It has been hypothesised that the A1 allele of the dopamine D2 receptor (DRD2) gene has an association with alcoholism (Blum et al., 1990). The current study tested this hypothesis.
The use of the polymerase chain reaction and the TaqI restriction enzyme made it possible to determine the TaqIA genotype in individuals. The genotype of 160 control subjects, 142 Burmese people and 66 alcoholic people was determined.
There were no significant differences in allele frequencies or phenotype distribution in the current study's controls when compared to the alcoholics. The A1 allele frequency in the control population was lower than in the alcoholic population, although not significantly so.
The control population is the current study was significantly different in both allele frequency and distribution from the Burmese population. This was not unexpected since the A1 allele is known to vary in frequency between different ethnic populations (Barr and Kidd, 1993).
The present results do not support an association between the A1 allele of the DRD2 gene and alcoholism. There are several reasons why this might be so.
The samples used for the current study were racially matched in that all subjects were Caucasian. However the subjects were not ethnically matched. The Caucasian population is made up of many ethnic subgroups that may differ in their A1 allele frequencies. If more of these populations were incorporated more often into either the alcoholic or the control group, then any association may be masked by sampling error.
A major difficulty with an association of A1 with alcoholism is that the TaqI A polymorphism is located more than 10 kb downstream from the coding region of the DRD2 gene. A mutation at this site would not be expected to lead to any structural change in the dopamine receptor (Turner et al., 1992). The TaqI A polymorphism is located at such a distance downstream from the coding region that it would be expected that A1 would be in linkage equilibrium with the receptor or 5' areas involved in gene regulation (Uhl et al., 1993).
The possibility remains that there really is no association, and that the A1 allele of the DRD2 gene has nothing to do with the risk or etiology of alcoholism. However a meta-analysis on previously published data suggests that there is an association.
Since the initial study by Blum et al. (1990), several other association studies have been performed. These studies have had conflicting results. Studies by Blum et al. (1991), Noble et al. (1991), Parsian et al. (1991b), Comings et al. (1991), and Smith et al. (1992) suggested an association between the A1 allele of the dopamine D2 receptor with alcoholism, while studies by Bolos et al. (1990), Gelernter et al. (1991), Schwab et al. (1991), Goldman et al. (1992), Turner et al. (1992), Cook et al. (1992), and Amadeo et al. (1993) failed to find an association. Parsian et al. (1991) presented data supporting an association limited to medically complicated alcoholism. Combining results in a meta-analysis could provide enough information to identify small genetic effects when large environmental influences are involved (Uhl et al., 1992).
When 13 published studies were combined and compared, highly significant differences were found between the total alcoholic and the total control population in the phenotype distribution, A1 allele frequency and A1 allele prevalence. Significant differences were also found between the present study's controls and others' controls, while no significant differences were found between the present study's controls and others' alcoholics. No differences were found between the present study's alcoholics and either the total controls or the total alcoholics.
The most likely explanation for an association is that the TaqI A polymorphism is in linkage disequilibrium with an upstream regulatory element, or a 3' flanking element, or another gene which confer susceptibility to alcoholism (Turner et al., 1992; Gejman et al., 1994). The A1 allele would have to be associated in linkage disequilibrium with a functional difference at DRD2 or at a neighbouring locus. Several linkage disequilibrium studies have found strong linkage disequilibrium between the TaqI A1 allele and the TaqI B allele and the SSCP 1 allele (Goldman et al., 1993; Hauge et al., 1991; Bolos et al., 1990; O'Hara et al., 1993).
The possibility remains that DRD2 modifies the expression of alcoholism without actually causing it. This would mean that other genes were the primary cause of alcoholism, and A1 was a modifying gene that causes the progression of the disease in individuals genetically predisposed to alcoholism (Gelernter et al., 1993a; Conneally, 1991; Comings et al., 1991). However if this were true, there would expected to be evidence of linkage in families, and complications in offspring with the A1 allele would be more likely than in offspring in the same family without A1 (Parsian et al., 1991b).
If the TaqI A polymorphism were a major gene underlying alcoholism, it would be expected to show linkage in family studies (Cook and Gurling, 1994). However the two linkage studies performed by Bolos et al. (1990) and Parsian et al. (1991b) have both been negative, even when the analysis was on severe alcoholism only.
