|Home | About | Journals | Submit | Contact Us | Français|
Poor response inhibition has been implicated in the development of alcohol dependence, yet little is known about how neural pathways underlying cognitive control are affected in this disorder. Moreover, endogenous opioid levels may impact the functionality of inhibitory control pathways. This study investigated the relationship between alcohol dependence severity and functional connectivity of fronto-striatal networks during response inhibition in an alcohol dependent sample. A secondary aim of this study was to test the moderating effect of a functional polymorphism (A118G) of the µ-opioid receptor (OPRM1) gene. Twenty individuals with alcohol dependence (6 females; 90% Caucasian; mean age = 29.4) who were prospectively genotyped on the OPRM1 gene underwent blood oxygen level-dependent (BOLD) functional magnetic resonance imaging (fMRI) while performing a Stop Signal Task (SST). The relationship between alcohol dependence severity and functional connectivity within fronto-striatal networks important for response inhibition was assessed using psychophysiological interaction (PPI) analyses. Analyses revealed greater alcohol dependence severity associated with weaker functional connectivity between the putamen and prefrontal regions (e.g., the anterior insula, anterior cingulate, and medial prefrontal cortex) during response inhibition. Further, the OPRM1 genotype was associated with differential response inhibition-related functional connectivity. This study demonstrates that individuals with more severe alcohol dependence exhibit less frontal connectivity with the striatum, a component of cognitive control networks important for response inhibition. These findings suggest that the fronto-striatal pathway underlying response inhibition is weakened as alcoholism progresses.
Increased impulsivity is implicated in the development and maintenance of substance use disorders (e.g., Volkow and Baler, 2012). The mechanisms underlying the association between alcohol use and impulsivity are complex and while high levels of impulsivity may serve as a predisposing etiological factor, long-term exposure to drugs of abuse can lead to neuroadaptation in the prefrontal cortex (PFC) and striatum, diminishing the brain’s ability to stop an impulsive response (de Wit, 2009; Jentsch and Taylor, 1999).
The fronto-striatal system has a prominent role in inhibitory control of impulsive responses (Jentsch and Taylor, 1999). Dysfunction of this system is associated with multiple pathological states involving excessive impulsive behavior, including Attention Deficit Hyperactivity Disorder (Rubia et al., 2005) and Tourette’s Syndrome (Marsh et al., 2007). Neuroimaging studies of Stop-Signal Tasks (SSTs) have identified several brain regions subserving the ability to stop an impulsive response, including components of fronto-basal ganglia circuitry implicated in the development of substance abuse. SSTs are designed to elicit a “race” between two response tendencies, “going” and “stopping” (Logan, 1994) with the Go process activating fronto-striatal-pallidal regions and the Stop process activating the inferior frontal gyrus (IFG), subthalamic nucleus (STN) region, pre-supplementary motor area (pre-SMA), globus pallidus pars interna, parietal cortex, and insula (Aron and Poldrack, 2006). Human and animal studies suggest response inhibition is mediated by a fast hyperdirect pathway in the right hemisphere connecting the IFG, the pre-SMA, and the STN (Aron and Poldrack, 2006). Moreover, evidence from preclinical rodent models (Eagle and Robbins, 2003) and functional connectivity analyses in healthy adults (Duann et al., 2009) provide support for the involvement of an indirect pathway connecting the cortex and caudate in successful response inhibition. Preclinical data suggests functional activity in the striatum is also directly influenced by input from the cortex (Brown, Smith and Goldbloom, 1998). Effective connectivity analyses found evidence supporting the complimentary actions of the hyperdirect and indirect pathways to implement successful inhibition of a prepotent response. These findings indicate that successful inhibition of the primary motor cortex relies on efficient fronto-basal ganglia communication and implicate a top-down controlled inhibitory process (Jahfari et al., 2011). Likewise, fronto-putamen connectivity has been identified as important for successful response inhibition (Zandbelt and Vink, 2010). Specific to addiction, the transition from PFC to striatal control over responding, and from ventral to more dorsal striatal subregions, has been proposed as the mechanism subserving the transition from voluntary/goal directed to more habitual drug seeking behavior (Everitt and Robbins, 2005). Thus, SST performance and related neural markers may be promising measures of risk for substance abuse (Nigg et al., 2006).
