BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on October 2, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from

Comparing the variants of takotsubo syndrome: an observational study of the electrocardiogram and structural changes.

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2018-025253

Article Type: Research

Date Submitted by the Author: 05-Jul-2018

Complete List of Authors: Watson, George; Christchurch Hospital, Cardiology Chan, Christina; Christchurch Hospital, Cardiology Belluscio, Laura; Christchurch Hospital, Biostatistics Doudney, Kit; Canterbury District Health Board, Molecular Pathology Lacey, Cameron; Christchurch Hospital, Psychological Medicine Kennedy, Martin; University of Otago, Department of Pathology Bridgman, Paul; Christchurch Hospital, Cardiology

Echocardiography < CARDIOLOGY, Heart failure < CARDIOLOGY, Keywords: Cardiomyopathy < CARDIOLOGY, Takotsubo

http://bmjopen.bmj.com/

on October 2, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 9 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 Comparing the variants of takotsubo syndrome: an observational study of the electrocardiogram 4 and structural changes. 5 a a b c d e a 6 Watson GM , Chan CW , Belluscio L , Doudney K , Lacey CJ , Kennedy MA , Bridgman PG * 7 8 9 10 a 11 Department of Cardiology, Christchurch Hospital, Christchurch, New Zealand 12 b Biostatistics, Christchurch Hospital, Christchurch, New Zealand 13 14 c Molecular Pathology Laboratory, Canterbury District Health Board, Christchurch, New Zealand 15 d 16 Department of PsychologicalFor peer Medicine, University review of Otago, Christchurch, only New Zealand 17 e Department of Pathology and Biomedical Science, University of Otago Christchurch, Christchurch, 18 New Zealand. 19 20 21 22 * Corresponding author: Dr Paul G Bridgman, Department of Cardiology, Christchurch Hospital, 23 Christchurch, New Zealand, [email protected] 24 25 26 The authors report no conflicts of interest related to this work. 27 28 29 30 31 G Watson is a medical student and was supported by the Heart Foundation of New Zealand through 32 33 a University of Otago Summer Studentship. http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 9

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 Objectives: 4 5 In takotsubo syndrome, QTc prolongation is a measure of risk of potentially fatal arrhythmia. It is not 6 known how this risk, or derangement of other markers, differs across the echo variants of takotsubo 7 syndrome. As the region of affected myocardium is usually larger, we speculated that patients with 8 the classic apical-ballooning form of takotsubo syndrome would have more severe derangement of 9 their markers. Therefore, we sought to explore whether apical takotsubo syndrome differs from the 10 variants of the syndrome in more ways than just regional wall motion pattern. 11 12 Design: 13 14 Observational study of patients gathered from a prospective database (2010-2018) and by 15 retrospective review (2006-2009). 16 For peer review only 17 Setting: 18 The sole tertiary hospital for a region in which case clusters of takotsubo syndrome were 19 20 precipitated by large earthquakes in 2010, 2011, and 2016. 21 Participants: 22 23 222 patients who met a modified version of the Mayo criteria for takotsubo syndrome were 24 included. All patients had digitally archived echocardiograms that were over-read by a second 25 echocardiologist blinded to the clinical report. 26 27 Primary outcome measures: 28 29 Ejection fraction, peak troponin, and QT interval. 30 31 Results: 32

Patients with the apical form of the syndrome were older (p=0.011), had a lower initial LVEF (35% vs. http://bmjopen.bmj.com/ 33 44%, p<0.0001), and a higher peak hsTnI (p=0.01) than those with variant forms. There was no 34 35 difference in the electrical abnormalities between the variants (QTc interval, HR, PR interval, QRS 36 duration, or T-wave axis). There was also no correlation between any of peak hsTnI, peak QTc, and 37 ejection fraction. QTc interval increased on day two and peaked on day three before falling steeply 38 (p<0.0001). 39 40 Conclusions: 41 on October 2, 2021 by guest. Protected copyright. 42 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 43 differ in their electrical abnormalities. There is a dissociation between the structural and electrical 44 abnormalities. QTc peaks on day 3 and then falls steeply. 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 9 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 Strengths and limitations of this study 4 5 - A key strength of this study is its size – it’s one of the largest studies to date in takotsubo 6 syndrome research. 7 - The dataset is remarkably complete. The lack of missing data allows for robust comparisons 8 9 and correlations and increases confidence in our conclusions. 10 - It is a weakness of our study that we did not include late imaging results or other modalities 11 such as MRI. 12 13 - Not all patients were followed up until QTc normalised – future studies should continue 14 serial recording of ECGs for a longer period to establish the later time course. 15 16 For peer review only 17 18 19 20 21

22 23 24 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 9

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 Introduction 4 5 There is widespread interest in the interaction between the mind and the body, and it is well 6 accepted that psychological factors influence physical illnesses. The purest example of psychological 7 stress causing a potentially fatal disease is takotsubo syndrome, also known as stress 8 cardiomyopathy. In this syndrome, the stress is typically acute and severe (and therefore readily 9 identifiable) and the symptoms can develop within moments. The incidence of takotsubo syndrome 10 is increasing, but it is agreed by experts that even in the most obvious of cases, we do not 11 understand how the psychological stress triggers the heart illness.1,2 Cardiologists have reported on 12 the wide range of psychological and physical triggers that can precipitate takotsubo syndrome and 13 14 recognise that there are variants within the syndrome. 15 The variants are defined by the specific pattern of wall-motion abnormality in the left ventricle. We 16 have noted how heterogeneousFor peer the clinical reviewpresentation of the conditiononly is and that different 17 18 subsets of patients can behave in different ways. Takotsubo syndrome has an early mortality rate 3 19 comparable to that of an acute coronary syndrome. Characterising the differing subsets will be 20 important in helping us move towards improved understanding of the condition and could lead to 21 better-tailored management of patients and perhaps insights into aetiology. Therefore, we sought to 22 explore whether the echocardiographic variants of the syndrome differ in more ways than just their 23 regional wall motion pattern. 24 25 In everyday clinical practice, clinicians use left ventricular ejection fraction (LVEF) as a measure of 26 risk and the degree of troponin elevation as an indicator of myocardial loss. Less widely recognised 27 in takotsubo syndrome is that QTc interval is a measure of risk of potentially fatal arrhythmia.4 It is 28 not known if this risk varies across the echo patterns of the condition or indeed if other clinical, 29 30 biochemical, and electrical features that cardiologists deal in vary between patterns either. As the 31 region of affected myocardium is usually larger, we speculated that patients with the classic apical- 32 ballooning form of takotsubo syndrome would have more severe derangement of their markers than 33 patients with one of the variant forms. Therefore, we compared patients with apical takotsubo to http://bmjopen.bmj.com/ 34 those with a variant. 35 36 37 38 Methods 39 Approval for the study was obtained from the Health and Disability Commission Ethics Committee. 40 on October 2, 2021 by guest. Protected copyright. 41 Christchurch Hospital is the only tertiary hospital in Canterbury. Unprecedented case clusters of 42 takotsubo syndrome were precipitated by large earthquakes in Canterbury in 2010, 2011, and 2016, 43 leading to the development of a prospectively-maintained registry of cases since 2010.5,6 In addition, 44 45 we retrospectively identified cases extending back to the installation of a digital echocardiography 46 archive in 2006. 47 All patients in the database meet a modified form of the Mayo criteria for takotsubo syndrome, 48 49 which allows inclusion of patients with one of the recognised characteristic patterns of wall motion 3 50 abnormality. Patients are not excluded if they have coronary artery disease that doesn’t explain 51 their wall motion abnormality, or if they die before resolution can be documented on follow up 52 echo. All echocardiograms are held in a digital archive and were over-read by a second 53 echocardiologist to confirm the wall motion pattern and the measurements. Discrepancies in the 54 reporting were settled by joint review. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 9 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 High-sensitivity troponin I (hsTnI) was used as a marker of myocardial damage. Prior to 2011, 4 Christchurch Hospital used a standard TnI assay – this was converted to hsTnI using the conversion 5 factor hsTnI = 880 * TnI + 1096, which was derived from a comparative study of the assays done at 6 the Canterbury Health Laboratory. 7 8 The electrocardiogram (ECG) machine performs automatic electronic measurement for heart rate 9 (HR), PR interval, QRS width, and QT interval. These were all checked by manual measurement. 10 Where the manual measurement showed that the electronic measure was incorrect the manual 11 measure was used. Borderline discrepancies were resolved by third-party measurement. QTc 12 interval was calculated using Bazett’s formula. The ECG machine also calculates T wave axis. For this, 13 14 manual measurement and calculation were not performed, and the electronic measure was always 15 used. LVEF, hsTnI, and ECG measurements were compared between the apical and variant (basal, 16 focal, mid-wall) forms,For and betweenpeer days over review a four-day period fromonly the time of admission. 17 18 Normality was assessed using Shapiro-wilk. Means of normally distributed variables were compared 19 using the two-sample t-test, while the medians of variables that do not follow a normal distribution 20 were compared by Wilcoxon rank-sum. Non-parametric analysis of rank correlation was done using 21 Kendall’s Tau. 22 23 24 25 Patient involvement 26 Patients were not involved in the design of this study. 27 28 29 30 Results 31 32 This study included 222 cases of takotsubo syndrome. 178 (80%) were apical, whilst 44 (20%) were 33 variant. http://bmjopen.bmj.com/ 34 Overall, the patients with the apical form of the syndrome were significantly older (68 vs 64 years, 35 36 p=0.011). However, unlike the variants, there appeared to be a bimodal age distribution in the apical 37 takotsubo patients. This is illustrated in Figure 1 and suggests that there may be a distinct younger 38 subset. We noted that these younger patients all had major psychological stressors, and in four 39 cases there was also a concomitant physical stressor. 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 9

