OUP user menu

Immunohistological basis of the late gadolinium enhancement phenomenon in tako-tsubo cardiomyopathy

Andreas Rolf, Holger M. Nef, Helge Möllmann, Christian Troidl, Sandra Voss, Guido Conradi, Johannes Rixe, Holger Steiger, Katharina Beiring, Christian W. Hamm, Thorsten Dill
DOI: http://dx.doi.org/10.1093/eurheartj/ehp140 1635-1642 First published online: 23 April 2009


Aims Tako-tsubo cardiomyopathy is characterized by transient contractile dysfunction after emotional or physical stress. Only few patients show late gadolinium enhancement (LGE) in cardiovascular magnetic resonance imaging (MRI). It was the purpose of this study to elucidate the histological basis of this phenomenon.

Methods and results The study included 15 patients. Tako-tsubo cardiomyopathy was diagnosed by coronary angiography and ventriculography. Cardiac MRI was performed within 24 h of admission. Endomyocardial biopsies were taken during the acute phase and after recovery. The content of fibrosis was determined by immunohistochemical staining of collagen-1. In the acute phase, cardiac MRI revealed LGE in five patients. This was completely reversed at follow-up [14, inter-quartile range (IQR) 11–14.5 days]. All patients showed a significant increase of collagen-1 compared with control tissue. Moreover, the amount of collagen-1 was significantly higher in LGE positive patients (LGE positive: 18.84, IQR 13.82–19.75 AU/μm2; LGE negative: 7.57, IQR 5.41–9.19 AU/μm2, P = 0.001). The presence of LGE was not associated with poorer left ventricular function.

Conclusion The presence of LGE cannot rule out tako-tsubo cardiomyopathy. Instead it defines a special subgroup of patients with a disproportionate increase of extracellular matrix.

  • Late gadolinium enhancement
  • Fibrosis
  • Cardiac magnetic resonance imaging
  • Apical ballooning
  • Tako-tsubo cardiomyopathy
  • Stress cardiomyopathy


Tako-tsubo cardiomyopathy, also referred to as apical ballooning or stress cardiomyopathy, has been increasingly recognized over the past 17 years. Dote et al.1 were the first to describe this new cardiac entity, which is characterized by transient hypokinesia or akinesia of the apical and mid-portions of the left ventricle (LV) without significant coronary artery stenosis. The syndrome is accompanied by acute chest pain, dynamic reversible ST segment abnormalities, and marginal elevation of cardiac enzymes disproportionate to the extent of akinesia. Emotional or physical stress usually precedes this cardiomyopathy.24 The pathomechanism has not yet been fully clarified. The most frequently proposed hypothesis includes excessive catecholamine levels leading to structural alterations.57

In the absence of coronary artery stenosis, several differential diagnoses with a very similar phenotype must be considered. The differentiation of embolic myocardial infarction, myocarditis, and tako-tsubo cardiomyopathy can be difficult.8,9 Cardiac magnetic resonance imaging (MRI) is an excellent tool in this setting. There is no scar tissue to be found, and the LV dysfunction in tako-tsubo cardiomyopathy is completely reversible. So the absence of late gadolinium enhancement (LGE) in tako-tsubo cardiomyopathy patients as documented by cardiac MRI studies has been described in many cases10 and is a common diagnostic criterion in most MRI centres. Recently, several studies challenged this notion by reporting delayed hyperenhancement in tako-tsubo cardiomyopathy patients. However, a morphological backdrop for LGE could not be evaluated in these studies, because no endomyocardial biopsies (EMBs) were taken.1113

Late gadolinium enhancement depends on differences in the volume of distribution of gadolinium in normal and pathologically altered myocardium.14 Several morphological changes within the myocardium may contribute to a larger volume of distribution. In general, the distribution of gadolinium is confined to the extracellular and interstitial space. The interstitial space constitutes only about 15% of the whole myocardium. Therefore, any changes to the interstitium, such as oedema or fibrosis, increase the volume of distribution causing LGE.15,16 Furthermore, myocardial necrosis with the loss of membrane integrity causes intracellular accumulation of gadolinium, which contributes greatly to an expansion of the volume of distribution and LGE.17 Therefore, in order to explain the presence of LGE in tako-tsubo patients, differences in the extent of oedema, necrosis, and fibrosis between LGE positive and LGE negative patients have to be examined systematically.