Several reports implicate decreased response inhibition in alcohol dependence, as measured behaviorally by the Continuous Performance Task (Bjork et al., 2004), Go/No-Go tasks (Kamarajan et al., 2005), and time required to cancel an initiated response in SSTs (Goudriaan et al., 2006); but see less supportive studies (Courtney et al., 2012; Lawrence et al., 2009). To date however, only one SST brain activation study has been conducted with alcohol dependent participants. In this study, the neuroimaging measures of response inhibition, post-error behavioral adjustment, and error processing, were found to differentiate abstinent alcohol dependent patients and healthy controls. More specifically, the authors observed decreased dorsolateral PFC activation during response inhibition and post-error slowing which is in accord with the transition from PFC to striatal control over responding and further implicates the role of efficient fronto-striatal connectivity in supporting response inhibition (Li et al., 2009).
Studies have found support for the modulating role of specific genetic variants of candidate genes subserving dopaminergic neurotransmission (i.e., DRD2, DRD4, and COMT) in the neural bases of impulsivity in heavy drinking and alcohol dependent samples (Boettiger et al., 2007; Filbey et al., 2011). Genes that affect the endogenous opioid system (e.g., OPRM1) represent plausible modulators of impulsivity, as increased levels of endogenous opioids have been associated with increased impulsivity (Ray et al., 2012). The µ-opioid receptor, which is encoded by the OPRM1 gene, has been identified as the primary site of action for opiates with high abuse potential (Pasternack, 1993), and non-opioid drugs such as alcohol exert some of their effects through activation of these receptors (Herz, 1997). The A118G polymorphism (Asn40Asp substitution) affects receptor activity for the endogenous ligand β-endorphin, such that the G-allele binds β-endorphin three times more strongly than A-allele. Individuals with the G-allele may display behavioral differences in responses mediated by β-endorphins at the more sensitive µ-receptors (Bond et al., 1998), and have been shown to experience greater subjective reinforcing effects after alcohol consumption (Ray and Hutchison, 2004). Preclinical evidence also supports the role of µ-opioid receptors in the regulation of inhibitory control (Olmstead, Ouagazzal and Kieffer, 2009; Wiskerke et al., 2011). In particular, OPRM1 knockout mice were found to exhibit increased motor impulsivity on a nose poke task (Olmstead et al., 2009), further implicating the role of the endogenous opioid system in inhibitory control.
Studies of the opioid antagonist naltrexone (NTX), an FDA approved medication for the treatment of alcoholism, and performance on tasks assessing behavioral impulsivity found NTX reverses a morphine-induced increase in preference for small immediate rewards over larger delayed rewards in rats (Kieres et al., 2004) and attenuates a variety of impulse disorders in humans (Kim et al., 2001). To date, one study has directly assessed the relationship between endogenous opioids and impulsivity in humans with problematic alcohol use (Mitchell et al., 2007). Mitchell and colleagues (2007) found enhanced response control (reduced motor errors during a conflict task) in abstinent alcoholics treated with NTX. Individuals with an external locus of control showed a reduction in impulsive decision making when receiving NTX suggesting that endogenous opioids may impair response selection during impulsive decision-making, but that the effects of NTX on this process are personality-dependent and likely mediated by frontal dopaminergic transmission. Further investigation on the neural underpinnings of impulsivity, including the endogenous opioid system and frontal control systems, in alcohol dependence is warranted.
Together, preclinical findings and pharmacological manipulations have implicated the opioidergic system in inhibitory control. Therefore, this study examines the role of the A118G SNP of the OPRM1 gene on fronto-striatal connectivity during response inhibition.
