Selected summaries from the XVI World Congress of Psychiatric Genetics, Osaka, Japan, 11–15 October 2008

Abstract

The XVI World Congress of Psychiatric Genetics, sponsored by the International Society of Psychiatric Genetics took place in Osaka, Japan, October 2008. Approximately 600 participants gathered to discuss the latest molecular genetic findings relevant to serious mental illnesses, including schizophrenia, bipolar disorder, major depression, alcohol and drug abuse, autism, and attention-deficit disorder. Recently, the field has advanced considerably and includes new genome-wide association studies with the largest numbers of individuals screened and density of markers to date, as well as newly uncovered genetic phenomena, such as copy number variation that may prove to be relevant for specific brain disorders. The following report represents some of the areas covered during this conference and some of the major new findings presented.

Introduction

The World Congress of Psychiatric Genetics (WCPG) grew out of a need in the late 1980s for bringing together investigators in the new field of molecular genetics to discuss the application of accelerating new technology to the genetics of psychiatric disorders. Up until that time, the only studies being performed were of an epidemiological nature (i.e. family, twin and adoption studies) that showed a clear genetic predisposition for illnesses, such as schizophrenia, bipolar disorder and later substance abuse, autism, and other childhood conditions. In 1992, the International Society of Psychiatric Genetics (ISPG) was formed to provide a stable structure for continual congresses in this field with a board to oversee a rotating congress chairperson and program committee. The ISPG has grown from 200 charter members to well over 1000 international participants in the society and WCPG. Information about the society can be found on its website (www.ISPG.net). One of its activities is to establish an annual set of awards. The oldest award that began at the first WCPG in 1993 is the ‘Lifetime Achievement Award in Psychiatric Genetics’. In 2008, this award was given to Jurg Ott from The Rockefeller Institute, New York and Beijing University, China. Professor Ott was a pioneer in developing and applying statistical methods for linkage analyses to psychiatric disorders. In addition, to honor the first president of the ISPG, Theodore Reich, in 2004 the ‘Reich Junior Investigator Award’ was established to acknowledge the accomplishments and potential achievements of a promising young investigator in the field. In 2008, this award went to Shaun Purcell of the Broad Institute in Boston for his analytic work with the new genome-wide association studies (GWAS) and other data. Finally, each year, with the continued support of the National Institutes of Mental Health, Drug Abuse and Alcohol Abuse in the USA, young investigator travel awards have enabled students to participate in the congress. The following reports have been written by some of those students who volunteered to devote their time to summarize some of the important sessions they attended. Although not all of the 2008 congress is represented, a more complete outline can be found on the websites (www.ISPG.net or www.wcpg2008.umin.jp). A similar, but more complete, account of the 2007 congress was published approximately one year ago (Alkelai et al., 2008).

New analytical methods for genome-wide association studies

Reported by Inti Pedroso

Recently, there has been a great interest in the new generation of GWAS with large numbers of patients being screened using high marker density, particularly in illnesses with a non-Mendelian mode of inheritance. The latest GWAS results were presented at this meeting for both schizophrenia and bipolar disorder and are currently in various stages of publication. Genes being uncovered by GWAS frequently do not fall in the list of the ‘usual suspects’ or candidates genes (Burmeister et al., 2008; McCarthy et al., 2008). Some examples of this are the links between the complement system/type I diabetes and autophagy/Crohn’s disease revealed by GWAS (Massey and Parkes, 2007; Parkes et al., 2007; Wellcome Trust Case Control Consortium, 2007). In addition, the effect sizes of common risk variants found by GWAS have been much lower than anticipated, and have often needed large meta-analyses to reach genome-wide significance (Barrett et al., 2008; McCarthy et al., 2008). Combining very large GWAS samples poses additional problems because of factors such as subtle ascertainment and environmental exposure differences across centers (Psychiatric GWAS Consortium Steering Committee, 2009).

Two oral presentations at this congress described possible solutions to some of these issues. Both presentations focused on new methods for carrying out gene-set and network analysis in GWAS data. Gene-sets are groups of genes related by some common biological criteria such as co-expression, a metabolic pathway, genomic position, functional domains, or subcellular localization. Gene-set analyses are commonly used in gene expression microarray analysis (Subramanian et al., 2005; Nam and Kim, 2008).

Dr A Corvin (Dublin, Ireland) presented a method to carry out gene-set enrichment for GWAS. This method counts the number of nominally associated single nucleotide polymorphisms (SNPs) in an assigned pathway, producing a specific pathway score. The significance of this score is assessed using empirical permutations. Dr Corvin showed results from a reanalysis of a Parkinson’s disease dataset to illustrate how this method may be a promising alternative to integrate nominally associated variants in the analysis of GWAS (Holden et al., 2008).

When contrasted with gene-set analysis, network analysis does not rely on known biological pathways, but, instead uses information about the functional interaction between genes (e.g. protein–protein interactions and transcriptional regulation) to analyze data, allowing the use of relationships not represented in defined gene-sets during the discovery of the biological basis of the disease. Dr Pedroso (London, United Kingdom) presented a new gene-wide method, which uses permutations to estimate the significance of the best P value in the gene, and provides a list of genes having a P value better than what it is expected by chance. He used the Wellcome Trust Case–Control Consortium Bipolar Data set (Wellcome Trust Case Control Consortium, 2007) to show that genes with a gene-wide P value of less than 0.05 can be used for gene-set as well as network analysis. Significant genes showed enrichment in relevant categories such as neurological disease (P=1.6e–⁷), psychological disorders (P=6.95e–⁶), transmission of nerve impulse (P=1.3e–⁵), neurogenesis (P=6.95e–⁶); all significant after multiple testing correction.

He also showed that, using a network approach, a number of modular networks were uncovered, including γ-aminobutyric acid (GABA) and glutamate receptors and the calcium signaling pathway. These results suggest that new insights into the biology of neuropsychiatry disorders may come from analytical methods that will allow mining the ‘higher hanging fruit’ in WGAS data. Rapid development on this area is already in progress and the road for translational research may be facilitated by the identification of components in the pathways that are potential targets for new drug development.

Copy number variation in schizophrenia

Reported by Sarah E. Bergen

One major highlight of this congress was a new focus on structural variants in the genome, copy number variations (CNVs) that could contribute to disease. Jonathan Sebat from Cold Spring Harbor Laboratories, a pioneer in uncovering these variants, gave a plenary lecture reviewing the findings in autism and schizophrenia (reviewed in DeLisi, 2009).

One oral session on CNVs began with a presentation by Dr David St Clair (University of Aberdeen, UK) on large recurrent deletions associated with schizophrenia. He and his colleagues initially used a population-based Caucasian sample involving 9878 transmissions from parents to offspring and identified 66 de-novo CNVs. A two-stage design was implemented to explore their potential relationship to schizophrenia. In phase I, novel CNVs were assayed in a Caucasian sample of 1433 schizophrenia cases and 33 250 controls from the SGENE consortium. Three deletions at 1q21.1 (1.38 Mb), 15q11.2 (450 kb), and 15q13.3 (1.57 Mb) demonstrated nominal association with schizophrenia and were followed up in phase II using a sample of 3285 cases and 7951 controls from Europe and China. In the combined sample, all three deletions were significantly associated with schizophrenia with the larger deletions conferring the greatest risk [1q21.1: odds ratio (OR)=14.83, P=2.9× 10⁻⁵ and 15q13.3: OR=11.54, P=5.3× 10⁻⁴ vs. 15q11.2: OR=2.73, P=6.0×10⁻⁴]. These deletions also show evidence for negative selective pressure and are flanked by low copy repeat sequences, known to facilitate loss or gain of genomic segments through non-allelic homologous recombination.

Dr Shi YongYong (Shanghai University, China) also presented a study of rare structural variants in schizophrenia, from a Chinese Han Population. He began his talk by highlighting the difficulties encountered by schizophrenia GWAS in identifying common susceptibility variants, while also stating optimism that rare CNVs can be detected using microarray SNP data. In his study, CNVs were assessed in a small group of 155 schizophrenia cases and 187 controls genotyped on Affymetrix 500 K chips. There were no differences in the proportions of cases (70.3%) and controls (70.6%) with CNVs larger than 100 kb, nor were differences apparent when comparing only rare CNVs with a frequency of less than 1% (cases 21.3%, controls 23.5%) or rare CNVs impacting genes (cases 14.2%, controls 17.6%). Restricting these analyses to the 36 individuals with an age of onset below 19 years also revealed no significant differences. Next, he assessed the genes impacted by rare CNVs in terms of their prior association with schizophrenia as reported by the Schizophrenia Gene Database (SZGene; Allen et al., 2008). In cases, five of the 96 genes impacted by rare CNVs were in SZGene (5.2%) compared with eight of 399 genes in controls (2.0%) – a nonsignificant difference. Three of the five schizophrenia susceptibility genes impacted by rare CNVs in cases involved deletions, whereas only one of the eight in controls, suggesting that deletions may be more deleterious than duplications. Although no significant relationships between CNVs and schizophrenia were detected in this study, it was only an initial examination that will be followed by similar studies with much larger sample sizes.

