Functional MRI evaluation of cognitive effects of carotid stenosis revascularization

Abstract Introduction Severe internal carotid stenosis, if left untreated, can pose serious risks for ischemic stroke and cognitive impairments. The effects of revascularization on any aspects of cognition, however, are not well understood, as conflicting results are reported, which have mainly been centered on paper‐based cognitive analyses. Here, we summarized and evaluated the publications to date of functional MRI (fMRI) studies that examined the mechanisms of functional brain activation and connectivity as a way to reflect cognitive effects of revascularization on patients with carotid stenosis. Methods A PubMed and Google Scholar (covering the relevant literature until November 1, 2021) search yielded eight original studies of the research line, including seven resting‐state and one task‐based fMRI reports. Results Findings demonstrated treatment‐related alterations in fMRI signal intensity and symmetry level, regional fMRI activation pattern, and functional brain network connectivity. The functional brain changes were associated largely with improvement in cognitive function assessed using standard cognitive test scores. Conclusions These findings support the contribution of fMRI to the understanding of brain functional activation and connectivity changes revealing cognitive effects of revascularization in the management of severe carotid stenosis. The review also highlighted the importance of reproducibility through enhancing experimental designs and cognitive task applications with future research for potential clinical translation.

. Severe stenosis can either be symptomatic where it is characterized by the presence of transient or permanent neurologic or ischemia-like symptoms or asymptomatic where no obvious neurological dysfunction can be detected on the physical examination by a neurologist and the patient reports no stenosis attributable complaints (Lanzino et al., 2009). Common clinical treatments for severe carotid stenosis include medical therapy and carotid endarterectomy (CEA), for which surgery is performed to remove the plaque. Alternatively, carotid angioplasty and stenting (CAS) is used, in which a mesh-like stent is inserted into the affected artery to improve blood flow (Lanzino et al., 2009). The choice of treatment is dependent on factors including the severity of stenosis, age, and the presence of comorbidities (Lanzino et al., 2009). While CEA is generally considered as the preferred treatment for symptomatic patients, CAS is typically used for patients who are deemed high risk for invasive surgical procedures such as older and frail patients, those with poor neck structure conditions preventing surgical access to carotid arteries or those with a contralateral artery obstruction (Lanzino et al., 2009). Revascularization treatments for carotid stenosis remarkably reduce the risks of adverse ischemic events (Lanzino et al., 2009;Rothwell et al., 2007).
Depending on the most impacted blood vessels, severe carotid stenosis if not treated can result in significant cognitive impairments, notably in executive functions and working memory (Berman et al., 2007;De Rango et al., 2008). The cognitive impact of stenosis treatment is subject of several reviews, but so far inconclusive findings of revascularization have been reported (Berman et al., 2007;De Rango et al., 2008;Ghogawala et al., 2008;Lal, 2007;Plessers et al., 2014;Sztriha et al., 2009). Berman et al. (2007) found the effects of revascularization inconclusive: with 36% of the studies showing cognitive improvements postrevascularization, whereas the remaining reports seeing mixed (50%) or negative (14%) results. Similarly, Ghogawala et al. (2008) observed posttreatment improvements in 29% of the studies, mainly concerning verbal memory and attention, whereas the other studies reported either a decline (41%) or no change in cognition (29%).
In addition, Plessers et al. (2014) identified approximately 10% of patients in several studies showing cognitive improvements post-CEA, whereas about 10−15% of patients experienced a cognitive decline, highlighting the ambiguity of the cognitive effects of revascularization.
Mixed results shown in these studies may be attributed to several factors, such as cognitive lateralization of neurocognitive effects, heterogeneity of patients' presentation of symptoms, variability of neuropsychological testing methodologies, statistical analyses, timing of the assessments, and inconsistencies in study design including the lack of control groups (Berman et al., 2007;Ghogawala et al., 2008;Kolb et al., 2019). Most notably, the majority of previous research has solely used paper-based cognitive tests to examine cognitive impact. Neuroimaging methods on the other hand can more sensitively detect functional brain changes arising from revascularization, while also helping prognostically, to identify patients with the most risks of cognitive decline and dementia (Zhang et al., 2020).
In recent years, functional magnetic resonance imaging (fMRI) has begun to be used in characterizing functional brain changes following revascularization. The fMRI method studies functional brain changes based on its ability to detect brain functional activation utilizing the dynamic paramagnetic properties of capillary blood circulation (perfusion-weighted imaging) or during oxygen exchange (Blood-Oxygen-Level-Dependent or BOLD imaging) (Ogawa et al., 1990).
Detection of brain fMRI activation changes can be coupled with specific fMRI tasks (task-phase fMRI), where the patterns of fluctuation of the BOLD signals are observed in response to the task. Meanwhile, fMRI activation can also be observed during "task-free" resting-state recordings, where there is no implicit cognitive input/output (restingstate fMRI). With resting-state fMRI, the changes in the pattern of functional connectivity, that is, the temporal correlation of spontaneous BOLD activations among spatially distributed brain regions at "rest," can be monitored over time (Smitha et al., 2017). Functional connectivity is usually studied in clusters of neural cells (called networks) responsible for various brain activities. Analyses typically include the salience, default mode, and sensorimotor networks involved in regulating behavior and brain functions, enabling the resting phase and controlling for sensory and motor activities (Smitha et al., 2017).
Although fMRI activation in the brain and cognitive performance are not equal, fMRI can provide a view of how the brain works in response to cognitively demanding tasks, providing a measure that can be used to link with cognitive testing scores and allowing an inference about one's cognitive abilities. For instance, in a well-designed task-phase fMRI study, the difference in brain activation changes between task runs and baseline can provide insights into different cognitive states as well as their functional localization in specific brain regions, which can be corroborated when combined with task performance parameters such as accuracy and reaction time. Similarly, resting-state fMRI studies can be used to characterize normal and abnormal brain functional connectivity in clinical conditions, which can be indicative of cognitive decline (Lv et al., 2018).
FMRI has been widely used to study cognitive changes and treatment effects across disease conditions (Guo et al., 2018). However, there is limited knowledge about using fMRI in accessing cognitive effects of clinical revascularization in treating carotid stenosis. Here, we conduct a review study to identify and map the available fMRI evidence towards understanding the cognitive benefits of carotid stenosis revascularization. To the best of our knowledge, this is the first attempt to summarize fMRI research findings on the cognitive impacts of revascularization in patients with severe carotid stenosis.

