Article Text

Download PDFPDF

Original article
Slowly resolving global myocardial inflammation/oedema in Tako-Tsubo cardiomyopathy: evidence from T2-weighted cardiac MRI
  1. Christopher Neil1,2,
  2. Thanh Ha Nguyen1,2,
  3. Angela Kucia3,4,
  4. Benjamin Crouch1,
  5. Aaron Sverdlov1,2,
  6. Yuliy Chirkov1,2,
  7. Gnanadevan Mahadavan1,2,
  8. Joseph Selvanayagam5,
  9. Dana Dawson6,
  10. John Beltrame1,2,
  11. Christopher Zeitz1,2,
  12. Steven Unger1,2,
  13. Thomas Redpath6,
  14. Michael Frenneaux6,
  15. John Horowitz1,2
  1. 1The Queen Elizabeth Hospital, Adelaide, Australia
  2. 2The University of Adelaide, Adelaide, Australia
  3. 3The Lyell McEwin Hospital, Adelaide, Australia
  4. 4The University of South Australia, Adelaide, Australia
  5. 5Flinders Medical Centre, Adelaide, Australia
  6. 6The University of Aberdeen, Aberdeen, UK
  1. Correspondence to Professor John Horowitz, Cardiology Unit, The Queen Elizabeth Hospital, 28 Woodville Road, Woodville South, SA 5011, Australia; john.horowitz{at}


Objective Tako-Tsubo cardiomyopathy (TTC) is associated with regional left ventricular dysfunction, independent of the presence of fixed coronary artery disease. Previous studies have used T2-weighted cardiac MRI to demonstrate the presence of periapical oedema. The authors sought to determine the distribution, resolution and correlates of oedema in TTC.

Patients 32 patients with TTC were evaluated at a median of 2 days after presentation, along with 10 age-matched female controls. Extent of oedema was quantified both regionally and globally; scanning was repeated in patients with TTC after 3 months. Correlations were sought between oedema and the extent of hypokinesis, catecholamine release, release of N-terminal prohormone of B-type natriuretic peptide (NT-proBNP), and markers of systemic inflammatory activation (high-sensitivity C-reactive protein and platelet response to nitric oxide).

Results In the acute phase of TTC, T2-weighted signal intensity was greater at the apex than at the base (p<0.0001) but was nevertheless significantly elevated at the base (p<0.0001), relative to control values. Over 3 months, T2-weighted signal decreased substantially, but remained abnormally elevated (p<0.02). The regional extent of oedema correlated inversely with radial myocardial strain (except at the apex). There were also direct correlations between global T2-weighted signal and (1) plasma normetanephrine (r=0.39, p=0.04) and (2) peak NT-proBNP (r=0.39, p=0.03), but not with systemic inflammatory markers.

Conclusions TTC is associated with slowly resolving global myocardial oedema, the acute extent of which correlates with regional contractile disturbance and acute release of both catecholamines and NT-proBNP.

  • Tako-Tsubo cardiomyopathy
  • oedema
  • cardiovascular MRI
  • N-terminal proBNP
  • allied specialties
  • psychology/psychiatry
  • depression
  • myocardial disease
  • myocardial ischaemia and infarction (IHD)
  • imaging and diagnostics
  • heart failure
  • endothelium
  • basic science
  • free radicals
  • oxidative stress
  • cardiac function
  • tissue doppler
  • cardiomyoplasty
  • cardiac ultrasound
  • coronary angiography
  • exercise echocardiography
  • cardiac remodelling
  • CT scanning
  • echocardiography-exercise
  • syndrome x
  • myocardial viability
  • coronary artery disease
  • gender
  • coronary artery disease
  • spasm
  • coronary physiology
  • coronary flow
  • angina treatment
  • microvascular
  • MRI
  • heart failure treatment
  • hypertrophic cardiomyopathy
  • endothelial function
  • nitrites
  • oxidative stress
  • heart failure treatment
  • cardiomyopathy hypertrophic
View Full Text

Statistics from


Tako-Tsubo cardiomyopathy (TTC) is characterised by the presence of transient segmental, usually periapical, systolic left ventricular dysfunction. As is implicit in the rapid recovery of global systolic function in the majority of cases, TTC is associated with minimal myocardial necrosis, with modest elevation of cardiac biomarker levels1 ,2 and generally with the absence of late gadolinium enhancement on cardiac MRI (CMR).3 ,4 Once regarded as a rare form of acute left ventricular systolic dysfunction, TTC is now considered a relatively common occurrence, especially in postmenopausal women.1 ,2 There is general agreement that near-complete recovery occurs within 1–2 weeks.1 ,2

There is no current consensus as to the pathogenesis of TTC. While there is considerable evidence that catecholamines play a triggering role,5–7 it remains uncertain why this disorder affects mainly postmenopausal women and why the left ventricular dysfunction is segmental rather than global.

