Childhood maltreatment the corticotropin-releasing hormone receptor gene and adult depression in the general population.код для вставкиСкачать
RESEARCH ARTICLE Neuropsychiatric Genetics Childhood Maltreatment, the Corticotropin-Releasing Hormone Receptor Gene and Adult Depression in the General Population Hans J€orgen Grabe,1* Christian Schwahn,2 Katja Appel,1 Jessie Mahler,1 Andrea Schulz,1 Carsten Spitzer,3 Kristin Fenske,4 Sven Barnow,5 Michael Lucht,1 Harald J€urgen Freyberger,1 Ulrich John,6 Alexander Teumer,7 Henri Wallaschofski,2 Matthias Nauck,2 and Henry V€olzke8 1 Department of Psychiatry and Psychotherapy, University of Greifswald, Greifswald, Germany 2 Institute of Clinical Chemistry and Laboratory Medicine, University of Greifswald, Greifswald, Germany Department of Psychosomatic Medicine and Psychotherapy, University of Hamburg, Hamburg, Germany 3 4 Department of Biological and Clinical Psychology, University of Greifswald, Greifswald, Germany 5 Institute of Clinical Psychology, University of Heidelberg, Heidelberg, Germany Institute of Epidemiology and Social Medicine, University of Greifswald, Greifswald, Germany 6 7 Interfaculty Institute for Genetics and Functional Genomics, University of Greifswald, Greifswald, Germany 8 Institute for Community Medicine, University of Greifswald, Greifswald, Germany Received 5 March 2010; Accepted 1 September 2010 Dysregulations of the hypothalamic-pituitary-adrenal (HPA) axis have been implicated in the pathogenesis of depressive disorders and the corticotropin-releasing hormone (CRH) was found to modulate emotional memory consolidation. Recently, two studies have reported an interaction between childhood abuse and the TAT–haplotype of the CRH-Receptor Gene (CRHR1) connecting childhood adversities and genetic susceptibility to adult depression. We tested the hypothesis of an interaction of childhood maltreatment with single nucleotide polymorphisms (SNPs) and haplotypes of the CRHR1 gene not previously investigated. Caucasian subjects (n ¼ 1,638) from the German general population (Study of Health in Pomerania, SHIP) were analyzed. As in the previous studies, childhood abuse and neglect were assessed with the Childhood Trauma Questionnaire (CTQ) and depression with the Beck Depression Inventory (BDI-2). The CRHR1-SNPs were genotyped on the Affymetrix Genome-Wide Human SNP Array 6.0 platform. We identified an interaction between the TAT–haplotype and childhood physical neglect. The interaction with physical neglect showed significant (P < 0.05) results in 23 of the 28 SNPs, with rs17689882 (P ¼ 0.0013) reaching ‘‘gene-wide’’ significance. Although we did not replicate the specific interaction of abuse and the TAT–haplotype of the CRHR1 gene we confirmed the relevance of an interplay between variants within the CRHR1 gene and childhood adversities in the modulation of depression in adults. The largest effect was found for rs17689882, a SNP previously not analyzed. Relevant sample differences between this and prior studies like lower BDI-2 scores, less childhood maltreatment and higher psychosocial functioning may account for the differences in gene–environment interaction findings. 2010 Wiley-Liss, Inc. 2010 Wiley-Liss, Inc. How to Cite this Article: Grabe HJ, Schwahn C, Appel K, Mahler J, Schulz A, Spitzer C, Fenske K, Barnow S, Lucht M, Freyberger HJ, John U, Teumer A, Wallaschofski H, Nauck M, V€ olzke H. 2010. Childhood Maltreatment, the CorticotropinReleasing Hormone Receptor Gene and Adult Depression in the General Population. Am J Med Genet Part B 153B:1483–1493. Additional Supporting Information may be found in the online version of this article. Grant sponsor: Federal Ministry of Education and Research (joint grant together with Siemens Healthcare, Erlangen, Germany); Grant numbers: 01ZZ9603, 01ZZ0103, 01ZZ0403, 03ZIK012; Grant sponsor: German Research Foundation; Grant number: 1912/5-1. Hans J€ orgen Grabe and Christian Schwahn contributed equally to this work. *Correspondence to: Hans J€ orgen Grabe, M.D., Department of Psychiatry and Psychotherapy, Ernst-Moritz-Arndt-University of Greifswald, HANSE-Klinikum Stralsund, Rostocker Chaussee 70, 18437 Stralsund, Germany. E-mail: email@example.com Published online 18 October 2010 in Wiley Online Library (wileyonlinelibrary.com) DOI 10.1002/ajmg.b.31131 1483 1484 Key words: major depression; hypothalamic-pituitary-adrenal (HPA) axis; childhood neglect; general population; epidemiology; CRHR1 gene; gene–environment interaction INTRODUCTION As psychosocial and genetic factors contribute to the vulnerability to major depressive disorders (MDD) the investigation and understanding of possible interactions between environmental and genetic factors have increasingly gained importance [Sullivan et al., 2000; Zimmermann et al., 2008; Caspi et al., 2010]. There is growing evidence that dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis is an important mechanism in the pathophysiology of MDD [Holsboer et al., 1982; Heim et al., 1997; De Kloet, 2004]. The physiological stress response is primarily mediated by the release of hypothalamic corticotrophin-releasing factor, also known as corticotropin-releasing hormone (CRH). This stimulates adrenocorticotropin release from the anterior pituitary, which in turn induces the release of cortisol from the adrenal cortex. The activity of CRH at the CRH type 1 receptor (CRHR1) in extrahypothalamic regions is also thought to produce symptoms of anxiety and depression [Holsboer, 1999; Reul and Holsboer, 2002]. In addition, elevated CRH concentrations in the cerebrospinal fluid and altered CRHR1 messenger RNA expression were found in depressed patients [Nemeroff et al., 1984; Merali et al., 2004]. Some studies have suggested that early life stress precedes the HPA axis hyperactivity observed in MDD [Heim and Nemeroff, 2001; Carpenter et al., 2004]. Accordingly, Bradley et al.  investigated the association between childhood abuse and 10 single nucleotide polymorphisms (SNPs) of the CRHR1 gene in a sample of 422 subjects who were predominantly African American (97.4%) with low socioeconomic status and high rates of lifetime trauma. The subjects were evaluated while waiting for their medical appointments in a large, urban, public hospital. A distinct TAT haplotype formed by the three most significant CRHR1 SNPs (rs7209436, rs110402, and rs242924) appeared to moderate the effect of child abuse on the risk for adult depressive symptoms. The authors reported that these effects were supported by similar findings in a smaller independent sample (n ¼ 199) with subjects who were predominantly Caucasian (87.7%). Despite some differences in the genotype distribution in the different strata of the replication sample the authors did not provide any P values for the gene environment interaction in this sample. The appropriate method for the additive interaction analyses with a binary outcome measure would have been the calculation of the Relative Excess of Risk due to Interaction (RERI) [Knol et al., 2007; Grabe et al., 2009]. Thus, it remained