Commentary to: Carotid Stenting for Asymptomatic Carotid Stenosis: Call for Judicious Approach
Article information
The CREST-2 (Carotid Revascularization, Endarterectomy Versus Stent Trial-2) [1] attempts to address a decade long debate regarding the role and efficacy of carotid revascularization for asymptomatic carotid stenosis: do patients with asymptomatic high-grade carotid stenosis benefit from revascularization with either carotid stent assisted angioplasty (CAS) or carotid endarterectomy (CEA)? The trial reported that CAS combined with intensive medical therapy reduced the 4-year risk of ipsilateral ischemic stroke compared with medical therapy alone—2.8% versus 6.0%, a relative risk reduction of roughly half [1]. The 4-year risk of ipsilateral stroke of CAS combined with intensive medical therapy was lower than that of CEA combined with intensive medical therapy (2.8 vs. 3.7). An accompanying editorial suggested that CAS would be a reasonable choice for patients who accept the immediate procedural risk or who cannot tolerate optimal medical management. However, the trial was complex, effectively 2 trials in 1, thus the results deserve careful interpretation and judicious application.
STATISTICALLY SIGNIFICANT BUT FRAGILE
The tipping-point analysis reported by the investigators revealed that a shift in outcome for only 3 to 4 patients—out of 1,245 randomized to CAS arm—would have made the primary endpoint nonsignificant [1]. Also, the detailed review of the CREST-2 results showed relatively higher risks of strokes in the stenting group than the medical therapy group (1.3% vs. 0%) in early phase (day 0 to 44). However, after follow-up of 1,686 person-years in the medical therapy arm and 1,714 person-years in the stenting arm, the stenting group’s annual event rate was 0.4% and that of medical therapy group was 1.7%, indicating the overall difference appeared to rely on fewer later ipsilateral strokes in the stenting arm than in the medical-therapy arm. Thus, time-to-event analyses, which can account for unequal follow-up hazards over time and distinguish early procedural risk from later preventive benefit, would be more appropriate statistical methods. The protocol-specified targets of medical management for low-density lipoprotein cholesterol, systolic blood pressure control, and glycated hemoglobin level in patients with diabetes did not meet current guidelines [1]. In addition, patients undergoing stenting received peri- and post-procedural dual anti-platelet therapy, while patients undergoing CEA received only aspirin. Thus, disparity and suboptimal quality in medical management may have compromised the surgical outcomes.
EXTRAORDINARY PERFORMANCE OF TRIAL OPERATORS
One of the most impressive numbers in CREST-2 is the periprocedural stroke and death rate of 1.3% [1]. This figure is exceptional by any current standard. It reflects a rigorous credentialing process of CREST-2 investigators: operators were required to demonstrate procedural proficiency through documented case volumes, independently adjudicated complication rates, and, in some cases, proctored procedures before being permitted to enroll patients [2]. The very low periprocedural event rate achieved from selected operators for highly selected patients is unlikely to reflect real-world practice.
History provides a vivid precedent that this is not a hypothetical concern. Following the publication of NASCET (North American Symptomatic Carotid Endarterectomy Trial) and ACAS (Asymptomatic Carotid Atherosclerosis Study) in the 1990s, the rate of CEA among Medicare beneficiaries approximately doubled, as documented in the Dartmouth Atlas of Health Care [3]. That expansion of CEA was accompanied by periprocedural mortality rates in community settings that substantially exceeded those reported in the trials (0.6% in NASCET and 0.1% in ACAS) [4]. Specifically, the operative mortality varied from 0.6% at high-volume academic centers to 2.5% at low-volume hospitals [4]. Also, even institutions that have participated in those trials showed higher (1.4%) periprocedural mortality [4]. Therefore, the operators who participated in NASCET, ACAS, and now CREST-2 may not necessarily represent the broader procedural operators.
The 2-year interim results of ECST-2 (The European Carotid Surgery Trial 2), which randomized patients with asymptomatic or low-to-intermediate-risk symptomatic carotid stenosis to optimized medical therapy alone versus medical therapy plus revascularization, found no evidence of benefit from revascularization, which reinforced the position that contemporary medical management significantly reduced ipsilateral ischemic stroke risk in carotid stenosis patients.
