Abstract

The dynamic interaction between subendocardial and subepicardial fibre helices in the left ventricle (LV) leads to a twisting deformation, which has an important role in LV function. This study sought to assess the influence of cardiac shape on LV twist in the normal and dilated human heart. The study comprised 45 dilated cardiomyopathy (DCM) patients and 60 for age- and gender-matched healthy volunteers. Speckle tracking echocardiography was used to determine basal and apical LV peak systolic rotation (Rotmax) and instantaneous LV peak systolic twist (Twistmax). LV sphericity index was calculated by dividing the LV maximal long-axis internal dimension by the maximal short-axis internal dimension at end-diastole. A parabolic relation between the sphericity index and apical Rotmax or Twistmax was identified in the total study population (R2 = 0.56 and R2 = 0.54, respectively; both P < 0.001) and healthy volunteers (R2 = 0.39 and R2 = 0.25, respectively; both P < 0.001), whereas these relations were linear in DCM patients (R2 = 0.40 and R2 = 0.43, respectively; both P < 0.001). In a multivariate analysis, LV sphericity index was the strongest independent predictor of apical Rotmax and Twistmax. In conclusion, LV apical rotation and twist are significantly influenced by LV configuration. Taking the important function of LV twist into account, this finding highlights the vital influence of cardiac shape on LV systolic function.

  • cardiac mechanics
  • cardiac function
  • left ventricular twist
  • dilated cardiomyopathy
  • echocardiography

the dynamic interaction between subendocardial and subepicardial fiber helices in the left ventricle (LV) leads to a twisting deformation (14). This twisting deformation plays an important role in optimizing LV ejection fraction (LV-EF) (24). Recently, speckle-tracking echocardiography (STE) has been introduced as a new method for angle-independent quantification of LV twist (9, 22). Speckles are natural acoustic markers that occur as small and bright elements in conventional grayscale ultrasound images. The speckles are the result of constructive and destructive interference of ultrasound, back-scattered from structures smaller than a wavelength of ultrasound (4). This gives each small area a rather unique speckle pattern that remains relatively constant from one frame to the next. Therefore, a suitable pattern-matching algorithm can identify the frame-to-frame displacement of a speckle pattern, allowing myocardial motion to be followed in two dimensions.

Normally, looking at the heart directly from the anterior wall, the LV fibre helix angle varies from approximately −60° at the subendocardium to +60° at the subepicardium, with the mid-wall circumferential fibres at 0° (5, 13). Shortening of this counterdirectional mantle of muscle fibers results in a wringing movement of the LV that propels blood out of the LV cavity. In a theoretical model by Taber et al. (29), peak systolic twist approximately doubled with a change in de epicardial/endocardial fiber angles from +90°/−90° to +60°/−60°, underscoring the importance of the arrangement of myocardial fibers for LV twist. Furthermore, in patients with dilated cardiomyopathy (DCM), differences in short-axis and long-axis dilatation result in changes in fiber angles that may further impair LV twist and thus cardiac function (12). In the present study, it was hypothesized that a change in cardiac shape may indeed lead to a change in the arrangement of myocardial fibers and thereby to a change in LV twist. To gain insight into cardiac (patho)physiology, this study sought to assess the influence of cardiac shape on LV rotation and twist in the normal and dilated human heart.

METHODS

Study participants.

The study population consisted of 45 DCM patients (mean age 40 ± 14 yr, 22 men, LV-EF 33 ± 13%) and 60 age- and gender-matched healthy volunteers (mean age 38 ± 15 yr, 30 men, LV-EF 62 ± 7%) in sinus rhythm, with good echocardiographic image quality that allowed for complete assessment of LV rotation of all myocardial segments at both the basal and apical LV level. DCM patients were divided into three subgroups of 15 patients according to LV-EF (group I: 20–30%; group II: 31–40%; and group III: 41–50%). Healthy volunteers were without hypertension or diabetes and had normal left atrial dimensions, LV dimensions, and LV function. DCM was characterized by LV chamber enlargement and systolic dysfunction, based on current guidelines (18). All DCM patients had undergone coronary angiography to exclude significant coronary artery disease. An informed consent was obtained from all subjects, and the institutional review board approved the study.

Echocardiography.

