Proposed Section Structure
For this topic, a clinically appropriate structure is: Highlights; Clinical Background; Why Hypertension Matters in Cryptogenic Stroke; Study Design and Methods; Key Results; Mechanistic and Clinical Interpretation; Limitations and Generalizability; Practice Implications and Future Directions; Conclusion; Funding, Registration, and Citation.
Highlights
In this exploratory analysis of the ARCADIA randomized clinical trial, hypertension with high-risk features appeared to modify the association between antithrombotic strategy and recurrent thromboembolic events after cryptogenic stroke.
Among patients without high-risk hypertensive features, apixaban was associated with a lower risk of recurrent ischemic stroke or systemic embolism than aspirin, with a hazard ratio of 0.43 (95% CI, 0.22-0.85).
Among patients with high-risk hypertensive features, no benefit of apixaban was observed; the hazard ratio was 1.68 (95% CI, 0.78-3.62), suggesting possible heterogeneity in underlying stroke mechanisms.
The findings support the hypothesis that unrecognized hypertensive arteriopathy may dilute the effect of anticoagulation in trials of cryptogenic stroke or embolic stroke of undetermined source.
Clinical Background
Secondary prevention after cryptogenic ischemic stroke remains one of the more difficult areas in vascular neurology. Although a substantial proportion of apparently nonlacunar ischemic strokes are suspected to be embolic, randomized trials comparing direct oral anticoagulants with aspirin in broadly defined embolic stroke of undetermined source have not shown an overall benefit of anticoagulation. ARCADIA was designed to test a more selective hypothesis: that patients with cryptogenic stroke and evidence of atrial cardiopathy, even without documented atrial fibrillation, might have an atrial thromboembolic substrate responsive to anticoagulation.
The main ARCADIA trial did not demonstrate superiority of apixaban over aspirin overall. The present analysis asks whether one important source of biological heterogeneity was insufficiently accounted for: hypertension-related cerebrovascular disease. This is clinically plausible. Long-standing hypertension can produce small-vessel arteriopathy, white matter injury, deep infarction, vascular remodeling, and cardiac structural changes such as left ventricular hypertrophy. Some patients classified as having cryptogenic stroke may therefore have a nonembolic stroke mechanism that would not be expected to improve with anticoagulation.
This question is highly relevant because mechanism-based treatment is central to modern stroke prevention. If a subgroup with hypertensive arteriopathy is mixed with patients who have occult cardioembolism, any beneficial effect of anticoagulation in the latter group could be obscured.
Why Hypertension Matters in Cryptogenic Stroke
Hypertension is both a major vascular risk factor and a marker of underlying target-organ damage. In stroke medicine, severe or poorly controlled hypertension may signal cerebral small-vessel disease, while left ventricular hypertrophy reflects chronic pressure overload and broader hypertensive end-organ injury. Either finding may identify patients whose stroke pathobiology is less likely to be predominantly embolic from the atrium.
The ARCADIA exploratory analysis defined hypertension with high-risk features using clinically accessible markers: systolic blood pressure 160 mm Hg or greater at enrollment, left ventricular hypertrophy on echocardiography, or both. This definition is not a formal stroke subtype classification, but it is pragmatic and reproducible. It attempts to capture the subset in whom hypertensive arteriopathy may be sufficiently prominent to alter treatment response.
Study Design and Methods
This was an exploratory secondary analysis of the Apixaban to Prevent Recurrence After Cryptogenic Stroke in Patients With Atrial Cardiopathy randomized clinical trial. The parent trial was conducted from February 2018 through February 2023 across 185 sites in North America. This analysis was performed between April and August 2025.
The original trial enrolled patients with recent cryptogenic stroke and atrial cardiopathy. Participants were randomized to apixaban, 5 mg or 2.5 mg twice daily, versus aspirin 81 mg once daily. For the present analysis, investigators included 945 of 1015 randomized patients after excluding those with missing blood pressure or echocardiographic data.
The mean age of the analytic cohort was 68.0 years, 54.3% were female, and 351 patients, or 37.1%, met the definition of hypertension with high-risk features. Follow-up in the analytic cohort had a median duration of 1.6 years, with an interquartile range of 0.7 to 3.0 years.
The primary outcome was recurrent ischemic stroke or systemic embolism. Cox proportional hazards models were used to test for interaction between treatment assignment and hypertension with high-risk features, and to estimate treatment effects within subgroups defined by the presence or absence of these features.
Key Results
During follow-up, 67 patients experienced recurrent ischemic stroke or systemic embolism. The central finding was a statistically significant interaction between hypertension with high-risk features and the association of antithrombotic therapy with the primary outcome.
