Migraine frequency, duration, or onset with respect to menstruation can guide treatment recommendations. The International Headache Society66 characterizes menstrually related and pure menstrual migraine as migraine without aura; however, migraine with aura is a possibility, albeit less frequent. Treatment considerations are consistent with international guidelines for acute and daily preventive treatment67,68 and for short-term prevention.69,70 NSAIDs indicates nonsteroidal anti-inflammatory drugs.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Brandes JL. The Influence of Estrogen on Migraine: A Systematic Review. JAMA. 2006;295(15):1824–1830. doi:10.1001/jama.295.15.1824
Clinical Review Section Editor: Michael S. Lauer, MD. We encourage authors to submit papers for consideration as a Clinical Review. Please contact Michael S. Lauer, MD, at email@example.com.
Author Affiliations: Department of Neurology, Vanderbilt University School of Medicine, Nashville, Tenn; and Nashville Neuroscience Group.
Context Menstrual migraine affects approximately 50% to 60% of female migraineurs, but knowledge regarding the role of hormones, especially estrogen, appears incomplete.
Objective To conduct a systematic review to determine the role of hormones on menstrual migraine.
Evidence Acquisition MEDLINE (January 1966 through September 1, 2005) and EMBASE Drugs and Pharmacology (January 1991 through September 1, 2005) were searched for articles published in the English language using the keywords migraine, estrogen, menstrual migraine, pure menstrual migraine, true menstrual migraine, menstrually-associated migraine, menstrually-related migraine, pregnancy, breast-feeding, perimenopause, menopause, nitric oxide, and estrogen receptors. A total of 643 unique articles were reviewed for relevance, scientific rigor, and generalizability. For each relevant citation, the bibliography was reviewed to identify additional sources of pertinent data.
Evidence Synthesis The influence of estrogen on migraine is evident by a 3-fold greater prevalence among women compared with men, and by significant changes in migraine incidence with changes in female reproductive status. Menstrual migraines are usually more resistant to treatment, generally not associated with aura, of longer duration, and associated with more functional disability compared with attacks at other times of the month. Biochemical and genetic evidence suggest central and peripheral roles for estrogen in the pathophysiology of menstrual migraine, with potential interactions with excitatory circuits, including serotonergic components. Although evidence for estrogen as a preventive treatment for menstrual migraine is inconsistent, serotonin receptor agonists (triptans) provide acute relief and also may have a role in prevention.
Conclusions Epidemiological, pathophysiological, and clinical evidence link estrogen to migraine headaches. Triptans appear to provide acute relief and also may be useful for headache prevention. Clear, focused, and evidence-based treatment algorithms are needed to support primary care physicians, neurologists, and gynecologists in the treatment of this common condition.
More than 20 years ago, Welch et al1 noted that “the ‘femaleness’ of the migraine condition is inescapable.” Initial evidence for the role of estrogen in migraine headache stems from studies2,3 published in the early 1970s evaluating the influence of exogenous estrogen on women prone to migraines around the time of menses (menstrual migraine). Migraine attacks were correlated with declining plasma estradiol levels. This was particularly evident after prolonged estradiol elevations such as those observed before menses.4 No plasma estradiol threshold was identified,3 and migraine attacks were independent of plasma progesterone concentrations.5 These early findings provided the conceptual framework for future studies of estrogen-induced migraines. Although decades have passed, the pathogenesis and treatment of menstrual migraine remain elusive. In this review, pathophysiological, epidemiological, and clinical evidence will be evaluated across phases of the reproductive cycle, along with a discussion of future research directions.
MEDLINE (January 1966 through September 1, 2005) and EMBASE Drugs and Pharmacology (January 1991 through September 1, 2005) were searched for articles published in the English language using the keywords migraine, estrogen, menstrual migraine, pure menstrual migraine, true menstrual migraine, menstrually-associated migraine, menstrually-related migraine, pregnancy, breast-feeding, perimenopause, menopause, nitric oxide, and estrogen receptors. The search revealed 643 citations; each was analyzed for relevance, scientific rigor, and generalizability. For each relevant citation, the bibliography was reviewed to identify additional sources of pertinent data.
