Apolipoprotein E ε4 Polymorphism as a Risk Factor for Ischemic Stroke: A Systematic Review and Meta-Analysis

Introduction Rising studies indicate that the apolipoprotein E (APOE) gene is related to the susceptibility of ischemic stroke (IS). However, certain consensus is limited by the lack of a large sample size of researches. This meta-analysis was performed to explore the potential association between the APOE gene and IS. Methods To identify relevant case control studies in English publications by October 2020, we searched PubMed, Embase, Web of Science, and the Cochrane Library. Pooled odds ratios (ORs) with fixed- or random-effect models and corresponding 95% confidence intervals (CIs) were calculated to analyze potential associations. Results A total of 55 researches from 32 countries containing 12207 IS cases and 27742 controls were included. The association between APOE gene ε4 mutation and IS was confirmed (ε4 vs. ε3 allele: pooled OR = 1.374, 95% CI, 1.214-1.556; ε2/ε4 vs. ε3/ε3: pooled OR = 1.233, 95% CI, 1.056-1.440; ε3/ε4 vs. ε3/ε3: pooled OR = 1.340, 95% CI, 1.165-1.542; ε4/ε4 vs. ε3/ε3: pooled OR = 1.833, 95% CI, 1.542-2.179; and APOE ε4 carriers vs. non-ε4 carriers: pooled OR = 1.377; 95% CI, 1.203-1.576). Interestingly, APOE ε4 mutation showed a dose-response correlation with IS risk (ε4/ε4 vs. ε2/ε4: pooled OR = 1.625; 95% CI, 1.281-2.060; ε4/ε4 vs. ε3/ε4: pooled OR = 1.301; 95% CI, 1.077-1.571). Similar conclusions were drawn in the small artery disease (SAD) subtype, but not in large artery atherosclerosis (LAA) or in cardioaortic embolism (CE), by subgroup analysis. Conclusions These observations reveal that specific APOE ε4 mutation was significantly associated with the risk of IS in a dose-dependent manner, while APOE ε4 mutation was related to SAD subtype onset without a cumulative effect.


Introduction
Ischemic stroke (IS) is a disturbing problem worldwide, which is attributable to its leading role in disability and mortality worldwide, regardless of age, ethnicity, or gender [1]. Uncovering the etiology of IS is crucial for recognition and prevention of this disorder. Genetic elements and environmental components positively contribute to this multifactorial disease [2,3]. Genetic inheritance provides a guide to the identification of high-risk individual. It deserves to investigate candidate gene polymorphisms in IS pathophysiological pathways. The apolipoprotein E (APOE) gene locates on chromosome 19q13.2. Two single polymorphisms (rs7412 and rs729358), three common alleles (ε2, ε3, and ε4), and six genotypes (ε2/ε2, ε2/ε3, ε2/ε4, ε3/ε3, ε3/ε4, and ε4/ε4) generate in populations [4]. The product of the APOE gene is a polymorphic protein named apolipoprotein E, which modulates the translocation of the cholesterol and other lipids among highly diverse cells [5], involved with neuroinflammation [6] and myelin integrity maintenance [7]. A study indicated that the activated CypA-MMP9 pathway in APOE4 carriers facilitated pericyte injury, which caused blood vessel dysfunction [8]. APOE polymorphisms and its risk associations with coronary artery disease [9], hypertension [10], diabetes [11], and carotid arterial atherosclerosis [12] are widely debated. The abovementioned diseases place individuals at a potential serious risk of IS. Individual studies of the association between IS and APOE polymorphisms have been explored extensively. Clinical differences, ethnic diversities, and small sample sizes restricted the present finding to an inconsistent and controversial one. Previous meta-analyses concerning to this issue have been published several years ago [13] or limited to specific ethnicity [14,15]. Accordingly, researches from 32 countries are qualified to form our meta-analysis to clarify how APOE genotypes are associated with IS. Moreover, we firstly revealed the correlation of the APOE gene and three IS subtypes (large artery atherosclerosis (LAA), small artery disease (SAD), and cardioaortic embolism (CE)).

