A review of pharmacogenetic studies of substance-related disorders

Abstract

Background

Substance-related disorders (SRDs) are a major cause of morbidity and mortality worldwide. Family, twin, and adoption studies have demonstrated the substantial heritability of SRDs. To determine the impact of genetic variation on risk for SRD and the response to treatment, researchers have conducted a number of secondary data analyses and quasi-experimental studies that target one or more candidate gene variants.

Methods

This review examines studies in which candidate polymorphisms were examined as mediator variables to identify pharmacogenetic effects on subjective responses to drug administration or cues or outcomes of medication trials for SRDs. Efforts to use a meta-analytic approach to quantify these effects are premature because the number of available studies using similar methods and outcomes is limited, so the present review is qualitative.

Results

Findings from these studies provide preliminary evidence of clinically relevant pharmacogenetic effects. However, independent replication of these findings has been sparse.

Conclusions

Although this growing body of literature has produced conflicting results, improved statistical controls may help to clarify the findings. Additionally, the use of empirically derived sub-phenotypes (i.e., which serve to differentiate distinct groups of affected individuals) may also help to identify genetic mediators of pharmacologic response in relation to SRDs. The identification of genetic mediators can inform clinical care both by identifying risk factors for SRDs and predicting adverse events and therapeutic outcomes associated with specific pharmacotherapies.

1. BACKGROUND

1.1. Drug Use and Addiction

Addiction is a chronic disease characterized by compulsive drug seeking and use [National Institute on Drug Abuse (NIDA), 2014]. Substance-related disorders (SRDs) cause and contribute to the deaths of millions of people each year by worsening comorbid psychiatric symptoms, such as depression, and medical conditions, such as cirrhosis of the liver, while also aiding in the spread of infectious diseases such as HIV, hepatitis B and hepatitis C. Substance-related disorders are also linked to crime and disability [United Nations Office on Drugs and Crime (UNODC), 2014].

Regular tobacco use contributes to many of the world's leading causes of death, including heart disease, stroke, and cancer (WHO, 2011). Cannabis is the world’s most widely used illicit substance, with upwards of 177 million regular users worldwide (UNODC, 2014). Although the long-term effects of chronic cannabis use are debated, adverse effects on cognition and mental health have been demonstrated (Fergusson et al., 2008; Hall and Degenhardt, 2009). While not commonly considered to be “addictive,” the world-wide use of the mild psychostimulant caffeine attests to its ability to serve as a reinforcer (Grigg, 2002).

More robust psychostimulants, such as cocaine and amphetamine-type drugs, are the second most widely used illegal drugs, with ~75 million estimated global users (UNODC, 2014). A number of serious medical complications are associated with cocaine use, including disturbances in heart rhythm, myocardial infarction, and neurological impairments (NIDA, 2009). Amphetamines may have a variety of neurotoxic, cardiotoxic and adverse neuropsychological effects as well (Scott et al., 2007; Shrem and Halkitis, 2008; Yu et al., 2003).

The abuse liability of naturally occurring opiates (e.g., morphine, codeine) and synthetic opioids (e.g., heroin, oxycodone, buprenorphine) is well known (Comer et al., 2008; Moratti et al., 2010; see also Meyer et al., 2014 and Trigo et al., 2010 for reviews). An estimated 9.2 million people worldwide are regular users of heroin (UNODC, 2014). The abuse of other opioids, including analgesics like oxycodone, is also widespread (Cicero, 2005; Darke et al., 1996; Gilson et al., 2004; Kintz et al., 2001; Substance Abuse and Mental Health Services Administration (SAMSHA), 2013). Opioids can significantly depress respiration, making overdose the most common cause of death among heroin users (Degenhardt et al., 2011; White and Irvine, 1999).

Although SRDs are a global public health concern, we have just begun to understand how genetics affect the initiation, course, and recovery from these disorders. A clearer understanding of the genetic contributions to these phenomena would inform the prevention, identification, and treatment of SRDs. In the qualitative review that follows, we focus on genetic aspects of the more common drugs of abuse that contribute substantially to morbidity and mortality around the world.

1.2. Genetic Involvement in Substance-related Disorders

Twin and adoption studies provided the first strong evidence for a genetic contribution to the susceptibility to develop SRDs (See Goldman et al., 2005 for a review). These investigations revealed that the heritability, the proportion of observable differences in a trait between individuals that is due to additive genetic effects, of substance abuse is quite significant. Depending on the specific drug, it has been estimated that genetic factors contribute 40–80% of the vulnerability to addiction (Agrawal et al., 2012; Crabe et al., 2002; Kendler et al., 2000; Tsuang et al., 1996, 2001; Uhl et al., 1999).

Genetic studies have examined a variety of individual genes that could contribute to the development and maintenance of SRDs. Genetic linkage studies are family-based studies that aim to establish a link between a region of a specific chromosome and the expression of a behavior or trait of interest. Linkage analysis uses panels of markers to identify the chromosomal region that harbors a gene of interest. If a marker consistently segregates in families with the trait under investigation (in this case, drug abuse), it is likely that the gene of interest is located in the chromosomal region identified by that marker. Similarly, association analyses test for a correlation between disease status and genetic variation to identify candidate variants that either contribute to a specific disease, or are in linkage disequilibrium with a causative variant.

Genome-wide association (GWA) and whole exome/genome sequencing techniques have been extremely successful in identifying genetic contributors to a number of complex human traits and diseases (Hindorff et al., 2009). These “agnostic” studies (i.e., those without a prior hypothesis as to function or location) have identified multiple genes and polymorphisms for more targeted investigations into how they mediate vulnerability to abuse. Alternatively, candidate-gene association studies take a hypothesis-driven approach to identifying potential mediator genes. Positional candidate genes are identified through linkage analysis and fine mapping based upon their approximate chromosomal location. More commonly however, functional candidate genes are identified based on their known (or presumed), relevant biological function. Linkage, association, and candidate gene studies have identified a number of specific chromosomal regions, genes, and alleles for further investigation.

2. GOALS AND METHODOLOGY OF THE REVIEW

Scientists have begun to perform studies where the presence or absence of target gene variants is used as an independent variable in laboratory studies measuring the subjective effects produced by a drug or clinical trials measuring treatment outcomes. The present review focuses on this method, known as pharmacogenetics, of identifying how genetic variation contributes to the susceptibility and maintenance of SRD’s. In this review, preclinical, linkage and association studies are often cited to describe how a particular genetic variant could alter gene expression or neurobiological function. However, the primary aim of this paper is to provide an overview of clinical pharmacogenetic studies investigating genetic mediators of the drug’s subjective effects or pharmacotherapy treatment outcome.

