8. Migraine & melancholic depression

These commentaries are based on Dr Gillman’s peer reviewed scientific papers, see Publications


There is an association of migraine with serious depression, encompassing the linked domains of genetics, patho-physiology and treatment response. A careful personal and family history of both disorders should be sought and will help in guiding the treatment of those with such a history. It is interesting that drugs working in several different ways (re-uptake inhibitors like TCAs, MAOIs and GABA-ergic drugs like valproate) are clearly efficacious for both conditions — a suggestion of shared patho-physiology. This commentary deals with recent research findings (including CGRP) and considers the Bayesian rationale for re-considering the role of several drugs (guidelines are somewhat discrepant) and partly reprises the role of older drugs, particularly MAOIs, that have been in the shadows for decades.

A striking paradox is that despite the recognised importance of serotonin in migraine pathophysiology and the apparent efficacy of some TCAs, the TCA that has the best effect on serotonin (clomipramine) has never been properly tested in migraine.


It has been my view for decades that the association between migraine and manic-depressive illness (Bipolar disorder) is closer than is often thought, and I paid special attention to a history of both when assessing cases. Evidence has continued to accumulate that this closeness of association includes the genetics, the patho-physiology and the treatment of these two conditions.

Since I began updating this commentary what appear to be major developments concerning calcitonin gene-related peptide (CGRP) are in the news. The neurology community is abuzz with the news of similarly decisively positive trials of several different agents all antagonising the effect of CGRP [1]. The effect is being touted as substantially greater than any previous treatment and, at least so far, there seems to be a widespread view that such agents are likely to be a major breakthrough in the treatment of migraine — but they always say that don’t they!

CGRP has a crucial role in the trigemino-cervical neurovascular pathway that contributes to vasodilation and neurogenic inflammation, which generates the acute attack: it is elevated during the headache phase of migraine, with reduction following headache resolution.

Another development seems to be a consensus that there is now better evidence and more agreement that vasodilation does not play a part in migraine headache after all [2].

The TCAs

The modest effectiveness of the ‘TCAs’ in migraine prophylaxis has withstood the test of time, indeed they are the most commonly used medications in the USA second only to topiramate [3, 4]. Nevertheless, there are different recommendations in the burgeoning multitude of guidelines.***. Lumping TCAs together is not valid, researchers have persistently failed take account of the completely different properties of individual TCAs, e.g. doxepin and trimipramine are simply antihistamines, and desipramine is as close to being a selective NRI as any new drug like reboxetine (see my TCA review paper [5]). Lumping the TCAs together constitutes an elementary error of pharmacology and logic. It is highly unlikely that they are all equally effective, whatever condition is being treated.

***They all rely on meta-analysis of much the same sets of data – which may be biased and unrepresentative.

The only TCA with potent activity as a 5-HT reuptake inhibitor is clomipramine. Amitriptyline, which has very little effect on 5-HT reuptake [5] and minimal effect on platelet 5-HT compared to CMI [6], is the TCA most commonly used for migraine.

So, preferring AMI over CMI is not logical, nor based on good science.

Evidence indicates a patho-physiological role for low 5-HT in the vulnerability to migraine, and acute 5-HT release from platelets in the triggering of acute episodes — clomipramine remedies both those abnormalities. That suggests further quality study of its ability to act as a prophylactic would be worthwhile: my fairly extensive clinical experience indicates clearly that it is significantly more effective than amitriptyline for migraine.

Bayesian reasoning suggests that a reasonable choice, from the TCAs, is clomipramine rather than amitriptyline: and yet, even after 50 years, there are only a couple of inadequate trials [7, 8]. However, there is one Japanese paper which describes marked improvement in 8/8 patients with CMI [9].

There is no good reason for this — quite the reverse — it is just a historical accident, similar to the accident of CMI being used as a drug for OCD, rather than depression, when in fact it is the best AD of all the TCAs (another ‘fact’ that is also not supported by good clinical trial evidence).

