Article Text

Download PDFPDF

Case series
Incidence and mechanisms of stroke after permanent carotid artery occlusion following temporary occlusion testing
  1. Justin T Whisenant1,
  2. Yasha Kadkhodayan2,
  3. DeWitte T Cross III1,3,
  4. Christopher J Moran1,3,
  5. Colin P Derdeyn1,3,4
  1. 1Department of Interventional Neuroradiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri, USA
  2. 2Department of Interventional Neuroradiology, Abbott Northwestern Hospital Neuroscience Institute, Consulting Radiologists Ltd, Minneapolis, Minnesota, USA
  3. 3Department of Neurosurgery, Washington University School of Medicine, St. Louis, Missouri, USA
  4. 4Department of Neurology, Washington University School of Medicine, St. Louis, Missouri, USA
  1. Correspondence to Dr Colin P Derdeyn, Mallinckrodt Institute of Radiology, Washington University, 510 S. Kingshighway Blvd., St. Louis, MO 63110, USA; derdeync{at}


Background and purpose Patients who require sacrifice of the internal carotid artery (ICA) have a substantial risk of stroke, despite preoperative testing with temporary balloon occlusion (TBO). The purpose of this study is to examine the incidence and mechanisms of stroke after permanent carotid artery occlusion in this population.

Methods Consecutive patients undergoing TBO testing from March 2002 to December 2011 were identified. The protocol included 30 min of balloon occlusion, continuous intraprocedural neurological assessment, angiographic imaging of collateral flow during the occlusion, and perfusion imaging. Clinical records were reviewed for procedure results, procedural complications, and the incidence and causes of stroke, transient ischemic attack (TIA) and death over 6 months. Strokes were categorized as thromboembolic or hypoperfusion based on available clinical and imaging data.

Results One hundred and fifty carotid occlusion tests were performed during the study period, including 84 women and 66 men. No procedural strokes were recorded. Thirty-seven patients (25%) had permanent occlusion of the tested ICA. Six of the 37 patients had ipsilateral stroke (16.2%) and three experienced TIA (8.1%). Two strokes occurred in the immediate postoperative period (thromboembolic), two strokes occurred within days of ICA occlusion (hypoperfusion), and two strokes occurred at least 30 days from the time of ICA occlusion (thromboembolic).

Conclusions The rate of ischemic stroke following carotid sacrifice remains high and most strokes are thromboembolic in nature. Our testing protocol did not eliminate the risk of hypoperfusion-related stroke. Delayed venous phase by angiography may be a better indicator of hemodynamic tolerance than perfusion imaging.

  • Angiography
  • Balloon
  • Complication
  • Stroke

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Carotid sacrifice is necessary in certain clinical settings, primarily in head and neck surgery for tumors involving the common or internal carotid artery (ICA), and less frequently now in the case of large wide-necked ICA aneurysms with the advent of flow diverters.1 However, not all patients will tolerate unilateral carotid artery occlusion, depending on the adequacy of the circle of Willis collaterals. Without preoperative testing of collateral flow, the rate of ischemic stroke after permanent ICA occlusion ranges from 17% to 30%.2

A number of tests have been developed to assess the adequacy of collateral flow. Manual carotid compression at the bedside has been used,3–5 but most frequently endovascular temporary balloon occlusion (TBO) is performed with clinical, angiographic and/or imaging assessment during and after the procedure. The exact methodology varies considerably.2 ,6–9 Some investigators use induced hypotension or vasodilatory challenge.10–14 The location of the occlusion balloon is also variable (relative to potential collaterals), and there is variable use of perfusion imaging.6–8 ,15–26 Our protocol includes 30 min of occlusion time, intraprocedural neurological assessment, angiographic evaluation of the adequacy of collateral flow, and administration of Tc-99m bicisate (Neurolite; Bristol-Myers Squibb, North Billerica, Massachusetts, USA) during occlusion with single photon emission CT (SPECT) brain perfusion imaging performed immediately after the test occlusion.

