Changes in overall ventricular myocardial architecture in the setting of a porcine animal model of right ventricular dilation

Right ventricular dilation is often caused by pulmonary valve regurgitation, which is predominantly seen in congenital heart disease following balloon dilation of critical pulmonary stenosis, or perforation of valvar pulmonary atresia [1]. The most important, and well-described, clinical context for pulmonary regurgitation, however, is in patients with repaired tetralogy of Fallot, the most common cyanotic congenital heart disease [2].

Even though the treatment of tetralogy of Fallot has improved dramatically in recent times, serious complications, like right ventricular dilation as a result of longstanding pulmonary regurgitation, still develop as an almost unavoidable result of the treatment [3, 4]. These results in an increased risk of heart failure, severe arrhythmias, and sudden death [1, 2, 4]. Dilation is a natural consequence of right ventricular volume overload. The ventricles have to dilate in order to accommodate the increased blood load because the two other possible responses, namely decreased end-systolic volume or increased heart rate, are more difficult to achieve. As of today, it is not known why the right ventricle, in the long term, so poorly tolerates dilation, nor why this ultimately will result in heart failure. Rearrangement of the cardiomyocytes making up the ventricular walls has been shown in various cardiac diseases [5‐7] and may very well be a contributing cause of heart failure in the dilated right heart. To achieve better understanding of heart failure as a consequence of right ventricular dilation, it is axiomatic that further knowledge of the myocardial architecture of the right ventricle is needed in the setting of these pathological conditions. Previously, it has been manual dissection and histological examination of pieces of the myocardium that have been the methods of choice when assessing mural architecture [8‐10]. As we have pointed towards earlier, the major drawback of these approaches is that they fail to assess the microstructure of the myocardium as a three-dimensional entirety. Diffusion tensor cardiovascular magnetic resonance now provides the means to circumvent this problem, even though, as yet, the technique is incapable of visualising the individual cardiomyocytes. It has been extensively used to characterise myocardial architecture in autopsied [11] and beating hearts [12], in both health [13] and disease [6], and has been used in a variety of species. The technique uses the spontaneous self-diffusion of water as a surrogate measure of the average long-axis orientation of cardiomyocytes within defined portions of the myocardium contained within a voxel. The diffusion pattern within each voxel is described as a mathematical construct called a tensor. A tensor is defined by its three orthogonal eigenvectors [14]. It is generally accepted that the primary eigenvector of the diffusion tensor follows the orientation of the long axis of the aggregated cardiomyocytes (Fig. 1). When assessing the orientation of the tertiary eigenvector, it is found to change its orientation through the cardiac cycle [15]. In disease states, furthermore, this change is associated with ventricular thinning or thickening [6]. It has been suggested that the tertiary eigenvector reflects the orientation of anatomical substructures of cardiomyocytes that are mainly of a flattened nature, but otherwise of unknown extent and appearance. These substructures have been referred to as lamellar units [16, 17], sheets [18], or sheetlets [5]. All these names have advantages and disadvantages. As of today no study has provided a complete anatomical description of these substructures. It has been stated that they are approximately four to six cardiomyocytes in thickness [19], which is probably a gross oversimplification according to more recent studies [17], but the extent and size of the structures are still unknown. It is, therefore, difficult to assess the orientation of such poorly defined structures, and even more difficult to assign a proper name for them. In this light, in spite of having done so in previous contributions, we now believe it is preferable to refrain from characterising the aggregations by name at present.

We have previously investigated the alterations in the right ventricular myocardial architecture in two animal models of increased right ventricular pressure overload [6, 20]. We were able to show that the myocardial architecture does change as a consequence of increased afterload brought upon by persistent pulmonary hypertension in the newborn [6]. Since dilation arises from a different pathogenetic mechanism as compared to hypertophy, this raised the question as to whether such morphological changes are also to be found in the setting of right ventricular dilation. For this purpose, we have introduced a porcine model of right ventricular dilation [21] and heart failure [22]. In this study, we aimed to compare the orientations of the aggregated cardiomyocytes in the normal and dilated right ventricular myocardium.