Dr George Kirov (Cardiff University, UK) reported support for the involvement of rare CNVs (<1%) in schizophrenia in 471 affected Caucasian cases from Bulgaria compared with 2792 controls from the Wellcome Trust Case Control Consortium. All were genotyped using the Affymetrix 500K Mapping Array (Santa Clara, California, USA) consisting of two chips, NSP and STY . Analyses were restricted to CNVs with a minimum size of 1Mb, 10 or more SNPs within the CNV, and detection on both arrays. The 22q11.2 (velocardiofacial syndrome) deletion, known to confer substantial risk for schizophrenia, and a deletion at 17p12 were each observed in two cases but not in controls. The latter deletion has also been found to confer susceptibility to hereditary neuropathy with liability to pressure palsies and has also been found in two additional schizophrenia studies. The deletion was found in six of 4618 (0.13%) cases and six of 36 092 (0.017%) controls yielding an OR of 7.82 (P=0.001). The locus of Prader-Willi/Angelman syndrome at 15q11.2 was also significantly associated with schizophrenia in this study (P=0.026). A duplication on 16p13.1 that was previously associated with autism was enriched in cases (0.6 vs. 2% in controls) but not significantly (OR = 2.98, P=0.13).

Dr Jennifer Stone (Broad Institute, Boston, USA) presented data from the International Schizophrenia Consortium, which has ascertained 3391 cases and 3181 controls from multiple sites throughout Europe. Genotyping of these participants was performed on Affymetrix 5.0 and 6.0 arrays, and CNV detection was done using Birdseye software, which implements a hidden Markov model using intensity data from SNP and CNV probes. Restricting analyses to CNVs with less than 1% frequency and minimum of 100 kb resulted in an inclusion of 6753 CNVs of which 39% were deletions. PLINK v1.03 was used to perform statistical analyses in two ways: assessing risk for schizophrenia based on (i) the total number of CNVs in an individual and (ii) the number of genes intersected at specific loci. They found that cases had a 1.15-fold higher number of CNVs than controls (P=2×10⁻⁵), and a 1.14-fold increase in genes intersected by CNVs (P=2× 10⁻⁶). When very large CNV deletions (>500 kb) were analyzed, the well-known velocardiofacial syndrome deletion at 22q11.2 (OR=21.6, empirical P=0.0017), and two additional loci at 15q13.3 (OR=17.9, empirical P=0.0029) and 1q21.1 (OR=6.6, empirical P=0.0076) were also identified as schizophrenia-related deletions.

Dr David Collier (King’s College, London, UK) presented data showing multiple deletions affecting exons of the neurexin 1 gene (NRXN1) associated with schizophrenia. As previous reports already linked NRXN1 with autism and schizophrenia (Kim et al., 2008; Rujescu et al., 2008), Dr Collier and his group examined CNVs in this gene and the closely related NRXN2 and NRXN3 genes as candidates. Participants were selected from seven European countries, which included 2977 cases with schizophrenia and 33 746 controls. No CNVs were detected in NRXN2 or NRXN3, but 66 deletions (one de novo) and five duplications were found in NRXN1. In schizophrenia cases, there were 12 deletions and two duplications (0.47%), whereas in controls the corresponding values were 49 and 3 (0.15%; P=0.13; OR=1.73). In a secondary analysis restricted to CNVs affecting exons, significant association was found with a high OR (P=0.0027; OR=8.97).

In summary, the focus on CNVs in this congress was novel and gave conference attendees a preview of more extensive work to come in future years. CNVs were detected in approximately 2–4% of cases of schizophrenia so far, but this will likely increase with the identification of novel loci and development of higher-resolution platforms (St Clair, 2008). They have also been identified in other neuropsychiatric disorders, such as autism and these data were presented at the 2007 congress and appear in the literature.

microRNAs in human brain development and psychiatric disorders

Reported by Ravinesh A. Kumar

Human brain development is orchestrated by a complex set of genetic factors that include a recently discovered class of small RNA molecules called microRNA (miRNAs). miRNAs are short (approximately 19–25 nucleotides), single-stranded noncoding RNAs that function in posttranscriptional gene regulation by base pairing with target mRNAs, which can lead to mRNA cleavage and translation repression. At this year’s congress, four researchers presented their unpublished work on miRNAs and highlighted key developments in miRNA neurobiology, including the importance of miRNAs in the complex etiology of neuropsychiatric disorders.

In the first symposium, Dr Ray Kelleher of Harvard Medical School presented his work on ‘The functional roles of microRNAs in the developing and adult brains’. Dr Kelleher’s previous work showed an essential role for translational regulation in the establishment of long-lasting synaptic plasticity and memory. Their findings further identified the ERK and mTOR cascades as the major signaling pathways that couple synaptic activity to the dendritic translational machinery. To investigate the role of translational control by miRNAs in developing and adult brains, Dr Kelleher has recently used a conditional genetic approach in mice. His work shows that miRNA expression in neural stem cells is essential for normal brain development, whereas miRNA expression in excitatory neurons in the adult cerebral cortex is required for the maintenance of normal neuronal physiology and behavior.

In the next symposium on ‘Noncoding small RNAs in neurogenesis and cognitive disorders’, Dr Xinyu Zhao of the University of New Mexico School of Medicine highlights the importance of epigenetic regulation in childhood neurodevelopment disorders. Dr Zhao’s lab is focused on the function of two methylated-CpG binding proteins, MECP2 and MBD1, in neurodevelopment. Mutation of MECP2 results in Rett syndrome and Dr Zhao’s group found that Mecp2 deficiency leads to impaired neuronal maturation. They showed that mutation in MBD1 results in reduced neurogenesis and autism-like behavior in mice. To determine whether noncoding small RNAs epigenetically regulated by MECP2 and MBD1 are critical for brain functions, Zhao and colleagues used both miRNA profiling and high-throughput sequencing. They identified several noncoding small RNAs with altered expression in neural stem cells and brain tissues derived from MECP2 and MBD1 knockout mice. Using both cultured primary neural stem cells derived from adult brains and retrovirus-targeting of endogenous neural stem cells in the adult brains, they demonstrated that a few of these miRNAs could regulate stem cell proliferation and differentiation both in vitro and in vivo. These results show that crosstalk between noncoding RNAs and epigenetic regulation contributes to the modulation of postnatal neurogenesis and possibly pathogenesis of autism spectrum disorders.

Dr Chunyu Liu of The University of Chicago presented ‘Variants of human brain expressed miRNAs and disease association’. Dr Liu’s group resequenced nearly 300 brain expressed miRNA genes in 266 participants with bipolar disorder, schizophrenia and unaffected controls and identified nearly 100 novel variants in miRNA precursor genes. By using a case–control association design, the novel variants were examined for disease association in bipolar disorder and schizophrenia. Nominally significant associations were observed for bipolar disorder. Further investigations of common and rare variant associations of the newly identified miRNA variants in psychiatric diseases are ongoing.

Finally, Dr Murray Cairns of The University of Newcastle in Australia presented his findings on ‘miRNA and post-transcriptional regulation in schizophrenia’. On the basis of the observation that schizophrenia is associated with widespread alterations in cortical gene activity, Dr Cairn’s group tested the hypothesis that some of these alterations in gene expression could be caused by a schizophrenia-associated disturbance in posttranscriptional gene regulation. Towards this end, Dr Cairn’s laboratory has examined miRNA expression in postmortem cerebral cortex and has identified changes in miRNA biogenesis that could have implications for the dysregulation of many target genes through posttranscriptional gene silencing. The data emerging also suggests that this influence could have significance in the development of schizophrenia.

Together, the findings presented at the symposium on ‘MicroRNA in human brain and neuropsychiatric diseases’ emphasize the growing need to explore the role of miRNAs in psychiatric genetics. Further research in this area should also include the investigation of other small noncoding RNA molecules such as small interfering RNAs and piwi-interacting RNAs that will illuminate our understanding of the genetic basis of mammalian brain development and the molecular disturbances underlying mental illness and related phenotypes.

The genetics of substance abuse

Reported by Jingchun Chen and Lorna Houlihan

Dr Laura Bierut (Washington University, St. Louis, USA) spoke about nicotine addiction as a model to investigate genetic risk factors in a plenary lecture. A GWAS and a candidate gene analysis were performed in nicotine-dependent individuals, as defined by a Fagerström test for nicotine dependence ≥4, and controls were not dependent (Fagerström test for nicotine dependence=0). Association of nicotine dependency was detected with the α5 nicotinic receptor gene cluster on chromosome 15q25 (Saccone et al., 2007). Of the many SNPs that were associated across this region, one was a nonsynonymous SNP(rs16969968) in the α5 nicotinic cholinergic receptor (CHRNA5). This SNP was highly conserved across nonhuman species and the amino acid change alters receptor function, although she did not describe any difference in the expression levels in brain tissue. However, the allele frequency distribution varies worldwide, from 0% in African populations to 37% in European populations (Bierut et al., 2008). Meanwhile, three separate lung cancer and smoking GWAS have also found convergent findings on a chromosome 15q25 locus (Amos et al., 2008; Hung et al., 2008; Thorgeirsson et al., 2008). Future studies will determine whether independent causative variants exist at this locus. Ongoing work is focused on other addictions, as part of the genes and environment initiative, which has performed a large-scale whole genome scan, on over 4000 participants with alcohol, nicotine, or cocaine dependency. Initial results showing suggestive peaks await replication.

Dr George Uhl (NIDA, NIH, USA), in another plenary talk, spoke about other genes involved in the complex genetics for addiction by performing cluster-based analyses on 2343 pieces of evidence from peer-reviewed publications between 1976 and 2006 to link genes and chromosome regions with addiction through single-gene strategies, microrray, proteomics, or genetic studies. His results suggested an overrepresentation of genes related to cell adhesion molecules (CAM) (Li et al., 2008). Fine-mapping and mechanistic studies also have associated isoforms of the CAM NRXN3 with nicotine dependence (Hishimoto et al., 2007) and other CAM with the ability to quit smoking (Uhl et al., 2008a, 2008b, 2008c).