WHAT THE TRIAL CANNOT TELL US
CREST-2 enrolled patients based on the degree of luminal stenosis (≥70%), yet luminal narrowing alone is an imperfect surrogate for stroke risk. Accumulating evidence suggests that carotid plaque composition, including features of plaque vulnerability, adds prognostic information in addition to luminal stenosis severity and may improve identification of asymptomatic patients at higher risk of ipsilateral stroke. Intraplaque hemorrhage (IPH) is another potent predictor of future cerebrovascular events. For example, IPH on magnetic resonance imaging (MRI) conferred a hazard ratio of approximately 4 to 7 for ipsilateral stroke, independent of the grade of luminal stenosis. In asymptomatic men with moderate carotid stenosis, MRI-evidence of IPH predicted future ipsilateral events with a negative predictive value of 100%. And lipid-rich necrotic core and thin or ruptured fibrous cap carry similarly elevated ipsilateral ischemic stroke risk.
CREST-2 did not systematically capture carotid plaque morphology data. The trial therefore cannot answer the clinically consequential questions: which asymptomatic patients with high-grade stenosis harbor high-risk plaques that might benefit most from revascularization procedures, and which patients carry stable disease where high (although small) peri-procedural risk exceeds the risk with modern medical therapy? In other words, plaque-level risk stratification would be an essential future strategy to identify a subgroup of carotid stenosis patients who might truly benefit from revascularization procedures. However, any single imaging biomarker including IPH, spontaneous microembolization on duplex ultrasound, or impaired cerebrovascular reserve (CVR) should not be regarded as a stand-alone management decision maker for asymptomatic carotid stenosis.
THE CONVERSATION AT THE BEDSIDE
As with other medical decision-making processes, the decision to proceed with carotid revascularization for asymptomatic carotid stenosis cannot be a guideline-driven straightforward process; it must be a shared decision-making process that demands honest communication about uncertainties. Patients must understand that the benefit demonstrated in CREST-2 is real but modest, statistically fragile, and achieved with optimized procedural conditions that may not be replicated in their local hospital. They must understand that the 1.3% periprocedural complication rate is the very minimum, not maximum number. Also, they must recognize that intensive lipid-lowering therapy, blood-pressure optimization, antiplatelet strategies, glycemic control, smoking cessation, and lifestyle modification continue to improve the prognosis of medically managed asymptomatic carotid stenosis. And they must be told that plaque imaging could refine their individual risk estimate and inform whether carotid revascularization is truly warranted. In addition, institutions and individual practitioners should know their own procedural risks and outcomes through continuous, rigorous monitoring and audit. And they should disclose their local outcomes to patients during counseling alongside the CREST-2 results.
NECESSITY OF JUDICIOUS APPLICATION
Many of us who treat carotid disease have already taken conservative management strategy before CREST-2 was published. Intervention has been reserved for patients in whom imaging features suggest plaque vulnerability such as IPH, echolucent or ulcerated plaque, or in whom documented stenosis progression or contralateral occlusion raises the stakes, or in whom optimal medical therapy is genuinely untenable. CREST-2 does not appear to invalidate that approach.
The trial should be celebrated for its rigorous design, its long commitment to operator quality, and its substantial contribution to an evidence base that has been evolving for decades. However, it should not be read as a mandate. As we learned lessons from NASCET and ACAS, trial results can be distorted when applied on non-trial based real-world circumstances.
Notes
Fund
None.
Ethics Statement
This is a commentary to a published paper, thus, Institutional Review Board review was exempted.
Conflict of Interest
SKL has been the editor of the Neurointervention since 2022. No potential conflict of interest relevant to this article was reported.
Author Contribution
Concept and design: SKL, EL, and DL. Analysis and interpretation: SKL. Data collection: SKL. Writing the article: SKL. Critical revision of the article: SKL, EL, and DL. Final approval of the article: SKL, EL, and DL. Statistical analysis: none. Obtained funding: none. Overall responsibility: SKL.