Inflammatory activation, both systemic and intramyocardial, is well documented in TTC. Most studies have revealed elevation of plasma C-reactive protein concentrations,8–10 while intramyocardial inflammatory activation has been documented directly by myocardial biopsy studies11 as well as by technetium pyrophosphate imaging.12 Analogously, a number of investigators have documented the presence of intramyocardial oedema, presumably of inflammatory origin, by CMR.3 ,13–15

We now report an investigation of the regional distribution and intensity of intramyocardial oedema in a cohort of patients with TTC. We made the following hypotheses.

  1. The distribution of oedema during the acute phase of TTC would parallel that of regional hypokinesis and that oedema would resolve synchronously with recovery of left ventricular wall motion.

  2. The extent of oedema would correlate with plasma normetanephrine concentrations, as an index of individual catecholamine exposure.

  3. The extent of oedema, presumably a consequence of intramyocardial inflammation, would correlate with markers of both myocardial (N-terminal prohormone of B-type natriuretic peptide (NT-proBNP) release16–18) and systemic (C-reactive protein elevation and impairment of circulating platelet responsiveness to nitric oxide19 ,20) inflammatory activation.

Our results suggest that (a) TTC is associated with slowly resolving myocardial oedema, presumably of post-inflammatory origin, affecting the left ventricle globally, with a base-to-apex gradient, and (b) the acute extent of oedema correlates with that of regional contractile disturbance (hypothesis 1) and of release of both catecholamines (hypothesis 2) and NT-proBNP (hypothesis 3).


Patient selection

Patients with a diagnosis of acute TTC (onset within the preceding 48 h), inclusive of confirmatory diagnostic cardiac catheterisation, were eligible for study. Over a period from March 2009 to October 2011, 45 such patients were evaluated at two tertiary referral hospitals. The diagnosis of TTC was made if the following criteria were met: (1) presentation with abnormalities of ST segments/T waves, with or without chest pain or dyspnoea; (2) sustained elevation of cardiac troponin T levels; (3) demonstration of a characteristic periapical wall motion abnormality (or an accepted variant), not consistent with a single epicardial coronary territory; and (4) angiographic exclusion of obstructive coronary disease in the territory subserving the regional wall motion abnormality,1 ,21 as well as elimination of viral myocarditis by clinical and CMR criteria.22 Of these 45 patients, 13 were excluded from the MRI protocol because of claustrophobia (seven), prior device implantation (two), or inability to obtain informed consent (four). Thus, 32 patients participated in the investigation. Follow-up CMR was performed on 24 patients at a median of 105 (IQR 90–116) days.

Clinical protocol and systemic investigations

Patients were monitored for a minimum of 48 h, during which time intravenous sodium heparin was administered and routine blood chemistry tests were performed. C-reactive protein was assayed serially to determine the peak in each patient; in cases where C-reactive protein was <10 mg/l, the level of high-sensitivity C-reactive protein (Olympus au5400, Dallas, Texas, USA) was determined. Levels of NT-proBNP (Elecsys E 170, Roche Diagnostics, Mannheim Germany) were determined at presentation and serially thereafter to determine the peak for the acute episode. NT-proBNP concentration was re-evaluated 3 months after the index event.

Normetanephrine levels

Plasma normetanephrine, a non-acidic derivative of norepinephrine reflecting extraneuronal uptake and O-methylation, was selected as an index of acute catecholamine exposure in TTC, because of its relatively prolonged plasma elimination half-life.23 Within 24 h of diagnosis, venous blood was drawn into tubes containing potassium EDTA and centrifuged at 1680 relative centrifugal force, before storage at −80°C and subsequent assay by liquid chromatography/tandem mass spectrometry.24

