Long-term prognosis of de novo atrial fibrillation during acute myocardial infarction: the impact of anti-thrombotic treatment strategies

Abstract

Aims

While the prognosis of patients presenting with de novo atrial fibrillation (AF) during the acute phase of myocardial infarction has been controversially discussed, it seems intuitive that affected individuals have an increased risk for both thrombo-embolic events and mortality. However, profound data on long-term outcome of this highly vulnerable patient population are not available in current literature. Therefore, we aimed to investigate the impact of de novo AF and associated anti-thrombotic treatment strategies on the patient outcome from a long-term perspective.

Methods and results

Patients presenting with acute myocardial infarction, treated at the Medical University of Vienna, were enrolled within a clinical registry and screened for the development of de novo AF. After discharge, participants were followed prospectively over a median time of 8.6 years. Primary study endpoint was defined as cardiovascular mortality. Out of 1372 enrolled individuals 149 (10.9%) developed de novo AF during the acute phase of acute myocardial infarction. After a median follow-up time of 8.6 years, a total of 418 (30.5%) died due to cardiovascular causes, including 93 (62.4%) in the de novo AF subgroup. We found that de novo AF was significantly associated with long-term cardiovascular mortality with an adjusted HR of 1.45 (95% CI 1.19–2.57; P < 0.001). While patients with de novo AF were less likely to receive a triple anti-thrombotic therapy as compared to patients with pre-existing AF at time of discharge, this therapeutic approach showed a strong and inverse association with mortality in de novo AF, with an adj. HR of 0.86 (95% CI 0.45–0.92; P = 0.012).

Conclusion

De novo AF was independently associated with a poor prognosis with a 67% increased risk of long-term cardiovascular mortality. Intensified anti-thrombotic treatment in this high-risk patient population might be considered.

Introduction

The development of atrial fibrillation (AF) during the acute phase of acute myocardial infarction (AMI), also known as de novo AF (dnAF), mirrors a common complication during the acute phase with a prevalence ranging from 2% to 21%.^1,2 During the past decades, the use of modern revascularization therapies has led to a profound reduction in incidence rates of dnAF—nevertheless their prevalence and absolute number remains high within the broad population of patients presenting with coronary heart disease.³ Recent investigations were able to demonstrate that dnAF during AMI has been associated with increased in-hospital mortality and worse overall prognosis.^4–6 Despite the underlying mechanisms leading to these worse outcomes are still insufficiently understood, deterioration of coronary blood flow and ventricular function mirror a major driver for the development of fatal cardiac adverse events.^7,8 Besides the coronary ischaemic risk after AMI, also the AF-based thrombo-embolic risk represents a potential thread for patient outcome.

To prevent adverse events, the antithrombotic management of patients with AF who have undergone percutaneous coronary intervention (PCI) with stent implantation remains a major challenge in clinical practice. While oral anticoagulation (OAC) is indicated in AF patients based on an increased risk for stroke, dual antiplatelet therapy (DAPT) with a P2Y₁₂ inhibitor plus acetylsalicylic acid after stent implantation is needed to prevent ischaemic events, including stent thrombosis. Facing the dilemma that DAPT significantly reduces the risk of stent thrombosis but is inferior to OAC in the prevention of stroke and systemic embolism, clinical practice guidelines recommend to establish triple antithrombotic therapy (TAT; OAC plus DAPT) or dual anti-thrombotic therapy (DAT) as the combination of OAC with only one antiplatelet agent.^9,10

Unfortunately, data on the impact of dnAF and post-AMI treatment approaches on survival are scarce and there are no robust data in the literature available investigating both the long-term prognosis of dnAF and the associated outcomes of different anti-thrombotic treatment strategies. Moreover, there is currently no treatment recommendation available on the referral of dnAF individuals considering OAC including DAT or TAT after the first episode of AF during the acute phase of AMI. However, it can reasonably be assumed that the presence of dnAF—as a reflection of a more vulnerable patient population—has a poor long-term prognosis, which may be ameliorated by antithrombotic treatment strategies from a long-term perspective. Therefore, the aim of this study was to evaluate the prognostic value of dnAF from a long-term perspective in patients presenting with AMI and provide information on the predictive value of anti-thrombotic treatment strategies in this high-risk AF population.

