Brain serotonin synthesis in MDMA (ecstasy) polydrug users: an alpha-[11C]methyl-l-tryptophan study


  • Linda Booij,

    1. Department of Psychology, Queen's University, Kingston, Ontario, Canada
    2. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    3. Sainte-Justine Hospital Research Center, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Jean-Paul Soucy,

    1. McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Simon N. Young,

    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Martine Regoli,

    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Paul Gravel,

    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Mirko Diksic,

    1. McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada
    2. Department of Neurology and Neurosurgery, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Marco Leyton,

    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    2. McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada
    3. Department of Neurology and Neurosurgery, McGill University, Montreal, Quebec, Canada
    4. Department of Psychology, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Robert O. Pihl,

    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    2. Department of Psychology, McGill University, Montreal, Quebec, Canada
    Search for more papers by this author
  • Chawki Benkelfat

    Corresponding author
    1. Department of Psychiatry, McGill University, Montreal, Quebec, Canada
    2. McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada
    3. Department of Neurology and Neurosurgery, McGill University, Montreal, Quebec, Canada
    • Address correspondence and reprint requests to Chawki Benkelfat, Department of Psychiatry, McGill University, 1033 Pine Avenue, Montreal, QC, Canada H3A1A1. E-mail:

    Search for more papers by this author


3,4-Methylenedioxymethamphetamine (MDMA, ecstasy) use may have long-term neurotoxic effects. In this study, positron emission tomography with the tracer alpha-[11C]methyl-l-tryptophan (11C-AMT) was used to compare human brain serotonin (5-HT) synthesis capacity in 17 currently drug-free MDMA polydrug users with that in 18 healthy matched controls. Gender differences and associations between regional 11C-AMT trapping and characteristics of MDMA use were also examined. MDMA polydrug users exhibited lower normalized 11C-AMT trapping in pre-frontal, orbitofrontal, and parietal regions, relative to controls. These differences were more widespread in males than in females. Increased normalized 11C-AMT trapping in MDMA users was also observed, mainly in the brainstem and in frontal and temporal areas. Normalized 11C-AMT trapping in the brainstem and pre-frontal regions correlated positively and negatively, respectively, with greater lifetime accumulated MDMA use, longer durations of MDMA use, and shorter time elapsed since the last MDMA use. Although the possibility of pre-existing 5-HT alterations pre-disposing people to use MDMA cannot be ruled out, regionally decreased 5-HT synthesis capacity in the forebrain could be interpreted as neurotoxicity of MDMA on distal (frontal) brain regions. On the other hand, increased 5-HT synthesis capacity in the raphe and adjacent areas could be due to compensatory mechanisms.


Animal studies showed that MDMA (3,4-methylenedioxymethamphetamine, ecstasy) exposure alters brain serotonin neurotransmission. Whether these effects are permanent is unknown. The present human study found that, compared to controls, young adult MDMA users had lower serotonin synthesis in parts of the frontal cortex, and greater serotonin synthesis in other regions, including the brainstem. The strength of these effects correlated with severity of use. The findings may indicate that MDMA can be neurotoxic but the brain may also have the capacity to regenerate, depending on the specific brain region. Longitudinal studies are needed to test the clinical relevance of these findings.

Abbreviations used



5-hydroxyindoleacetic acid


5-hydroxytryptamine or serotonin


brodmann area


beck depression inventory


cerebrospinal fluid


full width at half maximum


blood-to-brain clearance/trapping


lysergic acid diethylamide






Montreal Neurological Institute


magnetic resonance


positron emission tomography


structured clinical interview for DSM-IV: patient edition


serotonin transporter


statistical parametric mapping


tryptophan hydroxylase


volume of interest

Use of the drug 3,4-methylenedioxymethamphetamine (MDMA, ecstasy) declined in the last decade in the United States (US), but consumption rates recently started rising again, possibly reflecting a reduced perception of risk (Johnston et al. 2012). The increase in prevalence may not be limited to the US, as there was also an increase in serious MDMA-related incidents in the Netherlands in 2011 (Krul et al. 2012). In 2010, the lifetime prevalence for MDMA use in the US was 13.5% for individuals between ages 19–30 (Johnston et al. 2011).

Although the incidence of death as a direct consequence of MDMA use is relatively low (Rogers et al. 2009), concerns remain about the drug's potential long-term harmful effects. Extensive animal studies demonstrate neurotoxicity following MDMA use, particularly in the 5-hydroxytryptamine (5-HT, serotonin) system. This includes distal axotomy in 5-HT neurons, reductions in 5-HT neuronal signaling, reduced regional density of the serotonin transporter (SERT), and lower activity of tryptophan hydroxylase (TPH) (Capela et al. 2009), occasionally occurring after a single dose (Mueller et al. 2013).

