Genetics of methamphetamine use disorder: A systematic review and meta-analyses of gene association studies

Abstract

Genetic susceptibility to methamphetamine use disorder is poorly understood. No twin or adequately powered genome-wide association studies (GWASs) have been conducted. However, there are a large number of hypothesis-driven candidate gene association studies, which were systematically reviewed herein. Seventy-six studies were identified, investigating markers of 75 different genes. Allele frequencies, odds ratios, 95 % confidence intervals and power were calculated. Risk of bias was also assessed as a quality measure. Meta-analyses were conducted for gene markers if three or more studies were available. Eleven markers from adequately powered studies were significantly associated with methamphetamine use disorder, with Fatty Acid Amide Hydrolase (FAAH) and Brain Derived Neurotrophic Factor (BDNF) representing promising targets. Limitations of these studies include unclear rationale for candidate gene selection, low power and high risk of bias. Future research should include replications to enable more meta-analyses, well-powered GWASs or whole exome or genome sequencing, as well as twin and family studies to further complement the findings of this review to uncover genetic contributions toward methamphetamine use disorder.

1. Introduction

Amphetamine-like stimulants are the second most used illicit drug in the world, behind cannabis (UNODC, 2017a). Among amphetamines, methamphetamine (meth) represents the greatest global health threat, with a rapidly increasing market and number of users worldwide (UNODC, 2017b). In their most recent report, the U.S. Substance Abuse and Mental Health Services Administration (SAMHSA) estimated that 13 million people in the U.S. used meth in their lifetime (~4% of the total population), with 569,000 people using meth in the past month (SAMHSA, 2015). Meth use appears to be increasing globally, most likely due to expanding markets (UNODC, 2017b), highlighting the need to better understand meth use disorder.

1.1. Heritability of meth use disorders

Evidence from family, twin and adoption studies suggests substantial heritability for substance use disorders (for comprehensive reviews see Ducci and Goldman, 2012; Goldman et al., 2005; Vanyukov and Tarter, 2000). Substance use disorders are regarded as highly polygenic, with many hundreds of genes likely contributing to their heritability, each of which contribute a miniscule fraction of the overall risk (Ducci and Goldman, 2012). While the heritability of stimulant use disorders is estimated to be close to 40 % (Goldman et al., 2005), no twin or family studies have been conducted in people with meth use disorder, making it challenging to uncover the heritability of risk for this particular drug. Notably, that 40 % figure seems consistent across numerous classes of abused drugs including stimulants, cannabinoids, opioids, nicotine and alcohol.

Another approach to investigate the heritability of disorders are genome-wide association studies (GWASs). To date, only three GWASs have been conducted in people with meth use disorder (Ikeda et al., 2013; Sun et al., 2019; Uhl et al., 2008b). While Uhl et al. (2008b) and Ikeda et al. (2013) did not detect significant genome-wide single nucleotide polymorphism (SNP) associations, Sun et al. (2019) more recently reported an association between meth use disorder and three SNPs. It should be noted that the first two studies used overlapping samples, and all three studies were gravely under-powered for GWASs. Specifically, study with the largest sample size included 1750 people with meth use disorder, compared to the tens or even hundreds of thousands of participants required for reliable genetic findings for similarly polygenic complex psychiatric disorders. Of note, the three studies exclusively recruited people of Han Chinese or Japanese ethnicity. There is preliminary evidence of ethnic divergence of gene variants for meth use disorder (Bousman et al., 2010).

While twin studies have been lacking and GWAS results have been inconclusive, there has been an accumulation of candidate gene association studies in people with meth use disorder over the past two decades. Though not without important limitations, candidate gene studies can identify specific sequence variations in genes of interest and point towards potential molecular pathways affected to inform future research. While an association of genetic variants of a given candidate gene with some functional endpoint can potentially teach us about the biological effects of that gene and its naturally occurring sequence variations, this is very different from claiming that those gene variants contribute to heritable risk for a given psychiatric disorder. The latter claim can only be verified with rigorous genome-wide approaches or the careful study of large family pedigrees.

The first aim of the present study was therefore to systematically review gene association studies in people with meth use disorder to identify any promising targets and possible pathways underlying meth use disorder. Genotype, allele frequencies, and haplotypes were assessed. For specific markers with 3 or more studies available, meta-analyses were also conducted. Importantly, risk of bias assessment was conducted as a quality measure for each study using the National Heart, Lung, and Blood Institute Quality Assessment of Case-Control studies (NHLBI, 2020).

Candidate gene studies can be classified into two broad categories: those that examine possible influencing factors predisposing to the development of substance use disorders, (e.g., poor inhibition and impulse control), and those that specifically examine putative genetic variants in neurotransmitter and receptor systems involved in the substance use disorder of interest. In addition, Uhl et al. (2008a) suggested that the genetics of substance use vulnerability can either be due to individual differences in specific drug targets, or individual differences in receptor drug responses which may be shared across different class of substances (Uhl et al., 2008a). In that regard, the second aim of this review was to examine genetic contribution of broad substance use in meth use disorder to understand polysubstance use. We also systematically reviewed other secondary outcomes that the genetic association studies explored, including sex differences and age of first meth use.

2. Method

2.1. Literature search

This systematic review was conducted following the PRISMA guidelines (Moher et al., 2009). The protocol was registered with PROSPERO (CRD42020177826) after preliminary searches were conducted, but before literature search. The entire PubMed, EMBASE, PsycINFO and MEDLINE databases were searched for studies from their date of inception to April 10, 2020. The following search terms were used: “methamphetamine” OR “methylamphetamine”; and a combination of the following terms: “gene” OR “allele” OR “genotype” OR “polymorphism” OR “DNA” OR “haplotype” OR “twin” OR “linkage” OR “expression” OR “protein” OR “RNA”. To limit results to studies in humans, terms “human” OR “participant” were also used. Duplicated papers were removed, then titles and abstracts were screened for inclusion by two authors (AAG and JHK) using Covidence (Veritas Health Innovation, Melbourne, Australia).

2.2. Inclusion and exclusion criteria

Studies were included if they: (1) included human participants with a formal diagnosis (DSM or ICD) of meth use disorder, dependence or abuse; (2) included a healthy comparison group; (3) collected genetic information using human tissues; (4) reported genotype/allele distribution; and (5) were published in English language in a peer-reviewed journal.

Candidate gene variant studies were included. Studies were excluded if they were: (1) other systematic reviews or meta-analyses; (2) conference proceedings published as abstract only; (3) case reports; (4) book chapters; or (5) animal studies.

2.3. Quality assessment

Each study was assessed by two authors (AAG and JHK) using the NHLBI Quality Assessment of Case-Control Studies, which consist of 12 items (NHLBI, 2020). Each item was given a score of 1 (low risk of bias) or 0 (high risk of bias) to result in the total score for each study ranging from 0-12, with 0 suggesting a very high risk of bias, and 12 representing little to no risk of bias. Each assessor scored each study included in the review, and any conflict between assessors was resolved through discussion. The percent agreement between the two assessors was calculated across all 12 items.

2.4. Data extraction

For each study, the following information were extracted: (1) sample size; (2) mean age, sex, and ethnicity; (3) diagnostic tool (DSM or ICD) used to confirm diagnosis of meth use disorder; (4) existing comorbidities (including polysubstance use and meth-induced psychosis); (5) drug use parameters for the meth use disorder group (age of onset, duration of use, and severity of dependence); (6) type of biospecimen collected; (7) genes studied, including their location, function, and variant type (e.g., SNP, deletion, repeat etc.) when relevant; and (8) genotype and allele frequency data.

2.5. Data synthesis and analysis

Chi-square analyses were used to calculate significance levels for genotype and allele distribution between controls and people with meth use disorder for each study. Odds ratios and 95 % confidence intervals were also calculated for minor alleles. Power for each association was calculated using the total sample size and effect size w, using the minor allele frequency and sample size reported in each study, as described in Bousman et al. (2009). When a study reported more than one association, frequencies and power were calculated for each association separately. When enough studies were available, a meta-analytic approach was used. The outcome measure used was the effect size (odds ratio) for association between minor allele and meth use disorder using a random-effects model. Weights for each study were calculated in Rev-Man5 (The Cochrane Collaboration, 2020) using sample size and number of minor alleles. Heterogeneity was estimated using the I² statistic, with I² ≤ 50 % indicating low to moderate heterogeneity, and I² > 50 % indicating moderate to high heterogeneity. Sensitivity analyses were conducted based on quality of studies by systematically excluding the study with the highest risk of bias of each analysis. Publication bias assessment was not conducted as there were fewer than 10 studies included in each analysis (Higgins and Green, 2011).

Statistical analyses were performed using IBM SPSS Statistics 26 (IBM Corp., NY, USA) for genotype/allele analyses, G*Power 3.1 (Erdfelder et al., 2009) for power analyses, and Review Manager 5.4 (The Cochrane Collaboration, 2020) for the meta-analyses. Statistical significance was defined as p < 0.05.

3. Results

3.1. Study selection and quality assessment

From the 2330 studies identified through database search, 1065 remained following removal of duplicates (Fig 1). Following title and abstract screening by two independent assessors, 239 articles underwent full-text assessment for eligibility. In total, 79 studies met the inclusion criteria, for a total of 44,184 participants (17,682 cases and 26,502 controls). Three studies were GWASs, and 76 studies were hypothesis-driven candidate gene studies, investigating 75 unique genes (Supplementary Table S1).

Fig. 1.

PRISMA Diagram illustrating stages of article processing for the systematic review.

Summary of eligible studies are found in Table 1 and Fig. 2. Eligible studies either used ICD-10, DSM-IV or DSM-5 criteria, with four studies using a combination of these (Fig. 2A). A large majority of studies reported co-morbidities, with the most common being meth-induced psychosis and polysubstance use disorder (Fig. 2B). The most studied populations were of Japanese and Han Chinese descent, with one study investigating both populations (Fig. 2C). Three studies recruited Malaysians participants of four different ethnicities: Malay, Chinese, Kadazan-Dusun and Bajau. Two studies reported predominantly Caucasian samples, which also included other ethnicities. Fourteen studies exclusively examined male participants, and three studies did not report sex distribution (Fig. 2D).

Table 1.

Studies characteristics listed alphabetically based on first author.

Study	N	People with meth use disorder		Controls		Ethnicity	Diagnosis	NHLBI Score
		Age (mean ± SD)	Sex (% female)	Age (mean ± SD)	Sex (% female)
Aoyama et al., 2006	655	37 ± 11.7	17.8	41.3 ± 14.4	59.8	Japanese	ICD-10 F15.2 and F15.5	5
Bousman et al., 2010	193	38.6 ± 8.2	0	39.6 ± 10.2	0	Caucasian	DSM-IV Dependence	8
Chanasong et al., 2013	203	29.5 ± 4.6	0	28.8 ± 6.4	0	Thai	DSM-IV Dependence	3
Chen et al., 2004	851	NR	36.8	NR	43	Han Chinese	DSM-IV Abuse	5
Cheng et al., 2005	225	27.8 ± 6.8	0	29.8 ± 8.5	0	Han Chinese	DSM-IV Dependence	7
Ezaki et al., 2008	363	37.13 ± 11.0	16.3	38.96 ± 13.0	20.8	Japanese	ICD-10 F15.2 and F15.5	7
Ghafouri-Fard et al., 2020	426	35.43 ± 11.77	0	36.76 ± 11.15	0	Iranian	DSM-5 Meth use disorder	6
Han et al., 2008	77	35.1 ± 6.8	13.6	34.4 ± 6.7	15	Korean	DSM-IV Dependence	6
Hashimoto et al., 2005	388	36.9 ± 11.9	21.1	37.2 ± 10. 5	21.1	Japanese	ICD-10 F15.2 and F15.5	7
Hashimoto et al., 2008	451	36.9 ± 12.0	19.3	38.7 ± 12. 6	19.7	Japanese	ICD-10 F15.2 and F15.5	5
Hong et al., 2003	215	28.7 ± 6.9	0	28.1 ± 6.4	0	Han Chinese	DSM-IV Dependence	5
Iamjan et al., 2015	202	29.5 ± 4.6	0	28.8 ± 6.4	0	Thai	DSM-IV Dependence	5
Iamjan et al., 2018	202	30.1 ± 5.1	0	28.6 ± 6.3	0	Thai	DSM-IV Dependence	5
Ide et al., 2004	351	35.7 ± 1.1	22.5	34.4 ± 1.6	25.8	Japanese	ICD-10 F15.2 and F15.5	6
Ide et al., 2006	360	35.9 ± 1.0	22.7	35.2 ± 1.8	22	Japanese	ICD-10 F15.2 and F15.5	7
Ikeda et al., 2006	619	36.7 ± 12.0	19.8	34.3 ± 13. 6	52.2	Japanese	DSM-IV Dependence and Abuse	3
Ikeda et al., 2007	723	36.8 ± 12.0	18.1	37.0 ± 14. 4	53.3	Japanese	DSM-IV Dependence/ICD-10 F15.2 and F15.5	5
Ikeda et al., 2013	1100	NR	28.6	NR	52.5	Japanese	ICD-10 F15.2 and F15.5	6
Inada et al., 2004	326	36	21.7	37	21.9	Japanese	ICD-10 F15.2 and F15.5	6
Itoh et al., 2005	391	36.6 ± 11.9	20.6	37.2 ± 10.6	21.8	Japanese	ICD-10 F15.2 and F15.5	7
Iwata et al., 2004	473	NR	NR	NR	NR	Japanese	ICD-10 F15.2 and F15.5	5
Jugurnauth et al., 2011	983	28.15	39	34.4 ± 10.0	42.9	Han Chinese	DSM-IV Abuse	6
Kanahara et al., 2009	465	37.18 ± 11.73	20.7	39.26 ± 12.62	19.3	Japanese	ICD-10 F15.2 and F15.5	4
Kinoshita et al., 2008	393	NR	NR	NR	NR	Japanese	ICD-10 F15.2 and F15.5	3
Kishi et al., 2008a	1004	34.6	20	37.6 ± 14.3	55.6	Japanese	DSM-IV Dependence	4
Kishi et al., 2008b	944	36.2	19.9	37.3 ± 14.3	56.7	Japanese	DSM-IV Dependence	4
Kishi et al., 2010	536	36.3 ± 11.4	19.6	37.6 ± 14.3	19.6	Japanese	ICD-10 F15.2 and F15.5	5
Kishi et al., 2011a	515	37.1 ± 12.1	16.8	36.7 ± 12.8	18.6	Japanese	DSM-IV Dependence	6
Kishi et al., 2011b	533	37.1 ± 12.1	18.6	36.7 ± 12.8	18.6	Japanese	DSM-IV Dependence	6
Kishi et al., 2011c	534	37.6 ± 12.2	16.8	37.6 ± 14.3	19.6	Japanese	ICD-10 F15.2 and F15.5	4
Kishimoto et al., 2008	440	38.1 ± 12.6	17.8	37.2 ± 12.0	20.6	Japanese	ICD-10 F15.2 and F15.5	7
Kobayashi et al., 2004	331	37.6 ± 12.0	18.8	38.6 ± 12.0	48.4	Japanese	ICD-10 F15.2 and F15.5	6
Kobayashi et al., 2006	430	37.5 ± 12.0	19.4	35.0 ± 11.5	29.6	Japanese	ICD-10 F15.2 and F15.5	6
Kobayashi et al., 2010	400	37.5 ± 12.0	19.3	41.2 ± 12.3	48	Japanese	ICD-10 F15.2 and F15.5	6
Kobayashi et al., 2011a	405	37.4 ± 12.0	19.8	35.4 ± 11.5	30.9	Japanese	ICD-10 F15.2 and F15.5	6
Kobayashi et al., 2011b	400	37.5 ± 12.0	19.3	41.2 ± 12.3	48	Japanese	ICD-10 F15.2 and F15.5	5
Koizumi et al., 2004	357	Average 35.2; 37.0 ± 11.0 (males); 28.0 ± 5.0 (females)	20.4	Average 36.8; 37.0 ± 11.0 (males); 36.0 ± 10.0 (females)	21.5	Japanese	ICD-10 F15.2 and F15.5	6
Kotaka et al., 2008	489	37.5 ± 11.9	19.5	37.6 ± 13.1	20.8	Japanese	ICD-10 F15.2	6
Li et al., 2004	851	27.2 ± 7.0	36.8	34.4 ± 10.0	43	Han Chinese	DSM-IV Abuse	5
Lin et al., 2003	685	Average 27.3; 28.73 ± 7.14 (males); 25.29 ± a6.23 (females)	42.7	Average 34.2; 35.13 ± 9.93 (males); 32.93 ± 10.14 (females)	43.2	Han Chinese	DSM-IV Abuse	6
Liu et al., 2004	218	29.7	11	29.6 ± 6.8	8.33	Han Chinese	DSM-IV Dependence	7
Liu et al., 2006	788	27.2 ± 7.1	37.5	34.5 ± 9.5	42.4	Han Chinese	DSM-IV Abuse	5
Loftis et al., 2019	106	NR	NR	NR	NR	Multiple	DSM-IV Dependence	7
Matsuzawa et al., 2007	426	36.9 ± 11.3	18.8	39.0 ± 12.3	19.7	Japanese	ICD-10 F15.2 and F15.5	5
Morita et al., 2005a	353	37.8 ± 12.1	19	37.3 ± 12.1	19	Japanese	ICD-10 F15.2 and F15.5	9
Morita et al., 2005b	353	37.8 ± 12.1	19	37.3 ± 12.1	19	Japanese	ICD-10 F15.2 and F15.5	9
Morita et al., 2008	414	37.4 ± 11.9	18.1	36.5 ± 10.6	22.4	Japanese	ICD-10 F15.2 and F15.5	7
	305	35.8 ± 11.0	18.1	37.3 ± 10.5	23.8	Japanese
Nakamura et al., 2006	339	27.3 ± 7.3	23.7	28.8 ± 7.4	47.1	Han Chinese	ICD-10 F15.2 and F15.5	3
Nakamura et al., 2009	317	35.8 ± 11.0	17.8	37.3 ± 10.5	22.6	Japanese	ICD-10 F15.2 and F15.5	6
Namvar et al., 2020	455	36.43 ± 9.79	0	37.76 ± 9.6	60	Iranian	DSM-5 Meth Use Disorder	4
Nishiyama et al., 2005	466	36.7 ± 12.0	19.1	33.6 ± 13.0	47.2	Japanese	DSM-IV Dependence and Abuse	6
Nomura et al., 2006a	352	35.5 ± 11.3	20.3	36.5 ± 10.6	22.5	Japanese	ICD-10 F15.2	6
Nomura et al., 2006b	406	36.8 ± 11.9	20	36.3 ± 10.5	22.6	Japanese	ICD-10 F15.2	5
Ohgake et al., 2005	398	36.7 ± 10.6	20.9	36.5 ± 11.9	22.2	Japanese	ICD-10 F15.2 and F15.5	7
Okahisa et al., 2009	510	36.9 ± 11.9	20.3	37.2 ± 13.2	22.2	Japanese	ICD-10 F15.2	5
Okochi et al., 2009	718	36.7 ± 11.6	17.4	37.5 ± 14.4	54.5	Japanese	ICD-10 F15.2 and F15.5	5
Okochi et al., 2011	450	36.7 ± 11.6	17.5	35.5 ± 14.4	18.7	Japanese	ICD-10 F15.2 and F15.5/DSM-IV Dependence	8
Otani et al., 2008	539	36.9 ± 11.8	17.3	37.2 ± 13.1	19.6	Japanese	ICD-10 F15.2	7
Payer et al., 2012	100	34.4 ± 9.4	41.5	32.1 ± 9.5	46.8	Multiple	DSM-IV Dependence	7
Šerý et al., 2001	225	21.6 ± 4.5 (SEM)	40.9	21.3 ± 3.4 (SEM)	52.3	Caucasian	DSM-IV Dependence	3
Sim et al., 2010	340	31.0 ± 8.2	0	33.0 ± 12.2	0	Multiple (Asian)	DSM-IV Dependence	6
Sim et al., 2013	473	31.0 ± 8.4	0	32.0 ± 11.0	0	Multiple (Asian)	DSM-IV Dependence	8
Sim et al., 2014	555	30.0 ± 8.0	0	32.0 ± 11.0	0	Multiple (Asian)	DSM-IV Dependence	8
Su et al., 2014	419	30.79 ± 7.99	16.5	33.68 ± 9.87	21	Han Chinese	DSM-IV Abuse	8
Su et al., 2015a	419	30.79 ± 7.99	16.5	33.68 ± 9.87	21	Han Chinese	DSM-IV Abuse	7
Su et al., 2015b	572	31.46 ± 8.40	17.5	45.99 ± 12.96	60.6	Han Chinese	DSM-IV Dependence	6
Sun et al. 2019 (Discovery)	4608	30.51 ± 7.40	28.24	34.17 ± 6.41	35.3	Han Chinese	DSM-5 Meth Use Disorder	5
Sun et al. 2019 (Replication)	2667	36.36 ± 8.52	0	33.71 ± 11.80	0	Han Chinese	DSM-5 Meth Use Disorder	5
Suzuki et al., 2006	343	35.8 ± 11.0	18.2	37.3 ± 10.5	22.5	Japanese	ICD-10 F15.2 and F15.5	8
Tsai et al., 2002	228	28.6 ± 6.6	0	28.1 ± 6.4	0	Han Chinese	DSM-IV Dependence	7
Tsunoka et al., 2010	998	37.0 ± 10.8	16.8	37.6 ± 14.3	56.2	Japanese	ICD-10 F15.2 and F15.5	5
Uhl et al., 2008b	380	32.5 ± 10.0	30	NR	NR	Han Chinese	DSM-IV Dependence	7
	200	39.9 ± 13.0	21	NR	NR	Japanese	ICD-10 F15.2 and F15.5
Ujike et al., 2003	284	35.8 ± 11.3	16.9	36.0 ± 10.2	21.9	Japanese	ICD-10 F15.2 and F15.5	9
Ujike et al., 2009	445	38.1 ± 12.6	17.3	37.2 ± 12.0	20.6	Japanese	ICD-10 F15.2	9
Veerasakul et al., 2016	28	34.1 ± 6.2	0	32.9 ± 6.4	0	Thai	DSM-IV Dependence	4
Veerasakul et al., 2017	202	30.1 ± 5.1	0	28.6 ± 6.3	0	Thai	DSM-IV Dependence	8
Yoon et al., 2005	262	34.7 ± 7.4	12.7	47.1 ± 10.0	21.1	Korean	DSM-IV Dependence	8
Yoshimura et al., 2011	602	37.1 ± 11.8	18.7	40.0 ± 14.6	58.2	Japanese	ICD-10 F15.2	3
Zhang et al., 2020	1061	28.42 ± 7.35	15.3	36.08 ± 10.21	26.8	Han Chinese	ICD-10 F15.2/DSM-5 Meth Use Disorder	8
Zhao et al., 2018	1239	35.96 ± 0.32 (SEM)	24.8	35.25 ± 0.46 (SEM)	29.9	Han Chinese	DSM-IV Dependence and Abuse	4

