Impact of Schizophrenia Candidate Genes on Schizotypy and Cognitive Endophenotypes at the Population Level Nicholas C. Stefanis, Thomas A. Trikalinos, Dimitrios Avramopoulos, Nikos Smyrnis, Ioannis Evdokimidis, Evangelia E. Ntzani, John P. Ioannidis, and Costas N. Stefanis Background: Aspects of cognitive function and schizotypy have been proposed as potential endophenotypes for schizophrenia. It is unknown whether the expression of these endophenotypes at the population level is modulated by the genetic variability of candidate susceptibility genes for schizophrenia. Methods: We examined the potential impact of 18 single nucleotide polymorphisms (SNPs) within the DTNBP1, NRG1, DAOA/G32, and DAAO genes, on cognition and self-rated schizotypy, in a representative population of 2243 young male military conscripts. Single SNP and haplotype associations were evaluated. Results: The DTNBP1 SNPs rs2619522 and rs760761 exhibited several single marker associations, the minor alleles being associated with lower attention capacity but also a decrease in positive and paranoid schizotypy scores. The DTNBP1 haplotype load had borderline associations with nonverbal IQ, paranoid schizotypy, and sustained attention. For individual NRG1 polymorphisms, isolated but weak signals of association were noted with sustained attention and working memory but not schizotypy. The risk allele of functional SNP8NRG243177 was associated with reduced spatial working memory capacity. An isolated effect of DAAO haplotype variability was noted on negative and disorganization schizotypy. No convincing association of DAOA/G32 variability was detected. Conclusions: The DTNBP1 and, less so, NRG1 and DAAO variants might exert gene-specific modulating effects on schizophrenia endophenotypes at the population level. Key Words: Cognition, DAAO, DAOA/G32, DTNBP1, endophenotype, NRG1, schizophrenia, schizotypy

A

promising approach to the study of the genetic vulnerability for multifactorial and heterogenous disorders such as schizophrenia is the identification of endophenotypes (Gottesman and Gould 2003). Endophenotypes are measurable components along the pathway between the genetic infrastructure and the presentation of a disorder. Both schizotypy and cognitive function might serve as promising endophenotypes for schizophrenia. Family and adoption studies suggest a greater prevalence of schizotypal features (Kendler et al. 1995; Tienari et al. 2003) in relatives of patients with schizophrenia than in comparison groups. Schizotypal features also exhibit modest heritability (Lin et al. 2006; Linney et al. 2003). Formulated either as self-rated subclinical psychotic symptoms (Stefanis et al. 2002) From the University Mental Health Research Institute (NCS, DA, NS, IE, CNS);Department of Psychiatry (NCS, NS), National and Kapodistrian University of Athens, Athens; Clinical and Molecular Epidemiology Unit (TAT, EEN, JPI), Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina; Biomedical Research Institute (JPI), Foundation for Research and Technology-Hellas, Ioannina, Greece; Department of Psychological Medicine (NCS), Institute of Psychiatry, King’s College London, United Kingdom; and the Institute for Clinical Research and Health Policy Studies (TAT, JPI), Department of Medicine, Tufts University School of Medicine, Boston, Massachusetts. Address reprint requests to Nicholas C. Stefanis, M.D., Department of Psychiatry, National and Kapodistrian University of Athens Medical School, Eginition Hospital, 74 Vas. Sofias Ave., Athens 11528, Greece; E-mail: [email protected]. Received August 3, 2006; revised November 10, 2006; accepted November 15, 2006.

0006-3223/07/$32.00 doi:10.1016/j.biopsych.2006.11.015

or as self-rated schizotypal personality traits (Stefanis et al. 2004), schizotypy can be quantitatively expressed at the population level and represented in factorial dimensions. One might thus view psychosis as a quantitative trait with a distribution extended and measured into the general population (Johns and van Os 2001). Endophenotypes of schizophrenia might also include neurocognitive traits. Attention (Cornblatt and Malhotra 2001), working memory (Cannon et al. 2000), and anti-saccade eye-movements (Ettinger et al. 2006) satisfy most criteria put forward for endophenotypes by Gottesman and Gould (2003). Several candidate genes have been proposed and repeatedly assessed for association with schizophrenia. It would be interesting to evaluate whether these genes are associated with the aforementioned endophenotypes. This might help in understanding potential mechanisms of action on schizophrenia liability. We aimed to address this question, targeting four of the most prominent candidate genes: Neuregulin1 (NRG1; 8p21-12) (Stefansson et al. 2002), Dysbindin (DTNBP1) (6p22.3) (Straub et al. 2002), DAOA/G32 (D-amino-acid oxidase activator, previously known as G72) (13q33), and D-amino-acid oxidase (DAAO; 12q24-11) (Chumakov et al. 2002). All these genes have been evaluated in gene– disease associations with promising results in more than one study. Moreover, NRG1 and DTNBP1 are also located within regions for which linkage studies have shown relatively consistent signals (Lewis et al. 2003). However, the verdict on these candidate markers is not final yet. Some of the published gene– disease association studies to date have failed to show statistical significant associations. Moreover, the processes by which these candidate genes increase risk for schizophrenia remain unknown. A plausible scenario would be that they modulate at a molecular level endophenotypes associated with schizophrenia such as schizotypal features and/or cognitive function. We set out to BIOL PSYCHIATRY 2007;62:784 –792 © 2007 Society of Biological Psychiatry

N.C. Stefanis et al. evaluate this hypothesis in a large study population, evaluating 18 single nucleotide polymorphisms (SNPs) in these four genes and their haplotypes.

