Error-Related Neural Activity and Alcohol Use Disorder: Differences from Risk to Remission

Abstract

Studies suggest that individuals with alcohol use disorder (AUD) display abnormal neural error-processing, measured via the error-related negativity (ERN). The nature of the error-related abnormalities in AUD is unclear, however, as prior research has yielded discrepant findings. In addition, no study to date has attempted to characterize the dispositional nature of the ERN in AUD and directly test to what extent ERN amplitude reflects a risk factor, disease marker, and/or scar of AUD psychopathology. The current study compared ERN amplitude across 244 adult volunteers in the following five groups: 1) current AUD (n=39), 2) AUD in remission (n=60), 3) at-risk for AUD (n=43), 4) psychiatric controls with comparable rates of internalizing psychopathology as the AU groups (n=53), and 5) healthy controls with no lifetime history of psychopathology (n=49). Risk for AUD was defined as a positive, first-degree family history. All participants completed a well-validated flanker task, designed to robustly elicit the ERN, during continuous electroencephalographic (EEG) data collection. Results indicated that individuals with current AUD displayed smaller ERNs compared with individuals at-risk for AUD, with AUD in remission, psychiatric controls, and healthy controls. There were no differences amongst any of the other groups. This suggests that a blunted ERN may be concomitant with current AUD psychopathology and relatedly, a novel neurobiological AUD treatment target and/or objective marker of AUD disease status.

1. Introduction

Neural error-monitoring has been identified as a biologically-based dimension of psychological function underlying key features of psychopathology (Hanna & Gehring, 2016; Weinberg et al. 2015). In fact, the National Institute of Mental Health (NIMH) Research Domain Criteria (RDoC) Initiative has included neural error-monitoring as a core construct within both the ‘Negative Valence: Sustained Threat” and ‘Cognitive Systems: Performance Monitoring’ categories (see https://www.nimh.nih.gov/research-priorities/rdoc/index.shtml). One way that neural response to errors is measured is using the error-related negativity (ERN), an event-related potential (ERP) that appears as a negative-going deflection in the electroencephalogram (EEG) waveform following the commission of an error in speeded response tasks (Falkenstein et al. 1991; Gehring et al. 1993). The ERN is well-validated and observed across various levels of task difficulty, age ranges, and species (Endrass et al. 2012; Riesel et al. 2013); although there is still some debate regarding the psychological mechanisms underlying the ERN. Of the several theories that have been proposed (Botvinick et al. 2001; Holroyd & Coles, 2002; Weinberg, Riesel, & Hajcak, 2012), most agree that the ERN indexes the functioning of a performance monitoring system designed to flexibly control and modify behavior after a mistake has been made.

To date, the ERN has been most extensively studied with regard to anxiety-related psychopathology. Studies show that individuals with obsessive compulsive disorder (OCD; Gehring et al. 2000), generalized anxiety disorder (GAD; Weinberg et al. 2010), and social anxiety disorder (SAD; Endrass et al. 2014), display an enhanced ERN relative to healthy controls. The same pattern is true for individuals with high trait anxiety (Olvet & Hajcak, 2009) and increased worry (Hajcak et al. 2003). Interestingly, an enhanced ERN may be an endophenotype for anxiety psychopathology as it is stable over time (Weinberg & Hajcak, 2011), moderately heritable (Anokhin et al. 2008), precedes disorder onset (Meyer et al. 2018), and is resistant to changes in anxiety symptoms (Riesel et al. 2015).

There is a much smaller, emerging literature also implicating the ERN in substance use disorders (SUDs), including alcohol use disorder (AUD). It has been shown that individuals with cocaine dependence (Franken et al. 2007), nicotine dependence (Luijten et al. 2011), and externalizing personality traits such as impulsivity and disinhibition (Hall et al. 2007; Potts et al. 2006), evidence smaller ERNs relative to controls. In a sample of individuals with current AUD, Fein and Chang (2008) also found that smaller ERNs were associated with greater family history density of alcohol problems. Similarly, Euser and colleagues (2013) reported that offspring of parents with SUDs, who were cannabis use-naïve, displayed smaller ERNs relative to a low-risk control group. Researchers have therefore hypothesized that diminished error monitoring may be a biomarker of for SUDs.

