آیا برخی از نوجوانان به تاثیر زورگویی بر افسردگی حساس هستند؟

Abstract Not all children react to being bullied in the same way. Some children may be more predisposed to depression when bullied than are other children. Using a G × E approach, this study examined the influence of a genetic polymorphism in the serotonin transport gene (5-HTTLPR) on the victimization-depression link. The validity of the diathesis-stress versus environmental susceptibility hypothesis was tested. A total of 157 adolescents (Mage = 12.21) took part in this study. For adolescents with the S,S/S,L variants, victimization was positively related to depression. No relationship between victimization and depression was found for children with the L,L variant. Findings further suggest that the influence of 5HTTLPR on the association between victimization and depression more closely follow an environmental susceptibility model.

1. Introduction Although a certain degree of exposure to aggressive behaviors is likely to be normal for most children and adolescents, some children are the repeated target of abuse by their peers. This constant exposure to the negative aggressive actions (e.g., teasing, threats, rumors, physical harm) of one or more peers or classmates is often termed peer victimization or bullying ( Andreou, 2001 and Olweus, 2001). Estimates of the percentage of American children who report being victimized by their peers have ranged from 10% to 30% ( Grills and Ollendick, 2002, Limber and Small, 2003 and Nansel et al., 2001). This statistic is exceedingly important when one considers the potential for harm that peer victimization may cause. Being bullied by peers has been associated with a myriad of long-term psychological and physical problems ( Arseneault et al., 2010, Cleary, 2000, Knack et al., 2011 and Miller and Vaillancourt, 2007). The current study examined whether being bullied predicted increases in depression and anxiety in early adolescence. Perhaps more importantly, this study focused on a potential genetic moderator to the victimization-depression association, namely the influence of the serotonin transport gene 5-HTTLPR polymorphism, which has been associated with physiological responsiveness to chronic stress (Moffitt, Caspi, & Rutter, 2005). We anticipated that young adolescent children with at least one S allele in 5-HTTLPR would be more susceptible to the negative influences of bullying on depression and anxiety than would adolescent children who were L,L homozygote (i.e., we anticipated a G × E interaction). Although people can be bullied at any point in the lifespan, adolescence is a particularly important developmental period to study. As of 2009, adolescents were a part of the largest group in the world at 1.2 billion, accounting for approximately 18% of the world population (UNICEF, 2011). The sheer number of adolescents sheds light on the significance of studying this age group. Of greater consequence, early adolescence is associated with significant changes in biological, cognitive, and social functioning that are rivaled only by those changes seen in infancy (Larson & Richards, 1994). For example, early adolescent children undergo a number of major physiological changes including puberty, changes in the endocrine and immune systems, and rapid brain maturation (Golub et al., 2008 and Walker et al., 2001). Adolescents also undergo a number of changes in their social environment (Larson & Richards, 1994). Peers and friends begin to play an increasingly important socialization role (Hartup, 1996). As friendship intimacy increases during adolescence and supplements intimacy with parents, adolescents also begin to rely more on friends to satisfy needs and solve problems that arise (Furman & Buhrmester, 1985). Both longitudinal and cross-sectional studies have found that peer victimization also peaks during this time (Nansel et al., 2001 and Nylund et al., 2007). Additionally, the onset of mental illnesses, such as depression, anxiety disorders, and mood disorders, increases substantially during adolescence as a result of a greater vulnerability to stress that occurs as the adolescent goes through major