The immunohistological background of tako-tsubo cardiomyopathy has been documented.18 In systematic EMBs, structural deteriorations characterized by disorganization of contractile and cytoskeletal proteins were detected, but no signs of oncotic or apoptotic cell death could be found.19 The level of extracellular matrix (ECM) proteins was elevated. The amount of collagen-1 and fibronectin was significantly raised in the phase of severe contractile impairment and normalized after functional recovery.18 After functional recovery, all described alterations showed a nearly complete reversibility.

Consequently, as there were no histological signs of necrosis, it is most likely that the increase of collagen-1 and fibronectin contributes most to an increase of the volume of distribution of gadolinium.

Thus, this investigation hypothesizes that LGE in tako-tsubo cardiomyopathy could be correlated with a remarkable increase of ECM.


Ethics and patients

The study was approved by the Ethics Committee of the University of Freiburg. Each patient gave written informed consent to participate in the study. The investigation conformed to the principles outlined in the Declaration of Helsinki.

A total of 33 patients who met the diagnostic criteria of tako-tsubo cardiomyopathy had been examined by cardiac MRI between September 2006 and July 2008. Fifteen of these patients, for whom myocardial biopsies were available, were included in this study. Of the remaining 18 patients, 2 were not eligible for biopsy because of haemodynamic instability, the others consented to MRI but not to biopsy. All patients showed an acute onset of chest pain associated with transient ST-elevation and mild increase of cardiac enzymes. Each patient was carefully assessed with a history and physical examination.


The diagnosis was based on the clinical presentation with dyspnoea and/or chest pain, electrocardiogram, and the typical wall motion abnormalities in the left ventriculography, despite the absence of coronary artery disease.

Coronary angiography and ventriculography

Coronary angiography was performed upon admission in Judgkins technique with a 6 Fr catheter. At least five different projections of the left coronary artery, two of the right coronary artery, and two of the LV were obtained. Coronary artery disease was defined as >50% reduction in the lumen diameter.

Cardiac magnetic resonance imaging

A 1.5 T scanner (Siemens Sonata®, Erlangen, Germany) and a six-element phased-array surface coil were used for all MRI studies. The patients were positioned supine, head first.

LV function

After survey localizer sequences, cine images of the heart were acquired in standard two-chamber, four-chamber, LV outflow tract, and short-axis orientations using a fast, ECG-gated, breath-hold, steady-state, free precession sequence (TE: 1.58 ms, TR: 4.8 ms, flip angle: >60°, and slice thickness: 6 mm).

Wall motion abnormalities of the LV were evaluated following the American Heart Associations 16-segment model, a wall motion score was averaged over all 16 segments. On a stack of short-axis views (SLT 10 mm, no gap) covering the entire LV, endocardial and epicardial contours were drawn in each slice, using commercially available software (CAAS, Pie Medical, Maastricht, The Netherlands), in order to analyse the functional parameters. Ejection fraction (EF), end-diastolic volume, end-systolic volume (ESV), and stroke volume were calculated by multiplying the area with slice thickness according to Simpson’s method. All volumes presented are normalized to body surface area.

Late gadolinium enhancement

Contrast agent (0.2 mmol gadolinium-DTPA/kg bodyweight, Omniscan®, Amersham, Amersham, UK) was injected for viability imaging. After 10–15 min, imaging was performed by the use of an inversion recovery three-dimensional-TurboFLASH sequence (TR: 440 ms, TE: 1.25 ms, Flip angle: 10°, optimized TI: 270–310 ms, SLT: 5–6 mm, 14 slices, 39 segments, and voxel size: 1.6 × 2.0 × 7 mm interpolated to 1.6 × 2.0 × 5 mm) covering the entire myocardium in two-chamber, four-chamber, and short-axis planes.

