The genetic overlap between Intellectual Disability and
Attention-Deficit/Hyperactivity Disorder
Anne van Pens
Master thesis, written in October and November 2014 in Nijmegen, the Netherlands Supervisors: Marieke Klein, Alejandro Arias-Vasquez, Barbara Franke
Affiliations:
1: Radboudumc, Nijmegen, The Netherlands 2; Donders Institute, Nijmegen, The Netherlands
Corresponding author: Anne van Pens, annevrens@hotmail.com
Abstract
Objective: Attention-Deficit/Hyperactivity Disorder (ADHD) and Intellectual Disability (ID) co-occur
more often than expected by chance, suggesting some genetic overlap. In four subprojects we investigated whether genes, affected by rare genetic variations in patients with ID, contribute to multifactorial ADHD.Methods: (1) Single nucleotide polymorphisms (SNPs) in 392 autosomal ID-related genes and (2) a subset involved in neurite outgrowth were tested for association with multifactorial ADHD risk, both on gene-set and gene-wide level, using data from the meta-analysis of the ADHD working group of the Psychiatric Genomic Consortium (PGC; 5, 621 cases and 13, 589 controls). (3) Twelve genes selected on frequent occurrence in copy number variants (CNVs) in patients with ADHD and ID and/orcongenital anomalies {PRODH, RBFOX1, PTPRD, CNTNAP2, NRXN1, XYLT1, PRIM2, FAM110C, SKI, NRG3, GRIN2A, NRG3 and ERBB4}and(4) two genes selected because of suggested involvement in ID and ADHD in the literature {CHRNA7 and NRXN1} were tested for gene-wide associations with multifactorial ADHD risk using the ADHD PGC meta-analysis data and with symptom counts using data from the International Multicenter ADHD Genetics project (IMAGE; 930 cases). Single-SNP and gene-wide association analyses for CHRNA7 and NRXN1 were performed with regional brain volumes in 1302 healthy participants of the Brain Imaging Genetics (BIG) cohort. Single-SNP association analyses were also performed forvoxel-wide structural connectivity measurements.
Results: SNPs in all autosomal ID-related genes, but not in the subset of neurite outgrowth genes, were significantly linked to ADHD as a group. The MEF2Cgene showed gene-wide association with ADHD risk. Other gene-wide and SNP-specific analyses did not yield significant associations.
Conclusion: SNPs in 392 genes, and specifically the MEF2Cgene, affected by rare genetic variations in patients with ID, contribute to multifactorial ADHD risk as a group. This contribution to ADHD risk does not seem to be driven by neurite outgrowth genes.
Keywords: Genetic Overlap, Attention-Deficit/Hyperactivity Disorder, Intellectual disability, Gene-set analysis, MEF2C, Brain Imaging
Introduction
Neurodevelopmental continuum
Neurodevelopmental disorders are often accompanied by developmental or psychiatric
comorbidities. For example, 68-87% ofAttention-Deficit/Hyperactivity Disorder (ADHD) patients have at least one co-morbid disorder (Ghanizadeh, 2009; Jensen & Steinhausen, 2014; Kadesjo & Gillberg, 2001; Kraut et al., 2013; Larson, Russ, Kahn, & Halfon, 2011). Because neurodevelopmental disorders share a common genetic etiology (Pettersson, Anckarsater, Gillberg, & Lichtenstein, 2013), it has been hypothesized that the same 'risk genes' often contribute to different
neurodevelopmental disorders (Moreno-De-Luca et al., 2013). Therefore, all neurodevelopmental disorders have been hypothesized to lie on a 'continuum', with a largely shared causality (Moreno-De-Luca et al., 2013; Owen, 2012). The model of developmental brain dysfunction by Moreno-de-Luca et al. (2013) predicts that each particular genetic cause can manifest as a spectrum of impairments of varying severity in the cognitive, neurobehavioral and neuromotor domain. What specific impairments reach the threshold for clinical diagnosis, however, depends on the genetic background of the individual and on environmental factors. This explains why patients with neurodevelopmental disorders often fulfill also part of the criteria for other neurodevelopmental disorders. For example, many studies support the idea that 30-50% of all children with autism spectrum disorder (ASD) also fulfill the criteria for ADHD (de Bruin, Ferdinand, Meester, de Nijs, & Verheij, 2007; Gadow, DeVincent, & Pomeroy, 2006; Leyfer et al., 2006; Schwenck & Freitag, 2014; Simonoffet al., 2008; Sinzig, Morsch, & Lehmkuhl, 2008; van Steenset, Bogels, & de Bruin, 2013). Martin et al. (2014) showed that there is substantial genetic overlap between the biological
processes ofASD and ADHD, as for example large rare copy number variations (CNVs) contributing to the disorders disrupt the same biological processes. This supports the neurodevelopmental
continuum hypothesis.
Overlap between ADHD and Intellectual Disability
Less well-studied than the overlap between ADHD and ASD, is the overlap between ADHD and intellectual disability (ID). The prevalence of ADHD in patients with ID seems to be twice as high as in the general population (Franke et al., 2012; Maulik, 2010). The real prevalence ofADHD in ID-patients may be much higher, though, because ADHD can be difficult to diagnose in children with ID. This is because ID-patients may not understand the questions of a psychiatric interview meant for people with a normal intelligence quotient (IQ) (Turygin, Matson, & Adams, 2014). In addition, one must take into account the patient's developmental age, rather than the biological age to assess if the observed hyperactive, impulsive and/or inattentive behavior is aberrant (Buitelaar, Kan, & Asherson, 2011). Apart from that, parents of a child diagnosed with ID, may easily assign ADHD-like symptoms of their child to ID, and never report them to a medical professional. In addition, the low IQ in ID-patients may cause inability to understand schoolwork, games, etcetera, and may therefore lead to decreased motivation to pay attention to these things. On the other hand, inattention problems in ADHD may lead to lower scores on IQ tests (Styck & Watkins, 2014). All in all, research is needed to investigate whether the larger-than-expected shared prevalence is due to genetic factors, non-genetic factors or a combination of both.
Attention-Deficit/Hyperactivity Disorder (ADHD)
ADHD is a psychiatric disorder characterized by inattention and/or hyperactivity and impulsivity (American Psychiatric Association, 2013). The prevalence ofADHD is 5-6% in children and 2. 5-5% in adults (Franke et at., 2012). Although the disorder starts in early childhood, it often persists into adulthood (Demian, 2011; Franke et al., 2012). According to the Diagnostic and Statistical Manual of Mental Disorders V (DSM-V), a first diagnosis of ADHD can also be made in adulthood, provided that the symptoms have been visible since age 12 when assessed in retrospect (American Psychiatric Association, 2013).
In both children and adults, ADHD is a heterogeneous disorder, since it is diagnosed by having six out of nine symptoms in at least one or two domains of the disorder (American Psychiatric Association, 2013). In addition, the majority of patients with ADHD has co-morbid disorders (Ghanizadeh, 2009; Jensen & Steinhausen, 2014; Kadesjo & Gillberg, 2001; Kraut et al., 2013; Larson et al., 2011), such as conduct/oppositional defiant disorders, language development disorders, motor development disorders, ASD and/or ID (Jensen & Steinhausen, 2014). When ADHD remains untreated, it can also lead to drop-out of school (Barbaresi, Katusic, Colligan, Weaver, & Jacobsen, 2007), mood disorders (Chen et al., 2014; Cubero-Millan et al., 2014) and drug abuse (Levy et al., 2014). Fortunately, early diagnosis and adequate treatment of ADHD can improve the quality of life of patients significantly. Insight in the etiology of ADHD and its co-morbidities can improve diagnosis and treatment of patients with ADHD. Since the heritability of ADHD is high, with an estimated 70 - 80% of the phenotypic variance explained by genetic factors (Franke et al., 2012), identification of the genes involved in ADHD can help us elucidate the pathogenesis ofADHD. Identifying ADHD-genes appears challenging though, since most cases ofADHD have a multifactorial etiology. This means that they are caused by the combination of multiple variants in many genes and by environmental factors, which all have a small effect on the risk for ADHD. Most of these gene variants are 'common', which means that they have a prevalence of at least 1% in the general population, as is the case for single nucleotide polymorphisms (SNPs), but also for certain CNVs. Interestingly, halfofthetop-ranked ADHD candidate genes, as found in five genome-wide association studies (GWASs) that were studied by Poelmans et al. (2011), contribute to a single biological process, i. e. neurite outgrowth.
