On the Origins of MAOI Misconceptions: Reaffirming their Role in Melancholic Depression

Abstract

The first monoamine oxidase inhibitors (MAOIs) used for the treatment of depression in the 1950–60s were credited with treating severe melancholic depression (MeD) successfully and greatly reducing the need for electroconvulsive therapy (ECT). Following the hiatus caused by the then ill-understood cheese reaction, MAOI use was relegated to atypical and treatment-resistant depressions only, based on data from insufficiently probing research studies suggesting their comparatively lesser effectiveness in MeD. The siren attraction of new ‘better’ drugs with different mechanisms amplified this trend. Following a re-evaluation of the data, we suggest that MAOIs are effective in MeD. Additionally, the broad unitary conceptualisation of major depressive disorder (MDD) in the DSM model diminished the chance of demonstrating distinctive responses to different antidepressant drugs (ADs) such as SSRIs, TCAs, and MAOIs, thereby further reducing the interest in MAOIs. More reliable categorical distinction of MeD, disentangling it from MDD, may be possible if more sensitive measuring instruments (CORE, SMPI) are used. We suggest these issues will benefit from re-appraisement via an inductive reasoning process within a binary (rather than a unitary) model for defining the different depressive disorders, allowing for the use of more reliable diagnostic criteria for MeD in particular. We conclude that MAOIs remain essential for, inter alia, TCA-resistant MeD, and should typically be used prior to ECT; additionally, they have a role in maintaining remission in cases treated with ECT (and ketamine/esketamine). We suggest that MAOIs should be utilized earlier in treatment algorithms and with greater regularity than is presently the case.

MAOIs and ‘Treatment-Resistant’ Depression: Their (Under)Use in Current Practice

Introduction

The irreversible and non-selective monoamine oxidase inhibitors (MAOIs) were the first drugs used for the treatment of depression, preceding the discovery of the antidepressant activity of the first tricyclic (TCA) antidepressants (AD) in the 1950s and multiple differing classes in more recent decades. Following the introduction of the TCAs, a model developed in the United Kingdom whereby the TCAs were viewed as the preferential choice for ‘typical’ or endogenous depression (which was melancholia in the main) and the MAOIs were the preferential choice for a residual ‘atypical depressive’ group (where features such as phobic anxiety, hyperphagia and hypersomnia and mood reactivity were more prominent).¹ This latter proposed ‘atypical depression’ subtype was subsequently elevated to the status of distinct diagnostic category by North American clinicians, with the clinical features set comprising a personality style of sensitivity to rejection, experiencing hypersomnia and hyperphagia, reporting ‘leaden paralysis’ and showing a selective response to MAOIs.²

As a reflection, presumably, of the myriad empirical studies suggesting relatively comparable efficacy of all AD classes in managing major depression as quantified in meta-analyses (e.g., Cipriani et al., 2018³), and due to concerns about the need for dietary and drug restrictions (i.e., the efficacy: acceptability/tolerability ratio), the MAOIs fell into progressive disuse. They are generally reserved, at present, for use for those with ‘treatment-resistant’ depressions (TRD)—at least in theory. In practice, there are many psychiatrists who have never prescribed a classic MAOI (phenelzine, tranylcypromine, isocarboxazid),⁴ even for patients who have shown insufficient response to innumerable drug trials with modern ADs (e.g., SSRIs, SNRIs, mirtazapine, bupropion) and TCAs, or to combination and augmentation therapies.⁵

Article Scope

Whereas clinical guidelines^6,7 and meta-analyses⁸ continue to underline the effectiveness of MAOIs in various psychiatric disorders, especially in TRD, these recommendations have had little effect on clinical practice. This translational failure may be explained in part by the lingering effects of the tumultuous early history of MAOIs (incomplete knowledge of salient dietary and drug–drug interactions), and by the diminished financial incentives for pharmaceutical companies to promulgate their utility.⁹ Within this broader context, we suggest that the two principal reasons for the low, and ever-declining prescription rates of classic MAOIs are (I) persistent misconceptions regarding their safety profile, and (II) a lack of awareness of their high effectiveness for treating various depressive disorders and, in particular, melancholia or melancholic depression (MeD).

The first issue is addressed elsewhere:^10–13 dietary and drug–drug interactions are no longer mysterious; they are generally avoidable with proper attentiveness and care. This article focuses on the second issue, and explores (i) more generally, the impact of structural flaws in the dominant DSM model for defining and categorizing depression (sub)types, which complicate clarifications on the effectiveness of any AD treatment, with specific attention paid to MeD, and (ii) more specifically, whether MAOIs have a continuing or even superior role in alleviating MeD.

