Author + information
- Received April 24, 1998
- Revision received December 15, 1998
- Accepted January 20, 1999
- Published online May 1, 1999.
- Frits W Prinzen, PhDa,* (, )
- William C Hunter, PhD∗,
- Bradley T Wyman, MSc∗ and
- Elliot R McVeigh, PhD∗
- ↵*Reprint requests and correspondence: Dr. Frits W. Prinzen, Department of Physiology, Cardiovascular Research Institute Maastricht, University of Maastricht, P.O. Box 616, 6200 MD Maastricht, The Netherlands
The purpose of this study was to determine the spatial distribution of myocardial function (myofiber shortening and work) within the left ventricular (LV) wall during ventricular pacing.
Asynchronous electrical activation, as induced by ventricular pacing, causes various abnormalities in LV function, perfusion and structure. These derangements may be caused by abnormalities in regional contraction patterns. However, insight into these patterns during pacing is as yet limited.
In seven anesthetized dogs, high spatial and temporal resolution magnetic resonance–tagged images were acquired in three orthogonal planes. Three-dimensional deformation data and LV cavity pressure and volume were used to determine midwall circumferential strain and external and total mechanical work at 192 sites around the left ventricle.
During ventricular pacing, systolic fiber strain and external work were approximately zero in regions near the pacing site, and gradually increased to more than twice the normal value in the most remote regions. Total mechanical work, normalized to the value during right atrial pacing, was 38 ± 13% (right ventricular apex [RVapex] pacing) and 61 ± 23% (left ventricular base [LVbase] pacing) close to the pacing site, and 125 ± 48% and 171 ± 60% in remote regions, respectively (p < 0.05 between RVapex and LVbase pacing). The number of regions with reduced work was significantly larger during RVapex than during LVbase pacing. This was associated with a reduction of global LV pump function during RVapex pacing.
Ventricular pacing causes a threefold difference in myofiber work within the LV wall. This difference appears large enough to regard local myocardial function as an important determinant for abnormalities in perfusion, metabolism, structure and pump function during asynchronous electrical activation. Pacing at sites that cause more synchronous activation may limit the occurrence of such derangements.
Asynchronous electrical activation of the ventricles, as during left bundle branch block (LBBB) and during ventricular pacing, is associated with abnormal, locally different contraction patterns (1–3)and deficiencies in perfusion and glucose uptake despite patent coronary arteries (4,5). In animal experiments, regional systolic shortening and work were determined quantitatively and proved to be reduced in early activated regions and increased in late activated regions (6–8). During ventricular pacing, local work was related to local blood flow (7,8), a finding that suggests abnormalities in perfusion and glucose uptake in pacemaker patients and LBBB patients could be due to an abnormal distribution of mechanical work in the left ventricular (LV) wall. However, studies determining mechanical performance quantitatively only used epicardial (and mainly LV) pacing sites, whereas the right ventricular apex (RVapex) is the most frequently used site in patients and causes an activation pattern similar to LBBB (9,10). A recent echocardiographic study suggests that wall motion and perfusion abnormalities are less pronounced during pacing at the right ventricular (RV) than at the LV wall (11). It is also unknown how large the regions of abnormal wall motion are, because until now, measurements on local function were limited to a low number of sites in the LV wall.
Better knowledge of the effect of asynchronous activation on regional ventricular function is not only relevant to understanding the acute derangements mentioned above, but also because asynchronous activation appears to have potentially deleterious effects over time. In patients, it has been shown that perfusion deficits and wall motion abnormalities, as induced by pacing, become more pronounced with increasing duration of pacing, and this is accompanied by impaired LV pump function (12). Ventricular pacing and LBBB also lead to structural changes, such as asymmetric hypertrophy (13,14)and, in immature dogs, fiber disarray (15).
The present study was designed to obtain detailed insight into the effects of ventricular pacing on local myocardial function. This was obtained by determining deformation of the entire LV wall at high spatial resolution with use of the magnetic resonance imaging (MRI) tagging technique. This technique uses markers (tags), magnetically induced into the myocardium at end-diastole, that persist during systole. Myocardial motion patterns were translated to myofiber work using data on LV midwall deformation, LV cavity pressure and LV cavity to wall volume ratio using a previously validated approach (8). Asynchronous activation was induced by pacing at two ventricular sites (RVapex and left ventricular base [LVbase]).
