OUP user menu

Transient impairment of coronary flow reserve in tako-tsubo cardiomyopathy is related to left ventricular systolic parameters

Patrick Meimoun , Dorothée Malaquin , Tahar Benali , Jacques Boulanger , Hamdane Zemir , Christophe Tribouilloy
DOI: http://dx.doi.org/10.1093/ejechocard/jen222 265-270 First published online: 1 January 2008


Aims Recent studies suggest that coronary flow reserve (CFR) is transiently impaired in tako-tsubo cardiomyopathy (TTC). Mechanisms by which such impairment occurs are still unknown. To assess the relationship between CFR obtained by transthoracic Doppler echocardiography (TDE) and parameters of left ventricular (LV) performance in patients with TTC.

Methods and results A total of 20 consecutive patients in sinus rhythm, with TTC (mean age 70 ± 9 years, 19 women) underwent serial evaluation of TDE–CFR, in the distal part of the left anterior descending coronary artery (LAD), at the acute phase and after recovery using intravenous adenosine infusion (140 µg/kg/min over 2 min). CFR was calculated as hyperaemic to basal mean diastolic coronary flow velocity (CFV). Average of the septal and lateral mitral annulus early diastolic (Ea) and systolic (Sa) tissue velocity, early (E) and late (A) diastolic transmitral velocity, the ratio E/Ea, wall motion score (WMS, 16 segment model), LV end-systolic volume index (ESV/m2) and LV end-diastolic volume index (EDV/m2, biplane-Simpson method) were serially measured by TDE. Basal CFV, LV mass index and haemodynamics parameters did not differ between acute phase and recovery, whereas hyperaemic CFV increased significantly after recovery (P < 0.01) leading to a greater CFR (2.9 ± 0.3 vs. 2.1 ± 0.4, P < 0.0001). At the acute phase, hyperaemic CFV was significantly correlated to WMS, ESV/m2, but not to E/Ea, whereas at recovery, hyperaemic CFV was not correlated to LV parameters. The improvement of CFR was closely correlated to the decrease of ESV/m2, of WMS, but not to diastolic parameters. No significant correlation was found between CFR and E/Ea or LV mass index at each stage.

Conclusion There is a transient impairment of CFR at the acute phase of TTC, which is due to a reduced vasodilating capacity. This impairment is closely correlated to LV systolic parameters. Diastolic compressive forces to the coronary microcirculation do not appear to play a critical role.

  • Coronary flow reserve
  • Tako-tsubo
  • Coronary microcirculation


The clinical features of tako-tsubo cardiomyopathy (TTC) are well described, affecting typically post-menopausal women, secondary to a stressful event.18 The main characteristic is transient left ventricular (LV) wall motion abnormalities in the mid-apical regions with hyperkinesis of the basal segments, giving a balloon-like appearance of the left ventricle.18 The coronary angiography is normal or shows coronary narrowing <50%.18 Despite these characteristics are being well recognized, the pathophysiology of this fascinating cardiomyopathy is unclear. Recent studies suggest that the coronary microcirculation is involved and transiently impaired at the acute phase of the syndrome.913 However, the mechanisms by which such impairment occurs are still unknown. Recent advances in transthoracic Doppler echocardiography (TDE) allow non-invasive evaluation of coronary flow velocity (CFV) and coronary flow reserve (CFR) with a high success rate, and previous studies have validated the feasibility and usefulness of non-invasive evaluation of CFR in various settings.1416 We have shown that the non-invasive CFR is transiently impaired in TTC.13 As mechanisms by which such impairment occurs are unknown, the objective of the current study is to assess the relationship between CFR, CFV, and parameters of systolic and diastolic LV function, in patients with TTC, at the acute phase and after recovery.


