Jonathan M. Coutinho

BACKGROUND: Andexanet reduces hematoma expansion compared with usual care for acute FXa (factor Xa) inhibitor-associated intracerebral hemorrhage but is associated with a risk of thrombotic events (TE). We pooled data from andexanet-treated trial participants to determine baseline predictors of TE. METHODS: Secondary analyses of andexanet-treated participants in ANNEXA-I (Andexanet Alfa, a Novel Antidote to the Anticoagulant Effects of Factor Xa Inhibitors in ICH) and ANNEXA-4 (Andexanet Alfa, a Novel Antidote to the Anticoagulation Effects of Factor Xa Inhibitors), which included adults with acute life-threatening hemorrhage on FXa inhibitors. Stratified Cox proportional hazards models were used to examine associations between baseline variables and arterial or all TE and mortality. A stratified Cox regression model with arterial events as time-dependent covariates was used to examine associations of arterial events with mortality, adjusted for clinical covariates. RESULTS: Of 741 participants (mean age, 78.5 years; 54.5% male), 593 (80.0%) had intracranial hemorrhage and 109 (14.7%) had gastrointestinal hemorrhage. Seventy-seven (10.4%) participants had ≥1 TE, and 147 (19.8%) died. Arterial and venous thromboembolism events occurred a median of 3.5 days (interquartile range, 2.0–12.0) and 14.0 days (interquartile range, 10.0–20.0), respectively, after randomization. No baseline factors were associated with all TE. Arterial TE were associated with prior myocardial infarction (hazard ratio [HR], 2.64 [95% CI, 1.33–5.26]), high-dose eligibility (HR, 2.44 [95% CI, 1.33–4.48]), female sex (HR, 1.78 [95% CI, 1.00–3.16]), and age (HR per year, 1.04 [95% CI, 1.00–1.08]). Mortality was associated with arterial TE (HR, 2.68 [95% CI, 1.64–4.36]), prior congestive heart failure (HR, 2.53 [95% CI, 1.76–3.63]), intracerebral hemorrhage or intraventricular hemorrhage (HR, 1.90 [95% CI, 1.20–3.01]), and age (HR per year, 1.06 [95% CI, 1.03–1.08]). The initiation of anticoagulation after randomization was associated with a reduction in TE. CONCLUSIONS: In FXa inhibitor-associated hemorrhage treated with andexanet, baseline factors are associated with arterial TE and can be used to inform treatment decisions. Initiation of anticoagulation reduces the occurrence of TE. REGISTRATION: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT02313909. URL: https://www.clinicaltrials.gov ; Unique identifier: NCT02329327.

Introduction: Electroencephalography (EEG) has shown potential as a prehospital triage instrument for anterior circulation large vessel occlusion (LVO-a) stroke. In this study, we determined the diagnostic accuracy of EEG-based machine learning algorithms for LVO-a stroke detection. Patients and methods: Anterior circulation large vessel occlusion stroke detection machine learning models were developed using EEG data from the ELECTRA-STROKE (NCT03699397) and AI-STROKE (NCT05437237) studies. Single 2- to 3-min EEG recordings were obtained using 8 dry electrodes in adults with suspected stroke in prehospital and in-hospital settings. Patients with a clean EEG segment of ≥ 5 s (visually selected) were included and EEG symmetry features were extracted. We assessed the performance of multiple machine learning models using nested cross-validation (CV), optimising models with 3-fold CV (inner loop) and evaluating them with leave-one-out CV (outer loop). Results: Of 1293 patients with suspected stroke, 336 (26%) were included (74% prehospital, 26% in-hospital). Median age was 72 (IQR 61–80) years, and 231 (69%) were male. Sixteen (5%) patients had LVO-a stroke, 140 (42%) non-LVO-a ischaemic stroke, 52 (15%) TIA, 13 (4%) haemorrhagic stroke and 115 (34%) a stroke mimic. Naïve Bayes achieved the highest observed area under the receiver operating characteristic curve (0.85 [95% CI, 0.77–0.93]), with a sensitivity of 69% (95% CI, 44%–86%) and specificity of 86% (95% CI, 82%–89%). Discussion and conclusion: Machine learning models for LVO-a stroke detection based solely on EEG showed moderately high diagnostic accuracy. These results support the potential of EEG-based LVO-a stroke detection, but improved EEG data acquisition is needed before clinical implementation can be considered.

