What Is a Cryptogenic Stroke and What Causes It?

A cryptogenic stroke is an ischemic stroke whose cause remains unknown after a standard diagnostic workup. Roughly a third of all ischemic strokes fall into this frustrating category, making it one of the most common diagnoses stroke neurologists deliver. The label does not mean nothing caused the stroke; it means the usual suspects did not show up on testing. In many cases, further investigation eventually reveals a hidden heart rhythm problem, a small hole in the heart, an unstable arterial plaque, or even an undiagnosed cancer. Understanding why so many strokes escape easy explanation matters because the treatment you receive depends entirely on identifying the underlying mechanism.

How Common Cryptogenic Strokes Are

About 30 to 40 percent of ischemic stroke patients end up with a cryptogenic diagnosis, depending on how aggressively the medical team investigates and which classification system they use.¹ Some centers report the proportion closer to one in three.² That variation hints at a core issue: “cryptogenic” is partly a reflection of how hard you look. A hospital with limited access to prolonged heart monitoring or advanced vascular imaging will classify more strokes as cryptogenic than a specialized stroke center with every tool available. The diagnosis is as much about the completeness of testing as about the biology of the stroke itself.

What “Cryptogenic” Actually Means in Practice

Stroke classifications typically follow a framework called TOAST, which sorts ischemic strokes into categories based on cause: large-artery atherosclerosis, cardioembolism (a clot from the heart), small-vessel disease, other determined causes, and undetermined. Cryptogenic stroke is a broad bucket within that last category. It sweeps in cases where testing was incomplete, where multiple possible causes compete without a clear winner, and where everything came back normal despite a thorough workup.³

In 2014, researchers proposed a narrower category called ESUS, or embolic stroke of undetermined source, to carve out the subset of cryptogenic strokes that look embolic on brain imaging but lack an identifiable embolism source. To qualify as ESUS, a patient must have a non-lacunar infarct on imaging, no significant artery narrowing feeding the affected brain region, no major heart-based source of clots like atrial fibrillation, and must have completed a minimum set of tests including brain and vascular imaging, at least 24 hours of heart rhythm monitoring, an echocardiogram, and basic blood work.³ The ESUS label was designed to create a more uniform group for clinical trials, since “cryptogenic” on its own mixes together patients whose strokes probably have very different underlying mechanisms.

Hidden Atrial Fibrillation

The single most common hidden culprit behind a cryptogenic stroke is atrial fibrillation that was not running at the time of initial testing. AF causes the upper chambers of the heart to quiver instead of contracting effectively, which allows blood to pool and form clots. Those clots can travel to the brain and cause a stroke. The problem is that AF can be paroxysmal, meaning it comes and goes unpredictably. A standard 24-hour heart monitor may catch it, or it may miss a rhythm that only fires for a few hours every couple of weeks.

When patients with cryptogenic stroke undergo prolonged monitoring with implantable cardiac monitors, the detection rate climbs dramatically. One study using implantable monitors found that nearly half of cryptogenic stroke patients were eventually diagnosed with device-detected AF over a median follow-up of about 15 months.⁴ That figure is striking: it suggests that a large share of cryptogenic strokes are not truly mysterious but simply under-monitored.

Atrial Cardiopathy Without Atrial Fibrillation

An emerging concept complicates the AF story. Researchers now recognize that the atrium itself can become structurally and functionally abnormal in ways that promote clot formation even before AF ever develops. This condition, called atrial cardiopathy, describes a diseased atrial substrate that raises stroke risk independently of any rhythm disturbance.⁵ In other words, the atrium can be sick enough to throw clots even while the ECG looks normal.

One large study found that each 10-millimeter increase in left atrial diameter roughly doubled the five-year risk of ischemic stroke, and this held true even after accounting for AF as a separate factor.⁶ This finding has shifted thinking about what causes some cryptogenic strokes. It may not always be the rhythm that matters but the underlying heart tissue damage that eventually leads to the rhythm problem. The atrium’s structure can be the source of clots before AF ever shows up on a monitor.

Patent Foramen Ovale

Nearly half of cryptogenic stroke patients have a patent foramen ovale, a small opening between the right and left upper chambers of the heart that normally closes shortly after birth but persists in roughly a quarter of the general population.⁷ Because PFOs are so common in healthy people, the challenge has always been figuring out when a PFO actually caused the stroke versus when it was an innocent bystander.

The presumed stroke mechanism involves paradoxical embolism: a blood clot forms in the veins, typically the legs, travels to the right side of the heart, and slips through the PFO to the left side, where it can reach the brain. This can happen when pressure on the right side of the heart briefly exceeds the left, such as during straining, coughing, or a Valsalva maneuver.⁸ Less commonly, a clot can form directly within the PFO itself, or the PFO’s anatomy can trigger atrial arrhythmias that lead to clot formation.⁹

Several factors make a PFO more likely to be the actual cause rather than coincidental: younger age, a larger PFO, a greater degree of right-to-left blood shunting, and the co-presence of an atrial septal aneurysm (a floppy, bulging wall between the atria).⁹ Stroke teams weigh these features when deciding whether a PFO is likely to blame and whether closure would help.