Platelet responsiveness to nitric oxide

Platelet responsiveness to nitric oxide was of interest both as a physiological marker of systemic redox stress19 ,20 and in view of the reported thrombogenicity associated with TTC.25 For platelet aggregation studies, blood samples were drawn by venesection from an antecubital vein (0–24 h from diagnosis). Blood was collected in plastic tubes containing 1:10 volume of acid citrate anticoagulant (2 parts of 0.1 mol/l citric acid to three parts of 0.1 mol/l trisodium citrate); acidified citrate was used in order to minimise deterioration of platelet function during experiments.26 Aggregation in whole blood was examined using a dual-channel impedance aggregometer (model 560; Chrono-Log, Havertown, Pennsylvania, USA) as described previously.20 Aggregation was induced with ADP (final concentration 2.5 μmol/l). In order to determine platelet responsiveness to nitric oxide, inhibition of aggregation by the nitric oxide donor, sodium nitroprusside, was used.19 ,20

Control subject selection

Female age-matched controls were used to determine physiologically normal ranges of T2-weighted signal intensity (T2-w SI) in the absence of oedema, as described in the online appendix. They were sought by advertisement and were eligible in the absence of a contraindication to CMR or known cardiac or renal disease. The institutional human research ethics committee approved the study, and all patients and controls provided written informed consent.

Cardiac MRI

Imaging protocol

For acute, follow-up and control imaging, CMR was performed on 1.5 T Philips Intera and Achieva systems (Philips Medical Systems, Best, The Netherlands), with a five-channel phased-array coil and electrocardiographic gating. After initial scout images, cine CMR was performed in standard short- and long-axis cardiac views, using Balanced Turbo Field Echo sequences. All short-axis imaging (cine CMR, T2-weighted and late gadolinium enhancement) used the same slice thickness (8 mm) and interval space (no gap), to allow consistency of regional comparison between image sets, via a 17-segment model.27 A triple inversion recovery pre-pulse (double inversion dark blood suppression, followed by selective fat saturation (SPAIR)) was applied for imaging of myocardial oedema in short axis, in accordance with current guidelines of the assessment for myocardial oedema.22 ,28 Typical imaging parameters were: field of view, 360 mm; repetition time (TR), 2 × R-wave to R-wave interval (RR interval); echo time (TE), 80 ms; turbo factor, 35; matrix, 240/560; echo planar imaging factor (EPI) factor, 1; number of signal averages (NSA), 1; a fat inversion pulse time of 150 ms; and signal homogeneity correction (CLEAR). Gating was prospective and timed to mid-diastole. Late gadolinium enhancement studies were obtained with parallel imaging, using a field of view of 340 mm, TR of 3.6 ms, TE of 1.1 ms, and an inversion time manually adjusted to null the signal from viable myocardium, after bolus administration of 0.2 mmol/kg gadobutrol (Bayer Schering Pharma, Berlin, Germany). A panel of experienced CMR readers reviewed all imaging. The methodology for T2-weighted signal quantification is described in the online appendix.

Regional wall motion and radial strain

Offline analysis was undertaken using Philips ViewForum software (release Endocardial and epicardial borders were traced in short-axis cine CMR images, throughout the cardiac cycle, allowing calculation of left ventricular volumes, ejection fraction and wall thickness. Presystolic and peak systolic thicknesses were derived automatically; from these, peak radial strain was calculated for each left ventricular segment, as a percentage of deformation. Subjective wall motion scoring and the derivation of a wall motion score index were also undertaken.27

Statistical analysis

Differences between groups, in categorical and continuous variables, were assessed with Fisher's exact test or Student t test, respectively, as appropriate. The degree of relationship between acute global T2-w SI and various systemic markers was assessed by Pearson's r correlation, or Spearman's rank correlation in the case of non-Gaussian data. Differences in myocardial T2-w signal distribution in the acute and recovery (3 months) phases were examined using two-way analysis of variance with repeat measures. GraphPad Prism software (V.5.0) was used for all analyses, and statistical significance was inferred with a p value of <0.05.


Patient/control characteristics

All but one of the subjects with TTC were female, while, of the female subjects, one was premenopausal; patients' ages ranged from 41 to 83 years. Other demographics and comorbid conditions are summarised in table 1. In all but two of the patients with TTC, clear precipitant stressors were identified (63% psychological, 31% physical stress). Eleven patients presented with ST elevation on electrocardiography, one with left bundle branch block and the remaining 20 without ST elevation (all of whom exhibited T-wave inversion). Sixteen showed typical apical ballooning, whereas 15 were found to have a mid-ventricular variant29; one patient had findings consistent with the basal variant, or so called ‘inverted TTC’.30 In all cases routine serial echocardiography showed partial or total resolution of left ventricular regional hypokinesis within 10–14 days.

