Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Mar 1.
Published in final edited form as: Addict Behav. 2020 Oct 10;114:106695. doi: 10.1016/j.addbeh.2020.106695

GABAergic polygenic risk for cocaine use disorder is negatively correlated with precuneus activity during cognitive control in African American individuals

Bao-Zhu Yang 1,2, Iris M Balodis 1,4, Hedy Kober 1,3, Patrick D Worhunsky 1, Cheryl M Lacadie 5, Joel Gelernter 1,2,6,7, Marc N Potenza 1,7,8,9,10
PMCID: PMC8299472  NIHMSID: NIHMS1643364  PMID: 33153773

Abstract

Impaired cognitive control has been implicated in cocaine use disorder (CUD). GABAergic treatments have been proposed for CUD. Here we examined relationships between GABAergic genes and neural correlates of cognitive control in CUD. We analyzed two independent African American cohorts: one of >3,000 genomewide-genotyped subjects with substance dependence and another of 40 CUD and 22 healthy control (HC) subjects who were exome-array genotyped and completed an fMRI Stroop task. We used five association thresholds to select variants in the reference cohort, yielding five polygenic risk scores (i.e., CUD-GABA-PRSs) for the fMRI cohort. At p<0.005, the CUD-GABA-PRSs, which aggregated relative risks of CUD from 89 variants harboring in 16 genes, differed between CUD and HC individuals in the fMRI sample (p=0.013). This CUD-GABA-PRS correlated inversely with Stroop-related activity in the left precuneus in CUD (r=−0.58, pFWE<0.05) but not HC participants. Post-hoc seed-based connectivity analysis of the left precuneus identified reduced functional connectivity to the posterior cingulate cortex (PCC) in CUD compared to HC subjects (p=0.0062) and the degree of connectivity correlated with CUD-GABA-PRSs in CUD individuals (r=0.287, p=0.036). Our findings suggest that the GABAergic genetic risk of CUD in African Americans relates to precuneus/PCC functional connectivity during cognitive control. Identification of these GABAergic processes may be relevant targets in CUD treatment. The novel identification of 16 GABAergic genes may be investigated further to inform treatment development efforts for this condition that currently has no medication with a formal indication for its treatment.

Keywords: Cocaine use disorder, cognitive control, GABA, substance-related disorder, Stroop, addictive behaviors, magnetic resonance imaging

INTRODUCTION

Impaired control has been proposed as a central component in models of addiction1–3. The Stroop task assesses cognitive control4. Neural underpinnings of cognitive-control impairment in cocaine use disorder (CUD) and other substance use disorders (SUDs) have been suggested in Stroop-related fMRI studies5. Poor cognitive control may facilitate the development and progression of CUD, contribute to relapse5, and relate to poorer treatment outcomes6.

During Stroop performance, regional cognitive-control-related activations in cortical (e.g., ventromedial prefrontal cortex (vmPFC), posterior cingulate cortex (PCC)), and subcortical (striatum, thalamus) have been linked to treatment outcomes in CUD7,8. With treatment, Stroop-related cortico-striatal activations become less pronounced in individuals with SUDs including CUD, suggesting possible improved neural efficiency after treatment9,10. Furthermore, cognitive impairment in CUD relating to Stroop performance may vary according to individual characteristics. For example, among CUD subjects with impaired insight, lower error-induced rostral anterior cingulate cortex (rACC) activity has been associated with more frequent cocaine use and lower levels of emotional awareness11.

Stroop-related circuits have also been linked to CUD and CUD-related treatment outcomes. CUD patients showed differences in Stroop-related network recruitment, with a fronto-cingular network associated with treatment retention and subcortical and ventral prefrontal networks related to abstinence during treatment12. Additionally, a fronto-parietal circuit that included the precuneus was linked to Stroop effects and reaction times differently in CUD and HC subjects12. More recent work has suggested a cocaine abstinence network initially identified using task-based reward processing data and more recently extended to Stroop data13,14 (supplemental materials).

Genetic contributions to CUD are well established (supplementary materials). Cognitive function is also substantially heritable; heritability estimates of cognitive control, inhibitory control, and response inhibition range from 0.38 to 0.6015–17. However, little is known regarding how CUD genetic risk influences brain activity (supplemental material). A study investigating genetic influences of a single nucleotide polymorphism (SNP) at proenkephalin gene (PENK) on neural correlates of Stroop performance in CUD and HC subjects found a diagnosis-by-allele interaction that implicated neural responses to errors in the putamen, rACC/OFC, and inferior frontal gyrus, where neural activities were increased in CUD individuals carrying the higher risk allele18. Thus, genetic factors linked to CUD may influence brain mechanisms underlying poor cognitive control in CUD. However, such genetic factors have rarely been studied and not with respect to polygenic risk scores (PRSs), a potentially powerful new approach for targeting treatment development efforts for complex disorders19. This is particularly relevant for CUD, a complex condition for which no medication has a formal indication.