Methods and Materials Subjects The ASPIS (Athens Study of Psychosis Proneness and Incidence of Schizophrenia) examined 2243 randomly selected young male conscripts, ages 18 –24 years, from the Greek Air Force in their first 2 weeks of admission to the National Basic Air Force Training Center. Eight separate waves of conscripts were assessed between January 1999 and March 2000. Individuals at this age are most likely to display the clinical and subclinical experiences of psychosis (Johns and van Os 2001). Military service is compulsory in Greece, and all healthy men are recruited and assigned to the different army corps by random assignment. Of the 2243 conscripts, 167 did not participate for genetic testing. No conscript was excluded owing to medical conditions. Assessment All conscripts had already received a standardized screening interview by a team of army medical doctors of different specialties, and major medical conditions had been excluded. Conscripts underwent an extensive interview of computerized neurocognitive abilities and a self-rated psychometric evaluation. After obtaining written informed consent, DNA was extracted from mouthwash samples. After a short pretest trial, each subject performed eye-movement tasks and cognitive tasks. These included assessment of sustained attention with the Continuous Performance Task-Identical Pair version (CPT-IP) (Cornblatt et al. 1988); verbal and spatial versions of the n-back task to assess verbal and spatial working memory (Gevins et al. 1996); and anti-saccade eye movement task (Smyrnis et al. 2003). We a priori decided to exclude data from further analyses, if the central index of performance (d’) on CPT-IP or 2-back was ⱕ 0; if there were ⱖ 3 unsuccessful trials (of 5) for verbal and spatial 2-back; and if ⱖ 50 invalid anti-saccade trials (of 90). Conscripts completed a psychometric battery of self-administered questionnaires. Amongst other tests, the assessment battery included the Raven Progressive Matrices (Raven 1982); the Schizotypal Personality Questionnaire (SPQ) (Raine et al. 1991; Stefanis et al. 2004); Perceptual Aberration Scale (PAS) (Chapman et al. 1978); and Community Assessment of Psychic Experiences (CAPE) (Stefanis et al. 2002). All questionnaires have been translated and standardized for use in the Greek population. The CAPE questionnaire was not administered to the first two waves of conscripts (n ⫽ 487). We used the four validity items of the Temperament and Character Inventory (TCI) questionnaire (Cloninger et al. 1993) to exclude subjects who responded incorrectly in at least one of these items (random responders). This study was approved by the Bioethics and Medical Deontology Committee of the University Mental Health Research Institute. DNA Extraction, SNP Selection, and Genotyping Mouthwash samples for DNA extraction were chosen as described previously (Avramopoulos et al. 2002) to obtain a better procedure acceptance rate. It has been shown in large samples that there are no genotyping discrepancies between DNA obtained from blood and mouthwash (de Vries et al. 1996). We chose which SNPs to test with a systematic approach: we

BIOL PSYCHIATRY 2007;62:784 –792 785 performed systematic PubMed searches (last search December 2004) on the potential genetic associations of the aforementioned genes with schizophrenia. We scrutinized all information from population-based and family-based association studies. We did not consider any micro-satellite markers, owing to possible technical difficulties in allele calling. For DTNBP1 we selected six of the most positive SNPs reported originally by Straub et al. (2002): rs2619539 (p1655), rs3213207 (p1635), rs1011313 (p1325), rs2005976 (p1757), rs760761 (p1320), and rs2619522 (p1763). For these SNPs, linkage with schizophrenia was demonstrated both in a sib-pair sample that included the original families and in a sample of independently ascertained triads. Schwab et al. (2003) reported positive associations with schizophrenia for rs3213207, rs1011313, and rs760761. We chose a seventh SNP that was identified as a tag SNP for a high-risk haplotype for schizophrenia (rs1018381 [p1578]) in a follow-up analysis of the Straub et al. (2002) data (van den Oord et al. 2003). Moreover, rs1018381 was significantly associated with schizophrenia in another study (Funke et al. 2004). We evaluated, regarding the NRG1 gene, the five SNPs included in the originally proposed Stefansson “core” haplotype (Stefansson et al. 2002): SNP8NRG221132, SNP8NRG221533, SNP8NRG241930, SNP8NRG243177, and SNP8NRG433E1006. After scrutinizing the available studies on the potential association between DAOA/G30 and schizophrenia and taking into consideration the available—at the time—linkage disequilibrium (LD) estimates (rs3918342 and M24 showed a D’ of .99, for example), we chose to genotype SNPs rs2391191 (M15), rs778293 (M22), and rs3918342 (M23). These were the most widely replicated in European descent populations (Addington et al. 2004; Chumakov et al. 2002; Hall et al. 2004; Korostishevsky et al. 2004; Schumacher et al. 2004). During the planning of our study, data on the potential association of DAAO and schizophrenia were limited (Addington et al. 2004; Chumakov et al. 2002; Liu et al. 2004; Schumacher et al. 2004). We chose to assess SNPs rs2111902, rs3918346, and rs3741775, which were genotyped in all available studies and often showed a strong signal for potential association with schizophrenia. Overall we chose 18 SNPs, on the basis of the number of genetic association studies published and the frequency of statistically significant findings. Alleles are named as they appear in dbSNP build 123. All genotyping was performed blind to phenotype measures by K-Biosciences (Herts, United Kingdom) (http://www.kbioscience. co.uk/) with a competitive allele-specific polymerase chain reaction system (CASP). Overall genotyping error rate has been estimated to be ⱕ .3% (on the basis of the repeatability of the genotype calls in the same samples). Outcomes We assessed whether there were any genetic determinants of Raven IQ, separately for each gene. We quantified schizotypal traits as continuous measurements with the total score from the SPQ instrument, the total CAPE score, and the total PAS score. We assessed separately the four schizotypal latent dimensions of SPQ (cognitive/perceptual, negative, disorganization, and paranoid factor) and the three latent dimensions of CAPE (positive, negative, and depression factor), especially for the former two instruments. Moreover, we measured neurocognitive performance on four tasks (sustained attention, short-term spatial and verbal working memory, and anti-saccade eye movements). www.sobp.org/journal

