Perioperative Stroke in Carotid Artery Stenting as a Surrogate Marker and Predictor for 30-day Postprocedural Mortality – A Pooled Analysis of 156,000 Patients with Carotid Artery Disease
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.273642
Source of Support: None, Conflict of Interest: None
Keywords: Carotid artery disease, carotid artery stenting, meta-analysis, perioperative stroke, perioperative transient ischemic attack, systematic review
Stroke is one of the top three causes of morbidity and mortality worldwide. Carotid artery disease (CAD), an atherosclerotic occlusive disease affecting the carotid artery, is estimated to be responsible for at least one-fourth of all stroke events worldwide. Further, it is estimated that CAD is responsible for more than 40,000 strokes occurring annually in the United States.
The buildup of atherosclerotic plaques, most commonly at the carotid bifurcation, causes progressive luminal narrowing which could ultimately impede cerebral blood flow. In addition, the ulceration of the atherosclerotic plaque with subsequent embolization could also lead to transient ischemic attacks (TIAs) or strokes. Medical management, carotid endarterectomy (CEA), and carotid artery stenosis (CAS) are treatment strategies for CAD which aim to minimize the risk of stroke.
CEA, when indicated, had been repeatedly proven to be highly effective in preventing ipsilateral strokes in both symptomatic and asymptomatic CAD patients. Nonetheless, CAS has recently been proposed as a less invasive and a reasonable alternative for CEA for stroke prevention. In many cases, the choice between the two treatment modalities may be unclear. However, CAS clearly becomes the preferred modality in the subset of patients with high surgical risk for carotid surgery.
The stenting technique involves opening the internal carotid artery by pressing the plaque against the arterial walls. This results in improved cerebral blood flow and prevents the artery wall from reclosing. In addition, it prevents pieces of the plaque from breaking off and causing cerebrovascular insults. However, CAS is not devoid of risks. During the revascularization procedure, embolization of thrombus or dislodging of atherosclerotic debris from the arterial wall might occur. These thrombi can travel and block the cerebral vasculature. Although perioperative stroke is seen in both CAS and CEA, the rate seems to be higher in CAS (4.8% vs <3.5%).,
Multiple large-scale trials have evaluated the perioperative stroke rate for CAS. The periprocedural period was defined as 30 days following the CAS procedure. In the carotid revascularization versus stenting trial (CREST), endarterectomy versus angioplasty in patients with symptomatic severe carotid stenosis (EVA-3s), stent-protected angioplasty versus carotid endarterectomy (SPACE) and international carotid stenting study (ICSS) clinical trials, the periprocedural stroke rates for CAS ranged from 4.3% to 9.1%. Out of all these trials, only the CREST  trial assessed the association between perioperative stroke and mortality. It focused on long-term mortality and showed that stroke following CAS is associated with a nearly 3-fold increase in 4-year mortality. To date, no studies have comprehensively examined this association. More specifically, no study has examined whether perioperative stroke occurring within 30-day of CAS confers an increased risk of short-term mortality. In this study, we performed a systematic review of the literature and a meta-analysis to assess whether periprocedural strokes, within 30-days of a CAS procedure, increases or predicts the 30-day mortality.
We followed the PRISMA and MOOSE guidelines when preparing this meta-analysis., All steps were performed in accordance with the Cochrane Handbook of Systematic Review and Meta-Analysis.
We conducted an electronic search on PubMed, Embase, and World Science Database for relevant articles from inception till October 2017. The following keywords were used to identify relevant publications: carotid artery disease, carotid disease, symptomatic carotid artery stenosis, asymptomatic carotid artery stenosis, carotid stenosis, cerebrovascular disease, stroke, transient ischemic attack, neurologic deficit, paralysis, and paresis. We also searched for potential studies by reviewing the reference lists of the included studies.
Inclusion and exclusion criteria
Studies satisfying the following criteria were included in our systematic review and meta-analysis:
Studies were excluded if they (i) were published in a language other than English, (ii) had a sample size of less than 100 patients, (iii) were case reports, reviews, conference papers, posters, thesis, or books (iv) included overlapping data sets or were duplicated articles.
Screening of records and study selection process
Search results from the databases were imported into Endnote (Thompson Reuters, PA, US) and duplicates were deleted. Two review authors independently screened all citations retrieved from the literature search. The screening was performed in two steps: the first step was to screen titles and abstracts of all citations and the second step was to retrieve and screen full-text articles of the selected abstracts. Titles and abstracts of included studies were independently screened by two reviewers. Any disagreements between the reviewers were resolved by discussion and consensus.
Two review authors independently extracted data of the included studies to a predefined extraction sheet. Extracted data included the number of patients, number of procedures/stents placed, asymptomatic population, symptomatic population, age and gender of patients, presence of contralateral CAD (>50% stenosis), comorbidities (coronary artery disease, hypertension, diabetes, hyperlipidemia/dyslipidemia, smoking status), technical success rate of stenting, TIA incidence, stroke incidence, myocardial infarction (MI) incidence, all-cause mortality, stroke-related mortality, and MI-related mortality.