In sum, regions within the hyperdirect and indirect pathways implicated in successful inhibitory control may be associated with alcohol-induced neuroadaptation through which response inhibition becomes impaired in alcohol dependent populations (Baler and Volkow, 2006). To test this hypothesis, this study employed a clinical neuroscience approach to investigate the relationship between alcohol dependence and the neural basis of response inhibition, focusing on fronto-striatal connectivity in particular, in individuals with and without the OPRM1 risk allele. A neuroimaging SST protocol was administered to 20 individuals with alcohol dependence who were prospectively selected based on OPRM1 genotype. Fronto-striatal functional connectivity during response inhibition was assessed with a psychophysiological interaction analysis using the right putamen as a seed region to identify brain areas that correlated with activation in this striatal region during successful stopping versus going. The right putamen was chosen as the a priori seed region due to its anatomical connections with the medial PFC, orbitofrontal cortex (OFC), and dorsolateral PFC (Draganski et al., 2008), as well as its functional role in motor control (Alexander, DeLong and Strick, 1986), including response inhibition (Aron and Poldrack, 2006; Zandbelt and Vink, 2010). The resulting contrasts and connectivity maps were then correlated with a measure of alcoholism severity, to assess for differential fMRI activation in brain regions involved in inhibitory control. Specifically, we hypothesized that fronto-striatal functional connectivity important for response inhibition would be weaker in participants with greater severity of alcohol dependence. The second aim of this study was to examine the moderating role of the A118G SNP on fMRI activation during response inhibition.
Participants were non-treatment seeking problem drinkers (N = 295) recruited from the Los Angeles community through flyers and online advertisements to investigate the effect of the OPRM1 gene on subjective responses to alcohol. The protocol was approved by the local Institutional Review Board, and following consenting procedures, participants were screened for alcohol dependence and prospectively genotyped. For the MRI portion of the study, a subsample of 20 alcohol dependent individuals was selected to ensure equal numbers of participants with and without the minor (G) allele of the OPRM1 gene (AA, n=10; AG/GG, n=10). Ethnicity was matched across groups to account for population stratification at the OPRM1 locus. Inclusion criteria were: (1) ages 21 to 55 years, (2) current alcohol dependence, (3) no major psychiatric disorders, (4) no current use of illicit substances (other than marijuana), verified by toxicology screening, and (5) no abuse or dependence on any illicit substance (including marijuana) as defined by the DSM-IV in the past 12 months. Abstinence from alcohol at least 24 hours prior to their scan time was required, verified by a Breathalyzer test (Dräger, Telford PA).
The OPRM1 genotype groups were found to be balanced across demographic variables, as analyzed by independent t-tests (ps>0.05; Table 1). Alcohol use was assessed using the 30-day timeline follow-back (TLFB; Sobell and Sobell, 1980), resulting in estimates of alcohol drinks per drinking day and percent drinking days. Alcohol dependence and the exclusionary psychiatric diagnoses were assessed using the Structured Clinical Interview for DSM-IV (SCID; First et al., 1995) under the supervision of a licensed clinical psychologist (LAR). DSM-IV symptoms of alcohol abuse and dependence were recorded for a total of 11 possible symptoms. All participants completed the Clinical Institute Withdrawal Assessment for Alcohol (CIWA-Ar; Sullivan et al., 1989); Alcohol Dependence Scale (ADS; Skinner and Allen, 1982), Drinkers Inventory of Consequences (DrInC-2R) questionnaire (Miller, Tonigan and Longabaugh, 1995), and the Penn Alcohol Craving Scale (PACS; Flannery, Volpicelli and Pettinati, 1999). No individuals reported clinically significant levels of alcohol withdrawal at time of assessment as indicated by CIWA-Ar score (scores ≤6).
