Psychostimulants as antidepressants: Their nuanced role?

Abstract

While effectively the ‘first antidepressants’, the psychostimulants are rarely prescribed as antidepressant drugs seemingly in light of their judged low effectiveness, side effects, tolerance as well as concerns about dependency and abuse. Recent meta-analyses do find some support for them as being effective antidepressants for those with major depression, but they have not been closely evaluated in terms of their specific nuanced role for treating treatment-resistant (unipolar and bipolar) melancholic depression. The author has so prescribed them for over a decade and offers a case for their benefits for a distinct percentage of those with such conditions and notes their relatively few side effects.

Stimulants such as amphetamine and methylphenidate (introduced to market in the late twenties and mid-fifties respectively) were viewed then as having multiple properties including antidepressant actions. Their current status as ‘antidepressants’ remains unclear, and in light of several concerns (e.g. their effectiveness, side effect profile, risk of dependency and abuse), few clinicians prescribe them. I seek to argue that they can be highly effective in a sub-set of depressive conditions.

Support for their use as antidepressants largely emerged from case reports (e.g. Fawcett et al., 1991; Feighner et al., 1985). In one study (Gurian and Rosowsky, 1990), remission or distinct improvement was quantified in some 80% of patients and with low rates (i.e. less than 10%) of side effects quantified. In relation to benefits, Orr and Taylor (2007) noted that they reduced fatigue and promoted alertness, thus encouraging their use to treat depression. Candy et al. (2008) suggested that rapid action and capacity to reduce fatigue (a common depressive symptom) were key factors in their favour compared to more orthodox antidepressants.

Their evaluation in meta-analyses has been relatively limited until recently. An earlier Cochrane review (Candy et al., 2008) identified 24 randomized controlled trials and found support for the short-term efficacy of five differing psychostimulants (methylphenidate, dexamphetamine, methylamphetamine, modafinil and pemoline) in that there was a statistically significant reduction in depression severity. They judged, however, that most studies were too brief (generally of 4 weeks duration) and that more high-quality controlled studies were required to determine their comparative efficacy and in which ‘clinical situations they are optimal’.

There has been limited consideration of their role within the Australasian region. Malhi et al. (2016) noted that high-level evidence (e.g. randomized controlled trials) evaluating their efficacy for managing unipolar and bipolar depressive disorders was relatively scarce, their safety and tolerability was not able to be readily evaluated in light of a lack of long-term studies, and that, while uncommon, there was a risk of them causing ‘switching’ into mania. They speculated that their off-label use reflected their expediency of response, as well as their induction of feelings of euphoria and increased activity, and that their rapid action compared to conventional antidepressants indicated that they exert their action via differing mechanisms. Specifically, they contemplated whether they ‘augment’ antidepressant drugs – or more ‘facilitate’ the action of antidepressants.

Salient meta-analyses have emerged in recent years. One meta-analysis of data reported in 21 articles was detailed by McIntyre et al. (2017), examining 4 psychostimulants (methylphenidate, dextroamphetamine, lisdexamfetamine and modafinil) prescribed for adults with ‘major depression’ or ‘bipolar depression’, albeit with many of the studies having low participant numbers and large effect sizes as quantified by odds ratios (ORs) and with the asymmetric distribution of their funnel plots being ‘suggestive of publication bias’ (p. 414). Their analyses quantified that the psychostimulants were associated with ‘statistically significant improvement in depressive symptoms’ in both those with ‘major depressive disorder (OR = 1.41) and bipolar disorder (OR = 1.42), but later (p. 416) observed that, as a consequence of a ‘surfeit of limitations ... a strong conclusion about efficacy cannot be made’. Efficacy impact was slightly higher when these drugs were evaluated as monotherapies (OR = 2.25) than when used adjunctively (OR = 1.39). They also judged that the psychostimulants have been insufficiently studied as monotherapies or as adjunctive medications in adults with mood disorders and that such drugs might better be evaluated in ‘select domains’ rather than as ‘broad spectrum’ antidepressants.

Bahji and Mesbah-Oskui (2021) reported a network meta-analysis of 37 studies (with about one half of the studies addressing their use as adjunctive agents and only three examining their benefits for those with treatment-resistant depression), evaluating the efficacy and side effects of psychostimulants for ‘depression’. They quantified that the psychostimulants were associated with improvement in depression response rates and depression severity, but that in the network meta-analysis, the only psychostimulant showing significant antidepressant efficacy was methylphenidate. They concluded, however, that a ‘lack of consistent evidence precludes a definitive hierarchy of treatments and points to a need for additional, highly-quality RCTs’ (p. 416). Thus, despite the psychostimulants being ‘off patent’ (a state which generally leads to few efficacy studies being undertaken), the referenced meta-analyses argue for a reasonably sized database.

