Articles

Endovascular thrombectomy for acute ischaemic stroke with established large infarct: multicentre, open-label, randomised trial

Martin Bendszus, Jens Fiehler, Fabien Subtil, Susanne Bonekamp, Anne Hege Aamodt, Blanca Fuentes, Elke R Gizewski, Michael D Hill, Antonin Krajina, Laurent Pierot, Claus Z Simonsen, Kamil Zeleňák, Rolf A Blauenfeldt, Bastian Cheng, Angélique Denis, Hannes Deutschmann, Franziska Dorn, Fabian Flottmann, Susanne Gellißen, Johannes C Gerber, Mayank Goyal, Jozef Haring, Christian Herweh, Silke Hopf-Jensen,

Vi Tuan Hua, Märit Jensen, Andreas Kastrup, Christiane Fee Keil, Andrej Klepanec, Egon Kurča, Ronni Mikkelsen, Markus Möhlenbruch, Stefan Müller-Hülsbeck, Nico Münnich, Paolo Pagano, Panagiotis Papanagiotou, Gabor C Petzold, Mirko Pham,Volker Puetz, Jan Raupach,

Gernot Reimann, Peter Arthur Ringleb, Maximilian Schell, Eckhard Schlemm, Silvia Schönenberger, Bjørn Tennøe, Christian Ulfert, Kateĭina Vališ, Eva Vítková, Dominik F Vollherbst, Wolfgang Wick, Götz Thomalla, on behalf of the TENSION Investigators*

Summary

Background Recent evidence suggests a beneficial effect of endovascular thrombectomy in acute ischaemic stroke with large infarct; however, previous trials have relied on multimodal brain imaging, whereas non-contrast CT is mostly used in clinical practice.

Methods In a prospective multicentre, open-label, randomised trial, patients with acute ischaemic stroke due to large vessel occlusion in the anterior circulation and a large established infarct indicated by an Alberta Stroke Program Early Computed Tomographic Score (ASPECTS) of 3–5 were randomly assigned using a central, web-based system (using a 1:1 ratio) to receive either endovascular thrombectomy with medical treatment or medical treatment (ie, standard of care) alone up to 12 h from stroke onset. The study was conducted in 40 hospitals in Europe and one site in Canada. The primary outcome was functional outcome across the entire range of the modified Rankin Scale at 90 days, assessed by investigators masked to treatment assignment. The primary analysis was done in the intention- to-treat population. Safety endpoints included mortality and rates of symptomatic intracranial haemorrhage and were analysed in the safety population, which included all patients based on the treatment they received. This trial is registered with ClinicalTrials.gov, NCT03094715.

Findings From July 17, 2018, to Feb 21, 2023, 253 patients were randomly assigned, with 125 patients assigned to endovascular thrombectomy and 128 to medical treatment alone. The trial was stopped early for efficacy after the first pre-planned interim analysis. At 90 days, endovascular thrombectomy was associated with a shift in the distribution of scores on the modified Rankin Scale towards better outcome (adjusted common OR 2·58 [95% CI 1·60–4·15]; p=0·0001) and with lower mortality (hazard ratio 0·67 [95% CI 0·46–0·98]; p=0·038). Symptomatic intracranial haemorrhage occurred in seven (6%) patients with thrombectomy and in six (5%) with medical treatment alone.

Interpretation Endovascular thrombectomy was associated with improved functional outcome and lower mortality in patients with acute ischaemic stroke from large vessel occlusion with established large infarct in a setting using non- contrast CT as the predominant imaging modality for patient selection.

Funding EU Horizon 2020.

Copyright © 2023 Elsevier Ltd. All rights reserved.

Lancet 2023; 402: 1753–63

Published Online October 11, 2023 https://doi.org/10.1016/ S0140-6736(23)02032-9

See Comment page 1724

*The TENSION Investigators are listed in the appendix

Neuroradiologie (Prof M Bendszus MD, C Herweh MD,

S Bonekamp DVM,

Prof M Möhlenbruch MD,

C Ulfert MD, D F Vollherbst MD), and Neurologie

(Prof P A Ringleb MD,

Prof S Schönenberger MD, Prof W Wick MD), Universitätsklinikum Heidelberg, Heidelberg,

Germany; Klinik und Poliklinik für Neuroradiologische Diagnostik und Intervention (Prof J Fiehler MD,

F Flottmann MD, S Gellißen MD), and Klinik und Poliklinik für Neurologie (B Cheng MD,

M Jensen MD, M Schell MD, E Schlemm MD,

Prof G Thomalla MD), Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany; eppdata GmbH, Hamburg, Germany (Prof J Fiehler); Service de Biostatistique, Hospices Civils

Introduction

Endovascular thrombectomy in patients with acute ischaemic stroke due to large vessel occlusion is safe and improves the functional outcome when compared with medical treatment alone.1–7 In most studies, this effect has been demonstrated in patients with minimal or moderate infarct size on imaging before endo­ vascular thrombectomy. Therefore, current guidelines recommend endovascular thrombectomy in patients with an Alberta Stroke Program Early Computed Tomographic Score (ASPECTS; range from 0–10, where smaller values indicate a larger area of infarction) of at least 6.8,9 Patients with larger brain infarcts are frequently excluded from endovascular thrombectomy, even though they constitute up to 25% of ischaemic strokes due to large vessel occlusion in routine clinical practice.10

Three recent trials and a meta­analysis, including patients with a large infarct, suggested a benefit in functional outcome after endovascular thrombectomy compared with medical treatment alone.11–14 Patient inclusion in these trials was based on MRI or volumetry of the infarct core using perfusion CT, either entirely or in substantial patient subgroups. These trials either represented selected patient populations in Asia or included patients based on criteria

de Lyon, Lyon, France

(F Subtil PhD, A Denis MSc); Laboratoire de Biométrie et Biologie Évolutive, Université de Lyon, Villeurbanne, France (F Subtil, A Denis); Department of Neurology (A H Aamodt MD), and Department of Neuroradiology (B Tennøe MD), Oslo University Hospital, Oslo, Norway; Department of Neurology and Stroke Center, Hospital La Paz Institute for Health Research—

La Paz University

Hospital-Universidad Autonoma de Madrid, Madrid, Spain (Prof B Fuentes MD);

Department of Neuroradiology, Medical University Innsbruck,

Innsbruck, Austria (Prof E R Gizewski MD); Department of Clinical Neurosciences, Hotchkiss Brain Institute, Health Science Centre, University of Calgary & Foothills Medical Centre,

Calgary, AB, Canada

(Prof M D Hill MD, Prof M Goyal PhD); Department of Radiology (Prof A Krajina MD, Prof J Raupach MD), and Department of Neurology

(E Vítková MD), Faculty of Medicine in Hradec Kralove,

Charles University, Czech Republic; Department of

Neuroradiology

(Prof L Pierot MD PhD, P Pagano MD), and Department of Neurology (V T Hua MD), Hôpital Maison-Blanche, Université Reims-Champagne- Ardenne, Reims, France;

Department of Neurology (Prof C Z Simonsen MD, R A Blauenfeldt MD), and

Department of Neuroradiology (R Mikkelsen MD), Aarhus University Hospital, Aarhus, Denmark; Clinic of Radiology (Prof K Zeleňák MD), and Clinic of Neurology (Prof E Kurča MD), Jessenius Faculty of Medicine, Comenius University, Martin,

Slovakia; Division of Neuroradiology, Vascular and Interventional Radiology, Department of Radiology, Medical University Graz,

Graz, Austria (Prof H Deutschmann MD); Klinik für Diagnostische und

Interventionelle Neuroradiologie, Universitätsklinikum Bonn,

Bonn, Germany (Prof F Dorn MD); Institute of Neuroradiology (J C Gerber MD),

Department of Neurology (Prof V Puetz MD), and Dresden Neurovascular Center

(J C Gerber, Prof V Puetz), Universitätsklinikum Carl Gustav Carus an der Technischen Universität Dresden, Dresden, Germany;

