LV) performance. The latter has to be considered because ventricular pacing reduces ventricular pump function in the short term.1 2 10 11 12
It was the aim of the present study to investigate the effect of asynchronous electrical activation of the LV on regional geometry and microscopic structure of the LV wall and on global ventricular geometry and performance. To this end, LV dimensions and regional ventricular wall geometry were determined by means of 2-dimensional echocardiography at various time intervals during chronic ventricular pacing (pace group) or during sinus rhythm (sham group). LV function, including LV pressure-volume analysis, was assessed at the beginning and end of the 6-month experimental protocol. Collagen content and myocyte dimensions were determined postmortem in tissue sections from the LV wall.
Animal handling was performed according to the Dutch Law on Animal Experimentation (WOD) and The European Directive for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (86/609/EU). The protocol was approved by the Animal Experimental Committee of the University of Maastricht.
Thirteen adult dogs were premedicated by an intramuscular injection of acepromazine 0.2 mg/kg, atropine 0.1 mg/kg, and oxycodone 2 mg/kg. Anesthesia was induced with thiopental 15 mg/kg IV and maintained by ventilation with halothane (0.75 to 1.5%) in a 1:2 mixture of O2 and N2O. The ECG was recorded from the limb leads.
During sterile surgery, a Medtronic CapSure sp 4423 lead was positioned into the right atrium and a Medtronic 4951 M unipolar lead was inserted with its fishhook tip into the epicardium of the free wall of the LV, 1 cm below the base. This site was chosen because with this electrode position, both early- and late-activated regions could be visualized in 1 echocardiographic cross section. In 8 dogs (pace group, weight 28.9±9.5 kg), a pacemaker (Medtronic Synergist H7027, H7071, Elite II, or Thera DR 7941) was implanted. In 5 dogs (sham group, weight 24.4±2.3 kg, not significantly different from pace group), no pacemaker was implanted, but for assessment of the short-term effects of pacing, the leads were temporarily connected to a pacemaker.
After closure of the pacemaker pocket and the thorax, LV cavity and ascending aortic pressures were measured with a dual-tip micromanometer catheter (Sentron), and cardiac output was measured by thermodilution. LV cavity volume was measured by use of a 12-electrode dual-field conductance catheter (7F, Sentron) advanced into the LV via the left carotid artery, connected with a Leycom Sigma 5DF signal conditioner processor (CardioDynamics).13 Parallel conductance was estimated by injection of 5 mL of hypertonic saline (8%) into the pulmonary artery.13 Ventricular function was estimated from the slope and intercept of the end-systolic pressure-volume relation. Preload reduction, necessary to derive these values, was induced by inflation of a balloon in the inferior caval vein.
Hemodynamics and ECG recordings were made under baseline conditions and 15 minutes after ventricular pacing was initiated. Hemodynamic signals were digitized with 12 bits at 200 Hz by use of a DASH 16 G2 A/D convertor and stored on a personal computer for additional off-line analysis.
In the pace group, ventricular pacing was started ≈2 weeks after implantation, when the dogs had fully recovered from surgery. The heart was stimulated at its own rhythm by AV sequential pacing (DDD mode, upper rate 175 beats/min). The AV stimulation interval was 25 ms to ensure complete ventricular capture. Proper pacemaker function and pacing thresholds were checked regularly and adjusted when necessary.
Two-dimensional echocardiographic images of the LV were made by means of a Hewlett Packard ultrasound system (model 77020A) with a 3.5-MHz transducer (model 21206A) and were recorded on Super VHS videotape. Recordings were made at 0, 0.5, 1, 2, 3, 4, 5, and 6 months after onset of pacing in the paced animals and at 0, 3, and 6 months in the sham animals while they were lying on their right side. Animals were sedated with a mixture of acepromazine (0.2 mg/kg) and oxycodone (1.2 mg/kg). Long-axis images were made as well as parasternal short-axis cross-sectional images, taking care that the LV appeared as circular as possible and that the tip of the papillary muscles and the pacing lead were visible.
