We conducted a single-center retrospective study enrolling consecutive patients admitted to the Cardiovascular Department of Fondazione Policlinico Universitario “Agostino Gemelli” IRCCS between 2012 and 2018 with a diagnosis of tako-tsubo syndrome according to international tako-tsubo diagnostic criteria.1 The exclusion criteria included: patients with prior myocardial infarction, a symptom-onset-to-admission time greater than 12hours, the pre-existence of structural cardiomyopathy (such as hypertrophic, dilated, arrhythmogenic, noncompaction) or any clinically relevant heart valve disease, patients with an implanted pacemaker, patients chronically treated with drugs affecting the QT interval, or any type of congenital long QT syndrome or other cardiac channelopathies.

Based on the occurrence of subacute VAs on continuous electrocardiogram (ECG) telemetry during hospitalization, our population was classified as follows: a) VAs group, and b) non-VAs group. VAs events included any of the following: VF, polymorphic VT, sustained monomorphic VT (life-threatening VAs), non-sustained VT, and premature ventricular contractions ≥ 2000 within a 24-hour window of telemetry monitoring (non-life-threatening VAs), as defined elsewhere.16 Subacute VAs were distinguished from those recorded during the first 48hours from admission (ie, acute VAs).

For each patient, medical records were reviewed to collect clinical characteristics, medications, blood testing data, ECG data, and echocardiography parameters on admission. Blood testing and ECG findings were also obtained 48hours after admission, in the subacute phase. Coronary angiography and ventriculography data were also obtained. The study flow chart is shown in figure 1.

Figure 1.

Flow chart of the study. ECG, electrocardiogram; VAs, ventricular arrhythmias.

(0.29MB).

The study protocol was approved by the institutional review board and ethics committee of the Fondazione Policlinico Universitario “Agostino Gemelli” IRCCS. The study complies with the ethical guidelines of the Declaration of Helsinki. All patients gave written informed consent.

A standard resting 12-lead ECG (25mm/s, 1 mV/10 mm) was recorded in all patients on admission and 48hours thereafter using an EL 280 Resting Electrocardiograph (Welch Allyn, USA). ECG traces were used to assess cardiac rhythm, atrioventricular and intraventricular conductions, any ST-segment deviation, and ventricular repolarization. Two blinded cardiologists manually annotated all 12-lead ECG measurements. The intraclass correlation coefficient was calculated for each parameter. ECG data were averaged using 5 consecutive beats during sinus rhythm, or 10 consecutive beats during atrial fibrillation.

ST-segment elevation was defined as an upward shift ≥ 0.1mV (except in V2–V3 where a ≥ 0.15mV was considered) from the J point in ≥ 2 contiguous leads. ST-segment depression was defined as a downward shift from the J point in ≥ 2 contiguous leads of at least−0.05mV in leads V2 and V3 and−0.1mV in all other leads. T-wave inversion was defined as a T-wave amplitude less than−0.1mV. If the J point was not easily identifiable, the subsequent isoelectric line between the T-wave and P-wave was conventionally used as a reference marker.17 Persistent ST-elevation was defined as ST-segment elevation at 48hours on ECG with ≤ 50% resolution from the reference ECG on admission.

QT intervals were assessed in each of the 12 leads, measured from the beginning of the QRS complex to the end of the T-wave at the isoelectric line for each patient.17 T-waves smaller than 1mm in amplitude were not measured due to intrinsic uncertainty. The QT dispersion for each standard ECG recording was also computed as the difference between the longest and shortest QT interval.17 Furthermore, the JT peak intervals were measured in all the 12 leads from the end of the QRS complex to the peak of the T-wave (if the T-wave was positive) or to the nadir of the T-wave (if the T-wave was negative or biphasic). The Tpeak-Tend interval was measured from the peak of the T-wave to the end of the T-wave. For both the JT and Tpeak-Tend intervals, we considered the mean value of the 12 leads (global), the precordial leads only (precordial) and the peripheral leads only (limb leads) (figure 2). The JT peak and the Tpeak-Tend dispersion were defined as the difference between the longest and shortest measured interval, as appropriate.18 All the repolarization intervals were corrected for heart rate using Bazett's formula, as reported elsewhere.18

Figure 2.

Electrocardiographic repolarization indices measurements. The 12 leads of the ECG are shown in sagittal and coronal plane (on the top left), with a schematic representation of repolarization intervals (on the bottom left) and their measurements in the 12 leads, with a description of what constitutes a “global”, “precordial” and “in limb leads” value.

(0.41MB).

Continuous 3-lead (aVF, V3, V5) ECG telemetry monitoring was recorded during the hospitalization period, starting after the diagnosis of the tako-tsubo syndrome and further maintained until hospital discharge. These data were used to identify and quantify VAs events.

