OUP user menu

Assessment of left ventricular twist mechanics in Tako-tsubo cardiomyopathy by two-dimensional speckle-tracking echocardiography

Patrick Meimoun, Pricila Passos, Tahar Benali, Jacques Boulanger, Frederic Elmkies, Hamdane Zemir, Jerome Clerc, Anne Luycx-Bore
DOI: http://dx.doi.org/10.1093/ejechocard/jer183 931-939 First published online: 28 September 2011


Aims To assess left ventricular (LV) twist mechanics in patients with Tako-tsubo cardiomyopathy (TTC).

Methods and results Two-dimensional strain and LV twist by speckle-tracking echocardiography was performed in 17 consecutive patients with typical TTC according to the Mayo clinic criteria [78 ± 8 years, 88% women, and mean left ventricular ejection fraction (LVEF) 45 ± 10%], at the acute phase (within 24 h after admission) and after recovery (1 month later). Seventeen control (C) patients matched for age and sex (mean LVEF 70 ± 7%), and 17 patients with acute anterior myocardial infarction (MI) treated by successful primary angioplasty 24 h before, matched for LVEF, age, and sex, were compared with TTC patients. LV twist was assessed using the parasternal basal and apical short-axis planes, and defined as the net difference in degrees of apical (Ar) and basal rotation (Br). Peak systolic and early diastolic, apical (As and Ad) and basal (Bs and Bd) rotation rate, and LV twisting rate (TR) and untwisting rate (UR) (in °/s) were derived from rotational and twist curves. The time sequences were normalized to the percentage of systolic duration. At the acute phase, Ar, As, Ad, Bs, LV twist (10 ± 9° vs. 23 ± 6°), LV TR, and LV UR were significantly impaired in patients with TTC when compared with controls (all, P< 0.05). Patients with MI displayed intermediate values (P= NS vs. TTC, and P< 0.05 vs. C). However, in the subgroup of patients with electrocardiogram ST-segment elevation at presentation (n= 12 TTC and 17 MI), several LV twist mechanics parameters were significantly reduced in TTC patients when compared with MI patients, LV twist, and LV TR being the most significant factors (all, P≤ 0.01). Abnormal reversed Ar (clockwise when seen from the apex) was seen in three patients (18%) with TTC vs. none in the other groups. A significant correlation between LV twist and LVEF, LV volumes, wall motion score index, and plasma NT-pro BNP was observed in the TTC group (all, P< 0.05). At follow-up, LV twist mechanics improved significantly in TTC patients (all, P< 0.05 vs. acute phase), who had final values similar to C (all, P= NS), whereas the magnitude of improvement was lower in MI patients (P ≤ 0.05 vs. TTC).

Conclusion LV twist mechanics is significantly impaired in patients with TTC mainly due to a severe reduction in apical function and is entirely reversible. Furthermore, in the subgroup of patients with ST-segment elevation, the early post-admission evaluation of LV twist mechanics is more severely impaired in TTC when compared with MI.

  • Tako-tsubo
  • Twist
  • Untwisting
  • Rotation
  • Cardiomyopathy
  • Speckle tracking

Tako-tsubo cardiomyopathy (TTC) has recently been described as a transient cardiomyopathy affecting typically post-menopausal women and triggered by emotional or physical stress.14 In the typical form, transient mid-apical left ventricular (LV) wall motion abnormalities (WMA) are present in TTC.14 Two-dimensional strain by speckle-tracking echocardiography (STE) is a new tool that enables the evaluation of global and regional LV deformation at the longitudinal, radial, and circumferential direction.5,6 Two-dimensional strain is very attractive particularly when one would like to assess the apical LV function where it has overcome the angle dependence of Doppler tissue imaging, allowing for measurement of cardiac rotation. Apical LV rotation is an important contributor of LV twist which has a critical role in maintaining not only LV systolic function, but also LV diastolic function by the subsequent LV untwisting.613 TTC mimics acute anterior myocardial infarction (MI) in its presentation.14 However, LV twist has not been assessed in TTC where transient apical WMA are the main feature. In contrast, it has been demonstrated that LV twist is impaired in MI11,14 and is ascribed as a prognostic parameter.14 Furthermore, there is a paucity of data about diastolic function in TTC, and the evaluation of LV untwisting could be useful in this setting. In addition, subclinical LV dysfunction has been suggested after recovery of LVEF and WMA in TTC by STE,15 which is more sensitive to detect LV dysfunction than traditional echo-Doppler parameters.16 It could be interesting to clarify this finding for LV twist mechanics. Therefore, our objective was to describe changes in LV twist mechanics in systole and diastole in patients with typical TTC and to compare it in patients with anterior MI, using STE at the acute phase and at follow-up.


