Article Text

Download PDFPDF

Acute stress-induced (takotsubo) cardiomyopathy
  1. Dana K Dawson
  1. Correspondence to Dr Dana K Dawson, School of Medicine and Dentistry, University of Aberdeen and Aberdeen Royal Infirmary, Aberdeen, UK; dana.dawson{at}


Acute stress-induced (takotsubo) cardiomyopathy has a dramatic clinical presentation, mimicking an acute myocardial infarction and is triggered by intense emotional or physical stress. In this paper, we review the current state of knowledge of the mechanistic physiology underlying the left ventricular ballooning. The pathophysiology of the recovery from this acute heart failure syndrome is presented. The short-term and long-term outlook puts this new syndrome on a different perspective compared with recently held views. Current knowledge on susceptibility and predisposition already define distinctive characteristics of patients with takotsubo compared with myocardial infarction. Gaps in knowledge and future directions of research are identified in order to best direct efforts for identifying specific therapies for this condition, in the acute setting, to mitigate postacute symptoms or to prevent recurrences, none of which exist.

  • takotsubo
  • acute stress induced cardiomyopathy
  • cardiomyopathy
  • broken heart syndrome

Statistics from

Request Permissions

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


Acute stress-induced (takotsubo) cardiomyopathy has a dramatic clinical presentation, mimicking acute myocardial infarction (MI).1 The hallmark of this increasingly recognised condition is the association with a major emotional or physical stress that appears to precipitate the onset. The former are primary takotsubo presentations, whereas the latter are secondary (complicating any other coexisting medical condition or its treatment). It is usually diagnosed when invasive cardiac catheterisation demonstrates unobstructed coronary arteries, and ventriculography shows characteristic left ventricle (LV) ballooning leading to various degrees of acute LV dysfunction.

Brief history and terminology

As the youngest diagnostic entity in cardiology, takotsubo has made a rapid transition from an initial curiosity proposed in Japan by Sato in 1990 to a reasonably frequent diagnosis in any cardiology department. Although referred to initially as ‘apical ballooning’, the term acute stress-induced cardiomyopathy is probably most accurately reflecting this neurocardiac condition. Its description by analogy with the peculiar LV shape resembling a Japanese octopus-fishing pot (figure 1) gave it the name of ‘takotsubo’, and it has also been widely presented in the media by the resonant layman term of ‘broken heart syndrome’.

Figure 1

Japanese octopus fishing pot, called a ‘takotsubo’ (artwork done by Dr David Northridge, Consultant Cardiologist, Edinburgh Royal Infirmary).

Incidence, acute presentation pathways and mechanistic pathophysiology

As the typical symptoms of takotsubo are sudden onset of chest pain, breathlessness or collapse, these patients have an initial belief that they experience an MI—so do most ambulance crews and front-door physicians as the similarities of symptoms, presenting ECG, and biomarkers between the two conditions continue to raise difficulties in distinguishing them on presentation. In most cases the diagnosis is established at coronary catheterisation that shows unobstructed coronary arteries and one of the typical patterns of LV ballooning. However, delays in imaging investigations or concurrent presence of coronary artery disease introduce a degree of uncertainty in establishing a clear diagnosis. A diagnostic algorithm for these clinical scenarios has been summed up by the European task force on takotsubo in a recent position paper.2 In this decade of growing awareness among clinicians and patients alike, our own (unpublished) data as well as other centres’3 demonstrate that ~7% of all patients with presumed MI are in fact takotsubo.

ECG insights

Given the acute nature of takotsubo, it has been tempting to speculate on ECG changes that may add in the specificity of the diagnosis. However, there is a considerable variation in the presenting ECG, similar to that seen in acute MI. Patients with takotsubo can present with a normal ECG (11%), ST/T wave changes, (39%), ST-elevation (39%), transient left bundle branch block (4%) or arrythmias (atrial tachycardias, heart block and ventricular arrhythmias) (7%). The head-to-head comparison between the 12-lead ECGs of the ST-elevation-presenting patient groups suggests a larger spread of ST-elevation, which is non-localising in acute takotsubo, whereas the amplitude of ST-elevation seems overall less than that of a typical MI ECG.4 No acute ECG findings alone or in combination are specific enough to obviate or delay urgent cardiac catheterisation. The most characteristic takotsubo ECG feature remains the prolongation of QT/QTc interval, seen at presentation and peaking 24–48 hours thereafter; however, larger comparative studies will need to inform if this is a reliable differentiating feature from MI after correcting for gender differences in QTc since most takotsubo patients are women and the reverse is the case for MI. QTc prolongation is one of the factors considered to increase the risk of early arrhythmic complications without a direct correlation between the magnitude of QTc and the arrhythmic risk. Takotsubo recovery is characterised by significant repolarisation abnormalities (deep T-waves), which are protracted, sometimes repeaking later5 and which suggest that the myocardial healing process in this condition is different from that of MI.

