Article Text

Download PDFPDF

Ischaemic stroke in anticoagulated patients with atrial fibrillation
  1. Christoph Stretz1,
  2. Teddy Y Wu2,
  3. Duncan Wilson3,
  4. David J Seiffge4,
  5. Eric E Smith5,
  6. M Edip Gurol6,
  7. Shadi Yaghi1
  1. 1 Department of Neurology, Brown University, Providence, Rhode Island, USA
  2. 2 Department of Neurology, Christchurch Hospital, Christchurch, New Zealand
  3. 3 Stroke Research Centre, UCL Institute of Neurology, London, UK
  4. 4 Department of Neurology and Stroke Center, Inselspital, University Hospital Bern, Bern, Switzerland
  5. 5 Department of Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada
  6. 6 Department of Neurology, Massachusetts General Hospital, Boston, Massachusetts, USA
  1. Correspondence to Dr Shadi Yaghi, Department of Neurology, Brown University, Providence, RI 02906, USA; shadiyaghi{at}


Anticoagulation substantially reduces the risk of stroke in patients with atrial fibrillation (AF). However, recent studies have shown that up to 22%–36% of patients on anticoagulation will suffer an ischaemic stroke (IS). In this narrative review, we provide an overview of risk factors, mechanisms, management of acute IS and strategies for secondary prevention for patients with AF with stroke despite oral anticoagulation. For this paper, we reviewed available literature from important studies (randomised clinical trials, meta-analyses, reviews and case series) on patients with IS despite anticoagulation. We focused on recent studies that examined safety and efficacy of acute stroke treatments and evaluation and management strategies for secondary prevention. The literature review suggests that patients with AF with IS despite anticoagulation are a heterogeneous group with several possible mechanisms, which may include reduced or non-adherence to anticoagulation, competing non-cardioembolic stroke aetiologies or cardioembolic mechanisms separate from AF. The identification of one or more possible mechanisms of stroke despite anticoagulation may allow for a more targeted and individualised approach for secondary prevention. There are limited data to guide management in such patients, and strategies to prevent recurrent strokes include strict risk factor control and therapies targeting the most likely stroke mechanism. In cases where AF is suspected to be the culprit, clinical trials are needed to test the safety and efficacy of left atrial appendage occlusion plus anticoagulation versus continued anticoagulation alone.

  • stroke

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Atrial fibrillation (AF) is the most common cardiac arrythmia with increasing incidence over the years, reaching about 9% prevalence in patients 80 years or older.1 The actual incidence may be higher as a relatively high proportion of AF is subclinical and remains undiagnosed. AF is associated with an increased risk of ischaemic stroke, systemic embolism, and death.2 For example, in the USA, about 10%–12% (~80 000) of ischaemic strokes occur in the setting of AF and carry higher morbidity and mortality than those occurring in patients without AF.3 In patients with AF, randomised controlled trials (RCTs) showed that oral anticoagulation with vitamin K antagonists (VKA) or direct oral anticoagulants (DOACs) substantially reduces the risk of stroke. Recent studies, however, have shown that 22%–36% of ischaemic strokes in patients with AF occur in patients on anticoagulation and that the rate of recurrence in these ‘OAC-failures’ is very high.4 5 Possible aetiologies for recurrent ischaemic stroke despite anticoagulation include reduced medication adherence, interruption of medication use and inadequate dosing. In addition, concurrent causes of stroke, including large or small-vessel atherosclerosis and hypercoagulability—due to blood disorders or in the setting of autoimmune disease or malignancy—can increase the risk of recurrent stroke on anticoagulation.4 5

The aim of this review is to provide an overview of risk factors, mechanisms and acute stroke treatment and secondary prevention options in patients with AF who develop an ischaemic stroke while on anticoagulation.

