Article Text

Download PDFPDF

Original research
Multicenter clinical experience in over 125 patients with the Penumbra Separator 3D for mechanical thrombectomy in acute ischemic stroke
  1. Daniel Behme1,
  2. Annika Kowoll2,
  3. Anastasios Mpotsaris3,
  4. Claudia Hader4,
  5. Lukas Hechelhammer4,
  6. Johannes Weber4,
  7. Werner Weber2
  1. 1Department of Neuroradiology, Georg-August-University Göttingen, University Medical Center, Göttingen, Germany
  2. 2Department of Radiology and Neuroradiology, Ruhr University Bochum, University Medical Center—Knappschaftskrankenhaus Langendreer, Bochum, Germany
  3. 3Department of Radiology and Neuroradiology, University Hospital Cologne, Cologne, Germany
  4. 4Clinic of Radiology and Nuclear Medicine, Diagnostic and Interventional Neuroradiology, Kantonsspital St. Gallen, St Gallen, Switzerland
  1. Correspondence to Dr D Behme, Department of Neuroradiology, Georg-August University Göttingen, University Medical Center, Robert-Koch-Str. 40, Göttingen 37075, Germany; daniel.behme{at}


Background The aim of this study was to assess reperfusion and clinical outcome of treatment with the self-expanding retrievable Separator 3D in revascularization of acute ischemic stroke. The three-dimensional (3D) device secures thrombus with direct aspiration and supports debulking of the clot.

Methods At two centers, 129 consecutive stroke patients with National Institutes of Health Stroke Scale (NIHSS) scores ≥5 were treated with mechanical thrombectomy using the Separator 3D as a component of the Penumbra System within 8 h of symptom onset; modified Treatment in Cerebral Infarction (mTICI) revascularization scores, NIHSS score on admission and discharge, mortality rates, and modified Rankin Scale (mRS) outcomes at 90 days were evaluated.

Results A total of 129 vessels in 129 patients were treated. Occlusions were located in the middle cerebral artery (MCA, 48%), internal carotid artery (ICA, 33%), cervical ICA–MCA (3%), and vertebrobasilar arteries (16%). Intravenous thrombolytic therapy with recombinant tissue plasminogen activator was given to 78% of patients. Median NIHSS was 15 prior to treatment. Reperfusion to mTICI 2b or 3 was successful in 96/129 (74%) target arterial lesions, with more than half of cases (51%) achieving mTICI 3. The mean time from arterial puncture to revascularization was 65 min. At 90 days, the symptomatic intracranial hemorrhage rate was 4%, all cause mortality was 32%, and 43/99 patients (43%) achieved functional independence with an mRS score of ≤2.

Conclusions The results suggest that the Separator 3D enables safe and effective revascularization of occluded large arteries in acute stroke intervention, leading to a high rate of functional independence at 90 days.

Statistics from

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.


One innovation recently introduced into the toolkit for acute ischemic stroke care is the stent retriever, a cylindrical grabber deployed within the thrombus to assist in its removal and restoration of blood flow. Effective rapid vessel revascularization is associated with improved functional outcomes and decreased mortality.1–4 From the Trevo versus Merci retrievers for thrombectomy revascularization of large vessel occlusions in acute ischemic stroke (TREVO) 2 open label randomized controlled trial for US device approval, the Trevo stent retriever was shown to improve 90 day good functional outcome, defined as a modified Rankin Scale (mRS) score of 0–2 (40% vs 22%) and Treatment in Cerebral Infarction (TICI) 2 or greater reperfusion (86% vs 60%) compared with the Merci device, with fewer vessel perforations (1% vs 10%).5 Both devices displayed similar 90 day mortality and symptomatic intracranial hemorrhage (sICH) rates.

