Interaction between catechol‐O‐methyltransferase polymorphism and childhood trauma in suicidal ideation of patients with post‐traumatic stress disorder

Abstract Introduction Suicidal behavior of post‐traumatic stress disorder (PTSD) patients is influenced by genetic and environmental factors. The catechol‐O‐methyltransferase (COMT) gene has been known to be associated with suicidal ideation. The present study aimed to explore the relationship of COMT polymorphism, childhood trauma, and suicidal ideation in patients with PTSD. Methods Fifty patients with PTSD and 62 healthy controls (HCs) were recruited, and COMT variants rs4680 and rs4633 were genotyped through peripheral blood. Psychological assessments such as the childhood trauma questionnaire (CTQ), the scale for suicidal ideation, the clinician‐administered PTSD scale for DSM‐5, and a PTSD checklist were administered. A regression analysis, the Johnson–Neyman technique, and a two‐way analysis of covariance were conducted. Results Interaction of COMT polymorphism (rs4680, rs4633) and childhood emotional abuse (subscale of CTQ) predicted suicidal ideation in patients with PTSD. Patients with the rs4680 Val/Val genotype, compared to Met carriers genotype, showed higher suicidal ideation when childhood emotional abuse was high. Patients with the rs4633 CC genotype, compared to T carriers genotype, showed higher suicidal ideation when childhood emotional abuse was high. Conclusion Our results suggest that vulnerability to suicide could be increased in the Val/Val genotype of COMT rs4680 and the CC genotype of rs4633 in patients with PTSD. Moreover, PTSD group with high childhood emotional abuse demonstrated a significantly higher suicidal ideation than did those with low childhood emotional abuse.


| INTRODUC TI ON
Exposure to trauma event has been found to be significantly associated with suicide (Stein et al., 2010). An increased risk of suicide and/ or suicidal ideation has been repeatedly reported among patients with post-traumatic stress disorder (PTSD; Bryan & Corso, 2011;Jakupcak et al., 2009;Pompili et al., 2013;Ramsawh et al., 2014). The increased suicidal behavior in PTSD patients remains significant, despite controlling for comorbid psychiatric disorders such as major depressive disorder (Krysinska & Lester, 2010;Wilcox, Storr, & Breslau, 2009).
The genetic component of suicide is an interesting topic and has become increasingly studied. Catechol-O-methyltransferase (COMT) is one of the candidate genes associated with suicide (Baud et al., 2007;Broekman, Olff, & Boer, 2007;Kia-Keating, Glatt, & Tsuang, 2007;Nedic et al., 2011) and reported as a risk factor for PTSD (Valente et al., 2011). The COMT enzyme is involved in the catalysis and inactivation of catecholamines, such as dopamine.
Childhood trauma is also strongly associated with suicide attempts later in life (Etain, Henry, Bellivier, Mathieu, & Leboyer, 2008;Lopez-Castroman et al., 2012. Childhood trauma may cause the patient with PTSD to experience flashbacks and other PTSD symptoms through helplessness (Wiedemar et al., 2008) or other psychophysiological and personality variables (Orr et al., 2012). Lopez-Castroman et al. (2012) suggested that PTSD and childhood abuse, particularly when combined, are associated with more severe suicidal behavior and higher suicidal intent. Tang et al. (2018) indicated that psychological maltreatment, such as emotional abuse and emotional neglect, is significantly associated with suicidality even after controlling for PTSD symptom severity. Specifically, emotional abuse influenced the risk of suicidality, directly and indirectly, in PTSD. These results indicate that childhood abuse might be a predictor of future PTSD (Cloitre et al., 2010;Kearney, Wechsler, Kaur, & Lemos-Miller, 2010;Ozer, Best, Lipsey, & Weiss, 2003).
Although there were a few studies reporting an interaction between the COMT Val158Met polymorphism and childhood abuse, they have focused on major depressive disorder, bipolar disorder, anxiety disorders, schizophrenia, and other related disorders (Choi et al., 2018). We could not find studies on suicidal behavior related to the interaction between COMT polymorphism and childhood trauma in patients with PTSD. To our knowledge, this is the first study that proposes the relationship between childhood trauma, COMT polymorphism, and suicidal ideation in patients with post-traumatic stress disorder and healthy controls.
We hypothesized that COMT polymorphism would interact with childhood trauma to affect suicidal ideation; moreover, we hypothesized that patients with PTSD who had higher scores of childhood abuse would show higher suicidal ideation than those with lower childhood abuse, particularly in those with COMT risk allele. It is anticipated that examining childhood trauma and COMT polymorphism as risk factors for suicide in patients with PTSD will provide noteworthy implications for its prevention.