Animal investigation complied with the “Position of the American Heart Association on Research Animals in the U.S.” and was approved by the animal experimentation committee of the Johns Hopkins University. Seven male mongrel dogs (weight 20 kg to 25 kg) were induced with thiopental (20 mg/kg), and anesthesia was maintained using IV infusion of midazolam (0.1 mg/kg/h) and sufentanyl (10 μg/kg/h). An MRI-compatible Millar cathetertip micromanometer was introduced into the LV via a carotid artery. The chest was opened to position nonferromagnetic bipolar pacing leads to the right atrium (RA) and the base of the LV free wall and in the RVapex.
In the MRI, scanner-specialized filters were used to eliminate the voltages induced in the pacing leads by the radiofrequency (RF) and magnetic field gradient pulses (16). Through these filters the pacing leads were connected to the pacing and monitoring system outside the scanning room. Electrograms and LV pressure signals were recorded on a four-channel recorder (Gould, Cleveland, Ohio) and A/D converted at 200 Hz with 10 bits. Left ventricular cavity volume was calculated from the endocardial contours in the end-diastolic short-axis MRI and from the deformation of the endocardial layers (see the following section).
The imaging procedure has been described in more detail previously (16). During stopped ventilation, MRI tagging images were acquired in a 1.5 T scanner (Signa, General Electric Medical Systems, Milwaukee, Wisconsin) with a segmented k-space acquisition. A 6-ms RF gradient tagging pulse sequence was used to produce parallel planes of saturation bands (Fig. 1). The tag pattern was induced at end-diastole, and a total of 12 to 16 images were acquired at 19 to 21-ms intervals, covering the entire systolic phase. Each measurement consisted of six to eight short axis sections with tags at 0°, the same set of sections with tags at 90°, and a set of nine radially oriented long-axis sections (Fig. 1)at 20° intervals around the LV long axis (16).
After instrumentation and stabilization, measurements were first performed during pacing at the RVapex, and subsequently during RA and LVbase pacing. This order was chosen because pilot measurements showed that heart rate increased most with RVapex pacing. Therefore, a heart rate high enough to overrule intrinsic sinus rhythm during RVapex pacing could usually be maintained during the rest of the experiment. During ventricular pacing, the RA was stimulated simultaneously with the ventricle to ensure complete ventricular capture. The LV pressure and electrograms were recorded between MRI tagging image acquisitions.
Finally, the dogs were sacrificed by IV injection of an overdose of pentobarbital. The heart was removed from the chest, and the atria and ventricles were weighed. The LV wall volume was calculated as LV weight/1.05 to correct for the specific mass of the myocardium.
The analysis of strains from the images has been described in detail elsewhere (16). Briefly, the tag lines and heart contours in the images were identified off-line using a semiautomated algorithm. In each set of sections, displacement in the direction normal to the undeformed tag planes was obtained along the taglines. Three-dimensional (3D) deformation was calculated from the three independent sets of one-dimensional displacement data (a total of 2,000 to 3,000 points) (17). Myocardial deformation was calculated in a 24 (circumferential) × 8 (longitudinal) × 3 (radial) mesh of material points. A material point is an infinitesimal volume element of the myocardium, which moves and deforms with the surrounding tissue. The mesh was defined in the first image, and the material points were subsequently tracked in 3D space using the information on the displacement of the tags (17).
In each region Lagrangian strain (fractional change in length as compared with the length at end-diastole, when taglines were exactly parallel) was calculated. Although MRI tagging enables analysis of complete 3D deformation and pacing leads to changes in 3D deformation (18), we focused on circumferential strain in the midwall as derived from the complete 3D motion analysis. Muscle fibers in the midwall run within 20 degrees of the circumference (19). Therefore, in the midwall, circumferential strain closely approximates strain along the fibers. We were interested in myofiber strain, rather than strain in other directions, because in the myocardium work is generated in the direction of the muscle fibers. Fiber strain in other layers than the midwall is harder to determine because fiber angle is more variable and deformation data are not always complete in these layers.
