Lessons from ACST-2

Abstract

The recent 130-center, international, second Asymptomatic Carotid Surgery Trial involving 3625 patients found that, regardless of whether a patient underwent stenting or endarterectomy, the periprocedural risk of disabling or fatal stroke was about 1% and the 5-year estimated risk of non-procedural disabling or fatal stroke was 2.5%. With advances in technique, technology, and patient selection, stenting done by appropriately trained and experienced operators can achieve safety and efficacy comparable to endarterectomy for asymptomatic patients. The ongoing Carotid Revascularization and Medical Management for Asymptomatic Carotid Stenosis Trial will clarify whether revascularization, by either stenting or endarterectomy, remains an important therapeutic goal in the setting of modern intensive medical therapy.

The recently reported second Asymptomatic Carotid Surgery trial (ACST-2) provides much needed updated information about the relative efficacy of carotid artery stenting (CAS) versus carotid endarterectomy (CEA).¹ This 130-center, international randomized trial of CAS and CEA enrolled 3625 patients from 2008 to the end of 2020 and followed patients for a mean of 5 years. The periprocedural risk of disabling or fatal stroke from either procedure was about 1%. The 5-year estimated risk of non-procedural disabling or fatal stroke was 2.5% for each treatment group. The 5-year estimates for any non-procedural stroke were 5.3% for CAS and 4.5% for CEA (p=0.33). There was a slight excess of non-disabling strokes after CAS (45 vs 32) and a slight excess of myocardial infarction after CEA (4 vs 13). Investigators concluded that both procedures were well tolerated and comparably effective and that these results of ACST-2 were in line with other asymptomatic carotid revascularization trials, including CREST, ACT-1, and SPACE-2.

The similarity in the stroke outcomes between CAS and CEA in ACST-2 is notable. The absolute differences were low for any peri-procedural stroke (1.2%), any post-procedural stroke (0.7%), peri-procedural disabling or fatal stroke (0.3%) or non-disabling stroke (1.1%), or post-procedural fatal or disabling stroke (0.1%) or non-disabling stroke (0.7%). None of these differences was significant. Prior large, randomized trials comparing CAS and CEA were mainly in symptomatic patients, and all showed a significant advantage for CEA with regard to periprocedural stroke (EVA-3S, SPACE, ICSS, CREST) and stroke overall.

Enrollment and randomization in those preceding randomized trials took place from 2000 to 2008. In ACST-2, randomization took place from 2008 to 2020. For the newer procedure, CAS, the decade-plus time interval coincided with improvements in patient selection,² technology,³ and training,⁴ and coincided with increasing clinical experience among interventionists and referring physicians. For the older procedure, CEA, stroke outcomes in the large, randomized trials comparing CEA to medical treatment also improved over the time. The periprocedural stroke and death rates for CEA for various symptomatic and asymptomatic carotid populations were 4.4% for the Veterans Affairs Symptomatic trial;⁵ 2.4% for the Veterans Affairs asymptomatic trial;⁶ 5.8%⁷ and 6.7%⁸ for the North American Symptomatic Carotid Endarterectomy Trial, and 2.3% for the Asymptomatic Carotid Atherosclerosis Study (ACAS).⁹ Population-derived studies of CEA spanning more than 30 years demonstrated improvements in clinical practice over time as well. For the Cincinnati/Northern Kentucky population the periprocedural stroke and death rate was 9.5% in 1980 and 6.5% for 1983–1984.¹⁰ A CMS-based study showed a stroke and death rate for CEA of 4.4% for 1999/2000 and 3.1% for 2013/2014.¹¹ The characteristics and treatments of the randomized patient cohorts also merit mention with regard to outcome improvements over time. For example, in ACAS enrollment was from 1987 to 1993; the mean systolic blood pressure was 146 mm Hg; 28% of the CEA cohort were cigarette smokers; and the use of lipid lowering drugs was not reported. In ACST-2, 36% of the CEA patients had a blood pressure >140 mm Hg, cigarette smoking was not reported, and 80–90% of patients were reported to be on lipid lowering agents at one-month after treatment.

