| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Clinical Physiology, Lund University Hospital, Lund, Sweden
Submitted 18 December 2006 ; accepted in final form 12 February 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
80% of the RVSV, compared with 60% for the LVSV. This difference explains the larger portion of THVV found on the left side of the heart.
magnetic resonance imaging; myocardial contraction; physiology; cardiac volume; ventricles
The portion of the LV stroke volume (LVSV) generated by the longitudinal function of the AVPD (LVSVAVPD) has been studied by echocardiography (9, 23, 35). Recently, in this journal, there has been a debate regarding the LVSVAVPD (3, 4, 38). The measurements by Carlhall et al. focused on the volume enclosed by the mitral annular movement during the cardiac cycle (3), and the authors conclude in their reply that this is not equal to the LVSVAVPD (4). Our group has recently used MRI to quantify the LVSVAVPD, and it was found to be 60% of the LVSV (7). The remaining 40% is the result of radial squeezing of the LV and is not caused by AVPD and thus not related to atrial filling during systole.
Moreover, there is an outer volume variation of the entire heart that can be quantified as the total heart volume variation (THVV). The relationship between outer volume variations, inner volume displacements, and atrial filling is schematically illustrated in Fig. 1. THVV appears to be related to radial function (39), and we have previously shown that the THVV is less on the right side of the heart than on the left side (5). This implies that radial function will be less in the RV than in the LV. Furthermore, the AVPD of the RV has been shown to be larger than the AVPD of the LV (14), which is in agreement with the suggestion that there is a larger longitudinal component of RV pumping than LV pumping (22, 29). However, the SV generated by longitudinal AVPD has been quantified for the LV (LVSVAVPD) (7) but not for the RV (RVSVAVPD), and neither method for volumetric determination of the LVSVAVPD or RVSVAVPD from anatomic images has been validated by independent methods.
|
Therefore, the aims of this study were to evaluate the validity of the proposed direct relationship between the SV generated by radial contraction and the left, right and total volume variation. Also, we sought to measure the proportion of the RVSV generated by AVPD and compare the movement of the center of volume in the LV and RV.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
The study was approved by the local ethics committee. Our study population has been studied as a part of a previously published study population (7). In summary, 11 healthy volunteers (mean age 23 yr, 5 women) were included in the study, and written, informed consent was obtained. All subjects underwent cardiac MRI in the supine position.
MRI
A 1.5-T MRI scanner with a cardiac synergy coil was used (Philips Intera CV; Philips, Best, The Netherlands). The LV short-axis and long-axis planes were identified as previously described (25). Short-axis images were acquired covering the whole heart from the atria to the apex. Long-axis images were acquired in the four-chamber and LV outflow tract views. All cine images were acquired during end-expiratory apnea. Flow images of all vessels leading to (caval and pulmonary veins) and from (aorta and the pulmonary trunk) the heart were acquired as previously described (5). Flow was measured through a plane positioned perpendicular to each vessel.
Cine imaging. A steady-state free precession sequence with retrospective ECG triggering was used, giving a temporal resolution of typically 30 ms and spatial resolution of 1.4 x 1.4 x 8 mm. The repetition time was 2.8 ms, echo time was 1.4 ms, and flip angle was 60°. The long-axis images were acquired with higher temporal resolution, typically 20 ms. The heart rate from cine images was determined for each subject based on the R-R interval of the short-axis cine images.
Flow imaging. A free-breathing fast-field echo velocity-encoded sequence with retrospective ECG triggering was used to obtain full coverage of the cardiac cycle. Typical imaging parameters were repetition time of 10 ms, echo time of 5 ms, flip angle of 15°, slice thickness of 6 mm, 35 phases, number of acquisitions = 1, no parallel imaging, and a velocity-encoding gradient of 200 cm/s. Flow images in the veins were obtained with a velocity-encoding gradient of 80 cm/s. The in-plane spatial resolution was typically 1.2 x 1.2 mm, and the temporal resolution was dependent on the heart rate and varied between subjects from 20 to 35 ms (mean of 28 ms). Because of the suboptimal image quality in the free-breathing images, flow measurements in three vessels in three different subjects were performed with a breath-hold turbo field echo velocity-encoded sequence with retrospective ECG triggering. Typical imaging parameters of the breath-hold flow sequence were repetition time of 4.5 ms, echo time of 2.9 ms, flip angle of 15°, slice thickness of 10 mm, 40 phases, number of acquisitions = 1, parallel imaging factor of 2, an in-plane spatial resolution of 1.2 x 1.2 mm, and temporal resolution of 24–28 ms (mean of 25 ms). The mean heart rate from flow imaging was determined for each subject as the mean heart rate based on the R-R interval of the flow images acquired in all vessels leading to and from the heart.
