Limitations of the study
A kind of limitation of the present study arises from different patterns of drug abuse within the ecstasy user group itself. Ecstasy users also consume additional drugs, such as cocaine or cannabis, and ecstasy itself comprises a heterogeneous composition of psychoactive drugs. Besides methylenedioxymethamphetamine (MDMA) ecstasy contains other entactogenes such as methylenedioxy-N-ethylamphetamine (MDEA), methylenedioxyamphetamine (MDA) and methylbenzodioxolylebutamine (MBDB) in varying doses [39, 40]. These drugs differ in their specifity for the serotonin transporter (MDA<MDMA<MDEA<MBDB). In addition, ecstasy contains amphetamines in varying doses [41]. Finally, there are falsifications of ecstasy tablets containing caffeine or atropine, and even placebo. For these reasons the detected effects may not strictly be ascribed to ecstasy or MDMA abuse, although MDMA is, and was, the main psychoactive component of ecstasy [42, 43].
Data on cumulative ecstasy dose and time since last dose were obtained by self-assessment of the ecstasy users. Plausibility of these data was evaluated by testing of hair samples. Although there was an obvious discrepancy between self-assessment and hair analysis in only a few ecstasy users, these data are subject to some uncertainties. The shortness of most of the user's hairs restricted objective information about drug abuse by toxicological analysis of hair samples to at most the last few months. Both uncertainties due to self-assessment and uncertainties about the composition of ecstasy tablets rendered the detection of a correlation between cumulative dose and/or time since last dose ingestion and FDG uptake difficult.
The age-matched control group consisted of oncology patients from clinical routine. In general, these were not exposed to the same environmental factors as ecstasy users (‘raves’ etc). In addition, their psychical condition during PET scanning was rather different from the condition of ecstasy users, who were quite relaxed. This may introduce confounding effects [44]. However, it should be noted that the results of the analysis of the relationship of brain glucose metabolism and drug history in the group of ecstasy users are not affected by limitations of the control group.
The PET imaging protocol differed for ecstasy users and controls. In ecstasy users a 20 min emission scan was performed at precisely 45 min after injection of about 230 MBq FDG, while in controls an 8 min emission scan was acquired about 60 min after injection of about 370 MBq FDG. While the effect of scan duration and FDG dose on statistical image quality cancel to a large extent, there might be a systematic effect due to the different time intervals between injection and PET scanning. It is known that FDG uptake in the brain is still increasing at 45 min post-injection [45]. Therefore, the absolute FDG uptake, as measured by the standardized uptake value (SUV), for example, is expected to be higher in controls than in ecstasy users. However, assuming the same kinetic in all gray matter areas [46] this effect is cancelled by the image normalization applied, and, therefore, should not affect the results.
A further difference in the imaging protocol concerns the position of the individuals from the time of injection until the acquisition. While ecstasy users lay on a couch in the waiting room until they went to the scanner room, controls sat on a chair in the waiting room. However, due to radiation safety reasons controls were not allowed to walk around. Therefore, there is very little difference in movement activity during the uptake period, which is not expected to cause the observed difference in FDG uptake in the basal ganglia between ecstasy users and controls.
For the purpose of attenuation correction a post-injection transmission scan was performed with three rotating 68Ge rod sources in ecstasy users, but not in controls. Thus, attenuation correction for both ecstasy users and controls was performed by calculations based on a simplified model consisting of a uniform brain surrounded by uniform skull of predefined, constant thickness. These simplifications led to an increased variance of the FDG uptake within the group of ecstasy users and in the control group, particularly at brain levels at which the true skull thickness in a transversal slice deviates from the predefined value, e.g. limbic structures (cingulate, amygdala, hippocampus). In these areas calculated attenuation correction renders detection of differences with sufficient statistical significance more difficult.
The VOIs of the Karolinska Computerized Brain Atlas (CBA) were defined by manual segmentation of the atlas brain. Since segmentation was performed for both brain hemispheres independently, VOIs corresponding to the same structure in left and right hemisphere are not symmetric. In particular, they differ in size for some extent, e.g. putamen and caudate are slightly larger in the left hemisphere than in the right hemisphere, 8.60 vs 7.76, and 7.22 vs 6.68 ml, respectively. This may cause evaluation of the left structure to be less sensitive to remnant mismatch between the transformed individual brain and the atlas brain. This in turn may reduce the variance of ‘left’ FDG uptake, and thus, explain that statistical significance of our findings is slightly higher in the left than in the right hemisphere.
Prior to VOI evaluation each brain image was individually normalized to the average FDG uptake in the 15% brain voxels with highest uptake. This normalization implies that voxels with highest FDG uptake are not affected by ecstasy induced effects. According to our findings this assumption seems to be not fulfilled completely since putamen and caudate, which show the most significant difference between ecstasy users and controls, belong to the 15% brain voxels with highest uptake. However, since the total volume of putamen and caudate (~30 ml) is only about 20% of the normalization volume, the effect of a slightly reduced FDG uptake in these structures on the normalization might be neglected.
In the present work ‘hottest voxel analysis’, i.e. sampling the voxel with the highest uptake, has been used to characterize glucose uptake within a given structure. This approach is known to be susceptible to statistical fluctuations [47]. However, if applied to sufficiently smoothed data it has been demonstrated to be the best method of discriminating between groups of subjects supposed to differ with respect to glucose utilization within small brain structures [48].
In the statistical evaluation no Bonferroni adjustment (alpha adjustment) was applied to take into account multiple testing (number of regions of interest (ROIs), number of correlations). However, the regions showing the statistically most significant difference in the group means, putamen and caudate, were affected in both hemispheres. This would not be expected if the P values were small due to type I errors.
Therefore what can any study that hopes to look at elicit drug use conclude ?
Ecstasy abuse may cause lasting effects on central nervous activity in humans, with teenagers being at higher risk than adults.
While some may argue these studies don't prove anything - because of their limitations ... they are likely to be the best evidence we have and are ever likely to get!