Using PubMed and PsychINFO, we searched for English-language articles published between 1970 and 2013. We included various combinations of the following search terms: genetics, polymorphism, SNP, pharmacogenetics, caffeine, opioids, heroin, nicotine, cocaine, and amphetamine. Using this method, we identified over 500 publications. After removing duplicates, we reviewed the titles and/or abstracts to ensure relevance. Data from approximately 150 articles are included here. Due to the extensive literature concerning the pharmacogenetics of alcohol, this drug could not be included in the current review and remain within the length requirements of the journal. Over 90 relevant publications related to the pharmacogenetics of alcohol and treatment of alcohol use disorder were identified. As such, we felt this body of work necessitated a review of its own (see Jones et al., in press).

3. GENETIC AND DOPAMINERGIC INTERACTIONS

Although the various classes of commonly abused drugs act upon a multitude of neurobiological systems, prevailing theory implicates the effects of drugs on mesolimbic dopamine (DA) transmission as the basis for their rewarding and euphoric effects. All major drugs of abuse have been shown to enhance dopaminergic transmission in the “reward centers” of the brain, including the ventral tegmental area, nucleus accumbens, and prefrontal cortex (Di Chiara and Imperato, 1988; Sulzer, 2011; Tanda et al., 1997). As such, we begin this review with a focus on genetic and dopaminergic interactions. Because dopamine plays a key role in reward-related behavior, it has been the most extensively studied neurotransmitter systems. Variation in a number of genes involving dopaminergic signaling (synthesis, transmission, and metabolism) has been studied with respect to their contribution to SRDs.

The five dopamine receptor subtypes are typically divided into two groups, D1-like (D1 and D5) and D2-like (D2, D3 and D4; Girault et al., 2004). Collectively, the D2 dopaminergic pathway appears to be a key neural substrate for drug-related reward (see Di Chiara et al., 2004 for a review). Accordingly, several studies have focused on the connection between the D2 receptor gene (DRD2, which is on chromosome 11) and various drugs of abuse (Li et al., 2011). For example, many studies have shown an association between drug abuse and the presence of a polymorphism in the ankyrin repeat and kinase domain containing 1 (ANKK1) gene adjacent to DRD2 (see Li et al., 2011 for a review; Neville et al., 2004). This Taq1A polymorphism (rs1800497) is a C>T substitution located in a non-coding region of ANKK1. It was originally believed to map to DRD2, though it is 10 kb downstream of the gene. Rs1800497 has been associated with dependence on cocaine and amphetamines (Noble et al., 1993; Persico et al., 1996), nicotine (Noble et al., 1994; Comings et al., 1996; Spitz et al., 1998), cannabis (Nacak et al., 2012) and various drug combinations (Comings et al., 1994; O’Hara et al., 1993). However, the findings have not been consistent; other reports show no association of the Taq1A polymorphism with drug abuse (Heinz et al., 1996; Miginini et al., 2012).

The Taq1A polymorphism has been shown to be associated with altered dopamine transmission. Several studies have shown that the polymorphism is associated with an up to 30% reduction in D2 receptor density, decreased dopamine receptor sensitivity, and elevated dopamine transporter (DAT) density (Jonsson et al., 1999; Laine et al., 2001; Pohjalainen et al., 1998; Schellekens et al., 2012). Although less studied than the ANKK1 single nucleotide polymorphism (SNP), a non-synonymous DRD2 polymorphism (C957T polymorphism; rs6277) has been shown to be functional. In healthy subjects, the rs6277*C allele has been associated with lower striatal D2 binding than the T-allele (Hirvonen et al., 2004). In contrast, some in vitro studies have shown the C allele to be associated with greater stability of DRD2 mRNA and protein (Duan et al., 2003). Two other functionally relevant DRD2 intronic SNPs (rs1076560, rs2283265) have been demonstrated to influence gene splicing leading to increased expression of the short (D2S) vs. long (D2L) receptor isoform (Moyer et al., 2011; Zhang et al., 2007 B). These SNPs have also been reliably associated with cocaine, alcohol and heroin addiction (Clarke et al., 2014 B; Moyer et al., 2011).

Above are just a few examples of the functional influence of various polymorphisms affecting dopamine. Their potentially significant influence has led to several prospective pharmacogenetic studies. Because genetic mediators of dopamine function have been the most extensively investigated, this is where we begin this review. However, exciting pharmacogenetics research with other neurobiological systems (e.g., serotonergic and opioidergic) are introduced and discussed later.

4. PHARMACOGENETICS OF DRUG EFFECTS AND TREATMENT OUTCOME

4.1. Nicotine

To date, pharmacogenetic research on nicotine has focused mainly on the impact of genetic variation on the subjective response to the drug and on treatment outcome in smoking cessation trials. In a randomized, placebo-controlled laboratory study, 72 smokers were assigned to smoke normal and de-nicotinized cigarettes during positive and negative mood induction (Perkins et al., 2008a). In addition to rating the positive affective value of the cigarettes, participants were given the option of smoking the test cigarettes ad libitum (ad lib). During negative mood states, Taq1A T-allele carriers [i.e., heterozygotes (A1/A2) or homozygotes (A2/A2)] reported greater “liking” of nicotine cigarettes and took more puffs than C-allele homozygotes. DRD2 C957T*C allele homozygotes also smoked more nicotine cigarettes during negative mood states than during positive mood states. No such effect was seen among individuals with the other C957T genotypes. In another study by Perkins et al. (2008b), the subjective effects of intranasal nicotine (0, 5, or 10 µg/kg) were assessed among 101 non-smokers. Men who were 957T homozygotes reported greater dose-dependent subjective effects of nicotine, including “feel effect,” “anger,” and “reduced fatigue.” No significant associations were seen with the Taq1A polymorphism. Although both studies showed a possible mediating effect of the two polymorphisms on the subjective and reinforcing effects of nicotine, their results were contradictory.

Wang et al. (2008) examined the interaction of C957T genotype, mood, smoking behavior and nicotine withdrawal. They recruited 13 smokers to undergo two MRI scans separated by 1–3 weeks following either smoking as usual or after 12–14 hours of nicotine abstinence. During abstinence, DRD2 957T homozygotes exhibited less change in regional cerebral blood flow in brain areas previously associated with cigarette craving than C957 allele carriers. Gilbert et al. (2009) observed differences in electroencephalographic activity in Taq1A1 allele carriers during nicotine abstinence, and concluded that mood during nicotine withdrawal may be mediated by the Taq1A SNP.

