Open questions — stroke¶

Last curated: 2026-10-27

Stable IDs are retained when questions are refined. Tier 1 questions could change practice and are designable now; Tier 2 questions require enabling measurement, mechanistic, or implementation work. “Open” means the retrieved evidence leaves a decision unresolved, not that no study exists.

Tier 1 — practice-changing and designable now¶

OQ-1 — Which distal or medium-vessel occlusions benefit from thrombectomy?¶

Why open: HERMES established a large disability benefit for predominantly proximal anterior-circulation occlusion (common OR 2.49, 95% CI 1.76–3.53), but smaller vessels combine lower untreated tissue burden with greater procedural perforation risk. Active trials test different anatomical and deficit thresholds (Goyal 2016, PMID 26898852; NCT06146790; NCT07185022). Sharpened 2026-10-20: DISCOUNT (244 patients, France) stopped for futility and excess hemorrhage. mRS 0–2 was 62% with thrombectomy versus 68% without (OR 0.73, 95% CI 0.40–1.31), and as-treated sICH was 11% versus 3% (Clarençon 2026, PMID 42485024). The 2026 SVIN focused guideline does not support routine EVT here (Majidi 2026, PMID 42404835). NCT06146790's registry status is now "unknown" (last verified 2024-08). The question is no longer whether routine EVT helps. It is whether any deficit- or territory-defined subgroup offsets the procedural hemorrhage.

Design: Pragmatic randomized comparison stratified by vessel, eloquent territory, NIHSS domain, core, collaterals, and rescue crossover; ordinal mRS plus cognition, language, vision, and safety.

OQ-2 — What is the net benefit of thrombectomy across the full large-core spectrum?¶

Why open: DAWN and DEFUSE 3 selected small-core mismatch phenotypes, leaving uncertainty at very large core volumes and across ASPECTS, age, edema risk, and desired outcome states (Nogueira 2018, PMID 29129157; Albers 2018, PMID 29364767).

Design: Individual-patient prospective meta-analysis with common imaging reprocessing, ordinal/utility outcomes, hemorrhage, decompression, and survival with severe disability.

OQ-3 — Does adjunct intra-arterial lysis improve outcome after successful thrombectomy?¶

Why open: Angiographic reperfusion does not guarantee microvascular tissue flow. Active phase 3 work tests intra-arterial alteplase after successful reperfusion, but hemorrhage may offset distal-thrombus benefit (NCT07727395; Jiang 2018, PMID 28987927).

Design: Double-blind randomized adjunct trial with eTICI-stratification, perfusion normalization, symptomatic ICH, ordinal mRS, and prespecified no-reflow phenotype.

OQ-4 — Which tenecteplase dose and population should replace alteplase?¶

Why open: AcT demonstrated pragmatic non-inferiority, but optimal dose and effects before thrombectomy, in non-LVO stroke, and in extended or unknown windows require consistent evidence (Menon 2022, PMID 35779553; Thomalla 2020, PMID 33176180).

Design: Harmonized individual-patient network comparison with dose, occlusion, imaging, time, reperfusion, symptomatic ICH, and ordinal function.

OQ-5 — Can a prehospital intervention safely preserve penumbra during transfer?¶

Why open: Collaterals determine infarct tempo and treatment delay erodes benefit, but collateral augmentation, conditioning, or neuroprotection has not shown a replicated patient-centered effect (Leng 2016, PMID 26579719; Paul 2021, PMID 33144066).

Design: Ambulance or spoke-hospital platform trial requiring confirmed occlusion, rapid target-engagement marker, and interaction with transfer/reperfusion delay.

OQ-6 — Which acute ICH care-bundle components add independent functional benefit?¶

Why open: INTERACT2's dichotomous primary outcome was neutral while ordinal outcome favored intensive BP lowering; hemostasis, temperature, glucose, airway, and surgery may produce additive effects (Anderson 2013, PMID 23713578; Greenberg 2022, PMID 35579034; NCT06429332).

Design: Factorial or response-adaptive platform initiated ultra-early, measuring achieved physiological separation, hematoma expansion, renal/ischemic harms, and ordinal mRS.

OQ-7 — Who benefits from minimally invasive ICH evacuation, and when?¶

Why open: Location, volume, intraventricular extension, neurological trajectory, timing, and surgical technique define biologically different groups; active studies test catheter and stereotactic approaches (Greenberg 2022, PMID 35579034; NCT07187687; NCT07471256).