Vascular, myocardial and humoural compartments

The coronaries’ patency together with the typical myocardial ballooning seen at invasive angiography are the most poignant objective findings. More recently, it has become accepted that bystander findings of coronary artery disease do not preclude takotsubo diagnosis2 as optical coherence tomography studies conclusively showed a high prevalence of atherosclerotic plaques (some of which are thin cap fibroatheromas), but a complete absence of ruptured plaques or intracoronary thrombi.6 The dyskinetic wall motion, exemplified in figure 2 (online supplementary videos 1–-4), can involve the apex or midventricular cavity or both together; much less frequently the base of the LV, and characteristically dyskinesia occurs in a non-coronary distribution.7 A focal (segmental) subtype has also been proposed.8 Concomitant dyskinesia of the apical right ventricular myocardium has been reported in apical and midcavity LV subtypes, ranging in incidence from 15% to 50%, echocardiography typically reporting less right ventricular involvement than magnetic resonance studies.9–11 Cardiac biomarker release (troponin) is disproportionately low relative to the area of myocardial ballooning (up to ~25× higher than upper laboratory ranges). While this points towards a largely preserved myocardial viability, it most likely represents stretch and non-apoptotic myocardial injury, as it has been suggested from acute myocardial biopsies revealing myocyte vacuolation and alteration of cytoskeletal and contractile proteins.12 Contrasting to low troponin levels, high levels of brain natriuretic peptide and its N-terminal precursor were reported in subgroups of patients13 14; since this is not a universal finding, questions remain on whether they are representative of the severity of the increased intracardiac pressures, or part of a generalised inflammatory response, or an adaptive/protective haemodynamic mechanism. Other biomarkers have been explored (copeptin,15 matrix metalloproteinases,16 proinflammatory cytokines,17 microRNAs (16 and 26a)),18 but a distinctive biomarker pattern immediately applicable in clinical practice has not yet emerged.

Figure 2

End-diastolic (ED, top row) and end-systolic (ES, bottom row) frames of LV angiograms demonstrating the four types of LV ballooning seen in takotsubo cardiomyopathy (arrows). Click on each image for cine loops (online supplementary videos 1–4). LV, left ventricle.

Cardiac Imaging

In most cases, further imaging is performed (figures 3–5, online supplementary videos 5–20). Echocardiography is particularly useful in identifying systolic anterior motion of the mitral valve leaflets in those with a hyperkinetic basal LV, or acute mitral regurgitation, or determining the estimated pulmonary artery pressure, all of which have been related to in-hospital mortaliy.19 Cardiac magnetic resonance (CMR) has consistently demonstrated intense myocardial oedema seen in both left20–23 and right11 ventricular myocardium and a relative lack of macroscopic fibrosis.24 It is now clear that the initial reports of late gadolinium enhancement represented a technical inability to null the myocardium due to presence of intense oedema. This oedema is so marked that it results in a measurable, characteristic increase in LV mass in the acute phase.22 24 The acute increase in LV mass and lack of late gadolinium enhancement add diagnostic confidence, since takotsubo remains a diagnosis of exclusion.

Figure 3

Cardiac magnetic resonance (CMR) (A, B, E and F) and echocardiography (C, D, G and H) of end-diastolic and end-systolic four-chamber view (top row) and three-chamber view (middle row) of an acute mid-to-apical ballooning. Click on all end-diastolic images for online cine loops: A and E correspond to online supplementary videos 5 and 6. C and G correspond to online supplementary videos 7 and 8. Bottom row: (I–J) corresponding four-chamber and three-chamber late gadolinium images demonstrating no macroscopic fibrosis. Spectrally adiabatic inversion recovery with fat saturation (T2W-SPAIR) imaging shows high signal intensity (K) indicating myocardial oedema and native T1 mapping (L) shows colour-coded high T1 values also indicative of oedema and able to quantify its extent (scale bar). All CMR imaging performed at 3T.