Search terms
We searched the current literature (from January 1, 1990-the initiation of the fMRI technology, until November 1, 2021) using the MEDLINE databases, the National Library of Medicine's premier bibliographic resource. This resource contains 27 million+ references of 5200 journals in life sciences chiefly in biomedicine. We cross-checked the search results with Google Scholar to ensure adequate inclusion of studies on the topic.

Inclusion/exclusion criteria
The search was streamlined by including original research on humans in the English language, producing a subset of 1143 articles (shown in Figure 1). The article titles were examined to exclude review, commentary, protocol, and opinion papers. Any studies that did not use fMRI were also excluded, narrowing to 36 papers.
These papers were screened via full-text analysis to meet the crite-
Sample sizes varied among the studies and ranged from a twopatient case report to a study of 25 patient participants, with a mean sample size of 15.4 ± 7.9 (median = 18.5). Only two studies included healthy controls (Cheng et al., 2012;Huang et al., 2018). The mean age of patients across the studies was 71.2 ± 7.7 years (median = 71.4), whereas the sex ratio of the research patients was male predominated, ranging from 70 to 100% (mean = 82.0 ± 12.5 %, median = 79.5%).
Various brain regions and cognitive domains were examined in the studies under review. The task-phase study involved the administration of a higher-level working memory task and examined the prefrontal cortex in a small sample (Chinda et al., 2021). Most of the resting-phase studies examined the default mode network components; although several studies also examined brain regions including the dorsal attention, somatosensory, and salience networks ( also used by some studies (Table 1B).

3.2
Resting-state fMRI studies Cheng et al. (2012) compared pre-CAS and 3 months post-CAS fMRI data of unilateral asymptomatic carotid stenosis patients with age-and education level-matched healthy controls. The authors observed that post-CAS, functional connectivity increased between brain regions ipsilateral to the treated internal carotid artery (Smitha et al., 2017).
These included areas in the default mode and frontoparietal networks including the hippocampus, cingulate cortex, and medial prefrontal cortex. These brain changes were correlated with improvements in dizziness and MMSE scores. The authors suggested that fMRI showed patterns of brain network disruptions arising from stenosis, and these patterns were consequently improved postintervention (Tables 1A   and 1B). Symbol Test (Chen et al., 2020;Larner, 2018;Nasreddine et al., 2005), and verbal memory and immediate recall tests. The authors concluded that successful CAS resulted in increased brain perfusion and connectivity, and thus, improved cognition (Tables 1A and 1B). to the neural challenges caused by decreased blood flow, some of which could be changed through CAS (Tables 1A and 1B).     (Tables 1A and 1B). Chinda et al. (2021) reported the initial and thus far the only fMRI findings with the use of a cognitive task in two carotid stenosis patients who underwent clinical CAS interventions. The study utilized a delayed match-to-sample working memory task with two difficulty levels and cognitive testing using the central nervous system (CNS) vital signs (Gualtieri & Johnson, 2006). Postrevascularization, there was increased fMRI activation in the treated frontal and temporal lobes, which was associated with improvements in accuracy and task completion rates and decreased activation in the contralateral (untreated)

Task-based fMRI studies
hemisphere. The degree of cognitive improvement was related to the degree of flow limitation of the stenosis and the CNS vital signs scores (Tables 1A and 1B).