Department of Neurology

(J Haring MD), and Department of Radiology (Prof A Klepanec MD), Faculty Hospital Trnava, Trnava, Slovakia; Institut für Diagnostische und Interventionelle Radiologie

Research in context Evidence before this study

We searched PubMed for randomised trials using the search terms “randomised trial” OR “randomized trial” AND “thrombectomy” AND “large core” OR “large infarct” OR “low ASPECTS” for randomised clinical trials published between database inception and Aug 31, 2023. We identified three trials of endovascular thrombectomy versus medical treatment in adults with acute ischaemic stroke and large vessel occlusion presenting with extended infarct. The RESCUE-Japan LIMIT trial randomly assigned 203 patients in Japan with acute ischaemic stroke and large vessel occlusion and a large ischaemic region, reflected by an Alberta Stroke Program Early Computed Tomographic Score (ASPECTS) of 3–5, to endovascular thrombectomy or medical management alone. Most patients (86%) were enrolled based on MRI with assessment of the ischaemic region by diffusion-weighted imaging (DWI).

Patients assigned to endovascular thrombectomy had better functional outcomes than those assigned to medical management alone. The ANGEL-ASPECT trial randomly assigned 456 patients in China with stroke from large vessel occlusion and large infarct, defined by ASPECTS 3–5, or infarct core-volume of 70–100 mL, based on perfusion imaging to endovascular thrombectomy or medical management alone, and observed better functional outcome with endovascular thrombectomy. The international SELECT2 trial randomly assigned 352 patients with acute ischaemic stroke and large vessel occlusion who had a large ischaemic core, defined by ASPECTS 3–5, or core volume of at least 50 mL on perfusion CT or DWI and reported better functional outcomes with endovascular thrombectomy than with medical management alone. In all three trials, patient enrolment was either entirely or in substantial patient subgroups based on MRI or perfusion CT involving commercial software for post-processing. Patient

related to advanced imaging and post­processing techniques, which limits generalisability to other populations. However, management of acute stroke in clinical practice worldwide frequently relies on visual evaluation of early ischaemic signs predominantly on unenhanced CT, supplemented by CT angiography (CTA) to determine the site of vessel occlusion.15,16

In a randomised trial including patients with a large infarct on admission, we aimed to assess whether endovascular thrombectomy improves functional outcome in acute stroke due to large vessel occlusion. Patient inclusion was in an extended time window of 12 h and based on clinical standard imaging with exclusively visual image assessment using either plain CT or MRI depending on local standards.

Methods

Study design and participants

TENSION (The Efficacy and Safety of Thrombectomy in Stroke with extended lesion and extended time window)

management for acute stroke in clinical practice worldwide mostly relies on visual evaluation of early ischaemic signs, predominantly on unenhanced CT, supplemented by CT angiography to determine the site of vessel occlusion.

Moreover, two of the trials were exclusively done in Japan and China. Thus, the generalisability of the findings from these three trials to other populations is limited.

Added value of this study

TENSION was the first clinical trial to randomly assign patients with acute ischaemic stroke due to large vessel occlusion in the anterior circulation and a large infarct without the use of extended imaging and based on standard-of-care stroke imaging, which was non-contrast CT in 82% of patients and MRI in 18%. Patients were randomly assigned to receive either endovascular thrombectomy or medical treatment (standard of care) alone up to 12 h after stroke onset. At 90 days, there was a shift in the distribution of scores on the modified Rankin Score towards better outcomes in favour of endovascular thrombectomy.

Mortality was lower with endovascular thrombectomy, which also represents a novel finding not observed in the previous trials of thrombectomy for stroke patients with large core. There were no safety concerns with endovascular thrombectomy.

Implications of all the available evidence

Taken together, the published trials provide strong evidence for the benefit of endovascular thrombectomy in patients with acute ischaemic stroke from large vessel occlusion, who present with a large ischaemic region. Patients with acute ischaemic stroke with large infarct achieved a better functional outcome and had a higher probability of survival with endovascular thrombectomy than with medical treatment alone.

Endovascular thrombectomy in these patients could be guided by non-contrast CT, MRI, or perfusion imaging.

was an investigator­initiated, randomised, open­label, blinded endpoint, two­arm, post­market trial in Europe and Canada. The trial was conducted in 40 hospitals in Europe (four sites in Austria, four sites in Czech Republic, two sites in Denmark, four sites in France, 20 sites in Germany, three sites in Norway, two sites in Slovakia, and one site in Spain) and one site in Canada (appendix p 14). An academic steering committee designed and supervised the trial supported by an independent data monitoring and safety committee, ethics advisory board, and scientific advisory board. The protocol has been published and the trial was approved by the institutional review board of the University of Heidelberg and all participating sites.17 Informed consent by the patients or legal representatives was obtained before enrolment into the trial.

All patients underwent a standardised imaging protocol, including unenhanced CT and CTA or MRI including diffusion­weighted imaging (DWI) and magnetic resonance angiography (MRA), according to institutional preference. Patients were eligible if they

presented with acute ischaemic stroke due to focal occlusion in the M1 segment of the middle cerebral artery or the intracranial segment of the distal internal carotid artery (ICA) on CTA or MRA, and if they had an ASPECTS of 3–5 on unenhanced CT or DWI, as assessed locally. Patients were randomly assigned within 11 h of symptom onset or last known to be well, with expected completion of endovascular thrombectomy within 12 h. Eligible patients were aged 18 years or older, had a maximum US National Institutes of Health Stroke Scale (NIHSS) score of less than 26 (with scores ranging from 0 to 42 and higher scores indicating worse neurological deficits), and were premorbidly independent, which was defined as a historically estimated score of 0–2 on the modified Rankin Scale (mRS; scores range from 0 to 6, with 0 indicating no disability, 1 no clinically significant disability, 2 slight disability, 3 moderate disability but able to walk unassisted, 4 moderately severe disability,

5 severe disability, and 6 death). Imaging exclusion criteria were known vascular disease preventing endovascular thrombectomy or high­grade extracranial stenosis expected to require acute stent placement and any acute intracranial bleeding or mass effect. A full list of inclusion and exclusion criteria is provided in the protocol (appendix pp 39–154). Information on sex (male or female) was collected by self­report or medical records.

Randomisation and masking

Patients were randomly assigned to undergo endovascular thrombectomy in addition to medical treatment or to receive medical treatment (standard of care) alone. Randomisation was stratified by time from symptom onset or last known well (<6 h and 6–11 h) and stroke severity (NIHSS ≤18 and NIHSS 19–25) to assure randomised balance within these key prognostic variables. Patients were randomly assigned in a 1:1 ratio using a central, web­based module, with a permuted block design (random block sizes of two and four). All follow­up examinations were conducted by investigators or trial personnel masked to the patients’ group assignment (ie, personnel not involved in random assignment and treatment of patients and unaware of the randomisation results).

Procedures

The technique of endovascular thrombectomy (ie, stent retriever, aspiration, or both, with or without balloon protection) was left to the discretion of the treating physician. Medical management of patients in both groups was done according to national and international guidelines, including intravenous thrombolysis if indicated.8,9

Clinical assessment was done at baseline, at 24 (±6) h, at 7 days or hospital discharge, and at 90 (±14) days. Baseline disease characteristics included pre­stroke mRS, presenting symptoms, and stroke severity according to the NIHSS. Follow­up assessment additionally included

patient­reported outcome measures of health­related quality of life (EuroQol­5 Dimension [EQ­5D] and Patient­ Reported Outcomes Measurement Information System 10­item [PROMIS­10]) and anxiety and depression (Patient Health Questionnaire­4 [PHQ­4]). Examinations were conducted by trained, certified investigators masked to the patients’ group assignment. If in­person assessment at 90 days was not possible, a telephone interview was done to assess the mRS. All patients underwent a standardised imaging protocol at baseline, including unenhanced CT supplemented by CTA or MRI with DWI and MRA. Rating of ASPECTS for trial inclusion was determined at the respective site on unenhanced CT images or DWI. Before site initiation, all investigators did web­based ASPECTS training.18 All imaging data were transferred to a central core lab (Eppdata, Hamburg, Germany) and assessed masked to the group assignment.

