Article Text
PDF
Abstract
Background and purpose To determine the initial factors, including patient characteristics, stroke etiology and severity, time factors, and imaging findings, that could affect the clinical outcome of patients with acute ischemic stroke (AIS) caused by basilar artery occlusion (BAO) where successful recanalization was achieved via mechanical thrombectomy.
Methods Between March 2011 and December 2014, 35 patients with AIS caused by BAO received MRI/MR angiography-based mechanical thrombectomies, and recanalization was achieved with a Thrombolysis In Cerebral Infarction score of >2b. The patients were divided into a good outcome group (n=19), defined as those with a modified Rankin Scale (mRS) score of 0–2 at 3 months after stroke onset, and a poor outcome group (n=16), defined as a mRS score of 3–6. The differences between the groups were analyzed.
Results Initial National Institutes of Health Stroke Scale (NIHSS) score (good vs poor: 17.9±8.9 vs 27.6±8.5, p=0.003), posterior circulation Acute Stroke Prognosis Early CT Score (pc-ASPECTS) based on initial diffusion-weighted images (DWI) (good vs poor: 7.8±1.6 vs 5.4±1.8, p=0.001), pc-ASPECTS based on contrast staining on the post-thrombectomy control CT (good vs poor: 9.2±1.5 vs 6.3±2.2, p<0.001), and presence of contrast staining in the brainstem on that CT (good vs poor: 15.8% vs 81.6%, p<0.001) were significantly different between the groups.
Conclusions Patients with AIS caused by BAO with a lower initial NIHSS score, fewer lesions on initial DWI, and less contrast staining on the post-thrombectomy control CT have higher probabilities of a good clinical outcome after successful recanalization via a mechanical thrombectomy.
- MRI
- Stroke
- Thrombectomy
Statistics from Altmetric.com
Introduction
Recently published randomized controlled trials of mechanical thrombectomy with or without intravenous (IV) thrombolytic therapy in patients with acute ischemic stroke (AIS) have shown its greater efficacy compared with IV thrombolytic therapy alone.1–5 Improved reperfusion rates, early neurologic recovery, and good long-term functional outcomes have been reported. However, the main concern about mechanical thrombectomy is that even a ‘successful’ recanalization does not always lead to a good outcome (‘futile recanalization’6), and those studies were performed in patients with occlusion in the anterior circulation.1–5
AIS due to basilar artery occlusion (BAO) represents 20% of all ischemic strokes and its mortality rate remains the highest, with poorer functional outcomes, among all the stroke subtypes.7 Indeed, more than 85% of patients with AIS due to BAO die and, even after treatment, death and dependency rates exceed 70%.7 Recent small series with various mechanical thrombectomy devices used to treat BAO reported high recanalization rates with a relatively high rate of good outcomes at 3 months and low mortality rates.8–15 This year the results of patients with BAO in the Endovascular Stroke Treatment (ENDOSTROKE) study were announced.16 This is the largest series of endovascular treatment for AIS due to BAO to date. However, large randomized trials have not been performed.
In the present study we retrospectively analyzed data of 35 patients with AIS caused by BAO who received MRI/MR angiography (MRA)-based mechanical thrombectomies. All patients included in this study were treated via mechanical thrombectomy using a Solitaire stent (Solitaire AB/FR; ev3, Irvine, California, USA) or the direct aspiration first-pass (ADAPT) technique using the Penumbra reperfusion catheter (Penumbra/Penumbra MAX; Penumbra, Alameda, California, USA) and achieved successful recanalization (Thrombolysis In Cerebral Infarction (TICI) score 2b or 3). To determine which initial factors—including patient characteristics, stroke etiology and severity, time factors, and imaging findings—could affect the clinical outcome of these patients, we analyzed the differences between the good and poor outcomes groups, categorized based on the modified Rankin Scale (mRS) score at 3 months after stroke onset.
