J Neurosurg 119:988–995, 2013 ©AANS, 2013
Investigating the mechanisms of perioperative ischemic stroke in the Carotid Occlusion Surgery Study Clinical article Matthew R. Reynolds, M.D., Ph.D.,1 Robert L. Grubb Jr., M.D.,1,3 William R. Clarke, Ph.D., 4 William J. Powers, M.D., 5 Gregory J. Zipfel, M.D.,1–3 Harold P. Adams Jr., M.D., 6 and Colin P. Derdeyn, M.D.,1–3 for the Carotid Occlusion Surgery Study Investigators Departments of 1Neurological Surgery, 2Neurology, and 3Radiology, Washington University School of Medicine, St. Louis, Missouri; 4Clinical Trials Statistics and Data Management Center, University of Iowa College of Public Health, Iowa City, Iowa; 5Department of Neurology, University of North Carolina School of Medicine, Chapel Hill, North Carolina; and 6Department of Neurology, University of Iowa Carver School of Medicine, Iowa City, Iowa Object. The Carotid Occlusion Surgery Study (COSS) was a large, prospective clinical trial that examined whether superficial temporal artery–middle cerebral artery (STA-MCA) bypass, in addition to best medical therapy, reduced the risk of ipsilateral ischemic stroke in patients with carotid artery occlusion and hemodynamic cerebral ischemia. Despite improved cerebral hemodynamics and excellent bypass graft patency rates, COSS failed to show a benefit for the surgical group with respect to ipsilateral stroke recurrence at 2 years after treatment. This was due to a lower than expected rate of recurrent ipsilateral stroke in the medically treated group and a high rate of perioperative ipsilateral strokes in the surgical group. Critics of the trial have cited surgeon inexperience and technical difficulties related to the performance of the bypass graft as a leading cause of failure of the trial. Methods. The authors retrospectively identified all patients from the COSS with an ipsilateral, perioperative (< 30 days) ischemic stroke after STA-MCA cortical branch anastomosis. Study records, operative notes, stroke adjudication forms, and imaging studies were reviewed. Ischemic strokes were characterized as bypass graft related or non–bypass graft related based on clinical and radiographic findings. Results. Fourteen of 93 surgically treated patients experienced an ipsilateral, perioperative ischemic stroke. Postoperatively, the mean oxygen extraction fraction (OEF) ratio between the symptomatic and asymptomatic cerebral hemisphere significantly improved in these patients (1.30 ± 0.18 preoperative vs 1.12 ± 0.11 postoperative; p = 0.02), but did not normalize. In this cohort, total MCA occlusion time during the anastomosis (54.3 ± 23.5 minutes) was no different from the MCA occlusion time in those surgical patients who did not have a perioperative stroke (45.4 ± 24.2 minutes, p = 0.2). Bypass graft patency rates in patients with a perioperative stroke were 92% at 30 days (11 of 12 patients with patency data) and 83% at last follow-up visit (10 of 12 patients with patency data). These patency rates were not significantly different from those achieved at 30 days (100%; 76 of 76 patients with patency data; p = 0.14) and at last follow-up (99%; 71 of 72 patients with patency data; p = 0.052) in patients without a perioperative stroke. Eighty-six percent (12 of 14 patients) of strokes were likely attributable to factors unrelated to the STA-MCA anastomosis. Only 21% of strokes (3 of 14 patients) were in the territory of the recipient vessel and likely related to technical performance of the anastomosis itself. One patient was thought to have dual stroke mechanisms. Conclusions. Only a small minority of ipsilateral, perioperative ischemic strokes in the COSS could be attributed to technical problems of the bypass anastomosis. The majority of ischemic strokes could not be ascribed to this cause and were most likely due to patient hemodynamic fragility and the inability of patients to tolerate surgery. (http://thejns.org/doi/abs/10.3171/2013.6.JNS13312)
Key Words • Carotid Occlusion Surgery Study • COSS • perioperative • ischemic stroke • EC-IC bypass • STA-MCA bypass • vascular disorders Abbreviations used in this paper: ACA = anterior cerebral artery; ACoA = anterior communicating artery; COSS = Carotid Occlusion Surgery Study; DWI = diffusion-weighted imaging; EC-IC = extracranial-to-intracranial; ICA = internal carotid artery; MCA = middle cerebral artery; OEF = oxygen extraction fraction; PCA = posterior cerebral artery; PCoA = posterior communicating artery; POD = postoperative day; STA = superficial temporal artery.