The imaging literature also suggests that substance users display reduced error-related neural activity. In humans, one of the primary sources of the ERN has been localized to the anterior cingulate cortex (ACC; Miltner et al. 2003), and studies using functional magnetic resonance imaging (fMRI) show that individuals who use cocaine (Goldstein et al. 2007), opiates (Forman et al. 2004), marijuana (Gruber & Yurgelun-Todd, 2005), and methamphetamine (London et al. 2005) display ACC hypoactivity in response to the commission of an error. It has also been shown in healthy individuals that administration of low and moderate doses of alcohol decreases neural error-reactivity (Easdon et al. 2005; Ridderinkhof et al. 2002). Thus, converging evidence suggests that substance use is associated with impaired error-processing.

In contrast to the findings above, a few studies examining AUD, in particular, suggest the opposite direction of effects – i.e., AUD is associated with an enhanced ERN. Padilla and colleagues (2011) reported that males with remitted AUD exhibited larger ERNs compared with controls. A previous study by our lab found that veterans with a diagnosis of post-traumatic stress disorder (PTSD) and current AUD had larger ERNs relative to veterans with PTSD without AUD (Gorka et al. 2016). Schellekens and colleagues (2010) similarly found that recently abstinent males with AUD, both with and without co-occurring anxiety disorders, displayed larger ERNs relative to healthy controls, and that the AUD participants with anxiety disorders had larger ERNs relative to the AUD participants without anxiety disorders. The authors therefore posited that anxiety and AUD had additive effects on the ERN. Lastly, in a sample of undergraduates, binge drinkers without a history of AUD demonstrated increased ERNs compared with non-binge drinkers (Lannoy et al. 2017).

Taken together, the majority of existing studies suggest that SUDs are associated with blunted neural error-reactivity; however, the AUD findings suggest the opposite pattern of results, raising the possibility that individuals with AUDs display unique neurobiological characteristics compared with the other SUDs. However, the existing literature is notably complicated by several factors which limits the conclusions that can be drawn regarding the direction of ERN effects in AUD. Most importantly, several studies failed to take into account comorbid anxiety disorders, which are known to enhance neural reactivity to errors and could drive exaggerated responding within certain AUD/SUD samples. In order to clarify the role of error-processing in AUD, it is therefore necessary to directly account for anxiety psychopathology in AUD versus non-AUD group comparisons.

In addition to understanding the direction of ERN effects in AUD, no study to date has compared individuals at various stages of AUD psychopathology, from risk to remission, and it is unknown the extent to which the ERN changes along with changes in AUD illness. It is therefore still unclear if the ERN is a risk factor, state marker, or scar (i.e., consequence) of AUD. The anxiety literature clearly indicates that an enhanced ERN is a stable risk factor that precedes disorder onset and is observed during times of illness and remission (Manoach, & Agam, 2013; Olvet & Hajcak, 2008). It is possible that the ERN functions similarly with regard to AUD; however, chronic alcohol exposure is known to cause neuroadaptions within emotion and cognitive control circuits of the brain, including frontocortical pathways involving the ACC (Koob & Volkow, 2010). AUD illness and alcohol exposure may therefore change neural error-processing and it is unknown if such alterations resolve upon AUD remission. Similarly, only two studies to date have directly addressed the topic of risk (i.e., Fein & Chang, 2008; Euser et al., 2013) but did not account for comorbid anxiety or directly compare the ERN across individuals at-risk for AUD, with current AUD, and AUD in remission. In order to truly clarify the functional role and potential clinical utility of the ERN in AUD, it is critical to compare individuals from risk to remission while also accounting for other factors known to impact the ERN, particularly anxiety.

The study was designed to address the gaps in the ERN AUD literature by comparing ERN amplitude across individuals in the following five groups: 1) current AUD, 2) AUD in remission, 3) at-risk for AUD, 4) psychiatric controls, and 5) healthy controls. Risk for AUD was defined as no personal lifetime diagnosis of AUD but a positive first-degree family history of AUD (Kendler et al. 1997). For the three AUD groups (current, remitted, and at-risk), comorbid internalizing psychopathology was not exclusionary in order to increase the external validity of our sample. Thus, the ‘psychiatric controls’ represent a group of individuals who were also permitted to have internalizing psychopathology and were selected to have comparable rates of lifetime anxiety and depression as the AUD groups but no personal or first-degree family history of AUD. Lastly, the ‘healthy controls’ include individuals with no lifetime history of any psychiatric disorder and no first-degree family history of AUD and thus, represent a comparison for all other groups. Given the discrepancies between the AUD and SUD literatures, we did not have specific hypotheses about the direction of effects though we speculated that all three AUD groups would evidence abnormal ERN amplitude relative to the psychiatric and healthy controls.

