CLINICAL STUDY
Relationship of Thrombus Length to Number of Stent Retrievals, Revascularization, and Outcomes in Acute Ischemic Stroke Gaurav Jindal, MD, Timothy Miller, MD, Ravi Shivashankar, MD, Jason Mitchell, MD, MPH, MBA, Barney J. Stern, MD, Karen Yarbrough, MS, CRNP, and Dheeraj Gandhi, MD
ABSTRACT Purpose: To study the relationship between intracranial thrombus length and number of stent retrievals, revascularization rates, and functional outcomes in stroke. Materials and Methods: Retrospective data were collected from consecutive cases of stroke treated with endovascular procedures at a single institution from April 2012–September 2013. Thrombus length was measured in the anterior cerebral circulation. Demographic and clinical details; involved vessels; and procedural details, including the number of devices used and number of retrievals used for each device, were recorded. Revascularization rates and 90-day functional outcomes were recorded. Results: Data regarding the length of thrombus in the anterior cerebral circulation were available for 28 patients. There was no significant association between thrombus length and number of stent retrievals (P ¼ .3780), final thrombolysis in cerebral infarction (TICI) score (P ¼ .4835), or 90-day modified Rankin Scale score (P ¼ .4146). There was a significant difference (P ¼ .0280) between number of retrievals and final TICI score, with lower number of retrieval passes corresponding to higher final TICI scores. Conclusions: The data suggest no relationship between thrombus length and number of stent retrievals, final TICI score, or functional neurologic outcomes at 90 days in stent retrieval thrombectomy for acute ischemic stroke. These results do not support a predictive value for thrombus length quantification in the evaluation of stroke.
ABBREVIATIONS AIS = acute ischemic stroke, ANOVA = analysis of variance, DSA = digital subtraction angiography, IV = intravenous, mRS = modified Rankin Scale, SWI = susceptibility-weighted imaging, TICI = thrombolysis in cerebral infarction
Endovascular thrombectomy in the setting of acute ischemic stroke (AIS) has evolved in recent years as a result of the introduction of stent retriever devices and highly flexible guiding and suction catheters. Stent
From the Department of Radiology, Division of Interventional Neuroradiology (G.J., T.M., R.S., J.M., D.G.), and Department of Neurology (B.J.S., K.Y.), University of Maryland, 22 South Greene Street, Baltimore, MD 21201. Received March 4, 2014; final revision received April 30, 2014; accepted May 6, 2014. Address correspondence to G.J.; E-mail: [email protected] G.J. received research grants from Stryker Neurovascular and Microvention. D.G. is a paid consultant for Covidien, Reverse Medical, and Microvention and received research grants from NIH, Stryker, and Arstasis Axera. None of the other authors have identified a conflict of interest. & SIR, 2014 J Vasc Interv Radiol 2014; 25:1549–1557 http://dx.doi.org/10.1016/j.jvir.2014.05.014
retrievers have been available in the United States since 2012 for intracranial thrombectomy and have demonstrated significantly higher recanalization rates compared with older thrombectomy devices (1–3). Rates of successful recanalization of occluded intracranial arteries in the acute setting are reported to be 80%–86% (1,2). Systemic administration of intravenous (IV) tissue plasminogen activator (tPA) has shown limited utility against relatively long thrombi in the setting of AIS (4,5). However, the relationship of length of intracranial thrombus to number of stent retrievals used (ie, the number of passes or deployments made with a stent retriever) has not been extensively reported. Additionally, there have been few previous studies on the relationship of thrombus length to recanalization rates and functional outcomes, and these reports have demonstrated varying results (6,7). For this reason, a review
1550
’
Effect of Thrombus Length in AIS
was performed to evaluate these relationships using stent retrievers for the endovascular treatment of AIS in 37 patients at a single institution over an 18-month period. The purpose of this study is to determine if longer thrombi affect the number of stent retrievals used, recanalization rates, and functional outcomes in the setting of AIS in the anterior cerebral circulation.
MATERIALS AND METHODS Patient Selection and Clinical Data A prospective database was maintained consecutively for all patients undergoing acute endovascular treatment of stroke at our institution from April 2012–September 2013 according to a protocol approved by the institutional review board. This database was supplemented by retrospective data collection as detailed subsequently. The following data were collected on admission: sex; age; administration of IV tPA for AIS; National Institutes of Health Stroke Scale; and history of smoking, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, coronary artery disease, peripheral arterial disease, mechanical cardiac valve, and previous stroke or transient ischemic attack. Neurologic outcomes at 90 days were measured using the modified Rankin Scale (mRS) by an independent neurologist. During the 18-month study period, 37 intracranial endovascular thrombectomy procedures for AIS using stent retriever devices were performed. No patients during the study period underwent intracranial thrombectomy or thrombolysis with other devices or with intraarterial tPA administration. Of 37 patients, 19 were women. Mean patient age was 58 years (range, 36–81 y; SD, 13.7 y). Mean National Institutes of Health Stroke Scale score on admission was 18.5 (range, 8–26; SD, 5.0). IV tPA was administered to 24 patients (65%) before endovascular treatment. Comorbidities included hypertension in 29 patients (78%), hyperlipidemia in 20 patients (54%), diabetes in 8 patients (22%), coronary artery disease in 8 patients (22%), atrial fibrillation in 8 patients (22%), prior history of stroke or transient ischemic attack in 8 patients (22%), history of smoking in 9 patients (24%), peripheral vascular disease in 2 patients (5%), and mechanical cardiac valve in 2 patients (5%).