Calculation of local fiber stress and work
According to the LaPlace equation and more recently developed mathematical models, wall stress in the LV wall is a function of cavity pressure and the ratio of cavity diameter and wall thickness (20). A robust approximation for fiber stress (σ) in a thick-walled, anisotropic structure, which yields more physiologic values than most other models, is obtained using the ratio of LV cavity and wall volume (Vc/Vw) (20): where Pevis LV cavity pressure. Although this formula assumes uniform fiber stress, nonuniform fiber stress also can be calculated if two assumptions are made: the fibers are incompressible (do not leak water during the cardiac cycle) and force is conducted freely along all myocardial fibers. The latter assumption is derived from anatomical studies (19,21)showing that myocardial fibers form rope-like structures, which are continuous throughout the LV wall. If force is equal throughout the ventricle, σ is inversely proportional to the cross-sectional area of the muscle fiber, and thus proportional to fiber length. Using this property local σ can be calculated using (8): where L/L0is the ratio of fiber length at time tand at zero cavity volume, respectively. The (extrapolated) length at zero cavity volume serves as reference situation. L/L0is calculated as (L/Led)(Led/L0), where Ledis length at end diastole. L/Ledwas derived from the strain values obtained with MRI tagging and Led/L0was calculated from the ratio of end-diastolic LV wall and cavity volume as determined from MRI using (8,20): 
Local mechanical work was determined using the local fiber stress–fiber length relations (8), in analogy to the pressure-volume area (PVA) concept for the ventricle as a whole (22). The PVA represents total mechanical energy per heart beat and is equal to the sum of external work and potential energy. The PVA is closely related to myocardial oxygen consumption, even during heart beats with negative external work (22). Analogously, regional external work, potential energy and total work were calculated from regional stress–length diagrams (Fig. 2). Potential energy was calculated assuming that no stress is developed at estimated zero cavity volume (8). In epicardial regions of the LV free wall, thus calculated total work was highly correlated with local oxygen consumption when pacing from various sites (8).
Differences in hemodynamic and deformation variables between the three modes of pacing were evaluated using analysis of variance (ANOVA) for repeated measurements. The distribution of local work was analyzed using cumulative histograms. Histograms from the various experiments were pooled by normalization of the three histograms per experiment to the mean external work during RA pacing during that particular experiment. A p < 0.05 was considered significant. Mean values and significant differences (SDs) are presented.
Heart rate was not significantly different between RA, LVbase and RVapex pacing (113 ± 13, 115 ± 12 and 118 ± 19 beats/min, respectively). Peak systolic LV pressure was significantly lower during RVapex pacing (96 ± 20 mm Hg; 7.5 mm Hg = 1 kPa) than during RA pacing (105 ± 12 mm Hg). Also, stroke volume was lower during RVapex pacing than during RA pacing (14.8 ± 6.0 and 19.9 ± 13.9 ml, respectively). Values during LVbase pacing (103 ± 20 mm Hg and 17.4 ± 9.5 ml) were not significantly lower than during RA pacing. End-diastolic pressure was 10 ± 2, 10 ± 2 and 9 ± 2 mm Hg during RA, LVbase and RVapex pacing, respectively; NS. Also, end-diastolic volumes, as determined from the MRI images, were not significantly different between the three pacing modes.
Figure 3presents midwall circumferential strain as a function of time in one experiment. During RA pacing (upper row), differences in time course and extent of shortening between the various regions were relatively small. During ventricular pacing close to the pacing site (asterisk in Fig. 3), a rapid onset of shortening (negative strain) during the early systolic phase was followed by a rebound stretch and a second phase of shortening. In regions remote from the pacing site (left and right panels of lower row, and middle panels of middle row of Fig. 3), considerable early systolic stretch was followed by pronounced shortening during the ejection phase. Strain patterns gradually changed from the pacing site to the most remote regions.
The stress-length loops from the experiment presented in Figure 4illustrate that during RA pacing these loops had a similar shape and counterclockwise course throughout the LV midwall. During both kinds of ventricular pacing, the loops in regions close to the ventricular pacing site had a clockwise course during the early part of systole. This implies negative work: the region deforms passively and other regions work on that region. In regions remote from the pacing site, the stress-length diagrams were wider than during RA pacing.
