Multicenter Experience on Eversion Versus Conventional Carotid Endarterectomy in Symptomatic Carotid Artery Stenosis
Observations From the Stent-Protected Angioplasty versus Carotid Endarterectomy (SPACE-1) Trial
Background and Purpose—Carotid endarterectomy (CEA) is beneficial in patients with symptomatic carotid artery stenosis. However, randomized trials have not provided evidence concerning the optimal CEA technique, conventional or eversion.
Methods—The outcome of 563 patients within the surgical randomization arm of the Stent-Protected Angioplasty versus Carotid Endarterectomy in Symptomatic Patients (SPACE-1) trial was analyzed by surgical technique subgroups: eversion endarterectomy versus conventional endarterectomy with patch angioplasty. The primary end point was ipsilateral stroke or death within 30 days after surgery. Secondary outcome events included perioperative adverse events and the 2-year risk of restenosis, stroke, and death.
Results—Both groups were similar in terms of demographic and other baseline clinical variables. Shunt frequency was higher in the conventional CEA group (65% versus 17%; P<0.0001). The risk of ipsilateral stroke or death within 30 days after surgery was significantly greater with eversion CEA (9% versus 3%; P=0.005). There were no statistically significant differences in the rate of perioperative secondary outcome events with the exception of a significantly higher risk of intraoperative ipsilateral stroke rate in the eversion CEA group (4% versus 0.3%; P=0.0035). The 2-year risk of ipsilateral stroke occurring after 30 days was significantly higher in the conventional CEA group (2.9% versus 0%; P=0.017).
Conclusions—In patients with symptomatic carotid artery stenosis, conventional CEA appears to be associated with better periprocedural neurological outcome than eversion CEA. Eversion CEA, however, may be more effective for long-term prevention of ipsilateral stroke. These findings should be interpreted with caution noting the limitations of the post hoc, nonrandomized nature of the analysis.
The value of carotid endarterectomy (CEA) has been well established in patients with symptomatic and asymptomatic carotid artery stenosis.1–5 Currently, 2 techniques for endarterectomy are used. Conventional carotid endarterectomy (C-CEA), performed through a longitudinal arteriotomy of the internal carotid artery, is the most frequently used technique. A patch closure of the arteriotomy is usually used and has been demonstrated to be associated with superior outcome over primary closure of the vessel.6–9 Eversion carotid endarterectomy (E-CEA) was initially described by DeBakey and later popularized by others.10–13 In a recent Cochrane Collaboration systematic review comparing E-CEA and C-CEA, the authors concluded that there is currently no evidence that 1 technique is superior to the other with regard to periprocedural stroke, death, restenosis, or local complications.14 Thus, the ideal surgical technique for CEA has yet to be determined and the choice of the technique continues to be based on the personal experience and preference of the operating surgeon.
Published studies, whether from randomized15–19 or single-arm trials, have included patients with symptomatic as well as asymptomatic carotid disease.20–24 By contrast, the Stent-Protected Angioplasty versus Carotid Endarterectomy (SPACE-1) study included only symptomatic patients, randomized to carotid artery stenting or CEA with a conventional or eversion technique. We present the 2-year outcome of E-CEA versus C-CEA as a subset analysis of the SPACE-1 CEA arm comprising a discrete population of closely followed symptomatic patients with extracranial carotid disease.
Patients and Methods
All patients in the present study were enrolled in SPACE-1 prospective, randomized, multicenter clinical trial. The study protocol was approved by the ethics committees at each of the 35 centers and informed consent was obtained in each patient. SPACE-1 was designed to test the hypothesis that carotid artery stenting is not inferior to CEA for the treatment of severe symptomatic carotid stenosis. A detailed delineation of the eligibility criteria for the study has been published elsewhere.25,26 Briefly, subjects were eligible for enrollment if they had amaurosis fugax, hemispheric transient ischemic attack, or a completed stroke within the previous 180 days and had a stenosis of the ipsilateral carotid artery ≥50% in diameter reduction according to criteria from the North American Symptomatic Carotid Endarterectomy Trial (NASCET)3 or ≥70% according to criteria from the European Carotid Surgery Trial (ECST).2 Carotid duplex ultrasound, performed by experienced and certified ultrasonographers, was used to assess the degree of carotid artery stenosis at the time of randomization into the SPACE-1 trial and during follow-up after intervention.