Imaging Protocol Thrombus length in the anterior cerebral circulation and thrombolysis in cerebral infarction (TICI) scores were retrospectively measured by two board-certified radiologists with 5 years and 4 years of experience in interventional neuroradiology. The images were read in three sessions during the week following the inclusion of the last patient in the study. The two observers read the images together and reached the final decision regarding length of thrombus. Discrepancies were adjudicated by consensus. If
Jindal et al
’
JVIR
length of thrombus could not be adequately calculated in the opinion of either of the two observers, the patient was excluded from analysis. For thrombus length measurement, susceptibilityweighted imaging (SWI) was primarily used. When SWI was unavailable or thought to be inadequate, one or more other imaging modalities were used to measure thrombus length. Imaging modalities used to measure clot length included SWI, computed tomography (CT) angiography, noncontrast CT, contrast-enhanced magnetic resonance (MR) angiography, and digital subtraction angiography (DSA). Thrombus lengths were measured using SWI in 16 cases (57%), using a contrast material void in the vessel in three planes on CT angiography and correlation with DSA in 4 cases (14%), using hyperdensity in the middle cerebral artery cistern on noncontrast CT and correlation with DSA in 3 cases (11%), using a contrast material void in the vessel in three planes on CT angiography alone in 2 cases (7%), using contrast material void in the vessel on DSA alone in 2 cases (7%), and using a void of contrast material in the vessel on both three-plane contrast-enhanced MR angiography and DSA in 1 case (4%). The following imaging equipment was used. MR imaging scanners were Siemens 1.5T MAGNETOM Avanto and 3T MAGNETOM Trio (Siemens AG, Erlangen, Germany) systems using software level VB19. CT scanners were Philips 256-slice Brilliance iCT and 64-channel systems (Philips Healthcare, Andover, Massachusetts). Biplane angiography units were Siemens Artis zee (Siemens AG). CT angiography was performed at 120 kVp, 350 mAs, 0.671 pitch, 0.5-second rotation time, 64 0.625 collimation, 1-mm slice thickness at 0.5-mm increments; an 80-mL bolus of contrast material was used for imaging after a 20-mL test injection. MR examinations consisted of axial susceptibility-weighted images (repetition time of 49 ms, echo time of 40 ms, base resolution of 256, field of view of 230), axial diffusion-weighted images (repetition time of 5,400 ms, echo time of 86 ms, base resolution of 192, field of view of 220), axial apparent diffusion coefficient map images, and coronal acquired contrastenhanced MR angiography of the head and neck (repetition time of 3.35 ms, echo time of 1.22 ms, base resolution of 512, field of view of 320 81) in the arterial and venous phases of enhancement. CT and MR angiograms were evaluated at 1-mm slice thickness and were reconstructed on a three-dimensional workstation (AQUARIUS WORKSTATION, TeraRecon, Foster City, California) to confirm measurements.
Thrombectomy Treatment According to an institutional protocol derived prospectively, patients with AIS treated with mechanical thrombectomy either were ineligible for IV tPA or did not clinically improve after receiving IV tPA (0.9 mg/kg) and
Volume 25
’
Number 10
’
October
’
2014
had imaging evidence of a large artery occlusion with either an Alberta Stroke Program Early CT Score of Z 6 or o 75 mL of infarcted brain tissue on diffusionweighted imaging or CT or MR perfusion imaging. All thrombectomy procedures with a known time of symptom onset were begun within 6 hours of symptom onset. Endovascular thrombectomy was performed by one of three board-certified interventional neuroradiologists. All endovascular procedures were done on one of two Siemens Artis zee biplane angiography units. In each case, 6-F guiding catheters were used through which manual suction was applied during each clot retrieval in all cases except two, in which machine aspiration (Penumbra, Alameda, California) was used through the guiding catheter. Balloon guide catheters were not used. Guide catheters were generally advanced to the region of the craniocervical junction. Stent retrievers used were Solitaire (ev3/Covidien, Irvine, California) in 36 cases and Trevo (Stryker Neurovascular, Freemont, California) in 8 cases. The Solitaire device was available at our institution throughout the entire study period and was predominantly used. The Trevo device was introduced in the middle of the study period. All devices were opened into the thrombus and kept in place for 5 minutes before each stent retrieval thrombectomy. All procedures except two were performed under general anesthesia. The following procedural data were collected: involved anterior circulation vessel, length of thrombus, stent retriever device, guiding catheters used, number of attempted retrievals with each device, time to recanalization, and TICI scores before and after thrombectomy.
Statistical Analysis Descriptive data are presented as mean and SD or 95% confidence intervals where appropriate. One-way analysis of variance (ANOVA) or contingency table analysis with χ2 testing was performed to determine associations between clot length, number of stent retrievals, TICI score after thrombectomy, and 90-day outcomes with the mRS where appropriate for ordinal and continuous measures. Any P value o .05 was considered statistically significant. Statistical analysis was performed using JMP Pro version 10.0.2 (SAS, Inc, Cary, North Carolina).
1551
artery and the M1 segment artery in 6 cases (21%). Clot lengths ranged from 7–33 mm (mean, 19 mm ⫾ 7). There were no discrepancies in thrombus length measurements of 4 3 mm between the two observers. Additionally, there were no discrepancies in thrombus length measurements of 4 3 mm among the various imaging modalities used.
Measurements of Other Variables The number of retrievals in each case ranged from 1–5 (mean, 2.4; SD, 1.4), using a mean of 1.2 stent retriever devices per case. Initial angiography demonstrated a TICI score of 0 in all 28 cases. Final angiography demonstrated TICI 0 in 2 cases (7.1%), TICI 1 in 1 case (3.6%), TICI 2a in 2 cases (7.1%), TICI 2b in 7 cases (25.0%), and TICI 3 in 16 cases (57.1%). In 23 cases (82.1%), the final angiography score demonstrated a TICI score of Z 2b. The time between cross-sectional imaging and initial DSA ranged from 22–86 minutes (mean, 44 min). Mean time to recanalization from initial femoral arteriotomy was 55 minutes (range, 19–111 min). Functional outcome assessment in the remaining cases revealed mRS score of 0 in 2 cases (7.1%), 1 in 9 cases (32.1%), 2 in 3 cases (10.7%), 3 in 1 case (3.6%), 4 in 4 cases (14.3%), 5 in 5 cases (17.9%), and 6 in 4 cases (14.3%). Of 28 patients, 14 (50.0%) had a mRS score of 0–2 at 90 days. TICI scores after thrombectomy and 90day follow-up mRS results are outlined in the Table.
Statistical Analysis of Relationship of Thrombus Length to Other Variables Using one-way ANOVA, there was no significant relationship (P ¼ .3780) between clot length and the number of stent retrievals used in each case. Also, there Table . Outcomes after Thrombectomy No. (%) TICI score after thrombectomy 0
1 (3.6)
2a 2b
2 (7.1) 7 (25.0)
3
RESULTS
2 (7.1)
1
Z 2b mRS at 90 days after procedure
16 (57.1) 23 (82.1)
Measurement of Thrombus Length
0
2 (7.1)
Thrombus length measurements in the anterior cerebral circulation were available for 28 of 37 patients. Five patients had basilar artery occlusion and were excluded from this analysis, and thrombus length was not measurable in four patients because of difficulty in clearly visualizing the proximal and distal ends of the thrombus. Of the 28 remaining cases analyzed, the involved vessel was the M1 segment of the middle cerebral artery alone in 22 cases (79%) and both the distal internal carotid
1 2
9 (32.1) 3 (10.7)
3
1 (3.6)
4 5
4 (14.3) 5 (17.9)
6 0–2
4 (14.3) 14 (50.0)
mRS ¼ modified Rankin Scale; TICI ¼ thrombolysis in cerebral infarction.