2. Methods

2.1. Participants and Procedure

Participants were drawn from two larger studies examining individual differences in threat responding. Both studies were conducted at the University of Illinois at Chicago, used similar recruitment techniques, enrolled only young adults, and had identical laboratory protocols, making them well-suited for combined analyses. All participants were recruited via advertisements posted in the Chicago community, local psychiatric clinics, and nearby college campuses. Of the 244 volunteers included in the current study, 212 came from Sample 1 and 32 came from Sample 2. Both protocols were approved by the university Institutional Review Board and participants provided written informed consent. In both studies, participants completed a set of laboratory tasks, a battery of questionnaires and a semi-structured clinical interview, and received cash as payment for participation. For both protocols individuals were instructed to abstain from drugs and alcohol at least 24-hours prior to the assessment and alcohol abstinence was verified via breath alcohol screens. For Sample 2 (only) a urine drug screen was also administered and all individuals tested negative the day of the psychophysiological assessment.

Sample 1 was designed to examine affective responding within families and thus, required the enrollment of biological, sibling dyads. Participants were required to be between the ages of 18 and 30, be able to provide consent, and have at least one biological sibling willing and able to participate. Exclusion criteria included a personal or family history of mania or psychosis, a major medical or neurological illness, a history of serious head trauma, and left-handedness. Participants were not required to have any specific diagnoses; however, various advertisements were used in an effort to enroll a diverse sample, including individuals with current and remitted AUD. Of the 212 individuals from Study 1, 37 were healthy controls, 52 were psychiatric controls, 43 were at risk for AUD, 55 had remitted AUD, and 25 had current AUD. Of note, the aims of the larger study dictated that ERN data was collected from two siblings per family. For the purposes of the current study, only one individual within the sibling dyad was assigned to one of the five groups (i.e., the proband) and the other was treated as a family member providing diagnostic data similar to mother, father, and additional siblings (thus, analyses did not violate the assumption of independence of observation). Individuals with current AUD, remitted AUD, and at-risk for AUD were assigned to groups first. Instances where both siblings met criteria for one of the AUD groups, one participant was chosen to be in ERN analyses at random and the other was assigned as a family member. Next, individuals with a current internalizing disorder, but no personal or family history of AUD, were assigned to the ‘psychiatric control’ group and only one participant was chosen at random in cases where both siblings met criteria. Lastly, individuals with no lifetime history of internalizing psychopathology, and no personal or family history of AUD, were assigned to ‘psychiatric controls’ or ‘healthy controls’ at random and the sibling was assigned to be a family member. Of note, the psychiatric control group also includes healthy controls as the goal was to create a group that was not necessarily ‘ill’, but had rates of internalizing psychopathology that were comparable to the AUD groups.

Sample 2 was designed to examine the role of threat responding in AUD, specifically. Participants were therefore required to belong to one of three group: 1) current AUD, 2) remitted AUD, and 3) controls with no personal or family history of AUD. Participants were also required to be between 21 and 30 years old, and able to provide written informed consent. Exclusion criteria included any serious medical condition, psychotropic medication use within the past four months, deafness, contraindication for neuroimaging, pregnancy, lifetime moderate or severe SUD (other than alcohol and nicotine), and psychosis. Of the 32 individuals from the second sample, 12 were healthy controls, 1 was a psychiatric control, 5 had remitted AUD, and 14 had current AUD.

A total of 244 individuals were therefore included in the present study. The sample size of each group was as follows: 39 current AUD, 60 AUD in remission, 45 at-risk for AUD, 53 psychiatric controls, and 49 healthy controls.

2.2. Assessment of Psychopathology and Substance Use

Lifetime diagnoses of Axis I disorders for both samples were assessed via the Structured Clinical Interview for DSM-5 Disorders (SCID-5; American Psychiatric Association [APA], 2015), in-person, by trained assessors, and supervised by a licensed clinical psychologist. Thus, AUD remission was defined as a period of ≥3 months without meeting AUD diagnostic criteria according to the DSM-5.With regard to family history data, for Sample 1, first-degree biological family members were contacted and offered the opportunity to enroll in the study and complete the SCID in-person or via telephone. In-person and telephone diagnostic interviews have high inter-method reliability (Rohde et al. 1997). On average, 1.9±0.9 additional family members completed the interview. For Sample 2, all participants completed the Family History Screen (FHS; Weissman et al. 2000). The FHS is a structured interview assessing lifetime history of 15 different psychiatric disorders, including AUD, in each of the subject’s first-degree family members.

