جنبه های ژنتیکی مولکولی نقص توجه/بیش فعالی

Two genome wide scans, one of which was subsequently extended, have led to the identification of different chromosomal regions assumed to harbour genes underlying attention-deficit/hyperactivity disorder (ADHD). Some of these regions were also identified in patients with autism and/or dyslexia. The only region for which both studies detected a LOD score >1 was on chr 5p13 which is in the vicinity of the location of the candidate gene DAT1. The candidate gene approach has revealed the most robust and replicated findings for DRD4, DRD5, and DAT1 polymorphisms. Meanwhile interesting endophenotype studies have also been conducted suggesting a genetic basis for different diagnostic and therapeutic criteria. Animal studies for ADHD have investigated especially hyperactivity and have focused mainly on knockout and QTL designs. In knockout mice models the most promising results were obtained for genes of the dopaminergic pathway. QTL results in rodents suggest multiple loci underlying different forms of natural and induced hyperactivity. The molecular results mentioned above are presented and discussed in detail, thus providing both clinicians and geneticists with an overview of the current research status of this important child and adolescent psychiatric disorder.

1.1. Definition of the syndrome and diagnostic criteria Attention-deficit/hyperactivity disorder (ADHD) is currently diagnosed according to the DSM-IV-TR criteria [1]. The predominantly inattentive, predominantly hyperactive/impulsive or combined type is diagnosed if a threshold number of symptoms of inattention and/or hyperactivity/impulsivity apply. As for any other psychiatric disorder, we need to consider the possibility that the diagnostic criteria including the delineation of these three different types might be suboptimal with respect to the elucidation of the molecular genetic basis of the underlying biologically relevant traits. This concern applies particularly to ADHD, because the most frequent combined type is based on symptoms, which represent the upper and lower ends of the seemingly unrelated quantitative distributions for activity and attention, respectively. Thus, from a genetic viewpoint, it might be argued that a separate analysis of these two quantitative traits might be more straightforward. Their joint analysis and hence of the disorder as such is, however, warranted because inattention and hyperactivity co-occur considerably more frequently than can be expected by chance. Thus, in unselected twins correlations of 0.6–0.9 have been reported for symptoms of inattention and hyperactivity [2]. Furthermore, both twin and family studies indicate that the type does not breed true. Thus, a specific type in one twin of a monozygotic twin pair (MZ) does not predict the type in the second twin [3] and [4]. Within pedigrees, affecteds can have any one of the types [5] and [6]; no familial clustering of a particular type occurs. As with any other psychiatric disorder, the reliability and validity of the diagnostic criteria are of crucial importance for studies attempting to identify the molecular basis of ADHD. The rater-effect, which has repeatedly been observed for ADHD [3], [7] and [8], could lead to different heritability estimates depending on the respective informant. Whereas the estimates based on different informants are largely within the same range [7] and [8], the fact that the same child can be rated very differently with respect to the core symptoms of ADHD by mothers, fathers, teachers and clinicians, underscores the need to use an as uniform as possible phenotypical assessment procedure. ADHD according to DSM-IV-TR is a categorical diagnosis. At the same time the use of a threshold number of symptoms to define hyperactivity, impulsivity and inattention clearly indicates that these core phenotypes are viewed dimensionally. For the initiation and interpretation of both formal and molecular genetic studies it is important to distinguish a categorical vs. a dimensional conceptualisation of ADHD and to realize the potential advantages and disadvantages of both approaches. Findings indicative of cross-cultural differences in prevalence rates of ADHD [9] potentially suggest that the frequency of predisposing (and/or protective) genotypes differs across the world. However, caution is warranted because such differences in prevalence rates might at least partially be due to culturally divergent ratings of ADHD symptoms [10] and/or to socio-cultural influences on relevant clinical symptoms. 1.2. Formal genetic studies and heritability estimates Several formal genetic studies have addressed the contribution of both genetic and environmental factors to the development of ADHD using both categorical and dimensional definitions. Twin studies, for example, have come up with concordance rates between about 50 and 80% for MZ twins vs. 30–40% for DZ twins [11]. MZ and DZ correlations for quantitative traits of ADHD of between 0.48 and 0.92 and −0.16 and 0.57, respectively, also indicate substantial heritability (for review see Ref. [11]). Based on these results, heritability of ADHD is estimated at approximately 0.8 [12]. The importance of genetic factors in the etiology of ADHD is also supported by the results of adoption studies: biological parents and sibs of an ADHD-child are significantly more often affected by ADHD (and comorbid disorders) than the adoptive parents and sibs [11], [13], [14] and [15]. According to DSM-IV, comorbid disorders are diagnosed separately. However, it is conceivable that the genetic factors underlying inattention and/or hyperactivity/impulsivity at the same time predispose to other psychopathological or cognitive symptoms in subgroups of ADHD patients. Hence the comorbidity might be useful for defining these genetically potentially more homogeneous subgroups. Indeed, formal genetic evidence suggests that in genetic terms ADHD with and without comorbid conduct disorder differ [16], [17] and [18]; conduct disorders, but not affective and anxiety disorders, cosegregate within families [18]. 