Methods

Study population

A detailed protocol of the study design has already been described elsewhere.¹¹ In short, patients presenting with AMI admitted between December 1996 and December 2009 to the Vienna General Hospital, a university affiliated tertiary care centre with a high-volume cardiac catheterization unit, were included consecutive within our clinical registry. Acute myocardial infarction was defined in accordance to the guidelines of the European Society of Cardiology (ESC) as a ST-elevation myocardial infarction (STEMI) or a non-ST elevation myocardial infarction (NSTEMI).^12,13 Blood samples were taken at time of hospital-admission prior to coronary angiography and processed in accordance to local laboratory standards (Department of Laboratory Medicine, Medical University of Vienna, Vienna, Austria).

There were no specific exclusion criteria for patient enrolment. The study protocol complies with the declaration of Helsinki and was approved by the local ethics committee of the Medical University of Vienna (EK 159/2011).

Data acquisition and follow-up

Patient-relevant characteristics were assessed via the patients’ electronic medical records of the Medical University of Vienna, as well during a standardized follow-up procedure. Data assessment was performed by specially trained chart reviewers that inserted predefined patient characteristics into a record abstraction form for further analysis of the registry. Patients were followed prospectively until the primary study endpoint (=cardiovascular mortality) was reached.

Electrocardiogram (ECG) analysis of all individuals was assessed at the time of hospitalization for the acute event. Patients were continuously monitored via ECG-telemetry for at least 3 days after the acute event as a standard operating procedure at the study centre. Additionally, patients received a 12-lead surface ECG in case of symptoms during the entire hospitalization. Atrial fibrillation was defined in accordance to the guidelines of the European Society of Cardiology. A pre-existing AF was defined as a history of AF prior to the acute AMI event. De novo atrial fibrillation was defined as a new onset of atrial fibrillating impulses at the time of admission or during the hospitalization, in AF-naïve individuals. Permanent AF was defined as presence of AF at the time of discharge and paroxysmal AF was defined as one or multiple AF episode of AF during the hospitalization but free of AF at the time of discharge.⁹ Discharge letters of all participants were screened for the respective medication at the time of discharge to evaluate anti-thrombotic therapy. Since both ticagrelor and prasugrel were introduced to the local institution after patient enrolment, the present registry only pictures independently the effect of clopidogrel in addition to both vitamin-K-antagonist (VKA) and/or aspirin. Triple antithrombotic therapy was defined as the therapeutic combination of VKA and DAPT. Dual anti-thrombotic therapy was defined as the combination of single anti-platelet therapy (aspirin or clopidogrel) and VKA.

To reach the study goal, cardiovascular mortality was chosen as primary study endpoint. The patients’ cause and date of death was assessed by screening the national registry of death until January 2017 via the Austrian Registry of Death (Statistics Austria, Vienna, Austria) that was validated by the national health-insurance system. There was no relevant loss of follow-up.

Statistical analysis

Continuous data are shown as median and interquartile range (IQR) and compared using the Kruskal–Wallis test. Categorical parameters are presented as counts and percentages and analysed using the χ² test. Cox-regression hazard analysis was used to assess the influence of AF on cardiovascular mortality. Results were presented as hazard ratio (HR) and the respective 95% confidence interval (CI). Continuous variables were log-transformed prior inclusion the regression analysis. The multivariate model was adjusted for potential confounders: age, gender, body mass index, hypertension, diabetes mellitus type II, hypercholesterolaemia, smoking, positive family history of coronary artery disease, STEMI, percutaneous coronary intervention, renal function failure, heart failure, stroke, and Nt-proBNP due to their association with mortality. The Kaplan–Meier charts were plotted to graphically illustrate the impact of dnAF compared to pre-existing AF (peAF) and compared using log-rank test. Statistical significance was defined by two-sided P-values <0.05. Statistical analyses were performed using SPSS 24.0 (IBM SPSS, NY, USA). Source data of the respective analysis is available via the corresponding author on reasonable request.

Results

Detailed baseline characteristics for the entire study population (n = 1372), stratified in individuals free of AF, pre-existing AF, and de novo AF, are summarized in Table 1. In short, out of the entire study population [1372 patients; median age: 57 years (IQR 42–80); 58.8% male gender] presenting with AMI, 149 patients (10.9%) developed dnAF during the acute phase of AMI and 90 (6.5%) had a peAF. The study population covered a representative number of participants presenting with STEMI (n = 692; 50.4%).