In addition to multifactorial ADHD, for an unknown percentage of ADHD cases the underlying genetic background is likely to be of a mono- or oligogenic nature. In these forms of the disorder, one or a few severe gene defects are sufficient to cause ADHD in an individual patient (Franke et al., 2012). However, less severe defects in the same genes may contribute to multifactorial forms ofADHD. The genetic heterogeneity of ADHD, i. e. the different combinations of risk genes in different patients with ADHD, may explain why symptoms and co-morbidities of patients with ADHD vary so widely
('phenotypic heterogeneity'). Genes may be associated with part of the symptoms or brain phenotypes of ADHD, rather than with all of them. It is assumed that abnormality of a certain
behavioral, functional brain or structural brain domain lies in between the risk gene and the resulting disease (Hoogman, Buitelaar, Franke, Cools, & Arias-Vasquez, 2012).
Intellectual disability (ID)
ID is defined by deficits in intellectual functioning and adaptive functioning that have become apparent during the developmental period (American Psychiatric Association, 2013). The deficit in intellectual functioning is defined as scoring lower than 97. 5% of people with the same age and same culture on Intelligence Quotient (IQ) tests (American Psychiatric Association, 2013).
The prevalence of ID varies widely in different studies and different countries, mainly due to
differences in diagnostic definitions and methods (Greydanus & Pratt, 2005). According to ten large
population-based studies reported by the International Encyclopedia of Rehabilitation (Maulik, 2010), the prevalence of ID varies widely due to differences in definitions, but ties around 3% in the United States, when ID is defined as an IQ below 70 (without other criteria). Many of these children and adults also suffer from co-morbidities, especially from hearing impairments or seizure disorders (Maulik, 2010), but also from other physical problems. In The Netherlands, the rate of physical health problems in Dutch adults with ID is twice as high as that in adults without ID (van Schrojenstein Lantman-De Valk, Metsemakers, Haveman, & Crebolder, 2000). In addition, the prevalence of psychiatric problems in patients with ID is four to five times as high as in the general population (Harris, 2006). Although it may be hard to diagnose co-morbid psychiatric problems in patients with ID, this is very important, since treatment of the co-morbid psychiatric problems may still improve the well-being of the patient (Turygin, Matson, Adams, & Williams, 2014). For example,methylphenidate, the most widely used drug to relieve ADHD symptoms, works also well in patients with both ADHD and ID (Lipkin, 2013).
ID is usually a monogenic or oligogenic disorder, which means that it is caused by one or a few mutations in one or a few genes (Chelly & Mandel, 2001; van Bokhoven, 2011). These genetic variations usually have large effect sizes and they occur with a low frequency in the general population.
Research questions and approach
We hypothesized that genes being affected by rare genetic variations in patients with ID, can also contribute to multifactorial ADHD risk. In case of ADHD, these genes might harbor common variations (e. g. SNPs). In this study, we investigated the genetic overlap between ID and ADHD by using four different approaches.
First, an explorative pathway analysis was used to investigate whether 392 ID-related genes are associated with ADHD risk. Subsequently, gene-wide association analyses were performed to identify associations of the individual genes with ADHD risk.
Second, a pathway analysis, now limited to ID-related genes known to be involved in the process of neurite outgrowth, was performed to test for association with ADHD.
Third, we selected 12 genes, based on frequent occurrence in CNVs of patients with ID and ADHD (symptoms) and/or on their relation to ID and ADHD described in the literature. For all 12 genes, gene-wide association analyses were performed to test for associations with ADHD risk, symptom counts and symptom severity for inattention and hyperactivity/impulsivity.
Fourth, two genes were selected based on suggestions for associations with both ADHD and ID in the literature. Gene-wide association analyses were performed for several ADHD-related brain
phenotypes, such as the volumes of the total brain, gray matter, white matter, prefrontal cortex (PFC), caudate nucleus, hippocampus and nucleus accumbens. Genotype effects of selected SNPs on gray and white matter differences were investigated by using voxel-based morphometry (VBM) and voxel-wise analysis of structural connectivity, respectively.
Materials and Methods
Description of the samples
ADHD meta-analysis of the Psychiatric Genomic Consortium (PGC)
Meta-analytic data from 9 study cohorts containing 5, 621 cases ofADHD and 13, 589 controls were available for analysis (Neale et al., 2012, unpublished). This is the largest dataset for association with ADHD that is currently available (Neale et al., 2012, unpublished). The meta-analytic data, from a case-control analysis, were available as summary statistics, including genome-wide SNP data (imputed with 1,000 Genomes (Abecasis et al., 2012)) with corresponding p-values, minor allele frequencies (MAFs), odds ratios and imputation quality scores (INFO). ADHD was diagnosed according to the DSM-IV criteria (American Psychiatry Association, 2000). Data were obtained through the PGCADHD working group after an official proposal for the data analysis had been approved by the contributing sites.
International Multicenter ADHD Genetics (IMAGE) Project
The IMAGE dataset contains genome-wide SNP data, that is 2, 182, 904 SNPs imputed with Hapmap II release 22 (Li, Wilier, Ding, Scheet, & Abecasis, 2010), for 930 Caucasian children with ADHD
(diagnosed according to DSM-IV criteria). Of these children, 88% were male, the average age was 10 years (range 5-18 years) and the average IQ was 101 (range 55-160). Symptom counts were assessed for inattention and hyperactivity/impulsivity separately by using the Parental account of childhood symptoms (PACS), a semi-structured, standardized, investigator-based interview (Brookes et al., 2006). Data were complete for 871 patients. The Parent and Teacher Conners' long version rating scales (Conners, Sitarenios, Parker, & Epstein, 1998a, 1998b) were used to obtain continuous measures of ADHD severity. Teacher Conners' scores were available for 916 patients and parental Conners' scores for 930 patients.
CNV database of the Radboudumc
The CNV database of the Genome Diagnostics division of the department of Human Genetics of the Radboudumc includes information about all CNVs found in patients Radboudumc receiving genetic testing between 2005 and 2013 for ID, congenital anomalies or suspicion of genetic syndromes (Hehir-Kwa, Pfundt, Veltman, & de Leeuw, 2013). Coded information on CNVs of patients from a subset of this database, comprising 252 patients (with 3, 255 CNVs) who had a diagnosis of ID and increased ADHD symptoms or a diagnosis of ADHD (77. 4%) or an unclear diagnosis including ADHD (13. 9%) was obtained from the Clinical Genetics division of the Human Genetics department in September 2013 and used in this study. Patients had a mean age of 15 years (range 5 - 69 years) and 180 patients were male.
The Cognomics Resource BIG
The sample of the Brain Imaging Genetics (BIG) cohort included in the current study consisted of healthy individuals of Caucasian origin, for which both genome-wide genotyping data (imputed with 1, 000 Genomes (Abecasis et al., 2012)) and structural magnetic resonance imaging (MRI) scans were available (Cousijn et al., 2014). Participants had no self-reported psychiatric or neurological history, had never abused drugs and did not use medication other than oral anti-conceptives. Thirteen
hundred and two participants (mean age 22 years, age range 18-40 years, of which 557 males) were scanned with either a 1. 5 Tesla or a 3.0 Tesla Siemens MRI scanner (Guadalupe et al., 2014). All of them underwent a structural MRI scan. After that, the structural MRI data was processed and segmented as described by Guadalupe et al. (2014). After that, the volumes of the individual brain structures were calculated. For this, the images were re-oriented to the MN1152 standard in FSL software (Jenkinson, Beckmann, Behrens, Woolrich, & Smith, 2012) using the parameters of the Enhancing Neuro-lmaging Genetics Through Meta-analysis (ENIGMA) protocol
(http://enigma.ini. usc. edu/protocols/) (Thompson et al., 2014)). Subsequently, the cortical volumes were segmented using FreeSurfer (Fischl, 2012), while the subcortical volumes were segmented using FSL software (Jenkinson et al., 2012).
Apart from that, a VBM analysis was performed using SPM8 (Ashburner et al., 2013). In this analysis, the structural MR images were smoothed and used in multiple regression analysis in order to test for volumetric differences of white matter or gray matter associated with SNPgenotypes, as has been described by Hoogman et al. (2014).
Association of an ID-related gene-set with ADHD
Gene selection
For the selection of the ID-related gene-set, we downloaded the 'Intellectual Disability Gene Panel' that was published by the Radboudumc department of Human Genetics' Genome Diagnostics division (downloaded from https://www.radboudumc.nl/lnformatievoorverwijzers/
Genoomdiagnostiek/Documents/ngs-intellectual_disability_panel_181213.
pdfon March 27 , 2014).
This gene panel listed 490 candidate genes for ID (shown in Supplementary Table 1), based on findings of de novo mutations in patients with ID visiting the Radboudumc and collaboratinginstitutes and based on the literature including the online database Online Mendelian Inheritance in Man (OMIM) (Hamosh, Scott, Amberger, Bocchini, & McKusick, 2005). This list forms the basis for diagnostic testing using exome sequencing at the department of Human Genetics of the
Radboudumc.
