Gender-specific association of functional prodynorphin 68 bp repeats with cannabis exposure in an African American cohort.

BACKGROUND: Cannabis use disorders (CUDs) cause substantial neuropsychiatric morbidity and comorbidity. There is evidence for gender-based differences in CUDs, for instance, a greater prevalence in males than in females. The main active component of cannabis is delta 9-tetrahydrocannabinol (delta 9-THC), a partial agonist of the cannabinoid type 1 receptor. Preclinical studies show that genetic or pharmacological manipulation of the kappa opioid receptor/dynorphin system modulates the effects of delta 9-THC.
METHODS: In this case-control study of adult African Americans (n=476; 206 females, 270 males), we examined the association of the functional prodynorphin 68 bp (PDYN 68 bp) promoter repeats with categorical diagnoses of cannabis dependence (Diagnostic and Statistical Manual of Mental Disorders-IV criteria), as well as with a rapid dimensional measure of maximum lifetime cannabis exposure (the Kreek-McHugh-Schluger-Kellogg cannabis scale).
RESULTS: The PDYN 68 bp genotype (examined as short-short [SS], short-long [SL], or long-long [LL], based on the number of repeats) was not significantly associated with categorical cannabis-dependence diagnoses, either in males or in females. However, in males, the PDYN 68 bp SS+SL genotype was associated with both greater odds of any use of cannabis (p<0.05) and earlier age of first cannabis use, compared to the LL genotype (ie, 15 versus 16.5 years of age; p<0.045). Males in the SS+SL group also had greater odds of high lifetime exposure to cannabis, compared to the LL group (p<0.045). Of interest, none of the aforementioned genetic associations were significant in females.
CONCLUSION: This study provides the first data on how the PDYN 68 bp genotype is associated with gender-specific patterns of exposure to cannabis. Overall, this study shows that PDYN 68 bp polymorphisms affect behaviors involved in early stages of nonmedical cannabis use and potentially lead to increasing self-exposure. These data may eventually lead to improvements in personalized medicine for the prevention and treatment of highly prevalent CUDs and neuropsychiatric comorbidities.

Introduction

Nonmedical use of cannabinoid compounds, obtained from cannabis plants, is associated with considerable morbidity, as well as with comorbidity due to psychiatric and other substance-use disorders.1–3 The main active component of cannabis is delta 9-tetrahydrocannabinol (delta 9-THC), a partial agonist of the cannabinoid type 1 receptor (CB1) receptor (CB1-r).4 It can be hypothesized that the health-related consequences of cannabis use are related to various dimensional aspects of cannabis exposure (eg, the amount, pattern, and duration of use), in addition to environmental and genetic factors.

Among the major neurobiological systems that are known to interact with CB1-r is the kappa opioid receptor (KOR) system, and its cognate ligands, the dynorphins (encoded by the PDYN gene).5 Preclinical studies show that pharmacological and genetic modulation of the kappa receptor and PDYN systems causes robust changes in the behavioral effects of CB1-r ligands, including delta 9-THC.6,7 The kappa receptor/PDYN system is also known to be strongly involved in the mediation of the rewarding and aversive effects of other drugs of abuse, as well as the expression of dysphoria and depression- and anxiety-like effects.8–12 It is also known that dynorphins and other kappa receptor agonists cause a decrease in striatal extracellular dopamine concentrations in vivo.13,14

The human prodynorphin (PDYN) gene consists of four exons. Exons 1 and 2 encode the 5′-untranslated region of the mRNA, while exons 3 and 4 encode the translated region.15 Numerous mutations have been identified in the human PDYN gene, including a tandem repeat polymorphism in the promoter region. This polymorphism consists of one to five tandem repeats of 68 bp length.15,16 The 68 bp repeat has been the subject of several functional analyses and genetic association studies. An in vitro reporter gene expression assay showed that one and two copies of the repeat cause lower drug-induced increases in expression of the reporter gene than do three and four copies.17,18 However, a more recent in vitro expression study using a construct with a longer PDYN promoter sequence,19 as well as in vivo studies, demonstrated a complex relationship between the polymorphisms and PDYN expression, which is influenced by many factors such as brain region, gender, and cell type.18–20

Studies in rats demonstrated gender- and strain-related differences in self-administration of the synthetic CB1-r agonist WIN 55,212-2,21 with female rats showing a higher self-administration than males. A recent review has focused on gender-based differences in terms of specific effects of cannabis, including subjective effects and signs of withdrawal.22

The PDYN 68 bp repeat polymorphism has been investigated extensively in association studies for several drug-dependence phenotypes but with inconsistent results across studies. Several studies showed that three or four copies of the repeat constitute a risk factor for the development of cocaine/alcohol codependence and heroin-dependence diagnoses.16,23,24 Other studies reported no association of this polymorphism with drug use phenotype.17,25,26 However, to our knowledge, there have not been reports on the association of the PDYN 68 bp polymorphism with nonmedical cannabis use.