Several studies have associated MDMA use with altered 5-HT function in humans. Positron emission tomography (PET) studies of MDMA users found evidence of altered in vivo binding in cortical and/or subcortical areas, relative to controls, for tracers labeling brain SERT sites (see Capela et al. 2009 for review). A recent PET study also reported that the recreational use of MDMA was associated with long-lasting increases in 5-HT2A receptor density (Di Iorio et al. 2012).

Whether 5-HT alterations in MDMA users are pre-existing, because of MDMA or the other drugs used by MDMA users, or a combination of these, remain unknown. Nonetheless, given the important role of 5-HT in brain development, mood, behavior, and cognitive function (Parrott 2013), these alterations could have significant consequences for human functioning. Using acute tryptophan depletion (ATD), we recently reported that women who are polydrug MDMA users may be more susceptible than men to the effects of lowering 5-HT (Young et al. 2014). This study focused on exploring further the likelihood of putative differences in 5-HT neurotransmission between MDMA polydrug users and controls, using an in vivo measure of regional 5-HT synthesis capacity with PET, in real-time.

In humans, brain regional 5-HT synthesis can be estimated in vivo with PET imaging of α-[11C]methyl-l-tryptophan (11C-AMT) as a tracer, using the blood-to-brain clearance/trapping (K*) of the tracer (Diksic and Young 2001). The specificity and validity of the method has been demonstrated in both healthy and patient populations (Leyton et al. 2001, 2005; Rosa-Neto et al. 2005, 2007; Lundquist et al. 2007; Berney et al. 2008, 2011; Booij et al. 2010). In this study, we used the 11C-AMT method to compare brain 5-HT synthesis capacity in regular MDMA polydrug users who were free of drugs at the time of scan, with non-MDMA users. Our primary hypothesis was that polydrug users of MDMA would have ‘altered’ regional 11C-AMT trapping in cortical and/or subcortical areas, compared to age- and gender-matched individuals who had never used MDMA. On the basis of previous reports emphasizing differences in brain 5-HT synthesis between males and females (Nishizawa et al. 1997; Sakai et al. 2006), we also examined the regional K* data in MDMA polydrug users for a putative gender effect. Finally, we examined whether 11C-AMT trapping would be modulated by specific characteristics of MDMA use, such as an estimate of the cumulative lifetime MDMA dose, the duration of use and the time elapsed since the last MDMA dose.



Participants using MDMA, as well as MDMA-naïve controls, were recruited through newspaper advertisements, posters distributed at local universities and flyers distributed at local clubs and raves. Following a telephone interview to assess initial eligibility, participants underwent a full screening procedure in person involving the following: (i) A psychiatric assessment, including a semi-structured psychiatric interview (structured clinical interview for DSM-IV: patient edition, SCID-NP) (First et al. 2002), the beck depression inventory (Beck et al. 1961), and a determination of the presence or absence of psychiatric illness in the family using the family history method (Andreasen et al. 1977). (ii) A complete physical examination, including laboratory testing (glucose, electrolytes, alanine transaminase, aspartate aminotransferase, creatinine, thyroxine, thyroid- stimulating hormone, and complete blood count) and an electrocardiogram. (iii) Estimations of past and current drug use, using a timeline follow-back procedure as in our previous PET studies on drug use (e.g. Boileau et al. 2006).

The main inclusion criterion for MDMA polydrug users was the use of MDMA on at least 25 occasions, a level of use which has been previously associated with evidence of brain alterations in the brain serotonergic system such as altered CSF 5-hydroxyindoleacetic acid (5-HIAA) levels, altered brain SERT binding and altered mood and neuroendocrine responses to the 5-HT agonist probe meta-chlorophenylpiperazine (mCPP) (McCann et al. 1999). These criteria were similar to those used in a previous study on MDMA polydrug users conducted in our laboratory, using ATD (Young et al. 2014). Exclusion criteria included the following: (i) a current or past major medical illness (determined by physical examination and laboratory tests); (ii) evidence for current axis I DSM-IV disorders; (iii) current use of any prescription psychotropic drug; (iv) a beck depression inventory score of 12 or above; (v) reports suggesting evidence of a positive family history in first-degree relatives for major depressive disorder; and (vi) being a sexually active woman and not using a reliable form of contraception.

Assessment of the controls was similar to that of the MDMA polydrug users. Potential control participants reporting some illegal drug use other than MDMA were not in principle excluded from the study; selecting controls reporting no drug use could have resulted in a control cohort that would be too different from that of our MDMA users. Participants with less than five total uses, and no use in the previous year, of illicit drugs other than cannabis, were deemed eligible. For cannabis, we only included controls, if their use in the previous year averaged less than once a month.

Participants were asked to refrain from drug use for any drug, 3 weeks prior to the study sessions. They were required to test negative on two consecutive urine drug tests for the detection of illicit drugs, one at screening and one on the morning of the PET session (Triage, San Diego, CA, USA). All participants who were suitable and willing to give informed consent were scheduled. Once a full description of the study had been completed, written informed consent from all participants was obtained. The study was in conformance with the code of ethics of the World Medical Association. The Research Ethics Boards of the McGill University Health Centre and the Montreal Neurological Institute (MNI) approved the study.

