The Interaction of the Oxytocin Receptor Gene and Subtypes of Child Abuse on Social Cognition in Euthymic Patients With Bipolar Disorder Type I


 Background: Most studies on cognitive impairment in Bipolar Disorder (BD) have neglected the role of early stress, despite the high frequency of childhood maltreatment in this clinical group. The aim of this study was to establish a connection between a history of emotional, physical, and sexual abuse in childhood and social cognition (SC) functioning in patients with BD type I in euthymia, and to test a possible moderating effect of the polymorphism rs53576 in the oxytocin receptor (OXTR) gene. Results: We found a high frequency of child abuse, indicating that BD patients who had been victims of physical and emotional abuse as children, and were carriers of the GG genotype at OXTR rs53576, displayed higher social cognition alterations, specifically in an emotion recognition test. Conclusions: The gene-environment interaction study presented here proposes a Differential Susceptibility model of a genetic polymorphism that can be plausibly associated with SC functioning. This evidence might help to identify at-risk clinical subgroups within a diagnostic category with well-established intra-group heterogeneity such as BD. Future research aimed at testing the inter-level impact of early stress constitutes an ethical-clinical duty in light of the high rates of childhood abuse reported in bipolar patients.


Background
Bipolar Disorder (BD) has been linked to social cognition (SC) impairments even during periods of euthymia (Samame, 2013). The importance of these studies derives from evidence connecting these de cits with psychosocial functioning in bipolar patients ( The few studies comparing bipolar patients with and without a background of child abuse have reported signi cant differences between these groups in both SC and neurocognitive tests Russo et al., 2015). Interestingly, the evidence for greater cognitive de cits in bipolar patients with a history of childhood trauma may be grounded in hypotheses with a reasonable degree of biological plausibility Hypotheses linking OXT with mood disorders (Cochran et al., 2013) have been scarcely studied in patients with BD. In this regard, authors have reported a higher serum level of OXT in bipolar patients compared to healthy controls in both symptomatic periods and euthymia (Turan et al., 2013). Determining the association between the serum level of a neuropeptide and BD may be complex, given the high phenotypic variability within this diagnostic category. In this sense, other studies have focused on measuring intermediate functioning dimensions, such as SC (Tas et al., 2015). The relevance of genetic studies focused on intermediate phenotypes may be strengthened by designs that analyze gene-environment interaction. Most studies on OXTR gene variants have shown that the G allele of rs53576 is associated with protective social traits, such as pro-social behavior (Kogan et al., 2011) and better empathic ability (Gong et al., 2017); however, authors who have evaluated its interaction with a background of early trauma have reported a pattern of greater vulnerability for this allelic variant. Studies on the interaction between OXTR gene, history of child abuse and emotional regulation in adult life have shown that carriers of the rs53576 GG genotype display less emotional regulation if they had been victims of early trauma (Bradley et al., 2011;Hiraoka et al., 2019). This pattern of differential vulnerability of the OXTR rs53576 G allele has also been observed in functional imaging studies, which report strong gray matter reduction in the bilateral ventral striatum, along with increased amygdala responsiveness to emotional facial expressions (Dannlowski et al., 2016).
Gene-environment interaction studies in patients with BD are scarce and generally focused on neurocognitive measures, evaluating the moderating effect of genetic polymorphisms other than OXTR gene variants (Savitz et al., 2007;Aas et al., 2013). To the best of our knowledge, this is the rst study to measure gene-environment interaction in patients with BD focusing on SC and OXTR. Based on the available evidence, we hypothesized that carriers of the OXTR rs53576 G allele will have a poorer performance on SC as they have experienced greater trauma. In addition, considering the possible differential effects on brain development linked to subtypes of child abuse (Teicher et al., 2016) and the evidence indicating that subtypes of child abuse are connected with SC alterations (Pope et al.2014; Russo et al.,2015), we sought to identify speci c trauma subtypes whose association with poorer performance on SC could be moderated by OXTR rs53576. Inclusion criteria were as follows: (a) bipolar I DSM-IV-TR diagnosis, (b) age 18-65 years, (c) ful lling euthymia criteria for at least 3 months, de ned as a score of ≤ 6 on the Young Mania Rating Scale (YMRS) and of ≤ 8 on the Hamilton Depression Rating Scale (HDRS) and (d) the capacity to provide written informed consent. We excluded patients who were active drug users and who had received electroconvulsive therapy over the last 3 months. Patients who did not fully completed the instruments used in this study were also excluded, resulting in a nal sample of 82 individiduals.
The participants were 46.5 years old on average (SD = 14,9 years); 65.7% of them were women. The age at onset of BD was 23.3 years (SD = 9.66), the number of hospitalization was 3,4 (SD = 4,3) and the number of suicide attempts was 1,2 (SD = 2.0) 40.2% of the patients reported being victims of at least one type of abuse as a child. The average of abuses was 0.8 ± 1.3. The frequencies per subtype of abuse were: 40,2% for sexual abuse, 26,8% for physical abuse, 32,9% for emotional abuse, 18.3% for emotional neglect and 18.3% for physical neglect. Our analysis by sex showed that 71% of the women and 51% of the men had suffered at least one type of abuse as children (p < 0.02).
We genotyped the rs53576 polymorphism located in the OXTR gene. This polymorphism was not in Hardy Weinberg equilibrium (χ 2 = 19.8, p-value = 0.000009), which is probably explained by the bias in the sample selection, i.e., patients with BD. Since this polymorphism has been linked to several behavioral traits, the selected group does not represent the general population. As for the distribution of the genotype frequencies of OXTR rs53576, 20% of the were GG, while 80% were GA + AA; we did not nd differences by sex or age. Likewise, there were no differences in the age of onset of BD or number of hospitalizations.

