Poster Session at the 5th International Imaging Genetics Conference

The Scientific Program Committee of the International Imaging Genetics Conference invited the scientific community to submit abstracts of their current research to be peer reviewed and considered for inclusion in the scientific program as poster presentations. All accepted abstracts are published on the conference website below and will be displayed in the Beckman Center atrium both days of the conference. The poster sessions are an important aspect of the scientific program as they promote greater interaction between the researcher and other scientists.

Poster Presentations:

A1. Title: Common genetic variants of transcription factors are associated with differences in cortical area development Keywords: cortex, arealization, association, structural, MRI, TOP 

Authors: Trygve E. Bakken, Cooper Roddey, Alexander H. Joyner, Cinnamon S. Bloss, Erin N. Smith, Lars M. Rimol, Srdjan Djurovic, Ingrid Melle, Ingrid Agartz, Nicholas J. Schork, Ole A. Andreassen, Anders M. Dale

Aim of Investigation: Genes play a crucial role in cortical development, and many are highly conserved from fruit flies to mice to humans. Several homeobox transcription factors (TFs) are critical in defining the identity of anterior versus posterior cortical regions. For example, EMX2 is expressed in a gradient from posterior-medial to anterior-lateral cortex, and it imparts a visual cortical identity to surrounding cells. Mice underexpressing EMX2 have a smaller visual cortex and correspondingly larger frontal cortex, and mice overexpressing EMX2 show the opposite effect. In humans, there have been reports of defects in EMX2 leading to schizencephaly, a rare disorder that is characterized by large cortical malformations. In this study, we investigated whether common polymorphisms in these TFs lead to common variations in cortical arealization, rather than gross structural defects. 

Methods: Subjects were enrolled in the Thematic Organized Psychoses Research Study (TOP) and included adult Norwegian men and women with normal aging (controls; n=119), schizophrenia (n=85), bipolar disorder (n=81), or other psychotic disorder (n=39).  Subjects underwent neurocognitive and structural neuroimaging assessment and were genotyped on the Affymetrix 6.0 Array.  Single nucleotide polymorphisms in several TFs likely to affect cortical arealization were tested for association with the distribution of cortical areas using several regression methods and controlling for age, sex, and diagnosis.

Results :Several SNPs in EMX2, COUP-TFI and Odz1 showed association with a shift in the anterior-posterior distribution of cortical areas. 

Conclusions :Humans have a wide diversity of sizes of cortical areas, and these results suggest that this may be caused partly by common variations in developmental genes. It will be interesting to investigate whether cognitive differences can also be explained partly by this genetic variability. 

A2. Title: Power in Whole-Brain Whole-Genome Association Studies

Author: Satoru Hayasaka

Aim of Investigation: In a typical genome-wide association study (GWAS) with 500,000 or so SNPs, often thousands of subjects are required to detect significant associations with sufficient power. On the other hand, in a typical neuroimaging study, despite a large number of multiple comparisons (~100,000 voxels), often a few dozen subjects are enough to yield sufficient power. Therefore, in a combined whole-brain whole-genome analysis, it is not unreasonable to expect that the number of required subjects would be somewhere in between the two. To answer this question, power and sample sizes for a whole-brain whole-genome analysis were determined.

Methods: Power was calculated based on a multiple comparison correction method controlling the false discovery rate (FDR). It was assumed that, to assess any significant SNP-voxel associations, an F-test is performed for each of all possible combinations of 500,000 SNPs and 100,000 voxels. A small number of the SNP-voxel combinations (M1) were assumed to be associated, with effect size Cohen’s f=0.1. Power and FDR-corrected thresholds were calculated theoretically for various values of M1 at FDR=0.05 level. Results With M1=500, 80% power could be achieved with 561 subjects. For M1=50,000, 462 subjects were needed to achieve 80% power. The FDR-corrected threshold for the M1=50,000 case was p=10E-7.374, which was comparable to a typical GWAS analysis threshold. The threshold became somewhat more stringent for M1=500, with p=10E-9.375.

Conclusions It was found that around 500 subjects would be required to achieve sufficient power in a whole-brain whole-genome association study. This is somewhat smaller than the sample size in a typical GWAS, even though the threshold is not overly stringent. The results from this work indicate that combined whole-brain whole-genome association studies are feasible in terms of statistical power. Acknowledgement Supported by the Transitional Scholar Award from Wake Forest University Translational Science Institute.

A3. Title: Asymptomatic carriers of a single mutant Parkin allele show overactivity of the left rostral putamen in the context of a visuospatial response conflict

Authors: Joyce van der Vegt, Michael Weiss, Bastiaan Bloem, Ferdinand Binkofski, Christine Klein, Hartwig R. Siebner

Objective: Asymptomatic carriers of a single mutant Parkin allele show a latent nigrostriatal dopaminergic dysfunction. We examined how the presence of a mutant Parkin allele influences task related acvtivity in the basal ganglia when subjects have to select a movement in the context of competing response tendencies.

Methods: Asymptomatic individuals carrying a single mutant Parkin allele (n=6) and healthy age-matched controls (n=10) performed a Simon task during fMRI at 3T. Responses are coded by two symbolic cues in this two-choice reaction time task. The position of these cues is either spatially compatible or incompatible with the response instructed by the symbolic cue. We chose the Simon task because patients with Parkinson´s disease are impaired at resolving the conflict induced by spatial incompatibility (Praamstra & Plat, 2001). Mean reaction times (RT) and error rates (ER) during fMRI were analyzed using ANOVA. Task-related BOLD signal changes were analysed using SPM2 software. Only correct responses were included in these analyses.

Results: Participants respond faster when relative spatial positions of stimulus and response match. Incompatible trials were associated with higher ER in both groups. Parkin mutation carriers made significantly more errors than healthy non-carriers in compatible and incompatible trials (P<0.001). No significant differences in RT were found between groups. Analysis of the fMRI data revealed an increased activation of the left anterior putamen during spatially incompatible trials in Parkin mutation carriers relative to non-mutation carriers. This hyperactivity was only present in right hand responses. No differences between groups were found during spatially compatible trials.

Conclusions: The fMRI results indicate that a mutant Parkin allele is associated with overactivity of the basal ganglia when a correct action has to be selected in the context of conflicting response tendencies, due to a latent dopaminergic dysfunction. Given the overall increase in error rate, this may indicate basal ganglia dysfunction associated with latent nigrostriatal dopaminergic degeneration.