Figure 5illustrates the mean values and SD of the values of systolic circumferential strain, external and total work near the two sites where the pacing leads were located: the RVapex and the LVbase. During RA pacing, circumferential strain, external and total work at these two sites were not significantly different. During ventricular pacing, systolic circumferential strain was positive (net stretch) at the site of pacing and extremely negative at the opposite site (Fig. 5). Longitudinal strain, i.e., strain perpendicular to the circumferential direction, behaved similarly to circumferential strain: during RA pacing longitudinal strain at the RVapex and LVbase were similar (−0.119 ± 0.07 and −0.086 ± 0.045, respectively), during RVapex pacing longitudinal strain was smaller at the RVapex (0.002 ± 0.078) than at the LVbase (−0.100 ± 0.049) and during LVbase pacing longitudinal strain was smaller at the LVbase (−0.034 ± 0.027) than at the RVapex (−0.163 ± 0.032).
External work was negative at the pacing site and increased up to supranormal values in regions remote from the site of pacing. Local differences in potential energy were not statistically significant (data not shown). As a consequence, at the pacing site total work was 38 ± 13% and 61 ± 23% of the normal (RA pacing) value during RVapex and LVbase pacing, respectively, and 125 ± 48% and 171 ± 60% at the opposite site, respectively (Fig. 5; p < 0.05 between RVapex and LVbase).
Figure 6displays polar plot maps of external work during the three pacing modes in one experiment. A fairly homogeneous distribution was found during RA pacing, but large variations were present during ventricular pacing. External work ranged from negative at the pacing site to supranormal values in remote regions. Note that the area with reduced external work was larger during RVapex pacing than during LVbase pacing.
Figure 7presents cumulative histograms of external work (normalized to the mean value during RA pacing in each experiment). The steep curve during RA pacing indicates a homogeneous distribution. During LVbase pacing the curve was less steep, but crossed the RA curve virtually in the middle, indicating that mean external work (2.70 ± 0.57 mJ/g) was similar to that during RA pacing (2.90 ± 1.24 mJ/g), but that the distribution was considerably more inhomogeneous. During RVapex pacing the cumulative histogram was shifted to the left. Mean external work was significantly lower than during RA pacing (1.70 ± 0.99 mJ/g). This was due to a large number of regions (approximately 40%, predominantly those in the septum) with relative external work values below 0.5 and the absence of regions with increased external work (Fig. 7).
The present study quantifies the degree and extent of local myofiber shortening and work during asynchronous electrical activation, evoked by ventricular pacing from the LVbase and the RVapex. Both pacing modes cause pronounced redistribution of midwall fiber shortening and work, with work values ranging from 50% of normal at the pacing site to about 150% in remote regions. Contraction patterns change gradually when moving from the early to the late activated regions. The number of hypofunctioning regions, surrounding the pacing site, is larger during RVapex pacing than during LVbase pacing. This is associated with a significant reduction in ventricular pump function during RVapex pacing.
Abnormal contraction patterns
The effect of abnormal electrical activation on LV deformation appears to depend on the spatial resolution used. At the level of the whole ventricle, the principal directions and magnitudes of deformation (23)and systolic torsion (24)were hardly affected by ventricular pacing. In contrast, the present and previous studies show dramatic changes in local deformation, characterized by reciprocal contraction patterns in opposing sites of the LV wall (6). The contraction patterns change gradually from earlier to later activated regions (7,8,25). With use of the MRI tagging technique and additional hemodynamic measurements, the present study demonstrates that ventricular pacing reduces mechanical work in the septum during RVapex pacing by 50% and increases it by 50% in the LV free wall. Previous studies showed that during ventricular pacing blood flow varied in parallel with regional function, presumably due to local metabolic regulation of myocardial blood flow (7,8). Accordingly, a good correlation was found between local mechanical work and oxygen consumption during ventricular pacing (8). Together with the present findings, this indicates that during RVapex pacing septal oxygen consumption is reduced by half.
The regionally different contraction patterns are presumably caused by regional differences in fiber length, induced by the asynchronous activation. Early activated regions start to shorten early and vigorously (by up to 10%) because their afterload is still very low. This forces later activated regions on the opposite LV wall to be prestretched (by up to 15%, Figs. 3 and 4). Consequently, when all regions have been activated and LV pressure rises, late activated fibers are longer than early activated fibers. By virtue of a local “Frank-Starling” relation, the later activated regions are stronger and shorten more during the ejection phase. Therefore, the regional differences in contraction pattern during ventricular pacing can be regarded as differences in effective local preload.
