Stroke
An untreated large-vessel stroke burns 1.9 million neurons a minute, so this is written to be read quickly. The numbers, the trials behind them, and where the evidence still has holes.
At a glance
- Burden (GBD 2021): 11.9 million new strokes, 93.8 million survivors, 7.3 million deaths (10.7% of all deaths), 160.5 million DALYs. Third leading cause of death.
- Lifetime risk: about 1 in 4 adults over 25.
- Inequity: 87% of stroke deaths and 89% of DALYs fall in low- and lower-middle-income countries. 53% of incident strokes are in people under 70.
- Preventable: about 84% of the burden is attributable to 23 modifiable risk factors (GBD). INTERSTROKE: 10 factors explain 90.7% of population attributable risk.
- Treatable: thrombectomy for large-vessel occlusion has an NNT of 2.6 for one-step mRS improvement. Yet only about 13% of patients in the Australian registry received thrombolysis in 2024.
Biggest take-home: systolic blood pressure. It accounts for 56.8% of stroke DALYs worldwide, more than any other factor, before and after the first stroke.
Definition
- Classic WHO (time-based): rapidly developing focal (or global) neurological deficit lasting more than 24 hours or causing death, with no apparent non-vascular cause.
- AHA/ASA 2013 (tissue-based): CNS infarction is brain, spinal cord or retinal cell death from ischaemia, shown on imaging or pathology. Symptoms of 24 hours or more stand in only when imaging is unavailable or negative.
- Ischaemic stroke: symptomatic CNS infarction. Silent infarct: imaging-proven infarct without attributable symptoms. It raises later stroke risk 2 to 4 times.
- Intracerebral haemorrhage (ICH): non-traumatic bleeding into brain parenchyma or ventricles. Cerebral microbleeds count as abnormal at any size.
- Subarachnoid haemorrhage (SAH): bleeding into the subarachnoid space, mostly from a ruptured aneurysm.
- TIA: transient focal brain, spinal cord or retinal ischaemia without acute infarction. A DWI-positive "TIA" is a stroke, however brief the symptoms.
- Cerebral venous thrombosis with infarction or haemorrhage also counts as stroke under the AHA/ASA definition.
History
- Hippocratic corpus: apoplexia, "struck down with violence". The word described the event, not the vessel, and survived unchallenged for two millennia.
- 1658, Johann Jacob Wepfer (Schaffhausen): autopsied patients dead of "apoplexy" and found intracranial bleeding, separating subarachnoid from intracerebral haemorrhage and proposing a ruptured artery as the cause. Harvey's circulation was still controversial at the time.
- 1689: William Cole put "stroke" into English medical writing as a synonym for apoplexy.
- 1852, Rudolf Virchow: split apoplexia sanguinea from apoplexia ischaemica and coined "thrombosis" and "embolism" — the first mechanistic account of ischaemic stroke.
- 1951–1965, C. Miller Fisher: a Canadian neurologist who spent about three and a half years as a prisoner of war, then described internal carotid occlusion (1951), TIA as a stroke warning (1952) and lacunar infarcts and their syndromes (Neurology 1965). He also linked atrial fibrillation to cerebral embolism.
- 1976–1981, the penumbra: worked out in Lindsay Symon's Queen Square laboratory, named by Astrup in Stroke in 1977 and set out by Astrup, Siesjö and Symon in 1981. Tissue damaged but not yet dead became the target of every reperfusion trial since.
- 1993: Camilo Gomez published "Time is Brain!", the slogan Saver quantified in 2006.
- 1995, NINDS rt-PA trial: patients treated within 3 hours were at least 30% more likely to have minimal or no disability at 3 months. FDA approval followed in 1996 — the first treatment for acute ischaemic stroke.
- 2015, the thrombectomy year: five trials in one year of the NEJM (MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, REVASCAT). Most stopped early once equipoise collapsed: EXTEND-IA, led from Melbourne, halted after 70 patients with mRS 0–2 of 71% vs 41%.
Pathophysiology
Ischaemic cascade
- Falling CBF leads to ATP depletion, then Na⁺/K⁺-ATPase failure and anoxic depolarisation.
- Glutamate release causes NMDA-mediated Ca²⁺ influx (excitotoxicity), mitochondrial failure and free radicals.
- Cytotoxic oedema comes first. MMP-9-driven blood–brain barrier breakdown then causes vasogenic oedema and haemorrhagic transformation risk.
- Microglia and neutrophils drive inflammation, and apoptosis continues for days.
- Reperfusion can still fail at microvascular level ("no-reflow"), one rationale for adjunctive neuroprotection.
Core, penumbra and collaterals
- Core: irreversibly injured. Penumbra: electrically silent but viable, and salvageable if reperfused.
- Leptomeningeal collaterals set how fast the core grows. With good collaterals ("slow progressors") penumbra survives for hours, which is the biology behind the 6–24 hour EVT windows (DAWN, DEFUSE 3).
- Perfusion imaging (CTP or MRI) estimates core and penumbra and selects patients beyond fixed time windows.
Time is brain (Saver 2006)
- Per minute of untreated large-vessel ischaemia: 1.9 million neurons, 14 billion synapses and 12 km of myelinated fibres lost.
- Per hour: 120 million neurons.
- Thrombolysis benefit falls steeply with time (see treatment).
