No association between the Ser9Gly polymorphism of the dopamine receptor D3 gene and schizophrenia: a meta-analysis of family-based association studies.

BACKGROUND
Previous studies found that Ser9Gly (rs6280) might be involved in the occurrence of schizophrenia. However, no consist conclusion has yet been achieved. Compared to the case-control study, the family-based study took into account stratification bias. Thus, we conducted a meta-analysis of family-based studies to measure a pooled effect size of the association between Ser9Gly and the risk of schizophrenia.


METHODS
The relevant family-based studies were screened using the electronic databases by the inclusion criteria. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to measure the correction between Ser9Gly polymorphism and schizophrenia susceptibility. Subgroup analysis was performed by stratification of ethnicity (i.e., East Asian, Caucasian, and other populations). Additionally, publication bias was evaluated by the funnel plot.


RESULTS
After literature searching, a total of 13 family-based association studies were included, which contained 11 transmission disequilibrium test (TDT) studies with 1219 informative meiosis and 5 haplotype-based haplotype relative risk (HRR) studies. No statistical significance of the heterogeneity was detected in TDT and HRR studies. Thus, the pooled effect size was calculated under the fixed effect model. The results found that the association was significantly protective in East Asian in TDT studies (204 informative meiosis, OR = 0.744, 95% CI = 0.564-0.980, Z-value = - 2.104, p = 0.035).


CONCLUSIONS
The meta-analysis based on the family study found a protective association of Ser9Gly in East Asian. In future, large sample molecular epidemiology studies are needed to validate our findings.


Background
Schizophrenia is a complex mental disorder with the incidence rate of about 1% in the word. Genetic and environmental factors are involved in its pathogenesis [1]. According to the report, heritability of schizophrenia is as high as 80% [2]. So far, there has been no consist outcome regarding the etiology of this mental disorder [3,4]. Recently, the studies have reported that the dysfunction of dopaminergic neurotransmitter may be involved in the development of schizophrenia [5][6][7][8]. Therefore, the genes participating in dopaminergic metabolism are the underlying susceptible genes in this disease.
Dopamine receptor D3 (DRD3) is localized to the limbic areas of the brain and are associated with cognitive, emotional, and endocrine functions [9][10][11][12]. It is encoded by DRD3 gene. Ser9Gly variant (rs6280) is a functional polymorphic site in the first exon of DRD3 gene, which corresponds to a serine to glycine amino acid substitution at position 9 in the extracellular N-terminal domain of DRD3 [13]. This variation can have a higher affinity for dopamine and amplify dopamine intracellular signaling [14]. Presently, Ser9Gly polymorphism has been reported to be related to neurological and psychiatric disorders [15][16][17][18]. For the association between Ser9Gly and schizophrenia, there are still conflicting results without a consistent conclusion [19][20][21][22][23]. These contradictory results may be due to small sample size, inclusion of various genetic backgrounds, and other potential confounding bias [24].
Meta-analyses can merge the various homogeneity studies and arrive at a comprehensive result [25,26]. Since 1998, the meta-analysis have been conducted to assess the association between Ser9Gly SNP and schizophrenia risk [27][28][29][30][31][32][33]. However, all of the pooled results were based on the case-control studies, but not the family-based studies. The family-based studies are more powerful to detect risk factors of schizophrenia, considering that the ability to exploit the cosegregation of variants with schizophrenia within families helps distinguish causal from noncausal factors [34]. Therefore, we carried out a meta-analysis of family-based association studies to better evaluate the relationship between DRD3 Ser9Gly SNP and the risk of schizophrenia.

Literature search
Three online electronic English databases (Medline, Embase, and Web of Science) and one online Chinese CNKI database were searched using the following key words: "DRD3", "dopamine receptor 3", "dopamine D3 receptor", "dopamine receptor D3", "schizophrenia", and "Ser9Gly". Additionally, the other possible studies were screened and retrieved by the reference lists of the included articles and the available reviews.

