Plaque Echolucency and Stroke Risk in Asymptomatic Carotid Stenosis
A Systematic Review and Meta-Analysis
Background and Purpose—Ultrasonographic plaque echolucency has been studied as a stroke risk marker in carotid atherosclerotic disease. We performed a systematic review and meta-analysis to summarize the association between ultrasound-determined carotid plaque echolucency and future ipsilateral stroke risk.
Methods—We searched the medical literature for studies evaluating the association between carotid plaque echolucency and future stroke in asymptomatic patients. We included prospective observational studies with stroke outcome ascertainment after baseline carotid plaque echolucency assessment. We performed a meta-analysis and assessed study heterogeneity and publication bias. We also performed subgroup analyses limited to patients with stenosis ≥50%, studies in which plaque echolucency was determined via subjective visual interpretation, studies with a relatively lower risk of bias, and studies published after the year 2000.
Results—We analyzed data from 7 studies on 7557 subjects with a mean follow-up of 37.2 months. We found a significant positive relationship between predominantly echolucent (compared with predominantly echogenic) plaques and the risk of future ipsilateral stroke across all stenosis severities (0% to 99%; relative risk, 2.31; 95% confidence interval, 1.58–3.39; P<0.001) and in subjects with ≥50% stenosis (relative risk, 2.61; 95% confidence interval, 1.47–4.63; P=0.001). A statistically significant increased relative risk for future stroke was preserved in all additional subgroup analyses. No statistically significant heterogeneity or publication bias was present in any of the meta-analyses.
Conclusions—The presence of ultrasound-determined carotid plaque echolucency provides predictive information in asymptomatic carotid artery stenosis beyond luminal stenosis. However, the magnitude of the increased risk is not sufficient on its own to iden tify patients likely to benefit from surgical revascularization.
See related article, p 7.
Two randomized controlled trials found that carotid endarterectomy can reduce the annual risk of stroke in asymptomatic patients with 50% to 99% carotid artery stenosis to 0.5% to 1.0%.1,2 However, the clinical relevance of these results has been questioned because progressive improvements in medical therapy have significantly reduced the annual stroke rate in asymptomatic carotid stenosis. For example, a meta-analysis3 demonstrated that when taking into account studies completing recruitment of asymptomatic carotid stenosis subjects between 2000 and 2010, the annual ipsilateral stroke rate is ≈1% and potentially even lower when only the most recent observational data included in this meta-analysis are considered. For this reason, and because of the marginal surgical stroke prevention benefit seen in the randomized trials, investigations have focused on improving risk stratification strategies beyond luminal stenosis measurements.
Ultrasound is an attractive potential tool for obtaining stroke risk information in carotid disease because it is widely available and has almost no contraindications. The use of carotid plaque echolucency as a potential marker for stroke risk is supported by histopathologic studies showing that plaque echolucency corresponds to lipid-rich necrotic core or intraplaque hemorrhage, more commonly found in symptom-associated carotid stenosis than in asymptomatic stenosis.4,5 However, there are conflicting data in the literature regarding the predictive value of carotid plaque echolucency in asymptomatic patients,6,7 and the small study samples studied result in wide confidence intervals for risk estimates. For these reasons, we performed a systematic review and meta-analysis evaluating whether ultrasound characterization of carotid plaque echogenicity is a predictor of ipsilateral stroke in asymptomatic carotid atherosclerotic disease.
This study followed guidelines presented in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.8
Study Eligibility Criteria
Studies with ultrasound characterization of carotid artery plaque echolucency in subjects subsequently followed for development of future ipsilateral stroke were eligible. Specific inclusion criteria were (1) English language manuscripts; (2) studies with ≥30 subjects; (3) studies of asymptomatic patients without histories of prior ipsilateral stroke or transient ischemic attack (TIA) at the time of imaging; (4) ultrasound determination of the presence or absence of carotid vessel plaque echolucency in subjects with carotid plaque (both stenosis-causing and nonstenosing plaques); (5) mean follow-up >12 months after plaque imaging; (6) clinical ascertainment of first time ipsilateral stroke during follow-up; and (7) nonsurgical management of patients with follow-up information on >85% of the initially asymptomatic cohort enrolled in the study. If a mixed cohort of previously symptomatic and asymptomatic patients or medically and surgically treated patients were presented, such a study was included if the ultrasound test and stroke outcome data for only the patients who were asymptomatic at baseline could be extracted from the article. In cases where the ultrasound test result or outcome data were not clear in the article, we attempted to contact the corresponding author for additional details. Furthermore, if ultrasound test data from a cohort was published more than once, only the originally published article was used for the testing data and outcome results to avoid pooling duplicate results. Finally, if >1 method of echolucency determination was presented in an article, the method with the greatest predictive ability was included in the meta-analysis.
