Article Text

Heart failure and cardiomyopathy
Longitudinal rotation: an unrecognised motion pattern in patients with dilated cardiomyopathy
  1. Z B Popović,
  2. R A Grimm,
  3. A Ahmad,
  4. D Agler,
  5. M Favia,
  6. G Dan,
  7. P Lim,
  8. F Casas,
  9. N L Greenberg,
  10. J D Thomas
  1. Division of Cardiovascular Medicine, Cleveland Clinic Foundation, Cleveland, Ohio
  1. James D Thomas, MD, Department of Cardiology, Desk F-15, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195, USA; thomasj{at}


Background: Heart failure patients who are candidates for CRT frequently display longitudinal rotation (LR) – a swinging motion of the heart when imaged in a horizontal long-axis plane.

Objectives: To identify the magnitude and predictors of LR in patients with ischaemic (ICM) and idiopathic dilated (DCM) cardiomyopathy, and to assess predictive value of LR in patients undergoing cardiac resynchronisation therapy (CRT).

Design and setting: A retrospective study in a tertiary heart care setting.

Methods: Echocardiography was performed in 45 ICM and 41 DCM patients who were CRT candidates and 16 control subjects. Global LR, segmental strains and segmental LR were assessed from echocardiograms using speckle tracking. Repeat echocardiography >40 days after the beginning of CRT was performed in 64 patients.

Results: While DCM patients with QRS duration of both <130 ms and ⩾130 ms displayed significant clockwise LR (p<0.001 for both vs 0), ICM patients and control subjects had LR that did not differ from 0. The most significant LR predictor was end-diastolic volume (p<0.001) followed by the absence of ischaemia (p<0.001) and QRS duration (p = 0.05). DCM patients with prominent clockwise LR had lower septal but higher lateral strains than DCM patients with minimal LR, or ICM patients with counterclockwise LR. LR correlated with decrease of end-systolic volume in DCM (r = 0.49, p = 0.004), while no relationship was observed in ICM.

Conclusion: Clockwise LR is linked to presence of DCM, with the small impact of QRS duration. LR is a moderately strong predictor of end-systolic volume decrease during CRT in DCM.

View Full Text

Statistics from

When a normal heart is imaged in any long-axis plane, it can be appreciated that the left ventricle contracts by two actions, thickening of its walls and shortening of its long axis dimension.1 As the long axis shortens, the base of the heart acts like a piston pushing itself towards the apex.2 During this period, septal and lateral sides move in concert.3 In contrast, the apex is relatively stationary, with very low velocities.3

Heart failure patients, however, often display a swinging motion, with the apex of the heart appearing to move in a clockwise rotation during systole as viewed from the apical four-chamber view by echocardiography. In contrast to short-axis rotation of the normal heart seen in cross-sectional images, this rotational motion occurs on either side of the cardiac long axis, and is therefore termed longitudinal rotation (LR). In our experience, patients in whom LR is observed have enlarged hearts, frequently with bundle branch block on the electrocardiogram, and are often candidates for cardiac resynchronisation therapy (CRT). We have also observed that LR often disappears during follow-up of CRT.

Our primary aim in this retrospective study was to identify the magnitude of LR in patients with ischaemic (ICM) or idiopathic dilated cardiomyopathy (DCM). Our secondary aims were to determine predictors of LR and to discern whether LR is associated with specific abnormalities in segmental contraction. Finally, we assessed whether LR carries any predictive value in patients undergoing CRT.


Study population

We performed a search of our echocardiographic database to identify patients who were candidates for CRT and in whom pre-implant echocardiogram of satisfactory quality was performed on a Vivid 7 (GE Medical) echocardiography instrument. We identified 86 patients, and they form the patient population of this study. Sixty-four out of 86 patients also had a late post-CRT follow-up echocardiogram performed 40 days to 18 months (median 192 days, first and third quartiles 112 and 272 days) after the procedure. Additionally, we performed the analysis of LR in 16 control subjects. Table 1 presents subjects’ clinical and echocardiographic data.

