Association between miR-499 rs3746444 polymorphism and ischemic stroke : A systematic review and meta-analysis

Background: To explore the generic association between miR-499 rs3746444 polymorphism and ischemic stroke (IS). Methods: We performed a systematic review and meta-analysis, odds ratio (OR) and 95% condence intervals (CIs) were used to estimate the association quantitively. Results: A total of 6 studies (involving 2569 IS cases and 2645 controls) were included. miR-499 (rs3746444) polymorphism showed a statistically signicant association with IS risk in the allelic model (G allele vs A allele), the dominant model (GG+AG vs AA), the recessive model (GG vs AG+AA) and the homozygote model (GG vs AA). ORs of the above 4 models were 1.20 (95%CI: 1.02, 1.40), 1.21 (95%CI: 1.01, 1.46), 1.40 (95%CI: 1.04, 1.88), 1.48 (95%CI: 1.10, 2.00) respectively. The I square of the allelic model and the dominant model was 58% and 59%, indicating large heterogeneity among included studies. By sensitivity analysis, I square of the two models dropped to 34.5% and 38.4%, the ORs were 1.26 (95%CI: 1.13, 1.42) and 1.28 (95%CI: 1.12, 1.46), there was still a statistical association between miR-499 (rs3746444) polymorphism and IS. The heterozygote model (AG vs AA) was not statistically signicant, the OR was 1.18 (95%CI: 0.99, 1.42), the I square was 54%. Notably, by sensitivity analysis, I square of the heterozygote model dropped to 34.6%, the OR was 1.25 (95%CI: 1.08, 1.43), indicating a statistically signicant association between miR-499 (rs3746444) polymorphism and IS. There was no publication bias for all the models by Egger's test. Conclusion: miR-499

Background microRNAs (miRNAs) play an important role in gene regulation, more than half of human genes are adjusted by miRNAs. miRNAs also in uence many pathological pathways and process, like in ammatory response [1,2] , cancer development [3,4] , as well as ischemic stroke (IS) formation.
Recently in 2010s, studies on miR-499 (rs3746444) polymorphism sprang up. However, the evidence from different studies was con ict with each other. Jeon YJ [10] found no difference on the allele frequencies of miR-499 polymorphism between IS patients and the control. Xiao Y [11] reported a similar nding by a meta-analysis based on 3 studies. On the contrary, Darabi H [12] reported that miR-499 A/G polymorphism was statistically associated with the increased risk of IS in both the homozygote model and the heterozygote model [12] . In a review, Zhu J [13] reported that homozygote AA was associated with a decreased risk of IS in the homozygote model based on 3 articles.
In this systematic review and meta-analysis, we aim to provide an updated assessment on the association between miR-499 polymorphism and IS risk. Five models including the allelic model (G allele vs A allele), the dominant model (GG+AG vs AA), the recessive model (GG vs AG+AA), the heterozygote model (AG vs AA) and the homozygote model (GG vs AA) would be applied to make quantitive assessment.

Inclusion and exclusion criteria
Studies were selected under the following consideration: (1) The study used a case-control design; (2) The subjects should be IS patients and controls; (3) There must be extractable data on allele and genotype frequencies; (4) The subjects should be human instead of animal models.
Besides, duplication studies and irrelevant studies were excluded.

Databases and Search strategy
Pubmed, Science Direct, and google scholar were searched from inception to September 15, 2019. The following terms were used: rs3746444, miRNA-499, microRNA-499, miR-499; with combination of stroke, ischemic, ischemia, ischemic stroke, cerebro vascular, cerebrovascular, cerebrovascular disease. Two researchers extracted data independently, the following information was extracted: (1) The rst author; (2) Publication year; (3) Study region (4) Participants; (5) Sample size of the case group and the control group; (6) Gene sequencing method; (7) Frequencies of genotypes in the case group and the control group; (8) Application of gender and age match; (9) p-value of Hardy-Weinberg equilibrium (HWE) of the control group.

Assessment of methodological quality
To assess the quality of the studies included in this analysis, the Newcastle-Ottawa scale [14] was used by two independent assessors. For the Newcastle-Ottawa scale, the full score is nine; a score of 5 to 9 indicates high methodological quality, whereas a score of 0 to 4 indicates poor quality. Any disagreements about Newcastle-Ottawa scores were resolved by other authors following a comprehensive reassessment. Low quality studies would be excluded in our meta-analysis.

Data synthesis and statistical analysis
We used odds ratios (ORs) with its 95% con dence interval (CI) to estimate categorical outcomes. The heterogeneity across studies was assessed using Q test and I 2 statistic. I 2 describes the percentage of variability caused by heterogeneity rather than by chance. An I 2 of below 50% indicates a small heterogeneity and consequently a xed-effects model would be used to perform meta-analysis, otherwise a randomeffects model would be used, followed by a sensitivity analysis [14] . We further used the Egger's test [15] to check for publication bias of each model. The signi cance level was 0.05, two-tailed. All the analysis was done using the open source R program (Version 3.6.0).

