Incidence

Cerebrovascular disease is the second leading cause of death worldwide and is responsible for approximately 9.5% of all deaths. At least 750,000 strokes occur annually in the United States. Approximately 15% of strokes are fatal, 15% to 20% are severely disabling, and another 15% to 20% of patients who have stroke and recover have a subsequent disabling stroke in the future. Approximately 80% of these strokes are ischemic, and atherosclerotic disease of the cervical carotid artery is responsible for at least 40% of ischemic strokes¹³ (see also Chapter 97).

Etiology

Strokes secondary to carotid artery stenosis are a consequence of atheroembolization, or thromboembolism in the case of an occluding artery and distal propagation of thrombus. Small pieces of the carotid bifurcation plaque break off and embolize to the intracranial vessels, usually to the middle cerebral artery (MCA) in the anterior circulation. These strokes can also result from lesions in the common carotid artery (CCA) or in the distal or intracranial portion of the internal carotid artery (ICA). Strokes can also be secondary to a low flow state through the carotid artery (see Chapter 97).

Beta Blockers

On the basis of early small clinical trials, it was believed that all patients with vascular disease would benefit from perioperative beta blockade. Several large trials have been published, however, that challenge this recommendation. The largest of these, the Perioperative Ischemic Evaluation Study (POISE trial), randomly allocated patients undergoing noncardiac surgery to receive metoprolol or placebo preoperatively as well as for 30 days postoperatively. The investigators found a significant reduction in the number of myocardial infarctions (MIs) in the metoprolol group (4.2% vs. 5.7%). However, the group also found that the patients in the metoprolol group had increased overall mortality and stroke rate (3.1% vs. 2.3% and 1.0% vs. 0.5%, respectively).²⁵ On the basis of the findings of this pivotal trial and other data, the AHA issued a revised statement on the issue of perioperatively beta blockade. The current recommendation is to continue beta blockers in patients who are already taking them. There is a weak recommendation to start beta blockers if a patient has a history of coronary artery disease (CAD) or more than one clinical risk factor. For patients who are undergoing vascular surgery (including carotid endarterectomy) who have a single risk factor in the absence of documented CAD, there is a very weak recommendation to start beta blockade.²⁶ Perioperative beta blockade, with a goal of achieving a maximum heart rate of 60 to 80 beats per minute, is recommended in the SVS carotid guidelines.¹⁴

Antiplatelet Therapy

One of the most significant improvements in the management of cardiovascular disease in general has been the aggressive use of antiplatelet therapy, and early studies indicated a benefit of antiplatelet therapy in the perioperative period among patients undergoing CEA. Meta-analyses of antiplatelet therapy trials published by the U.K. Antiplatelet Trialists’ Collaboration in 1994 and 2002 concluded that antiplatelet therapy significantly reduces the incidence of stroke in high-risk patients, with a resultant 25% reduction in strokes overall.^27,28 The majority of these studies involved the use of aspirin or aspirin plus another antiplatelet agent. The benefit of acetylsalicylic acid (ASA) in reducing postoperative stroke and death after CEA was first shown in a secondary analysis from the North American Symptomatic Carotid Endarterectomy Trial (NASCET), a randomized controlled trial (RCT) that demonstrated the benefit of CEA in symptomatic patients with at least 50% ICA stenosis.⁹ In that trial, patients were prescribed up to 1300 mg ASA daily if tolerated. A subsequent association was found between perioperative stroke and death and the amount of ASA taken before surgery. The risk for perioperative stroke and death was 1.8% in patients taking 650 to 1300 mg ASA daily versus 6.9% in patients taking 0 to 325 mg daily. An RCT was subsequently conducted in which 232 patients undergoing CEA were randomly assigned to receive either ASA 75 mg or placebo preoperatively and for 6 months postoperatively.²⁹ Stroke at 30 days and at 6 months occurred in 0 and 2 ASA-treated patients, versus 7 and 11 placebo patients, respectively (P < .01), with no difference in bleeding complications.

To determine the optimal daily dose of preoperative ASA in reducing stroke and death associated with CEA, the ASA and Carotid Endarterectomy (ACE) trial randomly allocated 2849 patients to ASA doses of 81, 325, 650, or 1300 mg started preoperatively and continued for 3 months.³⁰ The combined end point of 30-day stroke, death, or myocardial infarction after CEA occurred in 5.4% of patients in the low-dose groups (81 and 325 mg) versus 7.0% of patients in the high-dose groups (P = .03). In an efficacy analysis focusing on patients not previously taking high-dose ASA, the combined end point was seen in only 3.7% of patients in the low-dose groups versus 8.2% of patients in the high-dose groups at 30 days (P = .002). Fewer hemorrhagic complications were also seen in the low-dose ASA groups, but this difference was not significant. One should note that a placebo group was not deemed appropriate for this study because of the widely accepted benefit of ASA in preventing stroke.

