Twenty-nine Caucasian DAD men (25–65 years old) were recruited on admission for withdrawal or day-hospital units in the addiction departments of Paul Brousse and Emile Roux Hospital (Assistance Publique, Hopitaux de Paris). Thirty-one patients had previously participated in a study correlating cerebral volumetry and neuropsychological data (26). A senior psychiatrist interviewed patients, performed a clinical examination and reviewed patient’s medical records and biological data. IQ was not evaluated. Sociodemographic and clinical data are presented in Tables 1 and 2 (drinking history variables).

All patients met DSM IV diagnostic criteria for alcohol dependence (32). All patients had a history of maximum two prior hospitalizations for withdrawal to minimize potential cognitive deficits induced by supplementary withdrawals (33). All patients had been abstinent for at least 3 weeks as shown by normalized gamma-glutamyl-transferase (GGT), mean corpuscular volume (MCV), and carbohydrate-deficient transferrin (CDT) levels (see Table 2). During withdrawal, a standard treatment protocol (vitamin B1 and B6 and digressive diazepam) was used. At the time of testing, all sedation had ceased for at least 7 days. Nine patients identified at least one first degree relative with a drinking problem. Even though four patients reported lifetime maternal drinking, none reported any maternal alcohol use during pregnancy.

Control subjects who drank less than two standard units of alcohol per week (20 g) during the previous year and had a score of ≤5 on the Alcohol Use Disorders Identification Test (AUDIT) (34) were recruited from the neighboring community (n = 29). Controls were compensated (100 Euros) by the study’s sponsor (INSERM).

Patients and controls were excluded, after a clinical interview with a senior psychiatrist, according to DSM-IV criteria, if they had a prior history of substance dependence or abuse (except tobacco and caffeine), axis I disorder (particularly mood and/or anxiety disorders, psychosis), high scores on Hamilton anxiety and Hamilton depression scales (>5), any current or past disease which could impair cognitive functions: hepatic (cirrhosis, hepatitis), neurological (seizures, encephalopathy, Wernicke–Korsakoff syndrome), cardiovascular, renal, or head trauma (loss of consciousness for more than 30 min), stroke, signs or symptoms of malnutrition or major brain abnormalities on MRI scan. Axis II was not considered as an exclusion criterion. Two participants were excluded: one because of a microinfarctus in the perithalamic region on MRI scan; and another because he chose not to complete the neuropsychological testing. None refused to participate. Our final analysis included 29 alcohol-dependent subjects and 29 male controls.

All participants received verbal and written protocol information and signed a consent form prior to inclusion. This protocol was approved by the Bicêtre Hospital ethics committee.

On the day of testing, fasting blood samples were taken to investigate the somatic complications of chronic alcoholism. The panel of tests included liver function tests: albumin, alkaline phosphatase, bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), GGT, and CDT. In addition, blood chemistry (complete blood count, MCV, mean corpuscular hemoglobin concentration, hemoglobin, hematocrit, serum protein, and serum sodium levels) was assessed (see Table 2).

After eating, subjects were evaluated using the AUDIT to assess the degree of alcohol dependence. Social functioning was evaluated using the Self-reported social adjustment scale (SAS-SR) (35), a self-administered questionnaire including questions on social and leisure activities, relationships with the extended family, marital partner, offspring, and relationships within the family circle. It also provides information concerning socio-economic status. Each item is scored on a 5-point scale with higher scores indicating poorer functioning. The total SAS-SR score is calculated by averaging all items.

Nicotine dependence was evaluated using the Fagerström Test of Nicotine Dependance (FTND) (36) measuring cigarettes smoked per day, pack-years, and smoking duration. Pack-years were calculated as [(number of cigarette per day/20) × (duration of smoking in years)].

Body mass index (BMI) was calculated by dividing weight (kg) by height squared (m²) providing a good instrument to control health risk.

All participants underwent the Mini-Mental State Examination (MMSE) and a battery of neuropsychological tests conducted by a senior neuropsychologist, according to standardized procedures. Subjects were tested individually in a single session with a standardized test order.

The battery of neuropsychological tests assessed three cognitive domains and measured memory (M), WM, and executive functions (EF). The tests are briefly described below and detailed in the Supplementary Material.