1552
’
Effect of Thrombus Length in AIS
was no significant relationship (P ¼ .4835) between clot length and final TICI score after thrombectomy using one-way ANOVA. However, there was a significant difference (P ¼ .0280, Pearson χ2) among number of retrievals used in each procedure and the final TICI score after thrombectomy, with a lower number of retrievals used corresponding to a higher TICI score after thrombectomy by correspondence analysis, although the contingency table for this relationship is sparsely populated (Z 20% cell counts are o 5). Oneway ANOVA showed no significant relationship between clot length and 90-day mRS score (P ¼ .4146). Quantile box plots of clot length to number of device retrievals, TICI score after thrombectomy, and 90-day mRS score results are shown in Figures 1–3.
DISCUSSION There have been multiple reports on the relationship of length of intracranial arterial thrombus to recanalization rates after administration of IV tPA. One study found a limited response to systemic thrombolysis in patients with a clot length of 4 8 mm identified on an ultrathin noncontrast CT head scan (8). Another study demonstrated that clot length was predictive of response to IV tPA, with thrombi 4 8 mm never recanalizing (4). A third study demonstrated that clot length 4 10 mm infrequently disappears with IV tPA (5). These studies suggest a potential need for appropriately selected ancillary therapy in patients with clot lengths 4 8–10 mm. The purpose of the present study was to determine if acute endovascular therapy using modern stent retriever devices and patient outcomes also are
Jindal et al
’
JVIR
influenced by length of intracranial thrombus. The results from this study suggest that relatively short thrombi and thrombi measuring 4 10 mm have a similarly high likelihood of recanalization using stent retrievers; mRS score of 0–2 at follow-up was achieved in 50% of cases in this series harboring a mean clot length of 19 mm. The role of thrombus length in endovascular procedures for AIS has not been evaluated to a great extent. To our knowledge, the current study is the second report specifically assessing the relationship between thrombus length to the number of stent retrievals used in acute intracranial endovascular stent retrieval thrombectomy. No relationship between thrombus length and number of stent retrievals was shown. The other recent study on this topic also demonstrated that there was no difference in number of stent deployments between clots of 4 14 mm or o 14 mm (7). Such a finding may be initially counterintuitive in the sense that longer thrombi may be expected to require a greater number of stent retrievals for ultimate removal. However, the lack of a relationship demonstrated by the available data suggests that both short and long thrombi have a similar likelihood of becoming incorporated into the interstices of a stent retrieval device and then being removed by the device. No significant relationship was seen between length of thrombus and final recanalization scores. These findings have implications for patient management in that a relatively long thrombus in itself does not appear by our results to be a reliable predictor of increased procedural effort and time to final recanalization likelihood when using stent retrievers for thrombectomy in AIS. To our knowledge, the present study is the second study assessing the relationship of length of thrombus to
Figure 1. Quantile box plot of clot length versus number of stent retrievals.
Volume 25
’
Number 10
’
October
’
2014
1553
Figure 2. Quantile box plot of clot length versus final TICI score after thrombectomy.
Figure 3. Quantile box plot of clot length versus 90-day mRS.
functional outcomes after intracranial stent retrieval thrombectomy. Soize et al (7) demonstrated that thrombus length 4 14 mm was associated with poor outcome at 3 months after mechanical thrombectomy with a stent retriever for AIS, which is in contrast to the present study. Spiotta et al (6) found no association between clot length, likelihood of recanalization, time to recanalization, and functional outcome at 90 days in the setting
of AIS, although most patients from this series underwent thrombectomy using the Penumbra aspiration system rather than the stent retriever devices used in our series and in the series by Soize et al. The number of retrievals used in each procedure was shown in the present series to have an inverse relationship to the final TICI score. This finding is likely explained by the greater number of attempts used for recanalization of an
1554
’
Effect of Thrombus Length in AIS
occluded vessel refractory to one or two retrievals in conjunction with the idea that it may be more difficult to restore flow within such a vessel. Such a finding raises the possibility of limited recanalization success after multiple initial failed retrieval attempts in certain refractory cases. Given inherent risks of mechanical thrombectomy, limiting the number of stent retriever deployments in initially refractory cases warrants discussion and debate.
Jindal et al
’
JVIR
The imaging modalities used were variable but primarily consisted of SWI (57%), an MR imaging sequence known to demonstrate intraluminal thrombus (Fig 4) (9). SWI artifact may not be directly synonymous with exact clot length, although the limits of thrombus on SWI in this series were most often well defined with convex surfaces. SWI may underestimate intracranial thrombus length on vertical segments of arteries; however, most patients in this series had an occlusion
Figure 4. (a) Axial susceptibility-weighted MR image demonstrates susceptibility artifact in the expected location of the right middle cerebral artery extending from the M1 segment of the vessel into an M2 segment (arrows). (b) Anteroposterior view before thrombectomy of right internal carotid artery injection DSA image after deployment of the stent retriever (black arrow on distal tines in an M2 segment artery) confirms the presence and extent of thrombus in the right middle cerebral artery (white arrows). (c) Anteroposterior view after thrombectomy of right internal carotid artery injection DSA image shows recanalization of the right middle cerebral artery with nonocclusive distal thrombi within the right middle cerebral artery (arrows).
Volume 25
’
Number 10
’
October
’
2014
within the M1 segment artery, which is horizontal and easily measurable. In cases in which SWI was unavailable, two imaging modalities often were used to confirm thrombus length.
1555
CT and MR angiography were thought to be reliable tools for clot length measurement if both the distal and the proximal ends of the thrombus were outlined by contrast material in three planes (Figs 5,6). Delayed
Figure 5. (a) Anteroposterior view of contrast-enhanced volume-rendered three-dimensional MR angiography image demonstrates the proximal end (white arrow) of the left middle cerebral artery occlusion, M1 segment artery; the distal end of the occlusion (red arrow) is delineated by retrograde contrast material filling of distal middle cerebral artery branches to the level of the distal occlusion. (b) Delayed contrastenhanced coronal view MR angiography image demonstrates enhancement of the clot in the left M1 segment artery (arrows). Anteroposterior views of (c) early-phase and (d) late-phase left internal carotid artery injection DSA images demonstrate the proximal occlusion of the left M1 segment artery in the early phase (arrow, c) and retrograde contrast material filling to the distal end of clot in the late phase (arrow, d).