Augmenting GABAergic function has been proposed as a possible therapy for CUD20 that may ameliorate cognitive-control deficits21,22. GABAergic agents such as topiramate, tiagabine, baclofen, and vigabatrin have shown some promise in CUD treatment23. Several lines of evidence support the development of GABAergic agents in treating CUD via restoring GABA system homeostasis and increasing cognitive control24,25. Low GABA levels in PFC have been observed in CUD subjects26. Cocaine may impact GABAergic systems in the PFC and subcortical regions and thus influence sensitization, craving and relapse processes27. Abnormal GABA levels and GABAergic inhibitory circuits have also been linked to impaired cognitive control28,29, and lower GABA levels have been associated with greater impulsivity and poorer response inhibition28. GABAergic disruptions may be partially restored using GABAergic medications30. Taken together, data suggest that abnormal GABAergic function may underlie impaired cognitive control in CUD (supplemental materials).

Impaired cognitive control in CUD has been hypothesized to be a consequence of cocaine misuse27,31. However, other studies suggest that impaired cognitive control may precede compulsive drug-seeking and drug-taking and accelerate the progression from experimentation to compulsive use32,33. Findings among discordant sib-pairs with and without stimulant dependence and unrelated healthy control subjects (HCs) suggest that poor cognitive control may run within vulnerable family members and not directly be caused by chronic drug misuse34,35. Thus, pre-existing or genetic factors prior to substance use may contribute to impaired cognitive control in CUD.

Here, we aimed to identify GABAergic PRSs as related to CUD and apply this information in an independent sample to investigate relationships with neural correlates of cognitive control. We chose to focus on African Americans (AAs) because they are disproportionately impacted by CUD, and genetic predispositions are often population-specific. To identify genes of interest, we selected SNP variants mapped to the genes identified in GABAergic pathways and estimated relative risks of SNP variants with respect to CUD from a reference cohort of AAs in a genome-wide association study (GWAS) of CUD36. In an independent AA fMRI case-control sample with genotypes from exome microarrays, we computed CUD-related GABAergic PRSs using the relative risks estimated from the reference cohort37. The PRS approach has been used to predict disease risk and facilitate interpretation of the genetic architecture of complex diseases38.

We hypothesized that CUD subjects in the fMRI cohort would, on average, have higher GABAergic PRSs and that higher GABAergic PRSs would be related to lower brain activations during cognitive control processing in regions implicated in cognitive control in CUD (e.g., cingulate, frontal and parietal cortices, precuneus, striatum, and thalamus).

METHODS AND MATERIALS

Subjects and Ethics

Participants were recruited using advertisements and other established methods. The studies were reviewed by ethics committees at the participating universities, written informed consent was obtained after procedures had been fully explained, and procedures were conducted in accordance with the Declaration of Helsinki.

African American Cohort for GABA-PRS (Reference Cohort)

The cohort for determining GABA-PRS included 3,318 AA participants with SUDs (2,482 subjects with cocaine dependence, with relates to moderate to severe CUD in DSM-5), as detailed previously36.

African American fMRI cohort

The fMRI cohort originally consisted of 67 AA participants (44 CUD, 23 HC) who performed the Stroop task during fMRI as described previously7,8. Five participants were excluded from further analysis because of excessive motion or poor image quality (final N=62; See Table 1 for demographic characteristics). Eligibility criteria required that all participants be over 18 years old, with no contraindications for MRI scanning (e.g., no head injuries). Current CUD status (cocaine dependence) was assessed using a Structured Clinical Interview for Diagnosis (SCID) for DSM-IV39. Additional information is included in supplemental materials.

Table 1.

Characteristics of the fMRI Cohort

Cocaine Use Disorder Healthy Control p-value
N 40 22
Male (%) 67.5 36.4 <0.05
Age (mean/sd) 43.5/5.5 31.9/9.7 <0.01
Education years (mean/sd) 11.9/1.3 14.9/1.6 <0.01
Years of cocaine use (mean/sd) 13.4/7.6
Age of cocaine first use (mean/sd) 22.5/8.1
Open in a new tab

Exome-Arrayed Genotyping and Imputation for the fMRI Cohort

DNA was extracted from peripheral blood lymphocytes using the PAXgene Blood DNA Kit (QIAGEN, California USA). All participants in the fMRI cohort were genotyped using the Illumina HumanExome Array 12v1.1. The average genotyping call rate per sample subject was 0.962 for the fMRI cohort, which passed the 95% standard quality threshold for the genotyping call rate40. We imputed genotypes for the fMRI cohort by first combining genotypes of the two studied cohorts (fMRI cohort and the reference cohort) and used a high imputation accuracy cutoff (>0.9). Details of the laboratory genotyping and quality control (QC), imputation, pre- and post-imputation QC for data analysis, and the resulting variant distribution are presented in supplementary materials.

Analytic Approach

The analytic approach is depicted in Figure 1.

Figure 1.

Figure 1.

Open in a new tab

Flow of the analytical approach investigating the GABAergic polygenic risk score (PRS). Rounded rectangles indicate the observed data or derived statistics. Rectangles indicate the analytical approach. The large-scale substance use cohort with genomewide genotypes was combined with the independent and exome-arrayed fMRI cohort for genotype imputation. The imputed genetic variants which belong to the gene ontology GABA query set were selected and further gone through the LD-based SNP pruning before the association analysis to generate the summary statistics for each SNP in the LD pruned GABA SNPs. Borrowing the summary statistics from the large cohort, GABA polygenic risk scores for CUD for the fMRI cohort were generated. On the other hand, the fMRI cohort with the Stroop task had image pre-processing and within-subject analyses. Correlation analyses between BOLD signal and GABA polygenic risk scores were conducted to identify brain regions implicated in the Stroop effect correlated with the GABA polygenic risk for CUD. GABA, gamma-aminobutyric acid; GWAS, genomewide association study; fMRI, functional magnetic resonance imaging; LD, linkage disequilibrium; SNP, single nucleotide polymorphism; BOLD, blood-oxygen-level-dependent.