Outcome analyses were based on per-protocol data from the SPACE-1 CEA treatment arm with several caveats. Patients who underwent C-CEA without patch closure of the arteriotomy were excluded from the analysis, noting the current consensus that patch angioplasty is superior to primary closure with regard to early and long-term outcome.6–9,27 As well, patients in whom a common carotid to internal carotid artery (ICA) bypass graft was placed were not included in the analysis. Patient follow-up consisted of duplex ultrasound at 1, 7, and 30 days and at 6, 12, and 24 months after operation. The primary end point of the SPACE-1 trial was ipsilateral ischemic or hemorrhage stroke with symptoms lasting >24 hours or death, both assessed at 30 days of operation. The rate of intraoperative ipsilateral stroke, transient ischemia attack, ipsilateral carotid artery occlusion, the development of a wound hematoma that required re-exploration, cranial nerve injury, cardiovascular events, dysesthesia of the cervical wound, postoperative hypertension, headache, seizure, and wound infection were recorded. Cardiovascular events were defined based on clinical criteria; a systematic screening using electrocardiogram or laboratory values was not done. Transient ischemia attack was defined as an acute disturbance of focal neurological function with symptoms lasting <24 hours. Residual carotid stenosis was defined when a stenosis of ≥50% was documented on the initial postoperative duplex ultrasound study. Investigators were not blinded to the type of endarterectomy the patient underwent when assessing clinical outcome.
At the 2-year time point, the occurrence of ipsilateral stroke, any stroke, disabling stroke with a score on the modified Rankin Scale of at least 3, death from any cause, vascular and nonvascular death, and restenosis ≥70% were tabulated. Strokes and deaths were categorized as procedural (occurring within 30 days of operation) or nonperiprocedural (occurring after 30 days). Restenosis was classified by the time of detection.
With a sample size of 206 patients in the E-CEA group and 310 patients in the C-CEA group, there was 81% power at the 0.05 significance level to detect a 6% difference in the rate of the primary end point “ipsilateral stroke or death within 30 days after surgery” using a 2-sided Fisher exact test.
Vascular surgeons were required to be experienced in the procedure of CEA and board-certified. Each was qualified for participation in the study after review of the outcome of 25 consecutive CEA procedures by a quality control committee headed by one of the authors (H.-H.E.). There were no restrictions with respect to the type of anesthesia (local or general), type of neurological monitoring if any, heparin dose, use of patch closure of the arteriotomy, choice of patch material, or indications for shunting.
The E-CEA technique was performed with an oblique transection of the ICA off the common carotid artery, endarterectomy of the ICA through an eversion technique, endarterectomy of the stump of the carotid bifurcation and the external carotid artery, and reimplantation of the endarterectomized ICA onto the common carotid. C-CEA was performed through a longitudinal arteriotomy, running from the carotid bifurcation onto the anterolateral surface of the ICA. The arteriotomy was closed with an autologous or prosthetic patch at the discretion of the operating surgeon. In those subjects who underwent selective use of an indwelling shunt, intraoperative neurological monitoring was carried out with transcranial Doppler scan, somatosensory-evoked potentials, electroencephalography, and other techniques not specified by the study protocol.
For continuous variables, median, means, and SDs were calculated; numeric data were analyzed with unpaired t tests or with Mann-Whitney U tests as appropriate. Univariate testing used Yates-corrected χ2 and Fisher-Freeman-Halton exact test for comparison of proportions.28 For the analysis of the association of independent variables with the primary end point, multivariable stepwise backward logistic regression was used, removing variables with nonsignificant ORs and specifying the final model with ORs and 95% CIs. Two-sided Fisher exact test was used for power analysis for the detection of a significant difference in regard to rates of the primary end point. Probability values were 2-sided and a difference was considered statistically significant when the value was ≤0.05. Kaplan-Meier methodology was used for survival, analyzing differences with the Peto log-rank test. Cox regression analysis was performed when the log-rank test demonstrated a statistically significant difference. Statsdirect statistical software, Version 2.7.3 (Statsdirect Ltd, Cheshire, UK) was used for all analyses.