Periprocedural stroke/TIA events after CAS are infrequent events; therefore, these binary outcomes could be expressed as OR or risk ratios. Of that patient pool, we focused on 30-day mortality of these patients which is even a smaller pool. We used R statistical software V 3.4.1 (Foundation for Statistical Computing, Vienna, Austria) for statistical analysis. The OR and relative risk (RR) and their 95% confidence intervals were pooled in the random and fixed effects meta-analysis models using the Mantel-Haenszel method. Forest plots were used for presentation of RRs, ORs, sensitivities, and specificities. Further, a hierarchical summary receiver operating characteristic (HSROC) curve with a 95% contour ellipsoid was used to present sensitivities and specificities. We reported both the fixed and random effects models for our statistics.
We assessed the heterogeneity across the included studies for OR, RR, sensitivities, and specificities by the I2 statistic. An I2 statistic <25% indicates a low amount of heterogeneity, and I2 statistic >75% indicates a high amount of heterogeneity. A funnel plot was constructed for a visual check to further assess the presence of publication bias in OR and RR. Further, linear regress test of funnel plot asymmetry was used to assess for publication bias. All calculations were based on Meta-Essentials: Workbook for Meta-Analysis. A P value <0.05 was considered statistically significant.
Our literature search of PubMed, Embase, and World Science databases yielded a total of 1198 articles. Following title and abstract screening, 333 records were eligible for full-text screening. Of them, 146 articles meeting the inclusion criteria were included in the final analysis. The flow diagram of the selection process is shown in [Figure 1]. The published articles used in the analysis included both retrospective and prospective cohort studies and all reported on the perioperative neurologic status and 30-day mortality. The minimum follow-up duration of all included studies was ≥30 days.
The 146 included studies comprised a total of 156,854 patients who underwent CAS The baseline characteristics of the cohorts of included studies are summarized in [Table 1]. In studies reporting the presence of CAD symptoms, 26.5% of the patients were symptomatic while 73.5% of the patients were asymptomatic. The mean age of the analyzed population was 70.7 years and 57.6% of the patients were males. About 15.2% of patients had contralateral CAD—defined as having >50% stenosis. In terms of comorbidities, 82.4% of patients were hypertensive, 71.3% of patients had hyperlipidemia/dyslipidemia, 57.1% had coronary artery disease, 33.1% of patients were diabetic, and 45.4% of patients were current or former smokers.
Clinical outcomes and mortality
The rate of technical failure in performing the stenting was 4.84 per 1,000 surgeries (449 failures in 92,873 procedures). The perioperative incidence of TIA in CAS was 2.4 per 100 procedures (883 TIA, events in 36,546 stenting procedures), while the perioperative incidence of stroke was 2.7 per 100 procedures (4,224 stroke-events in 158,691 procedures). In terms of fatality, 11.8% of stroke-events were fatal (498 deaths in 4,224 events) while 88.2% were nonfatal strokes. The perioperative incidence of MI was 6.1 per 1,000 procedures (375 events in 61,150 procedures); 22.0% of MI events were fatal (78 deaths in 354 MI events). The all-cause mortality was 10.6 deaths per 1,000 stenting procedures or 10.7 per 1,000 patients undergoing stenting. The total deaths in our cohort were 1,677 in 156,854 patients who underwent 158,691 stenting procedures.
The risk of 30-day mortality
114 studies were included in this analysis. The pooled OR of 30-day mortality after perioperative stroke is 18.98 (95% CI, 16.88–21.34) based on the fixed-effect model and 24.58 (95% CI, 19.92–30.32) based on the random-effects model. We found modest heterogeneity in the OR with an I 2 of 30.2% (95% CI, 11.7–44.9%), P = 0.002. The forest plot of the pooled OR and 95% CI for each of the included studies is depicted in [Figure 2].
The pooled RR of 30-day mortality after perioperative stroke was 16.50 (95% CI, 14.86–18.32) under the fixed-effect model, and 21.65 (95% CI, 17.87–26.22) under the random-effect model. We found modest heterogeneity in the RR with an I 2 of 32.6% (95% CI, 14.9–46.7%), P = 0.001. The forest plot of the pooled RR and 95% CI of each of the included studies is illustrated in the forest plot in [Figure 3].
Sensitivity and specificity
145 studies with a cohort of 156,802 patients were used in this analysis. [Supplementary Figure 1] exhibits a forest plot of sensitivities and specificities of perioperative stroke in predicting 30-day mortality following CAS. Perioperative stroke exhibited a sensitivity of 42.0% (95% CI, 0.378–0.464) and a specificity of 97.0% (95% CI, 0.967–0.973) in predicting 30-day mortality. Univariate heterogeneity analysis indicated modest heterogeneity in sensitivities (I 2 = 28.3%; P = 0.001) but a substantial heterogeneity in specificities (I 2 = 84.71%; P < 0.0001).