Additional presentations on substance abuse focused on methamphetamine dependence, a growing problem in many regions of the world and a longstanding concern in Korea, Taiwan, and Japan. The first speaker in this session was Dr Kim from Catholic University of Korea, who extended his studies of the dopamine receptor D1 (DRD1) gene and alcoholism to other substance abuse such as methamphetamine dependence. To examine the genetic effects of the dopamine receptor D (DRD) gene family (DRD1–DRD5) in the Korean population, nine polymorphisms in the DRD gene family were genotyped in 275 methamphetamine-dependent individuals and 273 controls. Although none of the polymorphisms of DRD1–DRD5 genes were found to be associated with the risk of methamphetamine dependence, the C allele of DRD2 (+32806 C>T) and a related haplotype (block2-ht1) was associated with anxiety and mood disorder symptoms in methamphetamine-dependent individuals. He also presented data on a ghrelin precursor gene polymorphism and methamphetamine dependence in the Korean population. Ghrelin is a recently isolated brain-gut peptide that increases growth hormone levels and stimulates appetite. Several recent studies have suggested that ghrelin plays a major role in the pathophysiology of drug-seeking behavior and anxiety. Dr Kim genotyped a ghrelin precursor polymorphism in 118 patients with methamphetamine dependence and 144 healthy controls. No significant difference was found in the genotypic and allelic distributions of the ghrelin precursor polymorphism. However, Met72 carriers were more depressed and anxious than patients with the wild-type Leu72 allele. He also mentioned that in two of his studies, plasma brain-derived neurotrophic factor (BDNF) concentrations were significantly higher in methamphetamine users compared with controls, and that plasma BDNF levels in smokers significantly increased from baseline after 2 months of smoking cessation. These findings suggest that BDNF may play some role in the brain response to the neurotoxicity of addictive substances.

The next two presentations, delivered by Dr Lin from Taiwan and Dr Ujike from Japan, pertained to psychiatric disorders co-occurring with methamphetamine abuse. In Dr Lin’s studies, 445 methamphetamine abusers and 436 normal controls were interviewed with the Diagnostic Interview for Genetic Studies and the Family Interview for Genetic Studies. He noted that relatives of methamphetamine users with a history of methamphetamine psychosis had a significantly higher morbid risk for schizophrenia than the relatives of those probands who never became psychotic. He also claimed that the morbid risk for schizophrenia in the relatives of the patients with a prolonged methamphetamine psychosis was higher than in the relatives of those users with a brief methamphetamine psychosis. By analyzing several polymorphisms in genes encoding proteins of the dopamine system and the GABA system, they found an excess of the high-activity Val158 COMT allele in methamphetamine abusers, and significant interaction between polymorphisms in the catechol-O-methyltransferase and DRD4 genes. They concluded that genetic variation in the dopamine system might encode an additive effect on the risk of becoming a methamphetamine abuser. They also provided preliminary evidence that DRD4 genes play roles in methamphetamine abuse and that GABA subunit genes in 5q33 may preferentially contribute to methamphetamine-use disorder in females. These results also suggested that the Ala/Val polymorphism of the superoxide dismutase 2 gene could be associated with the risk of developing methamphetamine psychosis.

Dr Ujike presented a genetic association study of several schizophrenia-related genes with methamphetamine psychosis cases (N=197) and age-matched and sex-matched controls (N=243) based on similarities between the clinical symptoms of both disorders. In a case–control association analysis of the frizzled 3 gene, a gene that was previously found to be significantly associated with schizophrenia, haplotypes were also found to be associated with methamphetamine psychosis (P<0.00001). Having the G-A-T-G or A-G-C-A haplotype of rs2241802-rs2323019-rs352203-rs880481 was a potent negative risk factor (ORs were 0.13 and 0.086, respectively). These findings indicate that genetic variants of the frizzled 3 gene may affect susceptibility to two analogous but distinct dopamine-related psychoses, endogenous and substance-induced psychosis. In addition, Dr Ujike showed that the dysbindin (DTNBP1) gene, another gene that has been repeatedly shown to be associated with schizophrenia, has significant associations with methamphetamine psychosis at polymorphisms P1635 (rs3213207, P=0.00003) and SNPA (rs2619538, P=0.049) and the three-locus haplotype of P1655 (rs2619539)-P1635-SNPA (permutation P=0.0005). The C-A-A haplotype, which was identical to a protective haplotype for schizophrenia and psychosis as previously reported, was also a protective factor (P=0.0013, OR=0.62, 95% confidence interval: 0.51–0.77) for methamphetamine psychosis.

In the last presentation, Dr Uhl highlighted results from a GWAS of substance dependence and cigarette smoking. He pointed out that we can now analyze GWAS datasets that use 500k-1M SNPs and compare substance-dependent individuals with matched control individuals of different racial groups, such as European, African, and Asian ancestries. Data that compare smokers who are successful at abstaining from smoking with those who cannot quit successfully (500k SNPs) are available from clinical trials and in community settings. Analyses that use Monte Carlo simulations to evaluate the significance of overlapping clusters of SNPs that achieve nominally significant case versus control allele frequency differences in multiple samples provide strong evidence for remarkable overlaps between the different datasets. There is significant, but more modest, overlap between quit success and substance dependence GWAS datasets. The data reported herein provide molecular genetic support for the idea that a smoker’s ability to abstain from nicotine has polygenic genetic components. Taken together, the current data provide promising results that we may soon use molecular genetics to match the type and/or intensity of antismoking treatments with the smokers that are most likely to benefit from them.

Major depression and alcoholism – new insights from twin, longitudinal, and genome-wide studies

Reported by Martina Rojnic Kuzman

This session was a part of Young Science Track, a forum for young scientists to present their work in an oral session. It was comprised of five talks: the first talk was by Dr Harriet A. Ball (London, UK) on ‘The heritability of depression and the influence of gender and environmental differences in a Sri Lanka-based twin sample’. Although the differences in the prevalence of depression across different countries can, in part, be explained by the differences in symptom-reporting and measurement methods, the influence of different environmental factors might also be an explanation. Dr Ball studied how environmental exposures might influence the heritability of depression. In the study, 3908 adult twins from Sri Lanka were enrolled and depression was assessed using the Composite International Diagnostic Interview. Structural equation models were applied to examine the etiology of depression, the environmental factors, and their co-occurrence. She found depression to be less heritable in men than in women, whereas a proxy measure of living standard was associated with depression in men but not in women. The overlap between measured environments and depression in men was driven by nonshared environmental factors. She concluded that specific environmental factors had been identified, which contribute to the development of depression in Sri Lanka and may potentially explain some of the differences in heritability in Sri Lanka compared with other countries.

Dr Christel Middeldorp (VU, Amsterdam, The Netherlands) reported a study on the association between serotonergic and neurotrophic genes with anxiety and depression. This study included 608 boys, 632 girls, 767 men, and 1182 women from The Netherlands Twin Register who were longitudinally assessed for anxiety and depression. Maternal and parental ratings of symptoms were available from a group of children at 3, 6, 7, 10, and 12 years of age and a group of adolescents between 14 and 18 years of age. Self-reported measures of anxiety, depressive and neurotic symptoms were assessed in five time points in adult age. Association analysis of these measurements with 37 SNPs in genes encoding serotonin 1A, 1D, and 2A receptors, tryptophan hydroxylase type 2 (TPH2), BDNF, and plexin A2 were performed using the structural equation modeling. No SNPs were associated with the measured scores. Thus, the author concluded that these SNPs might not contribute to anxiety and depression.

Dr Andreas Menke (Munich, Germany) reported on a study of polymorphisms within the metabotropic glutamate receptor 1 gene on chromosome 6q24 and major depressive disorder. Linkage for bipolar disorder had already been reported in this chromosomal region and he surmised that a common genetic vulnerability exists for both unipolar and bipolar depressions. The sample comprised of 350 cases and 370 controls from Germany, followed by a replication sample of 904 patients and 1012 matched controls. In addition, the associations of the SNPs and the dexamethasone-suppression/corticotrophin-releasing hormone stimulation test and 1-H magnetic resonance spectroscopy were analyzed in a subset of patients and controls. After correction for multiple testing, variants within the metabotropic glutamate receptor 1-GRM1 gene were significantly associated with unipolar depression. GRM1 SNPs rs2268666 and rs9386151 showed the highest associations in the discovery sample. Six of nine SNPs were replicated in the second sample. Carriers of the protective genotype against unipolar depression also showed better treatment outcome, better improvement of hypothalamus-pituitary-adrenal (HPA) axis dysregulation, and reduced hippocampal glutamate levels measured by magnetic resonance spectroscopy.

Dr Penelope A. Lind (Brisbane, Australia) reported on a study of alcohol dehydrogenase (ADH) gene variation and alcohol dependence. She examined the association of 41 SNPs in aldehyde dehydrogenase type 2 (ALDH2) and seven ADH genes with self-reported sensitivity to alcohol, alcohol consumption, and dependence in 4597 participants of European ancestry. Although ALDH2 genetic variation was not associated with any of the tested variables, rs1229984 (ADH1B Arg48His) was significantly associated with alcohol-related flushing, frequency of alcohol use, and the number of alcoholic beverages consumed per day. Furthermore, rs1042026 in ADH1B was associated with overall alcohol consumption. Dr Lind concluded that the pattern of effects on alcohol-related flushing, alcohol intake, and alcohol dependence symptoms reinforces the analogy between ADH1B variation and the effects of ALDH2 deficiency.

