| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,11Imaging Section, Department of Cardiovascular Medicine, and 2Lerner Research Institute, Departments of Cardiovascular Medicine and Stem Cell Biology, Cleveland Clinic Foundation, Cleveland, Ohio
Submitted 26 October 2006 ; accepted in final form 23 January 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
infarction; collagen; contractility
Although STE was recently validated in a variety of settings (1, 5, 10, 17, 22), the relationship of strain to underlying muscle histology is uncertain. It is unknown as to what measure a decreased strain represents muscle loss, presence of ischemia, or a simple contractility decrease. Finally, the applicability of this method to small-animal echocardiography is unknown.
The aim of this study was to assess sensitivity of circumferential (Scirc) and radial (Srad) midwall strain components to LV remodeling and scarring induced by the myocardial infarction (MI) of the LV in a rat.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Echocardiographic-histological comparisons. The data for these comparisons were gathered from animals that belonged to control groups in three treatment protocols performed at our institution. Although the study design is retrospective to original experimental protocols, the echocardiography data acquisition was performed prospectively according to standard imaging protocol in the same manner in all experiments and was identical in the studies except for the variation in the time between infarct induction and the date of euthanasia (4 vs. 10 wk). Out of these three experiments, we analyzed data sets from a total of 34 rats.
To perform these comparisons, 34 rats underwent ligation of their left anterior descending artery. Echocardiography was performed 1 to 2 days after surgery to confirm the presence of MI. After a period of 4 to 10 wk, after coronary artery ligation (or sham surgery), repeat echocardiography was performed. After this echocardiogram (2 h to 2 days), the animals were anesthetized, a hemodynamic study was performed, and animals were euthanized.
Additionally, we assessed 10 normal rats as control echocardiographic animals and 10 normal rats for control hemodynamic data acquisition.
Rat echocardiography. We used a 14-MHz linear transducer coupled to a Vivid 7 echocardiography machine (GE Medical, Milwaukee, WI) to image rats lightly sedated with ketamine (80 mg/kg ip). M-mode and two-dimensional cross-sectional echocardiography data were acquired at the papillary muscle level and digitally stored for further analysis. End-systolic diameters were calculated from the M-mode data, whereas the segmental end-diastolic wall thickness was calculated from two-dimensional data after dividing the LV into six segments using as anatomic landmarks the two insertion points of the right ventricular free wall and the location of posteromedial and anterolateral papillary muscles (4).
To assess midwall Scirc and Srad and corresponding strain rates, we analyzed two-dimensional high-frame rate (frame rate range 92–123 frames/s) echocardiography data with at least 14 frames collected during each heart beat. Data were analyzed using a speckle-tracking algorithm incorporated into Echopac PC workstation (GE Medical). This recently validated (1, 5, 23) software detects and tracks the ultrasonic interference pattern (speckle) inherent to standard two-dimensional echocardiography after segmenting the ventrcular silhouette into six segments (4). Speckles are discrete hypo- and hyperechoic areas caused by the interference pattern of myocardial tissue inhomogeneities. Speckles continuously appear and disappear during the cardiac cycle due to changing tissue orientation. The size of the speckle is inversely related to the ultrasound frequency. To track speckles, a region of interest is drawn in the end-systolic frame of the cardiac cycle. Software divides the region of interest into subsegment kernels and then identifies individual speckles within every kernel by their position, size, and echogenicity. The software identifies the position of these kernels in adjacent frames by finding the region with the statistically greatest speckle similarity. Appearance and disappearance of speckles within the image are treated by Fourier autocorrelation. Local midwall tissue velocities are calculated as a shift of kernel divided by the time between frames. These two-dimensional velocity vectors are transformed from the global coordinate system of the initial image into a local one, based on the regional wall orientation, to obtain velocities in the circumferential and radial directions. Midwall strain rates are calculated by dividing the difference in velocity between two points (oriented within the local coordinate system) by their distance. Finally, strain is obtained by integrating strain rate between end diastole and end systole (defined as the occurrence of maximal and minimal LV cross-sectional area) (Fig. 1) (2, 11, 26). For each animal, end-systolic Scirc and Srad, and corresponding peak systolic and peak early diastolic strain rates were calculated. Peak early systolic strain rate was defined as a peak strain rate occurring between end diastole and end systole. Peak early diastolic strain rate was defined as the first peak occurring after end systole. Finally, data from at least two distinct cardiac cycles were averaged.
|
Hemodynamic study. Rats were anesthetized by isoflurane (2–2.5%) delivered through a nasal cone. A 2-Fr microtip pressure-volume catheter (SPR-838, Millar Instruments; Houston, TX) was inserted from the right carotid artery and advanced into the LV under echocardiographic guidance (14-MHz linear transducer coupled to either Vivid 7, GE Medical or Applio, Toshiba echocardiographic machines). Data were digitally acquired at a sampling rate of 2 kHz using an RPCU-200 control unit (Millar Instruments) coupled to the National Instruments analog-to-digital converter and a personal computer. The gain and the parallel conductance of the LV volume signal were calibrated by LV end-systolic and end-diastolic volumes calculated by bullet equation from echocardiography data. In six infarcted and nine control animals, the inferior vena cava was exposed by lower right lateral thoracotomy, and the data were also collected during vena cava occlusion. All hemodynamic data were analyzed by HemoSoft software (BioCat, Cleveland, OH). After the end of the experiments, rats were euthanized with a bolus of KCl.
