Skip to content

Cardioembolic and cryptogenic stroke

TL;DR — Atrial fibrillation (AF) changes secondary prevention from antiplatelet therapy to anticoagulation; direct oral anticoagulants (DOACs) generally preserve or improve efficacy while reducing intracranial hemorrhage compared with warfarin in non-valvular AF (Hart 2007, PMID 17577005; Granger 2011, PMID 21870978). Detection rises with monitoring duration: 30-day external monitoring found AF in 16.1% versus 3.2% with repeat 24-hour monitoring, and insertable monitoring reached 30% detection by three years in CRYSTAL-AF (Gladstone 2014, PMID 24963566; Brachmann 2016, PMID 26763225). But AF detection is not synonymous with causal attribution, especially after small- or large-vessel stroke. Empirical anticoagulation has repeatedly failed in embolic stroke of undetermined source (ESUS), including biomarker-defined atrial cardiopathy, so cryptogenic stroke is a diagnostic category rather than an indication for anticoagulation (Hart 2018, PMID 29766772; Diener 2019, PMID 31091372; Kamel 2024, PMID 38324415). PFO closure reduces recurrence in carefully selected patients aged approximately 16–60 years after a genuinely cryptogenic embolic stroke, with the largest rationale in high-risk anatomy, but adds procedural/device risk and atrial fibrillation (Saver 2017, PMID 28902590; Mas 2017, PMID 28902593).

Definitions that should not be collapsed

Term Operational meaning Main error if misused
Cardioembolic stroke A plausible high-risk cardiac source is established and fits the infarct Calling every cortical infarct “cardioembolic” without demonstrating a source
Cryptogenic stroke Cause remains unknown after an adequate evaluation Treating variable workups as equivalent
ESUS Non-lacunar embolic-appearing infarct without ≥50% supplying-artery stenosis or major-risk cardiac source after a specified minimum workup Assuming a single embolic mechanism or proven benefit from anticoagulation (Hart 2018, PMID 29766772)
Atrial cardiopathy Structural, electrical, or biochemical atrial abnormality proposed to confer embolic risk without documented AF Equating an association marker with an anticoagulant-responsive mechanism (Kamel 2024, PMID 38324415)
PFO-associated stroke Cryptogenic embolic stroke in a patient whose PFO is judged causally relevant after exclusion of alternatives Treating a common incidental PFO as proof of paradoxical embolism (Saver 2017, PMID 28902590)

The diagnostic workup must be commensurate with the label: brain and arterial imaging, ECG and inpatient telemetry, echocardiographic assessment when it can change management, and a monitoring strategy calibrated to pretest probability (Kleindorfer 2021, PMID 34024117).

Atrial fibrillation: why anticoagulation matters

A meta-analysis of 29 randomized trials and 28,044 participants found adjusted-dose warfarin reduced stroke by about 64% relative to control and by 39% relative to antiplatelet therapy; absolute benefit was greatest in those at highest baseline risk (Hart 2007, PMID 17577005).

BAFTA directly addressed older age: among 973 community patients aged ≥75 years (mean 81.5), warfarin approximately halved fatal/disabling stroke, intracranial hemorrhage, or systemic embolism compared with aspirin, without the large extracranial-bleeding excess that had been feared (Mant 2007, PMID 17693178).

DOAC evidence in non-valvular AF

Trial n Comparison Stroke/systemic embolism Major safety result
RE-LY 18,113 Dabigatran 110 or 150 mg twice daily vs warfarin 150 mg: 1.11%/y vs 1.69%/y; RR 0.66 (95% CI 0.53–0.82) 110 mg had less major bleeding; both doses less hemorrhagic stroke (Connolly 2009, PMID 19717844)
ROCKET-AF 14,264 Rivaroxaban 20 mg daily vs warfarin 1.7%/y vs 2.2%/y in per-protocol on-treatment analysis; noninferior Less intracranial/fatal bleeding, more gastrointestinal bleeding (Patel 2011, PMID 21830957)
ARISTOTLE 18,201 Apixaban 5 mg twice daily vs warfarin 1.27%/y vs 1.60%/y; HR 0.79 (95% CI 0.66–0.95) Major bleeding 2.13%/y vs 3.09%/y; mortality lower (Granger 2011, PMID 21870978)
ENGAGE AF–TIMI 48 21,105 Edoxaban high/low dose vs warfarin Both regimens noninferior Dose-dependent reduction in major bleeding; more GI bleeding with high dose (Giugliano 2013, PMID 24251359)
AVERROES 5,599 unsuitable for VKA Apixaban vs aspirin Stroke/systemic embolism 1.6%/y vs 3.7%/y; HR 0.45 (95% CI 0.32–0.62) Major bleeding not significantly different (Connolly 2011, PMID 21309657)

These trials were AF trials, not exclusively recent-stroke trials. Choice and dosing depend on renal function, age, weight, interacting drugs, adherence pattern, gastrointestinal bleeding, and access. Mechanical valves and moderate-to-severe rheumatic mitral stenosis remain warfarin indications rather than DOAC indications (Kleindorfer 2021, PMID 34024117).