Data extraction
For association analyses, we extracted SNPs and association p-values from the ADHD PGC meta-analysis. Since this meta-analysis only covered autosomal genes, we excluded the X-chromosomal genes and were left with 396 autosomal genes. All SNPs tying within these genes (according to base pair positions in UCSC HG19; (Kent et al., 2002)) were extracted. Flanking regions of 25 kilobase (kb) were used to capture regulatory regions. A total of 687, 602 SNPs with a MAF of at least 0. 01 were considered for further analysis.
Gene-set analysis
We used the Knowledge-based mining system for Genome-wide Genetic studies version 3. 5 (KGG
3. 5) software (Li, P. C. Sham, S. S. Cherny, & Y. Q. Song, 2010) in order to test whether the group of
ID-related genes is associated with ADHD. Within this software package we chose the Hybrid set-based test (HYST) (Li, Kwan, & Sham, 2012) for association testing. A text file listing all 396 autosomal ID-related genes and a text file listing all SNPs that were extracted from the ADHD PGC meta-analysis
(as described in the previous paragraph), were used as input for KGG. The genome-wide genotyping data of BIG (Guadalupe et al., 2014) was used as a reference to define the underlying linkage
disequilibrium (LD) structure. The LD upper limit was set to an r of 0.8, while the lower limit was set
to 0. 2. SNPs with an INFO score below 0. 3 were ignored. Apart from that, default settings of KGG were used. KGG makes use of UCSC Hgl9 Refgene (Kent et al., 2002) to assign base pair positions to genes. Next, a gene-based association scan was run with HYST, based on a hybrid test of Gates and a scaled Chi-square test, in which SNPs without LD information were ignored. Finally, a pathway-based association scan was run, to investigate the association between ADHD and our customized pathway, consisting of all 396 ID-related genes. SNPs without LD information were again ignored. The actual analysis in KGG could only include 392 genes, since 4 genes (GRIK2, GSS, INPP54 and NEU1] had too few SNPs available. Genes were considered significant if their gene-wide p-value was below 0. 05 after Bonferroni correction for testing 392 genes. Significantly associated genes were considered for further analysis. The pathway was considered significant if the pathway p-value calculated by theHYST was below 0. 05 after Bonferroni correction for testing two pathways (see also below).
Gene-wide analyses
We made use of the International Multicenter ADHD Genetics Project (IMAGE) to further investigate if SNPs in the MEF2C locus, found to be significant in the KGG analysis, are associated with symptom counts for inattention and/or hyperactivity/impuisivity in patients with ADHD (Bralten et al., 2013). SNPs lying within the MEF2C gene (according to base pair positions in UCSC Hgl8 (Kent et al., 2002)) and its 25 kb flanking regions were extracted from the IMAGE dataset. SNP data were pruned prior to analyses (using the command 'indep-pairwise 50 5 0.8' in PUNK), resulting in 35 SNPs on the MEF2C locus. Gene-based association analysis was performed using the 'linear' and 'mperm' commands PLINK (S. Purcell et al., 2007), according to the protocol described in (Bralten et al., 2013). Thousand permutations were used to compute empirical p-values. We tested whether the MEF2C locus was associated with the Blom-transformed (Blom, Ludwig, & Gunnar, 1958) symptom counts for inattention and hyperactivity/impulsivity, respectively. MEF2C associations were considered significant if the p-value was below 0. 05 after Bonferroni correction for testing two symptom domains.
Association of ID genes involved in neurite outgrowth with
ADHD
Gene selection and association testing
We aimed to investigate whether genes involved in neurite outgrowth as well as in ID show an association with ADHD as a group. For this, we followed the same procedure as above, but then with
a different gene selection. First, the Intellectual Disability Gene Panel of the Radboudumc (see above
for description) was downloaded. Second, a list of 788 genes involved in the GO-term 'neuron projection development', according to AmiGOZ (http://amigo.geneontology.org (Carbon et al., 2009) was created. Third, we selected 62 genes that were included both in the ID-gene list and in the 'neuron projection development' GO-term. After exclusion of X-chromosomal genes, we were left with 46 genes and 87, 241 SNPs for further analyses. We used KGG 3. 5 for gene-set association
analysis as described above. Genes were considered significant if their gene-wide p-value was below 0.05 after Bonferroni correction for testing 46 genes.
Association of genes affected by rare CNVs in ID-patients with
ADHD
Gene selection
Here, we set out to investigate if genes affected by rare CNVs (i. e. CNVs having a prevalence below 1% in the general population) in patients with ID from our own department were also associated withADHD. For the selection of genes, we used the CNV database of the department of Human Genetics of the Radboudumc (see above). All of the 252 patients had a double diagnosis of ID and ADHD. Genes within the CNVs were annotated according to Hgl9 RefSeq (Pruitt, Brown, & Tatusova, 2002). Subsequently, a selection was made of genes that a) occurred at least twice in CNVs with a prevalence below 1% in the general population (preferably single-gene CNVs), b) were related to ID according to the literature and c) had been suggested to play a role in ADHD according to the literature (especially through CNV studies in patients with ADHD (Elia et al., 2010; Elia et al., 2011; Lesch et al., 2011; Stergiakouli et al., 2012; Williams et al., 2012; Williams et al., 2010)). This procedure resulted in the selection of 19 genes.
Data extraction
Since the PGC meta-analysis results only contained data on autosomal genes, we had to exclude seven X-chromosomal genes. All SNPs lying in the selected gene loci (according to UCSC Hgl9 (Kent et al., 2002)), including flanking regions of 50 kb, were extracted from the summary statistics of the ADHD PGC meta-analysis, as described above for approach 1. We included SNPs that had a MAF > 0. 01 and INFO > 0. 6 for further analyses.
Gene-wide analyses with the PGC data
wide analyses were run for each of the 12 genes using the offline version of the Versatile Gene-based Association Study (VEGAS) software, a tool that combines p-values of all SNPs within a gene into one gene-wide p-value by making use of Monte Carlo simulations and permutations (Liu et al., 2010). A 1,000 Genomes reference population was used for LD correction. The default settings of VEGAS, i. e. flanking regions of 50 kb and maximal one million permutations were used. The input files contained all SNPs within gene loci, including flanking regions. Results were considered significant if the p-value was below 0.05 after Bonferroni correction for testing 12 genes.
Gene-wide analyses with the IMAGE data
We also tested whether the 12 genes were associated with Blom-transformed (Blom et al., 1958) symptom counts and Conners' scores for hyperactivity/impulsivity or inattention using the data from the IMAGE study (see above). Data were analyzed as described for approach 1. We used 1, 000 permutations to calculate empirical p-vatues. For results with borderline significant p-values, analyses were re-run with 10, 000 permutations to acquire a more precise estimation of the p-value. Analyses were performed with pruning (using the command 'indep-pairwise 50 5 0. 8' in PUNK).
Associations were considered significant if the p-value was < 0. 05 after Bonferroni correction for performing 72 tests. This correction was applied in addition to the permutation testing for individual findings.
Association of genes for ID and ADHD with brain phenotypes
Gene selection
One of the ways by which genes may influence disease risk is by altering brain structure or function. Here we aimed to investigate whether genes involved in both ID and ADHD according to previous studies can be linked to structural brain volumes that often show differences between patients with ADHD and healthy individuals. For the selection of genes, we used four criteria: 1) indications for involvement in both ADHD and ID according to the literature in PubMed; 2) location on an autosomal chromosome; 3) preferably involved in neuron projection development according to AmiG02
(Carbon et al., 2009); 4) being part of the ID gene panel of the Radboudumc. This resulted in the selection ofCHRNA7and NRXN1. CHRNA7was not involved in neuron projection development, but has been reported frequently in single- or two-gene CNVs in ADHD (Stergiakouli et al., 2012; Williams
et al., 2012; Williams et al., 2010), and was therefore chosen. A third gene satisfying selection
criteria, CNTNAP2, could not be considered given overlapping projects of others involving this gene.
Data extraction
SNPs in the CHRNA7 (27 SNPs) and NRXN1 (712 SNPs) gene loci, according to UCSC Hgl9 (Kent et al., 2002) and including 25 kb flanking regions relative to the largest isoforms, were extracted from the genome-wide genotype data of the BIG cohort (Cousijn et al., 2014).
