Abstract
Migraine is a common episodic headache disorder. A comprehensive headache treatment plan includes acute attack treatment to relieve pain and impairment and long-term preventive therapy to reduce attack frequency, severity, and duration. Circumstances that might warrant preventive treatment include: (i) migraine that significantly interferes with the patient's daily routine despite acute treatment; (ii) failure, contraindication to, or troublesome side-effects from acute medications; (iii) overuse of acute medications; (iv) special circumstances, such as hemiplegic migraine; (v) very frequent headaches (more than two a week); or (vi) patient preference. Start the drug at a low dose. Give each treatment an adequate trial. Avoid interfering, overused, and contraindicated drugs. Re-evaluate therapy. Be sure that a woman of childbearing potential is aware of any potential risks. Involve patients in their care to maximize compliance. Consider co-morbidity. Choose a drug based on its proven efficacy, the patient's preferences and headache profile, the drug's side-effects, and the presence or absence of coexisting or co-morbid disease. Drugs that have documented high efficacy and mild to moderate adverse events (AEs) include b-blockers, amitripty-line, and divalproex. Drugs that have lower documented efficacy and mild to moderate AEs include selective serotonin reuptake inhibitors (SSRIs), calcium channel antagonists, gabapentin, topiramate, riboflavin, and non-steroidal anti-inflammatory drugs.
Introduction
Migraine is a common episodic headache disorder with a 1-year prevalence of approximately 18% in women, 6% in men, and 4% in children (1). It is characterized by attacks that consist of various combinations of headache and neurological, gastrointestinal, and autonomic symptoms. The International Headache Society (IHS) calls common migraine ‘migraine without aura’ (1.1) and classic migraine ‘migraine with aura’ (1.2), the aura being the complex of focal neurological symptoms that most often precedes or accompanies an attack (2). Effective migraine treatment begins with making an accurate diagnosis that may include ruling out alternate causes and ordering appropriate studies, and certainly includes addressing the headache's impact on the patient. Migraine varies widely in its frequency, severity, and impact on the patient's quality of life. A treatment plan should consider not only the patient's diagnosis, symptoms, and any coexistent or co-morbid conditions, but also the patient's expectations, needs, and goals (3). Patients need to be informed of the goals of treatment, the purpose of the various components of their treatment plan, the need for follow-up care, and the adverse effects of medications. Patient education and commitment improve compliance and foster the patient–physician relationship.
A comprehensive headache treatment plan includes a combination of: (i) education and reassurance; (ii) preventing attacks by avoiding triggers; (iii) the use of non-pharmacological treatments such as relaxation, biofeedback, and life style regulation, such as maintaining a regular schedule, getting adequate sleep and exercise; (iv) acute attack treatment to relieve pain and impairment and stop progression; (v) long-term preventive therapy to reduce attack frequency, severity, and duration; (vi) the use of physical and alternative medicine when appropriate; and (vii) periodic reassessment and reconsideration of the treatment plan.
The pharmacological treatment of migraine may be acute (abortive) or preventive (prophylactic), and patients with frequent severe headaches often require both approaches. Acute treatment attempts to relieve or stop the progression of an attack, or the pain and impairment once an attack has begun. Acute treatment is appropriate for most attacks and should be used a maximum of 2–3 days a week. Preventive therapy is given, even in the absence of a headache, in an attempt to reduce the frequency and severity of anticipated attacks.
Preventive treatment
Preventive medications are usually taken, whether or not headache is present, to reduce the frequency, duration, or severity of attacks. The United States Evidenced Based Guidelines for Migraine (4) has suggested that circumstances that might warrant preventive treatment include: (i) recurring migraine that significantly interferes with the patient's daily routine despite acute treatment. (e.g. two or more attacks a month that produce disability that lasts ≥3 days or headache attacks that are infrequent but produce profound disability); (ii) failure, contraindication to, or troublesome side-effects from acute medications; (iii) overuse of acute medications; (iv) special circumstances, such as hemiplegic migraine or attacks with a risk of permanent neurological injury; (v) very frequent headaches (more than two a week), or a pattern of increasing attacks over time, with the risk of developing rebound headache with acute attack medicines; or (vi) patient preference, i.e. the desire to have as few acute attacks as possible. These rules are stricter during pregnancy: severe disabling attacks accompanied by nausea, vomiting, and possibly dehydration are required for preventive treatment to be prescribed (5).
The major medication groups for preventive migraine treatment (Table 1) include β-adrenergic blockers, anti-depressants, calcium channel antagonists, serotonin antagonists, anti-convulsants, non-steroidal anti-inflammatory drugs (NSAIDs), and others (including riboflavin, minerals, and herbs). If preventive medication is indicated, the agent should be preferentially chosen from one of the first-line categories, based on the drug's side-effect profile and the patient's coexistent and co-morbid conditions (6).
Preventive prescription drugs
∗Very suggestive early clinical data require confirmation.
Principles of preventive therapy
Start the drug at a low dose and increase it slowly until therapeutic effects develop, the ceiling dose for the chosen drug is reached, or side-effects become intolerable.
Give each treatment an adequate trial. A full therapeutic trial may take 2–6 months. In controlled clinical trials, efficacy is often first noted at 4 weeks and continues to increase for 3 months.
Avoid interfering, overused, and contraindicated drugs. To obtain maximal benefit from preventive medication, the patient should not overuse analgesics, opioids, triptans, or ergot derivatives. Re-evaluate therapy: migraine headaches may improve independent of treatment; if the headaches are well controlled, slowly taper and, if possible, discontinue the drug. Many patients experience continued relief with a lower dose of the medication and others may not require it at all.
Be sure that a woman of childbearing potential is aware of any potential risks and pick the medication that will have the least adverse effect on the fetus (5). Women taking valproate can easily add folic acid.
Involve patients in their care to maximize compliance. Take patient preferences into account when deciding between drugs of relatively equivalent efficacy. Discuss the rationale for a particular treatment, when and how to use it, and what side-effects are likely. Address patient expectations. Discuss with the patient the expected benefits of therapy and how long it will take to achieve them.
Consider co-morbidity, which is the presence of two or more disorders whose association is more likely than chance. Conditions that are co-morbid with migraine include stroke, epilepsy, mitral valve prolapse, Raynaud's syndrome, and certain psychological disorders, including depression, mania, anxiety, and panic (Table 2).
Migraine co-morbid disease
β-adrenergic blockers
β-blockers, the most widely used class of drugs in prophylactic migraine treatment, are 60–80% effective in producing a >50% reduction in attack frequency. Rabkin et al. (7) serendipitously discovered propranolol's effectiveness in headache treatment in patients who were being treated for angina (8, 9).
The Agency for Healthcare Policy and Research (AHCPR) Technical Report (10) and the United States Headache Consortium (11) analysed 74 controlled trials of β-blockers for migraine prevention. Propranolol was consistently effective for migraine prevention in a daily dose of 120–240 mg. No absolute correlation has been found between propranolol's dose and its clinical efficacy (12). One meta-analysis revealed that, on average, propranolol yielded a 44% reduction in migraine activity compared with a 14% reduction with placebo. Overall, one of six patients discontinued propranolol treatment (13).
The relative efficacy of the different β-blockers has not been clearly established, and most studies show no significant difference between drugs. One trial comparing propranolol and amitriptyline suggested that propranolol is more efficacious in patients with migraine alone and amitriptyline is superior for patients with the phenotypes of migraine and tension-type headache (14).