To appropriately model the shared variance between the alcohol dependence severity indices and minimize the number of statistical tests, principal components analyses were conducted on the full sample (n=295) to derive factor scores capturing alcohol dependence severity. The principal factor method (promax oblique rotation) revealed one meaningful factor (first Eigenvalue=2.749, second Eigenvalue=0.858) with each index loading onto the factor at 0.40 or greater (ADS=0.83, PACS=0.74, Symptom Count=0.75, DrInC-2R=0.85, and CIWA-Ar=0.48) and accounted for 55% of the total variance. Participants’ scores on the single factor (alcohol dependence severity) were used in subsequent analyses. Participants’ ADS scores were also used in subsequent analyses for comparison with the alcohol dependence severity factor.
Participants performed a six-minute variant of the SST developed for neuroimaging protocols (Cohen et al., 2010). Participants had prior experience with the SST as part of the larger study for which they were recruited, thus no practice trials were given. While scanning, each of the 128 trials began with a white circular fixation ring presented in the center of the screen for 500ms, followed by presentation of a left- or right-pointing arrow. Participants were instructed to quickly press one of two buttons on a response pad corresponding to the arrow direction (Go trial). During Stop trials (32 trials), an auditory stop signal (900Hz, 500msec) sounded at varying delays after Go stimulus onset, signaling the participant to attempt to inhibit their response. A jittered delay (0.5–4s, mean=1s, taken from an exponential distribution) followed each response. The time intervals between the go and the stop signals (i.e., the stop-signal delay [SSD]) were determined dynamically during scanning using a staircase approach with two independent ladders. Ladder one started at 250ms and ladder two at 350ms. The SSD was either increased or decreased by 50ms on the next trial depending on whether the participant succeeded or failed in withholding their response, respectively.
The presentation of all stimuli and response collection were programmed using MATLAB (Mathworks, Natick, MA) and the Psychtoolbox (www.psychtoolbox.org) on an Apple MacBook running Mac OSX (Apple Computers, Cupertino, CA). Visual and auditory stimuli were presented using MRI compatible goggles and headphones (Resonance Technologies, Van Nuys, CA).
Neuroimaging was conducted using a 3 Tesla Siemens Trio MRI scanner at the UCLA Ahmanson-Lovelace Brain Mapping Center. The protocol began with initial structural scans followed by a series of four functional runs, including the SST, an alcohol taste-cues paradigm, a delay-discounting paradigm, and a risky decision-making paradigm. A T2-weighted, high resolution, matched-bandwidth, anatomical scan (MBW) and a magnetization-prepared rapid-acquisition gradient echo (MPRAGE) were acquired for each subject to enable registration (TR, 1.9s; TE, 2.26ms; FOV, 250mm; matrix, 256×256; sagittal plane; slice thickness, 1mm; 176 slices). The orientation for MBW and echoplanar image (EPI) scans was oblique axial to maximize brain coverage. The SST scan included 184 functional T2*-weighted EPIs (slice thickness, 4mm; 34 slices; TR, 2s; TE, 30ms; flip angle, 90°; matrix, 64 × 64; FOV, 192mm; voxel size, 3×3×4mm3). The first six volumes collected were discarded to allow for T1 equilibrium effects.
FSL 4.1 (FMRIB's Software Library, www.fmrib.ox.ac.uk/fsl) was used for the imaging analyses. Motion correction was carried out using the Motion Correction Linear Image Registration Tool (McFLIRT, Version 5.0) with the estimated motion parameters entered as covariates in the general linear model. Non-brain tissue/skull removal was conducted with the Brain Extraction Tool (BET). The images were smoothed using a FWHM Gaussian kernel (5mm) and high-pass filtered (100s cutoff) in the temporal domain using a Gaussian weighted straight line with the FMRI Expert Analysis Tool (FEAT, Version 5.63). The EPI images were first registered to the MBW, then to the MPRAGE using affine linear transformations, and into standard (Montreal Neurological Institute, MNI avg152 template) space for between-subject analyses. Registration to standard space was refined by FSL’s FNIRT nonlinear registration (Andersson, Jenkinson and Smith, 2007). One subject (G-allele carrier) was excluded from further analyses due to excessive motion (exceeding 3mm of translation).