Late gadolinium enhancement was defined as signal activity greater than two standard deviations from remote tissue as described by Kim et al.20 To avoid over or underestimation of the enhancing regions by image noise, a large reference area was defined. The presence or absence of LGE was coded dichotomously (LGE positive vs. LGE negative). In addition, we quantified LGE using commercially available Software (CAAS) on 24 contiguous short-axis slices with semi-automatic infarct detection according to the method described by Kim (compare above).


T2-weighted inversion recovery turbo spin echo sequences (TR: 2 RR intervals, TE: 53 ms, SLT: 6 mm, and three long-axis slices) were used for semi-quantitative analysis of oedema. The average area of signal enhancement was computed using the method of two standard deviations of enhancement from predefined, remote myocardium (compare above). The percent area of oedema was averaged between all long-axis slices.

Laboratory measurements

At the timepoint of the MRI examination, all patients gave blood samples from an antecubital vein into gel-filled tubes. Serum concentrations of creatinine kinase (CK), CK-MB, and troponin were measured following standard procedures. The reference values were: <140 U/L for CK, <24 U/L for CK-MB, and <0.01 for troponin.

Endomyocardial biopsies

Endomyocardial biopsies were obtained from each patient within 48 h after the onset of symptoms in the phase of severely depressed contractile function and after functional recovery [14, inter-quartile range (IQR) 11–14.5 days after admission], which was documented by means of serial echocardiography and MRI. Six biopsies were taken from the anterolateral and/or apical region of the LV guided by real-time three-dimensional echocardiography (IE33, Siemens Medical Systems, Erlangen, Germany) and fluoroscopy. Using this technique, the biopsy’s point of origin could be determined as being exactly from the dysfunctional region.21,22 The tissue was immediately immersed in liquid nitrogen.

Electron microscopy

Small tissue samples were embedded in Epon following routine procedures. Ultra-thin sections were stained with uranyl acetate and lead citrate, and both viewed and photographed in a Philips CM 10 electron microscope.


The tissue samples were mounted with Tissue Tek (Sakura Finetek Inc., Torrance, CA, USA), and cryosections were incubated with the first antibody (collagen-1, 1:100, Sigma-Aldrich, Munich, Germany) followed by treatment with a biotinylated second antibody. The last incubation was carried out with fluoroisothiocyanate-linked streptavidin-cy2 (Rockland Immunochemicals Inc., Gilbertsville, PA, USA). The nuclei were counterstained with Draq5™ (Alexis Corporation, Lausen, Switzerland).

The quantification of fluorescence intensity was performed with a confocal laser microscope (TCS SP, Leica Camera, Solms, Germany) equipped with appropriate filter blocks using a Silicon Graphics Octane workstation (Silicon Graphics, Sunnyvale, CA, USA) and three-dimensional multichannel image processing software (Bitplane, Zurich, Switzerland). A full range of gray values from black to peak white (0- to 255-pixel intensity level) was set during measurements. The intensity of fluorescence was expressed as arbitrary units (AU/μm2). Fibrosis was quantified from collagen-1-stained representative sections of the biopsy using Image J Software version 1.35. Seven fields of vision ×40 were randomly chosen. Contents were expressed as percentage of myocardium.

Control tissue

The control tissue was obtained from three donated hearts for which suitable recipients were not found at the time of surgery. The ages of the donors were 26, 31, and 39. Two donors were female. The donors had not had chronic illness, but they died from traumatic car accidents. Histological analysis of these hearts showed normal myocardial morphology.


