The familial nature of addiction has long been recognized. Indeed, the Rev Robert Burton (1577–1640), in a subsection of his literary masterpiece The Anatomy of Melancholy entitled ‘Parents a cause by propagation’ quotes Plutarch the Greek historian, biographer and essayist (∼ 46–113) as stating: ‘Ebrii gignunt ebrios’ (‘one drunkard begets another’) [1,2]. While this familial nature of addictions has been accepted readily, the proportion attributable to genetics has remained more controversial. Indeed, at times the genetic contribution has been overstated by the media with such newspaper headlines as ‘Gene and tonic: science proves that alcoholics can’t help it' and ‘Born to the bottle’. This positing of an addiction engendering gene is very different to the current and complex view, held by many in the field, that multiple genes interact with environmental factors at different stages during the development of addiction. Such media over-simplifications, combined with the perceived lack of any robust gene findings melding with the residual belief that that addiction is a moral condition, has to a certain extent discredited the molecular genetic approach. The aim of this review is to explore the evidence that genes are involved in addiction and, having reviewed the evidence, to then trace the attempts to identify specific genes, thereby to validate this approach as a means of identifying contributing, but not causal, factors. In so doing I have attempted to communicate these concepts to the general readership rather than address a specialist genetic audience; as such I apologize to the latter for this focus and the former if I have failed to achieve this aim. Furthermore, I hope I have adopted a ‘balanced’ approach, based on the information available, and not been lured into expressing mere opinion.
Aim To assess the progress and impact of genetic studies in the addictions arena and to present this information in a form accessible to the general readership of Addiction.
Methods Review of the evidence that genes are involved in addiction, approaches to their identification, current findings and the potential implications.
Results Family, twin and adoption studies provide strong evidence that addiction runs in families and that this is determined in part by genetic factors. Two main molecular genetic approaches, namely linkage and association, have been adopted to identify the specific genes involved. Both methods are fraught with problems. Linkage is limited by issues of sensitivity, and association by false positives. Perhaps the strongest finding in psychiatric genetics to date is the impressive effect that a single genetic variant, in the aldehyde dehydrogenase 2 gene, has on drinking behaviour and reducing the risk of developing alcohol dependence. Other findings are currently less robust; however, the implications of elucidating the genetic underpinning of addiction will be profound.
Conclusions Addiction genetics is a developing science that has yet to prove its worth in the clinical setting
ADDICTION RUNS IN FAMILIES; HOW MUCH OF THIS IS DUE TO GENES?
Observations that addiction runs in families, such as those reported by Burton above, have been supported by formal family studies that indicate a clustering of addiction in families, with higher rates in parents of affected individuals and their relatives [3,4]. Indeed, one's risk of developing alcohol-related problems increases with both the proximity and number of affected relatives . However, this clustering in families or ‘familiality’, while being consistent with, is not proof of, a genetic mode of propagation. That genes may be responsible in part for this familiality has been given strong support by adoption studies. These natural experiments transport a high-risk individual from the assumed predisposing environmental influence of the affected biological parent to a novel and potentially protective adoptive environment . For example, the adoption study reported by Guze and colleagues indicated a fourfold increase in the risk of developing alcoholism conferred by possessing a biological parent thus afflicted over those without such a parent . The rate of alcohol dependence in adoptees is approximately the same as that in their siblings raised by the biological parent and therefore adoption away from that parent does not appear to confer any protective effect [7,8].
These adoption studies provide strong evidence for the role of genes in addiction and further natural experiments, in the form of twin studies, provide the opportunity to quantify their relative importance. Twin studies compare the rates of a co-occurrence termed ‘concordance’ in monozygotic, or identical, twins and dizygotic, or non-identical, twins. Monozygotic twins are derived from a single fertilized egg, the term originating from the Greek for single (µóνος) and yoke (ζνγóς); zygote being the term for a fertilized egg. As such these twins are genetic clones of each other. Dizygotic twins are derived from two (Greek δíς meaning twice) fertilized eggs and as such have on average 50% of their DNA, that is deoxyribonucleic acid the hereditary material from which genes are made, from a common parental origin as would be expected to be the case for siblings. If a condition is genetic, it would be anticipated that more co-twins of individuals with addiction problems would be affected in monozygotic twins when compared with dizygotic twins. Indeed, this has been reported in most of the twin studies that have examined addictive behaviours [9,10]. In addition, by comparing the differences in the concordance rates between the monozygotic and dizygotic twins it is possible to provide an approximate estimate of the relative strength of the genetic contribution or ‘heritability’. Such estimates vary considerably, but for alcohol dependence typical estimates are 50% for males and 25% for females [9,10]. Similarly, smoking initiation and persistence are also under a genetic influence that amounts to at least 50% . The Vietnam Era Twin (VET) registry provided estimates of heritability of between 25% and 44% for abuse and/or dependence for various substances including cannabis, opiates, psychedelics, phencyclidine, stimulants and sedatives [12,13], while the Virginia twin studies provided estimates approximately double that of the VET study, with figures for women close to men [14,15]. These discrepancies may be due in part to differences between the two populations; for example, drug exposure .