Gene Ontology Query for the GABA-related Genes and Linkage-Disequilibrium-based SNP Pruning

We queried GABA-related genes from AmiGO2, a set of web-based tools for browsing the Gene Ontology database41. We retrieved 101 entities in 43 unique genes from Gene and Gene Product searching for the taxon of Homo sapiens. Correspondingly, 61,114 SNPs (including variants within 3 kb of 3’/5’ UTRs) located in the 43 genes were identified for the imputed fMRI sample. Among these SNPs, the post-imputation QC metrics included individuals and SNPs with genotype call rates <98%, Hardy-Weinberg equilibrium p-values <10−5, imputation accuracy <0.9, and minor allele frequency (MAF) <1%. The QC resulted in 57,938 SNPs, of which 0.05% was covered in the exome array. To reduce inflated PRSs due to variants in linkage disequilibrium, a linkage-disequilibrium-based SNP pruning procedure in PLINK was implemented to access independent SNPs of association42. PLINK is a commonly used and open-source software toolset designed to perform basic and large-scale genetic analyses. Using a sliding window of 50 SNPs and a five-SNP shift at each step, the procedure removed variants in the current window with a variant inflation factor exceeding 2. Finally, 3309 SNPs passed the pruning for further analysis.

Association Analysis for CUD in the Reference Cohort

An association test was conducted in a logistic regression to model the DSM-IV diagnosis of cocaine dependence on each SNP variant (i.e., 3309 SNPs from the pruned set) and included covariates of age, sex, substance dependence on alcohol, nicotine, cannabis, and opioids, and five principal components for adjustment of population stratification43 using PLINK42.

GABAergic Polygenic Risk Scores

We computed five PRSs for each individual using SNPs selected according to five nominal p-value thresholds of 0.5, 0.1, 0.05, 0.01, and 0.005 from the association result in the large cohort of individuals with SUDs. Effect directions of SNPs was aligned to indicate risk for CUD. Each PRS was generated by summing all relative risks weighted by the genotype of each SNP for each fMRI participant. We used PLINK to generate risk scores, which were multiplied by 100 to acquire the PRSs.

fMRI Image Acquisition and Analysis

During fMRI, participants completed the Stroop color-word interference test as described previously44. Images were acquired using a 3T Siemens Trio magnetic resonance imaging (MRI) system (Siemens AG, Erlangen, Germany). Details of the task, image acquisition, and image preprocessing are described in supplemental materials. A subject-level general linear model was used to estimate brain responses associated with congruent (e.g., the word ‘red’ in red font) and incongruent (e.g., the word ‘red’ in blue font) stimuli using robust regression. Neural correlates of the Stroop effect were assessed as differences in neural responses between incongruent versus congruent trials (evaluated using linear t-contrasts) using NeuroElf45.

Correlations Between Polygenic Risk Score and Brain Activity

Neural correlates (assessed via BOLD signal changes) of responses to incongruent versus congruent stimuli were correlated with PRSs in CUD and HC groups separately. Age was included as a covariate. As CUD individuals self-reported lower education levels than HCs, effects were investigated post hoc, as were effects of sex in a stratification analysis. For correlation analyses between CUD-GABA-PRSs and fMRI Stroop effects, we corrected for multiple comparisons at a family-wise-error (FWE) rate of 5% (pFWE<.05). Smoothness was estimated from the data, and Monte-Carlo simulation in NeuroElf was used to obtain a combined voxel-wise threshold (of p<0.01) and cluster threshold (k=65).

Seed-based Functional Connectivity Analysis

Post-hoc seed-based connectivity of the left precuneus region identified in correlation analyses was performed using BioImage Suite46. The time course of the seed in a given participant across all 6 concatenated runs was computed as the average time course across all voxels in the seed. This time course was correlated with that in every other voxel in the brain to create a map of functional connectivity r-values between the seed and every other voxel. These r-values were transformed to z-values using Fisher’s transformation, yielding one map for each participant representing the strength of seed-to-whole-brain connectivity. Voxel-wise two-sample t-tests were used to compare connectivity between CUD and HC participants. To correct for multiple comparisons, we adopted Monte Carlo simulation using AFNI 3dClustSim (version 16.3.05) on the autocorrelation values of the residuals from the t-test. Results are shown at pFWE<0.05 corrected with an initial voxel-level p-threshold of p<0.001 using a cluster of k=29.

