Introduction

Most people do not suffer from migraine, whatever the provocation, and even migraine sufferers differ from each other in their susceptibility.

As Latham wrote in 1872,

‘The sufferers possess what is called the nervous temperament…the attacks are produced by prolonged mental work, protracted mental excitement, or any intense strain on the feelings such as grief, anxiety, passion, etc.…the depression that follows over excitement, a debauch, etc. are all predisposing causes.’

This passage contains two crucial observations. Certain types of individual are more likely to suffer from migraine than others, and an attack is frequently preceded by a period of stress.

Even if there is a final common pathway operating in all migraine attacks, which is quite possible but unproven, it seems clear that predisposition to migraine is multifactorial. Different subjects find their attacks to be triggered by different agents, such as stress, diet, or menstrual cycle, and the nature of the trigger is specific to particular individuals. In others, attack and refractory period seem to follow an endogenous rhythm with no obvious external initiator (see Table 19.1). This variability suggests that there are many ‘ways in’ to a migraine attack, many possible points of vulnerability in different people. Thus, the genetic basis of migraine predisposition (Davies and Clifford Rose 1986) is also likely to be multifactorial or polygenic. In this paper, we discuss some aspects of the biochemistry of migraine predisposition, and indicate how some of the initiating factors may interact with the central monoamine systems. Professor Lance and others in this volume discuss persuasively how these centres (particularly the raphe nuclei and locus coeruleus) may be involved in the generation of a migraine attack. It is likely that the more we understand about individual vulnerability, the more we may be able to tailor specific treatment to particular patients.

Diet and migraine

Many patients believe that dietary factors such as alcohol, red wine, chocolate, or cheese can provoke migraine attacks (Glover et al. 1984; Peatfield et (p. 229 )

al. 1984). However, there has also been much scepticism, particularly among clinicians, about such reports. We have recently shown, in placebo-controlled trials, that both red wine (Littlewood et al. 1988) and chocolate (C. Gibb, P.T.G. Davies, V. Glover, M. Sandier, and F.C. Rose, unpublished work) can initiate migraine attacks in particular individuals, who form a minority of the total population at risk. Both trials were conducted on carefully selected patients who were already convinced about their own individual triggering agent; in the former study, no effect of red wine could be demonstrated in migraine patients who had no history of susceptibility to this agent. The biochemical basis of these effects is still unclear. Despite earlier claims, neither tyramine nor phenylethylamine are likely to be involved, because their concentrations in red wine and chocolate are very low (Schweitzer et al. 1975; Hurst and Toomey 1981; Hannah et al. 1988). In wine, flavonoid phenols are plausible candidates; they are present in much higher concentration in red than in white wine (Littlewood et al. 1988). Even so, there is as yet no direct evidence for their involvement in migraine. Nor is there convincing evidence that allergic mechanisms play some part in dietary migraine (Merrett et al. 1983).

Platelet levels of the enzyme, phenolsulphotransferase P (PST P), one of the two forms of an enzyme that conjugates phenols with sulphate (Rein et al. 1982), have been found to be significantly decreased in patients with self-reported dietary migraine compared with a non-dietary group (Littlewood et al. 1982; Soliman et al. 1987), although there is considerable overlap in individual values. If platelet PST P activities reflect those elsewhere in the body, it is plausible that low levels can reduce the body's defences against dietary or environmental phenols because this enzyme provides a major detoxication mechanism (Sandier and Usdin 1981). PST P is under strong genetic control (Reveley et al. 1983) and low values might be a manifestation of one of the many possible forms of genetic predisposition to migraine.

Menstrual migraine

Many women report an association between time of menstruation and their migraine attacks although, in others, no such link is observed (Epstein et al. (p. 230 ) 1975; Peatfield 1986). Migraine often remits in pregnancy (Eadie and Tyrer 1985) and recurs post partum (Stein et al. 1984). One interpretation of these findings is that high levels of oestrogen, or progesterone, or both, protect against migraine, with the sudden fall that occurs at menstruation or parturition provoking an attack in susceptible individuals. Somerville (1971, 1972) has reported that oestrogen injections at the end of the menstrual cycle can postpone an attack, whereas progesterone has no such effect. Dennerstein et al. (1978) claim that oestrogen withdrawal is associated with an increase in headache intensity; conversely, oestrogen therapy appears to reduce migraine severity (Chaudhuri and Chaudhuri 1975). As yet, there is no firm evidence that menstrual migraine is linked with abnormal hormone concentrations. Women may well vary in their response to fluctuating hormone levels, rather than to absolute concentrations of the hormones themselves.