Although variation in DRD2 may mediate nicotine withdrawal, one prospective outcome trial in women (N = 593) found no association between smoking cessation and variation in a number of genes involved in dopamine synthesis, receptor activation, dopamine transporter reuptake activity and metabolism, including SNPs in COMT (rs4680), DRD2 (rs1799732 and rs1800497), and DRD3 (rs6280), and variable number of tandem repeat polymorphisms (VNTR) in DRD4, SLC6A3, and TH (Ton et al., 2009). The COMT Val158Met SNP (rs4680) is a well-studied G to A substitution that produces a valine (Val) to methionine (Met) exchange in the gene that encodes the enzyme catechol-O-methyltransferase, which degrades catecholamines (dopamine, epinephrine and norepinephrine). The Met allele is associated with a 40% reduction in enzyme activity (Diamond, 2007; Lotta et al., 1995; Mattay et al., 2003).

Other investigators used a similar methodology to investigate the effects of variation in ANKK1 on pharmacological treatment outcome. Cinciripini and colleagues (2004) found that the Taq1A polymorphism mediated the response to the dopaminergic antidepressant venlafaxine in 134 participants in a smoking cessation trial. During the period following smoking cessation, smokers without the A1 allele (A2/A2) treated with venlafaxine showed a substantial reduction in negative affect. A1 allele carriers also quit significantly less often than A2 homozygotes. Interestingly, however, no genotype by treatment interaction was found on abstinence rates. Although, these data suggest that carriers of the Taq1 A1 allele may have more difficulty quitting smoking, the hypothesis was not confirmed in a subsequent clinical trial. In 116 smokers, Robinson and colleagues (2007) found that the Taq1 A1 allele was not associated with scores on the Wisconsin Smoking Withdrawal Scale (WSWS) or the Minnesota Withdrawal Symptom Checklist (MWSC), or with daily smoking abstinence rates during the 6-week post-quit observation period.

In contrast to results with the Taq1A SNP, Jonsson et al. (1999) found that the less studied Taq1B SNP (rs1079597) was associated with a lower D2 receptor density (Hauge et al., 1991). Robinson et al. (2007) found that smokers with the ANKK1 Taq1 B1 allele reported significantly more daily nicotine withdrawal symptoms than those homozygous for the Taq1 B2 allele whose withdrawal symptoms decreased significantly over the same period. The study also showed that smokers carrying the Taq1 B1 allele were significantly less likely to be abstinent on a given day than those homozygous for the B2 allele.

Pharmacogenetic treatment trials of ANKK1 have provided mixed results. A study of bupropion (a weak dopamine receptor antagonist) for smoking cessation in 577 heavy smokers showed that the medication was particularly effective among Taq1 A1 homozygotes (Breitling et al., 2010). Specifically, A1 homozygotes treated with bupropion had a 28% greater likelihood of smoking cessation, compared to a 12% greater likelihood of cessation among A2 allele carriers (i.e., A1/A2, or A2/A2 genotypes). In contrast, in a study of 30 smokers randomized to treatment with bupropion or placebo, the active medication reduced nicotine withdrawal symptoms (craving, irritability and anxiety) only among A2 homozygotes (David et al., 2003), a finding that was later replicated in a larger, two-site study (David et al., 2007a). In that study, 722 smokers were randomized to treatment with bupropion or placebo. In A2 homozygotes, those treated with bupropion were more than three times as likely as placebo-treated individuals to be abstinent at the end of treatment and at 6-month follow-up. There was no significant effect of bupropion on smoking cessation outcomes in Taq1 A1 carriers (A1/A1 or A1/A2). The Taq1A polymorphism was also shown to mediate smoking cessation outcomes using the medication rimonabant (an inverse agonist for the cannabinoid receptor CB1; Fong and Heymsfield, 2009). Wilcok and colleagues (2011) found that A2 homozygote smokers were significantly more successful in completely abstaining from smoking while on rimonabant (total N = 76).

Multiple studies have also evaluated the mediating effect of the Taq1A polymorphism on the effectiveness of nicotine replacement therapy (NRT). Yudkin et al. (2004) found that A1 allele carriers showed less therapeutic benefit from the nicotine patch than A2 allele homozygotes. Similarly, Johnstone et al. (2004) genotyped 755 heavy smokers for Taq1A alleles following completion of a randomized trial of the nicotine patch. They concluded that, at Week 1, heterozygous individuals for the ANKK1 SNP (CT/TT or A1/A2) showed a better response to NRT than to placebo. However, the pharmacogenetic effect on abstinence did not persist at 12 weeks. There was significantly greater abstinence at 12 weeks among individuals with at least one A1 allele and an A allele of rs77905 in the dopamine beta hydroxylase gene (DβH). Dopamine beta hydroxylase is an enzyme that catalyzes the synthesis of norepinephrine from dopamine. Individuals with the DβH A allele tend to have lower dopamine levels, and smokers with the allele tend to smoke more cigarettes (McKinney et al., 2000; Zabetian et al., 2001; though see Shiels et al., 2008). Munafo et al. (2009) later failed to replicate these findings in a larger cohort. Similar prospective trials have also failed to find a significant mediating effect of the Taq1A genotype on the efficacy of NRT in smoking cessation (Berlin et al., 2005; De Ruyck et al., 2010; Stapleton et al., 2011).

Although Lerman et al. (2006) failed to find a relationship between the Taq1A SNP and response to pharmacotherapy for smoking cessation, using data pooled from two clinical trials they found a significant mediating effect of other DRD2 polymorphisms. The first study was a double-blind, placebo-controlled trial of bupropion 300 mg/day (N = 414) and the second was an open-label trial that compared transdermal nicotine with nicotine nasal spray (N = 368). In addition to ANKK1, these investigators examined the mediating effect of an insertion/deletion polymorphism in the DRD2 promoter region (−141C Ins/Del, rs1799732). The more common 141C Ins allele was previously shown to confer greater transcriptional efficiency than the 141C Del allele (Arinami et al., 1997). The possible influence of the aforementioned DRD2 C957T SNP was also assessed. At the end of the 6-month treatment phase, 141C Ins allele homozygotes showed a more favorable response to bupropion. In contrast, smokers carrying the 141C Del allele had significantly higher quit rates while on NRT, independent of NRT type. Individuals homozygous for the C957T T allele also demonstrated increased abstinence following NRT. Based on these data, the investigators concluded that bupropion may be a more effective pharmacologic treatment for smokers homozygous for the DRD2 14C1 Ins allele, while NRT may be more beneficial for those who carry the Del allele. Their data also suggest that T-allele homozygotes for the C957T variant may, in general, be better able to maintain abstinence. This is an interesting finding when considered in the context of the study by Wang et al. (2008), which found that individuals with this genotype may experience less nicotine withdrawal.