RESULTS

GABAergic PRSs

Among the five GABAergic PRSs for the fMRI cohort, only the PRS generated by the p-value threshold of 0.005 significantly distinguished CUD from HC subjects (p=0.013, Figure 2A). This PRS, designated as CUD-GABA-PRS, was higher in CUD relative to HC subjects (23.47±0.84 versus 22.93±1.08, p=0.013). This CUD-GABA-PRS was generated from the association analysis for CUD in the large cohort with 89 SNPs (p-value<0.005) from the 3309 SNPs identified in the GABAergic pathways. These 89 SNPs were mapped to 16 genes: ADORA2A, CNTNAP4, GABARAPL1, GABRA1, GABRB1, GABRB3, GABRG1, GABRP, LHX6, NF1, NLGN1, NLGN2, PLCL1, PRKCE, SLC6A13, and USP46. Details regarding the number of SNPs passing the p-value threshold in each gene, gene size, and gene function are presented in supplementary Table S1.

Figure 2.

Figure 2.

Open in a new tab

GABA polygenic risk scores (CUD-GABA-PRSs) and correlated brain activations during Stroop performance. (A) Boxplot of the CUD-GABA-PRSs in the individuals with cocaine use disorder (CUD) and healthy control (HC) participants. Boxplots (in black) are overlaid with scatter plots. (B) Correlation analyses between CUD-GABA-PRSs and the neural correlates of the Stroop effect (incongruent > congruent) identified a cluster in the left-precuneus in the CUD group. All contrast maps are thresholded (k=65) at a voxel-level of p<0.01 two-tailed and family-wise-error-corrected at p<0.05. (C) Scatter plot of the peak activation extraction of the left precuneus at Talairach (x, y, z)=(−18,−45,48) correlated with CUD-GABA-PRSs in CUD and HC participants. (D) Within-sex scatter plots of the correlations of CUD-GABA-PRSs and the Stroop-effect-related activation in the left precuneus in individuals with CUD.

GABAergic PRSs and Stroop-related Measures

Within-group correlation analyses between the CUD-GABA-PRSs and brain activations during Stroop performance identified one cluster in the left precuneus in the CUD group (Figure 2B). The CUD-GABA-PRS was negatively correlated with Stroop-related activation in the left precuneus in CUD subjects (r=−0.58; voxel level p<0.01, cluster-corrected pFWE<0.05, threshold k=65) (Figure 2C). This negative correlation was not observed in the HC group (r=0.054) (Figure 2C), and the correlation in CUD subjects differed from that in HC subjects (p<0.05). The correlation pattern between the CUD-GABA-PRSs and precuneus activity was similar in female (r=−0.68) and male (r=−0.53) CUD subjects (p>0.05; Figure 2D).

Out-of-scanner Stroop performance was also examined. CUD-GABA-PRSs and Stroop-effect magnitudes (i.e., incongruent minus congruent reaction times) were not correlated (Supplementary Figure S1).

Seed-based Functional-connectivity

We conducted a post-hoc functional-connectivity analysis using the left precuneus as a seed in the CUD and HC groups. Differential connectivity to the PCC (Brodmann areas 23 and 31; voxel level p<0.001, pFWE<0.05, cluster-corrected k=29) was identified between CUD and HC subjects (Figure 3A). A pattern of negative functional connectivity between the left precuneus and PCC was observed in HCs and was weaker for CUD than HC subjects (mean connectivity z-scores are −0.015 for CUD and −0.22 for HC (p=0.0062; Figure 3B). The functional connectivity was positively correlated with CUD-GABA-PRSs in individuals with CUD (r=0.29, p=0.036). A non-significant correlation of similar magnitude but in the opposite direction was observed in HC individuals (r=−0.33, p=0.069; Figure 3C).

Figure 3.

Figure 3.

Open in a new tab

(A) A seed-based connectivity analysis identified differential connectivity between the precuneus seed and the posterior cingulate cortex (PCC) between individuals with cocaine use disorder (CUD) and healthy control (HC) participants. Findings are reported at pFWE<0.05 with a voxel-level p<0.001 using a cluster of k=29. (B) Boxplots of the connectivity z-scores in CUD and HC groups are shown. Boxplots are overlaid on scatter plots (difference between CUD versus HC individuals, p=0.0062). (C) Relationship between the connectivity z-scores and the GABAergic polygenic risk scores (CUD-GABA-PRSs). Correlations r=0.29 (p=0.036) in individuals with CUD (p=0.036), and r=−0.33 in HC individuals (p=0.069).

DISCUSSION

We identified among AA individuals elevated GABAergic PRSs linked to CUD using genetic variants extracted from genomewide-genotyped data. These PRSs were inversely associated with precuneus activation in an independent AA CUD sample during fMRI Stroop performance. Further, differential precuneus/PCC connectivity was observed between CUD and HC subjects. Our findings suggest that GABA-related genetic factors may relate to neural underpinnings of cognitive control in CUD. These findings provide a foundation for a possible pharmacogenetic strategy for developing treatments for CUD that is relevant for AA people, a group disproportionately impacted by CUD.

Genes Contributing to the CUD-GABA-PRSs

It is a novel finding to identify a CUD-GABA-PRSs. The identified 16 genes that harbor the 89 SNPs for calculating CUD-GABA-PRSs can be divided into two groups based on biological functions. Group-I includes seven genes (GABARAPL1, GAT2, GABRA1, GABRB1, GABRB3, GABRG1, GABRP) that are directly involved in encoding members of the GABA-A receptor family (GABAAR) or related GABAergic signaling molecules. Group-II includes nine genes (ADORA2A, CNTNAP4, LHX6, NF1, NLGN1, NLGN2, PLCL1, PKCE, and USP46) that are involved in various functions regulated by GABA or related to GABAergic functions (Supplemental Table S1). These 16 genes contribute importantly to GABAergic inhibitory signal transduction and executive functioning potentially related to behavioral inhibition (further discussed in suppplemental materials)47–50.