Echocardiographic studies were performed by a single, experienced sonographer (W. B. Vletter) with a commercially available system (iE33, Philips, Best, The Netherlands) equipped with a broadband S5-1 transducer (frequency transmitted 1.7 MHz, received 3.4 MHz). All echocardiographic measurements were averaged from three heartbeats. From the second harmonic M-mode recordings, the following data were acquired: left atrial size, LV end-diastolic anteroseptal and inferolateral wall thickness, and LV end-diastolic and end-systolic dimension. The LV sphericity index was calculated by dividing the LV maximal long-axis internal dimension by the maximal short-axis internal dimension at end-diastole (Fig. 1) (16). In a subset of subjects (20 healthy volunteers and 10 DCM patients), LV sphericity index was measured both by two-dimensional and three-dimensional echocardiography. Analysis of three-dimensional echocardiography datasets was performed on a QLAB workstation using 3D-Advanced Quantification 6.0 (Philips) as described previously (27). Comparison of LV sphericity index measured by either two-dimensional or three-dimensional echocardiography and correlations between LV sphericity index measured by both methods and other parameters revealed no differences, most likely caused by the scrupulous attempts to acquire a non-foreshortened LV during two-dimensional echocardiography by our single, highly experienced sonographer (W. B. Vletter, over 30 years of experience). Therefore, for practical reasons, LV sphericity index by two-dimensional echocardiography was used in the main part of the study. LV mass was assessed with the two-dimensional area-length method. LV-EF was calculated from LV volumes by the modified biplane Simpson rule (15). The cavity-to-wall ratio was calculated by dividing the end-diastolic LV dimension by the sum of the anteroseptal and inferolateral wall thickness. From the mitral-inflow pattern, peak early (E) and late (A) filling velocities, E-to-A ratio, and E-velocity deceleration time were measured. Tissue Doppler was applied end-expiratory in the pulsed-wave Doppler mode at the level of the inferoseptal side of the mitral annulus from an apical four-chamber view to measure the velocity of the mitral annular early diastolic wave (Em). To acquire the highest wall tissue velocities, the angle between the Doppler beam and the longitudinal motion of the investigated structure was minimized. The spectral pulsed-wave Doppler velocity range was adjusted to obtain an appropriate scale. The degree of mitral regurgitation (grades I–IV) was assessed as the mid-systolic jet area relative to left atrial area in the apical four-chamber view (10).

Fig. 1.

Left ventricular sphericity index (maximal left ventricular long-axis internal dimension divided by the maximal short-axis internal dimension at end-diastole) in a healthy volunteer (sphericity index 1.9; left) and a dilated cardiomyopathy patient (sphericity index 1.4; right).

To optimize speckle tracking, two-dimensional grayscale harmonic images were obtained at a frame rate of 60–80 frames/s. Parasternal short-axis images at the LV basal level (showing the tips of the mitral valve leaflets) with the cross section as circular as possible were obtained from the standard parasternal position, defined as the long-axis position in which the LV and aorta were most in-line with the mitral valve tips in the middle of the sector. To obtain a short-axis image at the LV apical level (just proximal to the level with end-systolic LV luminal obliteration), the transducer was positioned 1 or 2 intercostal spaces more caudal, as previously described (36). From each short-axis level, three consecutive end-expiratory cardiac cycles were acquired and transferred to a QLAB workstation (Philips, Best, The Netherlands) for offline analysis.

Speckle tracking analysis.

Analysis of the datasets was performed by STE using QLAB Advanced Quantification Software version 6.0 (Philips), which was recently validated against magnetic resonance imaging for assessment of LV twist (8). To assess LV rotation, six tracking points were placed manually (after gain correction) on an end-diastolic frame on the mid-myocardium in each parasternal short-axis image. Tracking points were separated about 60° from each other and placed on 1 (30°, anteroseptal insertion into the LV of the right ventricle), 3 (90°), 5 (150°), 7 (210°), 9 (270°, inferoseptal insertion into the LV of the right ventricle), and 11 (330°) o'clock to fit the total LV circumference. After the tracking points were positioned, the program tracked these points on a frame-by-frame basis by use of a least-squares global affine transformation. The rotational component of this affine transformation was then used to generate rotational profiles.

Data were exported to a spreadsheet program (Excel, Microsoft, Redmond, WA) to determine basal and apical LV peak systolic rotation during ejection (Rotmax), and Twistmax (defined as the maximal value of instantaneous apical LV systolic rotation − basal LV systolic rotation).

Statistical analysis.

Statistical analysis was performed using programs available in the SPSS statistical package (SPSS, version 15.0, Chicago, IL). Measurements are presented as means ± SD. All variables were tested for normal distribution of the data. Means were compared using Student's t-test. Regression of LV rotation parameters against parameters of LV dimension and LV-EF was performed. A quadratic model was used to investigate the relation between LV sphericity index and basal and apical Rotmax and Twistmax in healthy volunteers, because a parabolic relation was expected. Other correlations were tested using a linear model. Multivariate regression analysis was performed to look for independent associations. Squared values of the LV sphericity index were used to adjust for the parabolic relations identified in the univariate analyses. A P value of <0.05 was considered statistically significant. Intraobserver and interobserver variability for LV twist in our center are 6 ± 6% and 9 ± 5%, respectively (33).

