Predictive Performance of the International Takotsubo Registry Score in the Diagnosis of Takotsubo Syndrome Among Women with Non-ST Segment Elevation Myocardial Infarction

Abstract

Background:

The International Takotsubo Registry (InterTAK) score was developed to assess the probability of takotsubo syndrome (TTS) and to distinguish it from acute coronary syndrome, which has a similar clinical presentation. A European Society of Cardiology (ESC) consensus statement suggests invasive coronary angiography may be deferred in patients with non-ST segment elevation myocardial infarction (NSTEMI) and high probability of TTS. We aimed to determine the predictive performance of the InterTAK score in a real-world population of NSTEMIs.

Materials and Methods:

The InterTAK score was retrospectively calculated for consecutive female patients with NSTEMI who underwent invasive coronary angiography at two academic medical centers in New York City from March 2016 to August 2018.

Results:

Among 375 women with NSTEMI, 15 (4%) had InterTAK score >70, indicating high probability of TTS. Ten (67%) met ESC TTS criteria for a noninvasive strategy, among whom seven had TTS, two had spontaneous coronary artery dissection (SCAD), and one had coronary artery disease requiring revascularization. A total of 48 women (13% of cohort) had a final diagnosis of TTS. The InterTAK score had an area under the receiver operating characteristics curve (AUC) of 0.82 (95% confidence interval, 0.75–0.88) for TTS, with 21% sensitivity and 98% specificity at score >70.

Conclusions:

In this real-world sample of consecutive female NSTEMI patients, the InterTAK score was predictive of TTS but a high score was rare, and use of the score to obviate invasive angiography could have resulted in delay of revascularization and missed diagnosis of SCAD for a small number of patients. The InterTAK score should be used in conjunction with other clinical variables for angiographic referral.

Introduction

Takotsubo syndrome (TTS) is characterized by a clinical presentation similar to acute coronary syndrome (ACS) and is associated with transient and reversible left ventricular dysfunction in the absence of causative obstructive coronary artery disease (CAD) or unstable plaque.^1,2 TTS predominantly affects postmenopausal women and is often preceded by stressful emotional or physical triggers.

Distinguishing TTS from ACS can be challenging, given similarities in the initial presentation.³ The International Takotsubo Registry (InterTAK) score was developed to aid clinicians in the risk stratification of patients presenting with a syndrome compatible with either ACS or TTS and includes seven parameters readily available in an emergency department that, when present, favor the diagnosis of TTS. The maximum attainable InterTAK score is 100 points, and a score of >70 conferred an ∼90% probability of TTS in initial evaluation.⁴

The European Society of Cardiology (ESC) statement on management of TTS recommends the use of the InterTAK score for stable patients presenting with non-ST segment elevation myocardial infarction (NSTEMI) and possible TTS based on echocardiography, to identify patients in whom coronary angiography may be deferred in favor of coronary computed tomography angiography (CCTA).⁴ However, the predictive performance of the InterTAK score has not been independently validated in patients with NSTEMI. The aim of this study was to investigate the performance of the InterTAK score in predicting TTS in a real-world consecutive cohort of female patients presenting with NSTEMI.

Materials and Methods

The NYU Registry of Women with myocardial infarction (MI), a retrospective cohort study that includes consecutive female patients with MI presenting to NYU Langone Health or Bellevue Hospital Center, was used to examine the discriminatory performance of the InterTAK diagnostic score. This study was approved by the NYU Institutional Review Board (IRB) and a waiver of consent was obtained owing to the retrospective nature of the analysis. The InterTAK diagnostic score was retrospectively calculated for all adult women ≥18 years of age who presented with a provisional diagnosis of NSTEMI, defined according to the fourth Universal Definition of MI,⁵ from March 2016 to August 2018. Only patients who underwent invasive coronary angiography were included in the analysis. Patients with ST-segment elevation MI (STEMI) were excluded from the analysis given that their InterTAK score would not have influenced referral for coronary angiography.