786 BIOL PSYCHIATRY 2007;62:784 –792 These measures were quantified with the corresponding sensitivity indexes d’ (d’-CPT-IP, d’-S2B, d’-V2B) and anti-saccade error rate, respectively. Statistical Analyses For every SNP in each gene, we assessed compliance with the Hardy-Weinberg (HW) law with an exact test. Main analyses excluded SNPs whose frequencies violated the HW law at p ⬍ .01, and these SNPs were also not considered in the construction of haplotypes. Secondary analyses including all SNPs gave similar results (not shown). We estimated the D’ coefficient for SNP pairs within each gene. We used Haploview 3.2 (Barrett et al. 2005). Haplotypes for each gene were reconstructed with PHASE 2.1.1 (Stephens et al. 2001; Stephens and Scheet 2005). Convergence of the Markov Chain Monte Carlo chains in PHASE was assessed by multiple (at least five) runs. Haplotypes with frequency ⬍ 5% were grouped in a miscellaneous (“rare”) category. We checked for systematic differences in genotype and haplotype frequencies between those conscripts who were finally eligible and those who were excluded on the basis of their random or incomplete questionnaire responses. Differences in the distributions were tested with ␹2 tests for individual SNPs and with a permutation test implemented in PHASE for haplotypes (Stephens et al. 2001). We assessed the impact of genetic factors on phenotype with regression models. Single SNP analyses and haplotypebased analyses were both performed. In single SNP evaluations, the distribution of each outcome/genotypic group was assessed with analysis of variance (ANOVA). We also examined with linear regressions whether the number of minor allele copies was associated with each outcome (allele-load or allele-based additive models). In haplotype-based assessments, each estimated haplotype was weighted according to its probability, as calculated by PHASE (haplotype-load or haplotype-based additive models). We did not use diplotype analyses, because there was no good reason to assume specific genetic models. Moreover, diplotypes would increase the multiplicity of potential comparisons, and power would be considerably lower for specific pairwise comparisons. We relied on a likelihood ratio test (the “LR p” value in the Results and Tables) to assess whether a regression model taking into account the genetic factors provided better fit (explained data better) than a constant only model. In the regression models, the most common haplotype was used as the reference category. This choice was arbitrary but is mathematically equivalent to choosing any haplotype as the reference category. The main analyses did not adjust for non-genetic factors. Owing to Mendelian randomization, it is probably not necessary to adjust for age, IQ, or other factors, traditionally held as potential confounders in the epidemiology of schizophrenia. However, in a secondary analysis, we adjusted for age, IQ, and their interaction. In the latter case, we based inferences on the presence of genetic effects by assessing with a likelihood ratio test whether a model that takes age, IQ, and gene haplotypes into account provides better fit than a model based on age and IQ only. Finally, for all outcomes, we tested for potential interaction between DAOA/G32 and DAAO haplotypes in a two-way ANOVA framework. This interaction between haplotypes was assessed on the basis of a priori reasoning, because functional data support that DAOA/G32 is the activator of DAAO (Chumakov et al. 2002). Association analyses were performed www.sobp.org/journal

N.C. Stefanis et al. with Intercooled Stata 8.2 (Stata Corp, College Station, Texas). The reported p values are two-tailed and uncorrected for multiple comparisons.

Results Descriptive Data In total, 2243 randomly selected young male conscripts (mean age, 20.7 ⫾ 1.90) entered the study. Many SNP pairs were in high LD (Table 1). The yield of the genotyping process on the basis of the mouthwash material ranged between 71.2% and 82.8% for the 18 SNPs (Table 2). Genotype distributions of rs2005976 of the DTNBP1 gene deviated from those expected by the HW law beyond chance (p ⫽ .0001) (Table 2). Supplementary Tables 1 and 2 summarize the pairwise correlations and the distributional characteristics of the demographics and the response variables. The most common haplotypes were GCGTG for NRG1 (32.9%), GTCGAG for DTNBP1 (49.5%), GCC (29.3%) for DAOA/ G32, and TGC (36.8%) for DAAO. Detailed haplotype frequencies are available upon request. Differences Between Eligible and Non-Eligible Responders The proportion of conscripts who gave eligible responses and measurements varied between 49% for CAPE and 90% for cognitive trait measurements. Approximately 60% of the conscripts gave eligible responses in the SPQ and PAS questionnaires. Compared with conscripts who gave eligible responses, those with non-eligible responses were younger (average age difference ranging from 6 to 10 months for the various instruments, p ⬍ .001 in all instances) and performed somewhat worse in the Raven nonverbal IQ scale (average Raven IQ difference ranged from 5 to 8 units, p ⬍ .001 in all instances). However, random responders did not differ from correct responders in psychopathology measures such as paranoid ideation, psychoticism, anxiety, or depression. Furthermore, there were no statistically significant differences in the genotype and haplotype distributions between eligible and non-eligible responders for all four genes (p values ranged between .10 and .95 for all analyses). Associations With Individual SNPs—Model Free Approach On the basis of ANOVA considering all three genotypes, 20 SNP– outcome associations crossed the .05 level of significance (Table 3). The majority (14 of the 20 signals) pertained to DTNBP1 markers; whereas 4 involved NRG1, only 2 involved DAAO, and no signal was seen for DAOA/G32. In particular, rs760761 and rs2619522, two DTNBP1 markers in high LD, had 6 and 5 formally significant associations, respectively. The most significant associations were seen with the paranoid dimension of SPQ (p ⫽ .005 and p ⫽ .007 for these two SNPs, respectively). Moreover, formally significant associations were seen with DTNBP1 SNPs for all SPQ dimensions except for the negative one as well as for Raven IQ and sustained attention. The signals for NRG1 were seen specifically for cognitive ability (sustained attention and spatial and verbal working memory). For DAAO, the significant signals pertained to the CAPE depression dimension and spatial working memory (Table 3). Associations With Individual SNPs—Allele-Load Models Table 4 shows in detail associations for DTNBP1 SNPs and various endophenotypes and the estimated effect sizes. As shown, there was a larger number of formally statistically significant signals (n ⫽ 17), when analyses used an additive allele-load model. The majority of signals pertained to rs760761 (6 significant signals), rs2619522 (7 significant signals), or rs1018381 (4

BIOL PSYCHIATRY 2007;62:784 –792 787

N.C. Stefanis et al.

Table 1. DTNBP1, NRG1, DAOA/G30, and DAAO Intermarker Linkage Disequilibrium Measured With r2 DTNBP1

rs2619539 rs3213207 rs1011313 rs760761 rs2619522 rs1018381

rs2619539

rs3213207

rs1011313

rs760761

rs2619522

rs1018381

—

.14 —

.12 .01 —

⬍.01 .49 .02 —

⬍.01 .50 .02 .98 —

.06 .01 .01 .40 .40 —

243177

433E1006

NRG1a

221132 221533 241930 243177 433E1006

221132

221533

241930

—

.07 —

.27 .22 —

.08 .68 .31 —

.01 ⬍.01 .02 ⬍.01 —

DAOA/G30

rs2391191 rs778293 rs3918342

rs2391191

rs778293

rs3918342

—

.03 —

.02 .48 —

rs2111902

rs3918346

rs3741775

—

.73 —

.14 .29 —

DAAO

rs2111902 rs3918346 rs3741775

a For NRG1 prefix “SNP8NRG” to the single nucleotide polymorphism (SNP) number; it was omitted because of space limitations in the Table.

significant signals), and the lowest p value was .001 (paranoid dimension of SPQ with rs2619522). The minor risk alleles were generally associated with lower sustained attention capacity (rs760761 and rs2619522) and with lower paranoid schizotypal traits and positive psychotic-like experiences (rs760761, rs2619522, and rs1018381). Effect sizes were modest, even for the most significant associations (Table 4). In allele-load models, three of the four NRG1 signals identified in the model-free analyses remained formally significant. The same was true for one of the two DAAO signals identified in the model-free analyses. The effect sizes were still modest (.12 points/NRG1 SNP8NRG433E1006 minor allele load on sustained attention and on verbal working memory performance, ⫺.10 points/NRG1 SNP8NRG243177 minor allele load on spatial working memory performance, and ⫺.47 points/DAAO rs3741775 minor allele load on the CAPE depression factor). Haplotype-Based Analyses Analyses of the haplotypes did not yield any further insight on potential associations. Unadjusted DTNBP1 haplotype load (LR p value) revealed signals of borderline significance with Raven IQ, the paranoid SPQ dimension, and sustained attention capacity (Table 5), consistent with the analyses of single polymorphisms. Adjusted analyses (for Raven IQ and age) did not yield any prominent associations, but a nominally significant effect of DTNBP1 haplotype load on the paranoid SPQ dimension was still observed (p ⫽ .044).