Notes: N = total study sample size; NR = not reported; ICD-10 = International Classification of Diseases 10th edition; DSM-IV/5 = Diagnostic and Statistical Manual of Mental Disorders 4th and 5th Edition; F15.2 = Other stimulant dependence; F15.5 = Other stimulant-induced psychotic disorder; NHLBI Assessment = National Heart, Lung, and Blood Institute Study Quality and Risk of Bias Assessment, where with 0 = highest risk of bias and 12 = little to no risk of bias.

Fig. 2.

Key characteristics of included studies. A. Diagnostic criteria. Studies predominantly used the ICD-10 and DSM-IV criteria. B. Comorbidities. Comorbidities were reported in 71 % of studies, with polysubstance use disorder and meth-induced psychosis the most common. C. Ethnicities. Participants of Japanese and Han Chinese descent were the most studied. Some studies belonged to more than 1 category. D. Sex. Most studies included both male and female participants, but 14 studies only included males. Three studies did not report participant sex. Numbers on each wedge represent the number of studies with the characteristic of interest.

The quality of the studies based on the NHLBI Quality Assessment of Case-Control Studies (see Methods) ranged from 3 to 9, with the lower the score, the higher the risk of bias (Table 1). The percent agreement between the two assessors was 86.0 %. The main limitations were poorly defined and implemented exclusion and inclusion criteria (item 5; 93.4 % of all studies), poorly defined and differentiated cases and controls (item 6; 75.0 %) and unjustified sample size (item 3; 72.4 %).

3.2. Candidate gene association studies

Table 2 provides the genotype and allele frequencies, calculated odds ratios (OR) and 95 % confidence intervals (95 % CI), p-values and power for each gene variant investigated. The most common type of variant studied were single nucleotide polymorphisms (SNPs). Deletions (Dels), variable number tandem repeats (VNTRs), restriction fragment length polymorphisms (RFLPs), and linked polymorphic regions (LPRs) were also examined. The characteristics of all genes investigated, including non-abbreviated full name, chromosomal location, and function are given in Supplementary Table S1. Note that the listed function of each gene was derived from each study’s rationale, and that genes investigated may be involved in various other functions and pathways.

Table 2.

Genotype and allele frequencies of candidate gene association studies in meth use disorder.