Image Analysis
All images were evaluated with freely available software (Segment 1.6, http://segment.heiberg.se) (16).
Stroke volume.
The epicardial and endocardial border of the RV was outlined in the short-axis images in all subjects, and the RV stroke volume (RVSV) was measured as the difference in endocardial volume between end diastole and end systole as previously described (31). The delineation of the LVSV was undertaken as previously reported (7) and was used for comparison with flow patterns and for center-of-volume calculations. The total SV of both ventricles (TSV) was calculated according to the formula
 | (1) |
The AVPD of the LV and the LVSVAVPD were measured as previously described (7). In short, the epicardial area in the short-axis view encompassed by the range of the AVPD is multiplied by the AVPD to obtain LVSVAVPD (see below, Eq. 3). The LVSVAVPD is divided by the LVSV to obtain the percentage of the LVSV generated by AVPD (LVSVAVPD%). The rationale for using the epicardial area rather than the endocardial area has been discussed previously in detail (7). In short, to explain the concept, one may use the analogy of a retractable telescope. When the telescope is retracted, the decrease in volume of the telescope will be the outer area multiplied by the shortening of the telescope. Also, given the same outer area and shortening of the telescope, the SV of the telescope will be the same regardless of the thickness of the inner wall of the cylinder, which is unchanged during retraction. By comparison, the volume of ventricular myocardium is known to be constant over the cardiac cycle (7, 15, 30, 34). The outer area of the telescope corresponds to the epicardial short-axis area, and the shortening of the telescope corresponds to the AVPD because of the constant position of the apex. Thus epicardial areas must be used to calculate the longitudinal component of the SV. If endocardial areas are used, then the longitudinal component would be underestimated. Furthermore, there would be differences in the longitudinal and radial components, with differences in myocardial thickness. The radial function would be overestimated in patients with thicker myocardium if the endocardial areas are used. This is related to the well-established constancy of the myocardial volume throughout the cardiac cycle (7, 15, 30, 34), which results in a myocardial thickening and thinning caused by myocardial rearrangement during ventricular systolic shortening and diastolic lengthening (8), respectively.
Center of volume for the ventricles. The center of volume was calculated for each ventricle in end diastole and end systole from the delineation of the epicardial borders of the ventricle. The center of volume was expressed as coordinates with an x, y, and z value, as described previously for the total heart (6). Also, the movement in three dimensions, combining the movement in the x, y, and z direction between end diastole and end systole, was calculated as previously described (6).
AVPD. The AVPD for the RV (AVPDRV) was measured in the lateral wall of the RV in the four-chamber view and the anterior wall of the RV in the LV outflow tract view (Fig. 2). The mean of the septal movement in the same images was obtained, and the means of the anterior, lateral, and the mean septal movements were calculated.
|
 | (2) |
The previously described derived method for calculating the SV generated by the AVPD in the LV (7) was applied to the RV. Thus the derived RVSVAVPD was calculated according to the formula
 | (3) |
THV.
The THV within the pericardium was volumetrically measured as previously described (5, 6). The THVV determined by the volumetric delineation of the pericardial contents (THVVvol) was calculated according to the formula
 | (4) |
Flow measurements. Flow quantification was undertaken for all great vessels leading to and from the heart (5). Adding all inflow (ml/s) into the heart (caval and pulmonary veins) and multiplying by time (s) gives the increase in THV (ml) for each time point. Likewise, adding all outflow from the heart (aorta and pulmonary trunk) gives the decrease in THV for the same corresponding time point. THVV determined by flow measurements (THVVflow) was calculated as the net change in THV for each time point throughout the cardiac cycle as previously described (5). All inflow and outflow to and from each side of the heart was also calculated separately to measure the left heart volume variation and the right heart volume variation.
The percent THVVflow (THVVflow%) was calculated according to the formula
 | (5) |
Comparison of radial function with heart volume variation.
The TSV is composed of a longitudinal and a radial component according to the formula
 | (6) |
 | (7) |
 | (8) |
 | (9) |
Thus we compared the TSVrad to the THVVflow to test the hypothesis that radial function determines outer heart volume changes during the cardiac cycle. Furthermore, this was undertaken for each ventricle separately, whereby the SV generated by radial function for each ventricle was compared with the volume variation of the left and right heart.
Comparison of radial and longitudinal function.
