| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of 1Bioengineering and 2Medicine, University of California, San Diego, La Jolla, California
Submitted 16 November 2006 ; accepted in final form 17 April 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
ventricular pacing; mechanical activation; fiber stretch; dyssynchrony
Thus the aim of the present study was to test the hypothesis that the late-activated regions of the left ventricle distend in the same manner as passive filling, and that the functional consequences of prestretch are the result of changes in length-dependent activation. To achieve this, midwall and transmural arrays of radiopaque markers and bipolar electrodes were implanted in the left ventricular (LV) free wall of open-chest dogs to describe myofiber deformation at high temporal and spatial resolution during atrial and ventricular activation.
In this study we show that, in contrast to passive distension, prestretch is uniform transmurally. The majority of prestretch occurs during isovolumic contraction, after the prestretched tissue has been electrically activated and the mitral valve has closed. Both of these findings support the conclusion that prestretch occurs in tissue that is actively generating force. We confirm that, at aortic valve opening, late-activated tissue is at a longer muscle length than early activated tissue. This dyskensis is widely thought to produce the decrease in function observed in epicardial pacing. However, at least at short atrioventricular (A-V) delays, the muscle length in late-activated tissue is not significantly different from normal contraction, whereas early activated fibers are shorter. This not only provides insight into the decrease in function occurring with epicardial activation but also emphasizes the importance of A-V delay for clinical approaches such as cardiac resynchronization therapy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Surgical preparation. Ten adult mongrel dogs (21–24 kg) were anesthetized with intravenous propofol (6 mg/kg), intubated, and mechanically ventilated with isoflurane (0.5–2.5%), nitrous oxide (2 l/min), and medical oxygen (2 l/min) to maintain a surgical plane of anesthesia. Five dogs were implanted with midwall marker arrays, and five were implanted with transmural marker arrays. All underwent the same surgical procedure, as follows. The heart was exposed via a median sternotomy and a left thoracotomy at the fourth intercostal space and then placed in a pericardial cradle. An 8-Fr pigtail micromanometer catheter with a lumen (Millar Instruments, Houston, TX) was inserted through a 9-Fr arterial introducer placed in the left femoral artery, and the catheter tip was advanced into the LV. LV pressure was recorded with the pigtail micromanometer catheter, and the calibration was adjusted such that the end-diastolic pressure and peak LV pressure matched those recorded from a fluid-filled transducer connected to the lumen of the catheter. LV end-diastolic pressure (LVEDP) was calculated by finding dP/dtmax and working backward in time until dP/dt reached 5% of the maximum value. In the case of atrial pacing, this coincided closely with the notch in the pressure trace and the peak of the R-wave on the electrocardiogram (ECG). A flow probe (model #T208; Transonic Systems, Ithaca, NY) was placed on the ascending aorta to assess the onset and end of ejection. Stroke volume was obtained by integration of the aortic blood flow signal.
In one group of five dogs, radiopaque markers were used as follows to measure two-dimensional myocardial deformation in each heart. Lead beads (2 mm diameter) were sutured to the apical dimple (apex bead) and on the epicardium at the bifurcation of the left anterior descending (LAD) and left circumflex coronary (base bead) arteries to provide end points for a LV long axis. Gold (0.9 mm) or lead (1.2 mm) beads were plunged in the anterolateral wall in a rectangular pattern, forming a surface approximately parallel to the epicardium (see Fig. 1).
|
Experimental protocol.
Atrial pacing was performed by stimulating LA electrodes, and LV pacing was performed by stimulating both LA and LV electrodes (A-V delay = 35 ms), using a square-wave constant-voltage electronic stimulator (model SD9; Grass Instruments, Quincy, MA) at a frequency 20% above baseline heart rate to suppress native sinus rhythm. Stimulation parameters (voltage 10% above threshold, duration 8 ms, and frequency) were kept constant in each animal. Each animal was positioned in a biplane radiography system, and synchronous biplane cineradiographic images (125 frames/s) of the bead markers were digitally acquired with mechanical ventilation suspended at end expiration. The LV pressure, arterial pressure, surface ECG, and bipolar plunge electrograms were recorded simultaneously with the cineradiographic images. Recordings were obtained during atrial pacing and then following 
2–5 min of ventricular pacing (to achieve a steady hemodynamic state). Atrial pacing was continued between ventricular pacing runs.
Histology.