IDENTIFYING THE GENES INVOLVED IN ADDICTION
Having established a role for genes in addiction, what approaches have been employed to identify the specific genes involved and what progress has been made? Two main molecular genetic approaches have been adopted to chase those genes that may be responsible for the genetic contribution to addiction: linkage and association . These approaches are not mutually exclusive, as the genetic regions identified by linkage may be refined subsequently by association and the causative gene difference(s) identified.
The first uses a pedigree-based approach and is called linkage. Generally for a linkage study DNA samples are needed from individuals in families that are multiply affected by addiction. Using approximately 300 markers, that is segments of DNA that may differ between individuals usually by size or sequence, the whole genetic complement can be screened for genes involved in the condition under study. As such, linkage can identify novel genes that are involved in conditions for which there has been no a priori reason to suspect any contribution. While this approach can be stated in such simple terms the actual analysis is much more complex, employing computer programs designed to identify a statistically significant coinheritance between a marker and the condition across the families. If linkage is detected this implies that there is a gene, or genes, implicated in addiction in the proximity of the marker and efforts can then be made to identify which of the many possible genes are important. Essentially linkage is systematic, covering the human DNA complement or ‘genome’, and has proved to be an excellent approach to identify genes that have large contributions to a condition, such as Huntingon's disease . However, it is less useful where genetic contributions are relatively small and masked by complex gene–environment interactions.
Linkage findings in addiction
Several major linkage studies of alcohol dependence have been undertaken in the United States. The multi-centre study entitled the ‘Collaborative Study on the Genetics of Alcoholism’ (COGA) used samples from 987 individuals from 105 multi-generational families. The study found suggestive evidence for genes implicated in alcohol dependence on chromosomes 1 and 7, with modest evidence for chromosome 2 and a protective locus on chromosome 4 in the region of the alcohol metabolizing genes of the alcohol dehydrogenase (ADH) family cluster . The standard chromosome complement in humans is 46, consisting of 22 pairs of autosomes and two sex chromosomes. These are large DNA molecules, in combination with proteins, which enable the hereditary material to be packaged and managed within the cells. However, these original findings have not been supported further by results from a replication sample .
A second study from the United States, while employing a smaller sample, enhanced the chances of identifying linkage by using a more homogeneous population, namely an American Indian population . This sample consisted of 152 subjects from 32 interrelated nuclear families. Evidence of linkage was identified on chromosome 4, near a cluster of gamma-aminobutyric acid (GABA) subunit genes and on chromosome 11 near the dopamine 4 receptor and tyrosine hydroxylase genes.
Another linkage approach employs sibling pairs, and using a doubly affected sibling method families were recruited from the Pittsburgh area and the subsequent linkage analysis gave strong support for genetic loci implicated in alcoholism on chromosomes 1, 2, 6, 7, 10, 12, 14, 16 and 17 . A more recent study used a sibling-pair method in an Irish population and dissected alcohol dependence into alcohol-related traits. There was evidence for two regions on chromosome 9 being implicated in age of onset. Suggestive regions involved in initial response to alcohol were reported on chromosomes 1 and 11, while tolerance was linked to regions on chromosomes 1, 6 and 22. Maximum drinking was linked to regions on chromosomes 12 and 18 and withdrawal symptoms with chromosome 2 [22–24]. Similarly, a further study in American Mission Indians did not identify any linkage with alcohol dependence but alcohol use severity was linked to regions on chromosome 4 and 12, while withdrawal symptoms were linked to regions on chromosomes 6, 15 and 16 .