Outcomes

The primary outcome was functional neurological disability scored on the mRS at 90 (±14) days. The primary endpoint was the difference across the entire range of the mRS at 90 (±14) days between groups in ordinal analysis (shift analysis) in the intention­to­treat population. Secondary outcomes included independent (mRS ≤2) and moderate (mRS ≤3) functional outcome, functional health status and quality of life evaluated by the EQ­5D and PROMIS­10 questionnaires, post­stroke anxiety and depression evaluated by the PHQ­4 questionnaire, all assessed at 90 (±14) days after stroke, and rates of decompressive craniectomy, assessed during in­hospital treatment.

Safety endpoints comprised death or dependency 90 (±14) days after stroke (mRS 4–6), rates of symptomatic intracranial haemorrhage and parenchymal haemorrhage type 2 at 24 (±6) h,19 frequencies of adverse and serious adverse events, mortality rates at discharge and 90 (±14) days post­treatment, stroke­related death, rates of space­occupying infarction, and new ischaemic strokes. Symptomatic intracranial haemorrhage was centrally adjudicated and defined as an increase in the NIHSS score of 4 or more points or an increase in the score for an NIHSS subcategory of 2 or more points as compared with baseline or the lowest value before deterioration, with the presence of parenchymal haemorrhage type 2. Parenchymal haemorrhage type 2 was defined as an intracranial haemorrhage that involved more than 30% of the infarcted area with a substantial space­occupying effect or that was remote from the original infarcted area19 and was assessed by the image core lab. Imaging endpoints included core infarct volume (mL) at 24 (±6) h after randomisation. For the endovascular thrombectomy group, we also assessed the rate of embolisation, either distal to the target occlusion or to previously unaffected territories. Restoration of flow was centrally evaluated using the modified Thrombolysis in Cerebral Ischemia (mTICI) scale, a 6­point scale

und Neuroradiologie, DIAKO Krankenhaus gGmbH,

Flensburg, Germany

(S Hopf-Jensen MD,

Prof S Müller-Hülsbeck MD);

Klinik für Neurologie

(Prof A Kastrup MD), and Klinik für Diagnostische und Interventionelle Neuroradiologie

(Prof P Papanagiotou MD), Klinikum Bremen Mitte, Bremen, Germany; Institut für Neuroradiologie, Universitätsklinikum Frankfurt, Frankfurt am Main, Germany (C F Keil MD); Klinikum Dortmund gGmbH, Klinikum der Universität Witten/Herdecke, Dortmund, Germany (N Münnich MD,

G Reimann MD); Department of Radiology, Aretaieion University Hospital, National and Kapodistrian University of Athens, Athens, Greece

(Prof P Papanagiotou); Vascular Neurology Research Group, German Center for Neurodegenerative Diseases (DZNE), Bonn, Germany

(Prof G C Petzold MD); Division of Vascular Neurology, Department of Neurology, University Hospital Bonn,

Bonn, Germany

(Prof G C Petzold); Institut für Diagnostische und Interventionelle Neuroradiologie, Universitätsklinikum Würzburg, Würzburg, Germany (Prof M Pham MD); St Anne’s University Hospital Brno, Brno, Czech Republic (K Vališ MD)

Correspondence to:

Prof Götz Thomalla, Klinik und Poliklinik für Neurologie, Universitätsklinikum Hamburg- Eppendorf, Hamburg 20246,

Germany

[email protected]

See Online for appendix

ranging from 0 to 3, on which higher scores indicate greater reperfusion and successful reperfusion was defined as mTICI 2b or better.

Statistical analysis

For the sample size calculation, simulations were done using an assumption for the possible true distribution of mRS from the literature18 and a proportional odds alternative with an odds ratio of 1·5 to be assessed using the primary endpoint mRS shift analysis. Under the assumption of these distributions, a total of 620 patients were required to achieve a power of 80% for a one­sided test at the 0·025 level. Assuming a 5% dropout rate of patients to assess the primary endpoint obtained 90 days after inclusion, an effective sample size of 665 was necessary. The study included two interim analyses after a third and two­thirds of the patients had completed the 90­day follow­up, for futility and early efficacy (see statistical analysis plan, appendix pp 155–220).

We did efficacy analyses in the intention­to­treat population and in the per­protocol population as sensitivity analysis for the primary endpoint. The intention­to­treat population included all patients who were randomly assigned to a trial group. The per­protocol population included all patients who received the assigned treatment and had no clinically meaningful deviations from the protocol, also excluding all patients in whom central evaluation of baseline images resulted in an ASPECTS value less than 3 or more than 5. We did

safety analyses in the safety population, which included all patients based on the treatment they received. In the safety analysis, the endovascular thrombectomy group included all patients in whom endovascular thrombectomy was initiated and groin puncture was attempted, including patients receiving endovascular thrombectomy outside the trial protocol. All patients in whom no endovascular thrombectomy was initiated were allocated to the medical treatment group.

For the primary efficacy outcome, we used a proportional­odds logistic regression model with adjustments for the randomisation stratification factors; mRS scores 0 and 1 were merged due to the very low numbers in the best medical treatment group. The proportional odds assumption was fulfilled, hence the effect size was quantified by the common odds ratio (OR; likelihood­ratio test) with the Wald 95% CI. Missing mRS values at 90 days were imputed (multiple imputation with five imputed datasets, combined using the Rubin’s method) based on a proportional­odds logistic regression model including baseline age, baseline NIHSS, treatment group, and NIHSS at hospital discharge, with the full conditional specification method, under the missing­at­random assumption, except for patients who died before 90 days, for whom a mRS score of 6 was assigned. In addition, we did sensitivity analyses including best­case and worst­case scenarios. We analysed the primary outcome in prespecified subgroups, by adding an interaction term

125 assigned to receive endovascular thrombectomy plus medical management in the intention-to-treat population 124 had information on mRS at 90 days
		28 excluded from per-protoc 27 failed to meet inclusio 1 randomly assigned 1 pre-stroke mRS >2 active participant infective endocardi

Figure 1: Randomisation and analyses

ol analyses n criteria

>11 h after stroke onset

in another trial tis

22 core lab ASPECTS value out of range

1 proximal ICA occlusion requiring stenting 1 met imaging exclusion criteria

1 imaging indicative of a high risk of SICH

128 assigned to receive medical management alone in the intention-to-treat population

121 had information on mRS at 90 days

253 patients randomly assigned

89 assigned to receive medical management alone and included in the per-protocol population

97 assigned to receive endovascular thrombectomy plus medical management and included in the per-protocol population

39 excluded from per-protocol analyses 37 failed to meet inclusion criteria

3 randomly assigned >11 h after stroke onset 1 NIHSS score >26

3 pre-stroke mRS >2

1 occlusion not deemed accessible to the thrombectomy device

28 core lab ASPECTS value out of range

1 core lab ASPECTS evaluation not possible due to missing images

3 protocol deviations

3 received thrombectomy outside the trial protocol

The intention-to-treat population included all the patients who were randomly assigned to a trial group. The per-protocol population included all the patients who had undergone randomisation, who had received treatment as assigned, and who had not been excluded because of a major protocol violation. ASPECTS values range from 0 to 10, with lower scores indicating larger infarction; NIHSS scores range from 0 to 42, with higher scores indicating greater neurological deficit; and mRS values range from 0 (no symptoms) to 6 (death). Numbers of patients excluded from the per-protocol analysis might not add up due to individual patients with multiple exclusion criteria. ASPECTS=Alberta Stroke Program Early Computed Tomographic Score. NIHSS=National Institutes of Health Stroke Scale. mRS=modified Rankin Scale. SICH=symptomatic intracranial haemorrhage.

