Why do stimulants not work in typical depression?

Given the energizing properties of psychostimulants, it was deemed promising to test their antidepressant effects, especially when antidepressants were not yet available. First studies on stimulants in depression were published in the 1930s, but later reviews concluded that stimulants neither had a primary effect on mood nor that they reversed the depressive process (e.g. Shaw, 1964). However, research in this field is complex because effects and side effects of stimulants strongly depend on both outcome variable (e.g. effects on mood vs effects only on drive or attention) and applied dose (Arnsten, 2006; Busardo et al., 2016; Wood et al., 2014). Additionally, long-term effects of stimulants in depression are scarcely studied, although even in attention-deficit/hyperactivity disorder (ADHD) it is not sure whether the established short-term effects of stimulants remain overlong periods of treatment (Mulder et al., 2016). Furthermore, numerous different stimulants had been tested over the decades, such as amphetamines, methylphenidate, pemolineand (ar)modafinil, with the latterdrug having strongly renewed the interest in stimulants in recent years. And indeed, the topic is still controversial, as can be exemplified in a recent plea, underpinned by personal experience, for the role of dextroamphetamine in treatment of depression (Horgan, 2016; comment to Malhiet al., 2016). Malhi et al. had, in contrast, raised some concerns about the continued out-of-label use of stimulants in depression despite lack of evidence for efficacy.

First, we want to note that companies, which spent a large sum of money in clinical trials for stimulants as add-on in depression, finally stopped their efforts in receiving regulatory approval due to lack of efficacy: Shire (2014) stopped their program for lisdexamfetamine add-on in unipolar depression after two recent phase 3 studies failed to meet the primary efficacy endpoint. The same in bipolar depression, where two phase 3 multicentre studies (NCT01072630, NCT01305408) with armodafinil failed to reach the primary outcome criterion, and therefore, Teva (2013) announced that the company would no longer proceed with regulatory filings. Additionally, two other recent studies with lisdexamfetamine augmentation in bipolar depression (NCT01093963, NCT01131559) were stopped by the sponsor. Thus, there is indeed a discrepancy between the common belief that stimulants might help in depression and the lack of evidence to support this belief.

In the following, we present the recently introduced arousal regulation model of affective disorders, which provides an explanation for the lack of efficacy of stimulants in typical depression and that might be helpful to identify possible responders. Arousal has been suggested as one dimension of the Research Domain Criteria (RDoC) project of the National Institute of Mental Health (NIMH; Cuthbert and Insel, 2013), and it seems probable that differences in arousal predict response to arousal-modulating stimulants.

The arousal regulation model of affective disorders (Hegerl et al., 2016; Hegerl and Hensch, 2014) suggests that upregulation of arousal plays a pathogenetic role in typical depression. In such patients, arousal-stabilizing stimulants are unlikely to be helpful.

The upregulation of arousal in typical depression has been demonstrated in older (Ulrich and Fuerstenberg, 1999) and more recent electroencephalography (EEG) studies (Hegerl et al., 2012; Schmidt et al., 2016; Ulke et al., 2016), with the latter studies applying the freely available Vigilance Algorithm Leipzig (VIGALL, http://research.uni-leipzig.de/vigall/; Huang et al., 2015; Sander et al., 2015). VIGALL is an EEG- and electrooculography (EOG)-based software, which objectively classifies arousal levels by automatically attributing one out of seven arousal states (on the continuum from active wakefulness to sleep-onset) to each EEG segment of a 15-minute resting EEG recording (segment length 1 second as default). The upregulated arousal found in unmedicated depression is in line with the hyperactivity of the hypothalamic–pituitary–adrenal axis and the self-reported inner restlessness and tension. Other depressive symptoms, such as withdrawal from all social interactions and sensation avoidance, can be interpreted as an autoregulatory reaction of the organism to the upregulated arousal (Hegerl and Hensch, 2014). Additionally, the therapeutic effect of antidepressants (Hegerl and Hensch, 2014; Hensch et al., 2015), sleep deprivation and anticholinergic drugs might partly be due to their arousal-reducing properties (Hegerl et al., 2016; Hegerl and Hensch, 2014). It is a consistent preclinical finding that antidepressants reduce the firing rate of neurons in the noradrenergic locus coeruleus, which plays an important role in arousal regulation (West et al., 2009).

The common use of stimulants in typical depression partly results from a lack of precision in psychopathological description. Only at first glance, patients with typical depression appear to be suffering from sleepiness and lack of drive, symptoms which might respond to psychostimulants. However, closer examination reveals that many of them are exhausted and weary, though not sleepy, and that they do not suffer from a lack but from inhibition of drive combined with high inner tension (Hegerl et al., 2013). Patients with typical depression have prolonged sleep-onset latencies during the day and suffer from insomnia during the night. Therefore, psychostimulants with their arousal-stabilizing and arousal-increasing properties are unlikely to be helpful in these patients.

However, there might be subgroups with downregulated arousal, such as atypical depression or depression with fatigue in the context of inflammatory or immunological processes (Hegerl and Ulke, 2016). Indeed, there is some preliminary evidence that stimulants might be helpful in depression in patients with severe co-occurring non-psychiatric illness and in geriatric patients, two largely overlapping subgroups (Lavretsky et al., 2015; Ng et al., 2014). Also in bipolar disorder, a higher proportion of patients display atypical features including hypersomnia, and one can speculate that these patients might also benefit from stimulants (Hegerl and Hensch, 2014). The fact that some secondary outcome variables, such as remission rate at the end of the trial, became significant in the recent armodafinil add-on trial in bipolar I disorder could point in this direction (Frye et al., 2015).

To separate patients with upregulated arousal from those with downregulated arousal might increase the pathophysiological and therapeutic validity of diagnostic classifications in future clinical trials.