Twenty consecutive patients with TTC were prospectively included. They underwent serial evaluation of non-invasive CFR and TDE during the same exam, at the acute phase of the syndrome (within 48 h of admission), and after recovery (25 ± 5 days apart), using commercially available machines (Philips IE33 system, and Sequoia 256 ultrasound system, Mountain View, CA, USA). TTC was diagnosed according to the following criteria: acute chest pain or dyspnoea, a stressful event, transient LV wall motion abnormalities (WMA) at the apex of the ventricle, electrocardiogram (ECG) changes (ST-segment elevation or T-wave inversion), and no luminal narrowing >50% in all epicardial coronary arteries on angiography.8,9 Exclusion criteria were contra-indication to adenosine, febrile or acute neurological disorder, history of myocardial infarction, and significant valvular disease. Apart from CFR and TDE performed simultaneously for the study, within 48 h of admission and 3–4 weeks later in an outpatient visit, each patient underwent standard transthoracic echocardiography at admission and at least one repeat echocardiogram at 1 week (before discharge). Each patient gave written informed consent to participate in the protocol. The first 12 patients have already been included in the previous report.13

Coronary flow reserve

Non-invasive CFR was performed as previously described,1417 using intravenous adenosine infusion (140 µg/kg/min over 2 min). Briefly, the mid-distal part of the left anterior descending coronary artery (LAD) was studied using a low multifrequency transducer (3V2C probe),13,14,17 and the artery was visualized by colour Doppler flow mapping guidance, in the modified parasternal view. For colour Doppler echocardiography, the velocity range was defined from 12 to 16 cm/s. Blood flow velocity was measured by pulsed wave Doppler echocardiography, using a sample volume of 3–4 mm, placed on the colour signal in the distal LAD. The ultrasound beam direction was aligned as closely as possible with the distal LAD flow. No angle correction was performed for the study given that CFR is the ratio between hyperaemic and baseline flow velocity, and it is not affected by the actual flow velocity. However, the angle was kept as small as possible. CFR was calculated as the ratio of hyperaemic to basal mean diastolic flow velocity. The spectral Doppler signal of LAD flow was recorded at an identical portion of the artery for a given patient on each examination. Blood flow velocity measurements were performed offline by an experienced investigator blinded to patient data, by contouring the spectral Doppler signals, using the integrated software package of the ultrasound system. Final values of flow velocity represented an average of three cardiac cycles. Furthermore, a coronary resistance index (CRI) was calculated at baseline and after adenosine administration, as the ratio of diastolic blood pressure and simultaneously measured mean diastolic CFV. A contrast agent (Sonovue; Bracco, Altana Pharma France) was used in five cases (25%) to improve visualization of the colour Doppler signal and/or to obtain clear spectral Doppler signals in the LAD and was administered intravenously at a concentration of 45 µg/mL, in a volume of 5 mL, as a 0.1 mL bolus.

Transthoracic Doppler echocardiography

Serial TDE measurements were performed during the same exam as the CFR evaluation in which the patient was in the left lateral decubitus position. For the LV, following parameters were measured: end-diastolic diameter, end-systolic diameter, in the parasternal long-axis view, end-systolic volume index (ESV/m2), end-diastolic volume index (EDV/m2), and ejection fraction (EF) by the Simpson biplane’s rule, wall motion score (WMS), and WMS index (WMSI) using the 16 segment model.18 Conventional diastolic transmitral flow Doppler parameters (E wave, A wave, E/A ratio, deceleration time of E) and tissue Doppler velocities were obtained according to a standardized examination.19 Diastolic (Ea, Aa, Ea/Aa) and systolic (Sa) tissue Doppler velocities were obtained by using the average of the septal and lateral mitral annulus velocities. The ratio of E/Ea, an index of LV filling pressure,20 and the ratio E/Ea/EDV, an index of LV stiffness were also measured. Left atrial diameter was measured in the left parasternal view, and 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 two other experienced observers blinded to patient data.

Coronary angiography

Selective coronary angiography was performed using standard techniques, at the acute phase in all patients, and demonstrated normal coronary anatomy in all but two patients, who had an intermediate stenosis of a small diagonal branch (50%) and a marginal branch (40%). LV angiography demonstrated a typical apical ballooning appearance like an octopus pot in all patients. The mean LVEF was 38 ± 9%.


