MAOI Antidepressants: Independent Evidence, Diet & Interactions

Table of contents

• Evidence summary
• What MAOI antidepressants are
• How they work
• What they are prescribed for
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions
• Who should avoid MAOIs
• Dosage and how they are taken
• Follow the money: who makes them and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

Independent evidence and credibility scorecard

Independence tiers: 1 = regulator or government body; 2 = academic or Cochrane team with no declared commercial funding for the work; 3 = independent authors with declared industry ties, or work with an author employed by or consulting for a maker; 4 = manufacturer-funded or manufacturer-authored. Credibility runs from A (highest) to D. Most sources are from the US, UK and Western Europe; we found little evidence from Asia, Africa or Latin America beyond a Brazilian supply letter.

What MAOI antidepressants are

Monoamine oxidase inhibitors block an enzyme (monoamine oxidase, MAO) that breaks down serotonin, noradrenaline, dopamine and dietary amines such as tyramine. The antidepressant members of the class differ in two ways that matter for safety: whether they block the enzyme irreversibly (the body must make new enzyme, which takes days to weeks) and whether they block both forms, MAO-A and MAO-B.

Oral selegiline and rasagiline, used for Parkinson's disease, are also MAO inhibitors but are not licensed as antidepressants and are outside the scope of this review. Linezolid (an antibiotic) and methylene blue (a medicine for methaemoglobinaemia) are not antidepressants but inhibit MAO-A, which is why they appear in the interaction section.

How they work

Labels are frank that the antidepressant mechanism is not fully understood. The Parnate label says it is "presumed to be linked" to boosting monoamine activity through irreversible MAO inhibition (FDA Parnate label 12.1). The Nardil label adds that it is unknown whether MAO inhibition itself, other actions, or both produce the clinical effect (FDA Nardil label).

What is well established in humans is the pharmacology behind the risks:

• Long-lasting effect: tranylcypromine is cleared within about 24 hours, but MAO activity takes 3 to 5 days to recover and inhibition "may persist up to 10 days" (FDA Parnate label 5.9, 12.2). This is why washout periods are needed after stopping.
• Gut enzyme and tyramine: MAO in the gut is mainly type A and normally destroys tyramine in food. When it is blocked, tyramine can reach the blood and release noradrenaline from nerve endings, raising blood pressure: the so-called "cheese reaction" (FDA Emsam label 12.2).
• Why the patch and moclobemide differ: the patch delivers selegiline through the skin rather than the gut, and the label's tyramine studies show far less potentiation at the 6 mg dose than with oral tranylcypromine (see Risks). Moclobemide binds reversibly, so tyramine can displace it; the UK SmPC says its propensity to interact with tyramine is "slight and short-lasting" (UK Manerix SmPC 4.5).

What they are prescribed for

Licensed uses

What works and what does not

Benefits by claim

Do MAOIs work for depression?

Yes, on independent analyses. The largest network meta-analysis to date pooled 83 double-blind randomised trials and found that isocarboxazid, phenelzine, tranylcypromine and moclobemide all showed "significantly higher efficacy" than placebo. Tolerability and acceptability were comparable to other antidepressants (Giménez-Palomo et al. 2024). A separate 2021 network meta-analysis of 52 trials and 14 drugs found phenelzine had the strongest efficacy evidence of all the drugs compared (Suchting et al. 2021).

Individual licensing trials were small. Isocarboxazid's US approval rests on two 6-week placebo-controlled studies; systematically collected safety data covered only 86 patients (FDA Marplan label). For the Emsam patch, a 6-week trial (n=176) showed a 2.5-point greater improvement than placebo on the 17-item Hamilton scale (95% CI −4.6 to −0.4), and an 8-week trial (n=265) showed a 2.4-point difference on the 28-item scale (−4.5 to −0.3) (FDA Emsam label). In plain terms, that is a real but modest average gain, similar in size to many antidepressants. Both trials were run for the product's developer.

Atypical depression

"Atypical" depression here means a specific pattern: mood that still brightens with good news, plus features such as oversleeping, heavy limbs or overeating. This is where MAOIs have their best-known niche. A 2006 meta-analysis found MAOIs beat tricyclic antidepressants (effect size 0.27, 95% CI 0.16 to 0.42) but were no different from SSRIs (0.02, −0.10 to 0.14). Its authors said the data were "insufficient" for a direct MAOI-versus-SSRI verdict (Henkel et al. 2006).

The atypical-depression network meta-analysis screened 21 eligible trials (20 entered the network). On response, phenelzine (RR 2.58), moclobemide (RR 2.16) and imipramine (RR 1.76) beat placebo. But no intervention outperformed placebo once high-risk-of-bias trials were removed, and the overall confidence was rated low or very low (Fornaro et al. 2025). Our reading: MAOIs, especially phenelzine, are a reasonable option for atypical depression, but the claim that they are clearly superior to SSRIs is not proven.

Treatment-resistant depression

MAOIs are often described as a last resort for depression that has resisted several treatments. The best independent test is STAR*D level 4. People who had not remitted after three prior treatment steps were randomised to tranylcypromine (n=58) or venlafaxine plus mirtazapine (n=51). Remission was 6.9% versus 13.7%, not a statistically significant difference. Tranylcypromine produced less symptom reduction and more dropouts due to side effects. The authors concluded the combination "may be preferred" because of fewer side effects, no diet and ease of use (McGrath et al. 2006).