Patients who pass a TBO remain at risk of ischemic stroke after carotid sacrifice due to both hemodynamic and thromboembolic mechanisms. This risk ranges from 4.7% to 25%, as demonstrated in several case series.5–7 ,13 ,14 ,27–30 The purpose of this study was to determine the risk of stroke with sacrifice after passing a balloon test occlusion using our protocol, and to investigate the possible mechanisms of stroke occurring in this population. This information will be useful for clinical decision-making in patients where other treatment options are available (eg, flow diversion), as well as to guide further investigation in reducing the risk of stroke after sacrifice.


All consecutive patients undergoing carotid artery TBO tests at our institution from March 2002 to December 2011 were identified from a prospective quality assurance database for all neuroendovascular procedures. The medical record was reviewed for the following information: indication for TBO, results of the intraprocedural neurological assessment, intraprocedural angiographic findings, SPECT results, complications directly related to the diagnostic procedure, and the date and method (surgical or endovascular) of permanent ICA occlusion, if performed.

Procedural technique and intraprocedural angiographic assessment

TBO procedures were performed by one of four experienced interventional neuroradiologists assisted by a fellow. Most procedures were performed with local and systemic analgesia only; light sedation was used in some cases as determined by the individual operators. Following placement of 5 or 6 Fr femoral artery sheaths bilaterally, initial angiography was performed in the test ICA in order to evaluate the ipsilateral cerebral circulation and to determine the position of the occlusion balloon. Intravenous heparin was used to maintain an activated clotting time of 250–350 s.

A compliant low pressure balloon (HyperForm; Covidien, Irvine, California, USA) was placed in the target ICA via a 5 or 6 Fr guide catheter (Envoy; Codman, Raynham, Massachusetts, USA). In most instances the balloon was placed across the ophthalmic artery, if possible. Occasionally the occlusion balloon was placed in the common carotid artery for technical reasons.

After balloon inflation and angiographic confirmation of complete test artery occlusion, the contralateral carotid and either one or both vertebral and/or subclavian arteries were selected for angiography. The presence of circle of Willis or other collateral sources of flow were identified and the qualitative presence or absence of cortical venous delay was documented.31 Mild delay was categorized as a delay of 1–2 s in the appearance of the first vein on the occluded hemisphere and moderate delay was categorized as 2–4 s. These parameters were reported qualitatively or quantitatively and could not be recalculated from saved images. After publication of the paper by Abud et al,31 quantitative measurements of venous delay were included more frequently.

Depending on patient tolerance, the balloon was left inflated for 30 min. After angiographically confirming continued occlusion at 30 min, the balloon was deflated and removed. Systemic anticoagulation was reversed with intravenous protamine sulfate. Hemostasis was achieved after sheath removal using manual compression or an arterial closure device. Patients were monitored in a recovery area for post-procedural neurological and groin site complications prior to discharge on the same day or return to the previous level of inpatient care. Patients were maintained on warfarin, antiplatelet agents, or no agent depending on the clinical situation.

Intraprocedural neurological assessment

Intraprocedural neurological assessment was routinely performed by a neurologist independent of those performing the procedure. On occasions when a neurologist was unavailable, the assessment was performed by the attending interventional neuroradiologist. An initial examination was routinely performed with the patient on the angiography table and the balloon in position within the test artery, but deflated. The initial examination included an assessment of motor and sensory function, visual fields, speech and cognition, and evaluation for cerebellar signs.32

After confirming the patient's baseline status, the balloon was inflated and the patient's level of consciousness was continually assessed through an ongoing verbal conversation. The neurological assessment was repeated intermittently over 30 min, if clinically tolerated. If at any time during the 30 min of balloon occlusion the patient developed a neurological deficit, the balloon was deflated and the remainder of the procedure aborted. A final neurological examination was routinely performed at 30 min, just prior to balloon deflation and removal.

Whole brain SPECT

Approximately 30 mCi Tc-99m bicisate was administered intravenously in the angiography suite during the period of balloon flow arrest. Immediately after TBO, patients were transported to the nuclear medicine department for whole brain SPECT perfusion imaging. If decreased activity (defined as >15% qualitative reduction in activity) was noted in the territory of the occluded ICA, patients returned for a baseline SPECT examination which was compared with the post-TBO examination to determine whether the decreased activity was attributable to the balloon occlusion.

