Gene-environment interaction and psychiatric disorders: Review and future directions

G Model

ARTICLE IN PRESS

YSCDB-2424; No. of Pages 11

Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Seminars in Cell & Developmental Biology journal homepage: www.elsevier.com/locate/semcdb

Review

Gene-environment interaction and psychiatric disorders: Review and future directions Elham Assary 1 , John Paul Vincent 1 , Robert Keers, Michael Pluess ∗ Department of Biological and Experimental Psychology, School of Biological and Chemical Sciences, Queen Mary University of London, London, E14NS, United Kingdom

a r t i c l e

i n f o

Article history: Received 19 June 2017 Received in revised form 16 October 2017 Accepted 16 October 2017 Available online xxx Keywords: Genetics Gene-Environment interaction Diathesis-Stress Differential Susceptibility Psychiatry

a b s t r a c t Empirical studies suggest that psychiatric disorders result from a complex interplay between genetic and environmental factors. Most evidence for such gene-environment interaction (GxE) is based on single candidate gene studies conducted from a Diathesis-Stress perspective. Recognizing the short-comings of candidate gene studies, GxE research has begun to focus on genome-wide and polygenic approaches as well as drawing on different theoretical concepts underlying GxE, such as Differential Susceptibility. After reviewing evidence from candidate GxE studies and presenting alternative theoretical frameworks underpinning GxE research, more recent approaches and findings from whole genome approaches are presented. Finally, we suggest how future GxE studies may unpick the complex interplay between genes and environments in psychiatric disorders. © 2017 Elsevier Ltd. All rights reserved.

Contents 1.

2.

3.

4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.1. Candidate gene-X-environment (GxE) studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.1.1. 5-HTTLPR and depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.1.2. MAOA and antisocial behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.1.3. COMT and schizophrenia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.1.4. DRD4 and attention deficit hyperactivity disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.2. Theoretical frameworks in GxE research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.2.1. Genetic sensitivity rather than vulnerability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 1.3. Limitations of candidate GxE studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Genome-wide approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2.1. Genome-wide environment interaction studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 2.2. Polygenic score-X-environment interaction studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.1. Obtaining better quality environmental measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.2. Focus on transdiagnostic phenotypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.3. Applying a developmental perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.4. New analytical approaches and study designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.4.1. Vulnerability genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.4.2. Vantage sensitivity genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 3.4.3. Differential susceptibility genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

∗ Corresponding author. E-mail addresses: [email protected] (E. Assary), [email protected] (J.P. Vincent), [email protected] (R. Keers), [email protected] (M. Pluess). 1 Authors contributed equally to manuscript. https://doi.org/10.1016/j.semcdb.2017.10.016 1084-9521/© 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

2

1. Introduction The burden of psychiatric disorders is substantial for both affected individuals and society more generally [1]. Mental health disorders currently account for around a third of all disabilities worldwide, with major depressive disorder, identified as the leading global cause of disability [2]. Research on the predictors of mental health disorders has highlighted the importance of environmental stressors such as childhood maltreatment [3–5]. However, since the application of quantitative behavioural genetics methods (i.e., twin studies) to the field of psychiatry several decades ago, most agree that psychiatric disorders are a product of both genetic and environmental influences [6], with heritability estimates ranging from 37% for major depression [7], to 65%-80% for schizophrenia [8,9] and 60%-85% for bipolar disorder [10–12]. Behavioural genetics research also provided first evidence to suggest that psychiatric disorders reflect the result of the interaction between genetic and environmental factors rather than independent main effects [13]. In other words, the effect of genetic factors (i.e., heritability) on a disorder can differ as a function of environmental factors (and vice versa). For example, Kendler et al. [14] found that individuals at lowest genetic risk of major depression (i.e., monozygotic twins with an unaffected co-twin) had a 0.5% probability of developing depression if they were not exposed to stressful life events but a 6.2% probability if they experienced adversity (i.e., environmental effects). These probabilities were 1.1% and 14.6%, respectively, for monozygotic twins with high genetic risk for depression (i.e., monozygotic twins with an affected co-twin) showing that genetic vulnerability for depression is moderated by environmental risk factors. Advances in technology over the last 20 years facilitated examination of gene-environment interplay at the level of an individuals’ measured DNA rather than statistical estimation based on twin designs. The first GxE study investigating the interaction between a specific gene and adverse environmental influences on the development of psychopathology was reported by Caspi et al.’s [15] seminal study on the interaction between a genetic polymorphism in the monoamine-oxidase A gene (MAOA) and childhood maltreatment in the prediction of antisocial behaviour. Results suggested that carriers of the genotype conferring low levels of MAOA gene expression showed higher levels of adulthood antisocial behaviour, but only if they also experienced maltreatment in childhood (in the absence of maltreatment they were no more likely to develop problems than those with less vulnerable genotypes). In what follows, we will present a concise but comprehensive review of the current state of GxE research in the field of psychiatric genetics. The article is organized into three main parts: first, we will present selected findings from early candidate gene studies of major psychiatric disorders together with an overview of theoretical concepts underlying GxE research; second, we will review more recent studies applying genome-wide methodological approaches; and third, we will conclude by providing several suggestions for future research in the field. 1.1. Candidate gene-X-environment (GxE) studies The serotonin transporter (SLC6A4), monoamine-oxidase A (MAOA), dopamine receptor D4 (DRD4) and D2 (DRD2), catechol-Omethyl transferase (COMT), and brain-derived neurotrophic factors (BDNF) genes are some of the most commonly examined candidate genes in relation to psychiatric disorders (i.e., depression, antisocial behaviour, schizophrenia, and bipolar disorder). Considering the high comorbidity of psychiatric disorders [16] and their shared genetic aetiology, it is not surprising that many of these candidate genes have been examined and associated with multiple disorders. BDNF, for example, has been examined and found to be related to depression and bipolar disorder, as well as schizophrenia [17–19].

In the following section we will present selected GxE studies as illustrative examples involving 5-HTTLPR, MAOA, COMT and DRD4 as four of the most commonly studied candidate genes implicated in depression, antisocial behavior, schizophrenia and attention deficit hyperactivity disorder (ADHD). 1.1.1. 5-HTTLPR and depression The serotonin-transporter-linked polymorphic region (5HTTLPR) is a genetic polymorphism in the promoter region of the serotonin transporter gene (SLC6A4) [20]. The protein product of this gene (5-HTT) is expressed in the central and peripheral nervous systems and plays a key role in transporting the neurotransmitter serotonin from synapses to presynaptic neurons. The polymorphism consists of a long (l-allele) and a short (s-allele) variant, based on the insertion or deletion of 44 base pairs close to the beginning of the gene’s transcription site. The s-allele has been associated with lower and the l-allele with higher levels of serotonin transporter mRNA transcription [21]. Caspi et al. [22] were the first to examine the moderating effects of 5-HTTLPR on depression within a GxE framework, hypothesizing that the 5-HTTLPR s-allele may be implicated in depression by moderating the serotonergic response to stress. In their longitudinal study of 1 037 individuals, Caspi et al. [22] showed that individuals with the 5-HTTLPR s-allele genotype were at higher risk of depression and suicidality compared to those with the l-allele genotype, but only if they had a history of stressful life events or childhood maltreatment. In the absence of these adversities, there was no difference in depression between those with the s-allele and those with the long-allele. More than fifty studies have aimed to replicate these findings, some with more success than others. For example, Eley at al.’s [23] study on 377 adolescent boys and girls showed that a high-risk family environment was associated with higher depressive symptoms, but only in girls with the short allele. In a longitudinal study of 127 adults, Wilhelm et al. [24] reported higher probability of major depression for s-allele carriers in response to adverse life events (but not in the absence of adversity). In a large Spanish prospective cohort study of 737 adults, Cervilla et al. [25] examined the interaction between 5-HTTLPR and the number of threatening life events in the past six months in the prediction of interview-ascertained diagnosis of depression. They found that individuals homozygous for the s-allele showed higher a propensity than other genotypes for severe depression, but only in the presence of stressful life events. Nevertheless, other studies have failed to replicate these results. For example, in a cross-sectional study of 1 206 adults Gillespie et al. [26] found no significant interaction between self-reported stressful life events and 5-HTTLPR genotype in the prediction of depression. Similarly, Surtees et al. [27] in their study of 4 175 adults found no significant interaction between 5-HTTLPR and self-reported adverse childhood or adulthood experiences in the prediction of past-year major depression. Subsequent metaanalysis studies have found both support for [28–30] and against [31–33] the proposed GxE interaction effect between 5-HTTLPR and stressful life events on depression. Some concluded that GxE effects involving 5-HTTLPR are likely false positive findings due to studies being underpowered [31,32]. Karg et al. [28], on the other hand, proposed that previous meta-analyses [31,32] were biased by stringent study selection criteria given that their meta-analysis which included all available studies at the time (N = 54), clearly supported GxE in relation to 5-HTTLPR and childhood maltreatment in the prediction of depression. Furthermore, Uher and McGuffin [30] provided evidence that the lack of robust replication in previous meta-analyses was, at least partly, the result of including studies with low quality measures of stressful life events (i.e., retrospective self-report). In the most recent meta-analysis which included 31 studies, Culverhouse et al. [33] did not find support for a sig-