Gene	Code	Variant	Reference	Meth				Control				p-value	Meth		Control		OR	95% CI	p-value	Power (1 − β)
				M/M	M/m	m/m	n	M/M	M/m	m/m	n		M	m	M	m
ACE	rs4646994	Del	Sery 2001	17	51	19	87	33	61	32	126	0.323	85	89	127	125	1.06	0.72–1.57	0.754	0.06
ADORA1	rs56298433	SNP	Kobayashi 2011b	166	2	0	168	222	2	0	224	0.772	334	2	446	2	1.34	0.19–9.53	0.772	0.06
	rs10920568	SNP	Kobayashi 2011b	132	36	3	171	162	63	4	229	0.332	300	42	387	71	0.76	0.51–1.15	0.196	0.26
	rs5780149	SNP	Kobayashi 2011b	108	55	8	171	150	69	10	229	0.718	271	71	369	89	1.09	0.77–1.54	0.642	0.07
	rs6427994	SNP	Kobayashi 2011b	128	38	5	171	181	46	2	229	0.248	294	48	408	50	1.33	0.87–2.04	0.183	0.26
	rs41264025	SNP	Kobayashi 2011b	162	9	0	171	215	14	0	229	0.937	333	9	444	14	0.86	0.37–2.00	0.721	0.07
	rs16851030	SNP	Kobayashi 2011b	80	67	24	171	89	116	24	229	0.071	227	115	294	164	0.91	0.68–1.22	0.522	0.10
ADORA2A	rs13306114	SNP	Kobayashi 2010	168	3	0	171	229	0	0	229	0.044	339	3	458	0	N/A	N/A	0.045	0.52
	Exon1 + 219	SNP	Kobayashi 2010	169	2	0	171	228	1	0	229	0.401	340	2	457	1	2.69	0.24–29.77	0.402	0.14
	rs13306116	SNP	Kobayashi 2010	162	8	1	171	219	10	0	229	0.504	332	10	448	10	1.35	0.56–3.28	0.507	0.10
	rs5751876	SNP	Kobayashi 2010	35	85	51	171	70	114	45	229	0.018	155	187	254	204	1.50	1.13–1.99	0.005	0.81
	rs34923252	SNP	Kobayashi 2010	144	26	1	171	181	43	5	229	0.257	314	28	405	53	0.6 8	0.42–1.10	0.116	0.35
AGT	rs699	SNP	Sery 2001	32	46	11	89	41	58	26	125	0.274	110	68	140	110	0.79	0.53–1.17	0.23	0.22
AKT1	rs3730358	SNP	Bousman 2010	86	26	5	117	50	24	2	76	0.32	198	36	124	28	0.81	0.47–1.39	0.433	0.12
			Ikeda 2006	136	43	3	182	364	68	5	437	0.048	315	49	796	78	1.59	1.09–2.32	0.017	0.67
	rs3803300	SNP	Ikeda 2006	63	91	28	182	124	234	79	437	0.286	217	147	482	392	0.83	0.65–1.07	0.149	0.30
	rs1130214	SNP	Ikeda 2006	128	51	3	182	315	108	14	437	0.42	307	57	738	136	1.01	0.72–1.41	0.965	0.05
	rs2498799	SNP	Ikeda 2006	40	98	44	182	121	211	105	437	0.3	178	186	453	421	1.12	0.88–1.44	0.348	0.16
	rs2494732	SNP	Ikeda 2006	86	79	17	182	212	192	33	437	0.756	251	113	616	258	1.07	0.82–1.40	0.594	0.08
	rs2498804	SNP	Ikeda 2006	63	92	27	182	142	206	89	437	0.275	218	146	490	384	0.85	0.67–1.10	0.215	0.23
	rs2494743	SNP	Ghafouri-Fard 2020	120	82	18	220	109	85	12	206	0.516	322	118	303	109	1.02	0.75–1.38	0.905	0.05
	rs2498794	SNP	Ghafouri-Fard 2020	41	120	59	220	50	112	44	206	0.235	202	238	212	200	1.25	0.95–1.64	0.105	0.36
ANRIL	rs1333045	SNP	Namvar 2020	48	94	63	205	73	129	48	250	0.016	190	220	275	225	1.42	1.09–1.84	0.009	0.74
	rs1333048	SNP	Namvar 2020	41	89	75	205	98	117	35	250	<0.001	171	239	313	187	2.34	1.79–3.06	<0.0001	0.99
ARRB2	rs4790694	SNP	Bousman 2010	84	31	2	117	48	25	3	76	0.36	199	35	121	31	0.69	0.40–1.17	0.166	0.29
			Ikeda 2007	138	39	0	177	470	74	2	546	0.02	315	39	1014	78	1.61	1.07–2.41	0.02	0.64
	rs1045280	SNP	Ikeda 2007	117	54	6	177	417	122	7	546	0.012	288	66	956	136	1.61	1.17–2.22	0.004	0.83
	rs2036657	SNP	Ikeda 2007	129	44	4	177	436	103	7	546	0.133	302	52	975	117	1.43	1.01–2.04	0.043	0.52
BDNF	rs6265	SNP	Bousman 2010	86	29	2	117	45	29	2	76	0.12	201	33	119	33	0.59	0.35–1.01	0.052	0.49
			Cheng 2005	31	57	15	103	23	68	31	122	0.046	119	87	114	130	0.64	0.44–0.93	0.019	0.65
			Iamjan 2015	38	42	20	100	23	54	25	102	0.057	118	82	100	104	0.67	0.45–0.99	0.044	0.52
			Itoh 2005	70	96	23	189	70	107	25	202	0.884	236	142	247	15 7	0.95	0.71–1.26	0.71	0.07
			Sim 2010	56	99	31	186	36	84	34	154	0.256	211	161	156	152	0.78	0.58–1.06	0.114	0.36
			Su 2014	64	96	40	200	51	118	50	219	0.136	224	176	220	218	0.79	0.60–1.04	0.095	0.39
			Su 2015b	62	93	39	194	94	189	90	373	0.2	217	171	377	36 9	0.81	0.63–1.03	0.085	0.40
	132C > T	SNP	Itoh 2005	175	14	0	189	183	19	0	202	0.477	364	14	385	19	0.78	0.39–1.58	0.487	0.11
	rs2030324	SNP	Su 2015a	52	97	51	200	67	102	48	217	0.492	201	199	236	198	1.18	0.90–1.55	0.233	0.22
	rs16917204	SNP	Su 2015a	64	97	39	200	53	119	47	219	0.206	225	175	225	213	0.82	0.63–1.08	0.157	0.29
	rs16917234	SNP	Su 2015a	132	59	8	199	144	66	8	218	0.976	323	75	354	82	1.00	0.71–1.42	0.989	0.05
CHRNA4	rs755203	SNP	Kishi 2008b	74	77	34	185	259	358	130	747	0.284	225	145	876	618	0.91	0.72–1.53	0.446	0.12
	rs2273506	SNP	Kishi 2008b	138	40	7	185	564	176	7	747	0.016	316	54	1304	190	1.17	0.85–1.63	0.338	0.16
	rs2273504	SNP	Kishi 2008b	56	84	45	185	232	351	164	747	0.786	196	174	815	679	1.07	0.85–1.34	0.585	0.09
	rs1044396	SNP	Kishi 2008b	107	61	17	185	393	297	57	747	0.225	275	95	1083	411	0.91	0.70–1.18	0.478	0.11
	rs1044397	SNP	Kishi 2008b	79	78	28	185	273	349	125	747	0.303	236	134	895	599	0.85	0.67–1.07	0.172	0.28
	rs2236196	SNP	Kishi 2008b	137	44	4	185	563	177	7	747	0.383	318	52	1303	191	1.12	0.80–1.55	0.516	0.10
	rs4522666	SNP	Kishi 2008b	59	94	32	185	247	357	143	747	0.737	212	158	851	643	0.99	0.78–1.24	0.907	0.05
CHRNB2	rs4845652	SNP	Kishi 2008b	141	48	2	191	607	132	10	749	0.062	330	52	1346	152	1.40	1.00–1.96	0.052	0.50
	rs2072658	SNP	Kishi 2008b	134	53	4	191	466	256	25	747	0.124	321	61	1188	306	0.74	0.55–1.00	0.047	0.51
	rs2072659	SNP	Kishi 2008b	128	55	8	191	559	179	13	751	0.034	311	71	1297	205	1.44	1.07–1.94	0.015	0.68
	rs2072660	SNP	Kishi 2008b	97	79	15	191	440	268	45	753	0.15	273	109	1148	358	1.28	1.00–1.65	0.054	0.48
	rs3811450	SNP	Kishi 2008b	130	59	2	191	580	163	8	751	0.028	319	63	1323	179	1.46	1.07–1.99	0.017	0.67
COMT	rs4680	SNP	Bousman 2010	23	65	29	117	18	45	13	76	0.43	111	123	81	71	1.26	0.84–1.90	0.261	0.20
			Jugurnauth 2011	262	166	46	474	179	170	37	386	0.021	690	258	528	244	0.81	0.66–0.99	0.046	0.51
			Li 2004	228	150	32	410	181	172	37	390	0.034	606	214	534	246	0.77	0.62–0.95	0.016	0.69
			Suzuki 2006	57	65	21	143	94	85	21	200	0.313	179	107	273	127	1.28	0.93–1.77	0.123	0.34
	rs165599	SNP	Jugurnauth 2011	93	194	92	379	109	210	109	428	0.833	380	378	428	428	0.99	0.82–1.21	0.958	0.05
CSNK1E	rs135745	SNP	Kotaka 2008	126	78	11	215	151	100	23	274	0.346	330	100	402	146	0.83	0.62–1.12	0.226	0.18
CYP2D6	N/A	SNP	Otani 2008				202				336		151	51	218	11 8	0.6 2	0.42–0.92	0.017	0.67
DRD1	−48A > G	SNP	Liu 2006	252	106	5	363	302	110	13	425	0.194	610	116	714	136	1.00	0.76–1.31	0.991	0.05
DRD2	rs1800497	RFLP	Chen 2004	141	209	66	416	137	228	70	435	0.749	491	341	502	368	0.95	0.78–1.15	0.583	0.08
			Han 2008	8	26	3	37	19	18	3	40	0.054	42	32	56	24	1.78	0.92–3.45	0.088	0.40
			Sery 2001	57	34	2	93	75	53	4	132	0.769	148	38	203	61	0.85	0.54–1.35	0.5	0.10
			Tsai 2002	48	52	16	116	39	52	21	112	0.464	148	84	130	94	0.78	0.54–1.14	0.208	0.24
			Ujike 2009	83	84	35	202	92	114	37	243	0.523	250	154	298	188	0.98	0.74–1.28	0.863	0.05
	141 Ins/Del	Del	Ujike 2009	143	51	8	202	181	60	2	243	0.08	337	67	422	64	1.31	0.90–1.90	0.152	0.30
	rs1801028	SNP	Ujike 2009	185	17	0	202	231	12	0	243	0.139	387	17	474	12	1.74	0.82–3.68	0.146	0.31
DRD3	rs6280	SNP	Chen 2004	206	172	34	412	217	206	44	467	0.552	584	240	605	259	0.96	0.78–1.18	0.702	0.07
			Ujike 2009	100	87	15	202	126	95	22	243	0.639	287	117	347	139	1.02	0.76–1.36	0.906	0.05
DRD4	exon 3	VNTR	Chen 2004				393				414		751	35	780	48	0.76	0.48–1.18	0.222	0.23
			Li 2004	196	179	40	415	220	161	34	415	0.244	571	259	601	229	1.19	0.96–1.47	0.106	0.37
			Tsai 2002				116				112		215	17	212	12	1.40	0.65–3.00	0.389	0.14
			Ujike 2009				202				243		191	11	229	14	0.94	0.42–2.12	0.885	0.05
	rs1800955	SNP	Ujike 2009	61	106	35	202	66	134	43	243	0.775	228	176	266	22 0	0.93	0.72–1.22	0.611	0.08
DTNBP1	rs3213207	SNP	Kishimoto 2008	175	22	0	197	239	4	0	243	<0.001	372	22	482	4	7.13	2.44–20.86	0.0001	0.98
			Sim 2014	237	14	3	254	283	17	1	301	0.499	488	20	583	19	1.26	0.66–2.38	0.481	0.11
	rs2619539	SNP	Kishimoto 2008	78	94	18	190	118	107	15	240	0.173	250	130	343	137	1.30	0.97–1.74	0.074	0.43
	rs2619538	SNP	Kishimoto 2008	182	15	0	197	225	7	0	232	0.031	379	15	457	7	2.58	1.04–6.40	0.034	0.56
FAAH	rs324420	SNP	Morita 2005a	105	43	5	153	139	58	3	200	0.541	253	53	336	64	1.10	0.74–1.64	0.64	0.08
			Sim 2013	100	101	31	232	149	82	10	241	<0.001	301	163	380	102	2.02	1.51–2.70	<0.0001	0.99
			Zhang 2020	227	186	17	430	426	192	13	631	<0.001	640	220	1044	218	1.65	1.33–2.03	<0.001	0.99
GABBR1	rs2076483	SNP	Zhao 2018	492	260	39	791	258	171	18	447	0.147	1244	338	687	207	0.90	0.74–1.10	0.302	0.18
	rs29221	SNP	Zhao 2018	373	347	71	791	212	189	44	445	0.82	1093	489	613	277	0.99	0.83–1.18	0.912	0.05
	rs715044	SNP	Zhao 2018	668	118	5	791	353	90	5	448	0.039	1454	128	796	100	0.70	0.53–0.92	0.011	0.72
GABRA1	rs2279020	SNP	Lin 2003	83	117	46	246	131	201	100	432	0.358	283	209	463	401	0.85	0.68–1.07	0.162	0.29
GABRA6	rs4480562	SNP	Lin 2003	130	87	30	247	205	177	53	435	0.335	347	147	587	283	0.88	0.69–1.12	0.29	0.19
GABRB2	rs2229944	SNP	Lin 2003	230	17	1	248	407	29	1	437	0.915	477	19	843	31	1.08	0.61–1.94	0.788	0.06
GABRG2	rs4480617	SNP	Lin 2003	222	25	1	248	410	26	1	437	0.128	469	27	846	28	1.74	1.01–2.99	0.042	0.53
	rs211014	SNP	Lin 2003	73	119	54	246	119	200	117	436	0.364	265	227	438	434	0.86	0.69–1.08	0.197	0.25
	rs211013	SNP	Lin 2003	110	118	20	248	217	175	46	438	0.133	338	158	609	267	1.07	0.84–1.35	0.597	0.08
	315C > T	SNP	Nishiyama 2005	87	75	16	178	157	113	18	288	0.363	249	107	427	149	1.23	0.92–1.65	0.164	0.29
	1128 + 99 > 4A	SNP	Nishiyama 2005	56	79	43	178	80	128	80	288	0.589	191	165	288	288	0.86	0.66–1.13	0.278	0.19
GAD1	rs769404	SNP	Veerasakul 2017	25	53	22	100	27	49	26	102	0.761	103	97	103	101	0.96	0.65–1.42	0.839	0.05
	rs701492	SNP	Veerasakul 2017	54	38	8	100	51	43	8	102	0.829	146	54	145	59	0.91	0.59–1.40	0.667	0.07
GAD2	rs2236418	SNP	Veerasakul 2017	22	50	28	100	44	40	18	102	0.005	94	106	128	76	1.90	1.28–2.83	0.001	0.89
GCLM	rs2301022	SNP	Hashimoto 2008	126	76	16	218	125	90	18	233	0.669	328	108	340	126	0.89	0.66–1.20	0.437	0.12
			Kishi 2008a	108	64	13	185	450	311	56	817	0.675	280	90	1211	423	0.92	0.71–1.20	0.534	0.10
	−588C > T	SNP	Hashimoto 2008	150	62	6	218	172	59	2	233	0.214	362	74	403	63	1.31	0.91–1.88	0.149	0.30
	rs60197536	SNP	Kishi 2008a	117	62	6	185	572	228	19	819	0.206	296	74	1372	266	1.29	0.97–1.72	0.081	0.42
	rs718875	SNP	Kishi 2008a	116	63	6	185	571	221	27	819	0.153	295	75	1363	275	1.26	0.95–1.68	0.111	0.36
	rs3170633	SNP	Kishi 2008a	107	73	5	185	532	250	37	819	0.047	287	83	1314	324	1.17	0.89–1.54	0.252	0.21
GDNF	rs2973033	SNP	Yoshimura 2011	112	72	21	205	202	155	24	381	0.147	296	114	559	203	1.06	0.81–1.39	0.669	0.07
	rs10941370	SNP	Yoshimura 2011	78	94	32	204	145	183	53	381	0.82	250	158	473	289	1.03	0.81–1.32	0.789	0.06
	rs12518844	SNP	Yoshimura 2011	85	91	29	205	154	183	44	381	0.567	261	149	491	271	1.03	0.81–1.32	0.789	0.06
	rs2910702	SNP	Yoshimura 2011	85	101	20	206	164	172	46	382	0.549	271	141	500	264	0.99	0.77–1.27	0.909	0.05
	rs2910704	SNP	Yoshimura 2011	88	94	24	206	143	190	48	381	0.471	270	142	476	286	0.88	0.68–1.13	0.297	0.18
	rs884344	SNP	Yoshimura 2011	128	69	8	205	219	140	22	381	0.402	325	85	578	184	0.82	0.61–1.10	0.185	0.27
	rs1549250	SNP	Yoshimura 2011	95	90	20	205	198	158	25	381	0.246	280	130	554	208	1.24	0.95–1.61	0.112	0.35
	rs2973049	SNP	Yoshimura 2011	51	111	43	205	109	178	93	380	0.241	213	197	396	364	1.01	0.79–1.28	0.96	0.05
GHRL	rs35683	SNP	Yoon 2005	89	28	1	118	103	33	8	144	0.114	206	30	239	49	0.71	0.44–1.16	0.171	0.28
GPX1	rs1050450	SNP	Hashimoto 2008	189	29	0	218	207	23	3	233	0.134	407	29	437	29	1.07	0.63–1.83	0.793	0.06
GRIA1	rs1428920	SNP	Iamjan 2018	40	51	9	100	48	46	8	102	0.599	131	69	142	62	1.21	0.80–1.83	0.378	0.14
GRIA2	rs3813296	SNP	Iamjan 2018	23	37	40	100	15	50	37	102	0.155	83	117	80	124	0.91	0.61–1.35	0.64	0.07
GRIA3	rs989638	SNP	Iamjan 2018				100				102		91	9	93	9	1.02	0.39–2.69	0.965	0.05
	rs3761554	SNP	Iamjan 2018				100				102		64	36	72	30	1.35	0.75–2.44	0.318	0.17
	rs502434	SNP	Iamjan 2018				100				102		45	55	60	42	1.75	1.00–3.05	0.049	0.50
GRIN1	rs11146020	SNP	Chanasong 2013	56	35	9	100	65	33	5	103	0.4	147	53	163	43	1.37	0.86–2.17	0.182	0.26
	rs1126442	SNP	Chanasong 2013	62	36	2	100	54	46	3	103	0.381	160	40	154	52	0.74	0.46–1.18	0.207	0.25
GRM2	rs3821829	SNP	Tsunoka 2010	181	14	1	196	731	67	4	802	0.856	376	16	1529	75	0.87	0.50–1.51	0.613	0.08
	rs12487957	SNP	Tsunoka 2010	106	79	11	196	346	378	78	802	0.013	291	10 1	1070	534	0.70	0.54–0.89	0.004	0.82
	rs4687771	SNP	Tsunoka 2010	100	82	14	196	300	401	101	802	0.001	282	110	1001	603	0.65	0.51–0.83	<0.001	0.94
GSTA1	−69C > T	SNP	Hashimoto 2008	158	56	4	218	180	50	3	233	0.492	372	64	410	56	1.26	0.86–1.85	0.239	0.22
GSTM1	Del	Del	Koizumi 2004	82	63	12	157	88	97	15	200	0.269	227	87	273	127	0.82	0.60–1.14	0.242	0.22
GSTO1	rs4925	SNP	Hashimoto 2008	163	53	2	218	166	55	12	233	0.035	379	57	387	79	0.74	0.51–1.07	0.104	0.37
GSTP1	rs1695	SNP	Bousman 2010	15	58	44	117	9	29	38	76	0.22	88	146	47	105	0.74	0.48–1.15	0.178	0.27
		SNP	Hashimoto 2005	144	41	4	189	167	32	0	199	0.038	329	49	366	32	1.70	1.07–2.73	0.025	0.61
GSTT1	Deletion	Del	Hashimoto 2008				218				233		109	109	140	93	1.51	1.04–2.19	0.031	0.57
GSTT2	rs1622002	SNP	Hashimoto 2008	149	63	6	218	157	73	3	233	0.485	361	75	387	79	1.02	0.72–1.44	0.921	0.05
HTR2A	rs6311	SNP	Tsunoka 2010	58	96	42	196	262	374	166	802	0.708	212	180	898	706	1.08	0.87–1.35	0.497	0.10
	rs6313	SNP	Tsunoka 2010	52	95	49	196	220	386	196	802	0.965	199	193	826	778	1.03	0.83–1.28	0.795	0.06
HTR6	rs6693503	SNP	Kishi 2011c	124	59	14	197	257	70	10	337	0.002	307	87	584	90	1.84	1.33–2.55	0.0002	0.96
	rs1805054	SNP	Kishi 2011c	90	92	15	197	162	153	22	337	0.816	272	122	477	197	1.09	0.83–1.42	0.55	0.09
	rs4912138	SNP	Kishi 2011c	55	93	49	197	87	170	80	337	0.764	203	11	344	330	0.98	0.77–1.26	0.879	0.05
	rs3790757	SNP	Kishi 2011c	115	70	12	197	220	105	12	337	0.179	300	94	545	129	1.32	0.98–1.79	0.067	0.45
	rs9659997	SNP	Kishi 2011c	146	45	6	197	243	82	12	337	0.868	337	57	568	106	0.91	0.64–1.29	0.581	0.09
MAOA	T941G	SNP	Nakamura 2009				117				198		64	74	123	120	1.19	0.78–1.80	0.426	0.13
	30bp	VNTR	Nakamura 2009				115				197		82	52	146	95	0.97	0.63–1.50	0.907	0.05
NOS3	rs1800779	SNP	Okochi 2011	140	43	0	183	219	46	2	267	0.139	323	43	484	50	1.29	0.84–1.98	0.248	0.21
	rs2070744	SNP	Okochi 2011	140	42	1	183	219	47	1	267	0.358	322	44	485	49	1.35	0.88–2.08	0.168	0.28
	rs1799983	SNP	Okochi 2011	152	31	0	183	221	44	2	267	0.5	335	31	486	48	0.94	0.58–1.50	0.797	0.06
	rs3918188	SNP	Okochi 2011	107	62	14	183	140	110	17	267	0.286	276	90	390	144	0.88	0.65–1.20	0.425	0.13
	rs743507	SNP	Okochi 2011	130	49	4	183	183	78	6	267	0.849	309	57	444	90	0.91	0.63–1.31	0.61	0.08
	rs7830	SNP	Okochi 2011	49	91	43	183	85	121	61	267	0.496	189	177	291	243	1.12	0.86–1.46	0.399	0.13
NPY	rs16147	SNP	Okahisa 2009	96	99	24	219	119	128	41	288	0.537	291	147	366	210	0.88	0.68–1.14	0.339	0.16
NPY1R	rs7687423	SNP	Okahisa 2009	55	126	41	222	90	129	65	284	0.04	236	208	309	259	1.05	0.82–1.35	0.693	0.07
NQO1	rs1800566	SNP	Ohgake 2005	69	97	25	191	85	95	27	207	0.572	235	147	265	149	1.11	0.83–1.48	0.467	0.11
NQO2	N/A	Del	Ohgake 2005	117	59	15	191	123	74	10	207	0.333	293	89	320	94	1.03	0.74–1.44	0.843	0.05
NRG1	rs35753505	SNP	Okochi 2009	50	104	30	184	144	262	118	524	0.162	204	164	550	498	0.89	0.70–1.13	0.329	0.16
	rs241930	SNP	Okochi 2009	139	43	1	183	405	100	5	510	0.476	321	45	910	110	1.16	0.80–1.68	0.431	0.12
	rs6994992	SNP	Okochi 2009	48	102	30	180	146	280	107	533	0.53	198	162	572	494	0.95	0.75–1.20	0.659	0.07
	rs3924999	SNP	Okochi 2009	103	58	10	171	339	171	24	534	0.653	264	78	849	219	1.15	0.85–1.54	0.364	0.15
OPRD1	IVS1 + 18	SNP	Kobayashi 2006	140	27	3	170	220	40	0	260	0.097	307	33	480	40	1.29	0.80–2.09	0.3	0.18
	rs2234918	SNP	Kobayashi 2006	101	63	6	170	181	71	8	260	0.089	265	75	433	87	1.41	1.00–1.99	0.051	0.50
	rs4654327	SNP	Kobayashi 2006	109	54	7	170	190	62	8	260	0.143	272	68	442	78	1.42	0.99–2.03	0.056	0.48
OPRM1	rs2075572	SNP	Bousman 2010	20	57	40	117	23	26	27	76	0.05	97	137	72	80	1.27	0.84–1.92	0.252	0.21
			Ide 2006	73	43	12	128	154	72	6	232	0.012	189	67	380	84	1.60	1.11–2.31	0.011	0.72
	rs1799971	SNP	Ide 2004	50	56	25	131	67	99	47	213	0.434	156	106	233	193	0.82	0.60–1.12	0.213	0.23
			Ide 2006	49	54	25	128	67	99	47	213	0.435	152	104	233	193	0.83	0.60–1.13	0.233	0.22
	IVS1–4980G	SNP	Ide 2004	130	8	0	138	177	10	0	187	0.99	268	8	364	10	1.09	0.42–2.79	0.863	0.05
	rs599548	SNP	Ide 2006	96	28	4	128	135	41	3	179	0.696	220	36	311	47	1.08	0.68–1.73	0.739	0.06
	rs598682	SNP	Ide 2006	101	26	1	128	154	23	2	179	0.207	228	28	331	27	1.51	0.86–262	0.146	0.31
PAI-1	rs1799889	Del	Iwata 2004	87	80	17	184	115	138	35	288	0.251	254	114	368	208	0.79	0.60–1.05	0.105	0.37
PDYN	68-bp	VNTR	Nomura 2006a	85	48	10	143	149	59	1	209	0.001	218	68	357	61	1.83	1.24–2.68	0.002	0.87
PICK1	rs737622	SNP	Matsuzawa 2007	85	93	30	208	89	107	22	218	0.355	263	153	285	151	1.10	0.83–1.45	0.513	0.10
	rs3026682	SNP	Matsuzawa 2007	85	93	30	208	89	107	22	218	0.355	263	153	285	151	1.10	0.83–1.45	0.513	0.10
	rs11089858	SNP	Matsuzawa 2007	167	39	2	208	180	37	1	218	0.727	373	43	397	39	1.17	0.74–1.85	0.491	0.11
	rs713729	SNP	Matsuzawa 2007	166	37	5	208	150	63	5	218	0.025	369	47	363	73	0.63	0.43–0.94	0.022	0.62
	rs3952	SNP	Matsuzawa 2007	85	93	30	208	89	107	22	218	0.355	263	153	285	151	1.10	0.83–1.45	0.513	0.10
	rs2076369	SNP	Matsuzawa 2007	73	99	36	208	82	111	25	218	0.228	245	171	275	161	1.19	0.91–1.57	0.211	0.24
PIK4CA	rs2072513	SNP	Kanahara 2009	54	130	48	232	60	120	53	233	0.618	238	226	240	226	1.01	0.78–1.30	0.949	0.05
	rs165862	SNP	Kanahara 2009	54	126	52	232	59	117	57	233	0.676	234	230	235	231	1.00	0.77–1.29	1	0.05
	rs165793	SNP	Kanahara 2009	86	117	29	232	90	107	36	233	0.525	289	175	287	179	0.97	0.75–1.27	0.827	0.06
	rs165789	SNP	Kanahara 2009	55	127	50	232	60	118	55	233	0.676	237	227	238	228	1.00	0.77–1.29	1	0.05
PLAT	Alu	Del	Iwata 2004	48	96	41	185	77	146	65	288	0.967	192	178	300	276	1.01	0.78–1.31	0.954	0.05
PPP3CC	rs10108011	SNP	Kinoshita 2008	58	60	10	128	124	119	22	265	0.933	176	80	367	163	1.02	0.74–1.41	0.888	0.05
	rs2461491	SNP	Kinoshita 2008	34	68	26	128	82	130	53	265	0.654	136	120	294	236	1.10	0.82–1.48	0.536	0.09
	rs2461490	SNP	Kinoshita 2008	42	72	14	128	99	131	35	265	0.443	156	100	329	201	1.05	0.77–1.43	0.758	0.06
	rs2449340	SNP	Kinoshita 2008	44	69	15	128	112	122	31	265	0.293	157	99	346	184	1.19	0.87–1.61	0.279	0.19
	rs1116085	SNP	Kinoshita 2008	35	67	26	128	72	137	56	265	0.982	137	119	281	249	0.98	0.73–.132	0.896	0.05
PROKR2	rs17721321	SNP	Kishi 2010	170	29	0	199	292	43	2	337	0.47	369	29	627	47	1.05	0.65–.170	0.847	0.05
	rs6085086	SNP	Kishi 2010	77	97	25	199	185	121	31	337	0.001	251	147	491	183	1.57	1.21–2.05	0.001	0.92
	rs3746684	SNP	Kishi 2010	59	96	44	199	110	153	74	337	0.75	214	184	373	301	1.07	0.83–1.37	0.617	0.08
	rs3746682	SNP	Kishi 2010	91	84	24	199	121	155	61	337	0.042	266	13 2	397	277	0.71	0.55–0.92	0.01	0.73
	rs4815787	SNP	Kishi 2010	61	109	29	199	158	142	37	337	0.001	231	167	458	216	1.53	1.19–1.98	0.001	0.91
SIGMAR1	rs1799729	SNP	Inada 2004	70	61	12	143	90	78	13	181	0.921	201	85	258	104	1.05	0.75–1.48	0.783	0.06
	rs1800866	SNP	Inada 2004	61	65	17	143	86	83	14	183	0.396	187	99	255	111	1.22	0.87–.1.69	0.245	0.22
SIRT1	rs12778366	SNP	Kishi 2011a	146	47	4	197	249	64	5	318	0.546	339	55	562	74	1.23	0.85–1.79	0.273	0.19
	rs2273773	SNP	Kishi 2011a	98	83	16	197	155	131	32	318	0.762	279	115	441	195	0.93	0.71–1.23	0.617	0.08
	rs4746720	SNP	Kishi 2011a	66	94	37	197	113	155	50	318	0.656	226	168	381	255	1.11	0.86–1.43	0.42	0.13
	rs10997875	SNP	Kishi 2011a	144	46	7	197	199	111	8	318	0.02	334	60	509	127	0.72	0.51–1.01	0.055	0.49
SLC22A3	rs509707	SNP	Aoyama 2006	92	93	25	210	226	175	41	442	0.192	277	143	627	257	1.26	0.98–1.66	0.069	0.44
	rs4709426	SNP	Aoyama 2006	59	108	44	211	165	199	78	442	0.061	226	196	529	355	1.29	1.02–1.63	0.031	0.58
	rs7745775	SNP	Aoyama 2006	116	84	7	207	246	146	33	425	0.055	316	98	638	212	0.93	0.71–1.23	0.622	0.08
	rs3106164	SNP	Aoyama 2006	118	74	15	207	205	198	37	440	0.046	310	104	608	272	0.75	0.58–0.98	0.032	0.56
	rs3088442	SNP	Aoyama 2006	50	92	68	210	119	208	111	438	0.167	192	228	446	430	1.23	0.98–1.56	0.08	0.42
SLC6A3	3′ VNTR	VNTR	Hong 2003	84	16	1	101	92	17	0	109	0.58	184	18	201	17	1.16	0.58–2.31	0.68	0.07
			Liu 2004	107	38	1	146	56	16	0	72	0.636	252	40	128	16	1.27	0.69–2.36	0.448	0.12
			Ujike 2003	105	18	1	124	138	20	2	160	0.834	228	20	296	24	1.08	0.58–2.01	0.803	0.05
	242C>T	SNP	Ujike 2003	101	2	0	103	104	2	0	106	0.977	204	2	210	2	1.03	0.14–7.38	0.977	0.07
	1342A > G	SNP	Ujike 2003	101	22	1	124	129	26	4	159	0.54	224	24	284	34	0.89	0.52–1.55	0.693	0.08
	2319G > A	SNP	Ujike 2003	69	48	7	124	92	57	8	157	0.882	186	62	241	73	1.10	0.75–1.62	0.629	0.08
SLC6A4	5-HTTLPR	LPR	Ezaki 2008	107	56	3	166	116	65	16	197	0.026	270	62	297	97	0.70	0.49–1.01	0.054	0.49
			Hong 2003	59	33	10	102	61	35	16	112	0.603	151	53	157	67	0.82	0.54–1.26	0.366	0.15
			Payer 2012	14	22	17	53	11	20	16	47	0.939	50	56	42	52	0.90	0.52–1.58	0.724	0.06
	intron 2	VNTR	Hong 2003	86	16	1	103	100	12	0	112	0.324	188	18	212	12	1.69	0.79–3.60	0.169	0.28
	5-HTTVNTR	VNTR	Payer 2012	23	20	10	53	23	18	6	47	0.688	66	40	64	30	1.29	0.72–2.32	0.389	0.14
SLC6A9	rs2486001	SNP	Morita 2008	106	82	16	204	139	63	8	210	0.009	294	114	341	79	1.67	1.21–2.32	0.002	0.87
	rs2248829	SNP	Morita 2008	119	75	10	204	110	76	24	210	0.049	313	95	296	124	0.72	0.53–0.99	0.042	0.53
	rs2248632	SNP	Morita 2008	122	70	12	204	113	81	16	210	0.442	314	94	307	113	0.81	0.59–1.12	0.199	0.25
SNCA	rs1372520	SNP	Kobayashi 2004	142	27	1	170	141	18	2	161	0.388	311	29	300	22	1.27	0.72–2.26	0.413	0.13
	rs3756063	SNP	Kobayashi 2004	140	29	1	170	141	18	2	161	0.263	309	31	300	22	1.37	0.77–2.42	0.279	0.19
	rs2870027	SNP	Kobayashi 2004	60	88	22	170	55	75	31	161	0.281	208	132	185	137	0.86	0.63–1.17	0.33	0.16
	rs3756059	SNP	Kobayashi 2004	141	28	1	170	141	18	2	161	0.322	310	30	300	22	1.32	0.74–2.34	0.341	0.16
	T10A	Var	Kobayashi 2004				170				161	0.677
SOD2	rs2758357	SNP	Nakamura 2006 (Japanese)	75	36	2	113	130	47	9	186	0.22	186	40	307	65	1.02	0.66–1.57	0.944	0.05
			Nakamura 2006 (Han)	79	53	2	134	126	70	4	200	0.677	211	57	322	78	1.12	0.76–1.64	0.577	0.09
	rs4880	SNP	Nakamura 2006 (Japanese)	84	30	2	116	152	33	4	189	0.211	198	34	337	41	1.41	0.87–2.30	0.164	0.28
			Nakamura 2006 (Han)	86	43	2	131	150	47	5	202	0.145	215	47	347	57	1.33	0.87–2.03	0.183	0.27
	rs2855116 &rs2758332	SNP	Nakamura 2006 (Japanese)	91	24	1	116	153	33	3	189	0.689	206	26	339	39	1.10	0.65–1.86	0.73	0.06
			Nakamura 2006 (Han)	90	41	2	133	151	48	4	203	0.338	221	45	350	56	1.27	0.83–1.95	0.268	0.20
	rs2758330	SNP	Nakamura 2006 (Japanese)	27	53	35	115	54	83	48	185	0.497	107	123	191	179	1.23	0.88–1.71	0.224	0.23
			Nakamura 2006 (Han)	42	63	29	134	54	105	41	200	0.59	147	121	213	187	0.94	0.69–1.28	0.684	0.07
	rs2842980	SNP	Nakamura 2006 (Japanese)	27	53	35	115	53	84	50	187	0.606	107	123	190	184	1.19	0.85–1.65	0.307	0.18
			Nakamura 2006 (Han)	41	62	31	134	55	104	41	200	0.59	144	124	214	186	0.99	0.73–1.35	0.953	0.05
TAAR1	rs8192620	SNP	Loftis 2019				75				31		37	38	17	14	1.25	0.54–2.89	0.606	0.08
TNFA	308G > A	SNP	Nomura 2006b	175	9	0	184	207	7	0	214	0.412	359	9	421	7	1.51	0.56–4.09	0.417	0.13
	857C > T	SNP	Nomura 2006b	123	52	8	183	134	68	17	219	0.267	298	68	336	102	0.75	0.53–1.06	0.103	0.37
TNFAR1	36A > G	SNP	Nomura 2006b	131	44	4	179	151	59	6	216	0.763	306	52	361	71	0.86	0.59–1.28	0.461	0.11
TPH2	rs11178998	SNP	Kobayashi 2011a	130	29	3	162	197	46	0	243	0.102	289	35	440	46	1.16	0.73–1.84	0.534	0.10
	Exon2+C18A	SNP	Kobayashi 2011a	146	16	0	162	222	21	0	243	0.673	308	16	465	21	1.15	0.59–2.24	0.68	0.07
	rs10879348	SNP	Kobayashi 2011a	136	26	0	162	206	37	0	243	0.823	298	26	449	37	1.06	0.63–1.79	0.83	0.06
	rs17110566	SNP	Kobayashi 2011a	123	35	4	162	173	64	6	243	0.552	281	43	410	76	0.83	0.55–1.24	0.351	0.16
	rs4760816	SNP	Kobayashi 2011a	49	85	28	162	65	121	57	243	0.313	183	141	251	235	0.82	0.62–1.09	0.176	0.27
	rs4290270	SNP	Kobayashi 2011a	53	80	29	162	79	115	49	243	0.841	186	138	273	213	0.95	0.72–1.26	0.728	0.06
	Exon11+(C3)500(C2)	SNP	Kobayashi 2011a	159	3	0	162	242	1	0	243	0.151	321	3	485	1	4.53	0.47–43.77	0.152	0.30
	rs17110747	SNP	Kobayashi 2011a	92	63	7	162	136	95	12	243	0.955	247	77	367	119	0.96	0.69–1.37	0.815	0.06
	rs41317114	SNP	Kobayashi 2011a	119	38	5	162	187	50	6	243	0.717	276	48	424	62	1.19	0.79–1.79	0.402	0.13
TSNAX	rs1630250	SNP	Kishi 2011b	70	100	45	215	84	165	69	318	0.293	240	190	333	303	0.87	0.68–1.11	0.267	0.20
	rs766288	SNP	Kishi 2011b	93	91	31	215	128	141	49	318	0.786	277	153	397	239	0.92	0.71–1.18	0.507	0.10
	rs6662926	SNP	Kishi 2011b	58	105	52	215	90	164	64	318	0.537	221	209	344	292	1.11	0.87–1.42	0.388	0.14
TSPO/PBR	rs6971	SNP	Rathitharan 2020				11				26		6	5	18	8	1.88	0.44–7.99	0.392	0.14
XBP1	−116C > G	SNP	Morita 2005b	72	64	17	153	80	88	32	200	0.274	208	98	248	152	0.77	0.56–1.05	0.1	0.38