GABA PRSs and Precuneus Activity in CUD

Precuneus involvement in GABAergic functioning in CUD has been suggested previously51. The precuneus is one of three regions with the largest differences in benzodiazepine-induced decrements in whole-brain metabolism in CUD51. Stroop-related precuneus activation changes with treatment in CUD8. The association between CUD-GABA-PRSs and reduced precuneus activity only in CUD subjects suggests a potential pharmacogenetic effect. The extent to which relates to specific aspects of CUD warrants additional investigation.

Precuneus-PCC Connectivity

We observed Stroop-related precuneus-PCC functional connectivity in CUD subjects. Cognitive control is regulated by numerous brain networks, each containing a set of anatomically distributed regions52. The precuneus has been proposed as a central hub of the default mode network53,54, with specialized hub regions interlinking the parietal and prefrontal regions53. PCC activation during Stroop task performance was correlated with longer durations of abstinence in CUD treatment7. In the same sample, the precuneus contributed to two circuits (fronto-parietal and fronto-cingular) that were linked to Stroop performance in healthy subjects more so than CUD patients, with the engagement of the latter circuit linked to treatment retention12.

In concert, the precuneus and PCC have been implicated in the neural substrates of frontoparietal networks associated with attentional control55–57 and inhibitory control in CUD12, 58. However, these networks also involve other regions, and the PCC and precuneus may contribute to multiple networks. For example, the PCC has been implicated in intrinsic cognitive-control networks59,60.

As we found that higher GABAergic PRSs of CUD were associated with decreased Stroop-related precuneus activity and precuneus-PCC connectivity, further study is needed to understand precise mechanisms underlying these findings with respect to brain functional architecture and prevention and treatment development. Collectively, our results suggest a GABAergic genetic risk for potential deficits in cognitive control in AAs with CUD and altered precuneus-PCC functional connectivity that may relate to such impairments. Further research on the identified 16 GABAergic genes may further reveal potential mechanisms of cognitive controls in CUD and possible treatment targets.

Pharmacogenetic Considerations

Pharmacogenetics consider how genetic factors contribute to the metabolism, response, and/or side effects of a given medication61. Pharmacogenetic considerations for CUD are at relatively early stages. In disulfiram treatment of CUD, individuals carrying at least one minor allele in either the ANKK1 or DRD2 gene responded better to disulfiram treatment compared to individuals carrying only major alleles62. DBH, encoding dopamine β-hydroxylase (DβH), is a functional candidate gene for mediating cocaine treatment response, in which the functional regulatory variant rs1611115 reduces DβH enzyme levels63. CUD patients with rs1611115*CC genotypes, producing higher DβH enzyme levels, responded to disulfiram treatment63. However, among patients carrying the rs1611115*T-allele (CT/TT), there may be better responses to drugs such as doxazosin, an α1-adrenergic antagonist, that may operate by blocking norepinephrine stimulation and reward from cocaine-induced norepinephrine increases64.

Pharmacogenetic research on CUD treatment has focused on candidate gene approaches on a relatively small scale (N<50 in each treatment group) compared to modern genomewide association analyses. Our current study is not a pharmacogenetic study, but we considered a complex trait approach to investigate genetic factors behind cognitive control. Based on data reviewed in the introduction, we speculate on possible pharmacogenetic mechanisms for CUD treatment that may involve restoring GABAergic function. We hypothesize that the reported CUD-GABA-PRS may relate to such responses, although this is currently speculative. Further studies are required to examine interactions between the selected GABAergic genes and medications and investigate relationships between the GABA-PRS and responses to medications in CUD treatment.

Strengths and Limitations

Study strengths include the multiple cohorts studied and tested across multiple methods, the large cohort used to define the CUD-GABA-PRSs and an independent fMRI-scanned cohort. Further, inclusion of solely AA participants reduces race-related variance and minimizes potential false-positive findings in that a mixture of African and European American samples could create confounding results based on population stratification65. Limitations include the small size of the fMRI cohort; thus, findings should be considered preliminary. In fMRI analyses, since years of education years were not correlated with Stroop-related effects (r=0.068, p=0.598), education levels were not controlled in whole-brain correlations with PRSs. However, the main effect was found only in CUD individuals and was not affected by group differences in education. The voxel-wise p threshold we set was relatively lax (p<0.01), When adopting a more stringent voxel-wise threshold (p<0.001), the cluster appeared in the same left precuneus region; however, it did not survive the whole-brain-correction threshold, and should thus be considered cautiously. Another limitation is that the GABA-related genes could also have non-GABAergic functions. For example, one of the identified GABA-related genes, ADORA2A encodes the adenosine A2A receptor, which forms a heteromer with dopamine receptors66. Other limitations include the studies’ cross-sectional nature (precluding causal inferences), identified precuneus voxels including both white and gray matter, and possibility of HCs having lifetime cocaine use. Also, given the focus on AA subjects, future studies involving other racial or ethnic groups are warranted to examine generalizability of the findings. We used a data-driven whole-brain analysis to investigate our neuroimaging hypotheses. Alternate approaches (e.g., region-of-interest approaches) could be used in future studies to investigate potential roles of specific brain areas (e.g., frontal and striatal regions) with respect to relationships between genetic factors and neural correlates of cognitive control in CUD. We also did not use Bonferonni corrections when considering the 5 CUD-GABA-PRS thresholds, consistent with prior studies67.

