Imaging features and safety and efficacy of endovascular stroke treatment: a meta-analysis of individual patient-level data
Luis San Román,Bijoy K Menon,Jordi Blasco,María Hernández-Pérez,Antoni Dávalos,Charles B L M Majoie,Bruce C V Campbell,Francis Guillemin,Hester Lingsma,René Anxionnat,Jonathan Epstein,Jeffrey L Saver,Henk Marquering,John H Wong,Demetrius Lopes,Gernot Reimann,Hubert Desal,Diederik W J Dippel,Shelagh Coutts,Richard du Mesnil de Rochemont,Dileep Yavagal,Jean Christophe Ferre,Yvo B W E M Roos,David S Liebeskind,Robert Lenthall,Carlos Molina,Fahad S Al Ajlan,Vivek Reddy,Dar Dowlatshahi,Nader-Antoine Sourour,Catherine Oppenheim,Alim P Mitha,Stephen M Davis,Christian Weimar,Robert J van Oostenbrugge,Erik Cobo,Timothy J Kleinig,Geoffrey A Donnan,Aad van der Lugt,Andrew M Demchuk,Olvert A Berkhemer,Anna M M Boers,Gary A Ford,Keith W Muir,B Scott Brown,Tudor Jovin,Wim H van Zwam,Peter J Mitchell,Michael D Hill,Phil White,Serge Bracard,Mayank Goyal,HERMES collaborators
DOI: https://doi.org/10.1016/S1474-4422(18)30242-4
Abstract:Background: Evidence regarding whether imaging can be used effectively to select patients for endovascular thrombectomy (EVT) is scarce. We aimed to investigate the association between baseline imaging features and safety and efficacy of EVT in acute ischaemic stroke caused by anterior large-vessel occlusion. Methods: In this meta-analysis of individual patient-level data, the HERMES collaboration identified in PubMed seven randomised trials in endovascular stroke that compared EVT with standard medical therapy, published between Jan 1, 2010, and Oct 31, 2017. Only trials that required vessel imaging to identify patients with proximal anterior circulation ischaemic stroke and that used predominantly stent retrievers or second-generation neurothrombectomy devices in the EVT group were included. Risk of bias was assessed with the Cochrane handbook methodology. Central investigators, masked to clinical information other than stroke side, categorised baseline imaging features of ischaemic change with the Alberta Stroke Program Early CT Score (ASPECTS) or according to involvement of more than 33% of middle cerebral artery territory, and by thrombus volume, hyperdensity, and collateral status. The primary endpoint was neurological functional disability scored on the modified Rankin Scale (mRS) score at 90 days after randomisation. Safety outcomes included symptomatic intracranial haemorrhage, parenchymal haematoma type 2 within 5 days of randomisation, and mortality within 90 days. For the primary analysis, we used mixed-methods ordinal logistic regression adjusted for age, sex, National Institutes of Health Stroke Scale score at admission, intravenous alteplase, and time from onset to randomisation, and we used interaction terms to test whether imaging categorisation at baseline modifies the association between treatment and outcome. This meta-analysis was prospectively designed by the HERMES executive committee but has not been registered. Findings: Among 1764 pooled patients, 871 were allocated to the EVT group and 893 to the control group. Risk of bias was low except in the THRACE study, which used unblinded assessment of outcomes 90 days after randomisation and MRI predominantly as the primary baseline imaging tool. The overall treatment effect favoured EVT (adjusted common odds ratio [cOR] for a shift towards better outcome on the mRS 2·00, 95% CI 1·69-2·38; p<0·0001). EVT achieved better outcomes at 90 days than standard medical therapy alone across a broad range of baseline imaging categories. Mortality at 90 days (14·7% vs 17·3%, p=0·15), symptomatic intracranial haemorrhage (3·8% vs 3·5%, p=0·90), and parenchymal haematoma type 2 (5·6% vs 4·8%, p=0·52) did not differ between the EVT and control groups. No treatment effect modification by baseline imaging features was noted for mortality at 90 days and parenchymal haematoma type 2. Among patients with ASPECTS 0-4, symptomatic intracranial haemorrhage was seen in ten (19%) of 52 patients in the EVT group versus three (5%) of 66 patients in the control group (adjusted cOR 3·94, 95% CI 0·94-16·49; pinteraction=0·025), and among patients with more than 33% involvement of middle cerebral artery territory, symptomatic intracranial haemorrhage was observed in 15 (14%) of 108 patients in the EVT group versus four (4%) of 113 patients in the control group (4·17, 1·30-13·44, pinteraction=0·012). Interpretation: EVT achieves better outcomes at 90 days than standard medical therapy across a broad range of baseline imaging categories, including infarcts affecting more than 33% of middle cerebral artery territory or ASPECTS less than 6, although in these patients the risk of symptomatic intracranial haemorrhage was higher in the EVT group than the control group. This analysis provides preliminary evidence for potential use of EVT in patients with large infarcts at baseline. Funding: Medtronic.