Dr Sarah W. Feldstein Ewing (Albuquerque, New Mexico) presented a GWAS of neural responses to alcohol cues. One of the potential markers, rs9690648, was in the presynaptic cytomatrix protein, which was associated with greater activation in the anterior cingulate and striatum. The authors tested the association of this polymorphism and alcohol-dependence symptomatology in an independent sample. They found that this locus was correlated with alcohol dependence, quantity of drinking, difficulty in sleeping, and greater reactivity during alcohol-cue exposure paradigm.

Genome-wide gene-finding studies on anxiety: where we are and where to go from here

Reported by Lorna Houlihan

Several genome-wide approaches have been applied to the search for genes associated with anxiety. First, a meta-analysis on linkage scans of anxiety-related phenotypes was presented by Dr John Hettema (Virginia Commonwealth University, USA). In this analysis, neuroticism was used as a genetic endophenotype for anxiety disorder. His aim was to localize genomic regions for neuroticism and then to examine the overlap between neuroticism and anxiety disorder. The genome-scan meta-analysis was performed by two methods; Fisher’s method of combining P values and rank-based genome scan meta-analysis. In addition, correlations between the two methods were also calculated. This method was successful in identifying linkage regions on chromosomes 1, 6, 10, and 11 for anxiety-related phenotypes. Future analyses will incorporate more linkage scans on anxiety and will weigh different studies according to their size.

Dr Christel Middeldorp (Vrije Universiteit, The Netherlands) presented a GWAS of comorbid anxiety in major depression. The sample comprised of 1738 patients with major depression, 70% of whom were also diagnosed with anxiety disorders, and 1802 controls. The analysis was approached in four ways: healthy controls versus cases with depression and anxiety; depression cases versus depression and anxiety cases; a linear regression of a social phobia subscale; and the presence or absence of agoraphobia. Association with comorbidity was detected in seven chromosomal regions (6, 11, 1, 2, 7, 12, 16) with the genes of interest in these regions being HLA, ME3, CMKLR1, with copy number variants on chromosomes 1 and 2. Replication is underway to validate these results.

The impact of the comorbidity between anxiety and depression on the design of GWAS was discussed by Dr Naomi Wray (Queensland Institute of Medical Research, Australia). To detect small effect sizes when investigating the comorbidity between anxiety and depression, Dr Wray suggested obtaining multiple longitudinal measures (to increase trait reliability) and decreasing noise by using ‘super-controls’ selected from the lower end of a normally distributed trait. As anxiety and depression are common, this method provides greater power to detect differences between cases (with high-level neuroticism) and controls (with below-average neuroticism, rather than the average level of neuroticism). Two candidate genes emerged from these association studies (PCLO and SLC6A4), which await replication in a current NIH-funded mega-analysis of a GWAS on 12 000 cases with major depression and controls.

Dr Jordan Smoller (Massachusetts General Hospital, Boston, USA) introduced another approach by conducting a genome-wide search in patients from the Systematic Treatment Enhancement Program for Bipolar Disorder study (n=786) for the anxiety-related personality traits neuroticism and extraversion in bipolar disorder. No results reached genome-wide significance; however, the LSAMP and CHRNA2 genes involved in neuronal function and behavior were highlighted. However, these candidate genes were not confirmed in a subsequent replication study. The general consensus at this session was that the current genome-wide searches for anxiety traits so far have been underpowered.

Genetics of alternative phenotypes for psychiatric disorders: the endophenotype or intermediate phenotype concept

Reported by Sarah E. Bergen and Alja Videtic

Intermediate phenotypes may be useful in genetics research under the assumption that an easily quantifiable behavior will have fewer genes affecting it and will be closer to the underlying genetic basis for the disorder than the parent clinical diagnosis based on symptoms. Defining intermediate phenotypes is a difficult task and the concept has been interpreted in many different ways.

Intermediate phenotypes for bipolar disorder

Dr Melvin McInnis (University of Michigan, Ann Arbor, USA) assessed indices of personality using the NEO-PI as well as cognitive processing in euthymic and depressed bipolar patients and unaffected controls. Composite scores were derived for verbal and visual memory, fine motor dexterity, emotion processing, executive functioning, conceptual reasoning, processing speed, and inhibitory control. Half of these measures showed adverse functioning that correlated with the number of years affected. The number of psychiatric hospitalizations also showed an inverse relationship with 5 of the 8 composite scores. Bipolar patients taking antipsychotic medications compared with those who did not, had more than twice as many hospitalizations, suggesting an underlying difference in severity. Thus, several of these measures mentioned may be useful as intermediate phenotypes in the subtyping and study of bipolar disorder.

Heritability and initial association analyses of 12 endophenotypes and 94 candidate genes for schizophrenia from the Consortium on the Genetics of Schizophrenia

Dr Tiffany Greenwood (University of California, San Diego, USA) and her colleagues in the Consortium on the Genetics of Schizophrenia collaborative, initially explored the heritability of 12 endophenotypes for schizophrenia in a sample of 297 affected individuals and their first-degree family members (N=991), minimally including both parents and one unaffected sibling. Heritability estimates ranged from 19 to 42% (mean 29%). Next, they genotyped 1536 SNPs covering 94 genes in a subset of 129 families of their full sample. Before accounting for multiple testing, 49 of the genes were associated with at least one endophenotype under investigation, with many of the genes residing in the GABA or glutamate signaling pathways. Almost half the genes genotyped (42 of 94) interact in some manner, and thus this information will be utilized in future interaction analyses. Furthermore, there was evidence of pleiotropy (>3 endophenotype associations) for 10 genes. Dr Greenwood pointed out that correlations exist between several of the endophenotypes, but substantial pleiotropy for some genes such as ERBB4 (11 associated phenotypes) and NRG1 (7 phenotypes) still strongly suggest that they confer liability to schizophrenia.

Detection of causal genetic effects and application to endophenotypes in attention-deficit hyperactivity disorder

Dr Christoph Lange (Harvard University, Boston, USA) began by discussing the ways in which a variety of phenotypes may be associated with the same genetic marker through direct genetic causality, in direct genetic effects through one of the phenotypes, or nongenetic/environmental links between traits. As an example, he illustrated how an SNP predisposing to smoking may show association with both smoking and lung cancer even though the smoking–lung cancer relationship is a non-SNP-related link. To distinguish between a direct biological effect of an SNP and an SNP’s influence through another correlated phenotype, Dr Lange suggested a general adjustment principle which can be easily incorporated into standard association tests and he validated it using a GWAS of attention-deficit hyperactivity disorder. According to his simulations, the proposed adjustment is capable of distinguishing direct and associated effects for both case–control and family samples. This will allow researchers to account for a kind of bias that would otherwise be difficult to detect.

Endophenotypes, subtypes, and cross-diagnostic studies: will genetic findings rewrite the diagnostic codes?

The session concluded on a more philosophical note by Dr Margit Burmeister (University of Michigan, USA), examining how features of intermediate phenotypes/endophenotypes could be used in gene hunting efforts. She discussed the concept that intermediate traits are believed to mark the genetic predisposition for a disorder and are ideally more heritable than the disorder itself. Endophenotypes should also be stable, unaffected by sex, age, or medication, and quantifiable. Identifying a population in which to examine the endophenotype is another important aspect. Some genes show increased association with broadening inclusion of diagnoses, whereas others may be more specifically associated with particular disease subtypes. Features of many psychiatric disorders blend together, and several Mendelian disorders confer liability to multiple psychiatric diagnoses. Unique families may also contain rare genetic risk factors for psychiatric illnesses, which can lead to clues for etiological mechanisms in other affected populations. For example, a balanced translocation impacting DISC1 in one large family conferred risk for schizophrenia, bipolar disorder, recurrent depression, and conduct disorder, but DISCI variants may be associated with psychiatric illness in other families where there is no translocation. Rare-inherited and de-novo genetic variants may thus be of interest as clues to the overall pathophysiology of the disease.

In an additional session on gene mapping of categorical phenotypes and endophenotypes, the work of five promising young scientists from The Netherlands, Canada, Russia, UK, and Slovenia were presented. The purpose of the session was to try and divide different behavioral patterns into more stable phenotypes with clearer genetic association. The first two presentations focused on analyzing genes putatively involved in schizophrenia. Dr Maartje Aukes (Utrecht, The Netherlands) reported on a study of microarrays that were used to scan more than 6000 SNPs and that was the first genome-wide scan for several candidate endophenotypes for schizophrenia. The results of the study gave some new suggestive linkages for loci on chromosomes 2, 5, 12, 15 and different traits. In another study, Dr Clement C. Zai (Ontario, Canada) showed an association of the diagnosis of schizophrenia as the phenotype with the GABA type A receptor γ2 subunit gene (GABRG2). However, this will need further confirmation.

The presentation of Dr Mikhail V. Monakhov (Moscow, Russia) illustrated an association of the dopamine D2 receptor gene promoter region and altruistic behavior, genotyping two SNPs in 172 families. These preliminary results suggested a common mechanism for hard wiring of altruistic behavior to the brain dopaminergic reward circuitry.

Another presenter, Dr Ana Catarina P. Petreira (London, UK), reported on characterization of a putative susceptibility locus on chromosome 12q24 for bipolar disorder (the gene Slynar). However, although it was shown to have alternative splicing, there was no association of variants in this gene with bipolar disorder.

The session concluded with a presentation on the association of suicide with serotonergic genes and environmental stressors by Dr Alja Videtie (Ljubljana, Slovenia). This was the first study to show a significant association of suicide with SNP 68G>C in the 5-HT2C gene. A further interesting result was the possible association of −1420C>T in 5-HT2A and suicide, a finding that had been previously linked with other psychiatric disorders. For the polymorphism −1019C>G in the 5-HT1A gene, a significant association of high environmental stress and the G allele with suicide was also found.