Both oestrogen and progesterone have many effects on the neurotrans-mitter systems of the central nervous system (Deakin 1988). There are receptors for both in many different brain regions but they are especially concentrated in brainstem monoamine-containing cell body groups (Maggi and Perez 1985). Oestrogen appears to affect both dopamine and 5-hydroxytryptamine (5-HT) systems. It has clear actions on dopamine receptors and turnover, and both oestrogen and progesterone augment the number of 5-HTx receptors in the cortex (Biegon et al. 1983). O'Connor and Feder (1985), working in guinea-pigs, found evidence to suggest that a decline in oestrogen level reduces functional 5-HT neurotransmission. Conceivably, this may form a possible basis for increased vulnerability to menstrual migraine.

Platelet monoamine oxidase (MAO) and migraine

Apart from a transitory decrease during an attack (Sandier et al. 1970; Glover et al. 1977) mean platelet MAO activity is significantly reduced in migraine patients between attacks, compared with controls (Sicuteri et al. 1972; Sandier et al 1974; Bussone et al. 1977; Glover et al. 1981). The finding is particularly marked in males (Glover et al. 1981). Platelet MAO activity is under genetic control (Sandier et al. 1981) and it seems possible that the low values represent another vulnerability point for migraine.

We have recently found a direct and significant correlation in migraine patients between platelet MAO activity and anxiety and depression level, as assessed by questionnaire (Glover et al. 1987; J.T. Littlewood, A. Prasad, C. Gibb, V. Glover, M. Sandier, R. Joseph and F.C. Rose, unpublished results). Others have observed that high anxiety levels are associated with both extremes of the platelet MAO range (Schalling 1987). High platelet MAO activity has also been found in several studies of depression (Sandier et al. 1981). It is, thus, of interest that only low and not high values have been identified in migraine patients. Both anxiety and depression are likely (p. 231 ) to be biochemically heterogeneous and only certain subtypes appear to be linked with migraine.

Low platelet MAO activity has been noted in certain other clinical states (see Sandier et al. 1981 for review). Of most interest, perhaps, in the present context are the low values recorded in the idiopathic pain syndrome (Knorring et al. 1984). Oreland et al. (1984) have shown that platelet MAO activity correlated with 5-hydroxyindoleacetic acid concentrations in the cerebrospinal fluid, and they suggested that low enzyme activities reflect a hypoactive 5-HT system. There is evidence that 5-HT-depleted rats have an increased sensitivity to pain (Willner 1985). It may well be that a low platelet MAO activity reflects an underactive 5-HT system and is a predisposing factor for migraine for this reason.

Anxiety and depression in migraine

There is now considerable evidence that migraine patients have more anxiety and depression than controls (Diamond 1964; Howarth 1965; Kashiwa-gi et al. 1972; Price and Blackwell 1980). When patients in a migraine clinic are investigated, there is always the concern that the patients are to some extent self-selected and perhaps more neurotic than those in the general population. There have been some thorough investigations of the characteristics of migraine sufferers in the general population which have, in fact, been in good agreement with findings at specialist clinics. The consensus of opinion, however, is that there is little evidence of a migraine personality as such (Hundleby and Loucks 1985).

A major study of psychological aspects of migraine patients in the British Civil Service was carried out by Henryk-Gutt and Rees (1973), who showed significantly increased levels of anxiety, compared with controls. Two more recent population studies, one in an English market town (Crisp et al. 1977) and another in New Zealand (Paulin et al. 1985), found a higher incidence of anxiety and depression in migraine patients than in the general population; in the latter, there was a positive relationship with headache frequency. Price and Blackwell (1980) also found that migraine sufferers, too, show higher levels of trait anxiety, by a variety of rating scales, than normal subjects, as did Howarth (1965).

The association of migraine and depression has been explored in depth rather more recently. Garvey et al. (1983) asked 116 patients with major depressive disorder about their headache pattern. They had a headache rate similar to that of controls during their non-depressed phase, but an increased rate during episodes of depression. Merikangas et al. (1988) studied the association between migraine and depression in a family study of prob-ands with major depressive disorder, and in community controls and relatives of both groups. They noted a significant association between migraine and depression in both probands and their relatives; they also observed (p. 232 )

symptoms of anxiety in depressed patients with migraine. However, while both migraine and depression were strongly familial, their association appeared to be only partly transmissible. One explanation for these results would be if the depression were secondary to the migraine.

In a recent study (J. Jarman, M. Fernandez, V. Glover, M. Sandier, P.T.G. Davies, T. Steiner, C. Thompson and F.C. Rose, unpublished results), we have used the Schedule for Affective Disorder — Lifetime version to obtain a more detailed psychiatric profile than had been obtained before of typical migraine patients who attend a specialist clinic. The results are given in Table 19.2. More than half the patients interviewed had a lifetime history of some degree of psychiatric disorder. Major depression was much the commonest diagnosis and, in all but one patient, was of the endogenous type. This finding is in marked contrast with the depression observed in other circumstances; postnatal depression, for example, predominantly corresponds to the neurotic subtype. Given the high anxiety incidence in migraine patients, and their high score on the Eysenck Personality Questionnaire (Henryk-Gutt and Rees 1973), it might have been predicted that they, too, would have manifested predominantly neurotic depression. It seems possible that the biological factors that predispose to endogenous depression also predispose to migraine.