Across both studies, individuals completed a standard health screen questionnaire which included items regarding recent frequency of illicit substance use. Specifically, participants were asked to indicate how many times they used a particular substance (e.g., cannabis, cocaine, heroin) within the past 30 days. Recent alcohol use was assessed during the clinical interviews using the Time-Line Follow-Back technique (Sobell & Sobell, 1992). Participants were presented with a calendar of the past 30-days and asked to indicate on what days they drank and how many drinks they had. Frequency of binge episodes was calculated using the binge definition of having ≥5 standard drinks for men, and ≥4 drinks for women, in one sitting (NIAAA, 2014).

2.3. Flanker Task

Participants completed a modified arrowhead version of the original flanker task (Eriksen & Eriksen, 1974) to measure neural activity to error and correct responses (i.e., the ERN and correct response negativity [CRN], respectively). For each trial, participants viewed five horizontally aligned arrowheads. For half of the trials, arrows were compatible (“>>>>>” or “<<<<<”) and for the other half, the arrows were incompatible (“>>▱>” or “<▱<<”). Participants were instructed to respond as quickly and accurately as possible to indicate the direction of the center arrow (left or right) by pressing the appropriate mouse button. Arrows were presented for 200ms and participants were given up to 1800ms to respond. Trials were followed by an intertrial interval (1000-2000ms) during which a fixation cross appeared. The task consisted of 11 blocks of 30 trials (330 total trials). Speed and accuracy were emphasized during task instruction, and throughout the experiment as participants received performance-based feedback at the end of each block. If accuracy was 75% correct or lower, the message “Please try to be more accurate” was presented; if accuracy was greater than 90%, the message, “Please try to respond faster” was displayed; in all other cases, participants saw the message, “You’re doing a great job”.

2.4. EEG Data Collection and Processing

For both samples, continuous EEG was recorded during the task using the ActiveTwo BioSemi system (BioSemi, Amsterdam, Netherlands). Sixty-four electrodes were used for the first sample, based on the 10/20 system, as well as two electrodes on the right and left mastoids. Thirty-four standard electrode sites were used for the second sample and one electrode was placed on each mastoid. Off-line analyses were identical across the two studies and performed using Brain Vision Analyzer 2 software (Brain Products, Gilching, Germany). Data were re-referenced to the average of the two mastoids and high-pass (0.1Hz) and low-pass (30Hz) filtered. Data were segmented beginning 500ms before each response onset and continuing for 1500ms. Standard artifact rejection procedures were used (see Gorka et al, 2016, 2017). Baseline correction for each trial was performed using the 500 to 300ms prior to response onset.

ERN data was considered unusable if the EEG data were contaminated by excessive artifact and/or the participant made fewer than six errors during the flanker task (Olvet & Hajcak, 2009). The ERN and CRN were scored as the average activity on error and correct trials, respectively, from 0 to 100ms after response at electrode Cz, as this was the electrode where the ERN was maximal across all subjects, consistent with prior studies (Gorka et al. 2017; Meyer et al. 2013). The ERN and CRN were correlated (r = .43, p < .01), as expected. Therefore, to quantify the difference between error and correct trials, we followed guidelines by Meyer et al. (2017) and calculated an ERN standardized residual score (ERN_resid) by saving the variance leftover in a regression where the CRN was entered predicting the ERN. The ERN_resid was used as the primary variable in subsequent analyses.

Although several issues regarding the use of ERP difference scores have recently been outlined (see Meyer et al. 2017), traditionally, the ERN has been quantified as the difference between ERN and CRN (i.e., ERN – CRN; ΔERN). In fact, almost all of the prior studies examining the ERN in the context of AUD have used the ΔERN and thus, in-order to facilitate comparisons across studies, we also present secondary results comparing the groups on ΔERN.

Behavioral data from the task included the number of error trials for each subject, accuracy expressed as a percentage of trials with correct responses out of total number of trials, and average reaction times (RTs) on error and correct trials (separately).

2.5. Data Analysis Plan

We first tested whether the groups differed in demographic and clinical variables using a series of planned analyses of variance (ANOVAs) and chi-square tests. Specifically, we examined group differences in age, sex, race/ethnicity, psychiatric medication use, and current and lifetime mood, anxiety, and SUDs.

