Article Text


Original research
Endovascular thrombectomy beyond 12 hours of stroke onset: a stroke network’s experience of late intervention
  1. Ronan Motyer1,2,
  2. John Thornton1,
  3. Sarah Power1,
  4. Paul Brennan1,
  5. Alan O’Hare1,
  6. Seamus Looby1,
  7. David J Williams3,4,
  8. Barry Moynihan4,
  9. Sean Murphy5,6,7
  1. 1 Interventional Neuroradiology Service, Department of Radiology, Beaumont Hospital, Dublin, Ireland
  2. 2 Department of Radiology, The Adelaide and Meath Hospital, Dublin, Ireland
  3. 3 Department of Geriatric and Stroke Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland
  4. 4 Department of Geriatric and Stroke Medicine, Beaumont Hospital, Dublin, Ireland
  5. 5 Department of Geriatric and Stroke Medicine, The Mater Misericordiae University Hospital, Dublin, Ireland
  6. 6 School of Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland
  7. 7 School of Medicine, University College Dublin, Dublin, Ireland
  1. Correspondence to Dr Ronan Motyer, Department of Radiology, Beaumont Hospital, Dublin 9, Ireland; ronanmotyer{at}


Background Selected patients with proximal anterior circulation ischemic stroke who demonstrate limited infarct and sufficient penumbra may benefit from endovascular thrombectomy (EVT) beyond conventional time limits.

Objective To perform a retrospective review of all cases of EVT performed at our institution for proximal anterior circulation acute ischemic stroke with onset >12 hours.

Materials and methods Patients were assessed with non-contrast CT brain and multiphase CT angiography, with Alberta Stroke Program Early CT Score (ASPECTS) and collateral grade informing patient selection. Data, including patient demographics, workflow, neuroimaging findings, procedural details, recanalization rates, and 90-day functional outcomes, were collected.

Results Of the 542 consecutive endovascular thrombectomy cases performed during the study period, 25 (4.6%) were >12 hours from stroke onset. Median age was 69 years (IQR 55–80), median National Institute of Health Stroke Scale score on presentation was 14 (IQR 11–18.5), median ASPECTS was 8 (IQR 8–9), and rate of moderate–good collateral status was 96% (n=24). Median time to groin puncture was 14 hours 40 min (IQR 12 hours 36 min – 16 hours 18 min). Rate of successful recanalization (modified Thrombolysis in Cerebral infarction 2b–3) was 88% (n=22). Rate of functional independence (90-day modified Rankin Scale score 0–2) was 52% (n=13). There were no cases of symptomatic intracranial haemorrhage and 90-day mortality rate was 12% (n=3).

Conclusion With the use of ASPECTS and collateral grade to guide patient selection, good functional outcome with acceptable safety parameters may be achieved in patients undergoing EVT beyond 12 hours from stroke onset.

  • endovascular
  • stroke
  • large vessel occlusion
  • thrombectomy
  • thrombolysis
  • recanalization
  • IV tPA

Statistics from


In 2016, a meta-analysis by the HERMES Collaboration firmly established endovascular thrombectomy (EVT) as the standard of care for proximal anterior circulation acute ischemic stroke (AIS). The five clinical trials included in the meta-analysis predominantly randomized patients to intervention up to 6 hours from symptom onset, although REVASCAT included patients up to 8 hours and ESCAPE up to 12.1 However, the treatment paradigms created in response to these trials fail to take into account a cohort of patients considered ‘slow progressors’ (≤30% AIS large vessel occlusion presentations), thereby denying certain patients a potentially effective intervention.2

Now, less than 2 years since the HERMES publication, latest level one data sets out to significantly extend the treatment threshold for EVT in selected patients. Using perfusion-based neuroimaging to select patients, DAWN and DEFUSE-3 have demonstrated EVT benefit 6–24 and 6–16 hours from time last known to be well, respectively.3 4 American Heart Association/American Stroke Association guidelines have been updated in response to this evidence; although recommendations apply only to those patients who meet trial criteria, necessitating the availability of perfusion imaging.5

At our institution, a tissue-based approach to patient selection is applied, with both clinical and neuroimaging findings considered a better determinant of individual pathophysiology than symptom duration alone. In contrast to DAWN and DEFUSE-3, calculation of Alberta Stroke Program Early CT Score (ASPECTS) and collateral grade is used to guide our selection. This was proved to be effective by ESCAPE up to 12 hours from symptom onset.6 7

In light of recent evidence and use of alternative advanced neuroimaging, we chose to perform a retrospective review of all EVT cases performed at our institution for proximal anterior circulation AIS with onset >12 hours.