questionable if the findings of the first sample have been replicated by the second ‘‘supportive’’ Caucasian sample. In an extended sample of the previous study Heim et al.  confirmed the protective effect of the rs110402 A-allele against the negative emotional consequences of childhood abuse in the male subsample only. This effect was carried by the exposure ‘‘physical abuse’’ only but not by the other exposure dimensions (sexual and emotional abuse). Polanczyk et al.  replicated the interaction between childhood maltreatment and the TAT haplotype of the CRHR1 gene in the E-Risk Study comprising 999 Caucasian females. This interaction was not confirmed in the Dunedin Multidisciplinary Health and Development AMERICAN JOURNAL OF MEDICAL GENETICS PART B Study comprising 899 males and female participants. Only in the ERisk Study the retrospective self-report Childhood Trauma Questionnaire (CTQ) [Bernstein et al., 2003] was used like in the study of Bradley and coworkers, but not the Dunedin Study that relied on five different sources of information generating a cumulative exposure of the first decade of life [Polanczyk et al., 2009]. Replication of findings in gene–environment interaction research is very important to the field but often replication is compromised by the use of different assessment procedures for life events and outcomes. We applied therefore, the same questionnaires for the assessment of childhood maltreatment and depression as in the study of Bradley et al. . We tested the hypothesis of an interactive effect between childhood maltreatment and variants of the CRHR1 gene in a large Caucasian general population sample from a rural region in Germany. Importantly, our sample showed less depression, lower rates of childhood maltreatment and higher psychosocial functioning as compared to prior studies. First, we aimed to replicate the specific finding on the interaction with the TAT–haplotype. Second, we selected the most effective childhood stressor for interaction effects from the TAT–haplotype interaction analyses and extended the analyses to other SNPs and haplotypes of the CRHR1 gene that were not fully covered by the previous studies. METHODS Sampling and Phenotyping Methods Sample and sample recruitment. Data from the Study of Health in Pomerania (SHIP) were used [John et al., 2001; Grabe et al., 2005]. The target population was comprised of adult German residents (20–79 years) in northeastern Germany living in three cities and 29 communities, with a total population of 212,157. The net sample (without migrated or deceased persons) comprised 6,267 eligible subjects, of which 4,308 Caucasian subjects participated in SHIP-0 (1997–2001). Follow-up examination (SHIP-1) was conducted 5 years after baseline and included 3,300 subjects. All participants gave written informed consent. Since 2007, the ‘‘Life-Events and Gene–Environment Interaction in Depression’’ (LEGENDE) study has been ongoing which is based on the SHIP-1 sample. Of the SHIP-1 sample (n ¼ 3,300), 80 subjects were deceased and 677 subjects refused participation in the LEGENDE study. N ¼ 1,683 subjects have already participated in this ongoing study, 860 subjects have not responded yet to invitation letters and are in the process of telephone recruitments and home visits. From 1,683 consecutively recruited subjects with complete data sets five subjects were excluded from the final analyses because of the diagnosis of bipolar disorder based on a diagnostic interview [Wittchen et al., 1998]. Preliminary analyses including these five subjects indicated however, that the inclusion or exclusion did not change the results. Forty subjects were excluded a priori because of an unreliable or inconsistent performance during the interview according to the judgment of the interviewer. Thus, 1,638 were eligible for the genetic analyses of the genetic main effect and of the haplotype block structure in our sample (Table I). Because of some missing values in the CTQ [Gast et al., 2001; Bernstein et al., 2003], abuse could be analyzed in 1,611 subjects, emotional neglect in 1,614 subjects and physical neglect in 1,616 subjects. The Beck GRABE ET AL. 1485 TABLE I. Description of the Sample (n ¼ 1,638) Participants Characteristic No. Sex Male 776 Female 862 (School) educationa <10 years/elementary school 525 10 years/junior high 804 High school 106 University degree 202 Employment statusa Education 21 Unemployed 175 Housewife/maternity leave 28 Retired 576 Part-time employed 161 Fully employed 676 Social welfarea No 1,629 Yes 8 Household monthly income (n ¼ 1,602)b <500 s 22 500–899 s 115 900–1,299 s 254 1,300–1,799 s 363 1,800–2,299 s 356 2,300–2,799 s 203 2,800–3,299 s 131 3,300–3,799 s 87 3,800 s 68 Marital statusa Single 185 Married 1,180 Divorced/living separately 159 Widowed 113 Partnership (married or living with partner)a No 274 Yes 1,363 Abuse (n ¼ 1,611)c No 1,468 Yes 143 Physical neglect (n ¼ 1,616) No 1,349 Yes 267 Emotional neglect (n ¼ 1,614) No 1,420 Yes 194 % 47.4 52.6 32.1 49.1 6.5 12.3 1.3 10.7 1.7 35.2 9.8 41.3 99.5 0.5 1.6 7.2 15.9 22.7 22.2 12.7 8.2 5.4 4.2 11.3 72.1 9.7 6.9 16.7 83.3 91.1 8.9 83.5 16.5 88.0 12.0 Phenotype measures. Current depressive symptoms were assessed using the BDI-2, which is a 21-item self-report questionnaire with high reliability and validity [Beck and Steer, 1987]. The BDI-2 was used as continuous outcome measure in all analyses. The CTQ was used for the self-report screening of childhood maltreatment including emotional, sexual, and physical abuse as well as emotional and physical neglect [Gast et al., 2001; Bernstein et al., 2003]. It has a total of 28 items that are rated on a five-point Likert scale with higher scores indicating a high degree of traumatic experiences. In addition to a dimensional scoring procedure, the manual provides threshold scores to determine the severity of abuse and neglect (none ¼ 0, low ¼ 1, moderate ¼ 2, and severe to extreme ¼ 3). The distribution of the BDI-2 scores dependent on the different qualities and levels of exposure (CTQ) is given in Table II. In independent studies, the CTQ was reported to have good reliability and validity; additionally, the five-factor model (i.e., the five subscales reflecting the different types of childhood trauma) has been empirically confirmed [Scher et al., 2001]. Dichotomized variables (0 and 1 vs. 2 and 3) were created for each dimension. A dichotomized composite score of abuse (emotional, physical, and/or sexual) was coded according to Bradley et al. . At least one ‘‘moderate to severe’’ score in any of these three abuse dimensions was required to yield a positive overall abuse rating. Given the relatively high rate of physical and emotional neglect both dimensions were analyzed independently from abuse. Genetic Methods SNPs and genotyping. DNA was isolated from leukocytes by standard phenol/chloroform extraction. The SHIP-0 sample was genotyped using the Affymetrix Genome-Wide Human SNP Array 6.0. Hybridization of genomic DNA was performed according to the manufacturer’s standard recommendations. The genetic data analysis workflow was created using the Software InforSense. Genetic data were stored using the database Cache (InterSystems). Genotypes were determined using the Birdseed2 clustering algorithm. For quality control purposes, several control samples where added. The overall genotyping efficiency of the GWA was 98.6%. From the Affymetrix 6.0 data set, all 37 SNPs covering the intronic and exonic regions of the CRHR1 gene were selected (HapMap II comprises 46 SNPs in the CEU sample). The SNPs were included in a detailed gene–environment analyses when the minor allele frequency (MAF) was >0.1, the Hardy–Weinberg equilibrium (HWE) was P > 0.01 and the genotype rate was >95%. These criteria were met by 28 SNPs, which were then entered in further analyses. a Data available for 1,637 subjects. 