Better-informed patient selection in ACST-2 likely contributed to the result that the two revascularization procedures were comparable in stroke outcomes. Using an uncertainty principle, patients were enrolled if both the doctor and patient were substantially uncertain about whether to treat with CAS or CEA. By 2008 and beyond, such decision-making was enriched by the increasing experience with CAS in treating the heterogeneous patient cohort presenting with high-grade asymptomatic carotid stenosis. As a specific example, patients were expected to have had some type of angiography, such as catheter, computed tomographic, or magnetic resonance angiography, to assure that they were anatomically suitable for either procedure prior to randomization. This requirement of angiography helped avoid randomization of patients with unfavorable anatomy for CAS, such as type III or bovine aortic arch.¹²

Avoiding procedural atheroembolism has been a focus of many technological advances in carotid revascularization. ACST-2 also incorporated evolving technology. Newer and possibly better stents and embolic protection systems were allowed as enrollment progressed from 2008 to 2020 -- more flexible closed-cell stents, micromesh stents, lower profile embolic protection devices, and flow reversal embolic protection techiques.¹ For intra-procedural embolic protection, the use of protection devices has been rapidly, but not universally, adopted. An interrogation of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database (2011 to 2018) found that one quarter of patients were treated without embolic protection, and this occurred despite a 4-fold increase in perioperative stroke without an embolic protection device.¹³ Atheroembolism can occur during intravascular procedures even before getting to the target lesion. Avoiding such emboli provides the rationale for reversing cerebral blood flow during stenting, which may be more effective than using filter-type devices.¹⁴ Reversal of flow, and possibly avoiding the aorta, as occurs with transcarotid artery revascularization (TCAR), may be more effective than transfemoral carotid stenting with filter devices in avoiding distal embolization.¹⁵ However, TCAR imparts a non-zero risk of post-operative neck hemorrhage.

ACST-2 enrolled 3625 patients, a remarkable achievement, but shy of the 5000 patients that was the investigators’ original intent. Recruitment can lag in randomized trials for numerous reasons, some of the more recognized reasons being overestimation of burden of disease and administrative, regulatory, personnel and funding challenges. Most of the attention given to evidence-based medicine focuses on technical and methodological issues around maximizing scientific validity. However, evidence-based medicine is, in fact, heavily influenced by values.¹⁶ Values drive decisions on which questions to address with clinical trials. Many clinicians treating patients with high-grade atherosclerotic stenosis may not have been convinced that more evidence was needed to conclude that CAS and CEA can achieve comparable periprocedural and postprocedural results. Some took the public stance that prior trials like CREST and others had sufficiently demonstrated the comparability of carotid artery stenting to endarterectomy.¹⁷ As more evidence accumulates, confidence intervals tend to collapse around an average estimate of relative effectiveness and randomizing more patients can feel like an exercise in diminishing returns. Recruitment into a carotid revascularization trial in patients with asymptomatic disease may lag because some physicians may believe that the benefits of revascularization, whether by CAS or CEA, provide diminishing returns as modern approaches to lipid and blood pressure lowering become progressively more effective and better tolerated. Finally, one must appreciate that trials exist in a medical-regulatory-economic ecosystem, which may or may not favor the attainment of the highest level of medical evidence through randomization in trials. A classic example of an ecosystem favorable for a trial was the Netherlands at the time MR CLEAN was being conducted.¹⁸ The regulatory system forced interventionists to prove rather than assume efficacy for recanalization of acute large-vessel occlusion ischemic strokes.¹⁹

As noted above, the outcomes were similar with CEA and CAS in ACST-2. The stroke outcomes were the same or not significantly different.¹ When comparing procedural death plus any stroke over 5-years, an outcome arguably of greater interest to patients, the rates were 8.5% for CAS versus 7.0% for CEA (p=0.09). When it came to procedural death or any fatal or disabling stroke, CAS and CEA were nearly identical (3.3 vs. 3.5%; p=0.86). We can conclude that when the treating team has equipoise about whether to perform CEA or CAS, either procedure will achieve roughly the same outcomes. Assuming that revascularization should be done in the first place, it is becoming increasingly difficult to justify different rules regarding reimbursement for one procedure but not the other for standard-risk patients with asymptomatic carotid stenosis. In the US, the issue of reimbursement for CAS has been a source of contention for years.²⁰