Results are expressed as median and IQR. Because of the limited sample size, non-parametric Mann–Whitney U tests were used to test for statistically significant differences between the study groups. Kruskal–Wallis tests were used, if more than two groups were examined. Serial biopsy specimens (acute to recovery) were compared using Wilcoxon's signed rank test for paired variables. Differences in categorial variables have been tested using χ2 test or Fisher’s exact test (where appropriate). The statistical analysis was performed with SPSS 16 for MAC (SPSS, Chicago, IL, USA). A two-sided P-value of <0.05 was considered statistically significant. To control for an inflation of the type I error, we made Bonferoni adjustations.


Patients characteristics

The study enrolled 15 patients who were 48–76 years old (63, IQR 57–72). The LGE negative and LGE positive subgroups did not differ significantly in age. Prior cardiovascular history was uneventful, with no coronary artery disease, chest pain, myocardial infarction, valvular heart disease, or heart failure (Table 1). Each patient experienced a stressful incident on the day of admission. All patients reported sudden onset of chest pain mimicking an acute coronary syndrome. Symptoms rapidly resolved within several hours after admission. The electrocardiogram revealed a significant ST-elevation initially in the anterior leads changing to T-wave inversion on Day 1 after admission in all patients. No statistically significant differences between the groups were found for any of the baseline variables (compare Table 1).

View this table:
Table 1

Baseline characteristics

LGE positive (n = 5)LGE negative (n = 10)P-value
Age (year) [IQR]57 [52–59]65.5 [57.5–72]0.99
Sex (male) [n, (%)]1 (20%)1 (10%)0.56
ST-elevation/T inversion [n, (%)]5 (100%)10 (100%)0.4
Preceding stressful event [n, (%)]5 (100%)10 (100%)0.4
Time onset of symptoms to MRI (h) [IQR]19.7 [17.2–20.5]23.1 [18.9–27.9]0.31
Time onset of symptoms to biopsy (h) [IQR]5.5 [5.2–16.5]20.5 [11.3–43.3]0.13
Time MRI to biopsy (h) (positive values indicate time after biopsy) [IQR]12.1 [4.0–15.3]1.0 [−4.1 to 7.6]0.37
  • LGE, late gadolinium enhancement; IQR, inter-quartile range, MRI, magnetic resonance imaging.

Coronary angiography and ventriculography

No patient had spontaneous vasospasm upon admission, according to coronary angiography, and all patients had either no coronary artery disease or only a diffuse coronary sclerosis without obstructive stenoses (>50%) (Figure 1A and B). Left ventriculography showed akinesia in the anterolateral, apical, diaphragmatic, and septal areas as well as a hypercontractile base (Figure 1C and D).

Figure 1

Coronary angiography (A: left coronary artery; B: right coronary artery) excluded obstructive coronary artery disease. In left ventriculography, the typical contractile pattern of tako-tsubo cardiomyopathy was detected (C: diastole; D: systole).

Magnetic resonance imaging

Magnetic resonance imaging was performed between 9.5 and 48 h after the onset of symptoms (20.5, IQR 18.1–26.7 h) and between 24 h prior to 36 h after biopsy [4.0, IQR −4.0 to 13.7 h (positive values indicate a time after biopsy)]. In the acute phase of tako-tsubo cardiomyopathy, all patients showed hypokinesia or akinesia of the apical and/or mid-ventricular segments with moderate-to-severe reduction of EF (44.7, IQR 36.2–55.5%).

Pathological signal activity was detected in the LGE sequences of five patients. Compared with myocardial infarction or myocarditis, the signal intensity was much lower. Late gadolinium enhancement was best visualized on four-chamber and short-axis slices. It appeared as multiple diffuse patches that were intramuraly spread over the apical portion of the septal and lateral wall in a cougar-like pattern (Figure 2). Cross-sections of these patches on short-axis slices were similar to a mid-wall sign. Generally, the mass of LGE was comparatively small (relative size 4.1%, IQR 2.35–5.85).