From the Solitaire With the Intention For Thrombectomy (SWIFT) randomized controlled trial for US device approval, the Solitaire stent retriever exhibited better 90 day good neurological outcome, defined as an mRS score of ≤2, or equal to the pre-stroke mRS score if the pre-stroke mRS was >2, or National Institutes of Health Stroke Scale (NIHSS) score improvement of 10 points or more (58% vs 33%) and a higher rate of Thrombolysis in Myocardial Infarction 2 or 3 flow without sICH (61% vs 24%), compared with the Merci device.6 The early post-market experience and the North American Solitaire Stent Retriever Acute Stroke Registry (NASA) reported favorable revascularization results of 88% and 83%, respectively.7 ,8

The Penumbra System (Penumbra Inc, Alameda, California, USA) is a thrombectomy device specifically designed to remove clot through a system of reperfusion catheters and separators to facilitate direct aspiration.9–16 A stent-like three-dimensional (3D) binary nitinol device known as the Separator 3D was recently launched in Europe as a component of this system. Using the Separator 3D, direct aspiration and clot capture are combined to facilitate removal of the primary occlusion. Recently published initial feasibility and outcomes appear promising.15

The aim of this study was to determine the safety, revascularization, and clinical outcome of the Separator 3D as an additional component of the Penumbra System in the reperfusion of large vessel occlusions in acute ischemic stroke at our institutions.


Data were collected with approval of the ethics committees at both of our respective institutions, and informed consent was obtained from patients or their legal representatives in accordance with the Helsinki Declaration of 1975, as revised in 2005. Consecutive patients presenting with acute ischemic stroke due to large vessel occlusion between July 2012 and September 2013 were triaged on admission for potential thrombolysis or mechanical thrombectomy treatments in accordance with clinical practice guidelines. Patients were followed clinically for 90 days after the intervention.

Patient selection

Inclusion criteria were 89 years old or younger, NIHSS score of ≥5, and treatment time within 8 h of symptom onset. When time of onset was uncertain, patients were eligible if there was salvageable brain parenchyma depicted on CT perfusion imaging based on a cerebral blood volume/blood flow mismatch concept. Patients could present with any brainstem syndrome. Exclusion criteria included the presence of cerebral hemorrhage or infarct region more than one-third of the middle cerebral artery (MCA) territory.

Revascularization procedure

According to standard practices of the governing bodies of the respective institutions, patients had immediate CT angiography without postponing appropriate IV recombinant tissue plasminogen activator (rtPA) therapy. In cases of occlusion of large arteries, mechanical thrombectomy was performed. A maximum of four clot removal attempts were made. All proximal high grade internal carotid artery (ICA) stenoses (≥70%) identified were stented prior to the thrombectomy procedure. At center A, stented patients were given IV eptifibatide for 24 h to prevent in-stent thrombosis, as its short half-life would allow for emergency decompressive craniectomy. After intracranial hemorrhage (on control CT) and decompressive craniectomy were ruled out, therapy was switched to a secondary prophylaxis of stent occlusion based on a dual antiplatelet inhibition: under continuous IV eptifibatide, patients received a loading dose of 300 mg clopidogrel and 500 mg aspirin (ASA). Patients receiving daily ASA medication prior to treatment did not receive an ASA loading dose. Eptifibatide was stopped 4 h after loading with clopidogrel (and ASA); patients continued with lifelong ASA (100 mg/day) and additional clopidogrel (75 mg/day) for 8 weeks. If Multiplate testing 72 h later resulted in insufficient efficacy of clopidogrel (so called non-responder), we used prasugrel (10 mg/day) for 8 weeks.

At center B, instead of eptifibatide, ASA 100 mg/day and clopidogrel 300 mg/day were started 24 h after the procedure and continued with ASA 100 mg/day for at least 3 months and clopidogrel 75 mg/day for 6 weeks. The thrombectomy procedure was performed under general anesthesia and has been described previously in detail.15 Briefly, the target arterial lesion was directly catheterized by a microcatheter following introduction of an aspiration catheter with a 6 F long femoral sheath or an 8 F guide catheter proximal to the thrombus. After deployment of the Separator 3D, the ensnared thrombus and retriever were withdrawn into an aspiration catheter (054, 5MAX, or 5MAX ACE; Penumbra Inc) under continuous aspiration either with the MAX pump or a 60 mL syringe. It was at the discretion of the operator to try direct aspiration through the aspiration catheter if his Separator 3D attempt had failed and the aspiration catheter was still in place.