The two very poor short-term trials of clomipramine for migraine prophylaxis [7, 8] are clearly inadequate to justify an assumption that it is ineffective.

SSRIs are seemingly not effective: but opinion is not unanimous on that point. However, if release of 5-HT from platelets is a key mechanism in triggering attacks, then one would expect them to be effective: if they are not, that would argue against a major role of platelet 5-HT release.

Science alone does not determine what treatments we use: that is influenced by history, chance, ‘fashion’, and money [10].

In the original version of this commentary, more than 20 years ago, I observed that recent research substantiated a significant genetic relationship, especially with manic-depressive illness (bipolar disorder). There is now further evidence, which indicates that the link is also significant with unipolar depression, in particular the form showing ‘melancholic’ depression characteristics (see below for discussion of this).

Also relevant to the 5-HT issue it the association of migraine with stroke. This relationship remains uncertain but suggestive [11-15].

Serotonergic mechanisms may play a part in trigeminal meningeal nociception [16].


There is no consensus concerning the patho-physiology of migraine, and the alternative possibilities advanced are diverse: the CGRP evidence is rapidly gaining ascendency (as most new things do, until reality intrudes).

Biochemical studies of migraine have suggested alterations in platelet 5-HT content, which seems to decrease early in a migraine attack, as serum levels increase. Early in an attack there seems to be a considerable depletion of platelet 5-HT (which returns to normal within 72 hours [17, 18]), causing a short-lived increase in plasma 5-HT, and a rise in urinary 5-HIAA. An inverse correlation between the urinary levels of 5-HIAA and platelet 5-HT seems to have been replicated [19, 20], which indicates that the source of 5-HT is the platelet-release, not neuronal activity (cf. serotonin toxicity). It would be nice to see more extensive replication of these key findings.

An infusion of a 5-HT-releaser such as reserpine (note it is non-specific releaser, being more potent as a dopamine releaser) has been reported to induce migraine-like headaches [21] — it is therefore interesting that there seem to be no reports of migraine following the ingestion of the widely used and more specific 5-HT releaser, MDMA [22]. Other 5-HT releasers have also been reported to precipitate migraine [19, 23].

However, in relation to the CNS, migraine is not a feature of serotonin toxicity. This can be stated very reliably because they have now been an enormous number of well-documented cases, of all degrees of severity, examined by expert toxicologists — no migraine. End of story, surely?

We can conclude confidently that large increases of CNS serotonin do not trigger migraine.

Intravenous infusion of 5-HT has been reported to reduce the pain in migraine patients [24], presumably a peripheral effect, since serotonin does not cross the BBB***.

***There appears to be little discussion of this apparent paradox: but note this ‘fact’ seems to be substantiated by Anthony [19, 25], but it is contradicted by Ostfeld [26], see Panconesi for detailed discussion [23].

When will medical science come to terms with the requirement for the independent replication of results? It is astonishing that such a central & fundamental finding, in such an important condition, has been brushed aside almost un-replicated.

Whether 5-HT release by platelets plays a significant causal role remains uncertain: these, and other aspects of recent research, are reviewed in detail by Deen et al. [27], however they conclude ‘the importance of serotonin in migraine pathophysiology is indisputable’. Panconesi et al state ‘A low central 5-HT disposition associated with an increase in 5-HT release during attack is the most convincing change of 5-HT metabolism implicated in migraine’ [28].

Clomipramine substantially depletes platelet 5-HT — amitriptyline only a little [6, 29]. Platelets express the 5-HT transporter (aka SERT, or SER-T, or 5-HTT) on their outer membrane and utilise it to garner 5-HT from the serum and accumulate it in vesicles, where it then plays a role in platelet aggregation and blood clotting. This is evidenced by the fact that all potent 5-HT reuptake inhibitors increase the bleeding tendency, this is an aspect of SSRI side-effects which has received only belated attention until quite recently [30-33].