We next examined group differences in the task behavioral data. We performed two analyses of covariance (ANCOVAs) where group (i.e., current AUD, remitted AUD, at-risk for AUD, psychiatric controls, and healthy controls) was specified as a between-subjects variable - one for task accuracy and one for RT. Lastly, to test our hypotheses, we conducted an additional ANCOVA examining the effect of group on ERN_resid (and ΔERN). In all ANCOVAs, current SUD diagnosis (yes/no) was included as a covariate given that the groups differed on rates of SUD (see below). Similarly, current smoking status (yes/no daily smoker) was included as a covariate We also created a variable reflecting current anxiety psychopathology known to influence the ERN (i.e., generalized anxiety disorder, social anxiety disorder, and/or posttraumatic stress disorder; yes/no; see Moser et al. 2013 for a review). Although OCD is not classified as an anxiety disorder, per se, we included current OCD diagnoses in the composite anxiety variable given the well-established relationship between OCD and enhanced ERN amplitude (Kathmann et al. 2016). Significant group effects were probed using post-hoc Fisher’s least significant difference (LSD) tests.

3. Results

3.1. Descriptives and Behavioral Data

Of the individuals in the current AUD group, 46.2% met criteria for mild, 48.7% for moderate, and 5.1% for severe AUD. For individuals in remission, 45% had a history of mild AUD, 36.7% had moderate, and 18.3% had severe. Individuals in the remitted AUD group had been in remission for 34.0 ± 29.1 months (range: 3–144 months). The groups did not differ on age, biological sex, or race/ethnicity. The four groups other than the healthy controls did not differ on current or lifetime major depressive disorder or any individual anxiety disorder (see Table 1). There were also no differences amongst the four non-healthy groups when current anxiety disorder diagnoses were grouped as a composite ‘any anxiety disorder’ (yes/no) variable (χ²[3]=0.30, p=.96) or when internalizing disorders were divided into fear-based disorders (χ²[3]=0.32, p=.96; i.e., panic disorder, social anxiety disorder, specific phobia) and distress/misery disorders (χ²[3]=2.46, p=.48; i.e., major depressive disorder, generalized anxiety disorder, post-traumatic stress disorder; Watson, 2005). The only difference between matched groups was for current and lifetime illicit SUD such that the current AUD group had greater rates of SUD than individuals in the at-risk for AUD and psychiatric control groups (Table 1).

Table 1.