This study examined the association of variations of the PDYN 68 bp polymorphism with dimensional measures of cannabis exposure in African Americans, based on two reasons: 1) the frequency of the PDYN alleles is different between African American and Caucasian populations;19 2) prior studies reported an association of this PDYN polymorphism with heroin dependence in African Americans only, and not in Caucasians or Hispanics.16,23 Of interest, a recent epidemiological report showed greater odds of moderate or severe cannabis use disorders (CUDs) in African Americans, compared to other groups in the USA, including Caucasians.1 These differences could be due to behavioral, genetic, socioeconomic, environmental, or biomedical factors, or a combination thereof.

Recent trends in psychiatric research and genetic association studies have increasingly focused on “dimensional” phenotypic measures, as opposed to only on categorical diagnoses.27,28 Dimensional measures are those that are based on quantification of specific attributes and may be especially useful for clinical presentations that vary along some form of a continuum.29 One such dimensional measure, studied herein with a rapid instrument (the Kreek–McHugh–Schluger–Kellogg [KMSK] scale), is the “maximum lifetime exposure”, operationally defined as the maximal frequency, amount, and duration of the heaviest use in a person’s lifetime, up to the date of ascertainment.30,31

This study was therefore designed to examine the association of PDYN 68 bp repeats with both categorical diagnoses of cannabis dependence, as well as with dimensional measures of cannabis exposure, in African American men and women.

Methods

Participants

This was a case-control study, with a cohort of sequentially ascertained African American participants, examined as outpatients in a research hospital setting. “Cases” are defined as participants with an addictive disease diagnosis (Diagnostic and Statistical Manual of Mental Disorders-IV [DSM IV] criteria), and “controls” are participants without an addictive disease diagnosis. This study is a part of a project on the association of genetic markers with specific drug addictions. The total sample reported herein consisted of 487 consecutively ascertained African American participants. After exclusion of relatives and subjects with African genetic contribution below a set threshold (<50%), 476 participants, 354 African American cases with cannabis-, alcohol-, cocaine-, or opioid-dependence diagnoses and 122 African American normal volunteers (controls) were used for further analyses (Table 1). The mean age at the time of ascertainment was 44.0 (SD ±10.0) years in the cases and 30.9 (SD ±9.6) years in the controls.

Table 1

Demographics of the African American cohort in the study

Characteristics	Controls	Cases	Total
Total	126	361	487
After exclusion*	122	354	476
Male	50 (41%)	220 (62%)	270 (57%)
Female	72 (59%)	134 (38%)	206 (43%)
Age, years (±SD)	31 ± 9.6	44 ± 10.0	–
Diagnosis groups
Cannabis dependent
Male	–	74 (67%)	110
Female	–	36 (33%)
Alcohol dependent
Male	–	130 (63%)	205
Female	–	75 (37%)
Cocaine dependent
Male	–	141 (64%)	219
Female	–	78 (36%)
Opioid dependent
Male	–	79 (59%)	135
Female	–	56 (41%)

Note:

Exclusion of close relatives and subjects with low African American genome contribution.

Assessment of ancestry contribution, genome sharing, familial relationship, and duplicates in the study cohort

As mentioned earlier, African American ancestry was determined from questionnaires on family origin and confirmed by analysis using the software Structure 2.232 of 155 ancestral informative markers, as described previously.33 Five self-reported African American participants with <50% African ancestry contribution were excluded. Familial relationships were assessed via pairwise identity by descent analyses using Plink version 1.9 software, as part of a different study.33 Six participants with >20% genome sharing (PI_HAT >0.2) with other participants in this cohort were excluded, as they were relatives.

Recruitment, inclusion and exclusion criteria

Participants were recruited through posted notices and newspaper advertisements in the New York City area; these notices and the overall protocol were approved by the Rockefeller University Hospital Institutional Review Board.