Table 1

Patient and control characteristics

Patients and control subjects did not vary significantly with regard to age, body mass index or clinical history (table 1). Four patients were diagnosed as having TTC in the presence of atherosclerotic coronary disease; however, in all cases the coronary disease location was not consonant with the site of wall motion anomaly.21 On CMR (table 2), all patients exhibited wall motion scores of ≥1.125, but global left ventricular ejection fraction was often normal; the latter was reduced by <45% in one-third of patients. A panel of experienced CMR readers reviewed all imaging, an example of which is provided as an online supplement. Focal mid-anterolateral late gadolinium enhancement was evident acutely in one patient, but was noted to be of low intensity and had resolved by the time of follow-up.

Table 2

Cardiac MRI (CMR) findings

CMR findings in TTC: T2-w SI quantification

Control values

Individual data from the 10 control subjects are shown in figure 1A. Across all slices, the mean±SD T2-w SI value was 0.47±0.04 arbitrary units. Data were homogeneous without evidence of any base-to-apex gradient. The radial intersegment coefficient of variability was 7.9%.

Figure 1

Whole-slice T2-weighted signal intensity (T2-w SI) data from normal controls (A) and patients with acute Tako-Tsubo cardiomyopathy (TTC) (B). Slices 1–7 extend from base to apex, here and throughout the paper. Mean (open circles) and SD are indicated only for the normal subjects, as the TTC data were skewed (see text). Colour-coded bull's-eye plots (C) compare uncorrected T2-w SI (median) in acute and recovery phases of TTC with that of controls. T2-w SI data were not collected from the extreme apex (see the Methods section).

Acute TTC

Individual uncorrected T2-w SI data for patients with TTC during acute presentation are shown in figure 1B. The data were not normally distributed; hence no mean values are shown. In all cases, the distribution of T2-w signal was neither confined to the territory of a single coronary artery nor concentrated focally (as described in viral myocarditis22). However, T2-w SI values exhibited a significant gradient from apex to base (eg, comparing slices 1 to 7, the mean gradient was 0.13±0.10 arbitrary units; p<0.0001). Furthermore, even at the base, T2-w SI values were significantly greater than for comparable control data (p<0.0001). In acute TTC, but not in control subjects, mean mid-anterior T2-w SI was ∼17% greater than the corresponding posterior wall signal (p<0.005). Given the high proportion of patients with the mid-ventricular variant, we compared apical/mid-zone T2-w SI for apical TTC (median ratio 1.06 (IQR 1.0–1.19)) and mid-ventricular TTC (median ratio 1.0 (0.93–1.04), p=0.06).

Three months after onset

At 3 months, oedema had decreased substantially, with loss of the apex-to-base gradient, but with overall incomplete resolution, as depicted in bull's-eye format in figure 1C. A comparison of mean corrected regional T2-w SI values at presentation and at 3 months is shown in figure 2A. There was a substantial decrease in T2-w SI, with abolition of the regional T2-w SI gradient (for difference in T2-w SI, F=470, p<0.0001; for regional heterogeneity of resolution, F=7.4, p<0.05 on two-way analysis of variance). Although there was abolition of a regional gradient of T2-w SI at 3 months, the data suggested incomplete resolution. Comparison of global T2-w SI for patients with TTC at 3 months with that of the control subjects confirmed that T2-w SI was abnormally elevated (p=0.02).

Figure 2

Regional variability in background-corrected T2-weighted signal intensity (T2-w SI) values at presentation (A) and the relationship between these T2-w SI values and regional radial strain (B). In (A), both acute (closed symbols) and 3-month (open symbols) values for T2-w SI are provided. Acute data were non-Gaussian and are therefore plotted as median and IQR, whereas data in recovery were normally distributed and are plotted as mean and SEM. T2-w SI was markedly different between time points (p<0.0001, two-way analysis of variance). In (B), strain versus T2-w SI correlations are shown for both mid-ventricular (r=−0.49, p<0.001; in blue) and apical (r=−0.39, p=0.04; in red) levels. a.u., arbitrary units.

Correlates of corrected T2-w SI values

Corrected regional T2-w SI versus wall motion (hypothesis 1)

The expected inverse relationship between mean peak radial strain and T2-w intensity was present at the mid-ventricular (r=−0.49, p<0.001) and apical apex (r=−0.39, p=0.04), but was not significant at the base (p=0.08). Figure 2B summarises these regional data for individual patients.

