Results

The three groups were similar in terms of sex distribution, age, and educational level (table 1). Although the level of education was slightly lower in the ecstasy group, the differences were not statistically significant (χ²=6.44, df=8, p=0.59, educational levels 1–3 and 4–5 were pooled together for testing). The data for ecstasy and cannabis use are given in table 2. The two user groups were similar for the extent of previous use of cannabis. Ecstasy users had used ecstasy over 27 months on average. The average estimated total dose was 93.4 ecstasy tablets. Hence the average extent of use was about 3.5 ecstasy tablets a month, which is a typical recreational and not a very heavy use. Most ecstasy users had taken the drug for the last time 2 to 8 weeks before the study. In four cases this time period was shorter (7–10 days) whereas in two cases it was considerably longer (6–12 months).

In the reaction time (RT) tasks there were only very few errors and anticipatory responses in all three groups and there were no statistical differences among the groups. Therefore, these data are not presented. The overall results (RT and performance data) showed mostly good cognitive levels in all three groups. However, the ecstasy users tended to perform poorer in most tests and there were some statistically significant differences between the groups (ANOVA, table3): in the test of selective visual attention (go/nogo) ecstasy users had longer RTs than both control groups and a larger variance of RTs than the non-user group. In the tests of divided attention and intermodal integration ecstasy users had significantly longer RTs than the cannabis users. In the verbal working memory task (digit span backward) ecstasy users performed poorer than non-users. In the verbal learning and memory test (VLMT) ecstasy users recalled less words after the first presentation of the learning list than non-users (immediate recall). They required more repetitions to learn the items than both control groups and they forgot more words after the presentation of the interference list than the non-users. In the visuospatial memory task (VIG) ecstasy users recalled less figures after the first presentation than both control groups (immediate recall). Finally, ecstasy users showed a poorer performance than both control groups in all three general intelligence tests. Scores on the questionnaire for the assessment of concentration and memory problems did not differ significantly between the three groups, although the two user groups tended to score higher than the non-user group. By means of the canonical discriminant analysis 90.36% of the subjects could be assigned correctly to one of the three groups according to their performance data (92.9% of the ecstasy users, 85.7% of the cannabis users, and 92.6% of the non-users).

In the group of ecstasy users long RTs in the divided attention task were associated with a long period of regular ecstasy use (r=0.411, p<0.05). A poor digit span performance (working memory) was associated with a higher estimated cumulative ecstasy dose (r=−0.382, p<0.05) and a young age of onset of cannabis use (r=0.518, p=0.01). Poor performance in the VLMT (verbal memory and learning) was associated with heavier ecstasy use (immediate recall/estimated cumulative dose:r=−0.387, p<0.05; interference effect by second word list/frequency of use: r=0.425, p<0.05; number of repetitions in the learning phase/average one night dose: r=0.489, p<0.01) and with heavier cannabis use (number of repetitions/frequency of use:r=0.577, p<0.01). No other correlation of any test score with any aspect of the pattern of ecstasy or cannabis use reached statistical significance.

In the analysis of covariance (ANCOVA) using the general knowledge score as the covariate group differences were still significant for all previously significant performance scores except for the interference effect by the second word list in the verbal memory task (VLMT). In this particular case the group effect only approached significance (p=0.092) in the covariance analysis. In addition, there was a significant, although not very high correlation between the three intelligence test scores (general knowledge/mosaic test:r=0.386, p=0.043; mosaic test/LPS-4:r=0.390, p=0.040), but no correlation between any of the intelligence tests and the other performance scores in the ecstasy user group (n=28).

Discussion

A comprehensive cognitive test battery was administered to 28 recreational ecstasy users with concomitant use of cannabis only and to two equally sized matched groups of cannabis users and non-users. The test battery included tests of attention, memory and learning, frontal lobe function, and general intelligence. Although all three groups performed well within the normal range ecstasy users performed worse than one or both control groups in the more complex tests of attention (go/nogo task of selective attention, divided attention, and intermodal integration), in memory and learning tasks, and in the tasks reflecting aspects of general intelligence. By contrast, performance in simple reaction time tasks of attention (tonic and phasic alertness) was unaffected. Poorer performance scores or longer reaction times in the working memory, verbal memory, and divided attention tasks were associated with heavier ecstasy and heavier cannabis use. These results raise the concern that ecstasy use even in typical moderate recreational doses and possibly in conjunction with cannabis use may lead to a subclinical cognitive decline in otherwise healthy young people. Although we cannot rule out the possibility that concomitant regular use of other legal and illegal drugs and/or heavier ecstasy use was concealed by subjects, we do not think that this was the case because they were not aware of our inclusion and exclusion criteria when first interviewed and there was no motivation for them to falsify their data once they agreed to participate in the study. In addition, our clinical impression of the ecstasy users did not fit the typical characteristics of heavy polydrug users.