A4. Title: Hormone-genotype interactions in PFC function

Authors: Emily Jacobs, Mark D'Esposito

Dopamine transmission within the prefrontal cortex is critical for tasks dependent on working memory. The PFC is extremely sensitive to fluctuations in DA—both insufficient and excessive dopaminergic activity impairs PFC function. Taking into account an individual’s basal level of PFC DA is essential for predicting DA’s effect on cognitive processes. Individual differences in PFC DA stem in part from genetic polymorphisms that alter the efficiency of DA’s metabolic pathway. A polymorphism within the COMT gene produces an enzyme with four times greater activity. Individuals with the less active enzyme (met/met) have increased PFC DA relative to individuals with the more active enzyme (val/val). Estrogen also impacts the DA system by amplifying DA transmission. Thus, performance on tasks that depend on precise levels of PFC DA may vary throughout the cycle as estrogen levels rise and fall. Importantly, these effects may not be measurable unless individual variation in baseline DA levels is accounted for. This study investigates the effects of estrogen on the performance of DA-dependent tasks as a function of COMT genotype. Female subjects were pre-selected for COMT genotype and tested on two occasions, when estrogen levels are at their peak and trough. Subjects completed a modified n-back WM task during an fMRI scan. The task varied in difficulty based on load (0- 2- and 3-back blocks) and trial type (nontarget, target and lure trials). Given evidence in animals for an increase in synaptic DA release with increasing cognitive load, it was hypothesized that the greatest estrogen/genotype effects would occur at high levels of cognitive load. Results indicate opposite effects of estrogen based on COMT genotype: val-val subjects showed improved performance on lure trial accuracy when estrogen levels were high, while met-met subjects were impaired. The results fit the “inverted-U” model of DA action in the PFC.

A5. Title: Expression and behavior of AVP treated male titis

Authors: M. R. Jarcho, W. A. Mason, S. P. Mendoza, K. L. Bales

Arginine vasopressin (AVP) has several peripheral and central functions, including an intimate involvement in social behavior. AVP is critical to the formation of a pair-bond in male prairie voles, but its role in pair-bonding has not been previously been investigated in a monogamous primate species. This study investigated the effects of intranasal administration of AVP on male titi monkey (Callicebus cupreus) social behavior and gene expression. Behaviors measured included latency to contact and time spent in contact with both a familiar female partner and an unfamiliar female stranger. Gene expression profiles were assessed in peripheral blood mononuclear cells with GeneChip microarray technology and confirmed with real-time quantitative PCR. Behavioral observations showed that when treated with AVP males showed reduced latency to approach their partner relative to a stranger and initiated more contact bouts with their partner relative to a stranger. These results support previous work conducted with rodents. Gene expression analysis showed that intranasal AVP was associated with a down-regulation of several immune system related genes. More specifically, these genes (e.g. IL1B, IL1RAP, NFIL3) are involved in inflammatory response. Interleukin 1β is a proinflammatory cytokine, and is known to have receptors on the paraventricular nucleus, a region crucial to the initiation of the HPA stress response, and is known to increase plasma ACTH concentrations. Down-regulation of these genes, therefore represents a suppression of the inflammatory response. Although these gene expression results were not predicted, they may suggest some intervening mechanisms by which social support can affect health and vice versa. Future research will investigate the role of intranasal AVP in pertinent social behaviors, such as establishing pair bonds or becoming a new parent.

A6. Title: A study of Imaging Geno-Phenotypes in dyslexia

Authors: Holger Kirsten, Carolin Ligges, Arndt Wilcke, Peter Ahnert, Johannes Boltze

Aim of Investigation: Analysis of Imaging Geno-Phenotypes (IGP) is a highly promising new method to identify new putative risk genes for neurological conditions which are at least partially heritable. The aim of this study is to identify new putative risk genes for dyslexia. In addition, the genetic influence on important aspects of dyslexia (phonological processing, auditory processing, and visual processing) will be assessed.

Method: A group of twenty dyslexic German children and twenty age- and IQ-matched controls (range 10-14 years) performed controlled auditory, visual and phonological tasks in an fMRI study accomplished in block-design. In addition, EEG data was also recorded. Genetic variation is analyzed using Affymetrix Genome-Wide Human SNP Arrays 6.0 encompassing more than 900,000 SNPs and additional copy number probes. For each of the three tested paradigms, networks of brain regions relevant for processing these controlled tasks have been described in literature. This processing seems to occur differently in dyslexia. IGPs are analyzed for each of these three performed controlled tasks.

Results: Analysis of data of the phonological task could confirm differences in processing between dyslexics and controls within a network consisting of the inferior temporal, inferior frontal, and the superior temporal region. The study is still in progress. Analysis of imaging geno-phenotypes will comprise all genetic data but will also focus on genomic regions known to be related to dyslexia (DYX 1-9, total size 200-300 Mb). Comparison of IGP resulting from the auditory, visual, and phonological tasks may reveal differences in the genetic basis of these three dyslexia relevant aspects.

Conclusion: IGP offer a promising new approach for identification of new candidate genes. We are adopting this approach in order to identify new candidate genes in dyslexia. Acknowledgment: This work was partially funded by „Interdisziplinaeres Zentrum fuer Klinische Forschung, Jena (Teilprojekt 1.2.“, ) of University Jena, Germany

A7. Title: Distinct Genetic Influences on Cortical Surface Area and Cortical Thickness

Authors: Matthew S. Panizzon, Christine Fennema-Notestine, Lisa Eyler, Terry L. Jernigan, Elizabeth Prom-Wormley, Michael Neale, Kristen Jacobson, Michael Lyons, Michael Grant, Carol Franz, Hong Xian, Ming Tsuang, Bruce Fischl, Larry Seidman, Anders Dale, William Kremen

Aim of Investigation: Structural neuroimaging studies of aging and neuropsychiatric disorders, as well as genetically informative studies of cortical structure have largely been based on measures of cortical volume. Given that cortical volume is a product of thickness and surface area, it is plausible that measures of volume capture at least two distinct sets of genetic influences. The columnar organization of the cortex itself suggests that this may be the case. Specifically it has been argued that the size of the cortical surface area is driven by the number of columns, whereas cortical thickness is influenced by the number of cells within a column. The present study aims to examine the genetic relationships between measures of surface area and thickness.

Methods: Participants were 404 men in the Vietnam Era Twin Study of Aging (VETSA; 110 monozygotic pairs, 92 dizygotic pairs). Age range was 51 to 59 years. MRI measures of cortical thickness and surface area were determined using Freesurfer software. Bivariate twin analyses were utilized in order to estimate the heritability of surface area and cortical thickness, as well as their degree of genetic overlap. Analyses examined the relationship between: 1) global and lobar measures of cortical surface area and average cortical thickness; and 2) surface area and thickness across 20 specific regions of interest (ROIs).

Results: Total cortical surface area and average cortical thickness were both highly heritable (.89 and .81, respectively), but were essentially unrelated genetically (genetic correlation = .08, ns). This pattern was similar at the lobar and ROI levels of analysis. Heritability estimates for surface area and thickness were roughly consistent within ROIs; however, significant genetic correlations were not observed. Within the narrow age range of this sample, age had no effect on global or lobar measure of surface area or thickness.

Conclusions: These results demonstrate that cortical volume measures combine at least two distinct sources of genetic influences: those affecting thickness and those affecting surface area. We conclude that using volume as an endophenotype, as is commonly done, may confound the underlying genetic architecture of the brain, making it more difficult to elucidate the genetic influences on aging and neuropsychiatric illness.

