Introduction
Moyamoya disease (MMD) is an abnormal cerebrovascular characterized by progressive stenosis or occlusion of the intracranial vessels, resulting in the formation of a fine vascular network (the “moyamoya” vessels) at the base of the brain [
1]. The current concept is to prevent cerebral ischemia and avoid cerebral hemorrhage [
2]. And revascularization surgery is the most effective treatment for MMD. Surgical revascularization can be mainly divided into indirect bypass, direct bypass and combined bypass [
3]. However, the incidence of postoperative complications, such as postoperative ischemic or hemorrhagic events and cerebral hyperperfusion syndrome (CHS), has been increasingly reported. These procedure-related complications can seriously affect the prognosis [
4].
Previous studies have attempted to determine risk factors that would predict postoperative complications [
5‐
12]. To identify the clinical factors can be more useful in preventing postoperative complications. And, this information regarding the management of adult onset MMD is important [
13]. However, the risk factors of postoperative complications were still unclear. Therefore, it is particularly important to identify these risk factors so as to ensure the efficacy of surgical treatment. Thus we conducted this systematic review with meta-analysis to clarify the risk factors for postoperative stroke in adult patients with MMD.
Methods
This systematic review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA) criteria [
14]. Ethical approval is not required by our institution for secondary research using published scientific studies.
Data sources and searches
We comprehensively searched MEDLINE/PubMed, Web of Science, and Cochrane Library for eligible published literature with regard to the risk factors and postoperative complications in adult patients with MMD. The relevant studies were included until Dec 2018. The key words used in the search included “moyamoya disease”, “risk factors”, “revascularization”, “postoperative complications “postoperative ischemic” and “hemorrhagic events”. We also manually searched the reference lists of all accepted papers so that no studies were overlooked.
Study selection
Initially, we identified all possible preoperative risk factors of postoperative complications on univariate and multivariate analysis. We then restricted the systematic review to seven preoperative risk factors, which were the most consistent and amenable to analysis: male sex, age at onset, preoperative ischemic events, past medical history, posterior cerebral artery (PCA) involvement, Suzuki stage [
1], and surgical type.
Two investigators independently reviewed abstracts and full-text articles against inclusion and exclusion criteria. Disagreements were resolved through discussion or consultation with a third investigator. Inclusion criteria were as follows: 1) Postoperative complications are related to ischemic or hemorrhage events; 2) Study design was retrospective or prospective observational study; 3) Studies reported odds ratio (OR) or hazard ratio (HR) with 95% confidence interval (CI); 4) Treatment of patients must be surgery; 5) Quality score > 5. Studies were excluded if they were not clinical study; if postoperative complications were not ischemia events or intracranial hemorrhage.
Data extraction and quality assessment
One investigator abstracted data from the included studies, and a second investigator checked data for accuracy. We abstracted study design detail, patients’ characteristics, adverse postoperative events and preoperative possible risk factors.
Preoperative ischemic event included transient ischemic attacks (TIAs) and infarction. And in our study postoperative complications included hemorrhage events and/or ischemic events, which were not separate statistics in some research. Therefore, to unify the studies, we defined stroke as a new neurologic deficit with radiographic correlation, including ischemic stroke (IS) and hemorrhagic stroke (HS). And past medical history in this study mainly contained hypertension and diabetes. Surgery types were categorized into direct bypass (DB), indirect bypass (IB) and combined bypass (CB). The direct bypass meant superficial temporal artery to middle cerebral artery (STA-MCA) anastomosis. The indirect bypass included encephalo-myo-synangiosis (EMS), encephaloduroarteriosynangiosis (EDAS) and pial synangiosis. And the combined bypass included direct bypass and indirect bypass. PCA involvement and Suzuki stage were confirmed with preoperative digital subtraction angiography (DSA) and/or MR angiography (MRA).
Two investigators independently assessed the quality of the included studies by using the Newcastle–Ottawa Scale (NOS) [
15]. The NOS allocates a maximum of 9 points to each of 3 categories: 1) patient selection (3 items), 2) comparability of the 2 study arms (2 items), and 3) assessment of outcome (2 items). Each studies was assigned a final quality rating of good (7–9 points), fair (5–6 points), or poor (0–4 points). Disagreements among investigators were resolved through discussion or consultation with a third investigator. We excluded studies as poor quality.
