Outcomes and Risk Factors in 1,609 Carotid Endarterectomies

Abstract

Severe carotid stenosis is typically treated with carotid endarterectomy (CEA), but there is debate about the safety of this procedure in patients with occlusion of the contralateral artery, previous CEA in the same artery, and other risk factors. To evaluate the association of these factors with outcomes in standard CEA with Dacron patch angioplasty, we examined the records of 1,609 consecutive isolated CEAs performed at our institution over a 10-year period on 1,400 patients (851 men and 549 women; mean age, 69.5 yr) with symptomatic or high-grade asymptomatic carotid lesions. Twenty-three patients (1.4%) had perioperative strokes, of which 2 were fatal. The overall same-admission mortality was 0.2% (4 patients). Same-admission stroke/death was more likely in patients with any history of tobacco use (odds ratio [OR], 4.6; 95% confidence interval [CI], 1.6–13.6), contralat-eral occlusion (OR, 3.3; 95% CI, 1.2–9.1), angina with a Canadian Cardiovascular Society classification of 2 or greater (OR, 3.2; 95% CI, 1.4–7.6), or transient ischemic attack within the 6 weeks before surgery (OR, 2.4; 95% CI, 1.05–5.3). A total of 9 patients (0.6%) died within 30 days of CEA; our multivariate analysis did not reveal any significant predictors of 30-day mortality. We conclude that standard CEA with patch angioplasty is associated with low rates of death and morbidity for most patients, but patients with any history of tobacco use, substantial angina, contralateral occlusion, or preoperative transient ischemic attack may have an elevated risk of adverse outcomes.

Stenosis of the carotid artery is an important cause of stroke and other adverse neurologic events. Since the 1st successful open carotid endarterectomy (CEA) was performed in the 1950s, this procedure has become the standard treatment for severe carotid stenosis. Several large, randomized trials have shown CEA to be associated with low perioperative stroke and mortality rates.^1–4 Nonetheless, some physicians have argued that carotid angioplasty and stenting may be preferable to CEA because it is less invasive, entails a shorter recovery period, and may be less likely to produce adverse outcomes in patients with such risk factors as advanced age, occlusion of the contralateral artery,⁵ previous CEA in the same artery,^6–9 and previous neck irradiation.^10–12

The goal of the present study was to examine the frequency of adverse CEA outcomes and the value of various risk factors in predicting these outcomes. We reviewed the records of 1,609 CEAs performed at our institution on 1,400 patients for extracranial cerebrovascular disease. The potential predictors that we examined included several characteristics that have been cited as risk factors in previous reports. Our primary endpoints were the combined same-admission stroke/death rate and 30-day mortality. The secondary endpoints were postoperative transient ischemic attack (TIA), cerebral nerve injury, myocardial infarction, and reoperation for hemostasis.

Patients and Methods

We reviewed the records of all CEAs performed by 3 surgeons at our institution from January 1993 through December 2002. Cases were excluded from the review if CEA had been performed simultaneously with other procedures, because these data have been reported previously.^13,14

Procedures

Patients were considered candidates for CEA if duplex ultrasonography, radiographic angiography, or magnetic resonance angiography indicated that they had lesions that occluded at least 70% of the carotid artery. All procedures were standard CEAs with Dacron patch angioplasty and were performed with the patients under general anesthesia. Shunts were inserted at the discretion of the individual surgeon. Patients were given 1 mg/kg heparin during the procedure, the effects of which were reversed with protamine sulfate afterward. Completion imaging was not done routinely. Patients were monitored in the recovery intensive care unit for 3 to 4 hours before being sent to the nursing unit.

Data Collection and Analysis

Two research assistants collected chart information on 1,612 CEAs performed at our institution. These data were then merged with additional data from the database maintained by our institution's Biostatistics department. The resulting data set included 1,614 cases. Two of these cases were excluded because the patients' charts had been lost, and 3 cases were excluded because they were actually implantations of Dacron interposition grafts, and not CEAs. Thus, the final data set included 1,609 cases.

Because the rates of stroke and death during the in-hospital recovery period were relatively low, these were combined into a single endpoint (same-admission stroke/death) for purposes of analysis. A stroke was defined as a permanent neurologic deficit confirmed by the surgeon and consulting neurologist, with computed tomographic findings in support of the diagnosis. Transient ischemic attacks or other neurologic events that resolved within 24 hours were not classified as strokes. Other endpoints tabulated were 30-day mortality (the only outcome for which 30-day follow-up data were available for every patient) and same-admission rates of postoperative TIA, cranial nerve injury (determined by a neurologist whenever the surgeon suspected such injuries), myocardial infarction, and reoperation for hemostasis.