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17

18 19 Figure 1. Comparison of age for patients with apical takotsubo vs. variant pattern. The horizontal line 20 in the middle of the box represents the median of each group, while the diamonds represent the 21 mean. The box upper and lower bounds represent the upper and lower quartiles respectively. The 22 whiskers represent the min and max values excluding those that are 1.5*IQR away from the median 23 24 (circles). * p=0.011 25 26 27 Patients with the apical form of the syndrome had a lower initial LVEF (35% vs. 44%, p<0.0001), and 28 a higher peak hsTnI (p=0.01) than those with variant forms, but there was no difference between the 29 forms in peak QTc interval (p=0.93) (Table 1). Apical and variant forms also did not differ significantly 30 in the other ECG measures. 31 32 33 http://bmjopen.bmj.com/ 34 Table 1. Comparison of LVEF, hsTnI, and ECG measures between apical and variant takotsubo. All 35 ECG variables are as at presentation. LVEF is mean and standard error, other data median and 36 interquartile range. 37 38 Apical takotsubo Variant takotsubo p-value 39 40 Initial LVEF (%) 36.67 (SEM 0.69) 43.52 (SEM 1.60) <0.0001 41 on October 2, 2021 by guest. Protected copyright. 42 Peak hsTnI (ng/l) 2308 (1244-3936) 1638 (289-2441) 0.01 43 Heart rate (bpm) 79 (70-94) 76 (66-95) 0.18 44 45 PR interval (ms) 166 (150-178) 165 (143-180) 0.49 46 47 QRS duration (ms) 88 (82-100) 88 (80-96) 0.90 48 QTc interval (ms) 465 (434-490) 459 (433-484) 0.93 49 50 T-wave axis 75 (54-106.5) 76 (52-98) 0.95 51 52 53 Heart rate, QTc interval, and T-wave axis all changed significantly with time (all p<0.0001), but not by 54 echo variant. QRS interval and PR interval did not differ by day or echo variant. There were no 55 56 associations between the structural and electrical measures (Table 2). While peak QTc against LVEF 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 9 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 was borderline significant (p=0.04), the Kendal Tau correlation was only -0.097. This is a very weak 4 correlation, and with marginal statistical significance is not meaningful, as demonstrated by the 5 scatter plot (Figure 2). We documented different time courses for the changes in QTc and LVEF, and 6 that QTc peaks on day 3 (Figure 3). 7 8 9 10 Table 2. Associations between LVEF, peak hsTnI, and peak QTc. 11 12 Correlation coefficient p-value 13 14 Peak hsTnI vs LVEF -0.091 0.0609 15 Peak hsTnI vs peak QTc -0.022 0.6498 16 For peer review only 17 Peak QTc vs LVEF -0.097 0.0457 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 Figure 2. Scatter plot showing the correlation between peak QTc interval and LVEF. 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 Figure 3. Change in hsTnI and QTc interval over time after admission. 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 9

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 Discussion 4 5 We have found that the differences between the variants of takotsubo syndrome go beyond the 6 echo features. Patients with the apical form are older and have a lower ejection fraction and higher 7 troponin. While there were differences in these two structural measures, QTc prolongation was not 8 different between the variants. The pathophysiology of the myocardial dysfunction and the QTc 9 prolongation in takotsubo syndrome are not fully understood. Our data clearly suggest that the 10 repolarisation changes are not dependent on the mass of myocardium involved and we speculate 11 the mechanisms must be separate from early in the pathophysiological process that follows the 12 triggering event. 13 14 We have also demonstrated a dissociation between the structural and the electrical features in both 15 severity and time course, including demonstrating that QTc peaks on day 3. A case series from 16 Auckland found QTcFor rises until peer day three but review did not look beyond only that point and so did not capture a 17 7 18 peak. A small series of 93 patients from two hospitals in Boston had previously reported a peak on 4 19 day 3. We have now confirmed that finding in a much larger cohort. The Boston group also found 20 that QTc correlates with the risk of ventricular arrhythmia. There is, therefore, a strong case for in- 21 hospital monitoring of patients whilst QTc is prolonged. It is important that patients are not 22 discharged after a normal angiogram if their troponin is falling, without attention to their 23 electrocardiogram and QTc interval. We have shown that typically QTc falls steeply between days 3 24 and 4. Daily ECGs should be performed as an inpatient until the QTc has normalised. 25 26 Our data highlight the dissociation of the structural and electrical features and a complex 27 relationship between the structural indices. Whilst the apical form typically presents with a higher 28 troponin level and a lower ejection fraction, there is no relationship between these two. This 29 30 suggests that they are reflecting different aspects of the condition. Typically, in apical takotsubo, a 31 very large area of myocardium is initially akinetic and while troponin does rise, the rise is relatively 32 small, reflecting that only a very low fraction of affected cells are dying. On standard 2D echo, the 33 vast majority of patients have a full-function recovery. However, a study of 52 patients from http://bmjopen.bmj.com/ 34 Scotland has found that on advanced echo and MRI studies it is possible to detect residual effects 35 and permanent damage.8 Our results are consistent with the concept that the degree of acute 36 troponin rise reflects acute cell death and that the number of cells affected in this way is 37 38 uncorrelated with the total area of myocardium affected. 39 Our finding that variant patients are on average younger is not new. It has previously been reported 40 3

in a large cohort by the International Takotsubo Registry. However, their report did not explore the on October 2, 2021 by guest. Protected copyright. 41 42 distribution of age with each variant. In our cohort there appeared to be a bimodal distribution 43 amongst the apical variant patients with the younger patients having the same syndrome 44 (comparable LVEF, hsTnI, and QTc prolongation) as the older patients, but usually with dual or very 45 severe precipitants. This suggests that there is a higher threshold for triggering the apical syndrome 46 in this group of younger patients. 47 48 It is a weakness of our study that we do not have follow-up MRI data. Future work should look to 49 correlate longer-term changes with the acute indices of syndrome severity. Our data show a steep 50 fall in QTc from day 3 to day 4, but further studies should also continue serial recording of ECGs for a 51 longer period to establish the later time course. 52 53 Conclusions 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 9 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 4 differ in their electrical abnormalities. There is a dissociation between the structural and electrical 5 abnormalities. QTc peaks on day 3 and then falls steeply. 6 7 8 9 Contributors 10 11 The authors contributed to the study as follows: design and conception (PB, MK, CL), acquisition of 12 data (GW, CC), data interpretation and manuscript preparation (all). 13 14 15 16 References For peer review only 17 18 1. Minhas AS, Hughey AB, Kolias TJ. Nationwide trends in reported incidence of takotsubo 19 cardiomyopathy from 2006 to 2012. Am J Cardiol 2015;116:1128-1131. 20 21 2. Otten A, Ottervanger J, Symersky T, Suryapranata H, de Boer M, Maas A. Diagnosis of takotsubo 22 cardiomyopathy is increasing over time in patients presenting as ST-elevation myocardial infarction. 23 Neth Heart J. 2016;24(9):520-529. 24 25 3. Templin C, Ghadri J, Diekmann J, Napp L, Bataiosu D, Jaguszewski M et al. Clinical features and 26 outcomes of takotsubo (stress) cardiomyopathy. N Engl J Med. 2015;373(10):929-938. 27 28 4. Madias C, Fitzgibbons T, Alsheikh-Ali A, Bouchard J, Kalsmith B, Garlitski A et al. Acquired long QT 29 syndrome from stress cardiomyopathy is associated with ventricular arrhythmias and torsades de 30 pointes. Heart Rhythm. 2011;8(4):555-561. 31 32 5. Lacey C, Mulder R, Bridgman PG, Kimber B, Zarifeh J, Kennedy M, Cameron V. Broken heart 33 syndrome – Is it a psychosomatic disorder? J Psychosom Res. 2014;77(2):158-160. http://bmjopen.bmj.com/ 34 35 6. Chan C, Troughton R, Elliott J, Zarifeh J, Bridgman P. One-year follow-up of the 2011 Christchurch 36 Earthquake stress cardiomyopathy cases. N Zeal Med J. 2014;127(1396):15-22. 37 38 7. Looi J, Wong C, Lee M, Khan A, Webster M, Kerr A. Usefulness of ECG to differentiate takotsubo 39 cardiomyopathy from acute coronary syndrome. Int J Cardiol. 2015;199:132-140. 40 41 8. Schwarz K, Ahearn T, Srinivasan J, Neil C, Scally C, Rudd A et al. Alterations in cardiac deformation, on October 2, 2021 by guest. Protected copyright. 42 timing of contraction and relaxation, and early myocardial fibrosis accompany the apparent recovery 43 of acute stress-induced (takotsubo) cardiomyopathy: an end to the concept of transience. J Am Soc 44 Echocardiogr. 2017;30(8):745-755. 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from

