When anticoagulant therapy is not enough: recurrent stroke

Abstract

Despite the use of oral anticoagulants, the residual risk of recurrent ischaemic stroke in patients with atrial fibrillation (AF) is ∼4% per year; however, in patients who experience a stroke despite oral anticoagulant therapy (OAT), this risk approaches 9% per year. It is therefore essential to identify patients at higher risk and to develop new strategies for secondary prevention. These strategies include the exclusion of non-AF-related causes, optimization of OAT and its early initiation after stroke, as well as an aggressive rhythm-control approach. Additional non-pharmacological options include surgical or percutaneous left atrial appendage occlusion, although its effectiveness in secondary prevention after ischaemic stroke is still under evaluation. Novel pharmacological therapies targeting factors XI and XIa are currently under investigation and may provide comparable efficacy with improved safety compared with direct oral anticoagulants. Conversely, switching from one anticoagulant to another or adding antiplatelet therapy is completely ineffective. The mechanisms underlying recurrent stroke despite anticoagulant therapy are therefore complex, making an individualized, patient-specific approach mandatory.

Cerebral thromboembolism is the most feared complication of atrial fibrillation (AF). Oral anticoagulant therapy (OAT) reduces the risk of stroke by ∼60%, resulting in an absolute annual incidence of only 1–2%, at least in the intervention arms of randomized clinical trials. The introduction of direct oral anticoagulants (DOACs) has further expanded the population of patients eligible for this therapy. However, in routine clinical practice, stroke recurrence in patients with AF despite OAT now accounts for almost 40% of AF-related strokes.¹

The exact magnitude of this ‘residual risk’ has not yet been clearly defined and varies according to the populations studied, being lower in randomized trials and higher in real-world studies. In a recent meta-analysis of 23 studies addressing this issue, including a total of 140 307 patients,² the overall annual incidence of recurrent ischaemic stroke was 3.75%. In the subgroup of patients who experienced the index stroke while already receiving OAT, the cumulative annual incidence of recurrent ischaemic stroke was 7.20%.

Based on these data, the estimated 5-year cumulative risk of a first recurrent ischaemic stroke in patients with AF is 17.4%, whereas in patients who experienced a stroke despite OAT, the risk of ischaemic recurrence increases to 31.2%. The temporal pattern of risk appears to be non-linear and particularly elevated during the first year after the index stroke.

The key findings of this analysis can therefore be summarized as follows:

1. Despite the widespread use of oral anticoagulants, the residual risk of recurrent ischaemic stroke is ∼4% per year; consequently, one in five patients with AF will experience a recurrent ischaemic stroke within 5 years.

2. The risk of stroke recurrence is approximately twice as high as that reported in randomized clinical trials.

3. Patients with AF who experience a stroke despite anticoagulant therapy represent a very high-risk group, with an annual probability of stroke recurrence approaching 9%.

Causes of stroke recurrence

Although stroke recurrence despite OAT is often perceived as a pharmacological treatment failure, it is important to emphasize that the aetiology of ischaemic stroke occurring despite OAT in patients with atrial fibrillation (AF) is heterogeneous. It may include both AF-related mechanisms—such as inadequate intensity of anticoagulation due to underdosing, poor adherence to therapy, failure to consider food interactions (for example, rivaroxaban must be taken with food to achieve optimal bioavailability), or drug–drug interactions (all DOACs interact with P-glycoprotein and CYP3A4 inducers, such as phenytoin, carbamazepine, St John’s wort, or rifampicin), inappropriate perioperative management, or cardioembolism despite adequately conducted OAT—as well as mechanisms unrelated to the arrhythmia (e.g. stroke caused by large-artery atherosclerosis, cerebral small-vessel disease, aortic arch disease, or the presence of malignancy).