Values are expressed as means ± SD and percentages. Two-tailed Wilcoxon signed rank test and Mac-Nemar test were performed to compare the two time points, and Mann–Whitney test to compare independent samples. The relationship between CFV, CFR, and various parameters was evaluated by Spearman’s rank correlation. These parameters included age, BMI, coronary risk factors, LV mass/m2, hemodynamic variables, WMS, WMSI, ESV/m2, LVEF echo, EDV/m2, E, A, E/A, DTE, Ea, Aa, Ea/Aa, Sa, E/Ea, E/Ea/EDV/m2 and pulmonary artery systolic pressure. P< 0.05 was considered significant. The inter- and intra-observer variability for CFR measurements in our experience have been previously reported (∼5 and 4%, respectively).14,17 Intra-individual variability of CFR has also been reported at our laboratory: the upper limit of agreement for serial measurements, using the Bland–Altman method, in a control group is 12%.13


The characteristics of the study population are listed in Table 1. The mean age was 70 ± 9 years, and all but one patient were women. A stressful event was detected in all but two patient (emotional, n = 16, and physical, n = 2). The treatments were left at the discretion of the treating physicians at each stage of the disease. At the acute phase, two patients received an inotropic support, and hemodynamic support with intra-aortic balloon conterpulsation was required for two patients for <48 h. However, at the time of Doppler study no patient had any hemodynamic or inotropic support. Four patients (20%) were already on beta-blocker at the time of the first Doppler study and 12 (60%) at the time of the second Doppler study (P < 0.05). No significant difference was seen between the acute and the recovery phases concerning the use of other usually prescribed cardiovascular drugs that could influence the coronary microcirculation (see Table 1).

View this table:
Table 1

Baseline characteristics

Age (years)70 ± 9
Stressful event18/20 (90%)
BMI (kg/m2)27 ± 8
Hypertension14/20 (70%)
Diabetes5/20 (25%)
Dyslipidaemia10/20 (50%)
Smoking2/20 (10%)
BSA (m2)1.73 ± 0.2
LVEF (angio)38 ± 9%
Troponin peak0.6 ± 0.5 µg/L
chest pain/dyspnoea14/6
ECG (admission)a
ST elevation8 (40%)
T waves inversion11 (55%)
ACEi/ARA II10/20 (50%)16/20 (80%)
Beta-blocker4/20 (20%)12/20 (60%)
Statin6/20 (30%)9/20 (45%)
  • aLeft bundle branch block in one case. BSA, body surface area; BMI, body mass index; LVEF, left ventricular ejection fraction; ACEi/ARA II, angiotensin-converting enzyme inhibitors/angiotensin receptor antagonists.

Adenosine was well tolerated and no serious adverse event occurred during adenosine infusion. The CFV results are listed in Table 2. Resting CFV did not change at follow-up. Hyperaemic CFV increased significantly at the recovery phase compared with the acute phase, and therefore CFR improve significantly at follow-up, compared with the acute phase: 2.9 ± 0.3 vs. 2.1 ± 0.4 (P < 0.01). All the patients increased their CFR at the recovery phase compared with the acute phase, with a mean improvement of 40%. However, all but two patients increased significantly their CFR above the level of 12% between the two phases of the disease.

View this table:
Table 2

Haemodynamic and coronary flow parameters

Acute phaseRecovery phase
Heart rate (bpm)80 ± 1588 ± 2268 ± 11*79 ± 20
Systolic blood pressure (mmHg)114 ± 18111 ± 16133 ± 23*124 ± 22
Diastolic blood pressure (mmHg)66 ± 1265 ± 1365 ± 964 ± 11
Rate-pressure product (bpm mmHg)9133 ± 27819646 ± 31378951 ± 21749912 ± 2884
Mean coronary flow velocity (cm/s)23 ± 548 ± 1123 ± 567 ± 16 **
Coronary resistance index (mmHg/cm/s)3 ± 1.11.4 ± 0.42.9 ± 0.70.96 ± 0.2**
  • * P < 0.01 vs. acute phase; ** P < 0.0001 vs. acute phase.