ICA sacrifice

Permanent ICA occlusion was achieved by carotid ligation or endovascular sacrifice, depending on the clinical situation. Patients with head and neck cancers undergoing resection had local ligation and did not generally receive any postoperative antithrombotic or antiplatelet medication for stroke prophylaxis. Similarly, patients undergoing intracranial procedures with clip ligation of the ICA had no postoperative antithrombotic or antiplatelet medication for stroke prophylaxis. Endovascular sacrifice was performed under conscious sedation and full systemic heparinization. Methods varied between attending physicians and over time. All procedures were performed with proximal balloon occlusion. A distal occlusion just below the origin of the ophthalmic artery was the goal, but could not always be achieved owing to the anatomy. A coil mass was initially created with detachable coils followed by pushable fibered coils. Aplatzer occlusion devices (St Jude Medical, St Paul, Minnesota, USA) were used in some cases, as was ethylene vinyl alcohol (Onyx; ev3 Covidien, Plymouth, Minnesota, USA). Heparin was not reversed. In most cases, particularly early in this series, patients were anticoagulated with warfarin sodium for 6 months after the procedure. More recently, patients have been treated with dual antiplatelet agents prior to and after the procedure instead of anticoagulation.

Outcomes after permanent ICA occlusion

For patients who went on to have permanent ICA occlusion after TBO, additional clinical data were collected over the following 6 months by reviewing the electronic medical record and cross-sectional imaging, if available. The occurrence of three primary outcomes was recorded: ipsilateral ischemic stroke, TIA and death.

For stroke patients, the following data were also collected: timing of symptom onset, location of infarct, and medical management of the patient preceding and during the ischemic event. Based on these factors, a mechanism of infarct was postulated by consensus of the investigators. Acute onset of symptoms that did not respond to increasing blood pressure goals, with cortical or deep gray matter infarcts in single vascular territories, were postulated to be of thromboembolic etiology. A waxing and waning clinical course responsive to changes in blood pressure, with infarcts affecting the cerebral white matter and relatively sparing the cerebral cortex and deep gray matter, were postulated to be of hemodynamic etiology.33


One hundred and fifty patients underwent anterior circulation TBO from March 2002 to December 2011. This included 84 women and 66 men, with a mean age of 55.3 years (range 12–89). Indications for TBO varied. Most patients had a head and neck tumor that involved the carotid artery (n=86). Many had aneurysms or pseudoaneurysms of the carotid artery (n=62). One patient had an aggressive malignant otitis externa and one patient had an ICA laceration during trans-sphenoidal pituitary surgery.

The majority of patients (n=113, 75%) did not go on to have a permanent ICA occlusion, either because an alternative treatment strategy was chosen or because of failure of one or multiple components of the TBO test.

Thirty-seven patients (24%) went on to have permanent endovascular (n=30) or surgical (n=7) sacrifice of the tested ICA. All patients who went on to ICA sacrifice had follow-up of up to 6 months unless they died or suffered a stroke. Six of the 37 patients had an ischemic stroke (16.2%), 3 experienced a TIA without stroke (8.1%), and 28 patients had no ischemic event. Three deaths (8.1%) occurred in the 6 months after ICA occlusion. All deaths were determined to be due to pre-existing malignancy and not related to carotid sacrifice.

TBO procedural complications

There were no neurological complications related to the TBO procedure. One asymptomatic cervical ICA dissection occurred with balloon inflation at the C1–C2 level leading to a focal intimal flap without luminal narrowing or flow limitation. The procedure was aborted and the patient was treated with daily aspirin. The TBO test was completed 2 weeks later when angiography confirmed normal flow and no thrombus despite a small persistent flap. There were no groin hematomas requiring blood transfusion or surgery or resulting in admission for further observation.

Permanent ICA occlusion without ischemic complications

Twenty-eight patients who underwent preoperative TBO followed by permanent ICA occlusion had no ischemic complications during 6 months of follow-up (table 1). The mean age of these 28 patients was 55.6 years (range 22–84). Twenty-two of the 28 patients passed all three components of the TBO test; three patients failed at least one component of the test but still went on to have permanent ICA occlusion without ischemic complications. Of note, three patients had a mild (1–2 s) cortical venous delay between hemispheres (TBO angiography column in table 1). Balloon placement was in the common carotid artery in one of the 28 patients. This patient had a robust anterior communicating artery with no angiographic evidence of inflow (ie, ophthalmic collateral). The ICA was sacrificed. Three patients had subjective reduction in perfusion by SPECT, two of whom also had mild venous delay.