A. (Melancholic) Depression (mis)conceptualized: A Modern State of Affairs

Principles of Diagnostic Categorization: Preliminary Considerations

Psychopharmacology is not an abstract science. Theoretical models that catalogue various (subtypes of) disorders are best construed as working models that are subject to refinement on the basis of rigorously verified clinical research and experience. It follows that a key function of any diagnostic categorization is to circumscribe a condition by its clinical features so as to allow its determinants, natural history, and response to candidate treatments to be evaluated and compared. Therefore, the subtyping of depressive disorders should be structured in a way that minimizes category overlap, so that research seeking to determine differential treatment responses can be reliably undertaken. In other words, if different symptomatic presentations of depression respond preferentially or exclusively to a specific AD class, this is a strong indication of distinctly differing underlying pathophysiological changes. To illustrate this principle, this article highlights the defining criteria of MeD, which has a biological (endogenous) basis and is characterized by the presence of psychomotor disturbances (generally retardation, sometimes agitation, sometimes both) and certain specific symptoms (e.g., an anhedonic and non-reactive mood, anergia, foggy and compromised concentration). MeD (specifically the TCA-resistant cases) has been found to respond preferentially to ECT¹² and MAOIs¹⁴ over other AD treatments. These findings may be construed as providing external validity to MeD as a fundamentally distinct subtype of depression by causally linking the triple-action profile of MAOIs, unique in their dopamine-elevating properties, with the putative pathophysiology and symptom profile of MeD (as buttressed by Martinot et al. (2001), whose research on left caudate dopamine function provided ‘direct evidence of a link between dopamine hypofunction and psychomotor retardation in depression’¹⁵). To clarify the presumable mechanism, a parallel may be drawn between the resolution of psychomotor disturbances in MeD and a prominent example from the neurological sphere, namely the exclusive response of ‘Parkinsonian bradyphrenia and bradykinesia’¹⁶ to dopamine-enhancing treatments (versus the non-response to drugs that affect other monoaminergic systems, such as serotonin).

The DSM Approach: Past and Present Flaws of a ‘Unitary’ Depression Model

The dominant model over the last forty years for conceptualizing and classifying the depressive disorders is a dimensional model weighting severity, with DSM-III introducing a diagnostic category of major depressive disorder or MDD (along with other ‘minor’ depressive conditions), and ICD-10 classifying depressive disorders as either ‘mild,’ ‘moderate’ or ‘severe.’¹⁷ The diagnosis of MDD rapidly achieved cachet value among clinicians and researchers, and has almost universally become the current diagnostic criterion for evaluating drug and non-drug therapies for depression, whether in single studies or meta-analyses. From this, the risk emerges that MDD is perceived as an entity, albeit with some authors (e.g., Kendler¹⁸) challenging that reification. As reviewed elsewhere,¹⁹ several post-implementation studies quantified poor inter-rater reliability of MDD diagnoses, challenging the claim that DSM-III had resolved the reliability problem. It is clear that the definition of MDD fails to meet orthodox criteria for validity²⁰ in lacking a clear-cut clinical picture (with a depressed mood and a loss of interest and pleasure being the only obligatory requirements for the current DSM-5 definition), and with its intrinsic heterogeneity making determination of natural history, underlying neurobiology, and response to treatment problematic. Thus, large meta-analyses generally have difficulty to demonstrate any differences across multiple drug and non-drug treatments for MDD (e.g., Casacalenda et al., 2002;²¹ Spielmans et al., 2011²²). Meta-analyses of AD comparisons and placebo-controlled AD trials trend towards low-moderate effect sizes for the drugs evaluated (e.g., Kirsch et al., 2008²³; Cipriani et al., 2018³; Parker et al., 2020²⁴). The latter likely is the consequence of significant within-group heterogeneity in the RCTs included, where different subtypes of MDD (like MD) might have different response rates relative to placebo. However, these subtypes have never been identified nor investigated separately in meta-analyses. This DSM-based nosological approach thus violates the previously outlined key principle of diagnostic categorization, in that a non-specific (unitary) model of depression ensures non-specific results with regard to any comparative evaluation of differing treatments.