Comparing the variants of takotsubo syndrome: an observational study of the electrocardiogram and structural changes from a New Zealand tertiary hospital. ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2018-025253.R1

Article Type: Research

Date Submitted by the 30-Oct-2018 Author:

Primary Subject Cardiovascular medicine Heading:

Secondary Subject Heading: Radiology and imaging http://bmjopen.bmj.com/ Echocardiography < CARDIOLOGY, Heart failure < CARDIOLOGY, Keywords: Cardiomyopathy < CARDIOLOGY, Takotsubo

on October 2, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 15 BMJ Open

1 2 3 Comparing the variants of takotsubo syndrome: an observational study of the electrocardiogram 4

and structural changes from a New Zealand tertiary hospital. BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 Watson GMa, Chan CWa, Belluscio Lb, Doudney Kc, Lacey CJd, Kennedy MAe, Bridgman PGa* 7 8 9 10 11 a 12 Department of Cardiology, Christchurch Hospital, Christchurch, New Zealand 13 b Biostatistics, Christchurch Hospital, Christchurch, New Zealand 14 15 c Molecular Pathology Laboratory, Canterbury District Health Board, Christchurch, New Zealand 16 17 d Department of Psychological Medicine, University of Otago, Christchurch, New Zealand 18 For peer review only 19 e Department of Pathology and Biomedical Science, University of Otago Christchurch, Christchurch, 20 New Zealand. 21 22 23 24 * Corresponding author: Dr Paul G Bridgman, Department of Cardiology, Christchurch Hospital, 25 Christchurch, New Zealand, [email protected] 26 27 28 The authors report no conflicts of interest related to this work. 29 30 31 32 33 34 G Watson is a medical student and was supported by the Heart Foundation of New Zealand through 35 a University of Otago Summer Studentship. 36 37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44 45 on October 2, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 15

1 2 3 Objectives: 4 5 In takotsubo syndrome, QTc prolongation is a measure of risk of potentially fatal arrhythmia. It is not BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 known how this risk, or derangement of other markers, differs across the echo variants of takotsubo 7 8 syndrome. As the region of affected myocardium is usually larger, we hypothesised that patients 9 with the classic apical-ballooning form of takotsubo syndrome would have more severe 10 derangement of their markers. Therefore, we sought to explore whether apical takotsubo syndrome 11 differs from the variants of the syndrome in more ways than just regional wall motion pattern. 12 13 Design: 14 15 Observational study of patients gathered from a prospective database (2010-2018) and by 16 retrospective review (2006-2009). 17 18 Setting: For peer review only 19 20 The sole tertiary hospital for a New Zealand region in which case clusters of takotsubo syndrome 21 were precipitated by large earthquakes in 2010, 2011, and 2016. 22 23 Participants: 24 25 222 patients who met a modified version of the Mayo criteria for takotsubo syndrome were 26 included. All patients had digitally archived echocardiograms that were over-read by a second 27 echocardiologist blinded to the clinical report. 28 29 Primary outcome measures: 30 31 Ejection fraction, peak troponin, and QTc interval. 32 33 Results: 34 35 Patients with the apical form were older (p=0.011), had a lower initial LVEF (35% vs. 44%, p<0.0001), 36 and a higher peak hsTnI (p=0.01) than those with variant forms. There was no difference in the 37 electrical abnormalities between the variants (QTc interval, HR, PR interval, QRS duration, or T-wave http://bmjopen.bmj.com/ 38 axis). There was also no correlation between any of peak hsTnI, peak QTc, and ejection fraction. QTc 39 40 interval increased on day two and peaked on day three before falling steeply (p<0.0001). 41 Conclusions: 42 43 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 44 45 differ in their electrical abnormalities. There is a dissociation between the structural and electrical on October 2, 2021 by guest. Protected copyright. 46 abnormalities. QTc peaks on day 3 and then falls steeply. 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 15 BMJ Open

1 2 3 Strengths and limitations of this study 4 5 - A key strength of this study is its size – this is one of the largest single centre studies to date BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 in takotsubo syndrome research. 7 8 - The dataset is remarkably complete. The lack of missing data allows for robust comparisons 9 and correlations and increases confidence in our conclusions. 10 11 - It is a weakness of our study that we did not include late imaging results or other modalities 12 13 such as MRI. 14 - Not all patients were followed up until QTc normalised – future studies should continue 15 16 serial recording of ECGs for a longer period to establish the later time course. 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44 45 on October 2, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 15

1 2 3 Introduction 4 5 Takotsubo syndrome is increasing in incidence, but it is agreed that even in the most obvious of BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 cases we do not understand how the psychological stress triggers the heart illness.1,2 Cardiologists 7 8 recognise that there are variants within the syndrome. The variants are defined by the specific 9 pattern of wall-motion abnormality in the left ventricle. We have noted how heterogeneous the 10 clinical presentation of the condition is and that different subsets of patients can behave in different 11 ways. Takotsubo syndrome has an early mortality rate comparable to that of an acute coronary 12 syndrome.3 Characterising the differing subsets will be important in helping us move towards 13 14 improved understanding of the condition and could lead to better-tailored management of patients 15 and perhaps insights into aetiology. Therefore, we sought to explore whether the echocardiographic 16 variants of the syndrome differ in more ways than just their regional wall motion pattern. 17 18 In everyday clinical practice,For clinicians peer use left review ventricular ejection only fraction (LVEF) as a measure of 19 risk and the degree of troponin elevation as an indicator of myocardial loss. Less widely recognised 20 in takotsubo syndrome is that QTc interval is a measure of risk of potentially fatal arrhythmia.4 It is 21 22 not known if this risk varies across the echo patterns of the condition. A report from the 23 International Takotsubo registry (InterTAK) suggests that outcome does not vary between the apical 24 and variant patterns, whereas the other clinical, biochemical, and electrical features that 25 cardiologists deal in do.5 InterTAK found that patients with the typical takotsubo pattern had lower 26 ejection fractions and higher brain natriuretic peptide levels than those with an atypical pattern. This 27 28 is consistent with the observation that the region of affected myocardium is usually larger in the 29 typical patients. Outcome has previously been suggested to be adversely impacted by the presence 30 of coexisting coronary artery disease.6 We hypothesised that patients with the classic apical- 31 ballooning form of takotsubo syndrome would have more severe derangement of markers of 32 myocardial involvement and more severe QT prolongation than patients with one of the variant 33 34 forms. Therefore, we compared patients with apical takotsubo to those with a variant. 35 36 http://bmjopen.bmj.com/ 37 Methods 38 39 Christchurch Hospital is the only tertiary hospital in Canterbury, New Zealand. Unprecedented case 40 clusters of takotsubo syndrome were precipitated by large earthquakes in Canterbury in 2010, 2011, 41 7,8 42 and 2016, leading to the development of a prospectively-maintained registry of cases since 2010. 43 In addition, we retrospectively identified cases extending back to the installation of a digital 44 echocardiography archive in 2006. 45 on October 2, 2021 by guest. Protected copyright. 46 All patients in the database meet a modified form of the Mayo criteria for takotsubo syndrome, 47 which allows inclusion of patients with one of the recognised characteristic patterns of wall motion 48 abnormality.3 Patients are not excluded if they have coronary artery disease that doesn’t explain 49 50 their wall motion abnormality, or if they die before resolution can be documented on follow up 51 echo. All echocardiograms are held in a digital archive and were over-read by a second 52 echocardiologist to confirm the wall motion pattern and the measurements. Discrepancies in the 53 reporting were settled by joint review. 54 55 High-sensitivity troponin I (hsTnI) was used as a marker of myocardial damage. Prior to 2011, 56 Christchurch Hospital used a standard TnI assay – this was converted to hsTnI using the conversion 57 factor hsTnI = 880 * TnI + 1096, which was derived from a comparative study of the assays done at 58 59 the Canterbury Health Laboratory. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 15 BMJ Open