1s to 1.3 $. Abuse includes moderate to severe sexual, emotional or physical abuse. b c Depression Inventory (BDI-2) mean score [Beck and Steer, 1987] of the sample was 6.3 (SD 7.0) and the mean age was 53.6 (SD 13.6) years (range, 25.4–85.2 years). SHIP and LEGENDE were approved by the local Institutional Review Board and conformed to the principles of the Declaration of Helsinki. Statistical Analyses CRHR1 gene and linkage disequilibrium (LD) structure. The gene encoding CRHR1 is located on chromosome 17q21.31 and contains 13 exons spanning 51 kb. We used Haploview [Barrett et al., 2005] to determine the LD structure of the 28 SNPs within the CRHR1 gene in our sample (n ¼ 1,638) and to test for HWE. Using the confidence intervals according to Gabriel et al.  and the solid spine of LD method one haploblock was defined. The Four 1486 AMERICAN JOURNAL OF MEDICAL GENETICS PART B TABLE II. Mean Scores (Standard Deviation) of BDI-2 Dependent on Levels of Neglect or Abuse (Childhood Trauma Questionnaire) Level of neglect or abuse (CTQ); BDI-2 (mean score SD) Physical neglect Emotional neglect Maximum of neglect Physical abuse Emotional abuse Sexual abuse Maximum of abuse None 5.2 6.1 (n ¼ 973) 5.5 6.4 (n ¼ 991) 5.0 6.1 (n ¼ 765) 6.1 6.8 (n ¼ 1,487) 5.8 6.4 (n ¼ 1,442) 6.1 6.8 (n ¼ 1,507) 5.7 6.4 (n ¼ 1,300) Mild 7.4 8.0 (n ¼ 376) 6.9 6.9 (n ¼ 429) 6.7 7.0 (n ¼ 512) 7.6 8.7 (n ¼ 70) 8.6 9.5 (n ¼ 132) 7.7 6.7 (n ¼ 56) 7.8 8.6 (n ¼ 168) Moderate 8.6 7.5 (n ¼ 198) 8.3 9.2 (n ¼ 85) 8.3 8.1 (n ¼ 215) 9.0 8.7 (n ¼ 41) 8.1 7.5 (n ¼ 29) 8.2 10.2 (n ¼ 43) 7.7 8.5 (n ¼ 88) Severe 8.3 8.5 (n ¼ 69) 9.6 9.2 (n ¼ 109) 8.7 8.6 (n ¼ 140) 10.4 9.9 (n ¼ 31) 17.2 12.2 (n ¼ 23) 14.2 12.3 (n ¼ 17) 11.5 10.6 (n ¼ 55) ANOVA, P-value <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 The maximum of neglect is the highest level of physical and emotional neglect. The maximum of abuse is the highest level of physical, emotional, and sexual abuse. Gamete Rule algorithm yielded a three-block structure spanning regions of 23 kb, 1 kb, and 16 kb (Figs. 1 and 2). Block 1 comprised the three top-SNPs from Bradley et al.  (rs7209436, rs110402, and rs242924) that constituted the TAT–haplotype. Thus, the three-block solution was most appropriate to compare our results to the previous studies and to extent the associations to other parts of the CRHR1 gene. Ten of the 13 exons of the CRHR1 gene were located in block 3. FIG. 1. CRHR1 linkage disequilibrium (LD) map (n ¼ 1,638) demonstrating the physical location and LD pattern of the selected CRHR1 SNPs from the Affymetrix Genome-Wide Human SNP Array 6.0. Initially, all 37 SNPs covering the CRHR1 gene were selected. The SNPs were included in detailed genetic analyses when the minor allele frequency (MAF) was >0.1, the Hardy–Weinberg equilibrium (HWE) was P > 0.01 and the genotype rate was >95%. Finally, 28 SNPs as indicated were entered into the genetic analyses. The LD plot was generated with Haploview using r2 as the measure of LD, which ranges from 1 (or complete LD, indicated by black squares) to 0 (or absence of LD, indicated by white squares). The numbers within the LD plot indicate D0 (if no number is indicated, D0 ¼ 1). In our sample, we identified three separate blocks based on the Four Gamete Rule algorithm. GRABE ET AL. 1487 FIG. 2. Haplotypes of the three blocks as proposed by Haploview [Barrett et al., 2005] based on our sample (n ¼ 1,638). Haplotype Analysis Haplotype analysis has become an essential step when investigating an association between several SNPs within a gene and a phenotype. Haplotype-based analysis may help to differentiate the true effect of an SNP from what is due to its linkage disequilibrium with other variants [Tregouet et al., 2003; Frere et al., 2006]. Haplotypes may serve as better markers for unknown functional variants than SNPs alone, and importantly, they may also define functional units in which the effects cannot be predicted from what is known of the individual effect of each SNP that enters into their combination. When investigating unrelated individuals, haplotypes in general cannot be deduced from genotypes and must be statistically inferred. We therefore used THESIAS [Tregouet and Garelle, 2007; Miyamoto et al., 2008] (http://genecanvas.ecgene.net/), a multipleimputation algorithm that never assigns haplotypes to individuals (unlike software that treats haplotypes as observed), and is therefore not subjected to type I error inflation [Curtis and Sham, 2006; Tregouet and Garelle, 2007]. The THESIAS algorithm includes the possibility of (1) testing the null hypothesis of no association of haplotypes with the phenotype by means of the likelihood ratio test (with k 1 degrees of freedom for k haplotypes), (2) testing for the absence of deviation from the (by default) assumption of additivity of haplotype effects on the phenotype, and (3) adjusting for covariates and testing for interactions between these covariates and haplotypes. First the previously described TAT–haplotype was analyzed with THESIAS. Then the haplotypes were selected based on the haplotypes under the LD generated by haploview [Barrett et al., 2005] for frequencies >1% for our data set (n ¼ 1,638) (Fig. 2) for detailed analyses. We present the direct effects of the haplotypes on the BDI score and, based on the results of the individual SNPs, confirmatory interaction analyses between the haplotypes and physical neglect. By linear combinations, the BDI mean scores for each haplotype in the interaction analyses were determined. Given the testing of 3 or 4 haplotypes per block, we chose an a-level equal to 0.0125 for the analysis of each block. The direct effects of each haplotype on the BDI score are included in Table III. Regression analyses for interaction effects. Linear regression models were used to estimate the interaction effect between childhood abuse, emotional and physical neglect and the SNPs. A maximum sample size of 1,638 subjects was informative for these analyses. First, interactions between the three environmental stressors (separately) for each SNP were calculated. In agreement with Bradley et al.  an additive genetic model was applied. TABLE III. Direct and Interaction Effects With Physical Neglect for the Haplotypes of the Three Blocks