A8. Title: The dynamic role of genetics on the relationship between intelligence and cortical thickness during pediatric neurodevelopment

Authors: J. Eric Schmitt, Gregory L. Wallace, Rhoshel Lenroot, Philip Shaw, Liv Clasen, Jonathan Blumenthal, Kenneth S. Kendler, Michael C. Neale, Jay N. Giedd

Prior analyses using the genetically informative NIMH pediatric twin imaging sample demonstrated that the heritability of cortical thickness (CT) in the pediatric brain is both dynamic and heterogeneous, with superior and medial prefrontal cortex exhibiting high heritabilities that increase over childhood. Interestingly, these regions have been shown to have statistically significant group differences in developmental trajectories depending on IQ in a large longitudinal study of typical pediatric neurodevelopment (Shaw et al, 2006). In the present study, we examined the interactions between genetics, age, cortical thickness, and IQ in a large (N=600) sample of twins, siblings, and singletons ages 5-18 using anatomic magnetic resonance imaging (MRI) and structural equation modeling. Preliminary bivariate analyses of 54 cortical regions of interest (ROI) demonstrated negative additive genetic and positive environmental correlations between IQ and CT. We subsequently generated an ROI encompassing the significant voxels from Shaw et al. (ROIx) in order to test the hypothesis that genetics play a substantial role on the relationship between CT age trajectories and IQ. Univariate genetic modeling found that the heritability within ROIx (.66, 95% CI [.34, .75]) was substantially higher than outside of it (.39 [.07, .55]). Interaction models were then constructed in which moderators allowed for age, IQ (dichotomized into low or high IQ groups by the median IQ of 110), and their interaction to influence the contributions of genetic and nongenetic factors on cortical thickness. Within ROIx, strong gene by age interactions were observed in the high IQ group, with no interactions observable in the low IQ group. This interaction was statistically significant (chi-square = 9.8, p < .002). The genetically mediated CT variance in the higher IQ group was dynamic, beginning high in early childhood and decreasing over time, while genetic variance began lower in the low IQ group and remained static over the age range of the sample. Although further large longitudinal studies are required for confirmation, the present study suggests that genetic factors dominate the relationship between age-related changes in CT and intelligence during childhood and adolescence.

A9. Title: Dopamine DRD2 polymorphism alters reversal learning and associated neural activity.

Authors: Jocham, Gerhard, Klein, Tilmann A., Neumann, Jane, von Cramon, D. Yves, Reuter, Martin, Ullsperger, Markus

Aim: In humans, presence of an A1 allele of the DRD2/ANKK1-TaqIa polymorphism is associated with reduced expression of dopamine (DA) D2 receptors in the striatum. Recently, it was observed that carriers of the A1 allele (A1+ subjects) showed impaired learning from negative feedback in a reinforcement learning task.

Methods: Here, using functional MRI, we investigated carriers and non-carriers of the A1 allele while they performed a probabilistic reversal learning task.

Results: A1+ subjects showed subtle deficits in reversal learning. In particular, this deficit consisted of an impairment in sustaining the newly rewarded response after a reversal and in a generally decreased tendency to stick with a rewarded response. Both genetic groups showed increased fMRI signal in response to negative feedback in the rostral cingulated zone (RCZ) and anterior insula. Negative feedback that incurred a change in behaviour additionally engaged the ventral striatum and a region of the midbrain consistent with the location of dopaminergic cell groups. The response of the RCZ to negative feedback increased as a function of preceding negative feedback. However, this graded response was not observed in the A1+ group. Furthermore, the A1+ group also showed diminished recruitment of the right ventral striatum and the right lateral orbitofrontal cortex (lOFC) upon reversals.

Conclusion: Taken together, these results suggest that a genetically driven reduction in DA D2 receptors leads to deficient feedback integration in RCZ. This in turn was accompanied by impaired recruitment of the ventral striatum and the right lOFC upon reversals, which might explain the behavioral differences between the genetic groups.

A10. Title: Anxiety and empathy in Williams Syndrome: An fMRI study of brain responses to emotionally charged images and faces

Authors: Tricia A. Thornton-Wells, Chris Cannistraci, Mariam Eapen, Adam Anderson, John C. Gore, Elisabeth M. Dykens

Background. Williams syndrome (WS) is a genetic, neurodevelopmental disorder marked by hypersociability, decreased fear of strangers and heightened empathy, along with anticipatory anxiety and non-social fears. The dissociation between social and non-social fears in WS makes the disorder a good model for studying the neural correlates of these phenomena.

Methods. In a block design fMRI study, young adults with WS (n=13) and age- and gender-matched controls (n=13) passively viewed images (positive, negative or neutral) and faces (happy, sad or angry). Analyses employed random effects general linear models.

Results. When viewing negative images, adults with WS showed activation primarily in the right amygdala, whereas in controls, the amygdala activations were more bilateral (p < 0.005). In direct contrast, when viewing angry faces, participants with WS showed bilateral activation of the amygdala, whereas in controls the pattern of activation was stronger in the right amygdala (p < 0.01). In response to viewing happy and/or angry faces, WS individuals showed bilateral activation of a region spanning posterior cingulate, precuneus and anterior cuneus; whereas controls showed a decrease in activation of this same region versus baseline (between-group p < 0.0053).

Discussion. Group differences in amygdala activation to viewing threatening faces and images may be associated with heightened non-social anxiety in WS. Group differences in activation of the posterior cingulate / precuneus / cuneus region might be related to the empathetic nature of WS individuals. This region has been implicated in empathy and “theory of mind” tasks and is thought to be involved when stimuli have high emotional content. By investigating the genetic pathways and neurobiological processes of empathy in persons with WS, we can also learn about the typical development of this socially adaptive skill, with implications for other disorders, such as autism and schizophrenia, in which it is disrupted.

A11. Title: Changes in Three Major Frontotemporal White Matter Tracts in Schizophrenia Across the Adult Lifespan: A Diffusion Tensor Tractography Study

Authors: Aristotle N Voineskos, Nancy J Lobaugh, Sylvain Bouix, James L Kennedy, Benoit H Mulsant, Bruce G Pollock, Martha E Shenton

Aim of Investigation: The progression of white matter deficits and their relationship with age in schizophrenia is not well understood. Fronto-temporal white matter tracts represent an important susceptibility network in schizophrenia. Here, we use a tractography based approach, that permits measurement of properties of the diffusion tensor along white matter tracts. The present study examines three frontotemporal tracts across the adult lifespan in schizophrenia patients and in healthy controls.

Methods: 30 individuals with schizophrenia or schizoaffective disorder and 22 healthy controls ranging from 25-78 years old received one DTI scan as part of a comprehensive structural imaging protocol. Groups were matched on parental socioeconomic status, gender, and age, and all subjects were right-handed. Diffusion gradients were applied in 23 non collinear directions, with two b=0 images, whole brain coverage, and the sequence repeated three times, on a 1.5 T GE system. Following co-registration, whole brain seeding and tractography, the uncinate fasciculus (UF), inferior occipito-frontal fasciculus (IOFF), and cingulum bundle (CB) were segmented. Fractional anisotropy, trace, radial, and axial diffusivity values were then calculated.

