Takotsubo cardiomyopathy, a two-stage recovery of left ventricular systolic and diastolic function as determined by cardiac magnetic resonance imaging

Abstract

Aims

Takotsubo cardiomyopathy (TTC) is an entity mimicking acute myocardial infarction, characterized by transient severe systolic heart failure. Echocardiographic studies suggest that diastolic dysfunction is present in TTC at presentation; however, no reports exist regarding the time course of left ventricular (LV) recovery. This study describes the recovery of LV systolic and diastolic function in TTC. We hypothesized that, in TTC, there is diastolic dysfunction at admission, and that recovery is delayed compared with systolic function.

Methods and results

We enrolled (consecutively 2010–12) 16 patients (mean age 66, range 39–84 years) diagnosed with TTC and 20 healthy matched controls. We performed cardiac magnetic resonance imaging (CMR) at admission, pre-discharge, and 3-month follow-up. Diastolic function was assessed by LV peak filling rate (LVPFR) and left atrial (LA) emptying volumes. At admission, LV ejection fraction was low, increased at pre-discharge (37 ± 6 vs. 58 ± 6%, P < 0.001), and normalized at follow-up (to 65 ± 5%, P = 0.01). LVPFR did not increase during hospitalization (80 ± 3 vs. 89 ± 4 mL/s/m², P = 0.21), but was normalized at follow-up (to 206 ± 19, P < 0.001; controls, 214 ± 13, P = 0.23). During hospitalization, LA passive emptying volume remained low (6 ± 2 vs. 8 ± 3 mL/m², P = 0.05) and LA active emptying volume remained high (17 ± 3 vs. 16 ± 3 mL/m², P = 0.71), whereas LA conduit volume increased (7 ± 3 vs. 23 ± 4 mL/m², P < 0.001). T₂-weighted imaging demonstrated non-coronary distributed apical oedema without contrast enhancement.

Conclusion

Patients with TTC undergo fast systolic recovery. However, at discharge, profound diastolic dysfunction is demonstrated by CMR. At follow-up, both LV systolic and diastolic function is normalized in patients with recovered TTC.

Introduction

Takotsubo cardiomyopathy (TTC) is a recognized syndrome characterized by transient but often severe heart failure most commonly seen in post-menopausal women.^1–4 Long-term prognosis is excellent, but in its extreme TTC is complicated by ventricular rupture and lethal arrhythmias.^5–9 Since TTC is most often seen after physical, pharmacological, or psychological stress, it is suggested that a sympathetic reaction and surge of catecholamines form the pathophysiological basis.^10,11 A characteristic feature of TTC is the initial left ventricular (LV) systolic dysfunction that quickly recovers. Echocardiographic studies have suggested that diastolic dysfunction is also present in the initial phase of TTC.^12–14 However, no data are reported regarding the time course of LV diastolic dysfunction recovery in patients with TTC. Cardiac magnetic resonance (CMR) is the gold standard for the assessment of ventricular systolic function. We have previously reported how minor changes in the left heart diastolic function can be demonstrated by analysis of CMR derived LV and left atrial (LA) time–volume curves,^15,16 and that impaired LA function following an acute myocardial infarction is strongly related to an adverse prognosis.¹⁷

Our objective was to describe the systolic and diastolic recovery during hospitalization and at 3-month follow-up. We hypothesized that despite pronounced improvement of LV systolic function, patients are discharged with persistent LV diastolic dysfunction, which recovers later and can be described by analysis of LV and LA time–volume curves.

Methods

Subjects

We prospectively enrolled consecutive patients with TTC at Rigshospitalet, Copenhagen University Hospital, Denmark (2010–12). Patients were diagnosed in accordance with the modified Mayo Clinic criteria for TTC.⁸ Coronary artery disease was ruled out by coronary angiography. Three CMR studies were performed as part of the protocol. An initial CMR was performed within 24 h after admission. A second CMR was performed before discharge. The final CMR was performed at 3 months of follow-up. Twenty healthy subjects free of medication were enrolled as age- and gender-matched controls. Controls were free of cardiovascular, pulmonary, metabolic, or other disease including arterial hypertension. This was ensured through screening in the Danish national patient registries and a clinical examination including measurement of blood pressure. All patients and controls gave informed oral and written consent, and the study protocol was approved by The Danish National Committee on Biomedical Research Ethics.