Epidemiology and risk factors

Defining the incidence of stroke in patients on anticoagulation requires some consideration. There are many different variables, which likely affect incidence rates. A meta-analysis of available trials on VKA with various comparison drugs shows the incidence of stroke in patients on VKA is roughly 2% per year over a mean follow-up of 1.5 years per patient.6 The vast majority (79%) of patients in these trials were receiving a VKA for primary prevention. In a more recent meta-analysis of trial data in patients taking DOACs over a mean follow-up time of 2.2 years, ischaemic stroke occurred in 2.3% of patients on DOACs and 2.5% of patients on a VKA. Seventy per cent of these patients were on oral anticoagulants for primary prevention and about one-third of patients enrolled into RELY (dabigatran) and ARISTOTLE (apixaban) had a very low CHADS2 (congestive heart failure, hypertension, age ≥ 75, diabetes, stroke or transient ischemic attack) score of 0–1.The external validity of these studies in real-world situations has been hotly debated due to enrolment of low-risk patients and other strict exclusion criteria, such as concern for bleeding, cognitive issues, fall risk or alcohol use.

Recent data suggest that up to one-third of patients with AF develop an ischaemic stroke despite oral anticoagulation therapy.4 In a large multicentre international study (n=5413 patients),5 patients with ischaemic stroke despite anticoagulation were older, more likely to have traditional risk factors such as hypertension, hyperlipidaemia and diabetes, and previous stroke when compared with patients with ischaemic stroke not on anticoagulation. Interestingly, despite these differences, the CHA2DS2 VASc (congestive heart failure, hypertension, age ≥ 75, diabetes, stroke or transient ischemic attack, vascular disease, female sex) score was not different.5 Another multicentre study in the USA showed that patients with AF who developed an ischaemic stroke despite anticoagulation were more likely to have hyperlipidaemia, diabetes, prior stroke, severe left atrial (LA) enlargement and lower left ventricular ejection fraction on transthoracic echocardiogram.6 In this study, the CHA2DS2 VASc score was higher in patients who developed an ischaemic stroke while on anticoagulation.4 These small differences between studies are partially explained by population differences and the oral anticoagulant predominantly used (DOAC vs VKA).

Another study investigated factors associated with ischaemic stroke in patients on VKA who were therapeutic versus non-therapeutic on anticoagulation. They found that patients who were therapeutic at the time of ischaemic stroke were more likely to have ipsilateral moderate to severe arterial stenosis and a small infarct volume (<10 mL).7 This suggests that some strokes due to OAC failure may be attributed to non-cardioembolic mechanisms (atherosclerosis or small-vessel disease), where anticoagulation use may not be an important factor in preventing strokes related to these mechanisms.

Furthermore, the RENo (causes and risk factors of cerebral ischemic events in patients with nonvalvular AF treated With NOACs for stroke prevention) study8 compared patients with AF who had an ischaemic stroke while on DOAC treatment versus those who did not. In this study, off-label low dose DOAC, atrial enlargement, hyperlipidaemia, high CHA2DS2-VASc score and increased AF burden were all associated with ischaemic stroke despite DOAC therapy. Interestingly, in this study, 36% of the ischaemic strokes had a competing non-cardioembolic mechanism.8

Therefore, based on the above data, patients who develop an ischaemic stroke despite anticoagulation comprise a heterogeneous group. Some strokes may be related to more advanced cardiac disease (increased AF burden and severe atrial enlargement), insufficient anticoagulation (off label use of low dose DOAC or subtherapeutic VKA), and non-cardioembolic mechanisms such as small or large-vessel atherosclerosis, or other defined mechanisms such as hypercoagulability of malignancy. Overall, the findings of increased rates of AF-specific factors such as AF burden and LA enlargement suggest that many of these OAC failures had AF-related cardioembolic strokes.