Methods
Patients
Between March 2011 and December 2014, 46 patients presented at our institution with AIS caused by BAO and received mechanical thrombectomy within 8 h of stroke symptom onset or had stroke of unclear onset and presented within 24 h from the ‘last normal’ time. We were able to open the occluded basilar artery (BA) in 38 of these patients (82.6%) to a TICI score of ≥2b. However, three of the 38 patients were excluded because they could not undergo MRI/MRA due to poor neurological status and thus received CT/CT angiography-based (not MRI/MRA-based) mechanical thrombectomy.
Acute stroke management protocol
A neurological evaluation was performed upon presentation in the emergency room. The National Institutes of Health Stroke Scale (NIHSS) was administered by at least two stroke neurologists. If a patient was in a complete coma, we recorded a NIHSS score of 36. Non-contrast brain CT scans were performed immediately after presentation to rule out intracerebral hemorrhage (ICH). If no ICH was evident on CT scans obtained within 3 h of symptom onset (within 4.5 h since January 2014), we began treatment with recombinant tissue plasminogen activator (rt-PA) (Actylase; Boehringer Ingelheim, Basel, Switzerland) to achieve IV thrombolysis. The indications/contraindications for IV thrombolysis followed the guidelines for early management of adults with ischemic stroke.17 After beginning rt-PA, we performed stroke setting MRI using a 1.5 T magnet (Magnetom Avanto; Siemens Medical Solutions, Erlangen, Germany) or a 3.0 T magnet (Philips Ingenia; Philips Medical Systems, Best, The Netherlands). The MRI procedure consisted of b0 and b1000 diffusion-weighted images (DWIs), susceptibility-weighted images (SWIs), fluid attenuated inversion recovery (FLAIR), gadolinium (gadobutrol, Gadovist; Schering, Berlin, Germany)-enhanced T1WI, gadolinium-enhanced extracranial and intracranial MRA, and perfusion MRI. To reduce MRI scanning time we downgraded the image resolution. Consequently, the MRI scanning time was approximately 13 min using 1.5 T MRI and 15 min using 3.0 T MRI. If a BA occlusion was evident on MRA, we prepared the patient for intra-arterial thrombectomy after obtaining informed consent. In patients with stroke of unclear onset, the decision regarding whether to perform a mechanical thrombectomy was discussed between the neurointerventionist and neurologist based on MRI findings. IV thrombolysis was not performed in patients with suspected AIS who presented >3 h after symptom onset (>4.5 h after January 2014). Informed consent for the performance of an intra-arterial thrombectomy was obtained before each procedure in the presence of at least two legal representatives. This study was approved by our institutional review board.
We have reported on the technical aspects of these patients previously,15 so only an abbreviated explanation of the mechanical thrombectomy technique used is given here. The first 11 consecutive patients in this study, from March 2011 to December 2011, underwent mechanical thrombectomy using the Solitaire AB stent. However, in January 2012 the Korean Food and Drug Association banned the use of the Solitaire AB stent as a thrombectomy device. At that time, the Solitaire FR stent for thrombectomy had not been introduced in Korea. In January 2012 we replaced the first-line Solitaire stent thrombectomy device with the Penumbra reperfusion catheter. Since then, we have performed mechanical thrombectomy using a conventional Penumbra reperfusion catheter with the ADAPT technique. Since January 2014 we have used the Penumbra Max catheter instead of the conventional Penumbra catheter. Since January 2012 the Solitaire stent has only been used as a rescue method if we failed to open using a Penumbra catheter. TICI scores were estimated by the final angiography after thrombectomy. A control CT scan was performed immediately after the thrombectomy to detect hemorrhagic complications. We focused on contrast staining as well as hemorrhagic complications on that CT. We defined contrast staining as high density (>40 HU) on the control CT that conformed to the anatomic boundaries of normal structures, lacked surrounding edema, and had no mass effect, following Amans et al.18
Data collection and statistical analysis
We collected data on baseline characteristics, stroke risk factors, initial neurological status, mechanical thrombectomy results, and clinical outcomes at discharge and at 3 months after stroke onset using the mRS score. Additionally, we calculated the posterior circulation Acute Stroke Prognosis Early CT Score (pc-ASPECTS)19 using lesions on the initial DWI (figures 1 and 2). The pc-ASPECTS is a 10-point grading system to quantify ischemic changes in the posterior circulation regions, where 10 points indicate no ischemic damage and zero points indicate extensive damage of the posterior circulation.19 We applied DWI to this scoring system by subtracting one point for each high signal intensity lesion observed in the left or right thalamus, cerebellum, or posterior cerebral artery, and two points for each high signal intensity lesion observed in any part of the midbrain or pons.19 Furthermore, we applied the same scoring system using contrast staining locations on the control CT (figures 1 and 2). We also recorded the presence of contrast staining in the brainstem using the same CT.