Adding routine antiplatelet therapy to anticoagulation increases bleeding and is generally avoided unless a separate indication, such as recent coronary stenting, justifies a time-limited combination (Kleindorfer 2021, PMID 34024117).

The timing tradeoff is recurrent embolism versus hemorrhagic transformation. Infarct size, hemorrhagic conversion, reperfusion treatment, blood pressure, renal function, and an alternative urgent indication matter more than a single calendar rule.

ELAN randomized 2,013 patients to early DOAC treatment—within 48 hours for minor/moderate stroke or day 6–7 after major stroke—or later treatment—day 3–4, 6–7, or 12–14 respectively. The 30-day composite of recurrent ischemic stroke, systemic embolism, major extracranial bleeding, symptomatic intracranial hemorrhage, or vascular death was 2.9% early versus 4.1% later; symptomatic ICH was 0.2% in each group (Fischer 2023, PMID 37222476).

TIMING randomized 888 patients to DOAC initiation ≤4 days or 5–10 days. Early treatment was noninferior for the 90-day composite of recurrent ischemic stroke, symptomatic ICH, or death; no symptomatic ICH occurred in either group (Oldgren 2022, PMID 36065821).

Stroke context What randomized evidence supports Residual uncertainty
Minor/moderate infarct without significant hemorrhage Starting within the first several days is not worse than later initiation in ELAN/TIMING (PMID 37222476; PMID 36065821) Very early treatment after thrombolysis/thrombectomy in specific imaging phenotypes
Major infarct ELAN tested day 6–7 vs day 12–14 (PMID 37222476) Mass effect, substantial hemorrhagic transformation, and unstable neurosurgical cases were not resolved
ICH rather than infarction Separate recurrence/bleeding problem Timing and selection remain much less certain; see intracerebral hemorrhage

Rhythm monitoring: yield rises with time

EMBRACE randomized 572 patients aged ≥55 years with recent cryptogenic stroke/TIA after standard testing to a 30-day event-triggered recorder or repeat 24-hour monitoring. AF ≥30 seconds was detected in 16.1% versus 3.2% (absolute difference 12.9 points; NNScreen ≈8), and anticoagulant prescribing increased (Gladstone 2014, PMID 24963566).

CRYSTAL-AF randomized 441 patients to an insertable cardiac monitor or conventional follow-up. Detection was higher with the implant at six and 12 months; by 36 months, cumulative AF detection reached 30.0% in the implant arm (Sanna 2014, PMID 24963567; Brachmann 2016, PMID 26763225).

PER DIEM compared 12 months of implantable monitoring with 30 days of external loop recording in 300 patients. AF/flutter ≥2 minutes was detected in 15.3% versus 4.7% by 12 months, absolute difference 10.7 points (Buck 2021, PMID 34061146).

STROKE-AF complicates causal interpretation. Among 496 patients whose stroke was attributed to large- or small-vessel disease, insertable monitoring detected AF in 12.1% versus 1.8% with usual care at 12 months (Bernstein 2021, PMID 34061145). Detection after an apparently non-cardioembolic stroke may reflect background atrial disease, a contributor, or the actual cause; the trial measured detection, not reduction in recurrent stroke.

Monitoring strategy Duration Approximate detection result Inference limit
Repeat 24-h monitor 24 h 3.2% in EMBRACE control Low sensitivity for intermittent AF (PMID 24963566)
30-d external recorder 30 d 16.1% in EMBRACE Requires wear and transmission adherence (PMID 24963566)
Implant vs 30-d external 12 mo 15.3% vs 4.7% in PER DIEM AF threshold was ≥2 min, not a proven causal-dose threshold (PMID 34061146)
Insertable monitor 36 mo 30.0% in CRYSTAL-AF intervention arm More detection does not prove every episode caused the index stroke (PMID 26763225)

ESUS: why empirical anticoagulation failed

NAVIGATE-ESUS randomized 7,213 patients to rivaroxaban 15 mg daily or aspirin 100 mg. The trial stopped early: recurrent stroke/systemic embolism was not reduced, while major bleeding was increased (Hart 2018, PMID 29766772).

RE-SPECT ESUS randomized 5,390 patients to dabigatran or aspirin. Dabigatran did not significantly reduce recurrent stroke and did not establish empirical anticoagulation as superior (Diener 2019, PMID 31091372).