Figure 4

CMR (A–C) and echocardiography (D–F) images of midcavity ballooning. The left-hand side columns represent end-diastolic frames; the right side columns represent end-systolic frames of 2, 4 and 3 chamber views. Click on end-diastolic images for cine loops: A–C correspond to online supplementary videos 9–11 and D–F correspond to online supplementary videos 12–14. CMR, cardiac magnetic resonance.

Figure 5

CMR (A–C) and echocardiography (D–F) images of basal ballooning. The left-hand side columns represent end-diastolic frames; the right side columns represent end-systolic frames of two-chamber, four-chamber and three-chamber views. Click on end-diastolic images for cine loops: A–C correspond to online supplementary videos 15–17 and M–O correspond to online supplementary videos 18–20 .

Invasive haemodynamics

However, despite this distinctively different pathophysiology of the takotsubo heart, invasive heamodynamic studies have been unable to identify any differences in indices of systolic (preload recruitable stroke work and ventricular elastance) or diastolic (end-diastolic pressure/volume relationship) performance between takotsubo and acute MI.25 This is quite surprising given that analyses of large, representative cohorts of patients have shown that the LV ejection fraction at presentation is comparably lower in acute takotsubo than in acute MI.8

A different patho-physiology

Thus, there are some unprecedented pathophysiological dissociations in the clinical presentation of patients with takotsubo: first, takotsubo appears to affect the entire ventricular myocardium in the sense that there is a striking degree of panmyocardial oedema, but despite this oedematous spread there is only localised—although severe—myocardial impairment; second, there appears to be a dissociation between the large extent of functional involvement and the less affected haemodynamic status. Third, there is a different resolution dynamic compared with classical ‘stunned’/hibernating myocardium: in stunning/hibernation the ECG and wall motion abnormalities resolve in tandem once coronary blood flow is restored. In contrast, in takostubo, there is a protracted, continuously evolving abnormal ECG despite quick restoration of wall motion and relatively preserved coronary blood flow reserve. Further  data is needed to fully characterise the takotsubo coronary flow and microcirculation.

It is tempting to speculate that the largely preserved myocardial viability (although not necessarily accompanied by functional integrity) is likely to play a part in these differences. Few studies attempted to explain the mechanistic pathophysiology seen in acute takotsubo and herein the complexity lies in what findings are causative versus consequential. There remains intense debate on the trigger pathway and mechanism of myocardial injury. The most accepted theory to date is an exaggerated sympathetic stimulation resulting in a cardiotoxic discharge of circulating catecholamines (epinephrine, norepinephrine and dopamine).26 Not all investigators were able to replicate these findings,27 which if present do not clarify if catecholamines are truly causative or a simple epiphenomenon or perhaps a consequence of takotsubo physiology. Anecdotally, takotsubo patients do not demonstrate unusual tachycardia or high blood pressure at initial contact with emergency services. Whatever the stressor that impacts so dramatically on the heart, there are several studies that probed into mechanistic pathways by examining the acute and early post-recovery phase of the condition.

Early recovery pathways

A fundamental characteristic of takotsubo is the spontaneous recovery of the LV ejection fraction, which returns to normal or near normal in all patients over a variable period of time (days to weeks). This process of recovery has been documented both at myocyte and microvascular level.

