|Home | About | Journals | Submit | Contact Us | Français|
Distal cerebral embolization is a known complication of carotid interventions. Here we prospectively investigate whether subclinical microembolization seen on post-operative MRI leads to cognitive deficits in a cohort of patients undergoing carotid revascularization procedures.
Patients undergoing carotid interventions and eligible for MRI scanning were recruited to participate. Among 247 patients who received both preoperative and postoperative MRI evaluations, a total of 51 patients also completed neuropsychological testing prior to and at one month following the procedures. Cognitive evaluation included the Rey Auditory Verbal Learning Test (RAVLT) for memory evaluation and the Mini-Mental State Examination (MMSE) for general cognitive impairment screening.
All 51 patients (16 CAS and 35 CEA) were male with a mean age of 71 years, ranging 54 to 89 years. Among them, 27 patients (53%) were symptomatic preoperatively including 11 patients who had prior stroke and 16 patients who had prior TIA. The majority of the patients had significant medical comorbidities including hypertension (96%), diabetes (31.3%), coronary artery disease (47%), and COPD (15.7%). Two patients (4%) had prior ipsilateral CEA and 8 had contralateral carotid occlusion (15.7%).
Memory decline evident on RAVLT was identified in 21 patients including 8 CAS patients and 13 CEA patients. Eleven patients had evidence of procedure-related microemboli. Although there was no significant difference in baseline cognitive function or memory change between CEA and CAS cohort, the CAS cohort had significantly higher incidence of microembolic lesions. Multivariate regression analysis showed that procedure-related microembolization was associated with memory decline (P=0.016) as evident by change in RAVLT. Prior history of neurologic symptom was significantly associated with poor baseline cognitive function (MMSE) (P=0.03) and overall cognitive deterioration (change in MMSE) (P=0.026) as determined by Wilcoxon Rank Sum test and linear regression analysis respectively.
Although both CEA and CAS are effective in stroke prevention with minimal neurologic complication, neurocognitive effects remain uncertain. Procedure-associated microembolization and pre-existing neurologic symptoms are associated with poor baseline cognitive function and memory decline following the procedures. Further comprehensive cognitive evaluation to determine the benefit of carotid interventions is warranted.
Management of stroke costs $45 billion annually and is responsible for greater than one million hospital admissions each year in the United States1. Prevention of stroke with safe treatment of carotid disease is undoubtedly an important health care goal in our society. Both carotid endarterectomy (CEA) and carotid stenting (CAS) have been shown to be safe and effective options for stroke prevention in appropriately selected patients with relatively low neurologic complication2–7. However, subclinical microembolization identified on diffusion-weighted MRI (DWI) is common. We and others demonstrated that 20–70% of CAS patients and 10–20% of CEA patients experienced procedure-related microemboli evident on postoperative MRI despite absence of neurologic symptoms8–15. A recent sub-study of International Carotid Stenting Study (ICSS) trial confirmed 3 times more DWI lesions in the CAS than the CEA group15. However, the clinical relevance and long-term cognitive effects of microemboli associated with carotid interventions are largely unknown.
While many studies have investigated the effects of microemboli on postoperative cognitive dysfunction following cardiac surgery, only a limited number of studies addressed the influence of microemboli in cognitive functions among patients undergoing carotid revascularization procedures16–18. The results among carotid patients are controversial and systemic evaluation of microemboli is generally lacking.19–21 We hypothesize that subclinical microembolization moderates the degree of cognitive changes following carotid interventions and certain risks are associated with microemboli. This study was to evaluate the cognitive effects of microemboli associated with carotid revascularization procedures through a prospective, multi-disciplinary team approach. The neurocognitive battery was adapted from several published studies in consultation with a clinical neuropsychologist; it included standard clinical measures and was consistent with the recommendations by NINDS and Canadian Stroke Network as well as the consensus statement on neurobehavioral outcomes after cardiovascular surgery.22, 23 The battery was designed to evaluate cognitive function across the major cognitive domains and the key outcome measure was Rey Auditory Verbal Learning Test (RAVLT), a verbal world list learning memory test that has shown to be a sensitive measure following carotid intervention24. Other relevant patient-related factors were also evaluated.