Between the 1 June 2009 and the 1 June 2011, 17 consecutive patients affected by a typical TTC underwent prospectively a comprehensive transthoracic Doppler echocardiography including STE at the acute phase (within 24 h after admission at our institution) and after recovery. During the same period, 17 patients with acute anterior MI treated successfully by primary percutaneous angioplasty 24 h before, matched for left ventricular ejection fraction (LVEF), age, and sex, and 17 controls, matched for age and sex, served as comparative groups. Each patient gave inform consent to participate to the protocol.

Patients groups

TTC was diagnosed according to the following criteria: acute chest pain or dyspnoea, a stressful event when present, transient LV WMA at the apex of the LV, electrocardiogram (ECG) changes (ST-segment elevation or T-wave inversion), no luminal narrowing >50% and no evidence of plaque rupture in all epicardial coronary arteries on angiography, and a minimal troponine release despite extensive WMA.14 Furthermore, no patient had a history of coronary artery disease, a febrile or acute neurologic disorder. Selective coronary angiography was performed using standard techniques, at the acute phase in all patients (at admission in most cases), and demonstrated no significant coronary artery disease in all. Left ventriculography demonstrated the typical pattern like an octopus pot14 in all. The follow-up evaluation for STE analysis was uniformly performed 1 month later in an outpatient visit for all but one patient (at 2 weeks for this case). The patients with a first acute anterior ST-elevation MI were successfully treated with primary coronary angioplasty within 12 h of symptom onset (with a final angiographic TIMI-flow grade 3 and a residual stenosis <30% of the culprit vessel: the left anterior descending artery). The diagnosis of MI was based on chest pain lasting >30 min, ST-segment elevation >2 mm in at least two contiguous precordial ECG leads, and increase in serum troponin T (normal <0.05 µg/L in our hospital) and/or creatine kinase to more than three-fold the normal value. The early evaluation of STE was performed within 24 h after angioplasty. The follow-up evaluation of STE was performed at 3 months in this group. The control group consisted of 17 age-, sex-matched patients with the TTC group who had a normal rest echocardiography. These patients had no clinical or history of structural cardiovascular disease. Patients with shock, more than mild valvular disease, or with poor echocardiographic window were criteria exclusion (n= 3).

Two-dimensional Doppler echocardiography

A comprehensive transthoracic Doppler echocardiography was performed using commercially available machine (Vivid E9 system, General Electrics) with a M5S probe. Left ventricular end-diastolic (EDV) and end-systolic volume (ESV) were measured from the apical four- and two-chamber view and the LVEF calculated from the modified biplane Simpson's rule. Wall motion score index (WMSI) was measured using the 16-segment 4-point scaling model, from the apical four-, two-, and three-chamber views.4,17 Briefly, segment scores were as follows: 1 = normal or hyperkinesis, 2 = hypokinesis, 3 = akinesis, 4 = dyskinesis; the WMSI was derived as a sum of all scores divided by the number of segments visualized. Left ventricular mass index was measured according to the ASE formula.17 Conventional Doppler parameters were measured according to a standardized examination: Early (E), and late (A), diastolic transmitral flow velocity, and deceleration time of E (DTE), average of the septal and lateral annulus mitral early diastolic (Ea), late diastolic (Aa), and systolic (Sa) tissue velocity, and the ratio E/Ea. The pulmonary artery systolic pressure was calculated using the modified Bernoulli equation from tricuspid regurgitant peak jet velocity, and estimated right atrial pressure (from respiratory variation of inferior vena cava diameter). All echocardiograms were digitized online on optical disks for subsequent offline analysis by one other experienced observer blinded to patient data.

Two-dimensional speckle-tracking echocardiography and left ventricular twist assessment