Imaging and clinical assessment

A CT scan was performed within 24 h after the revascularization procedure to detect the presence of ICH. NIHSS scores were assessed on admission and discharge, and mRS scores were assessed at admission or pre-stroke and 90 days post-procedure. Good functional outcome was defined as 90 day mRS score ≤2. Successful reperfusion was defined as angiographic evidence of modified TICI (mTICI) scores of 2b or 3, as described in consensus recommendations published recently.17 ,18 Good neurological outcome at 90 days was defined as 90 day mRS score ≤2 or having a score equivalent to or better than the pre-stroke mRS score if the pre-stroke mRS was >2, or discharge NIHSS score improvement of 10 or greater.6

Statistical methods

Categorical variables were expressed with their frequency distributions and continuous variables as mean (SD) or median (IQR). To examine the distributions of 90 day mRS scores, shift analysis was used with the Cochran–Mantel–Haenszel/van Elteren's shift test.17 ,18 Shift analysis was performed by ordinal logistic regression, taking the six categorized mRS score (ie, mRS scores 5 and 6 combined) as a dependent variable.19 To compare mRS scores by reperfusion status or test differences among groups, the Fishers exact test and the Wilcoxon rank sum exact tests were used. A two sided p value of <0.05 was considered statistically significant. SAS software V.9.3 was used to perform all of the analyses.


Baseline data

From 6 December 2011 to 29 September 2013, 129 patients were treated with the Separator 3D, 86% (111/129) at center A and 14% (18/129) at center B. Mean age was 69.8±12.6 years (table 1). There were 69 women (54%), and target lesions were in the anterior circulation for 108 patients and the posterior circulation for 21 patients. In 9/108 anterior circulation treatments (8%) and in 2/21 posterior circulation treatments (10%) the respective operator performed a single direct aspiration after a failed or incomplete Separator 3D pass. In 7/11 cases this led to successful recanalization; 5/7 successful cases were performed in the second half of 2013 and the aspiration catheter used was the 5MAX ACE. The level of occlusion was: MCA in 62/129 cases (48%), including 14 M2 cases; ICA in 42/129 cases (33%), including 17 carotid T and 21 carotid L cases; cervical ICA–MCA in 4/129 cases (3%); and vertebrobasilar arteries in 21/129 cases (16%).

Table 1

Baseline characteristics and outcomes

Medical history included hypertension (85%), atrial fibrillation (52%), hyperlipidemia (41%), and diabetes mellitus (13%), with an initial median international normalized ratio of 1.07 and glucose concentration of 112.5 mg/dL. Time from symptom onset to admission averaged 155.2±123.0 min for 128 patients with data. Patients presented with an admitting median NIHSS score of 15 (IQR 12–19). Prior to treatment, all patients had NIHSS scores ≥5 (range 5–35). Between initial imaging and mechanical thrombectomy using the 3D, 100/129 patients (78%) had received initial IV rtPA according to national neurological guidelines, independent of mechanical thrombectomy treatment. Pursuant to proximal stenosis or occlusion, acute a priori stenting of the ICA was required in 9/38 cases (24%). In our dataset, duration of the intervention, defined as time from arterial puncture to reperfusion, averaged 65.0±43.5 min (median 57 min). Overall mean time from symptom onset to mechanical revascularization was 296.1±135.9 min (median 267.5 min), including two patients with only last well time known. For the cohort with an unambiguous time of symptom onset, mean onset to revascularization was 289.0±126.0 min (median 266 min).

Technical success

Pre-procedure, patients had mTICI scores of 0 or 1 in all cases. Post-procedure, the overall revascularization rate of the Separator 3D to mTICI 2b–3 was 74% (table 1) and to mTICI 3, 51% (23% had an mTICI score of 2b, 9% had an mTICI score of 2a, 5% had an mTICI score of 1, and 6% had an mTICI score of 0); in the remainder (6%), direct aspiration alone after a failed or only partially successful 3D attempt led to successful recanalization. Revascularization grading by the target arterial lesion is shown in table 2. A score of mTICI 2b–3 was achieved in 73% of anterior circulation occlusions and in 81% of posterior circulation occlusions. Of note, 10/17 (59%) carotid T occlusions attained mTICI 3 flow. Use of the Separator 3D achieved complete reperfusion (mTICI 3) with one pass in 47 of 129 cases (36%); the median number of 3D passes for maximal reperfusion was 2.0. Median direct aspiration attempts was 3.0; as mentioned previously, the reason for this difference is that the operators made 11 attempts of aspiration at the thrombus with the aspiration catheter alone after a failed or only partially successful 3D attempt. In 7/11 cases (5 anterior circulation and 2 posterior circulation) this led to successful recanalization but they were not counted as a 3D success (intention to treat approach).

