The present results indicate that humans who report having used MDMA on at least 25 previous separate occasions (but no MDMA within the previous 2 weeks) have global reductions in binding of [11C]McN5652 and DASB to the brain SERT. Results obtained with the two selective SERT radiotracers were highly consistent, although, overall, [11C]DASB appears better suited for detecting statistically significant MDMA-associated reductions in brain SERT density in humans. When considered with the large body of preclinical literature demonstrating the toxic potential of MDMA toward brain serotonin neurons (see Introduction), the most parsimonious explanation for SERT reductions in MDMA users is that it is related to MDMA-induced toxicity to serotonin axons and axon terminals. However, it should be noted that subjects who participated in this study had used MDMA on at least 25 separate occasions and, on average, had used it on 97 separate occasions. Therefore, conclusions regarding MDMA-induced SERT reductions in individuals with lower levels of MDMA exposure than those included in this study cannot be reached.
The present results are largely in agreement with the recent voxel-based analysis of MDMA users by Buchert et al 2004
, who found significant reductions of [11
C]McN5652 binding in a variety of brain regions in MDMA users (who were abstinent at the time of scanning), as well as a correlation between [11
C]McN5652 binding and duration of abstinence (in current MDMA users as well as those with protracted abstinence). Notably, their study failed to find group differences in SERT binding between controls and MDMA users who, on average, had abstained from use of MDMA for 18 months. The reasons for the absence of detectable differences between long-term abstinent MDMA users and controls in their study (as compared to data from the present study) could be related to several factors. These include the longer average duration of abstinence in their subjects (18 vs
4 months, on average) and possible regeneration of serotonergic axons over time, differences in analytic methods (quantitative vs
nonquantitative), and differences in spatial resolution (SPM requires normalization of brain volumes and, therefore, has reduced spatial resolution).
Of note, PET measures of the SERT, whether obtained using [11
C]DASB or [11
C]McN5652, are known to underestimate the extent of MDMA-induced reductions in the SERT, and percent reductions of BP potential do not correspond 1 : 1 with reductions of the SERT measured in vitro
(Scheffel et al, 1998
; Szabo et al, 2002
). Measurements of the SERT in areas of the brain with relatively low SERT density, such as the neocortex, are particularly vulnerable to ‘noise’ introduced by more richly innervated subcortical regions, decreasing the ability to discern differences between subject groups. In addition to the problem of region-to-region ‘spillover,’ several additional sources of variability decrease the power to detect significant differences between groups. These include signal-to-noise variability of the radioligand, intersubject variability (ie there is a range of ‘normal’ SERT binding in control subjects, as well as a range of presumed MDMA-induced changes in SERT binding), interscan variability, and variability associated with each of the factors employed in the model used to estimate SERT binding, to name a few. That having been said, the finding that two distinct radioligands employed within the same subject yield highly consistent global and regional PET data speaks strongly to the validity of the results herein reported.
The biologic significance of the observed SERT reductions is uncertain. However, as noted above, percent reductions in PET-binding parameters () tend to underestimate the true extent of SERT loss, as evidenced by the fact that in baboon they are approximately 50% of those measured in vitro
in the same animals (Scheffel et al, 1998
; Szabo et al, 2002
). Recognizing that percent reductions in PET-binding parameters may not precisely gauge the extent of SERT loss, the issue of biological significance, to a large degree, remains to be determined. From a functional standpoint, the absence of overt clinical consequences in our subjects could be taken to suggest that MDMA-induced SERT loss does not lead to gross abnormalities in behavior or physiology. However, upon thorough testing, a number of psychometric studies have revealed subtle cognitive/memory deficits in abstinent MDMA users (Verbaten, 2003
). Although it remains to be established that these memory abnormalities, if due specifically to MDMA, are directly related to serotonergic deficits, controlled studies in which indices of serotonin neuronal integrity are assessed in conjunction with detailed cognitive assessments should help to address this question.
The reasons for the failure to detect significant differences between SERT-binding parameters in a number of subcortical brain regions high in SERT density is, at first, puzzling. Indeed, we (McCann et al, 1998
) and other researchers (Buchert et al, 2003
) have previously noted a reduction in SERT-binding sites in some of these regions. Evaluation of the data in subcortical regions (ie putamen, midbrain) in the present study indicates a high level of variability relative to other brain regions where significant differences were identified. The reasons for the high variability in subcortical regions are likely to be multiple (see above). Also, in MDMA subjects, possible regeneration of serotonin axons and axon terminals from raphe nerve cell bodies, which are known to be unaffected by MDMA (Green et al, 2003
; Hatzidimitriou et al, 1999
) may contribute to variability in subcortical regions. Thus, the potential for recovery over time, coupled with the fact that recovery is not uniform among various brain regions, may be a source of variability in abstinent MDMA users who have different durations of abstinence. When considered with the fact that, unlike experimental animals, most MDMA users do not have a single exposure regimen of MDMA but, instead, tend to use MDMA over months to years, interpretation of neuroimaging studies in MDMA users poses challenges.
Exploratory analyses suggested a positive relationship between SERT binding and duration of abstinence, and a negative relationship between SERT-binding parameters and MDMA use intensity (as measured by self-reported typical monthly MDMA consumption). These data are consistent with preclinical data (Hatzidimitriou et al, 1999
), which demonstrate recovery of serotonin axonal terminals in some (but not all) brain regions over time, and a dose-related MDMA-induced serotonin neurotoxic response (Green et al, 2003
). However, they must be interpreted with caution and considered preliminary because, as mentioned above, MDMA users in this study had multiple exposures to MDMA, and our study was not powered to demonstrate significant correlations between SERT-binding parameters and use parameters.
Limitations of this study should be recognized. First, as noted previously (and in ), MDMA users tended to use more ‘other’ recreational drugs (licit and illicit) than controls, despite considerable efforts to match the two groups on this factor. The difficulty in matching subjects on this feature stems from the fact that MDMA users who met criteria for study inclusion (ie use on at least 25 separate occasions), as a rule, used a number of other recreational drugs as well. Therefore, the possibility remains open that ‘other drug’ use, alone or in interaction with MDMA, may play a role in SERT reductions noted in MDMA users. Second, even if MDMA subjects’ retrospective accounts of their drug use histories are fully accurate, it is impossible to verify whether drugs believed to be MDMA were pure, or whether they were adulterated with other drugs that may have influenced SERT binding. Finally, it is possible that reductions in SERT binding in MDMA users pre-existed MDMA use. Although subjects with neuropsychiatric diagnoses and conditions in which serotonin has been implicated were excluded from participation, it is possible that an unidentified condition that predisposes individuals to use recreational drugs played a role in SERT reductions in MDMA users.
In conclusion, quantitative PET results from the current study extend previous observations obtained with [11C]McN5652 to [11C]DASB. Together, studies with first-and second-generation SERT radiotracers in the same individual demonstrate that abstinent MDMA users have reductions in regional brain SERT. Notably, results obtained using two SERT radiotracers in the same subjects were highly consistent, using a variety of SERT-imaging parameters. Although both tracers were found to be capable of detecting MDMA-related reductions in regional brain SERT binding, results obtained using [11C]DASB were more robust. Future studies are needed to determine the lower limits at which PET is capable of detecting loss of the SERT, and whether individuals with lower exposure to MDMA also demonstrate losses of the brain SERT. In addition, studies aimed at determining the potential relationship between brain measures of the SERT and functional consequences of MDMA may be useful in elucidating the role of the serotonin system in normal brain function and in various neuropsychiatric disease states.