Comparing characteristics of patients presenting with peAF and dnAF, we observed sicker patients in the dnAF group in terms of clinical presentation, cardiovascular risk profile, and laboratory measures. However, the overall thrombo-embolic risk assessed via the CHA₂DS₂-VASc score between groups was comparable [dnAF: 4 (IQR 3–5) vs. peAF: 4 (IQR 3–5) P = 0.104], while the fraction of patients presenting with chronic kidney injury (dnAF: 14.8% vs. peAF: 5.6%) and type 2 diabetes mellitus (dnAF: 28.2% vs. peAF: 13.3%) were higher in the dnAF subgroup. Similarly, we observed higher values of both creatine kinase (P = 0.031) and NT-proBNP (P < 0.001) in dnAF as compared to peAF patients.

Interestingly, individuals presenting with dnAF were less likely to receive a coronary intervention (dnAF: 50.3% vs. peAF: 75.6%) while the rate of thrombolytic therapy was balanced.

Survival analysis

After a median follow-up time of 8.6 years, corresponding to 11 617 patient years, a total of 418 [30.5%; including 25 patients (5.9%) with fatal cerebrovascular event] individuals died due to cardiovascular causes, with 16 individuals in the peAF subgroup [17.8%; including 1 patient (6.3%) with fatal cerebrovascular event] and 93 [62.4%; including 9 patients (9.6%) with fatal cerebrovascular event] in the dnAF subgroup, respectively.

We observed that dnAF proved to be a strong and direct prognosticator for long-term cardiovascular mortality with a crude HR of 3.59 (95% CI 2.83–4.54; P < 0.001). Even after a comprehensive adjustment for a large subset of confounders the predictive potential of dnAF remained stable with an adjusted HR of 1.45 (95% CI 1.19–2.57; P < 0.001). Of utmost interest, peAF proved to be inversely associated with long-term cardiovascular mortality, presenting with a crude HR of 0.71 (95% CI 0.53–0.95; P = 0.020). Within the multivariate model, peAF still presented as an independent prognosticator for favourable long-term cardiovascular mortality with an adj. HR of 0.70 (95% CI 0.35–0.98; P = 0.043; see Table 2).

The Kaplan–Meier survival plot and log-rank test clearly highlighted that the long-term cardiovascular mortality was significantly higher for fatal cardiovascular events in the dnAF group as compared to patients with peAF and the AF-free population (P < 0.001; see Figure 1).

Anti-thrombotic treatment strategies

Considering anti-thrombotic treatment strategies, we found that patients with peAF were more likely to receive any oral anticoagulation (P < 0.001). Interestingly, 6.1% of patients presenting with dnAF received DAPT in addition to short-term therapy with low-molecular weight heparin, while this approach was not observed within the peAF group. Most importantly, the fraction of patients that received TAT at the time of discharge was significantly lower in the dnAF subgroup (peAF: 65.6% vs. dnAF: 37.6%). While the fraction of individuals receiving DAT was balanced between groups (peAF: 13.3% vs. dnAF: 14.0%), the fraction of patients receiving DAPT at the time of discharge was higher in the dnAF group (peAF: 15.5% vs. dnAF: 21.5%) (see Table 1). Dual anti-thrombotic therapy highlighted a strong and independent inverse association with long-term cardiovascular mortality in individuals presenting with peAF, with an adj. HR of 0.85 (95% CI 0.67–0.96; P = 0.022), while TAT was not associated with mortality [adj. HR of 0.91 (95% CI 0.76–1.21); P = 0.321]. Most importantly—as in contrast to peAF—DAT did not show an effect on survival in dnAF [adj. HR of 0.97 (95% CI 0.65–1.57); P = 0.346], while TAT had a strong and independent inverse association with long-term cardiovascular mortality with an adj. HR of 0.86 (95% CI 0.45–0.92; P = 0.012) (see Table 3). A total of 165 (92.1%) patients with a CHA₂DS₂-VASc of ≥2 received any anticoagulation with 62 (91.2%) individuals in the peAF and 103 (92.7%) in the dnAF group respectively. Notably, there was no difference in the incidence of fatal bleeding events when comparing different treatment strategies [dnAF: 2.6% (n = 4) vs. peAF: 2.2% (n = 2); P = 0.824].

Discussion

The current analysis presents—to the best of our knowledge—the largest one with the longest follow-up period of patients presenting with dnAF that additionally pictured the associated outcome of anti-thrombotic treatment strategies. The present data clearly highlighted that the presence of dnAF was independently and directly associated with an increased risk for fatal cardiovascular events from a long-term perspective and that intensified anti-thrombotic treatment might be considered in affected individuals.