Demographics and clinical characteristics

	Current AUD (n=39)	Remitted AUD (n=60)	At-Risk for AUD (n=43)	Matched Controls (n=53)	Healthy Controls (n=49)
Demographics
Age (years)	23.2 (3.2)a	23.2 (2.8)a	22.4 (3.6)a	22.2 (2.5)a	22.9 (3.2)a
Sex (% female)	51.3%a	55.0%a	58.1%a	60.4%a	59.2%a
Race/Ethnicity
Caucasian	53.8%a	51.7%a	58.1%a	32.1%a	46.9%a
African American	7.7%a	8.3%a	5.4%a	15.1%a	8.2%a
Hispanic	25.6%a	23.3%a	16.3%a	24.5%a	14.3%a
Asian	5.1%a	8.3%a	14.0%a	18.9%a	22.4%a
Other/Biracial	2.2%a	8.4%a	6.2%a	9.4%a	8.2%a
Substance Abuse
AUD Age of Onset	19.6 (3.5)a	19.1 (1.9)a	--	--	--
Alcohol Binges in Past Month	6.3 (4.2)a	1.6 (2.0)b	1.3 (2.3)b	0.2 (0.7)c	0.4 (0.9)c
Drinks Per Week in Past Month	14.5 (12.0)a	5.4 (5.7)b	3.8 (4.2)b,c	1.2 (1.7)c	1.8 (3.9)c
Daily Cigarette Smoker	12.8%a	11.7%a	11.3%a	3.8%a	4.1%a
Used Illicit Drugs in Past Month (yes/no)	35.9%a	23.3%a,b	11.6%b	3.8%b	4.1%b
No. Times Used Cannabis in Past Month	12.9 (39.0)a	4.3 (8.8)a	1.8 (5.9)a,b	0.1 (0.5)b	0.9 (4.0)b
No. Times Used Drug Other than Cannabis in Past Month	1.8 (4.7)a	0.9 (4.6)a,b	0.1 (0.5)b	0.1 (0.3)b	0.1 (0.4)b
Comorbid Diagnoses and Meds
Current SUD	15.4%a	8.3%a,b	4.7%a,b	3.8%b	0.0%b
Lifetime SUD	35.9%a	28.3%a,b	14.0%b,c	9.3%c	0.0%c
Current MDD	2.6%a	1.7%a	0.0%a	3.8%a	0.0%a
Lifetime MDD	33.3%a	43.3%a	39.5%a	34.0%a	0.0%b
Any Current Anxiety Disorder	28.2%a	30.0%a	25.6%a	26.4%a	0.0%b
Current Panic Disorder	5.1%a	6.7%a	0.0%a	1.9%a	0.0%a
Lifetime Panic Disorder	7.7%a	11.7%a	7.0%a	5.7%a,b	0.0%b
Current SAD	10.3%a	13.3%a	9.3%a	7.5%a	0.0%b
Lifetime SAD	15.4%a	21.7%a	16.3%a	15.1%a	0.0%b
Current Specific Phobia	15.4%a	11.7%a	18.6%a	15.1%a	0.0%b
Lifetime Specific Phobia	23.1%a	16.7%a	23.3%a	22.6%a	0.0%b
Current PTSD	2.6%a	1.7%a	0.0%a	0.0%a	0.0%a
Lifetime PTSD	17.9%a	10.0%a	9.3%a	7.5%a	0.0%b
Current GAD	7.7%a	6.7%a	2.3%a	3.8%a	0.0%a
Lifetime GAD	10.3%a	21.7%a	9.3%a	11.3%a	0.0%b
Current OCD	2.6%a	5.0%a	4.7%a	3.8%a	0.0%a
Lifetime OCD	2.6%a	6.7%a	9.3%a	7.5%a	0.0%a
Taking Psychotropic Meds	7.7%a	10.0%a	11.6%a	9.4%a	0.0%a
Flanker Task Variables
Accuracy (% Correct)	88.0 (12.4)a	90.5 (3.9)a	90.1 (4.9)a	89.5 (4.4)a	89.9 (5.6)a
Correct Trial Reaction Time	403.3 (97.9)a	425.3 (63.8)a,b	433.5 (41.1)b	437.9 (48.8)b	406.8 (57.8)a
Incorrect Trial Reaction Time	330.8 (106.9)a	351.3 (49.3)a	356.5 (39.4)a	363.7 (47.7)a	334.6 (76.2)a
ERNresid	0.4 (1.2)a	<0.0 (1.1)b	-0.2 (0.8)b	<0.0 (0.8)b	-0.1 (0.8)b
CRNresid	-0.2 (1.2)a	0.1 (1.1)a	0.2 (0.8)a	<0.0 (1.0)a	0.1 (0.9)a
ΔERN (ERN – CRN)	-5.5 (7.1)a	-8.2 (6.6)b	-9.5 (4.9)b	-8.1 (5.2)b	-8.8 (4.8)b

Note. Means or percentages with different subscripts across rows were significantly different in pairwise comparisons (p < .05, chi-square test for categorical variables and Tukey’s honestly significant difference test for continuous variables). CRN_resid was scored by saving the variance leftover in a regression where the CRN was entered predicting the ERN. AUD = alcohol use disorder; SUD = substance use disorder; MDD = major depressive disorder; SAD = social anxiety disorder; PTSD = post-traumatic stress disorder; GAD = generalized anxiety disorder; ERN = error-related negativity; CRN = correct-response negativity.

On average, participants made 31.4±20.6 errors during the task, resulting in 89.6%±6.4 task accuracy. Reaction times were faster for errors than for correct responses, as expected (t[243]=23.8, p<.01; Table 1). Controlling for current SUDs, smoking status, and anxiety disorders, results indicated the groups differed on reaction times for correct trials (F[4, 235]=3.15, p< .05) such that individuals in the current AUD and healthy control groups responded quicker than individuals in the other three groups. There were no group differences in task accuracy or reaction times for error trials. Task accuracy and reaction times did not correlate with the ERN_resid or ΔERN (ps >.19).

3.2. Group Differences in ERN

For ERN_resid, individuals with current anxiety disorders displayed larger ERNs compared with individuals without anxiety disorders (F[1, 235]=5.51, p =.02, np² =.02). There was trend-level main effects suggesting that smokers (F[1, 235] =3.67, p =.06, np² =.01) and individuals with current SUD diagnoses (F[1, 235] =3.48, p =.06, np² =.01) had a more blunted ERN relative to non-smokers and individuals without SUDs. As hypothesized, there was a main effect of Group (F[4, 235] =2.98, p =.02, np² =.05). Post-hoc paired comparisons revealed that individuals with current AUD displayed a more blunted ERN_resid compared with individuals with AUD in remission (Cohen’s d=0.37), at-risk for AUD (Cohen’s d=0.65), psychiatric controls (Cohen’s d=0.39), and healthy controls (Cohen’s d=0.52). There were no significant differences amongst any of the other four comparison groups (see Figure 1).