RESULTS

Clinical and echocardiographic characteristics of the study population are shown in Table 1. Heart rate (74 ± 16 vs. 65 ± 12 beats/min; P < 0.001), LV end-systolic dimension (5.0 ± 1.1 vs. 3.3 ± 0.5 cm; P < 0.001) and volume (117 ± 57 vs. 45 ± 14 ml; P < 0.001), LV end-diastolic dimension (6.2 ± 0.8 vs. 4.9 ± 0.5 cm; P < 0.001) and volume (169 ± 58 vs. 115 ± 23 ml; P < 0.001), LV mass (227 ± 83 vs. 149 ± 53 g; P < 0.001), left atrial size (4.2 ± 0.7 vs. 3.6 ± 0.5 cm; P < 0.001), E-to-A ratio (1.8 ± 0.9 vs. 1.5 ± 0.5; P < 0.05), and E-to-Em ratio (12.2 ± 5.6 vs. 7.2 ± 1.9; P < 0.001) were increased, whereas LV sphericity index (1.5 ± 0.2 vs. 1.9 ± 0.3; P < 0.001) and LV-EF (33 ± 13 vs. 62 ± 7%; P < 0.001) were decreased in DCM patients compared with healthy volunteers.

View this table:
Table 1.

Clinical and echocardiographic characteristics of the study population

Relation of LV rotation to LV dimension and function in the total study population.

Regression analysis revealed a positive linear relation of apical Rotmax (R2 = 0.40; P < 0.001) and Twistmax (R2 = 0.39; P < 0.001) to LV-EF (Fig. 2). A parabolic relation between the sphericity index and apical Rotmax (R2 = 0.56; P < 0.001) or Twistmax (R2 = 0.54; P < 0.001) was identified (Fig. 3). The cavity-to-wall ratio showed a negative linear relation with apical Rotmax (R2 = 0.29; P < 0.001) and Twistmax (R2 = 0.31; P < 0.001). There were no relationships between LV mass and apical Rotmax or Twistmax.

Fig. 2.

Linear model of regression, showing the linear relation between left ventricular ejection fraction and peak left ventricular apical rotation (left) and twist (right). Triangles denote apical rotation, squares denote basal rotation, and circles denote twist (open symbols are healthy volunteers; closed symbols are dilated cardiomyopathy patients).

Fig. 3.

Quadratic and linear models of regression, highlighting the parabolic relation between left ventricular sphericity index and peak left ventricular apical rotation (left) and twist (right). Triangles denote apical rotation, squares denote basal rotation, and circles denote twist (open symbols are healthy volunteers; closed symbols are dilated cardiomyopathy patients).

Relation of LV rotation to LV dimension and function in healthy volunteers.

In healthy volunteers, no significant relation could be identified between LV rotation parameters and LV-EF (Fig. 2). However, the parabolic relation between the sphericity index and apical Rotmax (R2 = 0.39; P < 0.001) or Twistmax (R2 = 0.25; P < 0.001) remained present (Fig. 3). The cavity-to-wall ratio showed a negative linear relation with apical Rotmax (R2 = 0.16; P < 0.01) and Twistmax (R2 = 0.21; P < 0.001). There were no relationships between LV mass and apical Rotmax or Twistmax.

Relation of LV rotation to LV dimension and function in DCM patients.

In DCM patients, a positive linear relation between LV-EF and apical Rotmax (R2 = 0.11; P < 0.05) or Twistmax (R2 = 0.12; P < 0.05) was revealed by regression analysis (Fig. 2). Also, a positive linear relation between the sphericity index and apical Rotmax (R2 = 0.40; P < 0.001) or Twistmax (R2 = 0.43; P < 0.001) could be identified (Fig. 3). In the three LV-EF subgroups of DCM patients, these relationships remained observable (LV-EF 20–30%: R2 = 0.42 and R2 = 0.46, respectively; LV-EF 31–40%: R2 = 0.28 and R2 = 0.36, respectively; LV-EF 41–50%: R2 = 0.31 and R2 = 0.44, respectively; all P < 0.05) (Fig. 4). There were no significant differences in age [38 ± 12 vs. 39 ± 15 vs. 43 ± 1 yr; P = not significant (NS)], heart rate (76 ± 15 vs. 73 ± 14 vs. 73 ± 13 beats/min; P = NS), systolic blood pressure (116 ± 14 vs. 120 ± 16 vs. 120 ± 15 mmHg; P = NS), diastolic blood pressure (68 ± 16 vs. 68 ± 15 vs. 71 ± 16 mmHg; P = NS), mitral regurgitation grade (1.8 ± 0.6 vs. 1.6 ± 0.8 vs. 1.6 ± 1.1; P = NS), and E-to-Em ratio (14 ± 5 vs. 12 ± 4 vs. 12 ± 5) between the subgroups with LV-EF 20–30%, 31–40%, and 41–50%, respectively. There were no relationships between the cavity-to-wall ratio or LV mass and apical Rotmax or Twistmax.