Histological analysis. The hearts were harvested, perfused with 10% formalin under constant pressure for 10 min, fixed in HistoChoice (Amresco, Solon, OH), and sectioned into three equal divisions perpendicular to the LV long axis. The midventricular and apical segments were paraffin embedded according to established protocols. Tissue sections (5 µm thick) were applied to glass slides, which were dried in an oven, deparafinized in xylene, and stained with hematoxylin and eosin and Masson-trichrome stains (Fig. 1). Photomicrographs of each section were acquired using an ArtixScan 4000tf slide scanner (Microtek, Carson, CA) in conjunction with the PathScan Enabler (Meyer Instruments, Houston, TX) at a resolution of 4,000 dpi.
Infarct size was estimated from midventricular LV cross sections by quantitation of fibrillar collagen accumulation stained by Masson trichrome using ImagePro Plus (version 5.1) image analysis software. As the hearts were collected 4 to 10 wk after MI and the anterior wall had significantly thinned, the total fibrosis does not adequately reflect the infarct size induced by ligation of the left anterior descending artery. Therefore, to quantify infarct size, we measured LV circumference composed out of collagen at the epicardium and at the endocardium. After we measured total LV circumference at the epicardium and endocardium, we calculated LV infarct size as follows: Infarct size (%) = 1/2(Circcoll endo/Circtot endo + Circcoll epi/Circtot epi)·100, where Circ signifies circumference, coll denotes collagen, tot denotes total, and endo and epi denote endocardial and epicardial. We used the same software.
Segmental fibrosis was calculated by dividing the LV cross-sectional area at the papillary muscle level into six segments as described (4). Two independent observers graded the amount of fibrosis in percentage, and their values were then averaged. It has been previously shown that this type of fibrosis estimation corresponds well with more quantitative ways of its measurement (27). Finally, to obtain histological wall thickness, we used ImagePro Plus (version 5.1) software to first segment the LV wall as described and then to calculate average segmental wall thickness.
Statistical methods. Data are presented as means (SD). To assess registration of histology and echocardiography data, we performed within-animal correlations of end-diastolic wall thickness determined by echocardiography with both histological wall thickness and with the regional fibrosis in a total of 16 animals with transmural MI.
Statistical analyses for comparison of strains and strain rates between the animals were performed by repeated-measures analysis of variance. To assess predictors of segmental strains and strain rates, a forward stepwise multiple linear regression was used with segmental fibrosis, end-systolic diameter, and infarct age (4 vs. 10 wk) as possible predictors. The standardized 
-coefficients were used as an estimate of the impact of the individual predictors on dependent variable. Finally, for comparison of hemodynamic data, we used a simple unpaired t-test. P value of <0.05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
Figure 2, D and E, shows that peak systolic circumferential and radial strain rates were also significantly smaller in rats with transmural infarct (P < 0.0001 for both), with significant strain rate heterogeneity in infarcted animals (P < 0.001). Similarly, peak early diastolic circumferential and radial strain rates were smaller in rats with transmural infarct (P < 0.0001 for both), with strain rate heterogeneity in infarcted animals (P < 0.001) (Fig. 2, F and G).
Predictors of regional midwall segmental strain after MI.
Multiple regression analysis showed that both global LV dimensions (i.e., LV end-systolic diameter) and regional fibrosis were significant predictors for Scirc and Srad, as well as for corresponding diastolic and systolic strain rates (Table 1), with infarct age having no predictive value. Furthermore, as evidenced by the larger values of for all strain-related parameters, global LV dimension was either similar or a stronger predictor than fibrosis, arguing against the hypothesis that any strain-related parameter is specifically coupled to regional fibrosis. Figure 3, A and B, demonstrates a strong and independent impact of end-systolic diameter and regional fibrosis on regional Scirc and Srad.