Results: Using Pearson correlation coefficients, left side fronto-temporal tract FA for each of UF and IFOF and CB were significantly negatively correlated with age in schizophrenia (r= -.376, p=.019, r = -.498, p=.006 and r= -.341, p=.05 respectively) and in healthy controls (r= -.428, p=.004, r= -.398, p=.012, and r=-.375, p=.04 respectively). While there were no differences between the late life groups, the early adult life schizophrenia group showed decreased FA in the left uncinate fasciculus (p=0.035) and right cingulum bundle (p=.028), and decreased axial diffusivity in left cingulum bundle (p=.029) and increased radial diffusivity in right inferior occipito-frontal fasciculus (p=.031) when compared to the early adult life control group.

Discussion: Schizophrenia has been proposed as a dysmyelinating disorder. There is growing evidence that abnormalities of white matter integrity are progressive in schizophrenia, rather than occurring entirely prior to the first episode of psychosis. Our DTI tractography study across the adult lifespan in schizophrenia presents evidence that there are differences in white matter integrity in schizophrenia compared to controls, and that these differences are primarily detectable in the first half of adult life. Damage to fronto-temporal white matter tracts may not be progressive in late life in schizophrenia compared to controls. Other white matter tracts that we have not yet studied may continue to be susceptible to change in late life.

A12. Title: Solutions for the bioinformatic challenges of a large European multi centre genetic and imaging study

Authors: JB Poline, A. Barbot, Y. Schwartz, B. Thyreau, A. Strand Vestbø, A. J Klaassen

IMAGEN is a very large European Research Project which aims to identify and learn more about biological and environmental factors that might have an influence on mental health in teenagers. To this aim, the consortium is acquiring T1, T2, T2* and diffusion images for more than 2000 subjects at eight neuroimaging centres. A battery of behavioural tests have been set up, numbering up to 5 hours of testing, while blood samples are collected and the data will be processed to obtain 1M SNP per subject. At the time of writing of this abstract, data for more than 200 subjects had already been acquired across the 8 centres. The bioinformatics aspects of such a large and complex multi centric study are crucial for its success. These components face challenges for which no definite or complete solutions exist. Specifically, we have proposed solutions concerning the logistics of collecting the data. Our system includes data anonymisation, data security and safety, data transfer, and local quality checks using tools from two SMEs (Scito, France; NordicImagingLab, Norway). Secondly, we have focused on how the collected data should be organised to meet the security and functional requirements of IMAGEN. We have chosen to base our central database on the XNAT tools and we describe the database technology and ontology issues for the project. This integrated solution allows the use of a postgresql database for queries through a web interface with an appropriate level of security. Thirdly, we review the possible solutions for applying complex processing to a large amount of multimodal data, as well as allowing distant processing. Our current system uses processing tools such as SPM, FLS, Brainvisa, Nipy. The processing outputs are stored in the central database and serve as a second level of quality check. This implies that information about the processing is also stored in the database, and therefore has a representation in the ontology. The combination of neuroimaging variables such as the Bold activity in a brain ROI with genetic information (SNP) should permit a powerful analysis of the central database by the members of the consortium. Acknowledgements: Support was provided by the IMAGEN project, which receives research funding from the European Community's Sixth Framework Programme (LSHM-CT-2007-037286). This manuscript reflects only the author's views and the Community is not liable for any use that may be made of the information contained therein.

A13. Title: Glycogen synthase kinase-3beta polymorphisms are associated with brain structural changes in major depressive disorder

Authors: Becky Inkster, Thomas E Nichols, Philipp G. Saemann, Dorothee Auer, Florian Holsboer, Pierandrea Muglia, Paul M. Matthews

Aim of Investigation: Indirect evidence suggests that glycogen synthase kinase-3beta (GSK3β) might be implicated with major depressive disorder (MDD). The aim of this study was to evaluate 15 GSK3β single nucleotide polymorphisms (SNPs) to test for association with regional grey matter (GM) volume differences in a well-characterised group of recurrent MDD patients. We then tested for MDD-by-genotype interactions within the masks derived from significant association clusters in the MDD group and age-, sex- and ethnicity-matched healthy controls to identify structural variation distinguishing the MDD group

Methods: High resolution T1-weighted MR images were collected for recurrent MDD patients and healthy controls at the Max Planck Institute of Psychiatry, Munich, Germany. All GSK3β SNP genotypes were extracted from genotyping obtained using Illumina’s HumanHap550 BeadChip platform. After several quality control procedures were performed, 134 recurrent MDD patients and 143 age-, gender-, and ethnicity-matched healthy controls were included for analysis. We used general linear model (SPM5) with non-stationary cluster-based inference (thus allowing for valid inference on clusters while accounting for heterogeneous smoothness) to measure associations between GSK3β polymorphisms and regional GM volume differences. A separate analysis was performed for each SNP. Bonferroni correction was applied for testing multiple GSK3β SNPs.

Results: GM variation was associated with GSK3β polymorphisms: the most significant associations were found for rs6438552, a putative functional intronic SNP, which showed 3 significant GM clusters in the right and left superior temporal gyri (STG) and the right hippocampus (p=0.00003, p=0.021 and p=0.018 respectively, corrected for multiple comparisons across the whole brain). Similar results were obtained with rs12630592, a SNP in high linkage disequilibrium. A significant SNP-by-MDD status interaction was observed for the effect on GM volumes in the right hippocampus and STG (p=0.0001 and p=0.013, corrected, respectively).

Conclusions: This study supports the hypothesis that GSK3β may have a role in determining regional GM volume differences of the right hippocampus and bilateral STG. The association between genotype and brain structure was specific to the MDD patients, suggesting that GSK3β genotypes might interact with MDD status. We speculate that this is a consequence of regional neocortical, glial or neuronal growth or survival. In considering core cognitive features of MDD, the association of GSK3β polymorphisms with structural variation in the temporal lobe and hippocampus is of particular interest in the context of other evidence for structural and functional abnormalities in the hippocampus of MDD patients.

A14. Title: The Association of NRG1 rs3924999 Genotype with Lifetime History of Psychosis and Regional Grey Matter Volume in Bipolar Affective Disorder

Authors: Robert K. McClure, M.D., Jair Soares, M.D., Giovana Zunta-Soares, M.D., Michael A. Escamilla, Ph.D, Consuelo Walss-Bass, Ph.D, John P. Hatch, Ph.D, Charles Bowden, M.D.

