Comparison of long-term results of carotid endarterectomy for asymptomatic carotid artery stenosis

Between 1993 and 2003, a total of 3120 patients with asymptomatic >60% extracranial internal carotid artery stenosis were included in a surgical and a conservative arm of the multicenter randomized controlled ACST-1 trial [8]. The percentage of asymptomatic patients undergoing surgery 5 years after randomization was 92.1% in the surgical arm and 16.5% in the conservative arm, and 92.2% vs. 23.5%, respectively, after 10 years (intention-to-treat analysis). There was no significant difference in the percentage of patients under the best possible adjuvant drug therapy either at the time of randomization (antihypertensive drugs: 51% vs. 55%, platelet aggregation inhibitors: 91% vs. 88%, anticoagulants: 5% vs. 6%, statins: 11% vs. 7%) or in long-term 10-year follow-up (antihypertensive drugs: 87% vs. 89%, platelet aggregation inhibitors: 88% vs. 89%, anticoagulants: 11% vs. 11%, statins: 80% vs. 82%). Excluding periprocedural events and non-stroke-related mortality, the risk for any stroke was 4.1% in the surgical arm and 10.0% in the conservative arm at 5‑year follow-up of patients aged <75 years (absolute risk reduction, ARR 5.9%) and 10.8% vs. 16.9% at 10 years (ARR 6.1%) (Fig. 1). At a ratio of 0.54 (95% confidence interval CI 0.43–0.68, P < 0.0001) of non-periprocedural stroke rates in the surgical vs. conservative arm, CEA showed a 46% reduction in the incidence of all strokes in the long-term. The proportional reduction in disabling or fatal stroke was similar to the proportional reduction in all strokes. Of the strokes for which the affected hemisphere was known, the greatest ARR was observed on the ipsilateral side (38 vs. 92 events, stroke rate ratio 0.43 (95% CI 0.28–0.68); P < 0.0001). Subgroup analysis of those patients with and without statins revealed a significant effectiveness of CEA in stroke prevention both at 5 years (ARR 3.4%, 95% CI 1.5–5.2; P = 0.0005 and ARR 10.8%, 95% CI 6.6–15.1; P < 0.0001, respectively) and at 10 years (ARR 5.8%, 95% CI 2.1–9.6; P = 0.002 and ARR 6.2%, 95% CI 0.4–12.8; P = 0.07, respectively). It can be concluded from the results of the ACST study that low complication CEA (<3% perioperative stroke rate) promotes the positive effect of drug therapy in long-term stroke prevention; however, due to the evolving developments in drug therapy, these data can only be partially extrapolated to the present time.

Of the altogether nine RCTs [9‐17] published to date comparing CAS vs. CEA, four [11, 12, 15, 16] included patients with symptomatic and asymptomatic carotid stenosis, while one [17] included asymptomatic patients only.

The SAPPHIRE study [11], which primarily included high-risk surgical patients, was the only multicenter study in which the majority of randomized patients were asymptomatic. With the exception of the CAVATAS study [12], stents were used in the endovascular group in all RCTs. The CAVATAS study [12] also used only percutaneous transluminal angioplasty (PTA) in the period prior to 1994 and after 1994, either stenting or PTA was favored at the discretion of the interventionist. With >2500 patients, the CREST study [15] published in 2010 is the largest RCT yet to compare both treatment methods in medium-risk patients. Brooks et al. [18] (Lexington I) enrolled only symptomatic patients in the first part of their study and only asymptomatic patients in the second part [19] (Lexington II). The two study populations were merged for the purposes of evaluating long-term results in a further analysis [16]. No protection devices were used either in the CAVATAS [12] study or by Brooks et al. [16, 18, 19]. The median follow-up time of these studies was between 2 and 10 years. Details of study characteristics and results relating to asymptomatic patient populations can be found in Table 1.

The individual studies, including their long-term results, are presented in chronological order by year of publication and critically examined.

Of 334 randomized patients, n = 260 patients (77.8%) underwent long-term follow-up (completed 3‑years follow-up) [20]. At n = 117, the proportion of asymptomatic (degree of stenosis >80%) patients in the CAS group was comparable to that in the CEA group (n = 120). There was no significant difference in the primary combined endpoint defined as death, myocardial infarction, or any stroke within the first 30 days, or death or ipsilateral stroke between 31 and 1080 days (Table 1). In their discussion, the authors emphasize that due to the small number of cases, it is not possible to draw conclusions on the investigation of non-inferiority of the CAS method in asymptomatic carotid stenosis. The markedly lower life expectancy of their patient groups compared with other RCTs (survival at 3 years: CAS 80% vs. CEA 75.8%) reflects the high-risk profile of the study participants. Study participants needed to meet at least one of the following criteria in order to be included as a high-risk patient in the study:

Thus, no significant statement could be made about the representative general population of patients with asymptomatic carotid stenosis, in particular due to the selection bias.