Positron emission tomography and magnetic resonance (MR) imaging

All participants underwent a 60-min dynamic PET scan, conducted with an ECAT HR+ (CTI Molecular Imaging, Inc/Siemens, Knoxville, TN, USA) in the late morning or early afternoon. The participants received a low-protein diet the previous day and fasted overnight (limited water intake was allowed). Women underwent the PET scan in the follicular phase of their menstrual cycle. Two intravenous lines were secured, one in each arm, with one used for tracer injection and one for blood withdrawal. Before beginning the dynamic PET scan, transmission scans were performed using a 68Ga source for attenuation correction. Participants received approximately 10 mCi of 11C-AMT over 2 min, at the beginning of the dynamic PET scan protocol. Twenty-six frames were acquired using the following protocol: six frames of 30 s, seven frames of 60 s, five frames of 120 s, and eight frames of 300 s. All scanning was acquired with a Neuro-Shield® (Montreal, QC, Canada) placed around the participant's neck to reduce scattering. Blood samples were drawn throughout the PET scan from the ante-cubital vein at progressively increasing time intervals to obtain the 11C-AMT plasma time activity curves. The input function was estimated using a validated non-invasive procedure combining venous sinus activity (first 20 min) and venous plasma time-activity data (Nishizawa et al. 1997, 1998). Specifically, radioactivity-time courses from the venous samples were drawn, and corrected by the venous sinus from dynamic PET images, as reported in Nishizawa et al. (1998). The slopes K* were then calculated using normalized exposure time points (Nishizawa et al. 1998).

PET scans included 63 image slices at an intrinsic resolution of 5.0 × 5.0 × 5.0 mm full width at half maximum (FWHM). Images were reconstructed using a Hanning filter of 8.1 mm FWHM in the trans-axial direction.

All participants also underwent high-resolution MR imaging using a 1.5-T system (Philips Gyro-scan; Philips Medical Systems, Eindhoven, the Netherlands) for PET/MRI co-registration. MR imaging data were stored as 256 × 256 × 160 mm matrices with 1-mm3 isotropic voxels.

Tryptophan measurement

Three 2-mL venous blood samples were drawn from each individual during the PET scan. The samples were centrifuged, and the plasma was stored at −80°C for measurement of plasma-free tryptophan concentration using high-performance liquid chromatography (Nishizawa et al. 1997).

Statistical analyses

Differences in demographics, drug use, plasma tryptophan values, and behavior between the MDMA polydrug users and controls were tested using one-way anova or chi-square statistics in the case of nominal variables.

Both the co-registration and normalization were conducted according to standardized procedures used at the MNI (e.g. Leyton et al. 2001; Rosa-Neto et al. 2004; Booij et al. 2010, 2012; Berney et al. 2011). Briefly, coregistration of the individual PET and MRI images was performed using an automatic procedure (Woods et al. 1993), which uses averaged tissue activity images obtained during the time period of 5–60 min of dynamic PET data acquisition (Okazawa and Diksic 1998). Parametric K* images of 11C-AMT trapping were generated (Okazawa and Diksic 1998) and re-sampled into MNI305 2-mm isotropic stereotaxic space using a standard automatic algorithm (Collins et al. 1994). The images were subsequently smoothed to a 14-mm resolution FWHM, using an isotropic Gaussian filter. To cancel out the effects of individual global effects on regional 11C-AMT trapping (K*) values, statistical analyses with proportional scaling was used and regional K* values were normalized by the mean global K* of the gray matter to 100, as in our previous studies (Leyton et al. 2001; Rosa-Neto et al. 2004; Booij et al. 2010, 2012; Berney et al. 2011). The rationale for normalizing to the mean global K* of the gray matter is to remove changes because of brain size, higher blood flow, etc. within and between groups. The number 100 was chosen to express normalized K* in percentages.

A Student's t-test was applied voxel-by-voxel, to determine regional differences in the K* distribution pattern between MDMA polydrug users and controls. Statistical significance was defined by two criteria. First, the height threshold used to interpret the t-test in terms of probability level was set at p = 0.005, thereby, accounting for multiple testing, while keeping the chance of a type II error under control. Second, the extent threshold was set at 40 voxels. Further analyses examined regional K* for gender differences by conducting separate analyses for men and women and tested associations with specific characteristics of MDMA polydrug use.