The goal of this study was to examine the association of the NRG1 5’ promoter SNP rs3924999 A genotype, with lifetime history of psychosis (LHP), dorsolateral prefrontal cortex, and hippocampal gray matter volume in subjects with Bipolar Affective Disorder (BPAD). 110 adult and adolescent subjects with BPAD were used in the LHP phenotype analysis. 62 of 110 adults with BPAD and adequate MRI data were examined with voxel- based morphometry. Genotyping was performed with standard methods. Ethnicity was examined as a potential confound. Consistent with a priori hypothesise, the A genotype was more commonly present in bipolar subjects with a LHP. Contrary to a priori hypothesise, the A genotype was not associated with reduced prefrontal or hippocampal grey matter volume. Exploratory analysis showed an association of the G genotype with reduced right posterior cingulate gyrus grey matter volume. Although the NRG1 rs3924999 A genotype may convey increased for psychosis in bipolar disorder, replication in a larger sample with careful attention to population stratification is necessary. Further analysis of additional subjects recruited from this population is ongoing.

A15. Title: Genetic association with imaging phenotypes: A simulation study

Authors: Maria Vounou, Giovanni Montana, Thomas Nichols, Brandon Whitcher

Aim of Investigation: We propose and evaluate two methods for performing Genome Wide Association studies in the imaging genetics framework. The ultimate goal of our work is to find a combination of genes that have an effect on a common complex neurological disease. Due to the 100’s of 1,000’s of SNPs in the human genome such studies present a daunting statistical challenge. While there has been a growing literature of single locus associations with brain imaging phenotypes, such methods face some drawbacks due to the massive amount of tests that are required. In addition to this, modelling each SNP one at a time ignores any possible structure within the genome, which may result in less power to detect true associations. On the other hand, a multilocus association analysis where each SNP’s effect is tested for association with the disease taking into account the effect of the remaining SNPs, is expected to provide more accurate results. To perform multilocus association analysis with quantitative traits, such as the imaging data, ideally we could use a single linear model with say, 100,000 SNP regressors but this is infeasible since we have many fewer subjects than SNPs. To overcome this problem, we propose the use of regularized regression methods for variable selection. These incorporate shrinkage in regression by imposing constraints on the size of the coefficients. One particular such method implemented for this work is the ‘Least Absolute Shrinkage and Selection Operator’ (LASSO) which induces an L1 penalty on the coefficients of the regression model. The L1 penalization performs a form of variable selection as it shrinks the regression coefficients and forces some of them to be equal to zero. Such models could handle a large number of SNPs for association analysis with the disease and produce a sparse, interpretable model.

Methods: In this initial work, we concentrate on Alzheimer’s disease and use data form the Alzheimer’s disease Neuroimaging Initiative (ADNI) project. We have developed a simulation design to assess the performance of the proposed method and compare its results to the single locus tests using linear models. We reduce the dimensionality of the imaging data by recording the mean grey matter intensity values over 10 regions of interest (ROI) and remove possible non-genetic effects by regressing on the variables ‘age’ and ‘sex’. Assuming multivariate normality for the residuals and estimating their mean and covariance, we consider the corresponding multivariate normal distribution as our baseline distribution to represent the healthy status. A sample of ‘realistic’ genotype data is simulated using FREGENE software, incorporating effects like selection and recombination. Fixing 2 ROIs to represent the diseased ROIs and 3 SNPs to represent the causative SNPs, the phenotype of an individual is simulated from the baseline multivariate normal distribution incorporating an additional signal on the mean value of the diseased ROIs. This signal depends on the realizations of the causative SNPs for that individual. In this manner, we create a genetic effect on the diseased ROIs while keeping the original correlation structure of the ROIs fixed. The causative SNPs are then removed from the sample and the two methods are applied. The assessment of their performance depends on their ability to pick a SNP in LD with a causative SNP whilst controlling for the false positives.

Results: Repeating the method for different genetic effect sizes, some preliminary results illustrated that the proposed method outperformed linear models when the genetic effect was reduced to a 10% change on the mean value of the diseased ROIs.

Conclusions: Our simulation design will further be used to implement a Monte Carlo study and more formally assess the power of each method. An additional important aspect to examine would be to consider the performance of the proposed method when varying different parameters of interest, such as sample size, minimum allele frequencies for the causative SNPs and number of ROIs. Additional work for this project will involve comparison of other penalization methods such as the elastic net penalty and multivariate techniques that allow the response to be multivariate as well. Two such methods are Partial Least Squares and Canonical Correlation Analysis.

Acknowledgements: Disclosure: TE Nichols is an employee of GlaxoSmithKline, UK.

A16. Title: A novel parallel ICA approach to investigate differential linkage between brain function and genetic profiles in a cohort of control and schizophrenia patients

Authors: Shashwath A Meda, Jingyu Liu, Joel Gelernter, Vince D Calhoun,Godfrey D Pearlson

Aim of Investigation: Understanding genetic influences on both healthy and disordered brain function is a major focus in psychiatric neuroimaging. We utilized task-related imaging findings from an fMRI auditory oddball task as potential intermediate phenotypes (endophenotypes) to investigate genomic factors derived from multiple single nucleotide polymorphisms (SNP’s) from genes previously shown to be associated with schizophrenia array. Our major aim was to investigate the linkage of these genomic factors to normal/abnormal brain functionality between a group of controls and schizophrenia patients.

Methods: We studied 64 healthy controls and 43 schizophrenia patients (matched on age, ethnicity and gender). All subjects performed an auditory oddball task, which consists of detecting an infrequent sound within a series of frequent sounds. Each subject was characterized on 23 different SNP markers spanning different risk genes previously associated with schizophrenia. We used a recently developed technique named parallel independent component analysis (paraICA) to analyze this multimodal dataset (Liu et. al. 2007). The method aimed to identify simultaneously independent components of each modality (functional imaging, genetics) and the relationships between them. Scans were acquired at the Olin Neuropsychiatry Research Center at the Institute of Living on a Siemens Allegra 3 T dedicated head MRI scanner equipped with 40 mT/m gradients and a standard quadrature head coil, using gradient-echo echo-planar-imaging with the following parameters (repeat time = 1.50 s), echo time = 27 ms, field of view = 24 cm, acquisition matrix = 64 × 64, flip angle = 70°, voxel size = 3.75 × 3.75 × 4 mm3, slice thickness = 4 mm, gap = 1 mm, 29 slices, ascending acquisition). Data were preprocessed using SPM2 running on Matlab 7.1, which involved realignment, normalization and smoothing with a 12 mm3 Gaussian kernel. First-level statistics were run within the GLM framework to derive contrast images for subjects responding to auditory oddballs. These images were then carried over as the input to the parallel ICA algorithm. The number of independent fMRI and genetic components were estimated to be 5 and 4 respectively using the minimum description length (MDL) criteria.