The CAVATAS study [21] randomized 505 patients, 90% of which had been symptomatic in the preceding 6 months, into a stent study arm or a surgical study arm. The main criticism of this study is that 75% of all patients in the interventional group were treated without protection devices or stents, meaning that the data cannot be extrapolated to current daily practice. There are also no published data on the small number of asymptomatic patients (Table 1).

In May 2010, the CREST investigators reported that the primary combined endpoint of periprocedural stroke, myocardial infarction, or death, or ipsilateral stroke within the first 4 years following randomization did not differ in the overall evaluation of all study participants when comparing CAS (n = 1262) vs. CEA (n = 1240) (7.2% vs. 6.8%; hazard ratio (HR) with CAS: 1.11; 95% CI 0.81–1.51; P = 0.51) [15]. At n = 1190, the number of asymptomatic patients was higher than in all other RCTs. A separate analysis at 4‑year follow-up showed no significant difference for the asymptomatic patient group in all endpoints when comparing both treatment methods, even though the difference in favor of CEA in the endpoint all periprocedural strokes or postprocedural ipsilateral stroke was close to the significance threshold (Table 2).

Paraskevas et al. [23] published an article in 2013 criticizing the methodology used in CREST. The authors cite a 2004 article [24] of Hobson et al. showing that even within less than 3.5 years from the start of the study (December 2000), two thirds of the CAS population (789 out of 1262 patients, 62.5%) were included in the study as symptomatic patients. Only after 2005, once the recruitment of symptomatic patients had been suspended, were asymptomatic patients added. As such, the actual percentage of asymptomatic patients in the CAS group should not exceed 37.5%. The 47.1% reported is thus questionable. A further point of criticism was the definition of myocardial infarction as the primary study endpoint, which was put on a level with the events stroke and death. The comparatively high rate of myocardial ischemia (2.3%) in the surgical arm is likely due to the routine determination of heart enzymes 6–8 h following surgery and the laboratory value-based definition of infarction (creatine kinase-MB or troponin-T levels elevated to at least twice the upper limit and electrocardiogram changes or symptoms consistent with this).

Despite these criticisms, the significant difference in the overall evaluation of the periprocedural myocardial infarction rate in favor of the CAS procedure could not be confirmed in the analysis of only asymptomatic patients (Table 2). Much like the 4‑year results, there was no significant difference at 10-year follow-up in patients with asymptomatic carotid stenosis in terms of the primary combined endpoint ([22], Table 1; Fig. 2).

Between 1998 and 2002 Brooks et al. [19] randomized 85 high and medium-risk patients with asymptomatic carotid stenosis. All patients in the CAS group were treated without protection devices but with stents, in contrast to the CAVATAS study. No cerebral events were observed in either group both during the periprocedural period and at 4‑year follow-up (Table 1). Long-term results in the combined endpoint ipsilateral stroke, fatal and non-lethal myocardial infarction, including the first 30 postprocedural days showed cardiac events to be significantly more frequent in the CEA group (Table 1), while rates of ipsilateral stroke did not differ (p > 0.05). The authors saw a correlation between the significantly more frequent cardiac events and elevated cardiac enzymes in the periprocedural period in the CREST study CEA group, and therefore suggested a negative predictor that possibly manifests in increased cardiac morbidity and mortality in the long term; however, this interpretation is purely hypothetical, particularly since the authors of the Lexington II study did not determine or analyze any heart enzymes.

The ACT-1 study [17] is the only multicenter controlled study to date that has investigated the value of CAS vs. CEA in the treatment of only asymptomatic patients at low to moderate risk (<79 years of age). In total 1453 patients were included with remarkably disparate group sizes (CAS n = 1089; CEA n = 364). Similar to the CREST study, no differences were seen in terms of the primary combined endpoint death, any stroke, or myocardial infarction within the first 30 days, or ipsilateral stroke at 1 year, as shown by the almost identical Kaplan-Meier curves in Fig. 2. Neither group showed any difference in the stroke rate during the periprocedural period (Table 1) or in long-term follow-up (Fig. 3). In line with the separate analysis of asymptomatic patients in the CREST study, the ACT-1 study found no difference in periprocedural myocardial infarction rates between CEA and CAS in (0.5% vs. 0.9%; P = 0.41).