COMT and MAO-A Polymorphisms and Obsessive-Compulsive Disorder: A Family-Based Association Study

Objective Obsessive-compulsive disorder (OCD) is a common and debilitating psychiatric illness. Although a genetic component contributes to its etiology, no single gene or mechanism has been identified to the OCD susceptibility. The catechol-O-methyltransferase (COMT) and monoamine oxidase A (MAO-A) genes have been investigated in previous OCD studies, but the results are still unclear. More recently, Taylor (2013) in a comprehensive meta-analysis of genetic association studies has identified COMT and MAO-A polymorphisms involved with OCD. In an effort to clarify the role of these two genes in OCD vulnerability, a family-based association investigation was performed as an alternative strategy to the classical case-control design. Methods Transmission disequilibrium analyses were performed after genotyping 13 single-nucleotide polymorphisms (eight in COMT and five in MAO-A) in 783 OCD trios (probands and their parents). Four different OCD phenotypes (from narrow to broad OCD definitions) and a SNP x SNP epistasis were also analyzed. Results OCD, broad and narrow phenotypes,were not associated with any of the investigated COMT and MAO-A polymorphisms. In addition, the analyses of gene-gene interaction did not show significant epistatic influences on phenotype between COMT and MAO-A. Conclusions The findings do not support an association between DSM-IV OCD and the variants of COMT or MAO-A. However, results from this study cannot exclude the contribution of these genes in the manifestation of OCD. The evaluation of broader spectrum phenotypes could help to understand the role of these and other genes in the pathophysiology of OCD and its spectrum disorders.


Introduction
Obsessive-compulsive disorder (OCD) is characterized by repetitive thoughts (obsessions) and repetitive behaviors (compulsions) that are unwanted, time consuming, egodystonic, and result in significant functional impairment [1]. Various studies have demonstrated that the etiology of OCD has a genetic component [2] with reported heritability rates ranging between 27% and 65% [3]. Although more than 140 candidate gene studies have been conducted, the findings have been inconclusive due to small sample size and few study replications [4]. The etiology of OCD seems to involve an interaction between environmental factors and several genes of small effect [2]. Studies employing different methodological strategies (e.g., cerebral spinal fluid metabolites measurements, pharmacological challenges, and drug treatment responses) all suggest that abnormal serotonergic neurotransmission is one of the most consistent biological findings in OCD [5]. Studies have also reported dopaminergic abnormalities in the basal ganglia and nucleus accumbens, as well as altered glutamate transmission in other locations in OCD samples, thus suggesting that complex interactions between multiple neurotransmitter systems may contribute to the phenotypic presentation of OCD [6]. This could be understood by a potential dysfunction in enzymes that metabolize central nervous system neurotransmitters in OCD. Two important enzymes of this type are catechol-O-methyltransferase (COMT) and monoamine oxidase A (MAO-A).
COMT is an Mg 2+ -dependent enzyme involved in the inactivation of certain catecholamines (norepinephrine, epinephrine, and dopamine). The important role that COMT plays in neuropsychiatric disorders has previously been described [7]. The most widely studied COMT polymorphism is a single-nucleotide polymorphism (SNP) leading to a valine-to-methionine substitution at codon 158 (val 158 met or rs4680), which results in a low-activity thermolabile and a high-activity thermostable forms of the enzyme-the met (158) and val (158) alleles, respectively [8].
A deletion within the 22q11 region, which includes the COMT region, causes velocardiofacial syndrome, which increases the risk for many psychiatric disorders [9], including OCD [10]. Although various studies have evaluated the association between COMT polymorphisms and OCD, the findings have been inconclusive [4].
The monoamine oxidases (MAO) are enzymes that catalyze the oxidation of monoamines, which exist in two forms: MAO-A and MAO-B. Both forms are found bound to the outer membrane of mitochondria in most cell types in the body but show different specificities. MAO-A is present in catecholaminergic neurons in the brain and plays a preferencial role in the metabolic degradation of several neurotransmitters, including serotonin, norepinephrine, epinephrine, and dopamine. The MAO-A gene maps to the p11.3 region on the X chromosome. The first evidence that MAO-A could play an important role in human behavior was the description of a large Dutch family that presented with a new form of mental retardation with prominent behavioral abnormalities linked to the X chromosome. Several males in the family were affected by this disturbance and exhibited aggressive behavior, with significant impulsivity, pyromania, suicide attempts, and sexually aberrant behavior [11,12].
Some studies that have investigated the association between MAO-A polymorphisms and OCD, and they have produced controversial results [4]. The various methodologies, different phenotype definitions, small sample sizes and diverse ethnic backgrounds of the populations studied could be sources of bias contributing to the discrepant results obtained in previous studies.
Recently, Taylor [13] performed a comprehensive meta-analysis of OCD genetic association studies. A total of 230 polymorphisms from 113 genetic association studies were identified. In the main meta-analysis, COMT and MAO-A polymorphisms were found to be associated with OCD in males.
In this study, the transmission disequilibrium of COMT and MAO-A SNPs was analyzed in OCD patients, using a family-based association approach. Given that family studies have shown familial aggregation between OCD and different psychiatric disorders (e.g., tic disorders, body dysmorphic disorder, skin picking, trichotillomania and anxiety disorders) [14,15,16,17,18,19], these disorders were included as broader phenotypes in the analyses.