Critics note that the mean tranylcypromine dose was only 36.9 mg a day, well below the US maximum of 60 mg (FDA Parnate label), and that the trial was open-label and small. The 2017 tranylcypromine meta-analysis (Ricken et al.) is one of the few reviews focused on the drug; we re-checked its abstract, which reports that tranylcypromine beat placebo in four controlled studies and matched other antidepressants in ten; we did not read the full text. We could not verify the authors' affiliations or conflicts. The 2024 network meta-analysis noted "a lack of specific studies" in treatment-resistant depression (Giménez-Palomo 2024). The honest summary: MAOIs remain a reasonable specialist option after other drugs fail, but the evidence that they rescue people other drugs cannot is thin.

Social anxiety disorder

The 2017 Cochrane review found benefit on treatment response for MAOIs (RR 2.36, 95% CI 1.48 to 3.75; 4 trials, 235 people) and for RIMAs such as moclobemide (RR 1.83, 1.32 to 2.55; 8 trials, 1,270 people), both low-quality evidence. In an NIH-funded trial of 166 randomised adults (128 in the intention-to-treat analysis), 12-week response rates (clinician-rated improvement scale) were 33.3% on placebo, 47.1% on group CBT, 54.3% on phenelzine and 71.9% on phenelzine plus group CBT (Blanco et al. 2010).

Moclobemide: the gentler MAOI

A 1999 meta-analysis found moclobemide somewhat less effective, albeit better tolerated, than older MAOIs (Lotufo-Neto et al. 1999). That review was partly supported by a US NIMH grant listed in Europe PMC (the grant number was not confirmed). In the 2024 network meta-analysis, moclobemide was also among the drugs that beat placebo (Giménez-Palomo 2024).

Are MAOIs underused?

Many reviewers say so. An Ontario cohort found the yearly rate of new irreversible MAOI prescriptions in older adults fell from 3.1 to 1.4 per 100,000 over a decade. 18.1% of users were also exposed to a serotonergic drug (Shulman et al. 2009; funded by the Canadian Institutes of Health Research). Underuse arguments from one of the tyramine-diet authors are noted below, and that author has a declared company tie.

Risks and all side effects

The tyramine question: how strict does the diet need to be?

The labels still list foods to avoid. For tranylcypromine the avoid column includes aged cheeses and tap or unpasteurised beer. The acceptable column lists processed cheeses, mozzarella, ricotta, cottage cheese and yogurt, and describes bottled and canned beers as containing little or no tyramine. The older Nardil label is stricter, listing beer, wine and yoghurt among things to avoid for two weeks after stopping (FDA Nardil label).

Reviews of modern food data argue that most foods now contain far less tyramine than in the 1960s, because of starter cultures and hygiene rules, and that even matured cheese is usually safe in healthy-sized portions (Gillman 2018). A prescriber's guide concludes that only "a small number of risky foods" need limiting, while stressing that sensitivity varies a lot between people (Van den Eynde et al. 2022). These reviews are detailed and cite food chemistry, but at least one author (Gillman) has a declared company tie (see Follow the money). The regulator-approved labels remain the safer guide, and the variability point cuts both ways: some people react to small amounts.

The clearest regulator-reviewed data come from the Emsam programme. The dose of tyramine capsules needed to raise systolic pressure by 30 mmHg (without food) was 270 mg on the 6 mg patch versus 10 mg on tranylcypromine 30 mg a day. The 6 mg result comes from a small study (10 subjects, mean values), so individual responses vary. The label requires tyramine-restricted eating from the first day at 9 and 12 mg, and the no-diet point applies only at 6 mg.

Common side effects

• Tranylcypromine: most common (over 30%) were dry mouth, dizziness, insomnia, sedation and headache; over 10% had overexcitement, constipation, blurred vision and tremor. Postural hypotension is a "major adverse reaction" above 30 mg a day (FDA Parnate label).
• Phenelzine: dizziness, headache, drowsiness, sleep disturbance, fatigue, tremor and muscle twitching, constipation, dry mouth, weight gain, postural hypotension, oedema, and sexual problems such as anorgasmia and erectile dysfunction (FDA Nardil label).
• Isocarboxazid: nausea, dry mouth and dizziness were the main drug-related events; dropouts were mostly for dizziness, low blood pressure on standing, fainting and dry mouth (FDA Marplan label).
• Selegiline patch: side effects are listed in the FDA label; we did not re-read the full table in this review.
• Moclobemide: very common sleep disorder, dizziness, headache, dry mouth and nausea; overdose of moclobemide alone generally causes mild, reversible effects (UK Manerix SmPC).

Rare but serious

• Liver injury: very rare fatal liver damage with phenelzine; hepatitis with tranylcypromine (Nardil label; Parnate label).
• Mania in people with bipolar disorder; screening is advised before starting.
• Blood disorders, low sodium (SIADH), seizures and low blood sugar in people on diabetes medicines (FDA Parnate label).
• Withdrawal and dependence: stopping tranylcypromine suddenly has caused withdrawal including delirium, even at low doses; symptoms can start within 1 to 3 days and last weeks. Abuse has been reported, sometimes in people with past substance misuse. Phenelzine withdrawal can range from vivid nightmares to psychosis and convulsions (Parnate label 9.3; Nardil label).
• Overdose of irreversible MAOIs can be fatal and effects may be delayed; the Parnate label advises observation for at least a week.

All interactions

The interaction list is long, and it applies during treatment and for a washout period after stopping. The table summarises the main groups from the US Parnate label (FDA Parnate label 4.1, 7.1); other MAOI labels are similar.