Across most published studies, the menstrual migraine window is defined inconsistently.6 Currently available diagnostic criteria stem from the Headache Classification Committee of the International Headache Society guidelines in a separate appendix to the formal guidelines7; this represents the best description of menstrual migraine based on clinical evidence. Both menstrually related migraine and pure menstrual migraine are defined as migraine attacks without aura occurring on day 1±2 of menstruation. Menstrually related migraine can occur at other times of the cycle, whereas pure menstrual migraine occurs strictly with menses. For simplicity in this review, menstrual migraine encompasses both menstrually related migraine and pure menstrual migraine.
The diagnostic criteria and characteristics of menstrual migraine appear in Box 1. It is unclear if menstrual migraine attacks are more severe than nonmenstrual migraine attacks because the classification and methods of data collection (prospective vs retrospective) have been inconsistent. In a prospective, population-based study, Stewart et al9 reported that menstrual migraines were not necessarily more severe. In contrast, a retrospective, self-report survey characterized menstrual migraine as more severe than other types of migraines,8 and a prospective analysis revealed that women with migraine on days 1 through 7 of the menstrual cycle had increased migraine severity compared with migraines occurring at other times of the month.11Quiz Ref IDIn a prospective study that assessed migraine severity on diary cards, the relative risk of a severe migraine was increased by 3.4-fold (95% confidence interval [CI], 2.6-4.5; P<.001) on days 1 through 3 of menstruation compared with other times of the month.12
Migraine Without Aura7†
At least 5 long-lasting (4-72 hours) migraine attacks that are characterized by 2 or more features (unilateral location, moderate to severe pain intensity, pulsating, or aggravated by routine physical activity) and are associated with 2 or more of the following during an attack: nausea, vomiting, photophobia, phonophobia, or osmophobia. The attacks cannot be attributed to any other disorder.‡
Pure Menstrual Migraine7
Migraine without aura that occurs exclusively on day 1±2 of menstruation (ie, days −2 to +3) in at least 2 of 3 menstrual cycles. The first day of menstruation is day 1 and the preceding day is day −1; there is no day 0. No migraine occurs at other times of the cycle.
Menstrually Related Migraine7
Migraine without aura that occurs on day 1±2 of the menstrual cycle in at least 2 of 3 consecutive menstrual cycles. Additional attacks of migraine with or without aura occur at other times of the cycle.
Menstrual Migraine Characteristics
Migraines are usually more resistant to treatment,8 generally not associated with aura,9,10 of longer duration,9 and are associated with more functional disability compared with attacks at other times of the month.8
*Menstrual migraine includes both pure menstrual and menstrually related migraine. Adapted from the Headache Classification Committee of the International Headache Society.7
†The International Headache Society characterizes menstrual migraines as those occurring without aura. However, menstrual migraines also may occur with aura.
‡As noted in the alternative criteria listed in Appendix A.1.7
Correlations have been identified between premenstrual syndrome and menstrual migraine.13 In a post-hoc analysis, premenstrual syndrome severity was significantly correlated with headache severity on days –3 to +6 of the menstrual cycle, a time of heightened incidence of menstrual migraine.14
Several reproductive milestones correlate with a change in migraine frequency. Prepubertal girls and boys have an approximately equal 4% prevalence of migraine.15 As girls mature, however, the lifetime prevalence of migraine increases to 18% for women and 6% for men, suggesting a hormonal link between female sex and migraine.16,17
Nearly 80% of female migraineurs report onset of migraine between the ages of 10 and 39 years, but prevalence figures during this time vary greatly depending on reproductive status and use of hormonal supplementation, method of data collection, headache definition, and demographics.10,18-20 Review of epidemiological data concluded that among female migraineurs the true prevalence of pure menstrual migraine was 3.5% to 12% and of menstrually related migraine was approximately 50%21 (Table).
