Abstract
The antidepressant, agomelatine, has a novel pharmacological profile, with agonist properties at M1 and M2 receptors and antagonist properties at 5HT2C receptors. Whether the antidepressant effects of this treatment are mediated by the drug’s effects at the M1 and M2 receptors or the 5HT2C receptor or a synergy between these actions remains unclear. In the present study, a healthy volunteer model of emotional processing, which discriminates between effective and non-effective antidepressant compounds, was used to assess the contribution of melatonin agonism to the efficacy of agomelatine. Fifty-eight healthy volunteers were randomised to receive 7 days of once daily treatment with either 1 mg melatonin, 3 mg melatonin or placebo. Seven days treatment with 3 mg melatonin resulted in earlier bedtimes consistent with a phase advance in circadian rhythm. Some marginal effects of melatonin were observed on emotional processing; however, these were neither consistent with nor comparable to those seen following conventional antidepressant treatment or with agomelatine itself. These data suggest that the antidepressant action of agomelatine cannot be accounted for solely by its action at the M1 and M2 receptors.
Introduction
The antidepressant, agomelatine, has a novel pharmacological profile, which includes agonism of melatonin M1 and M2 receptors, as well as antagonism of serotonin 2C (5HT2C) receptors. Recent meta-analyses suggest agomelatine is an effective antidepressant (Taylor et al., 2014; though see Koesters et al., 2013).
Chronobiotic effects may underlie the mechanism of action of agomelatine in depression (Bunney and Potkin, 2008), but whether the antidepressant effect of agomelatine is mediated via agonism of M1 and M2 receptors or the antagonism of 5HT2C receptors or a synergy between these effects remains unclear. The synergy between melatonin agonism and 5HT2C receptor blockade, however, is believed to underlie an increase in catecholamine release in prefrontal cortex and reduced glutamate signalling in response to stress in animal models following agomelatine administration (Racagni et al., 2011; Stahl, 2014).
Agomelatine is a naphthalenic analogue of melatonin, which shows high affinity for cloned human melatonin M1 and M2 receptors (PKi values of 9.92 and 9.42 for the M1 and M2 receptors respectively; Ettaoussi et al., 2013). Unlike melatonin, agomelatine is reported to be an antagonist at the 5HT2C receptor, although its affinity for this receptor is lower than for the melatonin receptors (PKi value of 6.15; Millan et al., 2003) and whether functional blockade of these receptors occurs following agomelatine treatment in humans has been debated (Norman, 2012; Sharpley et al., 2011). This is interesting as some effective antidepressants (e.g. mirtazapine) include 5HT2C antagonism in their pharmacological profile. Whereas a role for melatonin receptors in antidepressant treatment is implied by the disruption of circadian rhythm, which is common in depression (Kronfeld-Schor and Einat, 2012). Indeed, given the high affinity of the drug for the M1 and M2 receptors, and the ability of melatonin to shift circadian rhythm (Leproult et al., 2005), one prediction would be that the antidepressant mechanism of agomelatine in humans might be entirely attributable to effects at M1 and M2 receptors. However, current evidence suggests that melatonin itself has only weak antidepressant effects (Dolberg et al., 1998, Serfaty et al., 2010).
A recently developed healthy volunteer model of emotional processing has been demonstrated to discriminate effective antidepressant compounds from those that failed in large-scale clinical trials (Harmer et al., 2009). This model has the potential to be more sensitive to the detection of antidepressant action than conventional measures of mood in depressed patients and its use in healthy volunteers allows the consideration of the neuropsychological effects of compounds free from the confound of negative mood state. The model uses behavioural measures of facial affect recognition, emotional categorisation and memory and emotional memory and has been validated using a range of antidepressants with differing pharmacological profiles, including, for example, the tetracyclic mirtazapine (Arnone et al., 2009). Agomelatine shows positive effects in this emotional model, reducing the perception of sad facial expression and enhancing positive emotional memory (Harmer et al., 2011). We therefore reasoned that the model would be a useful assay of how far melatonergic agonism can account for the antidepressant action of agomelatine.
Method
Participants
We recruited a total of 58 healthy volunteers who gave informed consent to participate in two studies (independent samples in each study) investigating the effects of 7 days treatment with melatonin at doses of 1 mg vs. placebo (study 1: 34 volunteers, 17 female, mean age 21.82, range 18–36) and 3 mg vs. placebo (study 2: 24 volunteers, 13 female, mean age 22.7, range 20–41). Participants were screened to be free of current or previous Axis I psychiatric disorder on the Structured Clinical Interview for DSM-IV. None of the participants was currently taking medication except for the contraceptive pill; the premenstrual week was avoided for treatment and testing. The study was reviewed by the local ethics committee.