Sex

Female Male

56 (45%)

59 (55%)

67 (52%)

51 (48%)

Median interval between symptom onset and groin puncture (IQR), h

4·2 (3·4–5·9)

··

Anaesthesia performed (%), n/N Conscious sedation or none

General anaesthesia

69/123 (56%)

54/123 (44%)

··

··

Median age (IQR), years

Endovascular thrombectomy (N=125)

73 (65–81)

Medical treatment (N=128)

74 (64–80)

Median NIHSS score at hospital arrival (IQR)*

19 (16–22)

18 (15–22)

Level of consciousness on hospital arrival (%), n/N Fully awake

Somnolent

Coma

76/123 (62%)

44/123 (36%)

3/123 (2%)

79/126 (63%)

45/126 (36%)

2/126 (2%)

Transfer to centre with endovascular thrombectomy capabilities 70 (56%)

76 (59%)

Median interval between symptom onset and randomisation for known onset strokes (IQR), h

2·0 (1·2–3·5)

2·1 (1·2–3·6)

Median interval between randomisation and recanalisation (IQR), h

2·4 (1·8–3·0)

··

(Table 1 continues on next page)

Symptom onset known 73 (58%) 71 (56%)

Median pre-stroke modified Rankin Scale (IQR)† 0 (0–1) 0 (0–1)

between the treatment group and the baseline characteristics into the main model.

We analysed differences in the secondary binary outcomes between the groups by a logistic regression model, adjusted for randomisation stratification factors, and quantified by OR with the associated 95% CIs, or Fisher tests. For the outcome of death within 90 days, a Cox proportional­hazards model with the same adjustments as previous models was used to estimate the hazard ratio (HR) with the 95% CI; proportionality for this analysis was confirmed. The infarct volume at 24 h was compared between groups with a log­linear model.

Some secondary outcomes are not available because calculations have not been completed (difference of infarct volume at 24 h from infarct volume as predicted by pretreatment imaging), or because not all patients have completed the 12­month follow­up period (functional neurological outcome assessed by the simplified mRS questionnaire, patient­reported functional health status and quality of life, and post­stroke depression and anxiety at 12 months after stroke). These will be reported in a follow­up publication.

Medical history (selected; %), n/N
Previous ischaemic stroke	11/115 (10%)	18/118 (15%)
Hypertension	99/123 (80%)	98/121 (81%)
Diabetes	27/119 (23%)	29/121 (24%)
Dyslipidaemia	43/115 (37%)	44/115 (38%)
Atrial fibrillation	31/117 (26%)	48/118 (41%)
Myocardial infarction	14/114 (12%)	17/117 (15%)
Heart failure	16/113 (14%)	16/114 (14%)
Coronary artery disease	34/116 (29%)	26/115 (23%)
Extracranial carotid artery disease	8/115 (7%)	7/113 (6%)
Occlusion site (%), n/N
Internal carotid artery	41/125 (33%)	37/127 (29%)
Middle cerebral artery, M1 segment‡	83/125 (66%)	88/127 (69%)
Middle cerebral artery, M2 segment‡	0/125 (0%)	1/127 (1%)
Middle cerebral artery plus anterior cerebral artery	1/125 (1%)	1/127 (1%)
Additional extracranial internal carotid artery occlusion (ie, tandem occlusion)	8 (6%)	7 (5%)
Imaging method used for enrolment
CT	104 (83%)	104 (81%)
MRI	21 (17%)	24 (19%)
ASPECTS value (local evaluation for randomisation)§
3	36 (29%)	48 (38%)
4	45 (36%)	39 (30%)
5	44 (35%)	41 (32%)
ASPECTS value (core lab evaluation)§ 0–2	15/125 (12%)	23/127 (18%)
3–5	103/125 (82%)	99/127 (78%)
6–10	7/125 (6%)	5/127 (4%)
Intravenous alteplase administered 49 (39%) 44 (34%)

Because the statistical analysis plan, available with the protocol, did not include a provision for correcting for the secondary outcomes or multiple comparisons in the subgroup analyses, the CIs should not be used for hypothesis testing. Statistical analyses were done using SAS (version 9.4). This trial was monitored by an independent data and safety monitoring board. This trial is registered at ClinicalTrials.gov, NCT03094715.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

From July 17, 2018, to Feb 21, 2023, 253 patients were randomly assigned. We stopped the trial early after the boundaries for efficacy were crossed at the first interim analysis on Feb 21, 2023, when 222 patients had reached the primary outcome (appendix p 17, 24). Follow­up was continued for all patients who had been randomly

assigned, and we report the results for the entire included population. Of 253 patients randomised, 125 (49%) were assigned to receive endovascular thrombectomy plus medical treatment, and 128 (51%) were assigned to receive medical treatment alone (figure 1). Three patients assigned to medical treatment alone received thrombectomy outside the trial protocol. Thus, 128 patients who received thrombectomy and 125 patients who received medical treatment alone were included in the safety population. For the per­protocol analysis, 28 patients from the thrombectomy group and 39 patients from the medical treatment group were excluded. Thus, 97 patients who received thrombectomy and 89 patients who received medical treatment alone were included in

Endovascular Medical treatment thrombectomy (N=128)

(N=125)

(Continued from previous page)

Type of device used for thrombectomy (%), n/N Aspiration catheter alone

Stent retriever alone

Both aspiration catheter and stent retriever Median attempts for thrombectomy (IQR), n

21/121 (17%)

28/121 (23%)

72/121 (60%)

3·0 (2·0–5·0)

··

··

··

··

Data are n (%) unless specified. NIHSS=National Institutes of Health Stroke Scale. *Scores on the National Institutes of Health Stroke Scale with scores ranging from 0 to 42 and higher scores indicating greater neurological deficit. †Scores on the modified Rankin scale range from 0 to 6, with higher scores indicating greater disability. ‡The M1 segment is the main trunk of the middle cerebral artery and the M2 segment is the first-order branch of the main trunk of the middle cerebral artery. §Alberta Stroke Program Early Computed Tomography Score (ASPECTS) values range from 0 to 10, with lower values indicating larger infarction.

Table 1: Demographic and clinical characteristics of participants at baseline and treatment

Score on the Modified Rankin Scale at 90 days

0–1 2 3 4 5 6

Endovascular treatment 8·1 8·9 14·5

19·4

12·1

37·1

Medical treatment 10·7 13·1

19·7

54·1

0·8

1·6

0 10 20 30 40 50 60 70 80 90 100

Proportion of patients (%)

Figure 2: Distribution of Modified Rankin Scale scores at 90 days

Intention-to-treat population analysis. A score of 0 on the modified Rankin Scale indicates no symptoms, a score of 1 indicates no clinically significant disability, a score of 2 indicates slight disability, a score of 3 indicates

moderate disability, a score of 4 indicates moderately severe disability, a score of 5 indicates severe disability, and a score of 6 indicates death. Information on the primary outcome measure was missing in one patient in the endovascular thrombectomy group and six patients in the medical treatment group. Missing values were imputed.

the per­protocol population. Data on the primary outcome were missing for seven patients (one in the thrombectomy group and six in the medical treatment group). Among survivors, information on follow­up at 90 days was collected by in­person assessment in 28 (36%) of 78 patients assigned to endovascular thrombectomy and 13 (22%) of 58 patients assigned to medical treatment and by telephone interview in 50 (64%) of 78 patients assigned to endovascular thrombectomy and 45 (78%) of 58 patients assigned to medical treatment.