Other antiplatelet agents have not been as extensively studied in terms of their benefit in stroke reduction after CEA, but later studies of clopidogrel provide evidence that perioperative embolization is decreased. In one trial, 100 patients undergoing CEA who were taking 150 mg ASA daily were randomly assigned to receive in addition either clopidogrel 75 mg or placebo the night before surgery.³¹ The number of emboli detected by transcranial Doppler ultrasonography (TCD) within 3 hours of CEA was the main outcome measure. There was a tenfold reduction in the number of patients who experienced more than 20 microemboli in that period (odds ratio [OR] 10.2; 95% confidence interval [CI] 1.3 to 83.3; P < .01) in the group taking clopidogrel combined with ASA. No increase in bleeding complications or transfusion requirements was noted. The embolization rate was used as a surrogate for stroke because stroke occurs infrequently and the embolization rate has been well correlated with stroke risk during CEA. Other antiplatelet agents, such as ticlopidine and glycoprotein IIb/IIIa antagonists, though demonstrated to reduce stroke risk in nonsurgical patients, have not been specifically studied during CEA and generally have higher risk profiles than ASA and clopidogrel. On the basis of this level I evidence, and in light of the overall cardiovascular benefits, it is possible to recommend ASA, clopidogrel, or both for use before and after CEA.

On the other hand, a 2004 meta-analysis published several years ago found that antiplatelet therapy was not effective in preventing stroke and other vascular events after CEA.³² Further, a survey of the members of the Vascular Society of Great Britain and Ireland as well as the European Society for Vascular Surgery found for symptomatic patients, 5% (20/392) would stop aspirin and 43% (170/392) would stop clopidogrel before surgery. For asymptomatic patients, 12% (47/392) would stop aspirin and 55% (217/392) would stop clopidogrel.³³ Nevertheless, the preponderance of evidence strongly supports the efficacy and safety of antiplatelet therapy in the perioperative period for patients undergoing CEA. For example, in a multistate quality improvement study of Medicare patients undergoing 10,561 CEA procedures, failure to administer antiplatelet therapy was identified as a risk factor for perioperative stroke.³⁴ Likewise, in a report of 2714 patients undergoing 3092 CEA procedures in 11 hospitals in northern New England, the administration of preoperative antiplatelet therapy was associated with a 60% reduction in the risk of perioperative stroke.³⁵ These findings were corroborated in another study that demonstrated an absolute 7% reduction in the incidence of stroke in the early postoperative period.³⁶ In fact, on the basis of a review of the totality of evidence, the 2011 Clinical Practice Guidelines of the SVS recommend that all patients undergoing CEA should receive aspirin (81-325 mg) perioperatively.¹⁴

Heparin

Heparin has been administered as therapy for acute stroke or crescendo TIAs to patients before they undergo CEA. The International Stroke Trial did not find any benefit of routine heparin administration for acute stroke because of increased numbers of hemorrhagic stroke and fatal extracranial bleeding.³⁷ However, a report from the Oregon Health Sciences University found that crescendo TIAs could be controlled with intravenous heparin, thereby allowing urgent CEA to be delayed.³⁸

Unfractionated heparin (UFH) is routinely used intraoperatively to prevent carotid thrombosis despite a lack of level I evidence to support this practice. The combination of aspirin and intraoperative heparin administration appears to be especially effective in preventing thrombosis.³⁹ The University of Rochester group found that low-dose intraoperative heparin is safe and can obviate the use of protamine to reverse heparin intraoperatively.⁴⁰

Several studies have looked at the effects of UFH on platelet aggregation. Investigators found that platelet aggregation increased tenfold in response to arachidonic acid after administration of UFH even though the patients were taking aspirin.⁴¹ With this finding in mind, the investigators then studied low-molecular-weight heparin (LMWH) as an alternative to UFH. A randomized controlled trial looked at the rate of perioperative embolization via TCD for patients given either UFH or low-molecular-weight heparin. They found that the odds ratio for experiencing a higher number of emboli in the first 3 hours after CEA was twice that of patients receiving UFH (P = .04). However, despite the fact that the UFH group experienced more emboli, they did not have a higher 30-day stroke or death rate.⁴² It should be noted that this was a small study and that the clinical end points for the two groups were not different.

Protamine Administration

Numerous publications have examined whether reversing heparin with protamine during CEA is safe. The only randomized trial that looked at this question was a small study consisting of 64 patients. It demonstrated significantly reduced wound drainage with protamine but a trend toward an increased rate of stroke and death from carotid thrombosis.⁴³ Treiman et al⁴⁴ retrospectively compared their experience with 328 patients who received protamine and 369 patients who did not over a 5-year period and found no difference in stroke rates but a significantly higher incidence of wound hematomas in patients who did not receive protamine. Similar results were reported in a retrospective study of 407 CEAs by Levison et al.⁴⁵ However, in another retrospective study involving 348 patients, Mauney et al⁴⁶ found a significantly higher stroke rate with protamine (2.6% vs. 0%, P < .045) and no significant increase in bleeding rate in patients who did not receive protamine. In a publication of observations of protamine use during the General Anesthesia versus Local Anesthesia for Carotid Surgery (GALA) trial, protamine was not found to be associated with stroke.⁴⁷ Likewise, the Vascular Study Group of New England, reporting on 4587 CEA procedures performed by 66 surgeons in 11 hospitals from 2003 through 2008, found that protamine administration was associated with a reduced incidence of serious bleeding requiring reoperation without increasing the risk of perioperative stroke, myocardial infarction, or death.⁴⁸ The preponderance of evidence now suggests that protamine administration does not increase perioperative stroke risk, although we prefer not to use it routinely unless there is truly unusual bleeding.

