Prognosis of Asymptomatic Carotid Artery Occlusion
Systematic Review and Meta-Analysis
Background and Purpose—The aim of this systematic review was to quantify the risk of ipsilateral stroke in patients with asymptomatic carotid artery occlusion (ACAO).
Methods—Studies reporting ipsilateral stroke risk in ACAO were identified by a search of MEDLINE, EMBASE, and study bibliographies. Study estimates were pooled using a random effects model, and heterogeneity was quantified using the I2 statistic. The primary outcome was the annual rate of ipsilateral stroke.
Results—Thirteen studies were identified, encompassing 718 patients with ACAO who were followed up for a median of 2.8 years. The annual rate of ipsilateral stroke was 1.3% (95% confidence interval, 0.4–2.1; I2=53%). The annual rate of ipsilateral transient ischemic attack was 1.0% (95% confidence interval, 0.3–1.8; I2=40%). The annual rate of death was substantially higher at 7.7% (95% confidence interval, 4.3–11.2; I2=83%). Correction for possible publication bias for the primary outcome suggested a lower risk of ipsilateral stroke (0.3% per year; 95% confidence interval, –0.4 to 1.1).
Conclusions—Stroke in ACAO is relatively infrequent, but patients face high mortality rates. This suggests the need for intensified medical therapy in ACAO.
Carotid artery occlusion accounts for 10% to 15% of all strokes and transient ischemic attacks (TIAs).1 Most cases of occlusion are because of atherosclerosis, with a smaller fraction comprising dissection, embolism, radiation vasculopathy, granulomatous disease, and moyamoya.2 The most common site for occlusion is at the origin of the internal carotid artery in the neck.
Asymptomatic carotid artery occlusion (ACAO) is most frequently discovered as an incidental finding during radiological workup for cerebrovascular disease or after auscultation of a neck bruit.3 Improvements in ultrasound technology and wide-scale diffusion of carotid ultrasound have led to a larger number of patients being diagnosed with ACAO. Data conflict as to the long-term prognosis of ACAO, with some studies suggesting high rates of events and other studies finding low rates.4–6 This was the impetus for the current systematic review, which synthesizes adverse outcomes across all available studies, with a particular focus on the risk of ipsilateral ischemic stroke.
This review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines (Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines checklist is given in Appendix I in the online-only Data Supplement).7
Studies were selected if they (1) enrolled patients with ACAO, even if only as a discrete subgroup; (2) collected follow-up information on the occurrence of ipsilateral ischemic stroke among patients with ACAO; and (3) were published in the English language. Studies enrolling patients who underwent carotid revascularization were excluded as revascularization may alter the natural history of carotid occlusion.8 There were no restrictions by sample size, duration of follow-up, the number of events, or publication status.
A 2-pronged search strategy was used for this review. First, OVID MEDLINE (from 1946) and EMBASE (from 1947) were searched using relevant keywords (ie, carotid, occlu*, and asymptomatic). The specific databases searched included MEDLINE, OLDMEDLINE, MEDLINE In-Process & Other Non-indexed Citations, MEDLINE Daily, EMBASE Classic, and EMBASE. A validated prognosis filter was used to narrow the search to prognostic studies (Appendix II in the online-only Data Supplement).9 The search was deduplicated to remove identical citations found in multiple databases. All resulting search hits were then screened for relevance by viewing titles, abstracts, and keywords. Potentially relevant articles were retrieved as full text. Second, the internal reference lists of relevant studies were perused to obtain additional search hits for review. In iterative fashion, the reference lists of the additional search hits were also reviewed (snowballing). The search was kept updated using email autoalerts from the OVID interface. Articles were graded for inclusion based on the aforementioned study eligibility criteria.
Descriptive variables were extracted from each eligible study, including citation information (authors and publication year), baseline patient characteristics (total sample size and the number of patients with ACAO, inclusion and exclusion criteria, accrual interval, mean age, and sex distribution), study design (prospective versus retrospective, defined according to the study investigators), diagnostic imaging information (diagnostic technique and proportion undergoing angiography), follow-up information (mean and maximum follow-up), and events. The latter included ipsilateral ischemic stroke, ipsilateral TIA, total stroke, total TIA, death, stroke-related death, and cardiac death (defined as death caused by coronary artery disease or heart failure). In all cases where Kaplan–Meier event rates were available, these were used preferentially. However, in the absence of actuarial event rates, all studies contained sufficient information for calculating the absolute annual event rate for ipsilateral stroke and other end points. All data were extracted onto an electronic spreadsheet.