1.3. Heterogeneity in ADHD The recent results of genome wide linkage analyses have detected single chromosomal peak regions which overlap with those identified previously for autism and/or reading disorders [19], [20], [21], [22], [23], [24], [25], [26], [27] and [28]. Whereas it is currently unknown if these overlapping regions indeed indicate a gene(s) predisposing to more than one disorder, we need to keep this possibility in mind. Optimally, phenotyping should include careful assessment of comorbid disorders. Such extensive evaluations are however costly and thus have a negative impact on case numbers given a specified amount of research funds. Furthermore, future research might point to endophenotypes that from a current viewpoint do not seem a top priority. In this context the high ADHD prevalence rate of approximately 3–10% [29], [30] and [31] in itself suggests that the disorder is not homogeneous. As in other complex disorders, it is likely that in etiological and in particular in genetic terms distinct types of ADHD exist; this is particularly the case if infrequent monogenic forms of ADHD exist. The more polygenic the disorder is, the more alleles will not only overlap between affecteds; they will also occur with albeit lower frequency in unaffecteds. If and to what extent all these different (overlapping) types can be differentiated at the phenomenological level is a matter of debate. The recent elucidation of mutations in the melanocortin-4 receptor gene as a cause of obesity in 2–4% of obese children shows that such a delineation is not necessarily always possible [32]. Molecular analyses of quantitative traits can benefit substantially from ascertainment schemes which concentrate on those individuals with the most extreme concordant and/or discordant phenotypes [33]. Indeed, ADHD symptoms have been assessed dimensionally both in studies of heritability (e.g. [2]) and in gene localization studies [34]. In this context, it would seem helpful to know more about the quantitative distribution of the relevant traits in the general population in comparison to for instance sibs of ADHD probands. However, from a clinical viewpoint such studies are rendered difficult by the potential temporal instability of symptoms and the developmental changes during the course of ADHD. 2. Genome wide linkage analyses Genome wide linkage analyses offer the advantage that a priori hypotheses as to functional candidate genes that could influence the phenotype are not required. Thus, genes (and pathways) can be identified that had previously not been implicated in the respective disorder (such as NOD2 in Crohn's disease [35] and [36]). Model-free or non parametric linkage analysis also does not require specification of a genetic disease model and is based on the identity by descent (IBD) of marker alleles in pairs of affected relatives such as affected sib pairs. If a specific marker is not linked to a disease locus, the IBD score in sib pairs is on average 1, which corresponds to the expected rate of allele sharing (50%) between sibs: two sibs can share 0, 1 or both parental alleles, the probabilities of which are 25, 50 and 25%, respectively. The significance of an observed deviation from this expected distribution is often given as a logarithm of the odds (LOD) score [37]. Current genome scans are typically based on 350–1000 microsatellite markers spaced rather evenly throughout the genome with marker distances of about 10–3 cM, so a correction for multiple testing is necessary [38]. It is a matter of debate as to what LOD score actually constitutes a significant result for a complex disorder. Lander and Kruglyak [39] have proposed that for non parametric studies (respective values for parametric studies in parentheses) LOD scores ≥3.6 (3.3) and between 2.2 (1.9) and 3.6 (3.3) indicate definite and suggestive evidence of linkage, respectively; accordingly LOD scores below 2.2 (1.9) are not even suggestive of linkage. LOD scores ≥3.6 are only infrequently observed in genome scans for complex disorders; the probability of detecting such a high LOD score increases with the number of available relative pairs [40]. Some researchers prefer the usage of P-values instead of LOD scores particulary for non-parametric studies to delineate evidence (P≤2.2×10−5) and suggestive evidence for linkage (P≤7.4×10−4). It is important to understand the major limitations of genome wide linkage scans. Because genetic heterogeneity—different genes are operative in different families—applies to a complex disorder, both the sample size and the effect sizes of the underlying genes are of crucial importance. A genome scan based on 100 sib-pairs allows detection of definite linkage only if the underlying disease gene(s) is operative in several of these families and the effect size(s) is large [38]. Increasing the sample size will potentially allow detection of linkage regions harbouring infrequent major genes (i.e. gene variants with a major effect) or more common genes with lower effect sizes (oligogenes). Genes with a minor effect (polygenes) escape detection unless thousands of sibships are analysed [41]. This is due to the fact, that the probability that the same minor gene (among potentially up to several hundreds of other predisposing minor genes operative in the families) is inherited to affected sibs within a sibship is only minimally elevated above the expected 50%. Currently, there is no