Four trials comparing metoprolol with placebo had mixed results (15–18). Metoprolol was similar to propranolol (16, 19–21), flunarizine (22, 23), and pizotifen (24). Timolol (25–27), atenolol (28–30), and nadolol (31–36) are also likely to be beneficial based on comparisons with placebo or with propranolol.
β-blockers with intrinsic sympathomimetic activity (acebutolol, alprenolol, oxprenolol, pindolol) have not been found to be effective for migraine prevention (37–42). The only factor that correlates with the efficacy of β-blockers is the absence of partial agonist activity (16, 19, 43–46).
Mechanism of action
The mechanism of action of β-blockers is not certain, but it appears that their anti-migraine effect is due to inhibition of β1-mediated mechanisms (47). β blockade results in inhibition of norepinephrine release by blocking prejunctional β receptors. In addition, it results in a delayed reduction in tyrosine hydroxylase activity, the rate-limiting step in norepinephrine synthesis, in the superior cervical ganglia. In the rat brainstem, a delayed reduction of the locus ceruleus neurone firing rate has been demonstrated after propranolol administration (47). This could explain the delay in the prophylactic effect of the β-blocker.
The action of β-blockers is probably central and could be mediated by: (i) inhibiting central β receptors interfering with the vigilance-enhancing adrenergic pathway, (ii) interaction with 5-HT receptors (but not all β-blockers bind to the 5-HT receptors), and (iii) cross-modulation of the serotonin system (48, 49).
Schoenen et al. (50) have shown that contingent negative variation (CNV), an event-related slow negative scalp potential, is significantly increased and its habituation reduced in patients with untreated migraine without aura. CNV normalizes after treatment with β-blockers, consistent with central adrenergic hyperactivity in migraine. Migraineurs who have elevated CNV scores have a much better response to β-blocker therapy (80% effective) than migraineurs who have a low or normal score (22% effective), suggesting that the CNV may predict the response to β-blocker treatment (50). Migraineurs exhibit an enhanced centrally mediated secretion of epinephrine after exposure to light (51); this returns to normal after treatment with propranolol.
All β-blockers can produce behavioural side-effects, such as drowsiness, fatigue, lethargy, sleep disorders, nightmares, depression, memory disturbance, and hallucinations, indicating that they all affect the CNS. Adverse events most commonly reported in clinical trials with β-blockers were fatigue, depression, nausea, dizziness, and insomnia. These symptoms appear to be fairly well tolerated and were seldom the cause of premature withdrawal from trials (10). Common side-effects include gastrointestinal complaints and decreased exercise tolerance. Less common are orthostatic hypotension, significant bradycardia, impotence, and aggravation of intrinsic muscle disease. Propranolol has been reported to have an adverse effect on the fetus (52). Congestive heart failure, asthma, and insulin-dependent diabetes are contraindications to the use of non-selective β-blockers. At this time it appears that β-blockers are not absolutely contraindicated in migraine with aura unless a clear stroke risk is present. Whether this includes prolonged aura is uncertain. The reported adverse reactions to propranolol may be either coincidental or idiosyncratic, but the actual risk is uncertain.
Some authors have commented on continued improvement (53) and lack of rebound (54) after discontinuing propranolol. Others have found no carry-over effects (55). However, it seems more reasonable to slowly taper β-blockers, since stopping them abruptly can cause increased headache (17) and the withdrawal symptoms of tachycardia and tremulousness (56).
Propranolol is a non-selective β-blocker with a half-life of 4–6 h. It is also available in an effective long-acting formulation (57, 58). The therapeutically effective dose of propranolol ranges from 40 to 400 mg a day, with no correlation between propranolol and 4-hydroxypropranolol plasma levels and headache relief (59). The short-acting form can be given three to four times a day, although we recommend twice a day, and the long-acting form once or twice a day. Propranolol should be started at a dose of 40 mg a day in divided doses and slowly increased to tolerance. An advantage of the regular propranolol is its greater dosing flexibility. The dose in children is 1–2 mg/kg a day.
Nadolol is a non-selective β-blocker with a long half-life. It is less lipid-soluble than propranolol and has fewer CNS side-effects. The dose ranges from 20 to 160 mg a day given once daily or in split doses. Some authorities prefer it to propranolol since it has fewer side-effects (60).
Timolol is a non-selective β-blocker with a short half-life. The dose ranges from 20 to 60 mg a day in divided doses.
Atenolol is a selective β1-blocker with fewer side-effects than propranolol. The dose ranges from 50 to 200 mg a day once daily.
Metoprolol is a selective β1-blocker with a short half-life. The dose ranges from 100 to 200 mg a day in divided doses. The long-acting preparation may be given once a day.
Anti-depressants
Anti-depressants consist of a number of different classes of drugs with different mechanisms of action. Only tricyclic anti-depressants (TCAs) have proven efficacy in migraine; we cover the newer components for completeness and reader interest.
Clinical trials and use
A total of 16 controlled trials have investigated the efficacy of the TCAs amitriptyline and clomipramine, and the selective serotonin reuptake inhibitors (SSRIs) fluoxetine and fluvoxamine (14, 15, 61–75). Amitriptyline has been more frequently studied than the other agents, and is the only anti-depressant with fairly consistent support for efficacy in migraine prevention. Three placebo-controlled trials found amitriptyline significantly better than placebo at reducing headache index or frequency (64–66, 73). One of these trials, conducted in patients whose headaches were frequently severe or disabling in intensity, found no significant difference between amitriptyline and propranolol (73). Another trial reported that amitriptyline was significantly more efficacious than propranolol for patients with mixed migraine and tension-type headache, while propranolol was significantly better for patients with migraine alone (14). Similarly, a trial conducted in a group of patients with mixed migraine and tension-type headache found that amitriptyline was significantly better than timed-released dihydroergotamine (TR-DHE) at reducing headache index (63). However, an analysis of the data on headache duration, stratified by severity, showed that amitriptyline was significantly better than TR-DHE at reducing the number of hours of moderate and mild tension-type headache-like pain. In contrast, TR-DHE was significantly better than amitriptyline at reducing the number of hours of extremely severe and severe migraine-like pain. The evidence was insufficient to support the efficacy of clomipramine (15, 70) and fluvoxamine (75) for migraine prevention. Fluoxetine was significantly better than placebo in one (61) but not a second (71) migraine prevention trial.
Mechanism of action
TCAs, SSRIs, and serotonin norepinephrine reuptake inhibitors (SNRIs) increase synaptic norepinephrine (NE) or serotonin (5-HT) by inhibiting high-affinity re-uptake. Some are more potent inhibitors of NE, others of 5-HT re-uptake. Monoamine oxidase inhibitors (MAOIs) block the degradation of catecholamines. The most consistent neurochemical finding with anti-depressant treatment (including the TCAs, SSRIs, MAOIs, and electroconvulsive therapy) is a decrease in β-adrenergic receptor density and NE-stimulated cyclic AMP response. Increased α1 receptor system sensitivity is not seen as consistently with anti-depressant treatment. Long-term anti-depressant treatment decreases 5-HT2 receptor-binding and imipramine-binding sites (related to the 5-HT uptake system) but does not change 5-HT1 receptor binding. A strong interaction exists between the NE and 5-HT systems. Anti-depressant treatment β-receptor down-regulation is dependent on an intact 5-HT system, while lesions of the NE system block the decrease in 5-HT2 receptor binding (76). The decrease in 5-HT2 receptor binding sites does not correlate with a decrease in function; in fact, there may be enhanced physiological responsiveness.