All of these analyses were carried out in statistical parametric mapping (SPM) version 8 (SPM8; Wellcome Functional Imaging Laboratory, London, UK). To confirm and quantify the results obtained from whole brain SPM analyses, Volumes of interest (VOI) analyses were performed in those specific brain regions that showed significant group differences in the whole brain SPM analyses and were large enough to be reliably identified on each participant's MRI using an automatic segmentation method (Collins et al. 1999). Regional K* was expressed both as normalized and non-normalized values; the need to normalize the regional K* data in this study was supported by the observed trend in gender differences in the control group in plasma free tryptophan values (see below). Group differences in normalized K* extracted from the VOI analyses were analyzed using general linear models (GLM) for repeated measures, with hemisphere as a within subject factor, and group (MDMA polydrug user vs. control) as between subjects factor. Significant group by hemisphere interactions were further investigated using one-way GLM models, using the left and right hemisphere for the specific VOI investigated, as a dependent variable, and group, as a between subject factor. One-way GLM models, without a hemisphere term, were run for the analyses of the raphe and brainstem given that these are very small regions. As was the case for whole brain SPM analyses, all GLM analyses were rerun for males and females separately. The magnitude of group differences in normalized regional K* identified by VOI analyses was further quantified by calculating Cohen's d effect sizes using the G-power program (Faul et al. 2007).


Sample characteristics are shown in Table 1. Self-reported MDMA consumption varied significantly from 28 to 1015 tablets over the subject's lifetime (mean ± SD: 236 ± 282; median: 100), over periods of time ranging from 1 to 9 years (4 ± 2.3 years; median: 3.5). Compared to other samples reported in the literature, this corresponds to a ‘moderate’ to a ‘significant’ pattern of use. The controls all reported never having used MDMA. Consistent with what is normally observed in the general population, all MDMA users reported using other drugs as well (Table 2). There were no gender differences in reported lifetime MDMA intake, duration of MDMA use or time elapsed since the last MDMA dose, nor in the lifetime consumption of alcohol or of any other drugs. All participants were free of drugs when participating in the study, as verified by the urine drug tests obtained during screening and on the morning soon prior to the PET measurement.

Table 1. Characteristics of the sample
CharacteristicControlsMDMA polydrug users
  N N
Gender9 males, 9 females9 males, 8 females
  1. a

    One value (female control) was missing and thus not included in the analyses.

  2. BDI, beck depression inventory; n/a = not applicable. Difference in MDMA use cannot be tested statistically because both mean and SD in control group is 0.

Age (years)–29
BDI score2.32.60–93.93.20–10
Plasma-free tryptophan (nmol/L)a9.52.85.3–17.810.32.36.5–15
  7.1–10.5 0.49.3–10.5
Activity injected (mCI)9.31.1n/a9.828228–1015
  n/a 2.815–24
 n/an/a 10.51–33
Number of MDMA tablets, lifetime0n/an/a2362.31–9
Estimated age of first MDMA usen/an/a 19.2  
Time since last MDMA dose in weeksn/a  11.6  
Duration of use in yearsn/a  4.0  
Table 2. Self-reported lifetime drug/alcohol consumption other than MDMA, expressed as number of estimated uses
CharacteristicControlsMDMA polydrug users
  1. LSD, lysergic acid diethylamide; n/a, not applicable.

Alcohol intoxication1794590–20002692195–728
Cannabis, lifetime13180–6582794510–3500
Tobacco2126030–2300703095790–30 000

Plasma tryptophan

The mean plasma-free tryptophan concentrations ± SD were 9.5 ± 2.3 nmol/mL for controls and 10.3 ± 2.3 nmol/mL for MDMA polydrug users (Table 1). This group difference was not significant (t(32) = 0.91, p = 0.37). There was no gender difference (mean ± SD: 9.4 ± 2.5 nmol/mL for males and 10.4 ± 2.6 nmol/mL for females, t(32) = −1.2, p = 0.24), although in controls, there was a trend toward higher plasma-free tryptophan levels in females than in males (control males: 8.3 ± 2.0 nmol/mL; control females: 10.7 ± 3.2 nmol/mL) (t(16) = −1.92, p = 0.073).

11C-AMT trapping in MDMA polydrug users versus controls

Whole brain SPM analyses

MDMA polydrug users demonstrated focal areas of both increased and decreased regional normalized K* values, relative to controls (p < 0.005, clusters ≥ 40) (Fig. 1). MDMA polydrug users had decreased 11C-AMT trapping extending from the pre-frontal orbital–frontal regions all the way to posterior parietal regions. Notably, the main focus of increased uptake in MDMA polydrug users was in the brainstem, in the region of the periaqueductal gray matter, as well as in parts of the left lateral pre-frontal cortex and temporal lobe (Table 3).

Figure 1.

Normalized alpha-[11C]methyl-l-tryptophan (11C-AMT) trapping in 3,4-methylenedioxymethamphetamine (MDMA) polydrug users versus controls. The upper three color-coded images indicate brain regions in which there is lower 11C-AMT trapping in MDMA polydrug users than in controls (MDMA users < controls). The lower three color-coded images indicate brain regions in which there is higher 11C-AMT trapping in MDMA polydrug users than in controls (MDMA polydrug users > controls).