Figure 2

(A) Diffuse, patchy late enhancement distributed over the septal and lateral wall as indicated by arrows [left: original picture; right: semi-automatic enhancement detection (CAAS)]. (B) Septal enhancement from another patient as indicated by arrows [left: original picture; right: semi-automatic enhancement detection (CAAS)]. (C) Right oedema within apical region (arrow) [left: apical akinesia and ballooning (end-systolic frame)].

The EF was not significantly different for the LGE positive and negative groups (LGE positive: 50.1, IQR 45.3–59.2%, LGE negative: 41.2, IQR 35.9–48.4%, P = 0.13). The number of segments (16-segment model of the American Heart Association) showing impaired regional contractility ranged from 4 to 10 (7.5, IQR 5.5–9.5). The mean wall motion score was not significantly different between groups (LGE positive: 1.62, IQR 1.5–2.2, LGE negative: 1.65, IQR 1.6–2.1; P = 0.91). The EDV (LGE positive: 70.7, IQR 62.2–74.2 mL, LGE negative: 78, IQR 70.3–96.2 mL; P = 0.25) and ESV (LGE positive: 30, IQR 24.2–32.9 mL, LGE negative: 41.3, IQR 35.6–55.7 mL; P = 0.05) were within the normal range in all patients. Only ESVs were significantly smaller among LGE positive patients.

Oedema was observed in all patients on the T2-weighted TSE sequences. It was confined to the dysfunctional segments and showed no differences between the two groups (LGE positive: 23.1, IQR 20.9–25.6% area, LGE negative: 19.9, IQR 18.6–21.2% area; P = 0.27) (Table 2).

At follow-up (14, IQR 11–14.5 days), all functional parameters had returned to normal in all patients. Correspondingly, the LGE was completely reversed at follow-up in all patients.

Laboratory findings

The serum levels of myocardial CK and troponin were mildly elevated for all patients. There were no significant differences between patients with and without LGE (CK–LGE positive: 206, IQR 143.5–273.5 U/L; LGE negative: 203, IQR 143–259.8 U/L; P = 0.81. Troponin–LGE positive: 0.16, IQR 0.09–0.22 ng/mL; LGE negative: 0.15, IQR 0.02–0.5 ng/mL P = 0.57) (Table 2).

Electron microscopy

Cell swelling associated with a damage of the basal lamina, electron-lucent nuclei clumped with chromatin, or damaged mitochondria with flocculent densities as typical signs of oncotic cell death19 were absent in all specimens irrespective of the presence of LGE.


Myocarditis was excluded by means of immunhistochemistry and viral genome polymerase chain reaction (PCR).

Immunhistochemical analysis of collagen-1 in the EMB in the acute phase of tako-tsubo cardiomyopathy showed a significant increase in comparison to control tissue. The mean amount of collagen-1 was significantly more increased in patients with LGE than in patients without pathological signal activity (LGE positive: 18.84, IQR 13.82–19.75 AU/μm2; LGE negative: 7.57, IQR 5.41–9.19 AU/μm2; controls: 2.1, IQR 2.0–2.2 AU/μm2, overall test P = 0.03; post hoc test LGE positive vs. LGE negative P = 0.001, LGE positive vs. controls P = 0.04) compare Figure 3AC. However, this increase in extracellular collagen content was not evenly spread over the dysfunctional area, instead we found varying degrees of fibrosis in the six biopsies that were taken from the LV with significantly higher values in those patients with pathological signal activity.

Figure 3

Immunolabelling for collagen-1, specific labelling is green; F-actin (red) was visualized with TRITC-conjugated phalloidin, and nuclei (blue) were counterstained with Draq5™. Endomyocardial biopsies from patients with tako-tsubo cardiomyopathy showed a significant increase of collagen-1 in comparison to control myocardium (A). In late gadolinium enhancement negative patients (B), a significantly lower amount of collagen-1 was detected when compared with late gadolinium enhancement positive patients (C).