Acute treatment of ischaemic stroke on anticoagulation

Ischaemic stroke on warfarin

Intravenous thrombolysis

Patients with acute ischaemic stroke on anticoagulation with VKA are eligible for intravenous thrombolysis (IVT) if they present within the 4.5 hours time window and international normalized ratio (INR) ≤1.70.9 While patients presenting in the 3–4.5 hours window were excluded from the ECASS III trial if taking VKA regardless of INR,10 the use of intravenous alteplase may still be reasonable in this time window, provided INR ≤1.7 or prothrombin time (PT) <15 s.9 11In addition to alteplase, tenecteplase may be a reasonable alternative for IVT in patients with concomitant large-vessel occlusion (LVO), in the absence of contraindications for IVT.9 Although tenecteplase has practical advantages (single bolus dose administration between 0.25mg/kg and 0.40 mg/kg), tenecteplase is not approved by the Food and Drug Administration (FDA) for use in acute ischaemic stroke in the USA. Furthermore, tenecteplase has been compared with alteplase in patients without LVO in three phase II and one phase III superiority trial. While it was found to be equally safe to alteplase, the efficacy in comparison to alteplase remains unclear.9

Mechanical thrombectomy

Patients with acute ischaemic stroke and concomitant LVO are potential candidates for mechanical thrombectomy in accordance with American Heart Association/American Stroke Association guidelines.9 12 Randomised clinical trials showed efficacy of mechanical thrombectomy (within 12 hours from symptom onset) over standard medical care in patients with acute ischaemic stroke due to proximal LVO of the anterior circulation.9 12 Subsequently, for patients presenting in the extended time windows (6–16 hours) and (6–24 hours), mechanical thrombectomy using imaging-based criteria showed improved functional outcome.9 12 The use of anticoagulation was not a contraindication to enrolment in either trial. While the proportion of anticoagulated patients receiving mechanical thrombectomy was low (88/1872 or 4.7%),13 recent observational studies such as from the German Stroke Registry report proportions as high as 21.2%.14 Several studies that included patients on VKA at the time of stroke found comparable recanalisation and symptomatic intracranial haemorrhage rates when compared with non-anticoagulated patients.14–18

Ischaemic stroke on DOAC

Due to ease of use and efficacy in stroke prevention in non-valvular AF, DOACs are increasingly being used in preference to VKA. In contrast to VKA, DOACs have a relatively short half-life (6–14 hours) with rapid onset of anticoagulant effect.19 Unlike VKA, where IVT is considered reasonable with INR ≤1.7, current major stroke guidelines recommend against routine IVT within 48 hours of DOAC intake unless specific DOAC coagulation assays are normal.9 12 To complicate decision making in these patients, routine quantitative coagulations assays do not consistently correlate with anticoagulant effect of DOACs. Specific assays such as calibrated anti factor Xa-activity, dilute thrombin time or ecarin time are needed to more precisely measure the anticoagulant effect.19 Due to guideline recommendations and lack of universally accessible rapid and reliable laboratory measures of DOAC effect, many patients on DOAC, while potentially eligible, do not receive IVT due to concerns of haemorrhagic transformation.20

IVT in DOAC patients

Although randomised trial data are lacking, a growing number observational studies are investigating thrombolytic therapy in patients on DOAC treatment.21–23 The observational studies used different approaches to patient selection, often based on local standard operating procedures such as predefined levels of calibrated anti factor-Xa activity.22 24 A summary of observational studies including 10 or more patients treated with reperfusion therapies is presented in table 1.14–18 20 21 24–39

Table 1

Summary of observational studies with >10 patients describing reperfusion therapy in patients on DOAC

A study of Japanese stroke centres reported 71 DOAC patients who received IVT, including 15 (21%) of patients who also underwent endovascular thrombectomy (EVT), with no symptomatic intracranial haemorrhage (sICH).28 Where available, the time between last medication intake and reperfusion was mostly within 24 hours. The method for decision making for IVT was not reported, but it appears that clinicians used the INR cut-off for warfarin as all IVT patients had INR ≤1.7. A similar approach was likely adopted by the centres participating in the Get-With-the Guidelines Stroke Registry,11 which reported sICH rate of 4.8% in DOAC patients treated with IVT with a median INR of 1.1. Although the method for selection was not reported, it is possible the clinicians combined time from last ingestion (up to 7 days) and baseline INR level in their decision making.