Dr Stefan Kloiber (Max-Planck-Institute of Psychiatry, Germany) showed association results of polymorphisms in the TPH2 gene and elevated risk for metabolic disorders and the metabolic syndrome in patients with recurrent unipolar depression in a case–control study consisting of 1000 patients with depression and 1000 healthy controls. Risk genotypes have been previously linked to lower TPH2 mRNA expression and lower 5-HIAA levels in functional studies. If these results are replicated in further studies, they will have important clinical utility for identifying depressed patients at high risk for the metabolic syndrome and thus allowing individualized disease management and prevention strategies.

Dr Barbara C. van Munster (Academic Medical Center, Amsterdam, The Netherlands) presented data on serum proteomics by profiling in eight delirious and eight nondelirious patients after hip fracture surgery. In proteomics analysis using SELDI-TOF CM10 and Q10 ProteinChip arrays, three discriminating peaks were detectable, hemoglobin-β (15.9 kDa), the doubly charged ion of hemoglobin-β (7.97 kDa), and the glycosylated form of hemoglobin-β (16.0 kDa). The largest difference was confirmed in a validation group, providing preliminary data on a possible proteomic biomarker in delirious patients.

Data on association between the 5-HTT gene-linked polymorphic region and basal cortisol secretion was presented by Stefan Wuest (University of Trier, Germany). As the serotonergic system influences HPA-axis activity and vice versa, 5-HTT polymorphisms were analyzed with markers of HPA system function. Dr Wuest found 5-HTTLPR gene variants to be associated with morning cortisol levels and adrenocorticotropic hormone levels after dexamethasone administration. No association was detected in the cortisol and adrenocorticotropic hormone reaction to a social stress paradigm.

Dr Juho Wedenoja (Institute for Molecular Medicine Helsinki, Finland) presented a search for candidate genes for cognitive endophenotypes of schizophrenia in two chromosomal regions (2q33-37 and 4q13-26), previously identified to be associated with cognitive tasks in schizophrenic patients. About 1144 intragenic SNPs in 70 genes were used for fine-mapping in a Finnish schizophrenia family study (293 families). Associations of visual working memory with polymorphisms in the PAX3 gene and verbal learning/memory with polymorphisms in the SNCA gene were detected.

Dr Stephen J. Glatt (Syracuse, NY, USA) showed results of a gene expression post-mortem study performed in 13 patients with schizophrenia, 11 bipolar disorder patients, and 10 controls using the Affymetrix ‘Exon’ array. In a comparison of patients with psychotic features and individuals with no history of psychosis (including controls and five bipolar disorder patients), they were able to detect 156 significantly alternatively spliced genes and to verify these results by quantitative reverse transcriptase polymerase chain reaction. This preliminary data could possibly have biomarker potential in differential diagnoses by distinguishing psychotic and nonpsychotic patients.

Strategies for phenotype–genotype dissection in bipolar disorder

Reported by Fernando S. Goes

Although bipolar disorder is one of the most heritable complex diseases, elucidating its genetic underpinnings remains a formidable challenge, even in this era of GWAS. An inherent difficulty may be that the BP phenotype as defined by Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) may be too etiologically heterogeneous, representing a ‘group’ of bipolar disorders, each with potentially independent or overlapping genetic etiologies. This symposium focused on whether this presumed genetic heterogeneity may be simplified by considering reliable patterns of phenotypic heterogeneity.

Dr James Potash of the Johns Hopkins University began the symposium by introducing the hypothesis that certain clinical subphenotypes may reflect a simpler genetic architecture than the full bipolar phenotype, and hence be more amenable to genetic mapping. He went on to describe the utility of the recently completed Bipolar Phenome Project, a carefully organized and freely available database of almost 6000 participants collected by the NIMH Genetics Initiative BP Consortium with almost 200 clinical phenotypic variables (Potash et al., 2007). On the basis of the evidence of significant heritability, subphenotypes were selected for analyses in the recently genotyped Genetic Analysis Initiative for Bipolar Disorder (GAIN BP) GWAS sample. In preliminary analyses, Dr Potash showed GWAS results for the following heritable clinical phenotypes: psychosis, mood-incongruent psychosis, age of onset, and suicide attempts. Although no findings reached genome-wide significance, several findings were of sufficient interest to warrant future replication efforts. In mood-incongruent psychosis, for example, the strongest overall finding was in the NRG1 receptor ERBB4 (P=4.5 × 10⁻⁶), which has been identified as a strong functional candidate gene in schizophrenia. Furthermore, in the attempted suicide analysis, an intriguing cluster of SNPs (minimum P=3×10⁻⁷) was found to be associated with the muskelin 1 gene, a mediator of the extracellular matrix believed to play a role in synaptogenesis (Tagnaouti et al., 2007). However, given that neither of these findings met stringent criteria for genome-wide significance, Dr Potash emphasized the need for appropriate replication in similarly phenotyped samples.

The next speaker, Dr Nicholas Craddock from Cardiff University, described his ongoing work aimed at dissecting genotype–phenotype correlations in the Wellcome Trust Case Control Consortium sample. Dr Craddock began his talk by using an example from the type 2 diabetes field to emphasize the importance of careful phenotype selection in large-scale genetic analyses. In a review of the evidence for association of the FTO gene with T2D, Dr Craddock showed that subtle differences in ascertainment, or phenotype selection, between samples can inadvertently lead to a ‘failure’ to replicate, despite adequate power and a true causal initial association (Zeggini et al., 2007). Focusing on bipolar disorder, Dr Craddock spoke of the need to first explore available GWAS results to uncover subphenotypes which may be enriched for a particular gene or biological system and, subsequently, to test the association(s) in independent confirmatory hypothesis. For example, in a recently published study, Dr Craddock and colleagues found significant enrichment of an index polymorphism in the GABRB1 gene in cases meeting Research Diagnostic Criteria for schizoaffective disorder, bipolar type (Craddock et al., 2008). The same cases were used to independently confirm association with the biologically related GABA_A receptor family of genes with the Research Diagnostic Criteria for schizoaffective disorder, bipolar type phenotype. Dr Craddock also presented data to suggest that some of the more recently discovered ‘disease’ genes for bipolar disorder and schizophrenia may be better conceptualized as susceptibility genes to clusters of symptoms found across the Kraepelinian dichotomy. For example, data were shown that the ZNF804A schizophrenia susceptibility allele (O’Donovan et al., 2008) was also associated with bipolar disorder and, in particular, with those cases having prominent psychotic symptoms, suggesting that this risk allele may be more a marker of psychotic symptoms than a specific diagnosis. Similarly, Dr Craddock showed data to suggest that the strongest risk allele in the CACNAC1 gene from a recently published bipolar disorder meta-analysis (Ferreira et al., 2008) was particularly enriched in cases with bipolar II disorder and was also associated with recurrent unipolar depression, suggesting that this risk allele may be most strongly associated with depressive symptoms. In conclusion, Dr Craddock stressed that careful phenotypic exploration beyond traditional DSM-IV boundaries may be of particular help in uncovering more valid genotype–phenotype correlations and will be of crucial importance in ongoing replication efforts.

The final speaker in this symposium was Dr Thomas Schulze from the Mood and Anxiety Program at the National Institutes of Mental Health (Bethesda, Maryland). Dr Schulze presented subphenotype analyses of the bipolar GAIN GWAS study for episode frequency, polarity at onset, and the Global Assessment Scale score. Although no individual result reached genome-wide significance, Dr Schulze was cautious to emphasize that this did not mean that such phenotypes were not useful for genetic analysis. Indeed, in an extension of his previously described ‘reverse-phenotyping’ strategy (Schulze and McMahon, 2004), Dr Schulze proposed the use of an algorithm known as market basket analysis (MBA) to help uncover gene–gene and gene–phenotype associations. MBA is widely used in retail analyses and is less sensitive to missing data, which is common in clinical databases. The MBA algorithm initially searches for ‘association rules’ among individual data points, which are then used to create empirically derived data clusters. In applying MBA to the Bipolar Phenome Database, Dr Schulze found four such clusters, which were comprised broadly of patients with early-onset bipolar disorder, psychotic bipolar disorder, comorbid substance misuse, and no mental illness (controls). In his further discussion, Dr Schulze highlighted the uses of this method in ongoing analyses which will combine molecular data from GWAS with phenotypic data from the Bipolar Phenome Project in the hope of finding novel gene–gene and gene(s)–phenotype(s) correlations.

Studies in homogeneous ethnicities

Reported by Shiau Foon Tee

A series of linkage studies in the same family sets have identified promising candidate genes on chromosome 1p31, 2q14, 4q21, 15q14, and 10q21-22 for Taiwanese families with schizophrenia. In addition, significant associations between schizophrenia and DISC1, GNPAT (1q42), NOTCH4 (6p21), BMP6, TXDC5 (6p24), NRG1 (8p12), PPP3CC, DPYSL2, TRIM35, PTK2B (8p21-22), ANXA7, PPP3CB (10q21), CACNG2, RASD2, and MYH9 (22q12) have also been found in this ethnic group.

In contrast, in the Japanese population, genome-wide studies combining linkage and association analysis identified the genes DTNBP1, PDLIM5, NTNG1, and NRG1 to be positively associated with schizophrenia.

