The Molecular Genetics of ADHD: A View From the IMAGEProject

By Stephen V. Faraone, Ph.D., and Philip Asherson, MRCPsych, Ph.D.

Psychiatric Times

August 2005 Vol. XXII Issue 9

Although attention-deficit/hyperactivity disorder is frequently misunderstood as caused bynormal childhood energy, boring classrooms, or overstressed parents and teachers, severaldecades of research show ADHD to be a valid disorder with a neurobiological basis (Faraone, in press). Genetic studies have played a leading role in clarifying the biological basis of thedisorder. Family studies have documented familial transmission; adoption studies show thistransmission occurs through biological, not adoptive relationships; and twin studies show thatADHD is highly heritable such that genes account for about 75% of the disorder's variability inthe population (Faraone et al., in press). With a prevalence of 8% to 12% (Faraone et al., 2003),it is among the most common of psychiatric disorders.Given this strong evidence from epidemiologic studies, molecular genetic studies have begun thesearch for genes that increase susceptibility to ADHD. Two general approaches have been used.Genome scan linkage studies scan the entire genome in search of regions that might harbor susceptibility genes. They do not require a prior hypothesis about which genes cause thedisorder. In contrast, candidate gene studies nominate specific genes based on a biological theoryabout their putative role. They use the method of association to test these prior hypotheses.To date, there have been three research groups performing genome linkage scans of ADHD(Arcos-Burgos et al., 2004; Bakker et al., 2003; Fisher et al., 2002; Ogdie et al., 2004, 2003).

Table 1

summarizes their main results and gives their highest LOD scores (a statistic thatmeasures the evidence for linkage). With the exception of chromosome 17p11, genomic regionsimplicated by these studies do not overlap. Because three of the implicated regions (17p11,16p13 and 15q15) have also been implicated in studies of autism, these findings may suggest ashared genetic susceptibility for these disorders.For ADHD, the modest replication across studies completed so far suggests that genes of moderately large effect are unlikely to exist. For this reason, further studies are needed toincrease the power of available linkage data. Difficulties in replication may also be due togenetic heterogeneity (i.e., there are different forms of ADHD influenced by different genes).In contrast to the small number of linkage studies, numerous candidate gene studies have usedassociation to determine if biologically relevant genes influence the susceptibility to ADHD.Much of this work has focused on genes in catecholaminergic systems because the drugs thateffectively treat ADHD are either dopamine reuptake inhibitors (e.g., stimulants) or norepinephrine reuptake inhibitors (e.g., tricyclic antidepressants, atomoxetine [Strattera]).

Compared with dopaminergic and noradrenergic systems, serotonergic systems have receivedrelatively little attention in ADHD research. This is because measures of serotonin metabolismare minimally related to the clinical efficacy of the medicines that treat ADHD. Nevertheless,molecular genetic studies have examined serotonergic genes due to the well-known role of serotonin in impulsivity, one of the core symptoms of ADHD (Brunner and Hen, 1997).Another biological pathway to ADHD was discovered through study of coloboma mice, whichhave a hemizygous deletion of chromosome 2q (Wilson, 2000). This deletion region includes thegene encoding SNAP-25, a neuron-specific protein implicated in exocytotic neurotransmitter release. Coloboma mice show spontaneous hyperactivity, delays in achieving complex neonatalmotor abilities and learning deficiencies. These problems are not seen if the mice are given afunctioning

SNAP25

gene through a transgenic procedure. Treatment with mixed amphetaminesalts (Adderall) but not methylphenidate (Concerta, Metadate, Ritalin) reverses the hyperactivity,which is consistent with the mechanism of action of these medications.Methylphenidate treats ADHD by blocking the dopamine transporter. Mixed amphetamine salts block the dopamine transporter but also facilitates the non-vesicular release of dopamine throughreverse transport, which would be expected to reverse the deficits in exocytotic neurotransmitter release caused by the coloboma mutation.To test genes associated with these biological hypotheses, candidate gene studies have used case-control or family-based designs. Case-control designs compare allele frequencies between patients with ADHD and non-ADHD controls. Alleles that confer risk for ADHD should be morecommon among patients with ADHD. The family-based design compares the alleles that parentstransmit to children with ADHD to alleles parents do not transmit. If an allele increases the risk for ADHD, it should be more common among the transmitted alleles than the nontransmittedalleles. From both study designs, it is possible to derive an odds ratio (OR) that assesses themagnitude of the association between ADHD and the putative risk alleles. An OR of 1.0indicates no association; >1.0 indicates the allele increases risk for ADHD; and <1.0 indicatesthe allele decreases the risk for ADHD.Faraone et al. (in press) reviewed the ADHD candidate gene literature and examined pooled ORsfor candidate genes that had been examined in at least three case-control or family-basedassociation studies.