After 6 months, the dogs were operated on again, using the same anesthetic and catheter-implantation procedures. Hemodynamic measurements (see above) were performed with the pacemaker still functioning and 15 minutes after the pacemaker had been switched off.
After these measurements were taken, the heart was arrested in diastole by perfusion with ice-cold CdCl2 (0.1 mol/L). The heart was quickly removed, and the LV was weighed. For histological analysis, a transmural tissue block was taken from each heart from the LV free wall, at or near the pacing site, and 1 from the septum, opposite to that site. These blocks were immersion fixed in phosphate-buffered formalin 10% and embedded in paraffin. Morphometry was performed with a Quantimed 570 image analyzer (Leica). In a 5-μm-thick section, stained with a modification of the Azan technique,14 myocyte diameter and area were determined from 100 myocytes for each section by use of a final magnification of ×400. Only those myocytes in which the nucleus was centrally located within the cell were digitized and analyzed to ensure that the short axis of the myocyte was perpendicular to the microscope objective.15 In a 6-μm-thick section stained with Sirius Red16 (Polysciences), the collagen-positive area was determined in 45 fields (magnification ×250), excluding epicardial and endocardial as well as perivascular areas.17 Collagen content was expressed as fraction of the total area examined. The sample size for myocyte diameter and collagen assay was chosen on the basis of a progressive means test, indicating that with the sample sizes used, the mean values were within 3% and 8%, respectively, of the mean value obtained by use of a larger sample.
All histological measurements were performed while the observer was blinded for the experimental group and the wall sector from which the tissue was taken.
Hemodynamic Data Analysis
Hemodynamic data were analyzed off-line by use of software developed in our laboratory. The dedicated data acquisition and analysis software package CONDUCT-PC (CardioDynamics) was applied for conductance catheter–related data analysis. We calculated absolute LV cavity volumes by calibrating systolic conductance changes to stroke volume as determined from thermodilution cardiac output and heart rate.13 The time constant of monoexponential LV pressure decline (τ) was calculated using P(t)=P(0)×exp(−t/τ), where P(t) is LV pressure at time t and P(0) is LV pressure at time LV dP/dtmin.
Determination of Regional Ventricular Geometry
For each measurement, 3 consecutive end-diastolic video images were digitized off-line by use of a video frame grabber (8-bits gray scale, 768×578 pixels, model DT3155, Data Translation, Inc). The digitized images were analyzed by use of NIH Image software (version 1.52) by an experienced echocardiographer who was unaware of the specific time points of the images. Regional geometry (wall thickness and wall volume; see below) was determined within 6 wall sectors, as depicted in Figure 1⇓. Sectors 1 through 3 and 6 are situated at the LV free wall and sectors 4 and 5 at the interventricular septum. In the echocardiographic images, sector 6 was not always clearly visible and therefore was excluded from the analysis. In all animals, the location of the pacing lead fell within wall sector 2, and sector 5 was most remote from this sector.
In the analysis of the 2-dimensional echo images, a total of 50 to 70 contour points on the endocardial, epicardial, and papillary contours were marked manually (Figure 1⇑, A and B). Epicardial and endocardial contour coordinates were converted to a polar representation, with the center of the LV cavity used as the origin and the bisector of the angle between the papillary muscles and the center of the LV cavity used as 0° reference. Inner (ri) and outer (ro) radii in the sectors were determined by fitting the original epicardial and endocardial contour points to a model to limit the highest circular frequency to the fourth harmonic18 : where Rc and Rd are the calculated and measured radii (ri or ro), respectively, and a through h are constants. Wall thickness (WT) of a sector was calculated as WT=ro−ri. Wall sector area (Asector) was derived from the thus obtained ri or ro by integration over each sector (Figure 1D⇑). Sector wall volume (Vsector) was calculated as Asector×rm, assuming that growth in the radial and base-to-apex directions was equal. The median radius (rm)=[(ro2+ri2)/2]0.5. Intraobserver and interobserver variability for the measurements of regional wall thickness were 5.7% and 5.8%, respectively.