physiological changes (e.g., Paus et al., 2008). Adolescent anxiety and depressive disorders show an approximate two to three fold increased risk for adulthood anxiety (Pine, Cohen, Gurley, Brook, & Ma, 1998). Depression is the second leading cause of disability adjusted life years (DALYs; i.e., productive years lost due to premature death or physical disability) starting at 15 years (Murray and Lopez, 1997 and World Health Organization, 2011) and is often an early symptom of other physical health problems (e.g., Henningsen et al., 2003 and Penninx et al., 1999). Finally, depression is a major risk factor for suicide, which is the 3rd leading cause of child and adolescent mortality in the United States (Centers for Disease Control and Prevention (CDC), 2011). Ironically, it is during this period, when adolescents are going through major physiological changes and are the most vulnerable to depression and anxiety symptoms that they are also most likely to become victims of peer abuse (Eisenberg & Aalsma, 2005). A meta-analysis conducted by Hawker and Boulton (2002) found that being bullied was most strongly related to depression, more so than any other component of psychological distress. Bullying was also associated with increased risk of suicidal ideation, suicide attempts, and suicide (Hay and Meldrum, 2010, Kaminski and Fang, 2009 and Kim and Leventhal, 2008). For example, data collected by the Centers for Disease Control and Prevention as part of the Youth Risk Behavior Surveillance System revealed that there was a 73% increased likelihood that the adolescent would attempt suicide at least once if the adolescent was bullied compared to if the adolescent was not bullied (Centers for Disease Control and Prevention (CDC), 2009). Perhaps more ominous is the fact that the association between victimization and depression seems to hold steady over time. Depression and anxiety is also much higher for those individuals that continue to be bullied over time compared to those who are only bullied for a more limited period (Menesini, Modena, & Tani, 2009). Finally, this association between bullying and depression appears to be bi-directional in nature. Victimized children are more likely to become depressed, but depressed children are also more likely to be victimized because they exhibit distress that is often rewarding for the bullies, creating a viscous cycle of bullying and depression for these children (e.g., Arseneault et al., 2010 and Egan and Perry, 1998). Not all adolescents succumb to depression and anxiety when bullied. One possibility for this variability in adjustment outcomes is that individual characteristics of the child interact with environmental influences to produce outcomes. That is, some children may be constitutionally more at risk for developing depression and anxiety problems when exposed to bullying from peers than are other children. The notion that the developmental outcomes of a particular individual involve the interaction between hereditary and environmental factors is widely accepted (Caspi & Moffit, 2006; Plomin, Owen, & McGuffin, 1994). Recently, behavior geneticists have begun to examine the interaction between specific genotypes and environmental factors in relation to outcomes (Moffitt et al., 2005). Indeed, one specific functional polymorphism, namely a functional polymorphism in the promoter region of the serotonin transporter (5-HTT) gene has been associated with depression and anxiety in adults and children who have experienced adverse environmental influences. 5-HTTLPR codes for the serotonin transport protein 5-HTT that is responsible for the reuptake of extracellular serotonin back into the presynaptic neuron (Bah et al., 2007). The gene that codes for 5-HTT (SLC6A4) is found on the 17q12 chromosome (Bah et al., 2007, Holmes and Hariri, 2003 and Holmes et al., 2002). Humans have a 44-base pair insertion/deletion polymorphism found in the promoter region of the 5-HTT gene which is further up from the coding region of SLC6A4 (the serotonin transport linked polymorphic region, 5-HTTLPR) (Heinz et al., 2000 and Lesch et al., 1996). This gene can be comprised of a long variant (L) and a short variant (S) (Heinz et al., 2000 and Lesch et al., 1996). Those individuals who carry the L variant have 16 repeats within the gene while those who carry the S- variant have 14 repeats within the gene (Bah et al., 2007). About 30% of the human population carries the two L variant genes, 50% one S and one L, and finally, 20% of the human population carries two S variants of the gene (Caspi et al., 2003). The S variant is the less active allele; this short allele has been shown to decrease the transcriptional efficiency of the 5-HTT gene promotor (Bah et al., 2007 and Lesch et al., 1996). Individuals who have either one or two copies of the short (S) allele are known to have fewer serotonin transporters than those persons with an L,L genotype (Heinz et al., 2000, Holmes and Hariri, 2003 and Lesch et al., 1996). In addition, individuals with at least one S allele show greater amygdala activation to fearful stimuli (Hariri et al., 2002) and are at greater risk for developing depression, anxiety-related traits, and psychiatric disorders (Caspi et al., 2003, Hairiri et al., 2005, Lesch et al., 1996, Melke et al., 2001, Sen et al., 2004 and Serretti et al., 2006). Some of the first evidence for 5-HTTLPR influencing the onset of depression after stressful life events was reported by Caspi et al. (2003) as part of a longitudinal study. The polymorphism of the 5-HTTLPR gene moderated the link between stressful life events and depression. Specifically, those individuals who had one or two variants of the S-allele were found to exhibit more depressive symptoms, were more often diagnosed with depression, and had more suicidal thoughts as a result of stressful events than those individuals containing the two L alleles. At least fifteen independent studies have found a gene-environment interaction between the 5-HTTLPR polymorphism and environmental adversities that have been associated with depression (e.g., Caspi et al., 2003, Hairiri et al., 2005, Lesch et al., 1996 and Serretti et al., 2006). Several recent studies have begun to examine the influence of 5-HTTLPR on the bullying-depression association. For example, Benjet (2010) found that 5-HTTLPR moderated the influence of relational victimization on depression in a sample of 78 adolescent girls. The polymorphism of 5-HTTLPR alone did not significantly predict depression suggesting that some adolescents may be more susceptible to depression but only when bullied. Similarly, Sugden et al. (2010) found that victimized adolescents with the S,S genotype had more emotional problems than victimized adolescents with the L,L genotype. Again, they did not find a main effect for 5-HTTLPR. One potential problem with most of these G × E interaction explanations is that they have been framed exclusively in terms of risk. In other words, researchers have suggested that that individuals who carry at least one S allele may be more vulnerable to expressing specific negative developmental outcomes (e.g. depression) only if they additionally face environmental adversity such as being bullied (Monroe & Simons, 1991). Interestingly, research has not only found that individuals with at least one S allele report having more maladaptive problems than the L,L homozygote if they experience adverse environments (Fox et al., 2005 and Kochanska et al., 2011), but it also appears that these individuals with at least one S allele often appear to do better in more optimal environments than people with the L,L genotype (For examples, see Benjet, 2010 and Caspi et al., 2003). In other words, the G × E interaction appears to cross within the possible range of scores suggesting a disordinal interaction (see Aiken and West (1991, p. 22) for more details). As such, it is possible that having at least one short allele may simply make a person more malleable to both positive and negative environmental influences. Indeed, Boyce and Ellis (2005) concluded that the diathesis stress model did not take into account what they deemed as “differential susceptibility” in GXE interactions. This sentiment was also shared by Taylor et al., 2006, Belsky et al., 2009 and Belsky and Pluess, 2009 in their findings