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

nificant interaction effect. However, this study has been criticized for only including studies with sample sizes of 300 or more [34]. Hence, the field remains divided regarding GxE findings involving 5-HTTLPR. 1.1.2. MAOA and antisocial behaviour The Monoamine Oxidase A (MAOA) gene is located on the Xchromosome [35] and encodes mitochondrial enzymes that are involved in degrading of other amine neurotransmitters such as dopamine, serotonin and norepinephrine [36]. A functional variable number tandem repeat polymorphism (2, 3, 3.5, 4 or 5 repeats) exists in the promoter region of the gene [37], with short (3 and 3.5 repeat) versus long (4 repeat) associated with low versus high MAOA expression, respectively, and therefore higher or lower levels of amine neurotransmitters [37,38]. As noted earlier, the seminal study by Caspi et al. [15] examined the interaction between childhood maltreatment and MAOA. Findings suggested that the MAOA low-activity allele represents a genetic risk factor for antisocial behaviour, but only in combination with childhood maltreatment. The interaction findings involving MAOA have been replicated repeatedly. For example, in a longitudinal study of 235 individuals, Frazzetto et al. [39] reported that the low-MAOA-activity allele in interaction with traumatic life events (e.g., childhood maltreatment) in the first 15 years of life was associated with aggression and antisocial behaviour in adolescence and adulthood (but no association emerged in the absence of adversity). In another longitudinal study with 514 adolescent boys, Foley et al. [40] found that childhood adversity predicted higher probability of conduct disorder but only for those boys who also had the low-MAOA-activity allele. Finally, in a large cohort study of young children, Kim-Cohen et al. [41] found that the association between physical abuse exposure in boys assessed at age 5 and increased levels of mental health problems at age 7 (i.e., antisocial behaviour, attention-deficit hyperactivity, and emotional problems) was stronger for those with the low-MAOA-activity variants compared to those with other genotypes. However, not all replication attempts have proved successful. For example, Haberstick et al. [42] failed to find an interaction effect between MAOA and self-reported childhood and adolescent maltreatment in the prediction of criminal convictions and antisocial behaviour in a large longitudinal study of 774 Caucasian males. Overall, meta-analyses of GxE studies involving MAOA have consistently found support for the moderating effect of MAOA on the association between childhood maltreatment/adversity and antisocial behaviour [41,43,44]. However, this interaction is further moderated by gender, with effects being stronger in boys compared to girls [44]. 1.1.3. COMT and schizophrenia The catechol-O-methyltransferase (COMT) gene is a protein coding gene located on chromosome 22, the enzyme product of which is involved in degradation of catecholamine neurotransmitters such as dopamine, epinephrine and norepinephrine [45]. One of the most studied functional polymorphisms on this gene is a substitution of valine with methionine (Val158Met, rs4680) with those homozygous for the Val genotype having higher levels of dopamine degradation. GxE research in schizophrenia typically sought to investigate the role of COMT in relation to environmental factors that have consistently been linked to schizophrenia, such as obstetric complications at birth, regular cannabis use, and stress [46]. For example, Henquet et al. [47] examined the interactive effects of cannabis and COMT on psychotic symptoms in patients prone to psychosis and healthy controls. In patients with the Val allele hallucinations exacerbated following daily use of cannabis, but not in patients with the Met/Met genotype. Caspi et al. [48] found similar results in

3

their earlier study of 803 individuals, where cannabis use in adolescence was found to increase the risk of schizophreniform disorder in adulthood, but only in those with Val allele. With regards to stress, it is the Met rather than the Val allele that has been implicated as the risk allele. Collip et al. [49] found in their sample of psychosis patients and healthy controls that patients with the Met allele showed increased levels of psychotic experiences in response to stress, but not in patients homozygous for the Val allele (in healthy controls there no significant interaction emerged). Similar GxE results were found for psychotic experiences and delusions in response to daily stress for patients with psychotic disorders [50]. Not all studies have been able to replicate the reported interactions. For example, with regards to cannabis use, Zammit et al. [51] did not find a significant interaction between adolescent cannabis use and COMT in a sample of 493 individuals. In a further study, Zammit et al. [52] examined the interaction between cannabis use at age 14 and COMT genotype in the prediction of psychotic experiences in a large sample (N = 2 630), again unable to detect a significant interaction. A study of psychosis patients by Kantrowitz et al. [53] also failed to detect a significant interaction effect between COMT and cannabis use in the prediction of psychotic experiences. These inconsistent results may reflect important differences in study design and measurement [54]. For example, sample sizes range across studies from 35 to over 2 000 and include patients with diagnosis, patients at risk, healthy controls or a combination of these groups. 1.1.4. DRD4 and attention deficit hyperactivity disorder The Dopamine receptor D4 gene (DRD4) encodes the D4 subtype of the dopamine receptor, which is responsible for neuronal signaling in the mesolimbic system of the brain that regulates emotion and complex behaviour. This gene contains a polymorphic number (2–10 copies) of tandem 48 repeats. The 7 repeat polymorphism has been associated with decreased efficiency in dopamine reception [55]. The DRD4 gene has been associated with ADHD [56,57], a heritable disorder [approx 76%; 58] characterized by inattention and/or hyperactivity and impulsivity, and a precursor of a range of behavioural problems in adolescence and adulthood [59]. The genetic variant most commonly studied and associated with ADHD is the 7-repeat allele [58], which has been shown to confer increased risk of ADHD symptoms when exposed to prenatal smoking [60] or maternal stress [61]. In a recent longitudinal study (N = 308), Zohsel et al. [62] examined the relationship between children’s DRD4 genotype, mother’s self-reported prenatal stress and children’s externalising problems. In the context of high prenatal maternal stress, children with the DRD4 7-repeat genotype, but not other genotypes, were at higher risk to develop externalising behaviour. Despite successful replications, a study examining several different dopamine genes and their interaction with maternal smoking did not find a significant interaction effect for DRD4 genotype [63]. 1.2. Theoretical frameworks in GxE research Traditionally, the guiding conceptual framework for the majority of GxE studies has been the Diathesis-Stress model [64]. According to this model, the detrimental effects of adverse environmental influences only lead to psychopathology when combined with an inherent vulnerability of the individual. Such individual risk factors include those that pertain to biological differences between individuals, with genetic factors commonly regarded as the source of such variation. Specifically, the Diathesis-Stress model suggests that some individuals are more vulnerable to adverse environmental influences, as a function of their specific biological (e.g. genetic) and/or psychological (temperament) make-up, but that in

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11 4

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

Fig. 1. Models of differences in Environmental Sensitivity (figure adapted from Pluess, 2015). Differential Susceptibility (C) represents the combination of DiathesisStress (A) and Vantage Sensitivity (B). In the Diathesis-Stress model (A) Individuals with high genetic sensitivity (Genotype A) are at higher risk of developing negative outcomes when exposed to adverse environmental influences, compared to those with low genetic sensitivity (Genotype B) who prove resilient to environmental stressors. In the Vantage Sensitivity (B) model, high genetic sensitivity (Genotype A) is associated with an increased response to positive environmental influences with less sensitive individuals (Genotype B) benefitting less (i.e, showing Vantage Resistance).

the absence of such adversity, these inherent vulnerabilities are not sufficient to lead to psychopathology. Furthermore, individuals who do not succumb to the negative effects of environmental stressors, either as the function of not carrying these genetic vulnerabilities or due to the presence of other protective factors, are deemed resilient (see Fig. 1 model A). This understanding of GxE adopts a perspective of vulnerability, with only those individuals who carry certain alleles being at increased risk to develop psychopathology when faced with environmental adversity (e.g., negative life events). However, the Diathesis-Stress conceptualization of GxE has recently been challenged by the Differential Susceptibility Theory (DST) [65,66], which suggests that individuals differ in their general susceptibility to both negative and positive environmental influences. According to DST, those individuals who are genetically more susceptible to environmental influences may be indeed more likely to develop psychopathology in response to the adverse effects of stressors. However, higher genetic susceptibility may also make these individuals disproportionately likely to benefit from positive and supportive environmental influences (see Fig. 1 model C). Hence, higher susceptibility may function in a for better and for worse manner [65]. Consequently, resilience as observed in the Diathesis-Stress model may reflect “low susceptibility” to environmental influences and vulnerability “high susceptibility”. Building on DST, Pluess and Belsky [67] recently conceptualized individual differences in the response to positive environmental influences more specifically in the Vantage Sensitivity framework. According to Vantage Sensitivity, some people are more likely to benefit from the positive effects of supportive experiences than others as a function of inherent characteristics, including genetic differences [68] (see Fig. 1 model B). In contrast to Diathesis-Stress, the Differential Susceptibility framework is an evolutionary-inspired developmental model that considers potential disadvantages as well as advantages of individual differences in environmental sensitivity by examining effects on inclusive fitness. Such an evolutionary perspective may be better able to account for the observation that many of the genetic variants studied in candidate GxE psychiatric studies are “common” variants (i.e., they have a high frequency in the general population). If there were gene variants that are associated exclusively with an increased risk for the development of psychopathology