These drugs increase central CNS 5-HT levels — one of their proposed mechanisms of antidepressant action — which would be posited to lessen the hypothesised CNS 5-HT deficit in migraine. This central 5-HT under-activity being a putative link between migraine and depressive illness.

Incidental comment

The above is congruent with my own experience, from clinical practice, where I frequently switched people from amitriptyline to clomipramine, for the improved treatment of their depressive illness (CMI is a more potent anti-depressant drug than AMI), and at the same time observed a great lessening of the frequency of migraine attacks, over the medium to long term (one to six months). Indeed, it was not unusual to see patients who reported complete absence of migraine attacks once established on a therapeutic dose of clomipramine. I have seen that added benefit, of clomipramine over amitriptyline, in a sufficient number of patients over the years, to suppose that it deserves explanation.

Epidemiological & genetic studies

Migraine is now conceptualised as a neuro-vascular disorder with a strong genetic basis — some cases seem to be caused by single genes — common migraine exhibits definite familial clustering [34, 35].

There are causal mutations in ion channel genes (CACNA1A, ATP1A2, and SCN1A), which effect glutamatergic neurotransmission and the threshold for cortical spreading depression. In common polygenic migraine, genome-wide association studies have identified SNPs at many loci that are associated with migraine risk.

A recent Australian study suggests the risk for migraine in relatives of probands reporting depression is considerably higher than the risk for depression in relatives of probands reporting migraine [36]. Yang’s results indicate that the observed comorbidity between migraine and depression is explained almost entirely by shared genetic mechanisms [37]. Indications are that migraine in depressed patients represents a bipolar trait [38], cf. ‘The link’ below.

Oedegaard’s study further substantiates that unipolar cases with comorbid migraine resemble bipolar II patients [39].

The link: depression and migraine

At this point, a quick refresher concerning the genetics and classification of depression is pertinent. The form of depression that shows classic endogenous features, often now referred to as melancholic depression — to help separate it from the over-inclusive DSM category of major depressive disorder (MDD) — shows the classic biological features of anergia and anhedonia, combined with early-morning awakening, diurnal variation etc.; these demarcate a form of depression which is more likely to transmute, in time, into a picture of bipolar disorder [40]. Therefore, the genetic commonality between melancholic type ‘unipolar’ depression and classic manic-depressive illness (now referred to as bipolar disorder) is very significant, and it is this form of depression which is most strongly linked to migraine. The point to remember is that many index cases of depression, with the above melancholic characteristics, (and especially those who also suffer migraine) will turn out, in the course of a life-time, to exhibit a more typical manic-depressive course [40].

That observation has implications for optimum treatment strategies — like being careful of monotherapy with TCAs if patients are potentially manic-depressive (cf. MAOIs & valproate). Such considerations are not well addressed in guidelines.

It is also noteworthy that migraineurs, even without diagnosed depression, have a substantially increased risk of suicide [41, 42]. This is hardly surprising, since we would expect many missed cases of depression in people who are treated for migraine. They are unlikely to be examined carefully for depressive symptoms in all cases. This serves as an indication of the importance of detecting depressive symptoms in migraineurs.


Anyone, with even a passing acquaintance with my writing, will know my views about the regrettable underuse of MAOIs in serious depression [43, 44]. It will come as no surprise to learn that there are some studies indicating an excellent effect of MAOIs on severe migraine, although one would hardly think that from the current migraine literature, most of which fails to mention them at all. Because of my frequent use of MAOIs, and my long-standing interest in migraine and its connection with manic-depressive illness, I have observed many patients who, once put on tranylcypromine (TCP), had good long-term amelioration of previously long-standing and troublesome migraine headaches.