Background: Residual shunt is a known complication in percutaneous patent foramen ovale (PFO) closure, and moderate to large residual shunt has been associated with an increased risk of recurrent stroke. Device‐related factors, such as device size, have been associated with residual shunt, but comparative data per device type are scarce. Aims: We compared the prevalence of moderate to large residual shunt at 6 months between the Amplatzer and Occlutech devices. Methods: In this retrospective cohort study, patients undergoing successful percutaneous PFO closure with an Amplatzer or Occlutech dedicated closure device within the CAHAL centers Amsterdam UMC and Leiden UMC between 2018 and 2024 were included. The study population was divided into patients treated with an Occlutech and an Amplatzer device. The prevalence of moderate/large residual shunt at 6 months of follow‐up was compared between the two groups. As a sensitivity analysis, a logistic regression was conducted to determine the association between the treatment groups and moderate/large residual shunt. The logistic regression was conducted in a weighted population using inverse probability weighting (IPTW), correcting for age, sex, smoking status, shunt size at baseline, device size, and atrial septal aneurysm. Results: A total of 413 patients were included with an average age of 45.8 ± 11.3 years and 51.6% were male. Of the included patients, 279 were treated with an Amplatzer‐ and 134 with an Occlutech device. In the group treated with an Amplatzer, 22.2% of the patients had a moderate or large residual shunt at 6 months compared to 14.9% in the Occlutech group ( p = 0.11). The sensitivity analysis did not show an association between device type and moderate/large shunt OR 0.71 (95% CI: 0.39–1.28, p = 0.25). Conclusion: In this study, no significant difference was observed in moderate/large residual shunt prevalence at 6 months of follow‐up between the Occlutech and Amplatzer devices.

BACKGROUND: Periprocedural flat-panel computed tomography (FPCT) may complement real-time decision-making during endovascular treatment for stroke. This study aims to summarize current evidence on diagnostic findings, prognostic associations, and clinical applications of periprocedural FPCT in the angiography suite. METHODS: A systematic review and meta-analysis was preregistered (URL: https://www.crd.york.ac.uk/PROSPERO/ ; Unique identifier: CRD420251142156), conducted and reported according to MOOSE guidelines (Meta-Analysis of Observational Studies in Epidemiology). Medline, Embase, and Scopus were searched for studies published between January 2019 and September 2025 reporting on periprocedural FPCT in endovascular treatment patients (noncontrast FPCT, 3-dimensional–digital subtraction angiography/contrast-enhanced FPCT, or FPCT perfusion). Random-effects models were used to calculate pooled prevalences and odds ratios for associations with outcomes, primarily functional status assessed via the 90-day modified Rankin Scale score. RESULTS: Forty studies comprising 4139 patients were included. For noncontrast FPCT, 3 distinct imaging biomarkers were identified. Intraparenchymal hyperattenuations were frequent (pooled prevalence 49% [95% CI, 42%–56%]) and associated with any intracranial hemorrhage (odds ratio, 10.4 [95% CI, 4.5–24.1]), parenchymal hematoma (odds ratio, 4.3 [95% CI, 2.2–8.7]) and lower rates of good outcomes (eg, 90-day modified Rankin Scale score 0–2; odds ratio, 0.35 [95% CI, 0.24–0.50]). Subarachnoid hyperattenuations occurred in 23% (95% CI, 14%–35%) of patients and were associated with distal occlusions, markers of procedural aggressiveness, and lower rates of good outcome (eg, 90-day modified Rankin Scale score 0–2; odds ratio, 0.69 [95% CI, 0.53–0.91]). Intravascular hyperattenuations (distal occlusion tracker sign; pooled prevalence 35%, [95% CI, 24%–49%]) showed high specificity (93%, 95% CI, 87%–96%) for residual distal occlusions after endovascular treatment. Advanced vessel imaging improved thrombus conspicuity and technical efficacy, while postoperative flat-panel computed tomography perfusion shows added value for the detection of residual hypoperfusion, diagnosis of hemodynamic alterations, and prediction of final infarct. There was evidence for selection bias, suggesting slight overreporting of the above imaging signs. CONCLUSIONS: Periprocedural FPCT may offer real-time neurovascular and parenchymal assessment when conventional 3-dimensional imaging is unavailable. In our cohort, FPCT findings were associated with clinical outcomes and improved recognition of complications. FPCT may hence assist intraprocedural decision-making regarding adjunctive manoeuvres or drug administrations if further prospective validation can be demonstrated.