Study inclusion criteria

Participants had to be ≥18 years of age as well as competent enough to understand procedures and to sign and understand the informed consent in English.

Study exclusion criteria

Participants with uncontrolled schizophrenia or other psychotic disorders, which prevented them from understanding study procedures or informed consent, were excluded.34

Participants were excluded from the control category if they had any DSM IV addictive disease diagnosis (of abuse or dependence) or any of the following: 1) current or continuing abuse of alcohol, or at least one instance of drinking to intoxication during the previous 30 days; 2) any use of illicit drugs, including opiates, cocaine, and amphetamines (but excluding cannabis), during the 30 days prior to ascertainment; 3) if they had used illicit drugs (with the exception of cannabis) for at least three times a week for a period of at least 1 month, in their lifetime; and 4) if they had used cannabis on >12 days in the 30 days prior to ascertainment.34

As mentioned earlier, participants were also recruited through postings at several addictive disease clinics in New York City. Cases with addictive diseases were also ascertained as part of community recruitment in the same urban area. All ascertainments were completed during a standardized face-to-face interview with a trained licensed PhD psychologist, medical doctor, nurse practitioner, or registered nurse.

Measures

Structured Clinical Interview for DSM IV Axis I Disorders semistructured interview for DSM IV criteria35 and the KMSK scales for maximal lifetime exposure to specific drugs (focusing on cannabis in this study) were used.30 The cannabis KMSK scale is an ordinal integer measure of maximal self-exposure (scale range: 0–14) and can be rapidly determined in a clinical interview. A KMSK score of zero indicates that the participant had no lifetime use of cannabis, and this score increases ordinally up to a score of 14. The cannabis KMSK score is the composite sum of three subscores, characterizing the time in a participant’s life when use was the heaviest. These subscores are as follows: a) frequency of use (eg, per day or week; score range: 0–6); b) duration of use (eg, in months or years; score range: 0–3); and c) amount of use in a sitting (eg, in “joints”; score range: 0–5). The full text of the scale can be freely accessed at: http://lab.rockefeller.edu/kreek/kmsk. The scale has been used to measure drug exposure in different clinical populations and also as a dimensional phenotype in genetic association studies.31,36,37 A recent recalculation of the receiver operating characteristic (ROC) curve was carried out for a large sample in this urban area, for cannabis KMSK scores and the DSM IV cannabis-dependence diagnosis. This ROC analysis resulted in a cannabis KMSK score ≥10 as the “cut point” for optimal concurrent validity (unpublished data). The KMSK questionnaires also collected the age of first use and the age of onset of heaviest use for a substance (in whole years).

Genotyping of the PDYN 68 bp tandem repeat polymorphism

Genomic DNA was extracted from peripheral blood lymphocytes using a salt precipitation-based DNA extraction procedure. Approximately 100 ng of genomic DNA was amplified by polymerase chain reaction (PCR) using oligonucleotide primers for the promoter region of the prodynorphin gene. The primer sequences were as follows: forward primer 5′-CTGTGTATGGAGAGGCTGAGT-3′; and reverse primer 5′-AGGCGGTTAGGTAGAGTTGTC-3′; these flank the PDYN 68 bp tandem repeat region. A step-down PCR was performed for 35 cycles of 30 seconds at 94°C, 30 seconds at 66°C, 30 seconds at 63°C, 30 seconds at 60°C, 30 seconds at 57°C, 30 seconds at 52°C, and 30 seconds at 72°C, followed by 6 minutes at 72°C using an AmpliTaq Gold kit (Applied Biosystems, Foster City, CA, USA). PCR products were electrophoresed on a 2.5% agarose gel. PDYN genotypes were determined according to the size and number of PCR DNA fragments.

The lengths of the 68 bp tandem repeat PCR DNA fragments are expected to be 273 bp (one copy of the repeat), 341 bp (two copies), 406 bp (three copies), and 473 bp (four copies). For further analysis, genotypes were grouped as short/short “SS” (1,1; 1,2; 2,2 copies), short/long “SL” (1,3; 1,4; 2,3; 2,4 copies), and long/long “LL” (3,3; 3,4; 4,4 copies) repeat alleles.16 Several prior studies have used an “L recessive” model in genotype analyses for the PDYN 68 bp, ie, SS+SL versus LL.38

