Advances in pharmacogenetic research within OUD populations have been on the rise. Yet, no attempt has been made in quantitatively and qualitatively analyzing the literature and critiquing the quality of evidence reported by GWASs. This systematic review was able to summarize findings from GWASs with borderline genome-wide significance and the potential of being replicable in future studies. We have identified five studies that match our inclusion criteria, with three studies reporting significant results. SNPs associated with outcomes of daily heroin injection, methadone dose, and methadone and EDDP plasma concentration were found to be significant. SNPs corresponding to genetic regions of CNIH3 were reported to be more prevalent in daily heroin injecting patients. SNPs corresponding to or near OPRM1, TRIB2, ZNF146, and EYS were associated with methadone dose levels, depending on ethnicity. SNPs in an intergenic region on chromosome 9, SPON1, and an intergenic region on chromosome 3 were associated with differing plasma concentration of R-methadone, S-methadone, and R-EDDP, respectively. The quality of research and reporting of each study was assessed with the Q-Genie tool and no study was deemed to be of poor quality. Varying sample sizes were however observed, with some being too small for what is considered acceptable for GWAS analysis. With sample sizes of thousands required to produce adequately powered results in GWASs [26], sample sizes from Yang et al. (n = 344) and the African American population of Smith et al. (n = 383) fell short.

One gene related to the SNPs identified has been reported previously within candidate gene studies and has an established biological relevance within the genetics and pharmacogenetics of OUD research. The OPRM1 gene encodes the mu-opioid receptor, which binds endogenous and exogenous opioids [27]. Genetic variability in OPRM1 has been reported to have biological effects on the mu-opioid receptor function contributing to complex disorders. An in-vitro study showed that the OPRM1-G118 variant reduces OPRM1 mRNA and protein levels [28]. When studied in mice models, the equivalent point mutation OPRM1-G112 also resulted in decreased mu-opioid receptor mRNA and protein expression [29]. Findings showed that mice with the G112 allele had reduced morphine-induced antinociceptive responses [29]. Consistently, OPRM1 has been reported to be highly influential in opioid dependency, and, by some findings, OUD treatment outcomes, such as methadone dose and plasma concentrations, in European patients [30]. Therefore, it is not a surprise for SNPs in this gene to be associated with methadone dose at a GWAS significance level. Though, Smith et al.’s results are interesting because they found an OPRM1 association in patients of African American ethnicity but not of European ethnicity, as was expected. This incongruity calls for additional powered research in both ethnic populations to be conducted for a consensus.

Another gene identified has not been previously associated with OUD or MOUD outcomes in the literature but could be involved in biological pathways relevant to opioid use. The CNIH3 gene encodes the protein cornichon homolog 3, which regulates AMPA receptor trafficking [27]. This gene has been identified in schizophrenia studies by NCBI’s Gene database [31]. Therefore, it is possible that CNIH3 could be associated with the regulation of opioid use.

Most of the genes involving an identified SNP summarized in this systematic review do not seem to have been relevant to OUD or MOUD outcomes, nor could a biological relevance be identified for them. These genes include TRIB2, ZNF146, EYS, SPON1, as well as the intergenic regions for the SNPs located on chromosomes 3 and 9. The TRIB2 gene encodes the tribbles homolog 2 protein that regulates MAP kinase proteins’ activation [27]. This gene is evident in many tissues, most prominently in the ovaries, spleen, and nymph node tissues [31]. It has also been reported in the NCBI Gene Database to be identified in studies researching schizophrenia, neuropsychiatric disorders, autism, and aging [31]. ZNF146 encodes the zinc finger protein OZF, the primary function of which is to regulate DNA binding and transcription [27]. As such, it is present in a lot of tissues, including the brain, but is more prominent in the endometrium and thyroid [31]. In humans, EYS encodes the protein eyes shut homolog, which as deduced from the name, is involved in vision, more specifically, in maintaining the morphological integrity of photoreceptor cells through the possible involvement in channel regulations [27]. EYS is most prevalently expressed in fat and testis tissue [31], which shows no direct relation to methadone dose or metabolism as identified in Smith et al. Lastly, SPON1 encodes spondin-1, which is a cell adhesion protein within the nervous system [27]. SPON1 is mostly expressed in the gall bladder tissue [31], which does not provide a clear biological link to its function nor the outcome of methadone plasma concentration reported by Yang et al. [31]. Further research is required to make any conclusive statements concerning the biological relevance of SNPs in these genes to the observed MOUD outcomes.

In general, the results of this systematic review are able to inform future candidate gene studies and GWASs of key SNPs that require further research in larger cohorts as well as replications to solidify their associations to MOUD outcomes in indicated patients. The findings from such studies are able to inform the clinical and pharmacological response to patient doses and drug outcomes for administered MOUD.

Though rigorous, this systematic review has some limitations associated with the strict eligibility criteria predetermined in the protocol. It is important to note that in the process of including studies that were primary GWASs, GWAS meta-analyses were excluded. This could have affected the number, quality, and significance of the findings. An example is the exclusion of the GWAS meta-analysis findings from Nelson et al. that replicated original findings in a larger meta-analyzed sample, highlighting new SNPs that achieved significance (rs10799590, rs12130499, and rs298733) and SNPs that fell below our significance threshold in the process (rs1436175) [21]. However, since most GWAS meta-analyses reported associations using the same study populations and sample data, their inclusion would have made any reported findings redundant. Another limitation could be the exclusion of studies that reported genetic variance in the form of haplotypes. Though their inclusion might have made a meta-analysis possible, they did not satisfy the eligibility criteria of a SNP identified by a GWAS and would, therefore, not be very informative within the scope of our systematic review.

As stated previously, a meta-analysis was not feasible with the heterogeneity of the reported findings. This makes consensus more difficult to reach and the findings less generalizable, especially when considering differing ethnicities.

In addition, this systematic review was only able to highlight published GWAS associations. As a result, any findings that were not published due to inability to meet statistical thresholds might not have been included. Though efforts were made to include near genome-wide significant findings, the possible presence of publication bias should still be acknowledged.