Design: Technique-specific randomized trials with time-to-evacuation, residual volume, mortality, ordinal disability, and cognitive/communication outcomes.

OQ-8 — When and how should anticoagulation resume after anticoagulant-associated ICH?¶

Why open: Avoiding anticoagulation reduces recurrent bleeding exposure but leaves AF-related embolic risk; hemorrhage location, cerebral amyloid angiopathy markers, and competing risk vary (Greenberg 2022, PMID 35579034; NCT03243175).

Design: Randomized restart strategies stratified by lobar/deep site, microbleeds/siderosis, AF embolic risk, and left-atrial-appendage alternatives.

OQ-9 — Which aneurysmal-SAH strategy prevents delayed cerebral ischemia without causing harm?¶

Why open: Aneurysm securing prevents rebleeding, but delayed ischemia reflects vasospasm, microvascular dysfunction, spreading depolarization, and systemic complications; no single monitor or rescue pathway captures all (Claassen 2022, PMID 35985353; van Gijn 2007, PMID 17258671).

Design: Phenotype-triggered platform comparing monitoring/treatment bundles with infarction, cognition, quality of life, and treatment complication endpoints.

OQ-10 — Can mechanism-guided monitoring outperform fixed rhythm-monitoring duration after cryptogenic stroke?¶

Why open: Longer monitoring detects more AF, but detected short episodes may be incidental and do not uniformly establish embolic causality. TOAST also conflates incomplete and genuinely negative workups (Adams 1993, PMID 7678184; Amarenco 2013, PMID 23899749).

Design: Randomize monitoring intensity by atrial, imaging, age, and competing-source phenotype; primary endpoint should be recurrent stroke, not AF detection alone.

OQ-11 — Which PFO patients gain a causal—not merely associative—benefit from closure?¶

Why open: Closure trials enrich younger patients and high-risk anatomy after excluding competing mechanisms; extrapolation to older patients or low-risk anatomy is uncertain (Lee 2018, PMID 29544871).

Design: Prospective causal-risk model validated against closure interaction and recurrent embolic phenotype.

OQ-12 — What antithrombotic duration best balances early recurrence and bleeding after minor stroke/TIA?¶

Why open: CHANCE and POINT show early dual-antiplatelet benefit but accumulated bleeding with longer exposure; mechanism, infarct burden, genotype, and adherence may change the optimum (Wang 2013, PMID 23803136; Johnston 2018, PMID 29766750; Pan 2019, PMID 31424481). Widened 2026-10-20: OCEANIC-STROKE added a third axis to the trade-off. Asundexian on top of antiplatelet therapy lowered ischemic stroke (6.2% vs 8.4%; HR 0.74) without excess major bleeding (1.9% vs 1.7%), but with more lower-gastrointestinal bleeding (csHR 1.75) (Sharma 2026, PMID 41985132; Eikelboom 2026, PMID 42866879). The optimum now involves mechanism (antiplatelet versus factor XIa inhibition) as well as duration.

Design: Patient-level adaptive-duration trial with ischemic and major-bleeding utility weights.

OQ-13 — Can rehabilitation dose be assigned by response phenotype?¶

Why open: Motor, language, cognitive, and participation recovery differ by tract integrity, impairment, time, fatigue, and learning capacity; recovery biomarkers lack cross-validated treatment-response rules (Kim 2017, PMID 27503908; Bernhardt 2017, PMID 28697708).

Design: Sequential multiple-assignment trial using early within-person response to adapt task, intensity, and rest, with meaningful-change thresholds.

OQ-14 — Which caregiver intervention has a replicable effect on caregiver health and burden?¶

Why open: Nineteen RCTs had no pooled burden/strain effect, while seven individual educationally intensive studies were positive; remote interventions were also neutral with low certainty (Jammal 2024, PMID 39194341; Yu 2023, PMID 37696638).

Update 2026-10-27: A newer synthesis gives a different remote-intervention signal: its delivery-mode subgroup includes seven studies and 612 caregivers, with burden SMD 0.47 (95% CI 0.18–0.75; I² 64.45%). Positive values favor burden reduction in this review because change is baseline minus follow-up. Delivery-mode interaction was not significant (P=0.556); these exploratory across-study comparisons do not establish an optimal intervention package (Lin 2026, PMID 42909033; full text, Table 2). The different trial sets and effect-direction conventions need reconciliation before the earlier neutral result is treated as overturned.

Design: Multisite trial stratified by survivor dependency/cognition and caregiver baseline risk; specify intervention dose, respite access, mental health, work loss, and survivor outcomes.