on other G × E studies. As such, it may be better to conceptualize the S allele of 5-HTTLPR to be more involved with malleability or plasticity to environmental influences than simply increasing vulnerability or risk when exposed to adverse environments. As found by many studies, adverse or suboptimal environments are often more likely to yield negative effects for individuals with at least one S allele. However, these same individuals may be more likely to benefit from positive support and thrive more substantially than L,L homozygote individuals in similar optimal situations ( Uher & McGuffin, 2008). Kochanska, Kim, Barry, and Philibert (2011) were the first to explicitly examine whether 5-HTTLPR was associated with greater plasticity to environmental influences rather than simply vulnerability. They found that children with S,S/S,L alleles whose mothers were high in responsiveness were more competent than those children whose mothers were less responsive. Moreover, mother’s responsiveness had no effect on competence for children with the L,L homozygote genotype. Most interesting however, although S,S/S,L children whose mothers were less responsive reported worse outcomes than L,L homozygote children, S,S/S,L children who had responsive mothers, had better outcomes than L,L homozygote children. This finding suggests evidence of a differential susceptibility model (i.e., greater malleability) rather simply than a diathesis-stress model (i.e., greater vulnerability only). Despite the previous studies that have examined the G × E interaction of bullied children, no study to date has directly compared the two competing G × E interaction models: Diathesis-stress versus differential susceptibility (Kochanska et al., 2011). The Diathesis-Stress hypothesis would predict that bullied adolescents with at least one S allele would report higher levels of depression as a result of the social stress associated with being peer victimized. That is, peer victimization would act as the catalyst for depression, but only for adolescents with at least one S allele. Moreover, there would be no differences between adolescents with at least one S allele and adolescents with two L alleles in more optimal environments (i.e., low levels of bullying). Alternatively, the differential susceptibility hypothesis would predict adolescents with at least one S allele would be influenced by both positive and negative peer interactions. That is, not only would adolescents with at least one S allele be more likely to report higher levels of depression when bullied but they would also report lower levels of depression when exposed to positive peer environments (i.e., very low levels of peer abuse) compared to adolescents with the L,L genotype. In other words, the influence of the 5-HTTLPR gene variant on depression should be evidenced at both ends of the victimization spectrum. It is also possible that genetic factors may predispose some children to being the targets of victimization. Being bullied is not a completely random event (e.g., “being in the wrong place at the wrong time”). Rather, attributes of the child can influence the likelihood that he or she will become a peer victim. Such characteristics include poor self-control as well as exhibiting internalizing and externalizing problems; all of which are influenced by genetic factors (Arseneault et al., 2010 and Jensen-Campbell et al., 2009). More specifically, Ball et al. (2008) found genetic factors accounting for 73% of the variation in the victimization of children in their sample suggesting that genetic endowments may contribute to an individual being the recipient of aggressive peer acts. As such, we also explored the possibility that genetic variations of 5-HTTLPR may predispose individuals to experience peer victimization, which in turn leads to greater depression (i.e., a mediation model).

