Oxytocin Receptor Gene, Childhood Maltreatment and Borderline Personality Disorder among Male Inmates in China

Background Borderline personality disorder (BPD) is caused by a variety of biological and environmental factors. Accumulating evidence suggests that childhood maltreatment is a risk environmental factor in the development of BPD, but research on the genetic pathology of BPD is still in its early stages, and very little is known about the oxytocin receptor (OXTR) gene. The purpose of this study is to further explore the interaction between OXTR gene polymorphism and childhood maltreatment in BPD. Methods A total of 1804 prison inmates and 196 normal controls participated in this study. Childhood maltreatment was measured by a self-reported questionnaire, as was BPD. Venous blood was collected for the genotype test. Results Analyses revealed that the BP group had higher rs53576 AA genotype frequency and rs237987 AA genotype frequency than the non-BP group. Childhood maltreatment and its dimension could positively predict BPD risk. Among the prison samples, rs53576 GG genotype carriers had higher BPD scores at higher levels of emotional abuse, physical abuse and sexual abuse and lower BPD scores at lower levels of physical abuse and sexual abuse. maltreatment, we found a similar pattern of results. In a two-way interaction model with BPD as the dependent variable and the genotype of rs237897 as a moderator variable, childhood maltreatment and its dimensions were examined as independent variables. The analysis revealed that the interaction effect of genotype by childhood maltreatment total score was nonsignificant for BPD in the BP group and the non-BP group (R 2 = 0.033, t = 1.328, p = 0.185, R 2 = 0.045, t = 1.529, p = 0.127, respectively). Similar nonsignificant results were obtained for emotional abuse (R 2 = 0.024, t = 0.813, p = 0.417, R 2 = 0.045, t = 1.325, p = 0.186, respectively), physical abuse (R 2 = 0.019, t = 0.505, p = 0.614, R 2 =

2 Abstract Background Borderline personality disorder (BPD) is caused by a variety of biological and environmental factors.
Accumulating evidence suggests that childhood maltreatment is a risk environmental factor in the development of BPD, but research on the genetic pathology of BPD is still in its early stages, and very little is known about the oxytocin receptor (OXTR) gene. The purpose of this study is to further explore the interaction between OXTR gene polymorphism and childhood maltreatment in BPD.

Methods
A total of 1804 prison inmates and 196 normal controls participated in this study. Childhood maltreatment was measured by a self-reported questionnaire, as was BPD. Venous blood was collected for the genotype test.

Results
Analyses revealed that the BP group had higher rs53576 AA genotype frequency and rs237987 AA genotype frequency than the non-BP group. Childhood maltreatment and its dimension could positively predict BPD risk. Among the prison samples, rs53576 GG genotype carriers had higher BPD scores at higher levels of emotional abuse, physical abuse and sexual abuse and lower BPD scores at lower levels of physical abuse and sexual abuse.

Conclusions
The finding suggests that the interaction between childhood maltreatment and OXTR gene variations is an important mechanism for the development of BPD. The moderating role of the OXTR gene provides evidence for gene plasticity.