Ischaemic mechanisms (TOAST)
- Large-artery atherosclerosis: artery-to-artery embolism, in-situ thrombosis, branch occlusion, hypoperfusion. Extracranial carotid disease is commoner in European-ancestry populations. Intracranial atherosclerosis is commoner in Asian, African and Hispanic populations.
- Cardioembolism: AF, recent MI or LV thrombus, valve disease, endocarditis, cardiomyopathy.
- Small-vessel occlusion: lipohyalinosis or microatheroma of penetrating arteries, causing lacunar infarcts. Part of cerebral small vessel disease.
- Other determined: dissection (a leading cause in the young), vasculitis, RCVS, moyamoya, sickle cell, CADASIL, cancer-related hypercoagulability.
- Undetermined: cryptogenic, ESUS, or two or more competing causes.
Haemorrhagic mechanisms
- Hypertensive arteriopathy: deep ICH (basal ganglia, thalamus, pons, cerebellum).
- Cerebral amyloid angiopathy: lobar ICH in older adults, with lobar microbleeds and cortical superficial siderosis.
- Other causes: anticoagulants, AVM, cavernoma, tumour, venous thrombosis, sympathomimetic drugs.
- Secondary injury: early haematoma expansion, perihaematomal oedema, thrombin and iron toxicity, and intraventricular extension causing hydrocephalus.
- SAH: aneurysm rupture, then early brain injury and delayed cerebral ischaemia.
Epidemiology
Global (GBD 2021, WSO Fact Sheet 2025)
- Incidence 11.9 million; age-standardised 142 per 100,000.
- Prevalence 93.8 million.
- Deaths 7.3 million: second among non-communicable causes, third overall after ischaemic heart disease and COVID-19.
- DALYs 160.5 million, the fourth commonest cause.
- Subtypes (incidence): ischaemic 65.3%, ICH 28.8%, SAH 5.8%. ICH produces about half of all stroke DALYs despite being under a third of cases.
- Income split: ICH is 31.1% of strokes in LMICs vs 17.8% in high-income countries.
- Trends 1990–2021 (counts): incident strokes +70%, prevalent +86%, deaths +44%, DALYs +32%.
- Age-standardised rates have fallen, but the fall in incidence stalled from 2015. Rates are rising in people under 70, in South-East Asia, East Asia and Oceania, and in low-SDI countries.
- Age and sex: 53% of incident strokes are in people under 70, and 15% in those aged 15–49. Men have 53% of incident strokes.
- Cost: over US$890 billion a year (0.66% of global GDP), projected to nearly double by 2050.
Australia
- 45,785 strokes in 2023 (34,793 first-ever), about one every 11 minutes. Projected to reach about 72,000 a year by 2050.
- About half a million survivors.
- 9,147 stroke deaths in 2024; cerebrovascular disease was the 4th leading cause of death (ABS).
- Age-standardised stroke death rate fell 76% between 1980 and 2019 (104 to 25 per 100,000).
- Aboriginal and Torres Strait Islander people: stroke death and hospitalisation rates 1.5 times non-Indigenous rates.
- Lifetime cost of 2023's strokes: $15.7 billion.
Causes and risk factors
- Non-modifiable: age, family history, ancestry (intracranial atherosclerosis and ICH are commoner in Asian populations), sex.
- Ischaemic: hypertension, AF, diabetes, raised ApoB/LDL, smoking, carotid or intracranial atherosclerosis, obesity, inactivity, CKD, sleep apnoea. In the young: dissection, PFO, migraine with aura, drugs, thrombophilia.
- ICH: hypertension (more strongly linked to ICH than to ischaemic stroke in INTERSTROKE), CAA, anticoagulation, heavy alcohol, sympathomimetics.
- SAH: smoking, hypertension, heavy alcohol, family history, ADPKD.
- Women: pregnancy and postpartum, hypertensive disorders of pregnancy, oral oestrogen, migraine with aura.
- INTERSTROKE: smoking, diabetes, ApoB/ApoA1 and cardiac causes were more strongly linked to ischaemic stroke than to ICH.
Preventable factors
INTERSTROKE (32 countries, 13,447 cases, Lancet 2016)
- Ten factors explain 90.7% of population attributable risk (PAR), similar in men and women and in younger and older patients.
- PAR by factor: hypertension 47.9%, physical inactivity 35.8%, ApoB/ApoA1 26.8%, poor diet 23.2%, waist-to-hip ratio 18.6%, smoking 12.4%, cardiac causes 9.1%, alcohol 5.8%, diabetes 3.9%. Psychosocial stress was also significant.
GBD 2021 attributable DALYs
- High SBP 56.8%
- Ambient particulate matter 16.6%
- Smoking 13.7%
- High LDL 13.1%
- Household air pollution 11.2%
- High sodium 10.6%
- High fasting glucose 10.3%
- Kidney dysfunction 9.3%
- By cluster: metabolic 68.8%, environmental 36.7%, behavioural 35.2%.
- Fastest rising, 1990–2021: high BMI +88%, high ambient temperature +72%, high fasting glucose +32%, sugar-sweetened beverages +23%.
- Falling: smoking −13%, low vegetable intake −30%.
- Heat caused almost 2 million stroke DALYs in 2021, so climate is now a stroke risk factor.
Diagnosis essentials
- Pre-hospital: FAST recognition and an LVO screen. AHA/ASA 2026 advises taking suspected LVO direct to the nearest thrombectomy-capable centre where access is reasonable.