Dextran

Dextran is a polysaccharide that inhibits platelet aggregation.⁴⁹ It has been used to control embolic episodes both preoperatively and postoperatively. British investigators in 1997 showed that a 6-hour dextran infusion effectively controlled postoperative embolic events as measured by TCD, with a 0% stroke and mortality rate in a series of 100 patients.⁵⁰ In a follow-up study, the same group found that a 3-hour infusion was just as effective as a 6-hour infusion.⁵¹ In a study of 19 patients with recurrent or crescendo TIAs and high-grade carotid stenosis, dextran was used effectively preoperatively to control both clinical symptoms and embolic events, as measured by TCD, before CEA.⁵² The same investigators showed dextran infusion and TCD to be cost-effective in preventing stroke.⁵³ We favor dextran infusion after CEA to control platelet aggregation on the endarterectomy site and potential microembolization. However, dextran should be used cautiously, especially among patients with cardiac disease. Farber et al,⁵⁴ for example, found that postoperative dextran infusion was associated with a significantly increased incidence of perioperative MI and congestive heart failure but not a reduced rate of perioperative stroke.

Statins

The potential benefits of statins in patients with carotid artery disease are several. A reduction in cholesterol levels with statins may be associated with plaque regression (see Chapter 29).^55–63 For example, the Asymptomatic Carotid Artery Progression Study (ASAPS) demonstrated that lovastatin was associated with reduced carotid artery intima-media thickness and also a lower rate of combined cardiovascular events, although not a lower stroke rate.^64,65

However, numerous trials conducted since the mid-1990s have demonstrated statin medications to be highly effective in primary and secondary stroke prevention.^65–70 Furthermore, it appears that the stroke prevention benefits of statins are related to their pleiotropic effects rather than their cholesterol-lowering effects. In addition, several studies have demonstrated that statins are associated with reductions in the rate of perioperative cardiac morbidity and overall mortality in patients undergoing major vascular surgical procedures.^71–74

The demonstrated benefit of statins in primary and secondary stoke prevention and their positive impact on the incidence of perioperative cardiac complications and mortality in patients undergoing major vascular surgery stimulated our group to investigate the potential influence of statin medications specifically on the outcome of CEA. In a series of 1566 patients undergoing CEA, we found that statins were associated with reduced 30-day incidences of stroke (1.2% vs. 4.5%, P = .002), TIA (1.5% vs. 3.6%, P = .01), and death (0.3% vs. 2.1%, P = .002). Multivariable analysis demonstrated statins to be associated with a threefold lower risk of stroke (P = .019) and fivefold lower risk of death (P = .049) 30 days after surgery.⁷⁵ In a large administrative database analysis from Canada, Kennedy et al⁷⁶ reported a significantly lower rate of perioperative stroke and death in symptomatic patients who were taking statin medications at the time of CEA. Furthermore, Lamuraglia et al⁷⁷ reported that lipid-lowering agents were associated with significant protection against recurrent carotid stenosis or late anatomic failure after CEA. The evidence seems clear that all patients undergoing CEA should be taking statin medications at the time of surgery and long-term.

Anesthesia

A fundamental consideration in the conduct of CEA is selection of the anesthetic method. CEA may be performed with use of general anesthesia (GA), regional anesthesia (RA) with deep or superficial cervical block, and even pure local anesthesia (LA). Although early reports suggested a reduced length of hospital stay associated with CEA performed with RA, comparable lengths of stay are routinely documented today in patients who undergo surgery under GA. The majority of studies comparing the two techniques have reported improved perioperative cardiac stability with RA, but this improvement does not necessarily result in a reduced incidence of MI.^78,79

Disadvantages of RA include patient discomfort or anxiety, risk of seizure or allergic reaction, anxiety for the operating surgeon, and compromise of technique in a teaching setting. A meta-analysis by Tangkanakul et al,⁸⁰ as well as one by Rerkasem et al,⁷⁹ showed no clear benefit for CEA performed with use of LA. The GALA trial was a prospective European multinational study carried out in 95 hospitals in 24 countries, and involved 3526 patients who underwent CEA under general (1753) or local (1773) anesthesia from June 1999 through October 2007. There was no significant difference in the incidence of the primary outcome measure of stroke, death, or MI at 30 days, which occurred in 84 (4.8%) patients who underwent CEA under GA and 80 (4.5%) patients who underwent CEA under LA.⁸¹ There was also no difference in hospital length of stay between the anesthesia groups. There was a trend toward lower stroke rate among patients with a contralateral carotid occlusion having surgery under local anesthesia, although this difference was not statistically significant.