The protective effect of pre-existing atrial fibrillation

Of utmost interest, peAF was found to be associated with favourable outcome after AMI compared to AF-free individuals from a long-term perspective. Patients with peAF appear to benefit from their pre-existing cardiovascular therapy compared to patient without AF and anti-thrombotic therapy. Since the majority of peAF individuals have already received OAC at the time of the acute ischaemic event, the anti-atherothrombotic effect may temper the extent of the acute ischaemia. In this regard, the affected myocardial tissue seems to be smaller in peAF patient population—as highlighted by significant lower peak CK values, which closely correlate with infarct size. Similarly, the lower NTproBNP levels in peAF might additionally mirror a weak impact on both cardiac strain and left ventricular ejection fraction, as one of the most prominent prognosticators for outcome after AMI.^14,15

The outcome of AMI patients presenting with de novo atrial fibrillation

Aligned with a systemic review of McIntyre et al. this study could show that dnAF is frequent after AMI. While McIntrye and colleagues demonstrated incidence rates of dnAF up to 44% after medical illness, we could assess that dnAF occurred in 11% of patients after AMI.¹⁶

Previous studies that aimed to elucidate the impact of AF on long-term mortality demonstrated heterogeneous results. While some investigations showed that AF, either dnAF or peAF, is associated with higher rates of short- and long-term mortality, others have not highlighted any effect of this arrhythmia on patient outcome.^5,6 Notably, within the HORIZONS-AMI sub-study, Rene and co-workers found a higher risk for fatal cardiovascular events in individuals presenting with dnAF.¹⁷ Similarly Sanchez et al.¹⁸ observed an association of dnAF and short-term mortality. Therefore the results of the present investigation are in line with previous observations and even extended the current knowledge with a very long-term perspective. Considering the strong impact on patient outcome and the fact that many (paroxysmal) AF episodes remain undetected in clinical practice, patient characteristics that help to identify AMI individuals at risk for the development of dnAF should be considered in terms of a personalized secondary prevention.

The identification of patients at risk for de novo atrial fibrillation

Within the present investigation, patients in the dnAF group trended to present more frequent with STEMI but were less likely to receive PCI and stent implantation. One may assume that lower intervention rates may lead to an extended tissue damage. This assumption was mirrored by increased maximum CK levels. Extended tissue damage and scar formation, in turn might trigger AF impulses during the acute phase of AMI. Additionally, NTproBNP levels were also found to be significantly elevated in dnAF individuals, as a well-established risk-marker for development of AF based on increased cardiac strain. The present data suggest that cardiac strain might be deteriorated by wall motion abnormalities, which in turn occur as a result of increased tissue damage and scar formation. However, the exact pathophysiological association between NTproBNP and dnAF still needs to be clarified but regardless of their relationship elevated levels allow the identification of patients at risk for dnAF. Notably, the presence of elevated NTproBNP values—as an expression of poor left ventricular function—did not modulate the prognostic effect of dnAF within the present analysis.¹⁹

Aligned with the findings of Topaz and colleagues and Kinjo et al., patients developing dnAF were significantly older and tended to present with a higher burden of cardiovascular and metabolic comorbidities, such as type 2 diabetes mellitus (T2DM).^20,21 The influence of T2DM on kidney function is well known and is mirrored by increased rates of concomitant chronic kidney disease (CKD). Furthermore, factors associated with CKD, such as uraemic toxins, inflammation, fluid overload, and electrolyte imbalance were found to be associated with the development of first AF episodes.²² Considering the mentioned T2DM and CKD associated morbidity, it can reasonably be assumed that those factors impact on both the development of AF impulses and subsequently the prognosis of patients with dnAF. Since the afore mentioned patient characteristics can be considered as potential predictive values for identification of patients at risk for the development of dnAF, individuals presenting with those features may benefit from intensified clinical follow-up and prolonged ECG monitoring in order to detect AF episodes. Based on the associated impact on outcome and thromboembolic risk of AF, anti-thrombotic treatment strategies conquer a central position in terms of secondary prevention after detection of dnAF.