- Imaging: non-contrast CT within 25 minutes of arrival (AHA/ASA 2026), then CTA arch-to-vertex, with CTP or MRI for extended windows or unclear cases. Don't delay thrombolysis for CTA or CTP.
- ASPECTS on plain CT can select EVT candidates where perfusion imaging is unavailable.
- NIHSS (0–42) grades severity. Low scores can hide disabling deficits (aphasia, hemianopia, gait ataxia).
- Mimics: hypoglycaemia, seizure or Todd's paresis, migraine aura, functional disorder, sepsis unmasking an old deficit, tumour, subdural haematoma.
- Chameleons: posterior circulation stroke presenting as isolated vertigo. Use HINTS and a low threshold for MRI.
- Work-up: ECG and prolonged rhythm monitoring (longer monitoring finds more AF), echo (bubble study in the young), carotid and intracranial imaging, lipids, HbA1c. Screen for vasculitis, thrombophilia or dissection when indicated.
- TIA: NICE NG128 advises against using ABCD2 or other scores to triage. Assess suspected TIA within 24 hours.
- Where AI fits: automated LVO alerts, core and penumbra maps and ASPECTS scoring speed triage and transfer. Evidence of better patient outcomes is mostly observational, and medium or distal occlusions are easily missed. It supports a clinician's read but doesn't replace one.
Current treatment
IV thrombolysis
- Within 4.5 hours: tenecteplase 0.25 mg/kg (max 25 mg) single bolus, or alteplase 0.9 mg/kg (max 90 mg). Both are endorsed by AHA/ASA 2026 and the Stroke Foundation.
- Time-dependence (Emberson 2014, 6,756 patients, alteplase): odds of a good outcome were 1.75 at 3 hours or less, 1.26 at 3–4.5 hours, and 1.15 (non-significant) beyond 4.5 hours.
- Cost in bleeding (same analysis): fatal parenchymal haemorrhage 2.7% vs 0.4%.
- Tenecteplase vs alteplase: AcT (n=1,600): mRS 0–1 36.9% vs 34.8% (non-inferior), sICH 3.4% vs 3.2%, mortality 15.3% vs 15.4%. ATTEST-2 (n=1,858) and TRACE-2 were also non-inferior. Tenecteplase is easier: one bolus and no infusion during transfer.
- Extended windows: the Stroke Foundation allows alteplase up to 9 hours with perfusion mismatch. AHA/ASA 2026 allows thrombolysis up to 24 hours in selected imaging-selected patients.
- TRACE-III (tenecteplase at 4.5–24 hours for LVO with no EVT access): mRS 0–1 33.0% vs 24.2%, sICH 3.0% vs 0.8%. TIMELESS (tenecteplase before EVT at 4.5–24 hours) was neutral.
Endovascular thrombectomy (EVT)
- HERMES (5 RCTs, n=1,287): mRS 0–2 46.0% vs 26.5% (NNT about 5 for independence; 2.6 for one-step mRS shift). No excess mortality or sICH.
- Late window: DAWN mRS 0–2 49% vs 13%. DEFUSE 3 45% vs 17%, with mortality 14% vs 26%.
- Large core: MAGNA pooled RESCUE-Japan LIMIT, SELECT2 and ANGEL-ASPECT (n=1,009): mRS 0–2 23% vs 9%, mRS 0–3 41% vs 24%, sICH 1.8% vs 1.6%, mortality 27% vs 28%. Benefit extended up to an estimated core of about 150 mL. TENSION and LASTE also support EVT.
- AHA/ASA 2026 recommends EVT for ASPECTS 3–5 up to 24 hours (age under 80, NIHSS 6 or more, no significant mass effect).
- Basilar occlusion: mRS 0–3 46% vs 23% in ATTENTION (mortality 37% vs 55%) and 46% vs 24% in BAOCHE.
- Blood pressure after successful EVT: ENCHANTED2/MT target below 120 mmHg worsened function (OR 1.37). AHA/ASA 2026 lists intensive lowering below 140 mmHg after recanalisation as harmful.
Systems and stroke unit care
- Mobile stroke units: BEST-MSU (n=1,047) utility-weighted mRS 0.726 vs 0.657. 33% vs 3% were thrombolysed within 60 minutes of onset. AHA/ASA 2026 gives class 1 where available.
- Stroke units (Cochrane): death OR 0.81, death or dependency OR 0.79 vs general wards. The benefit lasts to 5 years.
- QASC (Australia): nurse-led fever, sugar and swallowing protocols cut 90-day death or dependency from 58% to 42% (NNT 6.4).
- AVERT (Australia): very early, high-dose mobilisation within 24 hours was worse: mRS 0–2 46% vs 50%, OR 0.73.
Intracerebral haemorrhage
- INTERACT3 bundle (n=7,036): SBP below 140 within 1 hour, glucose control, temperature 37.5 °C or less, and warfarin reversal to INR below 1.5 within 1.5 hours. Ordinal mRS OR 0.86 (0.76–0.97).
- Reversal: warfarin: 4-factor PCC plus vitamin K. Dabigatran: idarucizumab.
- Factor Xa inhibitors, ANNEXA-I: andexanet improved haemostatic efficacy (67.0% vs 53.1%) but increased thrombotic events (10.3% vs 5.6%), including ischaemic stroke 6.5% vs 1.5%, with no mortality difference.