1 2 3 The electrocardiogram (ECG) machine performs automatic electronic measurement for heart rate 4

(HR), PR interval, QRS width, and QT interval. These were all checked by manual measurement. BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 Where the manual measurement showed that the electronic measure was incorrect the manual 7 measure was used. Borderline discrepancies were resolved by third-party measurement. QTc 8 interval was calculated using Bazett’s formula. The ECG machine also calculates T wave axis. For this, 9 manual measurement and calculation were not performed, and the electronic measure was always 10 11 used. LVEF, hsTnI, and ECG measurements were compared between the apical and variant (basal, 12 focal, mid-wall) forms, and between days over a four-day period from the time of admission. 13 14 Normality was assessed using Shapiro-wilk. Means of normally distributed variables were compared 15 using the two-sample t-test, while the medians of variables that do not follow a normal distribution 16 were compared by Wilcoxon rank-sum. Non-parametric analysis of rank correlation was done using 17 Kendall’s Tau. 18 For peer review only 19 20 21 Patient involvement 22 23 Patients were not involved in the design of this study. 24 25 26 27 Results 28 This study included 222 cases of takotsubo syndrome. Four of the patients were male. The 29 30 echocardiographic pattern was apical in 178 (80%) and variant in 44 (20%). Amongst the variant 31 patients, that pattern was mid-wall in 32, inverted in 4, and focal in 8. In 37 cases (17%), the 32 identified stressor was an earthquake. Another identified emotional stressor was present in 75 33 patients (34%). There was a physical illness trigger in 38 patients (17%), and in 7 cases (3%) there 34 was a concurrent emotional and physical trigger. There was no identified stressor in 65 cases (29%). 35 36 Presentation rhythm was atrial fibrillation in 8 patients. 37 http://bmjopen.bmj.com/ 38 Overall, the patients with the apical form of the syndrome were significantly older (68 vs 64 years, 39 p=0.011). However, unlike the variants, there appeared to be a bimodal age distribution in the apical 40 takotsubo patients. This is illustrated in Figure 1 and suggests that there may be a distinct younger 41 subset. We noted that these younger patients all had major psychological stressors, and in four 42 cases there was also a concomitant physical stressor. 43 44 Patients with the apical form of the syndrome had a lower initial LVEF (35% vs. 44%, p<0.0001), and 45 on October 2, 2021 by guest. Protected copyright. a higher peak hsTnI (p=0.01) than those with variant forms, but there was no difference between the 46 47 forms in peak QTc interval (p=0.93) (Table 1). Apical and variant forms also did not differ significantly 48 in the other ECG measures. 49 50 Table 1. Comparison of LVEF, hsTnI, and ECG measures between apical and variant takotsubo. All 51 ECG variables are as at presentation. LVEF is mean and standard error, other data median and 52 interquartile range. 53 54 Apical takotsubo Variant takotsubo p-value 55 56 Initial LVEF (%) 36.67 (SEM 0.69) 43.52 (SEM 1.60) <0.0001 57 58 Peak hsTnI (ng/l) 2308 (1244-3936) 1638 (289-2441) 0.01 59 60 Heart rate (bpm) 79 (70-94) 76 (66-95) 0.18

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 15

1 2 3 4 PR interval (ms) 166 (150-178) 165 (143-180) 0.49

5 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from QRS duration (ms) 88 (82-100) 88 (80-96) 0.90 6 7 QTc interval (ms) 465 (434-490) 459 (433-484) 0.93 8 9 T-wave axis 75 (54-106.5) 76 (52-98) 0.95 10 11 12 Heart rate, QTc interval, and T-wave axis all changed significantly with time (all p<0.0001), but not by 13 14 echo variant. No patients were on medication known to prolong QT interval at the time of 15 admission. QRS interval and PR interval did not differ by day or echo variant. There were no 16 associations between the structural and electrical measures (Table 2). While peak QTc against LVEF 17 was borderline significant (p=0.04), the Kendal Tau correlation was only -0.097. This is a very weak 18 For peer review only 19 correlation, and with marginal statistical significance is not meaningful, as demonstrated by the 20 scatter plot (Figure 2). We documented different time courses for the changes in QTc and LVEF, and 21 that QTc peaks on day 3 (Figure 3). 22 23 24 25 Table 2. Associations between LVEF, peak hsTnI, and peak QTc. 26 27 Correlation coefficient p-value 28 29 Peak hsTnI vs LVEF -0.091 0.0609 30 31 Peak hsTnI vs peak QTc -0.022 0.6498 32 Peak QTc vs LVEF -0.097 0.0457 33 34 35 36 Discussion 37 http://bmjopen.bmj.com/ 38 We have found that the differences between the variants of takotsubo syndrome go beyond the 39 echo features. Patients with the apical form are older and have a lower ejection fraction, 40 corroborating findings reported by the InterTAK registry.3,5 Patients with the apical form also tended 41 to have a higher peak troponin. Of particular interest in our study is that QTc interval does not track 42 with our structural indices (left ventricular ejection fraction and peak troponin) and that the severity 43 44 of QTc prolongation does not differ between the variants. We do not fully understand the 45 pathophysiology of either the myocardial dysfunction or the QTc prolongation in takotsubo on October 2, 2021 by guest. Protected copyright. 46 syndrome. However, we note that in pluripotent stem cell-derived cardiomyocytes, oestradiol has 47 been shown to protect against action potential prolongation, suggesting that reduction in oestrogen 48 49 level may be a factor in increasing the risk of QT prolongation in these patients, who are 9 50 predominantly post-menopausal. We also note that magnetic resonance imaging of twenty patients 51 with apical takotsubo has found that QTc prolongation may correlate with an apical to basal gradient 52 seen on T2-weighted sequences for myocardial oedema.10 Our data, however, suggest that the 53 repolarisation changes are not dependent on the degree of myocardial involvement. We speculate 54 55 the mechanisms of electrical and structural derangement must be separate from early in the 56 pathophysiological process that follows the triggering event. This is an area that will require future 57 research. 58 59 We have demonstrated a dissociation between the structural and the electrical features in both 60 severity and time course, including demonstrating that QTc peaks on day 3. A case series from

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 15 BMJ Open

1 2 3 Auckland found QTc rises until day three but did not look beyond that point and so did not capture a 4 11

peak. A small series of 93 patients from two hospitals in Boston had previously reported a peak on BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 4 6 day 3. We have now confirmed that finding in a much larger cohort. The Boston group also found 7 that QTc correlates with the risk of ventricular arrhythmia. There is, therefore, a strong case for in- 8 hospital monitoring of patients whilst QTc is prolonged. It is important that patients are not 9 discharged after a normal angiogram if their troponin is falling, without attention to their 10 11 electrocardiogram and QTc interval. We have shown that typically QTc falls steeply between days 3 12 and 4. The implication of this for clinicians is that daily ECGs should be performed until the QTc has 13 normalised and that patients should not be discharged early. 14 15 Our data highlight the dissociation of the structural and electrical features and a complex 16 relationship between the structural indices. Whilst the apical form typically presents with a higher 17 troponin level and a lower ejection fraction, there is no relationship between these two. This 18 suggests that they areFor reflecting peer different aspects review of the condition. only Typically, in apical takotsubo, a 19 20 very large area of myocardium is initially akinetic and while troponin does rise, the rise is relatively 21 small, reflecting that only a very low fraction of affected cells are dying. On standard 2D echo, the 22 vast majority of patients have a full-function recovery. However, a study of 52 patients from 23 Scotland has found that on advanced echo and MRI studies it is possible to detect residual effects 24 and permanent damage.12 Our results are consistent with the concept that the degree of acute 25 26 troponin rise reflects acute cell death and that the number of cells affected in this way is 27 uncorrelated with the total area of myocardium affected. 28 29 Our finding that variant patients are on average younger is not new and has previously been 30 reported by the InterTAK registry.3,5 However, these reports did not explore the distribution of age 31 with each variant. In our cohort there appeared to be a bimodal distribution amongst the apical 32 variant patients with the younger patients having the same syndrome (comparable LVEF, hsTnI, and 33 34 QTc prolongation) as the older patients, but usually with dual or very severe precipitants. This 35 suggests that there is a higher threshold for triggering the apical syndrome in this group of younger 36 patients. 37 http://bmjopen.bmj.com/ 38 Features of our study population suggest that the study results should have good external validity. In 39 our cohort, 80% of cases were apical takotsubo and 20% were variant, similar to the InterTAK 40 registry cohort. In both our study population and in the InterTAK registry, the patients with a variant 41 pattern tended to be younger. 3,5 A minor difference between our case population and other case 42 43 series is in the ratio of males to females. Other multi-centre registries have reported that males 44 account for 5-15% of cases.13,14 In the current study, 2% of cases are male. The major clinical 45 implication of our study is that QT prolongation peaks late and therefore daily ECGs are required, on October 2, 2021 by guest. Protected copyright. 46 and inpatient care should not be shortened. In the absence of contrary evidence, we believe that it 47 48 is prudent that males are treated in this manner also. 49 It is a weakness of our study that we do not have follow-up MRI data. Future work should look to 50 51 correlate longer-term changes with the acute indices of syndrome severity. Our data show a steep 52 fall in QTc from day 3 to day 4, but further studies should also continue serial recording of ECGs for a 53 longer period to establish the later time course. 54 55 Conclusions 56 57 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 58 differ in their electrical abnormalities. There is a dissociation between the structural and electrical 59 abnormalities. QTc peaks on day 3 and then falls steeply. 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 15