Statistical analyses

Contingency analysis for categorical diagnoses of cannabis dependence and PDYN genotype were carried out with Fisher’s exact test. Cannabis KMSK scores (scale range: 0–14) were analyzed with Mann–Whitney U-tests, as they were not normally distributed. Cannabis KMSK scores were also cumulated into three simple “bins”, divided into scores of 0–4, 5–9, and 10–14. These “bins” thus provide a simplified ordinal measure of cannabis exposure as “low”, “medium”, and “large”, respectively, and were examined with chi-square analysis. Moreover, the “high”-exposure bin contains the participants with cannabis KMSK scores greater than or equal to the cut point for optimal concurrent validity with the DSM IV cannabis-dependence diagnosis (ie, cannabis KMSK scores ≥10; mentioned earlier). Logistic regressions were used to examine the ORs of cannabis KMSK scores in the “high-exposure” bin, using genotype as a predictor, with and without gender as covariate. All statistical analyses were carried out with Prism (Graphpad Software, Inc., La Jolla, CA, USA) or with Statistica (Tibco) software. The alpha level of significance was set at p≤0.05.

Results

Participant demographics

All 476 participants were African Americans, 270 males and 206 females, sequentially ascertained from April 4, 2002 to August 1, 2013. The demographic data of the cohort are shown in Table 1.

Cannabis KMSK scores in cases with cannabis dependence versus controls

As expected, cases with cannabis-dependence diagnoses (total n=110, KMSK data available from n=106) had higher cannabis KMSK scores than controls (normal volunteers, n=122). Median cannabis KMSK scores for cases with cannabis dependence were 14 (interquartile range [IQR]: 12–14) and for controls, they were zero (IQR: 0–0) (Mann–Whitney U=304.5; p<0.0001).

Ages of onset of heaviest use of cannabis and other drugs, in cases with cannabis dependence

In the cases with cannabis-dependence diagnosis, the age of onset of heaviest use of cannabis was in adolescence (mean: 16.8 years [95% confidence limits {CL}: ±2.03 years]), and preceded that for the other major drugs studied in this cohort (ie, alcohol, cocaine, and heroin), which occurred in early adulthood, from age 21 years onward (Friedman’s analysis of variance [ANOVA] statistic=72.83; p<0.0001 and Dunn’s post hoc tests; data not shown).

PDYN genotype and cannabis dependence as a categorical diagnosis

Genotypic frequencies in the control group (n=122) were 44% for LL and 22% for SS genotypes and, in the cannabis-dependent group (n=110), the frequencies were 36% and 29%, respectively. Chi-square analyses conducted either for the total group, or separately for men and women, did not show significant association of the PDYN 68 bp repeat genotype with cannabis-dependence diagnosis (Table S1).

Cannabis KMSK scores in males versus females

In the whole cohort, the median cannabis KMSK score was higher for males (median=11, IQR: 4–14; total n=270 with cases and controls combined; KMSK data available for n=268) than for females (median=3, IQR: 0–12; total n=206; KMSK data available for n=200) (Mann–Whitney U=18,659; p<0.0001). All the following genetic association analyses were therefore carried out after gender stratification.

Age of first use of cannabis and the PDYN 68 bp genotype in males

Using an “L recessive” model, we analyzed the ages of first use of cannabis in males with the SS+SL genotypes, versus the LL genotype, using a survival analysis. A log-rank test (Mantel–Cox) was significant (p<0.045), and the analysis indicated that the median age of cannabis first use for male participants with the SS+SL genotype was earlier than for those with the LL genotype (15 years versus 16.5 years of age) (Figure 1). The survival curves indicated that the occurrence of first use of cannabis after age of 25 years was uncommon for either genotype.

Figure 1

Survival curves for age of first use of cannabis, in all participants with the PDYN SS+SL genotype, compared to the LL genotype.

Note: Data are stratified by gender: (A) males; (B) females.

Abbreviations: LL, long–long; NS, not significant; PDYN, prodynorphin; SL, short–long; SS, short–short.

Age of onset of heaviest use of cannabis and PDYN 68 bp genotype in males

Using a similar rationale, we then compared the age of onset of heaviest cannabis use across genotypes. The result of this analysis was nonsignificant (not shown).

Genotype analysis of male participants who had never used cannabis, versus those who had used cannabis

We analyzed the PDYN genotype for participants who had never used cannabis (ie, cannabis KMSK score=0) versus those who had at least some use of cannabis in their life time (ie, cannabis KMSK score 1–14). A Fisher’s exact test for this analysis was significant (p<0.05), showing a greater proportion of participants who had ever used cannabis in the SS+SL versus the LL genotype (Figure 2).