Table 2

shows seven genes that provide statistically significant evidence of association with ADHD. Other genes have been studied, showing either negative (catechol O-methyltransferase, norepinephrine transporter) or unclear (D2 and D3 dopamine receptors;tyrosine hydroxylase; noradrenergic receptors ADRA2A, 2C and 1C; monoamine oxidase A;serotonin 2A receptor; the A4 and A7 acetylcholine receptor subunits, and the 2A glutamate)results.The ORs for the positive associations range from 1.13 to 1.45 (

Table 2

). These small ORs areconsistent with the idea that many genes of small effect mediate the genetic vulnerability toADHD. Moreover, they suggest one explanation for the frequent failure to replicate initialreports of association: Many individual studies are underpowered to find significant associationsif the effects are modest (Lohmueller et al., 2003). These small and sometimes inconsistent

effects emphasize the need for future molecular genetic studies to implement strategies that will provide enough statistical power to detect small effects.To address this project, we and colleagues from multiple sites in Europe conceived theInternational Multi-site ADHD Genetics (IMAGE) project. The main aim of IMAGE is togenerate a clinical and genetic resource of 1,400 sibling pairs and their biological parents. Thesibling pair design will enable the use of linkage analysis to identify chromosomal regionscontaining genes of moderate-to-large effect and association strategies to identify genes of smalleffect. The IMAGE group uses a novel approach by including ADHD probands and all availablesiblings in one dataset.In order to increase confidence in the diagnosis and decrease potential genetic and etiologicalheterogeneity, probands are recruited from ADHD treatment centers and selected for

DSM-IV

combined subtype. All siblings of probands are included in the study, whether or not they haveADHD, and continuous rating scale measures of symptoms will be used to map genes.The use of quantitative measures to map genes for common disorders is known as

quantitativetrait locus

(QTL)

analysis

and reflects the view that genetic influences on ADHD arecontinuously distributed throughout the population (Asherson and IMAGE Consortium, 2004).This means that genes that increase risk for ADHD are also expected to influence individualdifferences in ADHD symptoms throughout the entire population. Similar approaches have beenused for other common traits such as blood pressure (Harrap et al., 2002), cholesterol level (Lin,2003) and dyslexia (Cardon et al., 1994).The use of quantitative trait measures for ADHD genetic research is based on findings fromnumerous population-based twin studies that show high heritability for parent and teacher ratingsof ADHD symptoms (Thapar et al., 1999). We know, for example, that correlations for parent-and teacher-rated ADHD symptoms are 70% to 75% for identical (monozygotic [MZ]) twins(who share 100% of their genes) and 30% to 35% for nonidentical (dizygotic [DZ]) twins (whoshare 50% of their genes).In order to examine the genetic correlation between ADHD diagnosis and continuous rating scalemeasures, a method called

DF analysis

estimates group heritability from the differentialregression of identical and nonidentical co-twin trait scores to the population trait mean wheretwin probands are selected for extreme scores (DeFries and Fulker, 1988). In one study of 6,000 preschool twins, the group heritability for twins where probands were selected for extreme scoresranged from 0.83 to 0.93 (Price et al., 2001).Using the IMAGE dataset, we estimated the familial correlation between combined-type ADHD probands and continuous measures of ADHD symptoms among siblings. As expected, bothinattentive and hyperactive-impulsive symptoms show familial associations to combined-typeADHD of around 0.2 to 0.3, which is similar to sibling correlations for DZ twins estimated from population samples (

Table 3

).There are several potential advantages to our study design. The collection of DNA from a largesample of ADHD probands allows us to perform powerful tests of association that can identify