The 38-year-old with ipsilateral neck pain and a droopy eyelid is not a migraine until the cervical vessels have been seen. Cervical artery dissection is the diagnosis that changes three things at the bedside: the imaging order (neck vessels, not brain alone), the young-stroke workup, and — if they present as a code stroke — nothing about the lytic decision. The 2024 AHA scientific statement is explicit that otherwise eligible patients with dissection-related ischemic stroke are reasonable candidates for thrombolysis and thrombectomy.1 What it does not settle is anticoagulation versus antiplatelet therapy. CADISS and TREAT-CAD both ran that race. Neither crowned a winner.

Who dissects

Cervical artery dissection accounts for about 2% of all ischemic strokes and up to a quarter of ischemic strokes in adults under 50.1 Population incidence sits around 2.6 to 3.0 per 100,000 and is almost certainly an underestimate, because isolated pain is easy to miss.12 Mean age is in the mid-40s. Men are slightly over-represented; women present younger.1 In Olmsted County, internal carotid dissection was detected about twice as often as vertebral dissection across the full study window, with head or neck pain in 80%, cerebral ischemia in 56%, Horner syndrome in 25%, and a good outcome (modified Rankin 0–2) in 92%.2

"Spontaneous" and "traumatic" are a spectrum, not a binary. High-energy trauma belongs on a blunt cerebrovascular-injury pathway and is not the population CADISS, TREAT-CAD, or the 2024 AHA statement were written to cover.1 Most patients labeled spontaneous have a minor mechanical trigger — coughing, vomiting, sports, yoga, a roller coaster, cervical manipulation — acting on a wall that was already vulnerable.3 In CADISP, hypertension was more common than in healthy referents, while obesity and hypercholesterolemia were less common than in other young-stroke etiologies.4 Connective-tissue disease (vascular Ehlers–Danlos, Marfan) and fibromuscular dysplasia raise the pretest probability of recurrence and of multivessel disease; they do not have to be present for the diagnosis. Recent infection tracks more with carotid dissection; minor trauma tracks more with vertebral dissection.1 The 2014 AHA statement on cervical manipulative therapy found a statistical association that is stronger for vertebral arteries than carotid arteries and is not proof of causation — pain from an already-dissecting artery may be what sent the patient to the chiropractor.5

This page is the etiology chapter. The rest of a young-stroke evaluation — hypercoagulable testing, PFO, vasculitis, drug use — lives on the stroke workup checklist, and dissection itself is TOAST "other determined." A companion article on stroke in young adults is planned for this cluster; it is not live yet.

ICA versus vertebral: the local syndrome tells you where to look

Internal carotid dissection lives in the extracranial ICA, typically starting a couple of centimeters above the bulb — distal to where atherosclerotic plaque usually sits. The local syndrome is ipsilateral head or neck pain, a partial Horner (ptosis and miosis; anhidrosis is often absent because sudomotor fibers travel with the external carotid), pulsatile tinnitus, and sometimes lower cranial-nerve palsies.3 The ischemic syndrome is ipsilateral hemisphere or retinal ischemia, which is why a spell that "went away" in this setting is still a TIA, not a migraine that behaved.

Vertebral dissection prefers V2 and V3 as the artery climbs and turns at C1–C2. The local syndrome is occipital or posterior neck pain. The ischemic syndrome is brainstem, cerebellar, or other posterior-circulation territory — dizziness, ataxia, a Horner from lateral medullary infarction rather than from oculosympathetic fibers around the ICA — the misses that dizziness-as-stroke was written to catch.

