In this human laboratory study, we sought to determine whether the ABCB1 rs2235048 polymorphism would influence the pharmacokinetics of THC and self-reported subjective cannabis effects in young adults after smoking a single cannabis cigarette. Specifically, we hypothesized that the rs2235048 C allele would be associated with lower blood THC and metabolite concentrations, as well as lower subjective responses to smoked cannabis. This hypothesis was not supported, as we found no significant genotype difference in blood THC concentrations or subjective effects and found actually higher metabolite concentrations in C-allele carriers compared to the TT genotype.

Our blood cannabinoid data suggested that the rs2235048 C allele is associated with greater concentrations of THC and metabolites in blood after smoking a single cannabis joint. Though only the AUC of 11-OH-THC and THC-COOH were statistically greater in C-allele carriers compared to the TT genotype group, THC and metabolite concentrations were numerically higher in the CC/TC group at every time point. This finding contrasts with the results of Kebir et al. (2018) [16], where lower THC concentrations were reported in T-allele carriers of the rs1045642 polymorphism, which is linked to the C allele of rs2235048. Kebir et al. suggested that the lower blood THC concentrations in T-allele carriers could be due to the rs1045642 T allele (and by linkage, the rs2235048 C allele) leading to lower P-gp expression, thus resulting in less efflux of THC from the central compartment [16]. However, the participants included in their sample had diagnosed DMS-IV cannabis dependence, were consuming an average of 21 cannabis joints per week, and were on average 29.5 years old [16], which is a very different than our sample of young adults using cannabis on average 2.5 days per week who had never met criteria for cannabis dependence. In addition, in our study, cannabis was administered under placebo-controlled laboratory conditions, and thus our blood cannabinoid data were collected at known, fixed time points following cannabis exposure. In contrast, Kebir et al. sampled blood at a variable time after last personal use of cannabis [16]. Thus, while our findings may seem contradictory, it is possible that variation in the ABCB1 gene impacts THC pharmacokinetics differently depending on the cannabis use history of the population being sampled and on the timing of blood cannabinoid concentration analysis.

In order to account for the potential influence of ethnicity on the relationship between ABCB1 genotype and THC pharmacokinetics, we performed a secondary analysis in the subset of participants identifying as European Caucasian (EC). In the EC subgroup, the effects of rs1045642 genotype on THC and metabolite concentrations were even more pronounced, with AUC of THC, 11-OH-THC, and THC-COOH and Cmax of THC-COOH statistically significantly higher in C-allele carriers. These findings should be interpreted with caution as there were only 4 participants in the TT group in this subgroup. However, these results suggest that ethnicity is an important variable to consider in future work characterizing the role of ABCB1 in THC pharmacokinetics.

While our data suggest a potential association between rs2235048 genotype and THC pharmacokinetics, it is important to note that the C-allele carriers in our study did have a significantly greater mean weekly frequency of cannabis use (2.74 days/week, compared to 1.85 days/week in the TT genotype group). In addition, while the difference was not statistically significant, C-allele carriers had a greater estimated dose of THC in the ad libitum paradigm than the TT genotype group (84.18 mg compared to 76.60 mg). Thus, in lieu of a pharmacokinetic explanation for the genotype difference in blood cannabinoid concentrations, it is possible that C-allele carriers, on average, typically used cannabis with greater frequency and intensity than the TT group, and so chose to smoke more of the cannabis cigarette during the experimental session.

Across the seven subjective effects scales, we saw minimal evidence of meaningful genotype differences. We did see significantly higher AUC for the “liking” measure in C-allele carriers, but this is likely to be a spurious finding since we did not see a consistent pattern in subjective effects disaggregated by genotype as we have observed in other studies, e.g., associations with OPRM1 polymorphisms [8]. No previous studies were identified that examined ABCB1 influences on acute subjective cannabis effects. The most relevant finding is the association between the rs1045642 C allele and greater risk of cannabis dependence [15]. In their follow-up study that found greater THC concentrations in C-allele carriers, Kebir et al. [16] suggested that greater efflux of THC from the central compartment in C-allele carriers could encourage more frequent use of cannabis to maintain a drug high, and could also cause greater severity of cannabis withdrawal symptoms in individuals with a CUD. It could be that the putative association between ABCB1 polymorphisms and CUD is primarily related to pharmacokinetic, and not pharmacodynamic, differences, which would explain our finding of no genotype differences in self-reported subjective drug effects.

While these preliminary data support a role of ABCB1 genetic variation in moderating cannabis-related phenotypes, there are a few important limitations to highlight. We were limited by a relatively small sample size. We hope to address this limitation in future work as we recruit larger samples and can replicate and expand the current results (e.g., examine the influence of other ABCB1 polymorphisms, which could help to clarify inconsistencies between our results and previous findings). We believe our laboratory-based candidate gene approach is an important complement to population-based genetic designs such as GWAS, given our ability to measure a variety of well-described cannabis-related phenotypes such as blood cannabinoid concentrations and subjective effects. Replicating our findings in a larger sample could also allow us to address issues of confounding. In the present sample, we saw a higher mean frequency of cannabis use in C-allele carriers, which could be confounding the genotype difference we saw in blood cannabinoid concentrations. Thus, in a larger sample, we could control for group differences in cannabis use patterns. We were also limited in our capacity to investigate ancestry influences on our genetic association, as we had to dichotomize our ethnicity variable. There are known ethnicity/ancestry differences in rs1045642 allele frequency, e.g., T allele frequency of 56.1% in Caucasians, 20.2% in African Americans, 40% in Asian Americans, and 50% in both Mexican Americans and Pacific Islanders [23]. The study was also limited by the use of rs2253048 as a proxy for rs1045642 (as this SNP was not assayed by the genotyping array). Although these SNPs are in linkage disequilibrium, the correlation between alleles is not 100%. In other words, the rs2235048 C allele may not exactly correspond to rs1045642 T allele. Furthermore, there are other ABCB1 SNPs in linkage disequilibrium with rs1045642, and thus we cannot be sure of the true causal variant [17].

In conclusion, though our findings were mostly negative, we have provided the first preliminary data from a human laboratory paradigm suggesting an association between ABCB1 rs2253048 and acute cannabis measures in young adults, where the C allele was associated with some measures of blood cannabinoid concentrations (but no difference in self-reported subjective drug effects) after smoking a single cannabis cigarette. In addition, this relationship appears to be greater in individuals who self-identify a European Caucasian ethnicity, thus ethnicity is an important variable to consider in future research examining relationships between ABCB1 variation and THC pharmacokinetics. While these data are preliminary and require replication in a larger sample where potential confounds can be controlled, they support a small but growing body of literature that suggests the ABCB1 gene moderates a variety of cannabis-related phenotypes and may be involved in conferring risk to CUD.