consensus as to the extent to which major genes contribute to complex disorders. The identification of major genes detected via linkage studies has been successful in single disorders with complex inheritance such as Crohn's disease [35] and [36] and breast cancer [42] and [43]. If however the familial loading for a particular disorder is soley based on polygenic inheritance, genome wide linkage studies will prove futile. Accordingly, even LOD scores indicative of evidence for linkage obtained for such disorders would prove to be false positives. The sibling relative risk (λsib) [44] potentially provides a clue as to the role of major genes. It is calculated by dividing the rate for a specific disorder observed in sibs of an index patient by the population based prevalence rate. For ADHD, λsib is in the magnitude of 5 [18] and thus sufficiently high enough to base linkage studies on it. This of course only holds true if merely a limited number of genes underly the λsib, which accordingly would have a rather strong effect on the phenotype. The available formal genetic studies did not include enough data on more distant types of relationships to be able to estimate how many genes contribute to this λsib. The concordance rates observed in twin studies suggest that their number might indeed not be that large. Genome scans have only recently been initiated in ADHD [19], [20], [25] and [29]. The first genome scan was based on an US sample encompassing 126 affected sib pairs from 104 families [28], which was subsequently extended to 204 families with a total of 270 sibpairs [25]. An independent genome scan was performed in a Dutch sample [19] based on 164 sib-pairs from 106 families. Fischer et al. [20] and Odgie et al. [25] diagnosed ADHD according to DSM-IV criteria; ADHD was defined as ‘definite’ when at least six criteria of one of the two symptom lists were met or as ‘probable’ when the subject met only five criteria and addtionally showed significant impairment. All autistic phenotypes were specifically excluded. In contrast, Bakker et al. [19] also included sibs fulfilling DSM-IV criteria for both ADHD and an autistic spectrum disorder. For the narrow phenotype, all sib-pairs (n=117) met the full ADHD criteria without autism. The broad phenotype also allowed inclusion of sib-pairs, if only one sib met full ADHD criteria and the other sib(s) met five of the nine DSM-IV criteria for either inattention and/or hyperactivity/impulsivity. Both groups used a broad variety of diagnostic instruments. Bakker et al. [19] have previously compared the Dutch with the US sample: the US study group [25] was 80% white, with the average age of affected children being 11.1 years, and the average full-scale IQ being 105.9 (IQ≥70). The average age in the Dutch sample was 10 years and the IQ was >80. In the Bakker et al. [19] study all four grandparents of most of the affected sib-pairs were white and of Dutch descent. The rates for the inattentive type differed almost fourfold (12.6% in the Dutch sample and 45% in the enlargened US sample). Other differences between the two samples include the sex ratio (17% girls in the Dutch sample vs. 28% in the American sample; P=0.01), comorbidity with conduct disorder (6% in the Dutch vs. 15% in the US sample; P<0.01) and socioeconomic status (0 vs. 22% in the lowest socioeconomic class in the Dutch and US sample, respectively). Fisher et al. [20] Bakker et al. [19] and Ogdie et al. [25] each used about 400 polymorphic markers spanning all autosomes and the X chromosome (chr). In the US study, specific chromosomal regions were subsequently fine-mapped using additional microsatellite markers. The highest maximum multipoint LOD scores of 3.73 and 3.54 were obtained in the extended American [25] and the Dutch [19] samples for chr 16p13 and 15q, respectively. LOD scores between 2 and 3.3 were detected for an additional six chromosomal regions (Table 1). In the Dutch sample the narrowly defined phenotype despite the inclusion of a lower number of affected sib-pairs generally yielded higher LOD scores than the broad phenotype. In the US sample the broad phenotype likewise did not systematically increase the LOD scores observed for the narrow phenotype [20] and [25]. Both the American and Dutch groups did not find a major locus on the X chr as a potential genetic explanation for the predominance of males in ADHD.

Investigation of heritability in ADHD indicates a high genetic contribution. Association studies have come up with positive results for some of the candidate genes that were subsequently confirmed. This applies particularly to genes of the dopaminergic pathway, which at the same time currently represent the most frequently investigated genes; the candidate gene approach has revealed the most robust and confirmed findings for DRD4, DRD5, and DAT1 polymorphisms. Two genome wide scans have suggested different loci, some of which were also identified in patients with autism and/or dyslexia. The only region for which both studies detected a LOD score>1 was on chr 5p13; interestingly the candidate gene DAT1 is located at the telomere of chr 5p. Meanwhile interesting endophenotype studies have also been conducted suggesting a genetic basis for different diagnostic and therapeutic criteria based on genetic differences. Animal studies for ADHD have investigated especially hyperactivity and have focused mainly on knockout and QTL designs. In mice in which relevant candidate genes for ADHD were knocked out the most promising results were again obtained for genes of the dopaminergic pathway. QTL results in rodents suggest multiple loci underlying different forms of natural and induced hyperactivity. Further molecular genetic studies in humans will identify other genes relevant for ADHD.