Patients with high-grade carotid stenoses and scheduled for carotid interventions at the VA Palo Alto Health Care System (VAPAHCS) were recruited to participate. Both CEA and CAS patients were included. CEA with patch angioplasty was performed under general anesthesia using a standard protocol with routine usage of a shunt device; and CAS was performed in an endovascular suite (GE Medical Systems, Milwaukee, WI) with routine use of a distal embolic protection device, specifically Emboshield®. The indication for intervention was compliant with the standard clinical practice guideline and revascularization procedures were offered to patients with >60% symptomatic lesions or >80% asymptomatic lesions. Diagnosis of carotid stenosis was determined based on carotid ultrasound velocity criteria and NASCET criteria, and confirmed with preoperative MRA evaluations. CAS was offered to the patients who were high risk for CEA based on criteria established at consensus conferences and included various anatomical considerations (presence of tracheostomy, history of ipsilateral neck irradiation, prior radical neck dissection, or CEA)25. Nearly all patients routinely underwent persantine-thallium nuclear stress test (P-thal) and cardiology evaluations were obtained for those with an abnormal stress test. The medical risk was stratified collaboratively among the vascular surgeons, cardiologists, and internists. Both CEA and CAS procedures were performed by experienced vascular surgeons. Exclusion criteria included inability to undergo MRI, poorly controlled or untreated psychiatric diseases, history of neurological or systemic illness affecting the central nervous system, non-English speakers, and the inability to complete all study procedures. The study was approved by Stanford University IRB and Palo Alto VA R&D committee.
Study participants underwent neuropsychological testing at 1–2 weeks prior to and at one month following their procedures. At each test a Mini-Mental State Examination (MMSE) was performed to evaluate general cognition and then a series of tests were performed to test the following different cognitive abilities:
At the first testing session, the Weschler Test of Adult Reading (WTAR) was also given to assess general intelligence and to screen for gross abnormality prior to carotid procedures.
Two key neuropsychological measures were MMSE and RALVT. The MMSE is a brief mental status examination designed to provide a general clinical measure of overall cognitive function by assessing performance on the following cognitive domains: orientation, language, calculation, memory, and visuospatial reproduction. The score on the MMSE is the total number of items answered correctly. The total score can range from 0–30 points. Patients with MMSE<24 were considered severely cognitive impaired. Scores of 24–28 indicated mild cognitive impairment and >28 was considered relatively normal. The MMSE has high test-retest reliability and has moderate sensitivity. The MMSE was included as a brief measure of global cognitive functioning and screen for obvious dementia (cut score below 24). RAVLT is a reliable and valid measure of verbal memory and has psychometrically sound alternate forms that can reduce practice effects between follow up appointments. RAVLT is designed to evaluate verbal word list learning, for which the patients are presented with an original list of 15 words five times and are then asked to recall them immediately following each presentation. The patients are then presented with and asked to recall a second unrelated list (interference list) of words. Following the interference list, the patients are asked to recall the original list again. Lastly, each patient is asked to recall the original list after a 20-minute delay without further presentation of those words. Previously, RAVLT has been shown to be extremely sensitive (sensitivity 90.2% and specificity 84.2%) in discriminating healthy older adults from non-demented older adults with focal memory deficit26. In our study, RAVLT score was calculated as the sum of the words recalled in the first five trials and the change in score was determined by the difference between scores from pre- and post-procedural tests; patients with negative change in RAVLT sum were compared to those with no change or positive change on RAVLT.
This neurocognitive battery was selected by a clinical neuropsychologist (AR) and consistent with recommendations by NINDS and Canadian Stroke Network as well as the consensus statement on neurobehavioral outcomes after cardiovascular surgery22, 23. It consisted of standard clinical measures in part because these tests had been widely validated in diverse older adult populations and could be generalized for clinical use. The battery was also designed to be briefly and reliably administered by trained personnel.
Patients were scanned on a 1.5T magnet (Signa Excite HD 12.0, GE Medical Systems, Milwaukee, WI, USA) equipped with a head coil27. The brain was scanned utilizing multiple pulse sequences in the axial, sagittal, and coronal planes according to a standard clinical stroke protocol. MRI with DWI sequences were performed for each patient pre-operatively and within 48 hours following their procedures in order to identify incidence of microemboli. DWI was acquired with an echo-planar sequence (EPI) as we described previously28. Apparent diffusion coefficient (ADC) map was automatically generated. Scans were interpreted by a board-certified neuroradiologist who was blinded to the clinical status of the patients. A microembolus was identified as a new hyperintensity on postoperative DWI with corresponded hypointensity on ADC map. Number and location of microemboli were recorded. The intracranial vessels were imaged using 3-dimensional (3D) time-of-flight (TOF) MRA techniques, and the extracranial carotid and vertebral arteries were imaged utilizing pre-contrast 2D TOF and post-contrast 2D MRA techniques.
