Ischemic stroke (IS) is a most common cause of death and long-term disability which accounting for approximately 85% of all types strokes (Kotlęga et al. 2017). Being one of the most devastating consequences of atherosclerotic disease, ischemic stroke can be caused by any or the combination of multifactorial disease, including age, hypertension, diabetes mellitus, smoking, obesity and so on (Munshi and Kaul 2008). Of all the risk factors, genetic susceptibility to ischemic stroke was considerate to play a vital role in the development of IS since first shown in twin pedigree studies in 1992 (Bak et al. 2002), however, until now, the genetic pathogenesis of stroke remains unclear.

Phosphodiesterase 4D (PDE4D) is a member of the superfamily of cyclic nucleotide phosphodiesterases and is involved in degradation of cyclic adenosine monophosphate (cAMP) which is a key signaling molecule involved in the inflammatory responses of vascular cells (Das et al. 2016). Researches showed that inhibitors of PDE4D can increase cAMP levels and adhesion in vascular endothelial cells and decrease migration of vascular smooth muscle cells (Jorgensen et al. 2015). Thus it had been postulated to contribute to vascular diseases in the pathogenesis of atherosclerosis. Since Gretarsdottir et al. identified SNP83 (rs966221) in PDE4D as a susceptible gene for ischemic stroke in Caucasian population, a large number of studies reported the association of single nucleotide polymorphisms across this gene with the disease (Gretarsdottir et al. 2003). Xue et al. had also identified that SNP83 as a genetic risk factor for atherothrombotic strokes in a Chinese population (Xue et al. 2009). Nevertheless, results have been contradictory. Quarta et al. reported that there were no evidences of association between SNP83 and IS in a genetically homogenous population from Sardinia (Quarta et al. 2009). Meantime, Matsushita et al. also reported the same results in Japanese cohort (Matsushita et al. 2009). Therefore, to confirm the association between SNP83 and the risk of IS, we performed a meta-analysis of case-control studies by pooling all eligible studies to evaluate the overall risk and influence of ethnic factors to this disease.

Materials and methods

To identify all relevant publications focus on the risk for Is and SNP83 polymorphism, we conducted a comprehensive literature search of electronic databases, including the Pubmed, Embase, China National Knowledge Infrastructure (CNKI) and Wan Fang Data. Eligible case-control studies were extracted with the last search update on April 1, 2017. The following terms were used: “ischemic infarction OR cerebrovascular disease OR stroke OR cerebrovascular disorders OR ischemic stroke” in combination with “PDE4D OR phosphodiesterase 4D OR rs966221 OR SNP83” in combination with “polymorphism OR variant OR mutation”. The reference lists of retrieved studies and recent reviews were also manually searched for further relevant studies. In addition, if the genotype data for the SNP83 polymorphism was not illuminated in the original studies, we send an email to obtain full data for the meta-analysis.

Inclusion and exclusion criteria

Studies were considered eligible if they met the following criteria: (1) evaluation of the association between PDE4D SNP83 and risk of ischemia stroke; (2) the study provided sufficient information of allele or genotype frequencies; (3) the study was a case-control study; Exclusion criteria: (1) duplication of previous publications; (2) comment, review or editorial; (3) study without detailed genotype data. A study reporting the results for different subpopulations was treated as separate studies.

Data extraction

The data of the eligible studies were extracted in duplicate by two investigators (Peng Wang and Fei Yang) independently with a standard data-collection form. The following data was collected from each study if available: (1) first author’s name; (2) country of origin; (3) years of publication; (4) ethnicity of the individuals (categorized as Caucasians or Asians); (5) Gender and mean age in cases and controls; (6) Hardy-Winberg equilibrium; (7) numbers of cases and controls; (8) genotyping method; (9) counts of cases and controls for each genotype. If there were multiple publications from the same group, only the one with largest study was included. The discrepancies were resolved through discussion. If the dissent still existed, the third investigators would be invited to resolve the dispute.

Quality assessment

The quality of the included studies was evaluated by using the quality scoring criteria modified from Newcastle–Ottawa scale (NOS) for genetic association studies (Zhang et al. 2017). Total score of quality assessment ranged from zero to nine stars, and six or more stars are rated as high quality. Disagreement was settled through discussion among the investigators.