No significant change occurred for baseline haemodynamic parameters between the two phases of the disease, except for heart rate and systolic blood pressure. Hyperaemic CRI increased significantly at the recovery phase compared with the acute phase (Table 2).

TDE results are summarized in Table 3. The echocardiography performed at admission demonstrated a typical mid-apical akinesia with hyperkinesis of the basal segments of the LV in all patients. However, WMA improved gradually, and the WMS and WMSI measured within 48 h, at the same time to the CFR test, are not necessarily the same for all patients (one patient, for example, recovered WMA very early, within 48 h). Systolic anterior motion of the mitral valve was observed in three patients at the acute phase, with significant LV outflow tract obstruction, unmasked after the Valsalva maneuver, in two cases (peak gradient at 44 and 66 mmHg). After recovery, no patient exhibited any LV obstruction.

View this table:
Table 3

Echocardiographic Doppler parameters

Acute phaseRecovery phase
EDV/m2 (mL/m2)53 ± 849 ± 9*
ESV/m2(mL/m2)30 ± 814.5 ± 5*
Stroke volume (mL/m2)23 ± 435 ± 6*
LVEF (%)43 ± 769 ± 5*
WMS31 ± 516 ± 0.6*
WMSI1.94 ± 0.31 ± 0.03*
EDD (mm)47 ± 446 ± 4
ESD (mm)27 ± 528 ± 4
LV mass (g/m2)80 ± 1280 ± 14
LA diameter (mm)33 ± 633 ± 6
E (cm/s)63 ± 2474 ± 25
A (cm/s)76 ± 2287 ± 23
E/A0.9 ± 0.50.8 ± 0.2
DTE (ms)198 ± 60204 ± 35
Ea (cm/s)7.3 ± 2.59 ± 2**
Aa (cm/s)10 ± 310.5 ± 2
E/Ea9.4 ± 38.8 ± 3
Sa (cm/s)7.4 ± 29 ± 2**
E/Ea/EDV/m20.18 ± 0.070.19 ± 0.08
PASP (mmHg)38 ± 836 ± 10
  • *P < 0.01, **P < 0.05. EDV/m2, end-diastolic volume index; ESV, end-systolic volume index; LV, left ventricle; EDD, end-diastolic diameter; ESD, end-systolic diameter; LA, left atrium; DT, deceleration time of mitral E wave; PASP, pulmonary artery systolic pressure.

The systolic LV parameters improved significantly at the recovery phase compared with the acute phase, and there was also a significant improvement of the EDV/m2. Concerning the Doppler parameters, there was a significant change for only tissue Doppler velocity parameters such as Ea and Sa. The ratio E/Ea and E/Ea/EDV did not change significantly between the two phases of the disease.

Influence of treatments

At the acute phase, compared with patients without beta-blockers (n = 16), patients already on beta-blockers (n = 4) had a slightly but not significantly higher CFR (median 2.4 vs. 2.1, P = 0.06) and a slightly but not significantly better LV systolic function (WMS, median: 28 vs. 33, P = 0.21; LVEF, median: 48 vs. 43%, P = 0.3). At the recovery phase, patients on beta-blockers (n = 12) did not differ to patients without beta-blockers (n = 8) concerning CFR (median 2.9 vs. 2.96, P = 0.9), LV systolic, and diastolic parameters. Furthermore, the patients who started a beta-blocker during hospitalization (n = 8) did not differ significantly to patients who did not receive beta-blockers (n = 8), with regard to the final CFR and echo-Doppler parameters. The same comparisons were performed in patients with and without statin, or angiotensin-converting enzyme inhibitors/angiotensin receptor antagonist, and no any significant difference was found for each treatment, at any stage of the disease, with regard to CFR, LV systolic, and diastolic parameters.