CI = confidence interval; Del = deletion; LPR = linked polymorphic region; M = major allele; m = minor allele; OR = odds ration; RFLP = restriction fragment length polymorphism; SNP = single nucleotide polymorphism; Var = sequence variation; VNTR = variable number tandem repeat. Values in bold are statistically significant (p < 0.05).

3.2.1. Genotype and allele frequencies

Forty markers in 29 different genes displayed significant genotypic or allelic association with meth use disorder. Full names of genes are available in Table S1. Characteristics of each significant marker (including location) are given in Table 3. Of these markers, 11 were located in coding regions of a gene. Five of them were missense mutation leading to a change in amino acid sequence (BDNF rs6265; COMT rs4680; FAAH rs324420; GSTO1 rs4925; and GSTP1 rs1695), whereas the other six were synonymous mutations (ADORA2A rs5751876; ARRB2 rs1045280; CHRNA4 rs2273506; GRIA rs502434; PROKR2 rs3746682; and SLC6A9 rs2248829). In addition, four markers were located in the 3′ untranslated region (UTR; CHRNB2 rs2072659 and rs3811450; GCLM rs3170633; and PROKR2 rs3746682), three markers in the 5′ UTR (ADORA2A rs13306114; GABRG2 rs4480617; and GAD2 rs2236418), and two in the promoter region of their respective gene (PDYN 68-bp repeat; and SLC6A4 LPR). Ten markers were located within an intron (AKT1 rs3730358; ANRIL rs1333045; DTNBP1 rs3213207; GABBR1 rs715044; OPRM1 rs2075572; PICK1 rs713729; PROKR2 rs6085086; SLC22A3 rs3106164 and rs4709426; and SLC6A9 rs2486001). One of the markers investigated was a deletion (GSTT1). The last nine markers were not directly located within a gene. Six of them were downstream (ANRIL rs1333048; ARRB2 rs4790694 and rs2036657; GRM2 rs4687771 and rs12487957; and SIRT1 rs10997875) and three of them upstream (CHRNB2 rs2072658; DTNBP1 rs2619538; and HTR6 rs6693503) of the targeted gene. The effect of most variants on their gene was unknown, and therefore any significant association with meth use disorder should be taken with caution.

Table 3.

Characteristics of markers significantly associated with meth use disorder.