Figure 2

Number of participants who reported any lifetime use of cannabis, versus no lifetime use of cannabis, in the PDYN SS+SL genotype, compared to the LL genotype.

Note: Data are stratified by gender: (A) males; (B) females.

Abbreviations: LL, long–long; NS, not significant; PDYN, prodynorphin; SL, short–long; SS, short–short.

“Low”, “medium”, and “high” cannabis exposure bins and the PDYN genotype in males

We divided the cannabis KMSK scale into three bins, denoting “low”, “medium”, and “high” exposure (ie, scores 0–4, 5–9, and 10–14, respectively). As mentioned earlier, the high-exposure bin (KMSK scores 10–14) also contains all scores greater than or equal to the cut point for optimal concurrent validity with the cannabis-dependence diagnosis. A contingency analysis of these three cannabis exposure bins and the PDYN genotype (ie, SS+SL versus LL) was significant; χ2=6.08 (p<0.05) (Figure 3). The proportion of participants in the high cannabis exposure group was greater in the SS+SL versus LL genotype.

Figure 3

Number of participants in the “low”, “medium”, and “high” cannabis exposure bins, in the PDYN SS+SL genotype, compared to the LL genotype.

Notes: The bins were composed of participants with cannabis KMSK scores of 0–4, 5–9, and 10–14, respectively. The “high”-exposure bin contains all cannabis KMSK scores greater than or equal to the cut point for optimal concurrent validity with the DSM IV cannabis-dependence diagnosis. Data are stratified by gender: (A) males; (B) females.

Abbreviations: DSM IV, Diagnostic and Statistical Manual of Mental Disorders–IV; KMSK, Kreek–McHugh–Schluger–Kellogg; LL, long–long; NS, not significant; PDYN, prodynorphin; SL, short–long; SS, short–short.

Genotype analyses in females

Similar genotype analyses were carried out in females. None of these analyses reached significance when examining the female participants in this cohort (Figures 1–3).

Logistic regressions

The SS+SL genotype was a positive predictor of odds of having a cannabis KMSK score in the “high”-exposure bin, compared to the LL genotype as the reference group (OR: 1.47 [95% CL: 1.01–2.13]; p<0.045). However, this effect of genotype was not significant when gender was included as a covariate (p=0.142; not significant [ns]). In this regression analysis, males had greater odds of a cannabis KMSK score in the “high”-exposure bin, compared to females as the reference group (OR: 2.76 [95% CL: 1.90–4.03]; p<0.0001).

Discussion

We found that the PDYN 68 bp repeat genotype was not associated with odds of categorical diagnoses of cannabis dependence, in either males or females, in this cohort composed of African Americans. However, we found that in males, the PDYN 68 bp genotype was associated with a specific pattern of cannabis first use and of maximum lifetime exposure. Thus, male participants with the LL genotype had a lower probability of any use of cannabis, compared to those with the SS+SL genotype. Furthermore, when cannabis use did commence in males with the LL genotype, it did so at a later age, compared to those with the SS+SL genotype. We also found that males with the LL genotype had a decreased probability of reaching “high” cannabis lifetime exposure, compared to those with the SS+SL genotype. Studies have shown that earlier first use of cannabis is associated with greater risk of heavy use of this substance.39,40 Of note, none of the above genetic associations were detected in females. This study therefore provides the first data on genetic associations of the functional 68 bp PDYN genotype, suggesting that this genotype influences the early stages of cannabis use in males and, potentially, the progression to heavy exposure.41–43