Methods Participants
The sample consisted of 783 individuals meeting DSM-IV criteria for OCD, and their biological parents. The OCD probands were recruited from specialized OCD clinics at Harvard University, the University of California-San Francisco, the University of Michigan, the University of Toronto, as well as from two universities in Brazil-the University of São Paulo and the University of Pernambuco that participate from the Obsessive-Compulsive Foundation Genetics Collaborative Subgroup. Most of the sample was studied in a genome-wide association study [20]. The probands and their parents underwent structured clinical interviews. For subjects who were 16 years of age, we employed the Structured Clinical Interview for DSM-IV Axis I Disorders, Clinician Version [21]. For subjects who were < 16 years of age, we employed the Schedule for Affective Disorders and Schizophrenia for School Aged Children-Present and Lifetime Version (K-SADS-PL) [22]. In addition, all subjects were interviewed with the Yale-Brown Obsessive Compulsive Scale [23] and the Yale Global Tic Severity Scale [24]. All of the interviewers were trained mental health professionals (psychiatrists or psychologists). Reliability evaluations were performed and best-estimate diagnoses were made. The assessment procedures have previously been described in detail [25].

Ethics Statement
This work was performed in accordance with Declaration of Helsinki. The project was approved by the research ethics committees (IRB) at each of the participating institutions (

