The analyses were based on data from the national multicenter RETAKO (Registry on Takotsubo Syndrome) trial, supported by the Ischemic Heart Disease and Acute Cardiovascular Care Section of the Spanish Society of Cardiology. It is a partially retrospective and prospective (from January 1, 2012, onwards), voluntary, observational study that enrolled TTS patients from 38 centers in Spain. Data were collected at each hospital, electronically encrypted and anonymized, and transferred online to a central database. Its rationale and design have been previously described.9,10

The main inclusion criteria required a definitive TTS diagnosis based on the modified Mayo Clinic criteria11: (1) transient left ventricular dysfunction with apical, midventricular, or basal segmental alterations extending beyond the territory supplied by a single coronary artery; (2) angiographic absence of significant obstructive coronary disease (luminal narrowing >50%) or a complicated (ruptured/thrombosed) atheroma; (3) new electrocardiographic changes (ST-segment elevation and/or negative T waves) and moderate elevation of cardiac troponins. Complete normalization of wall motion abnormalities and left ventricular ejection fraction (LVEF) was required, except in cases that died before complete normalization of LVEF.

Our cohort (N=970) was obtained from 1095 patients with final diagnosis of TTS admitted between January 1, 2003, and July 31, 2018 (see Figure 1). We excluded 19 (1.7%) patients who died in the hospital; 14 (1.3%) with missing mortality data; 24 (2.2%) with pheochromocytoma; 21 (1.9%) with left ventricular outflow tract obstruction (LVOTO), defined as a resting peak instantaneous gradient >25 mmHg in absence of inotropics; 27 (2.5%) with sustained ventricular arrhythmias; and 20 (1.8%) with baseline heart rate ≤45 bpm or high degree atrial-ventricular heart block. For the present study, beta-blocker use was determined according to whether eligible patients had received beta-blockers on discharge from the hospital.

Figure 1.

Flow chart of study population. AV: atrioventricular; LVOTO: left ventricular outflow tract obstruction.

(0.33MB).

The study complied with the Declaration of Helsinki and was approved by the Institutional Ethics Committee of Hospital Clínico San Carlos (Madrid, Spain). All patients provided written informed consent.

All the possible triggers in each patient were collected. Patients were categorized into groups according to the types of preceding stressors – emotional stress, physical stress, and no identifiable trigger – based on the recently proposed classification.12 The pattern of LV dysfunction based on angiography or echocardiographic evaluation was classified as apical and non-apical types.8 Apical type is characterized by hypokinesia, akinesia or dyskinesia of midventricular and apical parts of the anterior, septal, inferior and lateral walls of the left ventricle, associated with hyperkinesia of basal segments. Non-apical pattern includes all the remaining types (basal, midventricular, and focal types).

Follow-up after hospital discharge was conducted either by outpatient clinical visits or structured telephone interviews with patients or relatives. Follow-up outcomes were pre-specified and defined as the first non-fatal TTS recurrence or the occurrence of all-cause death. Recurrence was considered if a documented new non-fatal episode met the diagnostic criteria for TTS.11 If a pre-specified outcome was observed, review of electronic medical records and consensus of two experienced local investigators were mandatory for event adjudication. Patients were censored if they did not experience the outcome of interest or were lost to follow-up.

Baseline characteristics according to treatment with beta-blockers were described by using number and percentage for categorical data and mean±standard difference for continuous data. Differences in characteristics were assessed by using Chi-square tests and two-sample Student t-tests, for categorial and continuous variables, respectively.

The association between beta-blocker therapy and survival free of TTS recurrence was studied in robust Cox regression analysis. Multivariable model risk adjustments were performed with all variables associated with the combined outcome of post-discharge mortality or TTS recurrence based on clinical plausibility or p-value <0.1 in the univariate Cox analyses (Table 1 of the supplementary data). Estimates were expressed as hazard ratios (HRs) with their 95% confidence interval (CI). Proportionality of risk assumption of survival models was evaluated by plotting Schoenfeld residuals or by studying the interaction of covariates with time for the Cox proportional hazard models.

Given the substantial differences in key baseline characteristics depending on prescription of beta-blocker therapy, we performed a sensitivity analysis using a propensity score (PS) analysis. PSs were estimated using a non-parsimonious multivariable logistic regression model, with beta-blocker therapy as the dependent variable and those characteristics that differed between patients treated and not treated with beta-blockers (Table 1) as covariates. Survival time inverse probability weighting (IPW) PS analysis was used to evaluate the association between beta-blocker therapy and survival free of TTS recurrence. The effect of beta-blockers was graphically represented in Kaplan-Meier curves adjusted by IPW, to balance the covariate distribution between the treatment and control observations.13 We complemented the analysis with an augmented IPW (AIPW) to estimate the average treatment effect (ATE) of beta-blockers in one-year survival free of TTS recurrence, using a doubly robust method that combined both the properties of the regression-based estimator and the IPW estimator.14 Subsequent PS matching (PSM) was performed to assemble a cohort in which all the measured covariates would be well balanced across the comparator group. PSM was performed with 1:1 nearest-neighbor matching without replacement and with a caliper width of 0.2 of the standard deviation of all PSs. Standard mean differences were estimated for all covariates before and after matching to assess pre-matching imbalance and post-matching balance; standard mean differences of <10% for a given covariate were considered as indicators of adequate balance (Table 2). In the PS-matched population, the reduction in mortality and/or TTS recurrence was compared using a stratified Cox regression model.