Ischaemic stroke in this setting can therefore be classified into three main categories:

1. Stroke not related to AF, accounting for approximately one quarter of cases.

2. Pharmacological treatment error, accounting for about one third of cases; and

3. Cardioembolism despite appropriate OAT, which represents the most common cause.

A comprehensive aetiological work-up is therefore mandatory to assess the adequacy of anticoagulant dosing and treatment adherence, as well as to identify other potential stroke mechanisms unrelated to AF. In particular, although on-label dose reductions are appropriate in selected patients with AF, the inappropriate use of reduced doses of DOACs is increasingly recognized in clinical practice as a common risk factor for AF-related stroke. Such an aetiological assessment is also crucial because the presence of large-vessel stenosis or small-vessel–related stroke may require a different secondary prevention strategy. Overall, the evidence available to date suggests that the current treatment paradigm for secondary stroke prevention in AF is suboptimal. Further research is therefore urgently needed to identify patients at highest risk and to develop novel therapeutic strategies aimed at reducing this residual risk.

Risk stratification may benefit from a detailed assessment of AF characteristics. Patients with persistent or permanent AF are at higher risk than those with paroxysmal AF, as demonstrated by post hoc analyses of the major non-vitamin K antagonist oral anticoagulant trials.³ Moreover, persistent AF is associated with AF-related stroke recurrence, whereas paroxysmal AF is more frequently associated with stroke recurrence unrelated to AF, suggesting that paroxysmal AF may not be the true cause of stroke in some patients. A further distinction appears to exist between AF diagnosed prior to the index stroke—which is more common in patients with recurrent ischaemic stroke than in those without recurrence—and AF frequently detected by cardiac monitoring during the acute hospitalization for stroke.⁴ Indeed, it remains unclear whether AF observed in this acute phase is the cause of the stroke or rather a consequence of it (so-called neurogenic AF).

Therapy

The optimal management of patients who experience ischaemic stroke despite anticoagulant therapy remains unclear, and current guidelines do not provide specific recommendations. The high rates of recurrence indicate that, in addition to optimizing currently available pharmacological treatments, novel approaches are required. These may include new drugs, such as factor XIa inhibitors, new strategies, such as aggressive management of atrial fibrillation (AF), or non-pharmacological interventions, such as percutaneous left atrial appendage (LAA) occlusion.

Optimization of anticoagulant therapy

In patients who experience recurrent stroke despite OAT, once underdosing, non-adherence, and potential drug–drug interactions have been excluded, five therapeutic strategies have historically been proposed:

1. Continuation of the same drug at the same dose.

2. Switching to another direct oral anticoagulant (DOAC).

3. Continuation of the same drug with dose escalation.

4. Switching to warfarin.

5. Addition of antiplatelet therapy.

All available studies consistently demonstrate that changing the DOAC or its dose is not superior to continuing the same agent, whereas switching from the index DOAC to warfarin or to another DOAC is associated with an increased risk of recurrent ischaemic stroke.⁵ In the case of switching to warfarin, this excess risk can be attributed to a lower time in therapeutic range in patients treated with this agent; in contrast, the mechanism underlying the increased risk associated with switching from one DOAC to another remains unclear. Finally, the addition of antiplatelet therapy is associated with an increased risk of major bleeding and, paradoxically, of ischaemic stroke, likely as a consequence of longer interruptions of antithrombotic treatment following bleeding events.

Timing of anticoagulant therapy

The risk of stroke recurrence is highest in the early period following the index event and subsequently declines over time. Historical data suggest that, in the absence of anticoagulation, the risk of early stroke recurrence may reach up to 1% per day during the first 10 days after an ischaemic stroke. Consequently, the potential benefit of anticoagulant therapy is particularly high during this early phase, given the elevated absolute risk of ischaemic stroke.

Unfortunately, pivotal phase 3 trials excluded patients with recent ischaemic stroke because of concerns regarding an increased risk of intracranial haemorrhagic complications, resulting in substantial uncertainty about the optimal timing for initiating OAT after stroke. The recent ELAN trial⁶ (Early vs. Later Anticoagulation for Stroke with Atrial Fibrillation), which randomized 2013 participants to early OAT (within 48 h after a minor or moderate stroke or on days 6 or 7 after a major stroke) or later OAT (days 3 or 4 after a minor stroke, days 6 or 7 after a moderate stroke, or days 12, 13, or 14 after a major stroke), estimated that the 30-day incidence of the composite outcome of stroke, systemic embolism, hemorrhage, or death ranged from 2.8% points lower to 0.5% points higher with early initiation of DOACs compared with delayed initiation.