At the acute phase, the hyperaemic CFV was significantly correlated to the WMS (r = −0.44, P = 0.05), the WMSI (r = −0.47, P = 0.04), and the ESV/m2 (r = −0.54, P = 0.01) (Figure 1), but not to other parameters (Figure 2). At the recovery phase, no significant correlation was found between the hyperaemic CFV and other parameters.

Figure 1

Scatter plots showing significant relationship at the acute phase between hyperaemic coronary flow velocity and end-systolic volume index, coronary flow reserve (CFR) and wall motion score (WMS) index, and delta CFR and delta WMS.

Figure 2

Scatter plots showing the absence of significant relationship at the acute phase between hyperaemic coronary flow velocity and E/Ea, CFR and E/Ea, and CFR and mitral E wave deceleration time.

At the acute phase, CFR was significantly correlated to ESV/m2 (r = −0.45, P = 0.05), to EDV/m2 (r = −0.57, P = 0.01) and to WMSI (r = −0.5, P = 0.03) (Figure 1). At the recovery phase, no significant correlation was found between CFR and other parameters.

The delta CFR (defined as CFR at the recovery phase minus CFR at the acute phase) was significantly correlated to the delta of only systolic parameters, such as ΔWMS (r = −0.86, P = 0.0001) (Figure 1), ΔWMSI (r = −0.87, P < 0.0001), and ΔESV/m2 (r = −0.53, P = 0.03). No significant correlation was found between CFR and E/Ea (Figure 2), between CFR and E/Ea/EDV/m2 and between CFR and LV mass index at each stage of the disease.


This study confirms that there is a transient impairment of the CFR at the acute phase of TTC. This transient impairment is due to a reduced vasodilating capacity and is closely correlated to LV systolic parameters, but the diastolic compressive forces to the coronary microcirculation do not appear to play a critical role in this setting.

Recent studies using nuclear cardiac imaging,9 invasive Doppler flow wire,11 coronary angiography with index of myocardial perfusion, such as the TIMI frame count10 and the TIMI myocardial perfusion grade,12 and CFR by TDE,13 suggested that there is a transient impairment of the coronary microcirculation at the acute phase of TTC. TDE allows simultaneous evaluation of LV performance and CFR. As the patients described in this study did not present any flow-limiting LAD stenosis, CFR in this setting explores the microcirculation.21,22 Adenosine acts substantially at the microcirculatory level, given that great epicardial arteries are relatively insensitive to this agent.23 Consequently, the variation of flow velocity recorded in the distal part of the LAD with adenosine is a surrogate marker of the variation of regional blood flow.

Given that CFR is influenced by several factors in general, with complex interplay between each over, no single mechanism can fully explain the entire spectrum of CFR, particularly in TTC where there is a dynamic improvement of both the coronary microcirculation and WMA over time. However, in this study, we found that the diastolic parameters were not significantly correlated to the blunted hyperaemic flow velocity at the acute phase, and the improvement of CFR was not significantly correlated to the change of diastolic parameters. Furthermore, at each stage of the disease, there was no significant correlation between E/Ea, which is an index of LV filling pressure,20 E/Ea/EDV/m2, which is an index of LV stiffness, and CFVs and CFR. These results strongly suggest that the diastolic compressive forces to the coronary microcirculation do not play a key role for explaining acute and transient impairment of CFR in TTC.