Table 1 Patients’ clinical and echocardiographic characteristics

The Internal Review Board of the Cleveland Clinic approved of, and all control subjects gave a written informed consent to, the study.

Database search and the definition of clinical terms

Patients were defined as having ICM if coronary angiography showed coronary artery stenosis of at least 50% in any of the major epicardial vessels, or if there was a documented history of prior myocardial infarction or coronary artery revascularisation. If ischaemic cardiomyopathy was present, location and presence of the myocardial scars were identified by, in the order of precedence: magnetic resonance imaging (n = 2), rubidium glucose positron emission tomography (n = 13), stress-rest single-photon computerised emission tomography (n = 10), or dobutamine stress echocardiography (n = 14); in seven subjects, scar data were unavailable. Patients with LV enlargement and normal coronary artery angiogram were defined as having DCM. All patients had LV ejection fraction of ⩽35% by echocardiography.

The QRS duration complex was determined from the last ECG obtained prior to biventricular device implantation. In patients in whom a biventricular pacing device was successfully implanted, the location of the LV electrode was determined by a chart review and/or by chest x-ray examination.

Echocardiography methods

We measured LV end-systolic and end-diastolic volumes by Simpson’s biplane echocardiography. The timing of aortic valve closure was determined from the pulsed-wave Doppler tracings of the LV outflow tract.

To obtain LR we analysed the apical four-chamber view using the speckle tracking software (EchoPac, GE Medical). The average frame rate for our data was 54 (SD 28) frames per second. The left ventricle was segmented into five segments: basal septal, mid-septal, apical, mid-lateral and basal lateral. For each of the segments, we obtained end-systolic values of segmental rotation, longitudinal strain and radial strain. In order to obtain these data, we integrated the strain rate (or rotation rate) starting from the mitral valve closure until the end-systole. End-systole was defined as the aortic valve closure time. We used the same software and the same view to obtain global rotation of the LV cross-section, which is LR. The absolute intra-observer and interobserver variability of global LR measurements was 0.9±0.6° and 1.8±1.6°, respectively, with r values for correlation of 0.97 and 0.85. We also used the same software to obtain global short-axis rotation at the apical level by analysing a parasternal cross-sectional view of the apex. In accordance with engineering notation, the negative sign means that there is a clockwise rotation, and vice versa.

We have previously validated this speckle tracking software for the measurement of global rotation against the “gold standard” of magnetic resonance imaging, and have demonstrated that it accurately tracks short-axis rotation.4 The software was also previously validated for the assessment of longitudinal and radial strains.5 6

Statistical analysis

To assess whether there were any differences in LR between subgroups of subjects, we performed a simple one-way analysis of variance followed by post hoc Tukey honest significant difference test. To assess possible predictors of LR, we performed forward stepwise multivariate regression with LR, end-diastolic volume, QRS duration and the presence of ischaemic heart disease as independent predictors.

To assess whether LR is associated with the abnormality of segmental contraction, we performed repeated-measures analysis of variance with subjects as random, and groups and segments as fixed factors.

To assess the impact of activation and LV size on LR, we selected DCM patients treated by CRT with the coronary sinus electrode implanted in the lateral or postero-lateral LV segments (ie initial targets of lead implantation), and in whom an echocardiogram was available before, immediately after, and at least 40 days after the beginning of CRT. We analysed data by one-way repeated measures analysis of variance followed by post hoc testing.

The predictive value of LR for the decrease of end-systolic volume during CRT was assessed by a simple linear regression.

Data are expressed as mean (standard deviation), unless otherwise noted. A p value<0.05 was considered significant.


We successfully obtained LR in all 102 subjects. However, short-axis rotation was obtained in only 70 out of 86 patients, due to poor image of the apex (n = 4) or unavailability of apical view (n = 12). Figure 1 represents end-diastolic, mid-systolic and end-systolic frames of a DCM patient with prominent clockwise LR, with rotation markers superimposed on the original two-dimensional echocardiography image. One can readily discern the clockwise marker motion during systole (supplemental video 1).