There was no compelling evidence supporting that the haplotypes of the selected polymorphisms in NRG1 and DAOA/G32 were associated with the examined quantitative traits, with the exception of a very marginal association of the latter with spatial working memory (Supplementary Tables 3 and 4). The most common five marker NRG1 haplotype in this population (reference haplotype GCGTG) overlapped with the original “core” haplotype identified by Stefansson et al. (2002). Conversely, we observed an association of DAAO haplotypes with the negative and disorganization dimensions of SPQ that had not been formally detected on the basis of the analyses of single markers. This persisted in adjusted analysis for age and IQ. The TGA haplotype carriers demonstrated significantly higher negative (⫹.014 points on average) and disorganization (⫹.024 points on average) schizotypal traits than individuals with the most common TGC haplotype (Supplementary Table 5). There was no evidence for interaction between DAOA/G32 and DAAO haplotypes (p ⬎ .19 for this interaction for all phenotypes).

Discussion Our analyses have focused on variants in four genes implicated in determining susceptibility to schizophrenia in previous studies. We have found a number of tentative signals for putative associations with endophenotypes that might be relevant to the pathogenesis of schizophrenia. Most of the detected signals www.sobp.org/journal

788 BIOL PSYCHIATRY 2007;62:784 –792

N.C. Stefanis et al.

Table 2. Genotype Frequencies Gene, SNP NRG1 221132 221533 241930 243177 44E1006 DTNBP1 rs2619539 rs3213207 rs1011313 rs2005976 rs760761 rs2619522 rs1018381 DAOA/G30 rs2391191 rs778293 rs3918342 DAAO rs2111902 rs3918346 rs3741775

Homozygotes, Minor Allele

Heterozygotes

Homozygotes, Major Allele

Exact pHWE

Successful Genotyping (%)

AA ⫽ 31 CC ⫽ 242 TT ⫽ 178 TT ⫽ 291 AA ⫽ 13

AG ⫽ 344 CT ⫽ 786 GT ⫽ 678 CT ⫽ 726 AG ⫽ 313

GG ⫽ 1402 TT ⫽ 726 GG ⫽ 797 CC ⫽ 586 GG ⫽ 1532

.08 .21 .07 .013 .58

79.2 78.2 73.7 71.5 82.8

CC ⫽ 316 CC ⫽ 14 TT ⫽ 11 TT ⫽ 19 AA ⫽ 54 CC ⫽ 53 AA ⫽ 12

CG ⫽ 909 CT ⫽ 270 CT ⫽ 275 CT ⫽ 462 AG ⫽ 527 AC ⫽ 524 AG ⫽ 257

GG ⫽ 589 TT ⫽ 1370 CC ⫽ 1563 CC ⫽ 1511 GG ⫽ 1247 AA ⫽ 1257 GG ⫽ 1586

.29 .88 .88 .0001 .94 .93 .62

80.9 73.7 82.4 71.2 81.5 81.8 82.7

AA ⫽ 201 CC ⫽ 311 TT ⫽ 447

AG ⫽ 767 CT ⫽ 761 CT ⫽ 918

GG ⫽ 847 TT ⫽ 556 CC ⫽ 465

.18 .08 .93

80.9 72.6 81.6

GG ⫽ 192 GG ⫽ 141 CC ⫽ 350

GT ⫽ 748 AG ⫽ 642 AC ⫽ 893

TT ⫽ 710 AA ⫽ 833 AA ⫽ 612

.87 .27 .45

73.6 72.0 82.7

All subsequent analyses exclude single nucleotide polymorphism (SNP) rs2005976 that overtly violated HardyWeinberg equilibrium (HWE). The SNP 243177 is considered in the reported analyses. Exclusion of this SNP did not change any of the results (not shown). pHWE, p value for HWE.

pertained to DTNBP1 variants—in particular, the highly linked rs760761, rs2619522, and rs1018381 SNPs. Some additional signals of interest were seen for DAAO and NRG1, whereas we found no clear associations for any of the endophenotypes with DAOA/G32. None of the p values were lower than .005, and thus they should be interpreted cautiously, given the number of comparisons performed. Most of the association signals pertained to DTNBP1. The minor alleles of rs760761, rs2619522, and rs1018381 were consistently associated with a small preferential decrease in positive (rather than negative) schizotypy scores and positive psychoticlike experiences. In contrast, the minor alleles of rs760761 and rs2619522 were also associated with decreased sustained attention capacity. Thus we found contradictory effects among alleles that were associated with schizophrenia in previous studies. Several explanations might be attempted. Firstly, there is considerable discrepancy as to which are the DTNBP1 risk alleles for schizophrenia (Mutsuddi et al. 2006). Mutsuddi et al. (2006) have shown that DTNBP1 haplotype frequencies are very similar across European countries, arguing against an allele stratification effect that could explain these discrepancies, and have provided a dense genetic map for DTNBP1 that might be used in future studies to try to reconcile the proposed associations with different markers in each study. Secondly, the presumed risk alleles (minor alleles of rs760761, rs2619522, and rs1018381) are included in the risk haplotype that has been associated with general cognitive impairment (Burdick et al. 2006) in healthy control subjects. The potential relevance to schizotypy is provided by Hallmayer et al. (2005), in which cognitively impaired patients that accounted for the linkage of schizophrenia to chromosome 6p25-24 were reported to have lower positive schizotypal scores than cognitively spared patients. This might imply that candidate genes within this or nearby regions, such as DTNBP1, might impact on “cognitive www.sobp.org/journal

