Anticoagulation in atrial fibrillation detected after stroke: Many questions,some answers

Oral anticoagulation (OAC) is the cornerstone of stroke prevention in atrial fibrillation (AF). However, pivotal randomized controlled trials (RCTs) which demonstrated the benefit of OAC almost exclusively selected patients with clinical AF, who experience high rates of stroke or systemic embolism (SSE) without OAC. Subclinical AF (SCAF) encompasses AF identified on cardiac monitoring devices including implantable loop recorders, permanent pacemakers, implantable cardioverter-defibrillators, and other wearables. SCAF is associated with an increased risk of stroke, but absolute rates are approximately 1% per year, which represents a substantially lower risk than that of clinical AF, raising uncertainty as to whether OAC is safe and effective for stroke prevention in SCAF.

Two recent RCTs have examined this question. NOAH-AFNET 6 enrolled patients with SCAF with the longest episode lasting >6 min with cardiovascular risk factors and compared edoxaban with usual care (53% aspirin). This trial was stopped early after a median of 1.8 years due to an increased risk of the safety composite outcome of major bleeding or death. The ARTESiA trial compared apixaban with aspirin and similarly included patients with CHA₂DS₂-Vasc ⩾3 and SCAF ⩾6 min. After a mean follow-up of 3.5 years apixaban reduced SSE by 37%, which was counterbalanced by an increased risk of major bleeding. Both RCTs enrolled older patients with a high cardiovascular risk factor burden, the median duration of the longest SCAF episode was 1.5–2.8 h, and the rate of clinically-detected AF was 6.3–8.07%/year. ¹ Importantly, almost all patients in had permanent pacemakers or implantable cardioverter-defibrillators and very few had SCAF detected by implantable loop recorder. The annual rates of ischemic stroke in the no OAC arm were 1.0%–1.1% and 0.6%–0.9% on OAC. Corresponding rates of major bleeding on OAC were 1.5%/year in ARTESiA and 2.1%/year in NOAH-AFNET 6. When both RCTs were pooled in a recent meta-analysis, anticoagulation was associated with a 30%–35% lower risk of ischemic stroke, all stroke, and SSE, with an increased risk of major bleeding. ¹ Extrapolating these data, OAC prevented an estimated 3 ischemic strokes, but caused approximately 7 additional major bleeds, for every 1000-patient years. These data came with the caveat of wide statistical uncertainty around the point estimate.

In the current edition of the ESJ, Palaiodimou et al. ² Performed a meta-analysis of the effect estimates in the subgroup of patients from both trials who had a history of prior stroke or transient ischemic attack. The combined sample of 599 patients had a high cardiovascular risk factor burden (median CHA₂DS₂-Vasc 6), frequently had co-existing atherosclerosis, and were mostly randomized >1 year after stroke. Most patients in the comparator arm were given aspirin (96%). The cumulative risk of any recurrent stroke was approximately 5% in patients receiving aspirin/placebo and 3% in those on OAC. The pooled analysis demonstrated that OACs reduced the risk of both any stroke and SSE by 50%, with a similar non-significant reduction in ischemic stroke. The cumulative risk of major bleeding was 5% on OACs and 2% on aspirin/placebo, representing a 2-fold increased risk, but without an increase in fatal bleeding, intra-cerebral hemorrhage or all-cause mortality. The authors estimated that the number needed to treat to prevent one stroke was 34 with a corresponding trade-off of a number needed to harm for major bleeding of 37.

ESO Guidelines have recommended the use of prolonged cardiac monitoring to detect SCAF in patients with cryptogenic stroke, but do not give a clear recommendation on the optimal duration or burden that warrants OAC initiation. Palaiodimou et al. provide important randomized evidence that OAC reduces the risk of recurrent stroke in patients with SCAF lasting ⩾6 min when compared with no anticoagulation. The data provide further support to the rationale that prolonged cardiac monitoring after ischemic stroke has the potential to lower recurrence risk by increasing rates of OAC use. This is particularly relevant considering several recent RCTs that demonstrated no benefit from routine OAC use after embolic stroke of uncertain source (ESUS). Taken together, the current analysis reiterates the importance of identifying SCAF before initiating anticoagulation in patients with ESUS.

Although this work will inform guidelines, there are several very important caveats for clinicians. AF detected after stroke (AFDAS) can be classified into three types. ³ AFDAS identified early after stroke which is high burden (type A) is likely causally related to the stroke and most clinicians would choose OAC over anti-platelets. AFDAS detected several months after stroke is often low burden (type B) and the casual relationship is uncertain. AFDAS found <2 weeks from stroke and present in low burden (type C) may be triggered by autonomic dysfunction after infarction, may not recur, may not predate the stroke, and thereby may represent a bystander finding. Patients enrolled in both trials do not relate in any way to Type A or Type C AFDAS and the current data should not be relied upon when considering treatment decisions for these cases. The findings in this work do shed some light on OAC for patients with Type B AFDAS. However, it is questionable whether the patient cohort enrolled in both trials is representative of the ESUS population in the usual scenario where AFDAS is identified on implantable loop recorder. Patients in the present work had a very high cardiovascular risk factor burden, were older, and frequently had established atherosclerosis. Moreover, these patients almost all had permanent pacemakers or implantable cardiac defibrillators thereby indicating a high prevalence of underlying atrial or ventricular cardiomyopathy. Therefore, the relative benefit in stroke reduction from OAC may not always sufficiently offset the risk of harm if the background stroke risk is lower than that represented in ARTESiA and NOAH-AFNET 6. This consideration is important for clinicians who are confronted with common scenario where a patient with AFDAS on implantable loop recorders has few risk factors and otherwise no other objective evidence of cardiopathy.

In summary, this important work provides some answers, but should encourage further research into the following unanswered questions:

1. What biomarkers can be used to risk stratify patients with AFDAS?

2. Can RCTs help provide support for OAC use in AFDAS with very low burden AF or Type C AFDAS or those with few CVS RFs?