TOAST Classification of Ischemic Stroke Subtypes
The TOAST classification of ischemic stroke subtypes - large-artery, cardioembolic, small-vessel, other, and undetermined - and how CCS and ASCOD refine it.
TOAST sorts ischemic stroke into five mechanistic buckets so that secondary prevention is matched to the actual cause, not to the syndrome.
- → Five subtypes: large-artery atherosclerosis, cardioembolism, small-vessel (lacunar) occlusion, other determined etiology, and undetermined.
- → The subtype is the lever for prevention: antiplatelet plus statin and risk-factor control for atherosclerosis and lacunes, anticoagulation for most cardioembolic sources, cause-directed therapy for the “other” group.
- → “Undetermined” is large and heterogeneous — cryptogenic, competing mechanisms, or incomplete workup — and CCS and ASCOD were built to shrink and sharpen it.
The Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification was published in 1993 to give a stroke trial a reproducible way to assign mechanism, and it has outlived the trial to become the default etiologic language of stroke medicine.1 Its appeal at the bedside is that it is mechanistic rather than syndromic: it answers “why did this artery occlude?” rather than “what does the deficit look like?” Because the answer dictates whether you reach for an antiplatelet, an anticoagulant, a statin, or a cause-specific therapy, getting the subtype right is the single most important diagnostic step in secondary prevention.
The five subtypes.
TOAST defines five categories, each anchored to clinical features plus the results of vascular imaging, cardiac evaluation, and a hypercoagulable or other workup.1
- Large-artery atherosclerosis (LAA): ≥50% stenosis or occlusion of a clinically relevant extracranial or intracranial artery, with a cortical, cerebellar, brainstem, or large (>1.5 cm) subcortical infarct that fits that vessel. Cardioembolic sources must be reasonably excluded.
- Cardioembolism (CE): a high-risk cardiac source — atrial fibrillation, mechanical prosthetic valve, left ventricular thrombus, recent (within ~4 weeks) anterior myocardial infarction, or infective endocarditis, among others. The infarct pattern is typically cortical or involves multiple vascular territories.
- Small-vessel occlusion (lacunar): a classic lacunar syndrome with a relevant subcortical or brainstem lesion generally <1.5 cm, no cortical signs, and no large-artery or cardioembolic source to explain it.
- Stroke of other determined etiology: a defined, non-atherosclerotic, non-cardioembolic cause — arterial dissection, vasculitis, an inherited or acquired hypercoagulable state, or a drug-induced mechanism, confirmed by appropriate testing.
- Stroke of undetermined etiology: three distinct situations collapse here — a complete workup that is negative (cryptogenic), two or more competing mechanisms (e.g., atrial fibrillation and an ipsilateral 70% carotid stenosis), or a workup too incomplete to decide.
Bedside pearl
Before you write “lacunar,” confirm the lesion is small (generally <1.5 cm), subcortical or brainstem, and in a territory that explains the syndrome — and that vessel imaging and a rhythm/cardiac source are clean. A small deep infarct with ipsilateral carotid stenosis or AFib is competing-mechanism “undetermined,” not a clean lacune.
Why the subtype changes management.
TOAST matters because each bucket points to a different prevention strategy, and current AHA/ASA secondary-prevention guidance is organized largely around these mechanisms.4 Large-artery atherosclerosis and lacunar disease are fundamentally antiplatelet-and-risk-factor diseases: high-intensity statin, blood-pressure and glycemic control, and antiplatelet therapy, with short-course dual antiplatelet therapy reserved for selected high-risk minor stroke or TIA and for symptomatic intracranial stenosis per guideline criteria. Cardioembolic stroke from atrial fibrillation or another high-risk source is generally an anticoagulation disease. The “other determined” group is the most cause-specific of all — dissection, vasculitis, and hypercoagulable states each carry their own treatment pathway. Calling a stroke “cryptogenic” is therefore not a diagnosis but a prompt to keep looking, including prolonged rhythm monitoring for occult AFib.
This page is educational and does not replace individualized clinical judgment; specific thresholds, monitoring strategies, and antithrombotic choices vary by guideline version and by institution, so confirm against your local protocol and the current AHA/ASA statement.4
Where TOAST struggles.
TOAST’s biggest weakness is that “undetermined” absorbs too much. In the original work, interobserver agreement was high once the diagnostic workup was available, but a large share of patients still landed in the undetermined group, and that fraction balloons when two plausible mechanisms coexist.1 The system also forces a single winner: a patient with both carotid disease and AFib is shoehorned into one box or dropped into “undetermined,” even though both diseases are real and both deserve treatment.
Pitfall
Do not let an incomplete workup masquerade as a settled subtype. A stroke called “lacunar” or “cryptogenic” before vessel imaging, an echocardiogram, and adequate rhythm monitoring are done is provisional. Premature closure here translates directly into the wrong antithrombotic — for example, missing paroxysmal AFib and leaving a patient on aspirin who needed anticoagulation.