Lerman and colleagues replicated their findings with DRD2-141C Ins/Del in another 6-month NRT trial (N = 363; Dahl et al. 2006), in which they observed an interaction between DRD2 141C and variation in the gene encoding the neuronal calcium sensor-1 protein (FREQ), the protein product that regulates D2 receptor desensitization (Burgoyne et al. 2004; Kabbani et al., 2002). They found that 62% of the smokers with at least one copy of the DRD2 141C Del allele and two copies of the FREQ A allele (rs1054879) were abstinent from smoking, compared to 29–38% abstinence among other smokers.

Variants in the serotonin transporter gene (SLC6A4) have also been shown to influence the effectiveness of bupropion and nortriptyline on smoking cessation rates (Quaak et al., 2012). SLC6A4 encodes the serotonin transporter, a membrane protein that takes serotonin back up into pre-synaptic neurons. In this study, 214 daily smokers were randomized and genotyped for the 5-HTTLPR variant (rs25531) that can result in a short (S) allele (14 repeats, associated with lower gene transcription) or a longer one (16 repeats, higher transcription, Nakamura et al., 2000). Primary outcome measures were prolonged abstinence from weeks 4–12, 4–26 and 4–52. The researchers concluded that bupropion and nortriptyline seem to be more effective in smoking cessation among SLC6A4 high-activity variant carriers. See Table 1¹ for a list of nicotine pharmacogenetic treatment outcome trials.

Table 1.

Pharmacogenetic Medication Trials for Nicotine Abuse

Gene	Polymorphism	Treatment Medication	Sample Size	Population	Treatment × Genotype Finding* ↑ More effective ↓ Less effective ↔ No interaction	Citation
ANKK1	rs1800497	Bupropion	577	German	↑ CC genotype	Breitling et al., 2010
	rs1800497	Transdermal Nicotine	755	Primarily Caucasian (98%)	↑ CT/TT genotypes	Johnstone et al. (2004)
	rs1800497	Transdermal Nicotine	752	European	↑ CT/TT genotypes (women only)	Yudkin et al. (2004)
	rs1800497	Bupropion	722	Caucasian /European Ancestry	↑A2/A2 (CC) genotype	David et al. (2007b)
	rs1800497	Transdermal Nicotine	908	European or Unknown Ancestry	↓ CT/TT genotypes	Munafo et al. (2009)
	rs1800497	Transdermal Nicotine	419	Primarily Caucasian/European	↔	Stapleton et al. (2011)
	rs1800497	Rimonabant	76	European Decent	↑A2/A2 (CC) genotype	Wilcox et al. (2011)
ARRB2	rs1045280 rs2036657	NRT (transdermal, nasal spray)	374	European Ancestry	↔	Ray et al. (2007)
CHRNA4	rs2236196	NRT (all formulations)	316	European Ancestry	↔	Hutchison et al. (2007)
CHRNB2	rs2072661	Bupropion	412	European	↔	Conti et al. (2008)
	rs2072661	Transdermal Nicotine	156	European	↑ GG genotype	Perkins et al. (2009)
COMT	Val158Met: rs4680	NRT (transdermal nicotine, nicotine spray)	290	Females of European and African ancestry	↑ Met/Met	Colilla et al. (2005)
	rs4680	Transdermal Nicotine	749	European	↑ Met/Met	Johnstone et al. (2007)
	rs737865 rs165599	Bupropion	511	Caucasian- and African-American	↑ A/G allele of rs165599 ↑ GG at both SNPs (AA participants only)	Berrettini et al. (2007)
CYP2A6		Bupropion	414	European	↑ Fast metabolizers	Patterson et al. (2008)
		Transdermal Nicotine	568	European	↓ Fast metabolizers	Schnoll et al. (2009)
	CYP2A6 CHRNA5: rs1696998, rs680244	Bupropion NRT (lozenge, transdermal) Bupropion + NRT	709	Caucasian -American	↑ Fast metabolizers (NRT) ↑ CYP2A6: Fast metabolizers + CHRNA5: GG/TT,GA/CT, AA/CC (NRT)	Chen et al. (2014)
CYP2B6	C1459T: rs3211371	Bupropion	426	Caucasian-European	↑Female CT/TT	Lerman et al. (2002)
	CYP2B6: rs3211371 ANKK1: rs1800497 SLC6A3 3’ VNTR: rs27072	Bupropion	291	European	↑ CYP2B6 CT/TT variant + ANKK1 A2/A2	David et al. (2007b)
	*6	Bupropion	326	European Ancestry	↑ CYP2B6*6 carriers	Lee et al. (2007a)
	*1, *4, *6	NRT (transdermal, nasal spray)	369	European Ancestry	↔	Lee et al. (2007b)
DβH	DβH: rs77905 ANKK1: rs1800497	Transdermal Nicotine	755	Primarily Caucasian (98%)	↑ Taq1A CT/TT + DβH GA/AA genotypes	Johnstone et al. (2004)
	rs77905	Bupropion	577	German (98%)	↔	Breitling et al. (2010)
DRD2	−141C Ins/Del: rs1799732	NRT (transdermal, nasal spray)	368	European	↑ Smokers with at least one copy of the Del C allele	Lerman et al. (2006)
	rs1799732 FREQ: rs1054879	NRT (transdermal, nasal spray)	363	European	↑ Smokers with at least one copy of −141 Del allele and two copies of FREQ	Dahl et al. (2006)
	rs1799732	Bupropion	414	European	↑ Smokers homozygous for the Ins C allele	Lerman et al. (2006)
	C957T: rs6277	NRT (transdermal, nasal spray)	368	European	↑ C allele	Lerman et al. (2006)
	rs6277	Bupropion	414	European	↔	Lerman et al. (2006)
	Exon 8 A/G: rs6276	Transdermal Nicotine	755		↔	Johnstone et al. (2004)
DRD4	VNTR	Transdermal Nicotine	720	European Ancestry	↔	David et al. (2008)
	C-521T: rs1800955	Transdermal Nicotine	720	European	↔	David et al. (2008)
HINT1	rs3852209 rs2278060	NRT (transdermal, nasal spray)	374	European Ancestry	↑ rs3852209 TT genotype	Ray et al. (2007)
OPRM1	Asn40Asp (A118G): rs1799971	NRT (transdermal, nasal spray)	320	European	↑ G carriers w/transdermal nicotine (no interaction with nasal nicotine)	Lerman et al. (2004)
	rs1799971	Transdermal Nicotine	710	European Ancestry	↑ AA carriers	Munafo et al. (2007)
	rs1799971	NRT (transdermal, nasal spray)	374	European	↑ G carriers	Ray et al. (2007)
	rs1799971	Transdermal Nicotine	598	European Ancestry	↔	Munafo et al. (2013)
SLC6A3 ANKK1	3′ UTR VNTR Taq1A: rs1800497	Bupropion	418	European	↔	Lerman et al. (2003)
	3′ UTR VNTR Taq1A: rs1800497	Bupropion	416	European	↓ ANKK1 A1 and SLC6A3-9	Swan et al. (2007)
SLC6A3	3′ UTR VNTR	NRT (all formulations) Bupropion	583	European	↔	O'Gara et al. (2007)
	3′ UTR VNTR	Bupropion	291	European	↔	David et al. (2007b)
SLC6A4	5-HTTLPR	Nicotine Replacement Therapy (NRT): (transdermal patch, nasal spray)	397	European	↔	Munafo et al. (2006)
	5-HTTLPR	Transdermal patch	741	European	↔	David et al. (2007c)
	5-HTTLPR	Bupropion Nortriptyline	214	Caucasian	↑ carriers L-variant (rs25531) Bupropion ↑ carriers of 3 combined high-activity variants Bupropion & Nortriptyline	Quaak et al. (2012)