3. Results 3.1. Does 5-HTTLPR predict victimization? First, we examined whether our participants’ age, ethnicity, and gender were associated with either 5-HTTLPR or victimization. There was no evidence that our participant demographics were associated with either victimization or 5-HTTLPR (see Table 1). Second, we examined the degree of relation between the 5-HTTLPR and victimization. It is possible that a person’s genetic make-up may make them more susceptible to bullying. That is, adolescents with at least one S allele may be more anxious or withdrawn initially, which makes them a greater target for bullying and in turn leads to greater depression and anxiety over time (Ball et al., 2008). We examined this possible mediation by examining the bivariate correlations between victimization and 5-HTTLPR. Higher numbers in 5-HTTLPR represented higher numbers of S alleles (i.e., L,L = 0, S,L = 1, and S,S = 2). In order for the mediation model to be plausible, there must be a relationship between 5-HTTLPR and peer victimization (see Baron & Kenny, 1986). In other words, 5-HTTLPR status (the independent variable) must be correlated with victimization (the mediator) for mediation to occur. There was no evidence that victimization was related to 5-HTTLPR; the average correlation among measures was −0.11, ns (Range: −0.03 to −0.13). Additionally, there was no evidence that the two victim groups were significantly different from one another on 5-HTTLPR status, χ2(2) = .597, p = .742. In sum, there was no evidence that variations in 5-HTTLPR influenced the likelihood of being peer victimized (see Table 2 for percentages). Table 2. Victimization membership by 5-HTTLPR status. 5HTT Total S,S S,L L,L Victims 9 (21.4%) 18 (42.9%) 15 (35.7%) 42 (100%) Non-victims 25 (21.7%) 56 (48.7%) 34 (29.6%) 115 (100%) Total 34 (21.7%) 74 (47.1%) 49 (31.2%) 157 (100%) Table options 3.2. Is victimization related to depression and anxiety? For the main analyses, types of victimization were treated as a continuous variable to (1) eliminate any reductions in power associated with treating victimization as a categorical variable and (2) prevent both the loss of information and the creation of additional sources of measurement error due to categorization (Aiken & West, 1991). Bivariate correlations revealed that being bullied was related to both depression and anxiety (see Table 3). Consistent with previous studies, type of victimization did not significantly influence the relationship between being bullied and depression or anxiety (e.g., Klomek, Marrocco, Kleinman, Schonfeld, & Gould, 2008). Table 3. Bivariate correlations: the influence of victimization on depression and anxiety. Gender Measures Achenbach DSM affective disorder Achenbach DSM anxiety disorder Depression (D-inventory) Depression (AD + D-inventory) Overall Overall Victimization 0.34⁎⁎ 0.22⁎⁎ 0.50⁎⁎ 0.47⁎⁎ Overt/Physical (CSEQ) 0.21⁎⁎ 0.15+ 0.35⁎⁎ 0.32⁎⁎ Relational (CSEQ) .031⁎⁎ 0.13+ 0.36⁎⁎ 0.37⁎⁎ Verbal (DIAS) 0.33⁎⁎ 0.24⁎⁎ 0.51⁎⁎ 0.47⁎⁎ Physical (DIAS) 0.16⁎ 0.12 0.33⁎⁎ 0.27⁎⁎ Indirect (DIAS) 0.41⁎⁎ 0.22⁎⁎ 0.50⁎⁎ 0.51⁎⁎ Male (Boy) Overall Victimization 0.17 0.22+ 0.47⁎⁎ 0.39⁎⁎ Overt/Physical (CSEQ) 0.12 0.11 0.32⁎ 0.27⁎ Relational (CSEQ) 0.23+ 0.25⁎ 0.40⁎⁎ 0.37⁎⁎ Verbal (DIAS) 0.15 0.26⁎ 0.50⁎⁎ 0.40⁎⁎ Physical (DIAS) 0.08 0.11 0.32⁎ 0.25⁎ Indirect (DIAS) 0.18 0.21+ 0.49⁎⁎ 0.41⁎⁎ Female (Girl) Overall Victimization 0.53⁎⁎ 0.27⁎ 0.54⁎⁎ 0.60⁎⁎ Overt/Physical (CSEQ) 0.39⁎⁎ 0.27⁎ 0.42⁎⁎ 0.45⁎⁎ Relational (CSEQ) 0.37⁎⁎ 0.06 0.32⁎⁎ 0.39⁎⁎ Verbal (DIAS) 0.53⁎⁎ 0.30⁎⁎ 0.54⁎⁎ 0.59⁎⁎ Physical (DIAS) 0.37⁎⁎ 0.24⁎ 0.38⁎⁎ 0.42⁎⁎ Indirect (DIAS) 0.50⁎⁎ 0.21⁎ 0.51⁎⁎ 0.55⁎⁎ Although the magnitude of the relationships appeared stronger for girls than for boys, t-tests for independent correlations revealed only three significant differences between boys and girls (overall-affective disorder, verbal-affective disorder, and indirect-affective disorder). ⁎ Correlation is significant at the 0.05 level (2-tailed). ⁎⁎ Correlation is significant at the 0.01 level (2-tailed). Table options However, the relationship between being bullied and depression was stronger for relational, verbal, and indirect forms of victimization compared to more physical forms of victimization. For example, when we collapse across measures of physical (overt-CSEQ and physical-DIAS) and relational (relational-CSEQ, indirect-DIAS, and verbal-DIAS) victimization and use the test for differences between dependent