Such an outcome is considered by some to be a stultified and self-referential system, wherein implausible conclusions²⁵ regarding the supposed near-equal efficacy of all AD classes in virtually all depression cases, are made to appear plausible—or even expected.^26,27 To quantify the approximate effectiveness of MAOI treatment in MDD, it is therefore important to consider the plethora of earlier research results, beginning from the 1950s and ‘60s, whilst being mindful of the wavering definitions of the various depressive disorders. In examining the continuing role of MAOIs in MeD, we may well ask ourselves, as the late Merton Sandler (co-originator of the monoamine hypothesis of depression with Michael Pare)²⁸ did in 1997: ‘Why did the three major breakthroughs in drug treatment all occur in the 1950s, and then nothing since then?’ As clarified by Shorter, it is likely that Sandler referred to chlorpromazine (antipsychotic), imipramine (TCA), and iproniazid (MAOI)—although we would add that lithium augmentation treatment is likewise a noteworthy advance. Shorter has opined²⁹ that the answer to Sandler’s question might be that ‘[t]hese breakthroughs occurred precisely because FDA-style trials using DSM diagnoses were not in force in the 1950s.’

Malady Analogy: Major Depression and Major Breathlessness

In essence, MDD is as informative a diagnosis as one of ‘major breathlessness.’ A patient receiving such a latter diagnosis would feel bewildered, at loss as to whether they had asthma, pneumonia, a pulmonary embolism, or some other condition. Knowing the integral condition might then dictate the clinician’s choice (in this instance) of a bronchodilator, an antibiotic, an anticoagulant, or some other specific treatment for the specific condition respectively.

Further, imagine that a new drug under development with high efficacy for managing asthma was evaluated in those with ‘major breathlessness’ (but not in those with asthma alone), but only a small minority in the overall ‘entity’ group actually had asthma. The drug would then appear ineffective (be it in single clinical studies or meta-analyses). Thus, the non-specificity of MDD disallows any condition-specific AD treatments being demonstrated as effective or superior to other options, leading to minimal separation between different ADs in the now numerous meta-analyses.

Admittedly, the lack at present of identifiable biological substrates to definitively recategorize the multitude of different depressive disorders encapsulated in the unifying MDD term poses considerable difficulties for any attempt at definitional remodelling. Further advancements in precision psychiatry (e.g., following the R-DOC approach)³⁰ will allow for further evidence-based differentiation. These reservations notwithstanding, it is sufficiently clear at present that the circular pattern of reasoning in the DSM model, from definitional overlap to undifferentiated treatment results back to definitional overlap—the theoretical validity of which is now considered buttressed by experiment—is especially evident in the DSM definitions of MDD with and without melancholic features.

Diagnostic Criteria For Melancholic Depression (MeD): DSM vs. SMPI Model

The DSM-5 diagnostic criteria for MDD with melancholic features³¹ (MeD) requires the presence of either significant anhedonia (which is notoriously difficult to reliably assess)³² or the absence of significant mood reactivity, with the ‘significant’ descriptor allowing considerable subjective variation. Additionally, the presence of at least three of the following symptoms is required: pronounced despair, diurnal mood variations (worse in mornings), early-morning wakening, psychomotor disturbances, weight loss and excessive guilt. As some are also criteria for major depression (i.e., weight loss, sleep disturbance, psychomotor disturbance, excessive guilt) there is no clear boundary between MDD and MeD, ensuring an imprecise definition of MeD.

Thus, the considerable overlap between the definitions of (a) MDD and (b) MeD in the DSM-model does not allow for adequate differentiation between the two³³ (as noted, for instance, in a recent study by Tondo et al. (2020), which used the DSM-5 definitions and found few symptomatic differences in melancholic versus non-melancholic subjects),³⁴ and thus complicates the matter of establishing differential treatment responses for MeD and non-MeD. Criticisms on the operationalization criteria of MeD centre around (a) the non-specificity of anhedonia and lack of mood reactivity, which are also commonly observed in other MDD cases, and (b) the required presence of diurnal mood variations, which may vanish in notably severe MeD cases, where a consistently despondent mood may be present.³⁵

Additionally, the DSM-5 criteria for MeD notably differ from other measures of melancholia, such as the Newcastle Index,³⁶ the CORE-based assessment of psychomotor disturbance,³⁷ and most prominently the clinician rating scale presented in the Sydney Melancholia Prototype Index (SMPI).^38,39 The SMPI addressed the limitations of the sign-based CORE system by acknowledging (a) the difficulties in assessing psychomotor disturbances in younger melancholic patients (in whom psychomotor disturbance is often less overtly present), and (b) the need to assess symptoms ‘at or near the nadir’ of a patient’s depressive episode.²⁰ Indeed, historically, melancholia was viewed primarily as a movement disorder,³⁵ reflecting the key feature of psychomotor disturbances^40,41 (retardation or agitation or both, along with cognitive impairment described as foggy thinking⁴²) as the defining criteria.⁴³ These criteria are comprehensively evaluated in the SMPI—it has 24 items assessing clinical symptoms and illness correlates, with a pre-established cut-off score accurately assigning 98% of those with MeD and 97% of those with non-MeD;⁴⁴ future research may be aimed at identifying neurobiological markers as a third reference point (in addition to the present diagnostic approach utilized by the SMPI, which focuses on the identification of MeD-specific symptoms and illness-associated factors)³⁹ to further bolster the reliability of the SMPI.