Episodic memory was assessed by the Rey–Osterrieth Complex Figure (ROCF), Free and Cued Selective Reminding Test. The information subtest of the Wechsler Adult Intelligence Third Revision (WAIS III) was used to assess semantic memory.

Working memory was assessed by Digit Span subtest, Spatial Span subtest, and Letter-number sequencing subtest (WAIS III).

We used the Trail Making Test part A and B (TMT A/B), Wisconsin Card Sorting Test (WCST), Stroop test, and Letter Fluency test to evaluate executive functions.

All tests were scored according to standard published procedures (see Supplementary Material).

Statistical analyses were conducted using R software (http://www.R-project.org/). The normality of variable distribution was assessed by the Shapiro–Wilk test. Two-sided comparisons between groups were made for age, education, BMI, SAS-R, and tobacco consumption variables using either Student’s t-test for normally distributed values or Mann–Whitney U test for non-normally distributed values, with a level of 0.05 or less. Then, Z-scores for the psychological tests were calculated to compare alcoholic patients with controls, adjusted by years of education. The mean and SD of each test was calculated for the control group, donated m and s, and the z-score for an alcoholic subject was defined as z = (score − m)/s.

Our final analysis included 29 DAD men and 29 male controls. The DAD and control groups were comparable in terms of age, years of education, BMI, SAS-SR, and MMSE (Table 1). The patients had greater tobacco consumption than controls as shown by FTND. The DAD group showed a significantly greater dependence score versus the control group (respectively, 3.85 ± 3.34 versus 0.89 ± 2.02; p < 0.001). DAD tobacco consumption was significantly greater than in the control group as shown by years of tobacco use (p < 0.0001) and the number of cigarettes per day (p < 0.0001).

All subjects were employed full-time and were in a stable relationship. Patients were either married (n = 20) or cohabiting (n = 9). All had a fixed address and lived in a house or flat. Our patients belonged to various socio-professional groups: artisans, small business owners (n = 7), teachers and scientific professions (n = 9), administrative and trained personnel (n = 7), and blue collar workers (n = 6).

All patients had undergone a maximum of two prior hospitalizations for withdrawal. Among our final patient sample, 7 had undergone two prior alcohol detoxifications, while the other 22 had only one prior detoxification. All patients had been abstinent for at least 3 weeks as shown by normalized GGT, MCV, and CDT levels (see Table 2).

Detoxified alcohol-dependent participants have less efficient executive functioning. Compared to controls, DAD patients had significant impairments on set-shifting and reactive mental flexibility (TMT B, p = 0.0041), and a significant difference was found between the two groups on criteria generation (WCST, p = 0.035). We also observed significant impairments in planning ability and organization as assessed by the figure copy (ROCF, p < 0.00001). No significant differences were found between the two groups in performance on inhibitory control (Stroop), spontaneous mental flexibility (Letter fluency), and perseverative errors (WCST) (see Table 3).

No differences were found for patients compared to controls for immediate cued recall, total delayed recall and recognition in the verbal episodic memory test. However, DAD subjects had lower scores than controls in free recall (p < 0.00001), total recall (p = 0.0002), and free delayed recall (p < 0.0001).

Visual episodic memory capacities were impaired in DAD patients group versus controls group (ROCF recall copy, p < 0.0001). Semantic memory score was also lower in DAD subjects compared to controls (WAIS III, information subtest, p = 0.0016) (see Table 3).

Detoxified alcohol-dependent subjects presented alterations in most WM tasks. Forward digit span was not significantly altered (p = 0.0803), whereas backward digit span was significantly diminished (p = 0.0019) (WAIS III). In addition, DAD alcoholic subjects performed significantly worse than controls on letter-number sequencing (p = 0.006). Spatial span was also diminished (Wechsler Memory Scale Third Edition, p < 0.001) (see Table 3).

Many studies have described a dysexecutive syndrome in alcoholic patients (20, 23, 25, 38–40). Our results showed reactive cognitive flexibility impairments and visual and verbal memory retrieval deficits. Similar to Sullivan et al. (11), we observed impairment in planning ability and organization as assessed by the ROCF copy. DAD patients used a copying strategy with a “bits and pieces” approach rather than a global structured approach. These results are consistent with a previous study suggesting that alcoholic patients use less efficient strategies to perform visuoperceptual learning tasks (41). Those impairments persist even after 6 months abstinence, as described by Munro et al. (10).