Fig 1.

On the left, topographic map of neural activity (error minus correct). On the right, response-locked ERP waveform for correct and error trials, as well as the difference waves (error-related negativity; ΔERN) for each of the five groups: current AUD (A), remitted AUD (B), at-risk for AUD (C), psychiatric controls (D), and healthy controls (E). AUD = alcohol use disorder.

For ΔERN, individuals with current anxiety disorders displayed larger ERNs compared with individuals without anxiety disorders (F[1, 235] =5.80, p =.02, np² =.02). There was a trend-level effect of current SUD (F[1, 235] =3.90, p =.05, np² =.02) such that individuals with current SUD displayed a decreased ERN relative to individuals without current SUD. There was no main effect of smoking status (F[1, 235] =2.60, p =.11, np² =.01). There was a significant effect of Group (F[4, 235] =3.22, p =.01, np² =.05). The pattern of results was identical to the ERN_resid results such that individuals with current AUD displayed a more blunted ΔERN compared with individuals with AUD in remission (Cohen’s d=0.41), at-risk for AUD (Cohen’s d=0.66), psychiatric controls (Cohen’s d=0.42), and healthy controls (Cohen’s d=0.54). There were no significant differences amongst any of the other four comparison groups.

4. Discussion

No prior study to our knowledge has examined whether individuals at various stages of AUD psychopathology evidence differences in ERN amplitude. The current study was designed to address these gaps and clarify the potential role of the ERN in AUD. Results revealed that individuals with current AUD displayed smaller ERNs relative to individuals with remitted AUD, individuals at-risk for AUD (by virtue of their family history), psychiatric controls, and healthy controls. There were no differences in ERN amplitude across the four, non-current AUD groups, and the results were identical when examining differences in the ERN_resid and ΔERN, and when controlling for current illicit substance abuse and anxiety disorders. This suggests that a blunted ERN is a marker of AUD illness.

The present study is the first to demonstrate that individuals with current AUD display diminished neural error-reactivity, similar to individuals with other SUDs (e.g., Forman et al. 2004; Franken et al. 2007) and externalizing traits (Hall et al. 2007). As some have previously speculated, it is possible that all (or several) forms of externalizing or impulse control problems are characterized by reduced error-processing (Hajcak & Olvet, 2008). However, a handful of past studies have suggested that AUD may not fit the broader externalizing literature such that individuals with AUD evidence enhanced ERNs (i.e., Gorka et al. 2016; Padilla et al. 2011; Schellekens et al. 2010). Our findings are therefore inconsistent with these other AUD studies; although there are several important points to consider. First, Padilla et al. 2011 did not account for comorbid anxiety disorders which may have contributed to an enhanced ERN. Second, both Schellekens et al. 2010 and Padilla et al. 2011 enrolled men with AUD who were recently abstinent (approximately 1-2 months), and Schellekens et al. 2010, in particular, was comprised of men admitted into an alcohol detoxification program. Thus, some AUD subjects may have been experiencing symptoms of sustained alcohol withdrawal, which potentiates anxiety and perhaps in-turn, increases ERN amplitude (Heilig et al. 2010). Our current AUD subjects were neither treatment-seeking and likely not in a state of sustained withdrawal as they were not instructed to alter their drinking behaviors other than providing a negative alcohol breath screen the day of the assessment; although notably, withdrawal symptoms were not measured directly. The same is true for our remitted AUD group as participants were in remission for an average of 2.8 years. Lastly, our prior study found that individuals with PTSD+AUD displayed greater ERNs than individuals with PTSD-AUD; however, there were no differences between participants in the PTSD+AUD group and a non-PTSD control group (Gorka et al. 2016). Therefore, an enhanced ERN related to AUD may have been a finding relatively specific within PTSD/anxiety populations.

Our results indicate that individuals with current AUD display diminished neural error-reactivity and it is important to consider the potential implications of this individual difference factor. A small, or blunted, ERN implies dampened activation of the brain’s conflict monitoring system, a neural circuit responsible for recognizing that a mistake has been made and adjusting behavior to avoid subsequent mistakes (Botvinick et al. 2001). AUD is a disorder characterized by habitual alcohol use even in face of drastic negative consequences (APA, 2015), which may reflect a failure to self-monitor and/or respond to negative consequences of one’s behavior. In other words, reduced neural error-reactivity may reflect deficient self-monitoring, which could facilitate ongoing alcohol use despite adverse interpersonal, occupational, financial and/or legal problems. Reduced neural error-monitoring may also more broadly support a tendency to engage in risky behaviors and the abuse of other drugs in addition to alcohol. Indeed, many have suggested that the broad externalizing dimension of psychopathology involves a deficit in neural error-reactivity and the on-line monitoring of behavior.