Saliva samples were collected using Oragene saliva collection kits (Kanata, Ontario, Canada) and sent to the UCLA Genotyping and Sequencing Core for genotyping. Polymerase chain reaction was performed on Applied Biosystems (Carlsbad, CA) dual block PCR thermal cyclers. Single-nucleotide polymorphisms were run on an AB 7900HT Fast Real-Time PCR System and analyzed using the Sequence Detection Systems software (Version 2.3). Genotypes were automatically scored by the allele calling software, verified by visual inspection.
SST performance was assessed for the following criteria: average percent inhibition on Stop trials between 40 and 60%, response on greater than 80% of Go trials, less than 10% arrow direction errors, and greater than 50ms SSRT (Congdon et al., 2010). All subjects’ SST performance met the specified criteria and the staircase procedure employed resulted in participant performance at the desired 50% inhibition level (Table 2).
Whole-brain statistical analysis was performed using a multi-stage approach to implement a mixed-effects model treating participants as a random-effects variable. Explanatory variables for the SST paradigm were created by convolving stick functions representing the onset of relevant experimental events with a double-gamma hemodynamic response function in FEAT. The events modeled included: Go, Successful Stopping, and Unsuccessful Stopping, all with 1.5 s duration from stimulus onset. Temporal derivatives were included as covariates of no interest to improve statistical sensitivity. Null events, consisting of the jittered inter-stimulus interval when the screen was blank, were not explicitly modeled and therefore constituted an implicit baseline. The following contrasts indexed response inhibition: (a) Successful Stopping versus Go, (b) Unsuccessful Stopping versus Go, and (c) Successful Stopping versus Unsuccessful Stopping. To examine activation related to the Go and Stop processes, (a) Go versus baseline and (b) Successful Stopping versus baseline contrasts were also computed.
Second-level group analyses were conducted on contrast images transformed into standard space. Z-statistic images were thresholded with cluster-based corrections for multiple comparisons based on the theory of Gaussian Random Fields with a cluster-forming threshold of Z=2.3 and a cluster-probability threshold of p<0.05 (Worsley, 2001). Alcohol dependence severity factor scores and ADS scores were modeled as explanatory variables on the whole-brain contrast maps as specified above. Anatomical localization of peak voxels within each cluster (maximum Z statistics and MNI coordinates) was obtained by searching within maximum likelihood regions from the FSL Harvard-Oxford probabilistic atlas. OPRM1 genotype (i.e., AA and AG/GG) was entered as a second level predictor variable and examined in relation to brain activation within SST contrasts using a whole-brain approach.
Functional connectivity was assessed using psychophyisiological interaction (PPI) analysis (Gitelman et al., 2003) which measures coupling of brain regions during specific task conditions. To examine fronto-striatal functional connectivity during response inhibition, we examined the coupling of the right putamen and the rest of the brain within the Successful Stopping versus Go contrast. Both anatomically and functionally defined right putamen seed regions were used. To determine the anatomically defined seed for each participant, we used the high-resolution MPRAGE anatomical images, segmented on a subject-specific basis in native space using FMRIB's Integrated Registration and Segmentation Tool (FIRST) in FSL. For the functionally defined seed, we included voxels from the thresholded Z-statistic images for the Successful Stopping versus Go contrast that fell within the boundaries of the anatomically defined putamen. The anatomically and functionally defined left putamen seed regions were also determined using the same methods. The average time course of the right and left putamen were extracted from motion-corrected, high-pass filtered image data (same pre-processing steps as outlined above). The PPI analysis was conducted using both SPM and FSL’s FEAT. The model was identical to the first-level model described above with the inclusion of three additional regressors: ‘psychological’, ‘physiological’, and ‘psychophysiological interaction’. These regressors were generated separately by computing three vectors using the PPI algorithms implemented in SPM5 (http://www.fil.ion.ucl.ac.uk/spm/software/spm5). The psychological vector was specified by a delta function with Successful Stopping events represented by 1 and Go events represented by −1 (zero centered), the physiological vector estimated ‘neural’ activation of the right putamen via hemodynamic deconvolution of the average pre-processed time course, and the psychophysiological interaction vector was the product of these two. The three vectors were convolved with a hemodynamic response function prior to inclusion in the FEAT model. A whole-brain contrast image for the PPI was computed from this model and submitted for second-level group analyses described above.