Despite these mild cognitive impairments, other cognitive functions would seem to be preserved in our DAD participants. Indeed, inhibitory control can allow them to resist acting impulsively, and more specifically, control and curb alcohol cravings. Moreover, self-control is having the discipline to stay focused and to complete a task, for example, at work.

By having preserved spontaneous flexibility abilities, DAD participants seem to be able to adapt their relationship, adjust their behavior in interaction and respond to unanticipated situations despite reactive cognitive flexibility impairment.

In addition, being able to maintain rules without any perseverative tendency, as controls, DAD patients are enable to plan for a solution by selecting efficient strategies, to sustain goals or instructions, and to dynamically adjust their response to a feedback from the environment.

Thus, our results show that alcohol-dependent patients who have maintained a professional integration show partially preserved executive functions that enable them to stay effective in their work and adapted in social interactions.

Our results show episodic memory deficits for visual material and spatial processing among DAD subjects, compared to controls. More precisely, our results suggest that consolidation or retention are preserved but retrieval is impaired (total recall and free delayed recall deficits associated with normal cued delayed recall). Other authors have also shown that alcohol-dependent subjects presented lower performance than controls in verbal episodic memory (19, 42, 43).

These memory deficits may be partially explained by executive dysfunction. Many authors have shown that executive functions are involved in encoding and retrieval (25, 44). Other authors have hypothesized that this may be due to inefficient retrieval strategies in alcohol-dependent subjects and not to storage capacities (45, 46).

Semantic memory and verbal knowledge tested by the information subtest of the WAIS III is significantly altered in DAD patients. Pitel et al. (44) showed that alcohol-dependent subjects presented impaired novel semantic learning (shallower encoding) and that alcohol-dependent patients used inefficient cognitive strategies to compensate for cognitive deficits.

Detoxified alcohol-dependent subjects presented significant impairments in most WM tasks. The deficits in forward digit span or spatial span, and in more complex information processing such as backward restitution and letter-number sequencing, suggest that alcohol-dependent patients have a limited ability to manipulate information due to impaired WM. Noel et al. (23) observed a similar result on the alpha-span task.

Working memory involves systems implicated in short-term maintenance and manipulation of information necessary to accomplish complex tasks. Baddeley’s model included two slave systems to ensure temporary information maintenance, the phonological loop and the visuo-spatial sketchpad, and a central executive system considered to be similar to the executive functions (47).

Thus, our subjects seem to present deficits in key systems at all levels of Baddeley’s model even though they maintain preserved psychosocial functioning. These findings suggest that alcohol-dependent subjects developed compensatory mechanisms, so they can preserve socio-professional adaptation.

All subjects recruited in this study are men. Thus, it is impossible to generalize our results to all alcoholic subjects. This study evaluated specific controls with particularly low alcohol consumption levels. Thus, the results from our controls may not reflect results in the general population. Among DAD patients, we can notice a wide variability length of abstinence (from 3 weeks to 3.5 years) that could influence the results.

This could also be explained by the relatively few detoxifications in this sample of patients. It has been shown that the number of reported alcohol withdrawals during the prior year was a significant predictor of poorer cognitive performance following detoxification (48, 49).

Another important question is whether the improvement of certain memory and executive cognitive functions would predict greater motivation to change drinking?

To answer these questions, it would be interesting to prospectively investigate this population following long-term detoxification with an evaluation of drinking behavior change, relapse, employment, and the social adjustment outcomes.

There was also a significant difference in tobacco consumption between the DAD population and the control group in this study. The literature often shows a strong correlation between the consumption of alcohol and tobacco in alcohol-dependent subjects (50). However, we did control for this variable in our analyses, and thus our results on neuropsychological variables were independent of this difference. Other studies have shown that continued tobacco use may have a significant impact on cognitive function as well as future dementia (51, 52). Further studies should examine an eventual synergistic role between continued alcohol and tobacco use in cognitive decline in DAD patients, as well as the potential impact withdrawal.

We have interpreted the cognitive functioning alteration as resulting from alcohol abuse. An alternative explanation is that neuropsychological deficits may precede alcohol abuse and represent vulnerability or predisposition to alcohol dependence. The question of whether cognitive deficits in alcoholic patients are a consequence or a predisposition to the disease is still a matter of controversy and this could be the object of other prospective studies.