Gene	Marker	Location	Alleles (M > m)	Variant effect
ADORA2A	rs5751876	Exon 2	C > T	Synonymous variant (Tyr > Tyr).
	rs13306114	5′ UTR	C > T	No known effect.
AKT1	rs3730358	Intron variant	C > A	No known effect.
	rs3803300	8kb upstream of AKT1	C > T	No known effect.
	rs1130214	Intron variant	C > A	No known effect.
	rs2498799	Exon 8	C > T	Synonymous variant (Glu > Glu).
ANRIL	rs1333048	5 kb downstream of ANRIL	A > C	No known effect.
	rs1333045	Intron variant	T > C	No known effect.
ARRB2	rs1045280	Exon 11	G > A	Synonymous variant (Ser > Ser).
	rs4790694	1.5 kb downstream of ARRB2	A > C	No known effect.
	rs2036657	500 bases downstream of ARRB2	G > A	No known effect.
BDNF	rs6265	Exon 9	G > A	Missense variant (Val > Met). Met is associated with reduced secretion of BDNF protein (Egan et al., 2003).
	rs16917204	Intron variant	G > C	No known effect.
	rs16917234	Intron variant	T > C	No known effect.
	rs2030324	Intron variant	A > G	No known effect.
CHRNA4	rs2273506	Exon 2	G > A	Synonymous variant (Leu > Leu).
	rs1044397	Exon 5	G > C	Synonymous variant (Ala > Ala).
CHRNB2	rs2072659	3′ UTR	C > G	No known effect.
	rs3811450	3′ UTR	C > G	No known effect.
	rs2072658	2kb upstream of CHRNB2	G > A	A-allele associated with reduced gene expression in vitro (Hoft et al., 2011).
COMT	rs4680	Exon 4	G > A	Missense variant (Val > Met). Val associated with greater protein abundance and stability, and greater enzyme activity (Tunbridge et al., 2019).
	rs165599	3′ UTR	A > G	No consistent effect reported across studies (Tunbridge et al., 2019)
CYP2D6	N/A	Combination of alleles	N/A	A range of allele combinations are association with increased, reduced, or no enzymatic activity (Bertilsson et al., 2002).
DRD2	rs1800497	downstream of DRD2	A2 > A1	A2 allele is associated with increased D2R binding (Tunbridge et al., 2019).
DRD4	48bp-repeat	Exon 3	2–11 repeats	No consistent effect reported across studies (Tunbridge et al., 2019)
DTNBP1	rs3213207	Intron variant	A > G	No known effect.
	rs2619538	2kb upstream of DTNBP1	A > T	No known effect.
	rs2619539	Intron variant	C > A	No known effect.
FAAH	rs324420	Exon 3	C > A	Missense variant (Pro > Thr). Thr associated with reduced enzymatic catalytic activity and reduced protein stability (Chiang et al., 2004; Dincheva et al., 2015).
GABRA1	rs2279020	Intron variant	G > A	No known effect.
GABBR1	rs715044	Intron variant	G > T	No known effect.
	rs2076483	Intron variant	A > G	No known effect.
	rs29221	Intron variant	C > G	No known effect.
GABRG2	rs4480617	5′ UTR	C > T	No known effect.
	315C > T	Exon 3	C > T	Synonymous variant (Asn > Asn).
	1128 + 99C > A	Intron variant	C > A	No known effect.
GAD1	rs769404	Exon 2	T > C	Synonymous variant (His > His).
	rs701492	Intron variant	C > T	No known effect.
GAD2	rs2236418	5′ UTR	A > G	G-allele associated with 6-fold increase in transcription levels (Boutin et al., 2003).
GCLM	rs3170633	3′ UTR	C > T	No known effect.
GDNF	rs2910702	Intron variant	T > C	No known effect.
	rs2910704	Intron variant	C > G	No known effect.
	rs884344	Intron variant	A > C	No known effect.
GRIA3	rs502434	Exon 10	T > C	Synonymous variant (Asn > Asn).
GRM2	rs3821829	4kb upstream of GRM2	C > T	No known effect.
	rs4687771	2.5kb downstream of GRM2	T > A	No known effect.
	rs12487957	3.5 kb downstream of GRM2	T > C	No known effect.
GSTO1	rs4925	Exon 4	C > A	Missense variant (Ala > Asp). Asp associated with reduced catalytic enzymatic activity (Menon and Board, 2013).
GSTP1	rs1695	Exon 5	A > G	Missense variant (Ile > Val). Val is associated with reduced catalytic enzymatic activity and reduced protein stability (Johansson et al., 1998).
GSTT1	Deletion	N/A	N/A	Deletion of the GSTT1.
HTR6	rs6693503	4kb upstream of HTR6	A > G	No known effect.
	rs1805054	Exon 1	C > T	Synonymous variant (Tyr > Tyr).
	rs4912138	Intron variant	G > A	No known effect.
NPY1R	rs7687423	Intron variant	A > G	No known effect.
OPRM1	rs2075572	Intron variant	G > C	No known effect.
PDYN	68-bp repeat	5′ UTR	1–5 repeats	Three or four repeats associated with increased PDYN expression compared to one or two copies (Zimprich et al., 2000).
PICK1	rs713729	Intron variant	T > A	No known effect.
	rs737622	5′ UTR	C > G	G-allele associated with increased gene expression (Matsuzawa et al., 2007).
	rs3026682	5′ UTR	G > A	A-allele associated with increased gene expression (Matsuzawa et al., 2007).
	rs11089858	Intron variant	G > A	No known effect.
	rs3952	Intron variant	A > G	No known effect.
	rs2076369	Intron variant	T > G	No known effect.
PROKR2	rs6085086	Intron variant	G > A	No known effect.
	rs4815787	3′ UTR	G > A	No known effect.
	rs3746682	Exon 2	C > A	Synonymous variant (Thr > Thr).
	rs3746684	Exon 2	G > A	Synonymous variant (Leu > Leu).
	rs17721321	Intron variant	G > A	No known effect.
SIRT1	rs10997875	1.5 kb downstream of SIRT1	C > T	No known effect.
SLC22A3	rs3106164	Intron variant	C > T	No known effect.
	rs4709426	Intron variant	C > G	No known effect.
	rs509707	Intron variant	C > T	No known effect.
	rs7745775	Intron variant	T > G	No known effect.
SLC6A3	3′ VNTR	3′ UTR	9–10 repeats	Some evidence of association of 10 repeats with dopamine transporter binding (Tunbridge et al., 2019).
SLC6A4	LPR	Promoter region	Long > Short	Short variant is associated with reduced gene expression, transporter levels and serotonin uptake in vitro (Greenberg et al., 1999; Lesch et al., 1996).
SLC6A9	rs2486001	Intron variant	T > C	No known effect.
	rs2248829	Exon 7	G > A	Synonymous variant (Asp > Asp).
SNCA	rs1372520	Intron variant	G > A	No known effect.
	rs3756063	Intron variant	C > G	No known effect.
	rs3756059	Intron variant	G > A	No known effect.
SOD2	rs2758357	Intron variant	C > T	No known effect.
	rs4880	Exon 2	A > G	Missense variant (Val > Ala). Val is associated with reduced enzymatic activity and increased oxidative stress (Sutton et al., 2005).
	rs2855116	Intron variant	A > C	No known effect.
	rs2758332	Intron variant	C > A	No known effect.
	rs2758330	Intron variant	G > T	No known effect.
	rs2842980	3′ UTR	A > T	No known effect.
TSNAX	rs1630250	2kb upstream of TSNAX	G > C	No known effect.
XBP1	−116C > G	Promoter region	C > G	G-allele associated with reducer transcriptional activity (Kakiuchi et al., 2003).

Ala = alanine; Asn = asparagine; Asp = aspartic acid; DRD2 = dopamine D2 receptor; Glu = glutamic acid; His = histidine; Ile = isoleucine; kb = kilobase; Leu = leucine; LPR = linked polymorphic region; M = major allele; m = minor allele; Met = methionine; Pro = proline; Ser = serine; Thr = threonine; Try = tyrosine; UTR = untranslated region; Val = valine; VNTR = variable number tandem repeat.

Minor alleles for markers in 14 genes (ADORA2A, AKT1, ANRIL, ARRB2, DTNBP1, FAAH, GABRG2, GAD2, GRIA3, GSTP1, GSTT1, HTR6, OPRM1, PDYN) were found to confer a significant risk for meth use disorder (ORs > 1). Of these genes, three had more than one significant marker (ANRIL rs1333045 and rs1333048; ARRB2 rs1045280, rs2036657 and rs4790694; and DTNBP1 rs3213207 and rs2619538), and one was explored in more than one study (FAAH rs324420). Minor alleles for markers in six genes (BDNF, COMT, CYP2D6, GABBR1, GRM2, and PICK1) were found to be protective (ORs < 1). One of these genes had more than one significant marker (GRM2 rs12487957 and rs4687771), and two genes were explored in more than one study (COMT rs4680; and BDNF rs6265). It should be noted that four genes (CHRNB2, PROKR2, SLC6A9, SCL22A3) had significant markers both conferring risk and protection. For CHRNB2, rs2072658 was protective (OR < 1), whereas rs2072659 and rs3811450 conferred a risk (ORs > 1). For PROKR2, rs3746682 was protective, whereas rs6085086 and rs4815787 were risk markers. For SLC6A9, rs2248829 was protective and rs2486001 conferred a risk. Lastly, for SLC22A3, rs3106164 was protective, while rs4709426 was a risk marker. The genotype of markers in five genes (CHRNA4, GCLM, GSTO1, SLC6A4, SIRT1) was significantly different between cases and controls, but failed to reach allelic statistical significance.

Power calculations revealed that a large majority of associations (~94 %) were underpowered (1 − β < 0.8) to detect appropriate genotypic or allelic associations. Power ranged from 0.05 to 0.99. For associations that were significant (ADORA2A rs5751876; ANRIL rs1333048; ARRB2 rs1045280; DTNBP1 rs3213207; FAAH rs324420; GAD2 rs2236418; GRM2 rs12487957 and rs4687771; HTR6 rs6693503; PDYN 68-bp repeat; PROKR2 rs4815787 and rs6085086; SLC6A9 rs2486001), power ranged from 0.16 to 0.99. Out of the 40 significant markers, a total of 14 were adequately powered (1 − β ≥ 0.80).

3.2.2. Haplotype analyses

Haplotype analyses revealed significant haplotypic associations between markers in 14 genes (AKT1, ANRIL, ARRB2, BDNF, COMT, DTNBP1, GABBR1, GABRG2, GAD1, GRM2, HTR6, PICK1, PROKR2, SLC6A9, SOD2) and meth use disorder (Table 4). For BDNF and GABBR1, all markers included in haplotype analyses were located in introns, whereas markers for the other genes were a combination of intronic, synonymous and up/downstream variants, with the exception of COMT which included the functional variant rs4680 (Table 3). None of the markers assigned to GRM2 were directly located within the gene, but rather at the 5′ and 3′ ends. GAD1 and SOD2 were the only two genes that did not show genotypic or allelic significance when markers where investigated individually.

Table 4.

Summary of haplotype studies.

Gene	Ref	Haplotype markers						Frequency		p-value
								Meth	Control
AKT1	Ghafouri-Fard 2020	rs2494743	rs2498794
		T	T					0.4	0.43	0.93
		T	C					0.33	0.3	0.94
		C	C					0.21	0.18	1
		T	T					0.06	0.08	1
	Ikeda 2006	rs3803300	rs1130214	rs3730358	rs2498799
		A	G	C	G			0.28	0.2	0.023
		A	G	T	A			0.074	0.048	0.049
		G	G	C	G			0.12	0.21	0.0032
ANRIL	Namvar 2020	rs1333045	rs1333048
		C	A					0.391	0.322	0.918
		T	A					0.026	0.303	<0.0001
		C	C					0.072	0.227	<0.0001
		T	C					0.51	0.146	<0.0001
ARRB2	Ikeda 2007	rs1045280	rs4790694	rs2036657				Global		0.0175
		T	C	A				0.802	0.854	0.0194
		C	A	G				0.086	0.052	0.021
BDNF	Su 2015a	rs16917204	rs16917234	rs2030324
		C	C	T				0.028	0.068	0.008
		C	T	T				0.381	0.393	0.772
		G	C	C				0.148	0.104	0.049
		G	T	C				0.316	0.325	0.826
		G	T	T				0.087	0.076	0.539
COMT	Jugurnauth 2011	rs4680	rs165599
		A	A					0.24	0.23	0.46
		A	G					0.03	0.08	0.0006
		G	A					0.26	0.27	0.59
		G	G					0.47	0.42	0.15
DTNBP1	Kishimoto 2008	rs2619539	rs3213207	rs2619538				Global		0.0005
		C	A	A				0.6046	0.7101	0.0013
		G	A	A				0.3315	0.2741	0.076
		C	G	T				0.0318	0.0022	0.0012
		C	G	A				0.0178	0.0023	0.11
		C	A	T				0.0055	0.0073	0.83
		G	G	A				0.0089	0	0.15
		G	A	T				0	0.0039	0.18
GABBR1	Zhao 2018	rs2076483	rs29221	rs715044
		C	G	A				0.081	0.108	0.024
		C	G	C				0.133	0.123	0.477
		T	C	C				0.309	0.312	0.884
		T	G	C				0.477	0.457	0.339
GABRG2	Nishiyama 2005	315C4T	1128+99C4A					Global		0.044
		C	C					0.275	0.314
		C	A					0.425	0.428
		T	C					0.262	0.186
		T	A					0.039	0.072
GAD1	Veerasakul 2017	rs769404	rs701492					Global		0.038
		C	C					0.46	0.385	0.125
		C	T					0.055	0.12	0.021
		T	C					0.27	0.326	0.218
		T	T					0.215	0.169	0.241
GRM2	Tsunoka 2010	rs3821829	rs12487957	rs4687771				Global		0.00746
		C	C	A				0.746	0.673	0.00822
		C	T	T				0.254	0.327	0.00822
HTR2A	Tsunoka 2010	rs6311	rs6313					Global		0.589
		A	T					0.083	0.0778	0.327
		G	T					0.465	0.467	0.468
		G	C					0.452	0.455	0.922
HTR6	Kishi 2011c	rs6693503	rs1805054
		A	C					0.519	0.574	0.0824
		A	T					0.287	0.292	0.863
		G	C					0.194	0.134	0.00931
		rs6693503	rs1805054	rs4912138
		A	C	A				0.444	0.488	0.227
		A	C	G				0.0744	0.0862	0.351
		A	T	G				0.284	0.292	0.0772
		G	C	A				0.0798	0.00161	0.0001
		G	C	G				0.118	0.132	0.646
OPRM1	Ide 2006	rs1799971	rs2075572	rs599548	rs598682			Global		0.3
		A	C	A	A			0	0.002
		A	C	A	G			0.005	0.024
		A	C	G	A			0	0.014
		A	C	G	G			0.336	0.329
		A	G	A	A			0	0.005
		A	G	A	G			0.115	0.08
		A	G	G	A			0.109	0.051
		A	G	G	G			0.028	0.029
		G	C	A	G			0.015	0.006
		G	C	G	G			0.382	0.441
		G	G	A	G			0.005	0.015
		G	G	G	A			0	0.004
		G	G	G	G			0.004	0
PICK1	Matsuzawa 2007	rs737622	rs3026682	rs11089858	rs713729	rs3952	rs2076369
		C	G	G	T	A	T	0.337	0.352	0.63
		G	A	G	T	G	G	0.323	0.323	0.85
		C	G	G	A	A	G	0.092	0.145	<0.02
		C	G	A	T	A	G	0.074	0.083	0.66
		C	G	G	T	A	G	0.089	0.055	<0.09
		G	A	G	T	G	T	0.035	0.007	0.01
		C	G	G	A	A	T	0.017	0.012	0.66
		G	A	G	A	G	G	0.004	0.01	0.4
PIK4CA	Kanahara 2009	rs2072513	rs165862	rs165793	rs165789
		G	C	T	A			0.457	0.451	1
		A	A	C	G			0.352	0.357	1
		A	A	T	G			0.112	0.105	0.996
		A	C	T	A			0.047	0.053	1
PLG	Iwata 2004	rs3757017	rs14224					Global		0.75
		D	D					0.47	0.48
		D	I					0.11	0.09
		I	I					0.01	0.01
		D						0.41	0.42
PROKR2	Kishi 2010	rs17721321	rs6085086	rs3746684	rs3746682	rs4815787		Global		0.000959
		G	G	G	C	G		0.351	0.488	0.00019
		G	A	A	G	A		0.403	0.281	0.000471
		G	G	A	G	G		0.171	0.164	0.791
		G	G	G	G	G		0.0745	0.0673	0.704
SLC6A9	Morita 2008	rs2486001	rs2248829
		C	G					0.48	0.54	0.08
		C	A					0.24	0.27	0.3
		T	G					0.28	0.16	0.000039
		T	A					0	0.027	N/A
SLC22A3	Aoyama 2006	rs509707	rs4709426					Global		0.178
		C	G					0.5104	0.566
		T	C					0.3158	0.2607
		C	C					0.1487	0.1432
		T	G					0.02513	0.02998
SOD2	Nakamura 2006 (Japanese)	rs2758357	rs4880	rs2855116 & rs2758332	rs2758330	rs2842980
		C	V	A	G	A		0.4663	0.4819	0.7244
		C	V	A	T	T		0.3872	0.3755	0.7887
		T	A	B	T	T		0.1202	0.1024	0.5212
		T	V	A	T	T		0.0262	0.0311	0.7495
		C	V	A	T	A		0	0.0091	0.0861
	Nakamura 2006 (Han)	rs2758357	rs4880	rs2855116 & rs2758332	rs2758330	rs2842980
		C	V	A	G	A		0.437	0.4374	0.9841
		C	V	A	T	T		0.3583	0.3852	0.4944
		T	A	B	T	T		0.1575	0.1038	0.0493
		T	V	A	G	A		0.0236	0.0489	0.0997
		T	A	A	T	T		0.0236	0.0246	0.9384
TNFA	Nomura 2006b	308G>A	857C>T
		G	C					0.79	0.75	0.22
		G	T					0.19	0.23	0.12
		A	C					0.025	0.017	0.42

3.3. Meta-analyses

Specific markers of seven genes (BDNF, COMT, DRD2, DRD4, FAAH, SLC6A3, SLC6A4) were studied in three or more studies, and a meta-analytic approach was used to determine any potential association between markers in these genes and meth use disorder. Summary of the key meta-analytic findings are presented in Table 5.