Anti-thrombotic management

While DAT—compared to all other antithrombotic treatment strategies—showed a clear survival benefit in peAF, it was not associated with an improvement in outcome in individuals presenting with dnAF. Interestingly, a TAT at discharge showed a strong and inverse association with mortality in dnAF, but no effect in peAF. Triple antithrombotic therapy therefore appears to be beneficial in patients with dnAF, but there may be bias in the choice of TAT, especially in patients with a lower risk of bleeding. Serval randomized controlled clinical trials (WOEST, PIONEER, RE DUAL, ENTRUST-AF PCI, and AUGUSTUS) comparing the efficacy and safety of DAT vs. TAT in AF patients with AMI and/or undergoing PCI and stenting.^23–27 In these trials, bleeding events were significantly lower with DAT compared to TAT. Despite, these studies were not sufficiently powered to rule out an increased risk for ischaemic events with DAT vs. TAT, a recent meta-analysis highlighted that DAT significantly reduces bleedings compared to TAT and seems to have a similar effect in preventing ischaemic endpoints in AF patients post-PCI or ACS.²⁸ However, the respective investigations did not give a profound subgroup analysis for the management of high-risk dnAF patients.

The choice of optimal antithrombotic therapy for patients with PCI and de novo AF is currently based on expert opinion. Since TAT at the time of discharge showed a strong and inverse association with mortality in dnAF, but no effect in peAF, this therapeutic approach appears to be particularly beneficial in patients with dnAF. However, a potential bias in the choice of TAT, preferably for patients with lower risk of bleeding, may be assumed. As indicated by the provided data, patients presenting with dnAF are at higher risk for fatal thrombo-embolic and atherothrombotic events which justifies a prolonged TAT, but decisions should be based on recent guidelines for evaluating the exact duration of TAT in dnAF patients. In addition, safer (in terms of bleeding risk) non-vitamin K antagonist oral anticoagulants (NOACS) are now available. Unfortunately, the present investigation only pictures the effect of VKA and not NOACs. While an even more beneficial effect of NOACs seems intuitive in dnAF patients receiving TAT, further studies are needed to evaluate the optimal antithrombotic approach in dnAF with NOACs.^9,10,29

The present data clearly highlighted that the presence of dnAF was independently and directly associated with an increased risk for fatal cardiovascular events from a long-term perspective. Considering the observation that DAT was not associated with a treatment benefit in dnAF, but a TAT strategy proved to be associated with a risk reduction during this long-term observation, an aggravated anti-thrombotic treatment approach via prolongation of DAPT in addition to OAC should be taken into account in this high-risk dnAF patient population to prevent fatal atherothrombotic and thrombo-embolic events. Even dough the present investigation provided results independently for the effects of VKAs, it can reasonably be assumed that both DAT and TAT using a NOAC agent provides a comparable efficacy in terms of prevention of atherothrombotic and thrombo-embolic events, and superiority in prevention of major bleeding events. However, future investigations need to confirm the present results from a NOAC perspective.

Limitations

The major limitation of the present analysis represents its single centre setting. However, via enrolment of a large patient population during an extensive observation period, we might overcome a potential selection bias. Moreover, we do not have data on ischaemic events or bleeding complications which were non-fatal, that might alter the overall picture of our results and would have allowed the calculation of a net clinical benefit of the respective anti-thrombotic treatment strategies. Additionally, the present investigation is limited to VKAs since enrolment was completed before NOACs were approved by the EMA and FDA. Moreover, due to the lack of personal follow-up visits, both the duration of DAT and TAT, as well the incident rates of AF episodes were not assessed during the observation period. However, based on available data in literature it can reasonably be assumed that frequencies of AF episodes intensified in both dnAF and peAF patients and a development of permanent AF on the basis of paroxysmal AF took place in several individuals.³⁰ Unfortunately, based on the study design we were not able to elucidate the respective rate or rhythm control strategy during the time of hospitalization. However, we may rely on recent evidence in literature that there is no difference in outcome when comparing both antiarrhythmic strategies, when considering the results of the present investigation.³¹

Conclusion

In summary, there is a lack of clear recommendations and clinical practice guidelines regarding the monitoring and anti-thrombotic treatment approach of patients presenting with dnAF during the acute phase of AMI. Within the present investigation, we found that dnAF was independently associated with a poor patient prognosis with an increased long-term risk for cardiovascular mortality by 67%. Considering the observation that DAT was not associated with a treatment benefit, but a TAT strategy proved to be associated with a risk reduction during this long-term observation, an aggravated anti-thrombotic treatment approach should be taken into account in this high-risk dnAF patient population.

Conflict of interest: P.S. reports grants from Daiichi Sankyo and grants from Boehringer-Ingelheim outside the submitted work. A.N. reports personal fees from Bayer, personal fees from BMS, grants and personal fees from Boehringer Ingelheim, grants and personal fees from Daiichi Sankyo and personal fees from Pfizer, outside the submitted work.

References