Reliable categorical distinction of MeD becomes possible when these more sensitive measuring instruments (CORE, SMPI) are used. In one application study, contrastingly low placebo response rates were found—10% in MeD vs. 40–60% in MDD—as well as significantly superior responses to physical vs. psychotherapy treatments (with SSRI<TCA⩽MAOI efficacy).^25,35 The presumable endogeneity of the disorder is underlined by the increased incidence of depressive disorders in close relatives of MeD patients when compared to non-MeD patients.³³ As MeD is typically severe and presents with a qualitatively different¹⁶ (‘morbid’) weighting to the depressed mood,⁴⁵ it is best conceptualized as both categorically and (most commonly) dimensionally distinct from non-MeD.⁴⁶ Thus, the impetus gathers towards (a return to) a ‘binary’ depression model which differentiated MeD from a residual set of non-melancholic depressive disorders.

Melancholia and the Appeal for a Return of the ‘Binary’ Depression Model

The DSM conception of MDD as an entity—albeit one that is attended by subcategory specifiers³¹—remains a topic of considerable contention.⁴⁷ As previously explicated, the most prominent challenge to this unitary model of depression,⁴⁸ comes from authors who emphasize that depression is not a homogeneous entity, and that the DSM model conflates the different clinical manifestations of different depressive disorders under the umbrella term of MDD. This is not, in essence, a new assertion. Prior to DSM-III (1980), and for many centuries (with references going back to Hippocrates and to St. Paul in 2 Corinthians), a binary model was dominant. Two principal types of depressive disorders were postulated:⁴⁹ (I) endogenous/endogenomorphic/type A/vital/melancholic depression, and (II) neurotic/exogenous/reactive/non-melancholic depression.

Despite the dominance of the DSM-model, the case for viewing MeD as a categorical and separate depressive subtype has been strengthened in this era of biological psychiatry, and has garnered substantial interest from various international authors,⁵⁰ with research findings indicating a strong genetic contribution and intrinsic biological perturbations such as compromised HPA axis functioning,⁵¹ as well as a low placebo response rate and a superior response to ADs than to psychotherapies.⁵² ‘Other’ depressions may in turn be conceptualized as being more heterogeneous in ‘type’, and as constituting a residual set of non-melancholic depressive conditions⁵³ that reflect the causal and differential impact of stressors as well as variable predisposing personality styles. Within this set there are some ‘conditions’ (e.g., anxious worrying depression) where the sufferer might respond well to the serenic effects of a selective serotonin reuptake inhibitor (SSRI); other conditions such as depression underpinned by the individual being highly introverted, so that cognitive behavioural therapy (CBT) might be preferential; or ‘hostile depressions’, where no AD (or any other drug class) has ever been demonstrated to be effective (unless causing sedation) and where anger management might be the cogent intervention.

When a binary subtyping model is used to evaluate treatment response, treatment gradients are often evident. For instance: a trial of AD versus cognitive behavioural therapy (CBT)⁵⁴ required only 39 patients to demonstrate the statistical superiority of the antidepressant over CBT in MeD patients in a 12-week study period. Such gradients in treatment effectiveness stand in sharp contrast to studies in MDD subjects, where differential response rates to differing therapies are minor or non-existent. Therefore, the return to a more high-resolution working model and subtyping of depression is vitally important, and may be achieved by aligning theory and practice via an inductive reasoning process within a binary model; this implies an increased focus on differential treatment responses, and a critical reassessment of concept validity on the basis thereof. That way, theory informs practice, and practice informs theory; this recoupling is essential. In challenging the DSM model of MeD as a MDD specifier rather than as a categorically distinct depressive disorder, we may do well to remind ourselves that it does not matter how beautiful your theory is. It does not matter how smart you are or what your name is. If your theory does not agree with experiment, it’s wrong. ∼ Richard P. Feynman.

B. The Case for MAOIs in Melancholia: The Evidence Reassessed

A Foray into MAOI History: Their Serendipitous Discovery

The history of MAOIs is fraught with irregularities that elude the purview of pure scientific reasoning, and enter into the realm of power and politics. The complete unfoldings are described by Shorter,²⁹ and are now summarised.