Quiz Ref IDEstrogen levels in pregnant women increase throughout each trimester and sharply decline postpartum. These hormone fluctuations correlate with the incidence of migraine attacks, with nearly 80% of 47 women with migraine without aura in one study reporting no migraine attacks in the third trimester.24 These data were confirmed in prospective analyses.25 De novo migraine during pregnancy is rare (<3%) and typically occurs during the first trimester.26 Nearly all women (94%) reported the return of migraines after delivery.18,27 Factors accelerating the return of migraines postpartum included bottle-feeding and age of 30 years or younger.24 These data are also consistent with reports that the incidence of migraine is low during breastfeeding.28
Quiz Ref IDAs women enter menopause and estrogen levels decline, the prevalence of migraine attacks also declines. Headache was reported in 76 (14%) of 556 naturally menopausal women.29 Headache symptoms were reported generally before menopause in 82% of these women, and 62% of the women who experienced migraine or tension-type headache before menopause reported abatement of symptoms after menopause. Comparable figures were collected via self-report in a large cohort of 17 107 postmenopausal women in the Women's Health Study, in which 11% reported having a migraine within the past year.30 Characteristics of women who experience migraine in menopause typically include younger age at menopause onset, surgical menopause, current smokers, daily use of alcohol, previous use of oral contraception, and current use of hormone therapy.30
Genetic predisposition for migraine appears equal in men and women,31,32 and several genetic polymorphisms in hormone receptor genes have been assessed as risk factors for migraine.33-35 A single nucleotide polymorphism in the estrogen receptor 1 gene G594A exon 8 positively correlated with an increased incidence of migraine.36 Polymorphisms of this gene are particularly interesting because estrogen receptors may directly affect nitric oxide production, thereby regulating vascular tone.37
In a population-based, cross-sectional association analysis of 1150 men and women, an androgen receptor gene polymorphism known to be associated with neurodegenerative diseases was not associated with an increased incidence of migraine attacks.38 In the same study, the presence of PROGINS, a polymorphism of the progesterone receptor known to negatively impact the expression of progesterone, positively correlated with an increase of 1.8-fold (95% CI, 1.2-2.6) in migraine attacks. An additive genetic effect is underscored by both estrogen receptor 1 and progesterone alleles in women being associated with an increase of 3.2-fold (95% CI, 1.9-5.3) in the risk of migraine attacks. Twin studies support additive genetic factors and nonshared environmental factors, providing a best-fit model for migraine risk.32
The role of estrogen in the pathogenesis of migraine is substantiated by studies of abrupt estrogen withdrawal.3 Among 98 women undergoing in vitro fertilization, a gonadotropin-releasing hormone analog was administered to down-regulate estrogen levels before controlled ovarian hyperstimulation. Low 17β-estradiol levels correlated with a spike in migraine attacks, and 82% of the women experienced migraine attacks that were debilitating.39
In a retrospective evaluation, 16 menopausal women with a previous history of migraine all experienced a migraine attack when supplemental estrogen was withdrawn, a finding in direct contrast to a cohort of 12 menopausal women without a previous migraine history.40 Women with a history of migraine retained their sensitivity to estrogen in menopause, and higher initial doses of estrogen appeared to prime the response. These data support Somerville's observations3 that prolonged exposure to estrogen followed by a sudden drop in plasma concentrations may precipitate a migraine attack.
In a study of 17 107 postmenopausal women, current hormone therapy use significantly increased the risk of experiencing a migraine within the previous year compared with women without hormone therapy use (13% vs 9%; P<.001).30 Significantly more women with current hormone therapy use had a history of oral contraceptive use, were younger (a mean 3.7-year difference), and underwent surgical menopause. In a multivariable analysis of 18 221 women who were treated with hormone therapy, those who used an intermediate dose of estrogen (0.625 mg/d) had a lower odds ratio of 1.28 (95% CI, 1.10-1.48) for migraine than those who used higher doses (odds ratio, 1.72; 95% CI, 1.39-2.13; P = .002 vs the intermediate dose) or lower doses (odds ratio, 2.00; 95% CI, 1.51-2.65; P = .001). There was no correlation with type or dose of progestin.30 For women with a history of migraine headaches, continuous estrogen and sequential (either continuous or cyclical) progesterone may increase the frequency and severity of migraine attacks.41 This supports the theory that women with a history of menstrual migraine have an increased sensitivity to estrogen, and low levels of estrogen supplementation may paradoxically induce a migraine.40 This paradox can be explained by instability of estrogen levels, causing frequent migraine attacks between cycles. Quiz Ref IDHigh estrogen levels or complete withdrawal of estrogen protects against migraine attacks.42 Fluctuations from higher to lower estrogen levels during estrogen supplementation, in association with migraine attacks being time-locked with decline, confirm estrogen withdrawal as the critical factor.