CCS and ASCOD: the refinements.
Two systems were developed to address TOAST’s limitations. The Causative Classification of Stroke (CCS) is an evidence-based, computerized algorithm (descended from the SSS-TOAST) that weights competing mechanisms and assigns the single most probable cause, while separately recording the level of diagnostic confidence.2 By design it markedly reduces the “undetermined-unclassified” fraction and shows excellent inter- and intra-rater reliability, which is why it is favored in multicenter research where consistent subtyping matters.2
ASCOD takes a different, phenotypic approach. Instead of forcing one cause, it grades every potential mechanism in each patient — A (atherosclerosis), S (small-vessel disease), C (cardiac), O (other), and D (dissection) — on a likelihood scale (1 potentially causal, 2 uncertain, 3 present but unlikely causal, 0 absent, 9 workup insufficient).3 ASCOD deliberately captures overlap, and it pegged the threshold for significant carotid or intracranial stenosis at 50%, aligning with common practice.3 The practical takeaway: TOAST remains the lingua franca for everyday communication and most trials, CCS is the reproducible algorithmic refinement, and ASCOD is the phenotyping tool when a patient has more than one disease worth treating.
Frequently asked questions.
What are the five TOAST subtypes?
Large-artery atherosclerosis, cardioembolism, small-vessel (lacunar) occlusion, stroke of other determined etiology, and stroke of undetermined etiology. The classification is based on clinical features plus vascular imaging, cardiac evaluation, and other targeted testing.
What stenosis cutoff defines large-artery atherosclerosis?
Under TOAST, ≥50% stenosis or occlusion of a clinically relevant extracranial or intracranial artery supplying the infarct, with cardioembolic sources reasonably excluded. ASCOD likewise uses a 50% luminal-stenosis threshold for a significant lesion.
How small does a lesion have to be to count as lacunar?
A small-vessel (lacunar) infarct is generally less than 1.5 cm, located in subcortical white matter or the brainstem, in a territory that explains a classic lacunar syndrome, with no cortical signs and no large-artery or cardioembolic source.
What goes into the “undetermined” category?
Three different situations: a complete but negative workup (cryptogenic), two or more competing mechanisms with no single best explanation, or a workup that is incomplete. The label is a signal to keep investigating, not a final mechanism.
How do CCS and ASCOD differ from TOAST?
CCS is a computerized, evidence-weighted algorithm that picks the single most probable cause and records diagnostic confidence, shrinking the undetermined group with high reliability. ASCOD instead grades every potential mechanism (atherosclerosis, small-vessel, cardiac, other, dissection) by likelihood, so it captures overlapping diseases rather than forcing one winner.
Why does the TOAST subtype matter for treatment?
Because prevention follows mechanism. Atherosclerotic and lacunar strokes are treated mainly with antiplatelet therapy, statin, and risk-factor control; most high-risk cardioembolic sources call for anticoagulation; and “other determined” causes such as dissection or vasculitis have their own cause-specific pathways. Specifics vary by guideline and institution.
References.
- Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke. 1993;24:35–41. PubMed
- Ay H, Benner T, Arsava EM, et al. A computerized algorithm for etiologic classification of ischemic stroke: the Causative Classification of Stroke System. Stroke. 2007;38:2979–2984. PubMed
- Amarenco P, Bogousslavsky J, Caplan LR, Donnan GA, Wolf ME, Hennerici MG. The ASCOD phenotyping of ischemic stroke (Updated ASCO Phenotyping). Cerebrovasc Dis. 2013;36:1–5. PubMed
- Kleindorfer DO, Towfighi A, Chaturvedi S, 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:e364–e467. PubMed
More clinical tools
Keep the stroke service moving.
- NIHSS Pocket Guide A bedside guide to the NIH Stroke Scale - all 15 items, the scoring conventions that trip people up, severity bands, serial use, and the posterior-circulation blind spot.
- Modified Rankin Scale (mRS) The modified Rankin Scale explained: all seven grades 0-6, the independence and walking boundaries, baseline-to-90-day anchoring, and the mRS 0-2 trial cut-point.
- ASPECTS ASPECTS explained: the 10-region MCA score on non-contrast CT, the regions most often missed, historical and large-core thrombectomy thresholds, and automated scoring.
- IV Thrombolysis Eligibility IV thrombolysis for acute ischemic stroke: the 4.5-hour and extended windows, tenecteplase and alteplase dosing, key exclusions, and where guidelines diverge.
- All clinical tools The full bedside reference index.
Related reading
From the articles.
- Types of stroke explained The plain-language version of stroke subtypes.
- What is a stroke? Stroke fundamentals behind the classification.
Reviewed by