^*Relationship between genotype and pharmacotherapy effectiveness outcome

4.2. Opioids

A number of studies have examined the effect of genetic variation on features of opioid dependence. One study investigated the mediating effect of the dopamine receptor gene (DRD4) 48-bp VNTR polymorphism (2, 4 and 7 repeat variants) on cueinduced craving in 420 heroin users (Shao et al., 2006). This polymorphism was selected based on its prior association with addictive disorders (Benjamin et al., 1996) and because of evidence that it is functional (Asghari et al., 1995; Van Tol et al., 1992). Following exposure to neutral and heroin-related cues, such as syringes, lighters, and tinfoil, participants completed a 5-item craving questionnaire. The study showed significantly greater cue-elicited heroin craving among individuals carrying the DRD4 VNTR 7 (longer) allele than in non-carriers. A similar study showed significantly greater cue-elicited heroin craving among Taq1 A1 allele carriers (A1/A1 or A1/A2 genotypes) than in A2 homozygotes (Li et al., 2006). These variants were identified as potential genetic risk factors for cue-induced drug craving and subsequent relapse to substance use. Though other findings relating DRD4 VNTR allele and cue reactivity have been contradictory (MacKillop et al., 2007; van den Wildenberg et al., 2007), Hutchison et al. (2002) found an association between this allele with greater cue-induced craving and arousal among smokers.

The effects of a number of opioid receptor gene polymorphisms have been implicated in subjective response to the first use of opioids. One such study assessed the association of several OPRM1 variants with positive vs. negative subjective responses to initial heroin use (Zhang et al., 2007a). Heroin users (N = 336) were divided into two groups based on their retrospective recall concerning their first experience with heroin (i.e., positive vs. negative groups). Association analyses were performed between these two groups and nine OPRM1 SNPs. An initially positive response upon first use of heroin was found to be associated with three OPRM1 SNPs (rs696522, rs1381376, and rs3778151), although the functional significance of these SNPs has yet to be determined. These data highlight the need for more studies into the mediation of subjective response to drugs by genes and how this consequently affects susceptibility to abuse.

Additional research has shown an association between gene variants and treatment outcome among opioid users. Methadone and buprenorphine are two common and effective maintenance medications for opioid use disorder (Ball and Ross, 1991; Joseph et al., 2000; Mattick et al., 2008). Lawford and colleagues (2000) reported a four-fold increase in the frequency of the Taq1 A1 allele among a sample of 95 treatment-seeking opioid users who responded poorly to methadone treatment compared to those with successful treatment outcomes. However, the association of the Taq1 A1 allele with the treatment effectiveness of opioid substitution therapy was not confirmed by a later study (Barrat et al., 2006). No significant differences in A1 allele frequency were observed between methadone- and buprenorphine-treated patients with poor and successful treatment outcomes (N = 116). Interestingly, the researchers observed significantly less opioid withdrawal among carriers of the A1 allele who had a more successful treatment outcome on methadone. Crettol et al. (2008) also failed to find an association between the presence of the Taq1 A1 allele and response to methadone treatment in a sample of 238 methadone-maintained patients. They found that the Taq1 A1 allele was not associated with response to methadone treatment, as measured by use of heroin or cocaine, withdrawal symptoms, program attendance and required daily methadone dose. They also found a lack of an association with the common variants of the D1 dopamine (Ddel, −48A>G) and mu opioid Asn40Asp (A118G) receptor gene, though they did find an association with the other DRD2 polymorphism C957T: patients with the C/C genotype were more frequently nonresponders to methadone treatment.

A later study failed to find an association between OPRM1 alleles and methadone treatment response, but did report significant interactions with other genes implicated in the pharmacodynamics of methadone (Fonseca et al., 2010). In addition to OPRM1, researchers assessed the mediating effects of genes encoding the metabotropic glutamate receptors (GRM6, GRM8), the nuclear receptor (NR4A2), the photolyase enzyme cryptochrome 1 (CRY1), and the transcription factor myocardin (MYOCD), which have previously been associated with the risk of opioid abuse (Nielsen et al., 2008). A sample of 116 opioid-dependent patients were classified as methadone responders (n = 83) and nonresponders (n = 33) according to illicit opioid use detected with urinalysis. Participants carrying the CRY1 AA genotype at (rs1861591) had a higher risk of being nonresponders. Participants carrying the MYOCD A allele (rs1714984) had an increased risk of being nonresponders but only if they were also carriers of the GRM6 AG genotype (rs953741).

Subsequent research has shown pharmacogenetic interactions between methadone pharmacokinetics and dosing requirements in a sample of 312 opioid-dependent patients (Hung et al., 2011). This study examined SNPs (G516T, A785G) in CYP2B6, which encodes the liver enzyme that is partially responsible for methadone metabolism and the activity of which is known to alter methadone plasma levels (Crettol et al., 2006). Also assessed were a common SNP (C3435T) of the highly polymorphic ABCB1 gene (rs1045642) that has been associated with disease severity and treatment response (Coller et al., 2006; Levran et al., 2008a) and the A118G SNP in of the gene encoding the mu opioid receptor (OPRM1). Investigators also assessed several polymorphisms associated with D2 expression or function: DRD2 C957T, DRD2 C939T, DRD2 A214G, ANKK1 C2137T (Crettol et al., 2008; Doehring et al., 2009; Lawford et al., 2000). Carriers of the variant ABCB1 3435C>T and CYP2B6 516G>T alleles were significantly more likely to require higher methadone doses than non-carriers. In contrast, carriers of the DRD2 -214A>G and 939C>T alleles required a significantly lower methadone dosage than non-carriers (Hung et al., 2011, see also Levran et al., 2013 and Barratt et al., 2012). A recent meta-analysis of the literature concerning a number of ABCB1 and CYP2B6 failed to find an overall effect of allelic variation and treatment response to methadone (Dennis et al., 2014). Their analysis did find that the CYP2B6*6 haplotype did significantly influence methadone metabolism. Specifically, methadone metabolism was significantly slower in *6 homozygous carriers.

