Key findings
The primary objective of this study was to determine the total burden and timing of in-hospital major arrhythmias and of the combined endpoint of ventricular arrhythmia, SCD/AD or resuscitation after successful primary PCI for transmural AMI within 12 h of symptom onset. Malignant ventricular tachyarrhythmia occurred in 2.8% of the patients, bradyarrhythmia occurred in 1.1%, PEA occurred in 2.6%, asystole occurred in 0.8% and SCD occurred in 0.3%. The combined endpoint was met in 7.4% of patients. We observed a biphasic distribution of events, with 76.7% of endpoints occurring within 96 h of symptom onset and 12.6% occurring 240–360 d after AMI. Occasional events were, however, observed throughout the hospital stay.
The secondary objective was to identify clinical predictors associated with the occurrence of the combined endpoint, ventricular arrhythmia, SCD/AD or resuscitation in the early phase of acute transmural infarction. Due to the low number of events, we were unable to reliably identify clinical predictors but observed positive associations between the combined endpoint and age, severely impaired LVEF, peak serum CK concentration, leucocytosis and coronary thrombus in the multivariable regression analysis.
Comparison with other studies
It is difficult to compare our findings with other studies on PCI, which were designed differently and had different endpoints. In some studies, more time elapsed between symptom onset and PCI than in our cohort. In other studies, the occurrence of endpoints before and during cardiac catheterization were taken into account, and the length of observation time and the differentiation of arrhythmias were not the same in all studies. The studies were, however, broadly comparable regarding study populations, concomitant medications, techniques and stents used.
The incidences of sustained VT or VF and SCD in our registry were 2.8 and 0.3%, respectively. The total burden of in-hospital malignant ventricular arrhythmias in our registry was therefore similar to or lower than that described in other reports of transmural AMI treated by primary PCI. In a similarly designed study by Giglioli et al., only episodes of VF were recorded, which occurred in 0.6% of patients after cardiac catheterization; however, the absence of reports on other endpoints makes a direct comparison with their findings difficult. In our study, the time to reperfusion was less than 4 h after symptom onset in 89.3% of the patients, and the results are probably best compared with the following two studies, which both included patients with STEMI treated with primary PCI within 6 h of symptom onset. Mehta et al. reported in the APEX-AMI trial that 2.0% of patients developed VT or VF after cardiac catheterization [
13]. Furthermore, Mehta et al. undertook an analysis of outcomes from the HORIZONS-AMI trial and reported that 5.2% of patients developed VT/VF after PCI [
14]. Only a limited comparison is possible with the following studies because the time to primary PCI after symptom onset was longer in those studies. An analysis by Ohlow of an observational registry of patients with STEMI treated with primary PCI within 24 h of symptom onset revealed that the incidence of malignant arrhythmia was 4.7%; however, the investigators did not state where the arrhythmias occurred, and they observed endpoints only during the CCU stay [
12]. A single-centre retrospective cohort study of patients with STEMI treated with primary PCI within 24 h of symptom onset undertaken by Cricri and colleagues reported a comparable number of patients (2.6%) who developed VT or VF after cardiac catheterization [
11].
There are limited data on the potential therapeutic benefit of primary PCI compared to thrombolysis in terms of the incidence of in-hospital malignant arrhythmias in patients with acute STEMI. The incidence of malignant ventricular arrhythmia in our cohort was less than the VF or sustained VT incidence of 10.2% reported in the GUSTO-I study, a large randomized clinical trial investigating thrombolysis with streptokinase in patients with STEMI within 6 h of symptom onset [
15]. This observation corroborates the hypothesis of PCI being superior to thrombolysis.
