Peripheral Expression of Dopamine Genes as a Biomarker for Cocaine Addiction in Humans

Abstract

Introduction

Objective measures for the severity of addiction are needed to identify and appropriately stratify patients into effective treatment groups for intervention and recovery. Our aim was to determine if blood-derived, dopamine-related biomarkers hold predictive validity in patients with cocaine use disorder (CUD).

Methods

A total of 349 participants (163 patients with moderate-to-severe CUD and 186 healthy control participants without a history of substance use disorder) were recruited for a cross-sectional case-control study in the East of England, UK. Data on individual differences in CUD-related traits, e.g., impulsivity, sensation-seeking, were collected. Measurements included mRNA transcription of dopamine-related genes (DRD2, DRD3, DRD4, and COMT) and risk alleles (rs6277, rs1800497, rs12364283, rs6280, and rs4680).

Results

CUD patients showed elevated DRD2 mRNA in peripheral blood (B₀ = 0.43, 95% CI [0.13, 0.73], p_FDR = 0.020) and the DRD2-related rs6277 risk allele was associated with CUD diagnosis (odds ratio = 2.16, 95% CI [1.15, 4.17], p = 0.02). DRD2 mRNA levels related to self-reported impulsivity (B₀ = −0.17, 95% CI [−0.32, −0.03], p = 0.013).

Conclusions

Blood-based expression of DRD2 may be a useful biomarker related to CUD diagnosis and related traits.

Introduction

Changes in fronto-striatal dopamine transmission are closely linked with the psychopathology of cocaine addiction [1]. Converging evidence indicates reduced striatal dopamine D_2/3 receptor availability underpins increased impulsivity [2] and sensation-seeking [3], both hallmarks of cocaine addiction [4]. Since reduced dopamine neurotransmission is widely regarded as both a risk and a consequence of cocaine addiction [1], dopaminergic abnormalities may be a viable biomarker for cocaine use disorder (CUD).

Here, we sought to investigate whether peripheral dopamine-related biomarkers from blood samples reflect altered dopaminergic function in patients with CUD, which could provide a low-cost diagnostic alternative to, e.g., neuroimaging techniques. We examined the relative expression of dopamine D₂ (DRD2), D₃ (DRD3), and D₄ (DRD4) receptors, as well as that of catechol-o-methyltransferase (COMT), an enzyme responsible for dopamine breakdown. We also assessed five single-nucleotide polymorphisms (SNPs) associated with dopamine transmission [5]: rs6277 and rs1800497 (evidence suggests reduced striatal DRD2 density in C- or A-allele carriers, respectively), rs12364283 (C-allele carriers may have higher DRD2 expression), rs6280 (C-allele carriers may have altered DRD3 affinity for dopamine), and rs4680 (A-allele carriers show reduced COMT enzyme activity). We hypothesised that gene expression and SNPs related to dopamine signalling would reflect reduced dopamine sensitivity consistently observed in CUD patients [6], and that these molecular changes predict CUD-associated differences in impulsivity and sensation-seeking.

Methods

Study Sample

A total of 349 participants took part in this study, consisting of 163 patients with moderate-to-severe CUD (DSM-5) and 186 healthy control participants without a history of substance use disorder. Participants were excluded if they had a lifetime history of psychotic, neurological, or neurodevelopmental disorders, traumatic brain injury, or current prescriptions of antipsychotics or stimulant-based medications. Additionally, inclusion in the control group required low levels of drug and alcohol use and no prescriptions for psychoactive medication. Urine screens confirmed active cocaine use in all patients and the absence of illicit drug use in all control participants. All participants completed the Barratt Impulsiveness Scale, Version 11 [7] and the Sensation-Seeking Scale [8] to assess their trait levels of impulsivity and sensation-seeking, respectively. CUD patients additionally completed the Obsessive-Compulsive Drug Use Scale [9] as a measure of cocaine use severity. For details about DNA and RNA processing, please see online supplementary Material (for all online suppl. material, see https://doi.org/10.1159/000551119).