when faced with adversity, one would expect that the frequency of these genes decreased over time (and that gene variants associated with resilience would increase). However, the observation that many of these genes are common and some appear to be even under positive selection [69], suggests that these gene variants may have benefits that counteract the negative effects of heightened vulnerability (e.g., higher Vantage Sensitivity towards positive influences [67,68]. Hence, the high frequency of some risk alleles suggests that these variants may not merely infer disadvantage, as proposed by a Diathesis-Stress perspective, but that they also infer advantages, as suggested by models of Differential Susceptibility and Vantage Sensitivity [67,70]. For example, DST suggests that natural selection would favour both high and low susceptibility types: while low genetic susceptibility may predict resilience in the face of adversity–and therefore reproductive fitness–higher genetic susceptibility may also lead to increased reproductive fitness through enhanced adaptation to the environment, particularly positive ones [e.g., see, [71–73]]. In other words, and consistent with most GxE findings, whether a particular gene variants reflects a risk, resilience, or vantage sensitivity factor depends on the specific qualities of the developmental context [68]. 1.2.1. Genetic sensitivity rather than vulnerability While the studies reviewed in the previous section suggest, at first sight, that these candidate genes represent genetic vulnerability/risk factors for the development of psychiatric disorders in response to environmental adversity–consistent with a DiathesisStress perspective–interpreting the results from a Differential Susceptibility perspective reveals an alternative picture. What appears to have gone unnoticed in some of the studies reviewed earlier is that those individuals carrying the “risk” variant often show less negative outcomes compared to those without this variant in the absence of adversity. For example, while the s-allele in the 5-HTTLPR studies by Caspi et al. [22] and Eley et al. [23] was associated with higher risk for depression in the context of stressful life events and adverse family environment, the same genotype also inferred lower risk of these problematic outcomes in the absence of stressful life events and family problems. A closer look at the Caspi et al. [15] and other MAOA studies [40,41] shows a similar pattern: the putative vulnerability allele (i.e., low-MAOA-activity) infers lower risk for conduct disorder in childhood and lower levels of antisocial behaviour in adulthood in the absence of childhood maltreatment. Similar results are found in studies involving COMT, with Val allele carriers showing lower risk of schizophreniform disorder in the absence of cannabis use compared to Met allele carriers [48]. Such findings are more consistent with Differential Susceptibility rather than Diathesis-Stress [for a comprehensive review, see [70]] and suggest that these common gene variants may reflect general sensitivity to the environment rather than exclusively vulnerability in response to adverse experiences. In order to examine whether a genetic variant reflects general sensitivity to the environment, the measured environment must cover the full range of the spectrum (i.e., from the negative to the positive end) which is not the case for most of the GxE studies in the field of psychiatric genetics. However, recent GxE studies adhering to this criterion have found support for the notion of a general genetic environmental sensitivity rather than genetic vulnerability to adversity. For example, the 5-HTTLPR s-allele has been associated with higher neuroticism in the context of negative life events, but also linked to lower levels of neuroticism and higher life satisfaction in the context of positive life events [74,75]. The theoretical proposition that the reviewed candidate genes, and several others reviewed elsewhere [70,76], reflect general sensitivity to environmental influences is supported by studies showing that these genes moderate the influence of a large range of environmental effects that are relevant to normal development. For example, 5-HTTLPR

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model

ARTICLE IN PRESS

YSCDB-2424; No. of Pages 11

E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

has been found to moderate for better and for worse, the impact of maternal responsiveness on children’s moral development [77], the effect of parenting practices on children’s positive affect [78], perceived racial discrimination on conduct problems and of child maltreatment on antisocial behaviour [79], to name a few [for a meta-analysis, see [80]. With regards to COMT, Baumann et al. [81] found in their sample of 782 adults that COMT Val158Met genotype moderated the effects of childhood adverse experiences on anxiety sensitivity in adulthood, with Met allele inferring greater risk of anxiety for those who were exposed to adverse experiences but also lower scores in the absence of such events. Evidence for genetic sensitivity rather than vulnerability also emerged regarding DRD4. For example, Berry et al. [82] found that the DRD4 7-repeat was associated with higher inattention in the context of insensitive early maternal care, but also with lower levels of inattention in the context of more sensitive maternal care. Similar Differential Susceptibility interaction patterns have been observed regarding the effects of quality of child-care on the development of social competence [83], effects of parenting on pro-social behaviour [84], effects of positive changes in parenting practices on children’s externalizing behaviour [85], and of childhood adversity on emerging adulthood alcohol dependence [86], to name a few [for a review, see [87]]. In conclusion, although many of the reviewed genetic variants increase vulnerability for the development of psychopathology in response to environmental stressors, they also appear to predict increased sensitivity to positive aspects of the environment. Hence, it may be more appropriate to consider these gene variants as markers of Environmental Sensitivity [88], rather risk factors for psychiatric disorders. This view of course does not negate the possibility of the existence of gene variants that exclusively increase vulnerability for disorders without inferring advantages in positive environmental contexts, or indeed variants that only confer protective effects in positive environments. However, it is import to consider and evaluate alternative GxE models rather than implicitly assuming a Diathesis-Stress perspective in studies aimed at investigating gene-environment interactions [89]. 1.3. Limitations of candidate GxE studies Notwithstanding the important contribution of candidate GxE studies to the field of psychiatry, it is important to acknowledge several limitations. First, the candidate gene approach requires a strong biological hypothesis in order to select appropriate candidates, but knowledge regarding the specific biological mechanisms underlying psychiatric disorders remains rather limited. Therefore, the risk of selecting inappropriate candidates is high and publication bias for significant novel results over null or negative replication efforts [90,91] may imply that selected candidate genes are more robust than they actually are. Second, recent discovery in psychiatric genetics suggests that most psychiatric disorders are influenced by many thousands of gene variants of very small effects rather than by a few gene variants of large effect [33]. In other words, the genetic architecture of common behavioural traits is highly complex and polygenic [92]. Third, of particular concern is the difficulty to replicate candidate GxE findings. The widely observed replication problems may be a function of small underpowered sample sizes that are unable to provide the statistical power required to detect interactions in the first place and, consequently, increase the probability of false positive results [93–95]. Finally, the majority of GxE studies have been conducted from a Diathesis-Stress perspective, which requires re-evaluation in light of recent research suggesting that many of the common genetic variants in these studies seem to reflect generally heightened susceptibility to both negative and positive environmental influences, as suggested by Differential Susceptibility framework

5

[70,89], rather than exclusively vulnerability to adversity. The notion that these genetic variants may be associated with heightened sensitivity to both environmental risk and protective factors may also contribute to the observed heterogeneous results [96]. In light of these limitations, it is not surprising that GxE research regarding psychiatric outcomes is currently transitioning towards polygenic and genome-wide approaches in larger better-powered samples. 2. Genome-wide approaches Genome-wide association studies (GWAS) have made it possible to assess the whole genome for associations with psychiatric disorders by assaying upwards of 500 000 variants simultaneously. This increased coverage means that, in contrast to candidate gene studies, GWAS are able to take a hypothesis-free approach which does not require any a priori assumptions regarding the role of specific genes in a disorder [97]. This hypothesis-free approach brings with it a substantial multiple-testing problem which means that–in order to protect against false positives–the threshold for genomewide significance is very high with p < 5 × 10–8 [98]. Initial GWAS were limited by inadequate sample sizes to detect variants of small effect at genome-wide significance. However, the formation of consortia and the pooling of data have made mega-analysis and meta-analysis possible, resulting in substantial progress in identifying replicable disorder-associated variants. For example, the path of discovery for schizophrenia (SCZ) from GWAS has steadily increased as a result of larger sample sizes, from a single locus [99,100], to 7 loci [101]. Further sample size increases have captured 22 loci [102], 62 [103], and most recently, 108 [104]. Consequently, the variance explained by common genetic variants increased from 3% [99], to 7% [104]. Polygenic approaches, which consider the combined effects of variants across the entire genome, have been shown to explain more of the variance. Nevertheless, these estimates still fall substantially short of the estimates reported by twin studies. For example, while twin studies report heritabilities of 80% for schizophrenia and 37% for major depression, SNP-based heritabilities are approximately 20% for these disorders. One explanation for this substantial missing heritability is that while gene-environment interactions involving the shared environment contribute to heritability estimates in twin models, they are not captured in SNP-based heritability estimates [105] 2.1. Genome-wide environment interaction studies Appreciating the limitations of the candidate gene approach, GxE research in psychiatric disorders has begun to employ genomewide environment interaction studies (GWEIS) to search the entire genome for variants that moderate the effects of the environment on psychiatric disorders. The GWEIS approach considers GxE on a whole genome basis [106], and typically involves testing interactions between the environment and each SNP independently. To date, only a handful of GWEIS have been conducted, all of which examine depression [107–109]. In one of the largest GWEIS conducted to date including over 10 000 women of African American and Hispanic/Latina descent, Dunn et al. [108] investigated interactions between SNPs and social support and stressful life events (SLE) on symptoms of depression. Whilst a genome-wide significant interaction was found between the SNP rs4652467, located near the CEP350 gene and SLE within the African American sample, the result did not replicate in a smaller independent sample. This draws further attention to the need for greater sample sizes in GWEIS, and further highlights the statistical complexities that come with combining GWAS with environmental determinants [106,110]. A more recent GWEIS by Otowa