We observed a predominantly biphasic distribution of composite endpoint events, with 76.7% occurring within 96 h of symptom onset and 12.6% occurring between 240 and 360 h. This biphasic pattern differs from the more monophasic distribution observed in the thrombolysis era as well as in studies of major arrhythmias after successful primary angioplasty for acute STEMI. In the GUSTO-1 thrombolysis trial, 39 and 55% of in-hospital deaths occurred within 24 and 48 h of randomization, respectively, while 84% of malignant arrhythmias occurred within 48 h of randomization [
16]. In study settings similar to ours with patients treated within 6 h of symptom onset, Mehta and colleagues found, in retrospective analyses of the APEX-AMI study population and in the prospective HORIZONS-AMI trial, that 70 and 85%, respectively, of VT-associated fatal events occurred within the first 48 h of leaving the catheterization laboratory [
13,
14]. In the two studies that included patients with STEMI treated with primary PCI within 24 h of symptom onset, a different temporal distribution was observed. In the study of Cricri and colleagues, most of the malignant arrhythmias (sustained VT, VF or bradycardia necessitating cardiac pacing) developed in the catheterization laboratory, and nearly all of these arrhythmias occurred within 24 h [
11]. Ohlow and colleagues reported 90% of VTs occurring within the first 48 h [
12].
Our secondary objective was to identify clinical predictors associated with the occurrence of the combined endpoint of VT, SCD or arrhythmic death, and resuscitation in the early phase of acute transmural infarction. These predictors would a) identify patients at high risk for the combined endpoint at the time of hospitalization and b) identify patients at risk despite the apparent lack of established risk factors, e.g., cardiogenic shock.
The variables used for our logistic regression modelling were based on observations from prior studies of risk stratification and include patient demographic and clinical characteristics, measures of the acuity and angiographic presentation of the MI, and indicators of the type and extent of myocardial ischaemia and necrosis [
4,
5]. Our results indicate that age, severely impaired LVEF, peak serum CK concentration, leucocytosis and the presence of coronary thrombus were positively associated with the combined endpoint. In patients treated by primary PCI in the APEX-AMI trial as well as those in the study of Ohlow and colleagues, a post-procedural TIMI flow of less than grade 3 was associated with VT or VF [
12,
13]. In the APEX-AMI trial, leucocytosis was also a predictor of ventricular arrhythmia [
13]. A similar observation was also made by Rahimi et al. in patients with NSTEMI [
17].
In several studies from the thrombolysis era, age, severely impaired LVEF and peak serum CK concentration have also been consistently associated with a higher incidence of VF or VT during or immediately after AMI. An analysis of the Holter Registry data from the Cardiac Arrhythmia Suppression Trials showed that age and reduced LVEF were independent predictors of the incidence and frequency of VT [
18]. The analysis by Ruiz-Bailén and colleagues of the ARIAM Database also showed that age and peak CK concentration were associated with VF [
19]
. In the study by Mont and colleagues of patients with AMI who were referred to a CCU after thrombolysis, serum CK-MB fraction concentration, Killip class and bifascicular block were independent predictors of the development of sustained monomorphic tachycardia [
20].
We also reported on secondary outcomes and found that in our ‘real-world’ single-centre registry of patients with acute transmural infarction treated with primary PCI, total in-hospital mortality was 6.5%, which is consistent with other reports. In similarly designed retrospective single-centre studies by Giglioli et al. and Kozieradzka et al., in-hospital mortality was 5.9% and 30-d mortality was 6.3% [
4,
10]. A more recent small single-centre study in China showed that mortality was 8.6% in patients aged > 60 years compared with 1.5% in the non-elderly group [
21]. In our cohort, 15% of patients with sustained VT and 93% of patients with VF died, resulting in a mortality rate among those who developed a ventricular arrhythmia twice that in a retrospective cohort study of 2317 patients with AMI reported by Henkel et al. (mortality rate 38%) and the APEX-AMI study (mortality rate 33%) [
1,
13]. The incidence of bradyarrhythmia in our cohort was also lower than that in other reports; specifically, only 1.3% of the patients developed bradyarrhythmia (with 0.8% of the cases being CHB), while Giglioli et al. reported an incidence of 6.3% [
10].