| Participants
Fifty patients with PTSD (12 males and 38 females) and 62 healthy controls (HCs; 17 males and 45 females) were recruited from the Psychiatry Department of the Inje University Ilsan Paik Hospital in Goyang, Korea.
The diagnosis of PTSD was based on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (DSM), Fifth Edition (SCID-5) by a psychiatrist. Patients were excluded if they had any abnormal brain injury findings by computed tomography or magnetic resonance imaging. HCs were recruited from the local community through flier and posters, were evaluated using the SCID-5 for psychiatric disorders, and underwent a physical examination by a psychiatrist. HCs were required to have no history of major trauma, such as a serious car accident, combat experience, sexual assault, or serious physical injury, and were not taking medications with potentially psychological disorders. Each participant signed an informed consent form approved by the Institutional Review Board at Inje University Ilsan Paik Hospital before participating (IRB no. 2015-07-025).

| Childhood trauma questionnaire
The Korean-validated version of the childhood trauma questionnaire (CTQ) was used to measure childhood trauma (Yu, Park, Park, Ryu, & Ha, 2009). CTQ consists of five subscales of various childhood traumas, including emotional abuse, physical and sexual abuse, and emotional and physical neglect, as well as another scale for detecting minimization and denial. It is known to be useful in eliciting retrospective reports of childhood maltreatment from young adults (Everson et al., 2008). The CTQ consists of 28 items and is assessed with a 5-point Likert scale ranging from 1 ("never true") to 5 ("very often true"). The Korean-validated version of the CTQ showed adequate reliability and validity (Kim, Park, Yang, & Oh, 2011). The coefficient α of the CTQ was .79 in the previous study  and .91 in the current study. The coefficients of each subtype of the CTQ were .95 (emotional neglect), .93 (physical abuse), .90 (sexual abuse), .92 (emotional abuse), and .65 (physical neglect) in our study.

| Scale for suicidal ideation
The Korean-validated version of the scale for suicidal ideation was used to measure suicidality, which showed adequate reliability and validity (Shin, Park, Oh, & Kim, 1990). The scale consists of 19 items and is assessed with a 3-point Likert scale ranging from (0-2).
The coefficient α of this scale was .81 in the previous study (Shin et al., 1990) and .92 in our study.

| Clinician-administered PTSD scale for DSM-5
The Clinician-Administered PTSD Scale for DSM-5 (CAPS) is a structured diagnostic interview that corresponds with the DSM-5 diagnosis for PTSD, administered by a psychiatrist (Weathers et al., 2018).
It consists of 30 items and assesses information about the frequency and severity of PTSD symptoms. This was assessed by standardizing and simplifying the conversion of symptom frequency and intensity ratings into symptom severity ratings (from 0 ["Absent"] to 4 ["Extreme/incapacitating"]) and dichotomous scores ("Yes" or "No").
CAPS showed good internal consistency and validity (Weathers et al., 2018). The coefficient α of CAPS severity score was .88 in the previous study and .89 in our study.

| Post-traumatic stress disorder checklist
To examine the severity of PTSD symptoms, the Korean-validated civilian version of the PTSD checklist (PCL) was administered.
PCL is often used to diagnose individuals with PTSD or to identify individuals at risk. While CAPS and PCL are highly correlated (Oh et al., 2014;Weathers et al., 2018), discrepancies have also been reported between the overall PTSD symptoms measured by the two questionnaires (Henn-Haase, 2016). It was indicated that patients with PCL reported more re-experiencing and fewer symptoms of avoidance of thoughts/feelings, negative emotions, detachment, and restricted affect compared to CAPS. The Korean-validated version of the PCL showed adequate reliability and validity (Oh et al., 2014). It consists of 17 items and assessed with a 5-point Likert scale ranging from 1 ("not at all") to 5 ("extremely"). The coefficient α was .93 in the previous study and .97 in our study.