Table 3. The regions in which MDMA polydrug users and controls differ in normalized 11C-AMT trapping (p < 0.005; cluster threshold: ≥ 40)
MNI coordinatesBrain region/BANo. of voxelsPeak T
X (mm)Y (mm)Z (mm)
  1. L, left; R, right; BA, Brodmann area; MNI, Montreal Neurological Institute.

Normalized 11C-AMT trapping in MDMA polydrug users < controls
0−2060L. medial frontal gyrus; BA 64324.09
0−5656L. pre-cuneus; BA 73403.62
48−434R. pre-central Gyrus; BA6443.38
−34−3654L. post-central Gyrus, Parietal Lobe763.31
38−6044Parietal lobe, R. inferior parietal lobule; BA 401263.29
22−6444R. pre-cuneus; BA 71263.15
4428−6R. inferior frontal gyrus; R. inferior orbitofrontal gyrus403.26
−3834−14L. middle frontal gyrus; R. inferior orbitofrontal gyrus; BA11653.10
−36−650L. middle frontal gyrus; BA 6423.09
Normalized 11C-AMT trapping in MDMA polydrug users > controls
−2−26−20L. midbrain (brainstem)623.63
−204446L. superior frontal gyrus3433.56
−263650L. middle frontal gyrus3432.95
−324832L. superior frontal gyrus; BA 93432.63
−368−28L. temporal lobe; L. superior temporal gyrus553.35

VOI analyses

>MDMA polydrug users had, relative to controls, higher regional normalized K* in the superior temporal gyrus [F(1,33) = 9.54, p = 0.004, d = 1.06] and middle frontal gyrus [F(1,33) = 5.08, p = 0.03, d = 0.77] in both hemispheres. Trends of increased normalized K* in MDMA polydrug users relative to controls, were also observed across the brainstem [F(1,33) = 3.58, p = 0.07, d = 0.65] and for the raphe specifically [F(1,33) = 2.94, p = 0.096, d = 0.71]. There were significant group by hemisphere interactions for the superior frontal gyrus [F(1,33) = 4.53, p = 0.04]. Compared to controls, MDMA polydrug users had higher normalized K* in the left superior frontal gyrus (relative to the right superior frontal gyus) and lower normalized K* in the right pre-cuneus (relative to the left pre-cuneus), although not to a statically significant degree when analyzing hemispheres separately (d = 0.60 and d = 0.61, respectively).

Gender effects of MDMA polydrug use on 11C-AMT trapping

Whole brain SPM analyses

A separate whole-brain analyses for males and females indicated that lower normalized 11C-AMT in the frontal regions was especially noticeable in males (Table 4). Increases in normalized 11C-AMT trapping in the brainstem and parts of the frontal and temporal lobe as observed in the entire group of MDMA polydrug users were not observed when analyzing males and females separately. Sample sizes were insufficient to reliably examine correlations between specific characteristics of MDMA use and 11C-AMT trapping for males and females separately.

Table 4. The regions in which male and female MDMA polydrug users have lower normalized 11C-AMT trapping than controls (p < 0.005; cluster threshold: ≥ 40)
MNI coordinatesBrain regionNo. of voxelsPeak T
X (mm)Y (mm)Z (mm)
  1. BA, brodmann area; L, left; MNI, Montreal Neurological Institute; R, right.

Females: MDMA users < controls
2−6050R. pre-cuneus, BA7443.41
−3436−18L. inferior frontal gyrus; L. inferior orbitofrontal cortex603.30
6−2460R. medial frontal gyrus, BA 6613.29
Males: MDMA users < controls
−24−8−38L. limbic lobe, BA 283843.66
−38−10−48L. inferior temporal gyrus3844.05
−6−5258L. pre-cuneus823.99
50−232R. pre-central gyrus; BA6573.97
4624−6R. inferior frontal gyrus; Inferior orbitofrontal cortex653.63
−28−3456L. post-central gyrus; BA 31553.65
−32−1054L. pre-central gyrus; BA61553.25
−32−1852L. pre-central gyrus1553.32

VOI analyses

Overall, VOI analyses were largely consistent with results drawn for whole brain analyses. Male MDMA polydrug users had lower normalized K* in the pre-central gyrus, relative to control males [F(1,16) = 5.73, p = 0.029, d = 1.14]. Group by hemisphere interactions were observed in the pre-central gyrus [F(1,16) = 5.79, p = 0.029] and in the pre-cuneus [F(1,16) = 9.26, p = 0.008]. Univariate one-way GLM analyses for the left and right hemispheres for each of these regions indicated that male MDMA polydrug users had lower normalized K* for both the pre-central gyrus and pre-cuneus in the right hemisphere (d = 1.47 and d = 1.10, respectively), relative to male controls. Female MDMA polydrug users had lower normalized K* in the lateral orbitofrontal gyrus [F(1,15) = 5.45, p = 0.03, d = 1.14], relative to female controls. In addition, increased normalized K* in the raphe observed in the MDMA polydrug users, was mainly because of an increased normalized K* in this region in female MDMA polydrug users [F(1,15) = 8.55, p = 0.01, d = 1.43; Table 5].