1556
’
Effect of Thrombus Length in AIS
acquisition contrast-enhanced MR angiography was useful for delineation of the distal end of the thrombus in one case in our series. Noncontrast CT was uncommonly used to measure clot length. Riedel et al (10) demonstrated that acute thrombus can be visualized reliably using noncontrast CT data reconstructed to r 2.5 mm slice width. Only three patient lesions in the present series were measured using noncontrast CT (all done using 2.5-mm slice thickness), and these three lesion measurements all were confirmed on DSA. Measurement of clot length on DSA is limited by its two-dimensional view of the lesion, and only two patient lesions were measured using DSA alone in our series. There are several limitations to the present study. The main limitation is the small number of patients. Because
Jindal et al
’
JVIR
of the study’s small sample size, the relationship of symptomatic intracranial hemorrhage after the procedure to length of thrombus could not be determined reliably. Nevertheless, there is enough statistical power to identify multiple predictive factors within the study. Future studies with larger patient numbers are necessary to elucidate further the relationships among the variables presented here. IV tPA was administered before mechanical thrombectomy in 65% of patients, and its effects on recanalization rates and outcomes are not specifically accounted for or delineated in our study. A mean time of 44 minutes between initial cross-sectional imaging and DSA may have allowed for progression or regression of thrombus, particularly in patients who received IV tPA; although this mean time interval is
Figure 6. (a) Coronal maximum intensity projection CT angiography image demonstrates the proximal and distal ends (arrows) of the occluded right internal carotid artery terminus and right middle cerebral artery; the distal end of the thrombus is delineated as a result of retrograde contrast material filling within the right middle cerebral artery to the level of the distal end of the occlusion. Anteroposterior views of (b) early-phase and (c) late-phase right internal carotid artery injection DSA images show an occluded right internal carotid artery terminus in the early phase (arrow, b) and outlining of both proximal and distal ends of the occlusion (arrows, c) as a result of retrograde contrast material filling to the level of the distal end of occlusion in the right middle cerebral artery in the late phase. (d) Anteroposterior view of right internal carotid artery injection DSA image after endovascular stent retrieval thrombectomy demonstrates complete recanalization of the right internal carotid artery and right middle cerebral artery.
Volume 25
’
Number 10
’
October
’
2014
relatively short, this possibility was not accounted for in some cases because clot length could not be reliably measured on all DSA imaging. For example, in a few cases, a proximal occlusion without distal delineation of thrombus was seen on DSA. Additionally, the study is limited by imperfect clot length measurement tools as outlined earlier. Use of multiple imaging modalities to measure clot length introduces inhomogeneity in data collection. However, SWI was used in most cases, and two imaging modalities were used in most of the remaining cases to measure thrombus length. In our estimation, the imaging methodology used in this series was able to estimate clot length accurately; four cases were excluded because clot length was deemed by either of two reviewers to be suboptimal. The study is also limited by the potential variability of appropriateness criteria used from one operator to the next, especially regarding the number of stent retrievals used to restore flow within an artery difficult to recanalize. However, the three endovascular operators in the single-center group generally practice under similar guidelines and use similar procedural techniques, including rate and tension with which the operator withdraws a stent retriever, in treating AIS. In conclusion, the data suggest that there is no relationship between thrombus length and number of stent retrievals, final TICI score, or functional neurologic outcomes at 90 days in the setting of stent retrieval thrombectomy for AIS. These results do not support a predictive value for thrombus length quantification in the evaluation of AIS. Further investigation with a
1557
larger number of patients is necessary to gain a better understanding of these relationships.
REFERENCES 1. Pereira VM, Gralla J, Davalos A, et al. Prospective, multicenter, singlearm study of mechanical thrombectomy using solitaire flow restoration in acute ischemic stroke. Stroke 2013; 44:2802–2807. 2. Nogueira RG, Lutsep HL, Gupta R, et al. for the TREVO 2 Trialists. Trevo versus Merci retrievers for thrombectomy revascularisation of large vessel occlusions in acute ischaemic stroke (TREVO 2): a randomised trial. Lancet 2012; 380:1231–1240. 3. Saver JL, Jahan R, Levy EI, et al. for the SWIFT Trialists. Solitaire flow restoration device versus the Merci Retriever in patients with acute ischaemic stroke (SWIFT): a randomised, parallel-group, non-inferiority trial. Lancet 2012; 380:1241–1249. 4. Moftakhar P, English JD, Cooke DL, et al. Density of thrombus on admission CT predicts revascularization efficacy in large vessel occlusion acute ischemic stroke. Stroke 2013; 44:243–245. 5. Shobha N, Bal S, Boyko M, et al. Measurement of length of hyperdense MCA sign in acute ischemic stroke predicts disappearance after IV tPA. J Neuroimaging 2013; 24:7–10. 6. Spiotta AM, Vargas J, Hawk H, et al. Hounsfield unit value and clot length in the acutely occluded vessel and time required to achieve thrombectomy, complications and outcome. J Neurointerv Surg In press; available online July 31, 2013. 7. Soize S, Barbe C, Kadziolka K, Estrade L, Serre I, Pierot L. Predictive factors of outcome and hemorrhage after acute ischemic stroke treated by mechanical thrombectomy with a stent-retriever. Neuroradiology 2013; 55:977–987. 8. Riedel CH, Zimmermann P, Jensen-Kondering U, Stingele R, Deuschl G, Jansen O. The importance of size: successful recanalization by intravenous thrombolysis in acute anterior stroke depends on thrombus length. Stroke 2011; 42:1775–1777. 9. Rovira A, Orellana P, Alvarez-Sabín J, et al. Hyperacute ischemic stroke: middle cerebral artery susceptibility sign at echoplanar gradient-echo MR imaging. Radiology 2004; 232:466–473. 10. Riedel CH, Jensen U, Rohr A, et al. Assessment of thrombus in acute middle cerebral artery occlusion using thin-slice nonenhanced computed tomography reconstructions. Stroke 2010; 41:1659–1664.