ATTICUS enriched ESUS for AF predictors or PFO and mandated monitoring. Apixaban did not reduce new ischemic MRI lesions compared with aspirin, and the trial was terminated for futility (Geisler 2024, PMID 38320511).

ARCADIA tested the stronger mechanistic hypothesis that atrial cardiopathy without AF marks an anticoagulant-responsive embolic substrate. In 1,015 participants, apixaban did not reduce recurrent stroke compared with aspirin; both groups had an annualized recurrence rate of 4.4% (Kamel 2024, PMID 38324415).

The four negative trials dismantle “embolic appearance → anticoagulation.” ESUS contains occult AF, nonstenotic plaque, aortic disease, cancer-associated thrombosis, PFO, and mechanisms not yet identified; treatment must follow the discovered cause.

Patent foramen ovale

PFO is common, so the central question is causal probability. Age, absence of conventional vascular risk factors, embolic infarct topography, large shunt, atrial septal aneurysm, venous thromboembolism, and completeness of the competing-cause workup shape attribution.

Trial Selected population Stroke result Tradeoff
RESPECT extended follow-up 980, age 18–60, cryptogenic stroke + PFO Recurrent ischemic stroke lower with closure over median 5.9 y; HR 0.55 (95% CI 0.31–0.999) Device/procedural complications; selected population (Saver 2017, PMID 28902590)
CLOSE 663, age 16–60, large shunt or atrial septal aneurysm 0 strokes after closure vs 14 with antiplatelet therapy over mean 5.3 y Procedural complications 5.9%; AF 4.6% vs 0.9% (Mas 2017, PMID 28902593)
REDUCE 664, mostly moderate/large shunts Clinical stroke 1.4% closure vs 5.4% antiplatelet; HR 0.23 (95% CI 0.09–0.62) AF/flutter 6.6% after closure (Søndergaard 2017, PMID 28902580)
DEFENSE-PFO 120 with high-risk morphology Primary endpoint occurred only in medical arm during median 2.8 y Small, open-label trial; wide uncertainty (Lee 2018, PMID 29544871)

Closure evidence does not extend automatically to older patients, TIA without imaging-confirmed infarction, lacunar stroke, incompletely excluded AF, or low-risk incidental PFO. Shared decisions should express absolute recurrence risk, procedural risk, device-related complications, and post-procedure AF rather than “open versus closed” anatomy alone (Kleindorfer 2021, PMID 34024117).

Causal attribution before closure

The trial-eligible phenotype is not “PFO plus neurological symptoms.” It is an imaging-confirmed, non-lacunar ischemic stroke after adequate arterial, cardiac, and rhythm evaluation, with no more plausible competing mechanism. Venous thrombosis or a strong transient venous-thromboembolism exposure increases biological plausibility, while hypertension, diabetes, smoking, atherosclerosis, or a small deep infarct reduce the probability that the PFO was causal (Saver 2017, PMID 28902590; Mas 2017, PMID 28902593).

Post-closure follow-up must capture recurrent neurological events, device complications, residual shunt, and new AF. Because closure trials showed more AF in device groups, failure to monitor rhythm can misclassify a treatment-associated competing mechanism as another cryptogenic event (Søndergaard 2017, PMID 28902580).

Other potential cardiac and systemic sources

Transesophageal echocardiography can characterize left-atrial appendage thrombus, valves, atrial septum, and aortic arch more sensitively than transthoracic imaging for selected questions, but routine testing is valuable only when its result changes treatment (Kleindorfer 2021, PMID 34024117).

Mechanical valves, infective endocarditis, intracardiac thrombus, recent myocardial infarction with ventricular thrombus, severe cardiomyopathy, cardiac tumors, and cancer-associated thrombosis are distinct mechanisms. They should not be pooled into ESUS, and each carries treatment and hemorrhage considerations beyond the AF trials summarized here.

Open questions

  • What duration and burden of device-detected AF after stroke is sufficient to infer causality and net benefit from lifelong anticoagulation? STROKE-AF proves detection in non-cardioembolic labels, not treatment benefit (Bernstein 2021, PMID 34061145).
  • Can temporal coupling between AF episodes and embolic events, atrial imaging, or blood biomarkers distinguish causal AF from incidental age-related AF (Sanna 2014, PMID 24963567; Kamel 2024, PMID 38324415)?
  • Which major-infarct and hemorrhagic-transformation phenotypes can safely begin a DOAC earlier than conventional rules? ELAN included severity-based timing but cannot settle every imaging subgroup (Fischer 2023, PMID 37222476).
  • Is there any biologically coherent ESUS subgroup that benefits from anticoagulation, or should the construct be retired in favor of source-specific phenotyping (Hart 2018, PMID 29766772; Diener 2019, PMID 31091372)?
  • Does PFO closure benefit carefully selected patients older than 60 years, and can AF monitoring prevent closure of incidental PFOs? Existing positive trials predominantly capped age at 60 (Saver 2017, PMID 28902590; Mas 2017, PMID 28902593).