- Surgery, ENRICH: minimally invasive evacuation within 24 hours of 30–80 mL ICH improved 180-day utility-weighted mRS (0.458 vs 0.374). Benefit was confined to lobar haematomas.
Secondary prevention
- Short DAPT (minor stroke or high-risk TIA):
- CHANCE: 8.2% vs 11.7%, HR 0.68, no bleeding excess.
- POINT: 5.0% vs 6.5%, HR 0.75; major haemorrhage 0.9% vs 0.4%. Benefit was concentrated in the first 21 days (3.6% vs 5.6%).
- INSPIRES (start within 72 hours, atherosclerotic): 7.3% vs 9.2%, NNT about 53. Bleeding 0.9% vs 0.4%.
- CHANCE-2: in CYP2C19 loss-of-function carriers, ticagrelor plus aspirin gave 6.0% vs 7.6% with clopidogrel plus aspirin.
- Standard approach: 21 days of DAPT, then a single antiplatelet.
- AF: a DOAC, started early.
- ELAN (n=2,013): 30-day composite 2.9% early vs 4.1% later.
- OPTIMAS (n=3,648): 3.3% in both arms (non-inferior).
- CATALYST (pooled TIMING, ELAN, OPTIMAS and START, n=5,411): 2.12% vs 3.02% at 30 days, OR 0.70, with no excess sICH.
- ESUS: DOACs failed in NAVIGATE ESUS, RE-SPECT ESUS and ARCADIA (atrial cardiopathy; 4.4% recurrence in both arms). Look for AF; don't empirically anticoagulate.
- Lipids:
- SPARCL (atorvastatin 80 mg): stroke 11.2% vs 13.1%, HR 0.84, NNT 46 over 5 years. More haemorrhagic strokes (55 vs 33).
- TST (LDL target below 1.8 mmol/L): 8.5% vs 10.9%, HR 0.78.
- Stroke Foundation and AHA target: LDL below 1.8 mmol/L (70 mg/dL), adding ezetimibe and then a PCSK9 inhibitor if needed.
- Blood pressure: target below 130/80 for most (AHA/ASA 2021).
- PFO: individual-patient meta-analysis (6 RCTs, n=3,740): HR 0.41; annual stroke 0.47% vs 1.09%. By PASCAL category, "probable" HR 0.10 but "unlikely" HR 1.14. The Stroke Foundation recommends closure under 60 when the PFO is the likely cause.
- Symptomatic carotid stenosis 70–99%: NASCET showed 2-year ipsilateral stroke 26% with medical treatment vs 9% with endarterectomy (ARR 17%, NNT about 6). Operate early.
- Asymptomatic carotid stenosis, CREST-2: stenting plus intensive medical treatment 2.8% vs 6.0%. Endarterectomy 3.7% vs 5.3% (not significant).
Rehabilitation
- Early, task-specific, high-repetition therapy delivered by a stroke-unit team.
- Progressive resistance training for weakness is a strong Stroke Foundation recommendation.
- Avoid very early high-dose mobilisation (AVERT).
- Screen for mood, cognition, fatigue, dysphagia and secondary prevention adherence.
Emerging treatments and trials
Factor XI(a) inhibitors
- OCEANIC-STROKE (asundexian 50 mg with antiplatelets, n=12,237, NEJM 2026): ischaemic stroke 6.2% vs 8.4%, HR 0.74, NNT 53 at 1 year. ISTH major bleeding 1.9% vs 1.7% (not significant).
- Asundexian is not yet approved anywhere; it has FDA Fast Track status. OCEANIC-AF was stopped early because asundexian was less effective than apixaban.
- LIBREXIA-STROKE (milvexian) and LIBREXIA-AF are ongoing and expected to report in 2027. LIBREXIA-ACS stopped for futility (MACE HR 1.05).
Neuroprotection
- Nerinetide: ESCAPE-NA1 was neutral overall (RR 1.04), but in the no-alteplase subgroup mRS 0–2 was 59.3% vs 49.6%. Plasmin generated by alteplase is thought to cleave the peptide.
- ESCAPE-NEXT and FRONTIER were both neutral. A pooled analysis suggests benefit within 3 hours with reperfusion; ACT-42 is underway.
- ApTOLL (TLR4 antagonist): the phase 1b/2a APRIL trial (n=151, with EVT) gave encouraging safety and mortality signals. Phase 2b/3 is planned.
Medium and distal vessel occlusion (EVT)
- ESCAPE-MeVO (n=530): mRS 0–1 41.6% vs 43.1%; mortality 13.3% vs 8.4%; sICH 5.4% vs 2.2%.
- DISTAL (n=543): common OR 0.90; sICH 5.9% vs 2.6%.
- DISCOUNT was stopped early. Treat medium and distal occlusions outside trials only selectively.
Anti-inflammatory
- Colchicine: CONVINCE (n=3,154) HR 0.84 (0.68–1.05), missing its primary endpoint, with a per-protocol signal.
- CHANCE-3 (n=8,343): 6.3% vs 6.5%, HR 0.98.
Oedema
- IV glibenclamide, CHARM (n=535, large hemispheric infarction): primary endpoint neutral, with subgroup signals (lesion volume under 125 mL).
- CHARM was stopped early for business reasons.
Recovery
- Vagus nerve stimulation paired with rehab, VNS-REHAB (n=108, chronic stroke): Fugl-Meyer upper limb score improved 5.0 vs 2.4 points.