The ecstasy users reported an average time period of more than 3 weeks from the last intake of ecstasy until the study day. Therefore, purely pharmacological effects of ecstasy or the commonly reported “after effects” in the days after ecstasy ingestion21-23 are unlikely to have determined the poorer performance of ecstasy users compared with the control groups. By contrast, concomitant cannabis use may well have influenced the cognitive performance of ecstasy users. Noteworthy, the two user groups were similar for the extent of previous cannabis use, the only exception being the duration of regular use, which was longer in the group of ecstasy users. However, there was no association between performance data and duration of regular cannabis use, although there were associations between poorer performance and other aspects of the pattern of cannabis use (younger onset age and higher frequency of use). Finally, the time elapsed since the last use of cannabis tended to be somewhat shorter in the cannabis user group. In conclusion, concomitant cannabis use is unlikely to fully account for the poorer performance of ecstasy users compared with the cannabis user group, although cannabis use is likely to have affected cognition and to have contributed to some extent to the poorer performance of ecstasy users compared with the non-user group. However, the data of the cannabis user group demonstrate that cannabis use alone was not sufficient to impair the performance of subjects to a significant degree.

Finally, it might be hypothesised that the poorer performance of ecstasy users was due to accidental pre-existing differences in general cognitive capacity or intelligence. Ecstasy users performed worse not only in the two tests of fluid intelligence (mosaic test, LPS-4: logical thinking, and problem solving) which are likely to be affected by subtle or the start of more serious cognitive deterioration, but also in the general knowledge test, a classic test of crystallised intelligence, which is thought to be insensitive to early cognitive decline.35 Although there was no statistically significant difference in the level of education between the three groups, cannabis users and non-users tended to have slightly higher levels of education on average (highest school leaving examination qualifying for admission to university or university degree, see table 1). However, we do not think that possible pre-existing differences in intelligence and education are sufficient to explain the poorer performance of the ecstasy user group in tests of attention and memory. Firstly, there was a significant correlation between memory and attention scores and the extent of ecstasy use indicating an association of poorer performance with heavier use. Secondly, group effects remained significant when we reran the ANOVA using the score of the general knowledge test as a covariate. Thirdly, there was no correlation between the three intelligence test scores and the performance or reaction time data of the attention and memory tasks. The poorer general knowledge in the ecstasy user group may well be the consequence of the early socialisation within the dance and drug scene, which is likely to have somewhat narrowed the fields of interests and to have interfered with the acquisition of general knowledge. By contrast, the poorer performance of ecstasy users in the two tests of fluid intelligence may reflect either a start of, but still subclinical cognitive decline, or the slightly inferior level of education, or a combination of both factors.

In summary, our data suggest that ecstasy use over a period of months or a few years may cause long term impairment of cognitive performance even when ecstasy is taken in typical recreational and not necessarily very high doses. Concomitant cannabis use may contribute to the impairment. However, the nature of the emerging cognitive disturbance is not yet clear. Besides the possibility of multiple distinct impairments of attention, memory processes, and other cognitive skills it might be hypothesised that one common disturbance underlies or contributes to impairments in various tasks. Increased levels of cognitive and behavioural impulsivity have been associated with conditions with reduced serotonergic function17 37-40and have been shown in some19 41 but not in all studies42 with ecstasy users. An impulsive cognitive style with poor planning may well impair the performance in complex tasks of attention, as well as in learning tasks and tasks of fluid intelligence. However, our overall data are not supportive of an increased cognitive impulsivity in our sample, because ecstasy users did not demonstrate increased rates of errors and premature reactions (anticipation: RT<100 ms) and had longer rather than shorter reaction times compared with the control groups. The only hints towards this direction were the results in the visual scanning task. Ecstasy users exhibited a shorter processing time than the control groups in the non-critical trials of this task, whereas in the critical trials their processing time correlated poorly with the target position. However, these group differences did not reach significance. Alternatively, the underlying common cognitive disturbance might be a problem of working memory, which refers to the ability of holding information “on line” for short periods of time and manipulating it in the service of guiding behaviour.43 Working memory is a complex skill which cannot be entirely differentiated from processes of attention, memory span, and even general intelligence.35 In our test battery working memory is best reflected by the digit span performance (backward), but, in addition, it is involved in the memory and learning tasks, in both tests of fluid intelligence and in the task of selective attention with response selection (go nogo). Therefore, working memory may be a good candidate for the substrate of cognitive impairment in ecstasy users.

The cognitive disturbance in our sample of ecstasy users is likely to be related to the well recognised neurotoxic potential of ecstasy, which is restricted to the serotonergic system. Although no clear picture has emerged so far about the role of serotonin in cognition, it is involved in various cognitive tasks involving memory and speed of information processing.16 17 44 Moreover, cognitive deterioration in users may be related to secondary regulatory mechanisms involving other neurotransmitter systems that are not directly affected by the neurotoxic potential of the drug. Subclinical cognitive decline may not be noticed by the subjects themselves over a long period of time. Therefore, subjects are likely to continue using ecstasy and put themselves at substantial risk for further deterioration over the years. Theoretically, it is possible that the cognitive impairment becomes apparent only after many years when the effects of normal aging add to the possible neurotoxic damage. A most important question refers to the reversibility or permanence of the adverse cognitive effects after longer periods of abstinence. A recent primate study showing abnormal cerebral 5-HT innervation patterns 7 years after MDMA exposure11 raises concern about the possible irreversibility of functional consequences of serotonergic damage in humans. This question cannot be answered in cross sectional studies and must be considered in future longitudinal investigations. Given the popularity of ecstasy among young people the present data are clearly alarming and underline the need for further research in this field.