Allow me to give a little detail by way of an anecdote. Most published medical research is wrong [45], and one only needs to go back a few years and check, to find a myriad of new treatments which never subsequently enjoyed independent replication or validation. The false prophet of the double-blind trial continually proves that drug A > B > C > A — Penrose stairs with drugs, as I have previously described it — and its dominant weighting in the assessment of evidence is quite unjustified by its proven inadequacy as a methodology, both in theory and practice. However, it is a methodology which serves the purposes of pharmaceutical companies getting easy approval for, and marketing, new and expensive drugs. One does not need a double-blind trial for an effective migraine prophylactic treatment any more than for a trial of whether penicillin or parachutes work [46]. If one needs statistics to do that analysis, then the treatment is likely to be of only marginal efficacy.

For many years it has been my habit to contact researchers, who have published subsequently un-replicated new treatment reports, to ask why they have not published anything else since then. Many of them simply decline to answer such questions, the honest ones admit further work was a failure — but, the subsequent failures are almost never published.

The earliest paper that I was aware of (when I wrote the original commentary concerning MAOIs & migraine 20 years ago), was published by an eminent Australian neurologist, Prof James Lance [47]. When I saw that he had not written anything for some years after that paper, I contacted him, in his retirement, to ask why. His reply was handwritten with a fountain pen — the last such letter that I remember receiving. He told me that he had continued to use them over many years and found them of considerable efficacy, but that he had, regrettably, simply not got around to doing any more formal work, or trials, or writing more about it.

There are a couple of more recent reports using MAOIs [48, 49], as well as more older ones [24, 50-53].

If MAOIs are used for severe depression my experience, along with the above references, suggests one can expect useful prophylaxis of severe migraine.

Since MAOIs are safer and easier to use than most people imagine, as I have discussed elsewhere [43, 44], they would seem to be the underused ammunition in the therapeutic armamentarium. For patients who are bipolar they represent a possible choice, since many are of the opinion TCP is a good option for the depressive phase of BPD.

GABA and Valproate

Valproate enhances the action of GABA, which is the major inhibitory neurotransmitter in the CNS, and seems to be of established efficacy for migraine. One imagines this indicates that another of the contributing mechanisms involved in the poly-genic susceptibility to both migraine and BPD, involves some kind of sensitivity and over-reactiveness of the CNS. In that context, it is notable that Valproate is an effective anti-manic drug, also with probable prophylactic activity, for bipolar disorder itself. It therefore has a place in patients who are bipolar.

Guideline recommendations for migraine prophylaxis

It is appropriate to preface any comment about guidelines (see other comment here) by noting their often undeclared conflicts of interest [54], and biases [55, 56], and the considerably discrepant recommendations by different groups compiling guidelines, which attests to their inherent problems. Nonetheless, it is of use to be aware of what drugs are, and are not, recommended in such sources. Anyone familiar with my writing will know of my considerable scepticism about the backbone of guidelines, meta-analysis, which I refer to as the ‘phrenology of the new millennium’.

As examples, the following guidelines are not exactly in full agreement!

Canadian Headache Society, 2012: ‘Based on our review, 11 prophylactic drugs received a strong recommendation for use (topiramate, propranolol, nadolol, metoprolol, amitriptyline, gabapentin, candesartan, butterbur, riboflavin, coenzyme Q10, and magnesium citrate) and 6 received a weak recommendation (divalproex sodium, flunarizine, pizotifen, venlafaxine, verapamil, and lisinopril) [57].

The ‘Quality Standards Subcommittee of the American Academy of Neurology and the American Headache Society’ opine [58]: ‘Divalproex sodium, sodium valproate, topiramate, metoprolol, propranolol, and timolol are effective for migraine prevention and should be offered to patients with migraine to reduce migraine attack frequency and severity (Level A).’

A UK perspective from ‘NICE’ may be found here:


They suggest: topiramate or propranolol & to consider amitriptyline (according to the person’s preference, co-morbidities and risk of adverse events).

It is important not to allow guidelines to become a substitute for thought and clinical judgement. Even more insidiously, some doctors seem to be using guidelines as an excuse to be intellectually lazy and dictatorial about the treatment they offer to patients: one hears increasingly frequently that patients have been told they cannot have a treatment because ‘it is not in the guidelines’. That is poor practice, and inappropriate.