Cardiac magnetic resonance

CMR was performed using a 1.5-T magnetic resonance scanner with chest and back surface coils (MAGNETOM Avanto 1.5 T, Siemens, Erlangen, Germany). Balanced steady-state free precession (SSFP) end-tidal breath-hold cine images were acquired in the two-, three-, and four-chamber views followed by contiguous short-axis plane slices covering the entire LV (10–12 slices) and transaxial plane covering the LA (6–8 slices; echo time 1.5 ms, resolution matrix 192 × 192, field of view 300–360 mm, phases 25, slice thickness 8 mm without gap).

The detection of oedema was obtained using a T₂-weighted sequence (short-tau inversion recovery) (time echo 65 ms, slice thickness 8 mm without gap, resolution matrix 192 × 192, field of view 300–360 mm, inversion time 180 ms, repetition time 2 × R-to-R intervals). Contiguous slices in the short-axis image plane were acquired from the atrio-ventricular plane to the apex covering the entire left ventricle as well as in the two-, three-, and four-chamber views.

T₁-weighted inversion recovery gradient-echo images were used for the assessment of myocardial late gadolinium enhancement (echo time 1.4 ms, resolution matrix 192 × 192, and field of view 300–360 mm). Images were obtained 10 min after intravenous bolus injection of 0.1 mmol/kg body weight gadolinium-diethylenetriamine pentaacetic acid (Gadovist, Bayer Schering, Berlin, Germany). The inversion time was determined to null the signal from the normal myocardium. Multiple 8-mm slices in the short-axis image plane were acquired to cover the entire LV without gaps as well as in the two-, three-, and four-chamber views.

The full CMR protocol was performed at admission. The pre-discharge CMR was exclusively performed for volumetric assessment of the heart. At the third CMR at 3 months, T₂-weighted assessment of oedema was omitted.

Image analysis

Offline analysis was performed using the certified semi-automated CMR software (cvi⁴², Circle Cardiovascular Imaging, Inc., Calgary, Alberta, Canada).

Volumes of the LV and the LA were measured from the short-axis and the transaxial SSFP stacks, respectively, in 25 phases covering the cardiac cycle. LV volume measurements were performed by tracing endocardial borders in the short-axis stack images covering the entire LV,¹⁸ and the transaxial stack images were used for LA volumes. When measuring LA volumes the pulmonary veins were excluded, but the atrial appendage was included, and by convention, the inferior LA border was defined as the plane of the mitral valve annulus.^19,20 The reciprocal volumetric nature of the LV vs. the LA through the cardiac cycle and difference in wall thickness was used to carefully distinguish the two chambers in three ways. First, we used long-axis view images as references for the atrioventricular plane. Secondly, we defined thin wall myocardium as LA and thick wall myocardium as LV. Finally, we distinguished LA from LV by the volume changes through the cardiac cycle, i.e. systolic volume reduction of the LV and corresponding expansion of the LA.

Oedema was defined as the hyperintense myocardium when the signal intensity was higher than 2 SD of the signal intensity in the normal myocardium. In TTC, it is a characteristic feature that affected regions extend to more than one coronary perfusion territory, and oedema is often present circumferentially in the short axis.^21,22 Therefore, we additionally assessed oedema on long-axis views using the non-affected, basal myocardium as reference for normal signal intensity.

Delayed gadolinium enhancement was defined as the hyperintense myocardium with signal intensity higher than 5 SD of the signal intensity in the normal myocardium (threshold for myocardial infarction and myocarditis). As with oedema we used basal unaffected myocardium as reference on long-axis view images.