Gene-wide analyses
The SNPs in CHRNA7and NRXN1, respectively, were tested for associations with ADHD status by using the ADHD PGC meta-analysis data and for associations with ADHD symptom scores using the IMAGE project data. Gene-based association analyses were performed as described above for the previous approach. In addition, gene-wide analyses were performed to investigate if the SNPs within CHRNA7or NRXN1 were associated with regional brain volumes. The following regional brain
volumes were selected, based on reports of differences in volumes in patients with ADHD with ADHD and healthy individuals according to the literature: total brain volume, total gray matter, total white matter, PFC, caudate nucleus, hippocampus, and nucleus accumbens. For the latter four structures, averaged volumes of left and right hemisphere were considered. Total brain volume is the sum of
total gray matter and white matter. PFC volume was computed by summing all PFC substructures
(lateral orbitofrontal cortex, medial orbitofrontal cortex, pars opercularis, pars orbitalis, pars triangularis, rostral middle frontal cortex, superior frontal cortex and frontat pole) as described by Destrieux et al. (2010). Gene-wide association analyses were performed by using the commands 'linear' and 'mperm' in PUNK, according to the protocol of Bralten et al. (2013). Thousand
permutations were used to calculate empirical p-values. Gender, age, and magnetic field strength were used as covariates for all analyses. Total gray matter was used as a covariate for the analysis of
total white matter and vice versa, and total brain volume was used as covariate for analysis of the PFC, caudate nucleus, hippocampus, and nucleus accumbens. Analyses were performed with pruning
(using the command 'indep-pairwise 50 5 0. 8' in PLINK). Associations with p-values below 0.05 after Bonferroni correction for doing 14 tests were considered significant. For genes that showed
borderline significant p-values when uncorrected for multiple testing, the analysis was repeated with 10, 000 permutations. For PFC, where borderline significant p-values were observed, association tests were also performed for all the substructures (left and right separately). P-values below 0.05 were considered significant after Bonferroni correction for testing 16 PFC substructures.
Selection of SN Ps
For single-SNP analyses, we selected SNPs in NRXN1 and CHRNA7 genes that were most likely to increase the risk for ADHD. SNPs were selected based on a) low p-values in the ADHD PGC meta-analysis (Neale et al., 2012, unpublished), b) low p-values in the PGC meta-analyses for the psychiatric disorders schizophrenia (Schizophrenia Psychiatric Genome-Wide Association Study (GWAS) Consortium, 2011), major depression disorder (Ripke et al., 2013), bipolar disorder (Psychiatric GWAS Consortium Bipolar Disorder Working Group, 2011) and/or low p-values in the
PGC Cross-disorder study (Cross-Disorder Group of the Psychiatric Genomics Consortium, 2013) as
assessed through Ricopili (May 2014; http://www.broadinstitute.org/mpg/ricopili/), c) functional effects of these SNPs described in the literature (affecting gene expression, brain functioning, brain activation, etc. ), and/or d) functional effects of proxy-SNPs that lie in high LD (> 0. 8 according to SNAP (Johnson et al., 2008)) with the SNPs of interest. A total of five SNPs were selected for analysis
through these procedures.
SNP-specific analyses
P-values for associations between selected SNPs and total brain volume, total gray matter, total
white matter, PFC, caudate nucleus, hippocampus, and nucleus accumbens were retrieved from
previous genome-wide analyses using the BIG cohort. SNPs with p-values below 0.05 after Bonferroni correction for performing 35 tests were considered significant. A VBM analysis was done using SPM8 (Ashburner et al., 2013) using the BIG data analysis in order to test for volumetric differences of white matter or gray matter associated with SNP genotypes, as has been described by Hoogman et al. (2014). SNP genotypes were coded as a linear additive effect. Scanner field strength, age, and sex were used as covariates. In addition, we tested whether SNP genotypes were associated with fractional anisotropy (FA) and mean diffusivity (MD) (parameters measuring structural connectivity).
Again, the used regression models assumed an additive effect ofSNP genotypes and used scanner field strength, age, and sex as covariates. False discovery rate (FDR) control was used to correct for
the testing of multiple voxels. Associations were considered significant if their FDR-corrected peak level had a p-value below 0.05 after Bonferroni correction for testing four measures (gray matter,
white matter, FA and MD) for five SNPs.
Results
Association of an ID-related gene-set with ADHD
The ID-related gene-set of 392 genes (see Supplementary Table 1) was significantly associated with ADHD (P = 0.0000656). P-values for the ten toranked findings are shown in Table 1. Gene-wide p-values of all 392 genes are displayed in Supplementary Figure 1. MEF2C was the only gene that was
significant (uncorrected P = 0. 0000461) after correction for multiple testing (correcting for 392 genes). A region plot of the uncorrected p-values for the association with ADHD in the ADHD PGC meta-analysis for the individual SNPs in MEF2C and its flanking regions is shown in Figure 1. MEF2C
was not significantly associated with Blom-transformed symptom count for inattention (P= 0. 517) or
hyperactivity/impulsivity (P= 0. 078) in the IMAGE childhood ADHD sample. Two additional genes
were borderline significant after Bonferroni correction: ST3GAL3 (P=0. 000142) and TUSC3
(P= 0. 000329). In total, 41 of the 392 (10. 5%) were nominally associated with ADHD (P < 0. 05).
Table 1. Top ten genes from pathway analysis of ID-related genes. Length Gene MEF2C ST3GAL3 TUSC3 TRAPPC9 ATR BBS7 LIG4 ALG2 PNP AGA ARFGEF2 MLYCD ASL NDUFS3 KCNK9 P-value 0.0000461 0. 000142 0.000329 0.00105 0. 00117 0.00234 0. 00341 0.00535 0. 00633 0.00649 0. 00665 0.00705 0.00731 0. 00782 0. 00863 Chr. 5 1 8 8 3 4 13 9 14 4 20 16 7 11 8 Start 88183208 44201933 15397791 140999030 142279180 122779757 108859793 101979651 20937537 178358221 47538426 83932730 65554225 47603603 140624803 138266 193961 133555 19659 107171 3078 8096 3679 5194 2558 114805 17058 3717 1209 90497 ffSNPs 1198 1478 3011 5084 1073 512 427 284 491 413 1012 684 380 225 856
Gene-wide association p-values of the top ten genes from the pathway analysis oflD-related genes in
KGG. Bold font = significantly associated with ADHD after Bonferroni correction. The Bonferroni
p-value cut-off for significance was 0. 000128. Chr. = chromosome number; Start = start position of the gene in bp; Length = length of the gene in bp; ff SNPs = number ofSNPs used for this analysis.
PlottedSNPs Iliii!ffii!!iBljEeSl!iH;iEFEI!i"liB|!!SMIS!!ll"g6!iSISEIffillliillE[!l@ELiliili!i!lii 0.8 -0.6 -0.4 !- 0.2 rs190982 ... 4
i-
c-I (0 (5 s^' ' .'
.;r ''" -^
. .... <%
Os3^
0 . 0 . . .' 0 h 100 80.1
60 a 01 1-. -. -. % -.-.
f-40^
r-^- y*^- ^ l»i
'u'
-UNC004B1 -MEF2C 88. 15 88.2 Position on chr5 (Mb)Figure 1. Region plot of association p-values for SNPs on MEF2Clocus.
The region plot was made using Locuszoom (Pruim et al., 2010). Association p-values are shown for SNPs located within the largest isoform ofMEF2C (according to UCSC Hgl9) and its 25 kb flanking regions using data from the ADHD PGC meta-analysis. The purple SNP, rsl90982, is the reference
SNP. Colors of other SNPs indicate the correlation (linkage disequilibrium coefficient [f1)) with the
reference SNP.
Association of ID genes involved in neurite outgrowth with
ADHD
The gene-set of 46 autosomal ID genes involved in neurite outgrowth was not associated ADHD (P=0. 363). Gene-wide p-values are displayed in Supplementary Figure 2. None of the individual genes
was significantly associated with ADHD after Bonferroni-correction. Two out offorty-six genes (4. 3%) had a nominally significant p-value (P < 0. 05).
Association of genes affected by rare CNVs in ID-patients with
ADHD
None of the 12 selected autosomal genes was significantly associated with ADHD in the ADHD PGC meta-analysis data (Table 2). The lowest p-values were found for PRODH (uncorrected P= 0. 0283) and for RBFOX1 (uncorrected P=0. 0845). PRODH and RBFOX1 were neither associated with Blom-transformed symptom counts, nor with parental or teacher Conners' scores for inattention or hyperactivity/impulsivity (Table 3).
Table 2. P-values ofgene-wide associations with ADHD in the ADHD PGC meta-analysis data.
Gene P-value Chr. Start Length ffSNPs
PRODH RBFOX1 PTPRD CNTNAP2 NRXN1 XYLT1 PRIM2 FAM110C SKI GRIN2A NRG3 ERBB4 0.028 0.085 0. 176 0. 252 0. 283 0. 480 0. 492 0. 621 0. 681 0.826 0.858 0. 919 22 16 9 7 2 16 6 2 1 16 10 2 18900286 6069095 8314246 145813453 50145643 17195626 57179603 38814 2160134 9852376 83635070 212240446 23780 1694245 2298477 2304637 1114031 369112 333772 8056 81424 424235 1111865 1163119 45 2432 2132 554 3954 1244 177 13 189 808 727 4634
Gene-wide association p-valuesfor ADHD. Chr. = chromosome number; Start = start position of the gene in bp; Length = length of the gene in bp; ffSNPs = number ofSNPs used for this analysis.