Inclusion criteria
The studies reporting Ser9Gly polymorhism were included after meeting the following criteria: (1) familybased design (original transmission disequilibrium test (TDT) [35] or haplotype-based haplotype relative risk (HRR) [36]; (2) original data, or available data to pool an effect size. Finally, we collected 13 articles and the flow

Data extraction
The data extraction was independently conducted by two authors (XNL and BJW). Information collected included last name of first author, year of publication, country, ethnicity of study population, diagnostic criteria for schizophrenia, and numbers of transmissions.

Meta-analytic methods
The meta-analysis of the family-based association studies was divided into two parts: TDT and HRR. For the TDT study, every included study provided a 2-by-2 transmission disequilibrium table that classifies heterozygous parental alleles (informative meioses) based on the passing status (Ser9 allele passed to the schizophrenic offspring) and data type (the number of observed passing vs. the number of theoretic passing) [37]. For one informative meiosis, the expected transmitted number that the allele is transmitted from heterozygous parents to the proband is 0.5 and the expected untransmitted number that the allele is not transmitted from heterozygous parents to proband is also 0.5. For the HRR studies, every included study provided a 2-by-2 HRR table that classifies parental alleles by type of allele (Ser9 or Gly9) and passing status (passed to the offspring with schizophrenia or not) [37].  The Q test was used to measure the heterogeneity among the included studies and P < 0.05 indicated the presence of heterogeneity [38,39]. Additionally, I 2 was calculated to quantify the apparent inconsistency and its conventional interpretation for the existed heterogeneity was low (< 25%), moderate (approximately 50%), and high (> 75%) [40]. When there was the existing of heterogeneity (p < 0.05 and/or I 2 > 50%), a random effect model was selelcted; otherwise, a fixed effect model was selected using the Mantel and Haenszel method [38,41].
For the pooled analysis, odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to quantify the association in the two-by-two Tables. P > 0.05 indicated the absence of statistical significance, and P < 0.05 indicated statistical significance. When P < 0.05, OR < 1 meant the variation as a protective factor, and OR > 1 meant the variation as a risk factor. Pooled calculations of ORs were obtained and compared with the controls (observed transmission vs. expected transmission for TDT study or transmitted vs. untransmitted for HRR study) using test statistic z and 95% CIs. Moreover, subgroup analysis were conducted by ethnicity (i.e., East Asian, Caucasian, and other populations) and diagnostic criteria (i.e., DSM-III-R, DSM-IV, and CCMD-III). In addition, the funnel plot was generated to evaluate publication bias according to the previous study [37].
All the statistical calculations of the meta-analysis were performed by Comprehensive Meta Analysis V2 software (Biostat, Englewood, NJ, USA).

Results
A total of 13 articles were identified by database searches, which included 16 studies [27,[42][43][44][45][46][47][48][49][50][51][52][53]. Among them, 11 studies were for TDT and 5 studies were for HRR. Table 1 showed the pooled ORs and 95% CIs for the 11 original TDT studies with 1219 informative meiosis. There was no statistical significance for the heterogeneity (I 2 = 28.3%) and the fixed effect model was selected. The pooled results indicated that there were no association between Ser9Gly SNP and schizophrenia (1219   The studies distribution of the funnel plot was substantially symmetrical for the pooled effect size (Fig. 3). Thus, there was not enough evidence for publication bias for TDT studies. Table 2 showed the pooled ORs and 95% CIs for the 5 HRR studies with 1704 samples. There was no statistical significance for the heterogeneity (I 2 = 30.372%) and the fixed effect model was selected. The pooled results indicated that there were no association between Ser9Gly SNP and schizophrenia (1704 samples, OR = 0.869, 95% CI = 0.713-1.059, Z-value = − 1.395, p = 0.163). The forest plot was showed in Fig. 4. Furthermore, we performed the subgroup analysis to further explore the association of Ser9Gly in Caucasian population. The results indicated no significantly preferential transmission of DRD3 Ser9 allele in Caucasian (OR = 0.871, 95% CI = 0.604-1.254, Z-value = − 0.744, p = 0.457) ( Table 3).
The studies distribution of the funnel plot was slightly asymmetrical for the pooled effect size (Fig. 5). A small but significant effect of publication bias for HRR studies was detected.