Determinations of Global LV Dimensions
Cavity and wall volume of the entire LV were calculated from the cross-sectional images and long-axis dimensions by use of cylinder-ellipsoid model calculations.19 20 We correlated LV wall volume at t=6 months with the gravimetrically determined postmortem LV mass.
Paired hemodynamic data were analyzed by use of a Wilcoxon signed rank test; the Mann-Whitney U test was used to evaluate differences between groups. For morphometric data, the samples were first assessed for normality of distribution by the Kolmogorov-Smirnov test. Then, a nested ANOVA was used.15 ANOVA for repeated measurements was used to evaluate changes of echocardiographic variables during the course of the experiment. If significant differences were found, significant points were isolated by use of Bonferroni-Dunn correction. Data are presented as mean±1 SD. P<0.05 was considered significant.
No dog in this study showed signs of cardiac failure or other illnesses during the entire study period. In all dogs in the pace group, cardiac pacing was possible throughout the study period.
Figure 2⇓ shows representative echocardiographic images of a heart before and 6 months after onset of pacing at the LV free wall. These images illustrate that ventricular pacing leads to global enlargement of the LV cavity and wall, whereas the LV free wall (the early-activated region) becomes thinner and the septum (the late-activated region) becomes thicker.
In the sham group (n=5), LV cavity volume and wall mass remained constant during the experimental period (data not shown). In the pace group (n=8), LV cavity volume and LV mass significantly increased over time (Figure 3⇓). The LV wall–to–cavity area ratio decreased by 7±11% and 10±16% after 1 and 6 months of ventricular pacing, respectively (P<0.05 by ANOVA).
Within 1 month of ventricular pacing, wall thickness tended to decrease in the early-activated LV free wall and to increase in the late-activated septum (Figure 4⇓, top); the LV free wall/septum thickness ratio decreased significantly by 17±12%. Between 1 and 6 months of pacing, this ratio further decreased to 33±15% below baseline owing to a 23±12% increase in septal thickness and a 17±17% decrease in LV free wall thickness compared with baseline. In sham animals, no changes in regional geometry were observed (Figure 4⇓, top).
Compared with baseline, sector volume of the septum was significantly increased by 20±16% after 1 month and by 39±13% after 6 months of pacing, but sector volume of the LV free wall did not change significantly (Figure 4⇑, bottom). Sector volume did not significantly change in regions 1 and 3 (adjacent to the earliest-activated LV free wall region; −0.7±10.4% and 11.1±15.4%, respectively; see Figure 1⇑) but significantly increased in sector 4, adjacent to the most remote septal region (30.3±15.3%).
The echocardiographically determined LV wall volume (LVecho) was highly correlated with postmortem LV weight (LVpostmortem), and the relation could be described by a linear relation: The LV/body weight ratio was significantly larger in pace than in sham animals (6.16±0.85 and 4.91±0.47 g/kg, respectively).
In the pace group, myocytes were significantly thicker in the septum (24.2±2.6 μm) than in the LV free wall (20.6±1.7 μm). In the sham group, myocyte thickness was not significantly different in these regions (22.3±1.9 and 22.3±3.0 μm, respectively; Figure 5⇓). The free wall–to-septum ratio of myocyte diameter was significantly smaller in the pace than in the sham group (0.82±0.07 and 0.99±0.09, respectively).
Ventricular pacing did not influence the myocardial collagen fraction. The collagen fraction in the LV free wall and septum was 4.1±0.7% and 4.6±0.3%, respectively, in the pace group and 4.0±0.8% and 4.1±0.7%, respectively, in the sham group.
Electrophysiology and Hemodynamics
After 15 minutes of ventricular pacing, the duration of the QRS complex more than doubled as compared with sinus rhythm (Table⇓). After 6 months of pacing, the width of the QRS complex further increased significantly by 20±23% of the value after 15 minutes of pacing.