Figure 1 Clockwise longitudinal rotation occurring during systole in a dilated cardiomyopathy patient. Tracking markers of individual segments are superimposed over standard two-dimensional echocardiographic apical four-chamber views of the left ventricle in end-diastole (upper left), mid-systole (upper centre) and end-systole (upper right). Observe the changing angle of the mitral annular plane. The graph below represents segmental (thin coloured lines) and global (thick white line) longitudinal rotation throughout the cardiac cycle.

Predictors of longitudinal rotation

Figure 2A shows the average end-systolic volume in normal subjects, ICM subjects with QRS complex duration of <130 ms, ICM subjects with QRS⩾130 ms, DCM subjects with QRS <130 ms and DCM subjects with QRS⩾130 ms. End-systolic volume was larger in all four patient groups than in controls (p<0.001), without difference between patient groups. Somewhat similarly, short-axis apical rotation was smaller in all patient groups (p<0.01 compared with controls) (fig 2B); in addition, ICM subjects with QRS<130 ms had more prominent short-axis apical rotation than DCM subjects with QRS<130 ms, and significantly different from 0 (p = 0.02). In contrast to this, fig 2C shows that only patients with DCM had LR significantly different from 0 (p<0.001 vs 0 for these two groups). Furthermore, DCM patients with QRS>130 ms had clockwise LR larger than control subjects and both ICM subgroups (p<0.001 for all three comparisons), while DCM patients with QRS<130 ms had larger clockwise LR than control subjects and ICM patients with QRS<130 ms (p<0.05 for both; see also fig 2C). The most significant predictor of LR was end-diastolic volume (p<0.001), followed by the absence of ischaemia (p<0.001) and QRS duration (p = 0.05), with the multiple r value of 0.56 (p<0.001).

Figure 2 (A) Individual values, means and 95% confidence intervals for end-systolic volume in normal subjects, subjects with ischaemic cardiomyopathy (ICM) with QRS duration ⩽130 ms, subjects with ischaemic cardiomyopathy with QRS duration ⩽130 ms, subjects with dilated cardiomyopathy (DCM) with QRS duration ⩽130 ms, and subjects with dilated cardiomyopathy with QRS duration ⩽130 ms. (B) Apical short-axis rotation in the same five patient groups. (C) Longitudinal rotation in the same five patient groups.

Since these data show that ICM and DCM patients have marked differences in LR, we constructed the average LR curve using data from DCM patients only. The peak of LR curve coincided with end-systole (supplemental fig 1).

Longitudinal rotation and segmental contraction patterns

To analyse whether LR is associated with a specific pattern of segmental contraction, we selected eight subjects with the highest clockwise LR (all with DCM), eight DCM subjects with lowest or counterclockwise LR, and eight ICM subjects with largest counterclockwise LR. Figure 3A shows segmental LR in these three groups. As expected, there was a very significant difference between the three groups (p<0.001). Furthermore, in patients with prominent LR, the rotation was very prominent in the apical segments. Figure 3B shows a significant difference in longitudinal strains distribution, with patients with prominent clockwise LR showing low strains in the septum but high strains in the lateral walls (p = 0.002). Figure 3C shows similar differences in radial strains between the three groups (p = 0.02). These data indicate that patients with prominent clockwise LR have dysfunctional septal and prominent lateral contraction.

Figure 3 Segmental longitudinal rotation (A), longitudinal strain (B) and radial strain (C) in dilated cardiomyopathy subjects with prominent clockwise global longitudinal rotation (DCM/LR+; n = 8), dilated cardiomyopathy subjects with minimal global longitudinal rotation (DCM/LR−; n = 8), and ischaemic cardiomyopathy subjects with counterclockwise global longitudinal rotation (ICM/LR−; n = 8). Error bars represent standard errors.