correlations (Bruning & Kintz, 1997), the relationship between overall relational victimization and reporting symptoms associated with affective disorder (r = 0, … , 39) was significantly larger than the relationship between physical victimization and these same symptoms (r = 0.20), t (153) = 3.068, p < .01. Similarly, the association between overall relational victimization and overall depression (r = 0.51) was significantly larger than the association between overall physical victimization and the same depressive symptoms (r = 0.32), t (153) = 3.19, p < .01. There was no evidence of differences between sub-types of victimization and anxiety problems, (rs = 0.22, 0.15), t (153) = 1.09, ns. A moderated multiple regression analysis that included sex of participant, overall victimization, and their cross-product was also run. There was a significant sex of participant × victimization interaction, B = 0.20, SE = 0.07, t(147) = 2.93, p = .004. Although victimization was positively associated with depression for both boys and girls, the magnitude of the relationship between overall victimization and depression was stronger for girls (B = 0.72, SE = 0.10; t = 6.91, p < .001, sr = 0.60) than for boys (B = 0.31, SE = 0.09, t = 3.40, p = .001, sr = 0.393). 3.3. Does 5-HTTLPR moderate the link between victimization and depression? Next we examined whether 5-HTTLPR moderated the link between victimization and depression. Bivariate correlations revealed that the relationship between overall victimization and depression were significant only for adolescents who had at least one S variant of 5-HTTLPR (see Table 4). To further assess whether victimization, 5-HTTLPR, and their interaction predict depression and anxiety, a series of moderated multiple regression (MMR) were run. Each type of victimization was centered and treated as a continuous variable. 5-HTTLPR was coded using unweighted effects4 codes (Aiken & West, 1991). Finally, the cross-product between victimization and each polymorphism code was created. Victimization, the unweighted effects codes, and their cross-products were then entered in the equation. Sex of participant was also entered into the model as a control variable using unweighted effects codes.5 Dependent measures included overall depression (i.e., averaged affective problems (CBCL and YSR) and depression (both self- and parent-reports)) and anxiety (averaged CBCL and YSR anxiety problems). Following significant interactions, simple effects analyses for victimization were conducted using the appropriate dummy codes for 5-HTTLPR (see Aiken and West (1991, pp. 130–133), for a review). Table 4. Correlations among victimization and internalizing problem measures by 5HTTLPR status. DNA 5HTT alleles Measures Achenbach DSM affective disorder Achenbach DSM anxiety disorder Depression (D-inventory) Depression (AD + D-inventory) S,S N = 34 Overall Victimization 0.26 (0.45⁎) 0.27 (0.51⁎⁎) 0.62⁎⁎ (0.58⁎⁎) 0.54⁎⁎ (0.59⁎⁎) Overt/Physical (CSEQ) 0.24 (0.40⁎) 0.34⁎ (0.49⁎⁎) 0.49⁎⁎ (0.37⁎) 0.45⁎⁎ (0.45⁎) Relational (CSEQ) 0.23 (0.34+) 0.17 (0.16) 0.51⁎⁎ (0.35⁎) 0.46⁎⁎ (0.40⁎) Verbal (DIAS) 0.22 (0.35+) 0.27 (0.45⁎) 0.65⁎⁎ (0.59⁎⁎) 0.54⁎⁎ (0.55⁎⁎) Physical (DIAS) 0.17 (0.19) 0.18 (0.41⁎) 0.38⁎ (0.27) 0.34⁎ (0.27) Indirect (DIAS) 0.22 (0.30+) 0.18 (0.25) 0.52⁎⁎ (0.39⁎) 0.45⁎⁎ (0.39⁎) S,L N = 74 Overall Victimization 0.52⁎⁎ 0.17 0.52⁎⁎ 0.58⁎⁎ Overt/Physical (CSEQ) 0.39⁎⁎ 0.14 0.35⁎⁎ 0.41⁎⁎ Relational (CSEQ) 0.39⁎⁎ 0.01 0.33⁎⁎ 0.40⁎⁎ Verbal (DIAS) 0.50⁎⁎ 0.24⁎ 0.51⁎⁎ 0.57⁎⁎ Physical (DIAS) 0.31⁎⁎ 0.11 0.40⁎⁎ 0.40⁎⁎ Indirect (DIAS) 0.59⁎⁎ 0.19 0.53⁎⁎ 0.63⁎⁎ L,L N = 49 Overall Victimization 0.11 0.12 0.34⁎ 0.24 Overt/Physical (CSEQ) −0.02 0.07 0.25 0.12 Relational (CSEQ) 0.21 0.25+ 0.21 0.24+ Verbal (DIAS) 0.12 0.20 0.38⁎ 0.26+ Physical (DIAS) −0.05 0.10 0.15 0.04 Indirect (DIAS) 0.21 0.26 0.39⁎ 0.32⁎ Diagnostic analyses revealed a cluster of four outliers in the S,S variant group: two Hispanic and two Caucasian boys who reported extremely high levels of victimization, but moderate levels of depression and anxiety on the Achenbach measures. Without the outliers, the magnitude of the relationship victimization–depression