Relationship of Thrombus Length to Number of Stent Retrievals, Revascularization, and Outcomes in Acute Ischemic Stroke Gaurav Jindal, MD, Timothy Miller, MD, Ravi Shivashankar, MD, Jason Mitchell, MD, MPH, MBA, Barney J. Stern, MD, Karen Yarbrough, MS, CRNP, and Dheeraj Gandhi, MD
ABSTRACT Purpose: To study the relationship between intracranial thrombus length and number of stent retrievals, revascularization rates, and functional outcomes in stroke. Materials and Methods: Retrospective data were collected from consecutive cases of stroke treated with endovascular procedures at a single institution from April 2012–September 2013. Thrombus length was measured in the anterior cerebral circulation. Demographic and clinical details; involved vessels; and procedural details, including the number of devices used and number of retrievals used for each device, were recorded. Revascularization rates and 90-day functional outcomes were recorded. Results: Data regarding the length of thrombus in the anterior cerebral circulation were available for 28 patients. There was no significant association between thrombus length and number of stent retrievals (P ¼ .3780), final thrombolysis in cerebral infarction (TICI) score (P ¼ .4835), or 90-day modified Rankin Scale score (P ¼ .4146). There was a significant difference (P ¼ .0280) between number of retrievals and final TICI score, with lower number of retrieval passes corresponding to higher final TICI scores. Conclusions: The data suggest no relationship between thrombus length and number of stent retrievals, final TICI score, or functional neurologic outcomes at 90 days in stent retrieval thrombectomy for acute ischemic stroke. These results do not support a predictive value for thrombus length quantification in the evaluation of stroke.
ABBREVIATIONS AIS = acute ischemic stroke, ANOVA = analysis of variance, DSA = digital subtraction angiography, IV = intravenous, mRS = modified Rankin Scale, SWI = susceptibility-weighted imaging, TICI = thrombolysis in cerebral infarction
Endovascular thrombectomy in the setting of acute ischemic stroke (AIS) has evolved in recent years as a result of the introduction of stent retriever devices and highly flexible guiding and suction catheters. Stent
From the Department of Radiology, Division of Interventional Neuroradiology (G.J., T.M., R.S., J.M., D.G.), and Department of Neurology (B.J.S., K.Y.), University of Maryland, 22 South Greene Street, Baltimore, MD 21201. Received March 4, 2014; final revision received April 30, 2014; accepted May 6, 2014. Address correspondence to G.J.; E-mail: [email protected] G.J. received research grants from Stryker Neurovascular and Microvention. D.G. is a paid consultant for Covidien, Reverse Medical, and Microvention and received research grants from NIH, Stryker, and Arstasis Axera. None of the other authors have identified a conflict of interest. & SIR, 2014 J Vasc Interv Radiol 2014; 25:1549–1557 http://dx.doi.org/10.1016/j.jvir.2014.05.014
retrievers have been available in the United States since 2012 for intracranial thrombectomy and have demonstrated significantly higher recanalization rates compared with older thrombectomy devices (1–3). Rates of successful recanalization of occluded intracranial arteries in the acute setting are reported to be 80%–86% (1,2). Systemic administration of intravenous (IV) tissue plasminogen activator (tPA) has shown limited utility against relatively long thrombi in the setting of AIS (4,5). However, the relationship of length of intracranial thrombus to number of stent retrievals used (ie, the number of passes or deployments made with a stent retriever) has not been extensively reported. Additionally, there have been few previous studies on the relationship of thrombus length to recanalization rates and functional outcomes, and these reports have demonstrated varying results (6,7). For this reason, a review
1550
’
Effect of Thrombus Length in AIS
was performed to evaluate these relationships using stent retrievers for the endovascular treatment of AIS in 37 patients at a single institution over an 18-month period. The purpose of this study is to determine if longer thrombi affect the number of stent retrievals used, recanalization rates, and functional outcomes in the setting of AIS in the anterior cerebral circulation.
MATERIALS AND METHODS Patient Selection and Clinical Data A prospective database was maintained consecutively for all patients undergoing acute endovascular treatment of stroke at our institution from April 2012–September 2013 according to a protocol approved by the institutional review board. This database was supplemented by retrospective data collection as detailed subsequently. The following data were collected on admission: sex; age; administration of IV tPA for AIS; National Institutes of Health Stroke Scale; and history of smoking, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, coronary artery disease, peripheral arterial disease, mechanical cardiac valve, and previous stroke or transient ischemic attack. Neurologic outcomes at 90 days were measured using the modified Rankin Scale (mRS) by an independent neurologist. During the 18-month study period, 37 intracranial endovascular thrombectomy procedures for AIS using stent retriever devices were performed. No patients during the study period underwent intracranial thrombectomy or thrombolysis with other devices or with intraarterial tPA administration. Of 37 patients, 19 were women. Mean patient age was 58 years (range, 36–81 y; SD, 13.7 y). Mean National Institutes of Health Stroke Scale score on admission was 18.5 (range, 8–26; SD, 5.0). IV tPA was administered to 24 patients (65%) before endovascular treatment. Comorbidities included hypertension in 29 patients (78%), hyperlipidemia in 20 patients (54%), diabetes in 8 patients (22%), coronary artery disease in 8 patients (22%), atrial fibrillation in 8 patients (22%), prior history of stroke or transient ischemic attack in 8 patients (22%), history of smoking in 9 patients (24%), peripheral vascular disease in 2 patients (5%), and mechanical cardiac valve in 2 patients (5%).