Corrected global T2-w SI versus plasma normetanephrine concentrations (hypothesis 2)

Plasma normetanephrine levels correlated directly (r=0.39, p=0.04) with global T2-w signal (figure 3A).

Figure 3

Acute phase correlations between global T2-weighted signal intensity (T2-w SI) and plasma concentrations of (A) normetanephrine (r=0.39, p=0.04) and (B) N-terminal prohormone of B-type natriuretic peptide (NT-proBNP) (r=0.39, p=0.03). a.u., arbitrary units.

Comparison of T2-w SI with markers of myocardial and systemic activation (hypothesis 3)

A number of investigators have shown that TTC is associated with acute release of NT-proBNP,31 ,32 as well as BNP.8–10 ,33 ,34 We recently demonstrated that this anomaly persists for at least 3 months after the onset of symptoms32 and now sought to evaluate potential correlations with the extent of oedema. A significant correlation with T2-w SI was present acutely (r=0.39, p=0.03; figure 3B), but not at 3 months follow-up.

Median peak C-reactive protein was 11 (6.3–55) mg/l. There was no correlation between C-reactive protein and global T2-w SI. Mean inhibition of platelet aggregation, with the nitric oxide donor sodium nitroprusside, was 43±26%. There was no correlation between platelet response to sodium nitroprusside and global T2-w SI.


This study has evaluated the correlates of initial distribution and subsequent evolution of intracardiac oedema in a group of 32 patients presenting with TTC. It is important to recognise that only 11 of the 32 patients studied had initial ST segment elevation and that only two required initial inotropic support. Therefore our study population reflects the full clinical range of TTC (in contrast with most earlier studies, in which the patient population has tended to have initial ST segment elevation, to be initially severely ill, and to have greater initial impairment of left ventricular systolic function).5 ,6

In this context, using T2-weighted CMR, we have demonstrated that global oedema is a component of the myocardial injury of acute TTC. The presence of oedema per se has been described in previous MR series.3 ,13–15 However, we report, for the first time, that TTC is a global process, with abnormal T2-weighted signal seen in the basal myocardium, which has tended to be regarded as being ‘spared’.35 ,36 As postulated, the distribution in the acute phase exhibited a base-to-apex gradient of intensity, thus broadly parallelling the distribution of regional hypokinesis (hypothesis 1; figure 1). In the mid-ventricular and apical regions, the extent of local oedema predicted the extent of wall motion deficit (figure 2B), while the relationship also approached significance at the base. Consistent with these findings, data from the study of Mansencal et al showed substantial impairment of systolic peak velocity, strain and strain rate in basal segments of patients with acute apical ballooning.37 In addition, a recent series documented three cases of reversible global left ventricular dysfunction, resembling TTC in terms of precipitation by stress, the expected electrocardiographic and biomarker response, and the absence of coronary disease or alternative causes.38 Our results also reveal that, in patients with the mid-ventricular variant of TTC, the apex is still involved. A trend towards higher apical/mid-ventricular T2-w SI was observed (p=0.06) in the apical, compared with the mid-ventricular, variant. Consistent with the initial presence of oedema, TTC was also associated with reduced LV mass during follow-up (table 2). On the other hand, the current data do not permit determination of the extent to which fluid accumulation, detected by T2-weighted imaging, is interstitial (vasogenic) as distinct from intracellular (cytogenic).28

The persistence of global oedema is perhaps surprising, given that TTC is usually portrayed as a rapidly reversible and benign process.2 However, in other pathological processes, such as myocardial infarction,39 persistent foci of abnormal T2-weighted signal have been described at 3 months. The documented presence of T cells and macrophages in the cardiac interstitium in biopsy samples from subjects with TTC as late as 3 months after diagnosis11 supports this and, in fact, led to our selection of 3 months as a follow-up time point.