The main benefit of LA is that it facilitates efficient selective shunting if that is the surgeon’s preference. However, the benefit of selective shunting is unclear (see under “Shunting”). In addition, there may be subsets of patients with significant cardiac or pulmonary comorbidity for whom having CEA under local or regional anesthesia is associated with less risk. For example, in an analysis of CEA performed by 64 surgeons in six hospitals in New York State, there was a 70% risk-adjusted risk of perioperative stroke or death among patients undergoing CEA under LA.⁸² In another analysis of the National Surgical Quality Improvement Program (NSQIP) database, which included 13622 CEA procedures performed at 123 Veterans Administration and 14 private hospitals, RA was utilized in 2437 (18%) cases and was associated with relative risk reductions for stroke (17%), death (24%), and cardiac complications (33%).⁸³

Patient Positioning

Careful positioning of the patient is important to ensure patient comfort and adequate operative exposure. Positioning begins with placing a roll behind the scapulae to achieve some hyperextension of the neck. A padded ring is placed under the head to prevent neck injury from extreme hyperextension. If GA is used, the endotracheal tube should be taped to the corner of the mouth opposite the surgical field. If LA or RA is used, a Mayo stand is placed over the patient’s head to suspend the surgical drapes away from the patient’s face to prevent a sensation of claustrophobia. It is our practice to tape the fan for a Bair Hugger warmer (Arizant Healthcare Inc., Eden Prairie, Minnesota) on the underside of the Mayo stand, because blowing air on the patient’s face relieves discomfort. If GA is used, it is induced before placement of additional lines. A radial artery catheter is inserted for continuous blood pressure monitoring.

Skin Incision

One of two skin incisions may be used (Fig. 100-1). The standard incision is a longitudinal incision parallel to the medial border of the sternocleidomastoid muscle. The upper portion of the incision is angled posterior to the earlobe if cephalic exposure above the angle of the jaw is required. An alternative method is to place the incision in an appropriately located skin crease, usually 1 to 2 cm inferior to the angle of the jaw. This incision provides excellent cosmesis postoperatively; frequently, the resulting scar is all but invisible. However, if the incision is made in a suboptimal location, more cephalic and caudal exposure in the wound is difficult to obtain. Therefore, if the surgeon is not experienced with a skin crease incision, it is best to make a skin crease incision on the basis of the location of the carotid bifurcation, known either from preoperative DSA, CTA, or MRA or from intraoperative DUS examination, or to use a longitudinal incision. With experience the surgeon will become more comfortable making the incision according to the location of the carotid pulse, although this can be misleading at times. If the incision is made too low, more cephalic exposure can be obtained by extending the skin crease incision posteriorly. If the incision is made too high, more caudal exposure can be obtained by extending the incision more anteriorly.

The surgeon begins the dissection by dividing the platysma and mobilizing the medial border of the sternocleidomastoid muscle. The external jugular vein lies deep to the platysma and should be sought in this plane to avoid injury or to retrieve it if it is needed for patching. It is more commonly encountered with an oblique skin crease incision than with a longitudinal incision. The other structure located at this level is the greater auricular nerve; injury to this nerve leads to numbness of the earlobe, which is bothersome to patients, especially those who wear earrings (see under “Cutaneous Sensory Nerves”). The medial border of the sternocleidomastoid muscle is mobilized and retracted laterally with self-retaining retractors. The carotid sheath is entered and the medial border of the jugular vein dissected. The facial vein is identified crossing medially in the base of the wound and divided; sometimes it has an early bifurcation or trifurcation, and multiple branches need to be ligated. The jugular vein is then retracted laterally. The vagus nerve is identified at this point in the carotid sheath, usually located posteriorly between the jugular vein and carotid artery, although in a minority of patients it may lie anteriorly.

Dissection is continued down onto the distal CCA, which is controlled circumferentially with an umbilical tape and a Rumel tourniquet. At this point the ansa cervicalis nerve should be identified; it usually lies medial to the distal CCA. Identifying this nerve facilitates safe dissection of the carotid bifurcation and avoids injury to the hypoglossal nerve, which crosses medially from a superior to an inferior location. The surgeon dissects superiorly along the ansa hypoglossi and continues dissection along the posterior edge of this nerve. As he or she follows the nerve cephalad and encounters its junction with the hypoglossal nerve, the surgeon continues the dissection safely along the posterior border of the hypoglossal nerve with minimal risk of injury; if the surgeon were to dissect along the anterior border of the ansa, it is possible to inadvertently transect the hypoglossal nerve in the crotch of the junction of these two nerves before the hypoglossal nerve is identified (Fig. 100-2). It is important to note that the position of the hypoglossal nerve can be quite variable. Often it lies sufficiently cephalad and medial that it may not be in the operative field. In most cases, however, it crosses the ICA in a position where it may need to be mobilized by division of the sternocleidomastoid branch of the occipital artery, which tethers the nerve (see Figure 100-10).

The conventional technique for CEA consists of a vertical arteriotomy and closure by patch angioplasty. In this case a vertical arteriotomy is begun on the CCA and continued through the carotid bifurcation into the ICA. One should avoid making the incision too close to the flow divider at the ECA origin because doing so could distort the anatomy and make the closure more difficult. If a shunt is used, it is placed in the distal ICA and back-bled before its proximal end is placed into the CCA. Two commonly used shunts are the Pruitt-Inahara and Javid shunts. A third shunt that we prefer is a simple vinyl tube, originally described by Collins et al⁸⁴ (Fig. 100-3). Because this shunt lies entirely within the artery, it allows the surgeon to almost completely finish closing the arteriotomy before the shunt is removed. Its small diameter permits atraumatic placement in even small ICAs, whereas its short length offers less resistance to blood flow such that physiologic flow in the ICA is maintained. Wilkinson et al⁸⁵ compared the Pruitt-Inahara and Javid shunts in a randomized trial. Using TCD, they found that the Pruitt shunt was less likely to maintain physiologic flow in the MCA, whereas the Javid shunt was associated with a higher incidence of cerebral embolism at declamping.