Although variation in OPRM1 has been more extensively studied than the genes encoding other opioid receptors, variation in the delta opioid receptor gene (OPRD1) has been associated with the risk of opioid abuse and dependence (Mayer et al., 1997; Zhang et al 2008). Accordingly, researchers have attempted to assess whether OPRD1 variants may also mediate the response to opioid substitution therapy. In one such trial, patients (N = 643) were randomized to receive buprenorphine/naloxone or methadone maintenance treatment over the course of 24 weeks (Crist et al., 2013 a). The researchers’ primary measure of treatment effectiveness was illicit opioid use, measured by weekly urinalysis. Six OPRD1 SNPs were genotyped as predictor variables. Among African-American participants (N = 77), an intronic OPRD1 SNP (rs678849) predicted treatment outcome for both pharmacotherapies. Methadone patients with the C/C genotype were less likely to have opioid-positive urine tests than T-allele carriers (C/T and T/T genotypes). The opposite outcome was observed with respect to buprenorphine/naloxone treatment. Patients with the C/C genotype were more likely to have positive opioid drug screens than T-allele carriers. Though it is unclear what effect this SNP has on OPRD1 transcription or delta receptor function, its association with both methadone and buprenorphine/naloxone treatment outcome suggests that it either directly affects or is in linkage disequilibrium with a functional variant that mediates the effects of these medications. Interestingly, this group of investigators found similar associations between OPRD1 SNPs and cocaine abuse that were specific to African-Americans, highlighting the possible influence of population genetics in this type of research (Crist et al., 2013 b).

In a more recent study, investigators found that two other intronic OPRD1 SNPs (rs581111 and rs529520) predicted buprenorphine treatment outcomes. However, this interaction was found only in females. Women with the AA or AG genotypes at rs581111 had significantly worse outcomes compared to women with the GG genotype, while women with rs529520 AA genotype had a significantly worse outcome compared to women with the CC genotype (Clarke et al., 2014 A). These studies highlight the potential of pharmacogenetic studies to provide important information for the selection of treatment when different options are available. Table 2 lists opioid use disorder pharmacogenetic treatment outcome trials, and the nature of their findings.

Table 2.

Pharmacogenetic Medication Trials for Opioid Abuse

Gene	Polymorphism	Treatment Medication	Drug of Abuse	Sample Size	Population	Treatment × Genotype Finding* ↑ More effective ↓ Less effective ↔ No interaction	Citation
ANKK1	rs1800497	Methadone	Heroin	95	Caucasian	↓ T allele	Lawford et al. (2000)
	rs1800497	Buprenorphine	Heroin	25	Primarily Caucasian (88%)	↔	Barrat et al. (2006)
	rs1800497	Methadone	Heroin	46	Primarily Caucasian (89%)	↔	Barrat et al. (2006)
	rs1800497	Methadone	Heroin	238	Caucasian	↔	Crettol et al. (2008)
CYP3A4	*1B	Methadone	Opioids	245	Caucasian-European	↔	Crettol et al. (2006)
CYP2B6	*4, *5, *6, *9	Methadone	Opioids	245	Caucasian-European	↔	Crettol et al. (2006)
DRD1	rs4532	Methadone	Heroin	238	Caucasian	↔	Crettol et al. (2008)
DRD2	rs6277	Methadone	Heroin	238	Caucasian	↑ CC genotype	Crettol et al. (2008)
Multiple Variants	MYOCD: rs1714984, GRM8: rs1034576, CRY1: rs1861591, GRM6: rs953741, OPRM1: rs1074287, NR4A2: rs1405735, intergenic variants: rs965972, rs1867898	Methadone	Opioids	169	Caucasian	↓CRY1 AA ↓ MYOCD A allele + GRM6 AG	Fonseca et al. (2010)
OPRD1	rs2234918	Methadone	Heroin	238	Caucasian	↔	Crettol et al. (2008)
	rs678849 rs1042114 rs10753331 rs59520 rs581111 rs2234918	Buprenorphine	Opioids	566 Caucasian 77 African-American	Caucasian- and African-American	↓ African-American rs678849 CC genotype	Crist et al. (2013 A)
	rs678849 rs1042114 rs10753331 rs59520 rs581111 rs2234918	Methadone	Opioids	566 Caucasian 77 African-American	Caucasian- and African-American	↑ African-American rs678849 CC genotype	Crist et al. (2013 A)
	rs678849 rs1042114 rs10753331 rs59520 rs581111 rs2234918	Buprenorphine Methadone	Opioids	582	European-American	↓ Female rs581111 AA/AG genotype ↓ Female rs529520 AA genotype	Clarke et al. (2014 A)
OPRM1	rs1799971	Methadone	Heroin	238	Caucasian	↔	Crettol et al. (2008)
	G-172T: rs6912029 G-1510A: rs12205732	Oral Naltrexone	Opioids	138	Arab	↓ 172GG and 1510GG	Al-Eitan et al. (2012)

4.3. Cocaine

Cocaine dependence has a strong genetic component, with an estimated heritability of 72%, but unlike the other substances discussed here, lacks an FDA-approved pharmacotherapy (Goldman et al., 2005). Researchers have tested over 60 medications, and though some have shown promise, none so convincingly as to have received FDA approval (Grabowski et al. 2004; Vocci and Ling 2005). Pharmacogenetics may help to identify subsets of users who respond to a particular pharmacotherapy, which could then be developed and marketed to a patient population with a specific genetic profile.

The acetaldehyde dehydrogenase-inhibiting drug disulfiram (Antabuse, Antabus), has been used to treat alcoholism since the 1940’s (Hald et al., 1948). Disulfiram has also shown some potential in treating cocaine abuse, and research has begun to examine whether genetic variation mediates treatment response (Carroll et al., 1998; 2004). Some researchers have proposed that by inhibiting dopamine β-hydrolylase (DβH), disulfiram alters cocaine signaling and/or availability (Bourdelat-Parks et al., 2005; Schank et al., 2006; Vaccari et al., 1996). Consequently, genetic variants affecting the DβH gene were among the first to be examined as genetic mediators of the efficacy of disulfiram in treating cocaine dependence.