- Stem cells: TREASURE (MultiStem, n=206) was neutral (excellent outcome 11.5% vs 9.8%). MASTERS-2 is ongoing.
Theoretical frontiers
Hypotheses, not recommendations.
Causes: theories
- Clonal haematopoiesis as a vascular risk factor: CHIP alters macrophage inflammation. In 78,752 people from 8 cohorts, CHIP carried total stroke HR 1.14, haemorrhagic stroke HR 1.24, with TET2 the strongest driver (Bhattacharya, Stroke 2021). Test: does CHIP status predict small-vessel progression on MRI, and does anti-inflammatory treatment work better in carriers?
- Plastic particles in plaque: polyethylene was detected in 58.4% of carotid plaques from 304 endarterectomy patients, and detection carried HR 4.53 for MI, stroke or death over about 34 months (Marfella, NEJM 2024). Observational, single-country, and confounding by lifestyle is wide open. Test: standardised particle assays in multiple biobanked plaque cohorts.
- Infection as a trigger, not a background: stroke rates rose after herpes zoster (incidence ratio 1.63 in weeks 1–4, falling to 1.23 by weeks 13–26; n=6,584, Clin Infect Dis 2014), with lower ratios in those given antivirals. Test: a trigger-period design in linked immunisation and stroke registries.
- The AF substrate exists before the AF: a neural network read 12-lead ECGs in sinus rhythm and flagged people with paroxysmal AF (AUC 0.87, 79% accuracy; 180,922 patients, Lancet 2019), and ARTESiA showed subclinical AF is treatable (stroke or systemic embolism 0.78% vs 1.24%/year, HR 0.63, but major bleeding 1.71% vs 0.94%). Test: AI-selected sinus-rhythm patients randomised to monitoring or anticoagulation after ESUS.
- Lp(a) may be causal but not yet treatable: Mendelian data support causality, yet Lp(a)HORIZON (pelacarsen, over 8,300 patients) missed its primary endpoint in 2026. Olpasiran, lepodisiran and zerlasiran trials are still running. Test: whether a bigger Lp(a) reduction, or earlier treatment, changes stroke specifically.
Prevention: theoretical options
- Swap the salt, not the behaviour: SSaSS randomised 20,995 people in 600 Chinese villages to 75% NaCl/25% KCl: stroke RR 0.86, major vascular events 0.87, death 0.88, with no excess clinical hyperkalaemia over 4.74 years. Test: effectiveness and safety in CKD-heavy, Western food-supply settings, including Australia.
- Polypill as a population tool: TIPS-3 (primary prevention) gave HR 0.69 for the composite with polypill plus aspirin vs double placebo. Test: a polypill arm in a stroke-specific primary prevention trial, with adherence as the measured mechanism.
- GLP-1 receptor agonists as stroke drugs: pooled trial data give cerebrovascular event RR 0.83 and ischaemic stroke RR 0.73, with no signal for haemorrhagic stroke. All of it comes from diabetes and obesity trials. Test: a trial with stroke as the primary endpoint in non-diabetic high-risk patients.
- Vaccination as vascular prevention: in 102,766 recombinant zoster vaccine recipients matched to 411,064 controls, stroke hospitalisation ran 3.3 vs 5.5 per 1,000 person-years (adjusted HR 0.575; Clin Infect Dis 2025). Confounding by health-seeking behaviour is the obvious objection. Test: a target-trial emulation, or stroke endpoints pre-specified in vaccine trials.
- One-and-done gene editing of lipid risk: VERVE-102, a single infusion base editor targeting PCSK9, cut LDL by a mean 53% at the top dose in 14 patients, with no treatment-related serious events. No outcome data at all. Test: long-term durability and safety before any stroke-prevention claim.
- Clean air and heat as prescribable exposures: ambient particulates account for 16.6% of stroke DALYs and heat-attributable DALYs rose 72% since 1990, yet no trial has tested an air-quality or heat-adaptation policy against stroke incidence. Test: stepped-wedge evaluation of clean-air or cooling interventions in high-exposure cities.
Treatment: theoretical options
- Finish the job inside the artery: CHOICE gave intra-arterial alteplase after successful thrombectomy: mRS 0–1 59.0% vs 40.4% (JAMA 2022), but it stopped at about 60% of planned enrolment. The mechanism is microvascular no-reflow. Test: an adequately powered trial, with tenecteplase as the obvious comparator.
- Cheap oxygen as a bridge: OPENS-2 gave 100% oxygen at 10 L/min for 4 hours around thrombectomy (n=282): adjusted common OR 1.65 for mRS shift, with no excess sICH or death (Lancet 2025). Single-country and surprising, given decades of neutral oxygen trials. Test: multinational replication with core-growth imaging.
- Condition the limb to protect the brain: RICAMIS found remote ischaemic conditioning improved mRS 0–1 (67.4% vs 62.0%, JAMA 2022), while RESIST (n=1,500, prehospital) was flat (OR 0.95). Dose, timing and who responds are all unresolved. Test: conditioning started in the ambulance in confirmed LVO, with penumbral imaging endpoints.
- Neuromodulation for fixed deficits: cervical epidural spinal cord stimulation restored arm and hand function in 2 patients years after stroke, with some carry-over after stimulation stopped (Nat Med 2023). Two patients is two patients. Test: sham-controlled crossover with blinded upper-limb scoring.