Alcohol use, dependence severity, and task performance are presented in Table 2. Of note, the sample was found to have mild-to-moderate alcohol dependence as compared to other alcohol dependent samples reported in the literature. No group differences were observed across OPRM1 genotype for these variables (ps>0.10). Across all subjects, behavioral performance on the task was found to be uncorrelated with alcohol dependence severity (ps>0.10).
Consistent with previous reports, the main contrast examining response inhibition (Successful Stopping versus Go) activated a broad set of brain regions including the IFG, pre-SMA, parietal cortex, putamen, and insula (Aron and Poldrack, 2006) (Table 3, Figure 1). No genotype differences were observed within this contrast. Results from other task contrasts are reported in Table S1.
To assess whether alcohol dependence severity is associated with response inhibition, whole-brain correlation analyses were conducted within the Successful Stopping versus Go, Successful Stopping versus Unsuccessful Stopping, and Successful Stopping versus baseline contrasts. No significant correlations or genotype differences among correlations were found between alcohol dependence severity (factor scores or ADS scores) and these contrasts.
Fronto-striatal functional connectivity using PPI contrasts describing the connectivity between the right putamen and the rest of the brain from the Successful Stopping versus Go contrast was estimated for all subjects. The right putamen showed significant connectivity (positive correlation) with the subcallosal/anterior cingulate cortices (ACC) and the paracingulate gyrus (Table 4, Figure 2A). OPRM1 genotype was found to moderate fronto-striatal connectivity such that A-allele homozygotes exhibited less functional connectivity between the right putamen and prefrontal regions, including OFC and middle-frontal regions (parameter estimates for the cluster: A-allele homozygotes = −0.258, G-allele carriers = 0.182; Table 5).
Whole-brain correlations of alcohol dependence severity measures (i.e., severity factor scores and ADS scores) with the PPI contrasts were performed using the anatomically and functionally defined seed regions. Regions from the anatomically-defined right putamen seed analysis showing a negative correlation with severity factor scores were restricted to the PFC and included the left anterior insula, bilateral IFG, OFC, and ACC (Table 4, Figure 2). No regions showed a positive correlation with alcohol dependence severity. Similar results were obtained when using the functionally-defined right putamen as the seed region and when using ADS scores instead of the alcohol severity factor scores (see supplementary information). Whole-brain correlations of alcohol dependence severity factor scores with the PPI contrasts using the anatomically and functionally defined left putamen as seed regions were also performed; however, consistent with the previous findings of a largely right lateralized response inhibition network (e.g., Aron and Poldrack, 2006), no significantly correlated regions of activation were observed for either left seed region. The anatomically-defined right putamen seed region was used for all further PPI analyses as we sought to determine fronto-striatal connectivity that was unbiased by variations in striatal functional activation. In addition, the influence of participant age was investigated through an exploratory correlation of alcohol dependence severity with the Successful Stopping versus Go PPI with the age range restricted to 21–33 years old (resulting in n=16). The restricted age range results were found to be virtually identical to that of the full sample.
OPRM1 genotype was found to moderate the PPI and alcohol dependence severity correlation in bilateral fronto-polar cortex, superior frontal gyrus, middle frontal gyrus, IFG, right caudate, OFC and right occipital pole/fusiform gyrus, such that A-allele homozygotes exhibited a steeper regression slope describing the relationship between alcohol dependence severity and functional connectivity with the right putamen. In contrast, G-allele carriers showed a steeper slope in the superior frontal and parietal gyri, precuneus, and bilateral precentral gyri as compared to A-allele homozygotes (Table 5).