Non-Stenotic Arterial Plaques

Traditional stroke workups focus on arteries with significant narrowing, typically 50 percent or more blockage. But a growing body of evidence points to arterial plaques that do not narrow the vessel much yet are unstable enough to rupture and send debris into the brain. These non-stenotic plaques are increasingly recognized as a real cause of cryptogenic stroke rather than an innocent finding on imaging.

Studies have found that plaques with high-risk features, such as internal bleeding or a lipid-rich core, are about five times more common in the carotid artery on the same side as the stroke compared with the opposite side.¹⁰ One MRI-based study found that complicated carotid plaques were present in about 31 percent of cryptogenic stroke patients on the side of the infarct, compared with only 12 percent on the opposite side.¹¹ Researchers now estimate that more than 20 to 30 percent of patients with unilateral hemispheric embolic strokes of unknown source may have a stroke attributable to non-stenotic carotid plaque.¹²

The aortic arch is another vascular source. Large plaques in the aorta can shed emboli, though the evidence for aortic plaques as an independent stroke cause is more mixed. One study found that large arch plaques were associated with about double the risk of combined cardiovascular events but did not independently predict stroke alone in the statin era.¹³ Microembolic signals, tiny clot fragments detected by transcranial Doppler, have been recorded in patients with thick aortic plaques but not in those with thinner ones, supporting the idea that these plaques can send material to the brain even if they are not the most common culprit.¹⁴

Cancer and Cryptogenic Stroke

Up to about 20 percent of patients presenting with a cryptogenic ischemic stroke have an occult, previously undiagnosed malignancy at the time of their stroke.¹⁵ The connection between cancer and stroke runs through hypercoagulability: tumors can activate clotting pathways, producing a state where the blood forms clots far too easily. This acquired clotting tendency is the most frequent stroke mechanism in cancer patients.¹⁶

Two clinical red flags help identify cancer-related cryptogenic strokes: elevated D-dimer levels (a marker of active clot formation and breakdown) and infarcts scattered across multiple brain territories rather than confined to one arterial region.¹⁵ When those two features appear together in a cryptogenic stroke patient, screening for an underlying cancer becomes an important part of the workup. For some patients, the stroke is the first sign that a tumor exists.

Infections and Arterial Dissection

Recent infections appear to be a meaningful trigger for cryptogenic stroke, particularly in younger adults. A case-control study found that infections in the preceding week were associated with roughly 2.6 times higher odds of early-onset cryptogenic ischemic stroke. The study also found that von Willebrand factor, a blood protein involved in clotting, was significantly elevated in stroke patients who had recent infections, suggesting that infection-triggered changes in the coagulation system may be a bridge between a mundane illness and a stroke.¹⁷

Arterial dissection, where the inner wall of a neck artery tears and blood collects in the vessel wall, is another under-recognized cause that can initially present as a cryptogenic stroke. Dissections can be missed in the emergency department; one large study found a misdiagnosis rate of about 3 percent among dissection patients who had recently been seen in an ED.¹⁸ Because dissections are treatable and carry a risk of recurrent stroke if missed, they represent one of the higher-stakes diagnostic failures in the cryptogenic stroke population.

Obstructive Sleep Apnea as a Contributing Factor

Obstructive sleep apnea shows up repeatedly in the cryptogenic stroke literature as a factor that may push patients toward cardioembolic strokes. One study found that cardioembolic strokes were about twice as common among sleep apnea patients compared with controls, and the association remained significant even after adjusting for atrial fibrillation.¹⁹ The mechanisms may include both a higher rate of undetected paroxysmal AF triggered by overnight oxygen drops and direct effects of sleep apnea on cardiac structure and blood clotting. For cryptogenic stroke patients, an untreated sleep apnea diagnosis raises clinical suspicion for a cardiac source and can prompt more aggressive heart monitoring.

How Doctors Hunt for the Cause

The standard diagnostic workup for a cryptogenic stroke includes brain imaging, imaging of the arteries in the head and neck, an echocardiogram, at least 24 hours of continuous heart rhythm monitoring, and blood tests. When that initial evaluation comes back empty, the investigation widens.