BACKGROUND: Cerebral venous thrombosis is an uncommon cause of stroke that primarily affects young adults. Several clinical trials and observational studies have compared vitamin K antagonists (VKAs) with direct oral anticoagulants (DOACs) for treating cerebral venous thrombosis, but the small sample sizes limit the ability to draw firm conclusions. METHODS: We performed a systematic review and meta-analysis of randomized controlled trials and observational studies comparing DOACs with VKAs for the treatment of cerebral venous thrombosis. Studies published between January 2007 and July 2025 were included. Prespecified subgroup analyses were conducted by study design, DOAC class (anti–factor Xa inhibitors versus direct thrombin inhibitors), and VKA. Risk of bias was assessed using the Cochrane Risk-of-Bias Tool 2 for randomized trials and the Risk of Bias in Nonrandomized Studies of Interventions for observational studies. Pooled risk ratios (RRs) with 95% CIs were calculated using random-effects models. RESULTS: Fourteen studies (8 randomized clinical trials [RCTs] and 6 observational studies) were included. Among 3698 patients, 1669 (45.1%) received DOACs and 2029 (54.9%) received VKAs. Rates of recurrent venous thromboembolism (10.24% versus 10.99%), major bleeding (4.26% versus 6.18%), clinically relevant nonmajor bleeding (2.63% versus 1.66%), good functional outcomes (95.09% versus 89.25%), and mortality (3.37% versus 3.30%) were similar in DOACs and VKAs. Recurrent venous thromboembolism was similar between DOACs and VKAs in RCTs (RR, 1.06 [95% CI, 0.35–3.24]) and in the overall pooled analysis of all studies (RR, 0.92 [95% CI, 0.55–1.53]). Major bleeding did not differ between DOACs and VKAs in RCTs (RR, 0.73 [95% CI, 0.24–2.22]) or in the overall analysis (RR, 0.74 [95% CI, 0.44–1.25]; P =0.26). Clinically relevant nonmajor bleeding was also comparable in RCTs (RR, 1.06 [95% CI, 0.25–4.48]) and overall (RR, 1.22 [95% CI, 0.38–3.86]). Good functional outcomes (modified Rankin Scale score of 0–2) were similar between DOACs and VKAs in RCTs (RR, 1.00 [95% CI, 0.84–1.19]) and in the combined analysis of all studies (RR, 1.02 [95% CI, 0.93–1.11]). Mortality was similar between DOACs and VKAs in RCTs (RR, 1.15 [95% CI, 0.28–4.68]) and in the pooled analysis. When direct thrombin inhibitors were compared with factor Xa inhibitors, there were no significant differences in outcomes. CONCLUSIONS: DOACs demonstrated comparable efficacy and safety to VKAs in the treatment of cerebral venous thrombosis. These findings support the use of DOACs as a reasonable alternative.