Statistics analysis

Hardy-Weinberg equilibrium was evaluated for each study by chi-square test in the control groups and P value <0.05 was considered significant disequilibrium. The odds ratio (OR) and 95% confidence interval (95% CI) were calculated to evaluate the strength of the association between SNP83 polymorphisms and susceptibility to IS. In the overall and subgroup meta-analysis, Pooled ORs were obtained from combination of single study for the dominant model (CC + CT vs. TT), recessive model (CC vs. CT + TT), homozygote model (CC vs. TT), heterozygote model (CT vs. TT) and allelic model (C vs. T). Heterogeneity was evaluated by Q statistic (significance level of p < 0.1) and I2 statistic (greater than 50% as evidence of significant inconsistency). Once, Q-test >0.10 or I2 < 50%, the fixed effect model (Mantel–Haenszelmethod) was used to calculate the pooled ORs, otherwise, the random-effect model (DerSimonian–Laird method) was employed in the study. Subgroup analysis was conducted according to the different ethnicities in the study population. Publication bias was estimated by Begg’s funnel plot. Sensitivity analyses were performed to evaluate the effect of each study on the combined ORs. All statistical analyses were performed with the software Stata 12.0 (STATA Corporation, College Station, TX, USA).

Results

Characteristics of studies

A flowchart of the literature search is presented in Fig. 1. After a preliminary online search, 175 potentially relevant articles were identified for further detailed evaluation. Among which 23 duplicated citations were removed, and 152 citations were included for further review by title and abstract screening. Finally, 36 articles were selected for full-text review and 26 cautions were enrolled in the analysis. Of all the studies, 7 were Caucasians and 19 were Chinese population. These studies were published between 2003 and 2015. Multiple genotyping methods were performed in the studies, including PCR-RFLP, TaqMan, DNA sequencing and so on.

Fig. 1
figure 1

The flow sheet of identification of eligible studies

Full size image

A total of 9832 patients were from IS case group and 13,354 were enrolled in control group. The distribution of genotypes in the controls of all studies was consistent with Hardy-Weinberg equilibrium (HWE) except 6 studies. The NOS results showed that the score ranged from 3 to 8 which indicated that the methodological quality of these selected articles was generally good. The characteristics of involved articles were summarized in Tables 1 and 2.

Table 1 Distribution of SNP83 genotype
Full size table
Table 2 Characteristics of the studies in this meta-analysis
Full size table

Meta-analysis results

The heterogeneity was identified by Q-test and I-squared statistic among genetic models. As is showed in the Fig. 2, serious heterogeneity were found under dominant model (I2 = 71.5%), recessive model (I2 = 78.6%), homozygote model (I2 = 77.6%), heterozygote model (I2 = 53.4%) and allelic model (I2 = 84.2%), thus the random model was employed in the analysis. Our results reveled that there were no significant associations between SNP83 and IS under genetic model of CC + CT vs. TT (OR = 1.09, 95% CI:0.96–1.24), CC vs. CT + TT (OR = 1.1,95% CI: 0.91–1.34), CC vs. TT (OR = 1.11, 95% CI: 0.88–1.4), CT vs. TT (OR = 1.08, 95% CI:0.97–1.20) and C vs. T (OR = 1.12, 95% CI: 0.99–1.27).

Fig. 2
figure 2figure 2

Forest plots of the SNP83 polymorphism under different genetic models. A is the dominant model of CC + CT vs.TT; B is the recessive model of CC vs. CT + TT; C is the homozygote model of CC vs.TT; D is the heterozygote model of CT vs. TT; E is the allelic model of C vs. T

Full size image

Subgroups based on ethnicity were utilized to further analyze the relationship of polymorphism with IS. In Chinese populations, IS was proved to be correlated with SNP83 polymorphism under CC+ CT vs. TT (OR = 1.19, 95% CI:1.02–1.38), CT vs. TT (OR = 1.14, 95% CI:1.01–1.29) and C vs. T (OR = 1.25, 95% CI:1.06–1.48) Nevertheless, no significant association were found in CC vs. CT + TT (OR = 1.2, 95% CI:0.9–1.61) and CC vs. TT (OR = 1.26, 95% CI:0.91–1.75). In addition, we did not observed any correlation of SNP83 polymorphism with IS in all the five genetic models(CC+ CT vs. TT: OR = 0.87, 95% CI: 0.69–1.11; CC vs. CT + TT: OR = 0.95, 95% CI:0.84–1.07; CC vs.TT:0.87, 95% CI:0.66–1.13; CT vs.TT: OR = 0.9, 95% CI:0.72–1.12; C vs. T: OR = 0.92, 95% CI:0.8–1.05) in the Caucasian populations.

Sensitivity analysis

A sensitivity analysis by sequentially removing each eligible study at a time was used to assess the influence of each individual study on the pooled OR. Our results indicated that there was not any single study influence the quality of the pooled ORs (Fig. 3).