Follow-up biopsies were performed after functional recovery had been documented by serial MRI and or echocardiography (14, IQR 11–14.5 days). After functional recovery, the collagen-1 content decreased significantly and returned to normal values in both groups (LGE positive: 2.4, IQR 2.3–2.5; LGE negative: 2.2, IQR 2.0–2.5; controls: 2.2, IQR 2.1–2.2, overall test P = 0.28; post hoc tests: LGE positive vs. LGE negative P = 0.7, LGE positive vs. controls P = 0.1) (Table 2).

Moreover, we found the typical histological pattern described above, including contraction bands as a sign of catecholamine toxicity, in all patients (data not shown).


Although tako-tsubo cardiomyopathy has been gaining more and more attention in the cardiology community, the incidence of this new cardiac syndrome remains underestimated. Tako-tsubo cardiomyopathy is often misdiagnosed, due to its resemblance to an acute coronary syndrome. Cardiac MRI is emerging as the tool of choice for differentiating tako-tsubo cardiomyopathy, embolic infarction, and myocarditis, when significant coronary artery disease has been ruled out in the presence of typical wall motion anomalies (mid-ventricular or apical ballooning). The presence or absence of LGE is crucial to the diagnosis of tako-tsubo cardiomyopathy, as most reports support the position that LGE rules out tako-tsubo cardiomyopathy.13 In general, the presence of LGE is considered as indicative of myocarditis or embolic infarction, depending on the mural distribution of enhancement.23,24

In this study, we were able to demonstrate that LGE can be present in tako-tsubo cardiomyopathy. The signal intensity found was lower than that usually documented in cases of myocardial infarction or myocarditis. Also the extent of LGE as represented by the relative mass of enhancing tissue is less than usually reported in studies of myocardial infarction.2527 Both facts are not surprising as the underlying tissue alterations are different and changes to LV function are reversible in tako-tsubo cardiomyopathy.

A frequent cause of LGE is necrosis with disruption of sarcolemmal integrity and intracellular accumulation of gadolinium. Cardiac biomarkers CK and troponin were not different from patients without LGE, we therefore concluded that LGE cannot be attributed to differences in sarcolemmal integrity. This finding is supported by the fact that tissue samples were void of signs of oncotic cell death under electron microscopy in both groups.

Neither did we find differences in the extent of oedema between groups represented by the per cent area of high signal intensity on T2w measurements.

In brief, we found no evidence that LGE among tako-tsubo patients is caused by either necrosis or oedema.

Instead we found a significantly higher increase of ECM as represented by collagen-1 staining in the group of LGE positive patients. As this is the third tissue alteration that can cause LGE and as necrosis and oedema have been ruled out, it is most plausible to assume that LGE in tako-tsubo cardiomyopathy is due to transient fibrosis. To understand the effect on image contrast, it is important to mention that the increased extracellular collagen content was not evenly spread over the dysfunctional area. Instead we found varying degrees of fibrosis in the six biopsies taken from each patient's LV. We must, therefore, assume that we are looking at islands of markedly increased fibrosis that stand out against tissue with moderately increased fibrosis. This explains the patchy character of LGE that we found. This is another aspect that distinguishes tako-tsubo cardiomyopathy from acute myocardial infarction, where necrosis or replacement fibrosis is uniformly present over a certain vessel-dependent territory.

In addition, myocarditis was ruled out by immunohistological examination of EMB and virus PCR.

These findings are interesting in several respects. First of all, our results provide evidence that the presence of LGE cannot rule out tako-tsubo cardiomyopathy (Table 2). The histological background of LGE has been well documented in the case of ischaemic heart disease,20 but less is known about the mechanisms that cause LGE in non-ischaemic heart disease. Our results agree well with the findings from Moon et al. correlating cardiac fibrosis with the degree of LGE on cardiac MRI in patients with hypertrophic cardiomyopathy. They compared pathological sections of a human heart after transplantation with LGE images that had been acquired before the patient was transplanted. Interestingly, they found that a myocardial segment was more likely to be rated LGE positive, if the ECM exceeded 15% of the area per field of view. These findings agree well with our results, which showed median AU of LGE of 18.84, IQR 13.82–19.75 AU/μm2 for LGE positive patients and 7.57, IQR 5.41–9.19 AU/μm2 for LGE negative patients.