Methodological quality of studies was rated using the Newcastle Ottawa scale; originally, a 9-point scale used for assessing observational studies.10 The scale was adapted to this data set by removing 2 questions focused on cohort studies with therapies. The resulting scale was scored out of a maximum of six points.
The primary meta-analytic outcome was ipsilateral ischemic stroke. This was expressed as an annual rate. Studies were combined using a random effects model, given widely varying event rates for ACAO reported in the literature. Point estimates with 95% confidence intervals (CIs) were computed across all studies. Heterogeneity was measured using the I2 statistic.11 As a rule of thumb, I2 values of 25%, 50%, and 75% represent low, moderate, and high heterogeneity, respectively.11 Also synthesized were total stroke, ipsilateral TIA, total TIA, death, stroke-related death, and cardiac death.
Prespecified meta-regression analyses were planned should I2 exceed 50% for the primary outcome. Using a univariate mixed effects regression and an unrestricted maximum likelihood approach, the following independent variables were tested for their effects on ipsilateral stroke rate: sample size, publication year, study design (prospective versus retrospective), mean age, sex female, (%), mean length of follow-up, and Newcastle Ottawa scale. Because medical therapy for atherosclerosis has improved over time, a subgroup analysis compared ipsilateral stroke rates in studies published before 2000 versus those published on or after 2000.
Publication bias was tested with a Duval and Tweedie trim-and-fill test.12 This analysis imputes potentially missing studies to the left or right of the mean and tests for change in the outcome variable.
From a total of 1240 citations, 1002 were found to be unique; these were then screened by reviewing titles, abstracts, and keywords (Figure I in the online-only Data Supplement). After eliminating irrelevant reports, 73 studies were assessed in full-text format. The most common reason for exclusion among these studies was that no asymptomatic patients were included. As a result of review, 13 studies were included in the meta-analysis (Table 1; Table I in the online-only Data Supplement).4–6,13–22
In aggregate, the studies enrolled 4406 patients; of whom, 718 had ACAO (16%). The median age of patients with ACAO was 67, and 23% were women. Most studies were prospective (n=10), and the median Newcastle Ottawa scale score was 5 (range 4–6; Table 2). All but 2 studies used ultrasound to define ACAO diagnostically. However, use of angiography was also high overall (66% of subjects).
Median follow-up was 2.80 years with a total of 1982 person-years of observation time. The annual ipsilateral stroke rate was 1.3% (95% CI, 0.4–2.1%; Table 3). Two- and 5-year rates of stroke were 2.5% and 6.3%, respectively. There was substantial heterogeneity in the base estimate (I2=53%). Annual total stroke was 2.0% (95% CI, 0.9–3.0%; I2=40%).
Eleven studies reported on ipsilateral TIA. The annual rate of ipsilateral TIA was 1.0% (95% CI, 0.3–1.8%; I2=40%). Annual total TIA was 3.0% (95% CI, 1.9–4.1; I2=0). Seven studies reported mortality; the annual rate of death was 7.7% (95% CI, 4.3–11.2). Heterogeneity was pronounced (I2= 83%). Six studies reported on stroke-related death, with an annual rate of 1.1% (95% CI, 0.07–2.1; I2=63%). Cardiac death was more frequent: 3.3% per year (95% CI, 1.2–5.4; I2=83%). Heterogeneity estimates were substantial for most measures with the exception of TIA and total stroke.
Meta-regression analyses revealed that ipsilateral stroke was inversely related to sample size (β coefficient, −0.00005; P<0.001), publication year (−0.00059; P<0.001), and female sex (−0.10, P=0.013) but positively related to prospective study design (0.01442; P<0.001). There was no relationship with sample age, methodological quality, or duration of follow-up. In the prespecified subgroup analysis, studies published on or after the year 2000 had a lower aggregate ipsilateral stroke rate (0.9% per year) than studies published before 2000 (1.5% per year); this difference was statistically significant (P=0.003).
The funnel plot was asymmetrical (Figure) with evidence for missing studies to the left of the mean. Duval and Tweedie trim-and-fill analysis suggested a revised ipsilateral stroke rate of 0.3% per year (95% CI, −0.4 to 1.1) based on correction for publication bias. Details of study follow-up are provided in Table 4.