During the implantation procedure, hemodynamics were not significantly different between the sham and pace groups, and the hemodynamic effects of pacing were similar in both groups (Table⇑). Pacing significantly reduced stroke volume index, dP/dtmax, and dP/dtmin and significantly increased heart rate. Pacing increased end-diastolic LV pressure significantly in the pace group, but the increase did not reach the level of significance in the sham group. Systolic LV pressure and cardiac index did not change significantly compared with sinus rhythm (Table⇑). Pressure-volume analysis showed that ventricular pacing significantly increased the slope of the end-systolic pressure-volume relationship but also the volume at which end-systolic LV pressure reached a value of 75 mm Hg (see Figure 6⇓ for examples). Ventricular pacing did not acutely change end-diastolic LV volume (Table⇑).
After 6 months of pacing, hemodynamic variables except for heart rate and end-diastolic LV pressure during sinus rhythm and ventricular pacing were not significantly different during the terminal procedure from the corresponding values during implantation (Table⇑). In both groups, the hemodynamic changes due to the switch from ventricular pacing to sinus rhythm were not significantly different in the implantation and termination procedures.
The findings in the present study demonstrate that long-term asynchronous electrical activation, as induced by ventricular pacing, leads to increased LV cavity volume and wall mass and asymmetrical changes in LV wall thickness. The early-activated regions become thinner and the late-activated regions become thicker. This asymmetry in wall thickness is associated with unchanged sector wall volume in early-activated regions and increased sector wall volume in late-activated regions. The increase in sector wall volume in the late-activated regions results from growth of cardiomyocytes and a proportional increase in collagen content. These data indicate that long-term asynchronous electrical activation induces asymmetrical hypertrophy and ventricular enlargement. Because workload has been shown to be lower in early- than in late-activated regions,21 the findings also indicate that local cardiac load is an important regulator of local cardiac growth.
Asymmetry in hypertrophy is most likely related to pronounced regional differences in contraction pattern during ventricular pacing. In early-activated regions, rapid early systolic shortening is followed by strongly reduced shortening later in systole. In contrast, in late-activated regions, considerable early systolic prestretch is followed by pronounced systolic shortening.2 3 22 Although stretch has been applied frequently to evoke growth responses in isolated myocytes,7 8 the real stimulus for hypertrophy is as yet unknown. In mathematical model studies, Arts et al9 simulated structural adaptation of the LV wall to pressure and volume overload. They proposed that the development of hypertrophy of the entire LV can be explained by local myocyte growth that is regulated by early systolic stretch. On the basis of the local stretch patterns mentioned above, selective hypertrophy and wall thickening in the late-activated septum and the absence of hypertrophy in the early-activated LV free wall are in accordance with the theory of Arts et al.9 More in general, the present data comply with the idea that local cardiac load is an important determinant of local cardiac growth, as proposed by Cooper et al.6 These investigators showed papillary muscle atrophy after cutting its chordae tendineae In the experiments of Cooper et al and in the present study, all myocardial regions were subjected to the same plasma levels of potentially growth-promoting humoral factors like noradrenaline and angiotensin II.4 5 This is important because in many experimental and pathological conditions, potential growth-promoting actions of noradrenaline or angiotensin may have been confounded by their hemodynamic effects (see Reference 2323 for review). The results of the present study do not exclude a role of autocrine or paracrine angiotensin release in mediating stretch-induced myocyte growth.8 21
Locally different growth is demonstrated both echocardiographically and histologically. The relative changes in myocyte diameter, however, appear to be less pronounced than those in wall thickness. Therefore, other factors such as increased myocyte length and hyperplasia may have contributed to macroscopic growth. Although hyperplasia is usually confined to more severe degrees of hypertrophy,24 its presence cannot be excluded in the present study.
The early-activated LV free wall probably becomes thinner owing to LV cavity dilation. This dilation did not occur immediately after onset of pacing but became evident after 1 month of pacing. The cavity dilation may be a secondary stimulus for hypertrophy throughout the LV wall, which may have enforced growth in the late-activated regions and may have prevented atrophy from occurring in the early-activated regions.