Possible implications of the abnormal contraction patterns
Because regional work is a good predictor of regional myocardial oxygen demand (8), the abnormal contraction patterns occurring during RVapex pacing and LBBB most likely reduce metabolic demands in the septum. This may explain the reduced septal perfusion and 18-fluoro-deoxyglucose uptake frequently observed in LBBB (4,5). At higher heart rates, regional perfusion abnormalities are amplified, which has been attributed to impaired perfusion due to the abnormal contraction patterns (11,26).
Although adaptation of the heart as a whole to increased workload is well known, adaptations to locally different workload, induced by asynchronous activation, also become evident. Asymmetrical hypertrophy occurs in LBBB patients (13)and in dogs with chronic ventricular pacing at physiologic heart rate (14). In the LBBB patients and in the paced dogs, the late activated regions, which are the regions shown to have higher workload in the present study, were hypertrophied most pronouncedly. In immature dog hearts, pacing induced fiber disarray (15). In patients, septal perfusion and wall motion deteriorates with increasing duration of pacing. Such changes are associated with deterioration of ventricular pump function (12). Strategies to decrease the asynchrony of activation, like pacing high in the ventricular septum, ameliorated the reduction in pump function (27)and prevented the development of fiber disarray (28). The present study indicates that MRI tagging may be a powerful tool to study the effects of new, potentially more physiologic pacing regimes on regional cardiac mechanics.
Relation between local contraction patterns and ventricular function
In the present study systolic LV pressure and stroke volume significantly decreased during RVapex pacing, but not during LVbase pacing. Other studies have also shown a larger depression of ventricular function during pacing at the RVapex than at other sites (23,29,30). The exact cause of the dependency of ventricular function on the site of pacing is not clear. Grover and Glantz (23)reported a smaller end-diastolic volume, indicating disturbed filling during RVapex pacing. However, in the present study, end-diastolic pressure and volume were similar during the three modes of pacing. Burkhoff et al. (30)found evidence that contractile failure is due to ineffective contractions in the early activated regions, thus leading to a “loss of effective muscle mass.” The present data support the loss of contractile function in early activated regions but also demonstrate hyperfunctioning late activated regions. Because ventricular pump function was not impaired during LVbase pacing, hyperfunction in the late activated regions has presumably compensated the loss of function in the early activated regions. However, during RVapex pacing the size of the region with reduced function appears to be too large to be compensated. Because a direct relation exists between the moment of local electrical activation and systolic fiber shortening (25), the large septal hypofunctional area during RVapex pacing suggests that large clusters of regions in the septum are activated almost simultaneously. This may imply that RVapex pacing couples into the right Purkinje system but is only slowly conducted transseptally. Instead, pacing from the epicardium of the LVbase does not couple into the Purkinje system, or much later.
Evaluation of experimental approach
The present study highlights the ability of MRI tagging to reveal regional differences in myofiber shortening and work during abnormal electrical activation. Moreover, it is completely noninvasive and, therefore, directly applicable to patients. Pacing and measurement of local electrograms in the MRI scanner are also possible when using specialized RF filters. Development of such a system for humans is required before human studies can be performed.
In the present study, atrio-ventricular (A-V) synchronous pacing was used with an A-V delay of 0 ms to ensure activation of the entire ventricle by the ectopic stimulus and to prevent intermittent forward conduction through the A-V node. The latter could have induced arrhythmias, which spoil MRI quality. Although a short A-V delay may itself have deteriorated ventricular function (29), in the present study reduced ventricular pump function was only found during RVapex pacing and not during LVbase pacing.