Narrow and Broad Phenotypes categories
In addition to the association analysis between COMT and MAO-A and DSM-IV OCD, additional analyses were performed in the Brazilian subsample (83 trios) using the broader phenotypes described below.
Hypothesizing that OCD phenomenology (and not severity) could be influenced by COMT and MAO-A genes, we examined the association of COMT and MAO-A polymorphisms and a broadly-defined OCD, which includes cases with OCD symptoms even when they did not meet full DSM-IV criteria for OCD (subclinical OCD).Acknowledging that OCD is part of a spectrum [26,27] that includes other anxiety disorders [19,28], body dysmorphic disorder, and pathological grooming disorders (skin picking and trichotillomania) [17], these conditions were collectively referred to as OCD spectrum disorders. The term 'OCD spectrum disorders' includes DSM-IV OCD, subclinical OCD, together with the DSM-IV anxiety disorders, body dysmorphic disorder, and pathological grooming disorders (skin picking and trichotillomania).
In addition, since some studies have suggested that OCD could be grouped based on tic disorder comorbidity [29,30], analyses in the Brazilian subsample were also completed to examine an association between COMT and MAO-A polymorphisms and OCD with tic disorder (Tourette syndrome or chronic tic disorder) comorbidity.
Genotyping DNA from peripheral blood samples were extracted using the "salting-out" protocol [31]. On the basis of Tagger software [32], COMT and MAOA genomic regions were specified and tag SNPs were picked (a local copy of HapMap data was used) within them. Tagger produced a list of tag SNPs and corresponding statistical tests to capture all variants of interest, and a summary coverage report of the selected tag SNPs. Eleven SNPs in the COMT gene region and six SNPs in the MAO-A gene region were selected. These SNPs were genotyped at the Psychiatric and Neurodevelopmental Genetics Unit of Massachusetts General Hospital, a teaching hospital associated with Harvard University. Genotyping was performed in 384-well plates with the Sequenom MassARRAY platform (Sequenom, San Diego, CA, USA). Primers for polymerase chain reaction (PCR) amplification and single-base extension assays were designed using Assay Design software, version 3.1 on the basis of FASTA sequences surrounding the SNPs, derived from SNPper [33]. Multiplex PCR was performed, followed by a pooled single-base extension reaction (iPLEX Gold). Samples were analyzed with a MassARRAY RT mass spectrometer, in automated mode. The resulting spectra were analyzed with SpectroTyper software after baseline correction and peak identification (Sequenom). Information about the ten COMT SNPs and the six MAO-A SNPs was obtained from the dbSNP, Celera, and HapMap databases. Three markers were excluded because of quality control issues: COMT rs2239393, because the genotyping success rate was lower than 75%; MAO-A rs2179098, because the minor allele frequency was 0.025; and COMT rs1544325, because the Hardy-Weinberg p value was < 0.05. The thirteen remaining SNPs collectively covered the COMT and MAO-A gene regions: the eight selected COMT SNPs spanned 25.6 kb, with a density of 2.84 kb/SNP; and the five selected MAO-A SNPs spanned 63 kb, with a density of 12.6 kb/SNP. Single-marker and haplotype analyses were performed in order to identify associations.

Statistical analysis
Statistical analyses were performed with the programs PLINK [34,35] and Haploview 3.32 [36]. Using the Haploview software, SNPs were selected on the basis of quality control criteria. 95% confidence bounds on D prime are generated and each comparison is called "strong LD", "inconclusive" or "strong recombination". A haplotype block is created if 95% of informative (i.e., non-inconclusive) comparisons are "strong LD" [37]. Following this, the haplotype TDT analysis was performed.
For single markers association, the results of the standard transmission/disequilibrium test (TDT) and the`parental discordance test` [34] were combined in order to calculate the pvalue using PLINK [34,35]. No covariate was added in the TDT analysis model. The`parental discordance test`is based on a comparison between affected and unaffected parents in terms of the number of alleles they carry, treating each parental pair as a matched case-control pair. Thè parental discordance test`assumes homogeneity, in terms of population stratification, within rather than between nuclear families. This test can add power to family-based association analyses, as well as provide considerable protection against population stratification [38].
In addition to the association analysis between COMT and MAO-A and DSM-IV OCD, a secondary investigation of the COMT and MAO-A associations in our sample was completed. First, probands were grouped into gender-matched trios in order to evaluate the influence of gender on the results. This approach was based on the knowledge that COMT has influence on estrogens [39] and that MAO-A is located on the X chromosome.
Second, exploratory analyses were performed on the Brazilian subsample to consider as affected status, some OCD-related phenotypes such as broadly-defined OCD (clinical plus subclinical OCD), OCD spectrum disorders and OCD with comorbid tic disorders.
Third, since both COMT and MAOA are involved in dopamine catabolism, the gene-gene interaction was evaluated in contributing to OCD etiology. It was used as the case-only epistatic analysis under the SNP x SNP model, using PLINK software [34,35] in the whole sample (783 trios). Only SNPs that are more than 1 Mb apart, or on different chromosomes, are included in case-only analyses.
The statistical power of this study was calculated using Genetic Power Calculator [40]. The results were subject to permutation analysis (100,000 permutations) using Plink, in order to control for false-positive association, and to Bonferroni`s correction for multiple analysis (corrected p-value threshold = 0.0004).

