Clinical Risk Factors and CT Imaging Features of Carotid Atherosclerotic Plaques as Predictors of New Incident Carotid Ischemic Stroke: A Retrospective Cohort Study

Study Design

Clinical and imaging data, obtained as part of standard clinical stroke care at our institution, were retrospectively reviewed with the approval of the institutional review board. At our institution, patients with suspicion of acute stroke and no history of significant renal insufficiency or contrast allergy routinely undergo a stroke CT survey including the following imaging protocol: noncontrast CT, perfusion CT at 2 cross-sectional positions, CTA of the cervical and intracranial vessels, and a postcontrast cerebral CT.

We retrospectively identified 1000 consecutive patients admitted to our emergency department from January 2001 to January 2007 who had undergone a CT/CTA to evaluate their brain and carotid arteries. Only patients who also had a follow-up brain imaging study at least 14 days after baseline CTA were included. Our focus of interest was the carotid artery plaque CT features at baseline on the initial CTA, and whether or not the patient developed a brain infarct on the follow-up brain imaging that was not present on the baseline brain CT (Fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Graphic depiction of our study population and how it was selected and distributed into different groups for purposes of statistical analysis. CEA indicates carotid endarterectomy.

Patients with remote or old infarcts in a carotid distribution were excluded from our analysis because carotid atherosclerotic disease is an evolving process, and the carotid artery condition may have evolved in the time interval between when the remote infarct occurred and the time of our CTA study, which was used to characterize the carotid artery plaque features. This could have interfered with our identification of the carotid wall features predictive of future infarct.

Patients who had a carotid endarterectomy were also excluded. Their inclusion would have dramatically affected our results, as the risk of subsequent infarct would be confounded by the procedure. Further, infarct is itself a risk inherent to endarterectomies and invasive cardiac surgeries.

Finally, we excluded patients who had presented with diagnoses that may predispose them to infarcts, independent of carotid atherosclerosis, including carotid dissection, intracranial hemorrhage, and vasculitides. Although we used a comprehensive protocol to differentiate infarcts (ie, between carotid and noncarotid), the source of some infarcts could not be identified. These were labeled “uncertain” infarcts and were subsequently excluded from the study.

CTA Imaging Protocol

CTA studies of the carotid arteries were obtained on 16-section and 64-section CT scanners (GE Healthcare, Milwaukee, Wisconsin). The image acquisition protocol was as follows: spiral mode, 0.6- to 0.8-second gantry rotation; collimation: 16 or 64 × 0.625 mm or 1.25 mm; pitch: 1.375:1 or 0.982:1; section thickness: 0.625 mm; reconstruction interval: 0.5 mm; acquisition parameters: 120 kVp/240 mA. A caudocranial scanning direction was selected, covering the mid-chest to the vertex of the brain. Seventy mL of iohexol (Omnipaque; Amersham Health, Princeton, New Jersey; 300 mg/mL of iodine) was injected into an antecubital vein with a power injector at a rate of 4 mL per second. Optimal timing of the CTA acquisition was achieved by using a test bolus technique. Effective dose associated with this CTA protocol was approximately 7–10 mSv.

Image Postprocessing

CTA studies of the carotid arteries were processed automatically using a custom CT-based automated classifier computer algorithm that was validated by using histology derived from carotid endarterectomy specimens as a criterion standard. This algorithm automatically segments the inner and outer contours of the carotid artery wall and distinguishes between the histologic components of the wall (lipids, calcium) by using appropriate thresholds of CT attenuation.^23,29 The algorithm creates a color overlay affording a visual display of the composition of the carotid wall for each CTA image (On-line Fig 1).²³ It then automatically analyzes several CT features of the carotid arteries and quantifies them 3-dimensionally (not in a plane, as with B-mode sonography), independent of any subjective human interpretation.³² The location of the largest lipid and calcium clusters was described as a percent of the carotid wall thickness, with 0% indicating the center of the cluster immediately adjacent to the inner contour and 100% indicating the center of the cluster immediately adjacent to the outer contour. Measurements were recorded for the 2.5 cm of the CCA immediately proximal to the carotid bifurcation and the 2.5 cm of the ICA immediately distal to the carotid bifurcation.⁴¹ Maximal carotid wall thickness was measured as the greatest distance from the lumen-intima interface to the outer edge of the adventitia. The physician processing the CTA datasets was blinded to the clinical findings of the imaged patients and to the group to which they belonged. Further, the physician was not aware of the reason for imaging or the patients' medical histories.