Table 5.

Summary of meta-analytic investigation of minor allele frequency in people with meth use disorder compared to controls.

SNP	OR	95% CI	z-score	p-value	I2
BDNF (rs6265)	0.78	0.69–0.88	4.13	<0.0001	0%
DRD2 (rs1800497)	0.95	0.83–1.09	0.77	0.44	14%
DRD4 (Exon 3 VNTR)	1.10	0.92–1.32	1.05	0.29	20%
COMT (rs4680)	0.90	0.79–1.02	1.68	0.09	72%
SLC6A3 (3′ UTR)	1.17	0.81–1.69	0.83	0.41	0%
SLC6A4 (LPR)	0.78	0.61–0.99	2.01	0.04	0%
FAAH (rs324420)	1.64	1.40–1.92	6.21	<0.00001	66%

Notes: CI = confidence interval; OR = odds ratio; SNP = single nucleotide polymorphism.

3.3.1. BDNF

There were seven studies examining the rs6265 marker of BDNF (Bousman et al., 2010; Cheng et al., 2005; Iamjan et al., 2015; Itoh et al., 2005; Sim et al., 2010; Su et al., 2015b, 2014). rs6265 is located within the coding region of BDNF, in exon 9 (Fig. 3A). It is a missense mutation, substituting a valine (Val) for a methionine (Met) residue. The meta-analysis revealed a significant protective effect of the minor allele on meth use disorder (p < 0.0001), with an overall effect size of 0.78 [95% CI, 0.69, 0.88] (Fig. 3B). Heterogeneity analyses revealed low heterogeneity (I² = 0%). Sensitivity analyses revealed that removing the study with the highest risk of bias (NHLBI score = 5) had no significant impact on the results or heterogeneity.

Fig. 3.

BDNF rs6265. A. Schematic representation of BDNF and the location of rs6265. Numbers represent exons. Lighter shaded boxes represent non-coding regions. B. Forrest plot of the association between the rs6265 variant of BDNF and meth use disorder. There was a significant overall effect (p < 0.0001) of the variant, with the minor allele associated with protection against meth use disorder (OR = 0.78). Event = number of minor alleles; OR = odds ratio.

3.3.2. Dopaminergic signaling

There were five studies examining the rs1800497 marker of DRD2 (Chen et al., 2004; Han et al., 2008; Šerý et al., 2001; Tsai et al., 2002; Ujike et al., 2009), four studies examining the exon 3 VNTR of DRD4 (Chen et al., 2004; Li et al., 2004; Tsai et al., 2002; Ujike et al., 2009), four studies examining the rs4680 marker of COMT (Bousman et al., 2010; Jugurnauth et al., 2011; Li et al., 2004; Suzuki et al., 2006), and three studies examining the 3′ VNTR of SLC6A3 (Hong et al., 2003; Liu et al., 2004; Ujike et al., 2003). rs1800497 is located downstream of DRD2 (Fig. 4A), and is associated with reduced dopamine D2 receptor (DRD2) binding (Tunbridge et al., 2019). The exon 3 VNTR is located within a coding region of DRD4 (Fig. 4C) and consists of 2–11 48-bp repeats. rs4680 is located within the coding region of COMT, in exon 5 (Fig. 4E). It is a missense mutation, substituting a Val for a Met residue, with the Val resulting in greater protein abundance and stability (Tunbridge et al., 2019). The 3′ VNTR is located within the 3′ UTR of SLC6A3 (Fig. 4G) and consists 9–10 repeat alleles.

Fig. 4.

Makers in dopamine-related genes. A. Schematic representation of DRD2 and the location of rs1800497. Numbers represent exons. Lighter shaded boxes represent non-coding regions. B. Forrest plot of the association between the rs1800497 variant of DRD2 and meth use disorder. C. Schematic representation of DKD4 and the location of the exon 3 VNTR. D. Forrest plot of the association between the exon 3 VNTR of DKD4 and meth use disorder. E. Schematic representation of COMT and the location of rs4680. F. Forrest plot of the association between the rs4680 variant of COMT and meth use disorder. G. Schematic representation of SLC6A3 and the location of the 3′ VNTR. H. Forrest plot of the association between the 3′ VNTR of SLC6A3 and meth use disorder. There was no significant overall effect (ps > 0.05) of any of the variants on meth use disorder. Event = number of minor alleles; VNTR = variable number tandem repeat.

The meta-analysis showed no effect of any markers on meth use disorder (smallest p = 0.09), with effect sizes of 0.95 [95% CI, 0.83, 1.09] for DRD2 (Figure 4B), 1.10 [95% CI, 0.92, 1.32] for DRD4 (Figure 4D), 0.90 [95% CI, 0.79, 1.02] for COMT (Fig. 4F), and 1.17 [95% CI, 0.81, 1.69] for SLC18A3 (Fig. 4H). Heterogeneity analyses revealed low heterogeneity for DRD2, DRD4 and SLC6A3 (I² = 0–20%), whereas COMT had considerable heterogeneity (I² = 70%). Sensitivity analyses revealed that removing the study with the highest risk of bias for each gene (NHLBI score = 3 (DRD2) and 5 (DRD4, COMT and SLC6A3)) had no significant impact on the results or heterogeneity.

3.3.3. Serotonergic signaling

There were three studies examining the LPR marker of SLC6A4 (Ezaki et al., 2008; Hong et al., 2003; Payer et al., 2012). The LPR is located in the promoter region of SLC6A4 (Fig. 5A), with the short variant associated with reduced gene expression (Greenberg et al., 1999; Lesch et al., 1996). The meta-analysis revealed a moderate but significant protective effect of the long variant on meth use disorder (p = 0.04), with an overall effect size of 0.78 [95% CI, 0.61, 0.99] (Fig. 5B). Sensitivity analyses revealed that when the study with the highest risk of bias (NHLBI score = 5) was removed, the significant overall effect did not survive (p = 0.07). Heterogeneity analyses revealed low heterogeneity (I² = 0%).

Fig. 5.

A. SLC6A4 LPR. Schematic representation of SLC6A4 and the location of the LPR. Numbers represent exons. Lighter shaded boxes represent non-coding regions. B. Forrest plot of the association between the LPR variant of SLC6A4 and meth use disorder. There was a moderate overall effect (p = 0.04) of the variant, with the minor allele associated with protection against meth use disorder (OR = 0.78). Event = number of minor alleles; LPR = linked polymorphic region; OR = odds ratio.

3.3.4. Endocannabinoid signaling

There were three studies examining the rs324420 marker of FAAH (Morita et al., 2005a; Sim et al., 2013; Zhang et al., 2020). rs324420 is located within the coding region of FAAH, in exon 4 (Fig. 6A). It is a missense mutation, substituting a proline (Pro) for a threonine (Thr) residue. The meta-analysis revealed a significant risk of the minor allele on meth use disorder (p < 0.00001), with an overall effect size of 1.64 [95% CI, 1.40, 1.92] (Fig. 6B). Heterogeneity analyses revealed considerable heterogeneity (I² = 66%). Sensitivity analyses revealed that removing the study with the highest risk of bias (NHLBI score = 8) had no significant impact on the results or heterogeneity.

Fig. 6.

A. FAAH rs324420. Schematic representation of FAAH and the location of rs324420. Numbers represent exons. Lighter shaded boxes represent non-coding regions. B. Forrest plot of the association between the rs324420 variant of FAAH and meth use disorder. There was a significant overall effect (p < 0.00001) of the variant, with the minor allele associated with a risk for meth use disorder (OR = 1.64). Event = number of minor alleles; OR = odds ratio.

3.4. Secondary outcomes

3.4.1. Polysubstance use disorder

To investigate whether markers were specific to meth use disorder or common to multiple substances, the difference in genotype and allele distribution between people with polysubstance use disorder and people who exclusively used meth was examined (Table 6). Of the 34 studies reporting a co-morbid diagnosis of polysubstance use disorder, only 14 provided data on polysubstance use for a total of 38 markers in 17 different genes. Two markers (GSTP1 rs1695; and PICK1 rs713729) displayed significant genotypic or allelic association with meth use disorder only, three markers (NPY1R rs7687423; SLC22A3 rs509707, rs7745775 and rs3106164; and XBP1 – 116C > G) with polysubstance use disorder only, and three markers (ADORA2A rs5751876; GDNF rs2910704; and PDYN 68-bp repeat) with both diagnoses. Of these, the genotype distribution of two markers (GSTP1 rs1695; and NPY1R rs7687423) was significantly different between cases and controls but failed to reach allelic statistical significance. Characteristics for each significant marker are given in Table 3.

Table 6.

Genotype and allele frequencies of candidate gene association studies between non-polydrug users and polydrug users.

Gene	Code	Reference	Drug Use Status	Meth				Control				p-value	Meth		Control		OR	95% CI	p-value	Power (1 − β)
				M/M	M/m	m/m	n	M/M	M/m	m/m	n		M	m	M	m
ADORA2A	rs5751876	Kobayashi 2010	Meth Only	6	29	17	52	70	114	45	229	0.01	41	63	254	204	1.91	1.24–2.95	0.003	0.84
			Polydrug Use	26	54	32	112	70	114	45	229	0.127	106	118	254	204	1.39	1.01–1.91	0.046	0.52
BDNF	rs6265	Itoh 2005	Meth Only	21	27	8	56	70	107	25	202	0.812	69	43	247	157	0.98	0.64–1.51	0.928	0.05
			Polydrug Use	47	60	15	122	70	107	25	202	0.768	154	90	247	157	0.92	0.66–1.28	0.616	0.08
	132C > T	Itoh 2005	Meth Only	51	5	0	56	183	19	0	202	0.913	107	5	385	19	0.95	0.35–2.60	0.915	0.05
			Polydrug Use	114	8	0	122	183	19	0	202	0.369	236	8	385	19	0.69	0.30–1.59	0.379	0.14
FAAH	rs324420	Morita 2005a	Meth Only	30	17	1	48	139	58	3	200	0.642	77	19	336	64	1.30	0.73–2.29	0.371	0.14
			Polydrug Use	75	26	4	105	139	58	3	200	0.352	176	34	336	64	1.01	0.64–1.60	0.951	0.05
GDNF	rs2973033	Yoshimura 2011	Meth Only	26	21	10	57	202	155	24	381	0.012	73	41	559	203	1.55	1.02–2.34	0.38	0.54
			Polydrug Use	82	48	11	141	202	155	24	381	0.366	212	70	559	203	0.91	0.66–1.25	0.553	0.09
	rs10941370	Yoshimura 2011	Meth Only	22	23	12	57	145	183	53	381	0.312	67	47	473	289	1.15	0.77–1.71	0.499	0.10
			Polydrug Use	54	66	20	14 0	145	183	53	381	0.983	174	106	473	289	1.00	0.75–1.32	0.984	0.05
	rs12518844	Yoshimura 2011	Meth Only	24	23	10	57	154	183	44	381	0.35	71	43	491	271	1.10	0.73–1.65	0.655	0.07
			Polydrug Use	57	65	19	141	154	183	44	381	0.82	179	103	491	271	1.04	0.79–1.39	0.774	0.06
	rs2910702	Yoshimura 2011	Meth Only	34	19	4	57	164	172	46	382	0.057	87	27	500	264	0.59	0.37–0.93	0.021	0.64
			Polydrug Use	50	77	15	14 2	164	172	46	382	0.17	177	107	500	264	1.14	0.86–1.52	0.348	0.16
	rs2910704	Yoshimura 2011	Meth Only	15	30	12	57	143	190	48	381	0.111	60	54	476	286	1.50	1.01–2.23	0.44	0.52
			Polydrug Use	70	60	12	14 2	143	190	48	381	0.042	200	84	476	286	0.70	0.52–0.94	0.017	0.67
	rs884344	Yoshimura 2011	Meth Only	30	23	4	57	219	140	22	381	0.776	83	31	578	184	1.17	0.75–1.83	0.481	0.11
			Polydrug Use	94	43	4	141	219	140	22	381	0.112	231	51	578	184	0.69	0.49–0.98	0.037	0.54
	rs1549250	Yoshimura 2011	Meth Only	26	28	3	57	198	158	25	381	0.547	80	34	554	208	1.13	0.74–1.74	0.573	0.09
			Polydrug Use	65	60	16	141	198	158	25	381	0.155	190	92	554	208	1.29	0.96–1.73	0.091	0.39
	rs2973049	Yoshimura 2011	Meth Only	18	28	11	57	109	178	93	380	0.685	64	50	396	364	0.85	0.57–1.26	0.421	0.13
			Polydrug Use	33	77	31	141	109	178	93	380	0.274	143	139	396	364	1.06	0.81–1.39	0.689	0.07
GSTP1	rs1695	Hashimoto 2005	Meth Only	44	9	2	55	167	32	0	199	0.026	97	13	366	32	1.53	0.77–3.03	0.217	0.24
			Polydrug Use	87	28	1	116	167	32	0	199	0.085	202	30	366	32	1.70	1.00–2.88	0.053	0.51
NPY	rs16147	Okahisa 2009	Meth Only	29	27	3	59	119	128	41	288	0.138	85	33	366	210	0.68	0.44–1.05	0.078	0.42
			Polydrug Use	64	69	20	153	119	128	41	288	0.945	197	109	366	210	0.96	0.72–1.29	0.805	0.06
NPY1R	rs7687423	Okahisa 2009	Meth Only	12	32	15	59	90	129	65	284	0.217	56	62	309	259	1.32	0.89–1.97	0.169	0.28
			Polydrug Use	40	91	25	156	90	129	65	284	0.031	171	141	309	259	0.98	0.75–1.30	0.908	0.05
NQO1	rs1800566	Ohgake 2005	Meth Only	17	31	4	52	85	95	27	207	0.188	65	39	265	149	1.07	0.68–1.67	0.775	0.06
			Polydrug Use	41	53	19	11 3	85	95	27	207	0.559	135	91	265	149	1.20	0.86–1.67	0.286	0.19
NQO2	N/A	Ohgake 2005	Meth Only	35	12	5	52	123	74	10	207	0.13	82	22	320	94	0.91	0.54–1.54	0.734	0.06
			Polydrug Use	68	36	9	113	123	74	10	207	0.464	172	54	320	94	1.07	0.73–1.57	0.733	0.06
PDYN	68-bp	Nomura 2006a	Meth Only	21	16	2	39	149	59	1	209	0.01	58	20	357	61	2.02	1.13–3.59	0.015	0.68
			Polydrug Use	56	31	8	95	149	59	1	209	<0.001	143	47	357	61	1.92	1.26–2.95	0.002	0.86
PICK1	rs737622	Matsuzawa 2007	Meth Only	23	23	9	55	89	107	22	218	0.364	69	41	285	151	1.12	0.73–1.73	0.604	0.08
			Polydrug Use	58	63	19	140	89	107	22	218	0.547	179	101	285	151	1.06	0.78–1.46	0.694	0.07
	rs3026682	Matsuzawa 2007	Meth Only	23	23	9	55	89	107	22	218	0.364	69	41	285	151	1.12	0.73–1.73	0.604	0.08
			Polydrug Use	58	63	19	140	89	107	22	218	0.547	179	101	285	151	1.06	0.78–1.46	0.694	0.07
	rs11089858	Matsuzawa 2007	Meth Only	44	11	0	55	180	37	1	218	0.772	99	11	397	39	1.13	0.56–2.29	0.732	0.06
			Polydrug Use	112	26	2	140	180	37	1	218	0.563	250	30	397	39	1.22	0.74–2.02	0.434	0.12
	rs713729	Matsuzawa 2007	Meth Only	47	7	1	55	150	63	5	218	0.045	101	9	363	73	0.44	0.21–0.92	0.025	0.61
			Polydrug Use	109	28	3	140	150	63	5	218	0.163	246	34	363	73	0.69	0.44–1.07	0.092	0.39
	rs3952	Matsuzawa 2007	Meth Only	23	23	9	55	89	107	22	218	0.364	69	41	285	151	1.12	0.73–1.73	0.604	0.08
			Polydrug Use	58	63	19	140	89	107	22	218	0.547	179	101	285	151	1.06	0.78–1.46	0.694	0.07
	rs2076369	Matsuzawa 2007	Meth Only	15	30	10	55	82	111	25	218	0.226	60	50	275	161	1.42	0.93–2.17	0.101	0.37
			Polydrug Use	53	62	25	14 0	82	111	25	218	0.195	168	112	275	161	1.14	0.84–1.55	0.409	0.13
SLC22A3	rs509707	Aoyama 2006	Meth Only	32	25	2	59	226	175	41	442	0.317	89	29	627	257	0.79	0.51–1.24	0.31	0.17
			Polydrug Use	57	65	22	144	226	175	41	442	0.024	179	109	627	257	1.49	1.12–1.96	0.005	0.80
	rs4709426	Aoyama 2006	Meth Only	21	32	6	59	165	199	78	442	0.255	74	44	529	355	0.89	0.60–1.32	0.55	0.09
			Polydrug Use	40	73	20	133	165	199	78	442	0.134	153	113	529	355	1.10	0.83–1.45	0.499	0.10
	rs7745775	Aoyama 2006	Meth Only	30	24	4	58	246	146	33	425	0.576	84	32	638	212	1.15	0.74–1.77	0.54	0.10
			Polydrug Use	81	59	2	142	246	146	33	42 5	0.014	221	63	638	212	0.86	0.62–1.18	0.348	0.16
	rs3106164	Aoyama 2006	Meth Only	30	24	4	58	205	198	37	440	0.749	84	32	608	272	0.85	0.55–1.31	0.47	0.11
			Polydrug Use	84	48	11	14 3	205	198	37	440	0.036	216	70	608	272	0.72	0.53–0.98	0.038	0.55
	rs3088442	Aoyama 2006	Meth Only	13	25	21	59	119	208	111	438	0.239	51	67	446	430	1.36	0.93–2.01	0.117	0.34
			Polydrug Use	34	65	45	144	119	208	111	438	0.357	133	155	446	430	1.21	0.93–1.58	0.163	0.29
SLC6A3	exon 15	Ujike 2003	Meth Only	30	1	1	32	138	20	2	160	0.233	61	3	296	24	0.61	0.18–2.08	0.422	0.12
			Polydrug Use	74	15	0	89	138	20	2	160	0.377	163	15	296	24	1.13	0.58–2.22	0.712	0.07
	242C > T	Ujike 2003	Meth Only	33	0	0	33	104	2	0	106	0.427	66	0	210	2	NA	NA	0.428	0.12
			Polydrug Use	67	2	0	69	104	2	0	106	0.662	136	2	210	2	1.54	0.22–11.09	0.663	0.07
	1342A > G	Ujike 2003	Meth Only	28	7	1	36	129	26	4	159	0.898	63	9	284	34	1.19	0.55–2.61	0.658	0.07
			Polydrug Use	72	17	0	89	129	26	4	159	0.288	161	17	284	34	0.88	0.48–1.63	0.688	0.07
	2319G > A	Ujike 2003	Meth Only	21	13	2	36	92	57	8	157	0.994	55	17	241	73	1.02	0.56–1.87	0.948	0.05
			Polydrug Use	47	37	4	88	92	57	8	157	0.675	131	45	241	73	1.13	0.74–1.74	0.564	0.09
TAAR1	rs8192620	Loftis 2019	Meth Only				32				31		19	13	17	14	0.83	0.31–2.26	0.716	0.07
			Polydrug Use				43				31		18	25	17	14	1.69	0.67–4.28	0.27	0.20
TNFA	308G > A	Nomura 2006b	Meth Only	55	3	0	58	207	7	0	214	0.495	113	3	421	7	1.60	0.41–6.27	0.499	0.10
			Polydrug Use	125	7	0	132	207	7	0	214	0.351	257	7	421	7	1.64	0.57–4.72	0.356	0.15
	857C > T	Nomura 2006b	Meth Only	35	18	5	58	134	68	17	219	0.976	88	28	336	102	1.05	0.65–1.69	0.848	0.05
			Polydrug Use	87	37	6	130	134	68	17	219	0.4	211	49	336	102	0.76	0.52–1.21	0.168	0.28
TNFAR1	36A > G	Nomura 2006b	Meth Only	41	14	1	56	151	59	6	216	0.849	96	16	361	71	0.85	0.47–1.53	0.58	0.09
			Polydrug Use	91	34	3	12 8	151	59	6	216	0.956	216	40	361	71	0.94	0.62–1.44	0.78	0.06
XBP1	−116C > G	Morita 2005b	Meth Only	19	22	7	48	80	88	32	200	0.961	60	36	248	152	0.98	0.62–1.55	0.928	0.05
			Polydrug Use	53	42	10	10 5	80	88	32	200	0.129	148	62	248	152	0.68	0.48–0.98	0.037	0.55