Statistical Analyses

All statistical analyses were conducted in R (v4.2.0). Group differences in mRNA expression were determined with linear mixed effects models (lme4 package), with group (control versus CUD) as the fixed factor, and qPCR plate as the random factor to correct for different batches; age and biological sex were additionally included as covariates. Benjamini-Hochberg corrections were performed to control the false discovery rate from separate linear mixed effects models. We did not apply multiple comparison corrections for multivariable regression models as we had pre-specified hypotheses on the effects of dopamine-specific SNPs. We determined the odds ratio (OR) for CUD with a logistic regression model, entering all genotypes as predictor variables, and CUD diagnosis as the outcome variable; age, biological sex, and ethnicity (White versus non-White) were included as covariates. To understand the associations between peripheral markers and personality traits, we fitted two separate multiple regression models with questionnaire data (impulsivity and sensation-seeking) as outcome variables and CUD diagnosis, mRNA expression, and SNPs as predictor variables; biological sex, age, and ethnicity were again entered as covariates. Where possible, we computed the Bayes factor (BF₁₀) for each predictor variable with the BayesFactor package to determine the strength of evidence against the null model (i.e., nested model without the predictor). For all models, we report the standardised coefficients (B₀). Additionally, we examined the sensitivity of our Bayes Factor estimation to different priors, as reported in online supplementary Table S1. For sensitivity analyses, please see online supplementary Material.

Results

Sample Characteristics

Control and CUD groups were matched for age but significantly differed in gender, smoking status, and years of education (Table 1). While molecular data were unavailable for a subset of participants (online suppl. Fig. S1), we verified there were no differences in missingness for either RNA (${\chi }_1^2$ = 0.25, p = 0.672) or DNA (${\chi }_1^2$ = 0.39, p = 0.620) samples. As expected, CUD patients reported higher levels of trait impulsivity and sensation-seeking.

Table 1.

Sample demographics

Measures	Group		Statistic (t/χ2)	p value
	control	CUD
Group size, n	186	163	​	​
Age, years	36.2 (10.2)	36.6 (8.5)	−0.318	0.751
Gender (% men)	75	92	16.2	<0.001
Current smokers (%)	23	93	170	<0.001
Duration of education, years	13.2 (2.7)	11.3 (1.8)	7.93	<0.001
Impulsivity (BIS-11 total score)	60.9 (9.7)	76.0 (10.1)	14.0	<0.001
Sensation-Seeking (SSS-V total score)	19.0 (5.6)	23.0 (5.6)	6.7	<0.001
Duration of cocaine use, years	–	16.2 (6.9)	–	–
Age of cocaine use onset, years	–	20.4 (5.6)	–	–
Cocaine use severity (OCDUS total score)	–	24.4 (9.2)	–	–

Mean values reported with standard deviation in parentheses.

CUD, cocaine use disorder; BIS-11, Barratt Impulsiveness Scale, Version 11; SSS-V, Sensation-Seeking Scale; OCDUS, Obsessive-Compulsive Drug Use Scale.

Peripheral Gene Expression and SNPs

In peripheral whole blood samples, CUD patients had significantly more DRD2 mRNA expression compared with control participants (B₀ = 0.43, 95% CI [0.13, 0.73], p_FDR = 0.020, BF₁₀ = 6.9), but similar levels of expression for DRD3 (B₀ = 0.22, 95% CI [−0.08, 0.53], p_FDR = 0.48, BF₁₀ = 0.403), DRD4 (B₀ = −0.07, 95% CI [−0.38, 0.24], p_FDR = 0.99, BF₁₀ = 0.200), and COMT (B₀ = 0.01, 95% CI [−0.27, 0.30], p_FDR = 0.99, BF₁₀ = 0.176) (Fig. 1a). We also assessed DRD2 model performance via a 10-fold cross-validation procedure (online suppl. Table S2). Age of onset, duration, and severity of cocaine use did not predict increased DRD2 mRNA expression (online suppl. Table S3).

Fig. 1.

Peripheral biomarkers of dopaminergic genes in CUD, and associations with personality traits. a Relative mRNA transcription levels for DRD2, DRD3, DRD4, and COMT. Patients with CUD show significantly increased transcription of DRD2 in peripheral blood. b ORs for CUD diagnosis at SNP loci associated with altered dopamine transmission: DRD2-related rs6277 C>T, rs1800497 G>A, and rs12364283 T>C alleles; DRD3-related rs6280 C>T allele; and COMT-related rs4680 G>A allele. C allele carriers at the rs6277 locus had significantly increased odds of a CUD diagnosis. c Standardised regression coefficients of multiple regression models – adjusted for age, biological sex, and ethnicity – predicting impulsivity and sensation-seeking, respectively. Self-reported impulsivity was significantly predicted by both a diagnosis of CUD and DRD2 mRNA expression. In contrast, sensation-seeking was only predicted by a CUD diagnosis; no gene expression or SNP frequencies reached significance. CUD, cocaine use disorder; ORs, odds ratios; SNP, single-nucleotide polymorphism; BIS-11, Barratt Impulsiveness Scale, Version 11; SSS-V, Sensation-Seeking Scale. Error bars denote 95% confidence intervals. *p < 0.05; **p < 0.01; ***p < 0.001.