Baseline Image Review

The CT studies of the brain parenchyma obtained at baseline and the brain imaging studies obtained within the first 14 days after the baseline CT were reviewed by a neuroradiologist for the presence or absence of an acute infarct and its distribution (unilateral or bilateral, single or multiple vascular territories, and location of vascular territory). The 14-day delay was selected because an acute infarct may not show up conspicuously on the day 0 brain parenchymal imaging, especially if the initial imaging is a CT. However, in such a setting, the patient is likely to get a repeat CT or a MR imaging of the brain within the first 2 weeks that would show the acute infarct. The neuroradiologist reviewed the same studies of the brain parenchyma for old remote infarcts. The neuroradiologist also reviewed the intracranial portion of the baseline CTA of the carotid arteries for the degree of completeness of the circle of Willis.

The neuroradiologist assessed the degree of carotid stenosis on the cervical portion of the baseline CTA but did not record any information regarding the carotid wall. During the review, the neuroradiologist was blinded to both the results of the automatic analysis of the carotid wall produced by the computer algorithm, as well as any clinical variables or characteristics.

Follow-Up Image Review

The neuroradiologist evaluated all follow-up studies of the brain parenchyma obtained at least 14 days after the baseline CTA for the presence or absence of an acute infarct and its distribution (unilateral or bilateral, single or multiple vascular territories, and location of vascular territory). The assumption was that any brain parenchymal imaging study obtained more than 14 days after an initial stroke study would probably be triggered by a new clinical event. This comparison to the baseline study allowed verification that any infarcts observed were in fact new.

Medical Chart Review

The medical records of the patients were reviewed, independently of any imaging findings, for age, sex, ethnicity, smoking history, history of diabetes, hypertension, hyperlipidemia, drug treatment, and clinical diagnosis at the time of the follow-up imaging.

Patient Classification

The medical records of the patients were reviewed to determine the probable etiology of the stroke. Using a system based on the Causative Classification System for Ischemic Stroke⁴² developed by the A.A. Martinos Center, and on the review of the imaging studies of the brain parenchyma by the neuroradiologist,^43,44 as well as the degree of carotid stenosis, cardiac disease, and cardiac risk factors (independent of any carotid wall CT features), infarcts were categorized as “carotid infarcts” if they were in a carotid distribution, with the probable mechanism being large artery atherosclerosis. Nonacute strokes and acute strokes in a distribution not consistent with a carotid origin were categorized as “noncarotid infarcts.” Our classification method and the criteria used for classification are summarized in On-line Fig 2 and On-line Tables 1–4.

Statistical Analysis

Our statistical analysis consisted of 2 parts: univariate analyses of the clinical and imaging variables, followed by a multivariate analysis to determine which combination of clinical variables and carotid wall characteristics indicated an increased risk of subsequent new carotid infarct. For our univariate analyses, different types of statistical tests were used for different clinical or imaging variables to verify any association with new incident carotid stroke. Simple t tests were used for continuous clinical variables, while Mann-Whitney tests were performed for the dichotomous clinical variables. For the imaging variables, we used a mixed effect logistic model (with fixed effect for patient).

Variables that showed P values <.2 in the univariate analyses were considered for use in the multivariate analysis, which used .05 as a threshold for statistical significance. However, many of the variables (especially the carotid imaging characteristics) were collinear, indicating overlap (ie, NASCET degree of narrowing and minimal cross-sectional lumen area). In these cases of collinear variables, we only used the one variable that was the most statistically significant (from the univariate analyses) in the final multivariate analysis. This multivariate model allowed us to discover which combination of clinical and/or imaging variables was associated with the greatest risk for subsequent new carotid infarct. Using ROC analysis, we compared different thresholds of the relevant variables to understand which helped provide the most optimal model. This resultant model was then compared with the conventional NASCET rule (ie, luminal stenosis) to find which was more effective in predicting strokes on follow-up in our sample.