Discussion
We conducted a meta-analysis of family-based association studies (11 for TDT and 5 for HRR) to investigate the putative association of the Ser9Gly SNP in DRD3 with the risk of schizophrenia. Our overal results suggest that no association exists, except for the significantly preferential transmission of DRD3 Ser9 allele in East Asian in TDT studies.
Several previous meta-analyses have assessed the potential association of DRD3 Ser9Gly with the risk of schizophrenia in case-control studies [28,29,[31][32][33]54]. The latest meta-analysis, which included 73 studies comprising 10,634 patients with schizophrenia (cases) and 11,258 controls, suggested that the Ser9Gly SNP is not associated with schizophrenia [33]. Its finding was consistent with our study. Although the subgroup analysis of TDT meta-analysis observed the significant association between Ser9Gly and schizophrenia in East Asian population, it only included two studies with the limited Fig. 4 Forest plot of the association between DRD3 Ser9Gly and schizophrenia for HRR studies. a: the statistical significance and Ser9 as a protective factor; b: the statistical significance and Ser9 as a risk factor  [47,48]. The results of the significantly preferential transmission of DRD3 Ser9 allele in East Asian group showed that the serine allele appears to be protective against schizophrenia. Ser9Gly variant corresponds to a serine to glycine amino acid substitution at position 9 in the extracellular. The substituted glycine allele is thought to yield D3 autoreceptors having a higher affinity for dopamine and more robust intracellular signaling. Subsequently, the increasing DRD3dependant dopamine intracellular signaling may induce the occurrence of schizophrenia. Moreover, one study of HRR in East Asian also found the significant association, but its sample size was still small (404 samples) [47]. Thus, the positive results need to be interpreted cautiously and more work is required to validate the association in East Asian population. Additionally, it is reasonable that the genetic heterogeneity can lead to the differences in the subgroup analysis of Caucasian and East Asian. Actually, the genetic heterogeneity will complicate the etiology of schizophrenia because the allele distributions of DRD3 Ser9Gly vary in different ethnicity population. Gly9 allele frequencies vary almost as much in the Japanese control populations (22-34%) as they do in northern and western Caucasian control populations (30-44%) [29,33]. Therefore, in order to reduce the genetic heterogeneity, it is necessary to study the homogeneous populations. Presently, numerous candidate genes are involved in the susceptibility of the complex disease, such as schizophrenia. Family-based association studies can provide an informative way to investigate the putative susceptible genes. Unlike population-based tests for association, the family-based tests for transmission disequilibrium are protected against population stratification and the results can avoid the effects of genetic background heterogeneity effectively [55]. Compared with the case-control study with the same sample size, the family-based study is less prone to confounding. Methodologically, it uses a more rigorous approach than the population-based study [56]. Thus, although our previous meta-analysis of casecontrol studies did not find the significant association of Ser9Gly locus with the risk of schizophrenia, it was still necessary to perform the meta-analysis of family based association.
There were two limitations in our current metaanalysis. Initially, we detected a slight but significant publication bias in the HRR studies. This bias might be due to only English-and Chinese-language studies included. Subordinately, we just evaluated the role of Ser9Gly SNP in the risk of schizophrenia. Nevertheless, only one variation just plays a minute role in the overall genetic susceptibility of the disease. Regrettably, the gene-gene interactions and epigenetics were not assessed without the sufficient information.

Conclusions
In conclusion, our meta-analysis of family-based association studies found no association between DRD3 Ser9Gly SNP and the risk of schizophrenia. The large sample homogeneous population studies will be necessary to further explore the role of DRD3 in the etiology of schizophrenia.