In a Korean population, two regions that satisfy genome-wide criteria for linkage were found on chromosomes 2p24.3 and 6q27 for schizophrenia and alleles in the genes NRG1, DISC1, COMT, PRODH and ZDHHHC8 had significant associations with schizophrenia. In another Korean study of bipolar disorder, a GWAS comprising of 906 000 genome-wide SNPs showed only 582 SNPs that had significant differences between cases and controls with a P value of less than 10⁻⁴. In conclusion, genetic epidemiologic studies have suggested that genetic susceptibility to schizophrenia may differ across populations even within Asia. Future studies comparing cohorts from many other countries and geographic regions are warranted.

Functional studies, biomarkers, and epigenetics

Reported by Elif Dagdan

In the second Young Science Track set of oral presentations, five young investigators reported on their work. The session focused on functional studies, biomarkers, and epigenetics. The first presentation was from Dr Marlon P. Quinones, from the University of Texas Health Science Center (Dallas, USA) on the effect of genetic variation in the serotonin transporter on T cell responses in bipolar illness. He and his colleagues searched for differences on the immune response that could be associated with variation of the serotonin transporter promoter region in individuals with bipolar affective disorder (BPAD). They studied type I T cell responses (IL-12 and tumor-necrosis factor-β) that promote the clearance of intracellular infection and type II responses (IL-4, IL-13, and IL-10) that promote the production of antibodies. They had 40 DSM-IV individuals with BPAD type I and 26 healthy controls. In the BPAD group, LL homozygous had significantly reduced serum levels of type I and II cytokines compared with SL and SS carriers, whereas healthy controls showed no significant differences in type I and II cytokine levels among carriers of different allelic variants. These preliminary results suggest that variation in genes linked with mood disorders may influence the immune response system in individuals with BPAD.

The second presenter, Dr Melanie D. Klok, from the University of Leiden, The Netherlands, discussed functional consequences of haplotypes in the human mineralocorticoid receptor (MR) gene. She presented a study that investigated the regulation and dynamics of MR expression. Stress-related disorders such as depression and also healthy aging are associated with changes in HPA-axis activity. Brain MR regulates basal activity and stress reactivity of the HPA axis, both of which are associated with neuronal excitability, electrolyte balance, and human behavior. A disturbed regulation of expression or functioning of MR is hypothesized to be a risk factor for precipitation of stress-related disorders. Dr Klok and colleagues showed that the human MR I180V SNP influences activation of MR in vitro, as well as modulates stress-induced cortisol response in healthy individuals. An association has been also found between this polymorphism and depression in the elderly. In addition, genetic variants in the MR gene were examined for an effect on MR expression and for association with biological markers, personality traits, and psychopathology. Assays are being performed to investigate the influence of steroid and sex hormones and transcription factors on MR promoter activity; 4 kb of the MR promoter region was sequenced and three main haplotypes were tested for their promoter activity under various conditions. These haplotypes seem to be associated with behavioral phenotypes and psychopathology.

The third presentation by Dr Patrick O. McGowan (McGill University, Montreal, Canada) was on ‘Promoter-wide hypermethylation of the ribosomal RNA gene promoter in the suicide brain’. Using sodium bisulfite mapping of the rRNA promoter and quantitative realtime PCR of rRNA expression, McGowan and colleagues tested the hypothesis that epigenetic differences in critical loci in the brain are involved in the pathophysiology of suicide. Alteration in gene expression in the brains of people who have committed suicide has previously been reported in several genes. DNA methylation as an epigenetic regulator is thought to play a role. Thirteen individuals were selected who committed suicide and had an early childhood history of neglect/abuse, and compared with 11 control individuals who died of unrelated causes. rRNA, a crucial component of the protein synthesis machinery, was significantly hypermethylated throughout the promoter and 5′ regulatory gene regions in the hippocampus of the individuals who committed suicide. These results are consistent with reduced rRNA expression in the hippocampus. This difference in rRNA methylation was not evident in the cerebellum and occurred in the absence of genome-wide changes in methylation. It is thus concluded that aberrant regulation of protein synthesis may occur in people who commit suicide through epigenetic gene modulation.

Dr Elif Dagdan (Trinity College Dublin, Ireland) followed with a presentation on S100B as a susceptibility gene for psychotic bipolar disorder and received the ‘Young Investigator award’ for the best oral presentation. He and his colleagues reported the association of an SNP (rs3788266) in the promoter region of S100B with a psychotic form of BPAD. The calcium binding and glial cell-derived neurotrophic factor, S100B, has previously been implicated in the pathology of BPAD and schizophrenia and has been reported to be elevated in serum of patients with both disorders. S100B is neurotropic at pico and nanomolar levels, but apoptotic at micromolar levels. The disease-associated C allele disrupts Trex/MEF3 consensus recognition, which is bound by six family transcription factors, suggesting that SNPs for this allele may affect S100B expression. The functional effect of rs3788266 on S100B promoter activity was determined using the luciferase reporter system. Dagdan and colleagues also measured S100B RNA levels in post-mortem brain tissue and protein levels in serum to test for possible genotypic effects in vivo. The Luciferase reporter gene expression was significantly increased in the presence of the T allele compared with the C allele. However, preliminary data indicate that BPAD individuals with the TT genotype have lower mean serum S100B levels compared with those with the TC or CC genotypes. A similar pattern was observed at the RNA level but was not statistically significant. Rs3788266 may represent a functional susceptibility variant that contributes to the increased S100B levels observed in BPAD patients.

The final presentation was by Dr Sietske G. Helder (Institute of Psychiatry, London, UK) on ‘Biomarkers of anorexia nervosa’. Anorexia nervosa (AN) is characterized by below normal weight, persistent fear of weight gain, as well as obsessive–compulsive, harm-avoidant and anxious traits. Dr Helder and colleagues conducted a longitudinal proteome-wide analysis, aimed at identifying new protein biomarkers associated with AN and recovery. Such biomarkers could be used to monitor disease severity and to predict outcome, for example, measuring response to treatment in a clinical trial, as well as to identify etiological factors. They analyzed blood plasma from AN patients by two-dimensional gel electrophoresis and mass spectrometry. An overview of their initial results on adult AN outpatients is being prepared for publication and unavailable for this report.

Pharmacogenetics

Reported by Martina Rojnic Kuzman

Several posters at this congress had interesting pharmacogenetic results. In the studies on depression, a significant contribution was made by Dr Masaki Kato and colleagues (University of Bologna, Bologna, Italy) who presented a meta-analysis of antidepressant pharmacogenetic findings in major depressive disorder. Pharmacogenetic studies with SLC6A4, HTR1A, HTR2A, TPH1, GNB3, BDNF HTR3A, and HTR3B were analyzed. Better treatment response was observed with the SLC6A4 STin2 12/12 genotype, TPH1 218C/C genotype, BDNF 66 Met genotype (Asian), and HTR2A-1438 G/G (Asian). When analyzing side effects, pooled ORs of 5-HTTLPR l, HTR2A-1438 G/G, and HTR3B 129 Tyr/Tyr were associated with a significant risk modulation, especially in patients treated with selective serotonin reuptake inhibitors. The authors concluded that STin2, 5-HTTLPR, HTR2A, TPH1, BDNF, and HTR3B gene variants may modulate antidepressant response. Joanna M. Biernacka and colleagues (Rochester, MN, USA) reported a positive association of CYP2C19 gene variants with remission in a large sample of 1074 patients treated with citalopram. Although CYP2D6 did not show a significant association with treatment response, the interaction of CYP2C19 genotypes with tolerance and remission was significant. Masataka Wakeno and colleagues (Osaka, Japan and Bologna, Italy) found an association of the C allele of the C-1297G alpha 2A-adrenergic receptor gene polymorphism with response to milnacipran in 93 depressed individuals and concluded that polymorphic variants of the ADRA2A receptor might have predictive value for treatment response in patients with depression.

In the studies on schizophrenia, interesting results in a larger sample were reported by Rudi Hwang and colleagues (Toronto, Canada). They reported a significant association of three of 10 polymorphic D3 loci variants with the response to 6 months of clozapine treatment in 245 patients. In addition, in a pooled analysis of four other response studies (total N=549), the authors confirmed the association of the ser9gly, Ser/Ser genotype with poor clozapine response. Qinghe Xing and colleagues (Shanghai, China) found a positive association of one of the four tested SNPs -rs1173713 for the 5-HT3A receptor and better treatment response to risperidone, measured with negative and general subscales of Positive and Negative Syndrome Scale in 107 Chinese patients with schizophrenia. One of the haplotypes, C-A-G, contributed to better treatment response. These authors thus concluded that polymorphic variants of the 5HT3A receptor might have predictive value for treatment response in patients with schizophrenia.

Other studies focused on side effects of psychotropic medications. Interestingly, in one of the posters, a genetic framework responsible for the arrhythmic effects of antidepressants and antipsychotics was presented by Dr Antonio Drago and colleagues (Bologna, Italy). They provided a list of relevant rare mutations and Tag SNPs within the genes that are currently considered as potential candidates for the definition of a genetic proarrhythmic profile (SCN5A, SCN4B, CACNL1AC, KCNH2, KCNQ1, KCNE1, ANK2, ALG10, KCNJ2, KCNE2, RYR2, KCND3, KCND2, ACE, NOS1AP, CASQ2).