TCAs up-regulate the GABA-B receptor, down-regulate the histamine receptor, and enhance the neuronal sensitivity to substance P. Some TCAs are 5-HT2 receptor antagonists. TCAs also interact with endogenous adenosine systems at central and peripheral sites. They inhibit neurogenic uptake of adenosine and augment the electrophysiological actions of adenosine. The enhanced availability of adenosine and activation of adenosine receptors contributes to antinociception. Adenosine A1 receptor activation results in antinociception mediated by inhibition of adenylate cyclase, while adenosine A2 receptor activation is pronociceptive due to stimulation of adenylate cyclase within the sensory nerve terminal (77). Adenosine A3 receptors facilitate pain due to release of histamine and 5-hydroxytryptamine from mast cells (78, 79).
The mechanism by which anti-depressants effect headache prophylaxis is uncertain, but does not result from treating masked depression. Anti-depressants are useful in treating many chronic pain states, including headache, independent of the presence of depression, and the response occurs sooner than the expected anti-depressant effect (80–82). In animal pain models, anti-depressants potentiate the effects of co-administered opioids (83). The anti-depressants that are clinically effective in headache prophylaxis either inhibit noradrenaline and 5-HT re-uptake or are antagonists at the 5-HT2 receptors (84).
Pharmacology of the TCAs
There is wide individual variation in the absorption, distribution, and excretion of the TCAs, with a 10–30-fold variation in individuals' drug metabolism. A therapeutic window may exist above which the TCAs are ineffective, but this has been evaluated only for nortriptyline for treatment of depression. TCAs are lipid-soluble, have a high volume of distribution, and avidly bind to plasma proteins. The anti-histamine and anti-muscarinic activity of the TCAs account for many of their side-effects (85).
The TCAs most commonly used for headache prophylaxis include amitriptyline, nortriptyline, doxepin, and protriptyline. Imipramine and desipramine have been used at times. With the exception of amitriptyline, the TCAs have not been vigorously evaluated; their use is based on anecdotal or uncontrolled reports.
Principles of tricyclic anti-depressant use
The TCA dose range is wide and must be individualized. With the exception of protriptyline, TCAs are sedating. Start with a low dose of the chosen TCA at bedtime, except when using protriptyline, which should be administered in the morning. If the TCA is too sedating, switch from a tertiary TCA (amitriptyline, doxepin) to a secondary TCA (nortriptyline, protriptyline). If a patient develops insomnia or nightmares, give the TCA in the morning. SSRIs can be given as a single dose in the morning, although rigorous evidence for activity in migraine is lacking. They are less sedating than the TCAs and some patients may require a hypnotic for sleep induction. Bipolar patients can become manic on anti-depressants.
Side-effects are common with TCA use. Their adverse effects are due to their interaction with multiple neurotransmitters and their receptors. The anti-muscarinic adverse effects are most common; however, adverse effects related to anti-histaminic activity and α-adrenergic mediation were from cerebral intoxication, but cardiac toxicity and orthostatic hypotension can occur. Anti-muscarinic side-effects include dry mouth, a metallic taste, epigastric distress, constipation, dizziness, mental confusion, tachycardia, palpitations, blurred vision, and urinary retention. Anti-histaminic activity may be responsible for carbohydrate cravings, which contributes to weight gain. Adrenergic activity is responsible for the orthostatic hypotension, reflex tachycardia, and palpitations that patients may experience. Amitriptyline and other TCAs rarely will cause inappropriate secretion of ADH. Any anti-depressant treatment may change depression to hypomania or frank mania (particularly in bipolar patients). Ten percent of patients may develop tremors, and confusion or delirium may occur, particularly in older patients who are more vulnerable to the muscarinic side-effects. Anti-depressant treatments may also reduce the seizure threshold (86), although this is not generally a problem in anti-migraine treatment.
Tertiary amines
Amitriptyline is a tertiary amine tricyclic that is sedating and has anti-muscarinic activity. Patients with coexistent depression are more tolerant and require higher doses of amitriptyline. Start at a dose of 10–25 mg at bedtime. The dose ranges from 10 to 400 mg a day.
Doxepin is a sedating tertiary amine TCA. Start at a dose of 10 mg at bedtime. The dose ranges from 10 to 300 mg a day.
Secondary amines
Nortriptyline is a secondary amine that is less sedating than amitriptyline. Nortriptyline is a major metabolite of amitriptyline. If insomnia develops, give the drug earlier in the day or in divided doses. Start at a dose of 10–25 mg at bedtime. The dose ranges from 10 to 150 mg a day.
Protriptyline is a secondary amine similar to nortriptyline. Start at a dose of 5 mg a day. The dose ranges from 5 to 60 mg a day.
Monoamine re-uptake inhibitors
Selective serotonin re-uptake inhibitors
Evidence for the use of SSRIs is poor. They may be helpful in patients with co-morbid depression because their tolerability profile is superior to tricyclics. Fluoxetine, fluvoxamine, paroxetine, sertraline, and citalopram are specific SSRIs that have minimal anti-histaminic and anti-muscarinic activity. These drugs produce less weight gain (and in some cases weight loss) and have fewer cardiovascular side-effects than the TCAs (87). The most common side-effects include anxiety, nervousness, insomnia, drowsiness, fatigue, tremor, sweating, anorexia, nausea, vomiting, and dizziness or light-headedness. Headache was noted in 20.3% of patients on fluoxetine; however, it was also noted in 19.9% of patients on placebo (88). The combination of an SSRI and a TCA can be beneficial in treating refractory depression (89) and, in our experience, resistant cases of migraine. The combination may require dose adjustment of the TCA because levels may significantly increase.
The efficacy analysis summarized in the AHCPR Evidence Report did not indicate a clear benefit of the racemic mixture of fluoxetine over placebo. In contrast, a recent randomized controlled trial of S-fluoxetine indicated a possible clinical benefit in migraine prevention, as measured by a reduction in migraine frequency, as early as 1 month after initiation of therapy (90). Anecdotal reports (91) and our experience seem to indicate its benefit in migraine prophylaxis where coexistent depression is a prominent issue. Some researchers have reported that fluoxetine does not improve or may worsen headache (92). A recent single-centre, randomized, double-blind, parallel study of fluoxetine for the prophylactic control of migraine consisted of two phases: 30 days of pharmacological washout and 6 months of therapy and monthly follow-up (93). A comparison of the total pain index between basal values (calculated during the period of washout) and monthly follow-up (calculated monthly during the period of 6 months of the therapy) showed a significant reduction (P<0.05) beginning from the third month of treatment in the fluoxetine group and no significant reduction in the placebo group.
Fluoxetine: start at a dose of 10 mg in the morning. The dose ranges from 10 to 80 mg a day.
Monoamine oxidase inhibitors
MAOIs exists in two subtypes: MAO-A, which preferentially deaminates NE and 5-HT, and MAO-B, which preferentially deaminates dopamine. Phenelzine is a non-selective inhibitor of MAO-A and MAO-B. L-Deprenyl is a selective MAO-B inhibitor that may be effective in the treatment of Parkinson's disease.
The MAOI phenelzine at a dose of 15 mg TID was shown to be effective (in an open study) (94), but no placebo-controlled, double-blind trials exist. The dose of phenelzine ranges from 30 mg to 90 mg a day in divided doses. All patients on MAOI-A must be on a restricted diet and avoid certain medications to prevent hypertensive crisis. Meperidine, sympathomimetics (including Midrin), alcohol, and foods with a high tyramine content (cheddar cheese, fava beans, banana peel, tap beers, Marmite and Veggie-Mite concentrated, yeast extract, sauerkraut, soy sauce, and other soybean condiments) must be avoided (95–100). Assistance from a pharmacist when using MAOIs may be desirable to avoid drug interaction.