Actual images of one patient classified as having a good outcome. The patient presented to our hospital because of decreased mentality at 3 h 53 min after symptom onset. The initial National Institutes of Health Stroke Scale (NIHSS) score was 24. (A) Initial MR-diffusion-weighted images showed no infarction lesion in the posterior circulation territory. Thus, the MR posterior circulation Acute Stroke Prognosis Early CT Score (pc-ASPECTS) was 10. (B) Initial left vertebral angiogram showing the distal portion of the basilar artery completely occluded. (C) Post-thrombectomy left vertebral angiogram showing the occluded basilar artery completely recanalized (TICI 3). (D) Control CT obtained immediately after thrombectomy shows no contrast staining in the brain parenchyma in the posterior circulation territory. Thus, the CT pc-ASPECTS is also 10. The patient recovered completely.
Actual images of one patient classified as having a poor outcome. The patient presented to our hospital because of decreased mentality at 1 h 17 min after symptom onset. The initial National Institutes of Health Stroke Scale (NIHSS) score was 36. (A) Initial MR-diffusion-weighted images (MR-DWI) showed multiple infarctions in both cerebellar hemispheres, the pons, midbrain, left thalamus, and left occipital lobe. Thus, the MR posterior circulation Acute Stroke Prognosis Early CT Score (pc-ASPECTS) was 2. (B) Initial left vertebral angiogram showing stenosis in the ostium of the left vertebral artery. We therefore approached the right side initially, but the right vertebral artery terminated as posterior inferior cerebellar artery endings. We performed balloon angioplasty in the left vertebral artery ostium just before obtaining this image. The flow to the basilar artery is poorly visible due to occlusion more distally. (C) Post-thrombectomy left vertebral angiogram showing the occluded vertebrobasilar artery completely recanalized (TICI 3). (D) Control CT obtained immediately after thrombectomy shows multiple contrast staining in the brainstem and parenchyma, where the lesions are located in the initial MR-DWI. Thus, the CT pc-ASPECTS is also 2. This patient died 1 day after the thrombectomy due to loss of brainstem function.
We divided the 35 included patients into a good outcome group (n=19), defined as a mRS score of 0–2 at 3 months after stroke onset, and a poor outcome group (n=16), defined as a mRS score of 3–6. Differences between the good and poor outcome groups were analyzed using the SPSS software (V.19.0; SPSS, Chicago, Illinois, USA). We used the unpaired t test, χ2 test, and Fisher exact test, as appropriate. p Values ≤0.05 were considered to indicate statistical significance. To identify independent variables associated with good outcomes, a backward stepwise logistic regression analysis was performed including all variables with p≤0.05 in the univariate analyses.
Results
Baseline characteristics, stroke risk factors, and procedural and clinical results are summarized in table 1.
Comparisons between good outcome and poor outcome groups
Patient data
In the good outcome group, the mean age of the 19 patients was 65.7±9.3 years (range 47–83; median 68) and the male:female ratio was 10:9. The mean time from stroke onset to hospital presentation was 207.6±143.9 min (range 30–555; median 160) and the initial NIHSS score was 17.9±8.9 (range 5–34; median 14). Seven of the 19 patients (36.8%) underwent IV thrombolysis (door-to-needle time 44.4±37.5 min (range 12–92; median 18).