The histological measurements indicate that after 6 months of pacing, collagen fractions have not changed. This implies that locally the collagen content increased in proportion with myocyte growth. Hypertrophy with unchanged collagen fractions is also seen with volume-overload hypertrophy.25 In other forms of hypertrophy, however, collagen fractions are increased, presumably owing to high plasma levels of angiotensin or aldosterone.26
LV Pump Function
As shown by others,1 2 10 11 12 27 ventricular pacing acutely reduces global ventricular function. The present study demonstrates that this reduction is similar after 15 minutes and 6 months of pacing. The observation that ventricular function recovers as much when pacing is stopped after 6 months as it decreases when pacing is started during implantation indicates that the myocardium is not failing and that asymmetrical hypertrophy is still compensated.
Pacing with a 30-ms AV interval did not change end-diastolic volume and tended to increase end-diastolic LV pressure (Table⇑), indicating that ventricular preload was not reduced and that the observed decrease in contractility was due to asynchronous electrical activation. Ventricular pacing did reduce early ventricular relaxation, as indicated by the decrease in dP/dtmin and the increased τ.
In the present study, we used AV sequential pacing with a short AV interval (30 ms) to ensure activation of the entire ventricle from the ectopic site. This approach was preferred above induction of AV block, which may cause myocardial damage, potentially interfering with the structural adaptations to be studied. Moreover, the setup used enabled us to study LV function during ventricular pacing and sinus rhythm at termination of the experiment as well. In patients, changing the AV interval from 100 to 130 to 0 ms decreased cardiac output by ≈20%.12 28 29 In a separate series of experiments in 5 AV-blocked dogs, however, we found that switching the AV interval from 100 to 25 to 30 ms changed cardiac output only by −10 to 3% (M. Peschar, MSc, and F.W. Prinzen, PhD, unpublished data, 1997). This is in agreement with the observation of Rosenqvist et al30 that hemodynamic performance in dogs is not influenced when the AV interval is varied between 60 and 150 ms. Therefore, in dogs, pacing at short AV intervals seems to affect cardiac pump function only to a minor degree. Most importantly, even if the short AV interval had impaired global ventricular function, it is very unlikely that it would have caused the asymmetrical hypertrophy as induced by asynchronous electrical activation, the major finding in the present study.
The site of ventricular pacing in the present study (base of the LV free wall) is different from the one used clinically, that is, the RV apex. The LV free wall was chosen as the site of pacing because in this situation, myocardial wall thickness in the early-activated LV free wall and in the late-activated septum could be determined in 1 short-axis echocardiographic image. Despite this difference, the findings in the present study are clinically relevant because comparable degrees of asynchrony are obtained during pacing at the RV apex and at the LV free wall.1 Moreover, in the present study the duration of the QRS complex during pacing was similar to that during RV apex pacing in both dogs30 and humans.31 In addition, in a recent study using MRI tagging, we were able to show that both LV base and RV apex pacing create a more than doubling of the heterogeneity of regional workload compared with atrial pacing. Of course, the sites of early and late activation were at virtually opposite locations during the 2 modes of pacing.22
Our findings are the first to show that in vivo asynchronous electrical activation can lead to locally different growth responses within the same ventricle of adult hearts of a large animal species that are presumably quite comparable to human hearts. It would be of interest to know whether pacing at a site causing less asynchrony, such as the high ventricular septum,30 leads to a lesser degree of asymmetrical hypertrophy, especially because pacing from this site resulted in fewer histological abnormalities than pacing from the RV apex.32
The present study shows that chronic asynchronous activation of the ventricles leads to asymmetrical hypertrophy. This demonstrates that in the LV, local cardiac mass is a function of local cardiac load, which is higher in late- than in early-activated regions, and emphasizes the importance of physiological, fairly synchronous electrical activation of the LV.
This study was supported by a grant from the Bakken Research Center, Maastricht, The Netherlands. The authors are indebted to Ruud Kruger, Theo van der Nagel, and Ferenc van der Hulst for their technical assistance during the experiments and the department of Animal Care of the University of Maastricht (head: A. van de Bogaard) for continued care of the animals. The authors are grateful for the continuous interest and support of Ivan Bourgeois of the Bakken Research Center, Medtronic, Maastricht.