Presence of lateral scar and longitudinal rotation

To pursue this observation further, we dichotomised our ICM subjects as having scar in the lateral or posterior walls at the base or mid-ventricular level (n = 16), or not having the scar in these regions (n = 23). Despite having similar end-diastolic volumes (221 (SD 70) ml vs 234 (SD 75) ml, p = 0.57), patients with lateral scar had rotation in the opposite direction (0.7 (SD 3.0) vs –1.60 (SD 2.7), p = 0.02), that is, counterclockwise rotation with decreased strain in the lateral wall and marked strains in the septum. Interestingly, in ICM patients with lateral scar no decrease of end-systolic volume was observed during CRT (−0.12 (SD 28.0)%, p  =  NS vs 0) while in patients without lateral scar end-systolic volume significantly decreased by 20.5 (SD 26.2)% (p = 0.004 vs 0).

Early versus late effects of cardiac resynchronisation therapy on longitudinal rotation

We analysed the effects of CRT in 14 DCM patients in whom the LV electrode was implanted in lateral or posterolateral segments, and who had adequate echocardiography studies performed pre-implantation, a day after implantation, and after >3 months. Pacing did not significantly alter QRS duration (157 (SD 27) vs 164 (SD 20) ms, p = NS). There were no changes of end-systolic volumes immediately after start of biventricular pacing (from 220 (SD 76) to 220 (SD 79) ml, p  =  NS), while after >3 months LV volumes decreased to 159 (SD 86) (p<0.01 vs both pre and early post). There was also no difference in LR early after the start of CRT (−5.3 (SD 3.3) vs –5.7 (SD 3.1)°, p  =  NS), while LR significantly increased at the late follow-up to −3.5 (SD 3.5)° (p<0.05 vs both pre and early post). These data show that LR is more dependent on end-systolic volume than on the pattern of the electrical activation.

Prognostic value of longitudinal rotation in cardiac resynchronisation therapy

Figure 4 shows the prognostic value of LR in 64 CRT patients who had a late echocardiographic follow-up. Since we have already shown that ICM and DCM subjects have marked differences in the LR, we analysed them separately. There was a moderately strong association between initial LR and end-systolic volume decrease (r = −0.49, p = 0.004) in the DCM patients, while no such relationship was observed in ICM patients.

Figure 4 Relationship between end-systolic volume change during cardiac resynchronisation therapy in dilated cardiomyopathy subjects (filled circles) as opposed to ischaemic cardiomyopathy subjects (empty triangles). While a significant inverse predictive relationship was observed for longitudinal rotation to predict reverse remodelling with CRT in dilated cardiomyopathy, no relationship was observed in ischaemic cardiomyopathy.

Numerical simulations

To calculate energy expenditure relative to total work delivered by the heart during a single cardiac cycle, we assumed that a heart is a hollow thick-walled sphere with internal radius of 4 cm, wall thickness of 1 cm, wall mass of 221 g, duration of systole and diastole of 400 and 600 ms, and global rotation of 11 degrees. We also assumed that the sphere does not translate and that it rotates around its centre. Assuming a conservative system,7 for rotational bodies the energy delivered by rotation is:

E = ½I*ω2

where ω is angular velocity and I the moment of inertia of the sphere:

Embedded Image

where M represents LV mass, and R2 and R1 represent inner and outer radius of the sphere. Under these assumptions, total energy used up by the rotation is 1.5×10−5 J, which is less than 0.001% of the total work of the heart.


In this paper we demonstrate that clockwise LR occurs when LV enlargement is caused by DCM, with a small contribution from QRS duration. In contrast, patients with ICM show a wider array of rotational behaviour depending on the location of myocardial scar; ICM patients with lateral scar frequently displayed counterclockwise rotation. These findings, along with demonstrated differences in regional longitudinal strains in patients with and without LR, provide the mechanistic explanation for this phenomenon. To our knowledge, although this kind of swinging motion was frequently recognised during routine echocardiography, this is its first systematic description.

