Early Endarterectomy Carries a Lower Procedural Risk Than Early Stenting in Patients With Symptomatic Stenosis of the Internal Carotid Artery
Results From 4 Randomized Controlled Trials
Background and Purpose—Patients undergoing carotid endarterectomy (CEA) for symptomatic stenosis of the internal carotid artery benefit from early intervention. Heterogeneous data are available on the influence of timing of carotid artery stenting (CAS) on procedural risk.
Methods—We investigated the association between timing of treatment (0–7 days and >7 days after the qualifying neurological event) and the 30-day risk of stroke or death after CAS or CEA in a pooled analysis of individual patient data from 4 randomized trials by the Carotid Stenosis Trialists’ Collaboration. Analyses were done per protocol. To obtain combined estimates, logistic mixed models were applied.
Results—Among a total of 4138 patients, a minority received their allocated treatment within 7 days after symptom onset (14% CAS versus 11% CEA). Among patients treated within 1 week of symptoms, those treated by CAS had a higher risk of stroke or death compared with those treated with CEA: 8.3% versus 1.3%, risk ratio, 6.7; 95% confidence interval, 2.1 to 21.9 (adjusted for age at treatment, sex, and type of qualifying event). For interventions after 1 week, CAS was also more hazardous than CEA: 7.1% versus 3.6%, adjusted risk ratio, 2.0; 95% confidence interval, 1.5 to 2.7 (P value for interaction with time interval 0.06).
Conclusions—In randomized trials comparing stenting with CEA for symptomatic carotid artery stenosis, CAS was associated with a substantially higher periprocedural risk during the first 7 days after the onset of symptoms. Early surgery is safer than stenting for preventing future stroke.
Carotid artery stenting (CAS) has evolved as an alternative treatment for carotid artery disease. During the past 20 years, CAS has striven to prove its feasibility and efficacy in stroke prevention when compared with that of carotid endarterectomy (CEA) for patients with symptomatic internal carotid artery (ICA) stenosis. Because of the high risk of early stroke recurrence after plaque rupture, it is now accepted that intervention offers the greatest benefit when performed soon after the onset of neurological symptoms.1,2 The somewhat greater perioperative risk of rapid CEA is offset by a much lower risk of stroke recurrence.3
Timing of treatment could also influence the results of CAS. Unlike early surgery, CAS seems to be a higher risk procedure when performed soon after symptoms.
The 2012 analysis based on individual patient-level data from 3 randomized trials comparing CAS and CEA reported by the Carotid Stenosis Trialists’ Collaboration (CSTC) suggested that timing of intervention influenced the occurrence of outcomes. CAS between days 0 and 7 was associated with the highest number of procedural complications when compared with patients treated between 8 and 14 days or thereafter. In contrast, surgery during each of these time intervals was safer.4
In this updated analysis, we added data from individual patients with symptomatic carotid stenosis from the CREST (Carotid Revascularization Endarterectomy Versus Stenting Trial). This provided the largest group yet studied of patients with symptomatic carotid stenosis randomized between CEA and CAS and enabled us to investigate associations between periprocedural outcome and timing of treatment for both techniques.
Four randomized clinical trials with blinded adjudicated outcomes were included; EVA-3S (NCT 00190398), SPACE (ISRCTN 57874028), ICSS (ISRCTN 25337470), and CREST (NCT 00004732). In the first 3 of these trials, patients with symptomatic moderate-to-severe carotid stenosis (≥50% stenosis measured according to NASCET criteria5 [North American Symptomatic Carotid Endarterectomy Trial]), deemed suitable for both procedures, were randomly allocated to CAS or CEA.6–8 Our pooled analysis of individual patient data was prospectively agreed at the design stage of the 3 European trials.9 Data on symptomatic patients from CREST were added in 2015. CREST included patients with transient ischemic attack, amaurosis fugax, and minor nondisabling ischemic stroke. To be eligible for CREST, patients had to have a carotid artery stenosis of at least 50% on invasive angiography, ≥70% on ultrasound, or ≥70% on computed tomographic or magnetic resonance angiography if the stenosis was 50% to 69% on ultrasound.10
The primary outcome of the present analysis was the combination of any stroke or death occurring within 30 days after treatment. Secondary outcome events were any stroke and fatal or disabling stroke happening within the same time period. The analysis was done per protocol: patients were only included in the analysis if the randomly allocated treatment was the first initiated revascularization procedure and if either the date of the qualifying event (the last ischemic event ipsilateral to the carotid artery being randomized in the trial) or the interval between the qualifying event and treatment was known. Patients with missing data on delay between qualifying event and treatment were excluded from the analysis.