Several linkage studies of smoking behaviour have been reported with inconsistent results, implicating loci on multiple chromosomes .
Are linkage studies the best approach in addiction?
Thus far, linkage studies have not delivered on their early promise to identify the regions involved in addiction. However, as noted above, the linkage approach is ideally suited to those conditions that have genes of major effect, and the genes contributing to addiction may fall below the threshold for detection by this technique.
In many ways the second molecular genetic approach, using association studies, is more suited to identifying the genes that contribute to addiction, as they can pick up genes of relatively small effect, accounting for less than 2% of the variance . Association studies rely on a genetic change, occurring many generations before, that alters an individual predisposition to a condition. Gradually over the generations, through recombinations of the genetic material that occur during the process of meiosis that produces eggs and sperm, the genetic variants that are further away from the genetic difference, conferring the vulnerability to addiction, become dissociated from it. Those markers that are very close to the genetic variation are very rarely dissociated from it and therefore the versions of these markers remain the same. Thus these, along with the vulnerability causing genetic variation, will be inherited as a block and as a result may occur at a different frequency in those with the condition when compared with the unaffected. These differences can be tested for significance simply by use of a χ2 test.
Association samples consist of seemingly unrelated individuals, although the approach is based on individuals being related many generations previously. The distribution of the variants of different genetic markers is compared between those with and without the condition of interest. If a statistically significant difference is identified, and this finding is replicated robustly, this suggests that the marker concerned is very close to the causal gene change increasing the vulnerability to the condition. Association studies are easier and cheaper to undertake than linkage approaches, although obtaining a suitable control sample often proves difficult. For example, are the best controls for smokers those who have never smoked (‘supernormals’) or those who are able to smoke but do not progress to dependence (‘chippers’)?
A gene variant that profoundly affects drinking behaviour and risk of dependence
Perhaps one of the most robust findings, not only in addiction but psychiatric genetics, has been the association between an inactive genetic variant at the aldehyde dehydrogenase 2 (ALDH2) gene and protection against the development of alcohol dependence. However, in this case the biochemical lesion was appreciated long before the genetic mechanism was established. Thus in 1972 Wolff raised the possibility that the way the body metabolizes, or deals with, alcohol might be an important factor in the development of alcoholism . At that time, the lower rates of alcoholism in the oriental ethnic groups he was studying (Japanese, Taiwanese and Koreans) were attributed to socio-environmental factors. His study reported differences in the reaction to alcohol, namely a high frequency of marked facial flushing, which he suggested, correctly as it turns out, could be related to the propensity to develop alcoholism.
When alcohol is consumed it is absorbed rapidly from the stomach and small intestine and metabolized primarily in the liver. The first step in this process is catalysed, or enhanced, by alcohol dehydrogenase (ADH) . In fact, ADH represents a group of seven enzymes that are encoded by a cluster of genes on the long arm of chromosome 4 (already mentioned in the COGA study above) . At least three of these enzymes are implicated in alcohol metabolism at the physiological levels of alcohol achieved during drinking . These produce acetaldehyde, an unpleasant toxin, which is normally broken down rapidly by ALDH2. A large proportion of individuals of oriental ethnic origin possess an inactive variant of this latter enzyme and the result is an accumulation of acetaldehyde: this causes dysphoria, flushing and palpitations [27,28]. This is also the site of action of disulfiram, one of the few medications used to help maintain abstinence in alcohol-dependent subjects . Disufiram acts as a suicidal, or irreversible, inhibitor of ALDH2, thus when alcohol is consumed the result is an aversive and potentially fatal reaction . Not surprisingly, therefore, the low activity variant protects those carrying it from developing alcohol problems and dependence, as it is similar to taking disulfiram all the time [27,31–33].
The genetic cause of the low activity variant has been elucidated and it is the result of a single base change: a single typographical error, as it were, in the gene [27,28]. This change alters one of the amino acids that are the building blocks that make up the protein, and this change reduces dramatically the activity of the enzyme. Hence, when alcohol is consumed it is metabolized to acetaldehyde and this toxic compound accumulates due to its disrupted metabolism, with the resulting adverse reaction and flushing. This single base pair change in the DNA of the gene has a profound effect on reducing drinking behaviour and ultimately the risk of developing alcohol dependence. However, despite possessing the low activity variant it is still possible to develop alcohol dependence, and recently there was great excitement in the field when an individual attracted a diagnosis of alcohol dependence using DSM-III-R criteria, despite being homozygous for the low activity variant; that is, both gene copies were inactive. However, the daily alcohol consumption reported was a mere 50 g of ethanol, equivalent to approximately six UK standard drinks or units, which is relatively small for typical individuals dependent on alcohol . Furthermore, the inactive variant is not present in individuals of western European extraction and therefore does not contribute to any protective effect in such populations.