This study found an association between alcohol dependence severity and fronto-striatal functional connectivity, such that individuals with more severe alcohol dependence exhibited less fronto-striatal connectivity during response inhibition on the SST. Further, OPRM1 genotype was found to moderate functional connectivity between the right putamen and prefrontal regions, as well as the correlation strength of fronto-striatal connectivity with alcohol dependence severity in frontal and posterior regions during response inhibition.
The SST has proven useful in the study of addictive disorders since a loss of cognitive control, including diminished response inhibition, has been hypothesized to accompany the transition from voluntary to compulsive drug taking behavior (Kalivas and Volkow, 2005). Differential neural activation during SST performance has been found to relate to heaviness of smoking in adolescent smokers (Galván et al., 2011) and risk status for alcohol use disorders (i.e., low alcohol responders; Schuckit et al., 2011). It has also been found to discriminate individuals with alcohol use disorders from healthy controls (Li et al., 2009). Results from a sample of stimulant abusers and their non-drug using siblings indicate that self-control deficits on the SST in both groups are associated with white matter disorganization in right PFC. Furthermore, duration of drug exposure was found to have effects on white matter organization, albeit less anatomically extensive. Together, these findings suggest that deficiencies in response inhibition may represent substance-induced neuroadaptation as well as a heritable risk marker for substance abuse (Ersche et al., 2012).
In the present study, alcohol dependence severity was associated with diminished fronto-striatal functional connectivity during response inhibition. Reduced fronto-striatal connectivity has been previously associated with abnormal reward prediction error signaling during decision making in alcohol dependent individuals (Park et al., 2010), and de-synchrony of cortico-striatal-midbrain activation in alcoholics was observed during inhibitory control on a Stroop task (Schulte et al., 2012), suggesting an important role of this pathway in modulating impulsive behavior. Although PPI analyses preclude determining directionality of connectivity, evidence from an effective connectivity analysis points to a top-down controlled inhibitory process, mainly directed from prefrontal regions, during successful response inhibition (Jahfari et al., 2011), suggesting that the negative relationship between fronto-striatal connectivity and alcoholism severity reflects weaker prefrontal control of the striatum. This is consistent with the clinically observed transition to later stages of addiction, which is marked by loss of control of a prepotent response, namely alcohol use (Kalivas and Volkow, 2005).
Precisely how fronto-striatal functional connectivity becomes impaired in alcohol dependence remains unknown. The consistent finding of white matter damage, including demyelination and axonal subtraction (Kril et al., 1997), within chronic alcohol abusing samples represents one possible explanation for the weakened functional connectivity observed. To our knowledge, no studies have reported white matter damage directly related to fronto-striatal connectivity, however, alcohol is found to affect white matter brain regions differentially, with the frontal lobes identified as among the more compromised regions (Harper et al., 2003; Kril et al., 1997; Pfefferbaum et al., 1997). Thus, it is possible that alcohol induced damage to white matter tracks directly involved in the connectivity between the frontal lobes and striatum may exist, just not at detectable levels given our current approaches. Alternatively, alcohol’s desynchronizing effects within and between neural networks (Ehlers, Wills and Havstad, 2012) may be compromising fronto-striatal communication, and following prolonged use, this desynchrony may result in impaired functional connectivity between these brain regions.
While SST behavioral performance was uncorrelated with alcohol dependence severity in this sample, a larger sample may reveal behavioral differences that accompany the neural disparities observed. Conversely, the BOLD signal may represent a more sensitive marker of group differences than behavioral phenotypes (i.e., task performance) and may partially explain the discrepancies in behavioral response inhibition previously observed in similar populations (Courtney et al., 2012; Lawrence et al., 2009). Thus, the correlations observed between severity and connectivity highlight a role for connectivity measures as relevant markers for alcoholism.
