Association among cytotoxic T-lymphocyte antigen 4 gene, rs231775 polymorphism, and recurrent pregnancy loss risk

Cytotoxic T-lymphocyte antigen 4 (CTLA-4) is expressed constitutively on regulatory T cells. So far, several studies have focused on association between CTLA-4 gene polymorphisms and recurrent pregnancy loss (RPL). However, above association between the CTLA-4 gene polymorphism and RPL susceptibility is uncertain. Therefore, we performed a timely meta-analysis of all current publications to clarify this relationship. We located articles from the PubMed and Chinese language (WanFang) databases that were published up until July 25, 2018. Finally, we obtained six case–control studies, containing 2405 total cases and 2607 total controls, based on search criteria for abortion susceptibility related to the CTLA-4 +49 G/A polymorphism. The odds ratios (OR) and 95% confidence intervals (CIs) revealed association strengths. There was significantly decreased association between this polymorphism and whole population risk (e.g. AA vs. GG: OR = 0.56, 95% CI = 0.38–0.81, P=0.002). Additionally, in ethnicity subgroups, similar association was found both in China (e.g. AA vs. GG: OR = 0.49, 95% CI = 0.39–0.63, P=0.002) and non-China (e.g. AG vs. GG: OR = 0.46, 95% CI = 0.34–0.63, P<0.001). Current analysis suggested CTLA-4 +49 G/A polymorphism may weakly decrease RPL risk for women of childbearing age.


Introduction
A pregnancy loss (PL) is defined as the spontaneous demise of a pregnancy before the fetus reaches viability. It includes all PLs [unexplained recurrent spontaneous abortion (RSA), RSA, recurrent miscarriage (RM), idiopathic RM] from the time of conception until 24 weeks of gestation [1,2]. Approximately 15% of pregnant women experience sporadic loss, 2% experience two consecutive PL and 0.4-1% have three consecutive PL [3]. Recurrent PL (RPL), also named as recurrent spontaneous abortion (RSA), is defined as the loss of two or more pregnancies [1,2,4]. In addition, RM is classically defined as the loss of three or more consecutive pregnancies before the 20th weeks of gestation with or without previous live births [5]. In the same time, the definition of RPL and RM exists some discrepancy in opinions, broadly speaking, both are classified as PL or abortion.
A series of pathogenic mechanisms associated with PL has been described, including uterine abnormalities, endocrine and metabolic problems, genetic anomalies, acquired and inherited thrombophilia and immunological factors [6]. More and more studies have focused on the genetic factors, especially the single nucleotide polymorphism (SNP) [7].
The cytotoxic T-lymphocyte antigen 4 (CTLA-4, Gene ID: 1493, MIM number: 123890 also known as GSE, ALPS5, CD152) gene maps to band q33 of human chromosome 2, spans about 6.2 kilobases, and contains four exons and three introns [8]. It is well known that CTLA-4 expressed on human placental regulatory T (T reg ) cells in decidual and peripheral dendritic cells may induce the expression of an immune-suppressive enzyme indoleamine 2,3-dioxygenase (ID), particularly during early phases of pregnancy [9]. Furthermore, high expression of ID promotes maternal-fetal tolerance [9]. In addition, the expression of T reg cells and CTLA-4 in peripheral and decidual lymphocytes was down-regulated in human miscarriages in several in vivo studies [10]. We predicted CTLA-4 and its related T reg cells are protective factors for RPL. SNPs are known as the most common type of DNA variation in individuals [11], which may affect DNA promoter activity and influence the translation, and finally may be associated with the susceptibility about human diseases [11]. Therefore, we hypothesized that the reduced number and/or functional deficiency of T reg cells due to the genetic variations in CTLA-4 gene may increase the risk of RPL.
So far, many studies have investigated the association between CTLA-4 rs231775 G/A polymorphism (wild-type allele: A; polymorphic allele: G, 49A>G, Thr17Ala) and RPL risk. However, the results were not conclusive or consistent. Thus, we conducted this timely meta-analysis of six case-control studies to derive a more powerful estimation of the association between CTLA-4 rs231775 G/A polymorphism and RPL susceptibility [12][13][14][15][16][17].