All measurements were made by a CMR physician blinded to the chronology of the CMR examinations (admission, pre-discharge, and 3-month follow-up).

Assessment of size and function

The left ventricle

Time–volume curves were constructed from the 25 volumes covering the entire left ventricle in the short-axis plane throughout the cardiac cycle (Figure 1). From the LV time–volume curves, LV end-diastolic volume (LVEDV) and LV end-systolic volume (LVESV) were defined, and LV stroke volume (LVSV) and LV ejection fraction (LVEF) were calculated. Furthermore, from the time–volume curves, we calculated the LV peak filling rates (LVPFR; Δvolume/Δtime) (Figure 1). Cardiac output was calculated as the product of heart rate and LVSV. All volumes were indexed to body surface area (BSA).

The left atrium

Time–volume curves were constructed from the 25 volumes covering the entire left atrium in the transaxial plane throughout the cardiac cycle (Figure 1). From the LA time–volume curves, specific LA volumes were determined: the minimum volume (LA_min); maximum volume (LA_max); mid-diastolic volume (LA_mdv, volume after passive emptying but before mid-diastolic expansion), and the volume immediately before atrial contraction (LA_bac).²³ From the LA time–volume curve, we deciphered the three volumetric contributions to LVSV, that is, the passive, the active, and the conduit contributions to LV filling. The following volumes were calculated: LA passive emptying volume = LA_max − LA_mdv; LA active emptying volume = LA_bac − LA_min; LA conduit volume = LVSV − (LA passive emptying volume + LA active emptying volume). All volumes were indexed to BSA.

Statistical analysis

Continuous data were tested for normality using the Shapiro–Wilks test and are listed as mean with standard deviation, standard error of the mean, median, and range as appropriate. Data were tested using factorial analysis of variance. Data at follow-up were tested against controls using Student's t-test. For all tests, a two-sided P-value of <0.05 was considered statistically significant. For all analysis, we used the SPSS software, version 20 (SPSS, Inc., Chicago, IL, USA).

Results

In total, 16 patients and 20 healthy controls were enrolled in the study and characteristics are presented in Table 1. Coronary angiograms were without significant stenosis, and ventriculography showed apical akinesia and ballooning in all the patients. CMR data are presented in Table 2. All TTC patients were examined with CMR within 24 h of admission, pre-discharge (median Day 5, range 4–6 days), and 3-month follow-up (median 93, range 91–103 days).

Time–volume curves from the LV, the LA, and LVPFR from admission, pre-discharge, and follow-up are presented in Figure 1A–C. The LV systolic and diastolic change over time is presented in Figure 2.

Oedema and late gadolinium enhancement

Oedema seen on T₂-weighted images was observed across more than one coronary perfusion area. Oedema co-occurred with dyskinetic regions observed on cine images (Figure 3A and B). Delayed gadolinium enhancement was not observed in any of the patients.

The left ventricle

At admission, LVEF was profoundly reduced (37 ± 6%). During hospitalization, we observed a fast recovery with nearly normalized LVEF values at pre-discharge (37 ± 6 vs. 58 ± 6%, P < 0.001). At follow-up, we observed a minor but significant additional recovery with complete normalization of LVEF (58 ± 6 vs. 65 ± 5%, P = 0.01; controls 66 ± 5%, P = 0.55). The LVPFRs were initially severely reduced with no improvement at pre-discharge (80 ± 12 vs. 89 ± 15 mL/s/m², P = 0.21). LVPFRs at follow-up showed complete remission (89 ± 15 vs. 206 ± 19 mL/s/m², P < 0.001; controls 214 ± 13 mL/s/m², P = 0.23). Initially, cardiac index was low (2.4 ± 0.1 L/min/m²; controls 3.2 ± 0.1 L/min/m², P < 0.001) but increased significantly during hospitalization (2.4 ± 0.3 vs. 3.6 ± 0.6 L/min/m², P < 0.001; controls 3.2 ± 0.2 L/min/m², P = 0.002) and had decreased at follow-up (3.1 ± 0.7, P = 0.005; controls 3.2 ± 0.2, P = 0.90). All LV values at follow-up were similar to healthy controls (Table 2) except for LV mass, which was higher in TTC patients (71 ± 12 vs.55 ± 8 g/m², P < 0.001).