In centres with rapid access to calibrated anti factor Xa-activity levels, local standard operating procedures for IVT with predefined plasma drug-level cut-off have been successfully adopted and reported. In patients with normal renal function and recent DOAC ingestion, plasma DOAC levels of 20–50 ng/mL was shown to be safe levels for IVT administration, however, these data should be cautiously interpreted as it is based on observational studies.20 21 24 IVT may be reasonable after a careful risk and benefit assessment in patients with intermediate drug levels of 50–100 ng/mL, while IVT is generally withheld in patients with levels>100 ng/mL.22 The most recent published study using this method reported a sICH rate of 3.1% from 69 DOAC patients treated with IVT based on plasma DOAC levels over a 6-year period from the Swiss Stroke Registry.20 The low rate is consistent with prior reports from Europe with sICH rates of ~4% (1/2427 patients) reported in Germany and 4% (2/51 patients) in the multicentre European NOACISP (Novel Oral Anticoagulants in Stroke Patients) study group.21

The development of reversal agents provides another feasible avenue for IVT in DOAC patients. Idarucizumab is a humanised monoclonal antibody fragment binding specifically to dabigatran and its metabolites,40 reversing the anticoagulant effect of dabigatran within minutes. Idarucizumab administration provides a rapid and permanent reversal of anticoagulant parameters associated with dabigatran. The administration of idarucizumab in dabigatran pretreated patients prior to IVT has been described in small case series including data from national registries in Germany and New Zealand.35–37 By contrast, published data on the use of IVT following anticoagulation reversal with andexanet alpha is limited to a single case report.41 Andexanet alpha binds and sequesters factor Xa inhibitors rivaroxaban and apixaban. In addition, andexanet alpha inhibits the activity of tissue factor pathway inhibitor, increasing tissue factor-initiated thrombin generation.40 In clinical trials, antifactor Xa activity increased to placebo levels ~2 hours after completion of the infusion.42

Both reversal agents must be administered via infusion; andexanet alpha is more costly relative to idarucizumab (US$~18 000 vs US$2400, respectively).22

The sICH rate is similar to non-anticoagulated patients and patients on warfarin, occurring in ~≤4% of patients from the reported cohorts (table 1). Functional outcome data, however, were not uniformly reported in these studies. As the relative safety reported in published data was based on observational studies only and in light of the known thrombotic side effects of potential reversal agents for OACs, current major guidelines do not recommend administration of reversal agents to administer thrombolysis in patients with ischaemic stroke on DOAC therapy.9 However, some experts support IVT after idarucizumab reversal but not after andexanet alpha in the current European Stroke Organisation guideline.12 RCTs are definitely needed to understand the benefits/risks of systemic thrombolysis in patients with acute ischaemic stroke on VKA with INR >1.7 and patients on DOACs, with or without administration of reversal agents.


Similar to IVT, a number of observational cohort studies have reported feasibility and safety of EVT in DOAC patients with LVO (table 1) achieving similar successful reperfusion rates and safety outcomes to EVT trials.14 16 18 24 26 34 38 39

The largest cohort consisted of 827 patients from the German Stroke Registry, with successful recanalisation (modified thrombolysis in cerebral infarction (mTICI) score 2b-3) achieved in 85%. Although sICH was not reported, the rate of any intracranial haemorrhage at 24 hours in DOAC patients (11.7%) was comparable to patients on VKA (12.5%) and patients without anticoagulation (14.6%).21 IVT pretreatment also did not increase haemorrhage rates.

