Introduction

Atrial fibrillation (AF) is a major risk factor for stroke [1]. This is particularly relevant for elderly patients as the prevalence of AF increases with advancing age from 0.12–0.16% in people aged < 49 years to 1.7–4.0% among those aged 60–70 years; proportions may be as high as 13.5–17.8% among those aged > 80 years [2,3,4]. Both AF and coronary artery disease (CAD) are common cardiovascular conditions encountered in daily clinical practice in elderly patients. The diseases tend to coexist because of shared risk factors, such as hypertension, diabetes, advanced age, obesity, and smoking, and they have similar pathophysiological features, such as inflammation [5]. CAD is more common in patients with AF, ranging between 17% and 46.5% [6, 7]. Furthermore, several studies have reported that comorbid CAD and AF aggravate one another [8, 9]. AF is associated with a twofold increase in the risk of myocardial infarction (MI) [7]. According to the Framingham study, patients with AF and heart disease have a 2.2-times higher probability of developing new coronary events (CE) compared with patients with heart disease without AF [9]. CAD is the third leading cause of death worldwide, with approximately 18 million deaths annually attributed to CAD [10].

Patients with both CAD and AF pose additional challenges in terms of treatment—which includes rhythm management, anticoagulants, and antiplatelet agents (APA)—and may require more complex treatment strategies to mitigate possible increases in bleeding risk [5, 11]. The Japanese open-label AFIRE trial suggested that oral anticoagulant (OAC) monotherapy was superior for safety compared with OAC and APA combination therapy in patients with AF and CAD [12]. Additionally, a temporal association between major bleeding and subsequent cardiovascular events and death in patients with AF and stable CAD has been demonstrated [13].

Although clinical outcomes in AF patients based on the presence or absence of CAD (e.g., MI or percutaneous coronary intervention [PCI]) have been reported [10, 14,15,16,17,18], real-world data on the incidence and risk factors of CE in elderly patients—who are at the highest risk of events—are lacking, especially in the era of direct OACs (DOACs). The All Nippon Atrial Fibrillation In the Elderly (ANAFIE) Registry aimed to clarify the real-world clinical status and prognosis of elderly patients with non-valvular AF (NVAF) in Japan. Over 30,000 elderly (≥ 75 years of age) Japanese patients with NVAF were enrolled and followed up for 2 years to investigate anticoagulation therapy status and outcomes in routine clinical practice. The 2-year follow-up data [19] and several sub-analyses have been published [20,21,22,23,24].

The main objective of this prespecified sub-analysis of the ANAFIE Registry was to investigate the incidence and risk factors of CE in elderly Japanese patients with NVAF. The occurrence of bleeding events in CE patients was also examined.

Methods

Study design

The ANAFIE Registry was a multicenter, prospective, observational study conducted at 1273 sites across Japan between 2016 and 2020. Details of the study design and rationale have been published [25]. The trial was registered in the UMIN Clinical Trials Registry under the identifier UMIN000024006. The study was compliant with the Declaration of Helsinki and local requirements for registries. Ethics committees approved the study protocol. Written informed consent was obtained from patients or family members in case of communication disorders (i.e., aphasia) or cognitive impairment.

Patients

Enrolled outpatients were men and women ≥ 75 years of age, diagnosed with NVAF by electrocardiography, who were able to attend hospital visits. Patients were excluded from enrollment if they were participating/planning to participate in an interventional study; had a definite diagnosis of mitral stenosis or artificial heart valve replacement (either mechanical or tissue valve prostheses), or had experienced very recent cardiovascular events, including stroke, MI, cardiac intervention, heart failure requiring hospitalization, or any bleeding leading to hospitalization within 1 month prior to enrollment; life expectancy of < 1 year; or who were deemed inappropriate for participation by treating physicians.

Study endpoints

Specifically, in this pre-specified sub-analysis, we assessed the incidence for new-onset CE (defined as a composite of MI and cardiac intervention for coronary heart diseases other than MI), MI, cardiac intervention for coronary heart diseases other than MI, major bleeding, clinically relevant non-major bleeding (CRNMB), intracranial hemorrhage (ICH), and gastrointestinal (GI) bleeding during the 2-year follow-up period. Major bleeding was classified using the International Society on Thrombosis and Haemostasis definition.

Statistical analysis

The Kaplan–Meier method was used to estimate the probability of occurrence of CE and other clinical events. The incidences of CE and other clinical events were also estimated as incidence rates per 100 person-years with 95% confidence intervals (CIs). A multivariate analysis was performed to identify risk factors of CE calculated using the Cox proportional hazards model. This analysis was also performed by combining the history of CAD and APA use. For bleeding events, based on the presence or absence of CAD, odds ratios (ORs) were evaluated using a logistic regression model adjusted for prognostic factors. Statistical tests were two-sided, with a significance level of 5%. The statistical software used for these analyses was SAS version 9.4 (SAS Institute, Tokyo, Japan).