The current study also extends this literature by demonstrating that blunted ERNs may be a state marker for current AUD. We found no differences between individuals at-risk for AUD, individuals with remitted AUD, and psychiatric and healthy control groups. Therefore, a blunted ERN was only observed during active AUD illness and may be concomitant with the disorder. Related, it is possible that active alcohol abuse blunts the ERN via acute toxicity; especially given that alcohol intoxication has been shown to transiently decrease the ERN in healthy individuals (Barthalow et al. 2012). Another possibility is that a typical ERN is protective of long-term AUD such that within AUD populations, individuals with a blunted ERN are more likely to persist in the disorder whereas individuals with a typical ERN may more easily achieve remission. Nevertheless, our data indicates that the ERN functions differently with regard to anxiety disorders and AUD and consequently, may have different clinical uses. With respect to anxiety, studies show that the ERN is a risk-factor (Weinberg et al. 2015) and may be useful for the early identification of at-risk youth for prevention approaches. With respect to alcohol, the ERN may be better suited as a clinical target for AUD interventions and an objective, psychophysiological marker of disease status. Related, impaired neural error-reactivity may contribute to the core features of problematic alcohol use and thus, directly targeting ACC function via behavioral and/or pharmacological approaches may enhance the effectiveness of AUD treatments. It is also possible that the ERN could be used as a diagnostic marker and assessed both before and during AUD treatment to track clinical progress and achievement of AUD remission. More research is needed to explore the feasibility and utility of these clinical strategies but given that the ERN is a reliable measure that is easy-to-record, its suitability for the anxiety and AUD clinics should continue to be investigated.

Two prior studies have suggested that a blunted ERN is a risk marker for AUD/SUD (Fein & Chang, 2008; Euser et al., 2013); though the current findings are inconsistent with this hypothesis. The study conducted by Fein and Chang (2008) used a modified version of the Balloon Analog Risk Task (BART; Lejuez et al. 2002) to measure neural reactivity to the explosion of simulated balloons – indication that for a given trial, no money was won. The BART therefore involves explicit feedback that a mistake was made (resulting in lost potential for money) whereas the flanker task does not provide trail-by-trial feedback and each mistake does not directly correspond to a monetary outcome. Thus, it is possible that the ERN during the BART captures additional individual difference factors beyond basic error-monitoring that are related to AUD risk and interestingly, a prior study by our lab demonstrated that behavioral performance during the BART was also related to family history density of AUD (Gorka et al., 2015). The study performed by Euser and colleagues (2013) did, however, use a similar flanker task; though the at-risk adolescents were classified by positive parental history of any SUD (not just AUD) and only a subset of at-risk individuals displayed a blunted ERN, particularly the adolescents who were cannabis-use naïve. It is therefore possible that a blunted ERN is a risk marker for some SUDs, in certain subgroups of individuals, though additional research is needed directly comparing ERN amplitude in illicit SUD versus AUD risk populations.

4.1. Limitations

The current study had numerous strengths including the large sample size and the comparison of multiple AUD and control groups. There are also several limitations. First, the current and remitted AUD groups had greater rates of SUDs than the other three groups, which is to-be-expected given the highly comorbid nature of substance abuse issues (Burns & Teesson, 2002). Although SUDs were included as covariates in the analyses, and SUD does not entirely account for the pattern of results as there were differences in the ERN between the current and remitted AUD groups, it is unclear to what extent prior and ongoing illicit substance use contributed to the pattern of results. Second, the current AUD group was comprised of individuals who primarily had mild to moderate AUD and thus the impact of severe AUD on the ERN is still unclear. Third, the analyses were limited to group comparisons and due to the combining of two separate datasets, we did not have overlapping dimensional assessments of AUD psychopathology across our subjects. It may therefore be useful for additional research to compare patterns of results across both AUD categories and dimensions in the same sample to elucidate to what extent the ERN tracks AUD symptoms.

4.2. Conclusions

In sum, the present findings indicate that a blunted ERN may be a state marker of AUD psychopathology. The findings also converge with a broader SUD literature suggesting that multiple forms of externalizing psychopathology may be characterized by impaired error-monitoring. Given that the ERN is a reliable, easy-to-record, objective psychophysiological measure, its utility in both the anxiety and AUD clinic should continue to be explored.