|
|
|
Intraobserver variability and registration of echocardiography and histology data. Intra- and interobserver variability data are presented in Table 3. The average within-animal correlation coefficient for the relation between echocardiographic end-diastolic wall thickness and histological wall thickness was r = 0.59 (SD 0.17) (supplemental Fig. 1A; note: supplemental figure may be found through a link in the online version of this article). End-diastolic wall thickness underestimated histological wall thickness by 0.29 mm (SD 0.66) (P = 0.003). The average within-animal correlation coefficient for the relation between end-diastolic wall thickness and regional fibrosis was r = –0.75 (SD 0.14) (supplemental Fig. 1B).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Several articles have assessed the sensitivity of strain (or strain rate) measured either by magnetic resonance or Doppler tissue imaging to detect viable myocardium after MI (3, 28, 30). However, to our knowledge, this is the first study that shows the way that regional fibrosis interacts with LV remodeling to influence segment function. This study extends the findings of Park et al. (19) who showed that fibrosis and end-systolic stress adversely affect diastolic strain rates, although they did not quantify their relative contributions. In the current article we show this interaction in a relatively large series of small animals using a novel noninvasive technology.
Impact of infarct on strain distribution in the rat ventricle. The two fundamental processes that occur after large anterior MI and that are responsible for the occurrence of heart failure, arrhythmias, and death are replacement fibrosis and LV dilation through adverse remodeling. In theory, replacement fibrosis affects strain through loss of contractile elements, whereas remodeling affects strain through increased stress (afterload) (15) and decreased contractility of remaining myocytes (9). In concordance with our study, a recent article has demonstrated that peak-systolic and early diastolic strain rates, which closely correlate with strain, show strong dependence on the development of postmyocardial fibrosis (19).
In contrast, the relationship between end-systolic dimensions and strain post-MI is not well documented. However, this relationship is quite obvious because of the interplay among regional stress, strain, and systolic myocardial stiffness. Regional stress represents the total force in a given direction per unit area normal to that direction (21), whereas systolic myocardial stiffness (i.e., contractility) can be described mathematically as the change in stress divided by the change in strain (13, 16). Thus, with a constant contractility assumed, an increase in stress must lead to a decrease in strain. Stress, in turn, is determined by local radii of curvature (21, 29), which for radial and meridional stress can be approximated by end-systolic diameters (8). As a confirmation, Bogaert et al. (3) used magnetic resonance imaging to show that strains decrease in the remote regions after myocardial infarct and that this decrease is associated with increased circumferential and longitudinal radii of curvature (3). Interestingly, a recent clinical article found no correlation between end-systolic volume and longitudinal strain measured by Doppler tissue imaging post-MI. This possibly reflects limitations of tissue Doppler strain rate measurement or relatively preserved end-systolic volumes after reperfused MI (30). In our article, we clearly show the interdependence of strain and end-systolic diameter. Since systolic cross-sectional LV shape corresponds roughly to a circle, end-systolic diameter is a good approximation of the radius of circumferential curvature, which in turn is a major determinant of systolic stress.
The practical consequence of this finding is that one cannot use a single cutoff value for strain to determine the amount of fibrosis or viability or just as a simple surrogate for the visual assessment of wall motion; rather, any regional strain measure has to be corrected for global end-systolic LV dimensions.
Measuring strain components by STE. Unlike ejection fraction that expresses a deformation (decrease) of ventricular volume as a single number, strain is an antisymmetric 3 x 3 tensor, meaning that it is composed of six interdependent components that reflect tissue deformation occurring along specific axes of a specified coordinate system. STE is a recently developed tool that tracks interference patterns from subwavelength structures to characterize segmental muscle movement (25). STE simultaneously measures strain in two (and potentially three) (12) orthogonal directions from a single echocardiographic view, while using local coordinate systems defined by neighboring ventricular shape. This makes it superior to the strain assessment by Doppler tissue echocardiography, with its known angle dependence and the ability to assess only one component of strain at a time. Several articles validated various strain components by STE against sonomicrometry and magnetic resonance imaging (1, 5, 22). Of note, although in clinical echocardiography frame rates of >90 frames/s often lead to poor speckle tracking (23), they are necessary in rodents due to their high heart rate. Speckle-tracking quality depends on scan line resolution, which in turn depends on transducer frequency, sector width and depth, and the number of crystals within the transducer. Since our 14-MHz linear transducer generates images using high crystal density over a very small sector width and depth, it was possible to obtain good tracking at high frame rates without the loss of scan line resolution.
However, STE expresses LV function in a specific region, not as a single value but as a set of multiple, possibly divergent, measures. Thus the question emerges, Is it necessary to measure more than one component of strain since it is known that these components are cross-correlated? In addition, if the strain values are divergent, which one is the most sensitive to particular physiological process? Our multiple regression analysis data indicate that, in a same segment, circumferential strain is more sensitive to presence of fibrosis, whereas radial strain depends more on LV geometry (a surrogate of end-systolic stress).