Imaging Protocol Thrombus length in the anterior cerebral circulation and thrombolysis in cerebral infarction (TICI) scores were retrospectively measured by two board-certified radiologists with 5 years and 4 years of experience in interventional neuroradiology. The images were read in three sessions during the week following the inclusion of the last patient in the study. The two observers read the images together and reached the final decision regarding length of thrombus. Discrepancies were adjudicated by consensus. If
Jindal et al
’
JVIR
length of thrombus could not be adequately calculated in the opinion of either of the two observers, the patient was excluded from analysis. For thrombus length measurement, susceptibilityweighted imaging (SWI) was primarily used. When SWI was unavailable or thought to be inadequate, one or more other imaging modalities were used to measure thrombus length. Imaging modalities used to measure clot length included SWI, computed tomography (CT) angiography, noncontrast CT, contrast-enhanced magnetic resonance (MR) angiography, and digital subtraction angiography (DSA). Thrombus lengths were measured using SWI in 16 cases (57%), using a contrast material void in the vessel in three planes on CT angiography and correlation with DSA in 4 cases (14%), using hyperdensity in the middle cerebral artery cistern on noncontrast CT and correlation with DSA in 3 cases (11%), using a contrast material void in the vessel in three planes on CT angiography alone in 2 cases (7%), using contrast material void in the vessel on DSA alone in 2 cases (7%), and using a void of contrast material in the vessel on both three-plane contrast-enhanced MR angiography and DSA in 1 case (4%). The following imaging equipment was used. MR imaging scanners were Siemens 1.5T MAGNETOM Avanto and 3T MAGNETOM Trio (Siemens AG, Erlangen, Germany) systems using software level VB19. CT scanners were Philips 256-slice Brilliance iCT and 64-channel systems (Philips Healthcare, Andover, Massachusetts). Biplane angiography units were Siemens Artis zee (Siemens AG). CT angiography was performed at 120 kVp, 350 mAs, 0.671 pitch, 0.5-second rotation time, 64 0.625 collimation, 1-mm slice thickness at 0.5-mm increments; an 80-mL bolus of contrast material was used for imaging after a 20-mL test injection. MR examinations consisted of axial susceptibility-weighted images (repetition time of 49 ms, echo time of 40 ms, base resolution of 256, field of view of 230), axial diffusion-weighted images (repetition time of 5,400 ms, echo time of 86 ms, base resolution of 192, field of view of 220), axial apparent diffusion coefficient map images, and coronal acquired contrastenhanced MR angiography of the head and neck (repetition time of 3.35 ms, echo time of 1.22 ms, base resolution of 512, field of view of 320 81) in the arterial and venous phases of enhancement. CT and MR angiograms were evaluated at 1-mm slice thickness and were reconstructed on a three-dimensional workstation (AQUARIUS WORKSTATION, TeraRecon, Foster City, California) to confirm measurements.
Thrombectomy Treatment According to an institutional protocol derived prospectively, patients with AIS treated with mechanical thrombectomy either were ineligible for IV tPA or did not clinically improve after receiving IV tPA (0.9 mg/kg) and
Volume 25
’
Number 10
’
October
’
2014
had imaging evidence of a large artery occlusion with either an Alberta Stroke Program Early CT Score of Z 6 or o 75 mL of infarcted brain tissue on diffusionweighted imaging or CT or MR perfusion imaging. All thrombectomy procedures with a known time of symptom onset were begun within 6 hours of symptom onset. Endovascular thrombectomy was performed by one of three board-certified interventional neuroradiologists. All endovascular procedures were done on one of two Siemens Artis zee biplane angiography units. In each case, 6-F guiding catheters were used through which manual suction was applied during each clot retrieval in all cases except two, in which machine aspiration (Penumbra, Alameda, California) was used through the guiding catheter. Balloon guide catheters were not used. Guide catheters were generally advanced to the region of the craniocervical junction. Stent retrievers used were Solitaire (ev3/Covidien, Irvine, California) in 36 cases and Trevo (Stryker Neurovascular, Freemont, California) in 8 cases. The Solitaire device was available at our institution throughout the entire study period and was predominantly used. The Trevo device was introduced in the middle of the study period. All devices were opened into the thrombus and kept in place for 5 minutes before each stent retrieval thrombectomy. All procedures except two were performed under general anesthesia. The following procedural data were collected: involved anterior circulation vessel, length of thrombus, stent retriever device, guiding catheters used, number of attempted retrievals with each device, time to recanalization, and TICI scores before and after thrombectomy.
Statistical Analysis Descriptive data are presented as mean and SD or 95% confidence intervals where appropriate. One-way analysis of variance (ANOVA) or contingency table analysis with χ2 testing was performed to determine associations between clot length, number of stent retrievals, TICI score after thrombectomy, and 90-day outcomes with the mRS where appropriate for ordinal and continuous measures. Any P value o .05 was considered statistically significant. Statistical analysis was performed using JMP Pro version 10.0.2 (SAS, Inc, Cary, North Carolina).
1551
artery and the M1 segment artery in 6 cases (21%). Clot lengths ranged from 7–33 mm (mean, 19 mm ⫾ 7). There were no discrepancies in thrombus length measurements of 4 3 mm between the two observers. Additionally, there were no discrepancies in thrombus length measurements of 4 3 mm among the various imaging modalities used.
Measurements of Other Variables The number of retrievals in each case ranged from 1–5 (mean, 2.4; SD, 1.4), using a mean of 1.2 stent retriever devices per case. Initial angiography demonstrated a TICI score of 0 in all 28 cases. Final angiography demonstrated TICI 0 in 2 cases (7.1%), TICI 1 in 1 case (3.6%), TICI 2a in 2 cases (7.1%), TICI 2b in 7 cases (25.0%), and TICI 3 in 16 cases (57.1%). In 23 cases (82.1%), the final angiography score demonstrated a TICI score of Z 2b. The time between cross-sectional imaging and initial DSA ranged from 22–86 minutes (mean, 44 min). Mean time to recanalization from initial femoral arteriotomy was 55 minutes (range, 19–111 min). Functional outcome assessment in the remaining cases revealed mRS score of 0 in 2 cases (7.1%), 1 in 9 cases (32.1%), 2 in 3 cases (10.7%), 3 in 1 case (3.6%), 4 in 4 cases (14.3%), 5 in 5 cases (17.9%), and 6 in 4 cases (14.3%). Of 28 patients, 14 (50.0%) had a mRS score of 0–2 at 90 days. TICI scores after thrombectomy and 90day follow-up mRS results are outlined in the Table.
Statistical Analysis of Relationship of Thrombus Length to Other Variables Using one-way ANOVA, there was no significant relationship (P ¼ .3780) between clot length and the number of stent retrievals used in each case. Also, there Table . Outcomes after Thrombectomy No. (%) TICI score after thrombectomy 0
1 (3.6)
2a 2b
2 (7.1) 7 (25.0)
3
RESULTS
2 (7.1)
1
Z 2b mRS at 90 days after procedure
16 (57.1) 23 (82.1)
Measurement of Thrombus Length
0
2 (7.1)
Thrombus length measurements in the anterior cerebral circulation were available for 28 of 37 patients. Five patients had basilar artery occlusion and were excluded from this analysis, and thrombus length was not measurable in four patients because of difficulty in clearly visualizing the proximal and distal ends of the thrombus. Of the 28 remaining cases analyzed, the involved vessel was the M1 segment of the middle cerebral artery alone in 22 cases (79%) and both the distal internal carotid
1 2
9 (32.1) 3 (10.7)
3
1 (3.6)
4 5
4 (14.3) 5 (17.9)
6 0–2
4 (14.3) 14 (50.0)
mRS ¼ modified Rankin Scale; TICI ¼ thrombolysis in cerebral infarction.