Coronary angiography was performed in all patients to exclude the presence of any relevant coronary artery disease. Left ventriculography was also performed to characterize the Tako-tsubo variants, as reported elsewhere.1

After coronary angiography, all patients underwent a transthoracic echocardiography study with a Toshiba Artida system equipped with a 3.5MHz phased array transducer. The echocardiography exam included: quantification of left ventricular volume and left ventricular ejection fraction (LVEF); assessment of wall motion; assessment of mitral diastolic function; assessment of any valvular defect; assessment of the pericardial space; and estimation of the systolic pulmonary artery pressure, all done following current recommendations.19

All data were analyzed for normality of distribution using the Kolmogorov-Smirnov test. Continuous variables are expressed as mean±standard deviation or median [interquartile range], as appropriate, and categorical variables are expressed as percentages. A parametric (T-test) or non-parametric test (Mann-Whitney U test, chi-square test) was used to assess the statistical significance of group comparisons, as appropriate. Correlation analyses were performed using the Pearson correlation coefficient method.

A multivariate binary logistic analysis was used to identify independent predictors of subacute VAs and mortality during hospitalization. Only statistically significant objective variables after univariate analysis were included in the multivariate model. Effect modification by each covariate was evaluated by testing whether including the interaction term in the multivariate logistic model significantly changed the log likelihood of the model applying stepwise logistic regression. The coefficients obtained from the logistic regression were expressed in terms of odds ratio with 95% confidence intervals.

Receiver operating characteristic curves were used to assess the sensitivity, specificity, and cut-off values of different electrocardiographic parameters for the prediction of subacute VAs. The optimal cut-off value was defined as the value yielding the maximal Youden index (max [sensitivity+specificity−1]) or the best combined sensitivity and specificity.20 Comparisons of the area under the curve with the Z test were performed using MedCalc Statistical Software v 17.2 (MedCalc Software, Belgium).

All tests were 2-sided and statistical significance was set at P<.05. The statistical analysis was performed by using the SPSS software v23.0 (IBM Corp, USA).

The study population included 87 patients with confirmed tako-tsubo syndrome (age 72±12 years old, 92% females). Baseline clinical data and laboratory findings are reported in table 1 and table 1 of the supplementary data, respectively. The average LVEF was 46±10% on admission, with an LVEF value ≤ 45% in 35 patients (40%). During the acute phase, cardiac arrhythmias occurred in 25 patients (29%): supraventricular arrhythmias (atrial fibrillation, atrial flutter, or supraventricular tachycardia) were detected in 11 patients (13%), while VA events occurred in 14 patients (16%) (1 polymorphic VT, 1 sustained monomorphic VT, 12 patients with premature ventricular contractions ≥ 2000 in 24hours).

During a median of 8 [5-14] days of hospitalization, subacute VAs (≥ 48hours from admission) were documented in 22 patients (25%) after a median of 91 [75-149] hours from admission. Life-threatening VAs occurred in 5 patients (6%) (1 VF, 1 polymorphic VT, and 3 sustained monomorphic VT) and non-life-threatening VAs were present in 17 patients (19%) (6 non-sustained VT and 11 patients with premature ventricular contractions ≥ 2000 in 24hours).

In the overall population, heart rate on admission was 88±21 bpm and significantly decreased at 48hours to 75±18 bpm (P=.001). The QTc interval was 450±48 msec, the corrected global JT peak interval was 232±40 msec, and both significantly increased at 48hours (470±38 msec, P=.040 and 251±38 msec, P=.001, respectively). The corrected global Tpeak-Tend did not significantly change between the ECG on admission and at 48hours (107±26 msec vs 103±26 msec, P=.863). All electrocardiographic measurements on admission and after 48hours are reported in table 2 of the supplementary data and table 3 of the supplementary data. The intraclass correlation coefficient was superior to 0.90 for each parameter, with a mean of 1±2 ECG leads per patient not available for the analyses due to intrinsic uncertainty. Further coronary angiography and echocardiography data are also summarized in table 4 of the supplementary data.

In-hospital mortality occurred in 6 patients: 3 related to septic shock, 1 associated with refractory VF, 1 due to cardiogenic shock, and 1 after cardiac free wall rupture (table 5 of the supplementary data). On admission, none of ECG parameters were statistically different between patients who died and those who survived during hospitalization. However, at 48hours from admission, heart rate was significantly higher in the former than in the latter (94±16 vs 72±18 bpm, respectively, P=.004, figure 1A of the supplementary data), with moderate correlation with LVEF (34%±11% in non-survivors vs 47%±10% in survivors, P=.006; r-0.373, P=.004, figure 1B,C of the supplementary data). No other ECG parameters showed significant differences between the 2 groups. Data on clinical complications during hospitalization are summarized in table 6 of the supplementary data.