Regional strain heterogeneity. We have shown that even normal animals show strain heterogeneity as assessed by STE, with the inferior wall showing the lowest circumferential strain. In contrast, only a nonsignificant trend was noted with radial strain being slightly higher in inferior region. This is possibly due to technical reasons, and further studies are needed to elucidate this. However, Sun et al. (24) and Moore et al. (14) showed that normal volunteers have lower strains in the inferior and septal, as opposed to the anterior and lateral, walls. Since LV structure between mammalian species is essentially identical (6, 7), some strain heterogeneity should also be expected in rodents. This indicates that different "normal" values have to be used for different segments.
Are diastolic or systolic function parameters more sensitive to ventricular structure? Both contractility and relaxation are tightly coupled to ventricular structure and to each other through calcium handling. For example, dobutamine simultaneously increases contractility and decreases the time constant of relaxation. Still, there may be a difference in the sensitivity of contractility versus relaxation parameters. In this article, we show that the presence of chronic MI (and thus of replacement fibrosis) affected the peak-negative LV pressure derivative more than it did peak-positive LV pressure derivative. However, we have not corroborated the findings of Park et al. (19) that early diastolic strain rates (whether radial or circumferential) are more sensitive to fibrosis or LV remodeling than systolic strain rates or strains. A possible reason may be because of a fusion of E and A waves of the strain-rate signal in rats or to a slow frame rate.
Clinical implications. Several articles have stressed the correlation between regional wall motion and strain (5, 11) to facilitate wall motion scoring. This article indicates that specific cutoff values of strain cannot be used as surrogates of "viability" since low strain can occur in the presence of LV remodeling, even in the regions remote from the myocardial infarct. Another finding is that the areas of very prominent negative strain (i.e., dyskinesis) were not frequently observed. This probably reflects the fact that replacement fibrosis has a very high tensile strength, resisting any passive change in systole.
Limitations. Several limitations should be kept in mind when assessing the results of this study. Strain is a three-dimensional tensor, whereas our method quantifies myocardial deformation in a two-dimensional plane. Also, the frame rate relative to heart cycle duration used in this study was lower than in previous ones performed in humans or large animals (1). However, it should be pointed out that in small animals, since the relative duration of LV filling is shorter, systole lasts as long as or even longer than diastole (20). Thus we had seven to eight systolic frames to reconstruct a systolic part of the strain curve, a number that was very similar to the eight frames that were available to us for a similar task in a recent human magnetic resonance tagging study (18). Additionally, because we did not infuse the paraffin into the hearts under end-diastolic pressure and because some distortional effects of dehydration and shrinkage associated with embedding could not be avoided, our histological measures of wall thickness do not exactly reflect in vivo end-diastolic thickness. Still, satisfactory within-animal correlation between histological and in vivo measured wall thickness was obtained.
Depressed strain in remote regions of the infracted animals may be due to ketamine effects. However, ketamine dosage was kept to a minimum with the aim of inducing sedation only. Also, both normal controls and MI groups were treated in the identical manner, and thus differences between groups in noninfarcted segments were obtained under the same physiological conditions. Finally, rats are able to withstand huge myocardial infarcts, and thus, very frequently, the infracted territories in our experiments were large. It is questionable how this relates to human infarcts, especially in the early reperfusion era.
In conclusion, STE can correctly identify segmental LV dysfunction induced by scarring in a rat model of MI. Regional midwall strain is influenced by both global remodeling and regional fibrosis.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  Y. Peng, Z. B. Popovic, N. Sopko, J. Drinko, Z. Zhang, J. D. Thomas, and M. S. Penn Speckle tracking echocardiography in the assessment of mouse models of cardiac dysfunction Am J Physiol Heart Circ Physiol, August 1, 2009; 297(2): H811 - H820. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H.-J. Nesser, V. Mor-Avi, W. Gorissen, L. Weinert, R. Steringer-Mascherbauer, J. Niel, L. Sugeng, and R. M. Lang Quantification of left ventricular volumes using three-dimensional echocardiographic speckle tracking: comparison with MRI Eur. Heart J., July 1, 2009; 30(13): 1565 - 1573. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Marcucci, R. Lauer, and A. Mahajan New Echocardiographic Techniques for Evaluating Left Ventricular Myocardial Function Seminars in Cardiothoracic and Vascular Anesthesia, December 1, 2008; 12(4): 228 - 247. [Abstract] [PDF]
|
|
|
|
|
|
D. J. Kelly, Y. Zhang, K. Connelly, A. J. Cox, J. Martin, H. Krum, and R. E. Gilbert Tranilast attenuates diastolic dysfunction and structural injury in experimental diabetic cardiomyopathy Am J Physiol Heart Circ Physiol, November 1, 2007; 293(5): H2860 - H2869. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. S. Owens and J. F. Plehn Recognizing Unrecognized Risk: The Evolving Role of Ventricular Functional Assessment in Population-Based Studies Circulation, July 10, 2007; 116(2): 126 - 130. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