1 2 3 Figure legends 4

5 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from Figure 1. Comparison of age for patients with apical takotsubo vs. variant pattern. The horizontal line 6 7 in the middle of the box represents the median of each group, while the diamonds represent the 8 mean. The box upper and lower bounds represent the upper and lower quartiles respectively. The 9 whiskers represent the minimum and maximum values excluding those that are 1.5*IQR away from 10 the median (circles). * p=0.011 11 12 13 Figure 2. Scatter plot showing the correlation between peak QTc interval and LVEF. 14 15 Figure 3. Change in hsTnI and QTc interval over time after admission. 16 17 18 References For peer review only 19 20 1. Minhas AS, Hughey AB, Kolias TJ. Nationwide trends in reported incidence of takotsubo 21 22 cardiomyopathy from 2006 to 2012. Am J Cardiol 2015;116:1128-1131. 23 2. Otten A, Ottervanger J, Symersky T, Suryapranata H, de Boer M, Maas A. Diagnosis of takotsubo 24 25 cardiomyopathy is increasing over time in patients presenting as ST-elevation myocardial infarction. 26 Neth Heart J. 2016;24(9):520-529. 27 28 3. Templin C, Ghadri J, Diekmann J, Napp L, Bataiosu D, Jaguszewski M et al. Clinical features and 29 outcomes of takotsubo (stress) cardiomyopathy. N Engl J Med. 2015;373(10):929-938. 30 31 4. Madias C, Fitzgibbons T, Alsheikh-Ali A, Bouchard J, Kalsmith B, Garlitski A et al. Acquired long QT 32 syndrome from stress cardiomyopathy is associated with ventricular arrhythmias and torsades de 33 pointes. Heart Rhythm. 2011;8(4):555-561. 34 35 5. Ghadri JR, Cammann L, Napp C, Jurisic S, Diekmann J, Bataiosu DR et al. Differences in the clinical 36 profile and outcomes of typical and atypical takotsubo syndrome. JAMA Cardiol. 2016;1(3):335-340. 37 http://bmjopen.bmj.com/ 38 6. Bill V, El-Battrawy I, Schramm K, Ansari U. Hoffmann U, Haghi D et al. Coincidental coronary 39 artery disease impairs outcome in patients with takotsubo cardiomyopathy QJM. 2017;110:483–488. 40 41 7. Lacey C, Mulder R, Bridgman PG, Kimber B, Zarifeh J, Kennedy M, Cameron V. Broken heart 42 syndrome – Is it a psychosomatic disorder? J Psychosom Res. 2014;77(2):158-160. 43 44 8. Chan C, Troughton R, Elliott J, Zarifeh J, Bridgman P. One-year follow-up of the 2011 Christchurch 45 on October 2, 2021 by guest. Protected copyright. 46 Earthquake stress cardiomyopathy cases. N Zeal Med J. 2014;127(1396):15-22. 47 9. El-Battrawy I, Zkao Z, Lan H, Schunemann J, Sattler K, Buljubasic B, et al. Estradiol protection 48 49 against toxic effects of catecholamine on electrical properties in human-induced pluripotent stem 50 cell derived cardiomyocytes. Int J Cardiol 2018;254:195-202. 51 52 10. Perazzolo M, Zorzi A, Corbetti F, De Lazzari M, Migliore F, Tona F, et al. Apicobasal gradient of 53 left ventricular myocardial edema underlies transient T-wave inversion and QT interval prolongation 54 (Wellens' ECG pattern) in Tako-Tsubo cardiomyopathy. Heart Rhythm. 2013;10:70-7. 55 56 11. Looi J, Wong C, Lee M, Khan A, Webster M, Kerr A. Usefulness of ECG to differentiate takotsubo 57 cardiomyopathy from acute coronary syndrome. Int J Cardiol. 2015;199:132-140. 58 59 12. Schwarz K, Ahearn T, Srinivasan J, Neil C, Scally C, Rudd A et al. Alterations in cardiac 60 deformation, timing of contraction and relaxation, and early myocardial fibrosis accompany the

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 15 BMJ Open

1 2 3 apparent recovery of acute stress-induced (takotsubo) cardiomyopathy: an end to the concept of 4

transience. J Am Soc Echocardiogr. 2017;30(8):745-755. BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 13. Tornvall P, Collste O, Ehrenborg E, Jarnbert-Petterson H. A case-control study of risk markers and 7 8 mortality in takotsubo stress cardiomyopathy. J Am Coll Cardiol 2016;67:1931-6. 9 14. Shah M, Ram P, Lo K, Sirinvaravong N, Patel B, Tripathi B, et al. Etiologies, predictors, and 10 11 economic impact of readmission within 1 month among patients with takotsubo cardiomyopathy. 12 Clin Cardiol. 2018;41:916-923. 13 14 15 16 17 18 Footnotes For peer review only 19 20 Contributors: The authors contributed to the study as follows: design and conception (PB, MK, CL), 21 acquisition of data (GW, CC), data interpretation and manuscript preparation (GW, CC, LB, KD, CL, 22 MK, PB). 23 24 25 Funding: George Watson was funded as a University of Otago Summer Student by the National 26 Heart Foundation of New Zealand. 27 28 Competing Interests: None declared. 29 30 Patient consent: Not obtained. 31 32 Ethics approval: The study was approved by the relevant statutory authority in New Zealand, the 33 34 Health and Disability Ethics Committee. The approval number is URA/11/07/033/AM03. 35 Provenance and peer review: Not Commissioned; externally peer reviewed. 36 37 http://bmjopen.bmj.com/ 38 Data sharing statement: For this study we used data and electrocardiograms collected during 39 routine clinical practice. All relevant summary data are provided in the paper. The Canterbury 40 District Health Board does not grant permission for the raw data to be made publicly available due 41 to ethical and legal restrictions. However, all interested readers may request data without restriction 42 from Dr Paul Bridgman ([email protected]). 43 44 45 on October 2, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 15

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Figure 1. Comparison of age for patients with apical takotsubo vs. variant pattern. The horizontal line in the middle of the box represents the median of each group, while the diamonds represent the mean. The box 31 upper and lower bounds represent the upper and lower quartiles respectively. The whiskers represent the 32 minimum and maximum values excluding those that are 1.5*IQR away from the median (circles). * http://bmjopen.bmj.com/ 33 p=0.011 34 35 149x106mm (300 x 300 DPI) 36 37 38 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 15 BMJ Open

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 Figure 2. Scatter plot showing the correlation between peak QTc interval and LVEF. 28 29 142x90mm (300 x 300 DPI) 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 15

1 BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 Figure 3. Change in hsTnI and QTc interval over time after admission. 25 26 170x90mm (300 x 300 DPI) 27 28 29 30 31 32 33 http://bmjopen.bmj.com/ 34 35 36 37 38 39 40 41 on October 2, 2021 by guest. Protected copyright. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 15 BMJ Open

1 STROBE Statement—Checklist of items that should be included in reports of cohort studies 2 3 Item 4 No Recommendation

5 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 Done. Page 1 7 (b) Provide in the abstract an informative and balanced summary of what was done 8 9 and what was found 10 Provided. Page 2 11 Introduction 12 13 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 14 Included. Page 4 15 Objectives 3 State specific objectives, including any prespecified hypotheses 16 Hypothesis specifically stated. Pg 4, Second paragraph, penultimate sentence. 17 18 Methods For peer review only 19 Study design 4 Present key elements of study design early in the paper 20 Yes. Page 4, see methods. 21 22 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 23 exposure, follow-up, and data collection 24 Yes. Page 4. 25 26 Participants 6 (a) Give the eligibility criteria, and the sources and methods of selection of 27 participants. Describe methods of follow-up 28 Included. Page 4, methods paragraph 2. 29 (b) For matched studies, give matching criteria and number of exposed and 30 31 unexposed 32 Not applicable 33 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 34 modifiers. Give diagnostic criteria, if applicable 35 36 Outcomes defined. Pages 4-5. 37 Data sources/ 8* For each variable of interest, give sources of data and details of methods of http://bmjopen.bmj.com/ 38 measurement assessment (measurement). Describe comparability of assessment methods if there is 39 more than one group 40 41 Methods fully included. Pages 4-5. 42 Bias 9 Describe any efforts to address potential sources of bias 43 Methods page 4 documents that all echos were over-read 44