Although the current routine practice in LVO patients without anticoagulation presenting within 4.5 hours is to offer IVT prior to EVT, four recently completed RCTs and a meta-analysis suggest similar safety and functional outcomes between patients treated with or without IVT prior to EVT.43 The data from these trials and safety of thrombectomy in DOAC patients from observational data support bypass of IVT in LVO DOAC patients at thrombectomy—capable centres in whom EVT can be started without delay.

Outcomes in patients with ischaemic stroke on anticoagulation

Short-term disability

Several studies report similar functional outcomes at 90 days14 17 18 when comparing patients with ischaemic stroke with versus without anticoagulation at time of stroke. It should be noted that comparing groups with and without anticoagulation may introduce bias if not restricting patients to those with AF only. Based on Get-with-the guideline data, patients with AF and ischaemic stroke had better outcomes with prior anticoagulation use when compared with those without or subtherapeutic anticoagulation.44

When comparing patients on DOAC versus VKA, VKA usage was associated with worse functional outcome at 3 months.45 In a prospective cohort on patients who had a stroke on anticoagulation using univariate and multivariate regression analysis, preceding VKA as compared with DOAC was associated with lower rates of good functional outcome without excess mortality.46 Similarly, patients on DOAC had a higher odds of excellent functional outcome at 90 days when compared with those on VKA in the Endovascular Treatment in Ischaemic Stroke prospective registry.39 In the consecutively collected data cohort in the Swiss Stroke registry, when adjusting for confounders, pretreatment with DOAC, but not VKA was significantly associated with favourable outcome at 3 months.20 Considering the fact that VKAs are typically used in patients with more severe comorbidities such as valvular disease, mechanical valves, hypercoagulable states, renal failure, the validity of observational data discussed above should also be considered carefully. None of these important factors are captured by the currently used risk scores, and they are hard to identify correctly when using administrative datasets.

Risk of stroke recurrence

Several studies have shown a high rate of stroke recurrence in patients with AF suffering an ischaemic stroke. Specifically, observational studies found that high CHA2DS2-VASc score,47 high National Institutes of Health Stroke Scale,47 large ischaemic stroke47 and type of anticoagulant administered47 48 were independently associated with greater risk of stroke recurrence and bleeding. When compared with patients who had a stroke without prior anticoagulation, those with stroke on anticoagulation have a higher risk of recurrent stroke.4 5 49 Switching anticoagulation does not appear to reduce the risk of recurrent stroke,4 5 suggesting additional pathomechanisms at play in this high-risk patient group.

Secondary prevention strategies in patients with ischaemic stroke on anticoagulation

Determining potential reason(s)

Determining the mechanism or a combination of mechanisms of ischaemic stroke despite anticoagulant therapy in patients with AF is essential for the choice of adequate secondary prevention strategies.4 5

Recent data4 5 49 provide evidence that these patients are at increased risk of future ischaemic strokes, but reasons and mechanisms are poorly studied and understood. In theory, this may comprise the following categories: (1) OAC failure, (2) medication error (ie, stroke due to non-adherence or inappropriate dosing), (3) stroke from competing, non-AF stroke aetiology (ie, ipsilateral high-grade stenosis, cerebral small vessel disease or others) and (4) other cardioembolism despite anticoagulant therapy (competing non-AF cardioembolic aetiology). Patients with AF who have a stroke despite anticoagulant therapy should undergo a thorough workup to identify competing stroke aetiologies (figure 1).

Figure 1

Diagnostic workup for patients with atrial fibrillation and ischaemic stroke despite anticoagulation. DOAC, direct oral anticoagulant; DWI, diffusion-weighted imaging; FLAIR, fluid-attenuated inversion recovery; CSF, cerebrospinal fluid; TEE, transoesophageal echocardiogram.