Carotid Occlusion Surgery Study was a prospective clinical trial designed to test the hypothesis that EC-IC bypass—in addition to best medical therapy—reduces recurrent ipsilateral ischemic events in patients with recently symptomatic carotid artery occlusion and hemodynamic insufficiency.6 The results of this study showed that the 2-year rates for ipsilateral stroke he
J Neurosurg / Volume 119 / October 2013
Mechanisms of ischemic stroke in COSS recurrence were similar for the surgical and nonsurgical groups (21.0% vs 22.7%, respectively; p = 0.78). In addition, the 30-day event rate for ipsilateral ischemic stroke was significantly higher in the surgical group than in the nonsurgical group (14.4% vs 2.0%, respectively). A second report detailed the surgical results of the STA-MCA bypass procedures.5 This study showed that the surgical group exhibited 1) high rates of bypass graft patency (98% at the 30-day postoperative visit and 96% at the last follow-up examination), 2) improved cerebral hemodynamics as measured by OEF on PET, and 3) much lower rates of recurrent ipsilateral stroke after POD 2 as compared with the nonsurgical group (9% vs 22.7% at 2 years, respectively). No patient characteristics or intraoperative variables were predictors of ipsilateral, perioperative stroke.5 A recent review of the COSS voiced a number of criticisms.1 Primary among these criticisms was the high perioperative morbidity rate, which was attributed to surgeon inexperience. In an effort to delineate whether the perioperative ischemic strokes in COSS were due to poor technical performance of the bypass graft anastomosis, we reviewed the clinical and radiographic findings in these cases to better understand the nature and mechanism of these adverse events. The COSS
The COSS was a prospective, parallel-group, 1:1 randomized, open-label, blinded-adjudication treatment trial. The study design—as well as detailed statistical analyses of the relationship between clinical, procedural, and operator variables and the risk of recurrent ischemic events— has recently been published.6 Briefly, patients with either a transient ischemic attack or ischemic stroke in the territory of a completely occluded carotid artery within 120 days were eligible to undergo cerebral hemodynamic testing. Those patients with an increased OEF—as measured by PET in the symptomatic cerebral hemisphere—were randomly assigned to best medical therapy or best medical therapy in addition to STA-MCA cortical branch anastomosis. One-hundred ninety-five patients were randomized: 97 to the surgical group and 98 to the medical group. Four patients randomized to the surgical group did not receive surgery. The surgical patients underwent an STA-MCA cortical branch anastomosis at a COSS-approved center. Details of the EC-IC bypass procedure and the certification process for participating neurosurgeons have been described in earlier reports.5,6 Primary outcome measures were 1) all stroke and death within 30 days of surgery, and 2) recurrent, ipsilateral ischemic stroke after 2 years. All ischemic strokes were independently confirmed (based on clinical and radiographic findings) by a joint neurosurgical and neurological blinded adjudication committee. Informed Consent
All participants in the COSS provided written informed consent according to local institutional review board and human research protection office regulations and study protocol requirements.
J Neurosurg / Volume 119 / October 2013
Classification of Ischemic Strokes
The authors conducted a retrospective, post hoc analysis of data collected from the COSS. All patients with adjudicated, ipsilateral ischemic strokes within 30 days of STA-MCA cortical branch anastomosis were identified, and their records, progress notes, discharge summaries, operative details, stroke adjudication forms, as well as pre- and postoperative cerebral imaging studies (including CT, MRI, and cerebral catheter angiograms) were reviewed. Operative notes were reviewed for any evidence of technical problems during the procedure. Periprocedural ischemic strokes were categorized by consensus of the primary investigators based on careful assessment of the clinical and radiographic data. Ischemic strokes were categorized as 1) likely related to technical performance of the cortical branch anastomosis (occlusion of the bypass graft or infarction in the territory of the recipient MCA), or 2) likely not related to technical aspects of the cortical branch anastomosis (thromboembolic, hemodynamic, other). When more than one mechanism contributed to the clinical findings, the stroke was defined as mixed. Strokes were considered disabling if the modified Barthel Index score was less than 12/20 at the first scheduled return visit more than 3 months after the stroke occurred.