Results: We found 3 fMRI components significantly correlated with two distinct gene components. The fMRI components, along with their significant genetic profile (dominant SNP) correlations were as follows: 1) PCC-ACC-Cuneus-Lingual/Fusiform-Cerebellum & DrD2-CHRNA7 (Cholinergic receptor rs2337506 & rs868437) [r=-0.25; p=0.01], 2) STG-IPL-SPL-DLPFC-Cingulate & SLC6A4_PR (Seratonin transporter) [r=0.21; p=0.04], 3) Default Mode-STG & SLC6A4_PR (Serotonin transporter). Functional components comprised task-relevant regions (including DLPFC, ACC, STG, IPL) also identified as abnormal in schizophrenia. Further, gene-fMRI combinations 1 (z=2.62; p=0.004) and 2 (z=2.99; p=0.001) listed above showed a significant difference between controls and patients, based on their correlated loading coefficients.

Conclusion: In summary, we constructed and demonstrated a framework to identify interactions between brain functional and genetic information; our findings provide a proof-of-concept that genomic SNP factors can be investigated by using endophenotypic imaging findings in a multivariate format. We thank the research staff at the Olin Neuropsychiatry Research Center who helped to collect and process the data. This research was supported by the National Institutes of Health, under grants R01 EB005846 & 1 R01 EB006841 (to VDC), and 2 RO1 MH43775, 5 RO1 MH52886 (to GP) and a grant from the MIND Institute (NPB).

A17. Title: The role of 5-HTTLPR polymorphism on cingulate volumes in pediatric bipolar and unipolar disorders: Preliminary findings

Authors: Paolo Brambilla, Rene Olvera, Mark Nicoletti, Hongtu Zhu, John P. Hatch, Consuello Walss-Bass, Michael Escamilla, Jair C. Soares

Introduction: The serotonin transporter gene has a polymorphism within the promoter region (5-HTTLPR) with two allelic forms, the long (L) and the short (S) and is a candidate to be associated with affective disorders. Also, the cingulate cortex is a key structure modulating emotion in humans and plays a major role in affective disorders. However, to our best knowledge, the potential effects of 5-HTTLPR polymorphism on cingulate anatomy have not been explored in affective disorders. In this perspective, a pediatric population is of particular interest, as early onset cases are likely to involve more pronounced biological abnormalities.

Methods: A case-control study was performed to compare 5-HTTLPR genotype between 39 pediatric affective disorder patients (mean age±S.D=13.52±2.96; 14 females; 23 bipolar disorder, 11 unipolar disorder) and 22 healthy controls (mean age±S.D.=13.55±2.61; 15 females). Genotyping was performed using the Applied Biosystems Taqman Allelic Discrimination assay (Foster City, California). Brain images were collected on a Philips 1.5 T MR system. Statistical analysis was carried out using linear regression (PROC GLM) and the F-test was used. Genotype, diagnostic status, gender and ICV were used as covariates.

Results: Significant effects of group by genotypes (S-carriers vs LL) have been found for anterior caudal (F=6.80, p=0.011) and posterior cingulate volumes (F=12.80, p=0.0007) in the right side. Within the subgroups, our results show that both S-carrier (N=34) and LL-genotype patients (N=5) have significantly reduced volumes in comparison to LL-genotype controls (n=2) for both anterior caudal (t=2.41, p=0.019, t=2.33, p=0.023) and posterior cingulate volumes (t=3.24, p=0.0019; t=3.48, p=0.00098).

Conclusions: Our findings, although preliminary due to the small sample size, suggest that the volumes of anterior and posterior cingulate in pediatric affective disorder patients are independent of the 5-HTTLPR polymorphism, possibly being more affected by environmental/psychosocial stressors or other genes. Future studies should further investigate the gene-environment modulation on cingulate volumes in children with affective disorders, also trying to identify potential differences between unipolar and bipolar patients.

Acknowledgments: This research was partly supported by MH 69774, MH 068662, RR 20571, and the Krus Endowed Chair in Psychiatry (UTHSCSA).

A18. Title: Optimal Multiple SNP, Whole Brain Association Methods with Exact Control of False Positives

Authors : Thomas E. Nichols, Becky Inkster, Paul M. Matthews

Aim of Investigation Single Nucleotide Polymorphisms (SNPs) and high-throughput genotyping methods have provided a wealth of data for investigators looking for genetic associations with brain structure or function. Often a particular gene is of interest, and perhaps even a specific SNP within the gene is known to change protein structure. For most genes, however, there is no single SNP known to change cell function or risk of a disease. In such cases multiple SNPs within or near a gene must be examined. If each SNP is simply analyzed separately there is a growing multiple testing problem, and, due to correlation induced by linkage disequilibrium (LD), a Bonferroni Correction can be quite conservative. Instead, we consider five different statistic combining methods, where a statistic map of joint evidence of association is produced. Since LD-induced dependence between SNPs invalidates standard parametric results, we use permutation to find exact resampling P-values for each method, both uncorrected and corrected for searching over space.

Methods We compare: Multiple regression, Fisher's method, Stouffer's method, Max F-stat (equivalently, Min P), and the conjunction Min F-stat (Max P) method. Each method is relatively more sensitive to a particular type of signal; for example, Fisher's or Stouffer's method are best when several SNPs have a signal, while the Max F method is best when only one SNP has an effect; the conjunction is selectively sensitive to the case when all SNPs show an effect. To evaluate the methods we compare voxel-wise and cluster-wise significance on the combining statistic maps, with cluster-wise significance adjusted for non-stationarity found with VBM data. We evaluated these methods on a 278 subject VBM dataset of healthy controls and depressed patients. We examined four genes, WNT3a (11 SNPs), KRM1 (12 SNPs), DVL2 (3 SNPs) and ZEB2 (25 SNPs).

Results On all genes the Max F test assessed voxel-wise had smallest corrected P-values on 3 the genes, and Max F assessed cluster-wise was best on the 4th (DVL2). On all genes except DVL2, cluster size was most sensitive when used with Fisher's or Stouffer's method. Counts of uncorrected P-values less than 0.9, when they differed at all, were greatest for Max F and Fisher's.

Conclusions To our knowledge this is the first application of formal SNP combining methods to whole brain genetic association studies. These evaluations suggest the preferred way to combine imaging data on multiple SNPs within genes is with the Max F combining method. Further evaluation is needed with more genes, and with combining SNPs from different genes.

A19. Title: MECP2 Variants Associate with Measures of Brain Size and IQ

Authors: Alexander H. Joyner, Cooper Roddey, Anders Dale, Ole Andreassen, Nicholas Schork, Cinnamon Bloss, Trygve Bakken

Abstract:  Recent attempts to link brain structure, genetics, and cognition have suffered from inconsistent and unreplicated results.  Here we study an x-chromosome gene, MECP2, chosen for its involvement in a wide variety of neurodevelopmental disorders and its effect on both brain size and cognition.  Using structural measures of brain size and genetic (SNP) data we show that common MECP2 variants associate with measures of intracranial volume and cortical area, among other regions, in a population of schizophrenic, bipolar, and control subjects. The minor allele of SNP rs3027898 associated with reduced intracranial and brain volume (simulated p-val = 6.0x10-4) and cortical area (simulated p-val = .01), the most significant primary results. Secondary analysis showed that basal ganglia, cerebellum, and hippocampus all showed association with reduced volume among those with the minor allele.  We did not detect any gene-by-diagnosis interactions.  Cortical thickness did not associate with any variant, an interesting result given recent evidence for independent determinants of cortical area and thickness. Interestingly, there exists a striking gender difference, with females showing no association with any tested brain region.   We also examine the association of common MECP2 polymorphism on neurocognitive measures, and find a mild association (simulated p-val = .01) in males for phonetic fluency.  The recent association of MECP2 variants with autism, which exists primarily in boys, coupled with our finding that MECP2 variants associate with brain size measurements in males but not in females, raises interesting questions about the gender specific mechanism by which X-linked genes affect human brain development.   