A key limitation, however, to such meta-analyses is that contributing studies generally consider their efficacy for those with ‘major depression’ or, less commonly, ‘treatment-resistant depression’ or even simply those with ‘depression’. Major depression is not an entity as it subsumes multiple heterogeneous depressive conditions. The relevant consequential risk can be illustrated with a medical analogy. If a highly efficacious bronchodilator for asthma was evaluated in those with ‘major breathlessness’ (and only a minority of study participants actually had asthma), then that drug would risk appearing to have only marginal or no benefit as an ‘anti-asthma’ medication. Similarly, ‘treatment-resistant depression’ should not be viewed as a diagnosis or an entity but as a descriptor of clinical non-response and which might reflect multiple factors. Thus, there is a need to concede that the psychostimulants have benefit only to one or more depressive sub-types, so emphasizing the observation by McIntyre et al. (2017) that they may have a nuanced role in select domains only.

I now turn to some personal observations, which are anecdotal and thus run the risk of bias. For several decades, I might prescribe a monoamine oxidase inhibitor (MAOI) to patients with a melancholic depression that had failed to respond to several differing antidepressant classes, including a tricyclic. In light of their demanding dietary restrictions and the risk of hypertensive crises and other significant consequences, I commenced trialling psychostimulants for those with treatment-resistant (unipolar and bipolar) melancholic depression some 15 years ago. The logic was that (1) perturbed dopaminergic transmission is responsible to some degree for the mood changes and psychomotor disturbance in those with melancholic depression (Hickie, 1996), (2) methylphenidate is a reuptake inhibitor of dopamine and norepinephrine and dexamphetamine is a potent dopamine and noradrenalin releaser, (3) there is a close mechanistic link between MAOIs and amphetamine (Stahl, 2008) and (4) to the extent that the MAOIs are viewed as commonly effective for those with melancholia, then the psychostimulants should have similar antidepressant properties but a superior ‘cost’ profile in not requiring dietary limitations and incurring the risk of a hypertensive crisis.

Two reports were subsequently published (Parker and Brotchie, 2010; Parker et al., 2013) with the latter comprising a larger sample of 112 patients (61 unipolar and 51 bipolar II) with melancholic depression and who had received no benefit from two or more antidepressants and/or had experienced distinctive side effects from antidepressants. Both the unipolar and bipolar patients had previously received a mean of five antidepressants, while the total number of previously trialled psychotropic drugs was seven and eight, respectively.

All bar four patients (who received dexamphetamine) were prescribed methylphenidate (mean dose 25 mg), 15% as monotherapy and the remainder as an augmentor. Some 50% of the bipolar and 49% of the unipolar patients discontinued the psychostimulant, necessitated by significant side effects in 12%. Whether continuing or having discontinued the medication, the most commonly reported side effects were anxiety, irritability or agitation, feeling ‘buzzy’ or ‘jazzy’, impaired concentration, sedation or fatigue, tachycardia or mood worsening. Tolerance was reported by 8% of the bipolar and 7% of the unipolar subjects, while 6% and 7% respectively reported a rise in blood pressure. Of those with a bipolar disorder, 10% judged that the drug induced a ‘high’ (compared to 3% in the unipolar group) while 80% of the bipolar patients judged that the drug did not worsen the pattern or severity of their ‘high’.

Whether continuing or having discontinued the medication, comparable rates of improvement were reported by the unipolar and bipolar patients (at a mean follow-up period of 69 weeks), with some 20% reporting the medication as ‘very effective’, an additional one-half as ‘somewhat effective’, and some 30% in each group judging it as ‘not effective at all’. Some 45% described the psychostimulant as the best or equal to the best previously prescribed antidepressant. In terms of benefit, 46% reported a sustained benefit, 26% a relapse following a response, 25% no benefit at all and 3% that their condition had worsened; while 42% judged the medication as worthy of being maintained.

I have now prescribed a psychostimulant (principally methylphenidate but occasionally dexamphetamine or a long-acting preparation) and almost invariably as an augmentor of a formal antidepressant drug (as against a monotherapy) to some 300 unipolar or bipolar patients with a drug-resistant melancholic disorder and have assessed many patients with a non-melancholic disorder who have previously received a psychostimulant and now offer the following observations after acknowledging several risks of bias. First, my support for their role may be weighted by having reviewed principally only those who continued and continue with the medication. As documented here, the discontinuation rate in my practice is in the order of 50% and thus I may weight the positive experience of the other 50% who continue with their medication. In defence, improvement being reported by some 50% of those with a treatment-resistant melancholic depression wishing to continue with their medication strikes me as significant in and of itself. Second, I might generate expectancy effects. Third, my measurement of benefit is broad and not as refined as would occur in formalized trials. Fourth, there may be recall biases.

I do not believe that such psychostimulants have any distinct role in managing those with a non-melancholic depression, showing minimal efficacy (other than often improving concentration), and with rates of tolerance, dependency and abuse adding an additional concern.

I view their antidepressant contribution to be loculated to those with unipolar or bipolar II melancholic depression and thus nominate this group for the ‘select domain’ niche contemplated by McIntyre et al. (2017). In those with such a depressive sub-type, the efficacy rate is high and side effects probably comparable in rates (albeit differing in nature) to more formal antidepressant drug classes. I am only aware of a few such patients who have developed dependency and/or increased their dose above the recommended upper limit. In those who have benefitted (and including some with a complete remission) tolerance has occurred in less than 10%.