Direct factor XI and XIa inhibitors

A new generation of oral anticoagulants is currently being evaluated in phase 2 and phase 3 randomized clinical trials. Given the central role of factor XIa in thrombus amplification, but its subsidiary role in haemostasis, it has been hypothesized that inhibition of factors XI and XIa may prevent thrombus formation with only a minimal increase in major bleeding events. Agents currently under investigation include asundexian and milvexian, which are direct oral factor XIa inhibitors, and abelacimab, a highly selective monoclonal antibody administered once monthly via subcutaneous injection.

In the PACIFIC-AF trial⁷ (Safety of the Oral Factor XIa Inhibitor Asundexian Compared With Apixaban in Patients With Atrial Fibrillation: a Multicentre, Randomized, Double-Blind, Double-Dummy, Dose-Finding Phase 2 Study), 876 participants with AF were assigned to apixaban or to two different doses of asundexian (20 mg or 50 mg). The incidence of ischaemic events was similar across all three groups, whereas bleeding events were lower with both doses of the novel agent. Based on these promising safety results, phase 3 trials investigating stroke prevention in patients with AF using asundexian and milvexian were initiated; however, the trial evaluating asundexian was prematurely terminated after an interim analysis failed to demonstrate efficacy. Abelacimab is currently being studied in patients with AF who are considered unsuitable for anticoagulant therapy, although patients with a history of stroke are excluded from this study.

Aggressive management of atrial fibrillation

Rhythm-control therapy using antiarrhythmic drugs or catheter ablation has recently emerged as a novel therapeutic option, in addition to anticoagulation, for stroke prevention in patients with AF. The EAST-AFNET 4 trial⁸ (Early Rhythm-Control Therapy in Patients With Atrial Fibrillation) demonstrated that this strategy—primarily based on antiarrhythmic drugs with catheter ablation as second-line therapy—in patients with recently diagnosed AF at high-risk of stroke reduced the composite endpoint of cardiovascular death, stroke, or hospitalization for heart failure or acute coronary syndrome compared with usual care (predominantly rate control).

Notably, during the remaining follow-up period of 4.1 years, 81% of the treatment effect was explained by the presence of sinus rhythm at 12 months, whereas no reduction in cardiovascular outcomes was observed in patients in whom early rhythm-control therapy failed to restore sinus rhythm by that time point (hazard ratio 0.94). A prespecified subgroup analysis of the EAST-AFNET 4 trial⁹ (Early Rhythm-Control Therapy for Atrial Fibrillation in Patients With a History of Stroke) involving 217 patients with a prior history of stroke—therefore highly relevant to the present topic—confirmed a lower incidence of stroke (3% vs. 7%) in the early rhythm-control group, with a greater treatment effect than that observed in patients without a history of stroke.

Similarly, in the RAFAS study¹⁰ (Risks and Benefits of Early Rhythm Control in Patients With Acute Strokes and Atrial Fibrillation: a Multicentre, Prospective, Randomized Study), which included 300 patients with acute stroke and AF, early rhythm-control therapy (initiated within 2 months of the index stroke) was associated with a lower incidence of ischaemic stroke at 12 months.

Overall, the available evidence suggests that the benefits of early initiation of rhythm-control therapy may be greater in patients with a history of ischaemic stroke than in the general AF population, given their heightened risk of recurrence. Data from the RAFAS study¹⁰ and from the subgroup of patients with prior stroke in the EAST-AFNET 4 trial⁹ provide reassuring evidence regarding the safety and efficacy of this strategy; however, its precise role and optimal timing have yet to be definitively established.