Materials and methods

As a comprehensive stroke centre, our institution provides emergency EVT on selected cases of AIS, with patients presenting directly or referred from a stroke network of 24 regional hospitals, the furthest located 244 km away. All patients are managed after intervention in an acute stroke unit at our institution or at the referring hospital following repatriation once clinically stable. A full description of our service has previously been published.8

Since June 2010, a prospectively maintained database has been kept of all EVT cases performed. Our institution enrolled patients in the ESCAPE trial (randomized: n=34, interventional arm: n=17), however, no patients in this present series were included in that analysis owing to symptom onset >12 hours.6 We have demonstrated similar EVT benefit <6 and >6 hours from symptom onset; 11 patients in the >6 hour cohort were included in this series.9

We present relevant data from all EVT cases at our institution where groin puncture was performed >12 hours from onset of proximal anterior circulation AIS. Similar to ESCAPE, occlusions of the carotid terminus, middle cerebral artery (MCA), and immediate branches were included.6 The cases included in this analysis presented between April 2014 and July 2017, allowing for completion of a 90-day functional outcome assessment. Data based on patient demographics, workflow, neuroimaging findings, procedural details, and clinical outcomes was collected. Second-generation stent-retriever thrombectomy devices were used in all cases.

Patients were selected after initial evaluation by a referring physician and discussion with a neurointerventional radiologist at our institution. For cases with unknown symptom onset time, stroke onset was recorded as time last known to be well. Modes of presentation were categorised as known onset time (KOT), wake-up stroke (WUS), and last seen well (LSW).

Neuroimaging comprised non-contrast computed tomography (NCCT) of the brain and computed tomographic angiography (CTA), preferably multiphase (mCTA), to determine occlusion location and collateral status. ASPECTS was calculated for all patients using NCCT images obtained.

A moderate–good collateral grade was defined by ESCAPE as ≥50% filling of the MCA pial arterial circulation on any phase of CTA; a small infarct core was defined as ASPECTS 6–10.6 Continuing our experience with ESCAPE, we applied these definitions.

Although no prespecified selection criteria were used, both clinical and neuroimaging findings were central in assessing suitability for EVT. Using similar criteria to ESCAPE, a minimum threshold of moderate–good collateral status with small infarct core was generally applied.6 These data, in conjunction with clinical assessment, informed patient selection. No strict National Institute of Health Stroke Scale (NIHSS) score cut-off point was applied. There was no maximum time limit for intervention.

The main clinical outcome measures included 90-day functional outcome according to the modified Rankin Scale (mRS), rates of symptomatic intracranial haemorrhage (sICH), and 90-day mortality. The degree of recanalization after EVT was recorded according to the modified treatment in cerebral ischaemia (mTICI) score, with a score of ≥2 b defined as successful revascularization.

As this was a retrospective evaluation of current practice in our unit, without influencing individual patient management, this study was categorized as a service review and was exempted from requiring institutional ethics approval. The ongoing audit of our service is registered with the institutional Quality and Standards Department.


From June 2010 to July 2017, a total of 542 patients with AIS underwent EVT at our institution. EVT with groin puncture >12 hours from onset was performed in 25 patients (4.6%) who presented with proximal anterior circulation AIS between April 2014 and July 2017.

Twenty-two patients (88%) that underwent EVT >12 hours were transferred from a referring hospital to our institution. The mode of patient presentation was as follows: 64% were LSW (n=16), 24% were WUS (n=6), and 12% were KOT (n=3).