The endarterectomy is begun in the CCA in the plane between the media and the adventitia. The proximal end point in the distal CCA is established, and the plaque is trimmed in that location in a beveled manner. The endarterectomy is continued into the orifice of the ECA, first with a Freer elevator and then with a fine clamp that is passed up into the ECA in the plane of the endarterectomy. The clamp is spread apart to further mobilize the plaque away from the adventitia in the 6-, 9-, and 12-o’clock positions; it is usually hard to pass the clamp in this plane at the 3-o’clock position next to the flow divider. The vessel loop on the ECA is released transiently while the plaque is everted from within the ECA. The end point of the plaque is inspected; an ideal end point is gradually tapering and feathered (Fig. 100-4). In our practice, all loose bits of intima and media in the orifice of the ECA are removed to perform complete endarterectomy of the ECA. However, Ascher et al⁸⁶ have found in a large series that endarterectomy of the ECA may be neglected without compromise of results.⁸⁶

The endarterectomy is continued up into the ICA. A technically perfect end point in the ICA is critical to avoid perioperative stroke and recurrent stenosis. In our experience, it is virtually always possible to achieve a satisfactory end point in the ICA, although special maneuvers may be required to expose the distal ICA and make an extended arteriotomy in this vessel to facilitate extraction of a long endarterectomy specimen, as seen in Figure 100-5. Tacking sutures at the distal end point should be avoided unless absolutely necessary; such sutures are problematic and associated with an increased perioperative stroke rate.⁸⁷ The endarterectomy should be terminated in normal ICA with a gradual, tapered transition to normal intima; this is best accomplished by pulling the plaque transversely away from the artery with lateral traction. One should avoid pulling out or down on the plaque, which would be more likely to result in a step-off that can be difficult to correct without traumatizing the artery.

We believe that repairing the arteriotomy with a patch angioplasty represents the standard of care in contemporary practice (see Fig. 100-4). The patch is sewn in with running nonabsorbable suture. A variety of patch materials are available for use, including autologous vein, polytetrafluoroethylene (PTFE), woven polyester (Dacron), and bovine pericardium. Some studies have suggested that autologous vein may be superior to synthetic patches, but of the prosthetic patches, no material appears to be clearly superior to another (see “Patch Closure”).

Options for autologous vein include the external jugular and saphenous veins. The external jugular vein can be harvested through the same surgical incision and is generally used as a double-layer patch after the surgeon inverts an intact tubular segment of vein without filleting open the vein. Care must be taken to keep the inverted tube flattened out as a rectangular patch while it is being sewn onto the artery. If this is not done, the edges can sometimes roll over or under and lead to a tapered, asymmetric, or severely deformed patch. It is our practice to always start the suture line at the superior end of the arteriotomy in the ICA, which is typically the most difficult—and critical—part of an anastomosis. As the patch material is sewn to one side of the artery, the artery must be stretched out with gentle tension so that an appropriate length of patch is used before it is trimmed; otherwise, there may not be enough patch material available to sew to the other side of the artery in the arteriotomy. When the suture line is nearly completed, the CCA and ICA are reclamped and the shunt is removed. Both clamps are briefly released to flush air or debris (or both) out of the arteries. The clamps should be placed proximal and distal to the patch or endarterectomized surface of the artery because these surfaces can be thrombogenic. The carotid bifurcation is flushed vigorously with heparinized saline and inspected again for debris or intimal flaps before the arteriotomy is finally closed. Once again the clamp on the ICA is briefly released to fill the bifurcation with blood. It is then replaced while the clamps on the CCA and ECA are released so that any remaining air or debris will be flushed up the territory of the ECA rather than the ICA. At this point the ICA clamp is removed.

Any bleeding from the suture line is addressed at this time. However, a final important technical point in this step again relates to the thrombogenicity of stagnant blood in contact with the patch material or endarterectomized surface of the carotid bifurcation. One should avoid reclamping unless absolutely necessary to control bleeding at this stage and avoid the risk of formation of thrombus or a fibrin-platelet aggregate on the patch or endarterectomized vessel. This phenomenon should not be underestimated; in 2002, AbuRahma et al⁸⁸ noted a 5% carotid thrombosis rate with Dacron patches in a prospective trial with a resultant 7% perioperative stroke rate.⁸⁸ As a result of that trial, the makers of the Dacron patch reengineered the patch to make it less thrombogenic.

Eversion Endarterectomy

Eversion endarterectomy is an excellent alternative technique that is utilized in many centers throughout the world. Two different versions of eversion endarterectomy are performed. DeBakey et al² originally described eversion endarterectomy with partial transection of the anterior portion of the carotid bifurcation. Etheredge⁸⁹ improved on DeBakey’s technique with complete transection of the bifurcation, which allowed the origins of both the ICA and ECA to be everted for a longer distance. The endarterectomy is performed by mobilizing the entire circumference of the carotid adventitia off the plaque (described as a “circumcision” by Etheredge) and then everting the adventitia and mobilizing it upward while gentle caudad traction is applied to the plaque. This maneuver is performed distally into the orifices of the ICA and ECA and then proximally into the CCA. Once the endarterectomy is complete, the divided bifurcation is reunited with a simple end-to-end anastomosis.