A total of 601 eligible patients were randomized into the surgical arm of the SPACE-1 trial. After excluding 12 patients who withdrew consent before treatment, 24 patients who did not undergo per-protocol treatment (18 never underwent operation and 6 had carotid artery stenting instead of CEA), 2 patients who experienced a primary outcome event before surgery, 43 patients who had C-CEA without patch closure of the arteriotomy, and 4 patients with prosthetic bypass graft interposition to the ICA, 516 remaining patients (85.9%) were suitable for analysis. Among these, 310 (60.1%) underwent C-CEA and 206 (39.9%) had E-CEA.
The baseline variables in the 2 groups were similar with the exception of a greater number of females and a slightly greater severity of stenosis in the E-CEA group (Table 1). The majority of the operative procedures were performed under general anesthesia (E-CEA 81% [n=167]; C-CEA 94% [n=292], difference 0.13; 95% CI, 0.07–0.19; P<0.0001). Cerebral monitoring was performed with transcranial Doppler (E-CEA 12.6% [n=26]; C-CEA 13.5% [n=42]; P=0.86), somatosensory-evoked potentials (E-CEA 44.2% [n=91]; C-CEA 54.5% [n=169]; P=0.03), electroencephalography (E-CEA 2.9% [n=6]; C-CEA 3.2% [n=10]; P>0.99), and other methods (E-CEA 36.4% [n=75]; C-CEA 14.5% [n=45]; P<0.0001). A shunt was used more frequently in the patients undergoing C-CEA (65% versus 17%; P<0.0001). Intraoperative angiographic verification of the technical result of the endarterectomy was carried out in 81 patients (39%) in the E-CEA group and 202 patients (65%) in the C-CEA group (P<0.0001). The duration of the operative procedure was significantly shorter in the E-CEA group (72 [SD ±27 minutes] versus 96 [SD ±28 minutes]; P<0.0001; 95% CI, 19.4–29.2). Median ICA clamp time was 20 minutes (interquartile range, 15–25 minutes) in the E-CEA group and 8 minutes (interquartile range, 5–24 minutes) in the C-CEA group (95% CI for difference between medians, 7–11); P<0.0001).
Perioperative ipsilateral stroke or death was more frequent in the E-CEA group, 9% versus 3% (P=0.005; Figure 1). The overall perioperative death rate was 0.6% (n=3); 1 vascular and 1 nonvascular death occurred in the E-CEA group (2 of 206 [1.0%]) and 1 nonvascular death occurred in the C-CEA group (1 of 310 [0.3%]; P=0.57). Intraoperative ipsilateral stroke occurred more frequently in the E-CEA group (n=8 [4%] versus n=1 [0.3%]; P=0.0035; OR, 12.49; 95% CI, 1.55–100.59; Table 2; Figure 1), although there was a lack of power due to small numbers. Six of these 8 patients were not shunted, each of which had stroke as a preoperative qualifying event and positive findings on cranial CT. The only patient in the C-CEA group with intraoperative ipsilateral stroke was shunted and had a transient ischemia attack with negative findings on cranial CT scan before surgery.
Early postoperative reoperation was necessary in 10 patients (1.9%) without a difference between the 2 groups (5 [2.4%] in E-CEA and 5 [1.6%] in C-CEA; P=0.53). The early reoperations were performed for hematoma (9 [1.7%]) or thrombosis of the reconstruction (1 [0.2%]). Other perioperative outcome measures did not differ by treatment group (Table 2).