OQ-15 — Does scheduled long-term needs review improve participation and quality of life?¶

Why open: An average 73.8% of survivors reported at least one unmet need, but descriptive prevalence has not established the effective review-and-response service (Chen 2019, PMID 31110106; Guo 2021, PMID 33671734).

Update 2026-10-27: SAFE reports a WSO survey on Post-Stroke Checklist use, barriers, and language/cultural adaptation, closing 30 November 2026. This is implementation mapping, not evidence that a review-and-response service improves outcomes (SAFE — “Post-stroke checklist survey launched”, https://www.safestroke.eu/2026/10/01/post-stroke-checklist-survey-launched/, accessed 2026-10-27).

Design: Cluster-randomized primary-care/community pathway with repeated accessible assessment, funded response options, participation, caregiver, and cost outcomes.

OQ-16 — Which system redesign closes reperfusion and stroke-unit inequities?¶

Why open: Low-income settings have 3.6-fold higher age-standardized mortality, while implementation evidence for affordable stroke services remains limited (GBD 2019 Stroke Collaborators 2021, PMID 34487721; Pandian 2017, PMID 28848165).

Design: Stepped-wedge regional trial comparing transport, telemedicine, essential imaging, drug access, transfer, and stroke-unit bundles; analyze treatment and disability equity.

OQ-25 — Do the 2026 positive neuroprotection trials replicate outside China and alongside reperfusion?¶

Why open: After decades of neutral neuroprotection trials, LAIS (loberamisal, 997 analyzed; mRS 0–1 69.7% vs 56.3%; RR 1.24, 95% CI 1.12–1.36) and ERASE-STROKE (oral edaravone, 614 without reperfusion therapy; mRS 0–1 65.4% vs 47.1%; OR 2.12, 95% CI 1.53–2.94) both reported large benefits. Both were single-country Chinese trials, and ERASE-STROKE excluded reperfused patients (Li 2026, PMID 42721021; Xiang 2026, PMID 42705987). The full texts were not retrievable on 2026-10-20, so the protocol details that decide transportability (concomitant care, endpoint adjudication, and handling of the 48-hour window) are unread.

Design: Independent multinational replication with ordinal mRS as the primary or key secondary endpoint, stratified by reperfusion therapy and time from onset, with prespecified target-engagement or biomarker sub-studies.

OQ-26 — How should the new late-window and mild-LVO thrombectomy reports change selection?¶

Why open: LATE-MT and MILD-MT now have published randomized reports, so these questions cannot be described as lacking randomized evidence (Yang 2026, PMID 42929656; Liu 2026, PMID 42929671). Their registered populations are imaging-selected (NCT05326932; NCT06179017). Full-text and source-table appraisal remains pending; no treatment threshold is inferred from the abstracts.

Design: First appraise the complete reports and protocols, including selection, crossover, procedural harms and analysis populations; then specify replication and transportability questions. SciELO and LILACS could not be searched (HTTP 403 access limitation), so this sweep cannot exclude regional evidence outside the sources queried.

Tier 2 — enabling science and measurement¶

OQ-17 — What is irreversibly injured “core” at the individual level?¶

Perfusion and diffusion thresholds are model-dependent; DWI can be negative or reversible, and CTP misses small lesions (Edlow 2017, PMID 28615423; Biesbroek 2013, PMID 23736122; Guadagno 2004, PMID 15090879).

OQ-18 — What causes microvascular no-reflow after macrovascular reperfusion?¶

Pericytes, distal emboli, endothelial swelling, leukocytes, and microthrombi are plausible mechanisms (Jiang 2018, PMID 28987927; Candelario-Jalil 2022, PMID 35387495). A PubMed update through 2026-08-30 found emerging retrospective human prediction work—DSA-NRP reported AUROC 0.933 for an angiographic machine-learning model but explicitly requires larger independent validation—without a prospectively validated treatment-selection phenotype or proven phenotype-linked intervention (Athreya 2026, PMID 42184179).

OQ-19 — When does post-stroke inflammation switch from injury to repair?¶

Microglial and peripheral immune responses are time- and context-dependent; indiscriminate suppression may block recovery (Jayaraj 2019, PMID 31291966; Zhao 2017, PMID 28260801).

OQ-20 — Can a blood biomarker add a safe, actionable diagnostic decision?¶

Across 141 studies and 136 candidates, few markers had meaningful clinical value; large associations for MMP-9 and other candidates remain heterogeneous and threshold-dependent (Hasan 2012, PMID 22320313; Krishnamoorthy 2022, PMID 34569521).