Results

Patient disposition and characteristics

Of the 32,275 patients analyzed in the ANAFIE Registry, 287 developed CE (MI, cardiac intervention for coronary heart diseases other than MI) (0.89%). Table 1 shows the characteristics of patients with new-onset CE and those without CE. Significantly more men than women had CE vs no CE. Creatinine clearance (CrCL) was significantly lower, and CHADS2 and HAS-BLED scores were significantly higher in patients with CE vs those without CE. Similarly, significantly higher proportions of patients with CE had diabetes mellitus, dyslipidemia, a history of CAD including prior MI and/or angina, a history of cerebrovascular diseases including lacunar infarction and peripheral arterial disease and falls within 1 year.

Table 1 Background characteristics of patients at baseline by the presence of new-onset coronary events
Full size table

Incidence of events

Figure 1 shows the probability of event occurrence for each clinical outcome. The probability of occurrence of CE after 2 years of observation was 0.94% [95% CI: 0.83–1.05]), which was lower than that of stroke/systemic embolic events (SEE) (3.18% [2.99–3.39]).

Fig. 1
figure 1

Kaplan–Meier curves for coronary events and stroke/SEE. Each occurrence and 95% CI show the data at 2 years. CE coronary event; MI myocardial infarction; SEE systemic embolic event

Full size image

The incidence rate was 0.48 per 100 patient-years (95% CI: 0.42–0.53) for CE, 0.20 (0.16–0.23) for MI, and 0.29 (0.25–0.33) for that of cardiac intervention for coronary heart diseases other than MI (Table 2); these were lower than that of stroke/SEE, which was 1.62 (1.52–1.73) in the main analysis of the ANAFIE Registry [19].

Table 2 Incidence rates of coronary events and stroke/SEE
Full size table

Risk factors of CE

The risk factors associated with new-onset CE were male sex, systolic blood pressure ≥ 130 mmHg, diabetes mellitus with glycated hemoglobin (HbA1c) ≥ 6.0%, CAD history with and without APA use, APA use without CAD history, and CrCL < 50 mL/min (Table 3). Of note, prior CAD and APA use each were significant risk factors, and their co-occurrence increased the risk even further.

Table 3 Risk factors for new-onset CE, MI, and cardiac intervention for coronary heart diseases other than MI during follow-up, results of a multivariate analysis
Full size table

Risk factors for MI were systolic blood pressure between 130 and 140 mmHg, no history of CAD with APA and history of CAD with APA vs no history of CAD without APA, and CrCL < 50 mL/min.

Risk factors for cardiac intervention for coronary heart diseases other than MI were male sex, systolic blood pressure ≥ 140 mmHg, diabetes mellitus with HbA1c ≥ 6.0%, history of CAD/no APA, and history of CAD/APA (Table 3). Of note, OAC use was not associated with the risk of CE.

Analysis of patient backgrounds based on the history of CAD and use of APA

Table 4 summarizes the comparison of background factors of patients with a history of CAD, with and without the use of APA. Patients using APA were significantly more likely to be male, slightly younger, and with higher BMI, lower CrCL, and higher HAS-BLED score. They were more likely to have paroxysmal AF, use OAC therapy, and had a higher proportion of comorbidities, diabetes mellitus, dyslipidemia, atherosclerotic diseases other than CAD (i.e., atherosclerotic infarction or peripheral artery disease), and chronic kidney disease.

Table 4 Patient characteristics according to a history of coronary artery disease and the use of antiplatelet agents
Full size table

Bleeding events in patients with CE

All bleeding events were observed in 2,557 cases (7.9%). Bleeding events in patients with or without new-onset CE are shown in Fig. 2. In the univariate analysis, new-onset CE was significantly associated with a higher incidence of major bleeding (OR: 3.35 [95% CI: 2.06–5.43]), CRNMB (2.06 [1.15–3.70]), ICH (2.03 [1.00–4.13]), and GI bleeding (2.30 [1.48–3.56]) compared with those without CE. In the multivariate analysis both when adjusting for the presence of APAs and when adjusting for the type of OACs, the incidence of major bleeding, CRNMB, ICH, and GI bleeding remained significantly higher for patients with new-onset CE compared with patients without CE. Of 287 patients who developed new-onset CE, the incidence of CE was 5.2% (n = 15) after all bleeding episodes, 2.4% (n = 7) after a major bleeding event, and 1.0% (n = 3) after a CRNMB event.

Fig. 2
figure 2

Risk of bleeding events with new-onset coronary event versus no onset of a coronary event. CI confidence interval; CRNMB clinically relevant non-major bleeding; DOAC direct oral anticoagulant; GI gastrointestinal; ICH intracranial hemorrhage; OAC oral anticoagulant; OR odds ratio

Full size image

Discussion

The main findings of this sub-analysis were as follows. First, in the overall population, the incidence rates of CE, MI, and cardiac intervention for coronary heart diseases other than MI were 0.48, 0.20, and 0.29 per 100 patient-years, respectively, which were lower than that of stroke/SEE (1.62 per 100 patient-years) reported in the main analysis of the ANAFIE Registry [19]. Second, compared with patients without CE (n = 31,988), those with new-onset CE (n = 287) were more likely to have lower CrCL and higher CHADS2 and HAS-BLED scores. Third, risk factors significantly associated with the onset of CE were male sex, systolic blood pressure of ≥ 130 mmHg, diabetes mellitus with HbA1c ≥ 6.0%, history of CAD, APA use, and CrCL < 50 mL/min. Fourth, the incidence of major bleeding, CRNMB, ICH, and GI bleeding was significantly higher in patients with new-onset CE compared with those without CE.