Variables included in the InterTAK score are female sex (points = 25), presence of emotional trigger (points = 24), presence of physical trigger (points = 13), absence of ST-segment depression (points = 12), presence of underlying psychiatric disorders (points = 11), presence of underlying neurologic disorders (points = 9), and presence of QTc prolongation (points = 6).³

Emotional triggers were defined as events identified by patient self-report as precipitating the index event, abstracted from notes in the medical record and categorized according to the most common factors identified by Templin et al.¹ Physical triggers were also abstracted from clinical documentation in the medical record. Presenting ECGs were extracted and analyzed for QTc duration and presence or absence of ST depression. Prolonged QTc was defined as ≥460 milliseconds in this female cohort. Comorbid psychiatric or neurologic conditions were identified based on medical history as entered in the medical record.

Clinical, angiographic, and echocardiographic data were reviewed to confirm discharge diagnoses. A discharge diagnosis of provisional TTS was defined based on the modified Mayo Clinic criteria, which was the definition used by the InterTAK registry group³: (1) ECG abnormalities and elevated troponin, (2) LV wall motion abnormalities extending beyond a single epicardial territory, (3) absence of angiographically apparent plaque rupture or causative obstructive coronary disease, and (4) absence of known myocarditis/pheochromocytoma. Other discharge diagnoses included MI with obstructive coronary disease (>50% at angiography), MI with nonobstructive coronary artery (MINOCA) disease as defined by American Heart Association criteria,⁵ and spontaneous coronary artery dissection (SCAD) identified on angiography.⁶

Continuous measures were represented as means and standard deviations when following a normal distribution and as medians and interquartile ranges (IQRs) when following non-normal distributions. Continuous variables were compared using two-sample t-tests when following a normal distribution and nonparametric Wilcoxon rank-sum tests when following a non-normal distribution. Dichotomous variables were compared using chi-square tests. Sensitivity and specificity of the InterTAK score were calculated based on a discharge diagnosis of TTS using a threshold score of >70. A receiver operating characteristic (ROC) curve analysis with a 95% confidence interval (CI) was performed to demonstrate the predictive performance of the InterTAK score for TTS.

Results

Among 375 consecutive female patients presenting with a provisional diagnosis of NSTEMI, the median InterTAK score was 43 (IQR = 37–54). An InterTAK score ≤70 points, indicating low-to-intermediate likelihood of TTS, was present in 360 of 375 patients (96%). Clinical characteristics of patients with low/intermediate scores and those with high probability scores (>70) are given in Table 1.

Table 1.

Clinical Characteristics of Female Non-ST Segment Elevation Myocardial Infarction Patients Based on International Takotsubo Registry Score

	Low/intermediate probability of TTS (≤70 points), N = 360	High probability of TTS (>70 points), N = 15	p
Demographics
Age, median (IQR)	68 (60–78)	62 (57–75)	0.34
Race, n (%)			0.86
White	176 (49)	7 (47)
Black	47 (13)	1 (7)
Asian	20 (6)	1 (7)
Other/unknown	117 (33)	6 (40)
Ethnicity, n (%)			0.72
Hispanic	80 (22)	4 (27)
Non-Hispanic	228 (63)	8 (53)
Unknown	52 (14)	3 (20)
Comorbidities, n (%)
Hypertension	275 (76)	5 (33)	0.001
Hyperlipidemia	226 (63)	7 (47)	0.28
Diabetes mellitus	155 (43)	3 (20)	0.11
Prior myocardial infarction	68 (19)	4 (27)	0.50
Smoker	132 (37)	8 (53)	0.28
Chronic kidney disease or end stage renal disease	47 (13)	3 (20)	0.43
Psychiatric disorders	58 (16)	13 (87)	<0.001
Neurologic disorders	59 (16)	2 (13)	1.00
Trigger, n (%)
Emotional	16 (4)	14 (93)	<0.001
Physical	84 (23)	3 (20)	1.00
None	260 (72)	1 (7)	<0.001
Symptoms on admission, n (%)
Chest pain	236 (66)	9 (60)	0.78
Dyspnea	167 (46)	6 (40)	0.79
Nausea/vomiting/abdominal pain	105 (29)	5 (33)	0.77
Syncope	29 (8)	1 (7)	1.00
Asymptomatic	7 (2)	0 (0)	1.00
Laboratory data
Peak troponin, median (IQR)	1.91 (0.44–7.51)	2.63 (0.89–4.60)	0.29
ECG and echo data
ST depression not present, n (%)	270 (76)	14 (93)	0.13
QTc (milliseconds), median (IQR)	466 (444–487)	476 (465–490)	0.17
LVEF (%), median (IQR)	55 (35–65)	40 (30–58)	0.43
Cath data, n (%)
PCI/revascularization performed	136 (38)	2 (13)	0.06
Final diagnosis at discharge, n (%)			<0.001
MI-CAD	221 (61)	2 (13)	<0.001
MINOCA	69 (19)	0 (0)	0.08
Takotsubo	38 (11)	10 (67)	<0.001
Type 2 MI	12 (3)	1 (7)	0.42
SCAD	4 (1)	2 (13)	0.02