The discovery of the psychic energizing effects of MAOIs was a serendipitous occurrence.⁵⁵ In the early 1950s, word began to spread that the tuberculosis drug iproniazid, a hydrazine-derivative that was found to have MAO-inhibiting properties,⁵⁶ elicited ‘Mona Lisa smiles’ in otherwise gravely ill patients.⁵⁷ When a 1957 trial affirmed these mood-enhancing effects in MeD subjects, the general sentiment was enthusiasm; finally, there was a pharmacological treatment for endogenous depression⁵⁸ that could—so optimistic were the initial observations—rival the effectiveness of ECT. Robie (1958) reported remission rates of ∼61% (53 out of 87 cases) with the use of iproniazid chemotherapy in melancholia.⁵⁹ These hopes were soon tempered:⁶⁰ although there are reports of patients responding to MAOIs after failed ECT treatment,^61,62 ECT is still considered the most efficacious treatment for MeD, albeit not generally as a first-line option in light of the risk of cognitive side-effects.^63,64 Still, the more amenable side-effect profile of MAOIs secures them a place, typically prior to ECT, in the treatment algorithm for managing MeD⁶⁵ (and MDD).

A Further Foray into MAOI History: Their Disavowal

Not long thereafter, from the early 1960s onward, ‘the familiar passage from unbridled enthusiasm to increasing scepticism’ (Hollister)⁶⁶ began. Clinicians grew increasingly disenchanted with MAOIs, and the collective interest shifted to TCAs. We identify here six key reasons for this occurrence:

• (I) The hepatotoxic effects of iproniazid, both at the high doses required to treat tuberculosis, and at the lower doses used in depressed patients.^67,68 In 1961, the FDA ordered iproniazid withdrawn, and it was made available only as an ‘Investigational New Drug’. Despite its irremediable shortcomings, it had served its purpose as a kind of psychopharmacologic torch, ushering in the brief age of MAOIs.

• (II) The adoption of the ‘atypical’ moniker to denote the set of reactive, non-melancholic depressive disorders that was soon believed to respond preferentially to MAOI treatment.⁶⁹ Patients exhibiting these symptoms were described by West and Dally (1959) as ‘showing somewhat atypical depressive states, sometimes resembling anxiety hysteria with secondary depression’, with the accompanying statement that ‘iproniazid was found to have a less rapid and less complete effect in some typical endogenous depressive illnesses’.¹

• (III) The surge of the TCAs. Whereas iproniazid was described by Cole et al. (1959) as being ‘of value in both the neurotic and the psychotic depressions’,⁶⁸ Delay and Deniker asserted in the same year the relative superiority of imipramine (TCA) over iproniazid for treating ‘endogenous depression or melancholias’.⁷⁰ Thus, as overviewed earlier, the dominant paradigm was formed—albeit on the basis of studies characterized by low patient sample sizes and questionable methodological stringency—and further shaped throughout the 1960s, whereby ‘typical depression’ was generally seen as a synonym for MeD, and ‘atypical depression’ was ascribed to those patients suffering residual disorders (but principally capturing those with anxiety-based depressive conditions), with TCAs favoured for the former subset⁷¹ and MAOIs for the latter—those being the only two AD classes at the time.

• (IV) The potential threat of ‘hypertensive crises’, or rather the perception thereof, due to interactions with dietary tyramine. The cause-effect relationship of this ‘cheese effect’ was uncovered by Blackwell and Mabbitt in 1965,⁷² and removed the element of unpredictability: the seemingly random hypertensive effects were not at all random, but preventable (and certainly manageable). Regrettably, this revelation came too late and did not stem the tide of anti-MAOI sentiments.

• (V) The publication of several research papers with methodological flaws in the early years of MAOI use (e.g., Overall, 1962).⁷³ The most obvious example would be the 1965 Report to the Medical Research Council,⁷⁴ which utilized too low dosages of phenelzine (⩽60 mg/day)^8,27 and included psychotic MDD-patients.⁵⁷ The results, however, had a clear negative influence on the popularity of MAOIs, certainly in the treatment of MeD, by assessing phenelzine response rates as equal or inferior to placebo.⁷⁴ Thereafter, the reputation of MAOIs declined to such an extent that Tyrer (1973) posed the question ‘Are monoamine-oxidase inhibitors antidepressants?’⁷⁵ His own answer: while there is merit to the use of phenelzine in phobic anxiety states,⁷⁶ there is ‘abundant evidence that MAOIs are ineffective in severe depressive illness.’⁷⁵