The main baseline demographic and clinical characteristics of the patients were similar in the trial groups (table 1). Across both groups, the median age of patients was 74 (IQR 65–80) years, and 123 (49%) of 253 were women. The median NIHSS score at baseline was 19 (IQR 16–22) for patients in the endovascular thrombectomy group and 18 (15–22) in the medical treatment group. Baseline ASPECTS was assessed based on non­contrast CT in 104 (83%) of 125 patients in the

endovascular thrombectomy group and 104 (81%) of 128 patients in the medical treatment group. There was an imbalance in baseline ASPECTS, as assessed by the image core lab with a larger proportion of patients in the medical treatment group presenting with a low ASPECTS of 0–2 (23 [18%] of 127 patients in the medical treatment only group vs 15 [12%] of 125 patients in the endovascular thrombectomy group).

CT was used as an imaging modality for enrolment in

208 (82%) of 253 patients. In the endovascular thrombectomy group, vessel occlusion was located in the ICA in 41 (33%) of 125 patients and in the M1 segment of the middle cerebral artery (MCA) in 84 (67%) patients. In the medical treatment group, vessel occlusion was present in the ICA in 37 (29%) of 127 patients, in the M1 segment of the MCA in 89 (70%) patients, and in the M2 segment of the MCA in one patient (1%). In the endovascular thrombectomy group, 36 (29%) of

125 patients had an ASPECTS of 3, 45 (36%) had ASPECTS 4, and 44 (35%) had ASPECTS 5. Distribution in the medical treatment group was 48 (38%) with ASPECTS 3, 39 (30%) with ASPECTS 4, and 41 (32%) with ASPECTS 5. Intravenous alteplase was given in

49 (39%) patients in the endovascular thrombectomy group and 44 (34%) patients in the medical treatment group.

In the primary outcome analysis including all 253 patients who had been randomly assigned, there was a shift in the distribution of scores on the mRS at 90 days towards better outcomes in favour of endovascular thrombectomy versus medical treatment alone (adjusted common OR 2·58 [95% CI 1·60–4·15]; p=0·0001; figure 2, table 2, appendix p 25). At the time of the interim analysis including 222 patients, there was a shift in the distribution of scores on the mRS at 90 days towards better outcomes with endovascular thrombectomy (OR 3·05 [2·5% lower CI 1·82]; p<0·0001; appendix p 24).

In the secondary outcome analysis, 21 (17%) of 124 patients in the endovascular thrombectomy group achieved a score of 0–2 on the mRS at 90 days as compared with three (2%) of 122 patients in the medical treatment group (adjusted OR 7·16 [95% CI 2·12–24·21; p=0·0016; appendix p 31). 39 (31%) of 124 patients in the endovascular thrombectomy group had a score of 0–3 on the mRS at 90 days versus 16 (13%) of 122 in the medical treatment group (adjusted OR 2·84 [95% CI 1·48–5·47]; p=0·0018; appendix p 31).

Analysis of patient­reported outcomes revealed higher PROMIS­10 values for physical health (median T score 39·8 [IQR 34·9–50·8] vs 34·9 [29·6–37·4]; p<0·0008) and mental health (median T score 43·5 [IQR 36·3–50·8] vs 38·8 [32·6–43·5]; p=0·025) at 90 days in patients assigned to endovascular thrombectomy as compared with those assigned to medical treatment.

In the safety population, death at 90 days occurred in

49 (40%) of 122 patients receiving endovascular thrombectomy and 63 (51%) of 123 patients receiving

Primary outcome

Endovascular thrombectomy (N=125)

Medical treatment (N=128)

Treatment effect (95% CI)

p value

Median score on modified Rankin Scale at 90 (±14; IQR) days (%)* 4 (3–6) 6 (4–6) 2·58 (1·60–4·15) 0·0001

Secondary outcomes

Independent neurological outcome (mRS ≤2) at 90 (±14) days (%), n/N 21/124 (17%) 3/122 (2%) 7·16 (2·12–24·21) 0·0016 Moderate neurological outcome (mRS ≤3) with 90 (±14) days (%), n/N 39/124 (31%) 16/122 (13%) 2·84 (1·48–5·47) 0·0018 Decompressive craniotomy 11 (9%) 9 (7%) ·· 0·65

Successful recanalisation (mTICI 2b or better)† 104 (83%) ·· 83% (76–89) ··

Distal embolisation‡ 4 (3%) ·· 3% (1–8) ·· Mean infarct volume at 24 (±6) h (CT or MRI) after randomisation (SD), mL§ 205·8 (139·1) 227·7 (107·2) 0·93 (0·80–1·08) 0·35 Median EQ-5D index at 90 (±14) days (IQR) 0·6 (0·3–0·9) 0·4 (0·2–0·6) ·· 0·0060

Median health status VAS at 90 (±14) days (IQR) 50 (30–70) 40 (20–58) ·· 0·13 Median PROMIS-10 physical health at 90 (±14) days, T-score (IQR) 39·8 (34·9–50·8) 34·9 (29·6–37·4) ·· 0·0008 Median PROMIS-10 mental health at 90 (±14) days (IQR), T-score 43·5 (36·3–50·8) 38·8 (32·6–43·5) ·· 0·025 PHQ-4 anxiety (%), n/N 12/52 (23%) 18/35 (51%) ·· 0·011

PHQ-4 depression (%), n/N 14/52 (27%) 11/35 (31%) ·· 0·81

Data are n (%) unless specified. The treatment effect was reported for the primary outcome as OR (95% CI) for the ordinal shift in the distribution of scores on the modified Rankin Scale towards a better outcome with imputation of missing values. Infarct volume at 24 h after randomisation was reported as coefficient from an adjusted log-linear model. PROMIS-10 and EQ-5D were reported as Wilcoxon rank sum test. Other outcomes were reported as the adjusted OR (95% CI) or as Fisher’s exact test for small numbers. The rate of successful restoration of flow of the occluded target arterial vessel, defined as mTICI 2b or better, was reported descriptively for the group that received thrombectomy with 95% CIs based on the Wald method. The rate of embolisation, either distal to the target occlusion or to previously unaffected territories, was reported descriptively for the group that received thrombectomy with 95% CIs based on the Clopper-Pearson method. The CIs for the secondary outcomes were not adjusted for multiple comparisons and can not be used for hypothesis testing. EQ-5D=EuroQol-5 Dimensions questionnaire. PHQ-4=poststroke anxiety and depression evaluated by the Patient Health Questionnaire-4. PROMIS 10=Patient-Reported Outcomes Measurement Information System 10-item. VAS=visual analogue scale. *Scores on the modified Rankin Scale range from 0 to 6, with higher scores indicating greater disability. †Successful reperfusion was defined as grade 2b to 3 on the modified Thrombectomy in the Cerebral Ischemia system ranging from 0–3, with higher grades indicating increased reperfusion; grade 2b indicates reperfusion of ≥50% of the occluded middle cerebral artery territory; and grade 3 indicates reperfusion of 100% of the occluded middle cerebral artery territory at the end of the thrombectomy procedure. ‡Assessed by the image core lab. §Mean infarct volume was assessed using CT in 101 (83%) of 121 patients assigned to endovascular thrombectomy and 96 (85%) of 113 patients assigned to medical treatment.