A recent review43 assessed the mechanistic roles of several hormones in migraine. The evidence suggests a direct role for estrogen affecting the vasculature through stimulation of nitric oxide release. The estrogen receptor α increased nitric oxide synthase activity in endothelial cells44 by direct activation of the protein phosphatidylinositol 3-OH kinase in a nonnuclear and perhaps membrane-associated compartment location.37,45 When the nitric oxide and platelet L-arginine pathway was compared in 60 women with menstrual migraine, nonmenstrual migraine, and no history of migraine, women with a history of menstrual migraine exhibited a heightened activation of the nitric oxide and L-arginine pathway and an increase in nitric oxide, especially during the luteal phase.46 This corresponds with a 5-HT (serotonin) decrease in the luteal phase, a time of heightened migraine attack frequency.47,48
In the trigeminal ganglia of rat cultures, high levels of estrogen influence gene expression and intracellular signaling by the extracellular signal-regulated kinase.49 Intracellular signaling by the extracellular signal-regulated kinase is up-regulated following the binding of extracellular ligands to cell receptors and influences inflammatory and neuropathic pain. Additional evidence suggested that the levels of neuropeptide Y, a regulator of inflammation and central nociception, and galanin, a modulator of the gonadotropin-releasing hormone and the luteinizing hormone,50 are modulated in concert with estrogen level fluctuations during the reproductive cycle.51 These data support the bimodal theory that either abrupt estrogen decline or chronically high plasma estrogen concentrations can influence trigeminal pain.49
Recent neuropharmacological evidence from a primate model of surgical menopause examining effects in the dorsal raphe of macaques suggests a link between estrogen and serotonin synthesis.52 Serotonin synthesis and degradation and neuronal firing appear to be influenced by estrogen receptor–mediated mechanisms. This evidence is consistent with a small clinical study of women with status migrainosus. During the placebo week of oral contraceptive therapy, 10 women with and 6 without status migrainosus received a 5-HT agonist and either transdermal estrogen or placebo.53 In the absence of estrogen and compared with the control group, women with a history of status migrainosus experienced significantly impaired neuroendocrine responses as indicated by decreased cortisol and impaired prolactin secretion. The neuroendocrine response was restored when women received both the 5-HT agonist and the transdermal estrogen, accompanied by clinical improvements in duration and severity of migraine. A possible link between estrogen and serotonergic signaling has added significance because 5-HT receptors are important therapeutic targets in the acute treatment of migraine (reviewed below).
Changing central opioid tonus has been proposed as another mechanism that may induce migraine around the time of menstruation.47 In contrast to women without menstrual migraine, patients with menstrual migraine exhibit poor response to luteinizing hormone after being injected with the opiate antagonist naloxone during the luteal phase of the menstrual cycle. The disparity in migraine incidence between men and women may be explained by luteinizing hormone releasing hormone neurons being more sensitive in pubescent girls than in their male counterparts.54 In an analysis of opioid tonus in women with menstrual migraine, plasma β-endorphin and cortisol responses were impaired during the premenstrual period,55 indicating that premenstrual opioid hyposensitivity may contribute to the risk of menstrual migraine. These data suggest a possible central role for estrogen in pain-modulating pathways.