The paradox of longitudinal rotation

Systolic dysfunction results in less movement of ventricular segments: velocities decrease,8 along with wall thickening, strains, and short-axis rotation.911 It seems unusual that in this setting a previously unreported motion of the LV, the magnitude of which is proportional to the severity of LV dilatation, emerges. However, our simple LR energy consumption model demonstrates that the force needed to induce LR is small, and therefore the appearance of very small inequalities in the force production, if in the right direction, will lead to LR.

The origin of longitudinal rotation

The origin of longitudinal rotation and its association with end-diastolic volume are unclear. While DCM does not change myocardial fibre orientation,9 it changes the direction of principal strains in the epicardium.9 12 In a normal heart epicardial principal strains follow myocardial fibre orientation, but with the development of heart failure they become parallel to the longitudinal axis.9 13 This changed angle of principal strains may allow subepicardial muscle fibres to pull the apex towards the baso-lateral wall. QRS prolongation due to left bundle branch block may further highlight this by making the lateral wall activate last, thus pre-stretching it and improving its contraction through the Frank–Starling mechanism.14 Finally, higher end-diastolic volumes may additionally lead to higher regional end-diastolic stress (preload) in the basal lateral wall, leading to stronger contraction.

Findings that support this conjecture are the following: the association of LR with high longitudinal strains in the lateral wall of the heart; LR decrease in the presence of inferolateral or lateral scar; and its association with the LV size. It is also partly supported by its weak correlation with QRS complex duration, as well as by its decrease with the decrease of LV dilatation.

Clinical implications

There are several practical implications of our findings. Prominent clockwise longitudinal rotation is seen almost exclusively in DCM subjects, with its magnitude proportional to LV end-diastolic volume. The degree of clockwise longitudinal rotation in DCM is linked to decreased septal, but increased lateral, deformation. Finally, LR may predict success of CRT in DCM subjects, even in the presence of QRS <130 ms. Of note, as LR represents incoordinate LV contraction, its occurrence even in the absence of dramatic QRS prolongation provides support for the thesis that some subjects with normal QRS duration have contraction dyssynchrony. In contrast, counterclockwise LR or absence of clockwise LR in ICM is associated with decreased lateral deformation and with the presence of scar in the lateral wall. These two factors may predict failure of CRT.15


This was a retrospective study, with wide variety of follow-up duration. Of note, some of the patients had short follow-up because of rapid deterioration of their clinical status, which led to heart transplantation. Additionally, we have not used a three-dimensional imaging method to evaluate LV strains, and thus it is possible that our observations represent short-axis rotation due to foreshortened cross-section of the LV. We have shown, however, that our DCM patients with the largest amount of LR have the least amount of apical short-axis rotation, making this possibility highly improbable. We chose the value of 130 ms, instead of 120 ms, a priori as a cut-off value for classification of patients in a narrow or wide QRS complex category in order to obtain a balanced number of subjects in each of the groups. However, previous reports showed that prolongation of QRS duration in DCM is not a distinct, all-or-none event but a continuous process.16 17

In conclusion, we have shown that clockwise longitudinal rotation appears in the setting of non-ischaemic DCM, with the additional impact of QRS duration. In contrast, patients with ICM show a wider array of rotational behaviour depending on the location of myocardial scar, with some of them even having counterclockwise rotation.


This manuscript was supported through grants by National Institutes of Health grant AG17479-02, by the National Space Biomedical Research Institute through NASA NCC 9-58 (Houston, TX), the Department of Defense (Ft. Dietrick, Md, USAMRMC grant no. 02360007), by the National Institutes of Health, National Center for Research Resources, General Clinical Research Center (grant MO1 RR-018390), and by a National American Heart Association grant 0235172N.


View Abstract

Supplementary materials

  • web only appendices 94/3/e11

    Files in this Data Supplement:


  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.