Advantages of this technique are that the anastomosis can be performed rapidly and that it is not prone to restenosis, and therefore patching is not required. Disadvantages of this technique are that more extensive dissection is sometimes necessary to mobilize the vessels during the eversion, the procedure does not lend itself readily to shunting (although shunting is not precluded by this technique), and it can be difficult to visualize the end point in the ICA after the plaque has been removed—the artery tends to retract as soon as the plaque pulls away from the adventitia, and it can be difficult to expose and re-inspect this area of the artery again. Therefore, in our opinion, a completion study should be performed with this technique.

In 1985, Kieny et al⁹⁰ introduced a modification of eversion endarterectomy, in which the origin of the ICA is excised obliquely off the carotid bifurcation, the ICA is inverted on its own, and endarterectomy of the CCA and ECA is performed through an arteriotomy in the side of the carotid bifurcation. With this technique it is easier to manipulate the ICA by itself, so eversion of the ICA is less cumbersome. The ICA is reanastomosed to the carotid bifurcation primarily (Fig. 100-6). This technique allows rapid plaque extraction, the anastomosis is not prone to restenosis, and no prosthetic material is required.

This technique is particularly effective for dealing with a redundant, coiled, or kinked ICA. The ICA can be pulled down and straightened and the redundant portion excised. The remaining portion of the ICA is spatulated and reattached to the arteriotomy on the carotid bifurcation. This procedure typically is not performed with a shunt but does not preclude shunt use. Less dissection is required than with transection of the bifurcation. However, exposure for thorough endarterectomy of the CCA and ECA may be suboptimal.

Exposure for High Lesions

The carotid bifurcation can be located anywhere between the second and seventh cervical vertebrae, and a bifurcation located high in the neck poses technical challenges that can increase perioperative risk for stroke and cranial nerve injury. Ideally, a high bifurcation will be identified on the preoperative imaging study. This is a potential advantage of CTA, images of which always included bony anatomy. A conscious effort must be made to locate the bony anatomy with DSA if unsubtracted images are not provided, but the anatomy can still be defined from subtracted images. Bony landmarks are never provided with carotid DUS, but an astute vascular technologist will note a high bifurcation and should record it in the report.

The final maneuver that can be extremely effective in gaining cephalic exposure is resection of the styloid process (Fig. 100-7). After the posterior belly of the digastric muscle is divided, the insertions of the styloid apparatus—styloglossus, stylopharyngeus, and stylohyoid muscles—on the styloid process are excised with a scalpel with either a No. 15 blade or a Beaver blade. The stylohyoid muscle can be found posterosuperior to the ECA and inserts into the posterior portion of the styloid process.^99,100 In addition, the occipital artery runs along the inferior border of the posterior belly of the digastric muscle. Identification and division of the occipital artery can prevent bleeding secondary to trauma caused by retraction (see Fig. 100-7). The styloid process is carefully resected with a rongeur.¹⁰¹ This step can extend the exposure of the ICA by an additional 4 to 5 mm. However, damage to the underlying facial nerve is a potential complication of styloidectomy.¹⁰² In our experience with four cases in which resection of the styloid process was necessary, this maneuver alone permitted exposure of the ICA all the way to the skull base.

Two other options can be used to improve exposure of a high bifurcation, and both require preoperative planning and coordination with an oral or plastic surgeon (Fig. 100-8). The patient should receive nasotracheal intubation. Simonian et al¹⁰³ reported on anterior subluxation of the mandible for use in high carotid exposures in 1984. They described their technique for subluxation as an evolution from full-mouth arch bars with circumdentate wiring to unilateral subluxation by circummandibular/transnasal wiring.¹⁰³ For the patient with healthy dentition, anterior subluxation of the mandible can be accomplished by placing circumdental wires around the mandibular cuspid and bicuspid teeth on the side ipsilateral to the vascular lesion. Corresponding wires are placed around the corresponding contralateral maxillary teeth. The mandible is then subluxed anteriorly, and the wires are twisted together to hold the fixation. In the patient with poor dentition arch bars can be placed prior to wiring to facilitate placement of the wires.¹⁰⁴ An even more aggressive approach involves the use of a complete vertical osteotomy through the vertical ramus of the mandible and separation of the mandible to expose the ICA.

One of the issues of long-standing debate related to the performance of CEA concerns the use of intravascular shunts: routine nonuse of shunts, selective use of shunts, and routine use of shunts. The simplest way to perform CEA is to just clamp the carotid bifurcation and perform CEA without a shunt, and several large series have documented excellent results of CEA without shunts.^105–107 However, all these studies report at least a small incidence of stoke, and in at least some cases the etiology of the stroke is intraoperative cerebral ischemia during carotid artery clamping.

Alternatively, some surgeons routinely use shunts in all cases of CEA, and excellent results have been reported in several large series with this practice.^108–111 However, all these studies document an incidence of stroke, and in some of these cases the cause of the stroke was attributed to technical problems related to use of the shunt.