Table 5. Regional normalized K* (mean ± SD) in selected VOIs
Brain regionMDMA polydrug usersControls
  1. Lat., lateral; Med., medial; VOIs, volume of interests. For the brainstem, the difference between the MDMA polydrug user and control group was a trend [F(1,16) = 3.27, p = 0.089, d = 0.85].

Raphe84.0 (24.4)94.7 (14.1)81.6 (23.3)73.1 (16)
Brainstem92 (5.1)91.9 (3.6)86.5 (7.5)84.5 (17.8)
Pre-central gyrus105.9 (4.8)110.4 (5.6)112.8 (7.1)110 (15.3)
Post-central gyrus106.3 (5.6)109.9 (3.4)111 (7.9)103.5 (11)
Pre-cuneus119.1 (8.2)118.1 (11.7)120.8 (10.8)118.7 (5.9)
Med. frontal gyrus117 (4.3)119.5 (4.7)115.3 (7.9)122 (6.6)
Middle frontal gyrus115.5 (4.8)116.6 (6.0)112.1 (5.7)107.7 (13.4)
Superior frontal gyrus103.6 (10)105.9 (6.7)99.7 (16.3)104.0 (4.8)
Superior temporal gyrus124.8 (4.7)124.5 (5.6)121.1 (3.6)118 (5.7)
Superior parietal lobule113.1 (7)110.9 (6.1)113.4 (10.5)114.8 (15.4)
Med. Orbitofrontal gyrus129.4 (5.6)120.9 (10.4)117.7 (18.4)122.7 (27.3)
Lat. Orbitofrontal gyrus104.6 (12.4)89.3 (8.8)98.4 (11.9)100.8 (11.2)

Correlations between 11C-AMT trapping and characteristics of MDMA use

A greater lifetime MDMA intake was associated with decreased 11C AMT trapping in a very small frontal cortical area in the right hemisphere, although only at a marginally significant level (Peak T = 3.38, kmax = 41). A greater lifetime MDMA intake was also associated with higher normalized 11C-AMT trapping in clusters of voxels projecting in the raphe area or in close approximation to it (anterior and superior right cerebellar hemisphere adjacent to the brainstem) (Peak T = 3.48, kmax = 28). There was no significant correlation between the total dose of any of the other drugs of abuse and normalized 11C AMT trapping (Fig. 2).

Figure 2.

Correlation between alpha-[11C]methyl-l-tryptophan (11C-AMT) trapping and indices of drug use in 3,4-methylenedioxymethamphetamine (MDMA) polydrug users. The color-coded images represent the strength of correlations between 11C-AMT trapping and lifetime MDMA use (upper row), duration of MDMA use (middle row) and time elapsed since last MDMA use (bottom row). The first three images in each row (left side) depict positive associations with 11C-AMT trapping, the last three images in each row (right side) depict negative associations.

A longer duration of MDMA use was associated with lower normalized 11C-AMT trapping in anterior regions of the brain, particularly in the mid-portion of the cingulate gyrus (Peak T = 5.02, kmax = 933). Duration of MDMA use correlated positively with 11C-AMT trapping in cortical regions covering mostly the post-central regions of the brain (Peak T = 5.29, kmax = 201). The latter correlation was confirmed by VOI analyses for the left post-central gyrus (r = 0.695, p = 0.002).

A shorter MDMA-free time interval, defined as the time elapsed from the last MDMA dose and the date of the PET scan, was associated with lower 11C-AMT trapping in a few regions in the right hemisphere neo cortex (Peak T = 4.07, kmax = 93) as well as with higher 11C-AMT trapping in the right cerebellar hemisphere (Peak T = 5.73, kmax = 270) and the brainstem (including the raphe region). The latter association was confirmed by VOI analyses (r = −0.51, p = 0.05).


This study points to regional differences between MDMA polydrug users and matched controls for 5-HT synthesis capacity, with both increases and decreases in 11C-AMT trapping in MDMA polydrug users relative to controls. While decreases were primarily observed in pre-frontal–orbital and parietal regions, with somewhat more widespread decreases in males relative to females, increases in 11C-AMT trapping was also noticeable, in particular in the brainstem.

Our results are consistent with studies in rodents and non-human primates indicating that repeated administration of MDMA often results in long-term region-specific reduction in numerous 5-HT markers, including brain tissue concentrations of 5-HT, 5-HIAA levels, TPH enzyme activity and SERT density (Biezonski and Meyer 2011; Urban et al. 2012). These reductions in 5-HT markers have previously been interpreted as a result of distal axotomy of brain 5-HT neurons (Molliver et al. 1990). However, it is noteworthy that in our study, the spatial extent of focal declines in cingulate cortex and mesencephalon is less than might have been expected on the basis of SERT imaging studies in similar populations or in pre-clinical studies where a loss of small serotonin fibers in the forebrain (see also below) was exhibited as a result of MDMA. In addition to decreases in 11C-AMT trapping in the frontal regions, increased normalized 11C-AMT trapping was also observed, mainly in the brainstem as well as in the superior medial and temporal gyri.

