Sample characteristics are summarized in . No statistically significant differences in age or gender distributions were found between the MDMA and non-MDMA user groups (p>0.05, Mann–Whitney and χ2 tests). No statistically significant associations of age or sex with any activation measure within any region were found (p>0.05, Spearman's rank correlations). Additional sample characteristics are provided in Online Supplement in Section ‘Additional Sample Characteristics'.
Effect of MDMA. Activation produced by the high-intensity stimulation is shown in . The major finding from the within-subject correlation analysis was that greater lifetime MDMA use was statistically significantly associated with greater stimulus-evoked visual activation for signal intensity in LGN, BA 17, and BA 18 (; ), and for spatial extent of activation in BA 17 and BA 18 () (p>0.05, Spearman's Rank correlations). There was no statistically significant association of the duration of MDMA abstinence with any outcome measure for any ROI.
Figure 3 Relationship of lifetime 3,4-methylenedioxymethamphetamine (MDMA) use to signal intensity in lateral geniculate nucleus (LGN), Brodmann Area (BA) 17, and BA 18. Y axis is the rank of signal intensity for the ROI (calculated as mean percent BOLD signal (more ...) Effect of combined MDMA and methamphetamine.
Of the 20 MDMA subjects, seven also reported use of methamphetamine (METH). Because METH produces serotonergic toxicity similar to that caused by MDMA (Peat et al, 1983
; Woolverton et al, 1989
), we reasoned that both drugs might produce similar and therefore additive effects on brain activation, especially if the observed findings are due to altered 5-HT neurophysiology. To explore this possibility, we created a measure of substituted amphetamine exposure that combined MDMA and METH exposures. As we did not have a priori
data to suggest the mg equivalent 5-HT toxicity produced by each drug in human beings, we simply summed the lifetime quantity of MDMA and METH use (in mg), shown as MDMA+METH in . Overall, the composite measure accounted for 32% of the variance (as the square of rs
) for LGN signal intensity vs
35% for MDMA alone. For BA 17 signal intensity, the composite measure accounted for 35, vs
25% of the variance for MDMA alone, with similar results found for BA 17 spatial extent, and for both signal intensity and spatial extent in BA 18. As for MDMA, there was no statistically significant association of the duration of METH abstinence with any activation measure in the METH-exposed subgroup.
Other drug effects. Although our primary analysis revealed that greater lifetime MDMA use, and greater lifetime use of MDMA+METH was associated with greater activation in all visual regions, MDMA users in our cohort (consistent with worldwide use patterns) reported exposure to multiple recreational drugs other than MDMA. We therefore examined the data for the influence of other drugs. Because partial correlations require data for all three variables in a specific analysis (activation measure, MDMA use, specific other drug use) and to maintain a sample of size of at least 10, these analyses were conducted only for those drugs used by at least 50% of the MDMA cohort (; alcohol, cannabis, cocaine, codeine, LSD, opium, psilocybin, and sedatives). There were no statistically significant associations of lifetime quantity of drug use with activation outcome variables (Spearman's rank correlation; p>0.05).
Effects of Scanner/Stimulus Delivery Method
As noted in the Materials and Methods, we acquired data on two different scanners, each employing a different stimulus delivery method (goggle vs projection screen). It is theoretically possible that the use of the different scanner/stimulus delivery methods might have affected our results. We conducted an analysis of the potential effects of these differences on our findings.
Lifetime quantity of MDMA use by scanner/stimulus method. MDMA use levels showed mean differences and correlations with scanner type at a trend (p<0.10 level). Mean MDMA use was slightly greater for MDMA users studied on the GE scanner (p=0.076; Mann–Whitney) (Supplementary Table S1). Scanner type was nonsignificantly correlated with lifetime quantity of MDMA use in mg (rs=0.41, p=0.074; Spearman's rank correlation).
Relationship of scanner to correlation of lifetime MDMA use with signal intensity. Supplementary Figure S1 depicts scatter plots of data from GE and Philips scanners to show the degree of overlap and pattern of data from the two scanners. The MDMA users studied on the GE scanner had higher minimum use levels than the Philips group, but there was good overlap for use levels in the higher use ranges.
Partial correlation analysis controlling for scanner effects. Although the graphic analysis suggests no scanner/stimulus-specific differences in slope of the relationship, we conducted a statistical analysis of the data adjusting for scanner/stimulus condition. We anticipated that as MDMA use was correlated at a trend level with scanner/stimulus method and as the additional scanner covariate would reduce the available degrees of freedom, a partial correlation analysis adjusted for scanner/stimulus would yield a weaker relationship between MDMA use and regional signal intensity. However, despite these factors, the partial correlation of lifetime quantity of MDMA use with regional visual stimulus-evoked signal intensity remained positive, but significantly so only for LGN (LGN, r=0.60, p=0.006; BA 17, r=0.35, p=0.092; and BA 18, r=0.351, p=0.141).
As predicted by data from our earlier study, there was general overlap between control group and MDMA user group activation measures, and there were no statistically significant differences between the MDMA and control groups in average signal intensity within the LGN, BA 17, or BA 18 (; ). In addition, there were no statistically significant between-group differences in the spatial extent of activation for BA 17 and BA 18 (; ). As noted above, spatial extent analysis was not performed for the LGN.
Figure 4 Signal intensity by brain region and group. Plot of individual signal intensity values by group shows the degree of overlap between the control and 3,4-methylenedioxymethamphetamine (MDMA) groups. As shown in , mean signal intensity did not differ (more ...)
Figure 5 Spatial extent of activation by brain region and group. Plot of individual spatial extent of activation by group shows the degree of overlap between the Control and 3,4-methylenedioxymethamphetamine (MDMA) groups. As shown in , mean spatial extent (more ...)
Given the relationship between lifetime MDMA exposure and activation shown above for the within-group analysis, we anticipated that the lack of between-group differences may have been secondary to the fact that the MDMA user group was not homogenous—that is, individuals with greater MDMA exposure would have higher mean activation measures than those with lower exposure. To address this issue, we conducted a subgroup analysis using a median split procedure to divide the MDMA user group into high- and low-exposure groups. To aid in controlling for Type I error, we used a non-parametric ANOVA (Kruskal–Wallis), followed by a post hoc Mann–Whitney test. This analysis revealed that the high-exposure MDMA group had significantly greater spatial extent of activation when compared with the control group, but there were no significant between-group differences in signal intensity (). In contrast to the greater spatial extent of activation seen for the comparison of high-exposure MDMA users to the control group, the only significant difference between low-exposure MDMA users and control subjects was found in the LGN where the low-exposure MDMA subgroup had lower signal intensity than the controls (). Only the signal intensity difference in LGN met the adjusted α-level (p<0.017) after accounting for the number of comparisons.