Patients
Patients were either admitted directly to the Karolinska University Hospital or transferred from one of six referring hospitals within the region. In addition, patients were occasionally sent from more distant hospitals outside the primary catchment area.
All patients arriving from referring hospitals had an unenhanced computed tomography scan (CT) performed before transferring to the Karolinska University Hospital. In the later part of the study, most patients also had a CT angiography (CTA) before transfer. If a large infarction was revealed on the CT scan, with a clear hypodensity exceeding 1/3 of the MCA territory or with extensive brain stem involvement, the patient was not treated with IVT and not transferred to the Karolinska. If the patient was eligible for IVT, the infusion (0.9 mg/kg) was initiated for all patients except for some with basilar artery occlusion, which was in accordance with the stroke protocol at the time. In the early years of the study, i.e., before 2010, the result of the intravenous treatment was awaited, and the patient was re-evaluated after approximately 1 h, then transfer was organized and the patient was sent to the Karolinska. From 2010 and onwards, the patient was instead transferred immediately after initiation of IVT, with ongoing treatment in the ambulance. If a CTA had been performed, an additional prerequisite for transfer was the presence of a large vessel occlusion [principally internal carotid artery (ICA), middle cerebral artery (MCA); segment M1 or M2 or basilar artery (BA)]. Patients that were non-eligible for IVT were transferred if no large infarct was detected and endovascular treatment could be initiated within 8 h after symptom onset. In addition, only patients with a NIHSS score of 8–30 (from 2010 changed to 6–25) without severe comorbidities were considered. Exception was made for a few patients with NIHSS <6 if the symptoms included aphasia. The reason for the upper limit of NIHSS score was that, according to our experience, patients with such high scores also had large manifest infarctions and would gain little from endovascular treatment.
All patients arriving from outside hospitals were transferred directly to the Department of Neuroradiology, thus bypassing the Emergency Room (ER), where representatives for the Departments of Neurology, Anesthesiology, and Neuroradiology met up. A quick neurological examination was performed after which a non-enhanced CT was repeated and complimented with a CT perfusion (CTP) and a CTA. If a CTA had been executed already at the referring hospital, it was repeated only if the patient had dramatically improved clinically.
Patients admitted directly to the Karolinska University Hospital were examined clinically in the ER and brought to the CT scanner with minimal time delay where the same “stroke protocol” was performed (CT, CTA, and CTP). If these patients were eligible for IVT, it was initiated immediately after the scanning.
All patients were subsequently brought directly to the angio suite, some with ongoing infusion of alteplase, where preparation for thrombectomy began. After having read the scans, the decision whether to proceed with thrombectomy or not was made jointly by the stroke neurologist and the neurointerventionist on call. Mechanical thrombectomy was considered for all patients with remaining symptoms, a large vessel occlusion and with no large infarct as interpreted from the CT, CTP, and CTA source images. In order to reduce the risk of reperfusion hematomas, ongoing infusions of alteplase were terminated when a guide catheter had been positioned, and it was likely that the thrombectomy would be technically feasible. The angiography suite preparations were aborted in case of clinical improvement or if the presence of an already large infarct could be determined based on the CT, the CT source images, and the CTP.
The study was approved by the research ethics committee at the Karolinska Institute, and no individual consent was collected before inclusion in the study.
Thrombectomy technique
The Merci retrievers (Concentric Medical/Stryker Neurovascular, Mountain View, California, USA), including the X5, X6, L5, and L6 devices, as well as the V-series, were used in the beginning of the study period. Since February 2009, different so-called stent retrievers were utilized: Solitaire FR (ev3 Endovascular, Plymouth, MN, USA/Covidien, Irvine, CA, USA), Trevo (Concentric Medical/Stryker Neurovascular), IRIIS/Capture (Mindframe, Irvine, CA, USA/Covidien), and Opticell (MindFrame/Covidien).
For most cases, the standard thrombectomy procedure began with placing the tip of an 80-cm long introducer sheath, mostly an Arrow 8F (Arrow International, PA, USA) in the common carotid artery, just below the bifurcation, in which an 8F Merci balloon guide catheter (Concentric Medical/Stryker Neurovascular) was inserted with its tip positioned in the proximal ICA. In case of a BA occlusion, a 6F Envoy (Cordis Corporation International, Johnson and Johnson Medical NV/SA, Waterloo, Belgium) or a 7F Guider (Boston Scientific, Natick, Massachusetts, USA/Stryker Neurovascular) was most of the time used as guide catheter.
After confirming the presence of the thromboembolus on an angiogram, a microcatheter was advanced through the thrombus with the aid of a microguidewire. After release of the stent retriever, centered at the obstructive blood clot, it was left in place for approximately 5 min (3–10) before the actual thrombectomy was performed. The balloon was inflated with simultaneous aspiration in the guide catheter to create flow reversal in the ICA. For early cases in the series, where the Merci device was used, the device was positioned distal to the thromboembolus.
For patients with so-called tandem occlusions in the anterior circulation, i.e., a proximal carotid severe stenosis or occlusion due to acute dissection or atherosclerosis, in addition to the MCA occlusion, the procedure was modified to include angioplasty or carotid stenting as well as to sometimes contain the use of an intermediate catheter, mostly the so-called “distal access catheter” (DAC; Concentric Medical/Stryker Neurovascular).