In the poor outcome group, the mean age of the 16 patients was 70.3±9.8 years (range 52–85; median 70) and the male:female ratio was 13:3. The mean time from stroke onset to hospital presentation was 143.5±126.8 min (range 30–569; median 109) and the initial NIHSS score was 27.6±8.5 (range 8–36; median 30). Nine of the 16 patients (56.3%) underwent IV thrombolysis (door-to-needle time 43±23.0 min (range 19–90; median 39).
The good outcome group had a significantly lower NIHSS score at presentation than the poor outcome group (p=0.003).
Angiographic data
In the good outcome group, the time from stroke symptom onset to femoral puncture was 352.6±260.4 min (range 120–1240; median 289) and the door-to-femoral puncture time was 119±70.5 min (range 67–318; median 91). All 19 patients had TICI scores of zero on initial angiography. The femoral puncture to recanalization time was 66.1±36.4 min (range 28–140; median 50). Four of the 19 patients (21.1%) achieved TICI scores of 2b and the remaining 15 patients (78.9%) had scores of 3. The time from stroke symptom onset to achieving maximum recanalization was 418.7±258.7 min (range 163–1270; median 382).
In the poor outcome group, the time from stroke symptom onset to femoral puncture was 318.1±154.7 min (range 125–745; median 286) and the door-to-femoral puncture time was 110±56.0 min (range 60–240; median 92). All 16 patients had TICI scores of zero on initial angiography. The femoral puncture to recanalization time was 69.6±43.2 min (range 30–201; median 58). Six of the 16 patients (37.5%) achieved TICI scores of 2b and the remaining 10 (62.5%) had scores of 3. The time from stroke symptom onset to achieving maximum recanalization was 318.1±154.7 min (range 125–745; median 286).
Comparisons of the two groups showed no significant differences in these variables.
Clinical outcomes
In the good outcome group, the mean NIHSS score on discharge was 3.7±3.7 (range 0–11; median 3) and the mRS score was 1.6±1.1 (range 0–3; median 1). At 3 months, the mRS score had improved to 0.6±0.8 (range 0–2; median 0) (figure 3).
Modified Rankin Scale (mRS) scores at discharge and at 3 months after stroke onset. At discharge the mRS scores of patients are scattered. However, at 3 months after symptom onset the mRS score of the patients had moved to each end. We call this an ‘all or nothing’ phenomenon in patients with acute ischaemic stroke caused by basilar artery occlusion.
In the poor outcome group, the mean NIHSS score on discharge was 23.1±10.9 (range 5–36; median 24) and the mRS score was 4.7±1.3 (range 1–6; median 5). Four patients died during their hospital stay. The causes of death were infarction progression and loss of brainstem functions in all four. At 3 months the mRS score had worsened to 5.4±1.1 (range 3–6; median 6) (figure 3). Seven more patients died after discharge and prior to the 3-month follow-up. Thus, the mortality rate at 3 months after stroke onset was 31.4%. The mRS scores at discharge of the six patients who died after discharge were 4 in one patient and 5 in five patients. The causes of death were complications of stroke, such as pneumonia, in all patients. One other patient whose mRS score was 1 at discharge died due to cardiac problems.
Imaging data
In the good outcome group, the door-to-MRI time was 66.6±53.3 min (range 12–214; median 49.0). The pc-ASPECTS based on initial DWI was 7.8±1.6 (range 5–10; median 8) and the pc-ASPECTS using contrast staining locations on the control CT was 9.2±1.5 (range 4–10; median 10). Contrast staining in the brainstem was observed in three patients (15.8%).
In the poor outcome group, the door-to-MRI time was 52.7±32.5 min (range 22–140; median 44.5). The pc-ASPECTS based on initial DWI was 5.4±1.8 (range 2–9; median 6) and pc-ASPECTS on the control CT was 6.3±2.2 (range 2–10; median 6). Contrast staining in the brainstem was observed in 13 of 16 patients (81.6%).