Materials and Methods
We followed the rules of the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement to make this meta-analysis [16].
2.1. Data Availability. The data that contribute to the findings in our study are available and the corresponding authors can be contacted for data access.

Literature Search.
Online databases (PubMed, Embase, Web of Science, and the Cochrane Library) were comprehensively searched for studies potentially involved and published in English publications and prior to October 30, 2020. We used a combination of some search terms relevant for IS (stroke, cerebral infarct, brain infarct, ischemic stroke, cerebral ischemia, transient ischemic attack, and cerebrovascular accident) and for the APOE gene (apolipoprotein E, APOE polymorphisms, apolipoprotein E polymorphisms, apolipoprotein E gene, rs429358, rs7412, apolipoprotein E epsilon 4, APOE e4, apolipoprotein E epsilon 2, and APOE e2). The detailed search strategies were showed next.

Selection Criteria.
The selection of the studies was independently completed by two investigators, and any difference was resolved by discussion until an agreement was reached. We carefully selected case control studies that evaluated the relationship of the APOE gene and IS with definite IS diagnoses (using computed tomography, magnetic resonance, or autopsy) regardless of the ethnic background. The detailed inclusion criteria were (1) high-quality studies which explore the relationship between the APOE gene and IS, (2) explicit IS diagnostic criteria, (3) nonstroke individuals as the control group, and (4) original data including independent and sufficient APOE genotype data, to compute ORs and 95% CIs. The newest and largest studies were chosen to avoid duplicate or overlapped data information.

Data Extraction. Two investigators separately finished
full-text reading to extract the needed information from each selected study and resolved the controversial items through serious discussion. The extracted information was (1) research characteristics, including the first author's name, year of publication, and geographical location of the study; (2) participant details, such as the sex ratio, mean age, and the sample size of case and control groups; (3) diagnostic criteria for IS; (4) determination methods of the APOE gene; (5) each genotype frequency; (6) the sample sizes of IS subtypes according to TOAST norms and respective genotype frequency; and (7) HWE in controls.

Quality Assessment.
We performed the quality assessment through the Newcastle-Ottawa Scale (NOS) score considering selection, comparability, and exposure. It ranged from 0 (worst) to 9 (best) and high-quality studies were known as with a NOS score ≥ 7.
2.7. The Result of Trial Sequential Analysis (TSA). Insufficient sample size, continuous updating, and repeating " significance testing" could increase the risk of type I errors. Therefore, traditional meta-analysis that focuses on the specific topic may suffer an increased risk of random error. Trial sequential analysis (TSA) was used to reduce the risk of type I error and obtain important information regarding the required sample size for such trials. Set the time sequence of a single study as the research node, and then, perform an interim analysis between the new study that will be included in meta-analysis and existing data accumulation. The required information size (RIS), trial sequential monitoring boundary, and futility boundary are estimated using the TSA. As the sample size of meta-analysis reaching the RIS or the z-curve crossing the trial sequential monitoring boundary, we can conclude that the results of metaanalysis are quite stable and further studies were not needed. We accomplished TSA following the guidelines of the user manual and previous article [18] by setting a