Table 2: Efficacy outcomes (intention-to-treat population)

medical treatment alone. Endovascular thrombectomy was associated with lower mortality (HR for death censored at 90 days 0·67 [95% CI 0·46–0·98]; p=0·038; figure 3 and table 3). Death or dependency (modified Rankin Scale 4–6) at 90 days was observed in 88 (69%) of 127 patients receiving endovascular thrombectomy and in 103 (87%) of 119 patients receiving medical treatment only (adjusted OR 0·34 [95% CI 0·18–0·65]; p=0·0011; appendix p 33).

1·0

0·9

0·8

0·7

0·6

0·5

0·4

0·3

0·2

0·1

0

Medical care

Thrombectomy

HR 0·67 (95% CI 0·46–0·98) p=0·038

0 10 20 30 40 50 60 70 80 90

Number at risk

Days since randomisation

Probability of survival

Symptomatic intracranial haemorrhage occurred in seven (5%) of 128 of patients given endovascular thrombectomy and in six (5%) of 125 given medical treatment alone. Parenchymal haemorrhage type 2 was observed in 11 (9%) patients receiving endovascular

endovascular thrombectomy, at least one serious adverse event was reported in 71 patients (56%), compared with

thrombectomy and ten (9%) patients receiving medical	Medical care 125	82	68	66 64	63	63	62	60 60
treatment (appendix p 33). In patients receiving	Thrombectomy 128	96	87	83 82	81	80	80	79 73

88 patients (70%) in the medical treatment group. A detailed list of serious adverse events is provided in the appendix (pp 35–38).

The results of subgroup analyses were generally supportive of the primary analysis (appendix p 21, 30), although the trial was not powered for subgroup analyses.

Figure 3: Kaplan-Meier analysis for mortality at 90 days

Analysis of mortality at 90 days was done in the safety population, including 128 patients who had received endovascular thrombectomy and 125 patients who had received medical treatment alone. HR=hazard ratio.

The per­protocol analysis included 97 patients in the endovascular thrombectomy group and 89 patients in the medical treatment group, and results were consistent with those of the primary intention­to­treat analysis

	Endovascular thrombectomy (N=128)	Medical treatment (N=125)	Treatment effect (95% CI)	p value
	Death or dependency (modified Rankin Scale 4–6) at 90 (±14) days (%), n/N 88/127 (69%)	103/119 (87%)	0·34 (0·18–0·65)	0·0011
	Mortality (censored at 90 days; %), n/N 49/122 (40%)	63/123 (51%)	0·67 (0·46–0·98)	0·038
	Mortality at 7 days or discharge 30 (23%)	36 (29%)	0·75 (0·42–1·32)	0·31
	Stroke-related death (%), n/N 33/127 (26%)	36/119 (30%)	0·80 (0·46–1·41)	0·44
	New ischaemic stroke 17 (13%)	20 (16%)	0·80 (0·39–1·60)	0·52
	Space-occupying infarction 21 (16%)	17 (14%)	1·26 (0·63–2·54)	0·52
	Parenchymal haemorrhage type 2 (%), n/N* 11/124 (9%)	10/116 (9%)	··	0·95
	Symptomatic intracranial haemorrhage† 7 (5%)	6 (5%)	··	1·00
	Fatal symptomatic intracranial haemorrhage 3 (2%)	4 (3%)	··	0·72
	At least one serious adverse event 71 (55%)	88 (70%)	··	0·014

(appendix p 18, 27). The sensitivity analyses (best­case scenario and worst­case scenario) were also consistent with the primary outcome analysis (appendix pp 17–18, 26–27). Given the imbalance in baseline ASPECTS among and between groups, we did an exploratory (not prespecified) analysis of the primary endpoint in the intention­to­treat population, adjusting for baseline ASPECTS, which was also consistent with the primary outcome analysis (appendix pp 28–29).

Data are n (%) unless specified. The treatment effect is reported for death (censored at 90 days), as an adjusted hazard ratio (95% CI). Other outcomes were reported as the adjusted OR (95% CI) or as Fisher’s exact test for small numbers. NIHSS=National Institutes of Health Stroke Scale. *Parenchymal haemorrhage type 2 was defined as an intracranial haemorrhage that involved more than 30% of the infarcted area with a substantial space-occupying effect or that was remote from the original infarcted area.

†Symptomatic intracranial haemorrhage was defined as an increase in the NIHSS score of ≥4 points or an increase in the score for an NIHSS subcategory of ≥2 points with presence of parenchymal haemorrhage type 2.

Table 3: Safety outcomes (safety population)

Discussion

The TENSION trial conducted in Europe and Canada showed that endovascular thrombectomy in addition to medical treatment resulted in improved functional outcome and lower mortality compared with medical treatment alone in patients presenting with a large brain infarction due to large vessel occlusion of the anterior brain circulation. Importantly, presence of a large infarction was identified by visual assessment of the infarct size mostly on unenhanced CT images without applying advanced imaging techniques or post­ processing tools. The trial was stopped early for efficacy after the pre­planned first interim analysis of 222 patients. The primary outcome analysis showed a significant shift towards an improved functional outcome across the entire range of the mRS at 90 days in patients assigned to endovascular thrombectomy. Results of secondary outcomes, including proportions of patients with functionally independent outcome (modified Rankin Scale 0–2) and being able to walk unaided (mRS 0–3) at 90 days, were higher in the group assigned to endovascular thrombectomy. Additionally, mortality was lower in patients receiving endovascular thrombectomy than in those receiving medical treatment alone. No safety concerns emerged, and there was no increase in the rates of severe

or symptomatic intracranial haemorrhages with endovascular thrombectomy.

The benefit of endovascular thrombectomy in the patient population with large infarcts has been demonstrated in three recent studies.11–13 Patient selection in these trials was based either on predominantly MRI as a more sensitive method to determine the infarct size11 or on perfusion imaging and a dedicated post­processing software12,13 to quantify core volume for patient inclusion. However, the most common standard in clinical practice worldwide used to assign patients to endovascular treatment is unenhanced CT supplemented by CTA to demonstrate the site of vessel occlusion.15,16 The TENSION trial aimed to use the clinical standard imaging for patient management by exclusively relying on visual image assessment of infarct size on either plain CT or MRI depending on local standards. Our results using this simple diagnostic algorithm yielded a comparable benefit of endovascular thrombectomy to previous trials,20 and, for the first time, also demonstrated a significant benefit in survival with endovascular thrombectomy. However, the benefit in survival with endovascular thrombectomy should be interpreted with caution as the trial was not powered to assess this and was stopped prematurely; therefore, the risk of a type I error was inflated. As in the previous trials, the overall rates of independent functional outcome were low and mortality was high across both groups in this population of patients with severe strokes. Nevertheless, endovascular thrombectomy in addition to medical treatment resulted in an absolute increase of 18% in the proportion of patients being able to walk independently at 90 days and an absolute decrease of 11% in mortality at 90 days.

The results of sensitivity analyses and the analysis of subgroups support these findings. The trial was not powered for subgroup analyses, but effect estimates were

in favour of endovascular thrombectomy in almost all prespecified subgroups. Endovascular thrombectomy was associated with a high recanalisation rate of 83% (mTICI 2b/3), which is in the usual range of recanalisation rates observed in previous trials of thrombectomy for stroke.21 We observed no significant effect of endovascular thrombectomy on the infarct size quantified 24 h after the procedure. A similar finding (ie, a benefit in functional outcome together with the absence of an effect on infarct size at 24 h) has also been reported in previous trials of endovascular thrombectomy22 and intravenous thrombolysis.23 Other surrogates of outcome (eg, reduction of rates of decompressive hemicraniectomy and of symptomatic intracranial haemorrhage) did not differ between groups, possibly due to the relatively small sample size.