Background
Borderline personality disorder (BPD) is characterized by a pervasive pattern of emotional instability, interpersonal tension, high impulsiveness, identity disorder and cognitive impairment [1]. BPD has been estimated to have 10% mortality caused by suicide [2]. Therefore, BPD imposes enormous economic and psychological burdens, highlighting the necessity of elucidating the pathological mechanism of BPD. Decades of research have corroborated the concept that BPD is caused by a 3 variety of biological and environmental factors, and the interaction between these two factors has been emphasized recently [1].
Twin studies have suggested that 35%-67% of the variance in BPD risk is attributable to genetic factors [3][4]. Great efforts have been made to elucidate the possible genetic background of BPD, and the current polymorphism studies mainly focused on the serotonin receptor (HTR), tryptophan hydroxylase (TPH), monoamine oxidase A (MAOA), catechol-O-methyltransferase (COMT) and dopamine transporter (DAT) genes [5][6][7][8]. However, very few studies have investigated the relationship between oxytocin receptor (OXTR) gene and BPD.
Oxytocin is a peptide hormone synthesized in the paraventricular and the supraoptic nucleus of the hypothalamus. Oxytocinergic neurotransmission is mediated by the oxytocin receptors. The OXTR gene is located on chromosome 3p25 and is associated with encoding and activity of oxytocin receptors [9]. A wealth of studies revealed that oxytocin and its receptor play crucial roles in social cognition and behavior, such as trust, empathy, and maternal bonding [10]. Nevertheless, oxytocin and its receptor were implicated in several psychiatric disorders including BPD [11]. A case-control study found that oxytocin could reduce trust and cooperation in a monetary task in anxiously attached BPD patients [12].
The contradictory effect of oxytocin and its receptor may be attributable to environmental factors, which means the interaction between the gene and the environment (G×E) plays a role in the development of BPD. The OXTR gene can be classified as a 'plasticity gene', resulting in increased susceptibility of individuals to both positive and negative facets of environmental experiences, in accordance with the hypothetical pathogenesis model for BPD proposed by Amad [1]. A recent study has confirmed this plasticity of the OXTR gene, individual who carries OXTR rs53576 A-allele had higher levels of BPD symptoms under negative family conditions and lower levels under positive conditions [13].
Regarding the influence of environment factors on BPD, previous studies had investigated the association between family environment, childhood maltreatment, early life stress and BPD [14][15][16]. In the present study, we sought to shed more light on childhood maltreatment and its interaction with 4 OXTR gene. Several lines of evidence showed that there are interactions between the OXTR gene and childhood maltreatment; for instance, OXTR rs53576 G-allele carriers who had experienced maltreatment in childhood have been observed to have increased conduct problems in adolescence [17], and maltreated individual who carried the rs237885 TT genotype in OXTR had an increased risk of aggression [18]. In addition, the interaction between the OXTR gene and childhood maltreatment affects the structure of the brain, especially the amygdala [19]; the amygdala is responsible for emotional regulation, which is what BPD patients lack. However, research on the interaction between the OXTR gene and childhood maltreatment for the development of BPD is very scarce.
Previous study has shown that most inmates had a personality disorder when they committed the crime, and BPD is one of the most common personality disorders among them [20]. Although the prevalence of BPD in women is higher than that in men [21], the crime rate of men is higher than that of women [22]. Considering the high risk of this population and the absence of such study on this group in China, male inmates were chosen as our study subjects.
The present study aimed to determine whether OXTR genetic variants and childhood maltreatment were associated with BPD, and whether these polymorphisms showed interactive effects with childhood maltreatment on BPD level in Chinese male inmates.

Participants
A total of 1804 male inmates from an adult male prison in Jiangsu Province, China and 196 healthy male subjects participated in the present study. The study was approved by the ethics committee of Nanjing Brain Hospital, and a consent form for this study was prepared. After all the informed consent was obtained, all of them were investigated by a structure questionnaire asking about sociodemographic characteristics, BPD and history of childhood maltreatment. According to the interpretation of Personality Diagnostic Questionnaire-4 (PDQ-4) and the purpose of this study, the participants with BPD score greater than or equal to 4 were considered to have borderline personality features while those who were less than 4 were considered to have no borderline features. Those who had previously been diagnosed with chronic heart, liver and kidney diseases; had a history of nervous 5 system diseases or mental disorders other than BPD; or had a long history of medication use were excluded.

Assessment of BPD
The Personality Diagnostic Questionnaire-4 (PDQ-4) was administered to assess borderline personality which is a further revision to PDQ based on DSM-IV [23]. This self-reported questionnaire consisted of 85 items, which are used to assess 10 types of personality disorders (paranoid, schizophrenic, split, hysteric, narcissistic, borderline, antisocial, avoidant, dependent and compulsive). In this study, the BPD subscale (9 items) of the PDQ was used in place of clinical diagnosis to assess the borderline personality traits of the subjects, with higher scores reflecting correspondingly higher levels of BPD.
The PDQ-4 demonstrates good reliability for assessing BPD across clinical and nonclinical samples [24][25]. The internal consistency as reflected by Cronbach's alpha for the 85 items in the present study was 0.93, and the internal consistency for BPD subscale was 0.73.