OQ-21 — How should cerebral small-vessel-disease progression be measured?¶

Lacunes, white-matter hyperintensities, microbleeds, perivascular spaces, and atrophy reflect overlapping mechanisms; a single burden score may hide causal heterogeneity (Wardlaw 2019, PMID 31097385).

OQ-22 — What biology drives post-stroke fatigue?¶

Forty-six imaging studies found no consistent association with common lesion measures, while fatigue remains functionally important (Jolly 2023, PMID 37485902).

OQ-23 — Which outcomes represent recovery rather than compensation?¶

The mRS and impairment scales cannot distinguish restitution, strategy change, environmental support, or participation; standard definitions exist but require adoption (Bernhardt 2017, PMID 28697708).

Communication difficulty persists for years and affects service access, social networks, and research participation (Wray 2017, PMID 28988185).

Dots not yet connected¶

These are testable junctions between evidence bodies already represented in this condition. “Missing junction” is a scoped synthesis gap, not a claim that no study exists.

#	Dot A	Dot B	The missing junction	Powers
D1	In 409 ESCAPE-NA1 patients with large-vessel occlusion, last known well to randomization under 6 hours, and near-complete reperfusion, median infarct growth was 4.74 mL/h (IQR 1.25–14.84). On log-transformed growth, collateral status β was −0.81 (95% CI −1.20 to −0.41), the same sign as ASPECTS (β −0.34, 95% CI −0.46 to −0.23) and the opposite sign to glucose (β 0.09, 95% CI 0.02 to 0.16) and NIHSS (β 0.07, 95% CI 0.04 to 0.10) (PMID 34493575)	In 198 patients with large-vessel occlusion transferred from three Dutch primary centers, median time from the first emergency call to the comprehensive center was 162 minutes (IQR 137–190). Highest-urgency transfer dispatch was associated with a shorter time (adjusted β −27.6 minutes, 95% CI −51.2 to −3.9). Thrombectomy was performed in 135/198 (68%) (PMID 34512538)	During interhospital transfer, fast infarct growth (≥5 mL/h) occurred in 71% of patients with hypoperfusion intensity ratio ≥0.40 and in 3% when the ratio was <0.40 (n=233). That ratio is a collateral-flow surrogate, and the cohort tested no treatment (PMID 36748945). RESIST tested ambulance-started limb remote ischemic conditioning in 902 patients with a target stroke diagnosis and found no 90-day mRS benefit (OR 0.95, 95% CI 0.75–1.20) (PMID 37787796). A collateral-directed treatment given during interfacility transfer is a separate question. On 2026-10-07 the grounder's PubMed query returned 5 records, and the same intersection in Europe PMC title-and-abstract fields returned those same 5. A broader PubMed query (collateral terms, interhospital transfer, and trial terms) returned 10. A narrow ClinicalTrials.gov `query.term` for stroke, collateral, transfer, interhospital, and intervention returned 0 studies. None of these queries found a collateral-support treatment trial. Growth and transfer time do not establish treatment benefit	OQ-5, OQ-16
D2	In a review that judged 30 hemorrhagic-transformation biomarker articles eligible, baseline MMP-9 had a pooled diagnostic odds ratio of 29.57 (95% CI 17.75–49.27). The report links MMP-9 ≥140 ng/mL to a 29.5-fold higher risk of symptomatic hemorrhagic transformation and an AUC of 0.881, and places that accuracy before thrombolysis. MMP-9 is one of five markers meta-analyzed inside the 30-article set (PMID 34569521)	POST-TNK randomized 540 patients in China with proximal occlusion, no prior intravenous thrombolysis, and eTICI 2c–3 after thrombectomy to intra-arterial tenecteplase 0.0625 mg/kg or none. 90-day mRS 0–1 was 49.1% (132/269) versus 44.1% (119/270) (adjusted RR 1.15, 95% CI 0.97–1.36; P=0.11). Symptomatic intracranial hemorrhage was 6.3% versus 4.4% (adjusted RR 1.43, 95% CI 0.68–2.99) (PMID 39804681)	A 7-trial meta-analysis (2131 patients) found mRS 0–1 odds ratio 1.44 (95% CI 1.21–1.72) and symptomatic hemorrhage odds ratio 1.15 (95% CI 0.75–1.75) for adjunctive intra-arterial thrombolysis after thrombectomy, with larger functional signals in dose, eTICI, NIHSS, and etiology subgroups (PMID 40506219). Those subgroups are clinical. On 2026-10-07 the grounder's PubMed query returned 1 record, a flat-panel CT study separating hemorrhage from barrier disruption (PMID 24948498); Europe PMC title-and-abstract search of that intersection returned the same record. A PubMed query for MMP-9 or blood-brain barrier with intra-arterial alteplase or tenecteplase and selection or prediction returned 0. A narrow ClinicalTrials.gov `query.term` combining intra-arterial thrombolysis, thrombectomy, biomarker, and blood-brain barrier returned 0 studies. No query found a trial that used a pre-lysis barrier or hemorrhage-risk blood marker to choose adjunct intra-arterial lysis	OQ-3, OQ-20