The calculations of regional myocardial work used in this study are based on measured fiber strains and estimations of fiber stress, enabling noninvasive estimation of local work at multiple sites. The approach to calculate work has been validated previously (8). Although myofiber orientation in the midwall may not be exactly circumferential, values of midwall circumferential strain were virtually equal to values of epicardial fiber strain, as measured in previous studies using optical markers. Because, at least during a normal heart beat, fiber strain is uniform throughout the wall (31), this indicates that in the midwall circumferential strain is a fair estimate of fiber strain. Moreover, qualitatively the effects of pacing are the same for midwall strain in the circumferential and longitudinal direction, so calculating work with use of strain in another direction would only cause quantitative, but not qualitative, differences. Another validation of this approach is that global external work was estimated to be 280 mJ per beat by multiplying stroke volume and LV pressure (see Results). This value correlates favorably with a regionally based estimate of myocardial work, obtained by integrating average local work (2.8 mJ/g, middle panel of Fig. 7) over the entire LV (average weight 94 g): 263 mJ per beat. Moreover, the decrement in global systolic work seen from hemodynamic measurements during RVapex pacing versus LVbase pacing also correlated with the decrement of regional work integrated over the entire ventricle (Fig. 7). The mechanical analysis in this study is limited to the midwall. However, comparable patterns of midwall mechanics in the present study and epicardial mechanics in a previous study (8)indicate that transmural differences are small compared with the regional differences induced by ventricular pacing.
Pacing at the LVbase and at the RVapex leads to local functional abnormalities throughout the LV wall, ranging from virtually absent systolic fiber shortening and external work at the site of pacing, to values twice above normal in regions most remote from the pacing site. The reduction in systolic shortening and external work as well as the size of the hypofunctioning zone around the pacing site were significantly larger during RVapex pacing than during LVbase pacing. These local changes may be responsible for the depression in ventricular function during RVapex pacing.
We are extremely grateful to Dr. Joshua E. Tsitlik for his help in adapting the RF filter system to the requirements of this study, Dr. Rajasekhar Suribhotla for his assistance with the measurements and analysis of the electrophysiologic data and Dr. Elias Zerhouni for his support and stimulating discussions.
☆ This work was supported by National Institutes of Health Grant HL-45683 and a grant from the Bakken Research Center, Maastricht, The Netherlands. E.R. McVeigh is an Established Investigator of the American Heart Association.
- analysis of variance
- left bundle branch block
- left ventricular, left ventricle
- left ventricular base
- magnetic resonance imaging
- pressure-volume area
- right atrium, right atrial
- right ventricular, right ventricle
- right ventricular apex
- three dimensional
- significant difference
- Received April 24, 1998.
- Revision received December 15, 1998.
- Accepted January 20, 1999.
- American College of Cardiology
- McDonald I.G
- Bashore T.M,
- Stine R.A,
- Shaffer P.B,
- Bush C.A,
- Leier C.V,
- Schaal S.F
- Ono S,
- Nohara R,
- Kambara H,
- Okuda K,
- Kawai C
- Badke F.R,
- Boinay P,
- Covell J.W
- Prinzen F.W,
- Augustijn C.H,
- Arts T,
- Allessie M.A,
- Reneman R.S
- Vassallo J.A,
- Cassidy D.M,
- Marchlinski F.E,
- et al.
- Vassallo J.A,
- Cassidy D.M,
- Miller J.M,
- Buxton A.E,
- Marchlinski F.E,
- Josephson M.E
- Beppu S,
- Matsuda H,
- Shishido T,
- Miyatake K
- Tse H.-F,
- Lau C.-P
- Van Oosterhout M.F.M,
- Prinzen F.W,
- Arts T,
- et al.
- Karpawich P.P,
- Justice C.D,
- Cavitt D.L,
- Chang C.-H
- Waldman L.K,
- Covell J.W
- ↵Streeter DD Jr. Gross morphology and fiber geometry of the heart. In: Handbook of Physiology. Vol. 1. 1979:61–112.
- Suga H,
- Goto Y,
- Yasumura Y,
- et al.
- Grover M,
- Glantz S.A
- Buchalter M.B,
- Rademakers F.E,
- Weiss J.E,
- Rogers W.J,
- Weisfeldt M.L,
- Shapiro E.P
- Helmer G.A,
- McKirnan M.D,
- Shabetai R,
- Boss G.R,
- Ross J.J,
- Hammond H.K
- Daggett W.M,
- Bianco J.A,
- Powel W.J,
- Austen W.G
- Burkhoff D,
- Oikawa R.Y,
- Sagawa K
- Rademakers F.E,
- Rogers W.J,
- Guier W.H,
- et al.