- Received December 18, 1997.
- Revision received February 19, 1998.
- Accepted February 25, 1998.
- Copyright © 1998 by American Heart Association
Prinzen FW, Augustijn CH, Arts T, Allessie MA, Reneman RS. Redistribution of myocardial fiber strain and blood flow by asynchronous activation. Am J Physiol. 1990;259:H300–H308.
Long CS, Kariya K, Karns L, Simpson PC. Trophic factors for cardiac myocytes. J Hypertens. 1990;8(suppl 7):S219–S224.
Cooper G IV, Kent RL, Uboh CE, Thompson EW, Marino TA. Hemodynamic versus adrenergic control of cat right ventricular hypertrophy. J Clin Invest. 1985;75:1403–1414.
Sadoshima JI, Takahashi T, Jahn L, Izumo S. Roles of mechano-sensitive ion channels, cytoskeleton, and contractile activity in stretch-induced immediate-gene expression and hypertrophy of cardiac myocytes. Proc Natl Acad Sci U S A. 1992;89:9905–9909.
Kent RL, McDermott PJ. Passive load and angiotensin II evoke differential responses of gene expression and protein synthesis in cardiac myocytes. Circ Res. 1996;78:829–838.
Park RC, Little WC, O’Rourke RA. Effect of alteration of left ventricular activation sequence on the left ventricular end-systolic pressure-volume relation in closed-chest dogs. Circ Res. 1985;57:706–717.
Burkhoff D, Oikawa RY, Sagawa K. Influence of pacing site on canine left ventricular contraction. Am J Physiol. 1986;251:H428–H435.
Daggett WM, Bianco JA, Powell WJ Jr, Austen WG. Relative contribution of the atrial systole-ventricular systole interval and of patterns of ventricular activation to ventricular function during electrical pacing of the dog heart. Circ Res. 1970;27:69–79.
Baan J, Van Der Velde ET, De Bruin HG, Smeenk GJ, Koops J, Van Dijk AD, Temmerman D, Senden J, Buis B. Continuous measurement of left ventricular volume in animals and humans by conductance catheter. Circulation. 1984;70:812–823.
Urabe Y, Mann DL, Kent RL, Nakano K, Tomanek RJ, Carabello BA, Cooper G IV. Cellular and ventricular contractile dysfunction in experimental canine mitral regurgitation. Circ Res. 1992;70:131–147.
Van Krimpen C, Smits JFM, Cleutjens JPM, Debets JJM, Schoemaker RG, Struyker-Boudier HA, Bosman FT, Daemen MJAP. DNA synthesis in the non-infarcted cardiac interstitium after left coronary artery ligation in the rat: effects of captopril. J Mol Cell Cardiol. 1991; 23:1245–1253.
Marino P, Kass D, Lima J, Maughan WL, Graves W, Weiss JL. Influence of site of regional ischemia on LV cavity shape change in dogs. Am J Physiol. 1988;254:H547–H557.
Gaynor JW, Feneley MP, Gall SA, Maier GW, Kisslo JA, Davis JW, Rankin JS, Glower DD. Measurement of left ventricular volume in normal and volume-overloaded canine hearts. Am J Physiol. 1994;266:H329–H340.
Wyatt HL, Heng MK, Meerbaum S, Hestenes JD, Cobo JM, Davidson RM, Corday E. Cross-sectional echocardiography, I: analysis of mathematic models for quantifying mass of the left ventricle in dogs. Circulation. 1979;60:1104–1113.
Prinzen FW, Hunter WC, McVeigh ER. Mapping of contractile inhomogeneity during ectopic ventricular stimulation using MRI tagging. Circulation. 1996;94(suppl I):I-387. Abstract.
Brilla CG, Pick R, Tan LP, Janicki JS, Weber KT. Remodeling of the rat right and left ventricles in experimental hypertension. Circ Res. 1990;67:1355–1364.