Table 2

Selected outcome measures by target arterial lesion

In the study, the 3D was used adjunctively in 7 patients (5%) following unsuccessful revascularization with other devices (Solitaire clot retriever (ev3-Covidien, Irvine, California, USA), pREset (Phenox, Bochum, Germany), Revive (Codman and Shurtleff, Raynham, Massachusetts, USA)). In one case, the patient was initially treated with other devices; three attempts were performed using the Solitaire followed by one attempt using the Revive, followed by one 3D attempt, which resulted in revascularization to mTICI 3. Four patients were initially treated with two passes of the Solitaire and then the 3D, with varying resultant flow (2×mTICI 3, 1×mTICI 1, 1×mTICI 0). One patient was treated with one pass of the pREset (mTICI 0). In one patient, aspiration catheter placement occurred only once due to the difficult access to the target vessel (post-procedure mTICI 0); the following attempts were performed with the aspiration catheter positioned in the proximal intracranial ICA. In all other 122/129 (95%) patients, the 3D was used as the primary device. In these 122 cases a rescue device was deployed in only 6 cases (4.9%).

Clinical outcome and complications

At discharge, 43/109 (39%) patients had a favorable neurological outcome, defined as an improvement in NIHSS score of ≥10 points. Median NIHSS score was 6, and 35% of patients (39/111) had an mRS score of 0–2 on discharge. Overall mean length of hospital stay was 9.1 days. For anterior occlusion patients, mean hospitalization was 9.7 days and for posterior occlusion patients, 8.1 days. After the intervention, 5 (4%) patients experienced sICH, and procedural complications noted included subarachnoid hemorrhage (3%), treatment related embolization outside the target downstream territory (embolization to new territory (ENT), 4%), vasospasm (2%), and vessel dissection (2%) (table 3). For anterior occlusion patients, sICH was noted in 5 patients, ENT in 4, vasospasm in 2, and vessel dissection in 1. For posterior occlusion patients, sICH was observed in none, ENT in 1, vasospasm in none, and vessel dissection in 1.

Table 3

Procedure related complications

At 90 days, 43% (43/99) of patients showed good functional outcome (mRS ≤2), and all cause mortality was 32% at 90 days. NIHSS data at admission are needed to assess the stroke severity of the population treated, and are helpful to place into perspective the NIHSS discharge data (figure 1, table 4). In anterior occlusions, 45% (39/87) of patients achieved good functional outcome; in posterior occlusions, 33% (4/12) showed good functional outcome (table 2). In contrast with the dichotomized mRS comparison, shift analysis shows the changes in distributions over the full range of 90 day mRS scores taking into account activity at presentation. Shift analysis confirmed that patients showed significant reductions in disability comparing 90 day post-procedure to admission mRS scores (Cochran–Mantel–Haenszel/van Elteren=22.4, p≤0.0001, OR 3.5 (2.1, 5.8)). All cause 90 day mortality was 32/99 (32%). Furthermore, reperfused patients demonstrated increased freedom from disability (mRS median 3 vs 6 (TICI 2b–3 vs 0–2a), Wilcoxon rank sum=1344.0, p value ≤0.0116; 90 day mRS 0–2, 48.7% vs 23.8%, Fishers exact test p value=0.0493).

Table 4

Distribution of modified Rankin Scale scores at 90 days

Figure 1

Distribution of National Institutes of Health Stroke Scale (NIHSS) scores at baseline and discharge.


The aim of this study was to assess the clinical outcome of acute stroke therapy with the Separator 3D in over 125 patients at two institutions. Compared with previously published data, treatment using the 3D device showed that the reperfusion rate (mTICI 2b–3) of 74% was in the range reported for other retrievers, with similar complication rates and procedure times in the POST, SWIFT, TREVO 2, and Solitaire post-market registries5–9 (table 5). Utilization of a rescue device was very rare in this series (6/122 or 4.9%). Five ENT were observed in conjunction with the Separator 3D. Data on ENT are yet limited in the stroke literature and the concomitant aspiration concept (ie, lesional aspiration vs proximal aspiration with balloon guide occlusion) is not always given in each study.20 This is also true for the SWIFT cohort. Nevertheless, compared with available SWIFT data on ENT with a reported rate of 1 in 144 treatments (0.7%), the current rate is higher. This shows that we urgently need more data in order to improve our technical approaches.6 ,21