Table 1

Sacrificed without complication

Permanent ICA occlusion with ischemic complications

In the six patients who had ischemic strokes following permanent ICA occlusion, two patients passed all three components of the TBO test (table 2). The mean age of these six patients was 62.7 years (range 43–81). Two patients passed intraprocedural neurological assessment and SPECT but did not have angiography performed during balloon occlusion for technical reasons. One patient passed intraprocedural neurological and angiographic assessments but did not have a SPECT. One patient failed angiographically due to the presence of a 2 s cortical venous delay in the ipsilateral cerebral hemisphere, but passed by neurological assessment and SPECT. Balloon placement was in the ICA in all six patients. All three patients who had TIAs only passed all three components of the TBO test.

Table 2

Sacrificed with Ischemic Stroke

Etiology of stroke

In the six patients who had ischemic stroke following permanent ICA occlusion, two strokes occurred in the immediate postoperative period, two strokes occurred within days of ICA occlusion, and two strokes occurred at least 30 days from the time of ICA occlusion (table 2).

The two immediate postoperative complications were categorized as procedural complications. One of these occurred after clip placement near the right anterior choroidal artery for treatment of a large posterior wall ICA aneurysm. The patient developed immediate postoperative slurred speech and left arm weakness which improved with pressors and the patient was discharged with only a mild pronator drift. The second procedural complication occurred after endovascular ICA occlusion for treatment of a giant cavernous segment ICA aneurysm. The patient awakened from anesthesia with right hemiparesis and visual changes in the left eye. The symptoms improved, however, and there were no residual deficits at discharge. This patient had been fully heparinized throughout and after the procedure.

The two strokes that occurred within days of ICA occlusion were categorized as due to impaired cerebral hemodynamics. Both patients were normal after awakening from anesthesia but experienced severe neurological symptoms attributable to the occluded ICA in the early postoperative period. Symptoms waxed and waned with changes in blood pressure, improving with maneuvers to augment cerebral hemodynamics, as with hypertensive therapy. Both patients remained fully anticoagulated throughout these events. Catheter angiography showed no embolic occlusions. The infarcts affected the deep frontoparietal white matter on imaging. Finally, both patients failed to recover and ultimately had poor functional outcomes (figure 1).

Figure 1

Early hemodynamic stroke demonstrated by CT in a patient with progressive left-sided hearing loss over a 5-month period due to a petrous internal carotid artery pseudoaneurysm (patient 3 in table 2). The patient passed the temporary balloon occlusion clinically and by single photon emission CT (SPECT), but did not have angiography during the balloon occlusion for technical reasons. The patient underwent trapping of the internal carotid artery from below the ophthalmic artery origin to the internal carotid origin. The patient developed right hemiplegia, right facial droop, and aphasia in the early postoperative period. Symptoms waxed and waned and appeared to improve with pressors, but the patient continued to decline despite pressure support. Catheter angiography early in the clinical course showed no evidence of embolic occlusion. The patient slowly evolved confluent white matter infarction over a week by serial CT scans. A CT scan obtained one week after the procedure showed extensive white matter infarction.

The two delayed strokes were categorized as thromboembolic in etiology. The symptoms were relatively minor and were temporally related to changes in anticoagulation or antiplatelet therapy. One patient discontinued dual antiplatelet therapy due to non-compliance 2 months after the procedure (figure 2), and the other patient became symptomatic days after being switched from warfarin to aspirin 4.5 months after the procedure.

Figure 2

Late thromboembolic stroke to the internal capsule (patient 5 in table 2). The patient presented with facial nerve dysfunction. The patient passed all elements of the temporary balloon occlusion and underwent endovascular sacrifice from just below the ophthalmic artery origin of the internal carotid artery. The patient was discharged on aspirin and clopidogrel. Four and a half weeks after sacrifice the patient returned with acute onset of left-sided weakness and left-sided neglect. The patient had stopped taking clopidogrel one week prior to this event. (A) Symmetric arterial flow during balloon occlusion. A thalamotuberal branch was seen arising from a posterior communicating artery infundibulum that did not connect to the P1 segment of the posterior cerebral artery. (B) Magnetic resonance FLAIR images showed an internal capsule/anterior thalamic stroke attributed to a thromboembolic mechanism involving the thalamotuberal artery.