DISCUSSION
In this article, we summarized the fMRI findings to date that investi-  Tani et al., 2018;Wang et al., 2017). Task-phase fMRI identified increased strength/level of brain activation and increased hemispherical symmetry in the prefrontal cortex in response to higher-level cognitive stimuli (Chinda et al., 2021). Such effect was seen in the short term (i.e., within 3 months posttreatment) (Cheng et al., 2012;Lin et al., 2016;Porcu et al., 2019;Wang et al., 2017). In the long-term posttreatment (i.e., 1-year follow-up), increased regional neural activity in areas including the precentral gyrus, middle frontal, and the anterior cingulate gyri (Porcu et al., 2021) as well as increased functional connectivity symmetry in the sensorimotor and salience networks (Huang et al., 2018) were also reported. Importantly, such changes in fMRI activation and network connectivity were mirrored by improvements in cognitive performance using paper-based standard cognitive tests in most studies (Table 2). These findings confirm the recognized sensitivity of fMRI in detecting hemodynamic response, thereby providing a way to view the brain at work in handling challenges, as an effective neuroimaging modality for studying brain function.
Only one study investigated the influence of symptomatic events such as acute stroke, amaurosis fugax, or transient ischemic attack on the cognitive improvement seen postrevascularization. In the case study using task-phase fMRI, Chinda et al. (2021) showed the importance of the degree of stenosis prior to the revascularization in determining the prognosis, with a greater post-CAS increase in BOLD activations in the treated frontal and temporal lobes, in conjunction with improvements in accuracy and task completion rates in an asymptomatic patient with more severe stenosis (>95%) compared with a symptomatic patient with less severe stenosis (only 70%). Previous research has also suggested that symptomatic patients including those with very severe stenosis can experience a greater level of compromise in their cerebral hemodynamics preintervention (Schaaf et al., 2010).
How disease history, expression, and symptom severity individually and collectively affect the cognitive recovery following revascularization interventions warrant further research with an increased sample size.
Although all the studies under review suggested posttreatment improvements in brain fMRI activation, it is important to note that considerable heterogeneity exists among these studies in terms of study purpose and design. For example, the domains of cognition examined ranged from higher-level working memory capacities in the task-phase studies (Chinda et al., 2021), to visual and verbal memory in the resting phase studies (Table 2). Given the currently thin but growing literature of the research field, all the relevant original fMRI studies that investigated brain functional changes suggestive for cognitive recovery by comparing prerevascularization and postrevascularization have been included in this review paper, regardless of the exact cognitive domain they examined. As a result, a more general statement across all cognitive aspects is unavailable from the studies, although they helped provide data demonstrating the impact of revascularization treatment on cognition, highlighting the advantage of this clinical intervention procedure beyond stroke prevention to better benefit patient care.
The review study also informs future research efforts in this field to investigate brain functional responses with careful selection and implementation of tasks in testing targeted cognitive domains.
TA B L E 2 Main findings of the fMRI studies to date that investigated cognitive benefits of clinical carotid endarterectomy, carotid angioplasty, and stenting treatments

Main findings Studies (%)
Functional Connectivity A general increased functional connectivity (FC) to brain regions involved in attention, executive function, and working memory was observed posttreatment; that is, the frontoparietal and the default mode networks and the cingulate cortices.

Cognitive Correlations
The increased fMRI connectivity/activation in postcarotid angioplasty and stenting (CAS) and carotid endarterectomy (CEA) patients were correlated with improvements in global cognitive score, for example, MMSE. 7 (87.5) fMRI Activation Functional activations increased in both CAS and CEA patients, higher amongst those with a less-severe symptomatic stenosis (≤80%).

(50) Symmetry
Blood-oxygen-level-dependent (BOLD) activations became more symmetrical among hemispheres post-CAS or CEA, indicating functional recovery on the hemisphere ipsilateral to the stenosis.