• (VI) The FDA-organized restructuring of the pharmacopoeia.²⁹ In 1966, the United States Food and Drug Administration or FDA required from manufacturers evidence of drug efficacy in the form of randomly controlled trials, which did not then exist for most drugs. In the FDA’s Drug Efficacy Study (DES), panels then offered opinions about whether various agents should be withdrawn (an interesting, albeit not uncriticized way of approaching the entire pharmacopoeia). Of the MAOIs then on the market, two did not make it through this regulatory Star Chamber, and three got warnings that would have scared many prescribers away (e.g., the ‘cheese effect’ warning for phenelzine). At this point, a further decline began. The MAOIs staggered away from the DES with a tarnished reputation, and would easily be relegated to secondary status by the TCAs, and buried by the SSRIs.

The Weight of MAOI History on Modern-Day Misbeliefs about Melancholia

Much of the current literature resorts to the uncritical repetition of the merits of MAOIs in treating ‘atypical’ and ‘treatment-resistant’ depressions—the implication being that they are not suited for the treatment of ‘non-atypical’ depressions, e.g., MeD.⁷⁷ This is in strong contrast to the early findings (1950s), when there was little doubt regarding the effectiveness of iproniazid in treating catatonia, melancholia and psychotic depressions.²⁹ Tranylcypromine underwent a similar fate. Shorter (p. 148) recounts: ‘Today the FDA insists that [tranylcypromine] is not to be used for melancholia, whereas historically the NAS/NRC psychiatry panel thought it was “probably” useful precisely for “severe” depressions that did not respond to ECT.’ He adds (p. 195): ‘The consequences for American psychiatry, psychopharmacology, and patient treatment have been disastrous, as ineffective drugs are offered for an undifferentiated DSM diagnosis of depression that does not exist in nature.’²⁹

There was a brief resurgence of interest in MAOI use in MeD during the 1980s: Himmelhoch et al. (1982) reported on the superiority of tranylcypromine over placebo following a double-blind study in predominantly ‘anergic, bipolar, endogenously depressed populations’.⁷⁸ Similarly, Nolen et al. (1988) established the efficacy of tranylcypromine in TCA-resistant, endogenously depressed subjects (by Newcastle Scale II criteria).⁷⁹

Other prominent studies from the same time period are more difficult to interpret. For example, Ravaris et al. (1980)⁸⁰ directly compared phenelzine (MAOI) to amitriptyline (TCA) in a double-blind controlled trial comparing 105 subjects with undifferentiated depression. Both ADs were found to be similarly effective, albeit with superior outcomes for amitriptyline in alleviating symptoms of early-morning awakening and weight loss (endogenous profiles), whereas ‘atypical’ profiles responded preferentially to phenelzine. But there is a significant encumbrance in granting the validity of these and similar data; it pertains to the fact that MAOIs are rarely used in appropriate (i.e., sufficiently high) doses: for example, the maximal dose used in this study by Ravaris et al. (1980) consisted of only 60 mg/day of phenelzine⁸⁰ (see also next subtitle).

Thus, the assumption of the relative inefficacy of MAOIs in MeD was perpetuated on account of methodological oversights centred around an apparent physician-side reticence to prescribe sufficiently high MAOI doses. This is problematic: whether one considers MeD a distinct depressive disorder (which is argued in this paper) or merely a gradational specifier within a unitary MDD model, the view that robust AD effects require a correspondingly robust AD dose is uncontroversial. For some ADs, dose determination may be aided by readily assessable objective metrics (e.g., plasma test for nortriptyline, imipramine, amitriptyline).⁸¹ For other ADs, including the MAOIs, such diagnostic tests may require further investigation to assess specificity and sensitivity; moreover, clinical application may be considered impractical for reasons of cost (e.g., platelet MAO-inhibition test) or for reasons that remain unclear (e.g., urinary tryptamine excretion—a method frequently and reliably used by Bieck et al. to gauge MAO-inhibition levels throughout various studies in the 1980s,^82–84 but has since fallen out of fashion). For the purpose of this paper, the discussion is contained to the measurement of platelet MAO-inhibition levels, as the data are more robust for this test.