Concluding comments

When patients manifest both depression and migraine, as they may do, it is important to be aware of, and treat, both conditions. It is good practice to screen migraineurs & depressives for both past and family history of depression (especially bipolar disorder) and panic attacks and migraine. This will guide treatment.

Clomipramine is a forgotten gem for severe depressive illness. It may be considered when depression and migraine are co-morbid, which is not infrequently. However, as with AMI, caution is required if bipolarity is suspected. When that is the case, MAOIs and valproate deserve consideration, and they are compatible with each other and with lithium.

If lithium is not required or desired, valproate may be considered for possible or definite bipolar cases, and MAOIs may be the treatment of choice for depressed bipolar migraineurs.

The frequent co-occurrence of depression and anxiety means beta-blockers need to be used with due consideration of the possibility of their immediate or future presence. Some beta-blockers are used for symptomatic treatment of anxiety, although usually they are only efficacious for tremor related to performance anxiety (musicians, for instance), but they can exacerbate depression because they cause tiredness, apathy and bad dreams.

This is especially so because we know that even migraineurs without diagnosed depression have a substantially elevated suicide rate.

Some are of the opinion that magnesium deficiency may be relevant: it is common, and all one has to do to remedy that is to use sea salt, as opposed to ordinary table salt, which has less magnesium. See ‘Why all migraine patients should be treated with magnesium’ [59] and re the surprisingly high frequency of magnesium deficiency see [60-62].

I find it interesting that guidelines contain little substantive comment or discussion about the relevance of the co-occurrence of migraine and melancholic-type depression and bipolar disorder (it is mentioned en passant at best). These are relevant factors in guiding treatment.

The discrepancies between different guidelines must reflect on the complexity of migraine, and on the quality of research, and the procedures by which it is assessed and translated into everyday practice.


1. Robinson, A.J., Positive Trial Results for Anti-migraine Drugs Targeting CGRP Could ‘Change the Way Neurologists Practice,’ Experts Say. Neurology Today, 2017. July 6th: p. 31-32.

2. Shayestagul, N.A., et al., Measurement of Blood Flow Velocity in the Middle Cerebral Artery During Spontaneous Migraine Attacks: A Systematic Review. Headache: The Journal of Head and Face Pain, 2017. 57: p. 852–861.

3. Burch, R., Antidepressants for Preventive Treatment of Migraine. Curr Treat Options Neurol, 2019. 21(4): p. 18.

4. Woolley, J.M., et al., Migraine Prophylaxis and Acute Treatment Patterns Among Commercially Insured Patients in the United States. Headache, 2017. 57(9): p. 1399-1408.

5. Gillman, P.K., Tricyclic antidepressant pharmacology and therapeutic drug interactions updated. British Journal of Pharmacology 2007. 151(6): p. 737-48.

6. Lingjaerde, O., Inhibition of platelet uptake of serotonin in plasma from patients treated with clomipramine and amitriptyline. Eur J Clin Pharmacol, 1979. 15(5): p. 335-40.

7. Noone, J.F., Clomipramine in the prevention of Migraine. Journal of International Medical Research, 1980. 8: p. s 49-52.

8. Langohr, H.D., et al., Clomipramine and metoprolol in migraine prophylaxis–a double-blind crossover study. Headache, 1985. 25(2): p. 107-13.

9. Tsuji, M. and H. Iida, Treatment of headache with antidepressant. Folia Psychiatr Neurol Jpn, 1984. 38(2): p. 143-9.

10. Lewontin, R.C., Biology as Ideology. The Doctrine of DNA. 1991: HarperCollins.

11. Abanoz, Y., et al., Migraine as a risk factor for young patients with ischemic stroke: a case-control study. Neurol Sci, 2017. 38(4): p. 611-617.

12. Hu, X., et al., Migraine and the risk of stroke: an updated meta-analysis of prospective cohort studies. Neurol Sci, 2017. 38(1): p. 33-40.