Results
A total of six studies [10,12,[16][17][18][19] , involving 2569 IS cases and 2645 controls, were included in this meta-analysis. The characteristics of the included studies was shown in table1. Four studies were carried out in Asia, one in Iran and one in South-Korea. All studies enrolled patients in hospital. Three different genotyping methods including Taqman, PCR-RFLP and RT-PCR were used in these studies. All studies matched age and gender between the IS cases and controls. Only one study reported a p-value of HWE below 0.05 in the controls.
The overall quality of the included studies was adequate, and the mean Newcastle-Ottawa score for the included studies was 7 (Table2).
The allelic model showed a statistically signi cant association between miR-499 (rs3746444) polymorphism and IS risk ( gure2), the OR of the allelic model was 1.20 (95%CI: 1.02, 1.40). However, the I 2 was 58%, indicating a large heterogeneity among included studies, thus a sensitivity analysis was further done, it turned out that after omitting study of Jeon YJ 2013 [10] , the I 2 dropped to 34.5%, and the conclusion remained unchanged, with an OR of 1.26 (95%CI: 1.13, 1.42).
The dominant model also showed a statistically signi cant association between miR-499 (rs3746444) polymorphism and IS risk ( gure3), the OR of the dominant model was 1.21 (95%CI: 1.01, 1.46). However, the I 2 was 59%, indicating a large heterogeneity among included studies, thus a sensitivity analysis was further done, it turned out that after omitting study of Jeon YJ 2013 [10] , the I 2 dropped to 38.4%, and the conclusion remained unchanged, with an OR of 1.28 (95%CI: 1.12, 1.46).
The heterozygote model was not statistically signi cant ( gure5), the OR was 1.18 (95%CI: 0.99, 1.42). However, the I 2 was 54%, indicating a large heterogeneity among included studies, thus a sensitivity analysis was further done, it turned out that after omitting study of Jeon YJ 2013 [10] , the I 2 dropped to 34.6%. Notably, the conclusion changed after omitting study of Jeon YJ 2013 [10] , with an OR of 1.25 (95%CI: 1.08, 1.43).
There was no publication bias for all the above models by Egger's test (Table3).

Discussion
Our study is a meta-analysis aimed speci cally at investigating the association between miR-499 polymorphism and IS risk based on casecontrol studies. A previous review by Zou D and coworker [20] concluded that miR-499 (rs3746444) polymorphisms may not be associated with IS risk in Asian populations. To our surprise, in Zou D's study, the OR was non-signi cant in any of the allelic model, the dominant model, the recessive model, the homozygote model or the heterozygote. And due to that three articles in Chinese were included in Zou D's review, it's hard for other researchers to repeat the result.
Our main nding is that miR-499 polymorphism is associated with IS risk, the association was statistically signi cant in the allelic model, the dominant model, the recessive model and the homozygote model. Besides, the heterozygote model also showed a statistically signi cant association by sensitivity analysis. The association between miR-499 polymorphism and IS risk is explainable: it was reported that hsa-miR-499 played a role in cell apoptosis under ischemia condition [21] , the mechanism might be that miR-499 inhibits cell apoptosis by suppressing dephosphorylation of dynamin-related protein-1(Drp1) that mediated by calcineurin. miR-499 also affects the regulation of C-reactive protein [22] , which was reported to be a cause of cerebral ischemia [23] .
There are two limitations of our meta-analysis. On one aspect, all the included studies used a case-control study design, indicating that there was selection bias and the interpretation of the results might be in uenced. On the other hand, we did have a large sample size compared to single studies, however, due to that most studies were carried out in China, it was hard to perform subgroup analysis by ethnicity except for Chinese population. By the single study of Jeon YJ 2013 [10] based on South-Korea population, the association between miR-499 polymorphism and IS risk was non-signi cant in any model; by the single study of Darabi H 2018 [12] based on Iran population, the association was statistically signi cant in all the models. The opposite ndings indicate that there might be ethic difference of miR-499 polymorphism's effect on IS.
Notably, the ethnic difference could explain the heterogeneity among the included studies in our meta-analysis, because no matter for the allelic model, the dominant model, or the heterozygote model, the heterogeneity dropped sharply once the study of Jeon YJ 2013 [10] was omitted by sensitivity analysis. What's more, the non-signi cant heterozygote model changed to be signi cant after omitting study of Jeon YJ 2013 by sensitivity analysis. In the future, an updated meta-analysis providing subgroups analysis by ethnicity is needed.
In conclusion, miR-499 polymorphism is related with IS, the G allele increases the risk of IS.

Declarations
Ethics approval and consent to participate: Not applicable Notes: a When there was no significant difference in the response rate between both groups based on a chi-squared test, one point was awarded. b Total score was calculated by adding up the points awarded in each item.  Figure 1 Flow chart of paper selection Figure 2 Forest plot of the allelic model Forest plot of the recessive model Figure 5 Forest plot of the heterozygote model Figure 6 Forest plot of the homozygote model

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Appendix5Subgroupanalysisofthehomozygotemodel.tiff Appendix1Subgroupanalysisoftheallelicmodel.tiff Appendix4Subgroupanalysisoftheheterozygotemodel.tiff Appendix3Subgroupanalysisoftherecessivemodel.tiff Appendix2Subgroupanalysisofthedominantmodel.tiff