efficiency” (Hallmayer et al. 2005), which in turn might affect the ability to elaborate on complex psychic experiences such as delusions and hallucinations. This might partially explain why DTNBP1 risk alleles, which are associated with cognitive impairment (Burdick et al. 2006), are also associated with negative rather than positive psychotic symptoms (DeRosse et al. 2006; Fanous et al. 2005). The DTNBP1 risk alleles, impacting on cognitive function, might bias individuals to underreport paranoia and psychotic-like symptoms in this study. However, a simple alternative explanation would be that some of these findings are due to chance (false positives). We observed a nominally significant effect of DAAO haplotype variability on negative and disorganization schizotypal personality scores. The TGA haplotype was associated with stronger negative and disorganization schizotypal personality traits than the TGC haplotype. Association with schizophrenia has been reported previously with this three-marker haplotype (rs2111902–rs3918346 –rs3741775) in two ethnically different studies (Liu et al. 2004; Schumacher et al. 2004) replicating the original observation (Chumakov et al. 2002). Liu et al. (2004) found that, as in our study, individual polymorphisms were not independently associated with schizophrenia whereas the three marker haplotype was. If not a chance finding, DAAO gene variability might increase liability to psychosis via an effect on weaker phenotypes, namely self-reported negative and disorganization schizotypy. However, genome scans have not found significant linkage of chromosome12q24.11 (the region of DAAO) to schizophrenia. Moreover, recent case-control and family-based association studies on schizophrenia have produced mixed results, including negative results (Goldberg et al. 2006; Liu et al. 2006; Yamada et al. 2005). For NRG1, three of the four tested polymorphisms (SNP8NRG221533, SNP8NRG433E1006, SNP8NRG243177) were independently associated with measures of cognition, namely

BIOL PSYCHIATRY 2007;62:784 –792 789

N.C. Stefanis et al. Table 3. Nominally Significant Associations of Individual SNPs

Gene, SNP [Response Relative Direction/Genotypea] (p Value) NRG1b

Response IQ (RPM) SPQ, Total Score

DTNBP1

DAAO

rs760761 [2-1] (.041) rs760761 [-21] (.035) rs2619522 [2-1] (.024) rs760761 [-21] (.034) rs2619522 [2-1] (.034)

SPQ, Cognitive/Perceptual SPQ, Negative Factor SPQ, Disorganization Factor

rs3213207 [12-] (.036) rs760761 [-21] (.025) rs2619522 [2-1] (.018) rs760761 [-21] (.007) rs2619522 [2-1] (.005) rs1018381 [2-1] (.026)

SPQ, Paranoid Factor

PAS, Total Score CAPE, Total Score CAPE, Negative Factor CAPE, Positive Factor CAPE, Depression d’-CPT-IP (sustained attention) d’-S2B (spatial working memory) d’-V2B (verbal working memory) Anti-Saccade Error Rate

rs1018381 [2-1] (.043) rs3741775 [2-1] (.048) 221533 [21-] (.011) 433E1006 [1-2] (.048) 243177 [2-1] (.040) 433E1006 [1-2] (.044)

rs760761 [-21] (.029) rs2619522 [21-] (.031) rs3918346 [21-] (.033)

Empty cells refer to no associations with p ⬍ .05 in the individual single nucleotide polymorphism (SNP) analyses for the corresponding outcomes. The DAOA/G30 gene SNP genotypes were not significantly associated with any of the responses in these analyses. SPQ, Schizotypal Personality Questionnaire; PAS, Perceptual Aberration Scale CAPE, Community Assessment of Psychic Experiences. a The symbols in the brackets indicate the relative ordering (ranking) of the average value of the response variable in the genotypic groups, in the order they appear in Table 2: [2] ⫽ lowest average value, [-] ⫽ middle-ranking average value, [1] ⫽ highest average value. For example, for NRG1 rs221533 the average values of the sustained attention d’ index are lowest in the CC genotype group, highest in the CT genotype group, and in-between in the TT genotypic group: CC ⬍ TT ⬍ CT. The ranking does not inform on the magnitude of the differences in the average values. b Use prefix “SNP8NRG” for the NRG1 SNP numbers; the prefix was omitted owing to space limitations.

sustained attention capacity and verbal or spatial working memory performance but not with schizotypal features as in another study (Lin et al. 2005). These markers are all contained within the seven multiple marker haplotype that was originally associated with schizophrenia (Stefansson et al. 2002). The identified signals

for NRG1 associations were only marginally significant. Notably, the minor allele (T) of SNP8NRG243177 was associated with a reduction of spatial working memory capacity. This is of particular interest, because it has recently been reported that SNP8NRG243177 is a functional polymorphism, the risk allele (T)

Table 4. Effect Sizes (␤)/Minor Allele Copy for DTNBP1 Gene SNPs That Showed Statistically Significant Associations in Individual SNP ANOVA rs760761/A Allele Copy Response IQ (RPM) SPQ, Total Score SPQ, Cognitive/Perceptual Factor SPQ, Negative Factor SPQ, Disorganization Factor SPQ, Paranoid Factor PAS, Total Score CAPE, Total Score CAPE, Negative Factor CAPE, Positive Factor CAPE, Depression d’-CPT-IP (sustained attention) d’-S2B (spatial working memory) d’-V2B (verbal working memory) Anti-Saccade Error Rate

rs1018381/A Allele Copy

rs2619522/C Allele Copy

␤

pall-load (pANOVA)

␤

pall-load (pANOVA)

␤

pall-load (pANOVA)

⫺.79 ⫺1.68 ⫺.02 ⫺.01 ⫺.03 ⫺.03 ⫺.53 ⫺1.77 ⫺.41 ⫺.95 ⫺.32 ⫺.10 ⫺.03 ⫺.01 .01

.061 (.041) .015 (.035) .014 (.034) .266 (.522) .011 (.025) .005 (.007) .059 (.109) .124 (.305) .417 (.712) .035 (.102) .278 (.482) .019 (.029) .539 (.431) .798 (.326) .313 (.601)

⫺.97 ⫺1.86 ⫺.02 ⫺.01 ⫺.03 ⫺.04 ⫺.49 ⫺2.56 ⫺.36 ⫺1.63 ⫺.46 ⫺.01 .05 .04 .00

.101 (.226) .056 (.093) .044 (.105) .573 (.593) .036 (.060) .010 (.026) .224 (.273) .123 (.291) .621 (.875) .012 (.043) .287 (.435) .911 (.878) .458 (.496) .488 (.378) .978 (.948)

⫺.74 ⫺1.88 ⫺.02 ⫺.01 ⫺.03 ⫺.03 ⫺.57 ⫺1.65 ⫺.13 ⫺1.03 ⫺.36 ⫺.10 ⫺.02 ⫺.01 .01

.079 (.091) .006 (.024) .010 (.034) .219 (.432) .005 (.018) .001 (.005) .043 (.105) .151 (.289) .793 (.878) .023 (.075) .221 (.212) .017 (.031) .652 (.426) .849 (.499) .268 (.539)

SNP, single nucleotide polymorphism; ANOVA, analysis of variance; pall-load, p value for the allele-load model; pANOVA, ANOVA p value; other abbreviations as in Table 3.

www.sobp.org/journal

790 BIOL PSYCHIATRY 2007;62:784 –792

N.C. Stefanis et al.