13. Lantz, M., et al., Migraine and risk of stroke: a national population-based twin study. Brain, 2017. 140(10): p. 2653-2662.

14. Osler, M., et al., Migraine and risk of stroke and acute coronary syndrome in two case-control studies in the Danish population. Clin Epidemiol, 2017. 9: p. 439-449.

15. Mawet, J., T. Kurth, and C. Ayata, Migraine and stroke: in search of shared mechanisms. Cephalalgia, 2015. 35(2): p. 165-81.

16. Kilinc, E., et al., Serotonergic mechanisms of trigeminal meningeal nociception: Implications for migraine pain. Neuropharmacology, 2017. 116: p. 160-173.

17. Izzati-Zade, K.F., The role of serotonin in the pathogenesis and clinical presentations of migraine attacks. Neurosci Behav Physiol, 2008. 38(5): p. 501-5.

18. Borgdorff, P. and G.J. Tangelder, Migraine: possible role of shear-induced platelet aggregation with serotonin release. Headache, 2012. 52(8): p. 1298-318.

19. Lance, J., 5-Hydroxytryptamine and its role in migraine. European neurology, 1991. 31(5): p. 279-281.

20. Dalsgaard-Nielsen, J., P. le Fevre Honore, and I. Zeeberg, Changes in platelet function and blood serotonin level in migraine patients during treatment with femoxetine. Acta Neurol Scand, 1982. 66(2): p. 191-8.

21. Iversen, H.K. and J. Olesen, Experimental headache, in The headaches., J. Olesen, P.C. Tfelt-Hansen, and K. Welch, Editors. 1993, Raven Press: New York. p. 145–151.

22. Partilla, J.S., et al., Interaction of amphetamines and related compounds at the vesicular monoamine transporter. Journal of Pharmacology and Experimental Therapeutics, 2006. 319(1): p. 237-46.

23. Panconesi, A. and R. Sicuteri, Headache induced by serotonergic agonists–a key to the interpretation of migraine pathogenesis? Cephalalgia, 1997. 17(1): p. 3-14.

24. Kimball, R.W., A.P. Friedman, and E. Vallejo, Effects of Serotonin in Migrainous Patients. Neurology 10:107-111 (Feb) 1960., 1960.

25. Anthony, M., H. Hinterberger, and J.W. Lance, Plasma serotonin in migraine and stress. Arch Neurol, 1967. 16(5): p. 544-52.

26. Ostfeld, A.M., Migraine headache: its physiology and biochemistry. JAMA, 1960. 174: p. 1188-90.

27. Deen, M., et al., Serotonergic mechanisms in the migraine brain – a systematic review. Cephalalgia, 2017. 37(3): p. 251-264.

28. Panconesi, A., Serotonin and migraine: a reconsideration of the central theory. J Headache Pain, 2008. 9(5): p. 267-76.

29. Lingjaerde, O., Inhibitory effect of clomipramine and related drugs on serotonin uptake in platelets: more complicated than previously thought. Psychopharmacology (Berl), 1979. 61(3): p. 245-9.

30. Anglin, R., et al., Risk of upper gastrointestinal bleeding with selective serotonin reuptake inhibitors with or without concurrent nonsteroidal anti-inflammatory use: a systematic review and meta-analysis. Am J Gastroenterol, 2014. 109(6): p. 811-9.

Consider Donating to PsychoTropical

PsychoTropical is funded solely through generous donations, which has enabled extensive development and improvement of all associated activities. Many people who follow the advice on the website will save enormously on doctors, treatment costs, hospitalization, etc. which in some cases will amount to many thousands of dollars, even tens of thousands — never mind all the reduction in suffering and the resultant destruction of family, work, social, and leisure capability. A donation of $100, or $500, is little compared to those savings. Some less-advantaged people feel that the little they can give is so small it won’t make a difference – but five dollars monthly helps: so, do not think that a little donation is not useful.

– Dr Ken Gillman

Dr Ken Gillman