Information Search and Data Collection
We performed systematic searches of multiple medical literature databases between January and March of 2014 to find all eligible articles without regard to when they were published. Major search terms for all databases included carotid stenosis, plaque, atherosclerosis, ultrasound, sonography, doppler, stroke, and transient ischemic attack. The search methodology details and data extraction process are provided in the Methods in the online-only Data Supplement.
Assessment of Risk of Bias in Studies
We used the following bias assessment criteria which we adapted from published meta-analyses9,10 of imaging markers of stroke risk in carotid disease: (1) risk of outcome ascertainment bias, for which we assessed whether investigators were blinded to ultrasound testing results when stroke outcomes were assessed; (2) risk of confounding bias, for which we assessed whether potentially confounding stroke risk factors were collected and analyzed; (3) completeness of follow-up data, for which we assessed whether losses to follow-up were systematically recorded and reported.
We performed a meta-analysis of studies meeting 2 criteria: (1) a relative risk (RR) was calculable from the raw data either as published or obtained via direct correspondence from the study author(s); (2) echolucency test results were presented in a dichotomized fashion (eg, predominantly echolucent versus echogenic) or in a categorical fashion that could be dichotomized in the data extraction process. We used the Q statistic to determine study heterogeneity and the Begg test for publication bias. We performed all analyses using a random-effects model in which we made the conservative assumption that included studies that did not have exactly the same effect size (RR). Given the potential for heterogeneity between studies in terms of sample size, patient characteristics, and testing methods, we did not use a fixed-effects model because in this approach, an identical effect size is assumed across all studies.
We performed 2 primary analyses from the studies in which ultrasound test data could be extracted, 1 including all patients (with all stenosis severities from 0% to 99%) and 1 limited to patients with ≥50% stenosis. The analysis limited to patients with ≥50% stenosis was performed by extracting this data from details provided in the article or via direct correspondence with the author. We also performed 3 additional prespecified subgroup analyses limited to the following study characteristics: (1) studies in which authors were blinded to imaging test data during outcome ascertainment and analyzed potentially confounding vascular risk factors; (2) studies in which plaque echolucency determination was made by subjective visual interpretation without quantitative imaging postprocessing; and (3) studies published since 2000. We also performed 3 post hoc sensitivity analyses: (1) excluding Polak et al11 to evaluate the possibility that its cohort of 4886 subjects could be significantly driving the effect size of the overall meta-analysis; (2) excluding Silvestrini et al12 given its use of an imaging-based (time-independent) definition of stroke, unlike the time-dependent, clinical definitions used in all other studies; and (3) excluding Topakian et al7 given that a subset of its subjects developed TIAs before stroke outcome ascertainment.
We screened a total of 5409 abstracts from which 8 manuscripts6,7,11–16 were ultimately deemed to meet all inclusion criteria for the systematic review. Study selection steps are summarized in Figure I in the online-only Data Supplement.