CI = confidence interval; M = major allele; m = minor allele; OR = odds ration. Values in bold are statistically significant (p < 0.05)

The minor allele of an intronic variant in PICK1 (rs713729) was found to be protective (OR < 1) in people who exclusively use meth, suggesting a specific role for this gene in meth use disorder susceptibility. An intronic variant of SLC22A3 (rs3106164) and a marker in the promoter region of XBP1 (−116C > G) were found to be protective in polysubstance use only, while a different intronic marker in SLC22A3 (rs509707) was found to confer a risk (OR > 1) for polysubstance use disorder. This suggests that both genes may be involved in the development of substance use disorders other than meth. A marker in the 5′ UTR of ADORA2A (rs5751876) and the promoter region of PDYN (68-bp repeat) were found to confer a risk for both meth use and polysubstance use disorders, highlighting a potential shared genetic vulnerability for more than one substance use disorder. Subgroup analyses for ADORA2A and PDYN were adequately powered (1 − β = 0.80–0.86).

One marker in GDNF (rs2910702) was found to be protective in meth only (OR = 0.59), while a different marker (rs884344) was protective in polysubstance use only (OR = 0.69). In addition, a third maker (rs2910704) conferred a risk in people who exclusively used meth (OR = 1.5), whereas it was protective in polysubstance users (OR = 0.7). This suggests a complex involvement of GDNF in substance use disorder vulnerability. It should be noted that all three GDNF markers were located within an intron. In addition, none of the associations were adequately powered (1 − β = 0.52–0.67).

3.4.2. Sex differences

A total of 12 studies examined genotype and allele distribution between male and female participants separately for 45 markers across 18 different genes (Table S2). Notably, three markers (ADORA2A rs5751876; ARRB2 rs1045280; and GABRG2 rs4480617) previously shown to confer a significant risk for meth use disorder (Table 2) were only significant in females and not in males (Table S2). In addition, six markers (CHRNA4 rs1044397; GABRA1 rs2279020; SNCA rs1372520, rs3756063 and rs3756059; and TSNAX rs1630250) displayed significantly different allele distribution between people with meth use disorder and controls when split by sex, although significance was not reached when combined. The intronic variant of GABRA1 (rs2279020) was only protective in females, whereas the synonymous variant located in exon 5 of CFIRNA4 (rs1044397) was only protective in males. In contrast, three intronic markers of SNCA (rs1372520, rs3756063, rs3756059) were shown to confer a risk for meth use disorder in females only, whereas a marker ~2 kilobases upstream of TSNAX (rs1630250) was a risk marker in males only. It should be noted that nearly all associations were underpowered in this subgroup analyses (1 − β = 0.05–0.81), with the association between GABRG2 rs4480617 and meth use disorder in female participants the only study adequately powered (1 − β = 0.81). Characteristics for each significant marker are given in Table 3.

3.4.3. Age of onset of meth use

Nine studies investigated the difference in genotype and allele distribution between adolescent-onset (< 20 years of age) and adult-onset of meth use (≥ 20 years of age) in markers in 9 different genes (Supplementary Table S3). GDNF (Ts2910704)—discussed above—was the only significantly different marker at the genotypic level but failed to reach allelic statistical significance. It was also greatly underpowered (1 − β = 0.07).

4. Discussion

The present study systematically reviewed gene association studies in people with meth use disorder. A previous systematic review uncovered 38 studies and provided a thorough examination of the field from 2000 to 2009 (Bousman et al., 2009). In the present review, we provide a significant update of this expanding field and reviewed a total of 79 studies. While the number of candidate gene studies has almost doubled in the past decade, there is still a clear lack of twin and family studies, as well as adequately powered GWASs. Our meta-analyses of gene markers investigated in three or more studies show that candidate gene association studies can be informative. However, it is critical that the choice of candidate variant is well rationalized. For example, a focus on a functional variant with known biological significance would be important in understanding the role of a particular gene in meth use disorder. In the present study, we found that only ~25 % of the markers significantly associated with meth use disorder were located within a coding region, with only five markers leading to amino acid changes (Table 3). In addition, investigating several candidates from the same neurochemical pathway could reveal how these genes interact and may lead to susceptibility of the disorder of interest. Rationale for a candidate gene study can also include significant findings from preclinical studies involving rodents. There were many studies without a clear rationale.

Of the 75 genes examined in the literature to date, 29 had at least one marker displaying significant genotypic or allelic association with meth use disorder. Utilizing our present findings showing the power of each association experiment, we will focus our discussion on associations with adequate power across the entire sample (1 − β ≥ 0.80), and any related secondary outcomes. We will also discuss key limitations in the current literature and highlight future research directions to further understand the potential genetic contributions toward meth use disorder.

4.1. Genes associated with meth use disorder

Fourteen markers in 11 genes were adequately powered, with NHLBI risk bias scores ranging from 4 (high risk of bias) to 9 (relatively low risk of bias). The large majority of genes examined were involved in neurotransmission, highlighting the need for future studies to examine other emerging areas identified to be involved in meth use such as inflammation and oxidative stress (Hacimusalar et al., 2019; Kohno et al., 2019; Luikinga et al., 2018; Shin et al., 2018; Zhang et al., 2018).

4.1.1. Endocannabinoid system

The most robust evidence with the biggest effect size for a genetic susceptibility to meth use disorder was the rs324420 variant of FAAH. Two reports found that the minor allele (A) was a risk for meth use disorder in Malaysian and Han Chinese samples, respectively (Sim et al., 2013; Zhang et al., 2020). A third report found no association in a Japanese sample (Morita et al., 2005a), though the study was highly underpowered (1 − β = 0.08). Meta-analysis revealed a strong overall effect (p < 0.00001), with the variant conferring ~1.64 greater odds for risk of meth use disorder (Fig. 6B). All three studies had a relatively low risk of bias, and sensitivity analyses revealed that removing the study with highest risk of bias had no significant impact on the results.

rs324420 is a variant in exon 3 of FAAH (Fig. 6A). It is a missense mutation which converts a proline (Pro/C-allele) to a threonine (Thr/A-allele) residue (Pro129Thr) in the Fatty Acid Amide Hydrolase (FAAH) protein. The main function of FAAH is to break down anandamide, an endocannabinoid (Deutsch and Chin, 1993). The Thr variant leads to reduced enzymatic catalytic activity and reduced protein stability in cell lines and rodents (Chiang et al., 2004; Dincheva et al., 2015). This leads to increased levels of anandamide in the brain. A study reported that participants carrying the minor allele displayed heightened reward-related ventral striatal activity and impulsivity (Hariri et al., 2009). In addition, the variant may modulate the subjective effect of stimulants. In people with cocaine use disorder, minor allele carriers reported a stronger drug effect, and a higher “high” compared to homozygous major allele carriers when receiving a cocaine injection (Patel et al., 2018). The minor allele variant has previously been reported to confer a risk for comorbid illicit drug or alcohol use (OR = 4.88) (Sipe et al., 2002). This suggests that the association between the variant and substance use disorders is not specific to meth use. Consistent with this idea, we found no difference in allele distribution between people who exclusively use meth and polysubstance users (Table 6). Therefore, the rs324420 variant of FAAH may, in concert with other genetic and environmental factors, promote substance use disorders to a limited extent by increasing levels of anandamide in the brain, leading to heightened subjective response to drugs, higher impulsivity, and increased in reward-related activity in the ventral striatum. It would be informative to validate the effect of the minor allele on levels of anandamide in the brain of people with meth use disorder. New pharmacotherapies, such as the newly discovered inhibitor LEI-401 (Mock et al., 2020), could then be trialed to reduce levels of anandamide in minor allele carriers, and may be a promising approach to treat substance use disorders.

4.1.2. BDNF

While none of the seven studies exploring the BDNF rs6265 variant were adequately powered (range 0.07 – 0.52), the meta-analysis suggested an overall protective effect of the minor allele (A) on meth use disorder (OR = 0.78; p < 0.0001) (Fig. 3B). Sensitivity analyses revealed that removing the study with the highest risk of bias had no significant impact on the results. BDNF codes for brain-derived neurotrophic factor (BDNF), which plays a critical role in growth, differentiation and survival of neurons and synapses (Huang and Reichardt, 2001). Importantly, BDNF acting in the mesocorticolimbic dopamine system is a positive modulator of the rewarding effect of psychostimulants (evidence reviewed in Ghitza et al., 2010).

The rs6265 variant has been widely studied across a range of neurological and psychiatric disorders (Bruenig et al., 2016; Chen et al., 2017; Lin et al., 2014). It is a missense mutation substituting a valine (Val/G-allele) for a methionine (Met/A-allele) residue, with the Met variant leading to impaired intracellular trafficking and secretion of BDNF (Egan et al., 2003). The Met/A-allele may confer protection against meth use disorder by reducing levels of BDNF, and therefore the rewarding effect of meth. In addition, the Met/A-allele appears to be protective in people with heroin and nicotine use disorders (Haerian, 2013; Zhao et al., 2020), suggesting a shared genetic risk across substances.

Su et al. (2014, 2015b) investigated the effect of this variant on cognition and impulsivity in people with meth use disorder. While they found that meth users displayed poorer performance in memory, language, and visuospatial tasks, there were no genotype effects (Su et al., 2015b). In a second study, they reported an association between Val66Met and aspects of impulse control in people with meth use disorder, with homozygous Met/A-allele carriers reporting higher attentional impulsivity (Su et al., 2014). It should be noted that both studies were underpowered (1 − β = 0.39–0.40), and further research is needed to better understand the potential impact of rs6265 on cognitive function in people with meth use disorder.

4.1.3. Dopaminergic system

Considering the critical involvement of dopamine in the development of meth and other substance use disorders, it is not surprising that genes involved in mediation and modulation of dopaminergic signaling were among the most studied. Three markers in dopamine-related genes were adequately powered (ADORA2A rs5751876; ARRB2 rs1045280; and DTNBP1 rs3213207)

The gene with the largest effect size was DTNBP1, with the minor allele (G) of the rs3213207 variant conferring ~7.13 greater odds of developing meth use disorder in a sample of Japanese descent (Kishimoto et al., 2008). rs3213207 is an intronic variant. DTNBP1 codes for Dystrobrevin-binding protein 1, or simply Dysbindin, a protein found in certain axon terminals throughout the brain (Benson et al., 2001). Dysbindin deficiency leads to increased extracellular dopamine receptor D2 (D2R), and increase strength of D2R signaling (Iizuka et al., 2007). Given than meth leads to an increase of dopamine concentration in the synaptic cleft (Cruickshank and Dyer, 2009; Riddle et al., 2008; Sulzer et al., 2005), it is possible that higher levels of extracellular dopamine in combination with increased concentration of D2R may contribute to the pathophysiology of meth use disorder. The rs3213207 variant has previously been shown to be associated with bipolar disorder, first episode psychosis and schizophrenia (Gaysina et al., 2009; Varela-Gomez et al., 2015; Wirgenes et al., 2009). In addition, there is evidence that haplotypes including rs3213207 are involved in the susceptibility for bipolar disorder and major depressive disorder (Duan et al., 2007; Kim et al., 2008; Pae et al., 2009, 2007). This is particularly notable because every participant with meth use disorder in Kishimoto et al. (2008) also presented with a co-morbid diagnosis of meth-induced psychosis. It is possible that this particular variant may be more broadly associated with psychotic disorders rather than meth use specifically.

Kobayashi et al. (2010) reported an association between the rs5751876 variant of ADORA2A and meth use disorder, with the C allele conferring a risk in a Japanese sample (OR = 1.50) (Kobayashi et al., 2010). ADORA2A codes for the adenosine A2A receptor, and plays an important role in striatal function (Fredholm et al., 2005). It is reported to form heterodimers with D2R, which may contribute towards the reinforcing effect of psychostimulants (Ballesteros-Yáãnez et al., 2018; Fuxe et al., 2005). While the rs5751876 variant is a synonymous mutation (Table 3), it has been associated with the anxiogenic effect of amphetamines, with C-allele carriers reporting lower levels of anxiety (Hohoff et al., 2005). This has led to the suggestion that carrying this allele might lessen the adverse effect of amphetamine use, therefore promoting sustained use. Subgroup analyses revealed that the C allele was a risk allele in people who exclusively use meth as well as polysubstance users. Sex analyses revealed that the significant distribution was only found in female participants, although this secondary outcome analyses were underpowered (1 − β = 0.19 for males and 0.73 for females). There is other research to have illustrated sex-specific associations between the rs5751876 variant and anxiety and emotion processing (Domschke et al., 2012; Gajewska et al., 2013). However, at present there is no insight into how a change in DNA sequence that does not change protein sequence leads to a functional effect. Further work is needed to understand whether the C allele might alter the transcription, stability or translatability of the encoded RNA and hence receptor levels or turnover. Until such insight is available, weak associations between synonymous mutations and function must be viewed with extreme caution.