Hee Jung Nam and colleagues (Seoul, South Korea) reported an association of the glutamate transporter gene SLC1A1 (rs2228622 and rs3780412 loci) and atypical antipsychotic-induced obsessive–compulsive symptoms among 40 patients who developed these symptoms after treatment with different atypical antipsychotics compared with 54 patients who did not develop any symptoms after 2 years of treatment. However, significant associations of obsessive–compulsive symptoms were also observed with age, sex, and medication type. Nevertheless, they concluded that variations in this gene might be involved in the pathogenesis of obsessive–compulsiveness after treatment with atypical antipsychotics. Two studies reported no association of polymorphisms in either the monoamine oxidase genes (MAOA or MAOB) or the dopamine transporter gene 40 bp variable number repeat (VNTR) and antipsychotics-induced restless legs syndrome among Korean patients with schizophrenia (Seung-Gul Kang and colleagues and Jung-Eun Choi and colleagues, Seoul, South Korea). Both studies involved the same measurement instruments in 190 patients. Dr Arun K. Tiwari and colleagues (Toronto, Canada) reported a possible association of the promoter polymorphism -rs806378 in the cannabinoid receptor 1 (CNR1) gene with clozapine-induced weight gain in 68 patients. They concluded that variations in this gene might be predictors of clozapine-induced weight gain.

Dr Adolfo Quinones-Lombrana and colleagues (Madrid, Spain) reported on a study of the novel RIP kinase ANKK1 and its relationship to the dopaminergic system in a mouse brain cell culture. The authors found a significant upregulation of ANKK1 and DRD2 genes after treatment with D2 receptor agonists (quinelorane and aripiprazole), whereas both receptor antagonists (sulpiride and haloperidole) inhibited the DRD2 gene. Sulpiride did not affect ANKK1 expression, but haloperidol treatment upregulated this gene. The authors concluded that these results suggest a functional relationship between these genes.

Drs Yu-Chih Shen and Chia-Hsiang Chen (Hualien, Taiwan) reported the identification of 22 SNPs and six rare mutations in the glutamate receptor, ionotropic, N-methyl-d-aspartate 3A (GRIN3A) that were associated with schizophrenia. Although no differences in distribution of haplotypes were observed in cases (N=333) versus controls (N=369), four missense mutations (D133N, Y579C, R1024X, Q1091H) and two synonymous mutations (Y873Y and E889E) were found among patients but not controls, reaching statistical significance. The authors concluded that these variants might contribute to the risk of schizophrenia.

Can we predict adverse events with pharmacogenetic markers?

Reported by Andreas Menke

This symposium focused on the identification of pharmacogenetic biomarkers to predict drug-related side effects. Dr Anil Malhotra (Zucker Hillside Hospital, New York, USA) presented data from various studies focusing on adverse events induced by antipsychotic medication. Weight gain leading to further morbidity and poor adherence to treatment is one of the most serious side effects caused by many antipsychotics. As shown by Reynolds and colleagues, patients treated with chlorpromazine or risperidone who carried the − 759T allele of the 5-hydroxytryptamine 2C receptor had significantly less weight gain than patients without this variant (Reynolds et al., 2002). These findings were also confirmed for clozapine treatment (Reynolds et al., 2003). Biomarkers were not only reported for adverse events, but also for drug response. In a study by Lencz and colleagues (Zucker Hillside Hospital), patients treated with atypical antipsychotics exhibited a significantly faster time until response when they were G carriers (A–241G), whereas − 141C Del carriers took a significantly longer time to respond (Lencz et al., 2006). PGxHealth has developed a blood test for risk of clozapine-induced agranulocytosis, with patients having the low-risk allele(s) having a 20% lower risk than the overall patient population, whereas patients with the high-risk allele(s) have a significantly increased risk (www.pgxhealth.com).

Switching from schizophrenia to major depression, Dr Andreas Menke (Max Planck Institute, Germany) showed data on biomarkers predicting treatment emergent suicidal ideation (TESI). Although antidepressants are associated with a reduction of suicides, there is a subgroup of patients who develop TESI at the onset of treatment with antidepressants. In 2005, the US Food and Drug Administration issued a black box warning for antidepressants indicating pediatric and adult patients at risk for this side effect. To identify this subgroup of patients at risk, in the STAR*D sample, two SNPs located in glutamate receptors were significantly associated with TESI, rs4825476 (within GRIA3), and rs2518224 (within GRIK2) (Laje et al., 2007). In the Munich Antidepressant Response Signature Project (http://www.mars-depression.de), these results were partly replicated, although a different risk allele was found with SNP rs4825476 and no association was found with rs2518224. However, 15 other SNPs located in GRIK2 were significantly associated with TESI (Menke et al., 2008). Among important differences in design between the two samples were patient selection, medication use, and population structure. Next, based on a whole genome approach (300 k BeadChips), a subset of 18 SNPs was presented that could predict TESI.

The last speaker was Dr Gonzalo Laje (NIMH, Bethesda, USA), who presented findings from a whole genome association with TESI of the STAR*D sample. The strongest associations were in two SNPs located within IL28RA and PAPLN, with ORs of 2.7 and 4.7, respectively. Data from a combined analysis of the STAR*D sample and the Munich Antidepressant Response Signature sample was also presented, and, one of the associated genes was nuclear receptor coactivator 3, NCOA3, which directly binds nuclear receptors and stimulates transcriptional activities.

Genetic routes to drug discovery in schizophrenia

Reported by Sarah E. Bergen

Ultimately, genetic findings will be important to use for targeting new treatments for disease. This session was devoted to the work of some scientists focusing on genetic identification of potential drug targets and the generation of animal models on which to test drugs in development.

Dr Douglas Blackwood (University of Edinburgh, UK) explored the impact of translocations disrupting four genes (DISC1, PDE4B, GRIK4/KA1, and NPAS3) in pathways amenable to pharmacological manipulation. Each of these genes had previously showed associations with schizophrenia in case–control studies. DISC1 was initially identified in a Scottish family carrying a balanced translocation, t(1;11)(q42;q14) affecting this gene (Millar et al., 2000). A similar translocation, t(1;16)(p31.2;q21), disrupting DISC1 and PDE4B was identified in a proband with schizophrenia and a cousin with a chronic psychotic disorder. DISC1 interacts with PDE4B to modulate neuronal cAMP levels, and the cAMP pathway and subsequent behavior in mouse mutants. It is modified by both antidepressants and antipsychotics (Millar et al., 2005). Another gene, the glutamate receptor GRIK4/KA1, was shown to be disrupted in a patient with schizophrenia carrying a complex chromosomal rearrangement. Glutamatergic alterations are well documented in schizophrenia, and pharmacological manipulations may be possible based on evidence that GRIK4 variants may modulate the response to the antidepressant citalopram in people with depression (Paddock et al., 2007). The final gene considered, NPAS3, was disrupted in a mother and daughter with psychosis. NPAS3 knockout mice lack in neurogenesis in the adult hippocampus and show cognitive and behavioral deficits, which may be related to schizophrenia neuropathology. NPAS3 alleles are related to response to the antipsychotic drug lloperidone in subjects with schizophrenia (Lavedan et al., 2008). These rare translocations conferring risk for schizophrenia may thus open a window to potential drug targets applicable to treating psychoses arising from multiple pathological circumstances.

The development of animal models of schizophrenia is a vital part of the drug development pipeline. It is well known that people taking phencyclidine (PCP) exhibit behavioral features of schizophrenia. The work Dr Brian Morris (University of Glasgow, UK) presented explored the extent to which chronic administration of PCP (2.6 mg/kg i.p. for 28 days) in rats mirrors the neurobiological changes observed in humans with schizophrenia (Pratt et al., 2008). The PCP-treated rats underwent quantitative ¹⁴C-2-deoxyglucose imaging, which revealed reduced glucose metabolism in the frontal lobes, analogous to the hypofrontality documented in schizophrenic patients. They also exhibited reduced mRNA expression of the GABAergic interneuron marker parvalbumin in the frontal cortex and reticular thalamus as seen in human schizophrenics. Chronic clozapine administration was found to restore parvalbumin deficits in the frontal cortex of PCP-treated rats but did not reverse hypofrontality.

Transcriptional profiling was also performed on nine human postmortem samples of prefrontal cortex (BA10) from affected individuals using the Affymetrix 100K array. Results were compared with expression array data from rat frontal cortex using IGA software. Nonparametric ranking of GO pathways revealed that expression of myelin, synaptic, and GABA-related genes were altered in both species. Altogether, this work presents evidence for a promising model of schizophrenia. By assessing transcriptional changes in rats following PCP administration and potential antipsychotic medications, drug candidates could be better evaluated.

Dr Judith Pratt (University of Strathclyde, UK) also used the PCP-treated rat model of schizophrenia, but took a reverse translational approach. To do so, she performed transcriptional profiling on frontal cortex tissue from this model and back-translated the resulting information in relation to human postmortem expression data. This validated some of the transcriptional changes observed in the rat PCP model, as well as generated novel hits in the rat to be explored further in humans with schizophrenia. The cognitive deficits exhibited by affected humans have been difficult to ameliorate pharmacologically, but increasing support for the validity of the PCP model of schizophrenia suggests it may be useful in the development of new drugs for cognitive deficits and other aspects of this devastating psychiatric disorder.