Patients without high-risk hypertensive features
In the 594 patients who did not meet criteria for hypertension with high-risk features, apixaban was associated with a lower risk of recurrent ischemic stroke or systemic embolism than aspirin. The hazard ratio was 0.43, with a 95% confidence interval of 0.22 to 0.85. The annualized absolute rate difference was -3.4%.
This is a clinically important effect size. While exploratory and not definitive, it suggests that in patients with atrial cardiopathy whose phenotype is less confounded by severe hypertension-related vascular disease, anticoagulation may better address the predominant mechanism of recurrent events.
Patients with high-risk hypertensive features
In the 351 patients with systolic blood pressure of at least 160 mm Hg at enrollment, left ventricular hypertrophy on echocardiography, or both, no significant benefit of apixaban was seen. The hazard ratio comparing apixaban with aspirin was 1.68 (95% CI, 0.78-3.62), and the annualized rate difference was 2.4%.
The confidence interval was wide and includes both no effect and potential harm. This pattern should not be overinterpreted as proof that apixaban is harmful in this subgroup, but it does argue against a meaningful protective effect comparable to that seen in patients without these hypertensive features.
How should clinicians read these subgroup data?
The most important statistical result is the interaction, not just the within-subgroup hazard ratios. A treatment that appears effective in one biologically coherent subgroup but not another suggests heterogeneity of treatment effect. In this case, the difference aligns with a mechanistic hypothesis: some patients in ARCADIA may have had strokes driven more by hypertensive small-vessel disease or mixed vasculopathy than by occult atrial thromboembolism.
Mechanistic and Clinical Interpretation
The analysis offers a compelling explanatory model for the neutral results of multiple anticoagulation trials in cryptogenic stroke. If patients with covert hypertensive arteriopathy are included in substantial numbers, a targeted benefit from anticoagulation in patients with atrial cardiopathy could be diluted.
Several biological pathways support this interpretation. First, uncontrolled or severe hypertension damages penetrating cerebral arterioles and promotes lacunar or subcortical ischemic injury. Second, left ventricular hypertrophy is a marker of cumulative hypertensive burden and often coexists with diffuse vascular remodeling, impaired arterial compliance, and cerebral small-vessel disease. Third, although atrial cardiopathy may increase thromboembolic propensity, its clinical consequences may be overshadowed when the dominant recurrent stroke mechanism is intrinsic cerebrovascular disease rather than embolism.
This framework also intersects with the broader challenge of stroke classification. Contemporary systems can identify major mechanisms, but mixed-mechanism strokes are common. A patient may have atrial cardiopathy, hypertension, and nonstenosing atherosclerosis simultaneously. The present study argues that a simple marker panel of hypertensive target-organ injury may help refine the selection of patients for anticoagulation-based prevention strategies.
Context Within the Existing Literature
The findings are best interpreted against the backdrop of prior embolic stroke of undetermined source and cryptogenic stroke trials. NAVIGATE ESUS and RE-SPECT ESUS both failed to show overall superiority of anticoagulation over aspirin in broad ESUS populations. ARCADIA, despite selecting patients with atrial cardiopathy, was also neutral in its primary analysis. This repeated pattern has raised concerns that current clinical enrichment strategies are still insufficiently precise.
The current exploratory analysis suggests one reason why. Atrial cardiopathy may be necessary but not sufficient as a marker of a treatable embolic substrate. The exclusion, or at least careful stratification, of patients with strong signatures of hypertensive cerebrovascular disease may be equally important. In that sense, the study does not overturn prior neutral trials; rather, it provides a biologically plausible explanation for them and suggests a more nuanced direction for future trial design.
Limitations and Generalizability
Several limitations are important. First, this was an exploratory post hoc analysis rather than a prespecified primary hypothesis test. Such analyses are useful for generating mechanistic insights but are inherently more vulnerable to chance findings.
Second, only 67 primary outcome events occurred in the analytic cohort, limiting precision. This is reflected in the wide confidence interval in the subgroup with high-risk hypertensive features. Replication is essential before changing standard practice.
Third, the definition of hypertension with high-risk features, while pragmatic, is imperfect. Systolic blood pressure at enrollment may not fully capture chronic hypertension exposure, and echocardiographic left ventricular hypertrophy is not specific to hypertensive cerebrovascular disease. More direct measures, such as MRI markers of cerebral small-vessel disease, ambulatory blood pressure burden, or composite phenotyping models, may improve discrimination.
Fourth, because patients with missing blood pressure and echocardiographic data were excluded, selection bias is possible. Fifth, the parent trial population consisted of patients with recent cryptogenic stroke and atrial cardiopathy enrolled at North American sites, so applicability to other stroke populations is uncertain.
Finally, bleeding outcomes are not the focus of the abstracted analysis. Any subgroup-specific efficacy signal for anticoagulation must always be weighed against hemorrhagic risk, particularly in patients with hypertension, who may also be more susceptible to intracranial bleeding.