Categorical variables were displayed as counts and rates, and continuous variables were displayed as means ± SDs. Univariate analysis was used to compare baseline values and intraoperative variables between those patients who experience a perioperative, ipsilateral ischemic stroke and those patients who did not. We compared the 2 groups using generalized Fisher exact tests for categorical variables and Student 2-tailed t-tests for continuous variables that were approximately normally distributed. The level of statistical significance was set at p < 0.05.
Fourteen patients in the COSS experienced an ipsilateral, perioperative (< 30 days) ischemic stroke (Figs. 1–3), resulting in a 30-day ischemic event rate of 15% (ischemic events occurring in 14 of 93 patients who were randomized to and underwent surgery). In this cohort, the patients’ mean age was 60.3 ± 7.2 years, and the male/ female ratio was 11:3 (Table 1). One patient (Case 14) had bilateral ICA occlusion on preoperative angiography, and 1 patient (Case 5) harbored minor (< 50%) contralateral ICA stenosis. While only 2 patients had an intact circle of Willis (that is, patent ACoA and PCoA), most of the patients seemed to have adequate collateral circulation to the symptomatic cerebral hemisphere (Table 1). Most ipsilateral, perioperative ischemic strokes (12 [86%] of 14 cases) occurred within 2 days of surgery (mean 1.7 ± 3.8 days after surgery; range 0–15 days after surgery) (Table 2). Postoperatively, the mean OEF ratio significantly improved in all 14 patients (preoperative OEF ratio 1.30 ± 0.18, postoperative ratio 1.12 ± 0.11; p = 0.2), but the ratio did not normalize (normal OEF ratio 989
M. R. Reynolds et al.
Fig. 1. Representative postoperative brain imaging for Cases 2–5. In all figures, the case numbers correspond to the numbers used in Tables 1–3. A and B: Case 2. CT scans obtained on POD 0 (the day of surgery) (A) and POD 2 (B) showing an evolving recipient artery territory ischemic infarct in the left temporal and parietal lobes (arrowheads). C and D: Case 3. Serial diffusion-weighted MR sections obtained on POD 1 illustrating multiple small areas of DWI hyperintensity within the cortical border-zone areas. E and F: Case 4. Serial diffusion-weighted MR sections obtained on POD 16 demonstrating small areas of DWI hyperintensity in the left frontal border-zone areas (E, arrowhead) and a larger lesion (2 × 3 cm) in the medial left occipital lobe (F, arrowhead). G and H: Case 5. Serial CT sections taken on POD 0 showing a large, confluent hypodensity in the right temporoparietal area (arrowheads).
Fig. 2. Representative postoperative brain imaging for Cases 6–9. The case numbers correspond to the numbers used in Tables 1–3. A and B: Case 6. Diffusion-weighted MR image (A) obtained on POD 1 depicting a large, confluent hyperintense region in the right ACA and MCA territories, and CT scan (B) obtained on POD 4 illustrating large, bilateral thromboembolic infarcts in the ACA and MCA territories. C and D: Case 7. Diffusion-weighted MR images obtained on POD 6 showing several small areas of hyperintensity (arrowheads) in the right posterior frontal and temporal regions. E and F: Case 8. Diffusionweighted MR image (E) obtained on POD 1 demonstrating a hyperintense area within the left frontal lobe, and cerebral catheter angiogram (F; left vertebral artery contrast injection, lateral view) obtained on POD 1 showing a large left PCoA (arrowhead) supplying the ACA branches. G and H: Case 9. Diffusion-weighted MR images obtained on POD 1 illustrating several small areas of periventricular left hemispheric white matter changes (arrowheads).