For each participant, age and gender were recorded as well as the following risk factors: history of smoking, hypertension (SBP >140mm Hg, and/or DBP>90mmHg), hyperlipidemia (serum total cholesterol >200mg/dL, LDL>130mg/dL, and/or triglyceride>200mg/dL), severe coronary artery disease (CAD) with reversibility on P-thal, chronic obstructive pulmonary disease (COPD), atrial fibrillation (A. fib), diabetes mellitus (DM), obesity (BMI>30), and peripheral vascular disease (PVD). Anatomical risk factors including history of ipsilateral CEA, ipsilateral neck surgery/radiation, and contralateral carotid occlusion (Contra-occlusion) were documented. Additionally, preoperative symptoms, such as stroke (prior stroke) or transient ischemic attack (prior TIA), and post-operative neurological symptoms were analyzed. For the CAS cohort, lesion calcification and aortic arch classification were also considered.
Data was collected for all patients on an electronic spreadsheet and analyzed using SAS software in consultation with a statistician (LS). Descriptive data was analyzed using a two-tailed Student’s t-test and Fisher’s exact test. Key independent variables included type of procedure, incidence of microemboli on postoperative DWI images (Post-DWI), prior stroke, prior TIA, aortic arch type, post-procedural neurological problems, and post-procedural administration of vasopressors. Key dependent variables were RAVLT and MMSE. RAVLT score was calculated as the sum of the words recalled in the first five trials (RAVLT sum) and the change in score was determined by the difference between scores from pre- and post-procedural tests; negative change in RAVLT sum indicated cognitive decline, while no change or positive change indicated no decline. Fisher’s exact test and a two-sample t-test was used to analyze the distribution of RAVLT change scores; fisher’s exact test was also used to look at the distribution of MMSE baseline scores and patients with scores less than 28 were identified as impaired. Wilcoxon Rank Sum Test and Kruskal-Wallis Test were used to examine the distribution of MMSE change. A Linear Regression was also performed to determine independent association between the key independent variables and the key outcome measures (change in RAVLT and MMSE).
A total of 247 patients who underwent carotid revascularization procedures received pre- and post-operative MRI scans with DWI at VAPAHCS from 2004 to 2011. From this cohort, a total of 69 patients were prospectively enrolled in the study and received neurocognitive battery from February 2009 to May 2011, when neurocognitive battery was available and approved. There was no difference in demographic characteristics or neurologic complications between the patients who received neurocognitive battery after February 2009 and those who did not in the earlier study period. Although there was a general trend of lower incidence of microemboli in the patients who received neurocognitive battery (23%) compared to the earlier cohort who did not undergo neurocognitive test (27.4%), the difference did not reach significance. Eighteen patients were excluded from analysis due to inabilities to complete the neurocognitive battery, failed screened tests, or withdrawal from the study, leaving a total of 51 patients who completed neuropsychological testing. Four patients who did not receive post-operative MRI scan but completed neuropsychological testing were excluded from the final analysis on cognitive outcomes. Among the 18 patients who were completely excluded from the study, the mean age was 74.5 years and the majority had a history hypertension (94.4%), hyperlipidemia (94.4%), diabetes (66.7%), and CAD (61%). Eight of the 18 patients underwent CEA and eight patients (44.4%) had symptomatic carotid stenosis. Microemboli occurred in 3 patients (16.7%), similar incidence to those who were included in the study.
All participants were male with a mean age of 71 years. Thirty-five participants received CEA (mean age, 72 years) and 16 received CAS (mean age, 68 years). Patient demographic characteristics are summarized in Table 1. A majority of the patients had a recent history of smoking (78%) and hypertension (96%). Over half of the patients (53%, n=27) were also symptomatic. There was no significant difference in general risk factors between CEA and CAS cohorts including age, smoking history, hypertension, hyperlipidemia, atrial fibrillation, COPD, DM, obesity, and PVD. There was also no significant difference in preoperative neurologic symptoms between the CEA and CAS groups. However, CAS patients had a trend of higher CAD (69% vs. 37%, P=0.068) and a significantly higher incidence of abnormal initial cardiac stress test with reversibility on p-thal (69% vs. 11%, P<0.001). CAS patients also had higher surgical risks, such as history of prior ipsilateral CEA (12.5% vs. 0%, P=0.094) and preoperative contralateral carotid occlusion (44% vs. 3%, P<0.001). Although there was no permanent neurologic complication, two CAS patients had transient neurologic attacks that were completely resolved at the time of the discharge. CAS patients also had a significantly higher incidence of procedure-related contralateral (19% vs. 0%, P=0.0019) and ipsilateral (50% vs. 9%, P=0.0269) DWI abnormalities indicative of microemboli (Table 1a). These DWI lesions were heterogeneous in distribution, consistent with our previous report28.
