Bipolar disorder risk alleles in adult ADHD patients

Download Bipolar disorder risk alleles in adult ADHD patients

Post on 21-Jul-2016




1 download

Embed Size (px)


  • Genes, Brain and Behavior (2011) 10: 418423 doi: 10.1111/j.1601-183X.2011.00680.x

    Bipolar disorder risk alleles in adult ADHD patients

    E. T. Landaas,,, S. Johansson,,,A. Halmy,,, K. J. Oedegaard,,,O. B. Fasmer,, and J. Haavik,,,

    Department of Biomedicine, University of Bergen, K.G.Jebsen Centre for Research on Neuropsychiatric Disorders,University of Bergen Centre of Medical Genetics andMolecular Medicine, Haukeland University Hospital,**Department of Clinical Medicine, Section for Psychiatry,Faculty of Medicine, University of Bergen, and Division ofPsychiatry, Haukeland University Hospital, Bergen, Norway*Corresponding author: J. Haavik, Department of Biomedicine,University of Bergen, Bergen, Norway.E-mail:

    Attention-deficit/hyperactivity disorder (ADHD) has anestimated prevalence of 35% in adults. Genome-wideassociation (GWA) studies have not been performedin adults with ADHD and studies in children have sofar been inconclusive, possibly because of the smallsample sizes. Larger GWA studies have been performedon bipolar disorder (BD) and BD symptoms, and severalpotential risk genes have been reported. ADHD and BDshare many clinical features and comorbidity betweenthese two disorders is common. We therefore wantedto examine whether the reported BD genetic variantsin CACNA1C, ANK3, MYO5B, TSPAN8 and ZNF804Aloci are associated with ADHD or with scores on theMood Disorder Questionnaire (MDQ), a commonly usedscreening instrument for bipolar spectrum disorders. Westudied 561 adult Norwegian ADHD patients and 711controls from the general population. No significantassociations or trends were found between any ofthe single nucleotide polymorphisms (SNPs) studiedand ADHD [odds ratios (ORs) 1.05]. However, a weakassociation was found between rs1344706 in ZNF804A(OR = 1.25; P = 0.05) and MDQ. In conclusion, it seemsunlikely that these six SNPs with strong evidence ofassociation in BD GWA studies are shared risk variantsbetween ADHD and BD.

    Keywords: ADHD, ANK3, BD, CACNA1C, genetics, GWAstudies, MDQ, MYO5B, TSPAN8, ZNF804A

    Received 21 October 2010, revised 11 January 2011,accepted for publication 25 January 2011

    Attention-deficit/hyperactivity disorder (ADHD) is a neuropsy-chiatric disorder characterized by hyperactivity, impulsivityand inattention. Initially it was considered a childhood condi-tion, but it has become increasingly evident that symptoms

    frequently persist into adulthood (Faraone et al. 2006), andthe prevalence of ADHD has been estimated to be inthe range of 35% in adults (Fayyad et al. 2007; Kessleret al. 2006).

    Affective symptoms are common in adult ADHD patients,constitute an important feature of the disorder (Reimherret al. 2005), and it has been suggested that such symptomsshould be among the diagnostic criteria in adults (Wenderet al. 1981). Comorbidity with other psychiatric disorders isalso common in adult ADHD patients (Haavik et al. 2010;Mcgough et al. 2005; Sobanski et al. 2007), and one of thefrequently reported co-occurring diagnoses is bipolar disor-der (BD) (Wingo & Ghaemi 2007). In addition to affectivesymptoms, individuals with BD and ADHD show overlappingsymptoms such as impaired impulse control and dysreg-ulation of energy and activity levels (Skirrow et al. 2009).We have previously reported that approximately 12% of ouradult ADHD patients have self-reported comorbidity with BD(Halmoy et al. 2010). However, 51% of the patients screenedpositive on the Mood Disorder Questionnaire (MDQ), ascreening instrument for bipolar spectrum disorders (BSD)(Hirschfeld et al. 2000), showing that symptoms of maniaand hypomania are highly prevalent in adult ADHD patientsalso in the absence of a diagnosed BD. Neuroimaging studiesare also compatible with partially overlapping pathogeneticmechanisms in these conditions (Passarotti et al. 2010).

    The heritability of childhood ADHD has been estimatedto be about 76% (Faraone et al. 2005). Although many link-age and candidate association studies have been performedin the search for susceptibility genes, findings have beeninconsistent and contradictory (Franke et al. 2009). Overrecent years, genome-wide association (GWA) studies haveresulted in a large number of genetic variants showing highlysignificant associations with traits in several medical spe-cialities (McCarthy 2010; Teslovich et al. 2010), althoughoften with relatively modest effect sizes. However, concern-ing ADHD, no gene region has been established at wholegenome significance so far (Franke et al. 2009). Althoughprogress has been slow even for most other common com-plex mental disorders, there have been some promisingresults, especially in BD, where the numbers of samplesstudied have been considerably larger than for ADHD.