Characteristics of Eligible Studies.
We collect a total of 55 studies from 32 countries containing 12207 IS cases and 27742 controls to make the meta-analysis . Figure 1 showed the detailed selection process. The selected studies and their main characteristics were exhibited in Table 1. Fifteen of the studies provided data about different subtypes (grouped by classification of cerebrovascular diseases III or TOAST classification) of IS: large artery atherosclerosis (LAA), small artery disease (SAD), and cardioaortic embolism (CE). We extracted them independently and specific information was showed in supplementary material table 1. There were seven studies (Koopal et al. 2016, Lai et al. 2007, Chowdhury et al. 2001, Kokubo et al. 2000, Ji et al. 1998, Couderc et al. 1993, Saidi et al. 2009) which deviated HWE obviously, and one study (Schneider et al. 2005) did not contain enough data to obtain HWE. Forty-eight studies used PCR-based method and seven researches (Slowik et al. 2003, Karttunen et al. 2002, Hachinski et al. 1996, Couderc et al. 1993, Brewin et al. 2020, Aalto-Setala et al. 1998, Schneider et al. 2005) used other methods to identify APOE genotypes. These studies used computed tomography or magnetic resonance to diagnose IS except that one research which used autopsy (Schneider et al. 2005). The NOS score mean value was 7.509, which suggested that the quality of included studies was reliable (supplementary material Table 2). PRISMA2020 checklist was provided to present our meta-analysis items (supplementary material Table 3).
3.7. The Result of Trial Sequential Analysis (TSA). The RIS was 8901 samples and the sample size of our meta-analysis reached it. Moreover, the cumulative z-curve crossed the trial sequential monitoring boundary before reaching the RIS as showed in Figure 4. The result of TSA guaranteed the stability of our meta-analysis results. Our sample size was proved to be enough for evaluating the relationship between APOE polymorphisms and IS risk.