The good outcome group showed significantly higher pc-ASPECTS based on initial DWI (p=0.001) and pc-ASPECTS using contrast staining locations on the control CT (p<0.001) and contained fewer patients with contrast staining in the brainstem (p<0.001) compared with the poor outcome group.
Backward stepwise logistic regression identified the pc-ASPECTS based on initial DWI as a predictor of good outcomes (p=0.044).
Discussion
The main findings of this study are that initial NIHSS score, pc-ASPECTS based on initial DWI, pc-ASPECTS based on contrast staining on the control CT, and the presence of contrast staining in the brainstem are initial factors that affect the clinical outcome, measured by successful recanalization via mechanical thrombectomy, in patients with AIS caused by BAO. However, because patients with more lesions on DWI usually show higher NIHSS scores, and given that contrast staining on the control CT represents tissue that was already damaged by ischemic insult,18 it can be concluded that lesions on the initial DWI are the most important predictor of the clinical outcome in patients with AIS caused by BAO, as shown in the backward stepwise logistic regression analysis.
The Basilar Artery International Cooperation Study (BASICS) is the largest prospective observational study published to date of patients who presented with an acute symptomatic and radiologically confirmed BAO.20 In that study, older age, absence of hyperlipidemia, history of a prior stroke, higher NIHSS score, and longer time to treatment (especially >6 h) were predictors of poor outcomes in patients with AIS caused by BAO.21 In the ENDOSTROKE study, age, hypertension, NIHSS score, collateral status, and the use of MRI prior to endovascular treatment predicted clinical outcome, and the latter three were independent predictors in multivariate analysis.16 Older age, time to treatment, and stroke severity are already well-established risk factors for poor outcomes after stroke.22 However, in the present study, only higher NIHSS score at presentation (stroke severity) was significantly different between the groups; the proportion of patients with hypertension, hyperlipidemia, initial cholesterol level, and prior stroke history were not significantly different. Although not statistically significant, the good outcome group was younger than the poor outcome group in this study (65.7±9.3 vs 70.3±9.8 years). Although also not statistically significant, the time from stroke symptom onset to achieving maximum recanalization was shorter in the poor outcome group than in the good outcome group (318.1±154.7 vs 418.7±258.7 min). This was an unexpected finding, but it may be explainable by collateral flows from the anterior circulation via the posterior communicating artery, as described by Strbian et al21 and also shown in the ENDOSTROKE study.16 To open the occluded artery more quickly, we skipped an angiogram of the anterior circulation and approached the posterior circulation directly. We also reviewed all MRA data of the study patients, but could not evaluate the collateral status using those images because the quality of images varied according to the patients’ cardiac function or motion artifacts. Thus, we could not judge the status of the collateral circulation in the patients included in this study. Good collateral circulation is a known factor affecting good clinical outcomes in stroke patients because it reduces infarction volume.23 If the patient has good collateral flows from the anterior circulation via the posterior communicating artery, he/she would cope with a longer time after BAO, would have fewer lesions on DWI, and would show a lower NIHSS score at presentation. Thus, we believe the status of the collateral circulation is more important than the time itself.
We think that the strongest point of this study was the results of MR-based mechanical thrombectomy. In the ENDOSTROKE study, the use of MRI prior to endovascular treatment predicted the clinical outcome (MRI vs CT: 57.7% vs 28.7%, p=0.005) and was one of the independent predictors of patients’ outcome.16 However, only 26 patients receiving MR-based endovascular treatment were included and 15 of the 26 patients showed good clinical outcomes.16 This was very similar to our result, with a rate of good clinical outcome of 54.3% (19 of 35 patients).