As mentioned earlier, dynorphins are endogenous KOR agonists and have countermodulatory roles on dopaminergic neurotransmission, a major neurobiological mechanism underlying the rewarding effects of drugs of abuse (including cannabis) and natural reinforcers.13,44–48 KOR/dynorphin systems also mediate other behavioral and mood functions,8,49 which could also underlie particular stages of cannabis use. In vitro and postmortem brain studies have demonstrated an effect of PDYN 68 bp repeat polymorphisms on PDYN mRNA levels.17–20 Earlier in vitro studies have shown that gene reporter constructs containing three or four copies of this 68 bp tandem repeat (compared to one or two copies) led to higher expression of the reporter gene in mouse neuroblastoma cells (NG108-15) when stimulated by the phorbol ester 12-O-tetradecanoylphorbol-13-acetate.17 Subsequently, we found cell-specific effects of the 68 bp repeats on the transcriptional activity of the PDYN promoter.19 Specifically, in human neuronal SK-N-SH and H69 cells, three or four repeats led to lower expression of luciferase than did one or two repeats. The opposite effect was found in the kidney-derived HEK293 cells.19 Interestingly, in previous studies, three or four copies of the PDYN repeats were found to be a risk factor for the development of cocaine-dependence diagnoses and dual cocaine/alcohol-dependence diagnoses.16,23 In contrast, in this study, we found that one or two copies of the repeats (SS genotype), presumably yielding higher levels of endogenous dynorphin peptides, are associated with increased probability of any use of cannabis, as well as earlier first use of cannabis, in males. A different genetic association study found that persons with one or two copies of the repeats exhibited an intermediate phenotype that included greater propensity for disinhibited behavior.38 Based on the latter study, it may be hypothesized that disinhibited behavior in the SS genotype could manifest as increased probability of any cannabis use, or earlier first use of cannabis, as found herein. Alternatively, it may be hypothesized that high PDYN expression can result in changes in mood and dysphoria, in addition to potentially increasing the risk of early cannabis use and more regular cannabis use.

Methodological considerations and limitations

This study had a relatively small cohort; however, this limitation may be mitigated by the fact that all participants were ascertained in a homogeneous setting, in face–to-face interviews with trained licensed clinicians. We also note that, as is common in this clinical field, many participants also had other DSM IV substance-dependence diagnoses, in addition to cannabis dependence.50 However, we found that the age of onset of heaviest cannabis use (in adolescence) preceded that for alcohol, cocaine, or heroin in this cohort (from young adulthood onward). Therefore, genetic associations to cannabis exposure found herein were unlikely to be confounded by heavy use of these other drugs, since this occurred later in the life span. Overall, the KMSK scores characterize the period in a participant’s life when use of a specific drug is heaviest, and this occurred for cannabis prior than for any of the other major drugs used by cases in this cohort. The age at ascertainment of cases was older than that of controls herein (44 and 30.9 years, respectively). However, herein, we found that first use of cannabis was uncommon after 25 years of age. Therefore, it is unlikely that the observed age-related differences between cases and controls would have resulted in miscategorizing of individuals who would have developed a diagnosable CUD later in their life span.

All the present associations were found in males but not in females, and this suggests that there are gender-specific factors that play a role in the impact of this functional PDYN polymorphism on cannabis use. It cannot be excluded that this lack of association in females was due to the lower number of female-versus-male cases with cannabis dependence found herein. Sex-specific effects of the KOR/PDYN and CB1-r receptor systems have been reported preclinically.49,51

Conclusion

This is the first study to examine the impact of functional PDYN 68 bp repeat polymorphisms on nonmedical use of cannabis. We did not detect genetic associations with categorical diagnoses of cannabis dependence, but only with specific aspects of cannabis use, including first use, and differential levels of exposure. Furthermore, in our study, these associations were observed only in males. This suggests that these PDYN 68 bp polymorphisms may affect behaviors involved in early stages of nonmedical cannabis use and potentially lead to increasing self-exposure. These data may eventually lead to improvements in personalized medicine for the prevention and treatment of highly prevalent CUDs.1,3

Table S1

Chi-square test on grouped PDYN genotype distribution in total and gender-specific African American groups

A)	Total cannabis-dependent cohort, n=110			Control, n=122			OR	p-value
	SS	SL	LL	SS	SL	LL
Genotype	32	38	40	27	42	53	1.18 (SS+SL) vs LL	NS
Frequency	29%	35%	36%	22%	34%	44%
B)	Male cannabis-dependent subsample, n=74			Control, n=50			OR	p-value
	SS	SL	LL	SS	SL	LL
Genotype	22	28	24	14	16	20	1.39 (SS+SL) vs LL	NS
Frequency	30%	38%	32%	28%	32%	40%
C)	Female cannabis-dependent subsample, n=36			Control, n=72			OR	p-value
	SS	SL	LL	SS	SL	LL
Genotype	10	10	16	13	26	33	1.06 (SS+SL) vs LL	NS
Frequency	28%	28%	44%	18%	36%	46%

Note: p-value was determined by the Fisher’s exact test.

Abbreviations: LL, long–long; NS, not significant; PDYN, prodynorphin; SL, short–long; SS, short–short.