The most common side-effects of MAOIs include insomnia, orthostatic hypotension, constipation, increased perspiration, weight gain, peripheral oedema, and, less commonly, inhibition of ejaculation or reduced libido. Insomnia can be reduced by giving most of the medication early in the day. The risk of hypertensive crisis may be reduced by having the patient take the MAOI 3–4 h before or after eating or taking the entire dose at bedtime, as gut MAO activity rapidly returns to normal (95). Sublingual nifedipine has been used to treat hypertensive crisis when it occurs in MAOI users (101).
Calcium channel antagonists
Calcium, in combination with a calcium-binding protein such as calmodulin or troponin, regulates many functions, including muscle contraction, neurotransmitter and hormone release, and enzyme activity. Its extracellular concentration is high; its intracellular free concentration is 10 000-fold smaller. The concentration gradient is established by membrane pumps and the intracellular sequestering of free calcium. When stimulated, the cell can open calcium channels in the plasma membrane or release intracellular stores of calcium (102).
Channel types and classes
Two types of calcium channels exist: calcium entry channels, which allow extracellular calcium to enter the cell, and calcium release channels, which allow intracellular calcium (in storage sites in organelles) to enter the cytoplasm. They include ryanodine and inositol 1,4,5-triphosphate receptors (103). Calcium entry channel subtypes include voltage-gated opened by depolarization, ligand-gated opened by chemical messengers, such as glutamate, and capacitative activated by depletion of intracellular calcium stores.
There are six functional subclasses of voltage gated calcium (Ca2+) channels that are named T, L, N, P, Q, and R. They fall into two major categories: high-voltage activated channels and the unique low-voltage activated T-type, which is activated at negative potentials (104).
Voltage-gated calcium channels are heteromers containing protein subunits of about 30–230 kDa that form calcium-selective pores across the cell membrane. These subunits have different functions, and subunit isoforms give rise to distinct channel subtypes. The α1 subunit, which consists of four homologous domains of six transmembrane segments each, forms the transmembrane pore and contains most of the channel's known drug-binding sites. Molecular cloning has revealed at least six α1 genes, which are associated with auxiliary subunits, including a membrane-spanning α2-δ complex that increases the amplitude of calcium currents and binds the anti-convulsant gabapentin, and a cytoplasmic β subunit that modifies the channel's current amplitude, voltage dependence, and activation and inactivation properties (103).
The three major classes of L-type Ca2+ channel blockers are the dihydropyridines (e.g. nifedipine), benzothiazepines (e.g. diltiazem), and phenylalkylamines (e.g. verapamil). Regions of the α1 subunit contain the binding sites for all these drugs (104).
Mechanism of action in migraine
The mechanism of action of the calcium channel antagonists in migraine prevention is uncertain. They were introduced into the treatment of migraine on the assumption that they prevent hypoxia of cerebral neurones, contraction of vascular smooth muscles, and inhibition of the Ca2+-dependent enzymes involved in prostaglandin formation. Perhaps it is their ability to block 5-HT release, interfere with neurovascular inflammation, or interfere with the initiation and propagation of spreading depression that is critical (105). The discovery that an abnormality in an α1a subunit (P/Q channel) can produce familial hemiplegic migraine (106) has led to a search for more fundamental associations.
Clinical trials
The AHCPR Technical Report identified 45 controlled trials of calcium antagonists, including flunarizine (25 trials), nimodipine (11 trials), nifedipine (five trials), verapamil (three trials), cyclandelate (three trials), and nicardipine (one trial) (10). Flunarizine was compared with placebo in eight migraine prevention trials and effect sizes could be calculated for seven studies (107–113) but not the eighth study (114). A meta-analysis of these seven heterogeneous trials was statistically significant in favour of flunarizine. Five comparisons of flunarizine with propranolol (115, 116), and two with metoprolol (23, 117), showed no significant differences between flunarizine and these β-blocking agents. There were no significant differences between flunarizine and pizotifen (118–120), or between flunarizine and methysergide (121). One trial comparing flunarizine and dihydroergokryptine (122) (DEK) reported mixed results, but suggested that differences in the effects of the two treatments were small. Nimodipine had mixed results in placebo-controlled trials. Three placebo-controlled studies suggested no significant differences (123, 124), while two reported relatively large and statistically significant differences in favour of nimodipine (123, 125). Nimodipine was not different from flunarizine (126), pizotifen (118–120) or propranolol (127). Our interpretation and our clinical experience is that nimodipine is ineffective in migraine prevention.
The evidence for nifedipine was difficult to interpret. Two comparisons with placebo yielded similar effect sizes that were statistically insignificant, but the 95% confidence intervals associated with these estimates were large and did not exclude either a clinically important benefit or harm associated with nifedipine (128, 129). Similarly ambiguous results were reported in one comparison with flunarizine (130) and two comparisons with propranolol (20, 131). One trial found that metoprolol was significantly better than nifedipine at reducing headache frequency (20). Our conclusion is that nifedipine is ineffective as a migraine preventive.
Verapamil was more effective than placebo in two of three trials, but both positive trials had high dropout rates, rendering the findings uncertain (46, 132). The single negative placebo-controlled trial included a propranolol treatment arm. This trial reported no significant difference between verapamil, propranolol, and placebo (46, 133). Our conclusion is that there is no rigorous, randomized, controlled trial evidence to support the use of verapamil in migraine.
Diltiazem (60–90 mg QID) was effective in two small open studies (134, 135). These studies are insufficient to recommend the use of diltiazem.
Side-effects of the Ca2+ antagonists are dependent on the drug, and include dizziness and headache (particularly with nifedipine), depression, vasomotor changes, tremor, gastrointestinal complaints (including constipation), peripheral oedema, orthostatic hypotension, and bradycardia. Patients frequently report an initial increase in headache. Headache improvement frequently requires weeks of treatment. Adverse events most commonly associated with flunarizine were sedation, weight gain, and abdominal pain. Symptoms reported with other calcium channel antagonists included dizziness, oedema, flushing, and constipation. Two trials of verapamil and one of nifedipine reported high dropout rates due to adverse events (10). Side-effects with nifedipine were frequent (54%) and included dizziness, oedema, flushing, headache, and mental symptoms (128).
Verapamil is available as a 40, 80, or 120-mg tablet or as a 120, 180, or 240-mg sustained-release preparation. Start at a dose of 80 mg two to three times a day, with a maximum of 640 mg a day in divided doses. The sustained-release preparation of verapamil can be given once or twice a day, but unreliable absorption reduces reliability. The most common side-effect is constipation; dizziness, nausea, hypotension, headache, and oedema are less common. Bioavailability is 20%. The absorbed drug is tightly protein bound. Peak plasma levels occur in 5 h; the half-life ranges from 2.5 h to 7.5 h.
Flunarizine is not available in the USA. The dose is 5–10 mg a day. The most prominent side-effects include weight gain, somnolence, dry mouth, dizziness, hypotension, occasional extrapyramidal reactions, and exacerbation of depression. The elimination half-life of flunarizine is 19 days.