To quantify basal (Br) and apical rotation (Ar), parasternal basal and apical short-axis planes were scanned and recorded using a M5S probe, at high frame rates (60–80 frames/s), during end-expiratory breath hold. The basal plane was defined as showing the tips of the mitral leaflets, whereas the apical plane was defined as just proximal to the level with end-systolic LV luminal obliteration.614 For obtaining this latter plane, the transducer was placed one or two intercostals spaces more caudally, as previously described.18 In each short-axis acquisition the LV cross-section was made as circular as possible. At each plane three consecutive cardiac cycles were acquired and digitally saved in a cine-loop format for offline analysis with the support of a dedicated software package (Echopac 7 version 108 for PC, GE Medical System). LV endocardial borders were manually traced at end-systolic phase for STE analysis and the region of interest was chosen to fit the whole myocardium. The reliability of the tracking was confirmed by the reliability parameter offered by the system and was again visually checked and readjusted if necessary. LV rotations at the basal or apical short-axis planes were determined as average angular displacement of six myocardial segments along the central axis. Counter-clockwise rotation as viewed from the LV apex was expressed as a positive value, whereas clockwise rotation was expressed as a negative value. LV twist was defined as the net difference in degrees between the Ar and Br (peak-to-peak). LV twist curve during the cardiac cycle was automatically built by the software from the Br and Ar curves (Figure 1A). As Br and Ar rotation were measured sequentially, care was taken to record the respective grey scale images at similar heart rates. LV torsion was defined as LV twist normalized for LV diastolic longitudinal length, and expressed in degree/cm.14 Peak systolic and early diastolic, apical (As, Ad) and basal (Bs, Bd) rotation rate, and LV twisting rate (TR) and untwisting rate (UR) (in degree/s) were derived from rotational and twist curves by the software and were also measured. Untwisting starts after peak LV twist and peaks around mitral valve opening. The first negative peak in early diastole was used for measuring peak LV UR. It is followed by two negative peaks corresponding to early and late filling phases of diastole (Figure 1B).1921 The systolic duration was defined as from the onset of QRS of the ECG to the aortic valve closure (AVC), which was determined automatically by the software and from appropriate pulsed-wave Doppler images. To account for variable heart rates between subjects, the time sequences were normalized to the percentage of systolic duration (onset of QRS, t= 0% and at AVC, t= 100%). With this method we measured time to Ar and Br, time to Ad and Bd. From the apical long-axis, four- and two-chamber views, global LV longitudinal strain (GLS) by STE was quantified as previously described.5,14,15 Briefly, the LV was divided in six segments in each apical view and the tracking quality was validated for each segment. Then, the myocardial motion was analysed by speckle-tracking within the region of interest. The automated algorithm used a 17-segment model and provided peak systolic longitudinal strain for each LV segment.

Figure 1

Left ventricular twist and twisting rate curves at the acute phase and at recovery in a patient with Tako-tsubo cardiomyopathy. (A) Left ventricular twist (white curve in each graph), the net difference between apical rotation (blue curve, positive), and basal rotation (purple curve, negative), is reduced at the acute phase. The exact time of peak left ventricular twist is uncertain because of the flat curve (severe depression of contractility), but it is around aortic valve closure and does not exceed 6°. A significant improvement of left ventricular twist is seen at the recovery phase with a final value at 23°. (B) Derived from the twist curve, peak left ventricular twisting rate (in systole) and left ventricular untwisting rate (in early diastole) are reduced at the acute phase. A significant improvement of these parameters is seen at the recovery phase. In addition to peak early diastolic left ventricular untwisting rate, there are two negative peaks corresponding to early and late filling phases of diastole (White curve in each graph). Owing to the complex myocardial architecture and sequential activation of its components, the twisting motion has two phases which translates in the apical rotation rate to two peaks particularly visible in this case at recovery; an early peak which peaks during early ejection period, and a later positive ejection period peak which is calculated for measuring As.


Results were expressed as mean ± SD and percentages. Categorical variables were compared using the χ² test or Fisher exact test as appropriate, and continuous variables were compared using one-way analysis of variance (ANOVA) or unpaired or paired student's t-test as appropriate. The relationships between LV twist mechanics parameters and LVEF, LV volumes, WMSI, and NT-proBNP (log), were evaluated using linear and non-linear correlations, and those exhibiting the best fit were retained. In addition, when possible, the influence of the underlying disease (TTC vs. MI) on the relationship between LV twist mechanics and LV function parameters cited above was tested by ANCOVA. Statistical analysis was performed using MedCalc for Windows, version (MedCal Software, Mariakerke, Belgium). A P-value < 0.05 was considered as significant. Intra-observer and inter-observer variability were evaluated from 10 randomly selected cases and calculated as the absolute difference divided by the average of the two observations, times 100, for each twist mechanic parameter.


The characteristics of the study population are summarized in Table 1. Patients with MI had a higher body mass index and troponin peak when compared with patients with TTC (all, P< 0.05). In addition to a better LVEF and WMSI, the control group had a lower heart rate, higher Sa and Ea, and a lower PASP, when compared with the other groups, (all, P< 0.05). Patients with TTC had a lower Ea when compared with the other groups (P< 0.05). In the TTC group, a stressful event was found in 15 cases, and the ECG at admission demonstrated ST-segment elevation and T-wave inversion in 12 and 5 cases, respectively. Reliable STE curves were obtained in all patients for Ar, whereas Br analysis was not possible in two cases with TTC (who had ECG T-wave inversion at presentation). Therefore, LV twist, LV TR, and LV UR were available in 15 cases with TTC (and in all cases in the other groups).