1552
’
Effect of Thrombus Length in AIS
was no significant relationship (P ¼ .4835) between clot length and final TICI score after thrombectomy using one-way ANOVA. However, there was a significant difference (P ¼ .0280, Pearson χ2) among number of retrievals used in each procedure and the final TICI score after thrombectomy, with a lower number of retrievals used corresponding to a higher TICI score after thrombectomy by correspondence analysis, although the contingency table for this relationship is sparsely populated (Z 20% cell counts are o 5). Oneway ANOVA showed no significant relationship between clot length and 90-day mRS score (P ¼ .4146). Quantile box plots of clot length to number of device retrievals, TICI score after thrombectomy, and 90-day mRS score results are shown in Figures 1–3.
DISCUSSION There have been multiple reports on the relationship of length of intracranial arterial thrombus to recanalization rates after administration of IV tPA. One study found a limited response to systemic thrombolysis in patients with a clot length of 4 8 mm identified on an ultrathin noncontrast CT head scan (8). Another study demonstrated that clot length was predictive of response to IV tPA, with thrombi 4 8 mm never recanalizing (4). A third study demonstrated that clot length 4 10 mm infrequently disappears with IV tPA (5). These studies suggest a potential need for appropriately selected ancillary therapy in patients with clot lengths 4 8–10 mm. The purpose of the present study was to determine if acute endovascular therapy using modern stent retriever devices and patient outcomes also are
Jindal et al
’
JVIR
influenced by length of intracranial thrombus. The results from this study suggest that relatively short thrombi and thrombi measuring 4 10 mm have a similarly high likelihood of recanalization using stent retrievers; mRS score of 0–2 at follow-up was achieved in 50% of cases in this series harboring a mean clot length of 19 mm. The role of thrombus length in endovascular procedures for AIS has not been evaluated to a great extent. To our knowledge, the current study is the second report specifically assessing the relationship between thrombus length to the number of stent retrievals used in acute intracranial endovascular stent retrieval thrombectomy. No relationship between thrombus length and number of stent retrievals was shown. The other recent study on this topic also demonstrated that there was no difference in number of stent deployments between clots of 4 14 mm or o 14 mm (7). Such a finding may be initially counterintuitive in the sense that longer thrombi may be expected to require a greater number of stent retrievals for ultimate removal. However, the lack of a relationship demonstrated by the available data suggests that both short and long thrombi have a similar likelihood of becoming incorporated into the interstices of a stent retrieval device and then being removed by the device. No significant relationship was seen between length of thrombus and final recanalization scores. These findings have implications for patient management in that a relatively long thrombus in itself does not appear by our results to be a reliable predictor of increased procedural effort and time to final recanalization likelihood when using stent retrievers for thrombectomy in AIS. To our knowledge, the present study is the second study assessing the relationship of length of thrombus to
Figure 1. Quantile box plot of clot length versus number of stent retrievals.
Volume 25
’
Number 10
’
October
’
2014
1553
Figure 2. Quantile box plot of clot length versus final TICI score after thrombectomy.
Figure 3. Quantile box plot of clot length versus 90-day mRS.
functional outcomes after intracranial stent retrieval thrombectomy. Soize et al (7) demonstrated that thrombus length 4 14 mm was associated with poor outcome at 3 months after mechanical thrombectomy with a stent retriever for AIS, which is in contrast to the present study. Spiotta et al (6) found no association between clot length, likelihood of recanalization, time to recanalization, and functional outcome at 90 days in the setting
of AIS, although most patients from this series underwent thrombectomy using the Penumbra aspiration system rather than the stent retriever devices used in our series and in the series by Soize et al. The number of retrievals used in each procedure was shown in the present series to have an inverse relationship to the final TICI score. This finding is likely explained by the greater number of attempts used for recanalization of an
1554
’
Effect of Thrombus Length in AIS
occluded vessel refractory to one or two retrievals in conjunction with the idea that it may be more difficult to restore flow within such a vessel. Such a finding raises the possibility of limited recanalization success after multiple initial failed retrieval attempts in certain refractory cases. Given inherent risks of mechanical thrombectomy, limiting the number of stent retriever deployments in initially refractory cases warrants discussion and debate.
Jindal et al
’
JVIR
The imaging modalities used were variable but primarily consisted of SWI (57%), an MR imaging sequence known to demonstrate intraluminal thrombus (Fig 4) (9). SWI artifact may not be directly synonymous with exact clot length, although the limits of thrombus on SWI in this series were most often well defined with convex surfaces. SWI may underestimate intracranial thrombus length on vertical segments of arteries; however, most patients in this series had an occlusion
Figure 4. (a) Axial susceptibility-weighted MR image demonstrates susceptibility artifact in the expected location of the right middle cerebral artery extending from the M1 segment of the vessel into an M2 segment (arrows). (b) Anteroposterior view before thrombectomy of right internal carotid artery injection DSA image after deployment of the stent retriever (black arrow on distal tines in an M2 segment artery) confirms the presence and extent of thrombus in the right middle cerebral artery (white arrows). (c) Anteroposterior view after thrombectomy of right internal carotid artery injection DSA image shows recanalization of the right middle cerebral artery with nonocclusive distal thrombi within the right middle cerebral artery (arrows).
Volume 25
’
Number 10
’
October
’
2014
within the M1 segment artery, which is horizontal and easily measurable. In cases in which SWI was unavailable, two imaging modalities often were used to confirm thrombus length.
1555
CT and MR angiography were thought to be reliable tools for clot length measurement if both the distal and the proximal ends of the thrombus were outlined by contrast material in three planes (Figs 5,6). Delayed
Figure 5. (a) Anteroposterior view of contrast-enhanced volume-rendered three-dimensional MR angiography image demonstrates the proximal end (white arrow) of the left middle cerebral artery occlusion, M1 segment artery; the distal end of the occlusion (red arrow) is delineated by retrograde contrast material filling of distal middle cerebral artery branches to the level of the distal occlusion. (b) Delayed contrastenhanced coronal view MR angiography image demonstrates enhancement of the clot in the left M1 segment artery (arrows). Anteroposterior views of (c) early-phase and (d) late-phase left internal carotid artery injection DSA images demonstrate the proximal occlusion of the left M1 segment artery in the early phase (arrow, c) and retrograde contrast material filling to the distal end of clot in the late phase (arrow, d).