Qualitative Assessment and Study Characteristics
Of the 8 articles meeting inclusion criteria for qualitative review and pooling (Table 1), all were prospective, longitudinal nonrandomized observational studies with 2 conducted as international multicenter studies,7,15 2 in the United States,11,16 and 1 each in Australia,13 Denmark,6 Italy,12 and Norway.14 All evaluated patients with similar mean ages (range 64.0–72.6 years) and all except 1 study11 had a preponderance of male subjects (range 43% to 84.2% male). We found considerable differences in the degree of extracranial carotid artery stenosis studied, with 5 studies11,13–16 including a combination of subjects with low (<50%), moderate (50% to 69%), and high-grade (≥70%) stenosis. Of these 5 studies, 2 included several subjects who did not have stenosed carotid arteries,13,14 and 1 study11 included several subjects without carotid plaque (stenosing or nonstenosing). The remaining 3 studies6,7,12 were limited to moderate and high-grade carotid stenosis subjects. Subjects in all studies were followed for ≥21.8 months (range 21.8–52.8 months) for ascertainment of clinically defined first-time stroke. The definition of asymptomatic carotid stenosis was not always explicitly given11,14,16 or sometimes included patients with remote ipsilateral symptoms (>12–24 months) as asymptomatic patients.7,13
Ultrasound test results and outcomes in each test group are summarized in Tables 2 and 3 with additional details about the cohort with >50% stenosis summarized in Table I in the online-only Data Supplement. Seven studies6,7,11,13–16 defined stroke clinically as ipsilateral hemispheric neurological deficit but only 3 of these 7 studies6,14,15 stipulated specifically that the deficit must be present for >24 hours to be defined as a stroke. One study, Silvestrini et al,12 used time-independent clinical features with brain imaging evidence of cerebral ischemia to define stroke. This meant that some transient events (most likely called TIAs in all the other studies we included) were classified as stroke in the study by Silvestrini et al. Of the 8 studies meeting inclusion criteria for the systematic review, 7 were amenable to the calculation of an RR of ipsilateral stroke in the presence of plaque echolucency. In the 1 other study,13 the authors presented only a composite outcome measure of TIA plus stroke, which prevented this study from being included in the pooled ipsilateral stroke RR calculation. In 6 of studies included in the systematic review, RR information for asymptomatic patients was either presented or could be calculated from the raw data provided in the article. In the remaining 2 studies,12,14 mixed cohorts of symptomatic and asymptomatic patients at baseline were presented for which we were able to obtain test and outcome data for the asymptomatic patients only after correspondence with the study authors.
Definitions of ultrasound testing methods, imaging equipment, abnormal test results, and outcome measures are provided in Table 4. For all studies, we were able to dichotomize test results into positive or negative for echolucency, using definitions provided by the study authors. All studies used standard clinical ultrasound equipment. In addition, all studies except 26,15 used subjective observer interpretation of plaque echolucency as the definition of a positive test result.
Assessments of Study Methods
In 5 of the 8 studies,6,7,11,12,14 the authors described a process of blinding of ultrasound results in the determination of clinical outcomes, although in the remaining 3 studies, blinding was not described. In 7 of the included studies, the authors recorded potentially confounding risk factors with only 1 study16 not including these data. Finally, in 2 studies, there was a description of the exact numbers of subjects who were lost to follow-up (12 subjects in 1 study15 and 2 in another),14 whereas in 1 study,13 there was mention that some loss to follow-up had occurred without presenting the actual numbers. There was no mention of loss to follow-up in the remaining studies. Furthermore, in only 4 studies7,12,14,15 was the proportion of subjects undergoing surgical revascularization, although asymptomatic, provided. Finally, in 1 study,7 patients who were asymptomatic at baseline were followed up until their first ipsilateral stroke, whether or not they had an ipsilateral TIA first. Therefore, in this study, a mixture of asymptomatic and symptomatic patients was included in stroke event rates and correlations with plaque characteristics.
Meta-Analysis Results for All Subjects Including Those With and Without Stenosis
In this primary analysis, we studied 7557 subjects with a mean follow-up of ≈37.2 months, yielding a total of 23 410.2 person-years of follow-up. No significant heterogeneity (Q=9.438, P=0.150) or publication bias (Kendall’s score=7, P=0.293) was present in this primary analysis. We found a significant positive relationship between plaque echolucency and the risk of future ipsilateral stroke with a random effects RR of 2.31 (95% confidence interval [CI], 1.58–3.39; P<0.001; Figure 1). Of the total study sample, 1741 subjects (23.0%) had a positive ultrasound test for echolucency, whereas 5816 (77.0%) had a negative test for echolucency. In the echolucent-positive test group, 100 ipsilateral strokes occurred compared with 141 ipsilateral strokes in the echolucent-negative test group. The cumulative incidence of ipsilateral stroke in the echolucent plaque cohort was 5.7% compared with 2.4% in the non-echolucent plaque cohort.
Meta-Analysis Results for Moderate and High-Grade Carotid Stenosis
In our analysis limited to subjects with ≥50% extracranial carotid stenosis,6,7,12,15,16 we accumulated 2095 subjects with a mean follow-up of ≈29.7 months, yielding a total of 5185.1 person-years of follow-up (Figure 2). No significant heterogeneity (Q=8.216; P=0.084) or publication bias (Kendall’s score=0; P=1.000) was present in this subgroup analysis. In patients with ≥50% carotid stenosis, we found a significant positive relationship between plaque echolucency and the risk of future ipsilateral stroke with a random effects RR of 2.61 (95% CI, 1.47–4.63; P=0.001; Figure 2). Of the moderate to high-grade stenosis patient sample, 649 subjects (31.0%) had a positive ultrasound test for echolucency, whereas 1446 (69.0%) had negative test for echolucency. In the echolucent-positive test group, 67 ipsilateral strokes occurred compared with 59 ipsilateral strokes in the echolucent-negative test group. The cumulative incidence of ipsilateral stroke in the moderate and high-grade stenosis echolucent plaque cohort was 10.3% compared with 4.1% in the non-echolucent plaque cohort.