The following parameters were used to compare the longitudinal and radial portions of the SV for the right and left side of the heart as well as for the total heart (see 
Fig. 5). The SVs for the right and left ventricle and total heart were determined by volumetric planimetry of the short-axis images. Radial function was calculated from the volume variation during systole by flow measurements. Longitudinal function of the RV was assessed by direct volumetry of RVSVAVPD and for the LV by the derived method for LVSVAVPD. The mean of the longitudinal contribution from the ventricles was used for calculating the longitudinal contribution to the SV of the total heart.
|
|
|
Measurements of volumes, times, and distances are expressed as means ± SE and the range. The Mann-Whitney test was used to test the significance of the differences between variables. P < 0.05 was defined as statistically significant. The relationship between variables was determined by Pearson's correlation coefficient. Differences between methods were expressed as means ± SD.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Subject characteristics together with the results for each subject are shown in Table 1. Heart rate did not differ between flow and cine imaging (P = 0.87). Previously unpublished tabular data for the LV (LVSV, AVPD, and LVSVAVPD and LVSVAVPD%) from a previous study (7) are included for comparison in Table 1 together with the RV values from this study. MR images from a typical subject are shown in Fig. 2.
|
The AVPD was larger in the RV (23.4 ± 0.8; P < 0.001) than in the LV (16.4 ± 0.5 mm). The RVSVAVPD is presented in Table 1 together with the contribution (in %) of the RVSV (RVSVAVPD%). The LVSVAVPD is presented in Table 1 for comparison with RVSVAVPD and RVSVAVPD%. The RVSVAVPD determined by the volumetric method did not differ from the derived method (P = 0.60), and the difference between the volumetric and derived method was 2.6 ± 7.2 ml (range of –11 to 9 ml). The regression line comparing the derived and volumetric method for RVSVAVPD (r = 0.92, P < 0.001) was described by the equation
 | (10) |
The short-axis area used for the derived method in the RV (37 ± 2 cm2, range 25–50 cm2) did not differ from the area used in the LV (42 ± 2 cm2, range 30–53 cm2; P = 0.13).
Center of Volume
The center of volume movement between end diastole and end systole in the base to apex direction was larger in the RV (13.0 ± 0.7, range 9.1–17.6 mm; P < 0.001) than in the LV (7.6 ± 0.4, range 5.0–9.3 mm) (Fig. 3). Similarly, the maximum center of volume movement between end diastole and end systole in three dimensions was also higher in the RV (13.0 ± 0.7, range 9.5–18.6 mm, P < 0.001) than in the LV (7.8 ± 0.4, range 5.3–9.6 mm). Also, there was no significant difference between the movement in the base to apex direction and the maximum movement in three dimensions for the LV or RV (P = 0.55 and P = 0.39, respectively).
Heart Volume Variation
Table 1 presents the THVV for all subjects. The THVVflow did not differ from THVV determined from volumetric delineation of short-axis images (THVVvol; P = 0.65). The relationship between THVVflow and THVVvol by regression analysis (r = 0.95, P < 0.001) was described by the equation
 | (11) |
The difference between THVVflow and THVVvol was 4.0 ± 7.8 ml. The proportion of volume variation of the right heart to THVV was less (38 ± 2%, range 17–46%; P < 0.001) than the proportion of volume variation of the left heart to THVV (62 ± 2%, range 54–83%).
Comparison of Radial SV and Heart Volume Variation
Figure 4A shows the relationship between heart volume variation and radial function for the left side (r = 0.81, R2 = 0.66, P < 0.01) and right side of the heart (r = 0.80, R2 = 0.64, P < 0.01). Figure 4B shows the relationship between THVVflow and the TSVrad (r = 0.92, R2 = 0.84, P < 0.001). The difference between THVVflow and the TSVrad was 3 ± 13 ml (range –10 to 16 ml), and THVVflow did not differ significantly from TSVrad (P = 0.66).
Comparison of Radial and Longitudinal Function
Figure 5 shows the SV generated by radial function and longitudinal function for each ventricle and the total heart. The longitudinal contribution to RVSV, LVSV, and TSV was larger than the radial contribution (P < 0.001 for all groups). The longitudinal contribution to RVSV was larger than longitudinal contribution to LVSV (P < 0.001), and the radial contribution to LVSV was consequently larger than radial contribution to RVSV (P < 0.05). The difference in pumping between the ventricles is shown in Fig. 6. The epicardial surface of each ventricle in end diastole and end systole was reformatted to a three-dimensional image, and a supplemental movie shows an animation of the epicardial movement of both ventricles. (The online version of this article contains supplemental data.)