Study size 10 Explain how the study size was arrived at on October 2, 2021 by guest. Protected copyright. 45 46 Methods page 4, consecutive cases were included 47 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, 48 describe which groupings were chosen and why 49 50 Not applicable 51 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 52 Described on page 5. 53 (b) Describe any methods used to examine subgroups and interactions 54 55 Subgroup analyses not performed. 56 (c) Explain how missing data were addressed 57 Completeness of data is a feature of this study. 58 (d) If applicable, explain how loss to follow-up was addressed 59 60 This is an inpatient study with complete follow-up to discharge (e) Describe any sensitivity analyses Not applicable. 1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 15

1 Results 2 3 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially 4 eligible, examined for eligibility, confirmed eligible, included in the study, 5 completing follow-up, and analysed BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 Included, page 5 results. 7 8 (b) Give reasons for non-participation at each stage 9 Not applicable 10 (c) Consider use of a flow diagram 11 Not necessary 12 13 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and 14 information on exposures and potential confounders 15 Included, page 5 results. 16 (b) Indicate number of participants with missing data for each variable of interest 17 18 ForData peer is complete, reviewpages 5-6 and figures only and tables. 19 (c) Summarise follow-up time (eg, average and total amount) 20 An inpatient study, detail included 21 22 Outcome data 15* Report numbers of outcome events or summary measures over time 23 Reported, results pages 5-6 24 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and 25 their precision (eg, 95% confidence interval). Make clear which confounders were 26 27 adjusted for and why they were included 28 Not applicable 29 (b) Report category boundaries when continuous variables were categorized 30 Not applicable 31 32 (c) If relevant, consider translating estimates of relative risk into absolute risk for a 33 meaningful time period 34 No relevant 35 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and 36 37 sensitivity analyses http://bmjopen.bmj.com/ 38 Not applicable 39 40 Discussion 41 Key results 18 Summarise key results with reference to study objectives 42 Done, page 6 43 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or 44 45 imprecision. Discuss both direction and magnitude of any potential bias on October 2, 2021 by guest. Protected copyright. 46 Included, page 7. 47 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, 48 multiplicity of analyses, results from similar studies, and other relevant evidence 49 50 Cautious interpretation given. Page 6-7. 51 Generalisability 21 Discuss the generalisability (external validity) of the study results 52 Done. Page 7. 53 54 Other information 55 Funding 22 Give the source of funding and the role of the funders for the present study and, if 56 applicable, for the original study on which the present article is based 57 Included on page 9. 58 59 60 *Give information separately for exposed and unexposed groups.

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 15 BMJ Open

1 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 2 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 3 4 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at

5 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 available at http://www.strobe-statement.org. 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 http://bmjopen.bmj.com/ 38 39 40 41 42 43 44 45 on October 2, 2021 by guest. Protected copyright. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from

Article Type: Research

Date Submitted by the 22-Dec-2018 Author:

Primary Subject Cardiovascular medicine Heading:

Secondary Subject Heading: Radiology and imaging http://bmjopen.bmj.com/ Echocardiography < CARDIOLOGY, Heart failure < CARDIOLOGY, Keywords: Cardiomyopathy < CARDIOLOGY, Takotsubo

on October 2, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 15 BMJ Open

1 2 3 Comparing the variants of takotsubo syndrome: an observational study of the electrocardiogram 4

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 15

1 2 3 Objectives: 4 5 In takotsubo syndrome, QTc prolongation is a measure of risk of potentially fatal arrhythmia. It is not BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 known how this risk, or derangement of other markers, differs across the echo variants of takotsubo 7 8 syndrome. Therefore, we sought to explore whether apical takotsubo syndrome differs from the 9 variants of the syndrome in more ways than just regional wall motion pattern. As the region of 10 affected myocardium is usually larger, we hypothesised that patients with the classic apical- 11 ballooning form of takotsubo syndrome would have more severe derangement of their markers. 12 13 Design: 14 15 Observational study of patients gathered from a prospective database (2010-2018) and by 16 retrospective review (2006-2009). 17 18 Setting: For peer review only 19 20 The sole tertiary hospital for a New Zealand region in which case clusters of takotsubo syndrome 21 were precipitated by large earthquakes in 2010, 2011, and 2016. 22 23 Participants: 24 25 222 patients who met a modified version of the Mayo criteria for takotsubo syndrome were 26 included. All patients had digitally archived echocardiograms that were over-read by a second 27 echocardiologist blinded to the clinical report. 28 29 Primary outcome measures: 30 31 Ejection fraction, peak troponin, and QTc interval. 32 33 Results: 34 35 Patients with the apical form were older (p=0.011), had a lower initial LVEF (35% vs. 44%, p<0.0001), 36 and a higher peak hsTnI (p=0.01) than those with variant forms. There was no difference in the 37 electrical abnormalities between the variants (QTc interval, HR, PR interval, QRS duration, or T-wave http://bmjopen.bmj.com/ 38 axis). There was also no correlation between any of peak hsTnI, peak QTc, and ejection fraction. QTc 39 40 interval increased on day two and peaked on day three before falling steeply (p<0.0001). 41 Conclusions: 42 43 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 44 45 differ in their electrical abnormalities. There is a dissociation between the structural and electrical on October 2, 2021 by guest. Protected copyright. 46 abnormalities. QTc peaks on day 3 and then falls steeply. 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 15 BMJ Open

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 15

1 2 3 Introduction 4 5 The reported incidence of Takotsubo syndrome is increasing.1 It is agreed that even in the most BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 obvious of cases we do not understand how the psychological stress triggers the heart illness.2 7 8 Cardiologists recognise that there are variants within the syndrome. The variants are defined by the 9 specific pattern of wall-motion abnormality in the left ventricle. We have noted how heterogeneous 10 the clinical presentation of the condition is and that different subsets of patients can behave in 11 different ways. Takotsubo syndrome has an early mortality rate comparable to that of an acute 12 coronary syndrome.3 Characterising the differing subsets will be important in helping us move 13 14 towards improved understanding of the condition and could lead to better-tailored management of 15 patients and perhaps insights into aetiology. Therefore, we sought to explore whether the 16 echocardiographic variants of the syndrome differ in more ways than just their regional wall motion 17 pattern. 18 For peer review only 19 In everyday clinical practice, clinicians use left ventricular ejection fraction (LVEF) as a measure of 20 risk and the degree of troponin elevation as an indicator of myocardial loss. Less widely recognised 21 4 22 in takotsubo syndrome is that QTc interval is a measure of risk of potentially fatal arrhythmia. It is 23 not known if this risk varies across the echo patterns of the condition. A report from the 24 International Takotsubo registry (InterTAK) suggests that outcome does not vary between the apical 25 and variant patterns, whereas the other clinical, biochemical, and electrical features that 26 cardiologists deal in do.5 InterTAK found that patients with the typical takotsubo pattern had lower 27 28 ejection fractions and higher brain natriuretic peptide levels than those with an atypical pattern. This 29 is consistent with the observation that the region of affected myocardium is usually larger in the 30 typical patients. Outcome has previously been suggested to be adversely impacted by the presence 31 of coexisting coronary artery disease.6 32 33 Our objective in this study was to explore whether apical takotsubo syndrome differs from the 34 variants of the syndrome in more ways than just the regional wall motion pattern. We hypothesised 35 that patients with the classic apical-ballooning form of takotsubo syndrome would have more severe 36 37 derangement of markers of myocardial involvement and more severe QT prolongation than patients http://bmjopen.bmj.com/ 38 with one of the variant forms. 39 40 41 42 Methods 43 Christchurch Hospital is the only tertiary hospital in Canterbury, New Zealand. Unprecedented case 44 45 clusters of takotsubo syndrome were precipitated by large earthquakes in Canterbury in 2010, 2011, on October 2, 2021 by guest. Protected copyright. 46 and 2016, leading to the development of a prospectively-maintained registry of cases since 2010.7,8 47 In addition, we retrospectively identified cases extending back to the installation of a digital 48 echocardiography archive in 2006. 49 50 All patients in the database meet a modified form of the Mayo criteria for takotsubo syndrome, 51 which allows inclusion of patients with one of the recognised characteristic patterns of wall motion 52 3 53 abnormality. Patients are not excluded if they have coronary artery disease that doesn’t explain 54 their wall motion abnormality, or if they die before resolution can be documented on follow up 55 echo. All echocardiograms are held in a digital archive and were over-read by a second 56 echocardiologist to confirm the wall motion pattern and the measurements. Discrepancies in the 57 reporting were settled by joint review. 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 15 BMJ Open