Table 3. P-values for associations of Blom-transformed symptom counts and Conners' scores with PRODH and RBFOX1
Gene
RBRFOX1
PRODH
ff SNPs Symptom
domain
Association with Association with Association
Blom-transformed parental with teacher symptom counts Connors'score
794 Inattention Hyperactivity /impulsivity Inattention 0. 434 0.419 0. 132 0. 128 0. 827 0. 658 0. 555 0. 012 Conners' score 0. 687 0. 439 0. 986 0. 681 Hyperactivity /impulsivity
Gene-wide association p-valuesfor RBFOX1 and PRODH with Blom-transformed symptom counts and Conners' scores. Bold font = nominally significant. # SNPs = number ofSNPs used for this analysis.
Association of genes for ID and ADHD with brain phenotypes
NRXN1 and CHRNA7were selected as genes of interests. Neither of the genes was associated with any of the tested brain volumes (total brain, total gray matter, total white matter, PFC, caudate nucleus, hippocampus, and nucleus accumbens) in the BIG dataset (Table 4). Because CHRNA7was
nominally associated with the PFC, we tested whether CHRNA7 is associated with the substructures
of the PFC (left and right separately). None of the associations was significant (Supplementary Table
2). Three SNPs were selected for CHRNA7 [rs2337980, rs904952, and rs6494223) and two for NRXN1 (rsl0191989 and rsll891766} for SNP-specific analyses. None of these SNPs was significantly
associated with total brain volume, total gray matter, total white matter, PFC, caudate nucleus,
hippocampus, or nucleus accumbens (Supplementary Table 3). Neitherofthevoxel-wise analyses did reveal any effect of SNP genotypes on brain structure (total gray matter or total white matter) or connectivity (fractional anisotropy or mean diffusivity) (Supplementary Table 4).
Table 4. Gene-wide association for CHRNA7 and NRXN1 with brain volumes
CHRNA7
NRXN1
ff Total Total Total
SNPs brain gray white
volume matter matter
27 0.265 0.883 0. 893
712 0. 194 0. 22 0. 747
Prefront Nucleus Hippo- Nucleus
alcortex caudate campus
accum-(PFC)
0. 044 0. 658 0.759 0. 785 0. 175 0. 895 bens 0. 844 0. 63Gene-wide association p-valuesfor CHRNA7 and NRXN1 with different brain volumes. Bold font = nominally significance.
Discussion
Summary of findings
SNPs in the ID-related gene-set of 392 genes were significantly associated with ADHD, whereas SNPs
in the neurite outgrowth subset were not. The MEF2Cgene was significantly associated with ADHD risk, but did not show a clear relation with domain-specific symptom counts for inattention or
hyperactivity/impulsivity. None of the other ID-related genes tested were significantly associated
with ADHD risk when tested individually, although ST3GAL3 and TUSC3 showed borderline significant
associations with ADHD risk after Bonferroni correction. The 12 individual genes that were selected
because of their frequent occurrence in rare CNVs in patients with both ADHD (symptoms) and ID, PRODH, RBFOX1, PTPRD, CNTNAP2, NRXN1, XYLT1, PRIM2, FAM110C, SKI, GRIN2A, NRG3, and ERBB4,
were not significantly associated with ADHD risk based on common genetic variants, nor with ADHD symptom count. Similarly, the two additional genes selected based on evidence from the literature
for involvement in both ADHD and ID, CHRNA7ar\d NRXN1, were not significantly associated with
ADHD risk, nor with selected brain volumes that might be thought of as endophenotypes for ADHD,
i. e. the volumes of the total brain, total gray matter, total white matter, PFC, caudate nucleus,
hippocampus, or nucleus accumbens. None of the individual SNPs in CHRNA7 (rs2337980, rs904952 and rs6494223), or in NRXN1 [rsl 0191989 and rsll891766) was associated with the aforementioned regional brain volumes either. These SNPs were also not associated with voxel-wise measures of gray- or white matter or of structural connectivity.
Interpretation of results in relation to the literature
In our first two approaches, we used gene-set association analyses, also known as 'pathway analyses'. There are two important reasons to perform gene-set analyses rather than gene-wide or
genes in one analysis can increase the explained phenotypic variance and improve the power of the study by allowing allelic heterogeneity (Bralten et al., 2013). Secondly, gene-set analyses can answer broadly oriented questions that give insight into the mechanisms and nature of disorders. Our results show that ID and multifactorial ADHD share an overlapping genetic background. While neurite outgrowth is a process involved in ADHD and ID (Poelmans et al., 2011; Trazzi et al., 2013; Van Maldergem et al., 2013; Wang, Moore, Adelmant, Marto, & Silver, 2013), SNPs in the ID-related genes involved in the process of neurite outgrowth could not be linked to ADHD risk, suggesting that
the overlap in ADHD and ID is not driven (primarily) by genes involved in this process (at least not
those that we tested).
To our knowledge, we are the first to have tried to find the overlapping genes between the rare variants in {monogenic or oligogenic) ID and the (common) SNPs in multifactorial ADHD. Most previous studies have focused on either rare genetic variations in monogenic or oligogenic disorders
or on common variations in multifactorial disorders, but not on a combination of both. For example,
Lee et al. (2013) studied the correlation ofSNPs in five multifactorial psychiatric disorders and found
that the SNPs in ADHD correlate moderately with those in major depressive disorder. In addition,
Doherty and Owen (2014) reported in a review article that SNP atleles overrepresented in patients with ADHD were also overrepresented in patients with schizophrenia or bipolar disorder. Apart from that, Doherty and Owen (2014) also focused on rare, large CNVs in monogenic or oligogenic
disorders. Their report included seven large CNVs associated with both ID and ADHD, suggesting involvement of certain genes in both ID and ADHD. Since these CNVs are rare, usually cause severe physical problems and have a high penetrance for ADHD, they are likely to be involved in monogenic
or oligogenic ADHD, but unlikely to play a role in multifactorial ADHD, which is the most common
form ofADHD. However, assuming that genes harboring rare variants involved in oligogenic ADHD
can also contribute to multifactorial ADHD through common genetic variations, we think that the
arguments of Doherty and Owen (2014) are in accordance with our finding of genetic overlap
between ID and multifactorial ADHD. In addition, previous studies have reported individual genes involved in both ID and ADHD (Hunter, Epstein, Tinker, Abramowitz, & Sherman, 2012; Jolly, Homan, Jacob, Barry, & Gecz, 2013).
We have shown that the ID-related gene MEF2C is significantly associated with the risk for
multifactorial ADHD. However, it does not seem to be associated with symptom count for inattention
(P=0.517). An association of /W£F2Cwith symptom count for hyperactivity/impulsivity might exist
(P=0. 078), even though our study was not able to prove this. To our knowledge, MEF2C has not been
associated with ADHD before, even though it has been studied extensively and has been linked to several disorders and characteristics. For instance, MEF2C, located on chromosome 5, has been previously found mutated in severe ID, and associated with Alzheimer's disease and epilepsy (Beecham et at., 2014; Novara et al., 2010; Novara et al., 2013; Nowakowska et al., 2010;
Paciorkowski et al., 2013). In addition, MEF2C haploinsufficiency has been linked to 1) autistic traits
such as lack of eye contact, absence of speech, impaired engagement with other people, and
stereotypic movements and to 2) severe motor problems, such as hypotonia, severely impaired fine
motor coordination and the inability to walk freely (Novara et al., 2010; Novara et al., 2013; Nowakowska et al., 2010; Paciorkowski et al., 2013). The latter is interesting, because ADHD
correlates strongly with autistic traits and moderately with motor coordination problems (Rommelse
et al., 2009), even though the motor problems in ADHD are usually far less severe than the motor
problems in patients with MEF2C haploinsufficiency (Fliers et al., 2008).