The incidence of CE in the present study was consistent with a recent report on the trends of antithrombotic therapy status and outcomes in Japanese AF patients (mean age ± standard deviation, 73.6 ± 10.9 years), in which the incidence of MI was 0.2% and that of stroke/SEE, 2.2% per patient-year during a 5-year follow-up [26]. The trends observed for the onset of CE in patients with a history of CAD were similar to those reported in a previous study overseas in which cardiovascular outcomes of patients with a history of CAD (i.e., MI) were worse than those of patients without a history of CAD [11]. However, the incidence of MI during anticoagulant therapy was lower in Japanese than in Western patients—a finding that might be attributable to ethnic differences [27,28,29].

The risk factors for CE for elderly patients with NVAF in the present analysis were also generally consistent with previous reports [30]. An unexpected finding of our study was that APA use was a risk factor for CE. Based on the comparison of patient background data, advanced atherosclerotic diseases are likely the basis for APA use. Therefore, patient background factors, for instance a history of atherosclerotic disease such as peripheral arterial disease, may be associated with high CE risk in patients taking APA. In these patients, it may be difficult to reduce the risk of CE, even with APA. Conversely, OACs were not associated with CE risk.

Another explanation is that bleeding associated with antiplatelet administration may have increased CE. In the present study, new-onset CE was significantly associated with a higher incidence of bleeding compared with patients without CE. The incidence of major bleeding in patients with new-onset CE was significantly higher than that in patients without CE (OR: 3.35 [95% CI: 2.06–5.43]). The findings of the multivariate analysis suggest that APA use or type of OAC at baseline may not have contributed to the incidence of bleeding in patients with new-onset CE. On the other hand, it has also been noted that bleeding complications are followed by ischemic events. In the present study, the incidence of CE after the onset of all bleeding was 5.2%, which is consistent with a previous report [13]. This may be because of the cessation of antithrombotic therapy, blood transfusion, or other interventions. As noted, adjusting for antiplatelet or anticoagulant drug use did not significantly change the OR. Therefore, these drugs do not appear to contribute to increased bleeding events in new-onset CE patients, which may instead be attributable to risk factors identified in the high bleeding risk criteria for PCI patients [31], although this remains unclear.

Concerning risk factors specifically for MI, systolic blood pressure between 130 and 140 mmHg, no history of CAD/APA, history of CAD/APA, and CrCL < 50 mL/min were significant risk factors, which is generally consistent with the previously reported risk factors for MI [32, 33]. Nevertheless, it was surprising that dyslipidemia and diabetes mellitus were not among the relevant risk factors identified. It is possible that the small sample size in that subgroup precluded these factors from reaching a significant difference. Additionally, if the administration of statins was high in this population, it is possible that the risk was modified. Nevertheless, data on statin administration were not collected in this study. Furthermore, it is possible that traditional risk factors, such as hyperlipidemia and diabetes, have become relatively unimportant in the elderly [34, 35]. Treatment history of dyslipidemia was also not recorded in this study.

Regarding the clinical relevance and implications of these findings, it is important to characterize patients at higher risk of CE to prevent and reduce CE. Our results clarify the risk of developing CE in elderly NVAF patients. Although these patients were at a lower risk of CE than stroke/SEE, strict management is required to prevent CE in patients with risk factors for CE, especially those with a history of CAD and those who require APA therapy. Another important finding was that aspirin use was a negative predictor of the development of CE. Therefore, concomitant antiplatelet therapy for primary and secondary prevention of CE may not be recommended for elderly patients with AF. These findings are consistent with the AFIRE trial [12] and are a further advance on the ASPREE trial, in which the use of low-dose aspirin in elderly patients without atrial fibrillation resulted in a significantly increased risk of major bleeding without a decreased risk of cardiovascular disease [36], although the lack of data on treatment changes after each event in the present study means that this requires further verification.

Limitations

The main limitations of the ANAFIE Registry have been published previously [25, 37]. Information such as the withdrawal or change of APA and anticoagulants and use of statins or other treatments for dyslipidemia was not evaluated during the study. Because there was a low prevalence of a history of MI in the study population (5.7%), our findings cannot be extrapolated to populations with higher MI prevalence.

Conclusions

This sub-analysis of the ANAFIE Registry is the first large-scale study to report that CE incidence was lower than that of stroke/SEE in elderly patients with NVAF. Risk factors for CE in elderly Japanese patients with NVAF were male sex, systolic blood pressure of ≥ 130 mmHg, diabetes mellitus, CE history, antiplatelet agent use, and CrCL of < 50 mL/min. New-onset CE was associated with a higher incidence of major bleeding than no CE. Thus, the current findings may contribute to the understanding of the management of elderly NVAF patients.