| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Physics and Medical Technology and 2Department of Cardiology, Vrije Universiteit University Medical Center, Amsterdam, The Netherlands
Submitted 21 June 2004 ; accepted in final form 11 October 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
heterogeneity; mechanical asynchrony; strain; atrial systole
Observations of considerable asynchrony in patients with narrow QRS widths imply that electrical phenomena are not the only determinant of the timing of shortening but that mechanical factors may also play an important role in the timing of myocardial shortening (10, 30). In addition, the observation in normal subjects that the pattern in the time to onset of shortening (Tonset) differs from the pattern of electrical activation suggests that the timing of shortening is not determined by electrical phenomena alone (31). However, direct evidence for a relation between mechanical factors and the timing of myocardial shortening in the in vivo heart is lacking. We hypothesize that heterogeneity in prestretch induced by the atrial contraction contributes to the heterogeneity in myocardial function and timing. Regional myocardial function is probably linked to LV prestretch by heterogeneity in wall stress [due to differences in architecture, geometry, and material properties (1, 2, 21)] or the dependence of the developed force on end-diastolic sarcomere length [a local manifestation of the Frank-Starling mechanism (13, 16, 24)]. In the same way, regional timing of myocardial shortening may be related to LV prestretch. Tonset occurs when the generated force overcomes the load against which the myocardium has to contract. In myocardium with higher prestretch, either the load (wall stress) is lower or the generated force is higher (due to a local Frank-Starling effect). Therefore, myocardium with higher prestretch is expected to show an earlier Tonset.
In this study, MRI with myocardial tagging was used to assess the circumferential strain (
c) of the LV with high temporal resolution and to determine the spatial distribution of LV prestretch induced by the atrial contraction. We explored whether the prestretch shows regional variations and whether these variations are related to the regional degree of shortening and to the regional timing of shortening.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Eleven healthy subjects (mean age 41 ± 12 yr; 7 men and 4 women) with no history of cardiac disease, normal ECG, ejection fraction > 50%, and blood pressures under 160/90 mmHg were studied after informed consent was obtained according to our institutional guidelines.
Imaging
The imaging protocol has been previously described in Ref. 31. In brief, myocardial tagging images with high temporal resolution (14 ms) were obtained by steady-state free precession imaging using the linearly increasing startup angle approach (32) to avoid artifacts. Complementary tagging (CSPAMM) was used for improved tag contrast and strain analysis (8, 17). Five short-axis image planes were acquired from base to apex, with 4 image series per plane: horizontally and vertically tagged images with both positive and negative sinusoidal tagging for CSPAMM. Breath hold misregistration was minimized using a multiple brief expiration breath hold scheme (5, 18). The four series of a single image plane were acquired in 
4.5 min. The tag spacing was 7 mm. Examples of the complementary tagged images are shown in Fig. 1.
|
45–55 min/subject. Postprocessing
Strain analysis.
Strain analysis was performed using the harmonic phase method, as described previously (31). c was calculated from the Lagrangian strain tensor as the percent change in length of a small line segment in the circumferential direction (9):
 | (1) |
where Ecc is the Lagrangian strain component in the circumferential direction. The circumferential strain data were averaged in six circumferential segments: inferoseptal, anteroseptal, anterior, anterolateral, inferolateral, and inferior. Because the fibers at the midwall are mainly oriented in the circumferential direction and lay in the image plane (11, 28), only strain data from the mid-50% of the myocardialwall was used. Hence, the c curves represent mostly the shortening of the underlying myocardial fibers.
The circumferential strain rate dc/dt was calculated numerically from a copy of the c curve that was filtered with a moving average filter of five sample width to reduce noise sensitivity. The filtered strain curve was used only for the calculation of the strain rates, not for the quantification of the amount or timing of shortening.
Prestretch.