Procedure
In a double blind procedure, participants were randomised to receive 7 days of melatonin (study 1: 1 mg, 17 volunteers allocated to melatonin; study 2: 3 mg, 12 volunteers allocated to melatonin; Pharma Nord) or placebo treatments. Treatment was taken at 6.00 pm each day. On the eighth day participants came into the laboratory at 9.00 am to complete the emotional task battery, which has been previously described (Eg Horder et al., 2009) and comprises a facial expression recognition test, an emotional categorisation task with surprise emotional recall and recognition, a dot-probe task of emotional attention with masked and unmasked conditions. Baseline and post treatment measures of mood and anxiety (Beck Depression inventory (BDI); Beck et al., 1961) and State and Trait Anxiety Inventory (STAI); Spielberger et al., 1970) and daily measurements of sleep and subjective state (Befindlichkeits Scale (BFS); von Zerseen et al., 1974) and Positive and Negative Affect Scale ((PANAS); Watson et al., 1988) were also taken.
Analysis
The placebo groups from both of the studies showed comparable performance on the emotional task battery (all p values >0.2). The effects of 1 mg and 3 mg melatonin are therefore compared both against each other and against the combined placebo groups from both studies. Demographic and baseline data were compared between the groups using one way analysis of variance (ANOVA). Day by day variations in subjective state (BFS and PANAS) were compared between groups using repeated measures ANOVA with treatment as the between-subjects factor and time as the within-subject factor. Time to bed and sleep duration were assessed using repeated measures analysis of covariance (ANCOVA; see results for details of covariates). Data from the emotional task battery were analysed using repeated measures ANCOVA (details of covariates given in Results) with dependent variables of accuracy, misclassifications reaction time and reaction time vigilance as appropriate (Horder et al., 2009).
Results
Demographic information, baseline characteristics and subjective state
At baseline, the groups were well matched for mood (BDI) and age (p values >0.1; Table 1). There were, however, between-groups differences in IQ (National Adult Reading Test; F(2,53)=8.26, p<0.01; Table 1), neuroticism (Eysenck Personality Questionnaire - Neuroticism subscale; F(2,53)=3.20, p=0.048; Table 1) and trait anxiety (STAI-T; F(2,57)=4.94, p=0.01; Table 1). IQ, trait anxiety and neuroticism were correlated with several of the behavioural measures on the emotional task battery and therefore included a covariates in these analyses. Trait anxiety and neuroticism were also correlated with some of the sleep measures and were included as covariates in this analysis. IQ was not correlated with the sleep measures. Day by day variations in subjective state (BFS and PANAS) did not vary between the groups (p values >0.1).
Demographic information, baseline characteristics and sleep variables.
NART: National Adult Reading Test; EPQ: Eysenck Personality Questionnaire; STAI: State and Trait Anxiety Inventory.
p value denotes value for the effect of group in one way ANOVA test.
Bed time is given as minutes after 6.00 pm.
3mg vs. placebo difference p=0.046.
Sleep onset and duration
Three mg, but not 1 mg, of melatonin resulted in earlier bedtimes compared to placebo on average across the week (main effect of group; F(2,53)=3.39, p=0.04; Bonferroni’s corrected pairwise comparison placebo vs. 3 mg p=0.046; Table 1). There was no effect of melatonin on sleep duration over the week (all p values >0.2, Table 1).
Emotional task battery
Effects of melatonin on emotional processing were assessed by considering emotion or valence by group interactions as well as main effects of group in repeated measures ANCOVAs. While there were no effects of the intervention on the emotion categorisation task in terms of accuracy or reaction time (all p values >0.4), there was an effect of group on the subsequent emotional recall task (F(2,26)=3.59, p=0.036). This effect of group was the result of increased recall of both positive and negative words in the 3 mg group compared to the 1 mg but not the placebo group (Bonferroni’s corrected pairwise comparisons 3mg vs, 1 mg: p=0.04; 3 mg vs. placebo p=0.62, 1 mg vs. placebo p=0.24; Figure 1). There was no effect in the recall task on false intrusions (all p values >0.09). In the emotion memory recognition task, while there were no effects in terms of accuracy or reaction times (all p values >0.1), there was a trend towards group differences in false intrusions across both valences (main effect of group: F(2,45)=2.96, p=0.06)), this appeared to be driven by reduced intrusions in the 3 mg group compared to placebo (placebo estimated marginal mean: 11.98 (SE 1.26); 1 mg: 8.53 (1.51); 3 mg: 6.35 (2.21); Bonferroni’s corrected pairwise comparisons 3 mg vs. placebo: p=0.10; 1 mg vs. placebo: p=0.28; 1 mg vs. 3 mg p=1.0). Dot probe reaction times were converted to vigilance scores of differences between congruent (probes appeared in the position of the emotional word) and incongruent (probes appeared in the opposite position to the emotional word) trials. Here there was a trend towards group differences in overall vigilance, which appeared to be driven by increased vigilance away from emotional and towards neutral cues following 3 mg melatonin (F(2,47)=3.10, p=0.055; placebo mean −2.00 (SD 49.26), 1 mg −2.37 (37.42), 3 mg −12.74 (45.66), all other p values on this task > 0.1).