Table 5. Associations With DTNBP1 Haplotype Load Rare (⬍ 5%) Response IQ (RPM) SPQ, Total Score SPQ, Cognitive/Perceptual Factor SPQ, Negative Factor SPQ, Disorganization Factor SPQ, Paranoid Factor PAS, Total Score CAPE, Total Score CAPE, Negative Factor CAPE, Positive Factor CAPE, Depression d’-CPT-IP (sustained attention) d’-S2B (spatial working memory) d’-V2B (verbal working memory) Anti-Saccade Error Rate

CCCACG

CTCGAG

CTTGAG

GTCACA

n

b

SE

b

SE

b

SE

b

SE

b

SE

LR p Value

1737 1237 1235 1235 1235 1235 1237 811 859 861 869 1872 1691 1617 1869

⫺3.432 2.359 .009 .028 .027 .043 ⫺1.169 10.562 5.263a 1.796 3.397a ⫺.138 ⫺.214 ⫺.260 ⫺.056

(2.382) (4.095) (.038) (.041) (.067) (.061) (1.708) (5.672) (2.593) (2.305) (1.515) (.253) (.279) (.244) (.058)

⫺.111 ⫺1.647 ⫺.016 ⫺.008 ⫺.028 ⫺.027a ⫺.525 ⫺.706 ⫺.004 ⫺.734 ⫺.100 ⫺.115a ⫺.051 ⫺.059 .010

(.551) (.920) (.008) (.009) (.015) (.014) (.384) (1.510) (.672) (.598) (.390) (.055) (.066) (.057) (.013)

.845a .382 .001 .006 .006 .003 .203 .237 .009 ⫺.195 .256 .074 .055 ⫺.003 ⫺.014

(.392) (.636) (.006) (.006) (.010) (.010) (.265) (1.097) (.491) (.432) (.285) (.039) (.046) (.040) (.009)

.710 .238 ⫺.003 .004 .001 ⫺.006 .157 .170 ⫺.068 ⫺.425 .192 .017 .047 ⫺.027 ⫺.011

(.597) (.970) (.009) (.010) (.016) (.014) (.405) (1.650) (.734) (.657) (.430) (.059) (.072) (.061) (.014)

⫺.495 ⫺1.826 ⫺.017 ⫺.006 ⫺.033a ⫺.036a ⫺.406 ⫺2.895 ⫺.525 ⫺1.843a ⫺.473 .007 .057 .063 ⫺.009

(.609) (1.006) (.009) (.010) (.016) (.015) (.420) (1.702) (.751) (.672) (.442) (.059) (.071) (.061) (.014)

.062 .12 .19 .61 .10 .052 .38 .22 .46 .11 .16 .039 .55 .56 .39

Shown are the coefficients of a regression model/copy of the corresponding haplotype. b, coefficient, expressing change in the response variable/ haplotype copy; SE, standard error of the coefficient b; LR p value, p value for a likelihood ratio test against a constant-only model; other abbreviations as in Table 3. a Significantly different (at the .01 ⬍ p ⬍ .05 level) from the reference haplotype GTCGAG (the most common in the population).

predicting higher levels of type IV NRG1 messenger RNA expression (Law et al. 2006) and associated with lower prefrontal (and temporal) activation and development of psychotic symptoms in high-risk individuals for schizophrenia (Hall et al. 2006). If not a chance finding, our result constitutes the first independent confirmation that functional SNP8NRG243177 impacts on aspects of human prefrontal brain function. Because spatial working deficits constitute an effective endophenotype for schizophrenia (Cannon et al. 2000), this finding also suggests a mechanism by which this NRG1 variant might confer risk for the disorder at an information processing level. In contrast to Hall et al. (2006), no association of SNP8NRG243177 with psychotic-like symptoms or IQ was detected in this study. Differences in population characteristics and in assessment instruments used might well account for these discrepancies. The identified signals for NRG1 associations did not persist in the haplotype-based analyses. Similar to DTNBP1, NRG1 associations in the literature have also highlighted different markers and haplotypes across studies. In contrast to DTNBP1, NRG1 haplotype frequencies might vary considerably across European populations (Gardner et al. 2006), making the replication of associations even more difficult. It is currently unknown whether population stratification might exist even within subgroups of the Greek population. Future studies should investigate whether our nominal associations with cognitive endophenotypes are replicated in other populations. Finally, no single polymorphism or haplotypes of DAOA/G32 were associated with any of the measured endophenotypes, except for a weak haplotype effect on spatial working memory. The significance of this is unknown, and recent emerging data do not as yet provide compelling support for the involvement of DAOA/G32 genetic variation on cognition (Goldberg et al. 2006; Ioannidis 2005). Given the large number of performed analyses, such an isolated marginal finding might easily be dismissed as being due to chance. Some caveats should be discussed. First, several of the identified signals might be false positives (Ioannidis et al. 2005; Wacholder et al. 2004). With 17 SNPs (excluding the HW equilibrium–violating SNP) and 15 outcome variables, there are www.sobp.org/journal

255 sets of analyses performed, even without consideration of haplotypes. Thus the interpretation of the modestly significant associations should be conservative. However, our approach was to target genes and variants that already had some indirect or direct support for involvement in the pathogenesis of schizophrenia and therefore the pre-study probability of significant associations was not negligible as in a hypothesis-free, discovery-oriented approach. Moreover, simple adjustment for multiple comparisons (e.g., Bonferroni adjustment of p values by 255fold) might be overly conservative, given the specific-hypothesis approach of our study and the strong LD between many of these markers and between interrelated endophenotypes. A confirmatory factor analysis approach that reduces schizotypy subscales to few latent factors was adopted a priori in this study. Further factor analyses on neurocognitive variables could potentially further reduce multiple comparison testing but at the expense of masking potential association to specific cognitive tasks, each representing a well recognized endophenotypic marker for schizophrenia. Second, it is possible that in haplotype analyses the power of our study might have been eroded by the cumulative impact of missing information across the constituent SNPs. Therefore, the inability of haplotype analyses to give further insights, with the exception of DAAO, might reflect mostly this loss of power. However, it is unlikely that major effects have been missed for any of these genes. Third, one should be cautious when extrapolating from these endophenotypes to schizophrenia. Some of these quantitative traits (e.g., neurocognitive ones) might be as pertinent or more pertinent for other diseases (e.g., cognitive syndromes and deficits). Overall, none of these four candidate genes for schizophrenia seems to strongly modulate the population variability of schizotypal features and cognitive ability, but isolated signals of association are plausible, in particular for DTNBP1 and NRG1. Risk genotypes might have quantitative rather than categorical effects and might influence milder or subclinical phenotypes (Fanous and Kendler 2005). Pleiotropic effects and phenotypespecificity need to be carefully replicated in additional large studies (Ioannidis et al. 2003) before any strong claims can be made. Despite the fact that our study is the largest conducted to