First, information to identify potential medication errors should be collected from patient and/or caregiver interviews and coagulation tests. Use of off-label low-dose DOAC therapy has been identified as one of the causes of stroke despite anticoagulation in patients on DOAC therapy.8 Concomitant medications with drugs influencing metabolisation and clearance of DOACs may also provide important information, although the list of drugs that significantly decrease the efficacy of DOACs is shorter than warfarin (eg, rifampicin, carbamazepine, phenytoin, phenobarbital and St. John’s wort). Adherence in patients taking DOACs must rely on evidence from patients and/or care givers about when and how they took the drug. In addition, drug-specific coagulation assays including calibrated antifactor Xa activity levels may provide further information.50 However, these levels are difficult to interpret, as there are no target levels (in contrast to INR in patients taking VKA). Absolute levels expected at different time points after last intake (peak/trough levels) have a wide CI51 52 and there is an important heterogeneity among patients who had a stroke, only in parts explained by dosage, renal function and other confounders.53 Antifactor Xa-activity should be used to determine non-adherence only if very low levels are measured on admission, for example, <30 ng/mL indicates that with very high likelihood, no direct factor Xa-inhibitor has been used in the last 48 hours. Overall, patients with OAC-failure due to medication use errors create a therapeutic challenge as there is no approved way of using monitoring, no data to support dose adjustment beyond the regular full dose, and no guarantee that a patient who missed doses will not do so again.

Second, a diagnostic workup should comprise additional testing to identify potentially competing, non-AF mechanisms of stroke. Patients with AF commonly harbour other non-AF related risk factors for stroke.4 6 The use of brain MRI allows for analysis of possible stroke mechanisms by taking into account patterns (embolic, deep, large or small scattered), distribution (one or more vascular territories) on diffusion-weighted imaging (DWI) and presence of prior strokes on fluid-attenuated inversion recovery imaging.54

Concomitant carotid artery stenosis may be as frequent as 12.5%55 in patients with AF. All patients should, therefore, undergo vascular imaging of the brain-supplying arteries either using CT/MR angiography and/or ultrasound to detect ipsilateral high-grade stenosis.

Clinical lacunar stroke syndromes and deep infarcts on DWI point to cerebral small-vessel/branch occlusive disease.54 56 By contrast, for patients with infarct patterns within one arterial system, additional vascular imaging such as vessel wall imaging or fat saturated imaging sequences may help detect dissection, unstable plaque or vasculitis.54 Cerebrospinal fluid (CSF) analysis may further be helpful in cases of suspected vasculitis. Infarct distributions that include multiple territories simultaneously or comprise single territories at different stages raise suspicion for aortic arch disease or paradoxical sources of embolism, in which case further imaging with CT angiography, transoesophageal echocardiogram (TEE), and transcranial duplex sonography may be helpful.54

Hypercoagulability due to cancer may be another important reason for stroke despite anticoagulation in patients with AF.54 D-dimer levels commonly >20× higher than in patients without cancer and multivascular patterns of small, scattered infarcts should raise suspicion for underlying malignancy and prompt cancer screening.54 57

In addition, coagulation testing for hypercoagulability including antiphospholipid antibody syndrome may be important in select patients as they should remain on VKA, which was found to be superior to DOACs in RCTs.

Third, cardiac workup should comprise structural cardiac imaging either using ultrasound (transthoracic or transoesophageal) or other adequate imaging modalities (cardiac CT or MRI) to detect intracardiac thrombus (which may require use of VKA long-term and immediate anticoagulation in patients where usually anticoagulation would be delayed for days or weeks), assess for infective endocarditis along with blood cultures based on clinical suspicion, or other rare cardiac conditions (ie, cardiac tumours) that may require dedicated treatments. In addition, cardiac imaging (eg, LA enlargement) and blood biomarkers may help provide evidence of severe atrial dysfunction as a potential mechanism of thrombus formation and identify high-risk patients.41 51

Both LA appendage (LAA) function and morphological features of shape, orifice size and fibrosis are associated with increased risk of ischaemic stroke,56 but the value of this information for management and secondary prevention is currently unknown.46 58 If LAA imaging using either TEE, cardiac CT/MRI shows LAA thrombus, this would suggest likely an AF-related embolism.