Genetic analyses found that G-allele carriers of the OPRM1 gene exhibited greater functional connectivity of the putamen to the caudate, orbitofrontal, and middle frontal regions, suggesting that the A-allele homozygotes may display weaker prefrontal inhibitory control through this pathway at matched levels of clinical impairments. The cluster parameter estimates would suggest that the genotype differences observed in the fronto-striatal functional connectivity during response inhibition act to cancel each other out, resulting in no observable connectivity for these regions in the main effect analysis. This result highlights the importance of considering individual differences variables, including genotype, in addicted samples. Furthermore, OPRM1 genotype was found to moderate the association between alcoholism severity and striatal connectivity in frontal and posterior regions. Specifically, A-allele homozygotes of the OPRM1 gene showed a stronger negative association between alcohol dependence severity and connectivity to the right caudate and OFC during response inhibition suggesting that the fronto-striatal pathway is more strongly affected by alcoholism in these individuals at similar behavioral levels of inhibitory control to the G-allele carriers. These results suggest that the G-allele of the OPRM1 gene, which is typically thought of as a risk allele for a host of alcohol phenotypes, may not be a risk factor through pathways of alcoholism severity and inhibitory control. Instead, the risk conferred by the G-allele may be unique to goal directed alcohol use mediated through neural reward processes (Ray et al., 2012). Thus, these different pathways influencing risk for the development and maintenance of alcoholism (i.e., reward sensitivity versus inhibitory control deficits) may account for the seemingly disparate findings.
This study must be considered in light of its strengths and limitations. A strength of this study is the well-ascertained sample of individuals with alcohol dependence. Although prospective genotyping in this study precluded the use of a control group, recent studies of response inhibition in substance users, suggested the majority of variance lies within the substance-using group (Filbey et al., 2011; Galván et al., 2011; Li et al., 2009). Thus, the within-group analysis performed here avoids extraneous sources of variation and allows investigation of the relationship between alcohol dependence severity and fronto-striatal connectivity. The use of a factor score comprised of various clinical dimensions of alcoholism represents another significant strength. The severity factor allowed for the global examination of several important dimensions of alcoholism, including craving, DSM-IV symptoms, and the experience of a host of negative consequences related to alcohol dependence, while minimizing the number of statistical comparisons. The small sample size represents a weakness of the current study. However, for the genotype analyses, the genotype–balanced sample mitigated some of the statistical power issues that may have arisen if post-hoc genetic groupings were used. It should be noted, however, that the results obtained from the prospective genotyping groups may be different in a genetically unselected sample. Lastly, the exclusion of treatment seekers and individuals endorsing significant alcohol withdrawal symptoms led to a sample consisting of individuals with mild-to-moderate levels of alcohol dependence. Future research is needed to validate these findings in larger and more severe samples of alcohol dependence.
In conclusion, the results of this study provide the first evidence for alcoholism-mediated differences in fronto-striatal connectivity during response inhibition on the SST. These disparities are likely related to findings of Ersche and colleagues (2012) showing poor prefrontal white matter organization related to response inhibition in substance abusing individuals and their siblings. Thus, the observed association between alcohol dependence severity and reduced fronto-striatal connectivity suggests a plausible pathway of addiction vulnerability that may be moderated by impairments in inhibitory control and genetic variation subserving those systems.
The authors would like to thank Andia Heydari, Pauline Chin, Katy Lunny, and Ellen Chang for their contribution to data collection and data management for this project.
This study was supported by a grant from ABMRF, the Foundation for Alcohol Research, awarded to the senior author, Dr. Lara Ray, and a grant from NIAAA (1R03-AA019569). Dr. Lara Ray is a paid consultant to Glaxo Smith Kline.
Author ContributionsLAR was responsible for the study concept and design. LAR and KEC contributed to the acquisition of MRI data. KEC and DGG assisted with data analysis and interpretation of findings. KEC drafted the manuscript. LAR and DGG provided critical revision of the manuscript for important intellectual content. All authors critically reviewed content and approved final version for publication.