- Regrow the cortex, carefully: astrocyte-to-neuron reprogramming looked like a route to replacing lost tissue until stringent lineage tracing showed the "new" neurons were pre-existing ones (Wang, Cell 2021). Any future claim needs genetic lineage tracing built in from the start.
- Stratify instead of broaden: neuroprotection, colchicine and ESUS anticoagulation all failed as all-comers strategies. The untested design is biomarker-gated: inflammation-high patients for colchicine, CHIP carriers for anti-inflammatory therapy, mechanism-defined ESUS subgroups for anticoagulation.
Gaps and open questions
Labelled hypotheses, not recommendations.
- Hypothesis: small vessel disease is under-treated. Lacunes, white matter disease and much vascular dementia share one pathology with no proven disease-modifying drug. Endothelial and blood–brain-barrier targets deserve definitive trials.
- Hypothesis: ICH needs its own drugs. ICH causes about half of global stroke DALYs, and close to a third of strokes in LMICs. Yet beyond the INTERACT3 bundle and lobar surgery, no drug has improved outcome. Haemostasis in non-anticoagulated ICH and CAA-specific therapy remain open.
- Hypothesis: "ESUS" is several diseases. Three neutral DOAC trials suggest a mixed group (non-stenotic plaque, occult cancer, atrial cardiopathy without AF). Mechanism-stratified trials using imaging and biomarkers are the logical next step.
- Hypothesis: neuroprotection failed on timing, not biology. Most agents were tested without reliable reperfusion. The nerinetide no-alteplase and early-window signals argue for testing agents given with or before EVT, including in the ambulance.
- Hypothesis: anti-inflammatory therapy works in a subset. CONVINCE's per-protocol signal and coronary colchicine data suggest benefit may be confined to atherosclerotic or high-inflammation phenotypes. This is untested prospectively.
- Hypothesis: AF after ICH has an untested best option. The choice between DOAC, left atrial appendage occlusion or nothing rests on small or ongoing trials.
- Excluded populations: the very old, people with pre-stroke disability, pregnancy, and patients from LMICs, where ICH is a bigger share, are under-represented. So are Indigenous populations, despite 1.5 times higher Australian stroke mortality.
- Implementation gap (Australia):
- About 13% thrombolysed (AuSCR 2024).
- Median door-to-needle 67 minutes; only 11 of 62 hospitals met the 60-minute target.
- 72% admitted to a stroke unit, but only 48% got 90% or more of their care there (Audit 2023).
- Regional patients were 76% less likely to reach a stroke unit.
- Rural access to EVT, not new drugs, may be the biggest unclaimed gain.
- Hypothesis: prevention needs population levers. High BMI, glucose, heat and air pollution are rising contributors. Population sodium reduction, polypills and clean-air policy probably have larger effects than any new drug, but trials of stroke outcomes are scarce.
- Hypothesis: post-stroke fatigue and depression are neglected. Fatigue affects about half of survivors and has no proven treatment. Depression treatment is under-tested as a target for functional recovery.
Prognosis and quality of life
- Short term (AuSCR 2024): about 1 in 5 patients died. At 3–6 months about a third had moderate to severe disability, and about half reported anxiety, depression or problems with usual activities.
- ICH: early mortality 30–40%, the deadliest stroke type.
- Recurrence after first stroke (Mohan 2011): 3.1% at 30 days, 11.1% at 1 year, 26.4% at 5 years, 39.2% at 10 years. Modern secondary prevention probably lowers this; WSO cites 10–16% at 1 year.
- Depression: 31% at any time point (61 studies, 25,488 people).
- Fatigue: 48% (35 studies).
- Cognition: cognitive impairment without dementia is common in the first year. Screen routinely.
- Return to work (among those employed before): 55.7% at 1 year, 67.4% at 2 years.
- Change over time: Australian age-standardised stroke death rate down 76% (1980–2019); event rates down 25% (2001–2018). Globally, age-standardised rates are falling but total burden is rising with population growth and ageing.
Weird trivia
- Moyamoya was named by an editor's error: Suzuki and Takaku submitted their 1969 paper as "A disease showing abnormal net-like vessels in base of the brain", with "moyamoya disease" as the subtitle. The editor, H. Houston Merritt, swapped the two. The Japanese word describes a hazy puff of cigarette smoke, which is what the collaterals look like on angiography.
- Pasteur worked with a hemiplegia: he had a left hemiplegia at 46, with lasting deficits and later speech disturbance (probably a cavernoma, Rev Neurol 1996). He vaccinated Joseph Meister against rabies in July 1885, long after the stroke.
- The most consequential blood pressure in history: Franklin Roosevelt's systolic was over 300 mmHg with a diastolic of 190 on the morning of 12 April 1945, when he died of a massive cerebral haemorrhage. His cardiologist had recorded 186/108 a year earlier. There was no autopsy, and at the time high blood pressure was widely thought to be compensatory.
- Two stroke drugs came from venom and bat spit: ancrod, from Malayan pit viper venom, gave favourable outcomes in 42.2% vs 34.4% in the STAT trial (JAMA 2000) before later trials sank it. Desmoteplase, from vampire bat (Desmodus rotundus) saliva, was neutral in DIAS-3 (51.3% vs 49.8%) and dropped in 2014.
- "Time is brain" is a 1993 coinage, by Camilo Gomez, and only got its famous numbers 13 years later when Saver counted 1.9 million neurons a minute.