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11 6

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

et al. [107], investigated the interaction between 534 848 SNPs and SLEs on depression in a sample of 320 Japanese white collar workers. The authors reported a genome-wide significant finding between SLEs and the SNP rs10510057. However, this finding did not surpass the more stringent significance threshold to take into account the number of hypotheses tested (i.e., main effects of G and GxE). However, the initial finding did replicate in an independent sample (N = 439), assessing work-related stress. However, a strict replication of this study is required as only a single stressor in relation to work-related stress was assessed in the independent sample–“fear of being unemployed”. In addition to examining GxE using standard approaches, GWEIS have also explored the usefulness of a joint test. This approach simultaneously tests G and GxE associations on outcomes and has been shown to increase power to detect genetic effects that are only expressed in individuals exposed to a measured environmental factor [111]. While the joint test has been used to successfully identify GxE in several complex disorders, few studies have applied this approach to psychiatric disorders [110]. Ikeda et al. [109] recently used the joint test to identify variants that moderate the effects of SLEs on depression in a sample of 1 112 hospital staff [109]. One genome-wide significant finding was reported for rs10485715, located downstream of the gene encoding bone morphogenetic protein 2 (BMP2). Further analyses suggested that SLEs were significantly more depressogenic in individuals with a TT genotype at this locus. Nevertheless, this association awaits replication in an independent sample. 2.2. Polygenic score-X-environment interaction studies It is well established that psychiatric disorders are polygenic, and are the result of the effects of multiple genetic variants across the genome. GWEIS, which take a SNP-by-SNP approach to GxE, may therefore be considerably underpowered to detect these effects. An alternative approach is to estimate genetic risk as a polygenic score (PGS) for a given individual and test the interaction between this score and an environmental variable. A PGS is simply the number of trait-associated alleles possessed by each individual in target sample, weighted by their effect sizes from a discovery sample [112]. PGSs can be limited to genome-wide significant variants, or series of more liberal thresholds (e.g., p < 0.0001, 0.001, 0.01, 0.1). Using this approach, two studies have examined the interaction between the PGS for major depressive disorder (MDD) and childhood trauma, in the prediction of MDD [113,114]. Although both studies did find a significant interaction between the PGS and childhood maltreatment, the nature of this interaction differed markedly for each of the studies. For example, Peyrot et al. [113] demonstrated that the effects of childhood trauma on MDD were greater for those with a higher PGS for MDD. However, in a later study, Mullins et al. [114] found that the effects of childhood trauma on MDD were greater for those with a lower PGS. The cause of these discrepant findings is unclear. However, there were several methodological differences between the two samples in the measurement of the environment. For example, in Peyrot et al. [113] data on childhood trauma was collected by trained clinicians using a structured interview, while in Mullins et al. [114] a self-report questionnaire was used. Further to childhood trauma, three studies have examined interactions between PGS and adult SLEs on and depression symptoms [115,116] and MDD [114]. The first of these studies reported that both SLEs and the depression PGS were positively associated with depression symptoms in the large Health and Retirement Study (N = 8 761). However, there was no evidence of a significant interaction between these factors [115]. Similar findings were reported in later case/control study of MDD, using the same polygenic score

[114]. Nevertheless, a recent reanalysis of the Health and Retirement Study using a more sophisticated longitudinal approach, revealed more promising findings [116]. This study investigated whether several polygenic scores (including those for MDD, depressive symptoms and subjective well-being) moderated the effects of a specific life event (i.e., death of a spouse) on changes in depressive symptoms. The authors reported that those who lost a spouse (N = 1 647) experienced a significant increase in depressive symptoms following their spouse’s death. However, these effects were significantly greater in those with a higher PGS for MDD or a higher PGS for depressive symptoms. Conversely, the PGS for subjective well-being was found to buffer against the depressogenic effects of spousal death. Those with higher scores experienced a smaller increase in depressive symptoms and experienced a more rapid return to baseline levels. These remarkably consistent findings support previous suggestions that a longitudinal approach, which more carefully considers the temporal relationship between environmental exposures and outcomes, and examines within-person changes in symptoms rather than cross sectional assessments, may be a more powerful approach to detect GxE [117]. The study also highlights the use of objective measures of the environment, which, unlike self-reported checklists of SLEs, are less likely to be a consequence of depression symptoms. While the majority of polygenic GxE studies have focused on depression, a handful of studies taken the same approach to investigate schizophrenia. A recent small pilot study (including 80 cases and 110 controls) explored the interaction between the schizophrenia PGS and childhood adversity, in the prediction of psychosis in the Genes and Psychosis (GAP) study [118]. Both a higher PGS for schizophrenia and the presence of childhood adversity predicted psychosis status additively. However, there was no interaction effect, meaning that the effects of childhood adversity do not differ as a function of the schizophrenia PGS. In a further study, researchers investigated whether the PGS for schizophrenia moderated the effects of cannabis use during adolescence on cortical thickness [119]. Findings indicated that cannabis use was associated with reduced cortical thickness and these effects were significantly greater in those with high PGS for schizophrenia. Nevertheless the moderating effects of the PGS appeared to be specific to females. In conclusion, although the GWEIS approach overcomes some of the previous criticisms of candidate gene studies of GxE, small sample sizes and limited measures of the environment have resulted in few replicated genome-wide significant GxE findings. The polygenic approach to GxE is more in keeping with current understanding of the genetic architecture of psychiatric disorders and may increase power to detect interactions by considering the aggregate effects of multiple variants. While still in its infancy, polygenic GxE has begun to provide some promising findings, particularly when more close attention is paid to measurement of the environment. Nevertheless, this approach relies on the untested assumption that genes implicated in case-control studies of psychiatric disorders are the same genes implicated in GxE.

3. Future directions In line with studies of genetic main effects, GxE research in psychiatric disorders is beginning the shift from a candidate gene to a genome-wide approach. The large samples required for genomewide research exacerbates several previously discussed challenges to GxE, such as the collection of high quality and more objective measures of the environment, and the need to consider developmental and life-course approaches. Genome-wide GxE also brings with it several further challenges including the multiple testing-

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

burden of GWAS and how to consider the aggregate effects of genotypes in polygenic approaches.

3.1. Obtaining better quality environmental measures Paradoxically, it is likely the lack of reliable assessments of environmental exposure and not the genome-wide genotyping that continues to hold back this type of research as evidenced from both the candidate gene and genome-wide GxE literature. For example, GxE findings for 5-HTTLPR and childhood maltreatment has been shown to be dependent on high quality and accurate assessments of childhood maltreatment [30]. Similarly, more recent polygenic approaches to GxE have been shown to be more successful in designs that consider prospectively assessed environmental measures on within-person changes in symptoms rather than selfreport cross-sectional measures of SLEs [117]. The experience sampling method (ESM) may offer a novel way of obtaining more accurate self-report data on an individual’s experience of their environments, and the impact they have on within-person changes in psychopathology. ESM has been shown to be an accurate measure of environments encountered during daily life, minimizing recall bias and mood congruency effects [120]. This approach has also been used to examine individual differences in sensitivity to both positive and negative environments and how this sensitivity relates to genetic risk for MDD [120] and response to antidepressants [121]. Furthermore, the rich longitudinal data collected through ESM can be used to better account for measurement error and increases the effective sample size, resulting in improved power. Despite the advantages of the ESM approach and its applicability to large scale data collection through the use of smartphone apps ESM has only been used in twin [122] and candidate gene studies [123] and is yet to be applied to genome-wide approach. The collection of high quality measures of the environment and phenotypes in samples large enough for genome-wide analyses is particularly challenging given the high cost and resources required. This is particularly poignant as past studies that involved large sample collections have often fallen short in this regard. As the logistics of obtaining high quality environmental data becomes increasingly difficult with larger sample sizes, this can lead to negative associations between study size, quality and non-replications of GxE in large samples [105]. It has been suggested that selective sampling of those at the extreme ends of distribution of a given PGS (high and low) from biobanked samples could represent a solution for future GxE research [124]. This extreme sampling approach would provide adequate power in smaller samples, allowing for more in-depth measurement of the environment and outcomes. This selective approach could be extended to genotype the extremes of environmental exposure to the same effect. The selection of environmental variables for assessment represents another challenge. Severe mental illnesses are potentially linked to many environmental exposures. For example, environmental exposures including season of birth [125] and vitamin D level [126] have been consistently linked with schizophrenia, as have childhood maltreatment [127], being socially disadvantaged in childhood [128,129], urbanicity [130], minority status [131], and cannabis use [132]. Overlap between environmental exposures and different disorders are also evident. For example, both childhood maltreatment [5,133] and being socially disadvantaged in childhood [128,134] have emerged in research as environmental risk factors for MDD. SLEs in adulthood are also a consistent risk factor for MDD [135] and have been linked with bipolar disorder [136]. To address this challenge future research should incorporate multiple environmental factors rather than specific ones such as minority status or childhood maltreatment alone.