If estrogen withdrawal precipitates a migraine attack, estrogen supplementation should prevent migraine attacks. Early acute studies56,57 using percutaneous estradiol gel showed some clinical benefit, especially when estradiol plasma levels were higher than 50 pg/mL. In an open-label study of 20 women who received 20 μg of ethinyl estradiol on days 1 through 21 of their menstrual cycle and 0.9 mg of conjugated equine estrogens on days 22 through 28, there was a 76% reduction in the number of days with headache per month.58 The results of percutaneous and oral estrogen were not replicated using an estradiol patch. In women with pure menstrual migraine, use of 50 μg of percutaneous estradiol via the Estraderm TTS patch (Novartis Pharmaceuticals, East Hanover, NJ) did not affect the incidence (59% for Estraderm TTS vs 69% for placebo), mean duration, or severity of migraine attack compared with placebo.59 One reason for the disparity among the trials may be that the trials tended to be small, often with fewer than 20 patients included. Furthermore, study designs, estrogen dosages, routes of administration, and definitions of menstrual migraine varied among the trials. A cautious approach to high-dose estrogen supplementation may be warranted to avoid an increased risk of ischemic stroke in at-risk women. Primary risk factors in these women include presence of aura, patient age, tobacco use, and other vascular risk factors including hypertension, diabetes, hyperlipidemia, and obesity.60
Quiz Ref IDWhile the level of evidence for menstrual migraine prevention is poor for estrogen, clinical trials using the 5-HT receptor agonists (triptans) for menstrual migraine treatment have been more robust and consistent. The receptors 5-HT1B and 5-HT1D play a significant role in migraine modulation via mechanisms known to mediate vascular smooth muscle vasoconstriction and inhibit vasoactive peptide synthesis. Prospective trials with the 5-HT1B/1D receptor agonists sumatriptan and zolmitriptan61-63 and retrospective trials with sumatriptan and rizatriptan64,65 have established the efficacy of acute treatment of menstrual migraine attacks. Management of acute migraine attacks with triptans is a valuable treatment strategy, particularly in patients with dysmenorrhea, infrequent migraine attacks, or when migraine onset is otherwise unpredictable (Figure). However, not all patients benefit from acute treatment, which may only provide sustained pain-free responses in approximately 20% to 30% of patients.71 Therefore, alternative treatments are warranted, particularly in patients with more severe, recurrent migraine.
With the understanding that menstrual migraine attacks are often predictable, longer in duration and severity than nonmenstrually related migraine attacks, and may not respond to acute therapy with triptans or nonsteroidal anti-inflammatory drugs, short-term prevention also has been assessed. In contrast to acute trials, short-term preventive trials focus on end points such as the reduction in menstrual migraine incidence, intensity, and disability (Box 2). A dose of 550 mg of naproxen sodium twice daily beginning 7 days before expected menses with continuation through day 6 of menstrual flow was assessed in a double-blind, placebo-controlled study of 40 women. Compared with placebo, headache intensity, duration, and analgesic use were reduced for patients receiving naproxen. In 33% of the women receiving naproxen, no menstrual migraine attacks occurred.72
Incidence of menstrual migraine in a given perimenstrual period
Severity and duration of menstrual migraine
Degree of functional impairment
Use of rescue medication
Patient ratings of treatment effectiveness and tolerability
One open-label, short-term trial with sumatriptan (25 mg orally 3 times per day) was conducted in which 20 women with menstrual migraine received treatment beginning 2 to 3 days before migraine onset with continued treatment for 5 days. Headache was absent in approximately 50% of treated cycles; during the treated cycles with headache, women reported reductions in headache severity compared with baseline.73 Based on these results, triptans with a longer half-life were studied for prevention of disabling migraine attacks.