One must consider how patients were managed medically in ACST-2 because there are many proven medical strategies to lower risk of stroke. Participants in ACST-2 were elderly and not unusually healthy: 30% had diabetes mellitus and 36.6% had a systolic blood pressure of >140 mm Hg. At baseline, nearly every patient was taking an antithrombotic medication, 87.3% were taking antihypertensive medication; and 84.6% were taking lipid lowering therapy. We are told that in ACST-2 long-term care was to be similar in both treatment groups. With medical treatment comparable for both groups, comparisons of stroke rates by treatment group remain valid. However, if medical treatments during the trial were not as intensive as they could have been, stroke rates for both treatment groups would be expected to be higher than they could have been. Modern medical therapy is not synonymous with intensive medical therapy. What is done is not necessarily what ought to be done. Of patients enrolled into CREST-2 taking at least one antihypertensive medication at baseline, the prevalence of guideline-adherent antihypertensive regimens was only 34%.²¹ Protocolized, centrally managed intensive medical care can drive down vascular risk factors. This approach is being used in CREST-2, which has systolic blood pressure <130 mm Hg and LDL cholesterol <70 mg/dl as the primary risk factor goals.²² Early indications are that protocolized and centralized management has favorably driven down vascular risk factors in the trial.²³

The commonly voiced recent concern regarding revascularization of patients with asymptomatic carotid stenosis is that the absolute risk of stroke is low with proper medical treatment without revascularization. The relative risk reduction achieved with revascularization would be more meaningful to patients and populations if the therapy were focused on high-risk individuals. Some have argued that revascularization should target patients with the highest degree of stenosis. A recent population-based study suggests that patients with 80 to 99% stenosis had significantly greater 5-year risk of ipsilateral stroke than did patients with 50 to 79% stenosis (18.3% vs. 1.0%).²⁴ Interestingly, randomized trials comparing revascularization to medical therapy have not shown degree of stenosis to modify the effect of revascularization on stroke outcomes. The reasons for this are unclear, but it may be that the highest risk patients are being steered away from enrollment in trials that include a group that does not get carotid revascularization. ACST-2 did not find substantial modification of efficacy by treatment group by degree of stenosis.

Some have argued that revascularization of asymptomatic carotid stenosis should be reserved for patients with high-risk plaque. The 2017 European Society for Vascular Surgery Guidelines advise revascularization of asymptomatic carotid stenosis patients in the presence of clinical and/or imaging characteristics besides degree of stenosis associated with an increased risk of late ipsilateral stroke.²⁵ Carotid plaque has been classified as unstable in several ways, including detection of: intraplaque hemorrhage, lipid-rich necrotic core, plaque ulceration, microemboli with transcranial Doppler, and plaque echolucency. About one quarter of patients with asymptomatic carotid stenosis will have plaque that can be classified as high-risk, and the presence of high-risk plaque increases risk of ipsilateral stroke by about threefold.²⁶ ACST-2 did not find plaque echolucency on ultrasound to substantially modify differential efficacy of CAS versus CEA. Echolucency might still be helpful in deciding which patients benefit the most from revascularization versus medical therapy alone.

ACST-2 considered myocardial infarction among its periprocedural outcomes. Though uncommon, periprocedural MI occurred in 0.2% of patients who had CAS first and 0.7% of patients who had CEA first. Investigators state that there was no evidence that myocardial infarction had been underestimated, however no tests for silent myocardial infarction were required. In CREST, which included patients with symptomatic carotid stenosis, myocardial infarction was significantly more common following CEA than following CAS.²⁷ Biomarker-only myocardial infarctions could not be counted in ACST-2, but CREST found that even these milder periprocedural myocardial infarcts confer increased mortality for 10 years.²⁸ Future trials comparing carotid interventions should include post-procedural cardiac biomarkers to detect systematically biomarker-only myocardial infarction as well as improve unbiased detection of symptomatic myocardial infarction. Blood biomarkers could also be used to improve detection of post-procedural neuroaxonal injury.²⁹ Such biomarkers can be measured uniformly in all sorts of clinical environments and can easily be assessed in a blinded fashion.