References

  1. Kleindorfer DO, Towfighi A, Chaturvedi S, et al. 2021 Guideline for the Prevention of Stroke in Patients With Stroke and Transient Ischemic Attack. Stroke. 2021;52:e364-e467. PMID 34024117
  2. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857-867. PMID 17577005
  3. Mant J, Hobbs FDR, Fletcher K, et al. Warfarin versus aspirin for stroke prevention in an elderly community population with atrial fibrillation (BAFTA). Lancet. 2007;370:493-503. PMID 17693178
  4. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361:1139-1151. PMID 19717844
  5. Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365:883-891. PMID 21830957
  6. Granger CB, Alexander JH, McMurray JJV, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365:981-992. PMID 21870978
  7. Giugliano RP, Ruff CT, Braunwald E, et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013;369:2093-2104. PMID 24251359
  8. Connolly SJ, Eikelboom J, Joyner C, et al. Apixaban in patients with atrial fibrillation. N Engl J Med. 2011;364:806-817. PMID 21309657
  9. Fischer U, Koga M, Strbian D, et al. Early versus later anticoagulation for stroke with atrial fibrillation. N Engl J Med. 2023;388:2411-2421. PMID 37222476
  10. Oldgren J, Åsberg S, Hijazi Z, et al. Early versus delayed non-vitamin K antagonist oral anticoagulant therapy after acute ischemic stroke in atrial fibrillation (TIMING). Circulation. 2022;146:1056-1066. PMID 36065821
  11. Gladstone DJ, Spring M, Dorian P, et al. Atrial fibrillation in patients with cryptogenic stroke. N Engl J Med. 2014;370:2467-2477. PMID 24963566
  12. Sanna T, Diener HC, Passman RS, et al. Cryptogenic stroke and underlying atrial fibrillation. N Engl J Med. 2014;370:2478-2486. PMID 24963567
  13. Brachmann J, Morillo CA, Sanna T, et al. Uncovering atrial fibrillation beyond short-term monitoring in cryptogenic stroke patients: three-year results from CRYSTAL AF. Circ Arrhythm Electrophysiol. 2016;9:e003333. PMID 26763225
  14. Bernstein RA, Kamel H, Granger CB, et al. Long-term continuous cardiac monitoring vs usual care after stroke attributed to large- or small-vessel disease: STROKE-AF. JAMA. 2021;325:2169-2177. PMID 34061145
  15. Buck BH, Hill MD, Quinn FR, et al. Implantable vs prolonged external electrocardiographic monitoring after ischemic stroke: PER DIEM. JAMA. 2021;325:2160-2168. PMID 34061146
  16. Hart RG, Sharma M, Mundl H, et al. Rivaroxaban for stroke prevention after embolic stroke of undetermined source. N Engl J Med. 2018;378:2191-2201. PMID 29766772
  17. Diener HC, Sacco RL, Easton JD, et al. Dabigatran for prevention of stroke after embolic stroke of undetermined source. N Engl J Med. 2019;380:1906-1917. PMID 31091372
  18. Geisler T, Keller T, Martus P, et al. Apixaban versus aspirin for embolic stroke of undetermined source. NEJM Evid. 2024;3:EVIDoa2300235. PMID 38320511
  19. Kamel H, Longstreth WT Jr, Tirschwell DL, et al. Apixaban to prevent recurrence after cryptogenic stroke in patients with atrial cardiopathy: ARCADIA. JAMA. 2024;331:573-581. PMID 38324415
  20. Saver JL, Carroll JD, Thaler DE, et al. Long-term outcomes of patent foramen ovale closure or medical therapy after stroke. N Engl J Med. 2017;377:1022-1032. PMID 28902590
  21. Mas JL, Derumeaux G, Guillon B, et al. Patent foramen ovale closure or anticoagulation vs antiplatelets after stroke. N Engl J Med. 2017;377:1011-1021. PMID 28902593
  22. Søndergaard L, Kasner SE, Rhodes JF, et al. Patent foramen ovale closure or antiplatelet therapy for cryptogenic stroke. N Engl J Med. 2017;377:1033-1042. PMID 28902580
  23. Lee PH, Song JK, Kim JS, et al. Cryptogenic stroke and high-risk patent foramen ovale: DEFENSE-PFO. J Am Coll Cardiol. 2018;71:2335-2342. PMID 29544871