Anti-convulsants
Anti-convulsant medication is increasingly recommended for migraine prevention because of placebo-controlled, double-blind trials that prove it to be effective. With the exception of valproic acid and phenobarbital, many anti-convulsants interfere with the efficacy of contraceptives (136, 137).
Nine controlled trials of five different anti-convulsants were included in the AHCPR Technical Report (138–144).
Valproic acid possesses anti-convulsant activity in a wide variety of experimental epilepsy models. Valproate at high concentrations increases GABA levels in synaptosomes, perhaps by inhibiting its degradation; it enhances the post-synaptic response to GABA; and, at lower concentrations, it increases potassium conductance, producing neuronal hyperpolarization. Valproate turns off the firing of the 5-HT neurones of the dorsal raphe, which are implicated in controlling head pain.
Disordered GABA metabolism during migraine has been reported (145). Imbalance in the plasma concentrations of GABA, an inhibitory amino acid, and glutamic acid, an excitatory amino acid, has also been observed (140–144, 146–148).
The mechanism of action of valproate in migraine prophylaxis may be related to facilitation of GABAergic neurotransmission (149–151). Valproate enhances GABA activity within the brain by inhibiting its degradation, stimulating its synthesis and release, and directly enhancing its post-synaptic effects. The concentration of valproate required to inhibit GABA transaminase is greater than that which occurs during therapy. However, active metabolites, one of which (2-en-valproic acid) accumulates in the brain, have an anti-convulsant effect and cause GABA accumulation in vivo (149). Other potential mechanisms of action include direct effects on neuronal membranes (it suppresses induced and spontaneous epileptiform activity), inhibition of kindling, and reduction of excitatory neurotransmission by the amino acid aspartate by blocking its release (150, 151). Valproate also attenuates plasma extravasation in the Moskowitz model of neurogenic inflammation (NI) by interacting with the GABAA receptor. The relevant receptor may be on the parasympathetic nerve fibres projecting from the sphenopalatine ganglia; in so doing, it attenuates nociceptive neurotransmission (149). In addition, valproate-induced increased central enhancement of GABAA activity may enhance central antinociception (152). Valproate also interacts with the central 5-HT system and it reduces the firing rate of midbrain serotonergic neurones (152).
Five studies provided strong and consistent support for the efficacy of divalproex sodium (140, 141, 153) and sodium valproate (138, 147). Two controlled trials of each of these agents showed them to be significantly better than placebo at reducing headache frequency (138, 141, 142, 147). A single study, reported in abstract form only, compared divalproex sodium with propranolol and found differences favouring divalproex sodium; however, the statistical significance of these results could not be determined (open-label study with high dropout rates) (140). A more recent study (published after December 1996 and therefore not included in the AHCPR Technical Report) found divalproex sodium more effective compared with placebo, but not significantly different compared with propranolol, for prevention of migraine in patients without aura (148). An extended release form of divalproex sodium demonstrated comparable efficacy to the tablet formulation (154).
Valproic acid is a simple 8 carbon, 2 chain fatty acid with 80% bioavailability after oral administration. It is highly protein-bound, with an elimination half-life between 8 h and 17 h.
Nausea, vomiting and gastrointestinal distress are the most common side-effects of valproate therapy. These are generally self-limited and are slightly less common with divalproex sodium than with sodium valproate. When the therapy is continued, the incidence of gastrointestinal symptoms decreases, particularly after 6 months. The fourth trial found significantly higher rates of nausea, asthenia, somnolence, vomiting, tremor, and alopecia when patients used divalproex sodium.
In an open-label study, Silberstein et al. (155) evaluated the long-term safety of divalproex sodium in patients who had completed one of two previous double-blind, placebo-controlled studies evaluating the safety and efficacy of divalproex in migraine prophylaxis. Of 163 patients enrolled, 46 had been treated with placebo and 117 had been treated with divalproex. The results, including data from the double-blind study, represented 198 patient-years of divalproex exposure. The average dose was 974 mg/day. Reasons for premature discontinuation (67%) included administrative problems (31%), drug intolerance (21%), and treatment ineffectiveness (15%). The most frequently reported adverse events were nausea (42%), infection (39%), alopecia (31%), tremor (28%), asthenia (25%), dyspepsia (25%), and somnolence (25%). Divalproex was found to be safe and initial improvements were maintained for periods of >1080 days. No unexpected adverse events or safety concerns unique to the use of divalproex in the prophylactic treatment of migraine were found.
Valproate has little effect on cognitive functions and it rarely causes sedation. On rare occasions, valproate administration is associated with severe adverse reactions, such as hepatitis or pancreatitis. The frequency varies with the number of concomitant medications used, the patient's age, the presence of genetic and metabolic disorders, and the patient's general state of health. These idiosyncratic reactions are unpredictable (156).
The risk of valproate hepatotoxicity is highest in children under the age of 2 years, especially those treated with multiple anti-epileptic drugs, those with metabolic disorders, and those with severe epilepsy accompanied by mental retardation and organic brain disease (157). The relative risk of hepatotoxicity from valproate is low in migraineurs. Valproate is potentially teratogenic and should not be used in pregnant women or women considering pregnancy (158). Hyperandrogenism, resulting from elevated testosterone levels, ovarian cysts, and obesity, is of particular concern in young women with epilepsy who use valproate (159). It is uncertain if valproate can cause these symptoms in young women with migraine or mania.
Because of valproate's potential idiosyncratic interactions with barbiturates (severe sedation, coma), migraine patients who are on valproate should not be given barbiturate-containing combination analgesics for symptomatic headache relief. If these drugs are used, they should be given with caution and at a low dose. Absolute contraindications to valproate are pregnancy and a history of pancreatitis or a hepatic disorder such as chronic hepatitis or cirrhosis of the liver. Other important contraindications are haematological disorders, including thrombocytopenia, pancytopenia and bleeding disorders.
Valproic acid is available as 250-mg capsules and as a syrup (250 mg/5 ml). Divalproex sodium is a stable co-ordination complex comprised of sodium valproate and valproic acid in a 1 : 1 molar ratio. An enteric-coated form of divalproex sodium is available as 125, 250, and 500-mg capsules and a sprinkle formulation. Start with 250–500 mg a day in divided doses and slowly increase the dose. Monitor serum levels if there is a question of toxicity or compliance. (The usual therapeutic level is from 50 to 100 mg/ml.) The maximum recommended dose is 60 mg/kg per day. An extended release form of divalproex sodium demonstrated comparable efficacy to the tablet formulation. The adverse effect profile in the clinical trial, however, showed almost identical adverse effect rates for the placebo and active treatment arms (160).
Gabapentin (600–1800 mg) was effective in episodic migraine and chronic migraine in a 12-week open-label study (161). Gabapentin was not effective in one placebo-controlled, double-blind study (162), but a more recent trial reported clinical efficacy for gabapentin in migraine prevention (163). Gabapentin 1800–2400 mg was superior to placebo in reducing the frequency of migraine attacks. The responder rate (50% decrease in attack frequency) was 36% for gabapentin and 14% for placebo (P=0.02). The two treatment groups were comparable with respect to treatment-limiting adverse events.
The most common adverse events reported in association with the gabapentin were dizziness or giddiness and drowsiness. Relatively high patient withdrawal rates due to adverse events were reported in some trials (10).