Because of prospectively gated triggering, no strain was measured during the last 150–200 ms of the R-R interval, in which the atrial contraction occurs. Prestretch was therefore measured as the difference in LV strain before and after atrial contraction. The prestretch was defined as the strain at the R wave minus the average plateau strain observed during diastasis, as illustrated in Fig. 2. For consistent assessment of the prestretch, the following algorithm was used. The onset of diastasis was defined regionally by the end of rapid relaxation and was determined as the zero crossing in dc/dt after the peak positive dc/dt (peak lengthening rate) in early diastole. To prevent inclusion of data points during atrial contraction, the end of diastasis was defined as 200 ms before the next ECG R wave, which was normally just outside the interval covered by the strain curve. If the strain curve did not cover the full diastasis, only the part covered was used, with a minimum duration of three strain samples. If the period of diastasis could not be determined for a certain segment, because of either a persistent positive strain rate or insufficient coverage of the cardiac cycle, the amount of prestretch for that segment was regarded as missing.
|
c(Tpeak) – c(Tavc) otherwise, where Tpeak is the time of peak circumferential shortening.
Timing of shortening.
The timing of shortening was quantified by Tonset and Tpeak. Tonset was defined as the beginning of the down slope of the c curve and was determined by an automated fitting algorithm as described in detail previously (31). Tonset is reported in milliseconds from the ECG R wave, and Tpeak is reported in ms from Tavc.
Investigated Relations
The prestretch data of the subjects were averaged per segment to construct a mean prestretch map for the normal heart. Missing values were omitted in the calculation of the average prestretch per segment. Similarly, maps of peak circumferential shortening, PCSR, amount of postsystolic shortening, Tonset, and Tpeak were obtained. Linear regression analysis on the averaged data was used to investigate the relation between the LV prestretch and parameters of regional myocardial function. To obtain insight in the individual variability in the observed relations, individual correlation coefficients were calculated between the prestretch and each of the investigated parameters of myocardial function.
In addition, the factors by which prestretch influences peak shortening were explored. In vitro studies have shown that an increase in prestretch increases 1) the duration of shortening and 2) the shortening velocity at a constant afterload (22, 26), which both may enhance peak shortening. Therefore, multiple regression analysis was used with the peak shortening as a dependent variable and Tpeak, PCSR, and the product of Tpeak x PCSR as predictors. Tpeak x PCSR accounts for the interaction between Tpeak and PCSR. The stepwise selection method was used to include or exclude a predictor from the model, using a probability < 0.05 as criterion to include a predictor and a probability > 0.1 as a criterion to exclude a predictor.
Statistics
Paired Student's t-tests were applied to compare the prestretch between opposite myocardial segments (using the mean value of the 5 slices) and to compare between the apex and base (using the mean value of the 6 circumferential segments). The regression analysis described above was performed using SPSS software (SPSS for Windows, Release 11.5.0, SPSS; Chicago, IL). Values are presented as means ± SE for regression coefficients and as means ± SD otherwise. P values of <0.05 were regarded as significant.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
All subjects had a normal QRS width of 83 ± 15 ms (range: 63–114 ms). Table 1 shows the global function parameters of the subjects. For all subjects included in this study, prestretch could be determined in at least 24 of 30 segments (average number of missing segments per subject: 2.3 ± 2.0).
|
An example of the c curves for all segments for a single subject is shown in Fig. 3. Note the regional differences in the plateau strain at diastasis and consequently in the amount of prestretch induced by the atrial contraction. The map of the average LV prestretch is shown in Fig. 4. Table 2 summarizes the prestretch for the six circumferential segments averaged over all five slices and shows in the clockwise direction a gradual increase in prestretch from the inferoseptal to the inferolateral segment. The difference between the apex and base was not significant. The highest prestretch was observed in the inferolateral segments (from apex to base) and the anterolateral and anterior segments at the apex. In each individual subject, the inferolateral prestretch was consistently larger than the anteroseptal prestretch.