The area under the ROC curve (AUC), which represents the accuracy of perioperative stroke in predicting 30-day mortality, was found to be 0.86. The HSROC, which presents a global summary of test performance, shows the trade-off between sensitivity and specificity. Graphs of the estimated HSROC along with summary points, 95% confidence ellipse, and prediction ellipse at each time point are shown in [Figure 4].
In order to investigate the existence of publication bias in the pooled estimates, funnel plots were constructed where the effect estimates were plotted against their standard error. By visual inspection of funnel plots [Supplementary Figure 2] and [Supplementary Figure 3], we found no asymmetry indicating no publication bias. We also used the linear regression test (Egger's test) of funnel plot asymmetry which demonstrated no significant asymmetry with a P value <0.001 to suggest publication bias. The funnel plots can be found in the supplementary figures.
Our study provides evidence that perioperative stroke is associated with a higher risk of 30-day mortality following CAS. Both the SNACC Consensus Statements  and the AHA/ASA Expert Consensus Document  define “perioperative stroke” as brain infarction of ischemic or hemorrhagic etiology that occurs during surgery or within 30 days of surgery. Similarly, we adopted this definition for perioperative stroke and performed data extraction on strokes happening within 30 days of carotid stenting. It is worth mentioning that although the definition incorporates both ischemic and hemorrhagic strokes, the majority of events were ischemic in origin.
In our study, we defined the “mortality outcomes” as any death occurring within 30 days of the stenting procedure. Although faced with many critics, the 30-day mortality rate is still considered to be the traditional measure for surgical quality. Further, this specific time window is commonly chosen because it is the standard “outcome of care” measured and publicly reported by the Centers for Medicare and Medicaid Services (CMS).
Huibers and colleagues investigated the underlying pathophysiological mechanisms for procedural strokes associated with CAS by analyzing patients within the ICSS trial. They concluded that although the mechanisms for stroke are diverse, a hemodynamic disturbance is one of the primary contributing etiologies to developing strokes in these settings. Actually, one-third of procedural strokes in their trial were directly attributed to periprocedural hemodynamic abnormalities. Uncontrolled hypotension during the procedure is likely the result of carotid sinus manipulation and baroreceptor dysfunction. In addition, technical flaws related to the delay in placing the shunt or prolonged balloon dilation might also contribute to the cerebral hypoperfusion. Hence, strict blood pressure monitoring and control, during and after the procedure, could potentially lower the incidence of periprocedural stroke.
In addition, carotid embolization is an important mechanism for periprocedural stroke and was found to be responsible for an additional 20% of stroke cases in CAS. Most of the time, cerebral deficits secondary to embolization occurs intraoperatively or in the early postoperative status (e.g., day 0–3) because of the manipulation of the carotid plaque. Actually, 85% of embolic events in the Huibers et al. cohort occurred on day 0 of CAS. Other potential etiologies for cerebral infarction include occlusion of the revascularized carotid artery, cardioembolic accidents, and cerebral hyperperfusion syndrome (CHS). It was previously suggested that the type of stent (closed- vs open-cell stents) utilized in stenting has an impact on the incidence of perioperative strokes. However, a recent meta-analysis by Texakalidis et al. debunked these claims as there was no statistically significant difference for perioperative stroke or TIA between the two types of stents.
Our study has many strength points: (1) we performed a comprehensive search on multiple databases, (2) we reported this manuscript in adherence with the PRISMA checklist, and (3) we performed all steps in strict accordance with the Cochrane Handbook of Systematic Reviews for interventions. Further, the large number of patients and studies included in the meta-analysis adds to the power of the evidence exhibited in this study. Nonetheless, there are few limitations in our study that need to be addressed. First, a large portion of the included studies is retrospective which raises the concern about “selection bias.” Second, heterogeneity analysis showed that a “moderate” heterogeneity exists among studies assessing OR, RR, and sensitivity, while a “substantial” amount of heterogeneity exists among studies assessing the specificity. Those heterogeneities might stem from the slight differences in the methodologies and definitions utilized in each of the included studies, especially how “perioperative stroke” was defined in each of them. Further, our analysis didn't account for the severity of stroke following surgery: we didn't distinguish between the mortality rates in minor neurologic deficits as compared to the permanently disabling strokes. Finally, “mortality” was the only metric used in our analysis to assess the burden of perioperative stroke following carotid stenting. In fact, quality is more than just keeping people alive, and hence, future research evaluating the impact of strokes should take into consideration other quality-of-life measures.
Perioperative strokes, occurring within 30 days following CAS, increases the risk of 30-day mortality significantly. These findings emphasize the importance of neurophysiologic monitoring to detect intraoperative cerebral ischemia during CAS. Further, this highlights the importance of optimizing post-procedural neurologic care for patients undergoing CAS and the imperative need to develop novel neuroprotective therapeutic approaches.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]