7

3.2. Focus on transdiagnostic phenotypes Psychological disorders are typically defined by diagnostic criteria used to assess whether people have a specific disorder or not. However, empirical research suggests that there is substantial aetiological overlap between the different disorders given that the same environmental and genetic factors have been found to contribute to various mental disorders–each defined by a different set of diagnostic criteria [137,138]. For example, childhood abuse has been implicated in schizophrenia, bipolar disorder, major depression, as well as anxiety [139,140]. In addition, these different disorders also appear to share a substantial proportion of the genetic risk [141]. Not surprisingly, studies suggest that effects of GxE are also transdiagnostic. For example, the 5-HTTLPR S-allele has been proposed to confer general vulnerability–when faced with daily life stress–for the development of a transdiagnostic emotion dysregulation phenotype rather than a specific psychiatric disorder [142]. Taken together, genetic sensitivity to the environment may reflect an aetiological pathway that underlies a wide range of mental illnesses. While this hypothesis remains to be tested more rigorously, it implies that transdiagnostic dimensions may be more suited as outcomes in GxE studies than categorical and narrowly defined clinical diagnoses [143]. Furthermore, targeting intermediate phenotypes that are shared across disorders such as emotion regulation or cognitive biases may also contribute to a better understanding of the general causal mechanisms of psychiatric disorders [144]. 3.3. Applying a developmental perspective GxE effects may differ across the life span, with effects being stronger in early development. In terms of the development of psychiatric disorders, several studies and meta-analyses of candidate genes including 5-HTTLPR, BDNF, DRD4, and FKBP5 have made the case for the importance of early environments in terms of GxE [80,87,145,146]. This has been further exemplified in a recent multi-wave longitudinal study assessing stress-sensitisation [147]. This study aimed to replicate and extend previous findings pertaining to gene-x-environment-x-environment interactions (GxExE) between candidate genes, childhood environments, and adult environments on psychopathology [148]. Using PGS of Environmental Sensitivity [88] from the aggregated effects of nine candidate gene variants, and an objective measure of childhood and adult material environments, the study reported significant evidence for ExE and GxExE in predicting adult psychological distress. Specifically, genetically sensitive children who experienced a poor early environment showed increased sensitivity to adversity during adulthood. Conversely, genetically sensitive children who experienced a good early environment were more resilient to adversity as adults. This study suggests that a life-course approach may be needed to fully understand the role of GxE in the development of psychiatric disorders and GxE studies should consider the role of genes but also early environments when examining response to adversity. 3.4. New analytical approaches and study designs Theoretical frameworks for GxE such as Diathesis-Stress, Differential Susceptibility and Vantage Sensitivity describe different mechanisms through which genes may moderate the effects of the environment on outcomes. However, these explanations are not necessarily mutually exclusive. It is highly likely that, in addition to the genetic main effects, genetic risk for psychiatric disorders includes genotypes that increase sensitivity to adversity (i.e., Diathesis-Stress), decrease sensitivity to protective factors such as social support (i.e., Vantage Resistance) and increase plastic-

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11 8

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

ity more generally (i.e., Differential Susceptibility). The traditional approach to GxE research is to first identify a statistically significant interaction between a genetic variant and environmental measure and proceed to characterize the interaction based on further criteria, such as regions of significance. While new study designs and approaches to genome-wide GxE may the increase power to detect interactions, the assumptions upon which these new approaches rely, means that they may be biased toward the discovery of genetic variants operating in a specific manner. This means that future GxE research should first consider the type of GxE under investigation, before employing an appropriate strategy to detect genes operating within that framework. 3.4.1. Vulnerability genes In the Diathesis-Stress model, the effects of adverse environments are accentuated for specific genotypes. This means that significant genetic effects of these genotypes may only be observed in those exposed to adversity. Given that known environmental risk factors for MDD are relatively common (e.g., stressful life events or childhood maltreatment) it is likely that genotypes operating in Diathesis-Stress models will display marginal genetic main effects. Studies searching specifically for vulnerability genes may therefore benefit from approaches that aim to reduce multiple-testing burden by only including variants showing with a marginally significant main effect in GxE analyses [149]. 3.4.2. Vantage sensitivity genes In the Vantage Sensitivity model of GxE, the effects of positive or supportive environments are accentuated in the context of specific genotypes [68]. Significant effects of these genotypes may therefore only be observed in those in supportive or positive environments. Assuming the positive environment is relatively common, it is likely that genotypes operating in a manner consistent with Vantage Sensitivity will display marginal genetic main effects. Much like in studies searching for vulnerability genes, Vantage Sensitivity genes may therefore be more efficiently detected in GWAS data by prioritizing variants showing a marginal association with the outcome of interest [149]. However, assessing genetic predictors of response to interventions represents a potentially more powerful approach to detecting Vantage Sensitivity genes [150–152]. This approach results in lower measurement error, requires fewer participants, and has greater statistical power than correlational studies [93]. Moreover, as interventions are randomly allocated, this approach also circumvents the confounding effects of gene-environment correlation (rGE). The power of this gene-x-experimental environment (GxeE) approach was illustrated in a recent meta-analysis which included the results of 22 GxeE studies including over 3 000 participants [152]. The study suggested that while effect sizes for the interventions were moderate and significant for in individuals with Differential Susceptibility alleles, for those with the alternative alleles, the interventions were no more effective than the control condition. Using a similar GxeE approach, Fox et al. [153] examined the 5-HTTLPR as a possible predictor of response to a computer based intervention similar to attention bias modification (ABM) that aimed to train attentional biases either towards positive or negative images. Findings provided further evidence for the s-allele of 5-HTTLPR as a marker of Differential Susceptibility as it appeared to accentuate the effects of both the positive and negative interventions. While the above candidate gene studies of GxeE provide an early proof of concept of this approach, genome-wide GxeE will be required to identify novel Vantage Sensitivity genes or those implicated in Differential Susceptibility. Although the sample sizes required for GWAS means that laboratory-based studies are not feasible, simple interventions such as attentional bias modifi-

cation can be administered to very large samples through the use of a home computer or smart phone app. Patient records of response to manualised psychological treatments, such as Cognitive Behavioural Therapy (CBT) represent yet another opportunity to conduct genome-wide GxeE studies. 3.4.3. Differential susceptibility genes In the Differential Susceptibility model of GxE, genotypes are proposed to increase responsivity to both negative and positive environmental factors. This means that the same genotypes can result in either negative or positive outcomes, dependent on the environment. In contrast to Diathesis-Stress or Vantage Sensitivity genes, Differential Susceptibility genes are therefore unlikely to show a marginal main effect on a given outcome. This means that approaches that prioritise SNPs based on their marginal main effects are unlikely to improve detection of novel Differential Susceptibility genes. It also means that Differential Susceptibility genes are unlikely to be represented in polygenic GxE analyses, which derive scores from case/control studies of psychiatric disorders. We recently reported on a novel approach to capture genes involved in Differential Susceptibility and Environmental Sensitivity more widely using monozygotic (MZ) twins [154]. As MZ twins are genetically identical, and share the same environment, discordance within MZ twin pairs on a measured outcome is considered to be the result of non-shared environmental effects. This means that twin pairs with variants associated with increased sensitivity to the environment will therefore have a greater intra-pair variability on a given outcome, due to their increased responsivity to unmeasured non-shared environmental influences [155]. Applying this approach to whole genome data we created a PGS of environmental sensitivity (PGSES ). This score significantly moderated the effects of parenting on emotional problems in a further unrelated sample of 1 400 children in a manner consistent with Differential Susceptibility. For individuals with a low PGSES , parenting had little effect on emotional problems. In contrast, in those with a high PGSES , negative parenting was a significant risk factor for emotional problems, while positive parenting was protective. Interestingly, the PGSES also significantly predicted differential response to psychological treatments in children with anxiety disorders. Specifically, individuals with a high genetic environmental sensitivity responded best to individual CBT moderately to group CBT and poorly to brief parent-led CBT. In contrast, those with a low genetic environmental sensitivity responded equally well to each treatment type. These effects are potentially clinically meaningful, with remission rates at the upper tertile of the PRS of 70.9%, 55.1% and 40.6% for Individual CBT, group CBT and brief parent-led CBT respectively. Nevertheless, replication of these findings in betterpowered samples is necessary to evaluate the usefulness of this approach in predicting sensitivity to the environment. 4. Conclusion A large number of candidate gene studies provide evidence for gene-environment interaction in the development of psychiatric disorders. However, in addition to difficulties with replication, single candidate gene studies generally fail to account for the complex polygenic architecture of most psychiatric disorders. Furthermore, the Diathesis-Stress perspective which guides most GxE research in psychiatry, does not consider that individuals who are genetically more vulnerable to the negative effects of adversity may also be more sensitive to the positive effects of supportive experiences as proposed by the Differential Susceptibility framework. In line with studies of the main effects of genes, GxE research has begun to shift towards new genome-wide and polygenic approaches, but these studies are currently limited by small samples and have yet

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

to yield robust results. In addition to adequate statistical power (i.e., larger samples), both candidate and genome wide GxE studies also require better and more accurate measurement of the environment. Hence, future studies should feature objective and reliable measures, including experience sampling methods, rather than relying on retrospective and subjective reports. Furthermore, it is important that future work considers developmental aspects by adopting a longitudinal life-course approach (i.e., taking effects of early developmental periods into account). Finally, although several new GxE study designs and analytical approaches are now available, future studies should carefully consider the specific type of interaction pattern they aim to investigate (i.e., Diathesis-Stress, Vantage Sensitivity, or Differential Susceptibility) and employ the appropriate strategy within the selected framework. GxE research that is guided by solid evolutionary theory on genetic sensitivity to the environment, featuring a polygenic and genome-wide approach with reliable and objective measures of contextual factors and set within a life-course perspective, is required in order to significantly advance knowledge on the interactions between genes and environment in the development of psychiatric disorders. A better understanding how genes and environment work together in the aetiology of psychiatric disorders will not only help us understand the origins of psychopathology but may also identify novel pathways for intervention and treatment.