Prolonged heart monitoring is the single highest-yield next step. Implantable loop recorders can monitor for years, catching intermittent AF that shorter monitors miss. Blood biomarkers are also gaining traction as triage tools. NT-proBNP, a hormone released when the heart is under stress, has shown promise in predicting which cryptogenic stroke patients will eventually turn out to have AF. A meta-analysis found that higher NT-proBNP levels were associated with roughly a threefold increase in the risk of subsequent AF detection, and low levels had a negative predictive value above 90 percent, meaning they are reasonably good at ruling AF out.²⁰ Another pooled analysis found that adding BNP or NT-proBNP measurements to clinical models significantly improved the ability to identify cardioembolic strokes within the first 72 hours of symptom onset.²¹

Vessel wall MRI is an emerging technique that lets radiologists look not just at whether an artery is narrowed but at the composition of its wall, revealing inflammation, bleeding within plaques, or subtle dissections that standard angiography might miss.²² This kind of imaging has the potential to reclassify some cryptogenic strokes by identifying non-stenotic vulnerable plaques or other vascular pathology that conventional imaging overlooks.

Artificial Intelligence in Stroke Diagnosis

One of the more promising developments is the use of AI to extract hidden information from routine ECGs. Even when a standard ECG shows normal sinus rhythm, subtle patterns in the waveform can reflect underlying atrial disease that predisposes to AF. A transformer-based AI model trained on over 700,000 ECGs achieved an area under the curve of about 0.81 for predicting which cryptogenic stroke patients would go on to develop AF lasting an hour or more during prolonged monitoring. When clinical variables were added, that number rose to 0.88.²³ A separate study confirmed that AI-enhanced ECG interpretation can help stratify which stroke patients would benefit most from prolonged cardiac monitoring, potentially sparing low-risk patients from months of wearing an implantable device.²⁴

Treatment When a PFO Is Found

For patients whose cryptogenic stroke is attributed to a PFO, percutaneous closure (threading a small device through a vein to seal the hole) has become the treatment of choice over antiplatelet medication alone. A systematic review of randomized trials found that PFO closure reduced the risk of recurrent stroke by about two-thirds compared with antithrombotic therapy.²⁵ The landmark CLOSE trial was particularly dramatic: no strokes occurred among the 238 patients who received PFO closure, compared with 14 strokes among the 235 patients assigned to antiplatelet therapy alone.²⁶ Other trials have confirmed the pattern, with consistently lower rates of both stroke and transient ischemic attack after closure.²⁷

The benefit is strongest in younger patients with high-risk PFO features. In older adults, where competing stroke causes like hypertension and atherosclerosis are more prevalent, the calculus gets murkier, and the PFO may be incidental rather than causal.

When No PFO Is Found and AF Remains Undetected

For the broader ESUS population, where neither a PFO nor AF has been identified, treatment decisions are harder. Several large trials tested whether blood thinners were better than aspirin for preventing recurrent strokes in ESUS patients, and the results were disappointing. A comprehensive meta-analysis found no statistically significant benefit of direct oral anticoagulants over aspirin for reducing stroke recurrence in this group.²⁸ A dedicated trial of apixaban versus aspirin in ESUS patients was stopped early for futility after showing no meaningful difference in new brain lesions at one year.²⁹

The failure of blanket anticoagulation for ESUS patients actually reinforces why accurate cause identification matters so much. ESUS lumps together patients with hidden AF (who would benefit from blood thinners), patients with unstable plaques (who might benefit from statins and possibly surgery), and patients with PFOs (who might benefit from closure). Treating them all the same way dilutes the benefit for any one subgroup. The current push in stroke medicine is to break ESUS apart into more specific categories so treatment can be tailored.

Prognosis and Recurrence

Cryptogenic stroke does not inherently carry a worse short-term prognosis than other ischemic stroke subtypes. A population-based study found that death or dependency at six months was comparable between cryptogenic stroke and other non-cardioembolic strokes, at roughly 23 percent versus 27 percent.³⁰ The ten-year recurrence risk in that same study was also similar across subtypes, at about 32 percent for cryptogenic versus 27 percent for other non-cardioembolic strokes.³⁰

In younger adults, the picture looks somewhat more favorable. A study of ischemic stroke in young adults found that the long-term cumulative recurrence rate was lowest in the cryptogenic group at about 6 percent, compared with roughly 23 percent for atherothrombotic stroke.³¹ This likely reflects the fact that cryptogenic strokes in young adults often involve one-time or reversible causes, like a PFO-mediated paradoxical embolism or a transient prothrombotic state, rather than progressive vascular disease.

Cognitive Outcomes After Stroke

Beyond recurrence, one concern patients and families have is whether a cryptogenic stroke affects long-term thinking and memory. Research on post-stroke cognitive decline shows that cognitive trajectories after ischemic stroke are broadly similar regardless of subtype. A study tracking stroke survivors over time found that small-vessel stroke survivors experienced measurable annual declines in global cognition, executive function, and memory, but these declines did not significantly differ from those in survivors of other ischemic stroke subtypes.³² In practical terms, the label “cryptogenic” does not appear to carry a unique cognitive penalty. What matters more for long-term brain health is stroke severity, location, and whether risk factors are managed to prevent a second event.