No significant difference was found in baseline cognitive function between CEA and CAS patient cohorts. A total of 26 patients were found to have low baseline cognitive function and 25 patients had relatively normal baseline cognitive function as determined by MMSE. Each group had two patients who did not receive MRI. There was no difference in the incidence of microemboli between the patients with low baseline cognitive function and those who did not (Table 1b) based on Fisher’s exact test. Memory decline as indicated by a decrease in performance on RAVLT was found in 21 (41%) patients including 13 CEA and 8 CAS patients. No difference was detected in memory change between CEA and CAS cohorts based on the two sample Student’s t-test when each individual score was considered or Fisher’s exact test when the two cohorts were compared (Table 1c). Changes in RAVLT and MMSE for each individual patient were displayed in Figures 1 and and2,2, respectively.
To evaluate predictors of decline on RAVLT or MMSE, six most relevant factors that have shown to affect cognitive function or incidence of microemboli in various studies were included in our multiple linear regression models: age, procedure type, presence of procedure-related DWI abnormality, preoperative TIA or stroke, and postoperative neurologic complication. For CAS patients, aortic arch anatomy was also evaluated. Eleven patients including 8 (50%) CAS and 3 (8.6%) CEA patients were found to have microemboli following the procedure and our regression analysis revealed that there was a trend (p=0.0525) towards memory decline for patients with microemboli. Patients who suffered procedure-related microemboli had a decrease in the mean RAVLT scores from 32.7±12 pre-procedurally to 29.7±9.3 postoperatively, while those who did not suffer microemboli had a slightly increased RAVLT mean score from 33.3±8.8 to 34.2±9.4. There was also a trend (p=0.0691) towards memory decline in patients who have experienced preoperative TIA symptoms. Using a backward model selection where variables with a p<0.15 were included in the model, post-procedure DWI lesions (post-DWI) was the only significant predictor of decline on RAVLT (p=0.0162, R2=0.31) (Table 2a). Similarly, prior stroke and age were significantly correlated with procedure-associated MMSE change (p<0.05) in our multiple linear regression model; and preoperative stroke symptom was the only factor associated with MMSE change (p=0.026) using a backward model selection (Table 2b).
The cognitive effects of microemboli following carotid intervention are largely unknown. As microembolization is increasingly recognized as an outcome measure of carotid interventions, understanding the clinical relevance of microemboli is essential. Our prospective evaluation highlighted that although microembolization is not significantly associated with overall cognitive change, it is an independent predictor of memory decline following carotid revascularization procedures. This study adds critical information to our limited knowledge on carotid revascularization- associated microembolization.
There are several studies that evaluated cognitive outcomes following carotid interventions and showed controversial results19–21. Lehrner et al showed that although overall cognitive functioning did not dramatically change in most patients at six months, significant improvement or deterioration in single neurocognitive domains were demonstrated20. Although the author postulated that magnitude of microemboli production might contribute to the discrepancy in cognitive domains, microembolization was not examined. Lal et al showed that carotid revascularization resulted in an overall improvement in cognitive function and that CEA was associated with a reduction in memory, while CAS patients showed reduced psychomotor speed29. Similar to these studies, we also prospectively evaluated each cognitive domain before and after carotid revascularization procedures. Furthermore, we systemically evaluated procedure-related microembolization and other significant risk factors.
The majority of the studies that examined carotid procedure-associated microembolization had limited neurocognitive evaluation. Gossetti et al identified microemboli in 44% of CAS vs. 4% of CEA patients and suggested that higher cognitive decline in the CAS group was likely due to higher embolic load, but comprehensive cognitive testing was not performed19. A study of 41 patients by Grunwald and colleagues showed improvement in cognitive speed but not in memory function following an unprotected CAS30. Unlike these studies, we adapted a test battery that targeted patients with mild cognitive dysfunction. The key outcome measure of memory used in our study is shown to be particularly sensitive for early Alzheimer disease and mild cognitive impairment amnestic subtype patients26.
