No Association of Catechol-O-Methyltransferase Polymorphisms with Schizophrenia in the Han Chinese Population

Introduction

Schizophrenia (SZ) is a complex genetic disease with a heritability of ∼80%-85% (Cardno and Gottesman, 2000). Dopamine (DA) signaling is strongly implicated in the pathophysiology of SZ, as evidenced by the fact that antipsychotic drugs target mainly on the DA receptors. Catechol-O-methyltransferase (COMT) catalyzes the degradation of the catecholamine neurotransmitters, including DA, norepinephrine, and epinephrine, thereby influencing the DA concentration in the dorsolateral prefrontal cortex (PFC). COMT maps at chromosome 22q11.2, a region that has been implicated in linkage areas of SZ (Lewis et al., 2003) and whose deletion causes the velo-cardio-facial syndrome. Thus, COMT is regarded as one of most promising candidate genes for SZ. It contains a missense polymorphism Val158Met (rs4680), resulting in a G to A substitution that changes Valine (Val) into Methionine (Met) at codon 108/158 in exon 4. rs4680 significantly affects the protein abundance and enzyme activity, with a higher COMT activity in individuals with Val homozygotes compared with those with Met homozygotes (Chen et al., 2004). This single-nucleotide polymorphism (SNP) has been widely studied for its connection with SZ susceptibility, but with conflicting results. Meta-analyses do not support the association with SZ (Fan et al., 2005; Munafo et al., 2005; Okochi et al., 2009). Several previous studies in the Han Chinese population failed to find a positive association with SZ (Fan et al., 2005; Yu et al., 2007; Kang et al., 2010; Chen et al., 2011); but these studies chiefly focused on only one SNP and the sample sizes involved were relatively small.

In this work, we aimed to evaluate the association of COMT with SZ in a two-stage study. In the discovery stage, we derived relevant data from our genome-wide association study (GWAS) data (Yue et al., 2011), involving nine SNPs in 768 SZ cases and 1348 healthy controls. In the validation stage, we tested the genetic distribution of rs4680 in 963 cases and 996 controls.

Materials and Methods

Results

Genotypic distributions of the 9 SNPs in the discovery sample did not deviate from Hardy-Weinberg equilibrium (HWE; p>0.001). rs4680 significantly deviated from HWE in the patient group of the validation sample (p=5.7e-4). Allele and genotype frequencies of the 9 SNPs did not significantly differ between patients and controls in the 768 SZ cases and 1348 controls (Table 1), nor did the haplotypes (data not shown). There was no association of rs4680 with SZ in the validation sample, nor in the combined sample (n=4071, p=0.11, odds ratio [OR]=1.08; Table 1). Overdominance test (GG+AA vs. GA) in the combined sample showed nominal significance (χ²=3.86, p=0.049), with GA being overrepresented in the cases.

Discussion

DA signaling is strongly implicated in the etiology of SZ. Because of its essential role in DA metabolism, COMT has been postulated to be involved in the development of SZ. The functional polymorphism, rs4680, leads to remarkable alteration of the enzyme efficiency. This functional relevance has stirred a big research interest for this marker in relation to cognitive function and psychiatric conditions. Although some studies supported an association of Val or Met allele with SZ, more studies failed to find any association (Fan et al., 2005; Munafo et al., 2005; Okochi et al., 2009). Cumulative evidence suggests that there is an inverted-U-shape relationship between DA activity in PFC and working memory performance (Goldman-Rakic et al., 2000), and that Val homozygotes may favor cognitive tasks demanding flexibility at the expense of lower stability, while Met homozygotes may perform better on tasks demanding more stability than flexibility (Durstewitz and Seamans, 2008). Therefore, the alleles of rs4680 might play a modulatory role rather than simply being “good” or “bad” in the onset of SZ (Tunbridge et al., 2006), and the precise effect of COMT activity on PFC function will be dependent on precisely where on the inverted-U curve the individual in question lies in any given environmental or genetic context.

Recently, Costas et al. (2011) proposed an overdominant model through their meta-analysis of Val158Met and SZ. In this model, they suggest that both too high and too low levels of DA signaling may be risk factors (i.e., heterozygotes perform better than homozygotes), with heterozygotes at rs6280 conferring a weak protective effect (OR=0.947). This result was in agreement with a Spanish study (Hoenicka et al., 2010). This model is intriguing, considering the inverted-U-shaped relationship of DA signaling and brain function, but has yet to be validated.

In this work, we extended rs6280 to nine common SNPs, spanning from intron 1 to the 3′UTR of COMT. Our analysis showed that there was no association of COMT polymorphisms or haplotypes with SZ in the Han Chinese population. In contrast to findings of Costas et al. (2011), we detected a nominally significant overrepresentation of rs6280 heterozygotes in the combined patient group compared with controls (p=0.049). Therefore, our data do not support the overdominant model of COMT rs6280 concerning SZ.

Of note is a strong deviation from HWE in the SZ cases in the validation sample (p=5.7e-4), and a nominal deviation in the combined sample (p=0.006). Although testing of HWE is commonly used for quality control in large-scale genotyping, there is little consensus regarding how to deal with the departure from HWE in the cases. Underlying reasons include genetic drift, population subdivision, genotyping error, and natural selection (Lachance, 2009). However, we genotyped another 22 SNPs in the same validation sample simultaneously and there were no severe departures from HWE (p>0.001) for these markers. These facts mitigate the likelihood of genetic drift, or population subdivision. Considering the calling rate of 94.8% we cannot exclude the possibility of genotyping errors. Otherwise, the lack of HWE for rs6280 in the cases might be selection-related.

There were several limitations to our study. Assuming a disease allele frequency of 0.25, and a type I error of 0.05, our sample has an 80% power to detect a variant with an OR larger than 1.22 (Dupont and Plummer, 1998). Although we tested nine SNPs within COMT, the selection of candidate SNPs was not unbiased but dependent on the design of the commercial array. The effect of rs4680 probably depends on the genetic background and environmental factors; therefore, testing rs4680 without accommodating other epistatic factors would result in reduced power and limited reproducibility in this work. Further studies taking into account the genetic context would be more revealing.

In conclusion, our study does not support the association of rs6280 with SZ in the Han Chinese population. The deviation from HWE of rs4680 in SZ cases calls for further investigation.