Some patients had multiple symptoms.

IQR, interquartile range; MI, myocardial infarction; MI-CAD, myocardial infarction with obstructive coronary artery disease; MINOCA, myocardial infarction with non-obstructive coronary arteries; PCI, percutaneous coronary intervention; SCAD, spontaneous coronary artery dissection; TTS, takotsubo syndrome.

There was no significant difference in age, race, or ethnicity between InterTAK score groups. Patients with a score >70 were less likely to have hypertension (33% vs. 76%, p = 0.001) and were more likely to have history of psychiatric disorders (87% vs. 16%, p < 0.001) and emotional triggers precipitating their event (93% vs. 4%, p < 0.001) than those with an InterTAK score ≤70. There were no differences between the groups in prevalence of other cardiovascular risk factors, symptoms on admission, peak troponin, left ventricular ejection fraction (LVEF), or ECG parameters including ST depressions and QTc interval. Women with NSTEMI and a high InterTAK score tended to be less likely to undergo coronary revascularization than women with NSTEMI and a low InterTAK score (13% vs. 38%, p = 0.06).

After echocardiography and coronary angiography, 223 (59%) women were diagnosed with type 1 MI secondary to obstructive coronary artery disease (MI-CAD), 69 (18%) were diagnosed with MINOCAs, 48 (13%) were assigned a provisional TTS diagnosis, 13 (3%) were diagnosed with type 2 MI secondary to supply–demand mismatch, and 6 (2%) had angiographic diagnosis of SCAD. Women with an InterTAK score >70 were more likely to be diagnosed with TTS (67% vs. 11%, p < 0.001) or SCAD (13% vs. 1%, p = 0.02) and less likely to have a discharge diagnosis of MI-CAD (13% vs. 61%, p < 0.001) than women with low or intermediate InterTAK scores (Table 1).

The ROC curve for the InterTAK score for the discharge diagnosis of TTS is given in Figure 1. The area under the ROC curve (AUC) of the InterTAK score was 0.82 (95% CI, 0.75–0.88), with 21% sensitivity and 98% specificity at a threshold of >70. The percentage of patients with a diagnosis of TTS for categories of InterTAK scores is given in Figure 2.

FIG. 1.

ROC curve for the InterTAK score in the diagnosis of TTS within a population of 375 consecutive female NSTEMI patients at two academic medical centers in New York City. The AUC of the InterTAK score was 0.82 (95% CI, 0.75–0.88), with 21% sensitivity and 98% specificity at a threshold of >70. AUC, area under receiver operating curve; NSTEMI, non-ST segment elevation myocardial infarction; CI, confidence interval; InterTAK, International Takotsubo Registry; TTS, takotsubo syndrome.

FIG. 2.

Percentage of patients with a diagnosis of TTS for categories of InterTAK scores within a population of 375 consecutive female NSTEMI patients at two academic medical centers in New York City. Total number of patients (takotsubo and others) in each category and percentage diagnosed with takotsubo are annotated on top of each bar. One patient had an InterTAK score between 81 and 90.

Among the 15 patients (4%) who had an InterTAK score >70, 10 had apical, midventricular, or basal wall motion abnormalities by echocardiography and therefore would have been recommended for a noninvasive strategy by ESC guidelines. Of these 10 patients, 7 (70%) had a discharge diagnosis of TTS, 2 (20%) had a discharge diagnosis of SCAD, and 1 (%) was found to have obstructive disease on angiography requiring revascularization.