One recent study by Wasser et al did not show lasting cognitive effects of microemboli when post-procedural testing was performed within 72 hours of the procedures31. To avoid the potential effect of anesthesia and to ensure that the observed cognitive changes would be enduring, we administrated postoperative neurocognitive battery at one-month follow-up. We also used parallel forms of the word list memory test to minimize the practice effect. We showed that procedure-related microemboli was the only predictor of memory decline at one month. Multiple studies on cardiac surgery population have confirmed that postoperative cognitive dysfunction is a significant predictor of long-term cognitive decline32–35.
Admittedly, there are several limitations of this study. Although NASCET and ACAS showed that surgical intervention was indicated for symptomatic patients with >50% and asymptomatic patients with >60% carotid stenoses, it is generally accepted that most of the patients, particularly asymptomatic patients, can be managed medically with improved medication regiments over the last decade. We and others have also adapted a more conservative approach in treating patients with carotid stenosis. It is our standard practice and the generally accepted practice within the vascular community to raise the interventional threshold. Therefore, our patient cohort had more severe diseases compared to the patients enrolled in NASCET and ACAS trials as well as the recent CREST trial. It is possible that severe carotid disease contributed to the general poor neurocognitive function at the baseline among our patients. Additionally, our CAS cohort had significantly higher cardiovascular risks (reversibility on P-thal) and surgical risks (contralateral carotid occlusion). The risk distribution was consistent with our standard clinical practice guideline as all patients underwent risk stratification and CAS was only offered to the high-risk patients. All patients who underwent CAS received a distal embolic protection device, namely Emboshied®. Although using the same embolic protection and stent devices limited technical variability and learning curve-related inconsistency, the effect of other embolic protection devices including flow reversal on microembolization was not evaluated in this study. Furthermore, although we found no difference in cognitive function or memory changes between the CAS and CEA cohorts, the CAS cohort had a significantly higher microembolic rate and the incidence of microemboli was shown to be the only independent predictor of memory decline using a linear regression model with backwards selection in our study. It is possible that differences in the effects of microemboli between the two patient populations may be seen in a larger sample size. We also showed trends of association between preoperative stroke symptoms and a decline in memory scores and that preoperative stroke correlated with decline in MMSE scores. These findings suggest that patients who had a history of neurological events were particularly at risk. Our results also show that symptomatic patients had correlated low baseline MMSE scores (p=0.0337), suggesting that these patients were at lower baseline cognitive functional status. We postulate that existing brain injury may pre-dispose patients to memory and cognitive function decline following these procedures.
In this study, we focused on changes in MMSE and RAVLT. MMSE was used as a general screen test for overall cognitive function and RAVLT as the primary outcome measure. Studies have demonstrated an association between memory decline and microemboli during cardiac surgery.16–18, and overall score of verbal memory on the RAVLT appeared to be a particularly sensitive measure of neurocognitive changes follow carotid interventions24. Our previous analysis showed that regions vulnerable to emboli included those implicated in memory (posterior cingulate), functions commonly affected in Alzheimer’s disease and vascular dementia (unpublished data). Therefore, we focused on our sensitive primary outcome measure and several important risk factors in this initial evaluation. Other risk factors and cognitive measures that aimed to comprehensively examine multiple cognitive domains not analyzed in this study may also be affected. Large prospectively collected data and long-term comprehensive cognitive evaluation are warranted in the future.
In summary, although there was no significant difference in baseline cognitive function or memory change between CEA and CAS cohorts, the CAS cohort had significantly higher incidence of microembolic lesions. Our study showed independent association between microemboli and memory deterioration following carotid interventions, signifying the clinical importance of microemboli. This study lays the foundation for further investigation. It is our hope that with a larger patient cohort we will be better able to characterize why changes in cognition may be occurring for these patients. Identifying a subgroup of patients susceptible to procedure-related cognitive deterioration enables better patient consultation and individualized patient care.
Funding Source: NIH R01NS070308 (WZ), and AHA10CRP2610312 (WZ)
Presented at Society for Clinical Vascular Surgeons, Las Vegas, March 15th, 2012
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.