Conclusions

We demonstrated that GABAergic PRSs relate to Stroop-related precuneus activity and connectivity in CUD by integrating two independent AA cohorts assessed through large-scale genetic screening and brain imaging. The findings extend pre-clinical reports of cocaine-related alterations in GABAergic neural function27,68, demonstrating for the first time GABAergic genetic links to neural correlates of cognitive control for individuals with CUD. Given that neural correlates of cognitive control have been linked to treatment outcomes for CUD, the findings suggest that GABAergic mechanisms may warrant further study in CUD treatment, particularly among AA populations.

Supplementary Material

1

  • GABA polygenic risk scores (PRSs) for cocaine use disorder (CUD) differ between cases and controls.

  • GABA PRSs relate inversely with Stroop neural correlates in the left precuneus in CUD.

  • Seed-based connectivity identified reduced precuneus-to-posterior-cingulate connectivity in CUD.

Conflict of Interest:

The authors report no conflicts of interest with respect to the content of this manuscript. Dr. Potenza has: consulted for and advised the Addiction Policy Forum, Game Day Data, AXA, Idorsia and Opiant/Lakelight Therapeutics; received research support from the Mohegan Sun Casino and the National Center for Responsible Gaming (now the International Center for Responsible Gaming); consulted for legal and gambling entities on issues related to impulse-control and addictive disorders; given academic lectures in grand rounds, CME events, and other clinical/scientific venues; and generated books or chapters for publishers of mental health texts. Dr. Gelernter is named as a co-inventor (with Dr. Henry Kranzler) on the PCT patent application #15/878,640 entitled: “Genotype-guided dosing of opioid agonists,” filed January 24, 2018. Dr. Kober has provided consultation for Invidior, Inc. The other authors report no disclosures. The views presented in this manuscript represent those of the authors and not necessarily those of the funding agencies.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  • 1.Brand M, et al. , The Interaction of Person-Affect-Cognition-Execution (I-PACE) model for addictive behaviors: Update, generalization to addictive behaviors beyond internet-use disorders, and specification of the process character of addictive behaviors. Neurosci Biobehav Rev, 2019. 104: p. 1–10. [DOI] [PubMed] [Google Scholar]
  • 2.Goldstein RZ and Volkow ND, Dysfunction of the prefrontal cortex in addiction: neuroimaging findings and clinical implications. Nat Rev Neurosci, 2011. 12(11): p. 652–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Noël X, Brevers D, and Bechara A, A neurocognitive approach to understanding the neurobiology of addiction. Curr Opin Neurobiol, 2013. 23(4): p. 632–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Botvinick MM and Cohen JD, The computational and neural basis of cognitive control: charted territory and new frontiers. Cogn Sci, 2014. 38(6): p. 1249–85. [DOI] [PubMed] [Google Scholar]
  • 5.Garavan H and Hester R, The role of cognitive control in cocaine dependence. Neuropsychol Rev, 2007. 17(3): p. 337–45. [DOI] [PubMed] [Google Scholar]
  • 6.Streeter CC, et al. , Performance on the Stroop predicts treatment compliance in cocaine-dependent individuals. Neuropsychopharmacology, 2008. 33(4): p. 827–36. [DOI] [PubMed] [Google Scholar]
  • 7.Brewer JA, et al. , Pretreatment brain activation during stroop task is associated with outcomes in cocaine-dependent patients. Biol Psychiatry, 2008. 64(11): p. 998–1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.DeVito EE, et al. , Functional neural changes following behavioral therapies and disulfiram for cocaine dependence. Psychol Addict Behav, 2017. 31(5): p. 534–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.DeVito EE, et al. , fMRI Stroop and behavioral treatment for cocaine-dependence: Preliminary findings in methadone-maintained individuals. Addict Behav, 2019. 89: p. 10–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.DeVito EE, et al. , A preliminary study of the neural effects of behavioral therapy for substance use disorders. Drug Alcohol Depend, 2012. 122(3): p. 228–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Moeller SJ, et al. , Functional, structural, and emotional correlates of impaired insight in cocaine addiction. JAMA Psychiatry, 2014. 71(1): p. 61–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Worhunsky PD, et al. , Functional brain networks associated with cognitive control, cocaine dependence, and treatment outcome. Psychol Addict Behav, 2013. 27(2): p. 477–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Yip SW, et al. , Connectome-Based Prediction of Cocaine Abstinence. Am J Psychiatry, 2019. 176(2): p. 156–164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Lichenstein SD, et al. , Dissociable neural substrates of opioid and cocaine use identified via connectome-based modelling. Mol Psychiatry, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Gagne JR and Saudino KJ, Wait for it! A twin study of inhibitory control in early childhood. Behav Genet, 2010. 40(3): p. 327–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Macare C, et al. , Preliminary findings on the heritability of the neural correlates of response inhibition. Biol Psychol, 2014. 103: p. 19–23. [DOI] [PubMed] [Google Scholar]