Topiramate is a structurally unique anti-convulsant that was discovered serendipitously. It was originally synthesized as part of a research project to discover structural analogues of fructose-1,6-diphosphate capable of inhibiting the enzyme fructose 1,6-bisphosphatase, thereby blocking gluconeogenesis, but it has not to date demonstrated clinical evidence of hypoglycaemic activity. Topiramate is a derivative of the naturally occurring monosaccharide D-fructose and contains a sulfamate moiety. Topiramate is rapidly and almost completely absorbed. In humans, the blood plasma concentration increases linearly as a function of dose over the pharmacologically relevant range (164, 165). It is not extensively metabolized and is eliminated predominantly unchanged in the urine. The average elimination half-life is approximately 21 h (165).
The anti-convulsant activity of most anti-epileptic drugs is thought to be due to a state-dependent blockade of voltage-dependent Na+ or Ca2+ channels, or an ability to enhance the activity of γ-aminobutyrate (GABA) at GABAA receptors (166, 167). Topiramate can influence the activity of some types of voltage-activated Na+ and Ca2+ channels, GABAA receptors, and the α-amino-3-hydroxy-5-methylisoxazole-4-proprionic acid (AMPA)/kainate subtype of glutamate receptors. Topiramate also inhibits some isozymes of carbonic anhydrase (CA) and exhibits selectivity for CA II and CA IV (168, 169).
Topiramate has been associated with weight loss, not weight gain (a common reason to discontinue preventive medication) with chronic use. Edwards et al. (170) enrolled 30 migraineurs in a placebo-controlled, double-blind prevention trial of topiramate. There was a 4-week baseline screening phase followed by a 6-week titration of 200 mg qd of topiramate in divided doses or placebo. This was followed by a one-half week steady state. Percent responders (patients with ≥50% reduction in 28-day migraine headache frequency) during the 18-week double-blind phase were topiramate 46.7% and placebo 6.7% (P=0.035).
Potter et al. (171) evaluated the efficacy and safety of topiramate in migraine prophylaxis. Forty patients who were 18–65 years of age and had migraine with or without aura were randomized to topiramate or placebo (1 : 1 ratio). The study duration was 20 weeks (baseline, titration, and maintenance phase of 4, 8, and 8 weeks). Nineteen patients were randomized to topiramate and 21 to placebo. Thirty-five patients completed the study. The mean topiramate dose was 125 mg (range 25–200 mg). The baseline 28-day headache frequency was 5.14 in topiramate patients and 4.37 in placebo patients. The mean 28-day headache rate over the entire double-blind phase was 3.31 (topiramate) vs. 3.83 (placebo; P=0.0025). The mean reduction in headache rate was 1.83 (topiramate) vs. 0.55 (placebo; P=0.0025). The median percent reduction in monthly headache rate was 33% (topiramate) vs. 8% (placebo; P=0.0061). Large-scale placebo-controlled studies are underway.
Serotonin antagonists
The anti-serotonin migraine-preventive drugs are potent 5-HT2B and 5-HT2C receptor antagonists, while mCPP, a 5-HT2B and 5-HT2C receptor agonist, induces migraine in susceptible individuals (172–175). Methysergide, cyproheptadine, and pizotifen, effective migraine prophylactic drugs, are 5-HT2B and 5-HT2C receptor antagonists, while ketanserin, a selective 5-HT2A and a poor 5-HT2B and 5-HT2C receptor antagonist, is not (176–178).
mCPP, a major metabolite of the anti-depressants trazadone and nefazodone, induces migraine hours after the immediate pharmacological response to the drug (monitored by elevation of plasma cortisol and prolactin) is over. Gordon et al. (173) found that mCPP induced headache in both migraineurs (five of eight) and in non-migraine controls (four of 10). No significant differences were found between the migraineurs and normal subjects in terms of their neuroendocrine or headache responses to mCPP, but there were highly significant associations between the cortisol responses and headache severity and duration.
Pizotifen and methylergometrine are potent rabbit jugular vein endothelial cell 5-HT2 receptor antagonists. 5-HT2B or 5-HT2C receptor activation by mCPP or endogenously released 5-HT could dilate cerebral vessels. Vasodilation, however, is neither necessary nor sufficient to cause headache (179, 180), but endothelium-derived nitric oxide (NO) can activate sensory trigeminovascular fibres resulting in calcitonin gene-related peptide (CGRP) release, which mediates pial artery vasodilation (181) and neurogenic inflammation (178, 180, 182). mCPP itself can produce extravasation in the dural membrane, which can be blocked by selective 5-HT2B antagonists (183).
Methysergide is also a 5-HT1 receptor agonist (184), but has lower affinity for the 5-HT1 than for the 5-HT2 binding site (185). Methysergide-induced contraction of dog isolated saphenous vein is also mediated by 5-HT1B receptors (186–188). Chronic, but not acute, treatment with methysergide attenuates dural plasma extravasation following electric stimulation of the rat trigeminal ganglion in the Moskowitz model (189). The difference between acute and chronic drug administration could be due to the accumulation of the active metabolite, methylergometrine. Methysergide (or methylergometrine) presynaptically could inhibit the release of CGRP from perivascular sensory nerves.
Methysergide
Methysergide is a semisynthetic ergot alkaloid that is structurally related to methylergonovine. It is a 5-HT2 receptor antagonist and 5-HT1B/D agonist. It was probably the first drug developed for migraine prevention (190), but its usefulness is limited by reports of retroperitoneal and retropleural fibrosis associated with long-term, mostly uninterrupted, administration (191).
The AHCPR Technical Report identified 17 controlled trials of methysergide for migraine prevention (19, 121, 192–203). Four placebo-controlled trials suggested that methysergide was significantly better than placebo at reducing headache frequency (198, 199, 201, 202).
Four comparison trials showed no statistically significant differences between methysergide and pizotifen (194, 196, 200, 201). Two trials that directly compared methysergide and propranolol failed to demonstrate any statistically significant differences between these treatments (19, 192). The only trial that compared methysergide with metoprolol reported an unusually low response to metoprolol (6%) and thus a misleading relative increase in methysergide efficacy (192).
Methysergide was associated with a higher incidence of adverse events than was placebo. Gastrointestinal complaints were most common and included nausea, vomiting, abdominal pain, and diarrhoea. Also frequently reported were leg symptoms (restlessness or pain), dizziness, giddiness, drowsiness, lassitude, and paraesthesia. Adverse events were no more common with methysergide than with pizotifen. The duration of the trials reviewed here was too short to detect the fibrotic complications sometimes observed with long-term methysergide use.
Side-effects of methysergide noted in clinical practice include transient muscle aching, claudication, abdominal distress, leg cramps, hair loss, nausea, weight gain, and hallucinations. Frightening hallucinatory experiences after the first dose are not uncommon (204). Curran and Lance (205) have treated leg claudication with vasodilators with some enhancement of methysergide's effectiveness, suggesting that its action on headache is not a result of vasoconstriction. The major complication of methysergide is the rare (1/5000) development of retroperitoneal, pulmonary, or endocardial fibrosis (206, 207).
Methysergide is indicated for the treatment of migraine. The dose ranges from 2 to 8 mg a day, with the higher doses being given two or three times a day. Many clinicians use higher doses, up to 14 mg a day, without adverse events and with higher efficacy (208). To minimize early adverse events, patients can start with a dose of 1 mg a day and increase the dose gradually by 1 mg every 2–3 days. (This can be accomplished by breaking the 2-mg tablets if 1-mg tablets are not available.) A trial of at least 2 months should be given, with doses up to at least 8 mg a day. Methysergide, in general, should not be taken continuously for long periods, since doing so may produce retroperitoneal fibrosis (191, 206, 209). Instead, the drug should be given for 6 months, stopped for 1 month, and then restarted. To avoid an increase in headache when methysergide is stopped, the patient should be weaned off the drug over a 1-week period. Some authorities use methysergide on a continuous basis with careful monitoring (208), which includes auscultation of the heart and yearly echocardiography, chest X-ray, and abdominal magnetic resonance imaging. The drug should be discontinued immediately on suspicion of pulmonary, cardiac, or retroperitoneal fibrosis (208).