We also examined the association between these outcomes and various demographic and perioperative factors that could potentially affect their likelihood (Table I). These included age, sex, occlusion of the contralateral artery, previous CEA in the same artery, previous neck irradiation, New York Heart Association (NYHA) classification of functional limitation related to heart disease, Canadian Cardiovascular Society (CCS) classification of functional limitation related to angina, preoperative serum levels of creatinine, previous coronary artery bypass grafting (CABG) surgery or coronary stenting, and symptoms of carotid stenosis (stroke, TIA, or both) within the 6 weeks before surgery. Histories of hypertension (blood pressure, >140/90 mmHg), coronary heart disease (CHD), tobacco use, diabetes, hyperlipidemia, pulmonary disease (including chronic obstructive pulmonary disease [COPD], asthma, emphysema, tuberculosis, and pulmonary embolism), oxygen-dependent COPD, myocardial infarction, and renal insufficiency were also examined as potential predictors of outcome. We used χ² analysis and the Fisher exact test with categorical variables and t tests with continuous variables. A 2-tailed P <0.05 was considered statistically significant. For each outcome, all variables that produced a result with P <0.15 in univariate tests were included in a stepwise binary logistic regression.

TABLE I. Patients' Characteristics and Risk Factors in 1,609 Carotid Endarterectomy Cases

Results

Patient Characteristics

One thousand six hundred nine CEAs were performed on 1,400 patients (851 men and 549 women), who had a mean age of 69.5 years (range, 34–93 yr). Of the 1,609 procedures, 734 (46%) were performed on patients who had experienced symptoms of carotid stenosis within the 6 weeks before surgery (Table I). Shunts were used in 1,426 cases (89%). The median time from surgery to discharge was 2 days (range, 1–112 days).

Primary Outcomes

In 23 (1.4%) of the 1,609 cases, patients had strokes during either the CEA procedure or the in-hospital recovery period. Twenty patients had strokes (1 fatal) in the cerebral hemisphere ipsilateral to the operated artery, 2 patients had strokes (1 fatal) in the contralateral hemisphere, and 1 patient had multiple infarcts in both hemispheres. In addition to the 2 stroke-related deaths, there were 2 other deaths during the same hospital admission—1 from myocardial infarction after an emergency CABG operation 2 days after CEA, and 1 from renal failure after the patient underwent aortic valve replacement 4 days after CEA—for a total same-admission mortality rate of 4 patients (0.2%). Therefore, the total incidence of the combined endpoint of same-admission stroke/death was 25/1,609 (1.6%).

Univariate analyses showed that 5 factors were associated with increased risk of same-admission stroke/death: NYHA functional class III, angina with CCS classification of 2 or higher, TIA within the 6 weeks before surgery, contralateral occlusion, and past or present tobacco use (Table II). When these factors were entered into a stepwise binary logistic regression analysis, the independent predictors of same-admission stroke/death were past or present tobacco use, CCS classification of 2 or higher, contralateral occlusion, and preoperative TIA (Table III).

TABLE II. Univariate Predictors of Adverse Outcomes in Carotid Endarterectomy (n = 1,609)

TABLE III. Independent Risk Factors for Adverse Outcomes in Carotid Endarterectomy (n=1,609)

Five patients died after hospital discharge but within 30 days of surgery, so the total 30-day mortality rate was 9/1,609 (0.6%). More than half (5/9) of the deaths occurred in patients who had undergone subsequent procedures unrelated to CEA, including the 2 non–stroke-related, same-admission deaths described above. Two patients who died after discharge (but within 30 days of CEA) had also undergone subsequent, non–CEA-related procedures during the same admission (1 renal artery and aortoiliac bypass and 1 aortomesenteric, aortorenal, and aortoiliac bypass). Another patient died during a subsequent admission for an aortocoronary bypass, 27 days after CEA. Of the remaining 2 patients who died after discharge but within 30 days of CEA, one died of myocardial infarction and the other of unknown causes.