A20. Gyrification and brain size: their genetics and relationship. A translational genetic imaging study in human and baboons

Kochunov P 1, Glahn D 1, Fox P 1, Zilles K 3, Shelledy W 2,3, Lancaster J., Thompson P 4, Blangero DJ 2,3 and Rogers J 2,3.

Introduction: Mammalian evolution produced larger, more gyrified brains. In mammals and especially in primates, phylogenetic factors caused an exponential increase in the number of neurons with a corresponding increase in brain size and gyrification (Figure 1). The across-species analysis shows that evolutionary increases in brain size and gyrification are highly and positively correlated(r=~0.99). However, within-a-species, two traits are uncorrelated, indicating a difference in phylogenic and ontogenic strategies. We studied the genetics of gyrification and brain size and the pleiotropy between two traits by performing genetic correlation. We performed this analysis using translational genetic imaging approach employing two evolutionary related species: humans and baboons.

Methods: Subjects: 193 healthy adult human volunteers consisting of (78/115 males/females), average age 48±18y/o and 180 adult baboons (68/112 males/females), average age of 16.0 ± 4.2 y/o were selected from their respective multi-generational pedigrees managed by SFBR. The genealogical relationship for each group were known and resulted in about equal power of statistical analysis.Allometric MR imaging protocol. Human and animal subjects were imaged using a 3-D-motion-corrected protocol structural MR imaging protocol optimized for high gray matter/white matter contrast. Spatial resolution was adjusted to obtain allometeric sampling in each group. Humans were imaged with the isotropic resolution of 800 m m, while animals were imaged with isotropic resolution of 500 m m. Measurement of gyrification index . Measurements of cerebral gyrification were performed from cortical surfaces extracted by BrainVisa. We adopted the classical 2-D GI methodology for GI measurements from 3-D surfaces by taking the ratio between the area of a gyrated surface and the area of its convex hull while excluding the area of the lateral ventricles (Figure 2). Two gyrification index (GI) values were calculated for each hemisphere: the GI out describing the gyrification of the external CSF/GM surface and the GI out describing the gyrification for the internal WM/GM interface. The values of GI out and GI in were calculated for each hemisphere and then averaged. Genetic analysis. Variance components methods, implemented in the SOLAR computational genetic package , were used to estimate the heritability of GI and brain volume and to perform genetic correlation between them.

Results: Heritability analysis of brain volume and cortical gyrification (GI out and GI in) reported highly significant and numerically-similar findings in both groups. Heritability of the brain volume was h 2 bv=0.74(p<1e-4) for humans and h 2 bv=0.81(p<1e-4) for baboons. Heritability of GI in humans was h 2 GI out=0.41; h 2 GI in=0.31 and in baboons it was h 2 GI out=0.44±0.20; h 2 GI in=0.29. Genetic correlation analysis reported a negative genetic correlation between brain volume and gyrification in both species. The correlation was significant (p<0.05) for GI in in humans and GI out in baboons and approached significance in GI out in humans (p=~0.09).

Discussion : We discovered that the both gyrification and brain volume are highly heritable and observed significant and negative pleiotropy between them. The later finding hints that gyrification could be a produced by a conflict between phylogenic and ontogenic strategies. While phylogenic factors favor bigger brains capable of accommodating more neurons, ontogenic factors favor smaller, less energy expensive brain. Negative genetic correlation between gyrification and brain size hints that increases in gyrification could be an ontogenic compromise to a phylogenic drive for increase in brain size.

A21. Genetics of cerebral aging. Heritability and linkage analysis of cerebral health markers in a large multifamily pedigree.

Kochunov P, Glahn DC, Cole S, Dyer TD, Almasy L, Winkler A, Lancaster JL, Fox PT, Kochunov V., Blangero J

Introduction: Individual genetic differences may be the source of many, if not all detrimental changes with aging. We studied genetics of commonly-used structural neuroimaging-based indices of declining cerebral integrity in normal aging. Three measurements of cortical (gray matter (GM) thickness, sulcal and intergyral spans) and two measurements of subcortical (DTI-derived fractional anisotropy (FA) and volume of hyperintense WM (HWM) lesions) cerebral integrity measured for a large, multi-family pedigree were subjected to genetic analysis. We analyzed phenotypical heritability, degree of shared genetic variance (pleiotropy) and possible gene linkage.

Methods:Subjects: 287 healthy adult human participants (107/187 males/females; average age 48±18y/o) in the Genetics Of Brain Structure (GOBS) study.

Antomical MR imaging protocol. Subjects were imaged on a 3T Siemens Trio scanner using the following protocol:

Anatomical T1 images were obtained using a 3-D-motion-corrected protocol structural MR imaging protocol optimized for high gray matter/white matter contrast with an isotropic resolution of 800 mm. Diffusion weighted images (DWI) were acquired using single-shot spin-echo, echo-planar gradient recalled echo, T2-weighted sequence was used to acquire diffusion weighted data with the spatial resolution of 1.7x1.7x3.0mm with 55 diffusion weighted directions. The T2-weighted data was acquired with a high-resolution (isotropic 1mm), 3D turbo-inversion recovery FLAIR with a non-selective inversion recovery pulse to prevent CSF pulsation artifacts

Data processing. Anatomical T1 images were processed to extract the following regionalized measurements of cerebral integrity: GM thickness, sulcal and gyral span (Figure 1). DWI data were processing using a Track Based Statistics Approach which provided Fractional Anisotropy measurements for individual white matter tracks. T2-weight data were manually traced to identify the volume of ventricular and core white matter lesions.

Results. Heritability analysis reported that all indices of cerebral integrity were significantly heritable with heritability values varying from 57% for HWM volume to 23% for intergyral span (Table 1). Genetic correlation analysis reported an intricate pattern of shared genetic variance among apparently disparate indices: sulcal span showed a significant degree of pleiotropy with FA (rhog=-0.49) and HWM volume (rhog=0.46).