The left atrium

LA conduit volume was initially low and increased significantly during hospitalization (7 ± 3 vs. 23 ± 4 mL/m², P < 0.001) with no changes at follow-up (23 ± 4 vs. 21 ± 8 mL/m², P = 0.46; controls, 21 ± 3 mL/m², P = 0.93). During hospitalization, LA passive emptying volume remained low (6 ± 2 vs. 8 ± 3 mL/m², P = 0.05), whereas LA active emptying volume remained high (17 ± 3 vs. 16 ± 3 mL/m², P = 0.71). At follow-up, we observed a complete normalization of LA emptying volumes, all being similar to healthy age- and gender-matched controls.

Discussion

This prospective consecutive study demonstrates the change of LV diastolic and systolic function from hospitalization to 3-month follow-up in patients with TTC. The main finding is that recovery of the LV diastolic function is delayed despite rapid systolic recovery (Figure 2). The first stage of LV recovery in TTC consists of a rapid systolic recovery with no diastolic recovery. The second stage consists of a delayed but complete diastolic recovery with only minor changes in the systolic function of the LV. LV diastolic dysfunction in patients with TTC has previously been reported using echocardiography; i.e. reduced LV early diastolic untwisting rate and reduced early diastolic longitudinal tissue velocities.^12,13 In the present study, the diastolic dysfunction and recovery is demonstrated by a delayed improvement of LVPFR and LA emptying dynamics.

The left ventricle

The profoundly reduced LVPFR during hospitalization demonstrate that patients with TTC suffer from compromised LV relaxation. Systolic performance plausibly affects the filling of the LV, i.e. a well contracting ventricle empties more and hence improves diastolic filling aided by an increased elastic recoil. Interestingly, even though systolic function increased markedly during hospitalization, LVPFR remained low. From admission to pre-discharge, the change in LV time–volume curves revealed prolonged early diastolic filling phase resulting in a shortened diastasis (Figure 1B). This relates to the one diastolic volume that increases during hospitalization; the LA conduit volume. Since filling rates remained low, the increased LA conduit volume represents improved LV filling related to the increased systolic emptying rather than improved diastolic function per se. The LV systolic function increases diastolic recoil and hence transmitral suction also increases. This explains the increased early LV diastolic filling volume at pre-discharge (Figure 1B).

During hospitalization, the patients' clinical status improved as systolic recovery occurred. However, most of the patients stated shortness of breath when compared with their normal state. This could cover the observed sustained LV diastolic dysfunction. Furthermore, we observed a significantly higher heart rate and cardiac index at pre-discharge when compared with follow-up. The elevated heart rate and cardiac index may represent a prolonged elevated adrenergic status at discharge. This is in concordance with Wittstein et al.,¹⁰ who have reported elevated levels of catecholamines as well as neuronal metabolites and neuropeptides in TTC patients at Days 7–9.

The left atrium

During hospitalization, the absent improvement in the LA passive emptying volumes was partly compensated by a sustained high LA active contribution to LV filling (Figure 1A and B). The shortened diastasis at pre-discharge seen in the LV is also mirrored in the LA as a prolonged passive emptying phase (Figure 1B). At follow-up, the LA emptying dynamics is similar to the age- and gender-matched control group (Figure 1C). However, LA size throughout the cardiac cycle tends to be higher in the TTC patients. These differences in LA size probably relate to the high prevalence of hypertension in the TTC group (50%), which over time affects LV mass and LA size. One could speculate that the observed delay in diastolic functional recovery may, in part, be due to LV hypertrophy. However, we observed no difference in the time course of recovery between patients with arterial hypertension and those without. Alternatively, it may represent subtle changes caused by TTC not fully recovered even after 3 months.