For a few patients, the underlying cause for the large vessel occlusion was an intracranial stenosis. Also, for these patients, the standard procedure was altered to include angioplasty and sometimes stenting. Placement of a stent, extra- or intracranially, was, however, avoided when possible due to the need for anti-aggregation, which in the context of a potential acute ischemic infarct may impose a risk for a severe reperfusion hematoma. If a stent was inserted, the patient received half the recommended bolus dose (0.125 mg/kg body weight) of abciximab (Reo-Pro; Eli Lilly, Sweden), but no infusion, followed by acetylsalicylic acid (Trombyl; Pfizer AB, Sollentuna, Sweden; bolus of 300 mg and then 75 mg daily for a minimum of 6 months) and clopidogrel (Plavix; Bristol-Myers Squibb AB, Bromma, Sweden; bolus of 300 mg and then 75 mg daily for a minimum of 3 months) after approximately 24 h when a routine follow-up CT scan did not reveal a significant hematoma.
As a supplement to the thrombectomy, for treating small distal emboli, alteplase was sometimes infused intra-arterially together with careful microcatheter and microguidewire manipulations. The dose of alteplase was maximized to 10 mg in case the patient had received IVT, whereas up to 20 mg could potentially be administered if no IVT had been given. Other adjunctive therapies included 2–5 mg of intra-arterial nicardipine (Cardene; PDL BioPharma Inc., Fremont, CA, USA) during the procedure if there was a tendency to focal vasospasm, and in the beginning of the series, acetylsalicylic acid (bolus of 300 mg and then 75 mg daily) and low molecular weight heparin (dalteparin sodium, Fragmin; Pfizer AB, 2,500 IU twice daily) were used after the procedure in non-stented patients to avoid re-occlusion. For patients later in the series (from 2010), usually, no anti-aggregation or low molecular weight heparin was administered post-procedure since re-occlusion in the absence of a stent was found not to be a significant problem.
Baseline and outcome data
Baseline and outcome data were retrieved from computerized records at the hospital. National Institute of Health Stroke Scale (NIHSS) score [
17] and the modified Rankin Scale (mRS) score [
18] were used as outcome parameters. A stroke neurologist assessed the NIHSS score before and after each intervention. In case NIHSS score pre- or post-treatment was not specifically noted in the records, but the criteria was described in the text, we assessed the score based on the text. In four cases of anterior circulation stroke, it was not possible to access enough information from the charts to assess a pre-treatment score, and in 13 patients, it was impossible to establish a post-treatment score. Information on the mRS score was for the majority (59 %) of patients obtained at follow-up 3 months after the procedure. For a minority, the scoring was postponed until 6 months post-treatment and for 9 % of the cases, it was delayed even more until maximum of 12 months after the thrombectomy procedure. The mRS scoring was done by an independent stroke neurologist or by a stroke research nurse certified for such evaluations, who also, in a few cases with missing mRS score estimated the value from the description in the charts. Favorable outcome was defined as a mRS score of 0–2, corresponding to independence in activities of daily living.
All the angiographies performed at the time of thrombectomy were re-evaluated by two of the authors (ÅKS and TA) to obtain definitive information on pre- and post-procedure mTICI score (thrombolysis in cerebral infarction) [
19,
20]. The mTICI score was defined as follows: 0 = no perfusion; 1 = penetration, but no distal branch filling; 2a = perfusion with incomplete (<50 %) distal branch filling; 2b = perfusion with incomplete (≥50 %) distal branch filling; and 3 = full perfusion with filling of all distal branches. A post-procedure mTICI score of ≥2b was considered to represent a successful revascularization. One of the authors (ÅKS) also re-examined all pre- and post-procedure unenhanced CT scans to obtain ASPECTS scores (Alberta Stroke Program Early CT Score) [
21] and to determine the presence of any post-interventional hemorrhage. Hemorrhagic transformation was classified into hemorrhagic infarction (HI) type 1 and 2 and parenchymal hematoma (PH) type 1, type 2, and remote parenchymal hematoma (PHr-2) in accordance with the ECASS definitions [
22]. We defined symptomatic intracranial hemorrhage according to the SITS-MOST definition: a PH-2 (blood clot exceeding 30 % of the infarcted volume with significant space occupying effect) or subarachnoid hemorrhage on the post-treatment imaging scan leading to a decline in NIHSS of ≥4 points or causing death (mRS 6) within 36 h [
23].
For the statistical analysis, we used SAS statistical software (V.9.1; SAS Institute Inc.). Ninety-five percent confidence intervals (95 % CI) were calculated assuming a binomial distribution. For testing independence between outcome and exposure variables, we used Pearson’s exact chi-squared tests. Observations with missing values were excluded. A p value <0.05 was considered statistically significant.
Since a delayed assessment of mRS score may lead to an overestimation of patients with good outcome, we assessed the potential impact of the delayed mRS assessments by doing sensitivity analyses where an extra point was added to the mRS score for all patients who were assessed after more than 6 months.
Test for trends were done by analyzing categories as continuous variables in logistic regression using the PROC LOGISTIC statement in SAS™.