View this table:
Table 1

Baseline characteristics

Controls (n= 17)Tako-tsubo (n= 17)Myocardial infarction (n= 17)
Age, years73 ± 878 ± 873 ± 12
Female, n (%)15 (88)15 (88)15 (88)
Hypertension, n (%)011 (64)8 (47)
Diabetes, n (%)03 (17)6 (35)***
Smoking, (%)005 (29)*
Dyslipidemia, n (%)05 (29)7 (41)
BMI, kg/m²25.5 ± 424 ± 430 ± 5.4*
LVEF, %70 ± 7*45 ± 1045 ± 8
WMSI1*1.87 ± 0.341.83 ± 0.16
LV mass, g/m²72 ± 1278 ± 1685 ± 21
Mitral E-wave, cms/s68 ± 1460 ± 2071 ± 26
Mitral A-wave, cms/s81 ± 2074 ± 2477 ± 24
DT E, ms210 ± 47207 ± 95177 ± 43
Sa, DTI, cm/s8 ± 0.5*6.3 ± 1.26.4 ± 1.3
Ea, DTI, cm/s8 ± 1.2*5 ± 1.86.1 ± 1.4**
Aa, DTI, cm/s9.5 ± 1.58.6 ± 2.28.6 ± 2
Average E/Ea9.5 ± 2.413.2 ± 6***12.4 ± 7
Septal E/Ea10.2 ± 2.815 ± 8.6***12.7 ± 8.4
Lateral E/Ea9 ± 2.611.8 ± 5.711.6 ± 6.9
Troponin peak, µg/L0.65 ± 0.55.8 ± 3.9**
NT-proBNP, pg/mL (log)3.7 ± 0.53.5 ± 0.5
PASP, mmHg29 ± 5*35 ± 934 ± 7
Systolic/diastolic blood pressure, mmHg130 ± 12/   70 ± 10115 ± 21/  65 ± 10118 ± 22/  70 ± 16
Heart rate, bpm63 ± 8*76 ± 1477 ± 12
  • BMI, body mass index; LVEF, left ventricular ejection fraction; WMSI, wall motion score index; DTE, deceleration time of mitral E-wave; Sa, Ea, Aa, DTI, average of the septal and lateral, systolic, early and late diastolic, mitral annular tissue velocities, using Doppler tissue imaging; PASP, pulmonary artery systolic pressure.

  • *P≤ 0.05 vs. other groups.

  • **P< 0.05 vs. Tako-tsubo group.

  • ***P< 0.05 vs. control group.

Acute phase

When compared with controls, Ar, As, and Ad, Bs, LV twist, LV torsion, LV TR, and LV UR were significantly impaired in patients with TTC at the acute phase (all, P< 0.05) (Table 2). Patients with MI exhibited intermediate values, not significantly different to patients with TTC (all, P= NS). Of note, three patients in the TTC group had values of LV twist in the normal range at the acute phase because they had partially recovered at the time of echocardiography with less extend WMA than the other patients. Abnormal reversed Ar was seen in three cases with TTC (18%), whereas no patient in the other groups exhibited this abnormal clockwise rotation of the apex during systole (example in Figure 2). In this subgroup of patients with abnormal reversed Ar, the median values of WMSI, EDV, and ESV were higher when compared with the remainder patients with TTC (2.13 vs. 1.91 mL, 107 vs. 76 mL, and 64 vs. 47 mL, respectively). In the subgroup of patients who had ECG ST-segment elevation at presentation (n= 29, 12 TTC and 17 AMI patients), several LV mechanics parameters differed significantly between TTC and MI patients (Table 3). LV twist and LV TR were the best parameters to discriminate TTC and MI patients in this subgroup (P-value < 0.01 for both).

View this table:
Table 2

Left ventricular twist mechanics and left ventricular global longitudinal strain at the acute phase