Before ACST-2, ACST demonstrated that CEA of asymptomatic carotid stenosis reduced stroke mainly by reducing ipsilateral carotid stroke, but CEA also significantly reduced contralateral carotid stroke.³⁰ The implication was that CEA can reduce stroke risk both by preventing artery-to-artery embolism and by improving blood flow. ACST-2 focused on prevention of all strokes rather than just post-procedural ipsilateral stroke. Clearly carotid plaque can cause stroke through embolism, hence the added risk of stroke in the setting of intraplaque hemorrhage or when microemboli are detected downstream. However, we should not dismiss the clinical significance of reduced cerebral blood flow without embolism. In a study of a subset of patients in the CREST-2 trial at baseline, carotid stenosis was associated with cognitive impairment, and cognitive scores for left and right carotid disease were similar.³¹ What remains to be seen is whether correction of hemispheric hypoperfusion with revascularization leads to improved cognitive trajectories. The ongoing CREST-H study is addressing this very point.³²

In summary, ACST-2 provides the highest level of evidence to date that, for asymptomatic patients who are good candidates for either procedure, CAS and CEA are comparable with regard to disabling and fatal events. Both procedures can be done safely, and post-procedural risk of stroke is low with either procedure. Newer procedures like TCAR have safety profiles comparable to, or perhaps slightly better than, CEA or CAS in selected patients, though this has never been tested in a proper randomized trial. It is unlikely that TCAR will lower the risk of post-procedural fatal or disabling stroke any more than CEA or CAS given the findings of ACST-2, where two very different approaches to revascularization generated very similar results. A crucial remaining question is whether any revascularization procedure is better than intensive medical management, particularly given the lower stroke rates being achieved without revascularization over the past 15 years.³³ The ongoing CREST-2 trial is addressing the efficacy of CAS relative to intensive medical therapy in patients who are good candidates for CAS and in a separate parallel trial is addressing the efficacy of CEA relative to intensive medical therapy in patients who are good candidates for CEA. ACST-2 has demonstrated that it remains appropriate that CREST-2 is testing both CEA and CAS against intensive medical therapy, as excellent results can be achieved with either technique.

Non-standard Abbreviations and Acronyms

ACAS

Asymptomatic Carotid Atherosclerosis Study

ACT-1

Carotid Angioplasty and Stenting versus Endarterectomy in Asymptomatic Subjects Who Are at Standard Risk for Carotid Endarterectomy with Significant Extracranial Carotid Stenotic Disease

ACST-2

Asymptomatic Carotid Surgery Trial-2

CAS

Carotid artery stenting

CEA

Carotid endarterectomy

CMS

Centers for Medicare and Medicaid Services

CREST

Carotid Revascularization Endarterectomy versus Stenting Trial

CREST-2

Carotid Revascularization and Medical Management for Asymptomatic Carotid Stenosis Trial

CREST-H

Carotid Revascularization and Medical Management for Asymptomatic Carotid Stenosis Trial - Hemodynamics

EVA-3S

Endarterectomy Versus Angioplasty in patients with Symptomatic Severe carotid Stenosis trial

ICSS

International Carotid Stenting Study

MR CLEAN

Multicenter Randomized Clinical trial of Endovascular treatment for Acute ischemic stroke in the Netherlands trial

NSQIP

National Surgical Quality Improvement Program

SPACE

Stent-Supported Percutaneous Angioplasty of the Carotid Artery versus Endarterectomy

SPACE-2

Stent-Supported Percutaneous Angioplasty of the Carotid Artery versus Endarterectomy-2

TCAR

Transcarotid Artery Revascularization