Association of a polymorphism of the dopamine transporter gene with externalizing behavior problems and associated temperament traits: A longitudinal study from infancy to the mid-teens

INTRODUCTION

The dopamine transporter gene (SLC6A3 or DAT1) has been a prime candidate for association studies with attention-deficit hyperactivity disorder (ADHD). Dopamine transporter levels in the brain have been found to be around 70% higher in adults with ADHD compared to controls [Dougherty et al., 1999]. Deletion of this gene in mice results in hyperactivity [Giros et al., 1996] and the dopamine transporter is the locus of action of stimulant drugs that have been used in the treatment of ADHD. The major polymorphism of the dopamine transporter gene so far studied involves a variable number of tandem repeats (VNTR) situated in the 3′ untranslated region of the gene [Vandenbergh et al., 1992]. The number of 40 bp repeat copies varies from 3 to 11 (PCR product length of 200 to 520 bp). The functional status of this polymorphism is uncertain. It had been thought that it might be in linkage disequilibrium with a functional polymorphism located in a nearby exon. However, systematic screening for DNA sequence variation in the coding region has found no common variants that could account for any associations [Grünhage et al., 2000; Vandenbergh et al., 2000]. Linkage disequilibrium with a polymorphism in the regulatory region of the gene is another possibility, but it has been found that linkage disequilibrium with the 3′ VNTR does not extend as far as the 5′ region [Grünhage et al., 2000]. It remains possible that the VNTR itself has functional significance. In support of this possibility, it has recently been reported that the 9-repeat/10-repeat genotype is associated with a 22% reduction in dopamine transporter protein availability in the putamen compared to the 10/10 genotype [Heinz et al., 2000].

In a number of family-based association studies, the 10-repeat allele has been found to be significantly associated with ADHD [Cook et al., 1995; Gill et al., 1997; Waldman et al., 1997; Daly et al., 1999]. However, there has been a negative finding as well [Swanson et al., 2000]. Homozygosity for the 10-repeat allele has also been associated with nonresponse to methylphenidate treatment in ADHD children [Winsberg and Comings, 1999].

All of these studies have been with children with clinically diagnosed ADHD. However, as pointed out by Thapar et al. [1999], hyperactivity is a continuous trait in the population. It may be profitable for genetic studies to search for susceptibility genes using continuous measures of hyperactivity or associated temperament traits and so avoid biases associated with clinic referral [Thapar et al., 1999]. A temperament trait hypothesized to be associated with ADHD is novelty seeking, which involves behaviors such as impulsivity and excitability [Thapar et al., 1999]. One study of adults did find that low novelty seeking was associated with the 9-repeat allele of the dopamine transporter gene [Sabol et al., 1999], but other studies have not found an association [Hill et al., 1999; Jorm et al., 2000a]. Temperamental traits in children have not been investigated.

In the present study, we report data on the dopamine transporter VNTR polymorphism in relation to hyperactivity, to other externalizing behavior problems, and to related temperamental traits in a general population sample. We collected DNA from a subgroup of children participating in the Australian Temperament Project, a longitudinal study with temperament data from 4–8 months to 15–16 years and behavior problem data from 3–4 to 15–16 years [Prior et al., 2000]. The advantage of the longitudinal data is that it allowed us to assess associations with the dopamine transporter VNTR at various stages of development. Gene expression may vary at different stages of development, so that associations are found at some ages but not others [McClearn et al., 1998]. We have previously reported data from the same study on the association of a polymorphism of the serotonin transporter gene with anxiety-related traits and behavior problems [Jorm et al., 2000b].

MATERIALS AND METHODS

RESULTS

The genotype frequencies were 362 with genotype 10/10, 256 with 9/10, 45 with 9/9, 10 with 10/11, 5 with 9/11, and 1 with each of 7/10 and 8/10. Collapsing the genotypes according to the presence or absence of allele 10 gave 362 with the 10/10 genotype, 268 with the 10/* genotype, and 50 with the */* genotype. The frequency of allele 10 was 73% in the current study, which is similar to other mixed European populations: 71% [Gelernter et al., 1998], 71% [Jorm et al., 2000a, 2000b], and 69% [Kang et al., 1999].

Table I shows the means and standard deviations for the behavior problem and temperament measures across the three genotypes. No differences were significant at the P < 0.05 level. An analysis of the data for males and females separately revealed similar results, with no significant group differences at P < 0.05. Because there may be ethnic differences in allele frequency even among European populations [Kang et al., 1999], we also analyzed the data excluding the 66 children who had at least one parent from southern Europe. This analysis showed an identical pattern of results, namely, no significant group differences.

DISCUSSION

This study showed no association of the dopamine transporter VNTR polymorphism with externalizing behavior problems or related temperamental traits at any age. Given that previous clinical studies have found that the 10-repeat allele is associated with ADHD, some associations were expected. There are a number of possible reasons for the discrepancy in findings.

The first is that the studies of ADHD involved clinic samples, whereas the present study involved a general population sample. However, if ADHD is the extreme of a continuum rather than categorically distinct [Levy et al., 1997], then consistent associations should be found in both types of sample. Nevertheless, it is possible that using an extreme clinical sample gives greater power to show small effects, even if the same underlying continuum is being studied. Unfortunately, because of the very different design in the present study, it is not possible to determine the power to detect the effects found previously. However, we can say that a sample of 660 has > 70% power to detect associations accounting for 1% of the variance, with P < 0.05 [Cohen, 1988]. So if there is a real effect not detected by the present study, it must be very small.

Another major difference is in the use of a community sample rather than a family clinic sample. It is often argued that family controls are superior because they are not subject to confounding due to population stratification. We did attempt to overcome this type of confounding by excluding children with southern European ancestry, but it is still possible that there is more subtle population stratification than we were unable to measure. On the other hand, general population samples have other advantages over family clinic studies that are often overlooked [Jorm and Easteal, 2000]. For example, family studies can be affected by transmission ratio distortion at some loci, which involves non-Mendelian segregation of alleles to live born offspring [Paterson and Petronis, 1999]. Furthermore, clinic samples are subject to many referral biases that general population samples are not. For example, clinic samples may be more chronic, have greater comorbidity, be particularly responsive or resistant to certain types of treatment, and have parents who are more distressed. All of these biases can lead to spurious associations. Because of their complementary strengths and weaknesses, family and population studies are both necessary for establishing replicable associations with candidate polymorphisms.

A particular strength of the present study is the availability of longitudinal data spanning the period from infancy to adolescence. Although no effects were found at any age, it is possible for allelic associations to be specific to particular points in development. Association studies are usually based on a restricted part of the life span and inconsistent findings could be due to developmental differences between samples. A developmental perspective should be more widely adopted.