| Hospital anxiety and depression scale
The Korean-validated version of the hospital anxiety and depression scale (HADS) was used to evaluate anxiety and depression symptoms. It consists of seven items for describing anxiety and seven items for determining depression (Oh, Min, & Park, 1999), and it is assessed with a 4-point Likert scale, ranging from 0 (no problems) to 3 (maximum distress). The Korean-validated version of HADS showed adequate reliability and validity (Oh et al., 1999). The coefficients of each subtype of HADS were .89 (anxiety) and .86 (depression) in the previous study and .93 (anxiety) and .86 (depression) in the current study.

| COMT genotyping
All participants had their blood sampled to extract DNA using a NanoDrop ® ND-1000 UV-Vis Spectrophotometer. Genomic were used for positive control. We calculated genotype frequencies for each individual polymorphism and evaluated the Hardy-Weinberg equilibrium to check the data quality and genotype error. The chisquared test was used to compare the observed numbers of each genotype with those expected for the population following chi-square distribution with one degree of freedom (Weir, & Cockerham, 1996).

All statistical tests and visualization of differentially expressed genes
were conducted using R version 3.3.3.
Twenty-three patients with PTSD were found to have Val/Val (GG) genotype, and 27 participants with PTSD were Met carriers (GA+AA) in COMT rs4680. In HCs with COMT rs4680, 34 participants were of Val/Val genotype, and 28 participants were Met carriers. Twenty-four patients with PTSD were of CC genotype, and 26 participants with PTSD displayed T carriers (CT+TT) in COMT rs4633. In HCs with COMT rs4633, 33 participants were of CC genotype, and 29 participants were CT+TT.

| Statistical analyses
Normality was tested using the skewness and kurtosis. Skewness <2.0 and kurtosis <7.0 were considered to be moderately normally distributed (Curran, West, & Finch, 1996). All variables in our results were within the range of normal distribution. After checking for normality, a regression analysis using SPSS Macro PROCESS for SPSS (version 2.16.3;Hayes, 2013) was performed to examine the moderating effect of childhood trauma on the path between COMT polymorphisms and suicidality. Years of education was controlled as a covariate. In contrast, sex and age were not included as covariates, since there were no differences of sex and age between PTSD and HCs (Table 1). Depression and anxiety scales were not controlled since negative mood is considered as a main symptom of PTSD (Dai et al., 2017;Flory & Yehuda, 2015).
Subsequently, the Johnson-Neyman technique (Hayes, 2013) was applied to investigate the regions in which the moderating variable had a significant effect. The Johnson-Neyman technique aligns the moderating variable in a continuous manner and computes the regions of significance for interactions by examining the significance between the predictor and outcome variables (Preacher, Curran, & Bauer, 2006).
Finally, a two-way analysis of covariance (ANCOVA) was used to probe the direction of the interactions. The moderating variable was divided into high and low groups at the point in which the moderating effect was first observed. Years of education was controlled for as a covariate. All significant levels were set at p < .05 (two-tailed), and all statistical analyses were performed using SPSS version 21 (IBM Inc.). vs. 13.71 ± 3.40, p = .020), and scores of PCL, CAPS severity, and some subscales on the CTQ (physical abuse and emotional abuse)

| Descriptive statistics
were significantly higher in the PTSD group than HCs. There was no significant difference in genotype frequency of COMT (rs4680 and rs4633) between the groups. Also, there was no significant interaction between COMT genotype and sex/age.
However, there were no significant interactions in other variables in patients with PTSD. Additionally, there were no significant interaction effects in HCs.

| Probing an interaction
The Johnson-Neyman analysis was executed. In COMT rs4680 (Table 3), the conditional effect was significant when the childhood emotional abuse score was above 9.233. It implies that the interaction between childhood emotional abuse and the rs4680 on suicidal ideation was significant only when the childhood emotional abuse score exceeded 9.233. Abbreviations: COMT, catechol-O-methyltransferase; HCs, healthy controls; PTSD, post-traumatic stress disorder.