After the functional data acquisition was complete, a 22- to 26-Fr cannula with side holes was inserted in the ascending aorta through a brachiocephalic artery, the animals were killed with pentobarbital sodium, and the heart was brought to arrest. The coronary arteries were perfused with normal saline, and the LV pressure was adjusted to the end-diastolic pressure observed during the study before fixing the heart with 2.5% buffered gluteraldehyde, as described previously (1). The fixed, Silastic-filled heart was skewered from the apical dimple to the mitral aspect of the aortic valve to define the LV long axis (18). It was then sectioned as shown in Fig. 2, yielding a total of 25 transmural samples from within the bead array. The transmural thickness of the block was measured, and the block was sliced into 1-mm-thick sections parallel to the epicardial tangent plane, forming a series of sections from the epicardium to the endocardium used to measure the fiber angles across the LV wall. After measuring all of the angles in one transmural section, least-squares linear regression was used to fit the transmural data. This process was repeated in each of the samples, for use in the strain analysis described below.
|
|
50 triangles (functional data points) per array. The coordinate axes were defined individually in each triangle, using the following method. First, the outward-facing normal vector (local radial direction) of the triangle was determined. Next, the circumferential direction was determined by finding a vector perpendicular to both the radial direction and the LV long axis, pointing toward the septum. Finally, the longitudinal direction was determined by finding a vector perpendicular to the radial and circumferential directions, pointing toward the apex. This process ensures that the circumferential and longitudinal directions lie in the plane of the triangle (see Fig. 4).
|
at the location of that triangle's centroid. This angle was calculated as a weighted average of the three nearest measurements to the centroid (weighting inversely proportional to distance). Thus the fiber and cross-fiber strains within each triangle were found. Three-dimensional triangle centroid locations were then projected on a two-dimensional grid such that the plane of the grid was perpendicular to the radial direction of an arbitrary triangle located near the center of the array. By representing all the strains from a single time point at their respective centroid locations on the grid, a spatial distribution of strain could then be fit using the Matlab function griddata (The Mathworks, Natick, MA) and visualized as a color map. Cubic interpolation was used to ensure smooth gradients between adjacent strain measurements.
Areas were considered prestretched if the maximum fiber strain before aortic valve opening (relative to the pacing stimulus) exceeded 0.02, the approximate resolution limit of the technique. Aortic valve opening strain was defined as the strain at the time of aortic valve opening during ventricular pacing with respect to that same hemodynamic reference during atrial pacing. This quantity was only defined in areas that were prestretched with ventricular pacing. Similarly, ejection strain was defined as the strain at the time when the aortic valve closes with respect to the time it opens, in the same areas used to define aortic valve opening strain.
Transmural arrays. A second group of five dogs was used to measure three-dimensional myocardial deformation. The same surgical preparation was used, except that the midwall bead array was replaced with three transmural columns of four to five 0.8-mm-diameter gold beads and a 1.7-mm-diameter surface gold bead above each column. The beads were placed within the anterior wall between the first and the second diagonal branches of the LAD. The experimental protocol was also the same, with the addition of epicardial pacing at the site of the array. The data analysis was performed by determining a continuous polynomial position field that mapped the beads in the undeformed reference configuration to those in the deformed configuration. Differentiation of this field with respect to undeformed position yields a continuous, nonhomogeneous transmural distribution of three-dimensional finite strains (1, 5).
Statistical analysis. Values are means ± SD unless otherwise specified. A one-factor repeated-measures ANOVA was used to assess the effect of time during the cardiac cycle on fiber strain. Tukey's test was used for post hoc comparisons. For aortic valve opening strains, a one-sample t-test with two-tailed P value was used. For all other statistics, a paired t-test with two-tailed P value was used. Statistics were performed using SigmaStat version 3.0 (SPSS, Chicago, IL). Significance was accepted at P < 0.05.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
To understand the deformation data in the context of the fiber anatomy, fiber angles throughout the array were measured histologically. At the depth of the array, myofibers were found to be oriented at an average of +27.5 ± 22.1° from circumferential, consistent with previous findings of fiber angle at this depth when compared at apical (1, 5) and basal sites (5). The average range of fiber angle within the array was 72.0 ± 5.0°.
Timing of prestretch. Figure 3 shows a typical time course of fiber strain in one triangle and the bipolar electrogram of the nearest electrode. Fiber strain data from the whole array for this animal are shown in Fig. 5, where the location of the triangle shown in Fig. 3 is outlined. The strain maps shown in Fig. 5 were obtained at the times marked by vertical lines in Fig. 3. Referring to Fig. 3, 85% of the prestretch occurs between end diastole (dashed line) and aortic valve opening (solid line); it peaks 59 ms after the time of local activation.