I usually only prescribe such a psychostimulant after the patient has failed to respond to two antidepressants of differing classes or has only shown a partial response at best. I slightly favour methylphenidate over dexamphetamine in terms of efficacy while there are data indicating that amphetamine is twice as likely to trigger psychosis than methylphenidate (Moran et al., 2019), while I have found modafinil to rarely offer benefit. My ‘soft’ observation in relation to favouring methylphenidate is not supported by the meta-analysis of McIntyre et al. (2017) which failed to quantify methylphenidate as significant – while ar/modafinil and dextroamphetamine emerged as significant in subgroup analyses but methylphenidate was the only psychostimulant to be formally significant in the analysis by Bahji and Mesbah-Oskui (2021). I have little experience with psychostimulant monotherapy, although the meta-analysis by McIntyre et al. (2017) suggested comparability to use as an augmenting agent and so use psychostimulants principally as an antidepressant augmentor. Short-acting methylphenidate appears far more effective than long-acting preparations and this presumably reflects its pulsatile dopaminergic action, but the long-acting preparations may offer benefit to some who report anxiety or agitation with short-acting preparations. The doses of methylphenidate and dexamphetamine rarely need to exceed 30 and 20 mg, respectively. Any antidepressant onset generally occurs rapidly (from day 1 to less than a week). Two antidepressant patterns are evident in those who report benefit. In the majority, the morning and noon dosing usually acts across the full day. In a minority, the benefits last only 2–6 hours and many patients thus spread their medication out across the day, so that any return of depression occurs at the time they seek to go to sleep. For those who benefit from the psychostimulant as an ‘augmentor’, most appear to need it over time as cessation while maintaining the base antidepressant drug is seemingly commonly associated with relapse or some level of mood deterioration.

The commonest side effects appear to be anxiety or agitation. I have not observed switches to hypo/manic states in unipolar melancholic depressed patients and only a small set of such switches or emergence of mixed states in those with a bipolar II disorder. I do not prescribe psychostimulants to those with a bipolar I melancholic or psychotic depression judging the risk of switching to be too distinctive. Blood pressure and pulse rate monitoring on initiation and over time may occasionally involve changes warranting medication cessation or advice of a cardiologist if the patient wishes to continue the medication. For individuals with a cardiac condition, I seek a cardiology assessment in relation to the safety risk before initial prescribing, with cardiologists seemingly regarding few cardiac conditions as contra-indicating a trial.

For those with melancholic depression who have failed to respond to a number of antidepressant drugs, a common long-standing strategy is to trial electroconvulsive therapy (ECT) and, in recent years, to trial transcranial magnetic stimulation (TMS) and ketamine. All are appropriate options but have their individual limitations and risks. ECT is a demanding process, commonly causing great distress to those who are receiving it for the first time, while a percentage of recipients experience ongoing cognitive problems. TMS is a time demanding intervention (and often necessitating hospital admission) and whether it is superior for melancholic as against non-melancholic depression is not clearly established – as is the case for ketamine and where benefit may only be short lived.

If some 20–40% (my clinical estimate) of those with treatment-resistant melancholia respond to a psychostimulant, then should this strategy not be trialled before such alternate strategies (perhaps apart from scenarios arguing for urgent ECT administration)? The effectiveness (or otherwise) of a psychostimulant is generally evident in the first 2 weeks and thus does not risk an extended delay in commencing such treatments if this strategy fails.

I suggest that the benefits of the psychostimulants have been under-recognized as a consequence of viewing them as broad antidepressants as against having specific benefits for those with a melancholic depression, and as a consequence of perceived risks of dependence and abuse. Recent meta-analyses (noted earlier) are encouraging in suggesting their general antidepressant effectiveness, but intrinsically limited in evaluating their role across heterogeneous depressive conditions as against their application for only those with a melancholic depression, and where the clinical ‘signal’ appears more distinctive. As Shorter (1997) observed, failing to respect what McIntyre et al. (2017) described as their select clinical scenario domain demonstrated that ‘the baby was thrown out with the bathwater’ (p. 31). Risks of dependence within those with melancholia appear low, while the risk of ‘abuse’ (taking higher doses than prescribed) is lowered by the need for the clinician to obtain a formal approval to prescribe such drugs for treatment-resistant depression from their state’s regulatory authority and rigorous requirements of dispensing pharmacists (e.g. checking with the prescriber about the script for any new patient attending the pharmacy, holding the script at the prescribing pharmacy and not dispensing ahead of prescribed repeat intervals unless so approved by the dispensing medical practitioner). Thus, in those with melancholia and who have failed to respond to two or more standard antidepressants (including a dual action one and a tricyclic antidepressant) and appear to have a treatment-resistant depression, I advocate a trial of a psychostimulant in light of the favourable cost/benefit efficacy compared to many other options. We nevertheless clearly need multiple formalized trials in such clinical populations to evaluate and quantify their effectiveness.