Using the same tracer and radio-autography 14 days after rats were administered a total of eight doses of MDMA, there were significant decreases in 5-HT in a variety of brain areas including parts of the cortex, striatum and hippocampus, and a significant increase only in the median raphe nucleus (Molliver et al. 1990). This suggests that, in addition to MDMA-induced neurotoxicity in humans, compensatory feedback mechanisms in the cell bodies of the 5-HT neurons might also occur. This observation was consistent with findings from a longitudinal study in MDMA polydrug users with repeated measurements of brain SERT binding, which reported increases in SERT binding after cessation of MDMA use (Buchert et al. 2006). Indeed, a large animal literature reports on the extent to which the effects of MDMA may be enduring/persistent, as opposed to reversible. For instance, although studies in rats have shown that there is almost complete recovery of 5-HT markers 1 year after drug exposure, as a result of sprouting of axons (Molliver et al. 1990), studies in monkeys reported that the recovery is incomplete, and innervation patterns remain altered (Scheffel et al. 1998; Hatzidimitriou et al. 1999). In squirrel monkeys that had been treated with MDMA and monitored for up to 7 years thereafter, some of the 5-HT ‘deficits’ ‘recorded’ at the end of the study were actually less severe than those observed at 18 months following exposure. Factors influencing recovery included the distance from the raphe nuclei to the purported affected site, the degree of initial axonal injury, and possibly the proximity of myelinated fibers (Hatzidimitriou et al. 1999). PET studies of SERT distribution in baboons also suggested that there is some recovery in regions closer to the raphe nuclei. Specifically, at 13 months after MDMA exposure, SERT levels had actually risen above control levels in the hypothalamus, but were still down respectively by 62%, 78% and 73% in the frontal, parietal and occipital cortex (Scheffel et al. 1998). A more recent autoradiography study in rats investigating the impact of MDMA on 5-HT biosynthesis found that 14 days after chronic MDMA administration, TPH2 enzyme immuno-reactivity was decreased, while TPH2 mRNA was increased in the cell bodies of 5-HT neurons in the Dorsal Raphe Nucleus, suggesting some compensatory mechanisms (Bonkale and Austin 2008).

The impact of polydrug use on brain 5-HT synthesis appeared to be more widespread in males than in females. Interestingly, there were no differences in reported lifetime MDMA intake, duration of MDMA use or time elapsed since the last MDMA dose, nor in the lifetime estimate of alcohol or of other drugs examined consumed. At first glance, these results seemed to be at odds with findings of previous studies, showing that women appear to be more susceptible to the neurochemical effects of MDMA on CSF 5-HIAA (McCann et al. 1994) and SERT binding (Reneman et al. 2001), than men. They are also at first glance at odds with our recent ATD study, indicating a greater mood-lowering response to ATD in female MDMA polydrug users than in males (Young et al. 2014). Yet, previous studies in animals did not find compelling evidence of gender effects of MDMA on the 5-HT system (5-HIAA or 5-HT concentrations) (Walker et al. 2007). Methodological differences across studies, genetic differences, polydrug use, hormonal factors, and MDMA dose might account for some of those different observations. On the other hand, in spite of the observation that the effects of MDMA polydrug use in males may be more widespread than in females, interestingly, female MDMA polydrug users in the present study had, relative to female controls, a relatively specific lowering in regional K* in the orbitofrontal regions, a brain region highly involved in emotion regulation. Alterations in regional K* in the orbitofrontal regions were absent in male polydrug users. Previous PET studies assessing cerebral metabolism following ATD have shown that the mood-lowering response to ATD in depression vulnerable populations correlates with alterations in metabolism in the orbitofrontal regions after ATD (Bremner et al. 1997; Smith et al. 1999; Neumeister et al. 2004). Taking the findings of our ATD study and from the current PET study together, it might be that the mood-lowering effect of ATD in female polydrug users as observed in (Young et al. 2014) are due to sex-specific, drug induced, 5-HT effects in the orbitofrontal regions. Only four female MDMA polydrug users participated in both the ATD study and the current PET study and thus correlations between the mood response to ATD and 5-HT synthesis capacity could not be reliably tested; yet, it is tempting to speculate that sex-specific effects of drug use on 5-HT neurotransmission in the orbitofrontal regions might make female users more susceptible to aversive behavioral manifestations, including clinical depression. Longitudinal studies in (former) drug users having consumed MDMA in significant quantities over long periods of time are needed to test such hypothesis.