TA B L E 2 Regression analysis examining moderating effects of childhood Emotional Abuse (EA) on the path between COMT polymorphism (rs4680, rs4633) and suicidal ideation
Bold value indicates a statistically significant correlation with a p-value less than 0.05.
In COMT rs4633 (Table 3), the conditional effect was significant when the childhood emotional abuse score was above 8.679. It implies that the interaction between childhood emotional abuse and the rs4633 on suicidal ideation was significant only when the childhood emotional abuse score exceeded 8.679.
The two-way ANCOVA was executed for probing the direction of the interaction between COMT polymorphism and childhood emotional abuse. In COMT rs4680, the genotype (Val/Val vs. Met carriers) was set and the childhood emotional abuse score was divided into high and low groups based on the score of 9.233. Subsequently, patients with PTSD were divided into four groups. In COMT rs4633, the genotype (CC vs. T carriers) was set and the childhood emotional abuse score was divided into high and low groups based on the score of 8.679. Subsequently, patients with PTSD were divided into four groups.
For PTSD patients with Val/Val genotype of rs4680, the results indicated that suicidal ideation was significantly higher in the high childhood emotional abuse group than the low childhood emotional abuse group (F = 11.237, p = .003). However, in Met carriers, suicidal ideation was not significantly different between high and low childhood emotional abuse groups (F = 0.537, p = .471; Figure 1a).
For PTSD patients with CC genotype of rs4633, suicidal ideation was significantly higher in the high childhood emotional abuse group than in low childhood emotional abuse group (F = 8.398, p = .008).
However, in T carriers, suicidal ideation was not significantly different between high and low childhood emotional abuse groups (F = 0.796, p = .381; Figure 1b).