Within the CRHR1 Gene With Replacement of Missing Genotypes Using THESIAS [Tregouet and Garelle, 2007] Genetic main factor; (n ¼ 1,616) Gene–environment interaction; physical neglect (n ¼ 1,616) Environment factor No. Haplotype Block 1 1 TCAAAATTG 2 CCAAGAGTG 3 CCAAGAGCA 4 CTGCGCGTG Block 2 1 ACC 2 GCG 3 GTG Block 3 1 AGAGCGCTGGCTGCTC 3 AGAACGCTGGCTGCTC 2 GATGTATGGATCATCT 4 AGAGCGCTAGCTGCTC a Haplotype; frequency; under LD (n ¼ 1,616) P-Value Mean BDI (95% CI)a Present; Absent; haplotype; haplotype; no/yes; no/yes; mean BDIa mean BDIa Coefficient for interaction (95% CI) P-Value 45.1 22.9 20.5 10.9 Ref. 0.3963 0.9876 0.1391 2.8 (2.5–3.1) 3.1 (2.6–3.5) 2.8 (2.2–3.3) 3.3 (2.7–3.9) 2.6/2.6 2.8/3.2 2.7/3.0 2.8/3.1 4.0/5.5 5.9/5.4 6.4/4.4 5.3/6.6 1.5 (0.3 to 2.7) 0.9 (2.4 to 0.6) 2.3 (4.0 to 0.6) 0.9 (0.5 to 2.4) 0.0156 0.2411 0.0098 0.2103 47.2 32.1 20.7 Ref. 0.5590 0.6492 2.9 (2.6–3.2) 3.0 (2.7–3.4) 2.7 (2.2–3.3) 2.7/2.7 2.9/3.1 2.8/3.0 3.9/5.5 5.8/5.7 6.5/4.3 1.6 (0.5 to 2.8) 0.3 (1.6 to 0.9) 2.4 (4.1 to 0.6) 0.0061 0.5923 0.0073 54.7 12.4 20.2 11.8 Ref. 0.9330 0.9295 0.1288 2.8 (2.5–3.1) 2.9 (2.2–3.5) 2.8 (2.2–3.4) 3.4 (2.7–4.1) 2.7/2.7 2.9/2.7 2.8/3.1 2.8/3.5 4.2/5.4 5.4/6.1 6.4/4.4 5.7/6.0 1.1 (0.0 to 2.3) 0.8 (0.7 to 2.3) 2.3 (4.1 to 0.6) 0.3 (2.2 to 1.6) 0.0523 0.2714 0.0086 0.7689 Mean estimated BDI value per haplotype; a combination (addition) of two haplotypes determine one BDI score. 1488 Depression (continuous BDI score) was the dependent variable and all models were adjusted for gender and age groups (10-year intervals). As the BDI scores were not normally distributed, we applied a bootstrap approach (10,000 bootstrap samples), which is robust against non-normal distributions, to obtain the P values and the CI of the regression coefficients or linear combinations. Because the bootstrap statistic was unbiased, percentile CI was chosen. Analyses were performed with STATA/MP software, version 10.1 (StataCorp LP, College Station, TX). Correction for multiple testing. The 28 SNPs within the CRHR1 gene were linked to a varying degree. It was therefore not appropriate to adjust for 28 independent tests. For the interaction analysis with all 28 SNPs, the a-level was adjusted for multiple testing according to the spectral decomposition techniques by Nyholt . Calculations were carried out using the program SNPSpD (http://genepi.qimr.edu.au/general/daleN/SNPSpD/). Based on this analysis, the a-level for the SNP analysis was corrected to be equal to 0.00386. Power Analysis We analyzed the power of our study in detail. We performed the power analyses with varying effect sizes (1–5 points of change in BDI score per allele), varying minor allele frequency (MAF) (0.1; 0.2; 0.45) and varying proportion of exposed subjects (9–16.5% according to the prevalence of exposure) using Quanto 1.2.3. (http://hydra.usc.edu/gxe) [Gauderman, 2002; Gauderman and Morrison, 2006]. For the power calculation, a sample size of 1,600 subjects, a mean outcome (BDI) of 6.3 with a standard deviation of 7.0, a marginal effect of 3.0 by the environment factor and of 0.0 by the genetic factor was assumed. The mode of inheritance was assumed to be additive. To specifically address the power of the replication of the findings of Bradley et al. , a two -sided a-level of 0.05 was chosen (Fig. 5). The power analysis for a two-sided a-level of 0.00386 (gene-wide significance) is presented in Supplemental Figure 1. RESULTS Gene–Environment Interaction of the TAT–Haplotype No interaction was observed between childhood abuse (THESIAS: coefficient for interaction 0.6 (95% CI 0.6 to 1.8); P ¼ 0.3381), emotional neglect (coefficient for interaction 0.7 (95% CI 0.4 to 1.9); P ¼ 0.1961) and the TAT–haplotype in the total sample. When analyzing female and male subjects separately, no interaction emerged (Tables I and II supplement). There was, however, an interaction between the TAT–haplotype and physical neglect in the total sample. Analyzing males and females separately females showed a somewhat weaker effect (Table III supplement). In the case of physical neglect, the estimated BDI score of carriers of this TAT–haplotype increased by 1.4 points (95% CI 0.2–2.6; P ¼ 0.0229) due to the interaction (for one copy of the TAT–haplotype). Based on the interaction between the TAT–haplotype and physical neglect in our sample all further analyses were performed with physical neglect. AMERICAN JOURNAL OF MEDICAL GENETICS PART B Gene–Environment Interaction of Haplotypes in the First, Second, and Third Haplotype Block The finding of the interaction between the TAT–haplotype and physical neglect was confirmed when analyzing the first haplotype block of the CRHR1 gene (Table III): The interaction between the first haplotype (comprising the TAT-sequence) and physical neglect increased the estimated BDI score by 1.5 (95% CI 0.3–2.7; P ¼ 0.0156) points (per copy of the haplotype). The interaction between the third haplotype (comprising the CGG-sequence) and physical neglect decreased the BDI score by 2.3 (95% CI 4.0 to 0.6; P ¼ 0.0098) points. Interactions between distinct haplotypes and physical neglect also emerged in the second and third haplotype blocks (Table III). The GTG-haplotype of the second block and the second haplotype of the third block (located at position 3 in Table III and Fig. 2) decreased the estimated BDI scores in interaction with physical neglect by 2.4 (95% CI 4.1 to 0.6; P ¼ 0.0073) and 2.3 (95% CI 4.1 to 0.6; P ¼ 0.0086), respectively. The interaction with the ACC–haplotype of the second block correspondingly increased the estimated BDI score by 1.6 (95% CI 0.5 – 2.8; P ¼ 0.0061) per copy of this haplotype. Gene–Environment Interaction of the SNPs The P-values for the interaction between the 28 SNPs of the CRHR1 gene and physical neglect are given in Figure 3. Assuming an a-level equal to 0.05, physical neglect showed statistically significant interactions in 23 of the 28 SNPs, with the top marker rs17689882 (P ¼ 0.0013) reaching gene-wide significance when adjusting for 28 SNPs (Figs. 3 and 4). Although interactions with childhood abuse and emotional neglect were not included in the further hypothesis, the corresponding P-values were indicated in Figure 3 for informative reasons. Direct Gene Effects None of the SNPs and none of the haplotypes were directly associated with the BDI score (P > 0.1). Gender Effects in Gene–Environment Interaction of the SNPs In direct response to the recent paper of Heim et al.  we further analyzed putative gender effects for our top SNP (rs17689882) and the top SNP of Heim et al. (rs110402) in interaction with physical neglect, abuse, and physical abuse. Although the interaction between rs17689882 and physical neglect was statistically present in males (P ¼ 0.0021) but statistically absent in females (P ¼ 0.1536), the P value for the three-way interaction was statistically not significant (P ¼ 0.4727). Analyzing the interaction between rs17689882 and abuse/physical abuse no significant three-way interaction with gender emerged (P > 0.3). Also in the gender stratified