SNP analysis revealed that carriers of the C allele for the DRD2-associated rs6277 locus showed markedly increased odds for receiving a diagnosis of CUD (OR = 2.16, 95% CI [1.15, 4.17], p = 0.018, BF₁₀ = 3.59). Neither DRD2-associated rs1800497 (OR = 0.53, 95% CI [0.16, 1.59], p = 0.27, BF₁₀ = 0.51), DRD2-associated rs12364283 (OR = 1.04, 95% CI [0.50, 2.15], p = 0.92, BF₁₀ = 0.19), DRD3-associated rs6280 (OR = 1.19, 95% CI [0.70, 2.03], p = 0.53, BF₁₀ = 0.07) nor COMT-associated rs4680 (OR = 1.13, 95% CI [0.62, 2.06], p = 0.69, BF₁₀ = 0.16) significantly predicted CUD diagnosis (Fig. 1b). We further verified that rs6277 model performance was acceptable (online suppl. Fig. S2). Although polymorphisms in rs6277 (C957T) and rs1800497 (Taq1A) are known to be in linkage disequilibrium and to have an additive effect on DRD2 bioavailability [10, 11], we were unable to include an interaction term in our models as all CUD patients carried at least one of the rs6277 or rs1800497 risk alleles; i.e., we could not identify an interaction term due to the null reference group. Nevertheless, we verified that participants carrying risk alleles at both loci did not differ from those carrying only one in either cocaine use duration (F(2, 124) = 0.51, p = 0.603) or severity (F(2, 121) = 0.02, p = 0.985).

Associations between Gene Transcription, Genotype, and Personality Traits

As expected, higher levels of self-reported impulsivity were significantly predicted by a CUD diagnosis (B₀ = 1.23, 95% CI [0.94, 1.53], p < 0.001, BF₁₀ = 6.1e11) (Fig. 1c). Reduced DRD2 mRNA expression was associated with impulsivity scores (B₀ = −0.17, 95% CI [−0.32, −0.03], p = 0.013, BF₁₀ = 3.7), but no other mRNA or genotype predictor reached statistical significance (online suppl. Table S4). Sensation-seeking was also significantly predicted by CUD diagnosis (B₀ = 0.58, 95% CI [0.25, 0.91], p < 0.001, BF₁₀ = 25.5), but not by any of the mRNA transcripts or SNPs examined (online suppl. Table S4).

Discussion

Using peripheral blood samples, we identified putative biomarkers associated with altered D2-like dopamine signalling in CUD patients [6], a finding largely consistent with the neuroimaging literature. For example, our finding that the odds for CUD diagnosis increased twofold in C-allele carriers at the DRD2-associated rs6277 locus agrees with previous data suggesting both a CUD diagnosis [12] and the C-allele [5] are associated with reduced striatal DRD2 density. In contrast, we found that peripheral DRD2 mRNA levels are increased in CUD, not reduced.

There are several plausible explanations for this finding. First, CUD participants in our study had recently used cocaine, verified by urinalysis. DRD2 internalisation occurs within 15 minutes of dopamine exposure [13]; it is therefore likely that compensatory expression of DRD2 mRNA arises following cocaine use to reestablish baseline DRD2 density. Indeed, this pattern, decreased Drd2 density and increased Drd2 mRNA transcription, was shown in the rat brain following prolonged cocaine self-administration [14]. Similarly, mRNA transcription of the D2-like D3 receptor increased in the mouse brain following persistent cocaine exposure [15], which increases the bioavailability of dopamine [6]. Future investigations in drug-abstinent CUD patients are needed to clarify these issues; accordingly, the gene expression data in CUD patients should be interpreted cautiously.