The logistic regression model identified E-CEA as the sole parameter associated with an increased risk of perioperative stroke or death (OR, 3.20; 95% CI, 1.41–7.28; P=0.005), excluding nonsignificant variables consisting of shunt use (OR, 0.68; 95% CI, 0.30–1.50; P=0.34), severity of the stenosis (OR, 1.03; 95% CI, 0.99–1.06; P=0.09), and sex (OR, 0.73; 95% CI, 0.30–1.77; P=0.88). There was no difference between sex-stratified proportions of the outcome event “periprocedural ipsilateral stroke or death” (male=5.5% versus female=4.6%; P=0.83). Within the E-CEA group, the use of a shunt during appeared to be associated with an increased risk of ipsilateral stroke, but this difference did not attain statistical significance (2 of 35 [5.7%] versus 1 of 201 [0.5%]; P=0.06). Transient cranial nerve palsies in early postoperative period were recorded in 17 (8.2%) patients after E-CEA and in 25 (8.1%) patients after C-CEA without a significant difference between the groups (Table 2; Figure 1).
There were no significant differences in the overall rate of stroke or death at the 2-year postoperative time point (Table 3; Figures 1 and 2). Compared with E-CEA, however, C-CEA was associated with a significantly higher risk of ipsilateral stroke beyond the 30-day perioperative period (C-CEA 2.9% versus E-CEA 0%; hazard ratio, ∞ 95% CI, 1.25–∞ log-rank P=0.017; Table 4; Figure 2). There was no difference in the rate of death after 30 days between the E-CEA- and C-CEA-treated patients (E-CEA 3.4%, C-CEA 2.9%; hazard ratio, 0.82; 95% CI, 0.30–2.24; log-rank P=0.69). As well, trends toward an increased rate of death or any stroke beyond the 30-day time point did not attain statistical significance (E-CEA 4.4%; C-CEA 9.0%; hazard ratio, 1.97; 95% CI, 1.02–3.82; log-rank, P=0.074; Table 4).
Restenosis and Reintervention
There was no statistically significant difference in the 24-month rate of restenosis after E-CEA versus C-CEA (Fisher-Freeman-Halton exact test, E-CEA n=5 [2.4%], C-CEA n=10 [3.2%]; P=0.79; OR, 0.75; 95% CI, 0.25–2.22; Figure 1). Similarly, there was no difference in the rate of reintervention; 2 patients undergoing C-CEA (0.4%) underwent revision for restenosis, 1 with percutaneous balloon transluminal angioplasty 6 months after the initial procedure and 1 with carotid artery stenting 12 months after the original endarterectomy procedure.
Analysis of the endarterectomy subgroups from the SPACE-1 multicenter trial demonstrates similar 2-year outcome after E-CEA or C-CEA without significant differences in the rate of stroke, death, or recurrent stenosis. There were, however, interesting differences in the frequency of early and later complications with the 2 techniques. Whereas E-CEA was associated with a higher risk of perioperative stroke, it appeared to offer greater protection from stroke between 30 days and 2 years after operation.
Numerous studies have compared C-CEA with patching with E-CEA.15–24,29 Perhaps the most important of these is EVEREST (Eversion carotid Endarterectomy versus Standard Trial study), a randomized prospective multicenter study first published in 1998.18 With the exception of the EVEREST and the findings of a few other reports,30,31 the majority of studies found E-CEA to be as effective or superior to C-CEA with regard to perioperative stroke and death.13,16,19,22,24,29,32–34
In the EVEREST trial, 1353 patients were randomly assigned to E-CEA or C-CEA with shunting and patch closure of the arteriotomy left to the discretion of the operating surgeon. The conventional group consisted of 419 patients with primary arteriotomy closure and 256 patients with patch angioplasty. There were no statistically significant differences in outcome between C-CEA and E-CEA, although a slightly higher incidence of perioperative complications was noted in the E-CEA group. A subsequent publication of the EVEREST trial appeared in 2000 with longer follow-up of the patients. Patients who underwent E-CEA had a lower incidence of restenosis than did those who underwent C-CEA (3.6% versus 9.2%; P=0.01), but a substantial number of C-CEA procedures was performed with primary closure of the arteriotomy. In fact, the C-CEA subgroup with patch angioplasty had the lowest incidence of ipsilateral stroke and the lowest risk of restenosis at 4 years.35
The findings of the current study corroborate the results of EVEREST. In our study, however, E-CEA was not associated with a significantly reduced incidence of restenosis in contrast to both EVEREST and a Cochrane Collaboration review of E-CEA versus C-CEA.14 Nevertheless, E-CEA performed better than C-CEA for stroke prevention between 30 days and 2 years after operation. This observation leads one to speculate whether a less stringent definition for recurrent stenosis might have resulted in the identification of lower profile yet still embologenic restenoses.