1556
’
Effect of Thrombus Length in AIS
acquisition contrast-enhanced MR angiography was useful for delineation of the distal end of the thrombus in one case in our series. Noncontrast CT was uncommonly used to measure clot length. Riedel et al (10) demonstrated that acute thrombus can be visualized reliably using noncontrast CT data reconstructed to r 2.5 mm slice width. Only three patient lesions in the present series were measured using noncontrast CT (all done using 2.5-mm slice thickness), and these three lesion measurements all were confirmed on DSA. Measurement of clot length on DSA is limited by its two-dimensional view of the lesion, and only two patient lesions were measured using DSA alone in our series. There are several limitations to the present study. The main limitation is the small number of patients. Because
Jindal et al
’
JVIR
of the study’s small sample size, the relationship of symptomatic intracranial hemorrhage after the procedure to length of thrombus could not be determined reliably. Nevertheless, there is enough statistical power to identify multiple predictive factors within the study. Future studies with larger patient numbers are necessary to elucidate further the relationships among the variables presented here. IV tPA was administered before mechanical thrombectomy in 65% of patients, and its effects on recanalization rates and outcomes are not specifically accounted for or delineated in our study. A mean time of 44 minutes between initial cross-sectional imaging and DSA may have allowed for progression or regression of thrombus, particularly in patients who received IV tPA; although this mean time interval is
Figure 6. (a) Coronal maximum intensity projection CT angiography image demonstrates the proximal and distal ends (arrows) of the occluded right internal carotid artery terminus and right middle cerebral artery; the distal end of the thrombus is delineated as a result of retrograde contrast material filling within the right middle cerebral artery to the level of the distal end of the occlusion. Anteroposterior views of (b) early-phase and (c) late-phase right internal carotid artery injection DSA images show an occluded right internal carotid artery terminus in the early phase (arrow, b) and outlining of both proximal and distal ends of the occlusion (arrows, c) as a result of retrograde contrast material filling to the level of the distal end of occlusion in the right middle cerebral artery in the late phase. (d) Anteroposterior view of right internal carotid artery injection DSA image after endovascular stent retrieval thrombectomy demonstrates complete recanalization of the right internal carotid artery and right middle cerebral artery.
Volume 25
’
Number 10
’
October
’
2014
relatively short, this possibility was not accounted for in some cases because clot length could not be reliably measured on all DSA imaging. For example, in a few cases, a proximal occlusion without distal delineation of thrombus was seen on DSA. Additionally, the study is limited by imperfect clot length measurement tools as outlined earlier. Use of multiple imaging modalities to measure clot length introduces inhomogeneity in data collection. However, SWI was used in most cases, and two imaging modalities were used in most of the remaining cases to measure thrombus length. In our estimation, the imaging methodology used in this series was able to estimate clot length accurately; four cases were excluded because clot length was deemed by either of two reviewers to be suboptimal. The study is also limited by the potential variability of appropriateness criteria used from one operator to the next, especially regarding the number of stent retrievals used to restore flow within an artery difficult to recanalize. However, the three endovascular operators in the single-center group generally practice under similar guidelines and use similar procedural techniques, including rate and tension with which the operator withdraws a stent retriever, in treating AIS. In conclusion, the data suggest that there is no relationship between thrombus length and number of stent retrievals, final TICI score, or functional neurologic outcomes at 90 days in the setting of stent retrieval thrombectomy for AIS. These results do not support a predictive value for thrombus length quantification in the evaluation of AIS. Further investigation with a
1557
larger number of patients is necessary to gain a better understanding of these relationships.
REFERENCES 1. Pereira VM, Gralla J, Davalos A, et al. Prospective, multicenter, singlearm study of mechanical thrombectomy using solitaire flow restoration in acute ischemic stroke. Stroke 2013; 44:2802–2807. 2. Nogueira RG, Lutsep HL, Gupta R, et al. for the TREVO 2 Trialists. Trevo versus Merci retrievers for thrombectomy revascularisation of large vessel occlusions in acute ischaemic stroke (TREVO 2): a randomised trial. Lancet 2012; 380:1231–1240. 3. Saver JL, Jahan R, Levy EI, et al. for the SWIFT Trialists. Solitaire flow restoration device versus the Merci Retriever in patients with acute ischaemic stroke (SWIFT): a randomised, parallel-group, non-inferiority trial. Lancet 2012; 380:1241–1249. 4. Moftakhar P, English JD, Cooke DL, et al. Density of thrombus on admission CT predicts revascularization efficacy in large vessel occlusion acute ischemic stroke. Stroke 2013; 44:243–245. 5. Shobha N, Bal S, Boyko M, et al. Measurement of length of hyperdense MCA sign in acute ischemic stroke predicts disappearance after IV tPA. J Neuroimaging 2013; 24:7–10. 6. Spiotta AM, Vargas J, Hawk H, et al. Hounsfield unit value and clot length in the acutely occluded vessel and time required to achieve thrombectomy, complications and outcome. J Neurointerv Surg In press; available online July 31, 2013. 7. Soize S, Barbe C, Kadziolka K, Estrade L, Serre I, Pierot L. Predictive factors of outcome and hemorrhage after acute ischemic stroke treated by mechanical thrombectomy with a stent-retriever. Neuroradiology 2013; 55:977–987. 8. Riedel CH, Zimmermann P, Jensen-Kondering U, Stingele R, Deuschl G, Jansen O. The importance of size: successful recanalization by intravenous thrombolysis in acute anterior stroke depends on thrombus length. Stroke 2011; 42:1775–1777. 9. Rovira A, Orellana P, Alvarez-Sabín J, et al. Hyperacute ischemic stroke: middle cerebral artery susceptibility sign at echoplanar gradient-echo MR imaging. Radiology 2004; 232:466–473. 10. Riedel CH, Jensen U, Rohr A, et al. Assessment of thrombus in acute middle cerebral artery occlusion using thin-slice nonenhanced computed tomography reconstructions. Stroke 2010; 41:1659–1664.