D7	A systematic review of return-to-work RCTs, searched through 4 January 2023, found 5 trials (626 participants). Three included mood and fatigue management in the same intervention, and 2 of those reported a higher percentage returning to work than their control groups (PMID 37674875)	RETAKE compared early specialist vocational rehabilitation for up to 12 months with usual care. Return to work for at least 2 hours per week at 12 months was 64.2% (165/257) versus 59.4% (117/197) (adjusted OR 1.12, 95% CI 0.75–1.68) (PMID 39614629). The development report for that intervention lists fatigue management among components added after piloting (PMID 41930667). In a separate high-dose walking booster study nested in a randomized trial, participants attended 91% (SD 13) of sessions. The results name fatigue as a barrier, and the implications text says therapists tailored sessions to physical ability and fatigue. Work was not an outcome (PMID 42803430)	POSITIF tested a telephone cognitive-behavioral fatigue intervention, collected return to work, and reported no between-group difference. The abstract gives no return-to-work percentages, and the intervention did not change motor-rehabilitation dose according to fatigue (PMID 35866206). The open junction is a trial that changes rehabilitation dose in response to fatigue and uses sustained employment as the outcome. On 2026-10-07 the grounder's PubMed query returned 5 records, including a heart-failure milrinone paper. Europe PMC title-and-abstract search returned 4 of those 5 and added no dose-scheduling employment trial. A broader PubMed query (stroke, fatigue, return to work or vocational, and trial terms) returned 13 records, including the RETAKE development report and POSITIF. A narrow ClinicalTrials.gov `query.term` for stroke, fatigue, vocational, return to work, rehabilitation, and dose returned 0 studies	OQ-13, OQ-22
D10	A review of 24 studies from 7 very-high-HDI countries reported age-standardized stroke incidence rate ratios of 1.7–3.2 for Aboriginal and Torres Strait Islander Australians, 1.2 for American Indians, 1.08–2.14 for Sámi populations, and 1.7–1.9 for Singaporean Malay populations, each versus its own comparator. Few included studies reported Indigenous stakeholder involvement or met CONSIDER reporting criteria (PMID 38354325)	Healing Right Way, a stepped-wedge cluster RCT in 4 urban and 4 rural Western Australian hospitals, enrolled 108 Aboriginal adults with stroke or traumatic brain injury (target 312). Cultural security training plus 6 months with an Aboriginal Brain Injury Coordinator showed no significant difference in adjusted mean 26-week EQ-VAS quality of life (p=0.83). Minimum processes of care favored the intervention by an adjusted difference in means of 6.8% (95% CI 0.40%–13.26%) at 26 weeks (PMID 41131785)	Healing Right Way tests post-injury cultural security and care coordination, and its primary outcome is a standard quality-of-life scale. On 2026-10-07 a PubMed search for Indigenous, Aboriginal, or First Nations stroke records that also mentioned thrombectomy, stroke unit, prehospital care, or systems of care, plus a trial term, returned 0 records. A broader PubMed search (those population terms AND stroke AND governance, community-led, or trial) returned 20 records, including Healing Right Way, the incidence review, and a World Stroke Organization scientific statement recommending community-led action (PMID 40415682). That statement does not count acute stroke-system trials by Indigenous governance. The original claim that such trials rarely use Indigenous governance had no denominator and is not restated. Cochrane CENTRAL, SciELO, and subscription databases were not searched	OQ-16

Maintenance rules¶

Retire a question only when evidence resolves the decision, not when one trial reports.
Record neutral and harmful results as answers.
Preserve the stable ID and add a resolution note when closing.
Re-query ClinicalTrials.gov before describing any NCT status.
Re-run absence searches before converting a “missing junction” into a claimed literature gap.