Procedure and instruments
The subjects' history of child abuse was evaluated using the Spanish language version of the Childhood Trauma Questionnaire -Short Form (Berstein et al., 1994; Behn et al., 2020). Brie y, this self-report instrument comprises 28 items that refer to ve subtypes of child abuse: sexual abuse (SA), emotional abuse (EA), physical abuse (PA), physical neglect, and emotional neglect. For ethical-clinical reasons, the self-report instrument was completed by each patient in a clinical setting, con dentially and in the presence of their treating physician. In this study, as recommended by Aas et al. (2013), we only considered the abuse scales.
Cognitive functioning was evaluated with social and non-social cognition. For the former dimension, we used the Spanish language version of TASIT (The Awareness of Social Inference Test) (McDonald, 2003), speci cally the emotion evaluation task. This test evaluates respondents' ability to recognize emotions by showing them 20 micro-videos in which actors simulate situations that represent the basic emotions of fear, disgust, surprise, sadness, and anger.
This study was approved by the ethics committee of the Valparaíso-San Antonio Health Service, Chile. All examinations were performed according to the tenets of the Declaration of Helsinki. All patients were informed about the study and gave their signed consent.

Genotyping
After the interview, blood samples (6 mL) were taken from the patients and placed in EDTA-coated tubes. The DNA extraction was performed using the NucleoSpin Blood kit (Macherey-Nagel, Germany) following manufacturer´s protocol. The OXTR rs53576 allelic variants were genotyped using a TaqMan® probe (SNP ID C___3290335_20, Applied Biosystems, USA) in a AriaMX thermocycler (Agilent, USA) following manufacturer´s protocol. Hardy-Weinberg equilibrium was tested by comparing the observed and expected genotypes using χ2.

Analysis
Simple moderation analysis was performed using Hayes' Process Macro v 3.0 (2018) for SPSS; this macro is widely used in research to perform moderation and mediation analysis. It has several bene ts over SEM models such as the need for less coding to generate same results o la capacidad de estimar cada ecuación del modelo de regression de forma independiente (see Hayes, Montoya & Rockwood, 2017). The analysis considered a bootstrapping sample of 5000, which allowed solving the lack of normal distribution of the sample (Currant-Everett, 2017), generating bootstrping con dence intervals for the interpretation of results.
First, the association between trauma scores of each of the ve subscales of the CTQ and performance in social cognition, moderated by OXTR rs53576, was evaluated as a way to identify speci c interactions according to types of trauma. For all analyses the moderating variable was coded as AA/AG (0) and GG (1).
Next, as a way of evaluating whether the cumulative effect of trauma is moderated by the genetic variable, a new independent variable was generated from the sum of CTQ scales that were on the cut-off score for "moderate" trauma. Because there are ve scales, the values range from 0 to 5.
For all analyses all continuous variables were mean centered.

Results
GxE and social cognition functioning by speci c types of trauma TASIT total score was regressed onto each of the ve subscales of the CTQ and its interaction with the genetic variable.

Discussion
Our study revealed that euthymic patients with BD type I that were victims of physical and emotional abuse as children and have the GG genotype at OXTR rs53576 displayed greater social cognition alterations, especially in an emotion recognition test. The present study adds evidence for gene-environment interaction, now focusing on an intermediate phenotype such as SC functioning, and thus it provides data that may improve our comprehension of previous hypotheses centered on more complex variables. In other words, alterations in emotional recognition tasks may well be related to greater demands on emotional regulation processes and their morpho-functional correlates in the limbic system. Such a clinical context may also shed light on why depressive symptomatology indexes are greater in carriers of the OXTR rs53576 GG genotype who have been affected by child abuse.
Our ndings also highlight differences on the role of the child abuse subtypes. Most studies have employed general measures of child abuse, leaving out the more speci c characteristics of the traumatic experience and their possible differential effects on the brain (Teicher et al., 2016). Our results revealed an interaction only in the presence of PA and EA, with no interaction with other types of abuse, despite the evidence indicating its undeniable impact on the life cycle of bipolar patients (Maniglio,2013). These results are consistent with studies that have reported an association between the PA and EA subtypes and SC performance, speci cally in emotional recognition tests (Russo et  This study has pioneered the exploration of SC in patients with BD using a gene-environment interaction model. Our aim was to provide data about SC de cits employing a hypothesis that presumes that adverse relational events in early life have an impact and that they interact with genetic variants of a plausibly related polymorphism such as OXTR rs53576. The description of a Differential Susceptibility genotype associated with the functioning of intermediate phenotypes may contribute to the identi cation of at-risk clinical subgroups within a diagnostic category with well-established intra-group heterogeneity such as BD.