View this table:
Table 2


LGE positive (n = 5) [IQR]LGE negative (n = 10) [IQR]P-value
EF (%)50.1 [45.3–59.2]41.2 [35.9–48.4]0.13
EDV (mL) (normalized to BSA)70.8 [62.2–74.2]78 [70.3–96.2]0.91
ESV (mL) (normalized to BSA)30 [24.2–32.9]41.3 [35.6–55.7]0.05
Wall motion score1.6 [1.5–2.2]1.65 [1.6–2.1]0.91
Creatinine kinase (U/L)206 [143.5–273.5]203 [143–259.8]0.81
Troponin (ng/mL)0.16 [0.09–0.22]0.15 [0.02–0.5]0.57
Oedema (% area)23.1 [20.9–25.6]19.9 [18.6–21.2]0.27
AU (% collagen-1)18.84 [13.82–19.75]7.57 [5.41–9.19]0.0001
  • EF, ejection fraction; EDV, end-diastolic volume; ESV, end-systolic volume; BSA, body surface area.

Likewise, a linear relationship of regional contractility and the degree of fibrosis has been reported.28 Our data, however, do not show a significant difference of contractility between the two groups.

Elsasser et al.29 studied the effect of regional fibrosis on regional contractility in patients with hibernating myocardium before and after bypass surgery. They found that the degree of fibrosis not only affected regional contractility but was also predictive of functional recovery after successful revascularization. A cut-off value of 32% fibrosis (fibronectin staining) per field of view was observed to preclude improvement of regional contractility after revascularization. This is in good agreement with our data as none of the LGE positive patients had a higher degree of fibrosis than the reported cut-off found by Elsasser et al. Correspondingly, all patients had completely recovered at follow-up, despite the disproportionate increase in ECM.

The presence of LGE is associated with a poorer prognosis in both ischaemic and non-ischaemic cardiomyopathies.15,30 In contrast, all LGE positive patients reported upon here were completely recovered at follow-up, and no lasting effect on ventricular function was observed. Different types of fibrosis have been described, ranging from interfiber fibrosis and perivascular fibrosis to plexiform fibrosis and replacement fibrosis.31 In the EMB of tako-tsubo cardiomyopathy, we found a transient increase of interfiber ECM, in the sense of a reactive fibrosis rather than a replacement fibrosis. This hypothesis is supported by ex vivo data from norepinephrine-stimulated rat fibroblast cultures that showed increased collagen content as early as 24 h after stimulation32 and data from Iwaki et al.33 who found up-regulation of early immediate genes as early as 2 h after norepinephrine stimulation. Endomyocardial biopspy after functional recovery showed a normal collagen-1 amount.

This supports our hypothesis that LGE in tako-tsubo cardiomyopathy is caused by an increase in ECM, it also explains, as this increase in ECM is transient, why the LGE phenomenon in tako-tsubo cardiomyopathy patients is not associated with long-term reduction of LV function.

Extrapolating from these findings to a more general approach to LGE, it is possible that any underlying disorder that is capable of causing reactive fibrosis might lead to transient LGE in other entities too.


This study is limited by its small sample size, so further reports are needed to confirm our findings. Also it is observational and not experimental in design. With that caveat in mind, the present study suggests that the presence of LGE cannot rule out tako-tsubo cardiomyopathy. Instead it appears to define a special subgroup of patients with a disproportionate increase of ECM. Tako-tsubo cardiomyopathy should always be considered if other criteria (such as typical wall motion pattern, oedema, or right ventricular involvement) are met.

Conflict of interest: none declared.


  • Both authors contributed equally to this work.


View Abstract