was stronger for S,S children. Numbers in parenthesis are the correlations with the outliers removed. + Correlation is significant at the 0.05 (1-tailed). ⁎ Correlation is significant at the 0.05 level (2-tailed). ⁎⁎ Correlation is significant at the 0.01 level (2-tailed). Table options Our overall measure of victimization was related to depression (B = 0.48, t (147) = 6.52, p < .001, sr = 0.44) and anxiety (B = 0.29, t (147) = 2.86, p = .005, sr = 0.23). As anticipated, victimized children had greater levels of depression and anxiety. Sex of participant was also related to depression (B = 0.16, t (147) = 2.94, p = .004, sr = 0.20) and anxiety (B = 0.18, t (147) = 2.44, p = .016, sr = 0.19). Girls reported more depression and anxiety than did boys. There was no evidence of a 5-HTTLPR main effect for either depression or anxiety, Fs (2, 147) = .94, .21, p = .394, ns p = .812, ns, respectively. There was no evidence of a 5-HTTLPR X overall victimization interaction for anxiety, F(2, 147) = .11, ns. Similar results were found for all sub-types of victimization (see Table 4 and Table 5). Table 5. Type of Victimization, adolescents’ 5-HTTLPR status, and their interaction as predictors of depression and anxiety. Predictors Achenbach DSM affective disorder Achenbach DSM anxiety disorder Depression (D-inventory) Depression (AD + D-inventory) ΔF ΔR2 ΔF ΔR2 ΔF ΔR2 ΔF ΔR2 Overall 8.13⁎⁎ 2.41⁎ 10.34⁎⁎ 11.41⁎⁎ Victimization 19.69⁎⁎ 0.10 8.16⁎⁎ 0.05 47.52⁎⁎ 0.23 42.57⁎⁎ 0.20 5HTTLPR 0.53 0.01 0.21 0.003 1.08 0.01 0.94 0.01 Victimization × 5HTTLPR 5.01⁎⁎ 0.05 0.11 0.001 3.12⁎ 0.03 4.96⁎⁎ 0.05 Overt 5.45⁎⁎ 2.22⁎ 5.18⁎⁎ 6.31⁎⁎ Victimization 12.15⁎⁎ 0.07 7.02⁎⁎ 0.04 23.41⁎⁎ 0.13 22.72⁎⁎ 0.12 5HTTLPR 0.67 0.01 0.36 0.004 1.27 0.01 1.27 0.01 Victimization × 5HTTLPR 4.24⁎ 0.05 0.98 0.01 1.99+ 0.02 3.77⁎ 0.04 Relational 4.92⁎⁎ 1.68 4.64⁎⁎ 5.58⁎⁎ Victimization 11.65⁎⁎ 0.07 2.81+ 0.02 18.89⁎⁎ 0.11 19.64⁎⁎ 0.11 5HTTLPR 0.59 0.01 0.34 0.004 1.77 0.02 1.34 0.02 Victimization × 5HTTLPR 1.11 0.01 0.99 0.01 1.11 0.01 0.89 0.01 Physical 4.41⁎⁎ 1.74 5.04⁎⁎ 5.62⁎⁎ Victimization 8.01⁎⁎ 0.05 4.51⁎ 0.03 18.93⁎⁎ 0.11 16.74⁎⁎ 0.09 5HTTLPR 0.8 0.01 0.43 0.01 1.88 0.02 1.64 0.02 Victimization × 5HTTLPR 2.97⁎ 0.034 0.08 0.01 2.55+ 0.03 3.69⁎ 0.04 Verbal 8.07⁎⁎ 2.92⁎ 11.03⁎⁎ 11.60⁎⁎ Victimization 18.97⁎⁎ 0.10 10.28⁎⁎ 0.06 52.28⁎⁎ 0.24 44.11⁎⁎ 0.2 5HTTLPR 0.39 0.004 0.14 0.002 0.93 0.01 0.7 0.01 Victimization × 5HTTLPR 4.92⁎⁎ 0.05 0.12 0.001 3.28⁎ 0.03 4.78⁎ 0.04 Indirect 8.70⁎⁎ 1.98+ 9.56⁎⁎ 11.41⁎⁎ Victimization 19.89⁎⁎ 0.10 5.27⁎ 0.03 42.43⁎⁎ 0.21 40.07⁎⁎ 0.19 5HTTLPR 0.54 0.01 0.22 0.003 1.2 0.01 0.92 0.01 Victimization × 5HTTLPR 5.97⁎⁎ 0.06 0.08 0.001 2.01+ 0.02 4.94⁎⁎ 0.05 + Significant at the 0.05 level (1-tailed). ⁎ Significant at the 0.05 level (2-tailed). ⁎⁎ Significant at the 0.01 level (2-tailed). Full-size table Table options As predicted, there was an interaction between 5-HTTLPR and overall victimization for depression, F(2, 147) = 4.96, p = .008, ΔR2 = 0.05. For children with the S,S variant, victimization was positively related to depression, B = 0.53, t(147) = 3.42, p = .001, sr = 0.23. For children with the S,L variant, victimization was also positively related to depression, B = 0.70, t(147) = 6.56, p < .001, sr = 0.45. For children with the L,L variant, there was no evidence that victimization was related to depression, B = 0.21, t(147) = 1.84, p = .07, sr = 0.125. As predicted, children with at least one S variant of the 5-HTTLPR were more susceptible to depression when victimized. Similar 5-HTTLPR X Victimization interactions were found for all sub-types of victimization, with the exception of relational victimization as assessed by the CSEQ (see Table 4 and Table 5). To examine whether this 5-HTTLPR × Victimization interaction suggested a diathesis –stress or differential susceptibility G × E model, we first collapsed our S,S and S,L group to represent adolescents with at least one S allele versus adolescents who were L,L for 5-HTTLPR. This approach to examining the 5-HTTLPR G X E interaction has been used previously (e.g., Kochanska et al., 2011). Using the two-group approach to victimization, the 5-HTTLPR × Victimization