Concerning safety, lower rates of severe adverse events were reported in patients receiving endovascular thrombectomy than in patients who received medical treatment alone. Overall, 7% of patients in the endovascular thrombectomy group had at least one serious adverse event related to the thrombectomy procedure, and there was no procedure­related mortality. Most importantly, neither rates of radiologically defined large parenchymal haemorrhages nor symptomatic intracranial haemorrhages were increased with endovascular thrombectomy.

Our trial supports the results of the previous trials of endovascular thrombectomy in patients with stroke and large core and extends these findings in some important aspects. TENSION was based mainly on patients in stroke centres in Europe, who were underrepresented in previous studies conducted mainly in Asia or the USA. TENSION also demonstrates a clinically relevant reduction of patients with severe disability or death in patients with large vessel occlusion and already extended infarcts. This reduction by endovascular thrombectomy represents important additional information to support treatment decisions in this patient population in clinical practice. In our trial, the rate of functional independence (mRS 0–2) at 90 days in the medical group was lower than in previous trials (2% vs 7–11%).11–13 The severity of symptoms in TENSION, assessed by the NIHSS at baseline, was similar to previous trials. However, patients in TENSION were mostly included based on non­contrast CT, whereas, in other studies, methods more sensitive to the extent of the acute ischaemic lesion were used, such as perfusion CT or DWI. Thus, the real size of the ischaemic lesion at baseline might have been larger in TENSION than in other trials, resulting in a lower rate of good clinical outcome.

Finally, regarding the translation of our findings to clinical practice, it is reassuring that a pragmatic approach to imaging evaluation based mainly on visual assessment of infarct size on non­contrast CT yielded a similar clinical benefit to the more sophisticated, but also more complicated and time­consuming, approaches involving MRI or perfusion imaging used in the previous

studies. We also observed a non­significant beneficial effect of endovascular thrombectomy in the subgroup analysis of patients with an ASPECTS value of 0–2 on post­hoc core lab analysis. Comparable effects have also been reported in previous studies.12,13 Future trials and meta­analyses should define the lower boundary of ASPECTS value for a clinical benefit of endovascular thrombectomy.

Our trial has limitations. First, as the trial was stopped early for efficacy, the overall patient number was smaller than planned and, therefore, the analysis of subgroups is limited. Moreover, halted trials are more prone to exaggerated effects due to reduced sample size.24 Second, interrater agreement of early ischaemic signs on CT was only moderate; therefore, patient inclusion based on visual assessment of ASPECTS on CT or MRI only without further functional imaging or post­processing might result in some variability. This was reflected by the results of the image core lab evaluation, which considered the ASPECTS values of 50 patients who were included in the study to be out of range for inclusion. However, this factor did not affect the overall clinical benefit of endovascular thrombectomy and the estimated treatment effect in the per­protocol population was not higher than in the intention­to­treat analysis. Enrolment in the trial was limited to patients in whom treatment could be accomplished within 12 h of symptom onset or last seen well, and to patients with an NIHSS of less than 26. We thus cannot generalise our findings to patients treated in a later time window or to those with more severe stroke symptoms. We also did not obtain data on race and ethnicity.

In this prospective randomised trial conducted in Europe and Canada, endovascular thrombectomy together with medical treatment was associated with improved functional outcomes and survival in patients with large vessel occlusions of the anterior circulation presenting with a large infarct on standard imaging compared with medical treatment alone, with no safety concerns.

Contributors

MB, JF, SB, AHA, BF, ERG, MDH, AK, LP, CZS, KZ and GT developed

the study protocol. MB, FS, and GT interpreted the data and drafted the manuscript. FS and AD analysed the data. All authors collected data and edited the manuscript. MB, FS, and GT accessed and verified the underlying data and all authors had access to the data and accept responsibility for submitting the article for publication.

Declaration of interests

MB reports funding from EU Horizon 2020 and Deutsche Forschungsgemeinschaft (payments to the institution); honoraria for lectures from Novartis, Boehringer Ingelheim, and Seagen; and consulting fees from NeuroScios and Boehringer Ingelheim and is an editor in chief of Clinical Neuroradiology (Springer). JF reports funding from the European Commission; personal consulting fees from Acandis, Cerenovus, Medtronic, Microvention, Phenox, Stryker, and Roche; consulting at Philips (no payments); payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events from Penumbra and Tonbridge; support for attending meetings or travel from Medtronic and Penumbra; stock or stock options from Tegus Medical, Eppdata, and Vastrax; and participation in a Data Safety Monitoring Board or Advisory Board at Phenox (personal fees) and Stryker

(personal fees) and is a past president of ESMINT. SB reports funding from the EU Horizon 2020 research and innovation programme (754640; payments made to the institution); and support for attending meetings or travel from Medtronic and Penumbra; AHA reports unrestricted research grants from Boehringer Ingelheim; honoraria for lectures from BMS/ Pfizer, Teva, Roche, Abbvie, Lundbeck, and Novartis; and participation in Advisory Boards for MSD, BMS/Pfizer, Lundbeck, Lilly, and Abbvie.

BF reports research grants from Carlos III Institute of Health; personal payment for educational lectures from Servicio Madrileño de Salud; payment for lectures from Euromedice to the institution; personal payment for educational lectures from Takeda; support for attending meetings from Daiichi Sankyo; receipt of materials for research from Abbot. MDH reports funding from Nil; grants to the University of Calgary for the TEMPO­2 trial from Boehringer Ingelheim, Biogen, NoNO (ESCAPE­NA1 trial and ESCAPE­NEXT trial), Canadian Institute for Health Research (ESCAPE­NA1 trial and ESCAPE­NEXT trial), Medtronic (HERMES collaboration), Alberta Innovates (QuICR Alberta Stroke Program); that some of the funds were used for the ESCAPE­NA1 trial from Alberta Innovates; consulting fees from Sun Pharma Brainsgate (paid work for adjudication of clinical trial outcomes);

US patents 62/086,077 (licensed to Circle NVI) and 10,916,346 (licensed to Circle NVI); private stock ownership from Circle and PUreWeb; participation as data and safety monitoring committee chair of the RACECAT trial (end 2020), the Oncovir Hiltonel trial (ongoing), and the DUMAS trial (ongoing); participation as a data and safety monitoring committee member of the ARTESIA trial (ongoing), and the BRAIN­AF trial (ongoing); and is president of the Canadian Neurological Sciences Federation (not for profit) and a Board member of the Canadian Stroke Consortium (not for profit). AK reports grants from the European Commission for the TENSION study (payment to the institution).

LP reports consulting fees from Balt, Microvention, and Phenox; and support for attending meetings or travel for the TENSION investigator meeting (transport and accommodation was reimbursed by the organization). CZS reports grants from Novo Nordisk Foundation and Health Research Foundation of Central Denmark Region. RAB reports speakers fees from Novo Nordisk and Beyer. HD reports financial compensation for the start­up fee and the obligatory payment to the hospital administration paid by the sponsor (Medical University Heidelberg) to the clinical division (no personal payments); personal consulting fees from Stryker; speakers honorary from Medtronic; support for attending meetings or travel from Medtronic; and past presidency of the Austrian Society of Interventional Radiology and past presidency Austrian Society of Neuroradiology. FD reports consulting fees from Cerenovus, Phenox, Balt, Cerus Endovascular, Stryker; payment for expert testimony from Cerenovus; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from Cerenovus, Stryker, Acandis, Asahi, and Penumbra; participation in a Data Safety Monitoring Board or Advisory Board at Cerenovus; and previously work as an associate editor for Journal of NeuroInterventional Surgery and Journal of Clinical Medicine. FF reports consulting fees from Eppdata and support for attending meetings or travel from Microvention, Medtronic, Cerebrovascular Research and Education Foundation (CREF), and Acandis. SG reports consulting fees from Eppdata. CH reports consulting fees from Brainomix and lecture fees from Stryker.