  • 17.Anokhin AP, et al. , Heritability of brain activity related to response inhibition: A longitudinal genetic study in adolescent twins. Int J Psychophysiol, 2017. 115: p. 112–124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Moeller SJ, et al. , Effects of an opioid (proenkephalin) polymorphism on neural response to errors in health and cocaine use disorder. Behav Brain Res, 2015. 293: p. 18–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Gibson G, On the utilization of polygenic risk scores for therapeutic targeting. PLoS Genet, 2019. 15(4): p. e1008060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Johnson BA, et al. , Topiramate for the treatment of cocaine addiction: a randomized clinical trial. JAMA Psychiatry, 2013. 70(12): p. 1338–46. [DOI] [PubMed] [Google Scholar]
  • 21.Filip M, et al. , Various GABA-mimetic drugs differently affect cocaine-evoked hyperlocomotion and sensitization. Eur J Pharmacol, 2006. 541(3): p. 163–70. [DOI] [PubMed] [Google Scholar]
  • 22.Sofuoglu M and Kosten TR, Emerging pharmacological strategies in the fight against cocaine addiction. Expert Opin Emerg Drugs, 2006. 11(1): p. 91–8. [DOI] [PubMed] [Google Scholar]
  • 23.Karila L, et al. , New treatments for cocaine dependence: a focused review. Int J Neuropsychopharmacol, 2008. 11(3): p. 425–38. [DOI] [PubMed] [Google Scholar]
  • 24.Huang YH, Schlüter OM, and Dong Y, Cocaine-induced homeostatic regulation and dysregulation of nucleus accumbens neurons. Behav Brain Res, 2011. 216(1): p. 9–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Wang J, et al. , Cascades of Homeostatic Dysregulation Promote Incubation of Cocaine Craving. J Neurosci, 2018. 38(18): p. 4316–4328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Ke Y, et al. , Frontal lobe GABA levels in cocaine dependence: a two-dimensional, J-resolved magnetic resonance spectroscopy study. Psychiatry Res, 2004. 130(3): p. 283–93. [DOI] [PubMed] [Google Scholar]
  • 27.Hearing MC, Zink AN, and Wickman K, Cocaine-induced adaptations in metabotropic inhibitory signaling in the mesocorticolimbic system. Rev Neurosci, 2012. 23(4): p. 325–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Silveri MM, et al. , Frontal lobe gamma-aminobutyric acid levels during adolescence: associations with impulsivity and response inhibition. Biol Psychiatry, 2013. 74(4): p. 296–304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.van den Wildenberg WP, et al. , Mechanisms and dynamics of cortical motor inhibition in the stop-signal paradigm: a TMS study. J Cogn Neurosci, 2010. 22(2): p. 225–39. [DOI] [PubMed] [Google Scholar]
  • 30.Sofuoglu M and Kosten TR, Novel approaches to the treatment of cocaine addiction. CNS Drugs, 2005. 19(1): p. 13–25. [DOI] [PubMed] [Google Scholar]
  • 31.Cass DK, et al. , Developmental disruption of gamma-aminobutyric acid function in the medial prefrontal cortex by noncontingent cocaine exposure during early adolescence. Biol Psychiatry, 2013. 74(7): p. 490–501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Belin D, et al. , High impulsivity predicts the switch to compulsive cocaine-taking. Science, 2008. 320(5881): p. 1352–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Homberg JR, Karel P, and Verheij MM, Individual differences in cocaine addiction: maladaptive behavioural traits. Addict Biol, 2014. 19(4): p. 517–28. [DOI] [PubMed] [Google Scholar]
  • 34.Ersche KD, et al. , Abnormal brain structure implicated in stimulant drug addiction. Science, 2012. 335(6068): p. 601–4. [DOI] [PubMed] [Google Scholar]
  • 35.Smith DG, et al. , Cognitive control dysfunction and abnormal frontal cortex activation in stimulant drug users and their biological siblings. Transl Psychiatry, 2013. 3: p. e257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Gelernter J, et al. , Genome-wide association study of cocaine dependence and related traits: FAM53B identified as a risk gene. Mol Psychiatry, 2014. 19(6): p. 717–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Purcell SM, et al. , Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature, 2009. 460(7256): p. 748–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Evans DM, Visscher PM, and Wray NR, Harnessing the information contained within genome-wide association studies to improve individual prediction of complex disease risk. Hum Mol Genet, 2009. 18(18): p. 3525–31. [DOI] [PubMed] [Google Scholar]
  • 39.First M, Structured Clinical Interview for DSM-IV Axis I Disorders: SCID-I Clinician Version. 1997: American Psychiatric Publishing Inc. [Google Scholar]
  • 40.Anderson CA, et al. , Data quality control in genetic case-control association studies. Nat Protoc, 2010. 5(9): p. 1564–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.The Gene Ontology Consortium. Gene Ontology database Available from: http://amigo.geneontology.org/amigo.