When periprocedural outcomes were included in the 2-year outcome parameters, no significant difference in stroke and death rates was detectable between the 2 groups.
Shunt placement has been perceived as problematic at the time of performing E-CEA,36 a view that may underlie the lower frequency of shunt use and consequential longer carotid clamp duration in the E-CEA group of the current study. The studies of Abu Rahma37 and others38,39 suggest that shunting is of particular importance in symptomatic patients such as those enrolled in the current trial, patients who appear predisposed to cerebral ischemia during carotid clamping. It is possible that this issue underlies the higher 30-day ipsilateral stroke rate in the E-CEA group, in which 6 of 8 patients who experienced intraoperative ipsilateral stroke did not have a shunt placed during their eversion procedure.
We were unable to detect a statistically significant association between outcome and sex, shunt use, or the severity of carotid stenosis. Although the failure to detect these associations may be a result of the relatively modest sample size of the current study, the Nationwide Inpatient Sample was similarly unable to detect a significant sex effect in symptomatic stenosis despite inclusion of 54 658 cases in the analysis.40 Even in the NASCET trial, neither sex nor severity of stenosis showed any influence on the perioperative ipsilateral stroke or death rates.3
Furthermore, although others have suggested that the more extensive dissection of the carotid bifurcation and the distal ICA necessary to obtain adequate exposure for E-CEA can result in a particularly high rate of cranial nerve injuries, we did not confirm a greater risk with the eversion technique.41
The current study suffers from all of the limitations inherent in any nonrandomized, post hoc analysis. In this context the sample size in the data set available was not sufficiently large to permit splitting the data into a training set (for model development) and a test set (for confirmation of model validity). Stepwise regression methods for variable selection are known to be unstable, especially in case of relatively small numbers of outcome events. Furthermore, the infinite hazard ratio CI for the ipsilateral stroke rate beyond 30 days and the wide OR CI for the death rate results indicate that a substantial effect in either direction has not been ruled out by the study findings. However, SPACE-1 was an independently monitored multicenter study, lending more credence to the accuracy of complication rates. Therefore, even secondary analysis of these data may have much more impact than single-center studies. The choice of the endarterectomy technique was left to the discretion of the surgeon, and some surgeons or centers might preferentially use 1 technique over the other; thus, an inordinately high or low surgeon- and center-specific complication rate could confound the analysis. Information on why 1 endarterectomy technique was chosen over the other was not captured and blinded outcome assessment was not performed in the study. As well, certain adverse events such as cranial nerve injury were not included in the predefined SPACE end points and, as such, their frequency might have been underestimated.
In patients with symptomatic carotid artery bifurcation disease, C-CEA with patch closure for symptomatic carotid artery stenosis is associated with a better early neurological outcome compared with E-CEA. However, E-CEA appears to be more effective than C-CEA for the long-term prevention of stroke such that outcome of the techniques is similar at the 2-year time point. These findings should be interpreted with caution noting the limitations of the post hoc, nonrandomized nature of the analysis.
- Received October 14, 2011.
- Revision received February 9, 2012.
- Accepted February 28, 2012.
- © 2012 American Heart Association, Inc.
- Moore WS,
- Mohr JP,
- Najafi H,
- Robertson JT,
- Stoney RJ,
- Toole JF
- Fields WS,
- Crawford ES,
- Debakey ME
- Cao P,
- De Rango P,
- Zannetti S,
- Giordano G,
- Ricci S,
- Celani MG
- Balzer K
- Eckstein HH,
- Ringleb P,
- Allenberg JR,
- Berger J,
- Fraedrich G,
- Hacke W,
- et al
- Bailey NTJ