Interestingly, MEF2C does not play a role in neurite outgrowth according to AmiG02 (Carbon et al.,
2009). Moreover, none of the individual ID-related genes involved in neurite outgrowth, nor the entire set of ID-related neurite outgrowth genes, could be linked to ADHD risk in our study. The tack
of a significant association of the set of ID-related neurite outgrowth genes and ADHD was
unexpected, because this gene-set seemed to have a higher o prior/ chance of being associated with ADHD risk than the set of all ID-related genes, since half of the ADHD candidate genes as found by GWASs were reported to be involved in neurite outgrowth in the study by Poelmans et al. (2011). Importantly, none of the ADHD candidate from the identified neurite outgrowth network was present in the ID-related gene-set that we tested. When Bralten et al. (2013) performed a gene-set analysis combining the dopamine/noradrenaline pathway, the serotonin pathway and the set of neurite outgrowth genes involved in ADHD as reported by Poelmans et al. (2011), they found a strong association with symptom count of hyperactivity/impulsivity, but not with symptom count of inattention. Since we already know from twin modeling that the genetic overlap of
hyperactivity/impulsivitywith inattention is incomplete (Moruzzi, Rijsdijk, & Battaglia, 2014), this emphasizes again the importance of testing hyperactivity/impulsivity separately from inattention. Apart from MEF2C, the SNPs of none of the other genes we tested showed significant gene-wide associations with ADHD risk. For most genes, this was in line with the literature, since no association withADHD has been previously reported for the majority of the genes that we tested. Although case reports ofADHD have been described for a minority of the tested genes, they describe CNVs or rare genetic variations, while our study only tested SNPs. For example, CNVs of CHRNA7(Hoppman-Chaney, Wain, Seger, Superneau, & Hodge, 2013; Stergiakouli et al., 2012) and deletions of (exons of)
the NRXN1 gene (Bradley et al., 2010; Curran, Ahn, Grayton, Collier, & Ogilvie, 2013) have been
linked to ADHD risk. However, our study did not show gene-wide associations of the SNPs within CHRNA7or NRXN1 with multifactorial ADHD risk, possibly because SNPs are likely to have smaller effects on gene functioning than CNVs or deletions of (a few exons of) the gene. Even if some SNPs in some of the tested genes were associated with multifactorial ADHD risk, their effects might have been too small to give a gene-wide effect that survives Bonferroni correction for testing multiple
genes.
CHRNA7 and NRXN1 were also not associated with any of the regional brain volumes that we tested in healthy individuals. The five SNPs in CHRNA7 and NRXN1 that we selected based on the literature describing psychiatric disorders and functional effects linked to these SNPs or their proxy-SNPs, were neither associated with the tested regional brain volumes, nor with the measures of structural connectivity in healthy individuals. For CHRNA7, this is in line with the literature. As far as we know, no previous studies have linked genetic variations in CHRNA7 to regional brain volumes. SNPs in NRXN1, however, have been linked to frontal white matter volume in healthy adults (Voineskos et al., 2011), which is likely to be related with two of the parameters we tested: total white matter volume and PFC volume. Importantly, even though the SNPs in CHRNA7 and NRXN1 could neither be linked to structural connectivity nor to any of the tested brain volumes in healthy individuals, they might still affect structural connectivity or regional brain volumes in patients with ADHD. This can occur if certain variations in other ADHD-retated genes are needed to establish an effect of CHRNA7
and/or NRXN1 on the brain.
Strengths and limitations
This study should be seen in light of several strengths and limitations. By our comprehensive project
of four different approaches, we investigated whether ID-related genes are involved in ADHD.
gene-set, gene-wide, and single-SNP level. In addition, we used different types of criteria to select the candidate ID genes, such as 1) involvement in neurite outgrowth, 2) frequent occurrence in CNVs in patients with ADHD (symptoms) and ID, and 3) indications for involvement in ADHD and ID in the literature. Finally, we did not only look at ADHD diagnostic status, but also at symptom counts and different brain volume and structural connectivity measures. Another strength of our research is that we used the largest meta-analysis containing genome-wide data on ADHD that is currently available, the ADHD PGC meta-analysis available (Neale et al., 2012, unpublished). This is crucial when
investigating multifactorial disorders, in order to be able to detect genetic associations with small
effect sizes.
The resemblance between all our approaches is that they all focus only on SNPs, only on autosomal genes and only on multifactorial ADHD. Therefore, we are not able to say anything about genetic variations other than SNPs, about X-chromosomal genes, or about monogenic/oligogenic forms of ADHD.
Our first two approaches used gene-set analyses testing the SNPs in the entire group of ID-related genes and the subset of ID-related genes involved in neurite outgrowth, respectively, for an association with multifactorial ADHD. To select the ID-related genes, we used an ID gene panel composed for diagnostic testing purposes by the Radboudumc. Although the genes in this panel have been carefully selected based on findings ofcfe novo mutations in patients with ID visiting the
Radboudumc and based on the literature by experienced geneticists, this list is no gold standard and differs substantially from other lists of ID-related genes (S. M. Purcell et al., 2014; van Bokhoven, 2011). In addition, the list is probably far from complete, since research continues to find new ID candidate genes. Similarly, research on gene ontology continues to find new genes involved in neurite outgrowth, and the list of neurite outgrowth genes used in the study was therefore also likely to be incomplete. The gene ontology browser that we used, AmiGOZ (Carbon et al., 2009) is updated every week.
About a quarter of genes present in the ID gene panel had to be excluded from analysis because they were located on the X-chromosome, which is not covered in the ADHD PGC meta-analysis. Because of this and because of the stringent Bonferroni correction we applied when testing all genes
individually, we might have missed important genes involved in both ID and ADHD in our gene-wide analyses.
Our third and fourth approach focused on individual genes. In the third, 12 genes were selected based on frequent occurrence in CNVs of patients with ADHD and ID and/or congenital anomalies. This third study might have been limited by the small sample size, because the CNV database
included CNVs of only 252 patients. In addition, we might have selected genes that occurred often in CNVs of patients with ID with ADHD, because of their association with ID (or congenital anomalies) and not due to their association with ADHD. To control for this, it would have been necessary to compare the frequency of the genes in CNVs in patients with ID with ADHD to that in patients with ID without ADHD. However, such a dataset was not available for this study, since the patients in the CNV database were not routinely tested for ADHD. In addition to that, ADHD can be difficult to diagnose or exclude in patients with ID (Buitelaar et al., 2011). In our fourth approach, the SNPs in CHRNA7 and NRXN1, two genes selected based on indications for involvement in ID and ADHD in the literature, could not be associated with any of the tested brain volumes or structural connectivity measures in healthy adults without ADHD from the BIG cohort. Testing in patients with ADHD was not possible because of time constraints.
Future research
It would be worthwhile to verify the association of ID-related genes with ADHD using data that
includes X-chromosomal genes, in order to get a more complete picture. Also, it would be nice to see what part of the ID-related gene-setcan be linked toADHDwhen looking at CNV studies ofADHD. Investigating with what regional brain volumes the subset of ID-related genes most strongly associated with ADHD risk (e. g. having an uncorrected p-value below 0.05) can be associated, has more power than testing this for individual genes, and could lead to more insight into the nature of the shared mechanisms of ADHD and ID. Moreover, it would be interesting to know what pathways are enriched in this subset, for instance by making use of a SNP-based pathway enrichment analysis (Weng et al., 2011). Finally, investigating if this subset can be associated with any other
neurodevelopmental disorder, can give us more insight into the mechanisms behind the suggested
existence of a 'neurodevelopmental continuum' (Moreno-De-Luca et al., 2013).
As for MEF2C, future research could focus on how SNPs in MEF2C affect brain imaging measures in individuals with and without ADHD. Many studies have shown that MEF2C haploinsufficiency affects brain structure, such as causing thinning of the corpus callosum (Nowakowska et al., 2010) and
enlargement of the lateral ventricles (Novara et al., 2013). It would be interesting to see if these
effects are also seen for certain SNP genotypes in MEF2C. In addition, one could investigate how variations in MEF2C affect performance on neuropsychological tasks (measuring attention,
impulsivity, short-term memory, etcetera), for example in the patients with ADHD and controls in the International Multicentre persistent ADHD CollaboraTion (IMpACT) study (Franke et al., 2010).
Final conclusion
Our gene-set analyses contribute to the understanding of the genetic overlap between ID and ADHD by showing that 1) the SNPs in at least part of the genes affected by rare genetic variations in ID
patients, contribute to ADHD, and 2) this genetic overlap between ID and ADHD seems to be not
primarily driven by the neurite outgrowth genes we tested.
More generally, they contribute to insight in the 'neurodeveiopmental continuum', by showing that multifactorial disorders might be caused by different variations in the same risk genes as monogenic
or oligogenic disorders.
Apart from that, one of our gene-wide analyses has identified MEF2C as a novel candidate for multifactorial ADHD, which to our knowledge has not been linked to ADHD before. Because the
association of MEF2C with multifactorial ADHD that we found was highly significant and because MEF2C haploinsufficiency has been linked to common co-morbidities of ADHD, we think MEF2C is a
promising candidate gene for ADHD.
Acknowledgements
I would like to thank Barbara Franke and especially Marieke Klein, who supervised me greatly. Apart
from that, I want to thank alt people of the department Multifactorial Diseases of the RadboudUMC
in Nijmegen for assisting in my project. In addition, I am grateful to the SURF Foundation support
team, since the gene-set analyses in KGG were carried out on the Dutch national e-infrastructure
with the support of SURF Foundation. Also, I am thankful to all people who contributed to the
development of the databases I used: the PGC ADHD working group, the people contributing to
(especially Nicole de Leeuw) and the people contributing to the Cognomics Resource BIG. Finally, I want to thank the developers of the programs KGG, VEGAS and PLINK, which I used to perform
gene-set and gene-wide analyses.