Assessment of childhood maltreatment
Childhood Trauma Questionnaire Short Form (CTQ-SF), compiled by Bernstein DP and Flink L [26], was administered to assess five categories of childhood abuse and neglect including emotional abuse, physical abuse, sexual abuse, emotional neglect and physical neglect. This instrument is a 28-item retrospective self-report questionnaire, and the items follow a Likert-type response style from 1 to 5, which is organized to reflect the frequency of abusive experiences (never true, rarely true, sometimes true, often true, very often true). Each dimension has 5 items, and the remaining 3 items are validity evaluation, so each dimension is scored between 5~25 and the total score is 25~125. The CTQ and its dimensions (other than physical neglect) have good reliability and validity in Chinese population [27]. The internal consistency as reflected by Cronbach's alpha for the 28 items in the present study was 0.77, and the internal consistency for each subscale was 0.79, 0.87, 0.88, 0.86, and 0.64, respectively.

DNA extraction and genotyping
DNA was extracted from 5 ml of venous blood in 367 BP subjects and 536 non-BP subjects. Based on 6 previous studies, eight SNPs in the OXTR was selected, including rs1042778, rs13316193, rs2254298, rs2268494, rs237889, rs237897, rs53576, rs6770632. Genotyping of the OXTR SNPs applied multiple PCR technology and high-throughput sequencing technology. Specific primers are designed for the loci that need to be detected. Multiple PCR amplification is carried out in a single tube. Different samples are distinguished by different Barcode primers, and the amplification is sequenced by high flux. The sequencing results were obtained using bioinformatics method to distinguish different samples.

Statistical analysis
Statistical analysis was performed using Statistical Product and Service Solutions (SPSS) 22.0 and process plug-in 3.0. First, the Hardy-Weinberg equilibrium for genotypes was assessed by a chisquared test to examine whether the sample was a genetic balance population. Second, descriptive statistics were used to presenting the sociodemographic characteristics of the subjects. Third, the differences in demographic characteristics and genotypes between the BP group and the non-BP group were examined by performing a series of t-tests and chi-squared tests. Fourth, association analysis between childhood maltreatment and BPD level was performed with linear regression. Lastly, the interaction between OXTR and childhood maltreatment was conducted using process plug-in 3.0.
All reported p values were two-sided, and p values less than 0.05 was considered to be statistically significant.

Subjects characteristics
According to the BPD scores all the subjects were divided into a BP group and a non-BP group, 575 cases and 1425 cases respectively. As presented in Table 1, groups were similar in age, education, and reproductive status (p > 0.05). Groups showed differences in marital status (p < 0.01). Compared with the non-BP group, the BP group had increased proportions of unmarried and married/cohabitating subjects, with decreased proportions of divorced/widowed subjects. Subjects in the BPD group obtained higher scores on the overall CTQ as well as its five subsets than those in the non-BPD group (p < 0.001). 7

Association analysis between individual SNP in OXTR and BP
Hardy-Weinberg genetic equilibrium test showed that, except rs1042778 in the non-BPD group (χ 2 = 7.807, p = 0.005), the SNPs conformed to Hardy-Weinberg equilibrium in both groups (χ 2 = 0.008-2.346, p > 0.05). As showed in Table 2, there were significant differences in the genotype frequencies of rs53576 and rs237897 between the two groups (p < 0.05). However, the five other markers, namely, rs13316193, rs2254298, rs2268494 and rs237889, rs6770632, were not found to be statistically significant.