In a randomized, placebo-controlled study of 206 women, the effects of a 5-day course of 2 doses of naratriptan were assessed as a prophylactic treatment for migraine. Patients were instructed to initiate treatment 2 days before the expected onset of menses. Over the course of 4 perimenstrual periods, more patients receiving 1 mg of naratriptan 2 times per day experienced headache-free days compared with patients receiving placebo (50% vs 25%; P = .003). However, this effect was not observed in the group receiving 2.5 mg of naratriptan 2 times per day.69 In a separate open-label study of 59 women, the impact of taking 1 mg of naratriptan 2 times per day on pure menstrual migraine was assessed.74 Patients initiated naratriptan 2 days before menstruation onset and continued taking it for 6 days total. The number of migraine attacks decreased by nearly 2 over the course of 3 months when 1 mg of naratriptan was taken 2 times per day compared with placebo.74
In a randomized, placebo-controlled trial evaluating the short-term prevention of menstrual migraine with frovatriptan, 579 women were randomized to receive 2.5 mg of frovatriptan or placebo (either once or twice daily) on days 2 to 5 after taking a loading dose of 5.0 mg of frovatriptan or placebo (once or twice daily) on day 1. The treatment regimen was initiated 2 days before anticipated menstrual migraine. Significant reductions in the incidence of menstrual migraine occurred over the course of 3 perimenstrual periods for women taking 2.5 mg/d of frovatriptan (52%) and 2.5 mg of frovatriptan 2 times per day (41%) compared with placebo (67%; P<.001). Similarly, fewer patients reported moderate to severe functional impairment with 2.5 mg/d of frovatriptan (29%; P<.01 vs placebo) or 2.5 mg of frovatriptan 2 times per day (18%; P<.001 vs placebo) compared with placebo (38%). Significant reductions also were observed in the frovatriptan-treated group for a reduction in headache severity, headache duration, the need for rescue medication, and menstrually related symptoms. A dose of 2.5 mg of frovatriptan administered 2 times per day was significantly better than 2.5 mg/d of frovatriptan for incidence of menstrual migraine, severity, and menstrual migraine-associated symptoms.70
Epidemiological, pathophysiological, and clinical evidence link estrogen to migraine headaches. Clear, focused, and evidence-based treatment algorithms are needed to support primary care physicians, neurologists, and gynecologists in the treatment of this common condition. Comorbidities associated with menstrual migraine also should be noted. Depression, panic disorder, and phobias are increased 1.9-fold to 3.4-fold in patients with migraine.75 The potential interaction between these comorbidities and estrogen in the pathogenesis of menstrual migraine remains to be established. Emerging evidence of the influence of estrogen on serotonergic activity may provide a mechanistic basis for the association between these comorbidities.
Migraine represents a substantial health care burden, both clinically and economically. Individuals with migraine and their families consume substantially more health care resources than those without migraine headache.76 Therefore, it is imperative that standardized diagnostic criteria be consistently implemented in clinical trials and used to develop appropriate treatment algorithms and provide governmental bodies (eg, National Institutes of Health) with a rationale to support research and education for menstrual migraine.
Pharmacological studies are needed to elucidate the interplay between estrogen and serotonergic signals. Standardized diagnostic criteria would be useful to further elucidate risks for estrogen-induced migraines, characterize the patterns of menstrual migraine in women, and determine its economic impact. Future clinical trials will be helpful in identifying treatment strategies for acute migraine attacks and may establish the role of short-term prevention in migraineurs. Armed with this knowledge, women and their physicians can better manage this disabling disorder through appropriate diagnosis and evidence-based treatment regimens.
Corresponding Author: Jan Lewis Brandes, MD, 300 20th Ave N, Suite 603, Nashville, TN 37203 (firstname.lastname@example.org).
Author Contributions: Dr Brandes had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Brandes was responsible for the study concept and design; acquisition of data; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content; administrative, technical, or material support; and for study supervision.
Financial Disclosures: Dr Brandes has received grants or research support from Merck, GlaxoSmithKline, UCB Pharma, Allergan, Johnson & Johnson, AstraZeneca, Pfizer, Bristol-Myers Squibb, Winston Laboratories, Sanofi-Aventis, Elan Pharmaceuticals, Novartis, Endo, Pozen, Vernalis, Ortho-McNeil, and Advanced Bionics; has served on the speaker's bureau for GlaxoSmithKline, AstraZeneca, Pfizer, Merck, Ortho-McNeil, Allergan, MedPointe Pharmaceuticals, Endo, and UCB Pharma; has served as a consultant to Merck, GlaxoSmithKline, Pfizer, AstraZeneca, Allergan, Ortho-McNeil, and Aradigm Corp; and has received educational funding from GlaxoSmithKline.
Acknowledgments: I thank Dana Franznick, PharmD (Complete Healthcare Communications Inc, Chadds Ford, Pa) for assistance in retrieving and collating literature search results and for helping to edit the manuscript to conform to submission requirements. Dr Franznick received support from Endo and Sanofi-Aventis but neither had any input into the preparation of the review, the analysis of the data, nor any role in the writing of the manuscript. I also thank Meredith Cucuel, BA, Ellen Koonce, BS, and Susan Hardy (all from Nashville Neuroscience Group, Nashville, Tenn) for providing administrative assistance in preparing the manuscript for submission.