Subgroup Meta-Analysis Results
No significant heterogeneity or publication bias was found in any of the subgroup or sensitivity analyses (Table 4). A statistically significant random-effects RR was preserved in subgroup analyses involving (1) only those studies where test result blinding and analysis of confounding stroke risk factors occurred (RR, 2.03; 95% CI, 1.26–3.27), (2) only those studies in which echolucency was determined by subjective visual interpretation (RR, 2.73; 95% CI, 1.76–4.22), (3) only those studies which were published after 2000 (RR, 2.14; 95% CI, 1.28–3.59). Likewise, a statistically significant random-effects RR was also preserved in the 3 post hoc sensitivity analyses performed, and the magnitude of the RR was not sizably different.
The degree of carotid stenosis criteria alone within the 50% to 99% range provides only a relatively weak means for clinically stratifying the risk for ipsilateral stroke in asymptomatic patients.18,19 In this systematic review and meta-analysis of over 7500 patients, we studied plaque echogenicity as an additional marker of stroke risk. We found that patients with predominantly echolucent plaques had an ≈2.3 fold higher risk of future ipsilateral stroke than those with predominantly echogenic plaques. In patients with 50% to 99% carotid stenosis, we also found an ≈2.6-fold higher risk of ipsilateral stroke if the plaques were predominantly echolucent compared with plaques, which were not predominantly echolucent.
Furthermore, in 3 additional prespecified subgroup analyses, we showed that the increased risk of ipsilateral stroke noted in patients with echolucent plaques was robust to differences in patient samples, study methodology, and study era. Specifically, we showed that the subjective visual interpretation of echolucency also identified higher risk subgroups in our pooled analysis without the use of quantitative gray-scale median analysis of plaque6 or post processing image normalization.15 However, the relative performance of subjective, visually determined echolucency versus quantitative, computer-based methods requires further investigation because only 2 studies6,15 in our meta-analysis used computer-aided methods, thereby preventing a meaningful comparison of the techniques. In addition, after excluding studies with a relatively higher risk of outcome ascertainment and confounding bias, there was a preserved statistically significant RR. Moreover, we found that the risk of stroke was increased with echolucent plaque even when the analysis was limited to publications after the year 2000, an era in which improvements to medical therapy for stroke risk prevention have been more widely implemented.3 Finally, our post hoc sensitivity analyses demonstrated that our results were robust to the following: (1) the exclusion of the largest cohort11 in our study; (2) the exclusion of a study16 with imaging-based (time-independent) definition of stroke, which has the possibility of overclassifying TIAs as strokes, which would not be done in all other studies using the time-dependent, clinical definitions; and (3) the exclusion of a study7 in which a subset of patients were followed for first-time stroke whether or not they had an ipsilateral TIA that preceded the stroke.