In 3 studies (EVA-3S, ICSS, and CREST), the date of the qualifying event was entered at baseline. In the SPACE trial, the date of the qualifying event was not prospectively assessed at study entry; however, the date of the qualifying event was retrospectively retrieved whenever possible for this pooled analysis. If the exact date was unknown, patients were included if information was available whether treatment had taken place within 7 days of the qualifying event or thereafter.
To obtain a combined estimate (risk ratio [RR] with 95% confidence intervals [CIs]), logistic mixed models were applied with the source study as random variable using a log link. The first model was unadjusted, whereas the second model accounted for age at treatment, sex, and type of qualifying event (retinal ischemia, transient ischemic attack, or stroke). Age at treatment was log-transformed based on the natural logarithm (ln) in the mixed-model analysis. In contrast to our previous analysis,4 just 2 timing groups were created, because results of patients treated between 8 and 14 days or thereafter were similar (Table I in the online-only Data Supplement), irrespective of the treatment technique. Therefore, the primary analysis compared patients treated within 7 days of neurological symptoms or thereafter. Second, an interaction between timing of treatment and treatment effect (CAS versus CEA) was tested by integrating a multiplicative interaction term in the logistic mixed model analyses. A P value of <0.10 for interaction terms was considered statistically significant. For all other statistical analyses, a P value of <0.05 was considered to indicate statistical significance.
The pooled data set for all 4 trials included 4754 patients with symptomatic ICA stenosis. 2361 patients were randomized to CEA (49.7%) and 2393 patients to CAS. For both treatment groups, many patients were excluded from data analysis because of missing information about their most recent neurological event or the treatment date (n=290 patients in the CEA group and n=274 in the CAS group). Another 52 patients (26 in each treatment group) were excluded because they did not receive the initially allocated treatment. Figure 1 gives detailed information about included and excluded patients by source trial. In total, 4138 patients (n=2045 in the CEA and 2093 in the CAS group) remained for per-protocol analysis. Table 1 summarizes the baseline characteristics of both treatment groups. The median delay between the most recent neurological event and treatment was 26 days (interquartile range: 11–61) for CAS and 29 days (interquartile range: 13–67) for CEA. Among 4138 patients, a small but relevant group underwent CAS and CEA within a week of their symptoms (14% in CAS versus 11% in CEA). Treatment groups did not differ for neurological parameters and comorbidities. Baseline characteristics were additionally provided for the 2 timing groups (Table 2).
Table II in the online-only Data Supplement compares patients included and excluded from the current analysis because of available or missing timing information. There were some minor and most probably random differences in baseline characteristics of these 2 groups.
Overall Outcome in the Study Population for Both Treatment Groups (CAS Versus CEA)
The risk of any stroke or death within 30 days after treatment was higher for the CAS compared with the CEA group for the entire study population: 7.3% versus 3.3%; crude RR, 2.29; 95% confidence interval (CI), 1.71–3.08. This association remained significant when the model was additionally adjusted for age at treatment, sex, and type of qualifying event (RR, 1.92; 95% CI, 1.50–2.47).
Outcome in Both Treatment Groups by Timing of Treatment (0–7 Days and >7 Days)
In the early period after the onset of neurological symptoms (0–7 days), CAS had the highest number and proportion of periprocedural strokes and deaths (n=24/287, 8.4%), compared with CEA (n=3/226, 1.3%). Patients in the CAS group had a higher risk of any stroke or death in the crude (RR, 6.51; 95% CI, 2.00–21.21) and adjusted models (RR, 6.74; 95% CI, 2.07–21.92; Figure 2; Table 3).