analyses, no significant results emerged for males or females with both exposures (P > 0.3). Analyzing gender effects for rs110402 in interaction with physical neglect, abuse, and physical abuse, no significant findings GRABE ET AL. 1489 FIG. 3. Interaction effects of CRHR1 single-nucleotide polymorphisms (SNPs) with physical neglect, emotional neglect and child abuse on adult depression. The y-axis shows the P values (log scale) for the interaction effects of the 28 SNPs for an additive genetic model. *Indicates the three top markers from Bradley et al.  forming the TAT–haplotype. $Indicates each of the three haplotype blocks (first block left side). #Indicates two exonic SNPs. Horizontal dashed lines indicate P ¼ 0.05 (nominally significant) and P ¼ 0.00386 (significance level after correction for multiple testing as described in the text). emerged for the three way interaction with gender (P > 0.4). Also in the gender stratified analyses, no significant results emerged for males or females with abuse/physical abuse (P > 0.5). However, rs110402 showed a trend for interaction with physical neglect in the gender stratified analyses (males: P ¼ 0.0794; females: P ¼ 0.1727; Table IV supplement). DISCUSSION We identified an interaction between the TAT–haplotype and childhood physical neglect in which the TAT–haplotype emerged rather as a risk than as a protective haplotype as has previously been reported [Bradley et al., 2008; Polanczyk et al., 2009]. Following our approach, we analyzed further haplotypes within a three block structure of the CRHR1 gene and 28 SNPs in total with physical neglect, the only CTQ dimension that interacted with the TAT–haplotype in our general population sample. In contrast to our best SNP (rs17689882) which was located in the third haplotype block, all three top markers (rs7209436, rs110402, rs242924) from the study of Bradley et al.  that constituted the TAT–haplotype were located within our first haplotype block and were only modestly linked to our top marker (r2 ¼ 0.21) in the third block. As far as we can conclude, our top marker has not been covered by the SNPs analyzed in the study of Bradley et al. . Therefore, we propose that rs17689882 is most closely linked to a causative variation. Further, we identified additional markers (including the two exonic markers rs16940674 and rs16940665) and haplotypes that conferred the effect of interaction in our sample compared to the African American sample of Bradley et al. . Most of our significant SNPs were located in intronic regions of the CRHR1 gene in the third haplotype block; however, 10 of the 13 exons were located in this haplotype block. This indicates that our relevant markers could be in LD with causative exonic markers in this region. Two SNPs (rs16940674 and rs16940665) in haplotype block 3 were exonic but did not change the amino acid sequence of the receptor protein. There is, however, accumulating evidence that synonymous SNPs can affect splicing or mRNA stability, thereby altering the gene products [Chamary et al., 2006]. Polanczyk et al.  replicated the interaction between childhood maltreatment and the TAT haplotype of the CRHR1 gene in the E-risk study comprising 999 Caucasian mothers of twins. However, this interaction was not confirmed in the Dunedin Multidisciplinary Health and Development Study comprising 899 male and female participants. Only the E-Risk study used the retrospective self-report CTQ [Bernstein et al., 2003] like in our study and the study of Bradley et al.  but not the Dunedin Study which relied on five different sources of information generating a cumulative exposure index by counting the number of maltreatment experiences during the first decade of life [Polanczyk et al., 2009]. As Polanczyk et al.  primarily analyzed the TAT–haplotypes no statement could be given on the role of the SNPs and haplotypes of the second and third block of the CRHR1 gene in their samples. Based on non-replications and critical meta-analyses [Munafo et al., 2009; Risch et al., 2009] there has been much debate on the validity and the comparability of findings in gene–environment interaction research. Besides the ethnical heterogeneity of different samples under investigation [Koenen and Galea, 2009; Caspi et al., 2010], the different statistical approaches [Grabe et al., 2009; Schwahn and Grabe, 2009] and the use of different clinical endpoints, the assessment of the complex nature of environmental exposures remains an unsolved issue. However, for the first time in G E interaction research several independent studies have used the CTQ. This allows for direct comparisons of the exposure between the studies. Why did we find an interaction with physical neglect but not with abuse? In our general population sample, a relatively low rate of 9% of childhood abuse (already combining emotional, physical, and sexual abuse) was found versus 12% of 1490 FIG. 4. Childhood physical neglect interacts with the SNPs of the CRHR1 gene to decrease the risk for depression in adults. A: Mean Beck Depression Inventory (BDI) scores (n ¼ 1,616) were associated with childhood physical neglect (Childhood Trauma Questionnaire) (P < 0.001). The error bars indicate robust SEM. B: Interaction effect between rs17689882 and childhood physical neglect on adult depression is presented. The sample sizes for no-to-mild childhood physical neglect and moderate-tosevere physical neglect are as follows: AA, 46, and 5, respectively; AG, 463, and 92, respectively; GG, 818, and 168, respectively. The presented P-value is based on the allelic analysis as described in the Materials and Methods Section. Note that by choosing this allelic model, the P-value is not based on a small group size. emotional neglect and 16.5% of physical neglect (Table II). In contrast, Bradley et al.  reported 37% of their sample had been exposed to childhood abuse that was identically measured with the CTQ [Bernstein et al., 2003]. In the extended sample (n ¼ 1,063) of the original Bradley et al.  paper Heim et al.  confirmed the protective effect of the rs110402 A-allele against the negative emotional consequences of childhood abuse in the male subsample only. Moreover, this effect was carried by the exposure ‘‘physical abuse’’ only but not by the other exposure dimensions (sexual and emotional abuse). These results point to gender differences and to differential effects of various exposures on the G E interaction, as found in our study. Our results provided some support for gender effects as the interaction between rs17689882 and physical neglect was statistically present in males AMERICAN JOURNAL OF MEDICAL GENETICS PART B (P ¼ 0.0021) but statistically absent in females (P ¼ 0.1536) in the gender-stratified analysis. However, the P-value for the three-way interaction (including gender) was statistically not significant. Also rs110402 showed a trend for interaction with physical neglect in the gender stratified analyses in males. Moreover, is it possible that physical neglect represents a more sensitive measure of general familial dysfunction in