1 2 3 High-sensitivity troponin I (hsTnI) was used as a marker of myocardial damage. Prior to 2011, 4

Christchurch Hospital used a standard TnI assay – this was converted to hsTnI using the conversion BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 factor hsTnI = 880 * TnI + 1096, which was derived from a comparative study of the assays done at 7 the Canterbury Health Laboratory. 8 9 The electrocardiogram (ECG) machine performs automatic electronic measurement for heart rate 10 (HR), PR interval, QRS width, and QT interval. These were all checked by manual measurement. 11 Where the manual measurement showed that the electronic measure was incorrect the manual 12 measure was used. Borderline discrepancies were resolved by third-party measurement. QTc 13 14 interval was calculated using Bazett’s formula. The ECG machine also calculates T wave axis. For this, 15 manual measurement and calculation were not performed, and the electronic measure was always 16 used. LVEF, hsTnI, and ECG measurements were compared between the apical and variant (basal, 17 focal, mid-wall) forms, and between days over a four-day period from the time of admission. 18 For peer review only 19 Normality was assessed using Shapiro-wilk. Means of normally distributed variables were compared 20 using the two-sample t-test, while the medians of variables that do not follow a normal distribution 21 22 were compared by Wilcoxon rank-sum. Non-parametric analysis of rank correlation was done using 23 Kendall’s Tau. 24 25 26 Patient involvement 27 28 Patients were not involved in the design of this study. 29 30 31 32 Results 33 34 This study included 222 cases of takotsubo syndrome. Four of the patients were male. The 35 echocardiographic pattern was apical in 178 (80%) and variant in 44 (20%). Amongst the variant 36 patients, that pattern was mid-wall in 32, inverted in 4, and focal in 8. In 37 cases (17%), the 37 http://bmjopen.bmj.com/ 38 identified stressor was an earthquake. Another identified emotional stressor was present in 75 39 patients (34%). There was a physical illness trigger in 38 patients (17%), and in 7 cases (3%) there 40 was a concurrent emotional and physical trigger. There was no identified stressor in 65 cases (29%). 41 Presentation rhythm was atrial fibrillation in 8 patients. At the time of admission, 9% of patients 42 were on medications that may prolong the QT interval and have a definite or possible risk of 43 44 ventricular arrhythmia. The rate of prescription of these medications was similar between patients 45 with the apical type (10%) and patients with a variant (7%). on October 2, 2021 by guest. Protected copyright. 46 47 Overall, the patients with the apical form of the syndrome were significantly older (68 vs 64 years, 48 p=0.011). However, unlike the variants, there appeared to be a bimodal age distribution in the apical 49 takotsubo patients. This is illustrated in Figure 1 and suggests that there may be a distinct younger 50 subset. We noted that these younger patients all had major psychological stressors, and in four 51 52 cases there was also a concomitant physical stressor. 53 Patients with the apical form of the syndrome had a lower initial LVEF (35% vs. 44%, p<0.0001), and 54 55 a higher peak hsTnI (p=0.01) than those with variant forms, but there was no difference between the 56 forms in peak QTc interval (p=0.93) (Table 1). Apical and variant forms also did not differ significantly 57 in the other ECG measures. 58 59 Table 1. Comparison of LVEF, hsTnI, and ECG measures between apical and variant takotsubo. All 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 15

1 2 3 4 ECG variables are as at presentation. LVEF is mean and standard error, other data median and

5 interquartile range. BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 6 7 Apical takotsubo Variant takotsubo p-value 8 Initial LVEF (%) 36.67 (SEM 0.69) 43.52 (SEM 1.60) <0.0001 9 10 Peak hsTnI (ng/l) 2308 (1244-3936) 1638 (289-2441) 0.01 11 12 Heart rate (bpm) 79 (70-94) 76 (66-95) 0.18 13 14 PR interval (ms) 166 (150-178) 165 (143-180) 0.49 15 16 QRS duration (ms) 88 (82-100) 88 (80-96) 0.90 17 18 QTc interval (ms) For peer465 (434-490) review459 (433-484) only 0.93 19 T-wave axis 75 (54-106.5) 76 (52-98) 0.95 20 21 22 23 Heart rate, QTc interval, and T-wave axis all changed significantly with time (all p<0.0001), but not by 24 echo variant. QRS interval and PR interval did not differ by day or echo variant. There were no 25 associations between the structural and electrical measures (Table 2). While peak QTc against LVEF 26 was borderline significant (p=0.04), the Kendal Tau correlation was only -0.097. This is a very weak 27 28 correlation, and with marginal statistical significance is not meaningful, as demonstrated by the 29 scatter plot (Figure 2). We documented different time courses for the changes in QTc and LVEF, and 30 that QTc peaks on day 3 (Figure 3). 31 32 33 34 Table 2. Associations between LVEF, peak hsTnI, and peak QTc. 35 36 Correlation coefficient p-value 37 http://bmjopen.bmj.com/ 38 Peak hsTnI vs LVEF -0.091 0.0609 39 40 Peak hsTnI vs peak QTc -0.022 0.6498 41 Peak QTc vs LVEF -0.097 0.0457 42 43 44 45 Discussion on October 2, 2021 by guest. Protected copyright. 46 47 We have found that the differences between the variants of takotsubo syndrome go beyond the 48 echo features. Patients with the apical form are older and have a lower ejection fraction, 49 corroborating findings reported by the InterTAK registry.3,5 Patients with the apical form also tended 50 to have a higher peak troponin. Of particular interest in our study is that QTc interval does not track 51 52 with our structural indices (left ventricular ejection fraction and peak troponin) and that the severity 53 of QTc prolongation does not differ between the variants. We do not fully understand the 54 pathophysiology of either the myocardial dysfunction or the QTc prolongation in takotsubo 55 syndrome. However, we note that in pluripotent stem cell-derived cardiomyocytes, oestradiol has 56 been shown to protect against action potential prolongation, suggesting that reduction in oestrogen 57 58 level may be a factor in increasing the risk of QT prolongation in these patients, who are 59 predominantly post-menopausal.9 We also note that magnetic resonance imaging of twenty patients 60 with apical takotsubo has found that QTc prolongation may correlate with an apical to basal gradient

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 15 BMJ Open

1 2 3 seen on T2-weighted sequences for myocardial oedema.10 Our data, however, suggest that the 4

repolarisation changes are not dependent on the degree of myocardial involvement. We speculate BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 the mechanisms of electrical and structural derangement must be separate from early in the 7 pathophysiological process that follows the triggering event. This is an area that will require future 8 research. 9 10 We have demonstrated a dissociation between the structural and the electrical features in both 11 severity and time course, including demonstrating that QTc peaks on day 3. A case series from 12 Auckland found QTc rises until day three but did not look beyond that point and so did not capture a 13 11 14 peak. A small series of 93 patients from two hospitals in Boston had previously reported a peak on 15 day 3.4 We have now confirmed that finding in a much larger cohort. The Boston group also found 16 that QTc correlates with the risk of ventricular arrhythmia. There is, therefore, a strong case for in- 17 hospital monitoring of patients whilst QTc is prolonged. It is important that patients are not 18 discharged after a normalFor angiogram peer if their review troponin is falling, onlywithout attention to their 19 20 electrocardiogram and QTc interval. We have shown that typically QTc falls steeply between days 3 21 and 4. The implication of this for clinicians is that daily ECGs should be performed until the QTc has 22 normalised and that patients should not be discharged early. Based on our data, this may require at 23 least a 4-day admission. 24 25 Our data highlight the dissociation of the structural and electrical features and a complex 26 relationship between the structural indices. Whilst the apical form typically presents with a higher 27 28 troponin level and a lower ejection fraction, there is no relationship between these two. This 29 suggests that they are reflecting different aspects of the condition. Typically, in apical takotsubo, a 30 very large area of myocardium is initially akinetic and while troponin does rise, the rise is relatively 31 small, reflecting that only a very low fraction of affected cells are dying. On standard 2D echo, the 32 vast majority of patients have a full-function recovery. However, a study of 52 patients from 33 34 Scotland has found that on advanced echo and MRI studies it is possible to detect residual effects 35 and permanent damage.12 Our results are consistent with the concept that the degree of acute 36 troponin rise reflects acute cell death and that the number of cells affected in this way is 37 uncorrelated with the total area of myocardium affected. http://bmjopen.bmj.com/ 38 39 Our finding that variant patients are on average younger is not new and has previously been 40 reported by the InterTAK registry.3,5 However, these reports did not explore the distribution of age 41 with each variant. In our cohort there appeared to be a bimodal distribution amongst the apical 42 43 variant patients with the younger patients having the same syndrome (comparable LVEF, hsTnI, and 44 QTc prolongation) as the older patients, but usually with dual or very severe precipitants. This 45 suggests that there is a higher threshold for triggering the apical syndrome in this group of younger on October 2, 2021 by guest. Protected copyright. 46 patients. 47 48 Features of our study population suggest that the study results should have good external validity. In 49 our cohort, 80% of cases were apical takotsubo and 20% were variant, similar to the InterTAK 50 51 registry cohort. In both our study population and in the InterTAK registry, the patients with a variant 52 pattern tended to be younger. 3,5 A minor difference between our case population and other case 53 series is in the ratio of males to females. Other multi-centre registries have reported that males 54 account for 5-15% of cases.13,14 In the current study, 2% of cases are male. The major clinical 55 implication of our study is that QT prolongation peaks late and therefore daily ECGs are required, 56 57 and inpatient care should not be shortened. In the absence of contrary evidence, we believe that it 58 is prudent that males are treated in this manner also. 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 15