Compared with those treated within 7 days, patients treated after 7 days had fewer strokes and deaths in the CAS group (n=129/1806, 7.1%), whereas the risk of stroke and death in the CEA group slightly increased (n=65/1819, 3.6%). The RR for CAS compared with CEA was still higher in this later treatment group: RRcrude, 2.00; 95% CI, 1.49–2.67 (Table 3); RRadjusted, 2.00; 95% CI, 1.50–2.68 (Figure 2). Results were almost identical for the outcome analysis of any stroke: RRcrude for CAS in the early treatment group 6.27; 95% CI, 1.92–20.44; RRcrude for CAS after 7 days 1.98; 95% CI, 1.47–2.67 (Table 3). Adjustment did not importantly change results (Figure 2). The analysis of fatal or disabling stroke outcome at 30 days also showed that the crude RR was higher for CAS than the CEA group within 7 days (RR, 8.29; 95% CI, 1.07–64.28) and after 7 days (RR, 1.77; 95% CI, 1.10–2.85; Table 3). Results were virtually unchanged after adjustment (Figure 2).
Interaction Between Time and Relative Risks of CAS Versus CEA
The test for interaction between timing of treatment and treatment effect (CAS versus CEA) revealed a P value of 0.07 in the crude and 0.06 in the adjusted model for the outcome any stroke or death. Comparable results were seen for the outcome any stroke at 30 days (P=0.07 for both models). There was no statistically significant interaction seen for fatal or disabling stroke (P=0.17; Figure 2).
CAS is not as safe as CEA in the treatment of patients with symptomatic stenosis of the ICA irrespective of the timing of treatment. The difference in safety between CAS and CEA is particularly potent in patients treated within 7 days of symptom onset.
There has been a heated debate as to whether early surgery in symptomatic patients is safe and meaningful. However, it is now widely accepted that early plaque removal effectively reduces stroke risk. Although early surgery may be associated with a slightly higher risk of perioperative complications, it still offers the best chance of a symptomatic patient avoiding future stroke.3,11 Recent literature on the risk of stroke recurrence after initial plaque rupture provides somehow controversial results. Most authors suggest that the risk of early recurrent stroke from symptomatic ICA stenosis remains high.12–14 In a recently published series of 377 patients with symptomatic ICA stenosis stroke recurrence rate reached 2.7% within the first day, 5.3% within 3 days, and 18.8% within 90 days after the qualifying event.14 Only one retrospective Swedish study reported a lower overall number of second events in 397 patients with symptomatic stenosis of the ICA (for recurrent stroke 2.0%; 95% CI, 0.6–3.4 by day 2, 2.4%; 95% CI, 2.0–5.9 by day 7, and 7.5%; 95% CI, 4.4–10.6 by day 30).15 In accordance with that, results from a medical intervention study revealed that the number of neurological events was significantly reduced with the application of best medical treatment after symptom onset. The intake of aspirin, clopidogrel, and a statin could relevantly decrease the number of recurrent neurological events in a series of 188 patients.16
Data from the Swedish Vascular Registry (SWEDVASC) were analyzed to investigate the time dependence of CEA outcome among >2500 symptomatic patients. The authors found that rapid surgery (between 0 and 2 days) was associated with a significantly higher frequency of perioperative complications (any stroke or death) when compared with patients treated between 3 and 7 days, 8 and 14 days, and thereafter (11.5% versus 3.6% versus 4.0% versus 5.4%; P<0.001, respectively).17 Only a small number of Swedish Vascular Registry (SWEDVASC) patients were treated in the early period (5.7%), which might limit the generalizability of these registry data. In contrast, their slight increase in perioperative complications was not replicated in 2 single-center studies.18,19 Both studies reported comparable perioperative complication rates for the same 4 surgical timing groups. A recent analysis of >56 000 patients with symptomatic ICA stenosis from the German nationwide statutory quality assurance registry also revealed no outcome difference between patients treated early by CEA (within 48 hours) and thereafter (any stroke or death 3.0% for CEA between 0 and 2 days versus 2.5% between 3 and 7 days versus 2.6% between 8 and 14 days versus 2.3% for CEA thereafter).20 In addition, data from the National Vascular Registry from the United Kingdom illustrated comparable outcomes for the 4 timing groups among >23 000 symptomatic patients.21 Complementary to this register data, which only contained CEA patients, results from the National (Nationwide) Inpatient