In univariate analyses, risk of death within 30 days of CEA was associated with history of CHD, a serum creatinine level greater than 2.0, and history of pulmonary disease (Table II). However, in the multivariate analysis, no variable was a significant independent predictor of 30-day death (Table III).

Secondary Outcomes

Twenty patients (1.2%) experienced TIA during postoperative recovery. Univariate analyses showed that postoperative TIA was most likely to occur in patients who had experienced a preoperative TIA or who had CEA on the left side (Table II). In multivariate analysis, both of these factors and previous coronary revascularization (via either CABG surgery or carotid angioplasty and stenting) predicted postoperative TIA (Table III). Thirteen patients (0.8%) were found to have postoperative cranial nerve injuries, all of which were transient. Previous ipsilateral CEA was the sole predictor of cranial nerve injury in both univariate and multivariate analyses (Tables II and III). Twenty-eight patients (1.7%) were returned to the operating room for bleeding after CEA was completed. Reoperation for bleeding was predicted only by male sex in both the univariate and multivariate analyses (Table II). Only 3 patients experienced perioperative myocardial infarctions, 1 of which was fatal. The small number of myocardial infarctions prevented any meaningful analysis of potential predictors of this outcome.

Discussion

We found a relatively low rate of adverse events after CEA, including same-admission stroke/death (1.6%), overall 30-day death (0.6%), TIA (1.2%), cranial-nerve injury (0.8%), and return to the operating room for bleeding (1.7%). Past or present tobacco use, contralateral occlusion, CCS classification of 2 or higher, and TIA during the 6 weeks before surgery all independently predicted same-admission stroke/death. History of coronary artery disease, a creatinine level of 2.0 or higher, and history of pulmonary disease predicted 30-day mortality individually, but no variable significantly predicted 30-day mortality in the multivariate analysis. Previous coronary revascularization, CEA on the left side, and preoperative TIA independently predicted postoperative TIA. Only previous ipsilateral CEA predicted cranial-nerve injury, and only male sex predicted the need for reoperation for hemostasis.

The present study examined several factors that have been said to complicate the CEA procedure and to increase the risk of postoperative death and morbidity. One such proposed risk factor is occlusion of the contralateral carotid artery, which can worsen cerebral ischemia during CEA. Although several studies^15–20 have found no significant differences in the incidence of stroke between patients with and without contralateral occlusion, pooled data from 14 studies published between 1981 and 1994 show that contralateral occlusion significantly increases the odds of stroke within 30 days of CEA (odds ratio, 1.91; 95% confidence interval, 1.35–2.69).⁷ Several more recent, large-scale studies have also found an increased stroke risk in patients with contralateral occlusion.5,6,8,9,21 The results of our study support these findings: approximately 5% of patients with contralateral occlusion had strokes or died during the postoperative period, whereas slightly more than 1% of patients without contralateral occlusion had these adverse events.

Previous ipsilateral CEA is also a risk factor in CEA patients, because scarring from the 1st procedure often disrupts the usual cleavage planes. A meta-analysis of 6 CEA studies has shown that patients who undergo repeat CEA have a higher risk of stroke and death than do patients who have primary CEA (odds ratio, 1.95; 95% confidence interval, 1.21–3.16).²² The present study, however, did not find significant effects of previous ipsilateral CEA on perioperative stroke/death or on 30-day death, nor was there even a trend in this direction. However, our results do suggest that scarring from previous CEA increases the likelihood of cranial-nerve injury during subsequent CEA; such injuries occurred in 5% of patients who had undergone previous ipsilateral CEA but in only 0.7% of patients who had not.

Patients with previous neck irradiation have also been considered to be at significantly increased risk for cranial nerve injury after CEA. Even more important, in patients with previous neck irradiation who develop symptomatic stenosis, the plaques that form in the carotid artery tend to contain more lipids and less fibrous tissue and collagen than do typical plaques.²³ These features make irradiation-related plaques unstable, increasing the likelihood of stroke and intraplaque hemorrhage and making the plaques difficult to detect with ultrasonography. Radiation can also damage the medial and adventitial layers of the artery or injure the artery indirectly by destroying the vasa vasorum,²⁴ making direct closure of the artery impossible in some patients.²⁵ The consistent use of patch angioplasty in the present study may have mitigated this problem, because there were no strokes or deaths among the 26 patients who had undergone previous neck irradiation. There were also no cranial-nerve injuries in these patients. However, because this subgroup of patients was relatively small (26 patients), these findings do not show definitively that previous neck irradiation is not a risk factor in CEA.