A total of 48 patients (13%) were given a provisional diagnosis of TTS. Clinical characteristics of patients with provisional TTS with low or intermediate as compared with those with high InterTAK scores are given in Table 2. Patients with a score >70 were more likely to have an underlying psychiatric disorder (90% vs. 21%, p < 0.001) and more likely to present following an emotional trigger (90% vs. 11%, p < 0.001). There were no differences in age, race/ethnicity, symptoms on admission, ECG findings, LVEF or peak troponin between those with low or intermediate versus high InterTAK scores. No patient given a provisional diagnosis of TTS underwent coronary revascularization.

Table 2.

Clinical Characteristics of Takotsubo Patients Based on International Takotsubo Registry Score

	Low/intermediate probability of TTS (≤70 points), N = 38	High probability of TTS (>70 points), N = 10	p
Demographics
Age, median (IQR)	71 (65–76)	61 (56–76)	0.72
Race, n (%)			0.52
White	26 (68)	6 (60)
Black	1 (3)	1 (10)
Asian	1 (3)	1 (10)
Other/unknown	10 (26)	2 (20)
Ethnicity, n (%)			0.65
Hispanic	9 (24)	4 (40)
Non-Hispanic	27 (71)	6 (60)
Unknown	2 (5)	0 (0)
Comorbidities, n (%)
Hypertension	17 (45)	4 (40)	1.00
Hyperlipidemia	22 (58)	5 (50)	0.73
Diabetes mellitus	6 (16)	2 (20)	0.67
Prior myocardial infarction	1 (3)	2 (20)	0.11
Smoker	14 (37)	7 (70)	0.08
Chronic kidney disease or end-stage renal disease	0 (0)	2 (20)	0.04
Psychiatric disorders	8 (21)	9 (90)	<0.001
Neurologic disorders	8 (21)	1(10)	0.66
Trigger, n (%)
Emotional	4 (11)	9 (90)	<0.001
Physical	19 (50)	3 (30)	0.31
Unknown	15 (39)	0 (0)	0.02
Symptoms on admission, n (%)
Chest pain	21 (55)	5 (50)	1.00
Dyspnea	10 (26)	4 (40)	0.45
Nausea/vomiting/abdominal pain	17 (45)	3 (30)	0.49
Syncope	5 (13)	1 (10)	1.00
Asymptomatic	2 (5)	0 (0)	1.00
Laboratory data
Peak troponin, median (IQR)	2.05 (1.21–4.12)	1.59 (0.44–4.28)	0.72
ECG and echo data
ST depression not present, n (%)	37 (97)	9 (90)	0.38
QTc (milliseconds), median (IQR)	487 (459–505)	476 (473–493)	0.50
LVEF (%), median (IQR)	35 (30–35)	33 (30–51)	0.44
Apical WMA	34 (89)	8 (80)	0.59
Cath data, n (%)
PCI/revascularization performed	0 (0)	0 (0)

Some patients had multiple symptoms.

WMA, wall motion abnormality.

Discussion

In this study of 375 consecutive women with NSTEMI who were referred for cardiac catheterization, 13% had TTS. Women with NSTEMI who had an InterTAK score >70 were more likely to have TTS and less likely to be diagnosed with MI-CAD than patients with a score <70 and specificity of this score threshold was 98%. However, a high score was rare (4%).

Ten of our patients had both a high InterTAK score and apical wall motion abnormalities, and thus would have been recommended for CCTA according to the ESC statement on management of TTS. Of these patients, one had obstructive CAD requiring percutaneous coronary intervention, and this patient would have had a delay in intervention if the ESC guidelines had been applied. This example serves as a reminder to exercise caution when applying the InterTAK score as it has the potential to bias providers against prompt revascularization in female patients presenting with ACS.

Women scoring >70 were also more likely to be diagnosed with SCAD, likely because patients with SCAD are predominantly women and more likely to present following an emotional trigger,⁷ characteristics included in the InterTAK score. Although patients with SCAD are not recommended for revascularization, a delay in diagnosis of SCAD could still lead to harm if patients are started on therapeutic anticoagulation before invasive angiography for a presumptive diagnosis of a type 1 MI.

