Integrating cognitive testing as an outcome in carotid revascularization trials

We read with considerable interest the systematic review by Dr. Kwok and colleagues regarding the effects of revascularization of extracranial carotid artery disease by carotid endarterectomy or carotid artery stenting on cognition¹. Their systematic review revealed only seven studies published from January 1985 to December 2022, of which none were randomized clinical trials. This is notable because, as the authors point out, cognitive testing has developed and improved in recent decades, in part driven by pharmaceutical industry funded clinical trials. In addition, the incorporation of cognitive testing is becoming more routine in clinical trials of new devices, particularly cardiovascular and cerebrovascular devices like those used in transcatheter aortic valve replacement.

The authors also point out the pitfalls of control groups used in existing studies that are composed of either patients with carotid stenosis who refused revascularization or who had no carotid stenosis. Patients without carotid stenosis are inappropriate controls to test the efficacy of treating carotid stenosis. Using those who do not consent to the procedure as controls is equally problematic. There can be bias by surgical indication. Patients that decline procedures will differ systematically and substantively from those that accept. The quoted risks of the procedure should and would be expected to be higher for patients with a co-morbidity burden, and these risk estimates will act as a deterrent to consenting to a procedure. We believe that a reliable and valid estimate of the efficacy of carotid revascularization on cognitive outcomes will come only from randomized trials. Previously, the major randomized trials of carotid revascularization have focused on stroke and death but have generally ignored cognitive outcomes². An exception to this is the ongoing multicenter, multinational randomized clinical trial known as CREST-2 (clinicaltrials.gov ID NCT02089217), which compares intensive medical management (IMM) with or without revascularization for patients with asymptomatic high-grade (≥70% diameter-reducing) carotid stenosis³.

As the review by Kwok et al shows, prior studies have suffered from small sample size. The mean size of the studies in their review was only 94 individuals. We believe that this number is too small to detect a signal of efficacy because of the confounders of participant by age, education, sex, race/ethnicity, depression, other medical co-morbidities, and degree of carotid stenosis. CREST-2 is ongoing and is assessing differences in outcomes of carotid endarterectomy and stenting compared to IMM alone. As of October 27, 2023, 96% of the CREST-2 recruitment goal of 2480 patients has been achieved. Kwok et al point out that prior studies include mixed populations of patients with symptomatic and asymptomatic stenosis. CREST-2 focuses on the asymptomatic population. To reduce variability in testing across clinical sites, CREST-2 utilizes a central core that administers a standardized cognitive battery by telephone.

The Kwok review focused on carotid revascularization studies that assessed both cognitive outcomes and cerebral blood flow. Cerebral blood flow information provides mechanistic insight into why revascularization might improve cognition. However, we believe that it is first important to establish reliably whether cognition is improved with mechanistic testing being complementary. The Hemmington et al study, included in the review, found that cognitive improvement could not be explained by changes in cerebral blood flow⁴. Carotid revascularization may improve cognitive outcomes by improving blood flow (hemodynamic hypothesis), but it could also improve cognition by reducing the risk of silent or subclinical hemispheric infarcts (embolic hypothesis). The two hypotheses are not mutually exclusive.

The more elaborately patients are assessed, either for baseline characteristics or outcomes, the less feasible it is to accomplish a large multicenter study. Large, simple trials illustrate this point. The sample size required to achieve a sufficient number of clinical outcomes, particularly infrequent outcomes in prevention studies like CREST-2 that seek to drive down risk of stroke and death in intensively managed patients with asymptomatic stenosis, prohibits testing blood flow on every subject. To that end, we have created a separate nested mechanistic study known as CREST-H (ClinicalTrials.gov ID: NCT03121209) that will test the hypoperfusion hypothesis using brain CT and MR perfusion scans⁵. This approach allows for a smaller sample size to achieve efficacy. We are hopeful to secure support to test an embolic hypothesis of cognitive impairment in the future. Patients who have CT angiography at baseline for assessment of eligibility or for clinical planning of a potential carotid intervention would have the images evaluated for stable or unstable plaque characteristics. This would allow testing of whether plaque stability status interacts with interventional groups and cognitive outcomes.

A major finding in the review by Krok et al is that heterogeneity in patient groups, cognitive test batteries, and blood flow study methods, precluded a formal meta-analysis. Including cognitive outcomes among the measures used in carotid interventional trials will help provide the needed consistency in methods and scientific rigor and reproducibility. Having greater harmonization of cognitive outcomes across studies would also be welcome.