Left atrial appendage occlusion

Approximately 90% of potentially embolic thrombi originate in the LAA. For this reason, LAA occlusion represents a reasonable therapeutic option for stroke prevention in patients with atrial fibrillation (AF). This can be achieved either by surgical ligation or by percutaneous occlusion. The procedure was initially developed for patients with AF who were not eligible for OAT, but it has subsequently been shown to be non-inferior to OAT in terms of efficacy in three randomized clinical trials (two compared with warfarin and one compared with DOACs).

However, no randomized clinical trial has yet investigated whether surgical or percutaneous LAA occlusion reduces the risk of stroke recurrence in patients with AF who experience ischaemic stroke despite OAT. Nevertheless, in the large multicentre LAAOS III trial¹¹ (LAA Occlusion during Cardiac Surgery to Prevent Stroke), 4770 patients with AF and a CHADSVASc score ≥2 undergoing cardiac surgery for valve replacement or coronary artery bypass grafting were randomized to undergo surgical LAA occlusion or no-occlusion, in addition to usual care, which included oral anticoagulation. After a mean follow-up of 3.8 years, ischaemic stroke or systemic embolism occurred in 4.8% of patients in the occlusion group and in 7.0% of those in the no-occlusion group (hazard ratio 0.67). This reduction in stroke and systemic embolism was observed despite 75–80% of patients in both groups receiving OAT throughout the duration of the study.

These findings suggest that surgical LAA occlusion may provide a substantial reduction in the risk of stroke and systemic embolism when used as an adjunct to OAT and highlight the promising potential of combining mechanical and anticoagulant therapies as a strategy to optimize stroke prevention in patients with AF who have already experienced a stroke despite anticoagulation. However, surgical LAA occlusion is too invasive for patients without other indications for cardiac surgery and is therefore not applicable to the majority of patients with stroke.

Percutaneous LAA occlusion represents a less invasive alternative to surgical ligation; however, it remains unclear whether the benefits observed with surgical occlusion can be extrapolated to the percutaneous approach. This uncertainty is compounded by the 2–5% risk of device-related thrombus formation and the not infrequent occurrence of incomplete LAA closure.

Future studies are therefore required to explore the potential role of percutaneous LAA occlusion in reducing the risk of stroke recurrence in patients with AF who experience ischaemic stroke while receiving anticoagulant therapy. In the coming years, results are expected from the ongoing ELAPSE trial¹² (Early Closure of LAA for Patients with Atrial Fibrillation and Ischaemic Stroke Despite Anticoagulation Therapy), which specifically addresses this clinical scenario.

Conclusions

Despite the widespread use of oral anticoagulants, the residual risk of recurrent ischaemic stroke in patients with AF is ∼4% per year; however, in patients who experience a stroke despite OAT, this risk approaches 9% per year. It is therefore essential to identify patients at higher risk and to develop new strategies for secondary prevention. These include the exclusion of stroke mechanisms unrelated to AF—given that approximately one third of recurrences have a non-arrhythmic aetiology—the optimization of OAT (including correction of non-adherence or inappropriate dosing), early initiation of anticoagulation after stroke, and an aggressive rhythm-control strategy, for which the optimal timing and most effective approach (antiarrhythmic drugs, electrical cardioversion, or catheter ablation) still need to be defined.

Additional non-pharmacological options include surgical or percutaneous LAA occlusion, although their efficacy in secondary prevention after ischaemic stroke is still under evaluation. Finally, novel pharmacological therapies targeting factors XI and XIa are currently being tested in phase 2 and phase 3 trials and may offer similar efficacy with improved safety compared with DOACs. In contrast, switching from one anticoagulant to another or adding antiplatelet therapy is entirely ineffective.

The causes of stroke recurrence despite anticoagulant therapy are therefore complex, making an individualized, patient-specific approach mandatory.

Funding

I have not received funding.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Disclaimer

This paper was originally published in the Italian language as ‘Quando la terapia anticoagulante non basta: l'ictus recidivante’, in the Volume degli Atti del Congresso “Conoscere e Cuare il Cuore 2026”, published by Centro per la Lotta contro l'Infarto for distribution at the CCC Conference. This paper was translated by Dr. Mario Albertucci, representative of the CLI Foundation, and republished with permission.