Carotid sacrifice is a useful and important procedure, primarily in the setting of head and neck malignancies. Balloon occlusion testing identifies a group of patients at lower risk for stroke with sacrifice, but it does not eliminate the risk.6–8 ,15–26 These data suggest that most strokes that occur during or after carotid artery sacrifice are thromboembolic and inherently not predictable from test occlusion. In addition, the protocol we used for TBO does not appear to eliminate the risk of stroke related to hemodynamic factors.

TBO testing can clearly be performed safely. There were no neurological complications in our series of 150 procedures. There was one non-flow-limiting asymptomatic dissection, making for a technical complication rate (0.7%) and permanent neurological complication rate comparable to that of diagnostic cerebral angiography.

There are various TBO protocols in the literature. The most universally performed is the angiographic assessment of collateral circulation during test balloon inflation (figure 3). Criteria have been proposed by several investigators evaluating the presence and timing of cortical venous delay between the ipsilateral test hemisphere and contralateral control hemisphere.31 ,34 ,35 Abud et al31 have reported that a threshold value of 3 s may be the maximum difference in cortical venous delay between hemispheres that is tolerable without increasing the risk of ischemic complications following permanent ICA occlusion. They found this cut-off accurate in a cohort of 60 patients who had both TBO and permanent ICA occlusion. Forty-four patients had exact symmetry of the venous phase, 10 had a delay of 1 s, and three had a delay of 2 s; none of these 57 patients had any periprocedural stroke. Three patients had a venous delay of 3 s. One of the three had a delayed watershed infarction. However, no follow-up beyond hospital discharge was reported. In our series, one of the two patients with probable hemodynamic stroke (patient 3 in table 2, figure 1) did not have angiographic assessment of collateral flow for technical reasons. In addition, many of the patients early in this series had qualitative evaluation of venous delay.

Figure 3

An example of venous delay. Anteroposterior images during temporary balloon occlusion of the right internal carotid artery and injection of the left common carotid artery. The method of Abud et al31 involves measurement of the time between the appearance of the first artery and the first vein in both hemispheres. (A) An early image from the angiographic run. Collateral flow is across the anterior communicating artery. The appearance of the first artery (chosen as the M2 branches) is captured on the same image 8.5 s into the run. The balloon (Hyperform; Microvention/Terumo Tustin, California, USA) has been placed across the ophthalmic artery and is positioned just proximal to the fetal origin of the posterior communicating artery. The microwire is visible in the proximal internal carotid artery. (B) A largely capillary phase in the left hemisphere with one definite vein (arrowhead) at 12.5 s, while contrast is still arterial on the right. The first vein on the right hemisphere appears at 16.0 s into the run, yielding a 3.5 s venous delay between the two hemispheres. This patient did not undergo sacrifice.

In an attempt to increase the sensitivity for hemodynamic impairment, brain perfusion imaging has been employed including whole brain SPECT, Xe-enhanced CT, positron emission tomography, MRI perfusion, as well as transcranial Doppler.7 ,11 ,21 ,22 ,24–26 ,36 These adjunctive techniques may be performed in conjunction with maneuvers that provoke ischemia, such as a hypotensive challenge. Subtle insufficiencies in cerebral hemodynamic reserve may be unmasked using hypotensive or acetazolamide challenge.10–12 Whether this additional information increases the predictive value of TBO is, however, unclear. The use of quantitative cerebral blood flow measurement has also been reported. Marshall et al reported on a series of 33 patients who had TBO testing followed by permanent ICA occlusion.25 During TBO the patients underwent standard neurological examinations, sustained attention testing, and quantitative cerebral blood flow measurements. Two scalp scintillation detectors recorded washout data after ipsilateral intracarotid injection of Xe-133 through a port at the tip of the ICA-occluding balloon. They found that cerebral blood flow of <30 mL/100 g/min during TBO was the only variable that predicted ipsilateral stroke over a mean follow-up period of 34 months after permanent ICA occlusion.