Platelet MAO-Inhibition Levels and the Importance of Dose in MAOI Treatment

Lower MAOI doses (e.g., 60 mg/day of phenelzine, as used in the Ravaris et al. (1980) study discussed earlier)⁸⁰ are not enough for many (non-geriatric)⁸⁵ patients to achieve the level of platelet MAO-inhibition that is needed for (or at the least correlates strongly with) a significant therapeutic response:⁸ Davidson et al. (1978) postulated that a minimum of 80% platelet MAO-inhibition might be required.⁸⁶ This assertion is buttressed by the earlier findings of both Ravaris et al. (1976), who described a non-superior response to phenelzine vs. placebo at 60% platelet MAO-inhibition levels,⁸⁷ and by Robinson et al. (1978), who performed an ‘analysis of percent platelet MAO inhibition as a function of phenelzine dose (mg/kg)’, and found ‘significant positive correlations at two, four, and six weeks of treatment (r = .29, .56, and .42 respectively)’.⁸⁸ Raft et al. (1982) added findings of a clinical trial in which ‘a linear relationship between the level of platelet MAO-inhibition and clinical improvement’ was established, the effect of ‘which was greatest if MAO-inhibition exceeded 80%’.⁸⁹ In fact, phenelzine (titrated up to 90 mg/day in some subjects) only became superior to placebo in week 4 of their trial (at which point average platelet MAO-inhibition was 85%); by week 5 (when MAO-inhibition exceeded 90%), phenelzine proved superior to amitriptyline. Thus, a generalised inference may be drawn from the platelet MAO-I trials to inform clinical practice, indicating the likelihood of greater therapeutic response with increased MAOI dose and treatment duration. (With the caveat that the relationship between peripheral and central inhibition with different MAOI drugs, e.g., phenelzine, tranylcypromine, is not established with any certainty.)

When gauging the effectiveness of MAOIs in MeD, the use of a sufficiently high dose becomes an important consideration. For instance: in 1982, Hamilton reported on an open, randomized trial with 130 melancholic patients receiving either imipramine or phenelzine at appropriate doses (⩽90 mg/day for phenelzine);⁴⁵ there was no significant difference between the two drugs, with both being inferior to ECT—indicating no clear gap in effectiveness between TCAs and MAOIs for treating MeD when a high MAOI dose is used. Vallejo et al. (1987) concluded similarly, following a double-blind comparison study between imipramine and phenelzine (75 mg/day) in melancholic patients that ‘Phenelzine should be considered as a good alternative to tricyclic antidepressants in cases when these drugs are contraindicated, cause major side-effects, or prove ineffective.’⁹⁰ Davidson et al. (1984) affirmed the effectiveness of the third remaining classic MAOI, isocarboxazid, in MeD,⁹¹ and drew specific attention to the importance of administering a sufficiently high dose: 30 mg/day of isocarboxazid proved effective for non-melancholic depressions, whereas a dose of 50 mg/day was required to treat melancholia cases effectively.⁹²

In line with the above, we conclude that the current evidence indicates that the TCAs with the most pronounced AD effects (e.g., imipramine, clomipramine) and the classic MAOIs appear roughly equipotent—but not necessarily interchangeable—in the treatment of MeD. Factoring in the dietary and drug-drug interactions, a reasonable conclusion might then be, in echoing Quitkin et al. (1979),⁹³ Vallejo et al. (1987),⁹⁰ and Thase et al. (1995),⁹⁴ that—within the realm of MeD treatment options—MAOIs are primarily indicated for the treatment of TCA-resistant MeD.⁹⁵

MAOIs in Melancholia: A Present-Day Perspective

There is a paucity of recent research quantifying the effectiveness of MAOIs in MeD (and MDD), so reassessing older data with enhanced knowledge of the underlying mechanisms is the most sensible approach to inform clinical practice. As discussed, the overlap between the DSM definitions of MDD and MeD continues to sow methodological confusion, with poor inter-rater reliability further complicating the interpretation of study results—e.g., McGrath et al. (1984) noted the effectiveness of tranylcypromine in treating MeD and melancholia in bipolar disorders as defined by the DSM-III criteria—but it is conceivable that the non-specificity of the diagnostic criteria may have resulted in uncharacteristically high placebo response rates for those with a melancholic disorder (25%),⁹⁶ thereby limiting the generalizability of these results to MeD as defined by CORE- and SMPI-criteria. Similarly, a later study by McGrath et al. (1986)⁹⁷ confirmed the effectiveness of phenelzine in MeD, albeit with comparable encumbrances to interpretability and generalizability (consequent to the non-specificity of the diagnostic criteria for MeD).