1 2 3 It is a weakness of our study that we do not have follow-up MRI data. Future work should look to 4

correlate longer-term changes with the acute indices of syndrome severity. Our data show a steep BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 fall in QTc from day 3 to day 4, but further studies should also continue serial recording of ECGs for a 7 longer period to establish the later time course. 8 9 Conclusions 10 11 The variants of takotsubo syndrome differ in more ways than just their echo pattern but do not 12 differ in their electrical abnormalities. There is a dissociation between the structural and electrical 13 abnormalities. QTc peaks on day 3 and then falls steeply. 14 15 Figure legends 16 17 Figure 1. Comparison of age for patients with apical takotsubo vs. variant pattern. The horizontal line 18 in the middle of the boxFor represents peer the median review of each group, whileonly the diamonds represent the 19 20 mean. The box upper and lower bounds represent the upper and lower quartiles respectively. The 21 whiskers represent the minimum and maximum values excluding those that are 1.5*IQR away from 22 the median (circles). * p=0.011 23 24 Figure 2. Scatter plot showing the correlation between peak QTc interval and LVEF. 25 26 27 Figure 3. Change in hsTnI and QTc interval over time after admission. 28 29 30 References 31 32 1. Minhas AS, Hughey AB, Kolias TJ. Nationwide trends in reported incidence of takotsubo 33 cardiomyopathy from 2006 to 2012. Am J Cardiol 2015;116:1128-1131. 34 35 2. Otten A, Ottervanger J, Symersky T, Suryapranata H, de Boer M, Maas A. Diagnosis of takotsubo 36 cardiomyopathy is increasing over time in patients presenting as ST-elevation myocardial infarction. 37 http://bmjopen.bmj.com/ 38 Neth Heart J. 2016;24(9):520-529. 39 3. Templin C, Ghadri J, Diekmann J, Napp L, Bataiosu D, Jaguszewski M et al. Clinical features and 40 41 outcomes of takotsubo (stress) cardiomyopathy. N Engl J Med. 2015;373(10):929-938. 42 4. Madias C, Fitzgibbons T, Alsheikh-Ali A, Bouchard J, Kalsmith B, Garlitski A et al. Acquired long QT 43 44 syndrome from stress cardiomyopathy is associated with ventricular arrhythmias and torsades de 45 pointes. Heart Rhythm. 2011;8(4):555-561. on October 2, 2021 by guest. Protected copyright. 46 47 5. Ghadri JR, Cammann L, Napp C, Jurisic S, Diekmann J, Bataiosu DR et al. Differences in the clinical 48 profile and outcomes of typical and atypical takotsubo syndrome. JAMA Cardiol. 2016;1(3):335-340. 49 50 6. Bill V, El-Battrawy I, Schramm K, Ansari U. Hoffmann U, Haghi D et al. Coincidental coronary 51 artery disease impairs outcome in patients with takotsubo cardiomyopathy QJM. 2017;110:483–488. 52 53 7. Lacey C, Mulder R, Bridgman PG, Kimber B, Zarifeh J, Kennedy M, Cameron V. Broken heart 54 syndrome – Is it a psychosomatic disorder? J Psychosom Res. 2014;77(2):158-160. 55 56 8. Chan C, Troughton R, Elliott J, Zarifeh J, Bridgman P. One-year follow-up of the 2011 Christchurch 57 Earthquake stress cardiomyopathy cases. N Zeal Med J. 2014;127(1396):15-22. 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 15 BMJ Open

1 2 3 9. El-Battrawy I, Zkao Z, Lan H, Schunemann J, Sattler K, Buljubasic B, et al. Estradiol protection 4

against toxic effects of catecholamine on electrical properties in human-induced pluripotent stem BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from 5 6 cell derived cardiomyocytes. Int J Cardiol 2018;254:195-202. 7 8 10. Perazzolo M, Zorzi A, Corbetti F, De Lazzari M, Migliore F, Tona F, et al. Apicobasal gradient of 9 left ventricular myocardial edema underlies transient T-wave inversion and QT interval prolongation 10 (Wellens' ECG pattern) in Tako-Tsubo cardiomyopathy. Heart Rhythm. 2013;10:70-7. 11 12 11. Looi J, Wong C, Lee M, Khan A, Webster M, Kerr A. Usefulness of ECG to differentiate takotsubo 13 cardiomyopathy from acute coronary syndrome. Int J Cardiol. 2015;199:132-140. 14 15 12. Schwarz K, Ahearn T, Srinivasan J, Neil C, Scally C, Rudd A et al. Alterations in cardiac 16 deformation, timing of contraction and relaxation, and early myocardial fibrosis accompany the 17 apparent recovery of acute stress-induced (takotsubo) cardiomyopathy: an end to the concept of 18 For peer review only 19 transience. J Am Soc Echocardiogr. 2017;30(8):745-755. 20 13. Tornvall P, Collste O, Ehrenborg E, Jarnbert-Petterson H. A case-control study of risk markers and 21 22 mortality in takotsubo stress cardiomyopathy. J Am Coll Cardiol 2016;67:1931-6. 23 24 14. Shah M, Ram P, Lo K, Sirinvaravong N, Patel B, Tripathi B, et al. Etiologies, predictors, and 25 economic impact of readmission within 1 month among patients with takotsubo cardiomyopathy. 26 Clin Cardiol. 2018;41:916-923. 27 28 29 30 31 32 Footnotes 33 34 Contributors: The authors contributed to the study as follows: design and conception (PB, MK, CL), 35 acquisition of data (GW, CC), data interpretation and manuscript preparation (GW, CC, LB, KD, CL, 36 MK, PB). 37 http://bmjopen.bmj.com/ 38 39 Funding: George Watson was funded as a University of Otago Summer Student by the National 40 Heart Foundation of New Zealand. 41 42 Competing Interests: None declared. 43 44 Patient consent: Not obtained. 45 on October 2, 2021 by guest. Protected copyright. 46 Ethics approval: The study was approved by the relevant statutory authority in New Zealand, the 47 48 Health and Disability Ethics Committee. The approval number is URA/11/07/033/AM03. 49 50 Provenance and peer review: Not Commissioned; externally peer reviewed. 51 52 Data sharing statement: For this study we used data and electrocardiograms collected during 53 routine clinical practice. All relevant summary data are provided in the paper. The Canterbury 54 District Health Board does not grant permission for the raw data to be made publicly available due 55 to ethical and legal restrictions. However, all interested readers may request data without restriction 56 from Dr Paul Bridgman ([email protected]). 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 15

1 STROBE Statement—Checklist of items that should be included in reports of cohort studies 2 3 Item 4 No Recommendation

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 15 BMJ Open

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2018-025253 on 5 May 2019. Downloaded from

Article Type: Research

Date Submitted by the 17-Mar-2019 Author:

Complete List of Authors: Watson, George; Christchurch Hospital, Cardiology Chan, Christina; Christchurch Hospital, Cardiology Belluscio, Laura; Christchurch Hospital, Biostatistics Doudney, Kit; Canterbury District Health Board, Molecular Pathology Lacey, Cameron; Christchurch Hospital, Psychological Medicine Kennedy, Martin; University of Otago Christchurch Department of Pathology Bridgman, Paul; Christchurch Hospital, Cardiology

Primary Subject Cardiovascular medicine Heading:

Secondary Subject Heading: Radiology and imaging http://bmjopen.bmj.com/