Sample were published. In this analysis, authors investigated the influence of ultraearly revascularizations (within 48 hours) on the outcome of CAS and CEA in >70 000 symptomatic patients. The comparison between CAS and CEA when performed within 48 hours after the onset of symptoms showed that CAS was associated with significantly more periprocedural complications, regardless of whether patients had a cerebral infarction on admission or not (OR, 3.45; 95% CI, 3.13–3.80; P<0.01 for CAS patients with infarct on admission compared with CEA under same conditions; OR, 2.53; 95% CI, 2.40–2.66; P<0.001 for CAS patients without infarct on admission compared with CEA again under same conditions).22 The authors did not find any outcome differences after later treatment for both CEA and CAS. Recently, the influence of timing on the outcome of CAS and CEA was also analyzed in the CREST data. The authors used 3 timing groups (CAS or CEA <15 days, 15–60 days, and thereafter) and did not see any time dependence for periprocedural outcome for both treatment techniques (HR for stroke or death in the CAS group comparing 15–60 days to <15 days 1.15; 95% CI, 0.48–2.75 and 1.12, 95% CI, 0.53–2.40 comparing >60 days to <15 days; both P=0.93). For the CEA group comparing 15–60 days to <15 days, the HR was 0.74; 95% CI, 0.22–2.49 and 0.91; 95% CI, 0.25–3.33, respectively, comparing >60 days to <15 days; both P=0.89).23 Differences in the findings between the CREST trial alone and the pooled CSTC data of all 4 randomized trials might be because of different time strata, making results more difficult to compare.
In the present report, we found that early CEA (in the period of 0–7 days after the onset of symptoms) had the lowest absolute risk for periprocedural complications for all 3 outcomes, whereas surgical risks in the later period were somewhat higher (eg, any stroke or death 1.3% for 0–7 days versus 3.6% after 7 days). In contrast, early CAS carried the highest periprocedural stroke or death risk, decreasing slightly in the later period (8.4% versus 7.1%). This suggests that the recently symptomatic ICA plaque with a ruptured plaque surface needs some time for stabilization to allow safer catheter passage in CAS.24,25 The clinical decision to perform early revascularization of symptomatic carotid stenosis is likely influenced by characteristics of the patient and the symptomatic event. For example, we saw in both treatment groups that the percentage of patients with a hemispheric stroke before inclusion was ≈10% higher in patients treated late than in patients treated early (Table 2). It is possible that clinically more stable patients were predominantly selected for early treatment.
In another analysis of the National (Nationwide) Inpatient Sample, the authors focused on patients with symptomatic ICA stenosis and recent cerebral infarction. Analyses were done in 4 timing groups: CEA or CAS within 48 hours after the onset of symptoms, between 48 hours and 4 days, between 5 and 7 days, and between 8 and 14 days. Among the 27 839 patients with recent cerebral infarction, patients treated between 5 and 7 days after symptoms had the lowest probability of periprocedural complications (OR, 0.64; 95% CI, 0.56–0.74; P<0.001) and mortality (OR, 0.63; 95% CI, 0.45–0.89; P<0.001), irrespective of the treatment technique.26 Without having details about cerebral lesions, we found in the CSTC population that CEA was most beneficial when performed between 0 and 7 days, whereas CAS was most harmful during the first week after the onset of symptoms. Because of small numbers of patients who were treated early after the onset of symptoms and relatively small numbers of periprocedural complications in both treatment groups, we could not build further time groups to determine an ideal time point for CAS and CEA among our study population.
Limitations of Our Analysis
Timing of treatment has to date never been a randomization criterion in larger trials. All information on the influence of timing of treatment is derived from post hoc analysis of randomized controlled trials comparing 2 treatment techniques (best medical treatment versus CEA or CEA versus CAS). Therefore, detailed information on patient selection and disease severity is lacking. This significantly limits the value of timing analysis to date. A randomized trial on timing of treatment would be mandatory in the near future.
Recent ischemic infarction on neuroimaging is reported to be a relevant risk factor for periprocedural complications after CAS and after CEA. Unfortunately, detailed findings from baseline CT or MRI were not available for the present analysis.