The index of microvascular resistance (IMR; the distal coronary pressure multiplied by the mean transit time of a saline bolus during maximal coronary hyperaemia) reflects the status of the microcirculation and has been shown to be elevated during takotsubo presentation at comparable levels with those seen in acute MI (well above the normal cut-off of 25 U).28 A clear time course recovery was documented, with IMR linearly decreasing form ~60 U at presentation to 25 U after the first 4 days.29 It remains unclear if this microcirculatory dysfunction is a vascular phenomenon per se or simply mechanical extravascular obstruction due to intense myocardial oedema. Persistence of endothelial dysfunction at variable times after an acute episode was shown in a small number of patients.30 Concomittantly, the oedematous process recovers, although it remains detectable on CMR imaging 3–4 months after an acute episode, at least in the areas that were dyskinetic during acute presentation.21 22 Cell swelling and widening of interstitial spaces was documented on electron microscopy of human myocardial biopsies in the acute phase with attenuated findings on recovery, in keeping with the non-invasive imaging findings.12 Cardiac energetics (the phosphocreatine/gamma-adenosine triphosphate ratio (PCr/γATP) ratio obtained non-invasively by 31P-magnetic resonance spectroscopy) is severely reduced acutely and part-recovers by 4 months.22 Despite this energetics reduction, upregulation of cardiomyocyte survival pathways (phosphorylated PI3K/Akt) have been clearly documented from human biopsies,31 which is in keeping with the lack of detection of macroscopic fibrosis on CMR in the vast majority of cases. However, CMR imaging of the extracellular compartment shows that this remains expanded,32 and LV biopsies showing an increase in collagen-1 subfraction suggest that microscopic fibrosis may play a part in this expansion.33 Finally, non-invasive indices of systolic and diastolic performance (LV twist, untwist and global longitudinal strain) continue to remain abnormal during this early recovery phase32 34 despite normalisation of LV ejection fraction and volumes. Therefore, contrary to initial beliefs, there is a large body of evidence that points to an incomplete recovery 3–4 months after a takotsubo episode, despite normalisation of ejection fraction. It remains unknown at this stage whether this recovery is just more protracted than initially assumed or if it is exhausted and results a new clinical phenotype that remains to be defined. Nevertheless, the EF normalisation remains clinically useful in consolidating the diagnosis at follow-up.

Insights from experimental models

Due to the likely complexity of takotsubo and the lack of knowledge of the precise mechanistic pathophysiology, it is hard to replicate this condition in animal models. However, it is possible to generate a non-ischaemic, non-pathogenic myocardial stress status, which is probably the closest approximation to a takotsubo-like model. Important insights can be derived from such experimental set-up; however, the findings must be interpreted and translated with the required caution into clinical interventions. Table 1 summarises the model constructs, the most relevant data and how these can be beneficially used in the human disease.35–38

Table 1

Experimental models of takotsubo-like disease.

The aftermath of a takotsubo episode

Establishing a correct diagnosis is important since it has become clear that takotsubo is not harmless. There is a recognised early, in-patient mortality of 3%–5%, with the mode of death attributed to ventricular arrhythmias, intractable pump failure, cardiac rupture or thromboembolic stroke.7 There is also a recognised recurrence rate, of 10%–15%, where the trigger is typically different and the interval of time to a recurrence is unpredictable.39 However, the two studies that challenged the way we viewed the prognosis of takotsubo until recently were the data emergent from the international takostubo registry8 and the Swedish cardiac catheterisation registry,40 which independently showed that the subsequent long-term morbidity and mortality of takotsubo is comparable with that of MI. The long-term takotsubo mortality is attributed to both cardiac and non-cardiac causes with the former having a significant contribution.8 The localisation of acute LV ballooning (basal, mid-cavity or apex) does not seem to impact differently on outcomes.41 In contrast, there have been some clear identifiers of poor in-hospital prognosis: a low LV ejection fraction,8 19 raised pulmonary artery pressure,9 10 severe mitral regurgitation19 and right ventricular involvement in an inverse McConnell pattern.42

To date, there is no specific therapy known to help the short-term recovery and protect against the early in-hospital mortality. Having said that, standard pharmacological and/or mechanical/electrical support to assist cardiogenic shock/arrhythmias should be promptly initiated according to local guidelines, with two notable exceptions: catecholamines should probably be best avoided until further evidence becomes available and intra-aortic balloon pump should be avoided in those with severe LV outflow obstruction due to systolic anterior motion of the mitral valve, as it can inadvertently lower the cardiac output. Equally there is no medication that is able to prevent recurrences or to mitigate the symptoms that a proportion of these patients continue to experience34 after they recover from their acute illness. Retrospective registry data suggest that beta-blockers are not useful in preventing acute in-hospital mortality43 or recurrences. However, randomised clinical trials with prespecified end-points in takotsubo have not taken place and remain premature until a more clear mechanism of disease is elucidated in humans. In the absence of this available evidence, some centres, such as ours, chose to not prescribe any medication to these patients on the basis of primum non nocere principles.