Background: Immune‐mediated inflammatory responses exacerbate brain injury and affect the prognosis in acute ischemic stroke. Methods: We characterized longitudinal complement activation after acute ischemic stroke using serial plasma samples from 10 patients and 8 controls. Initial markers of classical (C1q), lectin (mannose‐binding lectin), and alternative (complement factor Bb) pathways, as well as regulatory (complement factor H) and downstream markers (C3a, C5a, and C5b‐9) were measured by ELISA at 9 time points. Results: Patients with stroke showed broad complement activation, with increased C1q, mannose‐binding lectin, complement factor Bb, C3a, C5a, and C5b‐9, and reduced complement factor H compared with controls. Complement components showed distinct temporal patterns during follow‐up. Complement factor Bb levels were lower in patients with unfavorable functional outcome and in those with moderate‐to‐severe stroke. C1q and C3a trajectories differed according to functional outcome, whereas C3a trajectories also differed according to stroke severity. C1q trajectories differed according to age group and infarct volume. Conclusions: These findings suggest dynamic, pathway‐specific complement activation after acute ischemic stroke and highlight the need for larger studies to clarify its prognostic and therapeutic relevance.

BACKGROUND: Vitamin K antagonists have traditionally been used after cerebral venous thrombosis, whereas direct oral anticoagulants (DOACs) are increasingly prescribed. Whether DOACs are noninferior to warfarin for preventing 1-year symptomatic recurrent venous thromboembolism (VTE) remains uncertain. METHODS: We conducted a registry-based target trial emulation in a bidirectional multicenter Chinese cerebral venous thrombosis registry (retrospective phase, 2015–2022; prospective phase, 2023–2025). Eligible patients were adults with imaging-confirmed cerebral venous thrombosis who initiated either a DOAC or warfarin within 30 days after diagnosis. Time zero was the date of the first oral anticoagulant prescription. The primary efficacy outcome was symptomatic recurrent VTE within 1 year. Propensity score overlap weighting was used, and weighted Fine-Gray models treated death as a competing event. The clinically acceptable noninferiority margin was defined as an absolute increase of 1.5 percentage points in 1-year recurrent VTE risk, corresponding approximately to a relative margin of 1.50 for the primary Fine-Gray analysis. RESULTS: Among 2476 eligible patients, 942 initiated a DOAC (rivaroxaban, n=701; dabigatran, n=121; apixaban, n=113; edoxaban, n=7) and 1534 initiated warfarin. Within 1 year, symptomatic recurrent VTE occurred in 36 and 73 patients, respectively. The overlap-weighted 1-year recurrence risks were 3.74% and 5.47%, corresponding to a risk difference of −1.73 percentage points (95% CI, −3.76 to 0.30). DOAC treatment was associated with a lower subdistribution hazard ratio of 0.67 compared with that of warfarin ([95% CI, 0.37–1.21]; P =0.184), meeting the prespecified noninferiority criterion ( P for noninferiority=0.004). Major bleeding was rare within 1 year (0 versus 1 patient). Clinical nonmajor bleeding occurred in 18 versus 34 patients, and any bleeding occurred in 48 versus 75 patients. CONCLUSIONS: Among patients with cerebral venous thrombosis selected for oral anticoagulation, initial treatment with a DOAC met the noninferiority criterion relative to warfarin for the prevention of symptomatic recurrent VTE at 1 year. Findings regarding bleeding outcomes were exploratory and do not permit conclusions regarding comparative safety or equivalence.