ControlsTako-tsuboMyocardial infarction
Apical rotation, °15 ± 4**5 ± 57 ± 4
Time to apical rotation, % systole99 ± 5103 ± 1295 ± 12
Peak systolic apical rotation rate, °/s79 ± 27**32 ± 3547 ± 22
Early diastolic apical rotation rate (Ad), °/s−76 ± 35**−30 ± 42−43 ± 24
Time to Ad, % systole119 ± 11130 ± 17122 ± 21
Basal rotation, °−8 ± 4−6 ± 5−6 ± 3.5
Time to basal rotation, % systole96 ± 594 ± 1392 ± 12
Peak systolic basal rotation rate, °/s−78 ± 38*−48 ± 25−53 ± 20
Early diastolic basal rotation rate (Bd), °/s59 ± 2548 ± 2851 ± 24
Time to Bd, % systole116 ± 7121 ± 18113 ± 10
LV twist, °23 ± 6**10 ± 912 ± 5
LV torsion, °/cm3.2 ± 1**1.3 ± 1.21.4 ± 0.6
LV twisting rate, °/s124 ± 37**64 ± 4476 ± 20
LV untwisting rate, °/s−114 ± 43**−59 ± 35−71 ± 34
LV global longitudinal strain, %−21.3 ± 2**−8.2 ± 5−9.4 ± 2
  • In the TTC group, apical rotation was assessed in 17 patients; basal rotation and LV twist in 15 patients, whereas in the other groups data for LV twist were obtained in 17 patients.

  • *P< 0.05 vs. other groups.

  • **P≤ 0.01 vs. other groups.

View this table:
Table 3

In patients with Tako-tsubo cardiomyopathy and myocardial infarction with electrocardiogram ST-segment elevation at presentation, main left ventricular twist mechanics, left ventricular ejection fraction, and wall motion score index

Tako-tsubo, (n= 12)Myocardial infarction, (n= 17)
Apical rotation, °3 ± 47 ± 4*
Basal rotation, °−4 ± 4−6 ± 3.5
Peak systolic apical rotation rate, °/s23 ± 3547 ± 22**
Time to apical rotation, % systole106 ± 1395 ± 12**
LV twist, °6 ± 512 ± 5*
LV torsion, °/cm0.8 ± 0.61.4 ± 0.6*
LV TR, °/s46 ± 2576 ± 20*
LV UR, °/s−50 ± 35−71 ± 34
WMSI2.01 ± 0.251.83 ± 0.16**
LVEF, %41 ± 845 ± 8
  • LV, left ventricle; TR, twisting rate; UR, untwisting rate; WMS, wall motion score.

  • *P≤ 0.01.

  • **P< 0.05.

Figure 2

Transient abnormal pattern of apical rotation in a patient with Tako-tsubo cardiomyopathy. (A) At the acute phase, there is loss of left ventricular twist (white curve, difference between apical and basal rotation), due to a severe impairment of apical rotation (blue negative curve, abnormal clockwise rotation). (B) One month later, there is a total recovery of left ventricular twist due to recovery of apical rotation (blue positive curve, normal counterclockwise rotation). At each stage of the disease, there is a normal clockwise basal rotation (purple negative curve) AVC, aortic valve closure.

Time course of twist mechanics

At follow-up, nearly all patients were on beta-blockers in each subgroup (n= 14 and n= 17 in the TTC and the MI group, respectively). In addition to a total recovery of LV WMA, and a non-significant improvement for most of traditional diastolic parameters, a significant improvement of Ar, As, and Ad, Bs, LV twist, LV torsion, LV TR, LV UR, GLS, and LVEF was seen at follow-up when compared with the acute phase, in patients with TTC (all, P< 0.05) (Table 4), with final values not significantly different from the control group (all, P= NS). Concerning the subgroup of patients with abnormal reversed Ar, a total recovery of Ar with a normal pattern (counterclockwise rotation) was observed at follow-up (See Figure 2). In patients with MI, there was a significant improvement of Ar, Ad, LV twist, LV torsion, GLS, and LVEF at follow-up, when compared with the acute phase (all, P< 0.05), but the magnitude of improvement was lower when compared with the TTC group (Table 4). For each individual patient with TTC and MI, the final value for each LV twist mechanics parameter was compared with the lower limit of normal value as defined from the control group. Whatever the definition of the lower threshold of normal value, 5th or 95th percentile (Table 5), absolute lower value, above—2SD of the mean, all patients with TTC were above this limit for most parameters. Only one patient had final values of LV TR, LV UR, and Br below the 5th–95th percentile at follow-up. In comparison, according to the parameter tested, 18–41% of patients with MI were below the normal limit at follow-up.