Discussion
Recently, scholars explored more how gene polymorphisms were contributing to the occurrence and prognosis of diseases. And several previous publications had well explored how gene polymorphisms related to diseases onset and potential mechanisms [74,75]. As a heterogeneous multifactorial disorder, ischemic stroke could be regulated by certain gene synthesis and specific gene products. The genes involved in the pathological process of stroke are also worth of attention. Apolipoprotein E has been proven to affect atherosclerosis, neurodegeneration, and the process of nerve damage repair. That is why we explored the relationship between APOE gene polymorphisms and ischemic stroke risk. APOE is a 299-amino acid protein encoded by the APOE gene of three common polymorphisms, ε2, ε3, and ε4. The correlation of APOE gene polymorphisms and the risk of cerebral vascular and degenerative diseases have been investigated a lot, especially in Alzheimer's disease (AD) and cerebral amyloid angiopathy (CAA) [76]. APOE ε4 is associated with increased risk for AD whereas APOE ε2 is associated with decreased risk [77]. Mirza   25 Disease Markers the APOE ε2 allele might be associated with the pathophysiology and severity of cortical superficial siderosis in CAA [79]. As to IS, there existed quite many researches with inconsistent conclusions. Besides method differences, ethnic difference and unclarified pathophysiological mechanisms are probable reasons of the inconsistency.
There are tremendous researches and discussions focusing on the pathogenicity of ε4. An Indian research reported that VLDL and triglycerides levels were found to be significantly associated with ε2/ε4 and ε3/ε4 genotypes; the ε4 allele exerted a higher influence than the ε3 allele in plasma cholesterol levels [22]. As a lipid transport protein, APOE3 and APOE2 preferentially bind to the smaller, more phospholipid-enriched high-density lipoproteins (HDL), while APOE4 preferentially binds to the larger, triglyceride-rich very low-density lipoproteins (VLDL). Miyata and Smith demonstrated an antioxidant activity in the order APOE2 > E3 > E4, and other researchers also reported similar results that APOE4 was associated with increased oxidative stress [25,80], which might play a role in atherosclerosis and lead to increased risk of ischemic vascular diseases. Besides the above reasons, APOE4 was proved to be neurotoxic by assuming an abnormal conformation (the unique domain interaction between Arg-61 and Glu-255) which was highly susceptible to neuron specific proteolysis and generating neurotoxic fragments that escaped the secretory pathway and entered the cytosol [81]. Totally, from pathophysiological mechanisms to clinical research results, it seems that APOE4 is indeed related to a higher risk of IS, compared with other isoforms, both in ε4 heterozygote and homozygous. ε2 allele appears to be unclear and controversial in stroke [13]. In a meta-analysis of Martínez-González et al., compared with ε3/ε3, APOE ε2 was associated with intracerebral hemorrhage (OR = 1:32; 95% CI, 1.01-1.74); meanwhile, APOE ε2 was more related to lobar hemorrhage than deep hemorrhage [82]. As to the association of IS with APOE based on previous investigation, it is uncertain. Our estimates showed that both ε2/ε2 and ε2/ε3 genotypes exhibited no significant effects on IS risk, compared with ε3/ε3. Also, no differences were found in comparisons of ε2 allele vs. ε3 allele and ε2 vs. non-ε2 carriers. This result remained consistent with another meta-analysis in 2013 [14]. Interestingly, in subtype analysis, ε2/ε2 displayed significances in the CE group (OR = 4:290; 95% CI, 1.917-9.600; P < 0:0001) and SAD group (OR = 1:803; 95% CI, 1.037-3.134; P = 0:04). The largest meta-analysis of the APOE genotype with IS showed a positive linear association of increasing risk when ordered from ε2/ε2, ε2/ε3, ε2/ε4, ε3/ε3, ε3/ε4, and ε4/ε4 in European ancestry population [83]. The conclusion might explain why APOE4 brings a higher risk of IS but could not clarify that the CE and SAD subgroups in comparison of ε2/ ε2 with ε3/ε3 show significances. It is well known that all patients with type III hyperlipidemia (dysbetalipoproteinemia) were APOE ε2 homozygous, whereas most ε2/ ε2 subjects (>90%) were normolipidemic or even hypolipidemic, owing to reductions in LDL or HDL or both. Therefore, the APOE ε2 allele has both increased and decreased risks for atherosclerosis, which induced a comprehensive and undetermined result [84].
As to our subtype analyses, all LAA groups showed no significant difference among comparisons, which raised a question why isoforms of APOE, a lipid transport protein, seemed not to be related with IS caused by large artery atherosclerosis. Besides lipid metabolism and atherosclerosis, there might exist some other pathways underlying the relationships between APOE and risk of IS. Our estimates displayed that APOE isoforms were associated to risk of IS especially in the SAD subgroup. Hypertension was known to be an independent risk factor of SAD. Atherosclerosis, dyslipidemia, and hypertension have a complex interaction, and the causations with APOE need further investigation.
Our meta-analysis has several limitations. First, just as the abovementioned, heterogeneity between studies remains undeterminable. Second, results of our meta-analysis based on case control studies cannot provide a causal relationship, but only an association. Third, age variable and ethnicity can influence APOE frequencies in a population; we cannot obtain sufficient related information to perform further subdivided subgroup analyses. Fourth, other pathogenic factors about IS, a multifactorial disease, such as plasma lipid levels, hypertension, life-style, BMI, and gene-environment interactions, were unachievable. Fifth, the controls in accessible studies were not strictly defined; some were selected from healthy populations and others were from nonstroke people. The expected genotype distribution in controls was not in accordance with HWE in seven studies. Population selection in control groups failed to avoid certain diseases which might have a relation with the APOE gene, such as dyslipidemia, hypertension, other vascular diseases, and diabetes. Sixth, the case groups were not selected by a prospective process and the design of case control studies often caused abnormal gene frequency. 26 Disease Markers

Conclusions
In conclusion, our meta-analysis provides rational evidence that APOE ε4 mutation is a genetic risk factor for IS. Prospective studies of a large sample size, which concerns gene-gene and gene-environment interactions, should be carried out in the future to reach a more comprehensive outcome about the association of APOE gene polymorphisms and IS. What is more, future researches should be designed to elucidate the mechanism by which APOE ε4 mutation adds the risk of IS.

Data Availability
Data presented within the paper and the supplementary materials contributed to the findings in our study. They are all are available from our corresponding author for reasonable request.