Possible mechanisms and explanations
Our observation of a biphasic temporal distribution of the combined endpoint can be explained by the nature of our chosen endpoint, which comprised episodes of all major ventricular arrhythmias, SCD/AD and resuscitation not only during the initial phase when patients were continuously monitored in the CCU but also during the entire hospital stay. In addition, we did not consider any events that occurred before or during cardiac catheterization [
9,
11]. Other potential explanations are that the time to PCI was longer (up to 24 h) in other studies [
11,
12], and thus, myocardial necrosis may have been more pronounced in these studies. Furthermore, our population was unselected, unlike trials of study drugs/drug-eluting stents, such as the APEX-AMI and HORIZONS-AMI trials, in which some potential participants were excluded [
13,
14].
The multivariate analysis identified variables associated with the composite endpoint that differed from those of other studies, potentially because our endpoint included all ventricular arrhythmias, SCD/AD and resuscitation episodes, while other studies used only ventricular arrhythmias and CHB as the endpoint. Furthermore, leucocytosis, elevated CK concentration and severely impaired LVEF would not have been a consequence of staged or advanced infarctions in our cohort, as we only included patients with < 12 h of symptoms in whom necrosis and reactive inflammation would not have become established. In addition, 92.4% of our patients underwent primary PCI within 4 h of symptom onset.
It is challenging to explain the relatively high mortality rate of those who developed arrhythmia in our cohort. We cannot conclude that adverse outcomes were due to cardiogenic shock and VF alone, as > 40% of the patients in our cohort who died after an episode of VF did not exhibit symptoms or signs of low cardiac output.
Barron et al. concluded in the retrospective analyses of a TIMI 10 thrombolysis study that an elevated WBC was associated with reduced epicardial blood flow and myocardial perfusion, thromboresistance (arteries open later and have a greater thrombus burden), and a higher incidence of new congestive heart failure and death [
22]. Our data therefore seem to suggest that inflammation and the WBC itself may also be directly correlated with coronary thrombosis, impaired perfusion, and reperfusion injury in the PCI era.
Study strengths and limitations
Our cohort comprised 975 multi-ethnic patients with a clearly defined pathophysiological substrate (transmural AMI) and therapy (only primary PCI for reperfusion of the occluded vessel). In this cohort, we found a lower incidence of in-hospital major arrhythmia, SCD/AD and resuscitation but a higher mortality rate and biphasic temporal distribution of those who met the composite endpoint. Multivariable regression analysis showed positive associations between several factors and the combined endpoint. Due to the low number of events, we were unable to develop and calculate a risk score for the occurrence of the combined endpoint. The low number of events may be a consequence of the retrospective nature of this study.
A major concern may relate to the period in which data were collected (2005–2011) and the procedural aspects, i.e., the P2Y12 inhibition provided (Clopidogrel) and the very low percentage of implanted DES during primary PCI (12.6%) in our study. In the EUROMAX trial, the choice of prasugrel or ticagrelor over clopidogrel was not associated with differences in acute stent thrombosis or 30-day ischaemic outcomes after PCI [
23]. Furthermore, in the PRAGUE-18 study, prasugrel and ticagrelor were found to be similarly effective during the first year after MI, and economically motivated early post-discharge switches to clopidogrel were not associated with an increased risk of ischaemic events [
24]. The low percentage of implanted DES (all first generation) in our study should not be of concern because a meta-analysis [
25], cost analysis data [
26], cohort registries [
27] as well as single-centre studies [
28] have shown that the only benefit of DES is the reduction of target vessel revascularization. Total mortality and MACE or stent thrombosis are not superior in patients with STEMI receiving bare metal stents (BMS), although a trend toward lower mortality may be seen with everolimus-eluting stents (EES) [
29]. Furthermore, in the Norwegian Coronary Stent (NORSTENT) trial, patients (26% with STEMI) were randomized to the DES or BMS group. There were no differences in the incidence of the primary endpoint (a composite of death from any cause or non-fatal spontaneous MI) after a median follow-up of 5 years [
30].
Another limitation is the potential lack of generalizability of this single-centre study, which may not reflect outcomes achieved by other teams in different settings. Furthermore, observational studies may generate only a hypothesis that remains to be proven in a randomized controlled trial [
31].