|
|
|
The spatial distribution of peak circumferential shortening was very similar to the distribution of prestretch, as demonstrated in Fig. 4. Linear regression revealed a close correlation between the amount of prestretch of a segment and the peak c of that segment (r = 0.79, P < 0.0005; Fig. 5A). The correlation between the amount of prestretch and peak shortening rate was much weaker but still significant (r = 0.50, P = 0.005; Fig. 5B). Also, the amount of postsystolic shortening was correlated to the prestretch (r = 0.67, P < 0.0005; Fig. 5C). The regression parameters for the relations between prestretch and the parameters of the regional myocardial function can be found in Table 3.
|
|
|
Relation Regional Prestretch and Timing of Shortening
Tonset showed a weak but significant negative correlation with the amount of prestretch (r = –0.57, P = 0.001; Fig. 7A). This means that regions with more prestretch have an earlier onset of shortening. The duration of shortening was longer in regions with more prestretch, as can be seen from the positive correlation between Tpeak and prestretch (r = 0.71, P < 0.0005; Fig. 7B). The actual distributions of Tonset and Tpeak have been published previously (31). The relations between LV prestretch and timing of shortening were also observed on an individual basis, as can be seen from Fig. 6. For all subjects except one, at least two of five investigated relations were significant. None of the correlations that were opposite to the relations found for the average data were significant.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
LV Prestretch
The amount of prestretch ranged from 2.7 ± 1.3% in the septum to 5.5 ± 1.7% in the inferolateral segment. Ennis et al. (7) have reported strain measurements over the full cardiac cycle including the atrial contraction using a dedicated acquisition and reconstruction method. Although quantified in a different way, they found a similar pattern in the prestretch induced by atrial contraction, with lowest prestretch in the septum (2.8 ± 1.1%, n = 6) and highest prestretch in the lateral and inferior regions (3.8 ± 0.4% and 4.2 ± 1.0%, respectively). These highest values seem to be somewhat lower than found in this study, possibly because Ennis et al. used four instead of six circumferential segments, whereas the spatial variation in prestretch is relatively large in the lateral wall (Fig. 4 and Table 2).
Relation Regional Prestretch and Degree of Shortening
The spatial pattern in peak c (Fig. 4B), with the largest peak shortening occurring in the lateral wall, is comparable to the pattern reported in literature (1, 19). A close correlation was observed between peak shortening and prestretch. The multiple regression analysis between the peak shortening and both Tpeak and peak shortening rate supports the idea that the prestretch influences peak shortening via two pathways. Prestretch increases peak shortening not only by prolongation of the duration of shortening but also by increasing the rate of shortening.
Besides the peak shortening and peak shortening rate, the amount of postsystolic shortening (the amount of shortening developed after aortic valve closure) is also influenced by prestretch. The regression coefficient of the relation between postsystolic shortening and prestretch (1.0 ± 0.2) is comparable to that of the relation between peak shortening and prestretch (1.4 ± 0.2; Table 3). This implies for the regions with most prestretch that the extra peak shortening due to the extra prestretch is only slightly larger than the simultaneous increase in postsystolic shortening and consequently only partially benefits ejection.
Relation Regional Prestretch and Timing of Shortening
The negative correlation between the regional Tonset and LV prestretch implies that the force generated by the myocardial muscle overcomes the existing load earlier in a region with more prestretch than in a region with less prestretch. This is compatible both with local differences in wall stress and with a local manifestation of the Frank-Starling mechanism (length-dependent activation), as outlined in Fig. 8. In terms of local differences in wall stress (Fig. 8A), a lower wall stress will lead to more prestretch during diastole (as it makes the wall more compliant to the atrial contraction) and to a lower load against which the muscle has to contract during systole, which yields an earlier onset of shortening. In terms of a local Frank-Starling effect (Fig. 8B), an increase in initial segment length leads to a stronger contraction while peak force is reached at the same time (26), so that a given load is overcome earlier. Besides, prestretched regions can begin to shorten earlier by elastic recoil (25). The fact that the correlation between Tonset and prestretch was rather weak (r = –0.57) may be due to regional differences in electrical activation and to noise in the Tonset values.