Internal carotid Vertebral
Typical segmentExtracranial ICA, ~2 cm above the bulb — not the plaque zoneV2–V3, especially the C1–C2 turn
Local signsIpsilateral head/neck pain, partial Horner, pulsatile tinnitus, lower CN palsiesOccipital or posterior neck pain
Ischemic territoryHemisphere or retinaBrainstem, cerebellum, posterior circulation
Trigger patternInfection more often reportedMinor trauma / manipulation more often reported
Imaging trapHead CTA that never covers the neck; hematoma mistaken for slow flowLower MRA sensitivity than for ICA if fat-sat T1 is omitted; intracranial extension → SAH risk

Imaging pearls: the intramural hematoma is the diagnosis

The 2024 AHA statement considers CTA or MRA a reasonable first test.1 CTA is the emergency default: fast, widely available, and particularly useful for vertebral arteries. MRA with axial fat-suppressed T1-weighted images is the study that shows the intramural hematoma as a T1-bright crescent around a narrowed lumen — the finding that proves dissection when the lumen looks only "a bit irregular."1 Fat-sat T1 can be falsely negative in the hyperacute window, when blood products have not yet become bright, which is why a negative early MRI does not close the case if the story is still a dissection.1 If CTA is negative and clinical concern remains, fat-sat T1 MRA is the next move. DSA is not first-line; it is reserved for ongoing clinical concern after both noninvasive studies.1 Ultrasound is operator-dependent and blind to high cervical segments; it can help follow remodeling once the diagnosis is known.

Classic lumen signs — string sign, flame-shaped occlusion, intimal flap, double lumen, dissecting pseudoaneurysm — are helpful when present and absent in a large fraction of real cases. CADISS's central review failed to confirm dissection in 52 of 250 enrolled patients, a reminder that radiographic criteria are easy to over-call and easy to miss.6 A "normal CTA of the head" does not image the neck. That is the miss.

Pain plus ischemia, or pain plus a partial Horner, is a neck-vessel study — not a migraine cocktail and a discharge.

Anticoagulation versus antiplatelet: CADISS, TREAT-CAD, and the honest verdict

The mechanism of most extracranial dissection strokes is artery-to-artery embolism from thrombus at the injured intima, which is the rationale for an antithrombotic at all.8 Which one is the question this page exists to answer.

CADISS TREAT-CAD
DesignOpen-label RCT, 39 UK + 7 Australian sitesOpen-label non-inferiority RCT, 10 sites in CH/DE/DK
N / arteries250 (118 carotid, 132 vertebral)194 enrolled; 173 per-protocol; MRI-verified
WindowSymptoms within 7 days; mean randomization 3.65 daysSymptoms within 14 days
ArmsAntiplatelet (heterogeneous, including DAPT) vs heparin→warfarin, 3 monthsAspirin 300 mg daily vs VKA (INR 2–3), 90 days
Primary endpointIpsilateral stroke or death (clinical)Composite of stroke, major hemorrhage, death, and new MRI lesions
ResultStroke 4/250 (2%) at 3 months; 3/126 AP vs 1/124 AC for stroke or death. No difference. 1 SAH in the AC arm.Composite 21/91 (23%) aspirin vs 12/82 (15%) VKA; difference 8% (95% CI −4 to 21). Aspirin not shown non-inferior.
Later follow-up1-year stroke 6/250 (2.4%); recanalization no different by armMonths 3–6: rare events, exclusively hemorrhagic
VerdictUnderpowered for small differences; event rate far below older observational seriesDoes not make anticoagulation standard; does not make aspirin unsafe. Composite inflates the numerator.

CADISS is the trial that lowered everyone's estimate of early recurrent stroke.67 TREAT-CAD is the trial that keeps anticoagulation on the table, with the important caveat that many "events" were MRI lesions, not clinical strokes.9 An individual-patient meta-analysis of the two trials (n=444) found fewer composite events with anticoagulation (3 of 218 versus 10 of 226) that did not reach statistical significance, fewer ischemic strokes (1 versus 10), and more major bleeds (2 versus 0).10

STOP-CAD, a 3,636-patient observational cohort without major trauma, found no significant overall difference in ischemic stroke at 30 or 180 days; 87% of recurrent ischemic strokes had already happened by day 30. Anticoagulation was associated with fewer ischemic strokes in the occlusive-dissection subgroup (adjusted HR 0.40) and with more major hemorrhage by day 180 (adjusted HR 5.56).11 If anticoagulation is chosen, that 180-day bleed signal is the argument for switching before six months.