Excess catecholamines are present in TTC,¹⁰ resulting in increased LV afterload, increased LV wall stress,²⁴ and elevated LV diastolic filling pressure.²⁵ TTC reproduced in animal models have shown that the combined blockade of alpha- and beta-adrenoceptors can attenuate the cardiac changes.^26,27

In this study, all patients had the TTC apical ballooning variant. Mori et al.²⁸ have shown that while sympathetic innervation is greater in ventricular basis, the density of beta-adrenoceptors is higher in the apical myocardium which suggest that the humoral component is important in TTC.

Oedema, a marker of myocardial damage, had a non-coronary distribution in all our patients, which is a characteristic feature in TTC.²¹ Resolution of oedema at follow-up did not translate into lowered LV mass. Previously, it has been shown that resolution of oedema in myocarditis resulted in lowered LV mass.²⁹ However, this was not the case in our cohort of TTC. Even though the LV volumes normalized, the remaining elevated LV mass at follow-up despite resolution of oedema is most likely due to the high prevalence of hypertension in the patients. This also explains the elevated LA size at follow-up. At admission, we observed no difference in blood pressures when compared with the controls. This seemingly ‘normal’ blood pressure at admission may either cover well regulation or more plausible poor cardiac performance. Furthermore, persistent diastolic dysfunction despite systolic recovery could relate to calcium mishandling. Calcium handling in the failing heart is complex; however, proper cyclic calcium release and reuptake is crucial for preserving LV function. Diastolic lowering of cytosolic calcium enabling myocardial relaxation depends on SERCA2a and Na⁺/Ca²⁺ exchanger located in the sarcoplasmatic reticulum (SR) and the cell membrane, respectively. In the initial phase of TTC, it has been demonstrated that gene expression of SERCA2a is down regulated,³⁰ and in a rat model of TTC, calcium leakage was observed due to hyperphosphorylation of SR ryanodine receptor 2.³¹ Mann et al.³² suggested that excess catecholamines directly induce calcium overload and myocyte dysfunction; even without ATP depletion. In summary, cytosolic calcium overload could partly explain not only the initial systolic dysfunction, but also the delayed diastolic recovery observed in this study.

Perspectives and clinical implications

Patients with TTC are discharged at a time where diastolic impairment is still evident. Therefore, in the post-discharge management of patients with TTC, clinicians should be aware of a potential discrepancy between the observed cardiac recovery and patient's symptomatic status. One should keep in mind that symptomatic relief will not always fully reflect the degree of systolic recovery. The majority of the patients stated shortness of breath at discharge when compared with their normal state. This could cover a degree of heart failure with a preserved ejection fraction. The treatment of heart failure with the preserved ejection fraction is complex, but could be necessary in TTC—at least as a bridge until diastolic recovery has occurred.

Limitations

Only one follow-up measurement was performed at 3 months after hospitalization, and recovery of both systolic and diastolic function most likely occurs gradually over time with some variations.³³ Hence, serial CMR examinations after discharge would allow a more detailed description of LV recovery. Since all patients included in this study exhibited the apical ballooning variant of TTC, further studies are warranted to determine whether the presented findings apply to the other variants of TTC (basal, mid-LV, and biventricular). We did not measure catecholamines that are a central component in TTC. Myocardial biopsies would enable a more comprehensive assessment of the cellular changes and in particular calcium burden; however, this was beyond the scope of this study.

Conclusion

In patients with TTC, the LV recovery consists of two components. The first is a rapid systolic recovery with a minimal diastolic recovery, whereas the second component is a postponed diastolic recovery accompanied by a minor residual systolic recovery.

Conflict of interest: None declared.

Funding

The study was supported financially by the Danish Agency for Science, Technology and Innovation, the Strategic Research Council (grant no. 09-066994/DSF), The Beckett Foundation, and the Research Foundation of the Department of Cardiology, Rigshospitalet.

References