| D ISCUSS I ON
The present study aimed to explore the relationship between childhood trauma, COMT polymorphism (rs4680 and rs4633), and suicidal ideation in patients with PTSD and HCs. Our results showed that the interaction of COMT polymorphism and childhood emotional abuse predicted suicidal ideation in patients with PTSD. Patients with the Val/Val genotype of rs4680 and the CC genotype of rs4633 showed a higher suicidal ideation when the childhood emotional abuse was high. Our results suggest that vulnerability to suicide could be increased in the Val/Val genotype of rs4680 and the CC genotype of rs4633 in patients with PTSD.
Our results demonstrated that interaction between COMT polymorphism and childhood emotional abuse predicted suicidal ideation in patients with PTSD. Exposure to childhood trauma has been known to be an important risk factor for suicidal behavior (Brodsky & Stanley, 2008;Dube et al., 2001;Miller, Esposito-Smythers, Weismoore, & Renshaw, 2013). In addition, exposure to early-life trauma has shown to strongly increase the susceptibility to develop PTSD as adults (Yehuda et al., 2010). However, a genetic predisposition which is modulated or triggered by environmental factors could be a significant influencing factor on suicidal behavior (Mandelli & Serretti, 2013;Saveanu & Nemeroff, 2012). COMT polymorphism is one of the most common genes that have been studied in mental illness. Choi et al. (2018) found that rs4680 in the COMT gene was associated with suicide attempts in patients with mood disorders. Bernegger et al. (2018) found an association between COMT polymorphism and suicidal behavior in major depressive disorder and bipolar disorder patients who experienced childhood maltreatment.
COMT rs4680, as well as interactive effects between this gene and environmental adversities, has been associated with anxiety and panic symptom (Asselmann et al., 2018;Baumann et al., 2013). In sum, our findings support the previous notion that COMT gene-environment interactions could be related to suicidal ideation in various mental illnesses.
In our study, participants with the COMT rs4680 Val/Val genotype, compared to Met carriers genotype, showed a higher suicidal ideation when childhood emotional abuse was high. Previous studies also reported that the rs4680 Val allele is related with higher risk for psychopathologies and biological abnormalities (Baekken, Skorpen, Stordal, Zwart, & Hagen, 2008;Caspi et al., 2005;Craddock, Owen, & O'Donovan, 2006;Olsson et al., 2007;Thapar et al., 2005). Schulz-Heik et al. (2011) indicated that the rs4680 Val/Val genotype and the associated decrease in intrasynaptic dopamine might increase the vulnerability to dystrophic effects of trauma on the human brain.
The Val/Val genotype was associated with high levels of dissociation (Savitz et al., 2008) and anger (Perroud et al., 2010) (Choi et al., 2018;Lee & Kim, 2011). Our study considered the interaction of COMT rs4680 and childhood emotional abuse, but previous studies investigated only the main effect of independent variables on each COMT polymorphism such as childhood trauma, current alcohol consumption, and related depression scores (Choi et al., 2018;Lee & Kim, 2011).
Another possible interpretation is based on the mouse model (Scheggia et al., 2018), which demonstrates that mice with relatively increased COMT activity (i.e., mice with Val genotype) display enhanced aversive remote memories. It was known that COMT genetic variants affected prefrontal activation during long-term memory encoding and retrieval (Bertolino et al., 2006;Schott et al., 2006;Wimber et al., 2011). This study might indicate a possible role of COMT in psychiatric conditions associated with inappropriate retention of past aversive memories, such as PTSD. It suggests that individuals carrying COMT Val genotype could have higher vulnerability and more severe symptoms of PTSD.
In our study, the CC genotype of rs4633, compared to T carriers, showed a higher suicidal ideation when the childhood emotional abuse was high in patients with PTSD. A recent study reported that rs4633 was related to total PTSD symptom score based on DSM-5 (Goenjian et al., 2015). In particular, the symptom "negative changes in thoughts and mood" was regarded to be relevant. Our result is in line with this study in the sense that suicidal ideation is one of the severe negative thoughts and mood in PTSD symptoms. Although our results indicate the possibility that the CC genotype of rs4633 increases the vulnerability of suicidal ideation in patients with PTSD, there is a need for additional studies to enhance the understanding of this genotype effect.
Conversely, our results fail to demonstrate the interaction between COMT polymorphism and childhood emotional abuse on suicidal ideation in the HCs group. In the HCs group, childhood emotional abuse appeared to predict suicidal ideation, but the interaction effect with COMT polymorphism was not significant. This might be due to the difference in the ratio of COMT genotype between patients with PTSD (Val/Val; 40.4%) and HCs (Val/Val; 59.6%). In addition, these differences could be derived from the different levels of severity of childhood trauma between the two groups. The HCs F I G U R E 1 Interaction effects between the childhood emotional abuse and COMT polymorphism (rs4680, rs4633) on suicidal ideation. *p < .05; **p < .01 group, compared to PTSD, did not have severe childhood traumatic experiences or recent trauma to cause PTSD symptoms.
Additionally, there was neither a significant interaction between COMT genotype and sex, nor a significant group difference in sex between PTSD and HCs in current study. Previous studies reported sex-dependent effects related to COMT genetic variants (Papaleo, Sannino, Piras, & Spalletta, 2015;Sannino et al., 2014Sannino et al., , 2017. Such discrepancy may be due to the different methods used and the ethnicity-specific genetic difference. There are some limitations in this study. First, the CTQ and its subscales may not precisely reflect the participants' traumatic childhood experiences because it is a subjective and retrospective self-report. Second, suicidal ideation and suicide attempts were considered as one continuous process in our study, which is important to re-examine as suicidality could be a heterogeneous construct (Stein et al., 2010).
In conclusion, our study demonstrated that the interaction of COMT polymorphisms (rs4680, rs4633) and childhood emotional abuse predicted suicidal ideation in patients with PTSD. Our results suggested that vulnerability to suicide could be increased in the Val/ Val genotype of COMT rs4680 and the CC genotype of rs4633 in patients with PTSD. Moreover, PTSD group with high childhood emotional abuse demonstrated a significantly higher suicidal ideation than did those with low childhood emotional abuse.

ACK N OWLED G M ENTS
This study was supported by a grant from the Brain Research

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
A.K. conducted all statistical analyses and wrote the manuscript. designed the study, supervised its analysis, and edited the manuscript. All authors discussed the results and implications and commented on the manuscript at all stages.

PE E R R E V I E W
The peer review history for this article is available at https://doi. org/10.1002/brb3.1733.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author, S.H.L., upon reasonable request.