A functional DβH gene polymorphism that alters transcription and decreases DβH plasma levels was studied with respect to its interaction with disulfiram (Bhaduri et al., 2008; Zabetian et al., 2001). The C-1021T polymorphism (rs1611115) accounts for up to 52% of the variation in DβH enzyme levels, with individuals homozygous for the T allele having the lowest levels (Khonke et al., 2002; Deinum et al., 2004). In their assessment of the influence of this polymorphism, Kosten and colleagues (2013) randomized 72 cocaine- and opioid-dependent patients to receive disulfiram (250 mg/day) or placebo for 10 weeks. Overall, disulfiram treatment significantly reduced the number of cocaine-positive urines but the effect was only seen among subjects with the DβH gene variant that is associated with normal DβH levels.

Spellicy et al. (2013) also found that the aforementioned ANKK1 (rs1800497) and DRD2 (rs2283265) polymorphisms interacted with disulfiram treatment response. Patients homozygous (T/T) or heterozygous (C/T) for the minor alleles of both SNPs had a significantly lower number of cocaine-positive urines when maintained on disulfiram (250 mg/day), while their homozygous counterparts showed a marginal (rs2283265*CC) or no (rs1800497*CC) treatment effect (N = 68).

Other analyses of genetic mediators of disulfiram treatment have focused on a SNP in the gene that encodes 5, 10-methylene tetrahydrofolate reductase (MTHFR). In vitro studies have shown that the minor T allele of the C677T SNP (rs1801133) results in an enzyme with reduced activity and increased thermolability (Frosst et al., 1995). Because this enzyme affects a number of developmental and biochemical pathways, Spellicy and colleagues (2013) hypothesized that the polymorphism would be associated with the response to disulfiram treatment in cocaine users. The primary treatment outcome measure was the number of cocaine-positive urines. The percentage of cocaine-positive urines dropped from 73% to 52% for the combined C/T and T/T group pre- to post-disulfiram treatment. Subjects with the C/C genotype had a smaller reduction in cocaine-positive urines (from 81% to 69%) while on disulfiram, a small but significant treatment effect when compared to placebo.

Investigations into the interaction between disulfiram and a functional variant of the gene that encodes the α-1 adrenergic receptor (ADRA1A) are based on disulfiram’s effect on norepinephrine production (Shorter et al., 2013). This polymorphism (rs1048101) in exon 2 results in a substitution of arginine for a cytosine within the C-terminus of the protein (Cys347Arg). Data suggest that this single amino acid change alters the activity of the receptor and may impact cognition (Lei et al., 2005). Using the clinical trial methodology described above (disulfiram 250 mg/day, 69 cocaine and opioid co-dependent participants; Shorter et al., 2013), the investigators found a significant interaction between genotype and treatment, where the patients who were carriers of the minor T allele had a significantly lower percentage of cocaine-positive urines, while those homozygous for the major C allele showed no treatment effect.

Genetic mediation of treatment response has also been found for a novel immunotherapy that produces selective anti-cocaine antibodies (cocaine vaccine (TA-CD); Carrera et al., 1995). Nielsen et al. (2013) genotyped 66 of 114 cocaine- and opioid-dependent patients enrolled in a 16-week trial of the vaccine for a variant of the kappa subtype of opioid receptor, OPRK1 gene (rs6473797), an A to G transition, with the G allele previously shown to be a risk allele for opioid dependence (Levran et al., 2008b). Participants homozygous for the major (A) allele had a significantly better vaccine treatment response (i.e., decrease in cocaine-positive urines) than carriers of the G allele. A similar investigation by this group (Kosten et al., 2013) identified the C-1021T DβH SNP as a mediator of cocaine vaccine response (See Table 3 for pharmacogenetic medication trials for cocaine abuse trials).

Table 3.

Pharmacogenetic Medication Trials for Cocaine Abuse

Gene	Polymorphism	Treatment Medication	Sample Size	Population	Treatment × Genotype Finding* ↑ More effective ↓ Less effective ↔ No interaction	Citation
ADRA1A	rs1048101	Disulfiram	69	Caucasian (>50%); others unidentified	↑ carriers of at least one T allele	Shorter et al. (2013)
	rs1611115	Cocaine Vaccine	71	Primarily Caucasian (79% and greater)	↑ CT/TT genotype	Kosten et al. (2013)
	rs1611115	Disulfiram	74	Primarily Caucasian (67% and greater)	↑ CC/TT genotype	Kosten et al. (2013)
ANKK1	rs1800497	Disulfiram	68	Caucasian-, African- and Latino -American	↑ CT/TT genotypes	Spellicy et al. (2013)
DRD2	rs2283265	Disulfiram	68	Caucasian, African-American, & Latino	↑ GT or TT genotypes	Spellicy et al. (2013)
MTHFR	rs1801133	Disulfiram	67	Between 43% and 90% Caucasian, others unidentified	↑ CT or TT genotypes	Spellicy et al. (2013)
OPRK1	rs6473797	Cocaine Vaccine	114	Primarily Caucasian (80% and greater)	↑ AA allele carriers	Nielsen et al., (2013)

Cocaine-induced paranoia is another interesting area of pharmacogenetic research. Association studies have failed to consistently demonstrate heritability and genetic influence in the occurrence of this drug effect (Gelernter et al., 2005; Kalayasiri et al., 2006). However, candidate gene studies have found evidence of allelic association with cocaine-induced paranoia. Variants that have been implicated as possible mediators include: SLC6A3 3’ VNTR (Gelernter et al., 1994), COMT variants: rs737866, rs4680, rs174696, rs4680, rs933271, rs993883, rs740603 (Ittiwut et al., 2011), DβH rs1611115 (Kalayasiri et al., 2007), CNR1 rs806368 (Zuo et al., 2009) and MANEA rs9387522 (Farrer et al., 2009).

4.4. Amphetamines and Caffeine

Pharmacogenetic research on methamphetamine and caffeine is very limited. However, there is research to suggest that OPRM1 variation mediates subjective response to amphetamine. One hundred and sixty-two Caucasian participants completed three sessions receiving either placebo or d-amphetamine (10 and 20 mg). Several OPRM1 SNPs (rs510769, rs2281617, rs1799971, rs510769, rs1918760, rs2281617 and rs1998220) were associated with significantly higher subjective ratings of “Euphoria,” “Energy” and “Stimulation” after 10 mg amphetamine. These data imply that genetic variability in the μ-opioid receptor gene influences the subjective effects of amphetamine, though the investigators were later unable to replicate these findings (Duglos et al., 2011).