Strictly-defined DSM-IV OCD phenotype
The results of the OCD association analysis of the eight COMT SNPs and five MAO-A SNPs are described in Table 1. The TDT, the "parental discordance test", and the combined test were all not significant for the female (Table 2) and the male proband trios (Table 3).

Haplotypes
Two haplotype blocks were identified in each gene (Figs. 1 and 2). In the COMT gene (Fig. 1), SNPs rs737866 and rs933271 compose block 1; and SNPs rs4646316 and rs165774 compose block 2. In the MAO-A gene (Fig. 2), SNPs rs1465107 and rs1465108 compose block 1; and the SNPs rs6323, rs979606, and rs979605 compose block 2. None of these haplotype blocks were found to be associated with OCD ( Figs. 1 and 2).

Epistasis analysis
None of the COMT or MAOA SNPs showed an epistatic interaction in association with OCD (data not shown). Power calculation for whole sample The statistical power for this association analysis (783 trios, high risk allele frequency = 0.478; OCD prevalence = 0.025; alpha = 0.05) was 65% in average for COMT gene (1120 trios were needed to achieve 80% power) and 58% in average for MAOA (783 trios, high risk allele frequency = 0.257; OCD prevalence = 0.025; alpha = 0.05; and 1305 trios needed to achieve 80% power). A secondary analysis was performed, with the Brazilian subsample, as described below:

Strictly-defined OCD with comorbid tic disorders
When the Brazilian patients with OCD and comorbid tic disorders (Tourette syndrome or chronic tic disorders; N = 19 trios) were analyzed separately, OCD with comorbid tic disorder was not found to be significantly associated with any COMT or MAO-A SNP (S1 Table).

Broadly-defined OCD phenotype
When the Brazilian subjects meeting criteria for subclinical and clinical OCD are included together, there was preferential transmission of the A allele of the MAO-A SNP rs1465108 (p = 0.0469) (S2 Table). This finding did not remain significant after 100,000 permutations nor after Bonferroni`s correction.

Broadly-defined OCD spectrum disorders
Subsequent analyses revealed that the A allele of the MAO-A SNP rs979605 was associated with OCD spectrum disorders (as defined above: including patients with OCD, subclinical OCD, skin picking, trichotillomania, body dysmorphic disorder, or DSM-IV-defined anxiety disorders) (p = 0.027) (S3 Table). This finding also did not remain significant after 100,000 permutations nor after Bonferroni`s correction.