For the most part, our results were similar to those of large, multi-institutional studies such as the North American Symptomatic Carotid Endarterectomy Trial (NASCET), the European Carotid Surgery Trial (ECST), the Asymptomatic Carotid Atherosclerosis Study (ACAS), and the Asymptomatic Carotid Surgery Trial (ACST). For example, the 30-day mortality rate among our patients (0.6%) was not substantially different from that observed in ACAS (0.1%),²⁶ NASCET (1.1%),²⁷ ECST (1.0%),⁴ and ACST (1.1%).²⁸ In addition, although it is not possible to directly compare these studies with ours in regard to rates of postoperative stroke and TIA (because the studies recorded these outcomes for the 30 days after CEA, rather than between CEA and hospital discharge, as we did), a study of 2,024 CEAs by Rockman and colleagues²⁹ found that, of the perioperative neurologic events that occurred within 30 days of the operation, nearly 80% happened during the operation or within the first 24 hours afterward, and another 10% happened within 3 days. Thus, our 1.4% in-hospital stroke rate might be the equivalent of a 30-day stroke rate of 1.7% to 2.0%. This estimate seems to fit with the stroke rates found in multi-institutional studies: ACAS and ACST, which included only asymptomatic (and, therefore, lower-risk) patients, each found a 1.4% rate of disabling or fatal stroke, whereas NASCET and ECST, which studied only symptomatic patients, found rates of 2.4% and 3.2%, respectively. In our study, about half (46%) of the patients were symptomatic before undergoing CEA, so it seems reasonable that our 30-day stroke rate would be between these 2 extremes.

One notable difference between our findings and those of some multi-institutional studies is in the rate of postoperative cranial nerve injuries: reported rates of postoperative cranial nerve injury were considerably higher in both NASCET (8.6%)²⁷ and ECST (5.1%)³⁰ than they were in our study (0.8%). This finding seems particularly counterintuitive in light of the fact that previous ipsilateral CEA—the only independent risk factor for cranial nerve injury in our study—was uncommon among ECST patients³⁰ and was an exclusion criterion in NASCET.³¹ One possible explanation for the difference in cranial nerve injury rates is the variation in neurologic follow-up among these studies. In NASCET, all patients underwent neurologic evaluation at study entry and at several fixed time points after CEA, whether or not the patients had symptoms of cranial nerve injury.³ The follow-up protocol used in ECST was similarly rigorous, and the identification of cranial nerve lesions was systematically included in each neurologic evaluation. In the present study, on the other hand, a neurologist evaluated patients for cranial nerve injuries only if the patient had symptoms of such injuries. Thus, it is possible that transient, clinically insignificant cranial nerve injuries (which constituted the vast majority of cranial nerve injuries found in NASCET) occurred but went undetected in our patients.

Limitations of the Study

Some potential limitations of the present study must be acknowledged. First, some risk factors and outcomes may have been too rare to enable the drawing of meaningful conclusions, although our overall sample size was large. For example, the small number of deaths in the 30-day postoperative period (9/1,609) may have rendered difficult the detection of independent predictors of this outcome. Similarly, the small number of patients who had undergone previous neck irradiation (n=26) may have prevented us from finding any predictive value that this factor might have for CEA outcomes. Moreover, only 3 vascular surgeons were involved in the surgical procedures performed in these patients. There was no routine, detailed preoperative or postoperative neurologic examination, nor was there a systematic effort to identify myocardial infarction in this study, such as by measuring postoperative troponin or creatine kinase levels.

Conclusions

In conclusion, the present study—one of the largest single-center CEA studies yet conducted—found a relatively low rate of perioperative death and morbidity in 1,609 patients who underwent CEA with Dacron patch angioplasty. Stroke and death were more likely in patients with a history of CHD, tobacco use, moderate or severe angina (as indicated by CCS classification), TIA within the 6 weeks before the procedure, and contralateral occlusion. The size of this study and the diversity of risk factors in its population make these data a useful basis for comparison in studies of alternative treatments for carotid stenosis.

Acknowledgments

Stephen N. Palmer, PhD, ELS, contributed to the writing of this article. Dr. Palmer, William K. Vaughn, PhD, and Branka Kosarac, MD, provided statistical support.