Statistical analysis
Statistical analysis was conducted using Stata version 12.0 (Stata Corp). OR or HR with 95%CI were assessed for each risk factor. Heterogeneity of the studies was measured using I2 statistic. Adjusted OR from multivariate analyses were preferred, when multivariate analyses were not reported, the OR from univariate analyses was used. When I2>50% or P value<0.05 was identified for heterogeneity among studies, we used the random effect model; Otherwise, a fixed effects model was adopted. Potential publication bias examined by Begg’s rank correlation test and sensitivity analysis was also conducted. All significance testing was 2-sided, and the results were considered statistically significant at P < 0.05.
Discussion
MMD is a chronic cerebrovascular disorder. With recent advances in neuroradiological diagnostic modalities, the diagnosis of adult onset MMD has become more frequent than in the past. Postoperative stroke was common complication in patients with MMD. The previously reported rate of postoperative ischemia in MMD patients after revascularization varies from 1.5 to 11.4% [
16‐
19]. Some preoperative factors have been previously reported to be associated with increase of postoperative ischemic complications in MMD patients [
5‐
12,
16,
20‐
25]. Controversy still remains in the literature regarding which contribute as risk factors.
In this meta-analysis, preoperative ischemic event, PCA involvement, presence of diabetes and surgical type of IB and CB were verified to be independent risk factors associated with postoperative stroke. Preoperative ischemic event was likely to be an indicator of the instability of cerebral hemodynamics before surgical revascularization [
8]. Previous studies had reported that frequent occurrence of preoperative TIAs was an important indicator of a risk for perioperative ischemic [
22] and stroke [
26] complications. Our results showed that preoperative ischemic event had a significant increase in risk for postoperative stroke. Interestingly, in the subgroup analysis, this association only exit in Caucasian, but not in Asian. That may be due to our study results included not only TIAs but also symptomatic infarction. Our findings suggested that PCA involvement was an independent risk factor, possibly because patients with advanced stage MMD, the leptomeningeal collateral from the PCA was significantly important collateral blood flow source [
8]. Thus, PCA impaired may seriously affect cerebral hemodynamics, which may lead to postoperative cerebral ischemia. Besides, Jung et al. [
17] reported that direct bypass may result in cerebral infarction of contralateral hemisphere in patients with PCA involvement. In the past medical history of patients, several previous reports suggested that diabetes as a predictor of recurrent stroke [
27‐
29]. A recent meta-analysis reported that diabetes is an independent risk factor for stroke recurrence [
30]. In context of MMD, patients with diabetes may elevate expression of growth factors and cytokines, such as hepatocyte growth factor, transforming growth factor-β, vascular endothelial growth factor, and nitrotyrosine, which could lead to more collateral angiogenesis [
31]. Though our results support this hypothesis, it should be noted that only three included studies considered diabetes as a risk factor. For surgery types, a recent meta-analysis proves that direct bypass could reduce the risk of perioperative stroke than indirect bypass in MMD [
32]. Notably, we also found that the risk of postoperative stroke is higher in MMD patients who underwent indirect bypass and combined bypass compared with direct bypass. However, due to sensitivity analysis unstable (the result become insignificant when excluded Wonhyoung et al. [
8]), that conclusion should be interpreted cautiously.
Patients with higher Suzuki stage probably had poor collateralization pathways to compensate for the hemodynamic impairment [
5]. However, in the current study, sensitivity analysis of the association between Suzuki stage and recurrent stroke was instable, the result become significant after exclude the Xiangyang et al. [
12] We also found that older age of symptom onset was identified as a possible predictor of postoperative strokes. This may be because older patients are often accompanied by some underlying diseases, such as diabetes, arteriosclerosis, and hypertension and so on, these may aggravate postoperative complications. Therefore, further studies with a large sample size are still needed to confirm these factors.
There are some limitations in this meta-analysis. First, our research included three prospective studies, which reported HR value of different preoperative risk factors. Thus, pooled HR value of risk factors for postoperative stroke could not be obtained. Second, among different ethnic groups, there may be differences in postoperative stroke [
33], so racial differences may exist. However, in the current study, there is not enough data to explore the racial differences in other risk factors. Third, heterogeneity exists in the meta-analysis results of ischemic events and PCA involvement, it could be inter-institutional differences, such as clinical experience, diagnostic standard, operative techniques, and surgeon’s preferences, however, we did not find the cause of heterogeneity. Fourth, due to the limitation of data included in the studies, we only compared between DB and IB/CB (DB as a reference), so further research is still needed to determine the results with IB and CB. In addition, hyperperfusion syndrome was barely mentioned in the included studies, so it is impossible to analyze the influence of factors on hyperperfusion syndrome. Lastly, sensitivity analysis of surgery types and Suzuki stage are instable. Wonhyoung et al. [
8] and Xiangyang et al. [
12] have a serious impact on the results, respectively.