The use of embolic protection devices was mandatory in CREST but not in the 3 remaining trials. We could, therefore, not deliver detailed outcome analysis of CAS under the use of embolic protection devices compared with CEA.
CEA is effective at preventing stroke. Early plaque removal can be performed without relevantly increasing perioperative complications. In contrast, CAS during the early period after plaque rupture is associated with an increased risk of periprocedural complications. We could clearly demonstrate in this randomized and large population of symptomatic patients that risk differences between CAS and CEA were greatest in the early days after the index symptom. Early CEA was associated with the lowest risk of periprocedural complications. We, therefore, conclude that early CEA compared with early CAS after an initial neurological event offers the highest stroke prevention benefit for the patient at risk.
Dr Rantner wrote the first draft of the report and was supervised by Dr Fraedrich. Dr Kollerits, Dr Rantner, and Dr Fraedrich designed the statistical analysis plan, and Dr Kollerits undertook the statistical analyses. Dr Mas, Dr Ringleb, Dr Bonati, and Dr Howard extracted patients’ data from contributing trials. All the authors listed in the writing committee made substantial contributions to conception and design of the study, acquisition of data, or analysis and interpretation of data and also contributed to drafting the report or revising it critically for important intellectual content. Last but 3 author, last but 2 author, last but 1 author and last author contributed equally to the report. Dr Fraedrich had the final responsibility for the analyses and the content of the report.
Sources of Funding
Dr Bonati was supported by grants from the Swiss National Science Foundation (PBBSB-116873), the University of Basel, Switzerland, and The Stroke Association. Dr Brown’s Chair in Stroke Medicine at University College London is supported by the Reta Lila Weston Trust for Medical Research. Part of this work was undertaken at University College London Hospitals/University College London, who received a proportion of funding from the Department of Health’s National Institute for Health Research Biomedical Research Centers funding scheme. Dr Halliday’s research is funded by the NIHR Oxford Biomedical Research Center. G. Howard is funded by the National Institute of Health/National Institute of Neurological Disorders and Stroke. Detailed information about the Carotid Stenosis Trialists’ Collaboration and Acknowledgments are added in the online-only Data Supplement.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
↵Guest Editor for this article was Seemant Chaturvedi, MD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.016233/-/DC1.
- Received December 8, 2016.
- Revision received March 7, 2017.
- Accepted March 13, 2017.
- © 2017 American Heart Association, Inc.
- Goldstein LB,
- Bushnell CD,
- Adams RJ,
- Appel LJ,
- Braun LT,
- Chaturvedi S,
- et al
- Rantner B,
- Goebel G,
- Bonati LH,
- Ringleb PA,
- Mas JL,
- Fraedrich G
- 5.↵North American Symptomatic Carotid Endarterectomy Trial. Methods, patient characteristics, and progress. Stroke. 1991;22:711–720.
- Featherstone RL,
- Brown MM,
- Coward LJ
- Mas JL,
- Chatellier G,
- Beyssen B
- Hacke W,
- Brown MM,
- Mas JL
- Mantese VA,
- Timaran CH,
- Chiu D,
- Begg RJ,
- Brott TG
- Johansson E,
- Cuadrado-Godia E,
- Hayden D,
- Bjellerup J,
- Ois A,
- Roquer J,
- et al
- Shahidi S,
- Owen-Falkenberg A,
- Hjerpsted U,
- Rai A,
- Ellemann K
- Strömberg S,
- Gelin J,
- Osterberg T,
- Bergström GM,
- Karlström L,
- Osterberg K
- Tsantilas P,
- Kuehnl A,
- König T,
- Breitkreuz T,
- Kallmayer M,
- Knappich C,
- et al
- Loftus IM,
- Paraskevas KI,
- Johal A,
- Waton S,
- Heikkila K,
- Naylor AR,
- et al
- Meschia JF,
- Hopkins LN,
- Altafullah I,
- Wechsler LR,
- Stotts G,
- Gonzales NR,
- et al
- Gray WA,
- Yadav JS,
- Verta P,
- Scicli A,
- Fairman R,
- Wholey M,
- et al