The tyramine test in migraine patients

Evidence that the high incidence of endogenous depression of these migraine sufferers was at least partly due to a biological predisposition comes from the results of the tyramine test (J. Jarman, M. Fernandez, V. Glover, M. Sandier, P.T.G. Davies, T. Steiner, C. Thompson and F.C. Rose, unpublished results). Low tyramine sulphate conjugation after an oral load (Sandier et al. 1975; Bonham Carter et al. 1978) is a trait marker for (p. 233 )

endogenous depression, compared with the neurotic variant of the illness (Harrison et al. 1984; Hale et al. 1986). It is also likely to be a genetic marker, because some first-degree relatives of low tyramine sulphate excre-ters, who themselves have never been depressed, also excreted significantly lower amounts of conjugate than did controls (Hale et al. 1986). Despite substantial investigative effort, the biochemical basis for this finding, which appears to be unrelated to phenolsulphotransferase activity (Bonham Carter et al. 1981), remains unknown, as does its relationship to other putative markers of endogenous depression.

The migraine patients described in Table 19.2 had significantly lower tyramine sulphate excretion on oral tyramine challenge than controls. Figure 19.1 shows that almost all the migraine patients with a lifetime history of endogenous depression had a low tyramine sulphate output. This study gives strong independent support to the psychiatric diagnosis in these patients. It also has therapeutic implications. There is evidence (Hale et al. 1989) that depressed patients with low values in the tyramine test respond better to tricyclic antidepressant medication than those with normal values. It may be that the test will also identify a subgroup of migraine sufferers who respond well to these drugs (Kashiwagi et al. 1972; Couch et al. 1976).

Paradoxically enough, the tyramine conjugation deficit was first identified in a group of patients classified as having dietary migraine (Youdim et al. (p. 234 ) 1971), although a later attempt at replication proved unsuccessful (S. Bonham Carter and M. Sandier, unpublished results). In the light of the present findings (J. Jarman et al., unpublished results), it seems likely that the earlier group included, by chance, an enrichment of subjects with a lifetime incidence of endogenous depression. The present data fail to show any difference between dietary and non-dietary migraine in the tyramine test.

The tyramine conjugation study also provides evidence against the idea that depression is secondary to migraine in these patients (Merikangas et al. 1988), and rather supports the interpretation that a disturbance of particular biochemical systems predisposes to both migraine and depression. Whether monoamine systems are involved has been a subject of much speculation in published work.

Links between monoamine systems, anxiety, depression, and migraine

There is evidence from animal models that in anxiety as well as stress there is increased activity of the sympathetic nervous system, and release of catecholamines both peripherally (Gray 1982) and centrally, where there is increased firing of the locus coeruleus (Stone 1975). There is also evidence for enhanced noradrenergic function, with increased anxiety, in humans (Ballenger et al. 1984). As anxiety seems to predispose to migraine, and as stress is a major triggering factor, the release of catecholamines seems to be one possible cause of an attack.

Anthony (1981) noted a significant rise in plasma noradrenaline and dopamine β-hydroxylase levels during a migraine attack. Hsu et al. (1977) reported a rise in mean plasma noradrenaline levels three hours before a group of migraine subjects awoke, with a headache already present. Spierings (1985) suggested that activation of the sympathetic system may be a result of the headache, but the fact that stress often precedes headache makes it more likely that headache follows activation of catecholamine systems.

The biochemical basis of depression is unclear; despite many provisos (Stone 1983), the current evidence is still, on balance, compatible with early hypotheses that depression is associated with functionally /zypoactive catecholamine systems (Ballenger et al. 1984) and/or 5-HT systems (Van Praag 1984) in the brain. It is possible that a hypoactive 5-hydroxytryp-taminergic system is also associated with headache. 5-HT is released peripherally during a migraine attack (see Peatfield 1986, for review), and may well be released centrally also. There is evidence that reserpine, which depletes monoamines, causes headache in susceptible individuals (Curzon et al. 1969) whereas 5-hydroxytryptophan, the precursor of 5-HT, can alleviate it (Sicuteri 1972; Titus et al. 1986). (p. 235 )

A migraine attack follows a characteristic pattern: ‘The most important emotional colourings during the clinically recognised portion of a common migraine attack are states of anxious and irritable hyperactivity in the early portions of the attack and states of apathy and depression in the bulk of the attack’ (Sacks 1973). This progression, too, needs to be explained in biochemical terms.