A third option, founded on the notion that the use of a shunt conveys its own potential risks, is to use shunts selectively in the patient who would be at high risk for ischemic stroke if a shunt were not used. However, accurate identification of the small group of patients who require shunts has been difficult. Several techniques have been used to identify patients who truly need a shunt. In a patient under GA, such techniques include intraoperative measurement of carotid “stump pressure” after the CCA and ECA have been clamped, intraoperative neurologic monitoring via EEG or somatosensory evoked potentials (SSEPs), measurement of MCA flow by TCD, and monitoring with cerebral oximetry. An alternative method is to perform CEA with the patient under RA, with selection of patients for shunting based on alterations in the neurologic responses that develop after the carotid artery is clamped.

The etiology of intraoperative stroke may be ischemic or embolic. Ischemic stroke results solely from inadequate cerebral perfusion related to clamping of the carotid artery. Embolic stroke can occur during manipulation of the carotid artery before clamping, during the actual process of clamping of the carotid artery, or during reperfusion—either after placement of a shunt or after unclamping of the carotid bifurcation as the endarterectomy is completed. Embolic stroke can also occur as a result either of technical defects in the endarterectomy site or adjacent vessels related to damage from a clamp or a shunt or of a technically inadequate endarterectomy. This latter situation can lead to intraoperative stroke or (delayed) postoperative stroke. All the adjuncts described in the preceding paragraph are designed to detect intraoperative cerebral ischemia, whereas TCD has the added advantage of detecting intraoperative emboli.

Placement of a shunt has the capacity only to prevent ischemic stroke, but it could actually increase the risk for embolic stroke if performed poorly. Intraoperative cerebral ischemia is a relatively uncommon cause of intraoperative stroke.^112–114 This fact would support the argument of the “routine nonshunters,” who believe that cerebral ischemia is a rare cause of stroke and that shunting may do more harm than good (by causing embolic complications) such that even selective shunting is never justified. However, there is no denying that when cerebral ischemia does occur, it can lead to perioperative stroke.^115–117

It is clear that cerebral ischemia can be completely relieved by placement of a shunt. This should translate into prevention of stroke in the patient if embolic complications related to shunt placement are minimized. In fact, there is evidence that the benefits of shunting outweigh the risks in patients with cerebral ischemia.^118–120

A meta-analysis by AbuRahma et al¹⁰⁴ looked at all papers published on CEA from 1990 through 2010 and examined the perioperative outcome of routine shunting, routine nonshunting, and selective shunting. The mean perioperative stroke rate was 1.4% for CEA with routine shunting and 2%. for CEA with routine nonshunting. For selective shunting, the stroke rate varied depending upon the method used for intraoperative monitoring. It ranged from 1.1% for cervical block anesthesia, to 1.6% for EEG and carotid stump pressure, to 4.8% for TCD. The investigators concluded that both routine shunting and selective shunting are acceptable and should be employed at the surgeon’s discretion.¹⁰⁴

From a medical-legal point of view, routine nonshunting with use of GA would be hard to defend in the case of a patient who suffered an intraoperative stroke during CEA. Consequently, routine nonuse of shunts cannot be advocated, although it is still practiced by those who favor eversion endarterectomy,¹²¹ in which placement of a shunt is technically more difficult but not impossible.

Cerebral Monitoring

Selective shunting depends on using a technique to identify patients with intraoperative ischemia. The majority of CEAs are performed today with GA, and strategies to assess cerebral ischemia include stump pressure measurement; EEG, TCD, and SSEP monitoring; and measurement of regional blood and cerebral oxygen saturation. The problem is that none of these techniques is completely accurate. They do not uniformly predict the occurrence of intraoperative ischemia, nor do they prevent the unnecessary use of shunts. Finally, the selective use of shunting may actually increase the risk for an ischemic stroke that would not otherwise occur with routine shunting.

Awake Carotid Endarterectomy with Regional or Local Anesthesia

In view of these findings, performing CEA with RA is the most reliable method of predicting the need for selective shunting. It can be done with either a cervical plexus block or purely local anesthesia. In this case the decision to place a shunt is based solely on the development of hemispheric or global neurologic symptoms after the carotid artery is clamped. This criterion is considered the standard for selective shunting to which all other adjuncts should be compared. Shunt rates with this approach, on the order of 5% to 15%, are consistently lower than with other modalities.^146–153 In two prospective trials in which both stump pressure and EEG measurements were recorded before endarterectomy with RA but in which the need to shunt was ultimately determined by neurologic changes, both EEG findings and stump pressure were found to inaccurately predict the need for shunting.^154,155 In the study by Calligaro et al,¹⁵⁴ published in 2005, a cost analysis found that RA saved more than $3000 per case by avoiding EEG measurements.