Strengths and limitations of this study and suggestions for future research

The validity and significance of the observations presented in this study rest upon the following considerations: (i) Despite the criticism that the brain regional uptake of 11C-AMT reflects blood–brain transport of tryptophan, rather than 5-HT synthesis (Shoaf et al. 1998), experimental evidence accumulated over more than 15 years by us and other research groups (Leyton et al. 2005; Lundquist et al. 2007), now provides a firm basis for the claim that brain regional 11C-AMT trapping represents an acceptable proxy for the measurement of regional brain 5-HT synthesis. (ii) As is the case for other MDMA PET studies, we cannot rule out the possibility that some of the alterations reported could be pre-existing and related to characteristics particular to individuals who use MDMA. (iii) In addition to MDMA, many other drugs have been shown to influence 5-HT neurons, and the combination of MDMA and such drugs that are often used together with MDMA, e.g. methamphetamine, could have a greater effect on brain monoamines systems than either compound alone (Yuki et al. 2013). Thus, the group differences in 5-HT synthesis capacity may be due to the use of other drugs, either alone or in combination with MDMA. However, the fact that our results are consistent with those reported in rats administered MDMA (Muck-Seler et al. 1998) and studied with a similar technique, supports the general view that the effects depicted in this study are primarily because of MDMA. This interpretation is further strengthened by the observation that the pattern of effects on 5-HT systems observed in this study, was different from what was observed after administration of alcohol, nicotine, or cannabis (Jang et al. 2002; Nishikawa et al. 2009; Shen et al. 2011). (iv) Though our sample size was comparable with other PET studies in MDMA polydrug users, the results were not significant when corrected for Family-Wise Error. However, most of the regions exceeded largely the a priori determined and commonly used threshold criteria (mostly K > 100 and p < 0.001) and the VOI analyses generally did not contradict the results obtained with SPM, thereby controlling for type I errors. (v) Alterations in the midbrain of brain 5-HT synthesis, could have occurred in many different ways, unrelated to MDMA use. (vi) In our initial study, healthy females had overall a 52% lower 5-HT synthesis rates than healthy men (Nishizawa et al. 1997). The observation of lower 5-HT synthesis capacity in healthy females relative to healthy males have been replicated in a later study, using normalized regional K* data (Sakai et al. 2006), although these effects were region and hemisphere specific (Sakai et al. 2006); an observation we also noticed in this study, albeit to a non-significant degree. Our control sample was twice as smaller than Sakai et al. (2006) suggesting that the present study was underpowered to study reliably gender differences in regional K*. However, the influence of other factors, recently shown to be associated with subtle variation in brain 5-HT synthesis in carefully screened healthy populations, cannot be ruled out, including (but not limited to) genetic variation (Booij et al. 2012). Both males and females showed increases in normalized K* in the brainstem. (vii) Male MDMA polydrug users tended to have increases in regional K* across the brainstem, relative to male controls, while female MDMA polydrug users showed significant alterations specifically in the raphe. Indeed, even in a small brain structure such as the raphe, the effects of MDMA appear to vary within specific subregions (Muck-Seler et al. 1998).

Taken together, the data reported both in this and one recently published manuscript by our group (Young et al. 2014), examining in MDMA polydrug users the functional behavioral and/or neurochemical consequences of MDMA use, allow for the following observations: (i) Previous studies in humans assessing aspects of 5-HT neurotransmission in vivo, including while using 5-HT-related radio-ligands, combined with PET, have generally reported decreased 5-HT function in MDMA polydrug users, relative to controls. Some evidence is provided here that this observation can be further extended to brain 5-HT synthesis. Together with findings obtained from animal studies (Muck-Seler et al. 1998), the results also provide some evidence that MDMA has a possible neurotoxic effect on 5-HT fibers in brain regions involved in emotion-regulation. (ii) The ‘neurotoxic’ effects of MDMA appear to be different for males and females. Although the effects appear to be regionally more widespread in males across the brain, deleterious effects on 5-HT synthesis may occur in specific orbitofrontal regions in females (iii) The ‘neurotoxic’ effects of MDMA, indicated by the observation of a decreased regional 5-HT synthesis capacity, may coexist with evidence of increased activity in other sites, suggesting some capacity for regeneration of 5-HT systems. Unfortunately, this comment is only speculative, as our study was cross-sectional and not longitudinal.


In summary, this study provides evidence that MDMA polydrug users have regionally specific increases and decreases in brain 5-HT synthesis relative to controls. MDMA-associated decreases in 5-HT synthesis in the frontal regions could be interpreted as MDMA-induced axotomy effects, while increases in the brainstem might be explained by compensatory changes in cell bodies.

Acknowledgments and conflict of interest disclosure

The authors thank the staff of the Positron Emission Tomography and Cyclotron-Radiochemistry Units from the Montreal Neurological Institute for their valuable assistance. The study was funded by an operating grant from the Canadian Institutes of Health Research (CIHR) awarded to Drs S.N. Young, M. Leyton, M. Diksic, C. Benkelfat, and R.O. Pihl (MOP: 42502). Dr L. Booij is supported by a CIHR New Investigator Award. The authors also wish to thank Dr R. Palmour for proofreading the manuscript. All authors report no conflicts of interest.