With respect to methamphetamine, another area under investigation is possible genetic involvement in the occurrence of methamphetamine-induced psychosis. The majority of this research has focused on D2 receptor gene variants. A number of publications have identified DRD2 alleles as possible risk genotypes for methamphetamine psychosis (Harano et al., 2004; Ujike et al., 2009) with the body of literature expanding each year (Chanasong et al., 2013; Kishi et al., 2011; Kobayashi et al., 2004; Matsuzawa et al., 2007; Ujike et al., 2003).

Similarly, the aversive effects of caffeine consumption may have a genetic link. Caffeine produces mild psychostimulant and anxiogenic effects by antagonizing adenosine receptors and increasing dopaminergic transmission (Daly and Fredholm, 1998). Childs et al. (2008) examined functional polymorphisms in the genes for adenosine and dopamine receptors as possible mediators of caffeine-induced anxiety. They examined associations between self-reported anxiety following caffeine administration and variation in the genes for the A_2A receptor (ADORA2A) and D2 (DRD2) receptor. Healthy male and female individuals (N = 102), who consumed less than 300 mg caffeine per week, ingested capsules containing caffeine 0, 50, 150, or 450 mg under double-blind conditions, in four separate experimental sessions. Subjective anxiety was measured before and multiple times after ingestion of the capsules. They found a significant association between caffeine-induced anxiety and ADORA2A alleles: rs5751876, rs2298383, rs4822492, and DRD2 allele: rs1110976. This study provides evidence of mediation of aversive drug effects by genetic variation and suggests a mechanism by which these alleles may influence caffeine intake.

5. CONCLUSIONS

One cautionary tale concerning the reproducibility of pharmacogenetic research comes from one of the most prolific groups in this field, the Department of Human Genetics at the University of Chicago. Investigators there assessed acute response to d-amphetamine in 398 individuals with no history of drug abuse. Interim pharmacogenetic analyses using 99–162 participants from this data set found associations between a number of gene variants and amphetamine response: ADORA2A (Hohoff et al., 2005), BDNF (Flanagin et al., 2006), CSNK1E (Veenstra-VanderWeele et al., 2006), COMT (Hamidovic et al., 2010a), DRD2 (Hamidovic et al., 2009), FAAH (Dlugos et al., 2010), OPRM1 (Dlugos et al., 2011), SLC6A3 (Hamidovic et al., 2010b; Lott et al., 2005), SLC6A4 (Lott et al., 2006), and SLC6A2 (Dlugos et al., 2007). The promising data from these publications were cited throughout the psychiatric genetics literature. However, these researchers recently conducted a study with an additional 200 participants, in which they were unable to replicate the aforementioned findings (Hart et al., 2013).

The authors acknowledged that their failure to replicate indicated that the previous findings may have been false positives. The paper also cites insufficient correction for multiple testing (within each study and across the various studies) and publication bias as contributors to their erroneous findings. The influence of these factors requires that appropriate caution be used in pharmacogenetics research, which may warrant novel statistical or methodological approaches to control for their influence (Bosker et al., 2011; Munafó et al, 2007).

Though we have reviewed several studies showing a pharmacogenetic effect, the literature is often inconsistent about whether a particular gene alteration is related to the treatment outcome under observation. In addition to statistical sources of discrepant findings, other methodological differences in these studies may contribute to inconsistencies in the literature. Therefore, as the field of pharmacogenetics continues to develop, it would greatly benefit from standardization in measures such as the dosage of the pharmacological treatment being studied and how relapse is defined.

It has also been suggested that the source of this problem lies in the fact that our classification system for SRDs lacks etiological (causative) justification. Following this argument, the new DSM-V classification system, which combines substance abuse and dependence into an overarching disorder (Substance Use Disorder), does little to improve upon this issue. Some argue that the classification of patients with SRDs based upon endophenotypes related to disease pathogenesis may identify more distinct groups of affected individuals, which would be more suitable for experimental genetic approaches (see Ball et al., 1995; Basu et al., 2004; Chan et al., 2011; Kranzler et al., 2008 and Moss et al., 2007). For example, two possible endophenotypes of heroin users may be differentiated based their motivation to use the drug. The ability of heroin to activate the opioid and dopaminergic systems and produce pleasurable effects is undoubtedly an important contributor to its abuse liability. However, multiple studies have suggested that some individuals abuse opioids to self-medicate anxiety, depression, mania and psychosis (Fatseas et al., 2009). As such, the etiology of heroin abuse in some patient populations (initially at least) may be attributed to the positive subjective effects of the drug; these individuals may be genetically distinct from those whose abuse can be attributed to the negative reinforcing properties of the drug. As such, studies that examine pharmacogenetics of heroin abuse treatment response may yield different results, depending on the endophenotype of the sample that was tested. In addition to controlling for sample heterogeneity, employing larger samples sizes could improve replicability, leading to more clinically meaningful findings. In this respect, the field of substance abuse could learn from genetic studies investigating response to antidepressants (which utilize large sample sizes, control for population stratification, and employ sound statistical methodology) as a model for future studies (see Hunter et al., 2013 for an example).

The data presented in this review indicate that genetic variation as minimal as a single nucleotide change may alter drug effects, craving, withdrawal and abstinence. Because DA may constitute a pathway to addiction common to all drugs, the majority of this research has focused on genetic alteration in the dopaminergic pathway. However, the gene targets of this type of research are expanding each year. When combined with polygenetic approaches examining the combined effects of multiple variants, we continue to gain a more comprehensive view of the multitude of genetic factors involved in drug abuse. Despite the obstacles in this novel area of research, once refined, pharmacogenetics may allow for the identification of a subset of treatment-seeking patients most likely to respond to a particular medication and those who may be at an increased risk of more severe complications related to drug abuse. The development and course of drug addiction is a highly individualized experience. Genetics may allow us to determine why there is such variability in susceptibility, response, and prognosis. In the future, understanding the influence of genetic factors may change standard clinical practices and improve the lives of those suffering from drug abuse worldwide.

Highlights.

• We reviewed pharmacogenetic studies on drug subjective effects and medication trials

• These studies provide evidence of clinically relevant pharmacogenetic effects

• This body of literature has produced conflicting results that require resolution

• More rigorous procedural and statistical controls may reduce false positives

• Understanding genetic modulation may advance individualized Tx of drug use disorders