Fig. 3
figure 3

Begg’s funnel plot for publication bias analysis. A is the homozygote model and B is the heterozygote model

Full size image

Publication bias

Publication bias was evaluated using the Begg’s funnel plot. No significant evidence of publication bias was observed in the five genetic models (CC + CT vs. TT: p = 0.853; CC vs. CT + TT: p = 0.731; CC vs. TT: p = 0.544; CT vs. TT: p = 0.812; C vs. T: p = 0.692) (Fig. 4).

Fig. 4
figure 4

Sensitivity analysis examining the association between the SNP83 polymorphism and risk of IS under homozygote model and heterozygote model

Full size image

Discussion

PDE4D is a large gene spanning >1.5 Mb on chromosomal region 5q12 and has 8 splice variants, 22 exons and several hundreds of SNPs (Song et al. 2015; Song et al. 2017). It has been shown to effects on brain function, pulmonary hypertension and vascular smooth cells migration (Mika and Conti 2016). It is more of a clinical syndrome rather than a disease due to numerous clinical, genetic, and lifestyle risk factors. Work over the past few decades has shown that PDE4D are present in the brain regions such as the amygdala, prefrontal cortex, hippocampus, and nucleus accumbens (Yang et al. 2012). The distribution pattern suggests that they may serve distinct roles in the central nervous system and provide a theoretical basis for the separation of therapeutic and adverse effects of PDE4 inhibitors.

The association of specific PDE4D single-nucleotide polymorphisms (SNPs) with stroke is initially identified in an Icelandic population and further evaluated in both animal models and human beings (Liu et al. 2013). Since the first study reported the associations of PDE4D SNPs and ischemic stroke risk, an increasing amount of research has been subsequently published to study the relationship of PDE4D SNP with IS, especially SNP 83(Gretarsdottir et al. 2003). Yoon et al. retrieved 15 publications with 19,318 subjects, failed to find a linkage between SNP83 and IS in the overall population (Yoon et al. 2011). However, Yan et al. comprised of 8878 cases and 12,306 controls in a review had shown that association between SNP83 and IS in the overall population and in the Asian and Chinese populations, but not among Caucasians (Yan et al. 2014). Till now there was still not a consistency conclusion. In our study, we examined the association between SNP83 and ischemic stroke using case-control and population-based cohort. We had found that there were significant association of SNP83 with IS risk under the dominant model, heterozygote model and allelic model. Nevertheless, we didn’t observed any correlation of the polymorphism with IS in the Caucasian. Our results were in accordance with the former meta-analysis which was done by Yan et al. (Yan et al. 2014). However, other studies had also reported with opposite conclusions, such as Matsushita et al., who failed to replicate the results in Japanese populations (Matsushita et al. 2009). Whereas, Staton et al. and Gretarsdottir et al. found that SNP83 was associated with an increased risk of ischemic stroke in Caucasians (Gretarsdottir et al. 2003; Staton et al. 2006).

Although we had got the positive results in Asian population, however, the conclusions should be treated with caution. The HWE results showed that there were 6 studies with their p < 0.05 which included Chinese population studies and Caucasian populations. However, in the stratified analysis by HWE, we had observed that the results were mostly unchanged. In addition, the heterogeneity was detected under five genetic models in this meta-analysis. This could be caused by various reasons, such as the small sample size, different diagnostic criteria and disunity detection methods. Meantime, most studies (n = 19) enrolled in our analysis were done in Asian populations and there were only 7 articles explored the PDE4D polymorphism with IS in Caucasian. This can partly explain the reasons why we cannot found any correlations of SNP83 with IS in Caucasian populations. In addition, though significant difference was observed in at least 3 genetic models, the results were still worth deliberated. Most of the studies were cross with the invalid line in the analysis of dominant model, heterozygote model and allelic model, thus the significance drawn were without enough persuasion. Besides, there was a lack of sufficient data regarding the patients’ gender, environment or complicating disease such as hypertension or diabetes, which may modulate the relationship when discussing those confounding factors.

In summary, the current meta-analysis revealed that SNP83 displayed significant association with IS in Asian populations under the dominant model, heterozygote model and allelic model, which offering evidence that PDE4D may be involved in the pathogenesis of IS. Nevertheless, no correlation was observed in Caucasians. Additional, as with kinds of limitations, the relationships of SNP83 with IS susceptibility in Asian and Caucasian populations were worth further exploration. In future, well designed case-control studies with large sample sizes and stroke subtype analysis regarding the association of SNP83 with IS needed to be performed.