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
80%, and this is larger than the 60% of the LV pumping caused by longitudinal function. Furthermore, the present study presented novel methods that could be used in the assessment of cardiac pathophysiology. Longitudinal Function: Inner Volume Displacement and Venous Flow Patterns
The amount of RVSVAVPD% was 80%, which is higher than that for the LV. Our group (7) has previously shown that the LVSVAVPD% was 60% (7). This difference between the ventricles is explained by the larger AVPD of the RV than the LV, whereas the short-axis areas of the LV and RV did not differ. The difference in longitudinal function is also shown by the larger movement of the center of volume of the RV than of the LV. The similarity in movement of the ventricular center of volume between the base to apex direction and the maximum movement in three dimensions implies that AVPD explains most of the center of volume movement.
The flow in the caval veins showed a large peak in systole and a smaller peak in diastole in most subjects (data not shown). In other words, the caval veins exhibit a flow pattern that is more similar to that seen in the great arteries than that shown by the flow patterns in the pulmonary veins. This flow pattern in the caval veins has previously been described and is associated with normal right heart hemodynamics (32). The large amount of longitudinal contribution to RVSV means that most of the venous blood flowing into the right side of the heart will enter the right atrium through the caval veins during systole. In contrast, the LV has a lesser LVSVAVPD and will consequently have lesser venous filling of the left atrium during systole. Thus an effect of the larger longitudinal function of the RV than the LV is that the reservoir function (2, 12, 18, 24, 26, 36) of the right atrium is larger than for the left atrium, which is in line with our previous findings (5). AVPD has been shown to increase with increasing workload (27), and the coupling of AVPD and reservoir function shown in the present study is thus in concordance with previous studies showing an increased reservoir function of the left atrium (24) and right atrium (12) in exercise compared with at rest. The presents findings are also supported by early work in which radiopaque markers were used that showed a larger longitudinal component to the SV of the RV than the LV in dogs (29). Another study that used echocardiography in humans showed a larger AVPD of the RV than the LV (14). However, the present study is to our knowledge the first to quantify and compare the complete longitudinal components of pumping in both ventricles.
Radial Function: Outer Volume Variation
The present study has shown that the combined radial function of both ventricles corresponds closely to THVV. This study is the first to quantitatively do this by independent and concordant measurements of three-dimensional volumes by cine MRI and flow measurements in the great vessels. Furthermore, the present study has shown an excellent correlation between these measures for the total heart and good correlation for the left and right side of the heart separately. These findings are in agreement with previous authors who have proposed that there is a correlation between radial function and THVV based on two-dimensional MRI measurements (28, 39). The R2 of the relation between THVV and radial function was 0.84. This implies that 84% of the THVV depends on radial function of the ventricles. The remaining part of the THVV might be explained by either THVV at the atrial level or measurement error.
THVV in the present study was similar to previously reported results in humans (1, 5). Notably, the THVV differs between subjects, and the present findings indicate that this is related to different proportions of radial function and longitudinal function between individuals. Furthermore, THVV has been shown to be smaller in smaller species (11, 13, 19) than in humans (1, 5). When we consider the present findings, this implies that radial function is less prominent in smaller species and that longitudinal function is the major contributor to SV in these animals.
The macroscopic perspective of the coupling of radial and longitudinal function of the heart to outer volume variations is relatively straightforward. However, coupling the volume changes of the cardiac chambers to the function of the different myocardial fiber layers is somewhat more complicated. The contractile unit of the myocardium is the sarcomere, and the sarcomere can shorten by 10–12% (17). A consequence of the different angles between the myocardial fibers and the ventricular wall is that contraction results in myocardial shortening in the longitudinal and radial direction, as well as torsion of the myocardium around its long axis (20, 21, 29, 33). However, this study did not seek to investigate the intrinsic mechanics of the myocardium but rather the volume changes of the total heart and individual chambers; therefore, myocardial torsion or regional myocardial function was not studied. Furthermore, there are annular area changes during systole in the mitral valve (3), and in the present study these are regarded as a part of the radial function and thus included in the calculations of radial function.
Methodological Validation of Longitudinal Pumping
A previous study by our group (7) showed a close agreement between derived and volumetric quantifications of LVSVAVPD, and this has been repeated for the RV in the present study. The relationship between derived and volumetric quantification showed a slope that was slightly greater than unity, but there is also an intercept of –8 ml (Eq. 10), and the derived measurements did not differ from the volumetric measurements. This is also reflected in the low bias between the methods (2.6 ± 7.2 ml; see RESULTS). The slope, intercept, and mean ± SD difference between the methods reflect the small measurement error of the methods. Furthermore, derived and volumetric quantifications of the longitudinal components of SV are both obtained by cine images and thus are geometrically related. The present study shows an excellent agreement between direct volumetric methods and the flow-based determination of the radial and longitudinal components of cardiac pumping. Thus, the present study has used flow-based methods to independently confirm the validity of the volumetric method for using epicardial contours to measure the longitudinal components of left, right, and total heart pumping.