Fig. 4.

Linear model of regression, showing the linear relation between peak left ventricular apical rotation and twist in three subgroups, based on left ventricular ejection fraction, of dilated cardiomyopathy patients. Triangles denote apical rotation, squares denote basal rotation, and circles denote twist.

Multivariate analysis.

In a multivariate linear regression model applied to the total study population, age and LV sphericity index were identified as independent predictors of apical Rotmax (beta = 0.399, P = 0.059 and beta = 0.534, P = 0.001, respectively) and Twistmax (beta = 0.431, P = 0.071, and beta = 0.616, P = 0.000) (Table 2). Thus LV sphericity index was the strongest independent predictor of apical Rotmax and Twistmax.

View this table:
Table 2.

Univariate and multivariate regression analysis of variables related to left ventricular rotation and twist

DISCUSSION

LV twist describes the instantaneous circumferential motion of the apex with respect to the base of the heart and has an important role in LV ejection (14). To the best of our knowledge, the present study is the first to investigate the influence of LV shape, and presumably fiber orientation, on LV twist in the human heart. The main finding of the present study is that this twisting deformation of the human LV is influenced by LV configuration, both in normal subjects and in patients with DCM.

Influence of cardiac shape on LV twist.

In the late 1970s, Hutchins et al. (12) suggested that an efficient LV configuration would be a compromise between a spherical shape that would need the least energy for diastolic filling and a tubular shape that would permit maximal conversion of myocyte contraction into cavitary pressure increase. In models of LV mechanics, it has subsequently been shown that the LV myocardial fiber architecture is important for LV function (2, 13). Adequate pressure generation is probably primarily produced via circumferential mid-wall fibers but also by subendocardial and subepicardial spiraling fibers that generate LV twist and shortening of the LV long axis (13). In previous work, it has been shown that LV twist is of fundamental importance to systolic LV function (2, 14, 25, 26). In the present study, LV-EF was also positively correlated to Twistmax. The important role of apical Rotmax was underscored by the fact that LV-EF correlated with apical but not with basal Rotmax. LV twist may also be an important manner to equalize transmural differences in sarcomere shortening, end-systolic fiber stress, and contractile work during the ejection phase (2).

Myofiber morphology has either been described based on orientation of individual fibers or as multiple myocyte “sheet” arrangements separated by extensive “sheet-cleavage” planes (1, 3). The myofiber helix angle, representing the angle between the myofibers, as projected onto the circumferential-longitudinal plane, and the circumferential axis, was introduced for quantification of fiber orientation by Streeter et al. (28). This angle changes continuously from the subendocardium to the subepicardium, typically ranging from +60° at the subendocardium to −60° at the subepicardium (5, 13). In a theoretical model, Taber et al. (29) showed that Twistmax approximately doubles with a change in the myofiber helix angle from 90° to 60°. For the present study, it was hypothesized that the supposedly optimal myofiber helix angle of 60° is related to a certain LV sphericity index and that, therefore, either an increase or decrease in LV sphericity index would result in a decrease in LV twist, even in healthy volunteers.

Interestingly, in normal hearts, the LV sphericity index had a parabolic relation with apical Rotmax and Twistmax. A LV sphericity index of ∼2.1 was associated with the highest Twistmax, lower and higher sphericity indexes were associated with less Twistmax. Assuming that the LV sphericity index may be related to fiber orientation (29), a decreased LV sphericity index may be related to decreased Twistmax due to a decreased fiber angle, whereas an increased LV sphericity index may be related to decreased Twistmax as well, however, due to an increased fiber angle. Although definite proof is lacking because of the absence of direct exploration of fiber orientation, the findings of the present study seem to support the hypothesis that alterations in fiber orientation influence Twistmax.