J Neurosurg / Volume 119 / October 2013
Mechanisms of ischemic stroke in COSS
Fig. 3. Representative postoperative brain imaging for Cases 10–14. Patient numbers correspond to the numbers used in Tables 1–3. A: Case 10. Diffusion-weighted MR image obtained on POD 1 showing a small DWI hyperintensity (arrowhead) in the right frontal cortical border zone. B and C: Case 11. Preoperative noncontrast CT scan (B) showing old, left, deep white matter ischemic infarcts, and CT scan from POD 3 (C) demonstrating a new left posterior frontal hypodensity (arrowhead) in the recipient artery territory. D–F: Case 12. MR images (D and E) obtained on POD 3 depicting scattered cortical DWI hyperintensities in the left frontoparietotemporal border-zone areas (D); as well as a large, confluent temporoparietooccipital infarct in the territory of the recipient artery (E); and cerebral angiogram (F) obtained on POD 7 (left carotid artery contrast injection, lateral view) showing robust filling of the MCA branches through a patent graft with luxury perfusion of the posterior parietal region in the area of the recipient artery (arrowheads). G: Case 13. MR image obtained on POD 4 showing a large FLAIR hyperintense lesion involving the entire right MCA territory. H: Case 14. Diffusion-weighted MR image obtained on POD 3 illustrating a large, confluent, right-sided MCA core infarct.
1.062).4 This reduction in OEF ratio was similar to that achieved in the entire group of surgical patients (preoperative OEF ratio 1.26 ± 0.14, postoperative ratio 1.11 ± 0.10).5 Comparison of the mean preoperative and postoperative OEF ratios in the group of surgical patients who had a perioperative ischemic stroke versus the mean OEF ratios in patients who did not have a perioperative ischemic stroke showed no statistically significant betweengroups difference (preoperative: 1.25 ± 0.13 vs 1.30 ± 0.18, respectively [p = 0.2]; postoperative: 1.11 ± 1.0 vs 1.11 ± 0.10, respectively [p = 0.8]). The mean total MCA occlusion time during the STAMCA cortical branch anastomosis in the 14 patients with perioperative stroke was 54.3 ± 23.5 minutes (Table 2) and was no different from the mean MCA occlusion time in those surgical patients who did not have a stroke (45.4 ± 24.2 minutes, p = 0.2). Rates of bypass graft patency in those patients with a perioperative stroke were 92% (11 of 12 patients with patency data) at 30 days and 83% (10 of 12 patients with patency data) at the last follow-up visit (Table 2). These bypass patency rates are not significantly different from those achieved at 30 days (100%, 76 of 76 patients with patency data, p = 0.14) and at last follow-up (99%, 71 of 72 patients with patency data, p = 0.052) in patients who did not have a perioperative stroke. In only 21% of patients (3 of 14) were the perioperative strokes attributable to technical complications of the J Neurosurg / Volume 119 / October 2013
STA-MCA anastomosis, as evidenced by an infarct within the recipient MCA territory (Figs. 1–3, Table 3). In 86% of patients (12 of 14), the perioperative strokes were attributable to factors unrelated to surgical technical aspects of the STA-MCA anastomosis. For these patients, the pattern of infarction involved the cortical border-zone region (in 6 [50%] of 12 patients), a large MCA core region (in 4 [33%] of 12 patients), or large bilateral infarcts in multiple vascular territories (in 1 [8%] of 12 patients). The pattern of infarction in 1 patient (Case 1) could not be determined, but was likely due to a cortical border-zone infarct. One patient experienced dual stroke mechanisms that manifested radiographically as cortical border-zone infarcts (non–bypass graft related) and a confluent infarction within the territory of the MCA recipient branch (bypass graft related). One perioperative stroke was fatal and 2 were classified as disabling strokes at the first follow-up visit that was more than 3 months after the stroke occurred.
We identified the types and mechanisms of ipsilateral, perioperative ischemic strokes in the COSS in an effort to elucidate the underlying pathophysiological events leading to stroke and to determine whether these events were attributable to technical performance of the STA-MCA anastomosis. Only 21% were attributable to technical problems 991
65, M 58, M 61, F 62, M 65, F 39, M 55, M 57, M 66, M 60, M 69, F 62, M 61, M 64, M
1 2† 3 4 5 6 7 8 9 10 11† 12† 13 14
stroke: aphasia, weakness stroke: aphasia, weakness, sensory loss, visual field cut TIA: aphasia, weakness, ataxia stroke: aphasia, weakness stroke: dysarthria, weakness, sensory loss stroke: dysarthria, weakness stroke: dysarthria, weakness, sensory loss stroke/TIA: aphasia, weakness stroke: dysarthria, weakness, numbness stroke/TIA: dysarthria, amaurosis fugax, optic ischemia stroke/TIA: dysarthria, weakness, spasticity stroke/TIA: aphasia, weakness stroke/TIA: dysarthria, weakness stroke: dysarthria, weakness
Qualifying Event for COSS
* EC-OA = extracranial-to-ophthalmic artery; Pt = patient; TIA = transient ischemic attack. † Patients with ischemic infarcts likely related to failure of cortical branch anastomosis.