References [1] P.Y. Collins, et al., Grand challenges in global mental health, Nature 475 (7354) (2011) 27–30. [2] Organization, W.H, The Global Burden of Disease: 2004 Update, in The Global Burden of Disease: 2004 Update, World Health Organization, 2008. [3] M.H. Teicher, J.A. Samson, Childhood maltreatment and psychopathology: a case for ecophenotypic variants as clinically and neurobiologically distinct subtypes, Am. J. Psychiatry 170 (10) (2013) 1114–1153. [4] C.L. Battle, et al., Childhood maltreatment associated with adult personality disorders: findings from the Collaborative Longitudinal Personality Disorders Study, J. Pers. Disord. 18 (2) (2004) 193–211. [5] V. Nanni, R. Uher, A. Danese, Childhood maltreatment predicts unfavorable course of illness and treatment outcome in depression: a meta-analysis, Am. J. Psychiatry 169 (2) (2012) 141–151. [6] R. Plomin, M.J. Owen, P. McGuffin, The genetic basis of complex human behaviors, Science 264 (5166) (1994) 1733–1739. [7] P.F. Sullivan, M.C. Neale, K.S. Kendler, Genetic epidemiology of major depression: review and meta-analysis, Am. J. Psychiatry 157 (10) (2000) 1552–1562. [8] P. Lichtenstein, et al., Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study, Lancet 373 (9659) (2009) 234–239. [9] P.F. Sullivan, K.S. Kendler, M.C. Neale, Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies, Arch. Gen. Psychiatry 60 (12) (2003) 1187–1192. [10] P. McGuffin, et al., The heritability of bipolar affective disorder and the genetic relationship to unipolar depression, Arch. Gen. Psychiatry 60 (5) (2003) 497–502. [11] A. Bertelsen, B. Harvald, M. Hauge, A Danish twin study of manic-depressive disorders, Br. J. Psychiatry 130 (4) (1977) 330. [12] J.W. Smoller, C.T. Finn, Family, twin, and adoption studies of bipolar disorder, Am. J. Med. Genet. C Semin. Med. Genet. 123c (1) (2003) 48–58. [13] K.S. Kendler, L.J. Eaves, Models for the joint effect of genotype and environment on liability to psychiatric illness, Am. J. Psychiatry 143 (3) (1986) 279–289, http://dx.doi.org/10.1176/ajp.143.3.279. [14] K.S. Kendler, et al., Stressful life events: genetic liability, and onset of an episode of major depression in women, Am. J. Psychiatry 152 (6) (1995) 833–842. [15] A. Caspi, et al., Role of genotype in the cycle of violence in maltreated children, Science 297 (5582) (2002) 851–854. [16] R.C. Kessler, et al., Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the national comorbidity survey, Arch. Gen. Psychiatry 51 (1) (1994) 8–19. [17] S. Alemany, et al., Childhood abuse: the BDNF-Val66Met polymorphism and adult psychotic-like experiences, Br. J. Psychiatry 199 (1) (2011) 38–42. [18] G.M. Hosang, et al., Interaction between stress and the BDNF Val66Met polymorphism in depression: a systematic review and meta-analysis, BMC Med. (2014) 7. [19] G.M. Hosang, et al., Stressful life events and the brain-derived neurotrophic factor gene in bipolar disorder, J. Affect. Disord. 125 (1–3) (2010) 345–349.

9

[20] A. Heils, et al., Functional promoter and polyadenylation site mapping of the human serotonin (5-HT) transporter gene, J. Neural Trans. Gen. Sec. JNT 102 (3) (1995) 247–254. [21] K.-P. Lesch, et al., Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region, Science 274 (5292) (1996) 1527. [22] A. Caspi, et al., Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene, Science 301 (5631) (2003) 386–389. [23] T.C. Eley, et al., Gene-environment interaction analysis of serotonin system markers with adolescent depression, Mol. Psychiatry 9 (10) (2004) 908–915. [24] K. Wilhelm, et al., Life events, first depression onset and the serotonin transporter gene, Br. J. Psychiatry 188 (2006) 210–215. [25] J.A. Cervilla, et al., The risk for depression conferred by stressful life events is modified by variation at the serotonin transporter 5HTTLPR genotype: evidence from the Spanish PREDICT-Gene cohort, Mol. Psychiatry 12 (8) (2007) 748–755. [26] N.A. Gillespie, et al., The relationship between stressful life events: the serotonin transporter (5-HTTLPR) genotype and major depression, Psychol. Med. 35 (01) (2005) 101–111. [27] P.G. Surtees, et al., Social adversity: the serotonin transporter (5-HTTLPR) polymorphism and major depressive disorder, Biol. Psychiatry 59 (3) (2006) 224–229. [28] K. Karg, et al., The serotonin transporter promoter variant (5-HTTLPR), stress, and depression meta-analysis revisited: evidence of genetic moderation, Arch. Gen. Psychiatry 68 (5) (2011) 444–454. [29] C.F. Sharpley, et al., An update on the interaction between the serotonin transporter promoter variant (5-HTTLPR), stress and depression, plus an exploration of non-confirming findings, Behav. Brain Res. 273 (2014) 89–105. [30] R. Uher, P. McGuffin, The moderation by the serotonin transporter gene of environmental adversity in the etiology of depression: 2009 update, Mol. Psychiatry 15 (1) (2010) 18–22. [31] N. Risch, et al., Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression: a meta-analysis, JAMA 301 (23) (2009) 2462–2471. [32] M.R. Munafò, et al., Gene × Environment interactions at the serotonin transporter locus, Biol. Psychiatry 65 (3) (2009) 211–219. [33] R.C. Culverhouse, et al., Collaborative meta-analysis finds no evidence of a strong interaction between stress and 5-HTTLPR genotype contributing to the development of depression, Mol. Psychiatry (2017). [34] T.E. Moffitt, A. Caspi, Bias in a protocol for a meta-analysis of 5-HTTLPR, stress, and depression, BMC Psychiatry 14 (1) (2014) 179. [35] E.R. Levy, et al., Localization of human monoamine oxidase-A gene to Xp11. 23-11.4 by in situ hybridization: implications for norrie disease, Genomics 5 (2) (1989) 368–370. [36] J. Shih, K. Chen, M. Ridd, Monoamine oxidase: from genes to behavior, Annu. Rev. Neurosci. 22 (1) (1999) 197–217. [37] S.Z. Sabol, S. Hu, D. Hamer, A functional polymorphism in the monoamine oxidase A gene promoter, Hum. Genet. 103 (3) (1998) 273–279. [38] J. Deckert, et al., Excess of high activity monoamine oxidase A gene promoter alleles in female patients with panic disorder, Hum. Mol. Genet. 8 (4) (1999) 621–624. [39] G.et al. Frazzetto, Early trauma and increased risk for physical aggression during adulthood: the moderating role of MAOA genotype, PLoS One 2 (5) (2007) e486. [40] D.L. Foley, et al., Childhood adversity: monoamine oxidase a genotype, and risk for conduct disorder, Arch. Gen. Psychiatry 61 (7) (2004) 738–744. [41] J. Kim-Cohen, et al., MAOA, maltreatment, and gene-environment interaction predicting children’s mental health: new evidence and a meta-analysis, Mol. Psychiatry 11 (10) (2006) 903–913. [42] B.C. Haberstick, et al., Monoamine oxidase A (MAOA) and antisocial behaviors in the presence of childhood and adolescent maltreatment, Am. J. Med. Genet. B Neuropsychiatr. Genet. 135B (1) (2005) 59–64. [43] A. Taylor, J. Kim-Cohen, Meta-analysis of gene-environment interactions in developmental psychopathology, Dev. Psychopathol. 19 (4) (2007) 1029–1037. [44] A.L. Byrd, S.B. Manuck, MAOA, childhood maltreatment, and antisocial behavior: meta-analysis of a gene-environment interaction, Biol. Psychiatry 75 (1) (2014) 9–17. [45] M.H. Grossman, B.S. Emanuel, M.L. Budarf, Chromosomal mapping of the human catechol-O-methyltransferase gene to 22q11.1 → q11.2, Genomics 12 (4) (1992) 822–825. [46] J. van Os, G. Kenis, B.P.F. Rutten, The environment and schizophrenia, Nature 468 (7321) (2010) 203–212. [47] C. Henquet, et al., An experimental study of catechol-O-Methyltransferase Val158Met moderation of [Delta]-9-Tetrahydrocannabinol-Induced effects on psychosis and cognition, Neuropsychopharmacology 31 (12) (2006) 2748–2757. [48] A. Caspi, et al., Moderation of the effect of adolescent-onset cannabis use on adult psychosis by a functional polymorphism in the catechol-O-methyltransferase gene: longitudinal evidence of a gene X environment interaction, Biol. Psychiatry 57 (10) (2005) 1117–1127. [49] D. Collip, et al., COMT Val158Met-stress interaction in psychosis: role of background psychosis risk, CNS Neurosci. Ther. 17 (6) (2011) 612–619.