In addition to its benefit in predicting the need for shunting, other potential benefits of performing CEA on awake patients have been suggested. One study has demonstrated a lower rate of MI with RA that was statistically significant,¹⁵⁶ whereas others have not confirmed this benefit.^151,157,158 Bowyer et al¹⁵⁸ showed a statistically significant decrease in the neurologic event rate with RA, but not all patients were selectively shunted in this study, and no other study has duplicated this finding. Three studies have shown decreased length of stay with RA,^150,156,157 but in two of them the length of stay, approximately 5 days, was not reflective of modern practice.^150,157 This decrease does not necessarily translate into reduced cost,¹⁵⁹ but RA is probably the only adjunct for selective shunting that does not increase cost over that of routine shunting. Finally, as mentioned earlier for the multicenter randomized GALA trial, the definitive study of RA versus GA in CEA that included more than 3500 patients, there was no significant difference in the rates of perioperative stroke, death, or MI between the two anesthetic methods. There was, however, a higher rate of perioperative stroke among patients with contralateral occlusion who underwent CEA under general GA, although this difference was not statistically significant.⁸¹ Furthermore, in a post-hoc analysis of the GALA data, costs were less within the first 30 days after surgery in the RA patient cohort.¹⁶⁰

The disadvantages of RA are that not all anesthesiologists, surgeons, or patients are comfortable with performing CEA with use of RA. Cervical block is an advanced RA technique that requires considerable skill on the part of the anesthesiologist. Many surgeons find it stressful to have an awake patient who can behave unpredictably during the procedure. Some patients tolerate RA poorly because of claustrophobia or inadequate anesthesia with just a regional or local block. There is also a small risk of seizure or cardiac arrhythmia with inadvertent administration of local anesthetic into the carotid artery or jugular vein.

Routine Shunting

One limitation in the strategy of selective shunting is that with the exception of performing the procedure with RA upon the awake patient, none of the methods to assess the need for a shunt is absolutely accurate. When stump pressure measurement is utilized to determine the need for a shunt, instrumenting the carotid artery with a needle introduces the added risk of embolism. On the other hand, when the operation is performed in the awake patient, the need for shunt placement is determined through the patient’s neurologic reaction to a test clamp. When a test clamp response is positive, there are two options. One is to unclamp the vessels and reperfuse the brain while preparations are made to place the shunt. This maneuver exposes the patient to the risk of embolization. The other option is to proceed with the endarterectomy and placement of a shunt. In this instance, the brain is ischemic when the test clamp response becomes positive and is exposed to an additional period of ischemia while the shunt is being placed, whereas during routine shunting, shunt placement can usually be accomplished in 1 to 2 minutes after clamping, before the brain becomes ischemic. The additional period of ischemia after a positive test clamp response can be detrimental. Tempelhoff et al¹³⁵ reported that five of six postoperative neurologic deficits occurred in patients who had ischemic times longer than 9 minutes by EEG criteria.¹

Although there is no definitive clinical evidence to support one shunting strategy over another,^161,162 the routine use of carotid shunts eliminates these concerns. It allows CEA to be performed in a consistent manner, thereby eliminating surgeon and patient anxiety, without the added cost or complexity of monitoring equipment. The shunt is easy to place and facilitates performing CEA in a teaching environment. All studies consistently show that carotid shunting with either the Javid or Pruitt-Inahara shunt completely relieves the intracerebral ischemia caused by clamping, whether measured by neurologic status, EEG changes, MCA flow on TCD, or cerebral blood flow. The placement of a shunt introduces the risk of iatrogenic complications such as intimal dissection as well as of thromboembolic complications, but in the hands of experienced operators, the procedure is very safe. An absolute requirement for safe shunt placement is that the superior end of the plaque be positively identified and adequately exposed through the arteriotomy so that the distal end of the shunt does not “snowplow” into the plaque when it is placed, causing embolization or dissection. There is evidence that placing a shunt in the setting of severe ischemia decreases the stroke rate.¹⁶² As noted earlier, minimizing ischemic time in the brain by routine shunt placement has the theoretical advantage of limiting ischemia-reperfusion injury.

Pärsson et al¹⁶³ have shown that carotid shunting diminishes the inflammatory response of ischemic brain injury, as demonstrated by the production of various inflammatory mediators. This may be an important mechanism in the occurrence of delayed postoperative strokes, which can account for up to 70% of perioperative strokes.¹¹³ Numerous large series document excellent results with the use of routine shunting in CEA. Thompson¹⁰⁹ demonstrated superb results over a 15-year period spanning the 1960s and 1970s, with a stroke rate of 1.4% in 1107 CEAs. Javid et al¹¹⁰ reported similarly excellent results in a series of more than 1800 patients from the same era. In 1997, Hertzer et al¹¹¹ reported a series of more than 1900 CEAs performed at the Cleveland Clinic, virtually all of which involved routine shunt placement, with a perioperative stroke rate of 1.8%.¹¹¹ Hamdan et al¹⁰⁸ published a series of 1001 consecutive patients in whom routine shunting was used most of the time and in whom the combined stroke and death rate was 1.6%. Clearly, there are other reports documenting similarly outstanding results among patients undergoing CEA with selective shunting,¹⁶⁴ although selective shunting may somewhat increase the complexity of the procedure and the associated costs.

In summary, there is no definitive evidence to support the superiority of one shunting strategy over the others. As noted previously, a meta^–analysis has documented a mean perioperative stroke rate 1.4% for CEA with routine shunting and of 2% for CEA without routine shunting and from 1.1% to 4.8% for selective shunting, depending on the monitoring method.¹⁶⁵ Finally, despite acceptable results in some individuals’ hands, routine nonuse of shunts cannot be advocated in view of the abundant evidence that intraoperative ischemia can be a source of stroke that is preventable with shunts.

Patch Closure