The minor allele (G) of ARRB2 rs1045280 variant was found to confer a risk for meth use disorder (OR = 1.61) in Japanese people (Ikeda et al., 2007). Subgroup analyses revealed that the variant was only associated with meth use in female participants, although underpowered (1 − β = 0.72). While located in an exon, rs1045280 is a synonymous variant. ARRB2 codes for the beta-arrestin-2 protein, which plays a critical role in modulating dopamine and many other G protein-coupled receptors and downstream behaviors (Beaulieu et al., 2005). Along with rs1045280, Ikeda et al. (2007) found that two other SNPs located downstream of ARRB2 (rs2036657 and rs4790694) were also significantly associated with meth use disorder, though the associations were underpowered (1 − β = 0.52–0.64). In addition, they reported an association with two haplotypes comprising all three variants (Ikeda et al., 2007). In contrast, another report found a divergent trend in a Caucasian sample, with the rs4790694 variant seemingly protective (OR = 0.69; Bousman et al., 2010). As the role of these variants is still unknown, further work is needed to elucidate the potential influence of ARRB2 polymorphisms on meth use disorder, including examination of sex and ethnicity.

A meta-analytic approach was conducted on four markers in genes directly involved in dopamine signaling (DRD2 rs1800497; DRD4 Exon 3 VNTR; COMT rs4680; and SLC6A3 3′ VNTR). None of the variants were associated with meth use disorder (smallest p = 0.09; Fig. 4; Table 5), even after removing the study with the highest risk of bias for each analysis.

4.1.4. Serotonergic system

Seven studies investigated variants in serotonergic-related genes. The only significant and adequately powered allelic association was observed in HTR6, with the minor allele (G) of the rs6693503 variant conferring a risk for meth use disorder (Kishi et al., 2011a). rs6693503 is located ~4 kilobases upstream of HTR6, in the promoter region. The effect of this variant is unknown. In addition, two haplotypes (rs6693503(G)-rs1805054(C) and rs6693503(G)-rs1805054(C)-rs4912138(G)) were found to be more common in people with meth use disorder. rs1805054 is a synonymous variant located in exon 1 of HTR6, whereas rs4912138 is located within an intron. HTR6 codes for the 5HT₆ receptor (Kohen et al., 1996). Activation of the receptor leads to enhanced GABAergic activity, whereas inhibition produces an increase in extracellular levels of dopamine in the rat brain (Lacroix et al., 2004; Schechter et al., 2008).

Three studies investigated the association between the linked polymorphic region (LPR) repeat in the promoter region of SLC6A4 and meth use disorder (Ezaki et al., 2008; Hong et al., 2003; Payer et al., 2012). SLC6A4 codes for the serotonin transporter (SERT), which primarily regulates serotonergic signaling by serotonin reuptake from the synaptic cleft back into the pre-synaptic neuron (Coleman et al., 2016). The short variant of the LPR is associated with reduced gene expression, transporter levels and serotonin uptake in vitro (Greenberg et al., 1999; Lesch et al., 1996). While individually none of the studies reported a significant allelic association, meta-analysis revealed a moderate overall protective effect of the LPR on meth use disorder (OR = 0.78; p = 0.04, Fig. 5B; Table 5). However, the significance did not survive sensitivity analyses, with exclusion of the study with the highest risk of bias resulting in a non-significant result.

Despite the lack of evidence for a role of genetic variations in SLC6A4 in contributing to heritable risk for meth use disorder, the gene is implicated in the pathophysiology of meth addiction. There is evidence for a brain-region specific decrease in serotonergic activity and SERT expression in people with meth use disorder (Kish et al., 2009; Sekine et al., 2006). In addition, Sekine et al. (2006) found that greater aggression was associated with reduced SERT density in some brain regions (Sekine et al., 2006). Future replication studies will allow more definitive conclusions drawn on the involvement of serotonergic genes in meth use disorder.

4.1.5. GABAergic system

Several studies have investigated the potential association between meth use and gamma-aminobutyric acid (GABA) related genes. A report found that the minor allele (C) of rs2236418 variant of GAD2 conferred ~1.90 greater odds of developing meth use disorder (Veerasakul et al., 2017). Notably, the study had a low risk of bias (NHLBI score = 8). GAD2 codes for the shorter 65 kDa isoform of the glutamate decarboxylase enzyme (GAD65), and is widely expressed throughout the brain (Erlander et al., 1991). GAD65 plays a key role in the production of the inhibitory neurotransmitter GABA, and is activated in response to low levels of GABA at the synapse (Fenalti et al., 2007). The rs2236418 variant is located in the 5′ UTR of GAD2 and leads to increased transcriptional activity of GAD65 (Boutin et al., 2003), resulting in greater levels of GABA and GABAergic transmission. In addition to meth use disorder, the variant has been associated with alcohol use disorder in two independent samples of Russian descent, but this was not replicated in European-Americans (Lappalainen et al., 2007).

GABA can bind to two main types of receptors: ionotropic receptors (GABA_AR) and metabotropic receptors (GABA_BR). A study reported an association between meth use and the rs4480617 variant of the gamma2 subunit of GABA_AR (GABRG2) (Lin et al., 2003). The minor allele (G) conferred ~2.92 greater odds of developing meth use disorder. It should be noted that the association was only observed in female participants. While other GABRG2 polymorphisms have been shown to be association with heroin and alcohol use disorder (Li et al., 2014), very little is known of the rs4480617 variant. It is located in the 5′ UTR of GABRG2 and may affect the regulation of protein translation (Table 3). Future research should aim to investigate any effect of the variant on the receptor’s activity to further understand its potential sexually dimorphic role in the vulnerability for meth use disorder.

4.1.6. Glutamatergic system

Of the three studies reporting significant associations between glutamate-related genes and meth use disorder, two of them were adequately powered. The minor allele (T) of the rs2486001 variant of SLC6A9 and the rs2486001(T)-rs2248829(G) haplotype conferred risk for meth use disorder in a Japanese sample (Morita et al., 2008). SLC6A9 codes for the glycine transporter 1 (GlyT-1), which controls glycine neurotransmission and is implicated in modulating the N-methyl--d-aspartate (NMDA) receptor in the brain (Smith et al., 1992). The rs2486001 variant is an intronic variant without known function, whereas rs2248829 is located within exon 7 of SLC6A9. However, rs2248829 is a synonymous variant and has no effect on the structure of the GlyT-1 (Table 3). While silent mutations may have an effect on gene expression and rate of translation (Shabalina et al., 2013), this was never demonstrated for this variant of GLYT1. In addition, it is possible, though not yet demonstrated, that rs2486001 and rs2248829 are in linkage disequilibrium with other functional variants.

Another study reported that two variants (rs12487957 and rs4687771) of GRM2 were protective (ORs = 0.70 and 0.65, respectively), with people with meth use disorder less likely to carry the minor alleles (C and A) compared to controls (Tsunoka et al., 2010). The haplotypes rs3821829(C)-rs12487957(C)-rs4687771(A) were associated with a risk for meth use disorder. This suggests that, while carrying only one of the minor alleles is protective, carrying both rs12487957 and rs4687771 may confer a risk. GRM2 codes for the metabotropic glutamate receptor 2 (mGlu2), and is expressed in neurons throughout the brain (Flor et al., 1995). Activation of mGlu2 generally inhibits glutamate release at synapses (Niswender and Conn, 2010). Neither variants are directly located in GRM2, with rs4687771 and rs12487957 located ~ 2.5 kilobases and ~3.5 kilobases downstream of the gene, respectively. While these variants may modulate gene expression, this has yet to be determined. Preclinical evidence suggests that activation of mGlu2 can reduce reinstatement of meth seeking after extinction and abstinence (Caprioli et al., 2015; Kufahl et al., 2013). It is therefore possible that the minor alleles of rs12487957 and rs4687771 lead to increased activity of mGlu2, and thereby less sensitivity to the behavioral effects of meth.

4.1.7. Endogenous opioid system

The 68-bp repeat in the promoter region of PDYN has been shown to be associated with a 1.83 increase in the risk of developing meth use disorder in a Japanese sample (Nomura et al., 2006a, 2006b). PDYN codes for prodynorphin, a precursor protein that gives rise to dynorphins when cleaved (Day et al., 1998). Dynorphins are endogenous ligands for the kappa opioid receptor, and form part of the endogenous opioid system (Chavkin et al., 1982). Meth can increase expression of PDYN in the rodent brain (Adams et al., 2003; Cadet et al., 2016). The PDYN promoter region can contain one to five 68-bp tandem repeats (Rouault et al., 2011). Three or four copies of the 68-bp repeat increase PDYN expression compared to one or two copies (Zimprich et al., 2000). Another report suggests that changes in expression may be cell specific, with more repeats leading to increased gene expression in certain cell lines, and the reverse in others (Rouault et al., 2011). This highlights a complex relationship between the tandem repeat and PDYN expression. Subgroup analyses revealed that a greater number of tandem repeats were more common in both people who exclusively use meth and people with polysubstance use disorder compared to controls (Table 6),. This suggests a potential shared genetic vulnerability for more than one substance use disorder. This is consistent with studies reporting an association between the number of repeats and heroin use disorder, as well as comorbid alcohol and cocaine use disorders (Williams et al., 2007; Yuanyuan et al., 2018). It is also consistent with a large literature implicating dynorphinergic signaling as a homeostatic, negative-feedback adaptation to chronic stimulant exposure (Carlezon et al., 2005).

4.1.8. Other notable genes

A recent study by Namvar et al. (2020) reported a strong association between the rs1333048 variant (C allele) of ANRIL and meth use disorder (OR = 2.34) (Namvar et al., 2020). rs1333048 is located ~5 kilobases downstream of ANRIL, and its effect on ANRIL expression is still unknown. ANRIL, or antisense non-coding RNA in the INK4 locus, is a non-coding RNA involved in transcriptional silencing of other genes (Yap et al., 2010). Evidence suggests that ANRIL is only functional in higher primates (He et al., 2013), like many other long non-coding RNAs. It is associated with a wide range of disorders, including coronary heart disease, aneurysms, diabetes and several cancers (Pasmant et al., 2011). In their study, Namvar et al. (2020) also found an association between rs1333048 and a risk for bipolar disorders type 1 (OR = 2.33), type 2 (OR = 2.00), and major depressive disorder (OR = 2.42), suggesting a shared risk across neuropsychiatric disorders. The exact mechanism of action of ANRIL is still unknown. Results from a preclinical study suggest that ANRIL decreases apoptosis of hippocampal neurons and promotes memory recovery in a diabetic rat model (Wen et al., 2018). Memory deficits have been well documented in people with meth use disorder (Guerin et al., 2019; Potvin et al., 2018), and polymorphisms in ANRIL may be a contributing factor.

Kishi et al. (2010) reported an association between two variants (rs6085086(A) and rs4815787(A)) and two haplotypes of PROKR2 and a risk for meth use disorder (Kishi et al., 2010). rs6085086 is located within an intron, whereas rs4815787 is within the 3′ UTR of PROKR2 and may modulate gene expression. PROKR2 codes for the prokineticin receptor 2 (PKR2), which primary function in the nervous system is regulation of circadian rhythm (Cheng et al., 2002). While the authors previously reported an association between these variants and mood disorders (Kishi et al., 2009), the biological effect of these SNPs and the role of PKR₂ in meth use disorder remains obscure.

4.2. Limitations

While the evidence reviewed in this study provides some insight into the genetic basis of meth use disorder, prominent limitations should be noted, and any results discussed in the present review should be taken with great caution. Major limitations for most gene markers include lack of independent replication, unclear rationale for candidate gene selection, study population selection, low power, and high risk of bias, especially in the absence of genome-wide significance. It should also be emphasized that the contribution of any of the candidate gene variants studied to date to the heritable risk for meth use disorder is likely to be extremely small. In fact, with the absence of genome-wide significance or robust association in a large family pedigree, demonstrating a functional correlate of a genome sequence variation in one of these candidate genes is better interpreted as a contribution of that gene on the pathophysiology of meth use disorder and not specifically related to its genetic risk.

Of the 75 genes explored, 58 were only explored once. Replication in independent samples is critical to establish the validity of findings in individual studies. Also, in many instances there was no clear rationale for the selection of variants investigated, with ~65 % having no known or poorly understood biological function (Table 3). Further, none of the variants with a significant association have come up on GWASs. This may be because only three meth use disorder GWASs have been conducted to date, with small sample sizes and overlapping samples (Ikeda et al., 2013; Sun et al., 2019; Uhl et al., 2008b), highlighting the need for further research.

The present review also highlights that most studies were conducted in Japanese and Han Chinese samples. While historically China was one of the largest markets for meth production and use, recent evidence suggests that South-East Asia is the fastest growing meth market (UNODC, 2019). In addition, a recent analysis of metabolites in waste-waters revealed high use of meth in East and North-Central Europe, the United States, and Australia (González-Mariño et al., 2020). Preliminary evidence from Bousman et al. (2010) suggests that risk and protection conferred by putative gene variants may be ethnically divergent. It is therefore crucial to investigate the associations between gene and meth use disorder in a wider variety of populations.

Lastly, only a handful of studies were adequately powered to detect differences in genotype and allele distribution (Table 2). This was not surprising, as over 70 % of all studies failed to conduct appropriate power analyses and justify their sample size. In addition to low power, most studies reviewed had a high risk of bias. Over 90 % of all studies did not have clear inclusion and exclusion criteria, and the cases were not clearly differentiated from control in 75 % of studies. These bias risks undermine potential interpretation of the findings, as it is possible that genotype differences may arise from the heterogeneity between cases and controls. In addition, the lack of clear reporting of study design makes it challenging for future replication studies.

4.3. Implications and future directions

The findings summarized in the present review provide some evidence of the involvement of certain genes in meth use disorder. The most promising marker associated with meth use disorder is the functional variant rs324420 (Pro129Thr) of FAAH in people of East Asian descent, with a large overall effect size and studies with the highest quality and lowest risk of bias reviewed. While the exact role of Pro129Thr in susceptibility for meth use is unknown, it is possible that it plays a role in increasing the subjective effect of the drug, reducing impulse control and enhancing reward-related brain activity. Future studies should aim to replicate these findings in other populations. In contrast, while a meta-analysis of seven studies revealed that the rs6265 variant (Va166Met) of BDNF may confer protection against meth use disorder, the effect size was weak. It should be noted that both FAAH Pro128Thr and BDNF Val66Met have been implicated in other substance use and comorbid disorders.

To date, literature on the genetics of meth use disorder almost exclusively focuses on hypothesis-driven candidate gene association studies, and specific heritability scores are yet to be calculated. To uncover the heritability of the disorder, family and twin studies should be conducted, preferably across different populations. In addition, very few GWASs have been conducted in people with meth use disorder (see Limitations), and it appears clear that large-scale GWASs involving international collaborations are necessary to advance the field. Many studies reviewed reported significant associations with silent variants with no clear biological function. It is possible that genes involved in genetic susceptibility may be in linkage disequilibrium with the variants studied. In addition, due to the highly polygenic nature of substance use disorders (Ducci and Goldman, 2012), a well-powered GWAS would not only complement the candidate gene studies, but also identify alleles in linkage disequilibrium and allow for calculation of polygenic scores.

Of the 76 candidate gene association studies reviewed, 70.9 % of them reported a co-morbid diagnosis, with participants also diagnosed with either meth-induced psychosis, polysubstance use disorder, or both (Figure 2B). While substance-induced psychosis occurs in 36.5 % of people who use meth (Lecomte et al., 2018), and polysubstance use is common in this population (Connor et al., 2014), future studies should include a separately analysis of people exclusively diagnosed with meth use disorder to uncover genes specific to meth use.

5. Conclusion

In this study we systematically reviewed gene association studies in people with meth use disorder to identify any promising targets and pathways underlying the disorder. We assessed the quality and risk of bias of each study and calculated the power of each association. In addition, a meta-analytic approach was conducted when three or more studies were available for any given marker. We found that most significant associations were identified in neurotransmitter-related genes, with functional variants in FAAH and BDNF being the most robust targets. These findings highlight the utility of candidate gene studies conducted by different laboratories. We also identified key limitations in current literature. A large number of associations were underpowered to detect appropriate effect sizes, and quality assessment revealed that most studies had a high risk of bias, with poorly defined and differentiated cases and controls in nearly three quarters of them. Rationale for candidate gene selection was also poor, with only a small number of variants located in coding region of genes of interest. In addition, very few studies explored association in samples of non-Asian descent.