SH­J reports funding for data collection, payment, or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Terumo. MG reports research grants from Medtronic and Cerenovus (payments to the University of Calgary); royalties or licenses from Microvention (systems of intracranial access); personal consulting fees from Microvention, Medtronic, Stryker, Mentice, Philips, and Penumbra; and stock or stock options from Circle Neurovascular. CFK was chair of the German stroke registry (unpaid).

RM reports payments for a stroke lecture from TMC Academy.

MM reports grants from Balt, Medtronic, MicroVention, and Stryker; consulting fees from Siemens; and support for attending meetings or travel from Europa Group. SM­H reports consulting fees from Terumo and Boston Scientific Corporation; payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Terumo and Boston Scientific Corporation. NM reports the provision of study materials. PP reports support for attending meetings or travel for the TENSION investigator meeting (transport and

accommodation was reimbursed by the organisation). MP reports grants from the German Research Foundation (DFG SFB 1158 A10, DFG KFO 5001 P02, DFG KFO 5001 Z, and DFG SFB TR 240 B02); speaker

honoraria unrelated from Merck Serono and Bayer; and support for attending meetings or travel from Merck Serono (travel reimbursement) and Bayer (travel reimbursement). PAR reports consulting fees to the institution from Boehringer Ingelheim and Bayer; and payment or honoraria for lectures, presentations, speakers’ bureaus, manuscript writing, or educational events from Boehringer Ingelheim, Bayer, Pfizer, and BMS (all made to the institution). ES reports grants from Hamburg Innovation and Hertie Foundation. DFV reports research grants from MicroVention; consulting fees from Medtronic; and paid lectures from Cerenovus and Johnson & Johnson. WW reports consulting fees to the institution from Abbvie, BMS, GSK, and Servier. GT reports funding from the European Commission (EUHorizon 2020 research and innovation programme, 754640; payments to the institution); personal consulting fees from Acandis, AstraZeneca, Bayer, Boehringer Ingelheim, and Stryker; personal payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from Acandis, Alexion, Marin, Bayer, Boehringer Ingelheim, BristolMyersSquibb/Pfizer, Daiichi Sankyo, and Stryker; participation as DSMB member for the TEA Stroke Trial (no payments) and ReSCInD trial (no payments); work as a speaker of the Commission for Cerebrovascular Diseases of the German Society of Neurology (DGN; no payments); and membership of the Board of Directors of the European Stroke Organisation (ESO; no payments). All other authors declare no competing interests.

Data sharing

Individual participant data that underlie the results reported in this article, after de­identification, will be made available on request beginning 12 months and ending 36 months following article publication to investigators whose proposed use of the data has been approved by the TENSION steering committee. Requests can be made by email to the corresponding author.

Acknowledgments

TENSION was supported by the EU Horizon 2020 research and innovation programme (754640). We thank the patients and their families for participating in the trial, the members of the data and safety monitoring board (Kennedy Lees [University of Glasgow],

Karl Wegscheider [Universitätsklinikum Hamburg–Eppendorf], and Phil White [Newcastle University]), the European Society of Minimally Invasive Neurological Therapy (ESMINT), the International Consortium for Health Outcomes Measurement (ICHOM), and the Stroke Alliance for Europe (SAFE) for their collaboration on the TENSION project.

References

1. Berkhemer OA, Fransen PSS, Beumer D, et al. A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med 2015; 372: 11–20.
2. Saver JL, Goyal M, Bonafe A, et al. Stent­retriever thrombectomy after intravenous t­PA vs t­PA alone in stroke. N Engl J Med 2015; 372: 2285–95.
3. Jovin TG, Chamorro A, Cobo E, et al. Thrombectomy within

8 hours after symptom onset in ischemic stroke. N Engl J Med 2015;

372: 2296–306.

1. Goyal M, Demchuk AM, Menon BK, et al. Randomized assessment of rapid endovascular treatment of ischemic stroke. N Engl J Med 2015; 372: 1019–30.
2. Campbell BCV, Mitchell PJ, Kleinig TJ, et al. Endovascular therapy for ischemic stroke with perfusion­imaging selection. N Engl J Med 2015; 372: 1009–18.
3. Nogueira RG, Jadhav AP, Haussen DC, et al. Thrombectomy 6 to 24 hours after stroke with a mismatch between deficit and infarct. N Engl J Med 2018; 378: 11–21.
4. Albers GW, Marks MP, Kemp S, et al. Thrombectomy for stroke at 6 to 16 hours with selection by perfusion imaging. N Engl J Med 2018; 378: 708–18.
5. Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke:

2019 update to the 2018 guidelines for the early management of acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association.

Stroke 2019; 50: e344–418.

1. Turc G, Bhogal P, Fischer U, et al. European Stroke Organisation (ESO)–European Society for Minimally Invasive Neurological Therapy (ESMINT) guidelines on mechanical thrombectomy in acute ischemic stroke. J Neurointerv Surg 2023; 15: e8.
2. Sarraj A, Hassan AE, Savitz S, et al. Outcomes of endovascular thrombectomy vs medical management alone in patients with large ischemic cores: a secondary analysis of the Optimizing Patient’s Selection for Endovascular Treatment in Acute Ischemic Stroke (SELECT) study. JAMA Neurol 2019; 76: 1147–56.
3. Yoshimura S, Sakai N, Yamagami H, et al. Endovascular therapy for acute stroke with a large ischemic region. N Engl J Med 2022;

386: 1303–13.

1. Sarraj A, Hassan AE, Abraham MG, et al. Trial of endovascular thrombectomy for large ischemic strokes. N Engl J Med 2023; 388: 1259–71.
2. Huo X, Ma G, Tong X, et al. Trial of endovascular therapy for acute ischemic stroke with large infarct. N Engl J Med 2023; 388: 1272–83.
3. Román LS, Menon BK, Blasco J, et al. Imaging features and safety and efficacy of endovascular stroke treatment: a meta­analysis of individual patient­level data. Lancet Neurol 2018; 17: 895–904.
4. Nogueira RG, Haussen DC, Liebeskind D, et al. Stroke imaging selection modality and endovascular therapy outcomes in the early and extended time windows. Stroke 2021; 52: 491–97.
5. Kim Y, Lee S, Abdelkhaleq R, et al. Utilization and availability of advanced imaging in patients with acute ischemic stroke.

Circ Cardiovasc Qual Outcomes 2021; 14: e006989.

1. Bendszus M, Bonekamp S, Berge E, et al. A randomized controlled trial to test efficacy and safety of thrombectomy in stroke with extended lesion and extended time window. Int J Stroke 2019;

14: 87–93.

1. van Horn N, Kniep H, Broocks G, et al. ASPECTS interobserver agreement of 100 investigators from the TENSION study.

Clin Neuroradiol 2021; 31: 1093–100.

1. von Kummer R, Broderick JP, Campbell BC, et al. The Heidelberg bleeding classification: classification of bleeding events after ischemic stroke and reperfusion therapy. Stroke 2015; 46: 2981–86.
2. Goyal M, Demchuk AM, Hill MD. Endovascular therapy for ischemic stroke. N Engl J Med 2015; 372: 2366.
3. Goyal M, Menon BK, van Zwam WH, et al. Endovascular thrombectomy after large­vessel ischaemic stroke: a meta­analysis of individual patient data from five randomised trials. Lancet 2016; 387: 1723–31.
4. Albers GW, Marks MP, Kemp S, et al. Thrombectomy for stroke at 6 to 16 hours with selection by perfusion imaging. N Engl J Med 2018; 378: 708–18.
5. Thomalla G, Simonsen CZ, Boutitie F, et al. MRI­guided thrombolysis for stroke with unknown time of onset. N Engl J Med 2018; 379: 611–22.
6. Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta­regression analysis. JAMA 2010; 303: 1180–87.