Contribution of BDNF and DRD2 genetic polymorphisms to continued opioid use in patients receiving methadone treatment for opioid use disorder: an observational study

We selected the BDNF and DRD2 genes on the basis of evidence supporting their involvement in opioid dependence and addictive behavior. The rs6265 and rs1799978 SNPs were the preferred choices for the purpose of this investigation because of their association with substance use and psychiatric disorders among various clinical populations [33‐35, 50‐52]. We isolated DNA from whole blood and performed genotyping using the Applied Biosystems^® ViiA™ 7 Real-Time PCR System (Life Technologies Corp., Carlsbad, CA, USA) with Applied Biosystems TaqMan Genotyping Master Mix (Life Technologies Corp.), as described previously [49]. The genotype call rates were 97.7 and 99.2 % for BDNF rs6265 and DRD2 rs1799978, respectively.

All participants underwent qualitative and semi-quantitative urine analysis weekly/biweekly using the iMDx™ Analyzer and Prep Assay (NOVX Systems Inc., Richmond Hill, ON, Canada). The urine toxicology assays were implemented as part of the treatment model to monitor methadone adherence and to identify use of opioids. The iMDx™ test can differentiate between natural and synthetic opioids, allowing for easier identification of specific opioid use. Urine samples were collected and assayed at the respective methadone clinic sites.

Sample demographics were summarized using descriptive summary measures expressed as mean (standard deviation, SD) for continuous variables and number (percent) for categorical variables. Genotype and allele frequencies were computed and tested for Hardy–Weinberg equilibrium.

We performed univariate analysis on sample characteristics to evaluate differences between responders and nonresponders. Student’s t test was used for mean differences, and Chi square was used for categorical variables. We chose to use the results from these comparisons and include significant variables as covariates in our logistic regression model. We performed multivariable logistic regression analysis, with opioid use as the binary dependent variable and the two genetic variants, BDNF rs6265 (A/A vs. A/G vs. G/G) and DRD2 rs1799978 (G/A vs. A/A), as independent categorical variables adjusting for age, sex, ethnicity, methadone dose (mg), and duration of treatment (months). We also adjusted for the total number of urine screens to eliminate any effect of more frequent urine sampling, suggesting problematic behavior throughout treatment (change in outcome per opioid screen). We classified continued opioid use as having <80 % negative opioid urine screens (treatment nonresponders). This classification was based on data from our current sample and from previous literature demonstrating that 30–80 % of opioid urine screens generally test negative throughout the course of methadone treatment [53‐55]. Given the maximum value of this range, individuals with greater than 80 % negative screens (or alternatively, less than 20 % positive screens) are considered to be in good standing and, therefore, responding well to treatment. Given that 85 % of the participants were of self-reported European origin, we did not perform subgroup analyses based on ethnicity due to the small sample size of other ethnic groups in our study; this ethnic distribution is in keeping with our region population mix. In our regression, participants of European origin were compared to non-European origin, and men were compared to women.

Regression results, including model coefficients (odds ratio, OR), corresponding CIs, and associated p values are reported. The criterion for statistical significance was set at alpha = 0.05. There were no missing data in our analyses. We performed all statistics using STATA Version 12 (StataCorp LP, College Station, USA). The study is reported in adherence with the Strengthening the Reporting of Observational Studies in Epidemiology statement [56].

We confirmed the statistical power of this investigation post hoc using Quanto Version 1.2.4 (Morrison & Gauderman 2009, California, USA), with treatment response (continued opioid use) as the outcome variable. Using an additive, gene-only, unmatched case–control (1:2) model, including 240 methadone patients and a two-sided test with a p = 0.05 level of significance, we had 85 % power to evaluate the effect of BDNF rs6265 (minor allele frequency, MAF: 0.20) on methadone treatment response, with an OR of 1.5; and 70 % power to examine the effect of DRD2 rs1799978 (MAF: 0.05) on methadone treatment response, with an OR of 1.8.

Of the initial 260 participants recruited from methadone clinics, 20 participants were excluded from the study (duplicate entries = 5, buprenorphine treatment = 3, missing blood sample or urine data = 8, being prescribed opioids for chronic pain condition = 4). Therefore, 240 participants in total were included in the analysis (Fig. 1). The sample consisted of 144 (60.0 %) men and 96 (40.0 %) women, with a total mean age of 37.1 (SD = 10.4). Participants of European ethnicity made up 85 % of the sample. A majority of participants (81.3 %; n = 195) reported having a family history of mental illness or addiction. Responders and nonresponders were comparable across the majority of factors. Additional details of sample characteristics are shown in Table 1.

Genotype frequencies for rs6265 and rs1799978 are presented in Table 1. They did not deviate significantly from Hardy–Weinberg equilibrium (p = 0.21 for rs6265; p = 0.36 for rs1799978), and MAFs were consistent with previous literature (0.20 for rs6265 and 0.05 for rs1799978). Among our sample of 240 methadone patients, 36.3 % (n = 87) had at least one rs6265 risk allele, and 11.3 % (n = 27) had at least one rs1799978 risk allele.