View this table:
Table 4

Time course of left ventricular mechanics and left ventricular function, in patients with Tako-tsubo cardiomyopathy and anterior myocardial infarction

Acute phaseFollow-upAcute phaseFollow-up
Apical rotation, °5 ± 517 ± 8*7 ± 411 ± 6*,***
Time to apical rotation, % systole103 ± 1199 ± 595 ± 12***94 ± 6***
Peak systolic apical rotation rate, °/s32 ± 3581 ± 39*47 ± 2254 ± 24***
Early diastolic apical rotation rate (Ad), °/s−30 ± 42−77 ± 38*−43 ± 24−59 ± 32**
Time to Ad, % systole130 ± 17123 ± 10122 ± 21122 ± 13
Basal rotation, °−6 ± 5−9 ± 4−6 ± 3.5−7 ± 2.5
Time to basal rotation, % systole94 ± 1394 ± 1192 ± 1292 ± 10
Peak systolic basal rotation rate, °/s−46 ± 25−75 ± 20**−53 ± 20−53 ± 17***
Early diastolic basal rotation rate (Bd), °/s48 ± 2866 ± 3551 ± 2444 ± 20***
Time to Bd, % systole121 ± 18114 ± 19113 ± 10118 ± 11
LV twist, °10 ± 925.5 ± 10*12 ± 517 ± 6**,***
LV torsion, °/cm1.3 ± 1.23.3 ± 1.2*1.4 ± 0.61.9 ± 0.6**,***
LV twisting rate, °/s64 ± 44134 ± 36*76 ± 2080 ± 33***
LV untwisting rate, °/s−59 ± 35−128 ± 50*−71 ± 34−87 ± 40***
LV global longitudinal strain, %−8.2 ± 5−21 ± 2.4*−9.4 ± 2−15.2 ± 5*,***
LVEF, %45 ± 1072 ± 7*45 ± 851 ± 12*,***
LV EDV, mL84 ± 2176 ± 1196 ± 19119 ± 30**,***
(LVEDV/m², mL/m²)(50 ± 10)(45 ± 7)(51 ± 8)(63 ± 13)**,***
LV ESV, mL46 ± 1722 ± 7*54 ± 1959 ± 28***
(LVESV/m², mL/m²)(27 ± 9)(13 ± 4)*(29 ± 8)(31 ± 14)***
E/Ea13 ± 611 ± 412 ± 712.3 ± 7
WMSI1.87 ± 0.341*1.82 ± 0.161.52 ± 0.36*,***
Heart rate, bpm76 ± 1462 ± 10*77 ± 1260 ± 10*
  • Except for basal mechanics and LV twist, twisting, and untwisting, which were obtained in 15 patients in the Tako-tsubo group, all the other data were obtained in 17 patients for each subgroup.

  • *P≤ 0.01 vs. acute phase.

  • **P≤ 0.05 vs. acute phase.

  • ***P≤ 0.05 vs. Tako-tsubo.

View this table:
Table 5

Numbers of patients with Tako-tsubo cardiomyopathy and myocardial infarction at follow-up below the normal values of left ventricular twist mechanics as defined in the control group

5th or 95th percentile, control groupTTC, n (%)MI, n (%)
Br−4.5°1 (7)4 (24)
Bs−34°/s03 (18)
Bd26°/s1 (7)3 (18)
Ar07 (41)*
As45°/s06 (35)**
Ad−37°/s06 (35)**
LV twist13.5°06 (35)**
LV torsion1.8°/cm06 (35)**
LV TR87°/s1 (7)7 (41)
LV UR−63°/s1 (7)5 (29)
  • Br, basal rotation; Bs, basal systolic rotation rate; Bd, basal diastolic rotation rate; Ar, apical rotation; As, apical systolic rotation rate: Ad, apical diastolic rotation rate; LV TR, left ventricular twisting rate; LV UR, left ventricular untwisting rate. 5th percentile for positive values, 95th percentile for negative values.

  • *P≤ 0.01 vs. TTC.

  • **P< 0.05 vs. TTC.


In patients with TTC, at the acute phase, a significant correlation was found between LV twist and WMSI, LVEF, EDV, ESV, and LV UR (all, P < 0.05), between Ad and LV UR (P< 0.01). Also, a significant negative correlation was found between Ad and E/Ea at the septal site (r= −0.51, P= 0.04) and between NT-pro BNP (available in 15 cases) and Ar, LV TR, and LV twist (all, P< 0.05) (Table 6). Finally, the change of LV twist in patients with TTC was significantly correlated to the change of LVEF (r= 0.78) and WMSI (r= −0.66) (all, P z< 0.01). In patients with MI, the correlations between LV mechanics and the above parameters were also significant for most of them, except for NT-pro BNP and EDV (Table 5). Of note, a significant correlation was found between LV UR and Ea in the whole population (r= −0.41, P< 0.01). By ANCOVA, a significant influence of the underlying cardiac disease (TTC vs. MI, slopes comparisons) was found about the relationship between Ar and LV volumes, and concerning LV twist and ESV (all, P< 0.05), whereas this interaction was not significant about the relationship between LV twist and LVEF (and WMSI), and between Ar and LVEF (and WMSI) (all, P= NS). At follow-up, no interaction was found about these relationships. The intra-observer and inter-observer variability of the twist mechanics parameters were between 8–12% and 8–13%, respectively when considering all parameters; 8 ± 5 and 10 ± 7% for Ar, 10 ± 9 and 11 ± 10% for LV twist, and 11 ± 9 and 13 ± 10% for LV UR, respectively. Furthermore, the agreement between the two observers with respect to the direction of Ar and Br (clockwise or counterclockwise) was 100%.