A comprehensive workup including the aforementioned steps and using a multidirectional classification of stroke aetiologies (eg, atherosclerosis, small-vessel disease, cardioembolism, other, dissection (ASCOD)), medication errors, competing stroke aetiologies and traditional (uncontrolled/poorly controlled) risk factors is recommended to identify causes of ischaemic stroke on OAC.8 59 Following a thorough workup, secondary prevention strategies should be individualised to target the most probable stroke mechanism.

Choice of antithrombotic medication

When ischaemic stroke occurs despite adherence to therapeutic anticoagulation, one naturally considers whether the previous antithrombotic regimen should be changed. Unfortunately, there are no clinical trials to indicate the preferred course of action.

Two observational studies have shown that patients with AF and a breakthrough ischaemic stroke despite prior anticoagulation have 50%–60% higher risk of future ischaemic stroke than patients with AF and ischaemic stroke who were not previously taking anticoagulation.4 5 This suggests that at least some patients with breakthrough ischaemic stroke (OAC failures) have factors, not yet defined, that put them at higher future risk. Therefore, breakthrough ischaemic strokes should never be dismissed as stochastic events—that is, they are not only due to bad luck. This potentially motivates a change in antithrombotic strategy, although it is not clear what to do: whether to switch drug type, target a higher INR range with warfarin, or add an antiplatelet drug.

Of the head-to-head trials of DOACs compared with warfarin, only dabigatran at the 150 mg two times per day dose showed a reduction in ischaemic stroke, but this study was criticised for lack of blinding and inclusion of a very low-risk AFib population among other issues.60 However, network meta-analyses indirectly comparing the DOACs to each other have not found any differences in ischaemic stroke risks.61 Two observational studies of patients with AF and breakthrough ischaemic stroke despite anticoagulation failed to find an association between changing anticoagulants—most commonly, switching from warfarin to a DOAC—and lower future risk.4 5 However, these studies were not randomised and were under-powered to detect any but the largest differences.

Patients with AF with a breakthrough ischaemic stroke should have a complete work up to identify alternative sources of embolism or thrombosis unrelated to AF. Occasionally, one may identify an alternative cause of stroke with a clear indication for a specific antithrombotic strategy. Examples include antiphospholipid antibody syndrome, hypercoagulability of malignancy, or infective endocarditis. However, if an alternative cause is found, it is more likely to be a common stroke mechanism such as small vessel occlusion or unstable atherosclerotic plaque with arterial embolism, for which the optimal management in the setting of concurrent AF is unclear. Most of these stroke aetiologies such as intracranial atherosclerotic disease, cerebral small-vessel disease and other non-AF factors are better treated with antiplatelet therapy. The addition of an antiplatelet drug has the theoretical advantage of reducing platelet activation and aggregation by unstable plaque. However, in the Atrial Fibrillation and Ischaemic Events with Rivaroxaban in Patients with Stable Coronary Artery Disease trial, there was no benefit to adding aspirin to rivaroxaban in patients with AF and coronary heart disease more than 1 year after stenting, with an increase in major bleeding and mortality.62 The findings of this trial cast some doubt on whether adding an antiplatelet drug would be effective for patients with AF and atherosclerotic stroke.