Contraindications to methysergide use include pregnancy, peripheral vascular disorders, severe arteriosclerosis, coronary artery disease, severe hypertension, thrombophlebitis or cellulitis of the legs, peptic ulcer disease, fibrotic disorders, lung diseases, collagen disease, liver or renal function impairment, valvular heart disease, debilitation, or serious infection. Patients who receive methysergide should remain under the supervision of the treating physician and be examined regularly for development of the rare pulmonary/cardiac or retroperitoneal fibrosis or the development of vascular complications.
Methysergide is an effective migraine preventive medication that is an appropriate consideration in resistant headaches with a high attack frequency. All of the open and controlled studies attest to its efficacy. In addition to being effective in reducing attack frequency, it often acts synergistically with ergotamine for breakthrough attacks. Due to its side-effect profile, it should be reserved for severe cases in which other migraine-preventive drugs are not effective.
Cyproheptadine, an antagonist at the 5-HT2, histamine H1, and muscarinic cholinergic receptors, is widely used in the prophylactic treatment of migraine in children (95, 210, 211). Curran and Lance (205) found cyproheptadine more effective than placebo but less effective than methysergide. Cyproheptadine is available as 4-mg tablets. The total dose ranges from 12 to 36 mg a day (given two to three times a day or at bedtime). Common side-effects are sedation and weight gain; dry mouth, nausea, light-headedness, ankle oedema, aching legs, and diarrhoea are less common. Cyproheptadine may inhibit growth in children (212) and reverse the effects of SSRIs.
Pizotifen, a 5-HT2 receptor antagonist structurally similar to cyproheptadine, is not available in the USA. The United States Headache Consortium guidelines (11) found that evidence was inconsistent for the efficacy of pizotifen from 11 placebo-controlled trials (201, 222) and 19 comparisons with other agents (24, 118–120, 153, 194, 196, 200, 201, 213, 220, 223–230). Analysis of the placebo-controlled trials suggested a large clinical effect that was statistically significant. In direct comparisons with other agents known to be efficacious for migraine prevention, no significant differences were demonstrated between pizotifen and flunarizine (118–120), methysergide (194, 196, 200, 201), naproxen sodium (213), or metoprolol (24). However, in the 26 trials reviewed, pizotifen was generally poorly tolerated (10). Substantial weight gain, tiredness, and/or drowsiness were frequently reported. Pizotifen was associated with a high rate of withdrawals due to adverse events. Controlled and uncontrolled studies in Europe (231) have shown this drug to be of benefit in 40–79% of patients. The dose recommendation is 0.5–1 mg, one to three times daily by titration. Side-effects include drowsiness and weight gain (232).
Other drugs
Feverfew (Tanacetum parthenium)
Feverfew is a medicinal herb used in self-treatment of migraine. Four trials were conducted: two that were reported in the AHCPR and two that were conducted after the report was published. The AHCPR Technical Report listed two trials, distinctly different in design, that compared feverfew with placebo and no treatment. One trial was conducted in a self-selected group of feverfew users and showed that withdrawing feverfew led to a statistically significant increase in headache frequency (233). A pilot study of 17 migraineurs who ate fresh feverfew leaves daily was undertaken at the City of London Migraine Clinic. Patients were given capsules of freeze-dried feverfew or placebo. Those receiving placebo had a tripling in the frequency of migraine attacks. Patients on placebo reported increased nervousness, tension headaches, insomnia, or joint stiffness constituting a ‘post feverfew syndrome’ (perhaps another example of rebound).
The other, more conventional, trial was conducted in a larger group of migraineurs, most of whom (71%) had never used feverfew (234). This trial reported a smaller difference between feverfew and the control treatment than did the other trial, but still found the difference to be statistically significant in favour of feverfew (P<0.005). Two trials were not included in the AHCPR report. One was a double-blind, randomized, crossover trial that tested the efficacy of feverfew compared with placebo, and reported that treatment with feverfew was associated with a significant reduction in pain intensity and non-headache symptoms (nausea, vomiting, photophobia, and phonophobia) (235). The other trial reported no significant differences between feverfew given as an alcoholic extract and placebo for reducing migraine frequency (236).
Limited information indicates that adverse events were no more common with feverfew than with the control treatment (10). Feverfew's side-effects include mouth ulceration and a more widespread oral inflammation associated with loss of taste. Feverfew's mechanism of action is uncertain. It is rich in sesquiterpene lactones, especially parthenolide, which may be a non-specific NE, 5-HT, bradykinin, prostaglandin, and acetylcholine antagonist. The biological variation in the sesquiterpene lactone content and the long-term safety and effectiveness of feverfew are of concern (233).
Until recently it was generally assumed that parthenolides represent the active principle of feverfew. This hypothesis was supported by in vitro experiments that emphasized its biological activity. These studies have demonstrated that the plant has inhibitory effects on platelet aggregation and release of serotonin from blood platelets and leucocytes (237, 238). One trial that used feverfew extract with a standardized and constant concentration of parthenolides to treat migraine did not show any beneficial effect (236). Thus, the clinical effectiveness of feverfew for migraine prevention has not been established beyond reasonable doubt. More clinical trials are needed, both on a larger scale and with various feverfew extracts, including parthenolide-free sesquiterpene lactone chemotypes (239).
Riboflavin
A mitochondrial dysfunction resulting in impaired oxygen metabolism may play a role in migraine pathogenesis (240–243). Riboflavin (vitamin B2) is the precursor of flavin mononucleotide and flavin adenine dinucleotide, which are required for the activity of flavoenzymes involved in the electron transport chain. Given to patients with mitochondrial encephalopathy lactic acidosis and stroke (MELAS) or mitochondrial myopathies on the assumption that large doses might augment activity of mitochondrial complexes I and II, riboflavin improved clinical as well as biochemical parameters (244–247).
Based on the results of an open trial in migraine, a placebo-controlled, double-blind trial of high dose of vitamin B2 (400 mg) was performed and showed significant benefit (248). Schoenen et al. (248) compared riboflavin (400 mg) with placebo in migraineurs in a randomized trial of 3 months duration. Riboflavin was significantly superior to placebo in reducing the attack frequency (P=0.005), headache days (P=0.012), and migraine index (0.012). The proportion of patients who improved by at least 50% in headache days, i.e. ‘responders’, was 15% for placebo and 59% for riboflavin (P=0.002), and the number-needed-to-treat for effectiveness was 2.3. Only three adverse events occurred: two in the riboflavin group (diarrhoea and polyuria) and one in the placebo group (abdominal cramps). None was serious.
Magnesium supplementation was effective in one of two trials. One study enrolled 81 patients who had IHS migraine. Patients received 600 mg (24 mmol) of oral magnesium (trimagnesium dicitrate) or placebo daily for 12 weeks. In weeks 9–12 the attack frequency was reduced by 41.6% in the magnesium group and by 15.8% in the placebo group compared with the baseline (P<0.05). The number of days with migraine and symptomatic drug consumption also decreased significantly in the magnesium group. Adverse events were diarrhoea (18.6%) and gastric irritation (4.7%) (249).