A growing body of associative data (although not so in all studies8) links the injury of TTC with both endogenous5 ,6 and exogenous40 catecholamines. However, the precise mechanism by which myocardial damage occurs is not known. Possibilities include a direct interaction with cardiomyocyte function, energetics41 and integrity, as well as an indirect effect on the myocardium, secondary to coronary vasoconstriction and/or microvascular obstruction.42 Accurate determination of the magnitude of this putative catecholamine stimulus is inherently difficult, because of instability of concentrations; we therefore used plasma normetanephrine, in view of its longer plasma elimination half-life23 ,24 compared with norepinephrine. Notwithstanding a possible underestimation of the effect (eg, via sampling after peak catecholamine release), our data suggest that, in patients with TTC, a ‘dose–response’ relationship exists between greater norepinephrine exposure and greater acute myocardial injury (hypothesis 2; see figure 3A). This finding underscores the potential significance of neurogenic catecholamine release in the development of TTC, without clarifying the mechanism further, nor excluding the possibility that patients with TTC may constitute a ‘hyper-responsive’ population.43 Furthermore, we cannot exclude the possibility that a component of catecholamine release in TTC may represent a consequence, rather than a cause, of the disorder.

In order to explore the implications of intramyocardial inflammation and associated oedema, we examined global T2-w SI in relation to indicators of both myocardial and systemic inflammation (hypothesis 3). While the classical stimulus for the release of BNP/NT-proBNP is left ventricular wall stress (either systolic or diastolic44), remarkable elevation is also known to occur in response to cardiac inflammation16 ,17 and seems to be proportionally greater in the case of NT-proBNP18 versus BNP. Marked elevation of both BNP6 ,8–10 ,33 ,34 and NT-proBNP31 ,32 have previously been observed in TTC series. We showed a relationship between oedema and the peak NT-proBNP concentration during the acute phase of TTC, consistent with the hypothesis that the primary stimulus for NT-proBNP elevation may also be what promotes oedema—that is, inflammation. However, on the basis of data presented in this paper, a mechanical stimulus for natriuretic peptide synthesis and release cannot be excluded.

In contrast with the association between extent of oedema and that of NT-proBNP release, there was no significant correlation between T2-w SI and the two markers of systemic inflammatory activation studied. It appears that the majority of patients with TTC exhibit only minor systemic inflammatory activation; the occurrence of left ventricular mural thrombosis in TTC25 therefore probably reflects primarily local, rather than systemic, factors.

As CMR was performed only once during the acute presentation period, it is possible that, in some cases, scanning might have preceded the development of the maximal extent of intracardiac oedema. The utility of the spleen as a reference point is also subject to question, but if splenic inflammation were to occur in TTC, this would tend to decrease, rather than increase, the observed myocardial anomalies. In the present series, there was also a significant anterior-to-posterior T2-w SI difference in acute TTC. It is possible, in the light of previous reports,45 ,46 that this was partially modulated by coil sensitivity. On the other hand, the anterior wall of the human left ventricle is more densely innervated by sympathetic neurons than the inferoposterior wall.47 ,48 Therefore we cannot exclude the possibility that the difference in T2-w SI observed in acute TTC (but not controls) reflects the local impact on oedema generation of differential neurogenic catecholamine exposure, within the left ventricle. In either case, no marked effect on the conclusions of this study is likely. It may be pointed out that the use of a T2 mapping sequence49 would have allowed derivation of T2 times and thus direct quantitative appreciation of the oedema; this technique was not available at the time of this study. Nevertheless, this method may also be inherently unsuitable for current purposes, given that it yields consistently higher T2 times in periapical myocardium in normal subjects, when applied to short-axis studies. In calculating global T2-w SI by subtraction of patient versus control T2-w SI, we acknowledge that factors other than myocardial oedema may have differed between the two groups. However, it is likely that these effects are small in comparison with the effect due to increased free water content.

In conclusion, this study supports the understanding of TTC as a state of intramyocardial oedema secondary to a global left ventricular inflammatory response, early after the index event and persisting well beyond the resolution of segmental left ventricular contractile dysfunction. Furthermore, heterogeneity in the severity of oedema correlates with catecholamine release. These findings are of significance in our evolving understanding of TTC and present a partial basis for development of therapeutic intervention strategies for the disorder.


We wish to acknowledge the assistance of the medical, nursing and imaging staff of the hospitals involved.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Funding This study was supported in part by a research grant from the National Health and Medical Research Council of Australia. CJN was the recipient of an Australian Postgraduate Award scholarship and a CVL Cardiovascular and Lipid research grant. THN held a University of Adelaide Endeavour International Postgraduate Research Scholarship.

  • Competing interests None.

  • Ethics approval Ethics approval was provided by HREC, TQEH, Adelaide, Australia.

  • Provenance and peer review Not commissioned; internally peer reviewed.

  • Data sharing statement Imaging or clinical data could be shared, were a collaborative agreement in place with the requesting party.

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.