Discussion
The results presented here do not support the hypothesis that DSM-IV OCD diagnosis is associated with MAO-A or COMT. Results were also negative in the additional analyses performed: 1) one that examined a strictly defined phenotype that consisted of individuals with OCD and tic disorders; 2) one that examined a broadly defined phenotype that included OCD, subclinical OCD; 3) one that included other OCD spectrum disorders; and 4) one COMT-MAOA epistasis analysis. This result is consistent with the findings of a previous Brazilian study where no association between DSM-IV OCD and the COMT val-158-met (rs4680) variant in a case-control OCD study was observed (Meira-Lima et al. 2004).
As with any genetic association study, the present work is not without its limitations. The main limitation is the small sample size. The statistical power to refute the null hypothesis in this study was 23% (COMT) and 21% (MAOA) for the whole sample and 11% (COMT) and 10.5% (MAOA) for the analyses performed solely on the Brazilian subsample.
The most recent meta-analysis evaluated about 810 OCD cases and found a positive association between COMT rs4680 and OCD in males [41]. These analyses had a power of 49.9% for the total sample and 30.75% for the male sample. As mentioned previously, about 1800 OCD trios were needed to achieve 80% statistical power in a COMT rs4680 and OCD association study [40].
The same meta-analysis evaluated about 452 OCD probands regarding the MAOA EcoRV polymorphism and found a significant association between T allele and OCD in males [41]. This meta-analysis reached 17.2% statistical power for the MAOA polymorphism. The sample number needed to reach 80% power would be more than 2,000 trios [40].
Therefore, the accumulated empirical data to date on the association of COMT and MAOA polymorphisms and OCD are not enough to refute the null hypothesis and thus, additional association studies are needed to add power in future meta-analyses.
Another limitation was the number of tests performed in this study (more than 120 tests). The multitesting could increase the odds of false positive results. None of the results remained significant after Bonferroni`s correction or permutation analysis. Other investigators, who also had small sample studies, have reported associations between COMT and other psychiatric conditions, such as schizophrenia [42,43,44,45,46,47,48,49,50], bipolar disorder [47,50,51]; alcoholism [52,53], substance use disorders [54,55], depression [56,57], and anorexia nervosa [58]. Likewise, MAO-A has been associated with attention deficit hyperactivity disorder [59], anxiety disorders [60], major depressive disorder [61], and other psychiatric conditions [9,62,63]. Besides the odds of some false-positive associations, the non-specificity of the phenotypes associated with the COMT and MAO-A genes might be related to their metabolic functions. Because the COMT and MAO-A enzymes metabolize a number of important neurotransmitters in the limbic pathways, it is possible that, from a phenomenological perspective, impairment of their function could lead to various psychological presentations. Besides, underlying psychopathology of correlated mental disorders are hypothesized to be dimensional and continuous and to share genetic etiologic factors [64].
Additionally, the genetic heterogeneity of the sample could be another limitation, since the Brazilian population is an admixture of individuals of European, African and indigenous descent. However, the use of a family-based approach in conjunction with the`parental discordance test`minimized the odds of population stratification bias.
In conclusion, the present work investigated eight SNPs from COMT gene, five from MAO-A gene and the interaction between these two genes in 783 OCD trios using transmission disequilibrium linkage analyses. Analyses using three phenotype variations (from narrow to broad definitions of OCD phenotypes), showed only no significant association. Despite the lack of association, it appears that the broadly-defined OCD, as well as the OCD spectrum disorders, may be interesting phenotypes to be studied in future association analysis. Further genetic association studies with larger samples involving a more comprehensive coverage of polymorphisms in the COMT and MAO-A genes, as well as other meta-analyses, are needed in order to clarify the biological effect of those genes over the OCD susceptibility phenotypes.
Supporting Information S1  Table. Full dataset used in the analysis. Legend: FAMILY ID: Given code for each family; SUBJECT ID: given code for each subject; FATHER: subject's father code, 0 if it is not included in the sample; MOTHER: subject's mother code, 0 if it is not included in the sample; GENDER: subject's gender = 1 if male, and = 2 if female; DSM OCD: 1 = did not fulfill obsessive-compulsive disorder criteria according the Diagnostic and Statistical Manual for Mental Disordersforth version, and 2 = did fulfill those criteria; BroadOCD 1 = did not have at least 75% of obsessive-compulsive disorder criteria according the Diagnostic and Statistical Manual for Mental Disordersforth version, and 2 = did have at least 75% of obsessive-compulsive disorder criteria according the Diagnostic and Statistical Manual for Mental Disordersforth version; OCD+TIC: 1 = did not fulfill obsessive-compulsive disorder criteria plus Tourette's syndrome or chronic tic disorder criteria according the Diagnostic and Statistical Manual for Mental Disordersforth version, and 2 = did fulfill both obsessive-compulsive disorder criteria and Tourette's syndrome or chronic tic disorder criteria according the Diagnostic and Statistical Manual for Mental Disordersforth version; SpecOCD: 1 = did not fulfill any of the called Obsessive-compulsive spectrum disorders, which includes broad obsessive-compulsive diagnosis as well as the anxiety disorders, body dysmorphic disorder, and pathological grooming disorders (skin picking and trichotillomania) according the Diagnostic and Statistical Manual for Mental Disordersforth version. (XLSX)