In another multicentre, prospective, randomized, double-blind, placebo-controlled study, 20 mmol magnesium-L-aspartate-hydrochloride trihydrate given in divided doses was evaluated. An interim analysis was performed with 69 patients (64 women, five men). Of these, 35 had received magnesium and 34 placebo. There were 10 responders in each group (28.6% magnesium and 29.4% placebo). There was no benefit from magnesium compared with placebo in the number of migraine days or migraine attacks (250).
The studies differed in the amount of magnesium (24 mmol vs. 20 mmol) and in the salt (dicitrate vs. aspartate). Those differences may produce differences in bioavailability and efficacy and account for the reported difference.
Aspirin
O'Neill and Mann (251) and Masel et al. (252) found that aspirin (ASA) (650 mg a day) decreased headache frequency. Two major multicentre trials, however, clearly proved the efficacy of ASA in the prophylaxis of migraine: in 1988 ‘The British Physician Trial’ showed that a daily dose of 500 mg ASA reduced the frequency of migraine by an average of 30% (253). In a double-blind trial of low-dose ASA (325 mg every other day) in 22 071 United States male physicians (Physician Health Study), Buring et al. (254) found a 20% reduction in headache frequency. Although this is statistically significant, it may not be clinically significant. In a small open trial, Baldratti et al. (255) compared the efficacy of ASA (13.5 mg/kg) with propranolol (1.8 mg/kg). In this trial, both drugs were equally effective and reduced the frequency, duration, and intensity of attacks to the same extent. In a double-blind crossover trial, ASA (500 mg daily) was statistically less effective than 200 mg propranolol daily (22). High-dose ASA use may lead to overuse and the development of rebound headaches, although, in practice, ASA is usually implicated with other compounds in combination analgesics. Aspirin in low doses is indicated for the prophylaxis of myocardial infarction and transient ischaemic attacks. We would use aspirin only in patients with prolonged or non-visual aura.
Non-steroidal anti-inflammatory drugs
Some NSAIDs may be effective in migraine prophylaxis. These include sodium naproxen, fenoprofen, ketoprofen, and tolfenamic acid (256). Some headache disorders (paroxysmal hemicrania and hemicrania continua) are defined by their responsiveness to indomethacin (257–259). Although NSAIDs are effective, they must be used with caution because of their adverse effects on gastrointestinal and renal function (260).
α Agonists
The United States Headache Consortium guidelines and the AHCPR Technical Review included 17 controlled trials of α2 agonists for the prevention of migraine: 16 of clonidine (153, 261–275), and one of guanfacine (272). The evidence from these trials suggests that α2 agonists are minimally, and not conclusively, efficacious. Three of 11 placebo-controlled trials of clonidine found a significant difference in favour of the active agent, but the magnitude of the effect was small (265, 269, 270). Two comparative trials comparing clonidine with the β-blockers metoprolol (273) and propranolol yielded mixed results (274). Two comparative trials showed no significant differences among clonidine, practolol (265) and pindolol (275). One trial each found no significant differences between clonidine and pizotifen (153), or between clonidine and carbamazepine (275).
Clonidine's most commonly reported adverse events were drowsiness and tiredness. In studies comparing clonidine with β-blockers, adverse events occurred at similar rates for both interventions.
Newer treatments
Botulinum toxin type A (BotoxR)
Silberstein et al. (276) evaluated the safety and efficacy of pericranial Botulinum toxin type A injections as prophylactic treatment of chronic moderate-to-severe migraine. One hundred and twenty-three patients who had chronic IHS-defined migraine and a history of two to eight moderate-to-severe migraine attacks during a 1-month baseline were randomized to treatment with either 0, 25, or 75 U of Botulinum toxin type A injected symmetrically into glabellar, frontalis, and temporalis muscles. Diaries were kept for three months post-injection. At 12 centres, 41, 42, and 40 patients were randomized to 0, 25, and 75 U Botulinum toxin type A treatment groups and had baseline frequencies of migraine of 4.41, 4.45, and 3.95 attacks per month, respectively. The 25 U Botulinum toxin type A treatment group fared significantly better than the placebo group by the following measures: reduction in mean frequency of moderate-to-severe migraines during days 31–60; incidence of 50% reduction and incidence of two headache decrease in mild-to-severe migraines at days 61–90; and improvement by patient global assessment for days 31–60 post-injection. The 75 U Botulinum toxin type A treatment group was significantly better than the placebo group on patient global assessment for days 31–60, but not other parameters. Botulinum toxin type A treatment was well tolerated, but high-dose Botulinum toxin type A showed significantly more treatment-related adverse events than placebo. No serious treatment-related adverse events were reported. Pericranial injection of Botulinum toxin type A (25 U) showed significant differences compared with vehicle in reducing migraine frequency and associated symptoms during the 90 days following injection. Further studies are currently underway. Their results will allow us to ascertain the place of Botox in migraine prevention.
Setting treatment priorities (Table 3)
Preventive drugs
The goals of treatment are to relieve or prevent the pain and associated symptoms of migraine and to optimize the patient's ability to function normally. The medications used to treat migraine can be divided into four major categories: (i) drugs that have documented high efficacy and mild to moderate adverse events (AEs), which include β-blockers, amitriptyline, and divalproex; (ii) drugs that have lower documented efficacy and mild to moderate AEs, which include SSRIs, calcium channel antagonists, gabapentin, topiramate, riboflavin, and NSAIDs; (iii) drugs used based on opinion (a) mild to moderate Aes, (b) major AEs or complex management; (iv) drugs that have documented high efficacy but significant AEs or are difficult to use, which include methysergide and MAOIs; and (v) drugs that have proven limited or no efficacy, which include cyproheptadine, lithium, nifedipine, nimodipine, and phenytoin. Choose a drug based on its proven efficacy, the patient's preferences and headache profile, the drug's side-effects, and the presence or absence of coexisting or co-morbid disease (Table 2). Use the drug with the best risk-to-benefit ratio for the individual patient and take advantage of the drug's side-effect profile (277, 278). An underweight patient would be a candidate for one of the medications that commonly produce weight gain, such as a TCA; in contrast, one would try to avoid these drugs in the overweight patient and could consider the use of topiramate. Tertiary TCAs that have a sedating effect would be useful at bedtime for patients with insomnia. The older patient with cardiac disease or patients with significant hypotension may not be able to use TCAs or calcium channel or β-blockers, but could easily use divalproex, topiramate, or gabapentin. In the athletic patient, β-blockers should be used with caution. Medication that can impair cognitive functioning should be avoided in patients who are dependent on their wits (277, 278).
Co-morbid and coexistent diseases have important implications for treatment. The presence of a second illness provides therapeutic opportunities but also imposes certain therapeutic limitations. In some instances, two or more conditions may be treated with a single drug. When migraine and hypertension and/or angina occur together, β-blockers or calcium channel blockers may be effective for all conditions (279). For the patient with migraine and depression, TCAs or SSRIs may be especially useful (280). For the patient with migraine and epilepsy (281) or migraine and bipolar illness (158, 282) divalproex and topiramate are the drugs of choice. The pregnant migraineur who has a co-morbid condition that needs treatment should be given a medication that is effective for both conditions and has the lowest potential for adverse effects on the fetus. In individuals with more than one disease, certain categories of treatment may be relatively contraindicated. For example, β-blockers should be used with caution in the depressed migraineur, while TCAs, neuroleptics, or sumatriptan may lower the seizure threshold and should be used with caution in the epileptic migraineur.