Alternatively, differences in peripheral blood DRD2 mRNA levels could reflect changes in the proportion of white blood cells expressing DRD2. Consistently, increased CD4 T-cell levels in stimulant-dependent participants have been reported [16], which in turn reflect dopamine activation [17]. We further acknowledge that mRNA transcription may not perfectly reflect protein translation and that cell-specific variations may exist between peripheral and brain tissue; thus, further research is warranted to confirm whether the observed findings mirror an abnormality in central dopamine transmission. Further, CUD patients are more prone to infectious disease [16]; as this was not screened, it is possible that the changes in DRD2 mRNA reflect altered immunity due to comorbid infection.

Our observation that reduced DRD2 mRNA levels correspond to increased levels of self-reported impulsivity, irrespective of clinical diagnosis, is generally consistent with the role of DRD2 in trait impulsivity [2]. However, we did not find evidence for associations between dopaminergic markers and sensation-seeking, suggesting that variations in sensation-seeking are not reflected through peripheral expression of dopamine-related genes.

Finally, it should be noted that, although we did not observe an effect of ethnicity on either DRD2 mRNA expression or the CUD odds ratio, ancestry may be an important latent source of variation in our data. Future efforts to replicate our molecular findings in models incorporating ancestry adjustment, e.g., through the use of ancestry informative markers [18, 19], will be informative. In conclusion, our data provide preliminary evidence to suggest that peripheral expression of dopaminergic genes may be a relevant biomarker for CUD, which could complement existing clinical assessments for the diagnosis of CUD.

Acknowledgments

We thank Alex Hatton for assistance with blood sample processing. We thank all study participants, without whom this research would not have been possible.

Statement of Ethics

This study received ethical approval from the National Research Ethics Committee, Approval No. 08/H0308/310. All participants provided written informed consent prior to study enrolment and received financial compensation for their time.

Conflict of Interest Statement

E.T.B. has recently consulted Boehringer Ingelheim, SR One, Novartis, GlaxoSmithKline, Sosei Heptares, and Monument Therapeutics and holds equity in and is a cofounder of Centile Bioscience Inc. K.D.E. receives editorial honoraria from Karger Publishers, serves as a Trustee for the Society for the Study of Addiction, and was a member of the journal’s Editorial Board at the time of submission. The other authors have no conflicts to declare.

Funding Sources

This research was supported by grants from the Medical Research Council (G0701497) and the NIHR Cambridge Biomedical Research Centre Mental Health theme (NIHR203312). The views expressed are those of the authors and not necessarily those of the University of Cambridge, NHS, NIHR, or the Department of Health and Social Care. E.T.B. is supported by an NIHR Senior Investigator award. S.L.K. is supported by the NIHR Cambridge Biomedical Research Centre Mental Health theme (NIHR203312 and BRC-1215-20014). T.V.L. is supported by a Griffith Edwards Academic Fellowship from the Society for the Study of Addiction. At the time of the study, K.D.E. was supported by an Alexander von Humboldt Fellowship for Senior Researchers (GBR 1202805 HFST-E). The funders had no role in the design, data collection, data analysis, and reporting of this study.

Author Contributions

E.T.B., K.D.E., S.L.K., and T.V.L.: conceptualisation and writing – review and editing; E.T.B., K.D.E., L.T., S.L.K., and T.V.L.: methodology; S.L.K. and L.T.: investigation; S.L.K. and T.V.L.: formal analysis and writing – original draft; E.T.B. and K.D.E.: supervision; K.D.E.: funding acquisition.

Funding Statement

This research was supported by grants from the Medical Research Council (G0701497) and the NIHR Cambridge Biomedical Research Centre Mental Health theme (NIHR203312). The views expressed are those of the authors and not necessarily those of the University of Cambridge, NHS, NIHR, or the Department of Health and Social Care. E.T.B. is supported by an NIHR Senior Investigator award. S.L.K. is supported by the NIHR Cambridge Biomedical Research Centre Mental Health theme (NIHR203312 and BRC-1215-20014). T.V.L. is supported by a Griffith Edwards Academic Fellowship from the Society for the Study of Addiction. At the time of the study, K.D.E. was supported by an Alexander von Humboldt Fellowship for Senior Researchers (GBR 1202805 HFST-E). The funders had no role in the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

Regarding data and materials availability, further enquiries should be directed to the corresponding author. The sharing of data used in this study is restricted by the informed consent process. The code used to generate the analyses is available online (https://github.com/staceykigar/D2_SUD_project).