    As both BD and ADHD are highly heritable and often co-occur within families (Birmaher et al. 2010; Faraone et al.1997), one could hypothesize that the two disorders mightshare some common genetic risk factors (Hegerl et al. 2010).The aim of this study was to examine whether singlenucleotide polymorphism (SNP) alleles found associatedwith BD through recent GWA studies are more commonin patients with persistent ADHD than in controls recruitedfrom the general population. We chose to study six SNPs,of which five are located in or near the genes encodingankyrinG (ANK3) (Ferreira et al. 2008), myosin5B (MYO5B),

    418 2011 The AuthorsGenes, Brain and Behavior 2011 Blackwell Publishing Ltd and International Behavioural and Neural Genetics Society

  • Bipolar disorder risk alleles in adult ADHD patients

    tetraspanin-8 (TSPAN8) and the alpha 1C subunit of anL-type voltage-dependent calcium channel (CACNA1C) (Sklaret al. 2008). The sixth SNP is located in the zinc fingerprotein 804A gene (ZNF804A), a gene that was first foundassociated with schizophrenia in a GWA study, but wasfurther found associated with BD, because the P-valuebecame genome-wide significant only after individuals withBD were included in the original patient sample (ODonovanet al. 2008). Additionally, we wanted to test if there was anyassociation between these SNPs and scores on the MDQ.

    Material and methods

    SubjectsOur sample consists of 1272 Caucasians of Norwegian ancestry, allof more than 18 years of age. Of these, 561 were patients diagnosedwith ADHD according to ICD-10 research criteria (World HealthOrganization 1993), with two modifications: allowing the inattentivesubtype in DSM-IV as sufficient for the diagnosis and allowing for thepresence of comorbid psychiatric disorders, as long as the criteria forADHD were present before the appearance of the comorbid disorder(Johansson et al. 2008). These diagnostic criteria are very similarto the DSM-IV criteria for ADHD (American Psychiatric Association2000). The majority of the patients were recruited by responding toan invitation sent by letter to their addresses, as listed in a Norwegiannational registry of adult ADHD patients. The remainder was recruiteddirectly from psychiatrists or out-patient clinics (Johansson et al.2008). Patients who reported mental retardation were excludedfrom analyses. The control group consisted of 711 volunteers fromthe general population (aged 1840 years) recruited from all acrossNorway for the purpose of this study (described in Halmoy et al.2010). Controls were unselected, i.e. no controls were excludedbased on the presence of life-time psychiatric disorders or otherrelated traits. A written informed consent was obtained from allparticipants and the study was approved by the Norwegian RegionalMedical Research Ethics Committee West IRB #3 (FWA00009490,IRB00001872).

    MeasuresAll participants returned a questionnaire, where current and life-timepsychiatric morbidity was reported (Halmy et al. 2010). Additionally,they filled in the MDQ, which is a screening instrument designed tofacilitate the recognition of BSD (Hirschfeld et al. 2000). A positivescore is defined as 7 (of 13) positive items concerning life-timesymptoms of mania and hypomania, co-occurrence of at least two ofthese symptoms and functional impairment caused by the symptomsrated as moderate to severe. MDQ has been validated for use in bothhealthy and psychiatrically ill individuals (Hirschfeld et al. 2000, 2003).For the quantitative MDQ analyses, all individuals with missing itemswere excluded, resulting in analyses of 503 ADHD patients (90%)and 681 controls (96%). A few individuals fulfilled the criteria forMDQ positive or negative despite having missing items, and thusthe dichotomous MDQ analyses included 517 ADHD patients (92%)and 691 controls (97%).

    SNP selection and genotypingSelection of SNPs was based on a literature search as of January2010. SNPs that had been emphasized as likely to be associatedwith BD, at a level of genome-wide significance (P < 5.0 108)or near such values, either in BD GWA studies alone [rs9804190(Schulze et al. 2009); rs10994336 (Ferreira et al. 2008); rs1006737,rs1705236 and rs4939921 (Sklar et al. 2008)] or in combination withschizophrenia cases [rs1344706 in ZNF804A (ODonovan et al. 2008)]were selected as candidates for genotyping in our sample. Only oneSNP, with the strongest P-value, was chosen for each locus tolimit multiple testing issues, except for the ANK3 locus where two

    SNPs were selected as studies have pointed to two independentsignals. The DGKH (diacylglycerol kinase eta) SNP rs1012053 (Baumet al. 2008) failed in assay design and was not included in theanalysis.

    Samples of either whole blood or saliva were obtained fromall participants, and the Oragene DNA Self-Collection Kit (DNAGenotek Inc., Ontario, Canada) was used for DNA extraction. TheDNA was aliquoted into 96-well plates, each of which contained DNAfrom both cases and controls and a minimum of two blank samplesand two internal controls. SNP genotyping was performed using theMassARRAY iPLEX System (Sequenom, San Diego, CA, USA). Thegenotyping rate was 0.99 for all SNPs, and the concordance rate,found through use of internal controls and duplicates, was 100%.