To the best of our knowledge, this is the first study to show that right ventricular dilation from pulmonary regurgitation leads to alterations in overall ventricular mural architecture. Significant remodelling was found in both ventricles, despite the fact that pulmonary regurgitation is often considered a strictly right ventricular pathology. We found that the helical ang​les ​of ​right ventricular cardiomyocytes approach a more circumferential orientation as a consequence of right ventricular dilation (Figs. 3, 4 and 9). We also observed an increase in the proportion of intrusional angles close to zero degrees in the right ventricle, along with an increase in the proportion of E3 angles around −90° as seen in the diastolic phase. The latter finding is as anticipated for the alterations of right ventricular helical and intrusional angles. We found a similar tendency in terms of helical angles measured in the left ventricle, but here the mural architecture was seemingly changed in the opposite direction in terms of the intrusional and E3 angles. The latter attributes assumed a more systole-like orientation. In the left ventricle, furthermore, the proportion of intrusional angles around zero degrees decreased in the setting of right ventricular dilation, while the E3 angles changed towards a more horizontal alignment, again as seen in the systolic heart [5]. Ferreira and associates suggested this conformation to be brought upon by failure of diastolic relaxation. Hyldebrandt and co-workers assessed the current model by conductance catheter technique [22]. They found an unaltered left ventricular isovolumetric relaxation constant (tau) whereas left ventricular compliance was decreased as judged by a significant increase in the end-diastolic pressure volume relationship. Hence left ventricular diastolic function is certainly impaired as is often seen in hypertrophy [28]. It is, therefore, very interesting that we were unable to show hypertrophy when comparing myocardial thickness between groups. This suggests that intramural myocardial remodeling precedes detectable transmural thickening. We found little if any change in the architecture of the septum, apart from a strictly localised decrease in the intrusional angles measured in the right ventricular sub-endocardium. Diffusion tensor imaging studies of the right ventricle are a rare sight in the published literature. We are aware of only three studies, of which our group has contributed two [6, 20, 29]. Our current study is, to the best of our knowledge, the first to explore the influence of right ventricular dilation on myocardial architecture. In our previous study of right ventricular hypertrophy produced by pulmonary banding [20], we found no differences between the hypertrophied hearts and their controls in terms of helical angles, but in that study we did not assess either intrusional angles or E3 angles. Our scan resolution at that time was also 64 times lower than in both our newer studies, which might mask smaller differences. As we have previously elucidated right ventricular pressure and volume overload indeed leads to two different manifestations of heart failure [22, 30]. In our most recent study, of right ventricular pressure overload​ in​persistent pulmonary hypertension in the newborn [6], ​we found alterations in myocyte angulations with both preserved cavitary volumes and with universal myocardial hypertrophy. This is the inverse pattern compared with our present study of right ventricular volume overload, in which we found that the hearts with dilation had increased their right ventricular volume without changing either of their ventricular mural thicknesses. At present, we have investigated the thin right ventricular wall with an appropriate imaging resolution comparable to that of other studies, and we have utilised an animal model that is anatomically and physiologically well documented [21, 22]. We have also achieved a degree of pulmonary regurgitation in our model comparable with that of earlier studies. Dilated cardiomyopathy in the left ventricle has been assessed with diffusion tensor imaging by several groups [7, 31, 32]. Li and associates, using genetically modified hamsters, were unable to show any changes in helical and transverse angles, but their specimens were scanned using only 6 diffusion encoding directions, and with a very large slice thickness [31]. The alignment of the myocytes, therefore, is averaged over quite large myocardial areas, despite the use of a reasonable in-plane resolution. The same is the case in a recent study in humans by Nielles-Vallespin and associates [7]. They were, however, able to detect a significant change in the orientation of the aggregated cardiomyocytes in spite of low resolution while the helical and transmural angles were unaltered. Like in our study the myocardium of left ventricular dilated cardiomyopathy had a configuration towards diastole. Contrary to this finding the hearts in the study of von Deuster and co-workers seems to be configured in a state resembling systole and, moreover, they detect an overall increase in the helical angle [32]. This difference can most likely be attributed to the difference in baseline characteristics between the studies and to the use of strain correction in the von Deuster study. Here the hearts are barely dilated, but merely hypertrophied as opposed to the hearts in the Nielles-Vallespin study, which are significantly morphologically dilated. In our study with higher resolution, we found changes in all angles measured. In the right ventricle in particular, the helical angles approached a more circumferential orientation. This finding supports the “wicker basket” analogy originally proposed by Streeter and colleagues in 1973 [33], as shown in our Fig. 10. On this basis, we presume that the change noted in helical angulation of the cardiomyocytes in the right ventricular wall is simply a mechanical consequence of ventricular dilation. The ideal response of the ventricle to acute volume overload is to enlarge. An increased end-diastolic volume then causes a right-shift on the Starling curve, leading to altered working conditions for the individual cardiomyocytes. When this is normalised by secondary hypertrophy, a larger, but morphologically normal, right ventricle should cope with these alterations quite easily. A previously normal right ventricle, however, cannot expand isometrically while maintaining its normal shape, mainly because it is fixed to the left ventricle, which does not change its size. The “wicker basket” analogy shows that, if the chamber were allowed to lengthen as much as it widens, then the angles would not be disturbed (Fig. 10). In contrast, if the ventricle, as in this illustration, is obliged to undergo allometric scaling, the normal balance between the inner and outer helical angles will be disturbed. This can be dysfunctional, as it will lead to a mismatch between the contraction capabilities of the individual cardiomyocytes and the need for myocardial deformation required to maintain stroke volume [34]. As observed, the result will be a change in the architecture of the myocardial mesh. The dilation is effected mainly by changes in ventricular short-axis dimensions other than its length, and the myocardium is not capable of reforming itself in the timescale provided. Hence, the inevitable consequence of transverse dilation of the ventricle is that the aggregated cardiomyocytes will become more surface parallel in terms of decreased intrusional angles, with their helical angles flattened towards zero. This will produce an inevitable compromise between the normal balance between the helical angles of the inner and outer zones. An increased end-diastolic volume causes a right-shift on the Starling curve, leading to altered working conditions for the individual cardiomyocytes. To accommodate these changes, remodelling to some extent on a cellular level must also take place. We found right ventricular dilation with no thinning of the ventricular wall, meaning that the myocardium must have increased its mass. This augmentation of myocardial mass must be brought upon by cellular hypertrophy, because the number of cells in the myocardium cannot increase in number, at least not that much and that quick. Myocardial hypertrophy entails an increase in cell thickness, or cell length, or both. In this case the cells must mainly increase in length because the original wall thickness is maintained, but the radius of the ventricle is enlarged. Hypertrophy, however, does not allow any realignment of the myocyte orientation and hence normalization of the myocardial contractional forces. As described we have attempted to maintain the shape of the ventricles during fixation and we were unable to detect any difference in wall thickness. It is, however, impossible to reproduce the same loading conditions that apply in-vivo when imaging of the heart ex-vivo. In particular, when imaging the thin walled right ventricular myocardium, loading conditions could have a major impact. It is, therefore, not given that in-vivo imaging of the right ventricle under true loading conditions would produce the same results. This is an important limitation to this study. As previously discussed in-vivo imaging of the right ventricle is not feasible with sufficient resolution given the present developmental state of the technology. It has, however, been suggested that in-vivo diffusion tensor imaging of the interventricular septum can provide information on right ventricular function [35], but conclusions on its morphology and remodelling cannot be made from investigation of the interventricular septum. In this study we have analysed the changes in myocyte orientation in formalin fixed hearts. Given the well-known effects of formalin on tissues [36] it is reasonable to speculate that the process of fixation could skew our results. It is, however, well-documented that myocyte orientations as assessed with histology are comparable with those calculated from diffusion tensor imaging in both fresh [37] and perfusion fixed hearts [38]. We have, moreover, recently documented that the diffusion properties of cardiac tissue are comparable between fresh and perfusion fixed cardiac tissue [39].