Regarding the general therapeutic effect of MAOIs, pharmacological insights have equipped us with the tools to plausibly infer mechanistic correspondence between the ‘triple-action’ monoaminergic effects (serotonin, dopamine, norepinephrine) and the alleviation of depressive symptoms, whereby—as formulated in the Bradford Hill criteria for causation⁹⁸ and further concretized in Pearl’s causation theory⁹⁹—the temporal proximity of these events (within 1–2 months of treatment initiation) may serve to buttress the notion of a causal link. In relation more specifically to MeD, the dopamine-elevating properties of MAOIs are of particular importance (see also title A). The consideration that ‘[dopamine] is a credible aetiological candidate for the [psychomotor disturbance] in melancholic depression’¹⁶ has long been asserted (though is difficult to validate empirically), and appears to align with decades of clinical experience, thereby presenting a credible joinder of theory and practice. This is a prime illustration of why the collective merit of reliably reproduced observations remains a valuable source of evidence, and must be considered alongside the evidence produced by randomized controlled trials—indeed: viewed separately, each case study is an anecdote, but none is anecdotal¹⁰⁰ when the weight of evidence is assessed as a whole. Or, in the words of Blackwell and Taylor (1967: ‘[C]hance observation, widely felt need, and plausible biochemical rationale were sufficient to ensure easy acceptance of the MAOI without the added impetus provided by uncritical advertising and uncontrolled clinical trials.’⁶⁰

In addition to the dopaminergic effects of MAOIs—which do not notably occur with TCA treatment—the additional pharmacological effects common to both MAOIs and TCAs may be briefly discussed to further elucidate their effectiveness for treating MeD. This brings us to the hypothesis formulated by Gold and Wong (2021) study: their research indicates that melancholic patients present with elevated plasma and CSF levels of norepinephrine, which the authors link to the clinical picture of MeD as ‘a state of hyperarousal and a tortured sense of worthlessness’ that may be due to norepinephrine hypersecretion.¹⁰¹ They further posit that the attenuation of these classic melancholic symptoms by TCAs, MAOIs, and ECT, may be causally connected with the effect these treatments have on reducing the ‘basal and sensory-evoked firing rates of locus caeruleus neurons’;¹⁰¹ whereas this effect was also observed with SSRIs (fluoxetine and sertraline), the effect was greater with the TCAs (desipramine and imipramine) and with the MAOI phenelzine.¹⁰²

Summary Conclusion: MAOIs Remain Vital in the Treatment of Melancholia

Throughout the decades since their inception, the general—and initially highly positive—sentiments surrounding MAOI use have been influenced by methodologically impure research with inadequate dosing^74,80 (e.g., the STAR*D trial,¹⁰³ which used a low mean daily dose of 37 mg tranylcypromine and considered it rather ineffective as a result)¹⁰⁴ and by the success of marketing strategies for the selective-action ADs.⁹ The suffusion of erroneous beliefs regarding the limited effectiveness and the cumbersome side-effect profile of MAOIs jointly paved the way for their abandonment in favour of the newer and industry-backed SSRIs and SNRIs—a fate that the TCAs largely shared.¹⁰⁵ We are confronted with a relative scarcity of literature data on the use of MAOIs in MeD, with much of the most compelling primary source material stemming from the 1950s–‘80s.

Nevertheless, the current state of the research is sufficiently developed to allow for the statement that MAOIs have a clear and continuing role in treating SSRI- and TCA-resistant melancholic depression. It may be noted that such ostensibly ‘treatment-resistant’ depressive conditions automatically enter the purview of MAOI treatment, regardless of their subtyping as melancholic or non-melancholic.¹³

Conclusion

This review article examined the diagnostic validity of MeD as defined by DSM-criteria and by other criteria, partly based on preferential or exclusive treatment responses to ADs of the (TCA-) and MAOI-class(es). The first section (A) scrutinized the evidence for and against the presently dominant DSM-5 definition of MDD with melancholic features by assessing the aetiological and predictive value of this definition, and found it insufficiently precise due to category overlap with other MDD episode specifiers and with MDD itself (thus failing to provide a boundary between melancholic and non-melancholic depression). By contrast, the SMPI-criteria may present a more sensitive diagnostic instrument for identifying MeD as a categorically distinct depressive disorder with a low placebo response rate and superior outcomes for AD treatment (with inter-class differences: SSRI<TCA⩽MAOI) over CBT. Cognizant of the interpretational difficulties introduced by the plurality of definitions, the second section (B) then reassessed the data on the effectiveness of MAOIs in MeD, with the conclusion that MAOIs remain essential for use in at least TCA-resistant melancholia. Although it is unclear at present how intravenous ketamine and intranasal esketamine compare, as it stands, MAOIs may well be considered the second-to-last line of defence (before ECT) in difficult-to-treat melancholia cases; they may also have a role in maintaining remission in cases treated with ECT. We conclude that MAOIs should be utilized with far greater regularity than is the case at present.