N.C. Stefanis et al. date on genetic associations with schizotypy and cognitive endophenotypes, even larger studies should be encouraged, because power is limited, given the small magnitude of effect sizes (Ioannidis et al. 2006). Finally, effects of these genetic variants are likely to be subtle when seen in isolation. However, cumulatively they might have an important impact on the genetic background of these phenotypes. This work was supported by the Grant EKBAN 97 to NCS from the General Secretariat of Research and Technology of the Greek Ministry of Development. Intrasoft provided the technical support for this project. Supplementary material cited in this article is available online. Addington AM, Gornick M, Sporn AL, Gogtay N, Greenstein D, Lenane M, et al. (2004): Polymorphisms in the 13q33.2 gene G72/G30 are associated with childhood-onset schizophrenia and psychosis not otherwise specified. Biol Psychiatry 55:976 –980. Avramopoulos D, Stefanis NC, Hantoumi I, Smyrnis N, Evdokimidis I, Stefanis CN (2002): Higher scores of self reported schizotypy in healthy young males carrying the COMT high activity allele. Mol Psychiatry 7:706 –711. Barrett JC, Fry B, Maller J, Daly MJ (2005): Haploview: Analysis and visualization of LD and haplotype maps. Bioinformatics 21:263–265. Burdick KE, Lencz T, Funke B, Finn CT, Szeszko PR, Kane JM, et al. (2006): Genetic variation in DTNBP1 influences general cognitive ability. Hum Mol Genet 15:1563–1568. Cannon TD, Huttunen MO, Lonnqvist J, Tuulio-Henriksson A, Pirkola T, Glahn D, et al. (2000): The inheritance of neuropsychological dysfunction in twins discordant for schizophrenia. Am J Hum Genet 67:369 –382. Chapman LJ, Chapman JP, Raulin ML (1978): Body-image aberration in schizophrenia. J Abnorm Psychol 87:399 – 407. Chumakov I, Blumenfeld M, Guerassimenko O, Cavarec L, Palicio M, Abderrahim H, et al. (2002): Genetic and physiological data implicating the new human gene G72 and the gene for D-amino acid oxidase in schizophrenia. Proc Natl Acad Sci U S A 99:13675–13680. Cloninger CR, Svrakic DM, Przybeck TR (1993): A psychobiological model of temperament and character. Arch Gen Psychiatry 50:975–990. Cornblatt BA, Malhotra AK (2001): Impaired attention as an endophenotype for molecular genetic studies of schizophrenia. Am J Med Genet 105:11–15. Cornblatt BA, Risch NJ, Faris G, Friedman D, Erlenmeyer-Kimling L (1988): The Continuous Performance Test, identical pairs version (CPT-IP): I. New findings about sustained attention in normal families. Psychiatry Res 26:223–238. de Vries HG, Collee JM, van Veldhuizen MH, Achterhof L, Smit Sibinga CT, Scheffer H, et al. (1996): Validation of the determination of deltaF508 mutations of the cystic fibrosis gene in over 11 000 mouthwashes. Hum Genet 97:334 –336. DeRosse P, Funke B, Burdick KE, Lencz T, Ekholm JM, Kane JM, et al. (2006): Dysbindin genotype and negative symptoms in schizophrenia. Am J Psychiatry 163:532–534. Ettinger U, Picchioni M, Hall MH, Schulze K, Toulopoulou T, Landau S, et al. (2006): Antisaccade performance in monozygotic twins discordant for schizophrenia: The Maudsley twin study. Am J Psychiatry 163:543–545. Fanous AH, Kendler KS (2005): Genetic heterogeneity, modifier genes, and quantitative phenotypes in psychiatric illness: Searching for a framework. Mol Psychiatry 10:6 –13. Fanous AH, van den Oord EJ, Riley BP, Aggen SH, Neale MC, O’Neill FA, et al. (2005): Relationship between a high-risk haplotype in the DTNBP1 (dysbindin) gene and clinical features of schizophrenia. Am J Psychiatry 162: 1824 –1832. Funke B, Finn CT, Plocik AM, Lake S, DeRosse P, Kane JM, et al. (2004): Association of the DTNBP1 locus with schizophrenia in a U.S. population. Am J Hum Genet 75:891– 898. Gardner M, Gonzalez-Neira A, Lao O, Calafell F, Bertranpetit J, Comas D (2006): Extreme population differences across Neuregulin 1 gene, with implications for association studies. Mol Psychiatry 11:66 –75. Gevins A, Smith ME, Le J, Leong H, Bennett J, Martin N, et al. (1996): High resolution evoked potential imaging of the cortical dynamics of human working memory. Electroencephalogr Clin Neurophysiol 98:327–348.