To our knowledge, this is the first study to report the incidence of TTS specifically in a consecutive sample of female NSTEMI patients (13%). The relatively high prevalence of TTS in the current analysis may reflect increased awareness of the syndrome at our centers, and/or a population at greater risk for TTS.^8,9 As compared with female NSTEMI patients not diagnosed with TTS, those with a provisional diagnosis of TTS were less likely to have cardiovascular risk factors such as hypertension, diabetes, prior MIs, and CKD.

The inclusion of consecutive women with MI at two tertiary care centers provides important real-world validation of the InterTAK score. In our sample, the InterTAK score had a sensitivity of 21% for the diagnosis of TTS based on a threshold score >70, significantly lower than the published sensitivity of 94.7% at scores ≥50.³ Furthermore, patients with a high InterTAK score had a 30% chance of a diagnosis other than TTS, including MI-CAD and SCAD.

The difference in sensitivity of the InterTAK score between the derivation population and our study is likely because of the differences in the respective cohorts; the present analysis was restricted to women with NSTEMI and is smaller, whereas the InterTAK population consisted of a 2:1 ratio of matched patients with ACS to TTS.³ As >50% of ACS patients and >40% of TTS patients presented with ST segment elevations in the derivation cohort,³ the restriction of our analysis to NSTEMI patients likely explains a large part of the discrepancy in sensitivity that we observed. STEMI patients were not included in this study, as the InterTAK score would not have influenced their referral for invasive coronary angiography.

Although our population was restricted to women, the InterTAK population was 94.5% women,³ and as such this difference is unlikely to underlie the discrepancy in sensitivity that was observed. Theoretically, some patients diagnosed with TTS in InterTAK may have had SCAD.⁷ In a separate validation study conducted at a single center in Poland, an InterTAK score threshold of 45 points was associated with 75% sensitivity for TTS, but the study included men and patients with STEMI, and only 20 patients with a final diagnosis of TTS were included.¹⁰

Several limitations should be acknowledged. This was a retrospective study, potentially limited by observer bias. Data were extracted from clinical assessments in the electronic medical record, and are therefore limited by the quality of the provider documentation. This limitation holds particularly true for self-reported variables such as emotional and physical triggers. MRI was not routinely performed to exclude MI in patients with suspected TTS.

Although the sample size is relatively small and only 4% of our cohort had a score >70, we included all consecutive women with NSTEMI at two large academic medical centers. Only patients who underwent invasive coronary angiography were included in the present analysis, potentially leading to selection bias. These same criteria were applied to the derivation cohort, and were necessary to diagnose TTS according to the modified Mayo Clinic criteria³; as such, this limitation does not explain the difference we observed in the sensitivity of the InterTAK score.

Conclusion

In summary, among women with NSTEMI, a small proportion of women could experience delays to coronary revascularization for acute MI when the InterTAK score is used according to ESC recommendations to select patients for CCTA rather than invasive coronary angiography. The InterTAK score should be used to aid risk stratification for TTS in conjunction with other clinical variables.

Authors Contributions

T.A.: Data curation, formal analysis, investigation, writing—Original draft, writing—review and editing. A.H.: Conceptualization, methodology, data curation, formal analysis, investigation, supervision, validation, writing—original draft, writing—review and editing. N.R.S.: Conceptualization, methodology, writing—review and editing. L.B.: Investigation, writing—review and editing. M.A.: investigation, writing—review and editing. C.O.: Investigation, writing—review and editing. S.P.: Investigation, writing—review and editing. H.R.R.: Conceptualization, methodology, supervision, writing—review and editing, project administration.

Data Availability

The dataset used for this study are available from the corresponding author on reasonable request.

Author Disclosure Statement

The authors declare that they have no competing interests related to this study. Dr. Reynolds has received in-kind support from BioTelemetry, Inc., for an unrelated study in takotsubo syndrome, and in-kind support from Abbott Vascular and Siemens for an unrelated study in MINOCA.

Funding Information

This work is supported by AHA SFRN Grant No. 16SFRN27810006.