View this table:
Table 6

Correlations at the acute phase in the Tako-tsubo (n= 17) and myocardial infarction (n= 17) group

LV twist
 WMSI−0.640.01−0.65< 0.01
 LV UR−0.67<0.01−0.520.04
NT-proBNP (log)
 Apical rotation−0.50.05−0.10.7
 LV twist−0.570.02−0.30.2
 LV TR−0.540.040.20.4
  • EDV, left ventricular end-diastolic volume; ESV, end-systolic volume; LV, left ventricular; UR, untwisting rate; TR, twisting rate.

  • *For LV twist, n= 15 in the Tako-tsubo group, for Log NT-proBNP, n= 16 in the myocardial infarction group, and n= 15 in the Tako-tsubo group.


This study shows that LV twist mechanics involving the systolic and diastolic components are transiently impaired in TTC, mainly due to a loss of apical function. This impairment is entirely reversible contrary to what is seen in MI; and a significant correlation was found between LV twist mechanics and, NT-proBNP, and regional and global LV systolic function, and its change. Furthermore, LV twist mechanics tended to be more severely impaired in the TTC vs. MI group at the acute phase, the differences being significant in the subgroup of patients with ECG ST-segment elevation at presentation.

Apical function

LV twist has been characterized in several conditions and heart diseases 614,1828 but to our knowledge has never been evaluated in TTC where regional myocardial dysfunction is a main feature and may affect LV twist via an impairment of Ar. The LV myocardial architecture is complex with myofibres geometry changes from an oblique right-hand helix orientation in the subendocardium to a circumferential orientation in the mid-wall and a left-hand helix orientation in the subepicardium.22,23 LV twist originates from the dynamic interaction between these oppositely wound helices and its direction is governed by the epicardial fibres mainly because of their longer arm of movement.14,2428 Furthermore, LV twist and Ar are influenced by LV shape.26 There is a modification of LV apical shape at the acute phase of TTC (with a morphology like an octopus pot), and, in addition to a depressed contractility, might explain the impairment of Ar, and LV twist in patients with TTC. We found an intriguing pattern of transient abnormal reversed Ar in three cases with TTC, whereas this abnormal clockwise rotation of the apex was not seen in the other groups. An abnormal reverse Ar has already been described in patients with dilated cardiomyopathy, and was ascribed as a marker of more severe LV remodelling and dyssynchrony with poorer LV systolic and diastolic function.27 The patients with reversed Ar in the TTC group did not have QRS widening and bundle branch but exhibited an EDV, ESV, and WMSI, largely above the median value, suggesting that LV remodelling and extensive regional LV systolic dysfunction may be associated with a loss of LV twist in this setting. The significant negative correlations found in the current study between Ar, LV twist, and WMSI, and LV volumes emphasize these comments. This abnormal pattern of Ar also called ‘rigid body rotation’ has already been described in other settings such as non-compaction cardiomyopathy where it is proposed as a diagnostic criterion.28 However, this abnormal pattern is not specific of non-compaction cardiomyopathy. It has already been shown not only in dilated cardiomyopathy as described above,27 but also in MI,14 and now in TTC.


LV UR and Ad were transiently impaired in patients with TTC, whereas conventional parameters of diastole did not improve significantly, suggesting that untwisting could be a more sensitive marker of diastolic dysfunction in this setting. Indeed, the conventional parameters used for diastolic evaluation (E, A, Ea, Aa) are global parameters of diastolic function, while TTC is mainly characterized by regional functional abnormalities. These tools are influenced by the hyperkinesis of the basal segments, loading conditions, and heart rate, whereas Ad allows the regional evaluation of diastole, at the apex, and in normal settings, occurs during active relaxation, before mitral valve opening.1921 Furthermore, Ad is closely linked to LV UR which is a global diastolic parameter, and to E/Ea ratio which is an index of LV filling pressure. However, the determinants of Ad and its dependence of loading conditions warrant further investigations. More generally, the untwisting motion is related to the amount of potential energy release stored during the twisting activity, and UR is an important contributor of diastolic function, linked to the constant of relaxation, tau, and the intraventricular pressure gradient generated during early diastole (suction).1921The significant correlation found between LV UR and Ea, an index of LV relaxation, in the whole population, and between LV UR and LV twist in the TTC group, illustrates the potential utility of LV twist mechanics as a tool for the evaluation of diastole via LV untwisting. However, it is important to recall that LV twist and UR are influenced by LV volumes.6,17 A significant decrease in ESV at recovery in patients with TTC could be related to a significant increase in LV twist and LV UR without affecting diastolic function.