Identification and eligibility of relevant studies
Searches were conducted in PubMed and Chinese language (WanFang) databases using the key words 'cytotoxic T-lymphocyte antigen 4 or CTLA-4', 'spontaneous abortion or miscarriage or pregnancy loss' and 'polymorphism' or 'variant' . The last search was updated on July 25, 2018. In total, 18 articles were retrieved using the abovementioned terms, and six articles contained the inclusion criteria.

Inclusion criteria and exclusion criteria
Including studies had to meet following criteria: (1) address the correlation between RPL risk and the CTLA-4 rs231775 G/A SNP; (2) be a case-control study; and (3) have sufficient numbers of genotypes (AA, AG, and GG) for both the cases and controls. The following exclusion criteria were used: (1) lack of a control population; (2) lack of available genotype frequency data; and (3) duplicated studies.

Data extraction
The following items were collected: the last name of first author, the year of publication, the country of origin, the ethnicity of subjects, source of control (SOC), the total and number of each genotype frequency in the case-control groups, the Hardy-Weinberg equilibrium (HWE) of the controls, abortion type, control-type and the genotyping method. Ethnicity was categorized as Asian, China, and non-China.

Quality score assessment
The quality score assessment (Newcastle-Ottawa Scale, NOS) [18] was selected to assess the quality of each study. This measure assesses aspects of the methodologies used in observational studies, which are related to the study quality, including selection of cases, comparability of populations, and ascertainment of exposure to risks. The NOS rating ranges from zero stars (worst) to nine stars (best). Studies with a score of seven stars or greater was considered as a high quality.

Statistical analysis
Odds ratios (OR) with 95% confidence intervals (CI) were used to measure the strength of the association between the CTLA-4 rs231775 G/A SNP and RPL risk. The statistical significance of the summary OR was determined with the Z-test. A heterogeneity assumption was evaluated among studies using a chi-square-based Q-test. A P-value of more than 0.10 for the Q-test indicated a lack of heterogeneity among the studies [19]. If significant heterogeneity was detected, the random-effects model (DerSimonian-Laird method) was used. Otherwise, the fixed-effects model (Mantel-Haenszel method) was chosen [20,21].
We investigated the relationship between genetic variants of the CTLA-4 rs231775 G/A site and RPL risk by the allelic contrast (A-allele vs. G-allele), homozygote comparison (AA vs. GG), dominant genetic model (AA+AG vs. GG), heterozygote comparison (AG vs. GG), and recessive genetic model (AA vs. AG+GG). A sensitivity analysis was performed by omitting studies, one after another, to assess the stability of results. The departure of the CTLA-4 rs231775 G/A SNP from expected frequencies under HWE was assessed in controls using the Pearson chi-square test (P < 0.05 was considered significant). Funnel plot asymmetry was assessed using Begg's test, and publication bias was assessed using Egger's test [22], both of P-value less than is considered as significant. All statistical tests were performed using STATA Software (version 11.0; StataCorp LP, College Station, TX).

Network of gene interaction of CTLA-4 gene
The network of gene-gene interaction for CTLA-4 gene was utilized through String online server (http://string-db. org/) [23].

Study characteristics
In total, 18 articles were collected from the PubMed and WanFang databases via a literature search using different combinations of key words. As shown in Figure 1, 12 articles were excluded (two were duplications, 10 were irrespective articles). Finally, six different articles were included in current meta-analysis ( Figure 1). In total, there were 2405 cases and 2607 controls. Study characteristics from the published studies on the relationship between the CTLA-4 rs231775 G/A SNP and RPL risk are summarized in Table 1. In all the studies, the controls were women under normal pregnancy. Except one study, all studies were consistent with HWE. Finally, we checked the minor allele frequency (MAF) reported for the five main worldwide populations in the 1000 Genomes Browser [23]: East Asian (EAS), 0.3631; European (EUR), 0.3588; African (AFR), 0.3880; American (AMR), 0.4625; and South Asian (SAS), 0.3098 ( Figure 2). The MAF in our analysis was 0.4033 and 0.5213 in the case and control group, respectively, both higher than the results in the EAS from1000 Genomes Browser database.

Figure 4. Forest plot of RPL risk associated with CTLA-4 gene rs231775 polymorphism (AG vs. GG) in China population
The squares and horizontal lines correspond to the study-specific OR and 95% CI. The area of the squares reflects the weight (inverse of the variance). The diamond represents the summary OR and 95% CI.