Pt Age (yrs), Sex
Case No. normal not visualized normal normal yes ( 10 hours in operating room.
with the performance of the STA-MCA cortical branch anastomosis. While it is possible that the cortical borderzone infarcts and large MCA core infarcts seen in the rest of the patients were the result of cerebral hypoperfusion (hemodynamic insult) in the pre- or postoperative period, no significant episodes of hypotension were reported in these patients during surgery and in the immediate preand postoperative periods.5 The patient with large, bilateral confluent ischemic infarcts in multiple vascular territories was shown to have an embolus by postoperative cerebral catheter angiography and most likely experienced a thromboembolic event from a ruptured ipsilateral ICA atherosclerotic plaque that was seen on postmortem examination. Collectively, our observations suggest that despite excellent surgical technique, improved cerebral hemodynamics, and high rates of bypass graft patency, these patients still experienced a high rate of perioperative strokes. The COSS had rates of perioperative, ipsilateral ischemic strokes similar to those seen in previous EC-IC bypass trials,2,3 suggesting that despite advances in operative technique, neuroanesthesia, intraoperative hemodynamic monitoring/management, neurological critical care, and standardized postoperative management, patient outcomes remain similar. Thus, the tenuous cerebral hemodynamic state of these patients and their inability to tolerate surgery, rather than poor surgical techniques and inexperienced surgeons, seem to be the main cause of the perioperative ischemic events seen in the COSS. J Neurosurg / Volume 119 / October 2013
STA-MCA cortical branch anastomosis for patients with recently symptomatic carotid artery occlusion and hemodynamic cerebral ischemia was not effective in reducing recurrent, ipsilateral ischemic events in the COSS. One of the factors that led to the negative outcome of the study was the high rate of ipsilateral strokes in the first 2 days after surgery. The mechanism of these perioperative ischemic strokes was attributable to hemodynamic factors and the inability of patients with a symptomatic ICA occlusion and impaired hemodynamics in the cerebral hemisphere distal to the ICA occlusion to tolerate surgery in the majority of cases, rather than to poor surgical techniques and inexperienced surgeons. Disclosure Dr. Derdeyn reports an ownership interest in Pulse Therapeutics, a consultant relationship with W. L. Gore and Associates, and receiving support for the study described from Microvention, Inc. Dr. Adams reports consultant relationships with Merck, Medtronic, and Pierre Fabre. Author contributions to the study and manuscript preparation include the following. Conception and design: Reynolds, Grubb, Clarke, Powers, Derdeyn. Acquisition of data: Reynolds, Grubb, Powers, Derdeyn. Analysis and interpretation of data: all authors. Drafting the article: Reynolds, Grubb, Derdeyn. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Reynolds. Statistical analysis: Clarke. Study supervision: Grubb, Derdeyn.