The median age of patients undergoing EVT >12 hours was 69 (IQR 55–80). Median NIHSS on presentation was 14 (IQR 11–18.5), ranging from 5 to 26. NIHSS was not recorded on one intubated and sedated patient. No patient received intravenous thrombolysis. Median ASPECTS was 8 (IQR 8–9). Rate of moderate–good collateral status (50– 75% MCA pial filling) was 96% (n=24), and of good collateral status (75–99% MCA pial filling) was 76% (n=19). (table 1)

Table 1

Outcomes for endovascular thrombectomy >12 hours from symptom onset

Occlusion of the M1 segment occurred in 88% (n=22), of the M2 segment in 8% (n=2) and of the carotid terminus occurred in 4% (n=1). Additional occlusion of the internal carotid artery cervical segment occurred in 16% (n=4).

Median time to groin puncture was 14 hours 40 min (IQR, 12 hours 36 min – 16 hours 18 min), ranging from 12 hours 2 min – 20 hours 45 min. Successful recanalization (mTICI of 2b–3) was achieved in 88% of patients (n=22). An analysis of workflow is shown in table 2.

Table 2

Analysis of workflow

The rate of functional independence (90-day mRS 0–2) was 52% (n=13), while 68% of patients achieved an mRS of 0–3 (n=17). The 90-day mortality rate was 12% (n=3). No patient had sICH. Of those patients with a baseline mRS of 0–1, the rate of functional independence (90-day mRS 0–2) was 59% (table 1).

Functional independence, according to mode of presentation, was achieved in 2/3 (66.7%) strokes with KOT, 9/16 (56.3%) LSW strokes, and 2/6 (33.3%) WUS. The 90-day mortality rate was 0% for strokes with KOT (n=0), 12.5% for LSW strokes (n=2) and 16.7% for WUS (n=1).


Our outcomes highlight the positive experience of a comprehensive stroke centre in managing a specific cohort of late-presenting EVT cases.

Our experience may be compared with DAWN, a randomized controlled trial designed to evaluate EVT outcomes 6–24 hours from time last seen well. DAWN selected patients using a clinical-core mismatch paradigm with NIHSS and CT perfusion (CTP) or MR diffusion-weighted imaging (DWI). EVT was shown to be safe and effective, with better rates of functional independence than with standard medical therapy (49% vs 13%; 43% vs 7% in 12–24 hour cohort).3 Similarly, using CTP or DWI and a perfusion-imaging mismatch selection paradigm, DEFUSE-3 demonstrated favourable functional outcomes and acceptable safety parameters with EVT performed 6–16 hours from stroke onset.4

Data from DAWN and DEFUSE-3 is in keeping with our outcomes. 52% of our >12 hour cohort achieved functional independence (n=13) and 68% achieved a 90-day mRS of 0–3 (n=17). We have demonstrated acceptable safety parameters, with no cases of sICH and a 90-day mortality rate of 12% (n=3). These outcomes may also be compared with the HERMES meta-analysis, which used far narrower therapeutic time limits.1

Similarities in this patient cohort as compared with recent randomized controlled trials include proximal anterior circulation occlusion and treatment with second-generation stent retrievers.1 3 Median patient age and NIHSS in our series were 69 and 14, respectively, compared with 72 and 17 in the intervention arm of DAWN.3

Appropriate patient selection is central to achieving these outcomes and the variable rates of ischemic progression, largely due to collateral circulation, suggest a shift towards individualized clinical pathophysiological assessment.10 11 Multiple strategies have been adopted in response to this, including clinical-core mismatch, perfusion-imaging mismatch and our approach with NCCT ASPECTS to estimate infarct core, and mCTA to assess collateral status. This range of potential selection tools highlights current uncertainty about the best inclusion criteria.12

While perfusion-based selection has been shown to be effective, limitations do exist.13 A lack of standardized imaging parameters, uncertain thresholds for interpretation, interobserver variability, and a low signal-to-noise ratio limit the accuracy and reliability of CTP for estimating infarct core in clinical practice.14 This may be overcome with advances in postprocessing algorithms and the use of automated software such as RAPID (iSchemaView, California, USA).15 16 In contrast, MR DWI can accurately estimate infarct core, thereby aiding selection and limiting futile intervention, although its implementation as a selection tool may narrow inclusion criteria.17 A balance must be sought between maximising benefit while maintaining adequate safety. For instance, the CT perfusion-imaging mismatch paradigm has been suggested as overly restrictive without improving outcomes.18 However, for late intervention, best evidence does support perfusion-based neuroimaging to guide selection.3 4