Further Studies
Our group (6) has previously reported a similar THVV in controls and patients with a decrease in both ejection fraction and AVPD because of ischemic heart disease. This might be explained by a similar contribution of radial function to SV in patients compared with controls. We have also shown a similar LVSVAVPD% between controls and patients with dilated cardiomyopathy (7); however, further studies are needed to demonstrate differences between healthy subjects and patients with different cardiac diseases. The longitudinal function of the RV has been found to decrease with age, and it was suggested that this was compensated by an increased radial function (22). The subjects of the present study were young; therefore, further studies in older healthy subjects are warranted.
Limitations
Cine and flow images were not obtained simultaneously, and heart rate might change during the examination. Heart rate, however, was stable between cine and flow images, in all subjects except subject 1, and the differences in heart rate were not significant. Thus variation in heart rate probably is not a major limitation of this study. Cine images were acquired during end-expiratory breath hold, whereas all flow images except three were acquired during free breathing. The difference in intrathoracic pressure between these two states might influence the cardiac volumes and variations of heart volume variation. There was, however, no difference in THVV calculated by flow images and cine images, implying that this is not a major factor. Through plane motion limits any two-dimensional technique, when motion occurs in the plane perpendicular to the imaging plane. In the present study, the measurements of the ventricles were 1) long-axis images to calculate AVPD, 2) short-axis image stacks in end diastole and end systole to calculate either the SV in three dimensions or the largest end-diastolic short-axis area, and 3) three-dimensional reconstruction of the RV to calculate the SV obtained by atrioventricular-plane motion. Neither of these measurements is affected by through plane motion in a significant way; therefore, this is not a limiting factor in the present study. There is motion or strain in three directions at the level of myocardial tissue. However, examining the function of the myocardial tissue as such, i.e., radial, angular and longitudinal strain, was not the aim of the present study. The present study sought to assess the longitudinal and radial movements of the outer contours of the heart. Longitudinal motion occurs along the long axis of the heart and is by definition unidirectional. Radial motion is two dimensional in the short-axis plane of the heart. Together, longitudinal and radial motion account for all three dimensions of movement. Angular movement was not assessed, and it does not affect volumetric measurements.
Measurements of THVV by MRI is a relatively new technique (1, 5, 6, 10, 28, 39), and as such it is important to discuss the accuracy of this method. We have previously shown excellent agreement of volumetric measurements when performing quantification of the THVV in frontal, sagittal, axial, and short-axis views and minimal intraobserver variability of flow quantifications (5). There was also excellent agreement between volumetric and flow quantification of THVV (5), and similar results have been obtained in the present study. The fact that two different independent methods show excellent agreement of THVV (r = 0.95, P < 0.001, and bias of 4.0 ± 7.8 ml) supports the accuracy of the measurements of the THVV.
All contents of the pericardial sack (atria, ventricles, coronary vessels, and roots of the great vessels) were included in the measurements. If the volume of the root of either of the great vessels changed, the volumes of the pericardial content would be affected and would be measured. Also, flow in the vessels is affected by volume changes at the root of the vessels and would therefore be measured by the velocity-encoded images. Therefore, the volumes detected by both methods include volume changes of the vessels. Furthermore, there are variations in coronary flow throughout the cardiac cycle. However, the coronary flow originates in the aortic cusp and ends in the right atrium, and aortic flow was measured distal to the sinus of Valsalva and the origin of the coronary arteries. Thus the coronary circulation occurs within the volume of measurements and will not affect the measurements. The volumetric measurements were performed at the pericardial border, and both coronary arteries and veins are within the pericardium and would therefore be included in both the volumetric and flow measurements throughout the cardiac cycle.
Conclusions
This study has shown that the radial function of the ventricles causes the predominant part of the THVV. Furthermore, this study used independent and concordant measures to quantitatively show that the longitudinal contribution to RV function is 80% compared with 60% for the LV.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
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 METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
, Mean and SE (error bars).
) and left heart (
) and the radial function of the corresponding ventricle, respectively. Radial function was calculated by subtracting the longitudinal function (SVAVPD) from the SV. B: relationship between the THVV and the TSV generated by radial function. In both A and B, the solid lines indicate the line of identity and the broken lines indicate the regression line. Note that the right heart has lower volume variation and radial function than the left heart.