Association analysis between childhood maltreatment and BPD
The impact of childhood maltreatment on BPD was examined by linear regression analysis, in which BPD score was the dependent variable and total score for childhood maltreatment was the independent variable. The analysis revealed that childhood maltreatment is a significant positive 0.378, p = 0.707, R 2 = 0.018, t = 1.337, p = 0.182, respectively), emotional neglect (R 2 = 0.032, t = 0.733, p = 0.464, R 2 = 0.016, t = 0.002, p = 0.999, respectively) and physical neglect (R 2 = 0.033, t = 0.748, p = 0.455, R 2 = 0.026, t = 0.354, p = 0.724, respectively) all failed to interact with genotype to predict BPD scores in either group. However, the moderating role of OXTR rs53576 genotype on the relation between emotional abuse, physical abuse, sexual abuse and BPD was significant in the prison group (R 2 = 0.086, t = 2.037, p = 0.042, R 2 = 0.076, t = 2.803, p = 0.005, R 2 = 0.062, t = 2.952, p = 0.003, respectively). As shown in Figure 1, when subjects had lower emotional abuse scores, those who carried the GG genotype had a lower risk of BPD than those who carried the A allele. While, when subjects had higher emotional abuse scores, those who carried GG genotype had a higher risk of BPD than those who carried A allele. The interaction between OXTR rs53576 genotype and physical abuse, sexual abuse was similar. As shown in Figure 2, at lower levels of physical abuse, subjects who carried GG genotype had lower BPD scores compared with those carried A allele, while at higher levels of physical abuse, subjects who carried the A allele had lower BPD scores than those who carried the GG genotype. As shown in Figure 3, at lower levels of sexual abuse, BPD scores were lower among those with GG genotype, while at higher levels of sexual abuse, BPD scores were higher among those with GG genotype.

Interaction between OXTR rs237897 and childhood maltreatment
When we analyzed the interaction between OXTR rs237897 and childhood maltreatment, we found a similar pattern of results. In a two-way interaction model with BPD as the dependent variable and the genotype of rs237897 as a moderator variable, childhood maltreatment and its dimensions were examined as independent variables. The analysis revealed that the interaction effect of genotype by childhood maltreatment total score was nonsignificant for BPD in the BP group and the non-BP group