LAA closure (LAAC) with endovascular devices such as WATCHMAN FLX and AMPLATZER AMULET is a non-pharmacological approach for AFib-related stroke prevention. Close to 100% of thrombi found in patients with AF occur in LAA and the WATCHMAN FLX device is approved by FDA for use in non-valvular patients with AF with high embolic risk who have a reasonable rationale to consider an approach that does not rely on lifelong anticoagulation.63 More recently, the PRAGUE 17 study showed non-inferiority of LAAC with WATCHMAN FLX or AMPLATZER AMULET over DOACs in high-risk patients with AF.64 Another recent study showed much improved procedural safety with a periprocedural major complication rate of 0.5%, an advance that led to the approval of the new generation WATCHMAN FLX device in the USA.65 It should be remembered that LAAC caries a relatively low risk of upfront procedural risks and its availability should not stop the clinician from performing a full diagnostic workup. Patients with AF who sustain an ischaemic stroke while using OAC, an RCT comparing LAAC and long-term continued OAC to medical management with OAC might be a worthwhile effort given data showing that in patients undergoing cardiac surgery with nearly 75% continued on anticoagulation therapy, LAA occlusion was superior to standard of care in reducing the risk of ischaemic stroke.66 LAAC and long-term continued OAC strategy cannot be adopted at a large scale before such an RCT shows positive results. This may, however, be used in select patients based on shared-decision-making discussions between the patient, stroke neurologist and device implanting cardiologist.

In conclusion, while it is now clear that patients with AF and breakthrough ischaemic stroke despite anticoagulation are at higher risk of ischaemic stroke recurrence, there is little empirical evidence on which to base decisions on antithrombotic therapy. As previously discussed, a detailed medication history and assessment for concurrent non-cardiac and cardiac stroke aetiologies should be performed.

For patients previously taking VKA who had an ischaemic stroke with low INR, switching to a DOAC might be an option, but there is no strong evidence to suggest that this approach will reduce recurrence risk. Improved patient compliance is important in patients with reduced adherence, but the success rate of this approach is uncertain. If poor compliance is due to cost of DOACs, switching from a DOAC to VKA should be considered. Furthermore, patients with AF with concurrent antiphospholipid antibody syndrome, mechanical valves or left ventricular thrombus are not candidates for DOACs and should remain on or be switched to VKA.

The combination of anticoagulation plus an antiplatelet agent should probably be avoided given the higher risks of major bleeding,62 except possibly for short-term use in patients with arterial embolism from unstable atherosclerotic plaque.

Conclusions and future directions

The management of patients with AF who have a stroke despite anticoagulation is challenging. In terms of acute stroke treatment, VKA use with INR >1.7 or DOAC use in the past 48 hours would exclude them from IVT which has been shown to reduce disability12 in patients with AF and even reduce mortality in patients with ischaemic stroke in the setting of AF. Furthermore, they face a high risk of recurrent ischaemic stroke which does not seem to be reduced by switching to a different class of oral anticoagulants.4 5 Studies are needed to test the safety of IVT in patients with acute ischaemic stroke on DOAC treatment. Furthermore, recent evidence suggests equivalence of LAAC with endovascular devices to DOACs64 as well as a potential benefit of LAA occlusion in reducing ischaemic stroke in patients with AF undergoing coronary artery bypass surgery with nearly 75% of patients in the study remaining on anticoagulation.66 Therefore, randomised trials are needed to test the utility of LAAC plus continued anticoagulation versus anticoagulation only in patients with AF who develop an ischaemic stroke despite anticoagulation and whose stroke mechanism is thought to be related to anticoagulant failure.

Ethics statements

Patient consent for publication



  • Twitter @guroledip

  • Contributors CS, TYW, DW, DJS, EES, MEG and SY contributed to manuscript preparation, revision, concept and design.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests CS: received departmental funding from Massachusetts General Hospital for participation in the Neuro AFib study. SY: Medtronic (uncompensated). ESS: administers a research contract to the University of Calgary to provide brain MRI measurements to the Ottawa Heart Institute for the OCEAN clinical trial. MEG: received NIH grants and research funding to the hospital from AVID (a fully owned subsidiary of Eli Lilly), Boston Scientific (ISR) and Pfizer. DJS: advisory board fees (Portola/Alexion, Bayer Switzerland AG), unrestricted research grant (Portola/Alexion), academic research funds Swiss National Science Foundation, Bangerter Rhyner Foundation, Swiss Heart Foundation.

  • Provenance and peer review Commissioned; externally peer reviewed.