Number of valenced personality characteristic words (positive and negative) recalled in surprise memory test. Figure shows estimated marginal means for each group, which are corrected for the inclusion of the covariates of trait anxiety, neuroticism and IQ. Error bars show SE. *p<0.05.
Discussion
In summary, the current results support some neuropsychological effects of 3 mg but not 1 mg melatonin in healthy volunteers. Specifically, 7 days treatment with 3 mg (but not 1 mg) melatonin at 6.00 pm resulted in earlier bedtimes compared to placebo treatment. Although there were no selective effects of melatonin on the processing of positive vs. negative affective stimuli, the higher dose of melatonin tended to improve memory and reduce vigilance to emotional stimuli.
While melatonin 3 mg resulted in significantly earlier bedtimes, it did not affect total sleep duration. This suggests that, at the higher dose, the treatment may have resulted in a shift in circadian rhythm, consistent with other studies, which have demonstrated dose-dependent shifts in circadian rhythms after melatonin in healthy volunteers (Arendt et al., 1985, Deacon and Arendt, 1995, Rajaratnam et al., 2004).
Compared to 1 mg, 7 days’ treatment with 3 mg melatonin resulted in an increase in the number of both positive and negative valence words recalled in the emotional recall memory task; however, the difference between the placebo and 3 mg groups was not statistically significant. Consistent with a possible tendency for improved recall memory, there was a trend for a reduction in false intrusions on recognition memory also at 3 mg melatonin compared to placebo, suggesting greater discrimination in memory recognition. Such a finding implies that, while emotional processing biases may not be affected by melatonin treatment, there may be an overall modest improvement in memory processes, following 3 mg but not 1 mg treatment, although the interpretation of the recall finding is difficult given the lack of difference from the placebo group. In older adults with mild cognitive impairment, some studies have suggested that longer term melatonin treatment may improve cognition (Cardinali et al., 2012), although meta-analyses have failed to support this and the mechanism of any putative pro-cognitive effect is unclear (Jansen et al., 2006). Given that short-term treatment with 3 mg melatonin shifted bedtimes, it is possible that any small improvements in memory seen at this dose could be secondary to changes in sleep architecture (for example, rapid eye movement and/or short wave sleep duration and/or frequency). In a previous study, however, with lower dose melatonin, we found that melatonin treatment improved sleep efficiency without altering total rapid eye movement sleep or slow wave sleep time (Attenburrow et al., 1996).
The current study also revealed a small effect of melatonin on reaction time vigilance in the attentional dot probe, so that at 3 mg, but not at 1 mg or placebo, there was increased vigilance towards neutral probe words and away from emotional valenced probe words. Both antidepressants (generally those that are also useful in the treatment of anxiety) and anxiolytic treatments tend to result in reduced vigilance to negative stimuli (Murphy et al., 2008, 2009). Future studies are needed to assess if the effect seen with 3 mg melatonin can be replicated in a larger sample and whether it may be related to a potential anxiolytic action of melatonin treatment.
Together, these findings suggest that 7 days’ treatment with 3 mg of melatonin does result in earlier bedtimes consistent with a phase advance in circadian rhythm, presumably due to melatonin receptor agonism. However, there were only marginal effects on emotional processing, which were not comparable to those seen following conventional antidepressant treatments (Harmer et al., 2009) or agomelatine, where 7 days’ treatment resulted in a reduction of the perception of sadness and increased memory for positive emotional stimuli (Harmer et al., 2011). To the extent that the emotional processing model predicts clinical antidepressant activity, it therefore appears unlikely that the antidepressant effects of agomelatine can be solely attributed to melatonin agonism. A caveat to this conclusion is the use of a healthy volunteer model in the present study, a further study to confirm these findings in a depressed sample, where treatment may interact with pathological differences in the brain, for example, at the level of melatonin receptors.
Footnotes
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: C.J.H. has acted as a consultant for the following companies: Servier, GSK, Astra-Zeneca. Lundbeck and P1vital. She also holds shares in P1vital and is on the advisory board.
P.J.C. has been a paid member of advisory boards of Eli Lilly, Lundbeck, Servier and Wyeth and has been a paid lecturer for Eli Lilly, Lundbeck, GlaxoSmithKline and Servier. He has received remuneration for scientific advice given to legal representatives of GlaxoSmithKline
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the MRC and a Wellcome Summer Studentship awarded to MB.