  • 42.Purcell S, et al. , PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet, 2007. 81(3): p. 559–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Zhang Y, Guan W, and Pan W, Adjustment for population stratification via principal components in association analysis of rare variants. Genet Epidemiol, 2013. 37(1): p. 99–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Kober H, et al. , Cannabis abstinence during treatment and one-year follow-up: relationship to neural activity in men. Neuropsychopharmacology, 2014. 39(10): p. 2288–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Weber J NeuroElf. 2014; Available from: http://neuroelf.net/.
  • 46.Joshi A, et al. , Unified framework for development, deployment and robust testing of neuroimaging algorithms. Neuroinformatics, 2011. 9(1): p. 69–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Cheng CH, et al. , Resting GABA concentration predicts inhibitory control during an auditory Go-Nogo task. Exp Brain Res, 2017. 235(12): p. 3833–3841. [DOI] [PubMed] [Google Scholar]
  • 48.Dyke K, et al. , Comparing GABA-dependent physiological measures of inhibition with proton magnetic resonance spectroscopy measurement of GABA using ultra-high-field MRI. Neuroimage, 2017. 152: p. 360–370. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Mooney RA, Cirillo J, and Byblow WD, GABA and primary motor cortex inhibition in young and older adults: a multimodal reliability study. J Neurophysiol, 2017. 118(1): p. 425–433. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Nowak M, et al. , Driving Human Motor Cortical Oscillations Leads to Behaviorally Relevant Changes in Local GABAA Inhibition: A tACS-TMS Study. J Neurosci, 2017. 37(17): p. 4481–4492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Volkow ND, et al. , Enhanced sensitivity to benzodiazepines in active cocaine-abusing subjects: a PET study. Am J Psychiatry, 1998. 155(2): p. 200–6. [DOI] [PubMed] [Google Scholar]
  • 52.Marek S and Dosenbach NUF, The frontoparietal network: function, electrophysiology, and importance of individual precision mapping. Dialogues Clin Neurosci, 2018. 20(2): p. 133–140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Bullmore E and Sporns O, Complex brain networks: graph theoretical analysis of structural and functional systems. Nat Rev Neurosci, 2009. 10(3): p. 186–98. [DOI] [PubMed] [Google Scholar]
  • 54.Cunningham SI, Tomasi D, and Volkow ND, Structural and functional connectivity of the precuneus and thalamus to the default mode network. Hum Brain Mapp, 2017. 38(2): p. 938–956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Corbetta M, Frontoparietal cortical networks for directing attention and the eye to visual locations: identical, independent, or overlapping neural systems? Proc Natl Acad Sci U S A, 1998. 95(3): p. 831–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Markett S, et al. , Assessing the function of the fronto-parietal attention network: insights from resting-state fMRI and the attentional network test. Hum Brain Mapp, 2014. 35(4): p. 1700–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Zanto TP and Gazzaley A, Fronto-parietal network: flexible hub of cognitive control. Trends Cogn Sci, 2013. 17(12): p. 602–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Barros-Loscertales A, et al. , Lower activation in the right frontoparietal network during a counting Stroop task in a cocaine-dependent group. Psychiatry Res, 2011. 194(2): p. 111–8. [DOI] [PubMed] [Google Scholar]
  • 59.Leech R and Sharp DJ, The role of the posterior cingulate cortex in cognition and disease. Brain, 2014. 137(Pt 1): p. 12–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Pearson JM, et al. , Posterior cingulate cortex: adapting behavior to a changing world. Trends Cogn Sci, 2011. 15(4): p. 143–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Müller DJ and Rizhanovsky Z, From the Origins of Pharmacogenetics to First Applications in Psychiatry. Pharmacopsychiatry, 2020. 53(4): p. 155–161. [DOI] [PubMed] [Google Scholar]
  • 62.Spellicy CJ, et al. , ANKK1 and DRD2 pharmacogenetics of disulfiram treatment for cocaine abuse. Pharmacogenet Genomics, 2013. 23(7): p. 333–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Kosten TR, et al. , Pharmacogenetic randomized trial for cocaine abuse: disulfiram and dopamine β-hydroxylase. Biol Psychiatry, 2013. 73(3): p. 219–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Zhang X, et al. , Pharmacogenetics of Dopamine β-Hydroxylase in cocaine dependence therapy with doxazosin. Addict Biol, 2019. 24(3): p. 531–538. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Freedman ML, et al. , Assessing the impact of population stratification on genetic association studies. Nat Genet, 2004. 36(4): p. 388–93. [DOI] [PubMed] [Google Scholar]
  • 66.Gomes CV, et al. , Adenosine receptors and brain diseases: neuroprotection and neurodegeneration. Biochim Biophys Acta, 2011. 1808(5): p. 1380–99. [DOI] [PubMed] [Google Scholar]
  • 67.Milaneschi Y, et al. , Polygenic dissection of major depression clinical heterogeneity. Mol Psychiatry, 2016. 21(4): p. 516–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Kaufling J, et al. , gamma-Aminobutyric acid cells with cocaine-induced DeltaFosB in the ventral tegmental area innervate mesolimbic neurons. Biol Psychiatry, 2010. 67(1): p. 88–92. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1643364-supplement-1.docx (239.5KB, docx)

RESOURCES