Next, indices of cerebral integrity were subjected to a factor analysis to extract age-dependent and age-independent categories of cerebral aging. Age-depended category loaded on GM thickness, sulcal span and FA, while age-independent category loaded on HWM volume and intergyral span. Both categories were significantly heritable (Table 2) and showed a significant degree of pleiotropy (Rhog=0.65). Linkage analysis of individual measurements and factor scores identified several genomic regions harboring suggestive quantitative trait loci (QTLs) for the age-dependent category (LOD scores of 1.5-1.87) and one chromosomal location (chromosome 12q23) with significant evidence for a QTL (LOD score of 3.47) influencing the age-independent category.

Conclusion. The initial genetic analysis of indices of cerebral integrity provided insightful information regarding genetics of individual indices, pointed out at degree of shared genetics variance among them and identified prospective targets for more detailed analysis.

A22. Genetics of cerebral sulcation. Does genetics offer a new way of sulcal classification?

Kochunov P 1, Glahn D 1, Fox P 1, Cykowski M, Lancaster J., Zilles K 3 , Shelledy W 2,3, Coulon O, Blangero J 2,3 and Rogers J 2,3.

Introduction: Gyrogenesis of primary cortical structures begins in utero, and coincides with an explosive increase in cortical growth. Primary sulcal features appear first and are thought to be spatial landmarks and delimiters for highly differentiated functional subsystems. Knowledge of the genetic factors that are in control of sculpting external features of cerebral cortex is of fundamental importance to many aspects of neuroscience. In this study sulcal surface area and two linear phenotypes: sulcal length and depth were measured for eleven primary sulcal structures (Figure 1) and subjected to genetic analysis. The following a priori hypotheses of sulcal heritability were considered: dependence on structure’s ontogenic order and structure’s size and depth.

Methods: Animal subject. 180 pedigreed and genotyped adult baboons (Papio hamadryas) (68/112 males/females) average age 16.0 ± 4.2 y/o, were selected from a colony maintained by SFBR . The genealogical relationships included 414 parent-offspring pairs, 51 full sib pairs, 645 half-sib pairs.

Animal handling and MR imaging . Animal subjects were imaged using a 3-D-motion-corrected protocol structural MR imaging protocol optimized for high gray matter/white matter contrast, using an 8-element primate head coil. Animals were imaged with isotropic resolution of 500 m m, SNR of 25 and gray/white matter contrast of 25. The imaging protocol and all animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of the SFBR.

Measurements of sulcal area, length, depth. Phenotypical measurements of variability in cortical sulci were performed on sulcal surfaces generated by BrainVisa cortical segmentation pipeline. Sulcal area was calculated as the area of the sulcal surface. To measure sulcal length and depth, the sulcal surface was parameterized by solving a system of heat equations. The length of the sulcus was calculated as an average distance, along the sulcal surface, continuously collinear with the fundus and the top ridge. The depth of the sulcus was calculated as the average distance, along the sulcal surface, between the sulcal fundus and its top ridge.

Results: Genetic analysis of sulcal phenotypes reported a complex heritability pattern that varied by structure and phenotype, however, all sulci showed a statistically significant (p<0.05) heritability for at least one of the phenotypes. Testing of a priori hypotheses did not report significant findings. Heritabilities of sulcal area, depth and length were not significantly (p>0.05) correlated with structure’s ontological order, area or depth.

However, sulci known to separate functional areas (Central, Superior Temporal, Arcuate, Lunate and Cingulate), showed a significantly higher heritability in depth than in length (hd=0.60 vs. hl=0.25; p<0.01). Sulci known for being anchors and defining the size of functional areas e.g. Calcarine, Inferior Occipital, External Calcarine and Principal showed a significantly higher heritability in length than in depth (hl=0.56 vs. hd=0.32; p< 0.05).

Discussion. Our genetic analysis of sulcal morphology rejected the a priori hypotheses. However, a new finding has emerged. Sulcal structures known to divide disparate functional areas showed a much higher heritability in depth than in length. Sulcal structures known to define the size functional areas showed a much higher in length than in depth.

A23. The triallelic 5-HTTLPR polymorphism in the human serotonin transporter gene SLC6A4 may impact error-related activation in the medial frontal cortex in healthy subjects.

Agnes J. Jasinska 1, Emily R. Stern 2, William J. Gehring 3, Margit Burmeister 2,4,6, Kathrin Stossberg 6,7, Elżbieta Śliwerska 6, Gregory L. Hanna 2, Robert C. Welsh 5, Stephan F. Taylor 2.

Background: The low-functioning S allele (S or Lg) of the triallelic 5-HTTLPR polymorphism (5-HTTLPR & rs25531) in the human serotonin transporter gene SLC6A4 has been associated with anxiety-related personality traits, higher susceptibility to anxiety and mood disorders, increased amygdala reactivity to negative emotional stimuli as measured with functional magnetic resonance imaging (fMRI) and increased error negativity as assessed with event-related potentials (ERPs). Since experimental stress and anxiety have been shown to enhance error-related activation in the medial frontal cortex (MFC), we assessed the influence of the triallelic 5-HTTLPR polymorphism on this activation in healthy subjects using a monetary-incentive flanker-interference fMRI paradigm.

Methods: We report on preliminary data from 14 healthy subjects (all Caucasian; 6 females; mean age, 24.3, S.D., 5.6). Subjects were genotyped on the triallelic 5-HTTLPR polymorphism using standard procedures and performed a monetary-incentive flanker-interference task while fMRI BOLD was measured. After preprocessing, the functional data were analyzed at the subject level using fixed-effects models and at the group level using random-effects models, with the MFC as the a priori region of interest. At the subject level, the contrast of error – correct was used to measure the error signal. At the group level, two-sample t-tests were used to compare the error signal between the medium-functioning genotype group, i.e., carriers of 1 low- and 1 high-functioning allele (n = 6; La/S genotypes; 3 females; mean age 24.8, SD 4.2), and the low-functioning genotype group (n = 8; 1 Lg/S and 7 S/S genotypes; 3 females; mean age 23.9, SD 6.7). We had no high-functioning genotypes (La/La) in our sample.

Results: Both genotype groups showed relatively robust error-related activation along the medial frontal wall, but error detection elicited distinct patterns of activation in the two groups. The medium-functioning genotype group (La/S) showed a cluster of activation in the anterior/rostral MFC (cluster size k = 31; Z = 3.95 at 6, 48, 0), including bilateral medial frontal gyri. In contrast, the low-functioning genotype group (S/S or Lg/S) showed large clusters of activation localized more caudally along the medial frontal wall in the posterior medial frontal cortex (pMFC) (k = 68; Z = 3.72 at 6, -6, 57), including right supplementary motor area, left cingulate gyrus, and right medial frontal gyrus.

Conclusions: Despite small sample size, our preliminary data demonstrates the impact of the triallelic 5-HTTLPR polymorphism on human brain activity, further validating the imaging genomics approach to the study of human brain processes in vivo. Given the documented impact of the 5-HTTLPR on affective processing, the differences in regional patterns of MFC activation during error detection may be linked to different affective responsivity in the two genotype groups. However, the preliminary nature of the data and the absence of the LL genotype group warrant caution in interpreting the results.