References
Abecasis, G. R., Auton, A., Brooks, L. D., DePristo, M. A., Durbin, R. M., Handsaker, R. E.,. McVean,
G. A. (2012). An integrated map of genetic variation from 1,092 human genomes. [Research
Support, N I H , Extramural Research Support, Non-U S Gov't]. Nature, 491(7422), 56-65.
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5). Washington D.C. : American Psychiatric Publishing.
American Psychiatry Association. (2000). Diagnostic and statistical manual of mental disorders, 4th revised edition (DSM-IV-TR). Washington D.C. : American Psychiatric Publishing.
Ashburner, J., Barnes, G., Chen, C., Daunizeau, J., Flandin, G., Friston, K.,... Philips, C. (2013). SPM8 Manual. Retrieved from http://www.fil. ion. ucl. ac. uk/spm/doc/spm8_manual. pdf
Barbaresi, W. J., Katusic, S. K., Colligan, R. C., Weaver, A. L., & Jacobsen, S. J. (2007). Long-term school outcomes for children with attention-deficit/hyperactivity disorder: a population-based
perspective. [Research Support, N I H , Extramural Research Support, Non-U S Gov't]. J Dev
Behav Pediatr, 28(4), 265-273.
Beecham, G. W., Hamilton, K., Naj, A. C., Martin, E. R., Huentelman, M., Myers, A. J.,... Montine, T
J. (2014). Genome-Wide Association Meta-analysis of Neuropathologic Features of Alzheimer's Disease and Related Dementias. Plo5 Genet, 10(9).
Blom, G., Ludwig, 0., & Gunnar. (1958). Statistical estimates and transformed beta-variables.
Biometrische Zeitschrift, 3(4), 285-285. doi: 10. 1002/bimj. l9610030410,
Bradley, W. E., Raelson, J. V., Dubois, D. Y., Godin, E., Fournier, H., Prive, C.,... Belouchi, A. (2010).
Hotspots of large rare deletions in the human genome. [Research Support, Non-U S Gov't]. Plo5 One, 5(2), 0009401.
Bralten, J., Franke, B., Waldman, 1., Rommelse, N., Hartman, C., Asherson, P.,... Arias-Vasquez, A. (2013). Candidate genetic pathways for attention-deficit/hyperactivity disorder (ADHD) show association to hyperactive/impulsive symptoms in children with ADHD. J Am Acad Child Adolesc Psychiatry, 52(11), 1204-1212.el201.
Brookes, K., Xu, X., Chen, W., Zhou, K., Neale, B., Lowe, N.,... Asherson, P. (2006). The analysis of 51 genes in DSM-IV combined type attention deficit hyperactivity disorder: association signals in DRD4, DAT1 and 16 other genes. Mol Psychiatry, 11(10), 934-953.
Buitelaar, J., Kan, C. C., & Asherson, P. (2011). ADHD in adults: Characterization, Diagnosis, and
Treatment. New York: Cambridge University Press.
Carbon, S., Ireland, A., Mungall, C. J., Shu, S., Marshall, B., & Lewis, S. (2009). AmiGO: online access to ontology and annotation data. Bioinformatics, 25(2), 288-289.
Chelly, J., & Mandel, J. L. (2001). Monogenic causes of X-linked mental retardation. [Review] Nat Rev Genet, 2(9), 669-680.
Chen, M. H., Chen, Y. S., Hsu, J. W., Huang, K. L, Li, C. T, Un, W. C, ... Bai, Y. M. (2014). Comorbidity
of ADHD and subsequent bipolar disorder among adolescents and young adults with major
depression: a nationwide longitudinal study. [Journal article]. Bipolar Disord, 8(10), 12266. Conners, C. K., Sitarenios, G., Parker, J. D., & Epstein, J. N. (1998a). The revised Conners' Parent
Rating Scale (CPRS-R): factor structure, reliability, and criterion validity. [Research Support, U S Gov't, P H S]. JAbnorm Child Psychol, 26(4), 257-268.
Conners, C. K., Sitarenios, G., Parker, J. D., & Epstein, J. N. (1998b). Revision and restandardization of the Conners Teacher Rating Scale (CTRS-R): factor structure, reliability, and criterion validity [Research Support, U S Gov't, P H S]. J Abnorm Child Psychol, 26(4), 279-291.
Cousijn, H., Eissing, M., Fernandez, G., Fisher, S. E., Franke, B., Zwiers, M.,... Arias-Vasquez, A. (2014). No effect of schizophrenia risk genes MIR137, TCF4, and ZNF804A on macroscopic brain structure. [Journal article]. Schizophr Res, 10(14), 00418-00416.
Cross-Disorder Group of the Psychiatric Genomics Consortium. (2013). Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet,
381(9875), 1371-1379.
Cubero-Millan, I., Molina-Carballo, A., Machado-Casas, 1., Fernandez-Lopez, L, Martinez-Serrano, S., Tortosa-Pinto, P.,... Munoz-Hoyos, A. (2014). Methylphenidate Ameliorates Depressive Comorbidity in ADHD Children without any Modification on Differences in Serum Melatonin Concentration between ADHD Subtypes. IntJMolSci, 15(9}, 17115-17129.
Curran, S., Ahn, J., Grayton, H., Collier, D., & Ogilvie, C. (2013). NRXN1 deletions identified by array comparative genome hybridisation in a clinical case series -further understanding of the relevance of NRXN1 to neurodevelopmental disorders. Journal of Molecular Psychiatry, 1(1), 1-7. doi: 10. 1186/2049-9256-1-4.
de Bruin, E., Ferdinand, R., Meester, S., de Nijs, P. A., & Verheij, F. (2007). High Rates of Psychiatric Co-Morbidity in PDD-NOS. Journal of Autism and Developmental Disorders, 37(5), 877-886.
doi: 10. 1007/S10803-006-0215-X.
Demian, M. (2011). ADHD in Adulthood: Is it Mere Persistence? A Whorfian Dilemma. Journal of Undergraduate Life Sciences, 5(1).
Destrieux, C., Fischl, B., Dale, A., & Halgren, E. (2010). Automatic parcellation of human cortical gyri and sulci using standard anatomical nomendature. Neuroimage, 53(1), 1-15.
Doherty, J. L, & Owen, M. J. (2014). Genomic insights into the overlap between psychiatric disorders: implications for research and clinical practice. [Review]. Genome Med, 6(4).
Elia, J., Gai, X., Xie, H. M., Perin, J. C., Geiger, E., Glessner, J. T.,... White, P. S. (2010). Rare structural variants found in attention-deficit hyperactivity disorder are preferentially associated with neurodevelopmental genes. Mol Psychiatry, 15(6}, 637-646.
Elia, J., Glessner, J. T., Wang, K., Takahashi, N., Shtir, C. J., Hadley, D.,... Hakonarson, H. (2011). Genome-wide copy number variation study associates metabotropicglutamate receptor gene networks with attention deficit hyperactivity disorder. Nat Genet, 44(1), 78-84. Fischl, B. (2012). FreeSurfer. [Historical Article Research Support, N I H , Extramural Research
Support, Non-U S Gov't Review]. Neuroimage, 62(2), 774-781.
Fliers, E., Rommelse, N., Vermeulen, S. H., Altink, M., Buschgens, C. J., Faraone, S. V.,... Buitelaar, J. K. (2008). Motor coordination problems in children and adolescents with ADHD rated by parents and teachers: effects of age and gender. [Multicenter Study Research Support, N I H , Extramural]. J Neural Transm, 115(2}, 211-220.
Franke, B., Faraone, S. V., Asherson, P., BuitelaarJ., Bau, C. H., Ramos-Quiroga, J. A.,... Reif, A. (2012). The genetics of attention deficit/hyperactivity disorder in adults, a review. [Research Support, Non-U S Gov't Review]. Mol Psychiatry, 17(10), 960-987.
Franke, B., Vasquez, A. A., Johansson, S., Hoogman, M., Romanos, J., Boreatti-Hummer, A.,... Reif, A. (2010). Multicenter analysis of the SLC6A3/DAT1 VNTR haptotype in persistent ADHD
suggests differential involvement of the gene in childhood and persistent ADHD.
[Comparative Study Meta-Analysis Multicenter Study Research Support, Non-U S Gov't]. Neuropsychopharmacology, 35(3), 656-664.
Gadow, K. D., DeVincent, C. J., & Pomeroy, J. (2006). ADHD symptom subtypes in children with pervasive developmental disorder. [Research Support, Non-U S Gov't]. J Autism Dev Disord, 36(2), 271-283.
Ghanizadeh, A. (2009). Psychiatric comorbidity differences in clinic-referred children and adolescents with ADHD according to the subtypes and gender. J Child Neural, 24(6), 679-684.