Susceptibility and predisposition

The causative association with a type of emotion or stress has been widely accepted as the hallmark of this condition (mostly as a negative experience but in a minority a positive one,44 with a few remaining unable to recognise a trigger). Many triggers have been described, for example, bereavement, near-miss road traffic accident, arguments, conflict, divorce, public speaking, a severe fright, protracted stress and hardship or any physical illness such as cancer, administration of chemotherapy, infections, diarrhoea/vomiting, routine general anaesthesia, cardioversions and many others. There is also a clear gender predilection towards women (in proportion of 9:1—prompting speculation about a neuroendocrine predisposition or a significant under-diagnosis in men), and although takotsubo was initially thought to affect mostly postmenopausal women,3 it has become clear that younger patients are equally vulnerable. Data from single centre cohorts and registries suggest that takotsubo presenters are less burdened by the usual risk factors associated with coronary heart disease but in exchange, a self-declared history of mental health or neurological disorders appears to decorate the medical history of these patients.8 24 45 Since neither the acute onset nor the more persistent morbidity post-takotsubo cardiomyopathy can be immediately explained by any occult or intrinsic cardiac pathology that may be concurrent or predating the acute event, it is implicit to speculate that other predeterminants may be causally involved in subjects who develop this condition. Despite initial enthusiasm about a high surge of catecholamines,26 small isolated genetic studies failed to identify significant causative variants with a filtering process focused mainly on the adrenergic pathway.46–48 However, these studies are difficult to interpret as with a range of 28–95 patients studied, they were almost certainly underpowered to address this question. There remains therefore a need to describe the epidemiology of this condition in nationwide population studies in order to precisely define the characteristics of those susceptible. This would subsequently inform on the type of genetic studies needed to probe into a genetic predisposition, if any.

Research potential and future directions

There remain many unanswered questions in this complex syndrome. Perhaps the most urgent of all is to describe the  clinical physiology of the EF-recovered takotsubo patient in order to firmly establish what type of care these patients require long term, if at all.

The dynamic interplay of the microvascular abnormalities and the myocyte survival and recovery certainly deserves further attention. Small studies appear to suggest that microvascular perfusion remains functional in the malfunctioning myocardium at least to the degree where it does not jeopardise myocyte integrity both in the acute state and at 4 months follow-up. However, both reduced49 and increased50 glucose uptake has been reported in the ballooning segments, raising the possibility that the takotsubo myocyte may either have different degrees of insulin sensitivity or a swift ability to switch between substrates. If either or both should be the case, this may be exploited from a therapeutic standpoint. Furthermore, the unprecedented extent of myocardial oedema together with initial reports of cellular infiltrates12 questions whether inflammation could be either a cause or an effect in this pathology. Many intracellular and intercellular processes remain undefined and this type of information can be derived form a takotsubo-like experimental model. Both autonomic and central nervous system are likely to be involved. For example, studies suggest that 123I-metaIodoBenzyl-Guanidine (123I-mIBG), an analogue of norepinephrine, is reduced in takotsubo hearts shortly after presentation. However, norepinephrine is stored in presynaptic vesicles and released through an acetycholine-dependent mechanism located in the preganglionic neurons; it remains therefore unclear if the 123I-mIBG abnormalities are seen as a result of sympathetic overstimulation or vagal incompetence or both. The gender predilection requires further probing into cardiac–endocrine pathways to ascertain why this condition appears to afflict mostly women. Finally, large patient epidemiological studies are needed to define the physical and mental illnesses/personality profiles of this population and their families as this may provide important clues on patterns of genetic/environmental susceptibilities. Much remains to be uncovered before a mechanistic-guided therapeutic approach can be recommended.

Supplementary file 1

video 1

Supplementary file 2

video 2

Supplementary file 3

video 3

Supplementary file 4

video 4

Supplementary file 5

video 5

Supplementary file 6

video 6

Supplementary file 7

video 7

Supplementary file 8

video 8

Supplementary file 9

video 9

Supplementary file 10

video 10

Supplementary file 11

video 11

Supplementary file 12

video 12

Supplementary file 13

video 13

Supplementary file 14

video 14

Supplementary file 15

video 15

Supplementary file 16

video 16

Supplementary file 17

video 17

Supplementary file 18

video 18

Supplementary file 19

video 19

Supplementary file 20

video 20


The cardiovascular research unit team at the University of Aberdeen and the NHS Cardiology Department at Aberdeen Royal Infirmary.



  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.