BACKGROUND: OCEANIC-STROKE (XXX) showed that asundexian added to antiplatelet therapy reduced ischemic stroke without significantly increasing bleeding in patients with noncardioembolic ischemic stroke or high-risk transient ischemic attack. We characterized bleeding categories, sites, and key subgroups. METHODS: In the safety population, 12 254 participants received at least 1 dose of asundexian (n=6124) or placebo (n=6130). Bleeding was categorized by International Society on Thrombosis and Haemostasis criteria as major, clinically relevant nonmajor (CRNM), or minor. While-on-treatment analyses included events through 2 days after the last dose, with death and permanent discontinuation as competing risks. RESULTS: Major bleeding occurred in 117 (1.9%) participants with asundexian and 107 (1.7%) with placebo (cause-specific hazard ratio [csHR], 1.10 [95% CI, 0.85–1.44]; P =0.46). Major or CRNM bleeding occurred in 339 (5.5%) and 307 (5.0%), respectively (csHR, 1.12 [95% CI, 0.96–1.30]; P =0.16), and any bleeding in 763 (12.5%) and 752 (12.3%) (csHR, 1.02 [95% CI, 0.93–1.13]; P =0.65). Treatment effects for major and major or CRNM bleeding were consistent across prespecified subgroups, including planned dual antiplatelet therapy. Symptomatic intracranial hemorrhage occurred in 41 (0.7%) and 36 (0.6%) participants. Any gastrointestinal bleeding occurred in 186 (3.0%) and 147 (2.4%) (csHR, 1.28 [95% CI, 1.03–1.59]), and lower gastrointestinal bleeding in 102 (1.7%) and 59 (1.0%) (csHR, 1.75 [95% CI, 1.27–2.41]), primarily rectal bleeding. Any procedure-related bleeding and any mucosal bleeding did not differ significantly. CONCLUSIONS: Asundexian added to antiplatelet therapy reduced ischemic stroke without significantly increasing major, major or CRNM, or any bleeding. Lower gastrointestinal bleeding was more frequent with asundexian, whereas treatment effects on major and major or CRNM bleeding were consistent across prespecified subgroups. REGISTRATION: URL: https://www.clinicaltrials.gov ; Unique Identifier: NCT05686070.

Rationale: Effective prehospital stroke triage enables faster treatment and better patient outcomes. Although prehospital triage innovations are rapidly emerging, evidence regarding the balance between costs and benefits as well as implementation issues remains limited. Aims and Objectives: This review aims to synthesise published evidence on health economics and implementation determinants of prehospital triage tools for acute suspected ischaemic stroke patients. Method: A systematic search was conducted across five databases (CINAHL, Cochrane, EMBASE, PubMed, and Web of Science) between 1995 and 2025. Results were reported narratively, and reporting quality was assessed using CHEERS, STROBE‐simulation, COREQ, CONSORT, SQUIRE, and GRAMMS. Included were empirical studies of EMS‐operated prehospital triage innovations for acute suspected stroke patients. The main outcome measures were health economic outcomes (e.g., ICER) and system‐level implementation determinants (CFIR constructs). Results: From 6491 screened studies, thirty‐one were included: thirteen studies included health economic outcomes and eighteen reported implementation determinants. Prehospital triage tools assessed were: MSU ( n = 14), telemedicine ( n = 13), clinimetrics ( n = 1), prehospital MRI ( n = 1), biomarkers ( n = 1), and an AI‐based microwave tool ( n = 1). Most studies (68%) were of high reporting quality. Positive results regarding cost‐effectiveness were reported ( n = 11), including cost‐savings ( n = 1), as well as implementation determinants ( n = 11). However, evidence on sustained implementation was limited. Barriers included innovation design, complexity, workflow compatibility, work and IT infrastructure, and costs. Conclusion: Included studies suggest most prehospital stroke triage tools are cost‐effective, but implementation remains largely context‐dependent. Barriers, including innovation complexity, role delineation, workflow fit, design, and costs, highlight the need for future studies to integrate health economics into an implementation science framework alongside clinical outcomes.

Prehospital stroke trials enable assessments of ultraearly administration of treatments and management strategies, but face unique challenges in their design, conduct, and analysis. We outline important considerations and recommendations for planning prehospital stroke trials, focusing on patient identification, consent models, randomization strategies, end point selection, and statistical frameworks. As this field evolves, multidisciplinary collaboration, pragmatic study design, and scalable technology integration are critical to ensuring scientific rigor and equitable access. This framework is intended to inform and support future investigators in designing effective, ethically sound, and operationally feasible prehospital stroke trials in conventional emergency medical services settings.