Figure 5. Forest plot of RPL risk associated with CTLA-4 gene rs231775 polymorphism (AG vs. GG) in non-China population
The squares and horizontal lines correspond to the study-specific OR and 95% CI. The area of the squares reflects the weight (inverse of the variance). The diamond represents the summary OR and 95% CI.

Sensitivity analysis and bias diagnosis
We used a sensitivity analysis to determine whether modifying the meta-analysis inclusion criteria affected the results.
No other single study influenced the summary OR qualitatively ( Figure 6). Egger's and Begg's tests were performed to assess publication bias and the funnel plot symmetry was examined. Finally, no proof of publication bias was obtained (e.g. AG vs. GG: t = −0.23, P=0.892 for Egger's test; and z = −0.19, P=0.851 for Begg's test; Figures 7 and 8, Table  3).

Gene-gene interaction of online analysis
String online server indicated that MTR gene interacts with numerous genes. The network of gene-gene interaction has been illustrated in Figure 9.

Discussion
RPL is a common pregnancy complication affecting 1-3% of couples trying to conceive. Successful pregnancy is a result of maintaining the semi-allograft fetus from maternal immune responses [24]. The decision between tolerance and immunity determines the fate of each pregnancy. A network composed of different immune cells, numerous cytokines, growth factors, and adhesion molecules collaborate to reach the best outcome [25]. T reg cells are cellular components of natural self-tolerance and seem to reduce the chance of pregnancy failure by providing the tolerant environment in the endometrium, where a successful implantation can occur [26]. Confirming evidence, in this regard, came from a significant reduction in circulating and deciduas T reg cells among women with RPL [10,27]. One of the proposed mechanisms is the CTLA-4-dependent pathway using anti-CTLA-4-mAb disrupts the T reg activity in vivo, in which T reg provide this tolerance against the fetus [28].
To combine the importance of genetic etiology of RPL, it makes sense to deep study the CTLA-4 gene polymorphisms. Rs231775 variant is one of common polymorphisms in CTLA-4 gene. To our best of knowledge, it is the first time to select all published articles to analyze the association between CTLA-4 gene rs231775 polymorphism and RPL susceptibility. In current study, the major discover is that rs231775 may decrease RPL risk, in other words, individuals carrying A-allele may have a decrease association for RPL, or A-allele is a protective factor for RPL risk, on the other hand, the G-allele is a potential risk factor. We boldly guess that the A-allele may increase the expression of CTLA-4 protein, because CTLA-4 is a protective factor in promoting fetus toleration and RPL [17].
In addition, we used the online analysis system String to predict potential and functional partners (Figure 9). Finally, ten genes were predicted. The highest score of association was CD86 and CD80 (score = 0.999); however, ICOSLG and LYN had the lowest scores (0.963 and 0.949, respectively). CD86 and CD80 are both the natural B7 family ligands of CTLA-4, the level of CD86(+) was significantly higher in the RPL group than the normal pregnancy group [29]. Several observations have indicated that CD28/CTLA-4 and CD86/CD80 are involved in the maternal-fetal immune regulation, which might be potentially useful to immunotherapy for human RPL [30]. Most studies were focused on the association between FOXP3 gene and RPL, including SNPs and immune regulation (such as T reg , CD4+CD25+, Th17) [13]. Karim et al. reported that several novel CNVs/genes (such as 14q32.33/AKT1) in chromosomal abnormalities were associated with RPL risk [31]. Above information predicted that CD86, CD80, FOXP3, and AKT1 may influence CTLA-4 and regulate the RPL development, which may become intervention and treatment target genes in the future.
Limitations in the present meta-analysis include the suboptimal number of published studies for a comprehensive analysis. Second, interactions between different polymorphic loci of the same CTLA-4 may modulate RPL risk, which should be included in future research and analysis. In addition, our meta-analysis was based on unadjusted estimates. A more precise analysis should be conducted if individual data are available to adjust for other covariates including age, sex, family history, environmental factors, endocrine abnormalities, each type of PL, and lifestyle.
In summary, in the present meta-analysis, a significant decreased association was found between the CTLA-4 gene rs231775 SNP and RPL risk. To further confirm the results, larger scale case-control studies with different ethnic groups and multiple PL types are needed.