Genetics of sleep and circadian rhythms

Reported by Sandra Villafuerte

Dr Toru Takumi (Osaka Bioscience Institute, Japan) reported on molecular mechanisms of the biological clock and its output in circadian physiology and pathophysiology. The mammalian circadian system consists of three components: input (i.e. light, temperature, noise, social cues), pacemaker (suprachiasmatic nuclei) and output (supraventricular zone). The supraventricular zone, located just dorsal to the suprachiasmatic nuclei, represents the output pathway of the Circadian rhythm relaying signals to centers regulating behavioral and physiological activity such as hormone secretion and locomotion activity. Peripheral tissues and cultured cells have molecular clock mechanisms that are well conserved among species from Drosophila to human and provide a useful model for analyzing the effect of the circadian rhythm on behavior and disease. Microarray analyses in rodents have identified about 10% of genes rhythmically expressed in each tissue with few genes in common among different tissues (Storch et al., 2002). Circadian molecular mechanism is a series of interlocked transcriptional and translational feedback. Dr Takumi referred to three mechanisms underlying peripheral circadian control and the implications in human behavior. Three major regulatory conserved elements have been identified so far: E-box, RORE and DBPE when combined produce different rhythms. While CLOCK/BMAL1 heterocomplex is considered a positive regulatory element, PERs (Period) and CRYs (Cryptochrome) restrain the transcriptional activity. CLOCK and BMAL1 regulate gene expression by interacting with the promoter element E-box in PER2. Dr Takumi’s laboratory has identified double E-box-like elements required for the generation of cell-autonomous transcriptional oscillation of clock and clock-controlled genes across tissues and species (Nakahata et al., 2008). Mice with a deletion mutation in PER2 display aberrant circadian rest-activity cycles similar to that observed in human depressed patients (Nakamura et al., 2008). A second mechanism highlighted in this presentation was the Bmal1 positive regulation by the retinoic acid receptor-related orphan receptor Rorα as well as the negative regulation by the Rev-erbα orphan nuclear receptor (Akashi and Takumi, 2005). Rorα mRNA is widely expressed in most peripheral cells and is the most potent activator of BMAL1 transcriptional oscillation by directly acting on the Bmal1 promoter. REV-ERB acts as a transcriptional repressor of BMAL1. When REV-ERB protein levels decrease during the night, RORα can activate BMAL1 transcription (Akashi and Takumi, 2005). A closer look into the role of circadian rhythm on depression was done with the learned helplessness (LH) rat model for depression. At the behavioral level, the locomotor activity rhythm is changed in these rats and at the cellular level, the circadian transcriptional rhythm seemed to be correlated with each other. At the molecular level, the phosphorylated glycogen synthase kinase-3β (pGSK-3β) is likely to be the key molecule that connects the behavioral rhythm with the cellular ones. Downstream effects of GSK-3β include clock genes. Chronic treatment of Lithium (Li) lengthens the circadian period and delays phase of LH rat fibroblasts. Li, a glycogen synthase kinase-3 (GSK-3) inhibitor, may have downstream effects on clock genes. For researchers interested in investigating circadian phenotypes, the use of fibroblasts from patients correlates with in-vivo mechanisms, providing an accessible model as a diagnostic and therapeutic tool.

Dr Ebisawa (Department of Psychiatry, Tokyo Metropolitan Police Hospital, Tokyo, Japan) focused on different in-vitro tools available to study the effects of Li and other agents on circadian rhythm. Dr Ebisawa’s laboratory has established human retinal pigment epithelial (RPE1) cell lines expressing the luciferase gene controlled by the human BMAL1 promoter to study the circadian rhythm of human-derived cells in-vitro. The circadian period causes these cells to elongate and the amplitude decrease when 10 mmol/l Li is added into the medium. These established cells may be useful in the screening of drugs that affect circadian rhythm, which in turn may lead to treatments for human circadian rhythm disorders (Yoshikawa et al., 2008).

In-vitro assays have also been useful to study the effect of genetic variation on biochemical phenotypes such as protein phosphorylation, and enzyme activity. Variations in the clock gene seem to alter the CK1δ/ε-induced phosphorylation of the clock proteins. The tau mutation, a gain-of-function mutation in the CK1ε reduces its kinase activity in-vitro and shortens the circadian period. Peripheral cells cultured in-vitro can generate circadian rhythm in the same way as neuronal cells in the suprachiasmatic nuclei. When fibroblasts from human skin are cultured and stimulated with dexamethasone, clock mRNA expression fluctuates for several days reflecting the circadian rhythm in the SCN (Ebisawa, 2007). Fibroblasts from different individuals show widely varying circadian periods which may be due to different genetic factors (Brown et al., 2005). These were some examples of how in-vitro assays may be useful to analyze circadian rhythms at the molecular and physiological level.

Dr Malcolm von Schantz from the University of Surrey, UK reported on polymorphisms in circadian clock genes and their effects within and outside of the circadian oscillator. He focused mainly on the influence of genetic variants in clock genes on phenotypic variability in sleep, diurnal preference and sleep disorders.

A VNTR polymorphism in the coding region of PER3 is associated with both diurnal preference and delayed sleep phase syndrome (DSPS) and this association is affected by age, with the association being stronger in young people (Jones et al., 2007). Scores in the Horne-Ostberg questionnaire increase with age indicating a diurnal preference as people get older, especially in women. Interestingly, PER genes are located close to telomeres that are also shortened with age. Individuals homozygous for the PER3^5/5 repeat have increased scores on several markers of sleep homeostasis compared to individuals with PER^4/4. These markers include slow-wave sleep and electroencephalogram slow-wave activity in non-rapid eye movement sleep and theta and alpha activity during wakefulness and rapid eye movement sleep (Viola et al., 2007). Dr von Shantz’s laboratory showed that the shorter repeat PER⁴ protein product translocated more efficiently to the nucleus than the PER⁵ as determined by in-vitro analysis. Furthermore, the VNTR polymorphism in PER3 differentially influences the effect of sleep deprivation on executive and non-executive function in the early morning. Expression levels in leukocytes revealed individual phase-differences in PER3 expression during a constant routine and these levels correlate with sleep timing during entrainment (Archer et al., 2008).

A homeostatic process that regulates sleep is exemplified in sleep-deprived mice, which show that most diurnal changes in gene transcription are sleep-wake and tissue-dependent. Specifically, Homer1a and other genes involved in glutamate-induced neuronal activity are overexpressed after sleep loss in a transgenic mouse line suggesting a role for sleep in intracellular calcium homeostasis. Potential functional polymorphisms in PER2, another key player in the circadian oscillator, are associated with both advanced sleep phase syndrome and diurnal preference (Carpen et al., 2005). Among personality traits, conscientiousness was the major predictor of diurnal preference (Hogben et al., 2007). This diurnal preference has been linked to other physiological parameters and polymorphisms in circadian genes and may also be related to personality traits. Among mood disorders, Bipolar Disorder has been associated with the CLOCK gene. Mice carrying a mutation in the CLOCK gene display a behavioral phenotype similar to human mania and chronic administration of Li attenuated this behavior (Roybal et al., 2007).

Geriatrics and genetics

Reported by Lorna Houlihan

Substantial contributions have recently been made in the genetics of geriatrics, encompassing Alzheimer’s disease, delirium, and cognitive aging in healthy individuals.

In Alzheimer’s disease, Dr Akitoyo Hishimoto (Kobe University, Japan) showed that functional haplotypes at the NRXN3 splicing site alter the expression of NRXN3 isoforms and may also alter vulnerability to Alzheimer’s disease (Hishimoto et al., 2007). Dr Kenju Hara (University of Miami, USA) demonstrated that variants in the promoter region of the vitamin D receptor (VDR) gene are associated with late-onset Alzheimer’s disease and positively regulate the transcriptional activity of VDR. Alzheimer’s disease has also been investigated at the whole genome level with a large, well-characterized case-control sample. Dr Julie Williams (Cardiff University, UK) presented the stage 1 analysis of over 6000 cases and 13 000 controls genotyped on the Illumina HumanHap 610 platform. Preliminary analysis revealed association with the well-known Alzheimer’s disease genetic risk factor APOE and two other interesting novel loci on chromosome 8 and 12. Genotyping for stage two of the project comprises 2927 cases and 5686 controls on 20–30 000 of the most significant SNPs. Secondary analysis will incorporate APOE status, age of onset, biological pathways, clinical signs and symptoms, copy number variants and basic functional analysis.

Additional research presented by Dr Jonathan Haines (Vanderbilt University, USA) took a family-based approach to search for Alzheimer’s disease genes other than APOE. A high-density linkage screen for dementia loci was performed in a large Amish pedigree of 672 individuals, 103 of which were diagnosed with Alzheimer’s disease. Linkage analysis suggested four regions of linkage on chromosomes 1, 3, 5 and 18 (LOD>2) which are being refined with additional SNPs across the regions and a GWAS to localize the signals.

In the relatively new field of delirium studies, Dr Barbara van Munster (University of Amsterdam, The Netherlands) introduced the difficulties and possibilities of genetic research on delirium in the elderly. A delirium diagnosis is a disturbance of consciousness caused by any medical condition, and genetic studies of this acute neuropsychiatric syndrome are complicated by predisposing factors, precipitating factors, uncertainties on the underlying pathophysiology and confounding effects related to dementia. Genetic analysis has found an association of the APOE-e4 allele with longer-duration of delirium, though not with delirium in the elderly (van Munster et al., 2007). However, new study data added to a meta-analysis showed that the APOE ε4 allele is likely to be associated with delirium in the elderly, especially in a non-cardiac surgical population. Furthermore, variants in the DRD2 and SLC6A3 genes were associated with delirium but require validation in a comparable cohort.

Candidate gene analysis was also presented by Dr Lorna Houlihan (University of Edinburgh, UK) on cognitive aging in a large Scottish cohort. Nineteen SNPs in ten candidate genes were investigated in over 1000 Scottish cases from the Lothian Birth Cohort 1936, which measured cognitive ability at ages 11 and 70 (Deary et al., 2007; Houlihan et al., 2008). The study presented several nominal associations of COMT, KL, PRNP, PPP1R1B, SORL1 and WRN with cognitive abilities and/or age-related cognitive decline.