The pc-ASPECTS is a semiquantitative grading system for posterior circulation stroke suggested by Puetz et al in 2008.19 It is relatively simple and easy to apply, and is based on previous findings indicating that the numbers of regions involved and involvement of the pons and midbrain are the most critical issues in functional outcome in patients with posterior circulation stroke.19 Several studies that have applied pc-ASPECTS to CT angiography (CTA) source images or initial MR-DWI have validated its usefulness and demonstrated that it could predict functional outcomes in patients with AIS caused by BAO.22 ,24–26 It has been reported that a pc-ASPECT score of 8 points on the initial image, whether based on CTA or DWI, increases the probability of a good outcome.22 ,24–26 In the present study, 13 of 19 patients (68.4%) in the good outcome group had pc-ASPECTS scores of >8 points based on DWI compared with only one patient (6.3%) in the poor outcome group.
Mourand et al27 proposed another scoring system using only brainstem lesions on DWI because other lesions are less involved in functional recovery. They reported that a large cerebellar infarction may cause minimal symptoms and may not affect functional recovery.27 However, we disagree with their opinion, although we agree that the brainstem is a crucial organ in determining patient recovery. In the present study, compression of the brainstem due to large cerebellar infarction was the cause of death in some patients. Thus, we did not apply their scoring system.
In this study we focused on the value of the control CT, which is usually taken after an intra-arterial thrombolytic/thrombectomy attempt. Generally, the purpose of the control CT is to detect hemorrhagic complications that could have developed during the procedure. However, we believe that the control CT should provide more information than that. After taking an initial MRI, considerable valuable time passes in preparing for and performing the thrombectomy, and ischemic insults to the brain continue until recanalization is achieved. Amans et al18 reported that contrast-stained brain parenchyma on CT after an intra-arterial procedure in patients with AIS most often progressed to infarction because such contrast staining on the control CT means that damage by ischemic insult has already occurred. We think that contrast staining on the control CT, especially in the brainstem, is as important as lesions on initial DWI. In fact, in the backward stepwise logistic regression, the presence of contrast staining in the brainstem tended to be greater in patients with a poor clinical outcome (p=0.059).
During the collection of data for this study, we expected that stroke etiology would affect patient outcome. Actually, 73.7% of patients in the good outcome group were diagnosed with cardiogenic embolism (CE) while 50% of patients in the poor outcome group had large artery atherosclerosis (LAA); however, these differences were not statistically significant (p=0.098). Intra-arterial intervention in patients with AIS is a complex and time-sensitive procedure. In our experience, the patients with LAA required more complex techniques such as prior balloon angioplasty to approach the occlusion site, and it was more difficult to achieve successful recanalization in these patients. Thus, they had longer procedural times than the patients with CE. We believe that stroke etiology probably does affect patient outcome, but the small number of patients in this study clouded the issue.
Study limitations
This study has several limitations including retrospective data collection, single-center design, and the small number of patients. It is also known that the location of the occlusion in the BA is associated with the prognosis in patients with BAO,28 but we did not analyze this. Additionally, we could not judge the status of the collateral circulation in the patients included in this study.
Conclusions
To our knowledge, this is the first reported study to show that contrast staining on the control CT as well as lesions on the initial DWI can affect the clinical outcome in patients with AIS caused by BAO where successful recanalization is achieved. Patients with AIS caused by BAO with a lower initial NIHSS score, fewer lesions on the initial DWI, and less contrast staining on the post-thrombectomy control CT have a higher probability of a good clinical outcome after successful recanalization via mechanical thrombectomy.
References
Footnotes
Contributors Study conception and design: DSC and BHL. Data collection: All authors. Data analysis and interpretation: SS, Y-WK, DSC and N-CC. Literature research: SS and O-YK. Drafting the manuscript: SS. Revision of the manuscript for important intellectual content: SS, DSC, MK-O and K-JP. Approval of final version of manuscript: All authors.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Ethics approval was obtained from the Gyengsang National University IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional data can be requested by email from the first author (sonsn@hanmail.net) or the corresponding author (brevis1@nate.com).