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
20–25% of the LV (n = 5) and transmural arrays (n = 5). The results show that prestretch mostly occurs after time of end-diastolic pressure and differs from passive inflation. At short A-V delays, fiber length in prestretched regions was longer than early activated regions but not compared with atrial pacing, suggesting an overall decrease in preload that is consistent with a decrease in stroke volume. Ejection strain was increased, however, potentially as a secondary result of diminished pressure generation caused by the dyskinetic contraction. Badke et al. (2) were among the first to recognize prestretch, which we and others have defined as tissue lengthening occurring before aortic valve opening. Since then, it has been studied by several investigators using various techniques such as epicardial video markers (6, 16), MRI (17, 20), radionuclide angiography (3, 4), tissue Doppler imaging (11), and finite element modeling (9). The hemodynamic effects of ventricular activation are well known, and the results in this study are similar to many early publications. LV systolic pressure was reduced, within the range of 12–25% found in most previous studies (2, 6, 14). A few studies showed no change in systolic pressure, but this was probably because of unmatched heart rate (15) or differences in anesthesia (16). LVEDP was reduced in at least one study (14), but LVEDP was unchanged in most studies of A-V sequential pacing, when there is a component of atrial filling (2, 6, 15). These studies also found a decrease in stroke volume of between 13 and 23%, consistent with the 18% decrease seen here. In the current study, dP/dtmax was reduced by 18%, also similar to these other studies (8–27%).
The time course of prestretch has been reported but, with lower temporal resolution (15, 16, 21), has not been expressed in terms of fiber direction (20, 21) and has not been described relative to phases in the cardiac cycle. Because activation is asynchronous, the pattern of regional contraction relative to LV pressure generation is altered. During normal activation, the majority of the ventricle is activated within 20–30 ms of end diastole. With ventricular pacing, however, some areas are activated much earlier than this (during late diastole), and some areas are activated much later (during isovolumic contraction). Compared with other studies of ventricular activation, the delay to reach the latest activated region in the present study was comparable, on the order of 100 ms (15, 16). In this study, we compared the magnitude of stretch at the time of activation, the end of diastole, and the time of aortic valve opening. An MRI study by Wyman et al. (20) showed that the prestretch begins in late diastole and continues through the isovolumic contraction phase, reaching magnitudes of as much as 20%. However, the time of end diastole was not available. Also, midwall circumferential strains were used, which may have a different time course than fiber strain, with significant regional variation. In studies with epicardial markers (15, 16), 10% fiber stretch was seen during isovolumic contraction, the lower number most likely reflecting that late diastole was not included in those results. Our results clearly show that very little prestretch occurs before the time of ventricular end diastole, and perhaps more importantly when the tissue is lengthening it is also electrically depolarized and presumably capable of generating force.
From the current data, it is not possible to define the nature of the forces lengthening the prestretched tissue. However, it seems likely that the high intracavitary pressures during isovolumic systole play a major role in lengthening activated tissue. To determine if the three-dimensional deformation of prestretched tissues differs from passive inflation, we compared our transmural results with those of Omens and Covell (12). In those experiments, isolated, arrested canine hearts were passively inflated to a normalized volume change of 0.20 ml/g, corresponding to a pressure change from 0 to 10 mmHg. In the current experiments, the pressure changed from 3.9 mmHg (at stimulus) to 35.8 mmHg (at peak prestretch). Despite this larger pressure change, six of nine normal cardiac strain components were significantly smaller in magnitude, as shown in Fig. 10. The transmural gradients were also reduced in prestretch. For example, the gradient of radial strain (E33) was not significantly different from zero. Moreover, the transmural gradient during passive inflation (–0.088% depth–1) is >2 SD than during prestretch (–0.010 ± 0.033% depth–1), further emphasizing the difference in the mode of deformation. The reduced transmural gradients suggest that the activation sequence in these late-activated regions may progress transmurally from endocardium to epicardium, beginning the stiffening of these endocardial cells earlier and preventing them from stretching as much. In summary, the results of this study do not support the concept that prestretched tissue is passively lengthened in the same manner as diastolic filling of the ventricle.
|
In the present study, peak lengthening in prestretched areas occurred 44 ms following local depolarization. There was an increased magnitude of prestretch that was roughly proportional to the delay in local activation time. It has been previously shown that the activation time correlates with the local ejection strain (9) and mechanical activation time (20), but to our knowledge no one has shown the relationship with magnitude of stretch.
The question of whether ejection strain is augmented in late-activated areas has received substantial attention. Badke et al. (2) showed that late-activated segments had unchanged or depressed ejection strain, but larger "actual" shortening (total shortening regardless of phase in the cardiac cycle) when there was no contribution of atrial contraction to filling. Prinzen et al. (16), using A-V sequential pacing with a 30-ms delay, showed that ejection strain was larger in the late-activated region than the same areas during atrial pacing. However, the 150% increase over atrial pacing is larger than the 32% increase seen in the present study, likely because of the hemodynamic differences between the two studies discussed above. Despite these differences in magnitude, an increase was seen in both studies. The mechanism of this increase is unknown. However, several authors have suggested there is a length-dependent activation effect because of increased fiber/sarcomere length in the prestretched areas (9, 15, 17).