Discussion
This study investigates genetic (OXTR) and environmental (childhood maltreatment) factors of BPD and highlights G × E interactions relevant to BPD risk. In terms of gene test results, the BP group had an increased rs53576 AA genotype frequency compared with the non-BP group, which indicated that the increase in AA genotype frequency may increase the risk of BPD. This result is inconsistent with previous finding that the genotype groups for OXTR did not demonstrate difference in BPD scores [13], however, these variations may be because different analysis methods were used to explore the relationship between rs53576 and BPD. In the rs237987 of OXTR SNP, the higher AA genotype frequencies were found in BP group, while associations between the specific oxytocin receptor marker rs237987 and BPD susceptibility have not previously been reported. Thus, the result needs to be verified in a larger and more representative sample.
With respect to the environmental aspect, the results showed that childhood maltreatment and its dimension could positively predict BPD risk in the present study. This is not surprising given that evidence has accumulated suggesting that early childhood maltreatment is a risk factor for BPD. For instance, a retrospective study suggested that children who had been maltreated had an increased risk of presenting with borderline personality features [28]. Another study in adult reported the association between experiencing traumatic events in childhood and an increased clinical severity of BPD in adulthood [29]{Ferrer, 2016 #16;Ferrer, 2017 #17}.
Interactive effects between OXTR rs53576 and childhood maltreatment on BPD scores were not observed in both the BP group and the non-BP group, which is in contrast to prior literature reporting a significant moderation of the link between maltreatment and BPD symptoms by OXTR rs53576 was found in a sample of teenagers [30]. However, the above mentioned research took gender factors into account, and the interaction effects demonstrated differential patterns among girls and boys, while the present study included only male subjects. Considering the evidence that oxytocin is more influential in female than male behavior [31], gender differences may partially explain the nonsignificant interactive effect between rs53576 and childhood abuse. Additionally, the results may be due to the complexity of BPD phenotype, as a study reviewed the gene and environment studies relevant to BPD and found that most of them were associated with specific symptoms of BPD such as impulsivity and emotional sensitivity to the environment [32]. The interaction between OXTR rs237987 and childhood maltreatment was not significant could have the same explanations. Hence, it will remain for future research to more precisely evaluate the moderating role of OXTR polymorphisms on the relation between childhood maltreatment and BPD.
Notably, the present study found that patterns of OXTR gene×childhood maltreatment interaction were observed in prison subjects. In the OXTR rs53576, the GG carriers were strongly affected whether the emotional abuse, physical abuse and sexual abuse low or high, which conforms to the hypothesis of gene plasticity [1]. These results are consistent with the prior literature showing that GG carriers may be more sensitive to social cues and more prosocial than A-allele carriers, while this attunement for social cues may be a disadvantage under adverse circumstances such as early childhood maltreatment [33]. In contrast, there are also studies reported that A-allele carriers had higher levels of BPD symptoms under negative family conditions and lower levels under positive conditions, GG homozygotes had average levels of BPD symptoms regardless of their family quality [13]. Another study in low-income children of different ethnicities found that girls who carried the A allele were more susceptible than those without the A allele, while boys with the GG genotype were more susceptible than those without the GG genotype [30]. Thus, it may mean there are different patterns of OXTR gene×childhood maltreatment interactions in different genders, which can also interpret the contradictory results in various studies. Future research will be needed to clarify the exact mechanism.
The explanation for the significant interaction between genes and environment should be cautious.
The prior literature reported that more adolescents whose mothers had BPD experienced maltreatment than those whose mothers were normal and maltreatment could predict borderline features [34], which revealed that the intergenerational transmission of BPD may be through childhood maltreatment. Another study showed that mothers' maltreatment history is a more strongly prediction for mother-infant attachment disorganization score among mothers with more plasticity alleles of OXTR [35], which suggested that maternal maltreatment history would influence offspring depending on the maternal genetic characteristics. Hence, both genetic and environmental factors are contributed to the development of BPD. Whether adverse environment lead to changes in biomarkers or genetic basis increases susceptibility to adverse environment remain future research to explore.
This study has several strengths, including methodological advantages, comprehensive consideration of loci of OXTR, and the particularity of the selected subjects. First, the current study included a relatively large sample which increased the accuracy and rigor of the study. In addition, on the basis of previous research and the biological information analysis of BPD, eight loci of the OXTR gene were selected in this study. Finally, to our knowledge, this is the first study to demonstrate interactive effects between OXTR polymorphisms and childhood maltreatment subtypes on BPD levels in Chinese male inmates. Given the characteristics and high risk for BPD, this population is worth investigating.
The findings that the patterns of G×E interactions in prison sample contribute both to prison management and to treatment or even prevention of BPD.
This study has several limitations that need to be acknowledged. The current sample consists of male inmates, a population that is at risk for BPD. Thus, it is possible that the results would differ in the general population. Given that the different effects of oxytocin in different genders, the findings also cannot be generalized to females. Moreover, although CTQ-SF has a good reliability in this study, retrospective self-reporting cannot avoid recall bias, especially in the high-risk population. The subjects who had higher levels of BPD were likely to exaggerate the adverse events they experienced in early life, which would inflate the association between childhood maltreatment and BPD. In addition, the onset and duration of childhood maltreatment were not considered, so the interpretation and application of the results should be cautious.

Conclusions
In summary, the findings of the present study indicate that OXTR gene variations and childhood maltreatment are both risk factors for the development of BPD and that OXTR polymorphisms interact with childhood maltreatment subtype to predict BPD scores. Specifically, OXTR rs53576 GG carriers who had higher levels of emotional, physical or sexual abuse had higher BPD scores than A-allele carriers, while those with lower levels of emotional, physical or sexual abuse had lower BPD scores than A-allele carriers. These interaction findings reinforce the view that OXTR is a plasticity gene that is associated with increased sensitivity to the environment, not only for positively valenced but also for negatively valenced stimuli [36][37]. The results may contribute to understand the pathological mechanism of BPD, further guiding the early identification of personality disorders and the selection of interventions.   Figure 1 The interactive effect of OXTR rs53576 genotype and childhood emotional abuse on BPD scores.
19 Figure 2 The interactive effect of OXTR rs53576 genotype and childhood physical abuse on BPD scores.
20 Figure 3 The interactive effect of OXTR rs53576 genotype and childhood sexual abuse on BPD scores