We found an increase in the amount of circumferentially oriented cardiomyocytes in the right ventricle as was also found in the study of Sanchez-Quintana and co-workers achieved by dissecting human hearts with tetralogy of Fallot [9]. The physiological mechanisms of the myocardial remodelling observed in the presented model and in tetralogy of Fallot, however, are quite different. In this study, we have investigated a volume overload model, whereas tetralogy of Fallot is a pressure overload disease. Although the proportion of circumferentially oriented cardiomyocytes increases in the dilation model, the etiology is quite different from that observed in the setting of tetralogy. It could be argued that the increase of circumferentially oriented myocytes in tetralogy is a compensatory mechanism as part of myocardial hypertrophy, whereas in dilation the increase in circumferential myocytes is merely a disadvantageous mechanical consequence of expanding the cavity of the ventricle. Realignment of the cardiomyocytes must also lead to alterations in how the contraction of the myocytes affects the ventricle. Mathematical models as presented by Sallin and associates and ourselves have argued that the presence of a helical angle is mandatory in order to produce an ejection fraction within physiologically normal range [40, 41]. Moreover, a recent study in humans with situs inversus totalis by Khalique and co-workers found an altered helical angle pattern in the left ventricle leading to reduced torsion [42]. Even though all three of these works only study the left ventricle, it is highly likely that the myocardial rearrangement in the dilated right ventricle towards a helical angle of zero degrees is not beneficial for right ventricular cardiodynamics. It could very well be part of the explanation of the heart failure that will eventually result from ventricular dilation [22].