The finding of associations in the general population without linkage in pedigrees may be explained by several hypotheses (Parsian et al., 1991b). One is population stratification or admixture. If one or more subpopulations happens to have a high frequency of the A1 allele and also a high incidence of alcoholism, the result may be a spurious association, but no linkage. That is, if the subjects selected for alcoholism are made up of up a subpopulation with a higher frequency of A1 than those selected to be control subjects, then association may be found that is not a true association. Barr and Kidd (1993) suggested this was a strong possibility for the supposed association, although Cloninger (1991) noted that replication in independent studies makes population stratification and chance sampling unlikely. Another possible explanation is epistasis, in which the gene has multiple effects; or interaction among multiple genes (Parsian et al., 1991b). The A1 allele may not cause alcoholism itself, but interacts with genes and other causal factors to modify expression of the disease (Cloninger, 1991). Other possibilities include clustered sampling of related individuals, which may be ruled out if the subjects are unrelated; and chance findings in small samples, which requires comparison with other samples (Parsian et al., 1991b).
The subjects used in the present study were not classified as type I or type II alcoholics. One subset was classified as alcohol dependent (DSM-IV criteria, code 303.90). The other two subsets were classified by alcohol consumption data. Alcoholism is a heterogeneous set of disorders and ideally subjects should be chosen that are clinically matched in the type of alcoholism they have. This will be difficult, since alcohol abuse is not a single entity, with differences in clinical symptoms such as age at onset, severity of abuse, and alcoholism is associated with other forms of psychopathology such as antisocial personality and depression (Gilligan and Cloninger, 1989).
While the current study had 66 alcoholic samples, there were only 22 that were diagnosed according to DSM-IV criteria. This sample size is quite small, and future studies should ideally have higher numbers.
It will also be necessary to have shared definitions of alcoholism so that comparisons between studies will be possible using the same criteria for diagnosis.
Further association studies must consider the ethnic origin of their subjects to avoid spurious associations arising from incorporation of subjects from different populations which have different allele frequencies, and different levels of linkage disequilibrium between genes. Future studies should use subjects that are ethnically not just racially matched.
Other markers within the DRD2 gene should be studied. The use of a combination of polymorphisms at the DRD2 locus may give more statistical power (Barr and Kidd, 1993). Including the TaqIA polymorphism, there have been 8 informative polymorphisms described (Figure 1). They include TaqIB (Hauge et al., 1991), TaqIC (Parsian et al., 1991a), a CA microsatellite in the second intron (Hauge et al., 1991), two polymorphisms between in intron 6 detectable by PCR with allele-specific oligonucleotides (Sarkar et al., 1991), a SSCP in the 8th exon (Bolos et al., 1990) and a FokI RFLP in the exon 4/ intron 4 area (Lu et al., 1992). Because recombination in such a small region is so uncommon, a combination of alleles spread across the gene will be transmitted through families as a haplotype. The use of several polymorphisms greatly increases the power of statistical analysis. If there is an association, a haplotype should show a clearer association with the disease (Barr and Kidd, 1993). The current study attempted to examine the CA microsatellite marker in addition to the TaqI marker to study linkage disequilibrium within the gene. However due to time constraints, no useful results were obtained.
Family, twin and adoption studies have shown that alcoholism has a genetic component. In 1990, Blum et al. proposed an association between the A1 allele of the DRD2 gene and alcoholism. The DRD2 gene is the first candidate gene that has shown promise of an association with alcoholism (Gordis et al., 1990). However numerous replication attempts have had conflicting results. Indeed, the current study failed to find an association.
The A1 allele is not in the coding region of DRD2. For the polymorphism to have any effect, it must either be in linkage disequilibrium with a mutation in DRD2 or another nearby, unknown gene, that confers susceptibility to alcoholism, or it may be involved in a downstream regulatory role. The actual mechanism is not clear (Turner et al., 1992).
If the A1 gene modifies the course of alcoholism, it should still be expected to show linkage within families, but so far, this has not been shown.
A role for a variant DRD2 gene should really only be considered if a variant allele shows cosegregation with medical complications of alcoholism in informative alcoholic pedigrees. It would also be necessary to find either a variant of the encoded receptor protein, or else of a variant tissue distribution of that protein (Karp 1992).
Further work on the genetics of alcoholism will provide more understanding of the physiology of alcoholism and other neurophysiologic and neuropsychiatric abnormalities. This will lead to new approaches for identification of how dopamine receptor changes lead to increased substance abuse susceptibility and new approaches in the treatment, management and even prevention of alcoholism by identifying subtypes where genetic susceptibility is present. It may be possible to find individuals who are more vulnerable to alcoholism because they have a variant DRD2 gene.
Even if there is an association between A1 and alcoholism, the association will never be 100% and so therefore cannot be used as a diagnostic test. Environmental as well as genetic factors contribute to alcoholism, and other genes are critical for the predisposition to and subsequent expression of alcohol-seeking behaviour (Blum et al., 1990). Clearly, much more work is still to be done before it can be demonstrated what role genetic factors have in alcoholism.