As indicated above, the protection against alcohol dependence, conferred by this inactive genetic variant of ALDH2, is one of the most robust findings in the genetics of complex behaviours.
Other association findings
There are many other associations that have been reported both in alcohol and other drugs. For example, lower activity variants of ADH genes have been associated with an increased risk of alcohol dependence, again primarily in oriental populations, and this would again fit with the proposed role of acetaldehyde mentioned above . Apart from these findings the other reported associations are less robust, often with positive and negative attempts at replication. Because markers used in association have to be very close to the functional genetic variant implicated in the increased vulnerability, as explained above, very many would have to be examined to perform a comprehensive genome-wide scan. Consequently, a candidate gene approach has been adopted in which genetic markers are examined in systems that appear to be related to addiction. As a consequence, these candidate gene association studies have implicated the ‘usual suspects’ because these are precisely the gene systems studied. Thus, such studies of alcohol dependence have implicated genes in the dopamine, GABA and serotonin systems [35–37]. Similarly in opiate addiction, genes of the dopamine, noradrenaline, opiate, cannabinoid and serotonin systems have been implicated [9,38,39] while in smoking, possible associated genes have been described in the dopamine, serotonin, nicotine and nicotine metabolism systems [40–44]. It would be very tempting to provide a huge table of the multiple reported associations, with a list of the replication attempts, but this would be of limited value for the purposes of this review. Indeed, one of the major difficulties encountered with association studies is the potential for false positives.
Association studies are hampered by false positives
While this risk may be diminished by careful sample selection and characterization, related particularly to the control population, it is a risk inherent in the approach. For example, Buckland has attempted to estimate the probability that a reported association has occurred by chance in alcoholism . He recognizes the huge and inaccurate assumptions he makes, including assuming all genes are equally likely to be associated with alcoholism, that 20 genes are associated genuinely, there are 80 000 genes (which we now know is an overestimate as following the sequencing of the human genome the figure is now around 25 000) and the role of each gene can be tested using a genetic marker. From this model he estimates that ‘we would expect to find a true positive result in one in every 4000 experiments or 0.025% of the time’. In addition he calculates that ‘any result with P-value of 0.05 has a 99.5% likelihood of having occurred by chance’.
NEW DEVELOPMENTS IN MOLECULAR GENETIC APPROACHES
Addiction is a complex trait and perhaps best dissected down and studied as individual components of behaviour or endophenotypes (defined by Gottesman & Todd as ‘measurable components unseen by the unaided eye along the pathway between disease and distal genotype’) . Such endophenotypes include responses to alcohol, electrophysiological measures and personality traits such as impulsive sensation seeking. Indeed, promising results are being reported using this approach [47–50].
More recently molecular genetic techniques have been developed, such that it is now possible to propose a systematic association study of the whole genome using DNA chips or arrays. These use short fragments of DNA, spotted onto a surface such as glass or plastic, which identify different variants of a marker by their ability to pair or match up with an individual's DNA. As such these chips, that are a few centimetres square, can examine up to a million markers in one reaction, thus permitting genome-wide scans [51–54]. This approach can be made efficient and cost-effective by techniques that employ the pooling of DNA [54,55].
This is a systematic approach that can identify novel genes not predicted previously to be involved in addiction . However, the risk of false positives remains, compounded by the vast numbers of markers typed. The prospect of developing the experimental design and data management that will enable the ‘sifting of the wheat from the chaff’ is daunting. In addition to the ubiquitous nature of false positives, true positives may be obscured further by gene–environment and gene–gene interactions. It is therefore vital that all associations reported should be treated with extreme caution until replicated robustly.