In healthy volunteers, increased wall thickness, relative to the short-axis dimension of the LV cavity, was also associated with increased apical Rotmax and Twistmax. During the ejection phase, both the endocardial and epicardial spiraling fibers are electrically activated. However, the epicardial fibers govern the direction of LV twist, mainly owing to their longer arm of movement. It can therefore be anticipated that the epicardial fibers may become even more dominant when the LV walls are thicker, in particular relative to LV cavity dimension, because in such walls the differences in the arms of movement will be greater.

Clinical implications.

Alterations in LV geometry, as seen in DCM, may have several functional effects. As a consequence of dilation and systolic dysfunction, the LV takes on a more spherical geometry. In prior studies, it has been shown that increasing spherical geometry with apical and lateral displacement of the papillary muscles results in functional mitral regurgitation (6, 20). From the present study, it may be concluded that LV remodeling may also lead to decreased LV twist. The LV sphericity index, as a parameter of LV geometry, varied from 1.2 to 1.8 in DCM patients and showed a positive linear relation with apical Rotmax and Twistmax. Even when DCM patients with similar LV-EF were studied, the LV sphericity index remained positively correlated to both LV rotation parameters. In fact, the LV sphericity index was the strongest independent predictor of both apical Rotmax and Twistmax. Prevention of LV remodeling favorably impacts the untoward natural history of heart failure (17), which may be, at least partly, related to the preservation of LV twist. Therefore, assessment of LV twist may be of importance in the evaluation and guidance of therapies in DCM, especially if these therapies are supposed to be of influence on cardiac remodeling. Future studies are warranted to test this hypothesis.

Contemporary guidelines are clear on how to diagnose DCM (18). Furthermore, in the present study, there was considerable overlap of LV twist between healthy volunteers and DCM patients. Therefore, LV twist may not have any value as a diagnostic parameter with added discriminative value to distinguish DCM patients from normal subjects.

Apart from any potentially direct clinical implications, assessment of LV twist gives important insight into cardiac (patho)physiology (35). For example, as mentioned before, from the present study, it may be concluded that deterioration of LV twist may be one of the mechanisms through which LV remodeling effectuates its negative effect on prognosis. In addition, to properly interpret LV twist measurements, clinicians need a fundamental understanding of the underlying mechanics. Although findings of previous studies (7, 19) have suggested a correlation between abnormal function and abnormal LV twist, the basis of this relationship is not yet fully understood. The present study was performed to investigate the influence of cardiac shape on LV twist and to thereby provide increased rationale for interpreting clinical measures of LV twist. Our group recently studied LV twist in hypertrophic cardiomyopathy as well and found that apical Rotmax and Twistmax are dependent on the pattern of LV hypertrophy (32). Like in the present study, noninvasive assessment of LV twist by STE was used to gain insight into cardiac (patho)physiology. Conceptually, ideal therapeutic agents should target the underlying mechanisms that cause LV dysfunction. Therefore, expanding knowledge about cardiac (patho)physiology may be a first step in the process of developing new or improving current treatment strategies.

Limitations

In previous STE studies, Twistmax in control subjects varied widely, from 9° in a study by Takeuchi et al. (30) to 20° in a study by Tanaka et al. (31). In our study, a wide range of Twistmax in healthy volunteers was present as well. Apart from our new observation on the influence of LV configuration, (measured) apical Rotmax is significantly influenced by age (21, 23, 34) and correct visualization of the true LV apex (36). Therefore, all our studied patient groups (DCM vs. controls and the three DCM LV-EF groups) were matched for age, and a multivariate analysis was performed. Also, it seems reasonable to assume that the acquisition of the true LV apex will be equally successful in the different studied groups. The transducer position was optimized to acquire the true LV apex, even in the very elongated ventricles (36). In the near future, three-dimensional STE might provide a definite solution for this important latter limitation of two-dimensional STE. Furthermore, there may have been other, unidentified confounding factors in DCM patients, such as the length of the myocytes (i.e., length-tension relationship), myocyte loss, and/or fibrosis that may influence LV twist. True assessment of the independent influence of cardiac shape on LV twist would require an animal model or phantom study to be able to induce isolated changes in cardiac shape. Finally, as mentioned before, the identified influence of cardiac shape on LV twist may be mediated by the influence of cardiac shape on fiber orientation. To test this hypothesis, studies of pathological specimens or magnetic resonance myocardial fiber-orientation mapping (11) may be required, which were unfortunately not available in the present study.

In conclusion, LV apical rotation and twist are significantly influenced by LV configuration. Taking the important function of LV twist into account, this finding highlights the vital influence of cardiac shape on LV systolic function.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
View Abstract