|
1.9 µm [slack length (23)], then the observed increase in Tpeak of 19 ms/% prestretch is equivalent to 1,000 ms/µm increase in sarcomere length (95% confidence interval from the SE in Table 3: 614–1,384 ms/µm sarcomere length). This value is comparable with the slope between twitch force duration and initial sarcomere length reported for rat cardiac trabeculae (24, 26). Twitch force duration increases by about 400 ms/µm increase in initial sarcomere length for isometric contraction (derived from Fig. 4 in Ref. 24 and Fig. 1 in Ref. 12) and by about 670 ms/µm increase in sarcomere length for isosarcometric contraction (derived from Fig. 7 in Ref. 26). These literature values seem to be lower than the value found for Tpeak in this study (1,000 ms/µm). However, one should keep in mind that the data of the force duration are obtained from right ventricular rat trabeculae at 25°C, whereas our data of Tpeak are obtained from the human LV in vivo, ignoring regional differences in electrical activation time. Furthermore, twitch force duration may only be directly comparable with Tpeak if the afterload is constant during shortening, which is clearly not the case as quite a few segments reach peak shortening after aortic valve closure (27, 31). Clinical Implications
It has been shown that a simple wall stress model that incorporates only the geometry is insufficient to explain the regional variations in myocardial function in patients with nonischemic dilated cardiomyopathy (29). The close correlation between LV prestretch and peak shortening found in this study suggests that regional differences in material properties, myofiber structure, and contractile force should also be considered for a correct understanding of regional variations in myocardial function.
Previous work has shown that the onset of shortening has a pattern that is reversed to the pattern of electrical activation, with the earliest onset of shortening in the lateral wall (31). Our results show that the timing of shortening is related to the amount of LV prestretch due to atrial contraction and suggest that the influence of prestretch even dominates the influence of the electrical activation pattern in health. Consequently, it may be very difficult to predict the response to resynchronization therapy and the optimal pacing position in patients with heart failure and conduction abnormalities, without extensive modeling by means of computer simulations (15, 25).
Limitations
It was assumed that the two-dimensional c at midwall reflects myofiber behavior. For most slices this may be a good approximation, but especially toward the apex the midwall layer with circumferentially orientated fibers becomes thin, and an increasing amount of fibers will have a substantial longitudinal component (11). Because only the in-plane component of these fibers was observed in this study, the measured circumferential prestretch and shortening in the most apical slice may be less than the actual fiber stretch and shortening. Nevertheless, the relation between stretch and shortening is not affected as long as the angle of the fibers and the image plane is constant during the cardiac cycle.
Furthermore, we did not directly measure the strain during atrial systole but used the existence of the strain plateau during diastasis instead. The good agreement of our prestretch data with data obtained by Ennis et al. (7), who did measure strain evolvement during atrial systole, justifies this approach.
In conclusion, atrial contraction leads to a nonuniform pattern of prestretch in the LV, with the most prestretch in the lateral wall. This prestretch is closely correlated to the nonuniformity present in both the amount and the timing of shortening. The effect of prestretch on the timing of shortening shows that synchrony of contraction is not determined by electrical synchrony alone but that mechanical aspects also play an important role.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
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 |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  X. Monnet, L. Lucats, P. Colin, G. Derumeaux, J.-L. Dubois-Rande, L. Hittinger, B. Ghaleh, and A. Berdeaux Reduction in postsystolic wall thickening during late preconditioning Am J Physiol Heart Circ Physiol, January 1, 2007; 292(1): H158 - H164. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Ashikaga, S. R. Mickelsen, D. B. Ennis, I. Rodriguez, P. Kellman, H. Wen, and E. R. McVeigh Electromechanical analysis of infarct border zone in chronic myocardial infarction Am J Physiol Heart Circ Physiol, September 1, 2005; 289(3): H1099 - H1105. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