The continued use of opioids during methadone treatment was an indication of treatment nonresponse and can be measured objectively across samples. This allowed us to examine whether there was a genetic component to outcomes of methadone treatment. On average, 18.9 % (SD 24.1) of total urine screens throughout the duration of methadone treatment were positive for opioids in the total sample. Similar patterns were observed among genotype frequencies of responders and nonresponders (Table 1). Our logistic regression analysis showed that the minor alleles of BDNF rs6265 and DRD2 rs1799978 were not associated with continued opioid use during methadone treatment, while adjusting for age, sex, ethnicity, age of initial opioid use, methadone dose, duration of treatment, and total number of opioid urine screens (rs6265: OR 1.37, 95 % CI 0.792, 2.371, p = 0.260; rs1799978: OR 1.28, 95 % CI 0.511, 3.182, p = 0.603) (Table 2).

In this study, we explored the effect of BDNF rs6265 and DRD2 rs1799978 polymorphisms on an important methadone treatment outcome, continued opioid use, which represents an objective measurement of response to treatment. In our sample of 240 methadone patients of primarily European origin, we were unable to confirm a role for these specific SNPs in continued opioid use during treatment.

Our findings are in line with a study by de Cid and colleagues, the only other study to examine the influence of BDNF rs6265 in methadone treatment response [39]. They performed a haplotype analysis of 30 SNPs in the BDNF coding region, including rs6265, in a sample of 91 Caucasian individuals receiving methadone treatment for opioid use disorder. Grouping their sample into responders and nonresponders, they were unable to establish an effect of rs6265 on response to methadone treatment [39], but found that a haplotype block containing this specific SNP appeared more frequently in nonresponders compared with responders. However, the generalizability of these findings is limited by small sample size, large CIs, and short period of urinalysis testing (previous four urine screens, or approximately 1 month) [39]. With respect to other methadone outcomes, another study demonstrated no effect of rs6265 on methadone dose in a sample of 227 former heroin-dependent individuals in methadone treatment [38].

The rs1799978 SNP of DRD2 has only been examined in association with opioid dependence or methadone dose in two single-SNP and haplotype analyses. In both studies (by Hung et al. [40]; Doehring et al. [43]), the minor allele of rs1799978 was more common in opioid users. However, Hung et al. [40] demonstrated that carriers of the minor llele also required higher methadone doses in their sample of 321 methadone patients, which was not a consistent finding in the study by Doehring and colleagues for a sample of 85 German drug users [43], thus suggesting a potential ancestral influence of rs1799978 on methadone dose.

Although methadone dose was a significant predictor of continued opioid use in our regression analysis, this was an expected finding given the available evidence on methadone dosing in treatment [16, 57]. We included this variable to ensure that any effect found between the SNPs and continued opioid use was not explained by the relationship between methadone dose and continued opioid use.

Contrary to what the current literature suggests, although there may be a potential role for both BDNF rs6265 and DRD2 rs1799978 in susceptibility to opioid use disorder, the present study shows that these variants do not appear to exert large effects on continued illicit opioid use during treatment with methadone. Treatment response may, however, be influenced by the collective genetic risk conferred through multiple SNPs across several different genes (a polygenic effect). It is also possible that the continued illicit use of opioids during methadone treatment may be a result of other clinically relevant factors (i.e., medical or psychiatric comorbidity, social circumstances, life stressors, etc.).

Given that there is little conclusive evidence to support a genetic impact on methadone treatment, there is a need for well-designed, powerful, genome-wide association studies to identify specific SNPs that are relevant to methadone treatment response. Perhaps with this information, we will be able to simultaneously examine multiple candidate genes to understand the genetic composition of polygenic psychiatric disorders. Implementing the use of a gene score to assess an individual’s genetic load may prove to be a promising approach to predicting and identifying those patients who require closer monitoring or alternate treatment strategies to overcome their continued opioid use.

Future research should focus on investigating these questions in larger samples and in various populations to ensure validity. Furthermore, a thorough examination of other nongenetic determinants of continued opioid use may prove useful for identifying problematic areas that require modifications in treatment delivery. Additionally, the genetic effects of withdrawal symptoms and adverse methadone events may be a promising area of study.

To our knowledge, this is one of few studies to investigate the potential for an allelic effect of BDNF rs6265 and DRD2 rs1799978 on continued illicit opioid use among a large sample of methadone patients. These factors have not been thoroughly investigated in the context of methadone treatment response (as measured objectively through urine toxicology screens), which highlights a novel direction of research in the treatment of opioid use disorder with opioid-agonist treatments. Through this analysis, we aim to stimulate further research into potential polygenic influences on methadone treatment outcomes, as well as to confirm our findings in a larger sample of methadone patients. The uniformity of our cohort, attributed to consistency in delivery of treatment and standard of care across CATC clinics, ensures representativeness of the entire methadone patient population across Ontario, and likely throughout all of Canada. Our objective selection and definition of outcome measurements—specifically, continued opioid use—is also a noteable strength of this study.

Despite our negative findings, this study should be replicated in a larger sample using multiple genes in order to confirm a lack of association between BDNF rs6265 or DRD2 rs1799978 and methadone treatment response. Because the frequencies of the minor alleles were relatively low in our sample, a larger sample size may be required to estimate with confidence the effect of these variants on continued illicit opioid use.

In summary, the present study has demonstrated a lack of association between the two genetic variants (BDNF rs6265 and DRD2 rs1799978) and MMT response, contrary to what previously had been believed about the role of these variants in psychiatric disorders and addictive behavior. Further research with larger samples is needed to re-evaluate this question, as well as to investigate multiple genes simultaneously to assess polygenic effects on susceptibility to poor treatment response. Nevertheless, this study elucidates the potential for other nongenetic determinants that may contribute to continued opioid use during methadone treatment; it also brings attention to further questions regarding the role of genetics in addiction research.