ECG parameters on admission did not differ between the VAs and non-VAs groups. However, at 48hours both the QTc and corrected Tpeak-Tend interval (on global, precordial and limbs leads) showed significantly higher values in the VAs group than in the non-VAs group (all comparisons showed P<.05, figure 3A,C andtable 3 of the supplementary data). Moreover, comparisons between ECG intervals on admission and at 48hours showed that, although QTc and corrected JT peak values increased in both groups, the corrected Tpeak-Tend intervals (on global, precordial and limb leads) were only prolonged in the VAs group (all comparisons showed P<.05, figure 3 and table 2). Heart rate values significantly decreased in the non-VAs group between admission and 48 hour ECG (85±16 bpm vs 73±18 bpm, P=.001). Conversely, this decrease in heart rate was not statistically significant in the VAs group (91±28 bpm vs 83±17 bpm, P=.101). The VAs and non-VAs groups showed no differences in persistent ST-segment elevation at 48hours (7 patients [32%] and 17 [26%], respectively, P=.801). All electrocardiographic measurements and comparisons between the admission and 48-hour ECG in the VAs and non-Vas groups are reported in table 2, table 2 of the supplementary data and table 3 of the supplementary data.

Figure 3.

Differences of electrocardiographic repolarization parameters between the VAs group and non-VAs group. A: comparison between QTc on admission (in blue) and at 48hours (in red) in the VAs and non-VAs group and comparison between the 2 values at 48hours. B: same comparisons as in A using the corrected global JT peak interval. C: same comparisons as in A and B using the corrected global Tpeak-Tend interval. D: receiver operating characteristic curves of QTc and corrected global Tpeak-Tend at 48hours. E: distribution of subacute VAs according to the cut-off point of 108 msec for corrected global Tpeak-Tend at 48hours. AUC, area under the curve; QTc, corrected QT interval; VAs, ventricular arrhythmias.

(0.34MB).

Among clinical variables, acute VAs were significantly more frequent in the VAs than in the non-VAs group (11 patients [50%] vs 3 [5%]; P=.001), both considering life-threatening and non-life-threatening acute VAs (2 patients [9%] vs 0 [0%]; P=.040 and 9 patients [41%] vs 3 [5%]; P=.001, respectively). Conversely, supraventricular arrhythmias did not differ between the 2 groups during acute and subacute phases (3 patients [14%] vs 8 [12%]; P=.871 and 6 patients [27%] vs 16 [24.5%]; P=.804, respectively). The prevalence of New York Heart Association class III-IV on admission was also significantly higher in the VAs group compared with the non-VAs group (P=.003), although this was not associated with significant differences in N-terminal pro-B-type natriuretic peptide values (13 982±35 450 pg/mL vs 4847±5241 pg/mL; P=.174), LVEF (44±10% vs 47±8%, P=.789) or diastolic dysfunction (≥ 2nd degree according to current recommendations19: 9 patients out of 43 [21%] vs 14 out of 44 [32%], respectively; P=.681).

Serum potassium values on admission and at 48hours were lower in the VAs group than in the non-VAs group (3.5±0.4 mmol/L vs 3.9±0.4 mmol/L, respectively, on admission, P=.020, and 3.3±0.4 mmol/L vs 3.9±0.5 mmol/L, respectively, at 48hours, P=.003, table 1 of the supplementary data). Serum potassium levels showed a moderate negative correlation with prolongation of the corrected global Tpeak-Tend interval (r−0.374, P=.004). Further comparisons of other clinical, analytical, and imaging parameters are shown in table 1, table 1 of the supplementary data and table 4 of the supplementary data.

Univariate logistic regression analysis (table 3A) showed that New York Heart Association class III-IV on admission, the QTc interval at 48hours, corrected global Tpeak-Tend interval at 48hours and pericardial effusion on admission, were potential predictors of subacute VAs. Multivariate analysis (table 3B) further confirmed the corrected global Tpeak-Tend interval at 48hours as the only independent predictor of subacute VAs.

Receiver operating characteristic analyses of repolarization parameters showed that the corrected global Tpeak-Tend interval at 48hours provided a significantly larger area under the curve (0.740) compared with the QTc interval at 48hours (0.611) (Z test P=.040) (figure 3D). A cut-off value of 108 msec for the corrected global Tpeak-Tend interval at 48hours showed the best combination of sensitivity and specificity (71% and 72%, respectively). Subacute VAs occurred in 48% of patients with corrected global Tpeak-Tend interval>108 msec and in 24% with values<108 msec (P=.040, figure 3E).

Mortality rates were significantly higher in the VAs group than in the non-VAs group (n=4 [18%] vs n=2 [3%], respectively, P=.030). Further analysis showed no significant differences between life-threatening and non-life-threatening subacute VAs in such patients (1 patient with VF [20%] vs 3 patients with non-sustained VT [17.6%], P=1.0, respectively; see also table 5 of the supplementary data). Subacute VAs yielded an odds ratio of 7.01 after univariate logistic regression analysis (95% confidence interval, 1.1-41.3, P=.032). Acute VAs (within the first 48hours) were not significantly different between non-survivors and survivors (2 patients [33%] vs 12 [15%], P=.246, respectively).

Other relevant complications during hospitalization were similar between the VAs and non-VAs groups (table 6 of the supplementary data).