Greydanus, D. E., & Pratt, H. D. (2005). Syndromes and disorders associated with mental retardation. [Comparative Study Review]. Indian J Pediatr, 72(10), 859-864.
Guadalupe, T., Zwiers, M. P., Teumer, A., Wittfeld, K., Vasquez, A. A., Hoogman, M.,... Francks, C. (2014). Measurement and genetics of human subcortical and hippocampal asymmetries in large datasets. [Research Support, Non-U S Gov't]. Hum Brain Mapp, 35(7}, 3277-3289. Hamosh, A., Scott, A. F., Amberger, J. S., Bocchini, C. A., & McKusick, V. A. (2005). Online Mendelian
Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders. [Research Support, Non-U S Gov't Research Support, U S Gov't, P H S]. Nucleic Acids Res, 33(Database issue), D514-517.
Harris, J. C. (2006). Intellectual disability: understanding its development, causes, classification, evaluation, and treatment. New York: Oxford University Press.
Hehir-Kwa, J. Y., Pfundt, R., Veltman, J. A., & de Leeuw, N. (2013). Pathogenic or not? Assessing the clinical relevance of copy number variants. Clin Genet, 84(5), 415-421.
Hoogman, M., Buitelaar, J. K., Franke, B., Cools, R., & Arias-Vasquez, A. (2012). Imaging the effects of ADHD genes. Donders series, Donders Institute for Brain, Cognition and behavior.
Hoogman, M., Guadalupe, T., Zwiers, M. P., Klarenbeek, P., Francks, C., & Fisher, S. E. (2014). Assessing the effects of common variation in the FOXP2 gene on human brain structure. Front Hum Neurosd, 8(473).
Hoppman-Chaney, N., Wain, K., Seger, P. R., Superneau, D. W., & Hodge, J. C. (2013). Identification of single gene deletions at 15ql3. 3: further evidence that CHRNA7 causes the 15ql3.3
microdeletion syndrome phenotype. [Research Support, Non-U S Gov't]. Clin Genet, 83(4), 345-351.
Hunter, J. E., Epstein, M. P., Tinker, S. W., Abramowitz, A., & Sherman, S. L. (2012). The FMR1 premutation and attention-deficit hyperactivity disorder (ADHD): evidence for a complex inheritance. [Research Support, N I H , Extramural]. Behav Genet, 42(3), 415-422.
Inlow, J. K., & Restifo, L L. (2004). Molecular and comparative genetics of mental retardation. [Research Support, U S Gov't, P H S]. Genetics, 166(1}, 835-881.
Jenkinson, M., Beckmann, C. F., Behrens, T. E., Woolrich, M. W., & Smith, S. M. (2012). Fsl. [Historical Article Review]. Neuroimage, 62(2), 782-790.
Jensen, C. M., & Steinhausen, H. C. (2014). Comorbid mental disorders in children and adolescents with attention-deficit/hyperactivity disorder in a large nationwide study. [Journal article]. Atten Defic Hyperact Disord, 19, 19.
Johnson, A. D., Handsaker, R. E, Pulit, S. L, Nizzari, M. M., O'Donnell, C. J., & de Bakker, P. I. (2008). SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap. Bioinformatics, 24(24), 2938-2939.
Jolly, L. A., Homan, C. C., Jacob, R., Barry, S., & Gecz, J. (2013). The UPF3B gene, implicated in intellectual disability, autism, ADHD and childhood onset schizophrenia regulates neural progenitor cell behaviour and neuronal outgrowth. [Research Support, Non-U S Gov't]. Hum MolGenet, 22(23), 4673-4687.
Kadesjo, B., & Gillberg, C. (2001). The Comorbidity ofADHD in the General Population of Swedish School-age Children. Journal of Child Psychology and Psychiatry, 42(4), 487-492. doi: 10. 1111/1469-7610.00742.
Kent, W. J., Sugnet, C. W., Furey, T. S., Roskin, K. M., Pringle, T. H., Zahler, A. M., & Haussler, D. (2002). The human genome browser at UCSC. Genome Res, 12{6], 996-1006.
Kraut, A. A., Langner, I., Lindemann, C., Banaschewski, T., Petermann, U., Petermann, F.,... Garbe, E. (2013). Comorbidities in ADHD children treated with methylphenidate: a database study [Research Support, Non-U S Gov't]. BMC Psychiatry, 13(11), 13-11.
Larson, K., Russ, S. A., Kahn, R. S., & Halfon, N. (2011). Patterns ofcomorbidity, functioning, and service use for US children with ADHD, 2007. [Comparative Study Research Support, N I H , Extramural Research Support, U S Gov't, P H S]. Pediatrics, 127[3), 462-470.
Lee, S. H., Ripke, S., Neale, B. M., Faraone, S. V., Purcell, S. M., Perlis, R. H.,... Wray, N. R. (2013). Genetic relationship between five psychiatric disorders estimated from genome-wide SNPs. [Research Support, N I H , Extramural Research Support, Non-U S Gov't]. NatGenet, 45(9), 984-994.
Lesch, K. P., Selch, S., Renner, T. J., Jacob, C., Nguyen, T. T., Hahn, T.,... Ullmann, R. (2011). Genome-wide copy number variation analysis in attention-deficit/hyperactivity disorder: association with neuropeptide Y gene dosage in an extended pedigree. Mol Psychiatry, 16(S}, 491-503.
Levy, S., Katusic, S. K., Colligan, R. C., Weaver, A. L, Kitlian, J. M., Voigt, R. G., & Barbaresi, W. J. (2014). Childhood ADHD and risk for substance dependence in adulthood: a longitudinal, population-based study. [Research Support, N I H , Extramural]. PLoS One, 9(8).
Leyfer 0. T., Folstein, S. E., Bacalman, S., Davis, N. 0., Dinh, E., Morgan, J.,... Lainhart, J. E. (2006). Comorbid psychiatric disorders in children with autism: interview development and rates of disorders. [Research Support, N I H , Extramural Research Support, Non-U S Gov't]. J Autism DevDisord, 36(7), 849-861.
Li, Kwan, J. S., & Sham, P. C. (2012). HYST: a hybrid set-based test forgenome-wide association studies, with application to protein-protein interaction-based association analysis. Am J Hum Genet, 91(3), 478-488.
Li, Sham, P. C., Cherny, S. S., & Song, Y. Q. (2010). A knowledge-based weighting framework to boost the power ofgenome-wide association studies. [Research Support, Non-U S Gov't]. PLoS One, 5(12), 0014480.
Li, Wilier, C. J., Ding, J., Scheet, P., & Abecasis, G. R. (2010). MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. [Research Support, N I H , Extramural]. Genet Epidemiol, 34(8), 816-834.
Li, M. X., Sham, P. C., Cherny, S. S., & Song, Y. Q. (2010). A knowledge-based weighting framework to boost the power of genome-wide association studies. [Research Support, Non-U S Gov't]. PLoS One, 5(12), 0014480.
Lipkin, P. H. (2013). Methylphenidate reduces ADHD symptoms in children with severe ADHD and intellectual disability. [Comment]. Evid Based Ment Health, 16(4), 2013-101454.
Liu, J. Z., McRae, A. F., Nyholt, D. R., Medland, S. E., Wray, N. R., Brown, K. M.,... Macgregor, S. (2010). A versatile gene-based test for genome-wide association studies. Am J Hum Genet, 87(1), 139-145.
Martin, J., Cooper, M., Hamshere, M. L, Pocklington, A., Scherer, S. W., Kent, L,... Holmans, P. (2014). Biological overlap ofattention-deficit/hyperactivity disorder and autism spectrum disorder: evidence from copy number variants. [Research Support, Non-U S Gov't]. J Am Acad Child Adolesc Psychiatry, 53(7), 761-770.
Maulik, P. K. H. C. K. (2010). Epidemiology of Intellectual Disability. In S. J. H. B. M. (Ed.), International Encyclopedia of Rehabilitation.
Moreno-De-Luca, A., Myers, S. M., Challman, T. D., Moreno-De-Luca, D., Evans, D. W., & Ledbetter, D. H. (2013). Developmental brain dysfunction: revival and expansion of old concepts based on new genetic evidence. [Research Support, N I H , Extramural Review]. Lancet Neural, 12{4), 406-414.
Moruzzi, S., Rijsdijk, F., & Battaglia, M. (2014). A twin study of the relationships among inattention, hyperactivity/imputsivity and sluggish cognitive tempo problems. [Research Support, Non-U S Gov't Twin Study]. J Abnorm Child Psychol, 42(1), 63-75.
Neale, B. M., Medland, S., Ripke, S., Asherson, P., Franke, B., Lesch, K. P., & Faraone, S. V. (2012, unpublished). Meta-analysis ofgenome-wide association studies of attention
deficit/hyperactivity disorder