65, M 58, M
61, F 62, M
69, F 62, M
New Postop Cerebral Imaging Findings*
no new hypodensity large hypodensity in lt temporoparietal region; lesion also seen on FLAIR/DWI; lack of graft patency on 30-day postop angiogram multiple DWI/FLAIR hyperintensities scattered throughout lt hemisphere in cortical border-zone regions multiple DWI/FLAIR hyperintensities in lt frontoparietal area in cortical border-zone regions; larger DWI lesion in medial lt occipital lobe large confluent hypodensity in rt frontoparietal region; lesion larger than expected from infarction of recipient artery territory alone large confluent DWI/FLAIR hyperintensity in frontoparietotemporal region in ACA & MCA territories; subsequent imaging showed large hypodensities in bilateral MCA & ACA territories; angiogram showed embolus in lt M1, stenosis of lt A1, & occlusion of rt ICA lt face, arm & leg weakness; numbness of several small DWI/FLAIR hyperintensities in rt posterior frontal & parietal cortical border-zone areas lt hand, face aphasia large lt frontal DWI/FLAIR hyperintensity in MCA territory; graft patent on 30-day postop angiogram, but not on 6-mo postop angiogram rt face, arm & leg weakness; rt tongue several small DWI/FLAIR hyperintensities in lt frontal cortical border-zone regions deviation, aphasia, confusion lt face, arm & leg weakness; hemisensory multiple small DWI/FLAIR hyperintensities in rt frontal & parietal cortical border-zone regions loss rt hemiparesis, aphasia hypodense lesion in lt posterior frontal lobe underneath craniotomy bone flap rt face, arm & leg weakness; hemisensory multiple DWI/FLAIR hyperintensities in lt frontotemporoparietal cortical border-zone areas involving ACA & loss, aphasia MCA territories; also, large confluent DWI lesion in lt posterior temporoparietal area underlying craniotomy site; angiogram showed robust filling of distal MCA branches through a patent bypass graft & luxury perfusion to posterior parietal lobe near recipient artery lt face, arm & leg weakness; lt tongue large area of DWI/FLAIR hyperintensity in rt frontotemporoparietal region involving entire MCA territory; lesion deviation, dysarthria larger than expected from infarction of recipient artery territory alone lt hemiparesis, hemineglect, apraxia large confluent DWI/FLAIR hyperintense lesion in frontotemporoparietal area consistent w/ MCA core infarct; lesion larger than expected from infarction of recipient artery territory alone
rt facial, arm & leg weakness; dysarthria rt hemiparesis, hemisensory loss, speech difficulties lethargic, unresponsive, rt hemiparesis rt facial weakness, aphasia, gait ataxia, visual field cut lt face, arm & leg weakness; hemisensory loss, hemineglect, visual field cut lt hemiparesis, hemineglect
Postop Ischemic Stroke Clinical Signs
* Representative images are shown in Figs. 1–3. † Patients with ischemic infarcts likely related to failure of cortical branch anastomosis.
Pt Age (yrs), Sex
Stroke Related to EC-IC Bypass
TABLE 3: Postoperative ischemic stroke clinical signs, cerebral imaging characteristics, and proposed stroke mechanisms for those patients in the COSS who experienced ipsilateral, perioperative (< 30 days) ischemic strokes
M. R. Reynolds et al.
J Neurosurg / Volume 119 / October 2013
Mechanisms of ischemic stroke in COSS References 1. Amin-Hanjani S, Barker FG II, Charbel FT, Connolly ES Jr, Morcos JJ, Thompson BG: Extracranial-intracranial bypass for stroke—is this the end of the line or a bump in the road? Neurosurgery 71:557–561, 2012 2. Barnett HJ, Fox A, Hachinski V, Haynes B, Peerless SJ, Sackett D, et al: Further conclusions from the extracranial-intracranial bypass trial. Surg Neurol 26:227–235, 1986 3. Gibbs JM, Wise RJ, Thomas DJ, Mansfield AO, Russell RW: Cerebral haemodynamic changes after extracranial-intracranial bypass surgery. J Neurol Neurosurg Psychiatry 50:140– 150, 1987 4. Grubb RL Jr, Derdeyn CP, Fritsch SM, Carpenter DA, Yundt KD, Videen TO, et al: Importance of hemodynamic factors in the prognosis of symptomatic carotid occlusion. JAMA 280:1055–1060, 1998 5. Grubb RL Jr, Powers WJ, Clarke WR, Videen TO, Adams
J Neurosurg / Volume 119 / October 2013
HP Jr, Derdeyn CP: Surgical results of the Carotid Occlusion Surgery Study. Clinical article. J Neurosurg 118:25–33, 2013 6. Powers WJ, Clarke WR, Grubb RL Jr, Videen TO, Adams HP Jr, Derdeyn CP: Extracranial-intracranial bypass surgery for stroke prevention in hemodynamic cerebral ischemia: the Carotid Occlusion Surgery Study randomized trial. JAMA 306:1983–1992, 2011 (Erratum in JAMA 306:2672, 2011) Manuscript submitted February 15, 2013. Accepted June 18, 2013. Please include this information when citing this paper: published online August 2, 2013; DOI: 10.3171/2013.6.JNS13312. Address correspondence to: Matthew R. Reynolds, M.D., Ph.D., Barnes-Jewish Hospital, Department of Neurological Surgery, Campus Box 8057, 660 S. Euclid Ave., St. Louis, MO 63110. email: [email protected]