BIOL PSYCHIATRY 2007;62:784 –792 791 Goldberg TE, Straub RE, Callicott JH, Hariri A, Mattay VS, Bigelow L, et al. (2006): The G72/G30 gene complex and cognitive abnormalities in schizophrenia. Neuropsychopharmacology 31:2022–2032. Gottesman II, Gould TD (2003): The endophenotype concept in psychiatry: Etymology and strategic intentions. Am J Psychiatry 160:636 – 645. Hall D, Gogos JA, Karayiorgou M (2004): The contribution of three strong candidate schizophrenia susceptibility genes in demographically distinct populations. Genes Brain Behav 3:240 –248. Hall J, Whalley HC, Job DE, Baig BJ, McIntosh AM, Evans KL, et al. (2006): A neuregulin 1 variant associated with abnormal cortical function and psychotic symptoms. Nat Neurosci 9:1477–1478. Hallmayer JF, Kalaydjieva L, Badcock J, Dragovic M, Howell S, Michie PT, et al. (2005): Genetic evidence for a distinct subtype of schizophrenia characterized by pervasive cognitive deficit. Am J Hum Genet 77:468 – 476. Ioannidis JP (2005): Why most published research findings are false. PLoS Med 2:e124. Ioannidis JP, Trikalinos TA, Khoury MJ (2006): Implications of small effect sizes of individual genetic variants on the design and interpretation of genetic association studies of complex diseases. Am J Epidemiol 164: 609 – 614. Ioannidis JP, Trikalinos TA, Ntzani EE, Contopoulos-Ioannidis DG (2003): Genetic associations in large versus small studies: An empirical assessment. Lancet 361:567–571. Johns LC, van Os J (2001): The continuity of psychotic experiences in the general population. Clin Psychol Rev 21:1125–1141. Kendler KS, McGuire M, Gruenberg AM, Walsh D (1995): Schizotypal symptoms and signs in the Roscommon Family Study. Their factor structure and familial relationship with psychotic and affective disorders. Arch Gen Psychiatry 52:296 –303. Korostishevsky M, Kaganovich M, Cholostoy A, Ashkenazi M, Ratner Y, Dahary D, et al. (2004): Is the G72/G30 locus associated with schizophrenia? Single nucleotide polymorphisms, haplotypes, and gene expression analysis. Biol Psychiatry 56:169 –176. Law AJ, Lipska BK, Weickert CS, Hyde TM, Straub RE, Hashimoto R, et al. (2006): Neuregulin 1 transcripts are differentially expressed in schizophrenia and regulated by 5’ SNPs associated with the disease. Proc Natl Acad Sci U S A 103:6747– 6752. Lewis CM, Levinson DF, Wise LH, DeLisi LE, Straub RE, Hovatta I, et al. (2003): Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: Schizophrenia. Am J Hum Genet 73:34 – 48. Lin CC, Su CH, Kuo PH, Hsiao CK, Soong WT, Chen WJ (2006): Genetic and environmental influences on schizotypy among adolescents in Taiwan: A multivariate twin/sibling analysis. Behav Genet [Epub ahead of print]. Lin HF, Liu YL, Liu CM, Hung SI, Hwu HG, Chen WJ (2005): Neuregulin 1 gene and variations in perceptual aberration of schizotypal personality in adolescents. Psychol Med 35:1589 –1598. Linney YM, Murray RM, Peters ER, MacDonald AM, Rijsdijk F, Sham PC (2003): A quantitative genetic analysis of schizotypal personality traits. Psychol Med 33:803– 816. Liu X, He G, Wang X, Chen Q, Qian X, Lin W, et al. (2004): Association of DAAO with schizophrenia in the Chinese population. Neurosci Lett 369:228 – 233. Liu YL, Fann CS, Liu CM, Chang CC, Wu JY, Hung SI, et al. (2006): No association of G72 and D-amino acid oxidase genes with schizophrenia. Schizophr Res 87:15–20. Mutsuddi M, Morris DW, Waggoner SG, Daly MJ, Scolnick EM, Sklar P (2006): Analysis of high-resolution HapMap of DTNBP1 (Dysbindin) suggests no consistency between reported common variant associations and schizophrenia. Am J Hum Genet 79:903–909. Raine A (1991): The SPQ: A scale for the assessment of schizotypal personality based on DSM-III-R criteria. Schizophr Bull 17:555–564. Raven J (1982): Revised Manual for Raven’s Progressive Matrices and Vocabulary Scales. Windsor, United Kingdom: NFER-Nelson. Schumacher J, Jamra RA, Freudenberg J, Becker T, Ohlraun S, Otte AC, et al. (2004): Examination of G72 and D-amino-acid oxidase as genetic risk factors for schizophrenia and bipolar affective disorder. Mol Psychiatry 9:203–207. Schwab SG, Knapp M, Mondabon S, Hallmayer J, Borrmann-Hassenbach M, Albus M, et al. (2003): Support for association of schizophrenia with genetic variation in the 6p22.3 gene, dysbindin, in sib-pair families with linkage and in an additional sample of triad families. Am J Hum Genet 72:185–190.

www.sobp.org/journal

792 BIOL PSYCHIATRY 2007;62:784 –792 Smyrnis N, Evdokimidis I, Stefanis NC, Avramopoulos D, Constantinidis TS, Stavropoulos A, et al. (2003): Antisaccade performance of 1,273 men: Effects of schizotypy, anxiety, and depression. J Abnorm Psychol 112: 403– 414. Stefanis NC, Hanssen M, Smirnis NK, Avramopoulos DA, Evdokimidis IK, Stefanis CN, et al. (2002): Evidence that three dimensions of psychosis have a distribution in the general population. Psychol Med 32:347–358. Stefanis NC, Smyrnis N, Avramopoulos D, Evdokimidis I, Ntzoufras I, Stefanis CN (2004): Factorial composition of self-rated schizotypal traits among young males undergoing military training. Schizophr Bull 30:335–350. Stefansson H, Sigurdsson E, Steinthorsdottir V, Bjornsdottir S, Sigmundsson T, Ghosh S, et al. (2002): Neuregulin 1 and susceptibility to schizophrenia. Am J Hum Genet 71:877– 892. Stephens M, Scheet P (2005): Accounting for decay of linkage disequilibrium in haplotype inference and missing-data imputation. Am J Hum Genet 76:449 – 462. Stephens M, Smith NJ, Donnelly P (2001): A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68: 978 –989.

www.sobp.org/journal

N.C. Stefanis et al. Straub RE, Jiang Y, MacLean CJ, Ma Y, Webb BT, Myakishev MV, et al. (2002): Genetic variation in the 6p22.3 gene DTNBP1, the human ortholog of the mouse dysbindin gene, is associated with schizophrenia. Am J Hum Genet 71:337–348. Tienari P, Wynne LC, Laksy K, Moring J, Nieminen P, Sorri A, et al. (2003): Genetic boundaries of the schizophrenia spectrum: Evidence from the Finnish Adoptive Family Study of Schizophrenia. Am J Psychiatry 160: 1587–1594. van den Oord EJ, Sullivan PF, Jiang Y, Walsh D, O’Neill FA, Kendler KS, et al. (2003): Identification of a high-risk haplotype for the dystrobrevin binding protein 1 (DTNBP1) gene in the Irish study of high-density schizophrenia families. Mol Psychiatry 8:499 –510. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N (2004): Assessing the probability that a positive report is false: An approach for molecular epidemiology studies. J Natl Cancer Inst 96:434 – 442. Yamada K, Ohnishi T, Hashimoto K, Ohba H, Iwayama-Shigeno Y, Toyoshima M, et al. (2005): Identification of multiple serine racemase (SRR) mRNA isoforms and genetic analyses of SRR and DAO in schizophrenia and D-serine levels. Biol Psychiatry 57:1493–1503.