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11 10

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

[50] R. van Winkel, et al., Evidence that the COMT(Val158Met) polymorphism moderates sensitivity to stress in psychosis: an experience-sampling study, Am. J. Med. Genet. B Neuropsychiatr. Genet. 147B (1) (2008) 10–17. [51] S. Zammit, et al., Genotype effects of CHRNA7, CNR1 and COMT in schizophrenia: interactions with tobacco and cannabis use, Br. J. Psychiatry 191 (5) (2007) 402–407. [52] S. Zammit, et al., Cannabis: COMT and psychotic experiences, Br. J. Psychiatry 199 (5) (2011) 380–385. [53] J.T. Kantrowitz, et al., Adolescent cannabis use: psychosis and catechol-O-methyltransferase genotype in African Americans and Caucasians, Psychiatr. Q. 80 (4) (2009) 213–218. [54] G. Modinos, et al., Molecular genetic gene–environment studies using candidate genes in schizophrenia: a systematic review, Schizophr. Res. 150 (2) (2013) 356–365. [55] U.M. D’Souza, I.W. Craig, Functional polymorphisms in dopamine and serotonin pathway genes, Hum. Mutat. 27 (1) (2006) 1–13. [56] M. Burmeister, M.G. McInnis, S. Zollner, Psychiatric genetics: progress amid controversy, Nat. Rev. Genet. 9 (7) (2008) 527–540. [57] I.R. Gizer, C. Ficks, I.D. Waldman, Candidate gene studies of ADHD: a meta-analytic review, Hum. Genet. 126 (1) (2009) 51–90. [58] S.V. Faraone, et al., Molecular genetics of attention-deficit/hyperactivity disorder, Biol. Psychiatry 57 (11) (2005) 1313–1323. [59] S. Mannuzza, et al., Adult outcome of hyperactive boys. Educational achievement: occupational rank, and psychiatric status, Arch. Gen. Psychiatry 50 (7) (1993) 565–576. [60] R.J. Neuman, et al., Prenatal smoking exposure and dopaminergic genotypes interact to cause a severe ADHD subtype, Biol. Psychiatry 61 (12) (2007) 1320–1328. [61] N. Grizenko, et al., Maternal stress during pregnancy, ADHD symptomatology in children and genotype: gene-Environment interaction, J. Can. Acad. Child Adolesc. Psychiatry 21 (1) (2012) 9–15. [62] K. Zohsel, et al., Mothers’ prenatal stress and their children’s antisocial outcomes–a moderating role for the dopamine D4 receptor (DRD4) gene, J. Child Psychol. Psychiatry 55 (1) (2014) 69–76. [63] K. Langley, et al., Testing for gene × environment interaction effects in attention deficit hyperactivity disorder and associated antisocial behavior, Am. J. Med. Genet. B: Neuropsychiatr. Genet. 147B (1) (2008) 49–53. [64] S.M. Monroe, A.D. Simons, Diathesis-stress theories in the context of life stress research: implications for the depressive disorders, Psychol. Bull. 110 (3) (1991) 406–425. [65] J. Belsky, R.M.P.B. Fearon Bell, Parenting, attention and externalizing problems: testing mediation longitudinally, repeatedly and reciprocally, J. Child Psychol. Psychiatry 48 (12) (2007) 1233–1242. [66] B.J. Ellis, et al., Differential susceptibility to the environment: an evolutionary–neurodevelopmental theory, Dev. Psychopathol. 23 (1) (2011) 7–28. [67] M. Pluess, J. Belsky, Vantage sensitivity: individual differences in response to positive experiences, Psychol. Bull. 139 (4) (2013) 901–916. [68] M. Pluess, Vantage sensitivity: environmental sensitivity to positive experiences as a function of genetic differences, J. Pers. 85 (1) (2017) 38–50. [69] Y. Ding, et al., Evidence of positive selection acting at the human dopamine DF gene locus, PNAS 99 (2002) 309–314. [70] J. Belsky, M. Pluess, Beyond diathesis stress: differential susceptibility to environmental influences, Psychol. Bull. 135 (6) (2009) 885–908. [71] T.J. DeWitt, A. Sih, D.S. Wilson, Costs and limits of phenotypic plasticity, Trends Ecol. Evol. 13 (2) (1998) 77–81. [72] A. Sih, Effects of early stress on behavioral syndromes: an integrated adaptive perspective, Neurosci. Biobehav. Rev. 35 (7) (2011) 1452–1465. [73] W.E. Frankenhuis, K. Panchanathan, Individual differences in developmental plasticity may result from stochastic sampling, Perspect. Psychol. Sci. 6 (4) (2011) 336–347. [74] Y. Kuepper, C. Wielpuetz, N. Alexander, E. Mueller, P. Grant, J. Hennig, 5-HTTLPR S-allele: a genetic plasticity factor regarding the effects of life events on personality? Genes Brain Behav. 11 (6) (2012) 643–650. [75] M. Pluess, et al., 5-HTTLPR moderates effects of life events on neuroticism: differential susceptibility to environmental influences, Prog. Neuropsychopharmacol. Biol. Psychiatry. 34 (2010) 1070–1074. [76] J. Belsky, M. Pluess, Beyond risk, resilience, and dysregulation: phenotypic plasticity and human development, Dev. Psychopathol. 25 (4 Pt 2) (2013) 1243–1261. [77] G. Kochanska, et al., Children’s genotypes interact with maternal responsive care in predicting children’s competence: diathesis-stress or differential susceptibility? Dev. Psychopathol. 23 (2011) 605–616. [78] B.L. Hankin, et al., Differential susceptibility in youth: evidence that 5-HTTLPR x positive parenting is associated with positive affect ‘for better and worse’, Transl. Psychiatry (2011) 1. [79] D. Cicchetti, F.A. Rogosch, E.L. Thibodeau, The effects of child maltreatment on early signs of antisocial behavior: genetic moderation by tryptophan hydroxylase, serotonin transporter, and monoamine oxidase A genes, Dev. Psychopathol. 24 (3) (2012) 907–928. [80] M. Van I Jzendoorn, J. Belsky, M. Bakermans-Kranenburg, Serotonin transporter genotype 5HTTLPR as a marker of differential susceptibility? A meta-analysis of child and adolescent gene-by-environment studies, Trans. psychiatry 2 (8) (2012) e147.

[81] C. Baumann, et al., The interaction of early life experiences with COMT val158met affects anxiety sensitivity, Genes Brain Behav. 12 (8) (2013) 821–829. [82] D. Berry, et al., Gene-environment interaction between dopamine receptor D4 7-repeat polymorphism and early maternal sensitivity predicts inattention trajectories across middle childhood, Dev. Psychopathol. 25 (2) (2013) 291–306. [83] J. Belsky, M. Pluess, Genetic moderation of early child-Care effects on social functioning across childhood: a developmental analysis, Child Dev. 84 (4) (2013) 1209–1225. [84] A. Knafo, S. Israel, R.P. Ebstein, Heritability of children’s prosocial behavior and differential susceptibility to parenting by variation in the dopamine receptor D4 gene, Dev. Psychopathol. 23 (1) (2011) 53–67. [85] M.J. Bakermans-Kranenburg, et al., Experimental evidence for differential susceptibility: dopamine D4 receptor polymorphism (DRD4 VNTR) moderates intervention effects on toddlers’ externalizing behavior in a randomized controlled trial, Dev. Psychol. 44 (1) (2008) 293–300. [86] A. Park, et al., Interaction between the DRD4 VNTR polymorphism and proximal and distal environments in alcohol dependence during emerging and young adulthood, J. Abnorm. Psychol. 120 (3) (2011) 585–595. [87] M.J. Bakermans-Kranenburg, M.H. van IJzendoorn, Differential susceptibility to rearing environment depending on dopamine-related genes: new evidence and a meta-analysis, Dev. Psychopathol. 23 (1) (2011) 39–52. [88] M. Pluess, Individual differences in environmental sensitivity, Child Dev. Perspec. 9 (3) (2015) 138–143. [89] J. Belsky, et al., Vulnerability genes or plasticity genes? Mol. Psychiatry 14 (2009) 746–754. [90] F.J. Bosker, et al., Poor replication of candidate genes for major depressive disorder using genome-wide association data, Mol. Psychiatry 16 (5) (2011) 516–532. [91] A.L. Collins, et al., Hypothesis-driven candidate genes for schizophrenia compared to genome-wide association results, Psychol. Med. 42 (3) (2012) 607–616. [92] P. Donnelly, Progress and challenges in genome-wide association studies in humans, Nature 456 (7223) (2008) 728–731. [93] G.H. McClelland, C.M. Judd, Statistical difficulties of detecting interactions and moderator effects, Psychol. Bull. 114 (2) (1993) 376–390. [94] L.E. Duncan, M.C. Keller, A critical review of the first 10 years of candidate gene-by-Environment interaction research in psychiatry, Am. J. Psychiatry 168 (10) (2011) 1041–1059. [95] M.R. Munafo, J. Flint, Replication and heterogeneity in gene x environment interaction studies, Int. J. Neuropsychopharmacol. 12 (6) (2009) 727–739. [96] M. Pluess, J. Belsky, Conceptual issues in psychiatric gene-environment interaction research, Am. J. Psychiatry 169 (2) (2012) 222–232. [97] J.M. Kwon, A.M. Goate, The candidate gene approach, Alcohol. Res. Health 24 (3) (2000) 164–168. [98] I. Pe’er, et al., Estimation of the multiple testing burden for genomewide association studies of nearly all common variants, Genet. Epidemiol. 32 (4) (2008) 381–385. [99] S.M. Purcell, et al., Common polygenic variation contributes to risk of schizophrenia and bipolar disorder, Nature 460 (7256) (2009) 748–752. [100] J. Shi, et al., Common variants on chromosome 6p22: 1 are associated with schizophrenia, Nature 460 (7256) (2009) 753–757. [101] Schizophrenia Psychiatric Genome-Wide Association Study Consortium C, Genome-wide association study identifies five new schizophrenia loci, Nat. Genet. 43 (10) (2011) 969–976. [102] S. Ripke, et al., Genome-wide association analysis identifies 13 new risk loci for schizophrenia, Nat. Genet. 45 (10) (2013) 1150–1159. [103] H. Anderson-Schmidt, et al., Selected rapporteur summaries from the XX world congress of psychiatric genetics: hamburg, Germany, october 14–18, 2012, Am. J. Med. Genet. B: Neuropsychiatr. Genet. 162 (2) (2013) 96–121. [104] C. Schizophrenia Working Group of the Psychiatric Genomics, Biological insights from 108 schizophrenia-associated genetic loci, Nature 511 (7510) (2014) 421–427. [105] R. Uher, Gene?environment interactions in severe mental illness, Front. Psychiatry (2014) 48. [106] H. Aschard, et al., Challenges and opportunities in genome-wide environmental interaction (GWEI) studies, Hum. Genet. 131 (10) (2012) 1591–1613. [107] T. Otowa, et al., The first pilot genome-Wide gene-Environment study of depression in the japanese population, PLoS One (2016) e0160823. [108] E.C. Dunn, et al., GENOME-WIDE association study (gwas) and GENOME-WIDE by environment interaction study (gweis) of depressive symptoms in african american and hispanic/latina women, Depress. Anxiety 33 (4) (2016) 265–280. [109] M. Ikeda, et al., Genome-wide environment interaction between depressive state and stressful life events, J. Clin. Psychiatry 77 (1) (2016) e29. [110] L.M. Almli, et al., Correcting systematic inflation in genetic association tests that consider interaction effects: application to a genome-wide association study of posttraumatic stress disorder, JAMA Psychiatry 71 (12) (2014) 1392–1399. [111] P. Kraft, et al., Exploiting gene-environment interaction to detect genetic associations, Hum. Hered. 63 (2) (2007) 111–119. [112] F. Dudbridge, Power and predictive accuracy of polygenic risk scores, PLoS Genet. 9 (3) (2013) e1003348.