The mechanism underlying increased stroke risk in echolucent carotid artery plaque is not entirely understood, but is likely related to the echolucent appearance of high-risk elements of atherosclerosis, including lipid-rich necrotic core and intraplaque hemorrhage.4 The relative proportion of these tissues is also uncertain, though most histopathologic studies suggest that lipid may be the largest plaque element by volume in echolucent plaque.5 However, the precise differentiation of tissues in echolucent plaque may be of limited clinical significance because most of these presumed tissue types are features of more advanced atherosclerotic lesions.10
There are limitations of the ultrasonographic imaging methods used by the studies included in this meta-analysis. First, we found a lack of consistency in the methods used to determine plaque echolucency. More work is needed to standardize the definitions of echolucency and provide inter- and intraobserver variability measures of echolucency assessment. The role of quantitative methods, such as the use of gray-scale median values as well as the effect of differences in ultrasound equipment on diagnostic accuracy and prediction of outcome also requires additional study. In addition, future work is needed to understand how risk from plaque echolucency could be incorporated in a multifactorial risk assessment strategy in which other presumptive stroke risk markers, such as plaque ulceration, are synthesized to produce a composite risk score.18 It is important to remember that even assuming a 0.5% to 1% annual stroke risk in asymptomatic carotid stenosis ≥50%, that a 2.6× higher RR of stroke as seen with predominantly echolucent plaque, will still require additional risk factors to be taken into account to inform decisions with regard to a carotid procedure supplementing modern medical therapy. The most meaningful improvements in risk stratification in asymptomatic carotid stenosis may only occur when plaque echolucency is combined with other risk markers, such as clinical features, the degree of stenosis, and other imaging measures.18
Several additional limitations of our study are important to consider. First, we noted that most studies did not make clear what systematic efforts were used to assure complete subject follow-up, and thus it is unclear to what extent losses to follow-up may have contributed to bias in our estimation of the RR associated with echolucent plaques. Second, in the studies where blinding to test results was not performed, we think that the possibility of outcome ascertainment bias exists. Third, we studied only medically managed asymptomatic patients, and it is therefore unclear to what extent selection bias may have influenced the risk profile of this group compared with asymptomatic patients undergoing surgical revascularization. Fourth, we calculated unadjusted RRs. Because the covariate risk factors, individual patient follow-up times, and medical therapies used varied so widely across studies, the existing data are not amenable to the calculation of covariate-adjusted RRs or annualized stroke risks. Fifth, most studies included patients with a wide range of carotid stenosis, including 2 studies in which carotid stenosis was not present in all subjects13,16 and 1 study11 in which several patients did not have any carotid plaque (stenosing or nonstenosing). However, specific breakdowns of subjects with and without stenosis were not provided. Because some studies focused only on stenosis ≥50% or presented these data separately, we were able to estimate a pooled RR for this group. However, the total study data were not amenable to similar calculation of RR for only patients with low-grade carotid stenosis (<50%) or nonstenosing carotid plaque. Finally, additional heterogeneity of the studies in this meta-analysis arise from lack of clarity in reporting on asymptomatic carotid disease, including variability in the definition of stroke and symptomatic status, inconsistent reporting of the nature of medical therapy received by subjects, and imprecision about the classification of first-time versus recurrent ischemic events. Future studies in carotid disease should rely on standardized definitions for these basic clinical features and outcome measures so that studies can be understood, analyzed, and interpreted in a transparent and clear fashion.
In spite of these challenges and limitations, we think that there is sufficient evidence to conclude that ultrasound carotid plaque echolucency is a predictive risk factor for ipsilateral stroke in patients across a wide range of carotid stenosis severity. Despite the limitations of available research results, to our knowledge, our study is the best quality, most comprehensive analysis of the predictive value of detecting plaque echolucency. Ultimately, the validation of plaque echolucency and other risk markers to inform treatment decisions in patients with asymptomatic carotid stenosis will require examination in high quality, prospective longitudinal studies of patients receiving current optimal medical treatment. Finally, although nearly 30% of subjects with ≥50% stenosis demonstrated plaque echolucency, given the low absolute stroke risk in these patients, plaque echolucency alone is not a powerful enough risk factor to select asymptomatic stenosis patients likely to benefit from carotid endarterectomy or stenting.
Sources of Funding
Dr Gupta’s effort was supported by an Association of University Radiologists-General Electric Radiology Research Award Fellowship and the Foundation of the American Society of Neuroradiology Scholar Award. Dr Mushlin received funding from a National Institute of Health Clinical and Translational Science Award to Weill Cornell Medical College.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006091/-/DC1.
- Received May 9, 2014.
- Revision received September 9, 2014.
- Accepted September 11, 2014.
- © 2014 American Heart Association, Inc.
- Halliday A,
- Mansfield A,
- Marro J,
- Peto C,
- Peto R,
- Potter J,
- et al
- Grønholdt ML,
- Nordestgaard BG,
- Wiebe BM,
- Wilhjelm JE,
- Sillesen H
- Grønholdt ML,
- Nordestgaard BG,
- Schroeder TV,
- Vorstrup S,
- Sillesen H
- Gupta A,
- Chazen JL,
- Hartman M,
- Delgado D,
- Anumula N,
- Shao H,
- et al
- Gupta A,
- Baradaran H,
- Schweitzer AD,
- Kamel H,
- Pandya A,
- Delgado D,
- et al
- Mathiesen EB,
- Bønaa KH,
- Joakimsen O
- Nicolaides AN,
- Kakkos SK,
- Griffin M,
- Sabetai M,
- Dhanjil S,
- Thomas DJ,
- et al