an overall higher functioning sample with relatively low rates of childhood abuse. Another impressive difference between both samples is reflected by the respective BDI scores: in our general population sample the BDI score was much lower (6.3; SD 7.0) compared to the mean BDI score (14.4; SD 13.1) from the study of Bradley et al. . Further, the first sample from the study of Bradley et al.  was composed of 97.4% African Americans with low socioeconomic status who were recruited while waiting for their medical appointments in a large, urban, public hospital. In contrast, our Caucasian sample reflected the German general population living in a rural area. Given those major differences in the sample and exposure structure we consider it as plausible, that a different childhood stressor emerged in interaction analyses in our sample. We therefore conclude that we found clear statistical evidence that aversive childhood conditions interact with variants within the CRHR1 gene. From a biological perspective it is likely that variants of the CRHR1 gene influence memories with emotional content which are specifically relevant for the development of depressive disorders. In fact, CRH mediates the effects of emotional arousal on memory consolidation [McGaugh, 2004] and blockade of CRHR1 receptors has been shown to impair the consolidation of fear memory [Roozendaal et al., 2002; Hubbard et al., 2007]. Polanczyk et al.  suggested that subjects carrying the protective variants of the CRHR1 gene might have an impaired activation of the fear memory consolidation which could result in a relatively unemotional cognitive processing of memories of aversive childhood experiences. Moreover, the protective variants of the CRHR1 gene have been associated with deceased cortisol responses to the dexamethasone/ CRH test in adults who have been exposed to childhood maltreatment [Heim et al., 2009; Tyrka et al., 2009]. Both mechanisms could contribute to the protective interaction of the CRHR1 gene with childhood maltreatment against depression. It is important to consider that also G E interactions act within a biological network that is influenced by other gene variants too. One recent study points to a significant G G E interaction with the serotonin transporter promoter gene polymorphism (5-HTTLPR) [Ressler et al., 2010]. The protective variants of the CRHR1 gene exerted their effect only in the carriers of the 5-HTTLPR LL-genotype. Thus, unmeasured differences in the distribution of the 5-HTTLPR genotypes could also be responsible for different findings in the CRHR1 gene childhood maltreatment studies. Some potential limitations and strengths of our study need to be discussed. The power is always a matter of concern in genetic analyses. We therefore analyzed the power of our study in detail. The largest effects in the study of Bradley et al.  were associated with the SNPs of the TAT–haplotype (rs7209436, rs110402, and rs242924). All three SNPs had an MAF >0.45 in our sample (corresponding to Fig. 5A). Nine percent of our sample was exposed to sexual, emotional, or physical abuse (by the GRABE ET AL. definition used by Bradley et al. ). Thus, an effect of 2.5 points of BDI change per allele with a power >80% (a-level of 0.05) could have been detected in our study, which is much smaller than the effect of about seven BDI points per allele reported by Bradley et al. . The same applies to emotional neglect with an even higher power (12% of our subjects). Our top marker (rs17689882) had an MAF equal to 0.205, and the proportion of subjects exposed to physical neglect was 16.5% (corresponding to Fig. 5B). Thus, a BDI change as small as 2.2 points per allele could have been detected with a power >80%. Choosing an a-level of 0.00386 (‘‘gene-wide’’ significance) and an MAF equal to 0.205, a BDI change due to an interaction as small as 3.0 points per allele could have been detected with a power of 80% (Fig. 1, Supplement). Despite the acceptable power of our analyses, false negative and positive findings cannot be ruled out completely. Haplotype analyses can provide important information regarding the combined effects of SNPs. Based on the THESIAS software, we were able to include all SNPs within a haplotype block into one analysis. Thus, we were not forced to select a limited number of tagSNPs for the analyses to get a sufficient coverage of allelic variation within each haplotype block. The THESIAS software, however, does not provide the ability to run bootstrap analyses. Therefore, no robust estimates could be calculated in haplotype analyses to overcome the non-normal distribution of the BDI scores. In order to estimate the magnitude of a possible type I error of interaction analyses we made a priori comparisons between regression analyses with and without bootstrap at the level of individual SNPs. The differences between bootstrap and conventional CIs were small and only minor differences in P values with and without bootstrapping emerged (mean standard deviation: 0.0005 0.0020). We therefore consider our results in haplotype interaction analyses to be largely unaffected by the non-normal distribution of the BDI scores. It may be important to keep in mind that the CRHR1-SNPs were genotyped on the Affymetrix GenomeWide Human SNP Array 6.0 platform. We applied reasonable quality procedures and selected SNPs with a minor allele frequency > 0.1, a HWE with P > 0.01 and a genotype rate >95%. Bradley et al.  recruited their study subjects while waiting for their medical appointments or while waiting with others who were scheduled for medical appointments in the waiting rooms of a large, urban, public hospital. In order to avoid major selection effects (e.g., treatment samples) that may bias the results, we consecutively recruited the subjects from the general population who were participating in an ongoing epidemiological study. Due to refusal in participation (n ¼ 677) and the fact that the recruitment of subjects who were non-responders to the postal invitation is still ongoing in the LEGENDE study, our analyses may not be fully representative to the German general population. Based on the DSM-IV stem questions for depression [Wittchen et al., 1999] implemented in SHIP-0 and -1, 18.1% of the recruited subjects for LEGENDE versus 20.1% of the non-recruited subjects (deceased, refused, or non-responders) in LEGENDE reported depressive symptoms in the SHIP-1 survey (P ¼ 0.14). No differences emerged between the subjects who already participated in LEGENDE (N ¼ 1,683) and the other subjects from the baseline cohort (SHIP -0) in the rate of the screening diagnosis of depression at baseline (P > 0.8). Thus, when comparing the rate of the previous screening 1491 FIG. 5. Power simulation for gene–environment interactions in our sample with varying effect sizes b (1–5 points of change in BDI score per allele) and varying proportion of exposed subjects. The minor allele frequency was 0.45 (A), 0.2 (B), and 0.1 (C). We simulated the power of our analyses with varying effect sizes, varying minor allele frequency (MAF) (0.1; 0.2; 0.45) and (9–16.5% according to the prevalence of exposure). The mode of inheritance was assumed to be additive, and a two-sided alpha level of 0.05 was chosen (Quanto 1.2.3) [Gauderman and Morrison, 2006]. diagnoses of depression no major selection bias emerged between the participants of LEGENDE and the representative baseline (SHIP-0) and follow-up cohorts (SHIP-1). Comparing the mean BDI-2 score of our sample (6.3; SD 7.0) with another sample from Germany used in a validation study of the BDI-2 (n ¼ 315), only minor differences emerged (7.69; SD ¼ 7.52) [Kuhner et al., 2007]. We therefore suggest that the generalizability of our results at least to other Caucasian general population samples with relatively low rates of childhood maltreatment should be good. Although the mean BDI score in our sample was much lower (6.3; SD 7.0) compared to the mean BDI score (14.4; SD 13.1) from the study of Bradley et al. , interaction effects with the CRHR1 1492 gene were found. For homozygote carriers we found a change of 3–BDI points due to interaction, which is lower than previously reported [Bradley et al., 2008]. The changes of 3–4 BDI points due to interaction in our sample, however, reflected a relative change in BDI scores of about 50%, which we consider a relevant effect. As we analyzed a dimensional depression score, we cannot generalize our findings to the risk of having a lifetime episode of major depression or not. Such analyses need a different statistical approach, and the reduced power compared to the dimensional endpoint needs to be addressed [Knol et al., 2007; Grabe et al., 2009]. Although we did not replicate the specific interaction of childhood abuse and the TAT–haplotype of the CRHR1 gene we confirmed the interaction of variants within the CRHR1 gene with aversive childhood conditions in the modulation of the susceptibility for depressive symptoms in adulthood. ACKNOWLEDGMENTS There are no conflicts of interest. External financial support in the past 5 years: Hans Jo€rgen Grabe: German Research Foundation; Federal Ministry of Education and Research Germany; speakers honoraria from Bristol-Myers Squibb, Eli Lilly, Novartis, Eisai, Wyeth, Pfizer, Boehringer Ingelheim, Servier, and travel funds from Janssen-Cilag, Eli Lilly, Novartis, AstraZeneca, and SALUS-Institute for Trend-Research and Therapy Evaluation in Mental Health. Carsten Spitzer: Travel funds and speakers honoraria from JanssenCilag and Boehringer Ingelheim; research grant from the ‘‘Stiftung zur Aufarbeitung der SED-Diktatur.’’ Christian Schwahn, Katja Appel, Jessie Mahler, Andrea Schulz, Kristin Fenske: none. Sven Barnow: German Research Foundation; Federal Ministry of Health Germany. Michael Lucht: German Research Foundation; Federal Ministry of Education and Research; speakers honoraria from Sanofi-Aventis, Eli Lilly Co. and Sanofi-Synthelabo. Harald J. Freyberger: German Research Foundation; Social Ministry of the Federal State of Mecklenburg-West Pomerania; Family Ministry of the Federal Republic of Germany; speakers honoraria from AstraZeneca, Lilly, Novartis and travel funds from Janssen-Cilag. Ulrich John: German Research Foundation; German Cancer Aid; European Union; Federal Ministry of Education and Research Germany; Social Ministry of the Federal State of Mecklenburg-West Pomerania of Germany. Alexander Teumer: none. Henri Wallaschofski: Research grants from Pfizer, Novartis, Novo Nordisc, SanofiAventis and the German Research Foundation. Matthias Nauck: Research grants from the Federal Ministry of Education and Research Germany, BioRad Laboratories GmbH, Siemens AG, Zeitschrift f€ ur Laboratoriumsmedizin, Bruker Daltronics, Abbott, Jurilab Kuopio, Roche Diagnostics, Dade Behring, DPC Biermann and Becton Dickinson. Henry V€olzke: Research grants by SanofiAventis, Biotronik, the Humboldt Foundation, the Federal Ministry of Education and Research (Germany) and the German Research Foundation. SHIP is part of the Community Medicine Research net of the University of Greifswald, Germany, which is funded by the Federal Ministry of Education and Research (grants no. 01ZZ9603, 01ZZ0103, and 01ZZ0403), the Ministry of Cultural Affairs and the Social Ministry of the Federal State of MecklenburgWest Pomerania. Genome-wide data have been supported by the AMERICAN JOURNAL OF MEDICAL GENETICS PART B Federal Ministry of Education and Research (grant no. 03ZIK012) and a joint grant from Siemens Healthcare, Erlangen, Germany and the Federal State of Mecklenburg-West Pomerania. The University of Greifswald is a member of the ‘‘Center of Knowledge Interchange’’ program of the Siemens AG. This work was also funded by the German Research Foundation (DFG: GR 1912/5-1). H.J.G. and C.S. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. We would like to thank Daniela Becker, Varinia Popek, Frauke Grieme, Daniel Grams, and Andrea Rieck for their contribution to the study (organization, data collection, and data management). REFERENCES Barrett JC, Fry B, Maller J, Daly MJ. 2005. Haploview: Analysis and visualization of LD and haplotype maps. Bioinformatics 21(2):263–265. Beck AT, Steer RA. 1987. Beck depression inventory—Manual. San Antonio: The Psychological Corporation. Bernstein DP, Stein JA, Newcomb MD, Walker E, Pogge D, Ahluvalia T, Stokes J, Handelsman L, Medrano M, Desmond D, et al. 2003. Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse Negl 27(2):169–190. Bradley RG, Binder EB, Epstein MP, Tang Y, Nair HP, Liu W, Gillespie CF, Berg T, Evces M, Newport DJ, et al. 2008. Influence of child abuse on adult depression: Moderation by the corticotropin-releasing hormone receptor gene. Arch Gen Psychiatry 65(2):190–200. Carpenter LL, Tyrka AR, McDougle CJ, Malison RT, Owens MJ, Nemeroff CB, Price LH. 2004. Cerebrospinal fluid corticotropin-releasing factor and perceived early-life stress in depressed patients and healthy control subjects. Neuropsychopharmacology 29(4):777–784. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE. 2010. Genetic sensitivity to the environment: The case of the serotonin transporter gene and its implications for studying complex diseases and traits. Am J Psychiatry 167(5):509–527. Chamary JV, Parmley JL, Hurst LD. 2006. Hearing silence: Non-neutral evolution at synonymous sites in mammals. Nat Rev Genet 7(2):98–108. Curtis D, Sham PC. 2006. Estimated haplotype counts from case–control samples cannot be treated as observed counts. Am J Hum Genet 78(4):729–730; author reply 728–729. De Kloet ER. 2004. Hormones and the stressed brain. Ann NY Acad Sci 1018:1–15. Frere C, Tregouet DA, Morange PE, Saut N, Kouassi D, Juhan-Vague I, Tiret L, Alessi MC. 2006. Fine mapping of quantitative trait nucleotides underlying thrombin-activatable fibrinolysis inhibitor antigen levels by a transethnic study. Blood 108(5):1562–1568. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, et al. 2002. The structure of haplotype blocks in the human genome. Science 296(5576):2225– 2229. Gast U, Rodewald F, Nickel V, Emrich HM. 2001. Prevalence of dissociative disorders among psychiatric inpatients in a German university clinic. J Nerv Ment Dis 189(4):249–257. Gauderman WJ. 2002. Sample size requirements for association studies of gene–gene interaction. Am J Epidemiol 155(5):478–484. Gauderman WJ, Morrison JM. 2006. QUANTO 1.1: A computer program for power and sample size calculations for genetic-epidemiology studies. http://hydra.usc.edu/gxe. GRABE ET AL. 1493 Grabe HJ, Lange M, Wolff B, Volzke H, Lucht M, Freyberger HJ, John U, Cascorbi I. 2005. Mental and physical distress is modulated by a polymorphism in the 5-HT transporter gene interacting with social stressors and chronic disease burden. Mol Psychiatry 10(2):220–224. Nemeroff CB, Widerlov E, Bissette G, Walleus H, Karlsson I, Eklund K, Kilts CD, Loosen PT, Vale W. 1984. Elevated concentrations of CSF corticotropin-releasing factor-like immunoreactivity in depressed patients. Science 226(4680):1342–1344. Grabe HJ, Spitzer C, Schwahn C, Marcinek A, Frahnow A, Barnow S, Lucht M, Freyberger HJ, John U, Wallaschofski H, et al. 2009. Serotonin transporter gene (SLC6A4) promoter polymorphisms and the susceptibility to posttraumatic stress disorder in the general population. Am J Psychiatry 166(8):926–933. Nyholt DR. 2004. A simple correction for multiple testing for singlenucleotide polymorphisms in linkage disequilibrium with each other. Am J Hum Genet 74(4):765–769. Heim C, Nemeroff CB. 2001. The role of childhood trauma in the neurobiology of mood and anxiety disorders: Preclinical and clinical studies. Biol Psychiatry 49(12):1023–1039. Heim C, Owens MJ, Plotsky PM, Nemeroff CB. 1997. Persistent changes in corticotropin-releasing factor systems due to early life stress: Relationship to the pathophysiology of major depression and post-traumatic stress disorder. Psychopharmacol Bull 33(2):185–192. Heim C, Bradley B, Mletzko TC, Deveau TC, Musselman DL, Nemeroff CB, Ressler KJ, Binder EB. 2009. Effect of childhood trauma on adult depression and neuroendocrine function: Sex-specific moderation by CRH receptor 1 gene. Front Behav Neurosci 3:41. Holsboer F. 1999. The rationale for corticotropin-releasing hormone receptor (CRH-R) antagonists to treat depression and anxiety. J Psychiatry Res 33(3):181–214. Holsboer F, Liebl R, Hofschuster E. 1982. Repeated dexamethasone suppression test during depressive illness. Normalisation of test result compared with clinical improvement. J Affect Disord 4(2):93–101. Hubbard DT, Nakashima BR, Lee I, Takahashi LK. 2007. Activation of basolateral amygdala corticotropin-releasing factor 1 receptors modulates the consolidation of contextual fear. Neuroscience 150(4):818– 828. John U, Greiner B, Hensel E, Ludemann J, Piek M, Sauer S, Adam C, Born G, Alte D, Greiser E, et al. 2001. Study of Health In Pomerania (SHIP): A health examination survey in an east German region: Objectives and design. Soz Praventivmed 46(3):186–194. Knol MJ, van der Tweel I, Grobbee DE, Numans ME, Geerlings MI. 2007. Estimating interaction on an additive scale between continuous determinants in a logistic regression model. Int J Epidemiol 36(5):1111–1118. Koenen KC, Galea S. 2009. Gene–environment interactions and depression. JAMA 302(17):1859–1862. Kuhner C, Burger C, Keller F, Hautzinger M. 2007. Reliability and validity of the Revised Beck Depression Inventory (BDI-II). Results from German samples. Nervenarzt 78(6):651–656. McGaugh JL. 2004. The amygdala modulates the consolidation of memories of emotionally arousing experiences. Annu Rev Neurosci 27: 1–28. Merali Z, Du L, Hrdina P, Palkovits M, Faludi G, Poulter MO, Anisman H. 2004. Dysregulation in the suicide brain: mRNA expression of corticotropin-releasing hormone receptors and GABA(A) receptor subunits in frontal cortical brain region. J Neurosci 24(6):1478–1485. Miyamoto Y, Shi D, Nakajima M, Ozaki K, Sudo A, Kotani A, Uchida A, Tanaka T, Fukui N, Tsunoda T, et al. 2008. Common variants in DVWA on chromosome 3p24.3 are associated with susceptibility to knee osteoarthritis. Nat Genet 40(8):994–998. Munafo MR, Durrant C, Lewis G, Flint J. 2009. Gene environment interactions at the serotonin transporter locus. Biol Psychiatry 65(3):211–219. Polanczyk G, Caspi A, Williams B, Price TS, Danese A, Sugden K, Uher R, Poulton R, Moffitt TE. 2009. Protective effect of CRHR1 gene variants on the development of adult depression following childhood maltreatment: Replication and extension. Arch Gen Psychiatry 66(9):978–985. Ressler KJ, Bradley B, Mercer KB, Deveau TC, Smith AK, Gillespie CF, Nemeroff CB, Cubells JF, Binder EB. 2010. Polymorphisms in CRHR1 and the serotonin transporter loci: Gene gene environment interactions on depressive symptoms. Am J Med Genet Part B 153B(3): 812–824. Reul JM, Holsboer F. 2002. Corticotropin-releasing factor receptors 1 and 2 in anxiety and depression. Curr Opin Pharmacol 2(1):23–33. Risch N, Herrell R, Lehner T, Liang KY, Eaves L, Hoh J, Griem A, Kovacs M, Ott J, Merikangas KR. 2009. Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk of depression: A meta-analysis. JAMA 301(23):2462–2471. Roozendaal B, Brunson KL, Holloway BL, McGaugh JL, Baram TZ. 2002. Involvement of stress-released corticotropin-releasing hormone in the basolateral amygdala in regulating memory consolidation. Proc Natl Acad Sci USA 99(21):13908–13913. Scher CD, Stein MB, Asmundson GJ, McCreary DR, Forde DR. 2001. The childhood trauma questionnaire in a community sample: Psychometric properties and normative data. J Trauma Stress 14(4):843–857. Schwahn C, Grabe HJ. 2009. Gene–environment interactions and depression. JAMA 302(17):1859–1862. Sullivan PF, Neale MC, Kendler KS. 2000. Genetic epidemiology of major depression: Review and meta-analysis. Am J Psychiatry 157(10):1552–1562. Tregouet DA, Garelle V. 2007. A new JAVA interface implementation of THESIAS: Testing haplotype effects in association studies. Bioinformatics 23(8):1038–1039. Tregouet DA, Barbaux S, Poirier O, Blankenberg S, Bickel C, Escolano S, Rupprecht HJ, Meyer J, Cambien F, Tiret L. 2003. SELPLG gene polymorphisms in relation to plasma SELPLG levels and coronary artery disease. Ann Hum Genet 67(Pt6): 504–511. Tyrka AR, Price LH, Gelernter J, Schepker C, Anderson GM, Carpenter LL. 2009. Interaction of childhood maltreatment with the corticotropinreleasing hormone receptor gene: Effects on hypothalamic-pituitaryadrenal axis reactivity. Biol Psychiatry 66(7):681–685. Wittchen HU, Lachner G, Wunderlich U, Pfister H. 1998. Test-retest reliability of the computerized DSM-IV version of the Munich-Composite International Diagnostic Interview (M-CIDI). Soc Psychiatry Psychiatr Epidemiol 33(11):568–578. € un B, M€ Wittchen H-U, H€ ofler M, Gander F, Pfister H, Storz S, Ust€ uller N, Kessler RC. 1999. Screening for mental disorders: Performance of the Composite International Diagnostic-Screener (CID-S). Int J Meth Psychiat Res 8(2):59–70. Zimmermann P, Bruckl T, Lieb R, Nocon A, Ising M, Beesdo K, Wittchen HU. 2008. The interplay of familial depression liability and adverse events in predicting the first onset of depression during a 10-year follow-up. Biol Psychiatry 63(4):406–414.