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016

G Model YSCDB-2424; No. of Pages 11

ARTICLE IN PRESS E. Assary et al. / Seminars in Cell & Developmental Biology xxx (2017) xxx–xxx

[113] W.J. Peyrot, et al., Effect of polygenic risk scores on depression in childhood trauma, Br. J. Psychiatry 205 (2) (2014) 113–119. [114] N. Mullins, et al., Polygenic interactions with environmental adversity in the aetiology of major depressive disorder, Psychol. Med. 46 (04) (2016) 759–770. [115] K.L. Musliner, et al., Polygenic risk, stressful life events and depressive symptoms in older adults: a polygenic score analysis, Psychol. Med. 45 (08) (2015) 1709–1720. [116] B.W. Domingue, et al., Genetic heterogeneity in depressive symptoms following the death of a spouse: polygenic score analysis of the US Health and Retirement Study, Am. J. Psychiatry (2017) 16111209 (p. appi. ajp. 2017). [117] S. Sen, et al., A prospective cohort study investigating factors associated with depression during medical internship, Arch. Gen. Psychiatry 67 (6) (2010) 557–565. [118] A. Trotta, et al., Interplay between schizophrenia polygenic risk score and childhood adversity in first-Presentation psychotic disorder: a pilot study, PLoS One 11 (9) (2016) e0163319. [119] L. French, et al., Early cannabis use: polygenic risk score for schizophrenia and brain maturation in adolescence, JAMA psychiatry 72 (10) (2015) 1002–1011. [120] C. Menne-Lothmann, et al., Genetic and environmental causes of individual differences in daily life positive affect and reward experience and its overlap with stress-sensitivity, Behav. Genet. 42 (5) (2012) 778–786. [121] M. Wichers, et al., The dynamic interplay between negative and positive emotions in daily life predicts response to treatment in depression: a momentary assessment study, Br. J. Clin. Psychol. 51 (2) (2012) 206–222. [122] M. Wichers, et al., Evidence that moment-to-moment variation in positive emotions buffer genetic risk for depression: a momentary assessment twin study, Acta Psychiatr. Scand. 115 (6) (2007) 451–457. [123] M. van Winkel, et al., Impact of variation in the BDNF gene on social stress sensitivity and the buffering impact of positive emotions: replication and extension of a gene–environment interaction, Eur. Neuropsychopharmacol. 24 (6) (2014) 930–938. [124] R. Plomin, C.M. Haworth, O.S. Davis, Common disorders are quantitative traits, Nat. Rev. Genet. 10 (12) (2009) 872–878. [125] P.B. Mortensen, et al., Effects of family history and place and season of birth on the risk of schizophrenia, New Engl. J. Med. 340 (8) (1999) 603–608. [126] J.J. McGrath, et al., Neonatal vitamin D status and risk of schizophrenia: a population-based case-control study, Arch. Gen. Psychiatry 67 (9) (2010) 889–894. [127] F. Varese, et al., Childhood adversities increase the risk of psychosis: a meta-analysis of patient-control, prospective-and cross-sectional cohort studies, Schizophr. Bull. (2012) sbs050. [128] C. Muntaner, et al., Socioeconomic position and major mental disorders, Epidemiol. Rev. 26 (1) (2004) 53–62. [129] S.A. Stilo, et al., Social disadvantage: cause or consequence of impending psychosis? Schizophr. Bull. 39 (6) (2013) 1288–1295. [130] G.W. Brown, et al., Functional polymorphism in the brain-derived neurotrophic factor gene interacts with stressful life events but not childhood maltreatment in the etiology of depression, Depress. Anxiety 31 (4) (2014) 326–334. [131] P. Fearon, et al., Incidence of schizophrenia and other psychoses in ethnic minority groups: results from the MRC AESOP Study, Psychol. Med. 36 (11) (2006) 1541–1550. [132] T.H. Moore, et al., Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review, Lancet 370 (9584) (2007) 319–328. [133] J. Lindert, et al., Sexual and physical abuse in childhood is associated with depression and anxiety over the life course: systematic review and meta-analysis, Int. J. Public Health 59 (2) (2014) 359–372. [134] R. Paananen, et al., Social determinants of mental health: a Finnish nationwide follow-up study on mental disorders, J. Epidemiol. Commun. Health 67 (12) (2013) 1025–1031.

11

[135] G.H. Brown, TO, Social Origins of Depression: A Study of Psychiatric Disorder in Women, Tavistock, London, 1978. [136] D.L. Dunner, V. Patrick, R.R. Fieve, Life events at the onset of bipolar affective illness, Am. J. Psychiatry 136 (4-B) (1979) 508–511. [137] Cross-Disorder Group of the Psychiatric Genomics C, Genetic relationship between five psychiatric disorders estimated from genome-wide SNPs, Nat. Genet. 45 (9) (2013) 984–994. [138] Cross-Disorder Group of the Psychiatric Genomics C, Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis, Lancet 381 (9875) (2013) 1371–1379. [139] Harriet L. MacMillan, et al., Childhood abuse and lifetime psychopathology in a community sample, Am. J. Psychiatry 158 (11) (2001) 1878–1883. [140] J.L. Garno, et al., Impact of childhood abuse on the clinical course of bipolar disorder, Br. J. Psychiatry 186 (2) (2005) 121. [141] Cross-Disorder Group of the Psychiatric Genomics Consortium, Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis, Lancet 381 (9875) (2013) 1371–1379. [142] C.C. Conway, G.M. Slavich, C. Hammen, Daily stress reactivity and serotonin transporter gene (5-HTTLPR) variation: internalizing responses to everyday stress as a possible transdiagnostic phenotype, Biol. Mood Anxiety Disord. 4 (1) (2014) 2. [143] S. Ofrat, R.F. Krueger, How research on the meta-structure of psychopathology aids in understanding biological correlates of mood and anxiety disorders, Biol. Mood Anxiety Disord. 2 (1) (2012) 13. [144] G.A. Miller, B. Rockstroh, Endophenotypes in psychopathology research: where do we stand? Annu. Rev. Clin. Psychol. 9 (2013) 177–213. [145] M. Aguilera, et al., Early adversity and 5-HTT/BDNF genes: new evidence of gene–environment interactions on depressive symptoms in a general population, Psychol. Med. 39 (09) (2009) 1425–1432. [146] C.F. Gillespie, et al., Risk and resilience: genetic and environmental influences on development of the stress response, Depress. Anxiety 26 (11) (2009) 984–992. [147] R. Keers, M. Pluess. Childhood quality influences genetic sensitivity to environmental influences across adulthood: a life-Course gene-Environment interaction study. Dev. Psychopathol. (in press). [148] L.R. Starr, et al., Sensitizing effect of early adversity on depressive reactions to later proximal stress: moderation by polymorphisms in serotonin transporter and corticotropin releasing hormone receptor genes in a 20-year longitudinal study, Dev. Psychopathol. 26 (4pt2) (2014) 1241–1254. [149] C. Kooperberg, M. Leblanc, Increasing the power of identifying gene x gene interactions in genome-wide association studies, Genet. Epidemiol. 32 (3) (2008) 255–263. [150] M.H. van IJzendoorn, et al., Gene-by-environment experiments: a new approach to finding the missing heritability. Nature Reviews, Genetics 881 (2011) (author reply 881). [151] M.J. Bakermans-Kranenburg, M.H. van IJzendoorn, The hidden efficacy of interventions: gene × environment experiments from a differential susceptibility perspective, Annu. Rev. Psychol. 66 (2015) 381–409. [152] M.H. van Ijzendoorn, M.J. Bakermans-Kranenburg, Genetic differential susceptibility on trial: meta-analytic support from randomized controlled experiments, Dev. Psychopathol. 27 (01) (2015) 151–162. [153] E. Fox, et al., The serotonin transporter gene alters sensitivity to attention bias modification: evidence for a plasticity gene, Biol. Psychiatry 70 (11) (2011) 1049–1054. [154] R. Keers, et al., A genome-Wide test of the differential susceptibility hypothesis reveals a genetic predictor of differential response to psychological treatments for child anxiety disorders, Psychother. Psychosom. 85 (3) (2016) 146–158. [155] P.M. Visscher, D. Posthuma, Statistical power to detect genetic Loci affecting environmental sensitivity, Behav. Genet. 40 (5) (2010) 728–733.

Please cite this article in press as: E. Assary, et al., Gene-environment interaction and psychiatric disorders: Review and future directions, Semin Cell Dev Biol (2017), https://doi.org/10.1016/j.semcdb.2017.10.016