Practice Implications and Future Directions
These data are not yet practice changing in the sense of supporting routine anticoagulation for all patients with cryptogenic stroke and atrial cardiopathy who lack high-risk hypertensive features. However, they are highly informative for clinical reasoning and future research.
For clinicians, the study reinforces the need for careful phenotyping after cryptogenic stroke. Blood pressure control, review of chronic hypertensive burden, assessment for left ventricular hypertrophy, and MRI evaluation for small-vessel disease should all be part of a mechanism-oriented workup. A label of cryptogenic stroke should not imply that all residual uncertainty is embolic in nature.
For trialists, the findings support more granular enrichment strategies. Future studies may need to combine positive selection for atrial thromboembolic risk, such as biomarkers of atrial cardiopathy or prolonged rhythm monitoring, with negative selection against probable hypertensive arteriopathy. Imaging biomarkers, including white matter hyperintensity burden, cerebral microbleeds, enlarged perivascular spaces, and lacune patterns, could be integrated into future designs.
For health systems and guideline developers, this work highlights a broader principle: precision secondary prevention in stroke will require mechanism-based stratification that goes beyond conventional clinical labels. The failure of one-size-fits-all anticoagulation approaches in cryptogenic stroke should prompt refinement, not abandonment, of targeted antithrombotic strategies.
Conclusion
This exploratory ARCADIA analysis suggests that hypertension with high-risk features may meaningfully modify the effect of antithrombotic therapy in patients with recent cryptogenic stroke and atrial cardiopathy. In patients without these hypertensive markers, apixaban was associated with a lower risk of recurrent ischemic stroke or systemic embolism than aspirin. In patients with high-risk hypertensive features, no such benefit was observed.
The most important implication is conceptual. Neutral anticoagulation trials in cryptogenic stroke may partly reflect biological heterogeneity, especially the inclusion of patients whose events are driven by hypertensive arteriopathy rather than occult cardioembolism. The study therefore advances the field not by establishing a new treatment standard, but by sharpening the mechanistic questions that future trials must answer.
Funding, Registration, and Citation
Trial registration: ClinicalTrials.gov Identifier: NCT03192215.
Citation: Ridha M, Hailat R, Stanton R, Merkler AE, Elkind MSV, Longstreth WT, Tirschwell DL, Kronmal RA, Hannawi Y, Kamel H, Burke J, Woo D. Hypertension With High-Risk Features in Cryptogenic Stroke: An Exploratory Analysis of the ARCADIA Randomized Clinical Trial. JAMA Neurology. 2026 Jun 1;83(6):573-581. PMID: 42008258. URL: https://pubmed.ncbi.nlm.nih.gov/42008258/
References
1. Ridha M, Hailat R, Stanton R, Merkler AE, Elkind MSV, Longstreth WT, Tirschwell DL, Kronmal RA, Hannawi Y, Kamel H, Burke J, Woo D. Hypertension With High-Risk Features in Cryptogenic Stroke: An Exploratory Analysis of the ARCADIA Randomized Clinical Trial. JAMA Neurology. 2026;83(6):573-581. PMID: 42008258.
2. Kamel H, Longstreth WT Jr, Tirschwell DL, Kronmal RA, Broderick JP, Palesch YY, Meinzer C, Dillon C, Ewing I, Spilker JA, et al. The AtRial Cardiopathy and Antithrombotic Drugs In Prevention After Cryptogenic Stroke Randomized Trial: Rationale and methods. International Journal of Stroke. 2019;14(2):207-214. PMID: 30590944.
3. Hart RG, Sharma M, Mundl H, Kasner SE, Bangdiwala SI, Berkowitz SD, Swaminathan B, Lavados P, Wang Y, Wang Y, et al. Rivaroxaban for Stroke Prevention after Embolic Stroke of Undetermined Source. New England Journal of Medicine. 2018;378(23):2191-2201. PMID: 29766772.
4. Diener HC, Sacco RL, Easton JD, Granger CB, Bernstein RA, Uchiyama S, Kreuzer J, Cronin L, Cotton D, Grauer C, et al. Dabigatran for Prevention of Stroke after Embolic Stroke of Undetermined Source. New England Journal of Medicine. 2019;380(20):1906-1917. PMID: 31091372.
5. Kleindorfer DO, Towfighi A, Chaturvedi S, Cockroft KM, Gutierrez J, Lombardi-Hill D, Kamel H, Kernan WN, Kittner SJ, Leira EC, et al. 2021 Guideline for the Prevention of Stroke in Patients With Stroke and Transient Ischemic Attack: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2021;52(7):e364-e467. PMID: 34024117.