- The man who described the TIA was a prisoner of war: Miller Fisher was captured after his ship was sunk in 1941 and spent about three and a half years in a German camp before the work that produced lacunes, TIA and the carotid-stroke link.
- A quarter of a century separated the idea from the drug: the penumbra was named in 1977; the first treatment that could save it was approved in 1996.
Guidelines
- Stroke Foundation: Australian and New Zealand Living Clinical Guidelines for Stroke Management
- newsGP: key updates to the living stroke guidelines (tenecteplase, 9-hour window, DAPT, PFO, LDL)
- AHA/ASA 2026 Guideline for the Early Management of Acute Ischemic Stroke (replaces 2018 and the 2019 update)
- AHA/ASA 2021 Guideline for the Prevention of Stroke in Patients With Stroke and TIA
- ACC: ten points to remember from the 2021 secondary prevention guideline
- AHA/ASA 2022 Guideline for Spontaneous Intracerebral Hemorrhage
- European Stroke Organisation guidelines (incl. 2025 ICH, 2025 BP, 2026 SAH)
- NICE NG128: Stroke and TIA in over 16s, diagnosis and initial management
Key trials
- NASCET (1991): endarterectomy for symptomatic 70–99% stenosis, ARR 17% at 2 years.
- SPARCL (2006): atorvastatin 80 mg, stroke HR 0.84.
- QASC (2011): fever, sugar and swallow protocols, NNT 6.4.
- CHANCE (2013): 21-day DAPT, stroke 8.2% vs 11.7%.
- Emberson IPD (2014): alteplase benefit falls with time; none beyond 4.5 hours.
- AVERT (2015): very early mobilisation was harmful.
- HERMES (2016): EVT, mRS 0–2 46% vs 26.5%.
- INTERSTROKE (2016): 10 factors, PAR 90.7%.
- DAWN and DEFUSE 3 (2018): EVT to 24 hours with imaging selection.
- POINT (2018): DAPT benefit concentrated in the first 21 days.
- TST (2020): LDL below 1.8 mmol/L, HR 0.78.
- ESCAPE-NA1 (2020): nerinetide neutral overall.
- CHANCE-2 (2021): ticagrelor in CYP2C19 loss-of-function carriers, HR 0.77.
- PFO IPD (2021): closure HR 0.41, driven by PASCAL "probable" and "possible".
- VNS-REHAB (2021): Fugl-Meyer upper limb +5.0 vs +2.4.
- AcT (2022): tenecteplase non-inferior to alteplase.
- ATTENTION and BAOCHE (2022): EVT for basilar occlusion, mRS 0–3 46% vs 23–24%.
- ENCHANTED2/MT (2022): SBP below 120 after EVT was harmful.
- SELECT2, ANGEL-ASPECT, RESCUE-Japan LIMIT (2022–23): EVT for large core.
- ELAN (2023): early DOAC after AF stroke is safe.
- INSPIRES (2023): DAPT up to 72 hours, 7.3% vs 9.2%.
- INTERACT3 (2023): ICH care bundle, OR 0.86.
- ARCADIA (2024): apixaban no better than aspirin in atrial cardiopathy.
- ANNEXA-I (2024): andexanet improves haemostasis but more thrombosis.
- CONVINCE and CHANCE-3 (2024): colchicine neutral.
- ENRICH (2024): minimally invasive surgery for lobar ICH helps.
- OPTIMAS (2024): early DOAC non-inferior.
- TRACE-III (2024): tenecteplase at 4.5–24 hours without EVT, mRS 0–1 33% vs 24%.
- ATTEST-2 (2024): tenecteplase non-inferior.
- ESCAPE-MeVO and DISTAL (2025): EVT for medium and distal occlusion, no benefit.
- CREST-2 (2025): stenting beat medical treatment for asymptomatic stenosis; endarterectomy did not.
- OCEANIC-STROKE (2026): asundexian HR 0.74, no bleeding excess.