(12.5)
The well-needed task-phase studies can be used to directly view the response of the brain to explicit stimulations when it is at work. fMRI tasks designed on higher-level functions, such as executive functions and working memory, are appropriate choices in studying cognitive impairment and should be adapted for understanding cognitive recovery in carotid stenosis (Chinda et al., 2021;McDonald et al., 2018).
Behavioral data on completing the task such as reaction time and accuracy can also be collected and used in augmenting the fMRI results.
Another source of heterogeneity may revolve around the fMRI processing and analysis techniques utilized (Table 1B). Even though standard fMRI processing and analysis procedures were largely followed by different studies under review, subtle differences in preprocessing such as the size of the smoothing kernel (6 vs. 8 mm) may be related to the subsequent data presentation. Differences were also seen in the cluster size and model of analyses; for example, restingstate studies used independent component analysis, amplitude of lowfrequency fluctuation analysis, seed-based correlation analysis, or ROI voxel-wise analyses. Although each of the methods has advantages (Lv et al., 2018), their differences can make finding generalizations across the studies difficult.
One general criticism of the translational potential of fMRI findings is the dependence of the BOLD signal on efficient capillary gas exchange and respiration and thus can represent an indirect measure of brain activity due to the lag time among other attributing factors (Ogawa et al., 1990). Even so, fMRI offers a unique way to innovatively "see" what happens in the brain when it is at work with sensitivity, relatively high spatial and temporal resolutions, and lack of requirement for iodinated or radioactive substances. The long debate about fMRI, especially BOLD, can only be resolved with its continued development in applications research.
Other than the inherent limitations related to fMRI technology in general, certain limitations with the current studies may be addressed through improving experimental design and methodology. Even though resting-state fMRI can help understand the general status of functional brain networks through connectivity analysis, task-based fMRI is required for the identification of the patterns of activation and func- The interpretation of the fMRI results under review deserves some caution. Since BOLD fMRI depends on efficient gas exchange, the signal might be even impacted particularly in patients with widespread cerebrovascular disease such as carotid stenosis. Also, patients operated for symptomatic carotid stenosis may have experienced a cerebrovascular event that acutely impacted cognition and brain perfusion. Cognitive recovery in such participants may be experienced independently of treatment. Even so, many of the studies investigated asymptomatic patients (Cheng et al., 2012;Chinda et al., 2021;Huang et al., 2018;Lin et al., 2016;Porcu et al., 2019;Wang et al., 2017) and the improvements in cognition observed are likely reflecting the actual treatment effect as opposed to natural recovery following an ischemic event.
Our review study has some limitations. First, the search was MED-LINE databases based, Google Scholar supplemented, and targeted publications in English. Although the coverage was wide and highquality journals were included, work reported in non-peer-reviewed publications, conference papers, and in other languages was excluded.
A few recent fMRI studies were excluded as they did not include a treatment Goode et al., 2009;He et al., 2020;Liu et al., 2020;Porcu et al., 2020;Xiao et al., 2018), conduct both precognitive and postcognitive tests Muscas et al., 2019;Xiao et al., 2018), or examine the internal carotid arteries (Rosen et al., 2018). Similarly, a couple of task-phase fMRI studies that examined only motor recovery using a motor task without additional pre/postrevascularization cognitive evaluation were excluded (Jensen et al., 2008;Schaaf et al., 2010). Given the relatively small number of available data, the review study is not meant to be a meta-analysis. The selection of studies was based on the fulfillment of the inclusion criteria and not necessarily the quality of the original work, the specific revascularization procedures, or fMRI experimental and analysis methods. Variability existed in the reviewed studies in terms of sample size, participant condition, follow-up duration, and rate, impeding the generalizability of findings. The review also did not intend to target specific cognitive domains or brain regions involved and encompassed diverse measures.
Even with these limitations, this review can make a useful contribution to the literature. To the best of our knowledge, this is the first effort to identify, categorize, and summarize research on the cognitive impacts of revascularization using fMRI in patients with severe carotid stenosis. The review, albeit on a relatively small number of available studies, showed that cognitive benefits beyond stroke prevention may be actualized through revascularization. The review identified the need for more specific fMRI studies, especially those that employ proper designs, that is, applying appropriate cognitive tasks, multiple follow-ups, control conditions, and behavioral and standard cognitive measures.

CONCLUSION
fMRI is a unique way to provide valuable brain activation information to investigate mechanisms of cognitive effects of clinical revascularization for the treatment of severe carotid stenosis. Recent fMRI research has suggested positive cognitive effects of revascularization in treating severe carotid stenosis, though with notable heterogeneity. The literature review helps facilitate future development of the research field for potential translation.