Time course of left ventricular twist parameters

According to current knowledge, one indubitable diagnostic criteria of TTC is the entire recovery of LV systolic function at follow-up, which is based on the recovery of regional WMA and significant improvement of the LVEF.14,15,29,30 However, these parameters are subject to some limitations and 2D-strain is a proven more sensitive marker of LV systolic function in various settings.5,15,16,29 Therefore, based on a recent report using a longitudinal 2D-strain in patients with TTC, a concern has arisen about the possibility of incomplete recovery of LV systolic function in these patients.15 It was suggested that some subtle regional abnormalities persisted at the early follow-up of the disease despite the normalization of the WMA.15 However, the follow-up period was not standardized in that study ranging from 11 to 66 days, and the recovery is not uniform in TTC, varying from one patient to another, from 1 day to several weeks.14 We have chosen a relatively fixed time point for the evaluation of LV twist mechanics and GLS in our study, allowing a more uniform follow-up period. Except one case with borderline values for some parameters at follow-up, all patients improved the regional apical LV function and recovered entirely with final values not different from controls, even when using more subtle tools such as 2D-strain.

Tako-tsubo vs. acute myocardial infarction

The magnitude of LV twist and Ar is influenced by infarct size25 and LV systolic function11 after MI. Therefore, one could expect that TTC patients would have more marked depression of Ar, and LV twist than MI patients, because WMA extend far beyond one coronary territory in such disease.14 However, the patients with MI were matched for age and the LVEF, factors influencing LV twist mechanics, 6,10,11,17 and therefore no significant difference was found between TTC and MI patients, except in the subgroup of patients with ECG ST-segment elevation at presentation. In this subgroup, TTC patients had a more severely impaired LV twist mechanics when compared with MI probably because of a worse LV systolic function as suggested by a higher WMSI. However, our results do not allow definite conclusions about the differentiation of TTC and MI from LV twist mechanics, giving the small sample size. Another possible mechanism for explaining the relative difference between TTC and MI is related to myocardial salvage. The successful primary angioplasty may convert a transmural MI to a non-transmural MI leading to a paradoxical increase in Ar and LV twist due to the relative sparing of subepicardial fibres not counterbalanced by the impaired subendocardial fibres. Furthermore, at follow-up the LV volumes are significantly lower in TTC when compared with MI maybe influencing the differences of LV twist mechanics seen between the two diseases. According to myocardial fibres arrangements, the LV twist mechanics injury seen in TTC patients implies a transmural extent of myocardial impairment.614 However, contrasting to what is seen in MI where a severe impairment of LV twist is associated with adverse LV remodelling at follow-up,14 this impairment is entirely reversible in TTC.


In recent years, LV variants of TTC have been described including apical sparing and mid-LV form.30 Whether LV twist is similarly impaired in these variants warrants further investigations. Our measurements of LV twist was made using 2D-strain which has been validated against sonomicrometry and resonance magnetic imaging.7,8 However, it is a two-dimensional approach and LV moves in three directions. The through plane motion, particularly at the basal level, is one limitation of the technique in addition to the necessity to obtain a good image quality. Recent advances with 3D-strain should improve the feasibility and accuracy of the method. Our measures were performed at the acute phase and 1 month later in TTC. However, there is a dynamic improvement of LV function in TTC and serial measurements would have permit a more precise evaluation of the time course of LV twist dynamics in this setting. Furthermore, it is possible that the patients without ECG ST-segment elevation at presentation were seen later in the course of the disease and had already partially improved compared with patients with ST-segment elevation. Finally, there is no definite explanation about the significant decrease in Bs in patients with TTC when compared with controls at the acute phase. Apart technical considerations, and unless the decrease in Bs illustrated a subtle abnormality at the basal level despite no WMA, the significant impairment of the systolic and diastolic function in TTC vs. controls could have influenced a little bit the basal rotational function.

In conclusion, LV twist mechanics involving the systolic and diastolic components is transiently impaired in TTC, with in some cases an inverted pattern of Ar, and is entirely reversible. In the subgroup of patients with ST-segment elevation at presentation this injury is more significantly profound than in MI.

Conflict of interest: none declared.


View Abstract