Prediction of MDMA response in healthy humans: a pooled analysis of placebo-controlled studies

Abstract

Background: 3,4-methylenedioxymethamphetamine (MDMA, "ecstasy") is used both recreationally and therapeutically. Little is known about the factors influencing inter- and intra-individual differences in the acute response to MDMA. Effects of other psychoactive substances have been shown to be critically influenced by personality traits and mood state before intake.

Methods: We pooled data from 10 randomized, double-blind, placebo-controlled, cross-over studies performed in the same laboratory in 194 healthy subjects receiving doses of 75 or 125mg of MDMA. We investigated the influence of drug dose, body weight, sex, age, drug pre-experience, genetics, personality and mental state before drug intake on the acute physiological and psychological response to MDMA.

Results: In univariable analyses, the MDMA plasma concentration was the strongest predictor for most outcome variables. When adjusting for dose per body weight, we found that (a) a higher activity of the enzyme CYP2D6 predicted lower MDMA plasma concentration, (b) a higher score in the personality trait "openness to experience" predicted more perceived "closeness", a stronger decrease in "general inactivation", and higher scores in the 5D-ASC (5 Dimensions of Altered States of Consciousness Questionnaire) scales "oceanic boundlessness" and "visionary restructuralization", and (c) subjects with high "neuroticism" or trait anxiety were more likely to have unpleasant and/or anxious reactions.

Conclusions: Although MDMA plasma concentration was the strongest predictor, several personality traits and mood state variables additionally explained variance in the response to MDMA. The results confirm that both pharmacological and non-pharmacological variables influence the response to MDMA. These findings may be relevant for the therapeutic use of MDMA.

Keywords: MDMA; individual differences; prediction; safety; set and setting.

Figure 1.

Standardized regression coefficients and statistical significance of each predictor variable in the linear mixed effects models adjusting for drug dose per body weight. Smaller asterisks show the uncorrected statistical significance. Bigger asterisks show the significance after correction for multiple testing across all 20 × 25 = 500 significance tests using the Benjamini–Hochberg procedure (Benjamini and Hochberg, 1995). *p < 0.05, **p < 0.01, ***p < 0.001. AUC, area under curve; AMRS, Adjective Mood Rating Scale; ASC, altered state of consciousness; NEO-FFI, NEO Five Factor Inventory; STAI, State-Trait Anxiety Inventory; VAS, Visual Analog Scale.

Figure 2.

Size of the penalized regression coefficients and rank of importance of the predictor variables in the LASSO models. As one Lasso model was developed for each response variable, each column in the tile plot reports the results of one Lasso model. The rank of importance of each predictor for each outcome was determined by ranking the predictor variables according to their absolute size of the regression coefficients in each Lasso model. AUC, area under curve; AMRS, Adjective Mood Rating Scale; ASC, altered state of consciousness; NEO-FFI, NEO Five Factor Inventory; STAI, State-Trait Anxiety Inventory; VAS, Visual Analog Scale.