Sources
- GBD 2021 Stroke Collaborators. Burden of stroke and its risk factors 1990–2021. Lancet Neurol 2024 (abstract record)
- Feigin VL et al. WSO Global Stroke Fact Sheet 2025. Int J Stroke
- O'Donnell MJ et al. INTERSTROKE. Lancet 2016 (abstract record)
- The Lancet: INTERSTROKE press release with PAR by factor
- World Neurology: the AHA/ASA updated definition of stroke
- Saver JL. Time is brain, quantified. Stroke 2006 (HealthDay summary)
- AIHW: Heart, stroke and vascular disease, stroke
- Stroke Foundation: Economic Impact of Stroke report (2024)
- Stroke Foundation: ABS Causes of Death 2024
- NeuroNews: AuSCR annual report (2024 data)
- Stroke Foundation: National Stroke Audit 2023
- Prabhakaran S et al. 2026 AHA/ASA acute ischaemic stroke guideline (PDF)
- Emberson J et al. Alteplase IPD meta-analysis. Lancet 2014 (summary)
- Menon BK et al. AcT. Lancet 2022
- NeuroNews: ATTEST-2 and tenecteplase trials
- ACC: TRACE-III
- TheNNT: HERMES (Goyal M et al. Lancet 2016)
- TCTMD: HERMES
- ACC: DAWN
- ACC: DEFUSE 3
- The Hospitalist: TESLA and MAGNA large-core analyses
- TCTMD: ATTENTION and BAOCHE
- Practical Neurology: ESCAPE-MeVO and DISTAL
- NeuroNews: DISCOUNT
- TCTMD: ENCHANTED2/MT
- Healio: BEST-MSU
- Cochrane: organised inpatient (stroke unit) care
- Middleton S et al. QASC. Lancet 2011
- Bernhardt J et al. AVERT. Lancet 2015
- First10EM: INTERACT3 (Lancet 2023)
- ACC: ANNEXA-I (NEJM 2024)
- The Bottom Line: ENRICH (NEJM 2024)
- AHA: 2022 ICH guideline (early mortality)
- NHS SPS: CHANCE (NEJM 2013)
- Relias: POINT (NEJM 2018)
- TCTMD: POINT, benefit in first 21 days
- TCTMD: INSPIRES (NEJM 2023)
- ACC: CHANCE-2
- CTU Bern: ELAN (NEJM 2023)
- The Hospitalist: OPTIMAS and CATALYST
- TCTMD: ARCADIA
- GPnotebook: SPARCL (NEJM 2006)
- Healio: Treat Stroke to Target
- Healio: PFO closure IPD meta-analysis and PASCAL (JAMA 2021)
- NASCET. NEJM 1991 (abstract)
- ACC: CREST-2 (NEJM 2025)
- TCTMD: OCEANIC-STROKE (NEJM 2026)
- Bayer: asundexian regulatory status
- HCPLive: LIBREXIA-ACS and ongoing LIBREXIA trials
- ACC: ESCAPE-NA1 (Lancet 2020)
- News-Medical: ESCAPE-NEXT, FRONTIER and pooled nerinetide analysis
- Bellvitge Hospital: ApTOLL APRIL trial
- UCD: CONVINCE (Lancet 2024)
- World J Cardiol: colchicine trials table (CHANCE-3, CONVINCE)
- NeurologyLive: CHARM
- CGTLive: TREASURE (JAMA Neurol 2024)
- BioSpace: VNS-REHAB (Lancet 2021)
- Mohan KM et al. Cumulative risk of stroke recurrence. Stroke 2011
- Hackett ML, Pickles K. Frequency of depression after stroke. Int J Stroke 2014
- Prevalence of fatigue after stroke: meta-analysis. Eur Stroke J 2021
- Return to work post-stroke: meta-analysis. Arch Phys Med Rehabil 2019
- Sexton E et al. Cognitive impairment no dementia in the first year post-stroke. Eur Stroke J 2019
- Engelhardt E. Apoplexy, cerebrovascular disease and stroke: historical evolution of terms. Dement Neuropsychol 2017
- World Neurology: review of van Gijn, Stroke: A History of Ideas (Wepfer 1658, Virchow, Moniz)
- LITFL: C. Miller Fisher — carotid occlusion (1951), TIA (1952), lacunes
- C. Miller Fisher: biography, 1965 lacunes paper, wartime captivity
- Ischaemic penumbra: Symon's laboratory, Astrup 1977 and Astrup–Siesjö–Symon 1981
- Camilo R. Gomez: "Time is Brain!" (1993)
- NINDS: the 1995 rt-PA trial and 1996 FDA approval
- Endovascular Today: the 2015 thrombectomy trials, including EXTEND-IA
- Jirō Suzuki (1924–1990): how moyamoya disease got its name
- Bhattacharya R et al. Clonal haematopoiesis and risk of stroke. Stroke 2021
- Marfella R et al. Microplastics in carotid plaque. NEJM 2024 (summary)
- Langan SM et al. Risk of stroke following herpes zoster: self-controlled case series. Clin Infect Dis 2014
- Attia ZI et al. AI-enhanced ECG to detect AF during sinus rhythm. Lancet 2019 (summary)
- ARTESiA: apixaban in subclinical atrial fibrillation. NEJM 2024 (summary)
- TCTMD: Lp(a)HORIZON top-line results (pelacarsen, 2026)
- SSaSS: salt substitute and stroke. NEJM 2021 (summary)
- TCTMD: TIPS-3 polypill with or without aspirin (NEJM 2021)
- GLP-1 receptor agonists and cerebrovascular events: meta-analysis summary
- Recombinant zoster vaccine and stroke hospitalisation. Clin Infect Dis 2025 (summary)
- TCTMD: VERVE-102 base editing of PCSK9, phase 1b (2025)
- TCTMD: CHOICE, intra-arterial alteplase after thrombectomy (JAMA 2022)
- OPENS-2: normobaric hyperoxia with thrombectomy. Lancet 2025 (PDF)
- RICAMIS: remote ischaemic conditioning. JAMA 2022 (summary)
- RESIST: prehospital remote ischaemic conditioning (2023)
- Powell MP et al. Cervical spinal cord stimulation after stroke. Nat Med 2023 (summary)
- Wang LL et al. Revisiting astrocyte-to-neuron conversion with lineage tracing. Cell 2021
- STAT trial: intravenous ancrod (Malayan pit viper) for acute stroke. JAMA 2000
- Desmoteplase (vampire bat saliva): DIAS-3 and discontinuation
- Hektoen International: Franklin Delano Roosevelt's last illness
- Rumbach L et al. Pasteur's neurological disease. Rev Neurol 1996
Information only. Doses, time windows and eligibility vary by jurisdiction and product information. Check local protocols and the current guideline before treating.
maladywise.com/deep-dives/stroke/