The main finding of this review is that the risk of stroke from ACAO is low: 1.3% per year or 0.3% per year in the publication-bias adjusted analysis. Therefore, procedures to repair most cases of ACAO are not warranted because the potential benefits are low, and most procedures carry inherent risk for stroke and other adverse outcomes.
That does not mean that ACAO is a low-risk condition. The annual risk of death is high (7.7%); about half of which is caused by cardiac causes. Patients with ACAO often have atherosclerosis in other critical locations, such as the coronary and peripheral arteries.5 Affected patients have a high burden of vascular risk factors that merit treatment to reduce their risk of systemic atherothrombosis.2 These risk factors include smoking, hypertension, diabetes mellitus, and dyslipidemia.
The main limitation of this review is that for many end points—including ipsilateral stroke—heterogeneity was present. This was because of several factors, including publication year, which is a proxy for accrual era. Older studies had higher rates of ipsilateral stroke than more recent studies, probably because of less treatment in previous studies with contemporary agents, such as statins and angiotensin-converting enzyme inhibitors. If true, this suggests that the rate of stroke from ACAO in the modern era (0.9%) is even lower than the grand mean suggests. In addition, there is a strong suggestion that missing studies exist to the left of the mean, which if combined with the data set would further lower the mean rate.
On the other hand, total rates for stroke and TIA were higher than ipsilateral rates, which again suggests that ACAO serves as a surrogate marker for vascular disease in other beds. Carotid occlusion may also carry risk for progressive cognitive decline. Patients with this condition warrant intensive, lifelong management of vascular risk factors to reduce systemic risk. The pathophysiology of stroke in ACAO is uncertain and may involve hemodynamic compromise, atheroembolism, or impaired washout. A secondary analysis of the Stroke Prevention by Aggressive Reduction in Cholesterol Levels trial suggested substantial benefit for atorvastatin among patients with carotid stenosis (however, occlusion was not defined in these results).23 Of note, later carotid revascularization was reduced by 56% (hazard ratio, 0.44; 95% CI, 0.24–0.79; P=0.006) in the group randomized to atorvastatin. There have been no trials of antiplatelet therapy specifically in patients with carotid occlusion, and thus, results might be extrapolated from patients with carotid stenosis. In the Perindopril Protection Against Recurrent Stroke Study (PROGRESS), active blood pressure–lowering treatment reduced the risk of large-artery ischemic stroke by 39% (95% CI, 5–61); 7% of the study population had carotid disease at baseline (defined as previous carotid intervention or carotid stenosis, >50%).24
Another limitation was the use of ultrasound in all but 2 studies to define carotid occlusion. Overall, the rate of angiography was 66%. The use of ultrasound may carry high sensitivity but poor specificity for the diagnosis of carotid occlusion and may also contribute heterogeneity for outcomes. However, studies typically used standardized definitions and referred to reference standards. Another limitation is a lack of uniform data on hemodynamic impairment, which may identify a subgroup at greater risk of stroke.25 Of note, only 23% of study participants were women, suggesting a significant selection bias that complicates application of the results to women. Furthermore, some studies were small (median sample size, 30), which may lead to imprecise estimates with broader 95% CIs.
Sources of Funding
This work was supported by a research grant from the Program of Experimental Medicine, Department of Medicine, Western University, London, Ontario, Canada.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.012760/-/DC1.
- Received January 13, 2016.
- Revision received March 4, 2016.
- Accepted March 10, 2016.
- © 2016 American Heart Association, Inc.
- Baumann T,
- Steck AJ,
- Lyrer P
- Hennerici M,
- Hülsbömer HB,
- Rautenberg W,
- Hefter H
- Moher D,
- Liberati A,
- Tetzlaff J,
- Altman DG
- Wells G,
- Shea B,
- O’Connell D,
- Peterson J,
- Welch V,
- Losos M,
- Tugwell P
- Higgins JP,
- Thompson SG,
- Deeks JJ,
- Altman DG
- Countee RW,
- Vijayanathan T
- Markus H,
- Cullinane M
- Vernieri F,
- Pasqualetti P,
- Passarelli F,
- Rossini PM,
- Silvestrini M
- Vernieri F,
- Pasqualetti P,
- Matteis M,
- Passarelli F,
- Troisi E,
- Rossini PM,
- et al
- Sillesen H,
- Amarenco P,
- Hennerici MG,
- Callahan A,
- Goldstein LB,
- Zivin J,
- et al
- Chapman N,
- Huxley R,
- Anderson C,
- Bousser MG,
- Chalmers J,
- Colman S,
- et al