The SPECT protocol used in this series is an indicator of relative cerebral blood flow and probably reflects cortical blood flow rather than deep and periventricular white matter. Gray matter cerebral blood flow is four times the rate of white matter blood flow. The early non-procedural infarcts that occurred in our study population following permanent ICA occlusion predominantly affected the deep and periventricular white matter, the anatomical site of the end arterial beds of the cerebral circulation that are most susceptible to hemodynamic insufficiency.33 ,37 This has important implications for perioperative and postoperative patient management, which should include close attention to factors that may alter cerebral hemodynamics, specifically avoiding decreases in cerebral perfusion pressure and augmenting cerebral blood flow when necessary to reverse or reduce symptoms, should they occur. In addition, in this series relative decreases in SPECT were present in several patients who tolerated sacrifice.

The accuracy of TBO is dependent on technical anatomical considerations as well, specifically the site of balloon inflation. Lesley et al9 have shown that an additional 14% of patients intolerant to supraclinoid ICA sacrifice can be detected by performing the balloon occlusion at or beyond the ophthalmic artery origin. However, for patients being evaluated for cervical ICA sacrifice, procedural complexity can be minimized by test occluding the more easily accessible cervical ICA, most pertinent in patients with unfavorable para-ophthalmic ICA anatomy or in those who develop intolerable pain or cranial neuropathy upon para-ophthalmic balloon inflation. Of course, when test occluding the cervical ICA, it is important to evaluate for retrograde ophthalmic artery to ICA collateral flow during the balloon occlusion.

One limitation of our study is that cortical venous delay between hemispheres was not consistently quantified. However, three of the six patients in this series who had a stroke had no cortical venous delay at all. Furthermore, not all of the strokes were hemodynamic, and some were clearly related to a delayed thromboembolic phenomenon, emphasizing the importance of other factors such as anticoagulation or antiplatelet therapy which was not standardized in this study. Other limitations include the retrospective nature of the study and the small number of patients who underwent permanent ICA occlusion. A fundamental problem with all the literature on TBO is that patients who fail the test by one of many different protocols or techniques generally do not go on to sacrifice, making it very difficult to determine the accuracy of the procedure in predicting stroke risk.


TBO is a safe procedure, but the rate of ischemic stroke following carotid sacrifice in patients who have passed a TBO protocol of intraprocedural neurological assessment, angiographic testing, and SPECT brain perfusion imaging remains significant. Most of this stroke risk is from thomboembolic factors. This is important to consider when weighing the options of sacrifice and other treatment options. Early non-procedural cerebral infarcts affected the deep white matter, supporting a hemodynamic etiology. Angiographic parameters such as cortical venous delay may be more accurate than SPECT perfusion imaging in predicting tolerance to permanent occlusion.



  • Contributors JTW made substantial contributions to the conception and design of the work, the acquisition and analysis of the data, drafting of the manuscript, and final approval of the version to be published. YK made substantial contributions to the acquisition and analysis of the data, critical revising of the manuscript and final approval of the version to be published. DTC made substantial contributions to the acquisition and analysis of the data, critical revising of the manuscript and final approval of the version to be published. CJM made substantial contributions to the acquisition and analysis of the data, critical revising of the manuscript and final approval of the version to be published. CPD made substantial contributions to the conception and design of the work, acquisition and analysis of the data, drafting and revision of the manuscript, and final approval of the version to be published. In addition, CPD takes responsibility for the entirety of this work.

  • Competing interests JTW, YK and DTC report no competing interests. CJM serves as a consultant for Covidien Neurovascular, Stryker and Codman. CPD has the following relationships with industry: Pulse Therapeutics, Chair Scientific Advisory Board (stock options); Microvention, Angiographic Core Lab for LVIS trial (consultant); Penumbra, DSMB member for 3D separator trial (consultant); Silk Road, DSMB member for ROADSTER trial (consultant).

  • Ethics approval Washington University Human Research Protection Office.

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

  • Data sharing statement Any unpublished data related to this paper will remain with the corresponding author. They will not be available without additional IRB approval.