interaction was still significant, B = −.21, t (149) = −3.003, p = .003, sr = −.204. Simple effects analyses again revealed that victimization was positively associated with depression for adolescents with at least one S allele, B = 0.65, SE = 0.09; t (149) = 7.34, p < .001, sr = 0.50. Victimization was not associated with depression for adolescents who were homozygous for L alleles, B = 0.21, SE = 0.12, t (149) = 1.84, p = 0.07, sr = 0.21. Given the interaction was significant, we next examined whether our G × E interaction provided evidence for the diathesis-stress versus the environmental susceptibility hypothesis. Using regions of significance (RoS) analyses outlined in Aiken and West (1991, pp. 134–137), we attempted to identify where the two simple regression lines (S,S/S,L versus L,L) were significantly different from one another. In other words, we wanted to determine at what value(s) of victimization the two genotype groups (i.e., S,S/S,L versus L,L) differed significantly from each other on depression. If only the high end range cut-off value was within the possible range of victimization, it would provide support for the diathesis-stress model. If both the low and high end cut-off values are within the possible range of victimization, it would provide support for the environmental susceptibility hypothesis. In Fig. 1, we graphed the interaction to represent ±2 SD on peer victimization to illustrate where adolescents with at least one S allele will show significantly different levels of depression compared to adolescents with the L,L variant. Our simple regression slopes crossed at 0.36, suggesting that our interaction was disordinal (i.e., the simple slopes crossed within the possible range of victimization scores). Additionally, we computed the proportion affected (PA) index, which is the proportion of adolescents who fall above the crossover point. Roisman et al. (2012) suggest that larger the percentage that falls above the cutoff, the stronger the support for differential susceptibility compared to diathesis-stress, with at least 16% of the sample needing to be above the cutoff. In our sample, 23.7% of the sample scored about the cut-off of 0.36, suggesting more support for differential susceptibility versus diathesis-stress. In addition, approximately 6% of our sample was two SDs above the mean on our overall victimization measure. A total of 12% of our sample was one SD below the mean on overall peer victimization; none of our participants were two SD below the mean on peer victimization. Adolescents’ genotype moderate the influence of overall victimization on ... Fig. 1. Adolescents’ genotype moderate the influence of overall victimization on depression. Figure options Using RoS analyses, S,S/S,L adolescents showed significantly higher levels of depression at the value of 2.00 than did L,L adolescents, which was within the possible observed range of victimization given 6% of our sample scored above this value on our overall victimization measure. In addition, S,S/S,L adolescents showed significantly less depression at the values of −0.55 than did L,L adolescents did, which was again within the possible range of victimization. Approximately 31% of our sample scored below this value on the overall victimization measure. Finally, following procedures outlined by Roisman et al. (2012), we calculated the Proportion of the Interaction (PoI) index to examine whether sample size alone influenced our RoS findings. The PoI, which is independent of sample size, measures of the degree to which our GxE interaction conforms to the prototypic differential susceptibility interaction. Values on the PoI index that fall between 0.40 and 0.60 are thought to be consistent with the differential susceptibility hypothesis (see Roisman et al. (2012, pp. 394–395), for more details). Our PoI was 0.42, providing additional evidence for differential susceptibility. In sum, our data most closely resemble the environmental susceptibility model for G × E interactions; adolescent children with at least one S allele seemed to be more malleable to environmental influences (i.e., is bullying) at both ends of the continuum.