Our approach was demonstrated to be effective by ESCAPE, with subsequent analysis supporting its role in patient selection 6–12 hours from symptom onset. Advantages for resource availability, speed of acquisition, and applicability are noteworthy.6 7 13

ASPECTS informs patient selection by quantifying early ischemic change; a recent study suggests comparability with CTP in predicting infarct volume.19 Superior EVT outcomes may be achieved with ASPECTS >7, while scores ≤4 have been shown to be futile.20 Calculation of ASPECTS with CTA source images may lead to further accuracy, although we did not perform such analysis.21 Limitations do exist, with uncertain inclusion thresholds and significant interobserver variability reported.13 22 23 We acknowledge that further analysis is required before strict ASPECTS cut-off points can be defined. The lowest ASPECTS in a patient we treated who achieved functional independence was 6.

Evaluation of collateral status may be performed accurately and reliably with mCTA, thereby indirectly estimating infarct core and penumbra, while also confirming occlusion location. Collateral grade predicts EVT benefit, with superior EVT outcomes achieved in patients with good collateral flow.24–27 Grading can subsequently be used to aid accuracy of ASPECTS calculation, an approach proved effective by the ESCAPE study.6 13 We believe establishing collateral status and inference of penumbra to be of particular importance in the late-presenting cohort, limiting potentially futile intervention.

Ultimately, a randomized controlled trial is required to appropriately evaluate the various selection paradigms and ascertain the most effective and applicable neuroimaging techniques and inclusion thresholds, particularly in cases of late presentation.

This study is not without limitations. The analysis was observational and not a randomized controlled trial, as patients were selected for EVT on a case-by-case basis, at the discretion of the neurointerventionalist. However, we have demonstrated that in a specific cohort of late-presenting patients with AIS, EVT can be performed safely and with good functional outcomes. The majority of presentations were WUS or LSW stroke. While this potentially confounds the recorded onset times, it is reflective of clinical practice and typical of the late-presenting stroke cohort. The treatment arm in DAWN comprised 64.5% WUS and 25.2% LSW stroke.3 The outcomes, therefore, highlight the potential benefit that may still be achieved in these patient cohorts.

Importantly, time to treatment must remain a key focus, given its direct impact on patient outcome. Thus, despite emerging data showing positive outcomes for late EVT, emphasis on fast workflow continues to be a priority.28–30 While our >12 hour cohort represents a minority of the total pool of presenting AIS cases, achieving functional independence in approximately half of such patients is important, given the severe disability associated with non-intervention.


Recent data demonstrate the benefit of EVT in selected patients up to 24 hours from symptom onset following proximal anterior circulation AIS. Our institutional experience of EVT performed >12 hours from onset is reflective of this evidence, despite using alternative neuroimaging techniques to inform patient selection.

Considerable uncertainty remains around the assessment of late presenting AIS patients, particularly the role of advanced neuroimaging. Despite this, EVT is rapidly establishing itself as best practice beyond conventional time limits. It is essential that optimal neuroimaging and associated inclusion criteria be correctly defined so as to ensure maximum benefit and minimum harm to these patients.


The authors thank Sinead Duff and Chloe Carpenter for their contribution to this project and continued support.


View Abstract


  • Contributors RM is lead author, monitored data collection, interpreted and presented acquired data, drafted and revised the manuscript, approved submitted version and agrees to be accountable for all aspects of the work. JT is senior author, implemented and directed service, designed data collection tools, acquired data, monitored data collection, interpreted data, advised and revised the manuscript, approved submitted version and agrees to be accountable for all aspects of the work. SP, PB AOH, SL, DJW, BM implemented the service, designed data collection tools, acquired data, monitored data collection, interpreted data, advised and revised the manuscript, approved submitted version and agree to be accountable for all aspects of the work. SM is senior author, advised on service implementation, monitored data collection, interpreted data, advised and revised the manuscript, approved submitted version and agrees to be accountable for all aspects of the work.

  • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent Not required.

  • Ethics approval Beaumont Hospital Quality and Standards Department.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.