Our study has also shown that left ventricular myocardial remodelling is brought about by pulmonary regurgitation, as evidenced by the observed changes in the distribution of the E3 angles. The proportion of angles with a numerically low value around zero increases concomitant with right ventricular dilation. This phenomenon has been described previously, since it is the same pattern as seen when the heart approaches the systolic contractional state [5]. The systolic-like configuration of the E3 angles in the present study, however, is not associated with the anticipated mural thickening. The helical angles in the left, and especially the right, ventricle furthermore have a more diastolic configuration, with values closer to zero. Hence, there is a mismatch between the state of contraction and the configuration of the cardiomyocytes. This is in keeping with our findings in our sheep model of right ventricular pressure overload in persistent pulmonary hypertension of the newborn [6]. In this setting, we found the reverse situation, with the cardiomyocytes configured in a more diastolic state in spite of myocardial hypertrophy. In both studies, therefore, we describe types of mismatch between contraction state and myocyte architecture that could potentially aid in the explanation of heart failure in myocardial remodelling. There is great difference in literature on how to quantify the orientation of aggregations of cardiomyocytes. Some quantify them as absolute values [7], while others, such as ourselves consider the orientations with a sign. Our results underline why this is important. When contemplating Fig. 8 it is clear that positive and negative E3-angles are not equally distributed and, moreover, changes in the distributions caused by right ventricular dilation do not affect positive and negative angles equally. Differentiation between positive and negative angles is, therefore, indeed important although the functional interpretation of the differences between positive and negative E3 angles is far from clarified. In addition, we cannot unequivocally answer the question as to why right ventricular volume overload, in time, causes heart failure. Heart failure is well recognised to be a multifaceted disease, with a complex aetiology which probably also varies depending on the cause of failure [43]. We have now seen several cases of myocardial remodelling in the development of heart failure, but questions regarding the threshold relative to clinical heart failure, and the reversibility of the myocardial remodelling, have never been investigated. If this proved possible, then the specific role of remodelling in heart failure could potentially be more clearly elucidated.

Our present study indicates that remodelling as seen relative to mural architecture may play a part in the pathophysiology of heart failure in right ventricular dilation. This remodelling may simply come to pass by mechanical effects on the myocardium brought upon by pulmonary regurgitation. A mismatch is found between the alignment of the aggregated cardiomyocytes and the cardiac contractional state indicating that remodelling might not fully achieve the benefit intended.