On the contrary, there was a close correlation between the change of CFR and the change of WMS, WMSI, and ESV/m2. Furthermore, there was a blunted vasodilating capacity suggested by the impairment of the hyperaemic CFV at the acute phase of TTC. The close correlation, at the acute phase, between the hyperaemic CFV and systolic parameters such as WMS, WMSI, and ESV/m2, but not with E/Ea and E/Ea/EDV/m2, implies that passive diastolic forces do not explain this injury. However, the exact anatomic supports of CFR and hyperaemic CFV impairment at the acute phase of TTC are unknown. Secondary to a stressful event, an exaggerated transient catecholamine release giving rise to myocardial stunning in a predisposed myocardium is a mechanism strongly suggested in TTC.4,5,8,10 Furthermore, increased local release of norepinephrine from the heart of patients with TTC has recently been described, and the authors speculated that this increased local release of catecholamine might cause the transient LV apical ballooning in TTC.24 In addition, regional defects on cardiac 123-metaiodobenzyl-guanidine-enhanced imaging have been demonstrated in patients with TTC,25 strongly suggesting that the disturbance of cardiac sympathetic innervation was present at the acute phase. Taken together with impaired CFR, these findings suggest that the coronary microcirculatory impairment in TTC would be sympathetically mediated. Interestingly, mental stress that is associated with enhanced sympathetic activity induce endothelial dysfunction,26 and the integrity of the endothelium is crucial for the regulation of the coronary flow at rest and also during hyperaemia.27 Furthermore, as TTC affect preferentially post-menopausal woman, sexual hormones might play a role in this setting, as suggested by experimental studies.28 Interestingly, estrogens have a positive impact on the vascular bed, via endothelium dependent and independent mechanisms,29 and these hormones improve CFR, consistent with a protective role at the coronary microcirculatory level.30

The close correlation between the impairment of CFR and extend of WMA in this study strongly suggest that the injury of myocardial flow reserve might play a role in the pathogenesis of acute and transient WMA seen in TTC. A recent study using the TIMI myocardial perfusion grade during angiography showed that the impaired myocardial perfusion is frequently present in patients with TTC and correlates with the extent of myocardial injury.12 Sudden surges in circulating catecholamine levels could, however, induce both microvascular dysfunction and direct cardiotoxicity, as suggested by the transient structural alterations seen in myocardial biopsy samples of patients with TTC.31 At the acute phase, the moderate impairment of CFR, above the level of 2, compared with the extent of WMA seen in some patients, maybe imply this double target for catecholamine toxicity.


In our study, CFR at the acute phase was performed not immediately after admission but when the patients were in stable condition (within 48 h). As there is a gradual improvement of WMA in TTC, CFR could have been worst if performed immediately. However, inpatients without epicardial coronary stenosis, CFR explores the microcirculation under stable haemodynamic condition.21,22 This requirement was not met in the majority of patients at admission. Resting CFV, which is closely related to myocardial oxygen consumption and metabolic state, could influence CFR. However, resting CFV did not change between the two exams, neither the rate pressure product. Therefore, these factors are unlikely to play a role in the transient impairment of CFR seen in our patients. The maximal vasodilating capacity of the coronary microcirculation depends on the coronary perfusion pressure. Although the coronary perfusion pressure was not directly assessed in this study, we can speculate that it did not change between the two exams given that the diastolic blood pressure was similar for both CFR evaluations. Furthermore, baseline CRI did not change between the two exams, but hyperaemic CRI was significantly higher at the acute phase of TTC, reinforcing the hypothesis of blunted vasodilating capacity for explaining the transient impairment of CFR in TTC. Observations on flow reserve were limited to the distal LAD and do not necessarily reflect changes in other coronary arteries. And finally, the lack of correlation between Doppler parameters of diastolic dysfunction and CFR does not necessarily imply that diastolic compressive forces do not play any role in the impairment of CFR. In fact, CFR as measured by echocardiography assessed a transmural (both subepicardial and subendocardial) CFR, and cannot detect subtle change due to wall tension in a kinetic regions at the subendocardial level, where blood flow is directly exposed to LV diastolic pressure.

In conclusion, the transient impairment of CFR in TTC is not just a passive phenomenon related to the disturbance of haemodynamic parameters. Drugs that improve haemodynamic parameters and reduce fluid overload would not be expected to increase dramatically CFR in this setting. The close link between CFR and the extent of myocardial injury strongly suggests that the coronary microcirculation might play a role in the pathogenesis of TTC. Drugs that have a positive impact on the coronary microcirculation such as new generation beta-blockers would be useful to prevent, or at least attenuate the harmful consequences of excess catecholamine release in this setting.

Conflict of interest: none declared.


View Abstract