Table 5

Comparison of Separator three-dimensional (3D) study with Solitaire post-market studies and SWIFT trial

The cohort who achieved functional independence (mRS ≤2) at 90 days (43%) was slightly higher than the SWIFT study data and consistent with expectations from NASA6–9 (table 5). Among carotid T and carotid L occlusions, patients treated with the 3D achieved 71% reperfusion while 22% achieved good functional 90 day outcome. In Bhatia et al's2 study on recanalization rates with IV rtPA, similar results were documented in carotid T and carotid L patients who achieved 43.5% reperfusion and 17% good outcome. Mortality was significantly higher in the current 3D study compared with the SWIFT study, but not different from the two post-market studies.6–9 Factors which may have contributed to the difference include the number of vertebrobasilar patients in the 3D study (16% compared with 2% in the SWIFT study), and the level of occlusion (33% ICA in the 3D study (29.5% in the carotid T/L) compared with 21% ICA in the SWIFT study (breakdown not reported)).6 Within the carotid T cohort, 3 of 12 (25%) patients suffered from preceding high grade stenosis. The relatively high proportion of carotid T and carotid L patients in this study may have had an impact on the overall mRS score compared with other studies.

Revascularization to mTICI 3 with the Separator 3D was accomplished in 51% of cases, exceeding the 14% rate observed in the TREVO 2 trial (p<0.0001) in a cohort with only half the percentage (8%) of vertebrobasilar occlusions, or the rate of 40% in the NASA registry (p=0.007), comprising 10% vertebrobasilar occlusions.7 Additionally, the rate of ICA occlusions was higher in this study compared with TREVO 2, NASA, and the early post-market experience with the Solitaire stent. Moreover, the rate of sICH was significantly higher in the Solitaire post-market experience studies (p=0.04 and p=0.004)5 ,7 ,8 (table 5).

In light of the Interventional Management of Stroke (IMS) III endovascular treatment subgroup findings, the current 3D experience is consistent with the IMS III Penumbra rates of 85% revascularization and 33% 90 day mRS 0–2 in ICA and M1 occlusions.22 Shift analysis may be more sensitive than dichotomized mRS analysis to evaluate variables affecting outcome by examining the change in outcome distributions over the entire range of measured outcomes and also may increase study power to detect a treatment effect.19 As seen in our study of the 3D, good outcomes were demonstrated in both dichotomized and shift mRS analyses (dichotomized OR=4.7 (2.5, 9.0); shift analysis Cochran–Mantel–Haenszel/van Elteren=22.4, p≤0.0001). Moreover, reperfusion after treatment with the 3D was correlated with improved functional outcomes.

Limitations of this study included the two center retrospective investigation designed without a concurrent control group, lack of core laboratory adjudication, and possible selection bias for thrombolysis refractory patients or those who present with larger thrombi. Although these clinical results are positive, subsequent studies should include a randomized control study design.


Results from 129 patients in this two center study support the fact that the Separator 3D enables safe and effective revascularization in patients with large vessel occlusion. These clinical results, in addition to a low rate of rescue therapy, are encouraging, while complication rates were comparable to similar devices.


The authors would like to thank Dr Sophia S Kuo for her valuable assistance in manuscript preparation.



  • DB and AK contributed equally to this work.

    Some of the data were presented at ESC 2014, Nice (Dr Kowoll, Bochum), at ESC 2013, London (Dr Mpotsaris, Cologne), and at ABC WIN Seminar 2014, Val d'Isere (Dr Behme, Göttingen).

  • Contributors DB, AK, JW, and WW: interventions, writing, data sampling, and editing. AM: interventions, data sampling, and editing. CH, LH: interventions and data sampling.

  • Competing interests DB: minor speaking honoraria from Penumbra. AM: minor speaking honorarium and travel grant from Penumbra; travel grant from Covidien. JW and WW: minor speaking honoraria and travel grant from Penumbra.

  • Ethics approval Retrospective analysis from a prospectively captured database was gathered on consecutive patients undergoing acute ischemic stroke thrombectomy with the Separator 3D at the Recklinghausen and St Gallen institutions, using an institutional review board approved protocol (#4849-13 for the first institution).

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