Gray (1982) has discussed the biochemical effects of stress, and its relationship to anxiety and depression, in a way that may be relevant here. In rat models, after acute anxiety or stress has increased the activity of norad-renergic neurones in the central nervous system (Stone 1975), there may be depletion of noradrenaline from vesicles, and this may correspond to a state of functional exhaustion or depression. This, in turn, may be followed by transneuronal enzyme induction, occuring 16 to 18 hours after the original stimulus, and remaining elevated for a number of days (Thoenen 1975; Fillenz 1977). These three stages of neuronal and biochemical hyperactivity, underactivity, and reaction or recovery are possible biochemical correlates of the prodrome, the headache phase and the refractory period of a migraine attack.

Figure 19.2 shows a model that illustrates this hypothesis. The beginning of a migraine attack is associated with a surge of 5-HT release. 5-HT cells in the dorsal raphe nucleus are innervated by excitatory noradrenergic projections from the locus coeruleus (Willner 1985). The sequence can thus be triggered by stress, and occurs more readily in an anxious personality. This stage is followed by lowered activity of the 5-hydroxytryptaminergic system, possibly owing to 5-HT depletion or altered receptor sensitivity, associated with the headache itself. Low platelet MAO activity and the fall of oestrogen (p. 236 ) at menstruation may both be associated with a hypoactive 5-HT system which predisposes to this stage, as does a vulnerability to endogenous depression (as indicated by the tyramine test). The refractory period may involve a reactive enzyme or receptor induction. How dietary triggers enter this cycle is, at present, unknown.

The picture is clearly both speculative and an oversimplification and will be testable only by new methods such as positron emission tomography scanning. However, the model may be useful to focus attention on different potential points of vulnerability which, perhaps, need different treatments.

Directions for the future

The biochemical cascade of the type discussed above, or the one earlier envisaged by Sandier (1972) — which involved a dietary triggering agent, release of an intermediate agent from the gut wall to the portal and venous circulation, and a ‘spasmogen’ released into the arterial circulation from the pulmonary vascular bed — provide one possible scenario, but it can obviously be short-circuited. Recent observations of Brewerton et al. (1988) on m-chlorophenylpiperazine (m-CPP), a triazolopyridine metabolite of tra-zodone, appear to be important here. Oral administration of this compound was able to produce a typical, common migraine-like headache in 28 out of 52 subjects, eight to 12 hours after its administration. Tryptophan or a placebo failed to initiate attacks. There was a significantly higher headache incidence in the tested subjects who had a history or a family history of migraine and, indeed, 18 out of 20 of these developed severe symptoms. Thus, in this case at least, the postulated biochemical sequence was bypassed and the chemical substance responsible for the end-organ response probably travelled directly to its site of action.

The presence of acute peripheral concomitants of a migraine attack — a second transitory type of platelet MAO deficit present only across the acute episode (Sandier et al. 1970; Glover et al. 1977), platelet 5-HT release (Curran et al. 1965; Anthony et al. 1967); liberation of platelet β-thromboglobulin (Gawel et al. 1979); and increased platelet aggregability (Hilton and Cumings 1972) — have encouraged speculation that a circulating platelet-damaging agent might be present (Sandier 1978), perhaps being liberated from the pulmonary vascular bed (Sandier 1972). This substance may also be responsible for known blood-vessel calibre changes and, conceivably, for head pain. Where such a humoral agent might interact is, of course, unknown, although in the light of present information it is tempting to nominate the 5-HT₁like receptor system.

m-CPP itself seems to be a 5-HTi receptor ligand (Brewerton et al. 1988). It is noteworthy that most of the drugs that are clinically effective in migraine therapy bind to one or other central 5-HT receptor (Peroutka 1988). Do migraine-initiating foodstuffs bind similarly? It would be of great (p. 237 ) interest to see whether fractions of red wine, now shown objectively to provoke migraine attacks (Littlewood et al. 1988), attach themselves to these receptors. Chocolate, too, is another migraine-provoking agent identified objectively (C. Gibb, P.T.G. Davies, V. Glover, M. Sandier and F.C. Rose, unpublished work) and extracts of it might be similarly investigated.

The approach adopted by Peroutka (1988), to identify possible 5-HT receptor ligands that act as therapeutic agents, exploited their ability to displace other known ligands from human brain membranes. Whether receptor-binding kinetics undergo substantial post mortem changes in human brain and, indeed, whether brains from migrainous subjects bind ligands similarly to those from controls, are other urgent questions that must be tackled in the future. And do the antimigraine drugs which bind to 5-HT_r like receptors displace m-CPP from its binding sites? We shall try to provide some answers in the near future.