To investigate this possibility, we compared the change in fiber length at aortic valve opening in prestretched areas with the same areas during atrial activation. At an A-V delay of 35 ms, there was no statistically significant change in fiber length (mean fiber aortic valve opening strain = 0.005). It therefore seems unlikely that length-dependent activation is responsible for the enhanced ejection strain in this preparation. In fact, of the four well-known regulators of ventricular performance (heart rate, preload, inotropic state, and afterload), only afterload changed significantly. Although average systolic pressure is not a true measurement of local tissue afterload, this may be the best explanation for the observed increase in ejection strain.
Although fiber length at aortic valve opening was not different from atrial pacing, there was still a difference in fiber length between early and late-activated regions. This reflects the heterogeneous distribution of stretch in the ventricle. Coupled with the aortic valve opening strain results, this shows a net reduction in overall LV preload that is at least partially responsible for diminished hemodynamic function.
Aortic valve opening strain, the change in muscle fiber length from atrial to ventricular pacing in prestretched regions, was not significantly different from zero. This can be explained by the preexcitation that occurs at short A-V delays. As a result, at the time of ventricular stimulation, the fibers were shorter than at atrial-paced end diastole. We have anecdotal evidence in one dog with complete A-V block that shows that aortic valve opening strain increases with A-V delay, to a maximum of 100 ms. Despite this, the ejection strain was unchanged over this range of A-V delays. This likely reflects the balance between preload and afterload, both of which increase with A-V delay.
Study limitations.
The present study was designed to investigate the magnitude and timing of stretch in the late-activated region of the ventricular paced canine LV. The approach has several limitations. First, beads were plunged into the myocardium with the use of a trocar, which likely causes some local injury. However, the overall deformation of the array region is unlikely to be affected by small areas of injury. No evidence of ischemia was seen in the ECG in any of the animals. Second, the interpretation of these data was made with the assumption that all the beads formed a surface at a fixed transmural depth, while in reality there was a small variation in bead depth. Third, although the equations to calculate strain are exact, there are errors involved in bead digitization (the resolution of the radiographic system is 0.02 mm), as well as the three-dimensional reconstruction from the two-dimensional images. Fourth, the measured fiber orientations were registered with the strain data based on bead locations, which may have produced some inaccuracies. Averaging of the three closest fiber angle measurements presumably reduced these errors. Additionally, changes in segment length along the fiber direction were used for all fiber length comparisons rather than any direct measure of fiber or sarcomere length. Last, with the current technique, simultaneous strain measurements of the entire ventricle are not possible. Therefore, it is possible that the location of maximum prestretch is not encompassed by the array.
In this study, we used a fixed A-V delay of 35 ms. Because the experiments were performed in the presence of a functioning A-V node, a relatively low A-V delay was chosen to ensure that the activation wave front spread from the pacing site and activated the ventricles exclusive of native conduction. In a similar model, Faris et al. (7) showed that such fusion of activation sources did not occur at A-V delays <60 ms. However, this choice of delay likely results in suboptimal filling and a reduction of end-diastolic volume compared with atrial pacing (10).
In conclusion, during LV pacing, the mechanics of the late-activated tissue was studied using midwall and transmural marker arrays in the canine left ventricle. In late-activated tissue, the majority of prestretch was found to occur during isovolumic systole, probably because of a combination of intracavitary pressure and series and parallel force generation by early activated fibers. The deformation was different from passive inflation, reflecting the lengthening of already electrically activated tissue that resulted in reduced transmural gradients of strain. At short A-V delays, muscle fiber length in prestretched areas was unchanged compared with atrial activation, whereas fiber length in early activated areas was reduced. Although ejection shortening in late-activated areas was increased compared with atrial activation likely due to a decrease in afterload, overall ejection shortening (as reflected by stroke volume) was reduced due presumably to dyskensis and to an overall reduction in muscle length.
|
ACKNOWLEDGMENTS |
|---|
|
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 |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. P. Sengupta, B. K. Khandheria, and M. Belohlavek Letter by Sengupta et al Regarding Article, "Mechanisms of Preejection and Postejection Velocity Spikes in Left Ventricular Myocardium: Interaction Between Wall Deformation and Valve Events" Circulation, February 17, 2009; 119(6): e204 - e204. [Full Text] [PDF]
|
|
|
|
 |
|
 P. P. Sengupta Exploring left ventricular isovolumic shortening and stretch mechanics: "The heart has its reasons...". J. Am. Coll. Cardiol. Img., February 1, 2009; 2(2): 212 - 215. [Full Text] [PDF]
|
|
|
|
 |
|
 H. Ashikaga, B. A. Coppola, K. G. Yamazaki, F. J. Villarreal, J. H. Omens, and J. W. Covell Changes in regional myocardial volume during the cardiac cycle: implications for transmural blood flow and cardiac structure Am J Physiol Heart Circ Physiol, August 1, 2008; 295(2): H610 - H618. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
).