Analysis of follow-up mortality and TTS recurrence was also performed in subgroups defined according to age (<75 years or ≥75 years), sex, presence of coronary artery disease, atrial fibrillation/flutter, chronic obstructive pulmonary disease (COPD), Killip class, status of cardiogenic shock, and therapy with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers (ACEI/ARB) on discharge. Treatment-by-subgroup interactions were tested using a Cox proportional hazards model adjusted by IPW, containing the beta-blocker therapy, subgroup, and the treatment-by-subgroup interaction.

A two-tailed p<0.05 indicated statistical significance. Statistical analyses were performed using Stata 15.1 (StataCorp, College Station, Texas).

Among the 970 patients with TTS included in this study, 582 patients (60.0%) were treated with beta-blockers. There were significant differences in baseline characteristics between patients with and without beta-blocker treatment (Table 1). In particular, patients who did not receive beta-blockers were more frequently comorbid, including COPD (20.6% vs. 11.9%) and history of cancer (13.9% vs. 9.3%). Compared with patients untreated with beta-blockers, those who received beta-blockers tended to be treated with higher rates of aspirin (56.5% vs. 32.2%; p<0.001), ACEI/ARB (73.2% vs. 43.3%; p<0.001), and statins (59.3% vs. 34.0%; p<0.001).

For the entire cohort, during a mean follow-up of 2.5±3.3 years, there were 87 deaths (3.6 per 100 patients/year) and 29 TTS recurrences (1.2 per 100 patients/year). Unadjusted mortality rates were similar for patients who received beta-blockers compared with those who did not [3.3 (95% CI 2.4–4.3) vs. 4.2 (95% CI 3.1–5.7) per 100 patients/year, respectively; p=0.254)]. A similar result was found for TTS recurrence [1.4 (95% CI 0.9–2.1) vs. 0.9 (95% CI 0.4–1.8) per 100 patients/year, respectively; p=0.278)]. After multivariate adjustment (including age, sex, hypertension, diabetes mellitus, smoking, coronary artery disease, prior stroke, prior beta-blocker therapy, COPD, history of cancer, Killip≥II on admission, atrial fibrillation/flutter, creatinine ≥1.5 mg/dl, apical vs. non-apical TTS, emotional or physical trigger, and medication at discharge – aspirin, P2Y12 inhibitors, oral anticoagulation, ACEI/ARB, calcium channel blockers, diuretics, statins, anxiolytics, and antidepressants), there were no significant differences in the composite outcome of mortality or TTS recurrence between patients treated and untreated with beta-blockers (adjusted HR 0.95, 95% CI 0.57–1.53; p=0.794).

For the balanced PS analysis, 566 patients were removed (380 from the beta-blocker group and 186 from the non-beta-blocker group), leaving 404 patients (202 in each group). Overlap assumption assessment and balance checks were conducted and the results are summarized in Table 2 of the supplementary data. The area under the curve for the PS model was 0.87 (95% CI 0.85–0.89), which indicated a good discrimination for the model. Beta-blocker therapy was not associated with significant differences in dmortality or TTS recurrence neither after adjusting by IPW (Figure 2) nor after PS matching (HR 0.85, 95% CI 0.46–1.60, p=0.621). After weighting and adjustment, differences in one-year survival free of TTS recurrence between patients treated and untreated with beta-blockers were not found (average treatment effect (ATE) 95% −0.01 95% CI −0.07 to 0.04; p=0.621) (Table 2).

Figure 2.

Kaplan-Meier curves adjusted by inverse probability weighting (IPW) propensity score analysis.

(0.11MB).

Beta-blocker therapy was similarly prescribed in the four major TTS types (Figure 3A). Lack of benefit of beta-blockers in terms of follow-up mortality and/or TTS recurrence was consistent across different TTS types (Figure 4: central illustration). In the PS-matched population, the HR for mortality or TTS recurrence of beta-blockers in patients with apical TTS was 0.68 (95% CI 0.34–1.36, p-value=0.277), and in patients with non-apical TTS was 2.62 (95% CI 0.54–12.65, p-value=0.231). Regarding to the trigger, beta-blocker therapy was most frequently prescribed in patients with TTS related to emotional stress (68.4% vs. 52.5% in TTS related to physical stress, and 57.7% in TTS without an identifiable triggering factor) (Figure 3B). However, no differences in mortality and TTS recurrence were found between patients treated and untreated with beta-blockers, neither after PSM (Figure 4: central illustration) nor after IPW (Table 2).

Figure 3.

Rate of beta-blocker prescription according to the type and trigger of Takotsubo.

(0.25MB).

Figure 4.

Prognostic impact of beta-blocker therapy in Takotsubo syndrome according to location and trigger. HR: hazard ratio.

(0.76MB).

Subgroup analysis after adjusting by IPW has not found a benefit of beta-blockers in any group (Table 3). Even in patients with COPD, beta-blockers were associated with higher rates of mortality or TTS recurrence (HR 2.42, 95% CI 1.07–5.49; p=0.034) (Table 3).

Our results should be interpreted in the context of several limitations. First, it is possible that various subclasses, doses and durations of beta-blocker therapy were used in this study. This heterogeneity of beta-blocker therapy might have affected the outcomes. However, those data were not collected in our registry. In addition, we have no data about rates of discontinuation or new prescriptions during follow-up. Second, because of its observational nature, unmeasured confounders constrain causal inference in the present study. Although propensity scoring and further adjustments were made for many additional confounders in the survival models, residual confounding is likely to happen anyway. So cautious interpretation is encouraged and, therefore, prognostic associations should be considered exploratory. Nonetheless, our results are consistent with other non-randomized data. Clearly, a randomized controlled trial is necessary to further investigate the issues surrounding beta-blocker use in TTS.