We have highlighted previously the complex gene–environment interaction. Indeed, such interactions may obscure and hamper the identification of genes predisposing to addiction. Recently longitudinal studies have been used to explore these directly and have suggested interactions between childhood adversity, a monoamine oxidase variant and conduct disorder and adverse life events, a serotonin transporter variant and depression [56,57]. Similar approaches in addiction are in their infancy but there is emerging evidence of a similar interaction between maltreatment, the serotonin transporter and earlier alcohol use . In addition, a gene–environment correlation and interaction has been reported with GABRA2, marital status and alcohol dependence . Furthermore, it is important to remember that one gene may influence addiction in several different ways. For example, a genetic variant at the dopamine 4 receptor could influence addictive behaviour more directly through reward mechanisms but also through the personality trait of impulsive sensation seeking and possibly cognitive bias [49,50,60].
More recently, by combining linkage results across substances and integrating these with other approaches including genome-wide scans, some authors have reported a convergence in regions that may be implicated in the vulnerability to addiction to several different classes of substance, across races and animal models [61,62].
Limitations of current approaches
Both the molecular genetic approaches of linkage and association are limited. The former may not be sufficiently sensitive to identify the genes involved in addiction, while the latter may be dogged by false positives reported so enthusiastically that they may lead to the approach being discredited inappropriately. Currently neither has delivered on their promise of identifying the genes.
THE BENEFITS AND ETHICAL ISSUES OF FINDING THE GENES FOR ADDICTION
Identifying the genes involved in the predisposition to addiction could have a profound impact on the way in which this condition is viewed. It would provide support for the medical model, thereby endorsing the patient status of those affected and eliciting empathy from society and allocation of resources. Could such genetic findings be interpreted as absolving individuals of any responsibility for their addictive behaviour? Could they sue their parents for the genetic endowment that has left them so disabled? Could it give false security to those without risk-increasing genes regarding their drinking behaviour? Furthermore, would it encourage a nihilistic approach towards those at increased risk, such that they are considered by others and/or themselves, to be without hope of addressing their addiction? In addition, identifying the genes, their proteins and the respective functions at each stage of the process of addiction development will allow the dissection of the underpinning biological changes. This may allow the stratification of the apparent unitary diagnosis of addiction into subtypes of varying prognosis. Such biological knowledge will inform the development of novel interventions: pharmacological, psychological, spiritual and their combination. Furthermore, these may be targeted and tailored to the individual, possibly by using a genetic test.
Finally, screening and risk alteration using genetic testing may also be feasible. Such possibilities also raise concern regarding their misapplication. Indeed, in the not-so-distant past the commitment to eugenics has been used to justify appalling atrocities in the name of genetic improvement, such that it has been described as murderous science: a genetic utopia to some may represent a discriminative hell to others . Certainly these are both scenarios played out in the film Gattaca, in which a society stratified on the basis of genetic endowment is challenged . In truth, the increased rate of spontaneous abortion in first-cousin marriages suggests that each of us are carrying ‘defective genes’ with an average of 1.4 deleterious segments of DNA, termed lethal equivalents, that if they were not balanced by ‘good’ DNA would be incompatible with life . However, given the large number of genes likely to be involved in addiction and the relatively small effect size of each, it is unlikely that screening and risk alteration are serious potential uses. Perhaps the biggest fear is that the effect size of each gene will be so small as to render any clinical use insignificant. Should this be the case, the cost–benefit of this research would be extremely high. Indeed, it has been argued that effective public health measures to reduce addiction related problems are already known, for example reducing the availability of alcohol through price, and that the money would be better spent implementing them, although this may be deemed politically and personally unpopular.
There is considerable evidence from family, twin and adoption studies for the operation of genetic factors in the vulnerability to addiction. Attempts to identify the specific genes involved have employed primarily the molecular genetic techniques of linkage and association. The former approach is insensitive to genes of less than major effect, which are likely to be acting in addiction, while the latter is hampered by a high false positive rate and multiple testing. As such, these approaches have not yet delivered on their early promise. However, the genetic explanation of the previously recognized oriental flush reaction provides compelling evidence for the powerful effect that a single base pair change in a gene can have on drinking behaviour and risk of dependence. Currently, all other findings are less robust and are awaiting confirmation of their role, if any. New methods are being adopted to further chase the genes that predispose to addiction, including genome-wide association studies, DNA pooling, the use of endophenotypes and the exploration of gene–environment interactions. The potential clinical benefits of this research are great, and in contrast the risks small.