    Statistical analysesThe clinical data and the distribution of the MDQ scores wereanalysed with descriptive methods using 2 tables and t-testsperformed by the Statistical Package for Social Sciences version15.0.1 (SPSS Inc, Chicago, IL, USA). The genetic statistical analyseswere performed with the PLINK software version 1.07 (Purcell et al.2007), based on an additive allelic model and using linear/logisticregression with gender and ADHD status as covariates. Genotypedistributions for all SNPs were consistent with HardyWeinbergequilibrium, P 0.05. For the MassARRAY iPLEX analysis, twoindividuals were excluded because of low genotyping efficiency(missingness >0.3). These were further excluded from all clinicaland genetic analyses. All odds ratio (OR) estimates are presentedfor the allele found associated with increased BD risk in the BDGWA studies. A two-tailed level of P < 0.05 was chosen for nominalsignificance, and all P-values are presented without correction formultiple testing.

    Power calculations in the total sample were performed with thegenetic power calculator (, using a disease prevalence of 0.034 for adult ADHD, asignificance level of 0.05 and assuming an additive allelic model. Weused the allele frequencies detected in our control sample, and theORs were obtained from the original GWA studies.


    Clinical characteristics of the sampleSociodemographic and clinical characteristics of the 561 adultADHD patients and 711 controls from the general populationare summarized in Table 1. Significantly more patients thancontrols reported life-time episodes of depression or anxiety,with a frequency of 68.5% among the patients and 14.2% inthe control group. BD, problems with alcohol and problems

    Table 1: Clinical characteristics of the sample

    ADHD patients Controls

    N 561 711Males, % (N) 51.7 (290) 40.1 (285)Age (SD) 34.1 (10.4) 29.6 (6.5)MDQ score, mean (SE) 8.1 (0.18) 2.9 (0.12)MDQ positive, % (N) 48.7 (252) 6.1 (42)Self-reported morbidity, % (N)

    Depression/Anxiety 68.5 (401) 14.2 (128)BD 11.2 (63) 1.1 (10)Alcohol problems 23.1 (135) 2.1 (19)Problems with illegal drugs 26.4 (155) 2.2 (20)

    Five hundred and three patients and 681 controls with nomissing items.In total, 517 cases and 691 controls.

    Genes, Brain and Behavior (2011) 10: 418423 419

  • Landaas et al.

    with illegal drugs were also much more common in theADHD group than among the controls. ADHD patients scoredsignificantly higher than controls on the MDQ, with a meanscore of 8.1 in the patients and 2.9 in the controls; 48.7%of the patients and 6.1% of the controls were classified asMDQ positive.

    Genetic analysesThe results from the allelic association analyses betweenBD susceptibility SNPs and adult ADHD are shown inTable 2. For comparison, the ORs reported in the original BDGWA studies implicating the SNPs are also listed, togetherwith the estimated power to detect such effects at the0.05 significance level, given our sample size and allelefrequencies. None of the SNPs showed any association withADHD, and no trends in the same direction as the previousBD studies were seen (all ORs < 1.05).

    Table 3 shows the results from the SNP analyses in theentire sample with MDQ score used as the phenotype, bothanalysed as a continuous and a dichotomous measure. Theresults presented are obtained after correction for gender andADHD status, but without correction for multiple testing. Forthe SNP rs1344706 in ZNF804A, we found some evidencefor association between the BD risk allele (T) and a positiveMDQ score (dichotomous measure: OR = 1.25; P = 0.05).Effects of the same size were observed in both ADHDpatients and controls studied separately, when the datawere stratified by ADHD status and gender was used asa covariate (dichotomous measure patients: OR = 1.25;P = 0.09; controls: OR = 1.28; P = 0.28). The associationbetween the MYO5B locus SNP and MDQ was in theopposite direction compared with previous BD findings(dichotomous measure all participants: OR = 0.64; P =0.02; patients: OR = 0.62; P = 0.03; controls: OR = 0.73;P = 0.43).


    We observed no associations between adult ADHD andthe SNPs rs9804190 and rs10994336 in the ANK3 locus,rs1006737 in CACNA1C, rs1705236 in TSPAN8, rs4939921in MYO5B and rs1344706 in ZNF804A. Furthermore, notrends were observed that would support that any of the firstset of risk alleles from BD GWA studies are involved in ADHDsusceptibility. All ORs were

  • Bipolar disorder risk alleles in adult ADHD patients

    Table 3: The effect of BD risk alleles on MDQ score and the allelic distribution between MDQ-positive and MDQ-negative individuals,corrected for gender and ADHD status

    MDQ dichotomous

    MDQ continuous Frequency risk allele

    Chromosome Gene SNP Risk allele P MDQ positive MDQ negetive OR P

    2 ZNF804A rs1344706 T 0.20 0.16 0.61 0.57 1.25 0.0510 ANK3 rs9804190 C 0.14 0.43 0.78 0.79 0.96 0.7710 ANK3 rs10994336 T 0.02 0.95 0.07 0.07 1.17 0.4712 CACNA1C rs1006737 A 0.09 0.53 0.37 0.36 1.1...


View more >