Guidelines have not picked a winner, and they say so. The 2021 AHA/ASA secondary-prevention guideline gives a Class 1 (C-EO) recommendation that antithrombotic therapy for at least 3 months is indicated after extracranial carotid or vertebral dissection with stroke or TIA, and a Class 2a (B-R) recommendation that either aspirin or warfarin is reasonable in that window.8 The 2024 AHA scientific statement suggests individualizing the choice and continuing for at least 3 to 6 months.1 ESO states that clinicians can prescribe either option.12 DOACs were not the randomized intervention in CADISS or TREAT-CAD; they are used in practice, not proven here. The general antiplatelet-versus-anticoagulant fork after stroke — mechanism first — is the job of the antithrombotic selection page; dissection is the exception in which both answers remain reasonable.

How long, healing, and recurrence

Three months is the trial window. Three to 6 months is the AHA 2024 suggestion.18 Beyond 3 months, TREAT-CAD's extension found rare events that were exclusively hemorrhagic.13 A common approach is to treat through 3 months, restudy the vessel, and then de-escalate: if the lumen has healed, many clinicians stop or continue a single antiplatelet depending on residual irregularity and other indications; if a high-grade residual stenosis or a thrombus-bearing lumen remains, a longer course is individualized. Recanalization often occurs over 3 to 6 months and was not improved by anticoagulation in CADISS.7 Extracranial dissecting pseudoaneurysms are common and usually follow a benign course.1 Recurrent dissection is uncommon: a 2024 meta-analysis estimated 4% overall, 2% at 1 month, and 7% at 1 year in studies with longer follow-up, with ischemic events at recurrence around 2%.14 Family history, connective-tissue disease, and FMD identify the higher-risk tail. Put the restudy date and the antithrombotic stop or switch date in the same place as the rest of the discharge checklist.

Do not withhold reperfusion

Cervical artery dissection is not a contraindication to intravenous thrombolysis. The 2024 AHA statement considers thrombolysis and mechanical thrombectomy reasonable in otherwise eligible patients.1 ESO recommends alteplase within 4.5 hours if standard criteria are met, and thrombectomy for anterior-circulation large-vessel occlusion.12 A STOP-CAD secondary analysis of 1,653 patients with spontaneous dissection and acute ischemic symptoms found intravenous thrombolysis associated with better 90-day functional independence (aOR 1.67) without a significant increase in symptomatic intracranial hemorrhage.15 Tandem occlusion — extracranial dissection plus intracranial large-vessel occlusion — is the anatomy that should trigger a thrombectomy conversation, not a pause to "confirm the dissection first." The lytic checklist itself is unchanged: see IV thrombolysis and the 2026 AHA/ASA acute ischemic stroke guideline for eligibility, not for a dissection-specific carve-out.

When to refer

Vascular neurology belongs on the case when dissection is the cause of a young stroke, when there is tandem occlusion or intracranial extension (especially vertebral dissection with subarachnoid hemorrhage), when ischemia recurs on therapy, when a pseudoaneurysm is expanding or symptomatic, and when the vessel pattern suggests FMD, a genetic arteriopathy, or more than one dissected artery. Endovascular reconstruction of the dissected segment is not first-line; the 2021 guideline allows consideration (Class 2b) after recurrent events despite antithrombotic therapy.8 Classify the mechanism as TOAST "other determined" on the TOAST page, score the deficit with the NIHSS as for any ischemic stroke, and counsel early activity restriction — avoiding contact sports and extreme neck strain — more than a lifetime label.1

The bottom line

Image the neck. Do not withhold reperfusion. Pick an antithrombotic for at least 3 months, individualizing anticoagulation versus antiplatelet therapy with occlusive thrombus, bleeding risk, and intracranial extension in mind, and do not expect CADISS or TREAT-CAD to make the choice. Restudy the vessel. Most patients heal, most do not dissect again, and most of the stroke risk is front-loaded into the first weeks.