Abstract
Consistent evidence demonstrates that migraine is far more common in women than in men, but the explanations for this preponderance have not been systematically evaluated. We examined whether the female preponderance is attributable to genetic factors using data from a controlled family study which included 260 probands and their 1232 first-degree adult relatives. We found that although the risk of migraine was three times greater among the relatives of probands with migraine compared with controls, there was no differential risk of migraine among the relatives of male vs. female probands with migraine. Taking these data together with other family studies, we conclude that the increased risk of migraine in females is likely to result from increased exposure to non-familial endogenous or exogenous risk factors for migraine that lower the threshold for expression of migraine in women.
Introduction
Migraine is a common, debilitating disorder which places a significant burden on society and affected individuals. There is consistent evidence that it is far more common in women than in men. Figure 1 presents a summary of the sex ratios of the lifetime prevalence of migraine in international population-based studies that employed the International Headache Society (IHS) using direct interviews of adults (1–7). The sex ratio for lifetime migraine remains stable between two to threefold greater among women and is consistent across countries. In addition, 1-year prevalence rates in females ranging from 1.5 to 18.3% are also greater than those of males 0.6–9.5% in population-based studies worldwide (e.g. Ethiopia, Hong Kong, China, Peru, Hungary, Norway, Sweden, Austria, Puerto Rico, England) (8–17). The lifetime pattern of the sex difference varies with age—prior to puberty, migraine prevalence is equivalent or slightly greater in males, but during adolescence the incidence and prevalence increase more rapidly among females (18–20). The female preponderance over the lifetime continues to increase until its peak in the late 40s/early 50s and thereafter declines, but never returns to equal.
Sex Ratios of the Lifetime Prevalence of Migraine among Population based studies.
Numerous possible explanations for the sex difference in migraine have been proposed and there is abundant research on some of the mechanisms for the female excess in migraine. Following exclusion of possible artefactual explanations for the female preponderance (such as biased sampling, reporting, or confounding disorders, e.g. anxiety, depression), several categories of explanations have been considered and include: neurobiological factors (e.g. fluctuation of reproductive hormones, increased stress reactivity in women, sex difference in pain reactivity); greater exposure or sensitivity to environmental stressors (e.g. role of stress, life events, susceptibility to sensory stimuli); and genetic factors (e.g. greater genetic loading for migraine in women) (21–25).
One possible explanation for a sex difference in a disorder is that one sex has a greater propensity to express a disorder because of underlying genetic factors. Diseases with clear-cut Mendelian modes of transmission, i.e. autosomal dominant, autosomal recessive, or X-linked, have predictable ratios of affected status among male and female probands and relatives. The sex ratio is expected to be equal for most autosomal disorders, whereas sex differences are a characteristic of X-linked diseases, such as colour blindness and haemophilia. Patterns of transmission of more common migraine subtypes are contradictory and fail to support any specific Mendelian pattern of transmission over another (26–28). Even though current evidence does not support mitochondrial inheritance as an explanation for migraine, there are data that suggest maternal transmission characterizes early-onset migraine (29–37).
In addition to providing evidence regarding the possible role of transmissible genetic factors, family studies may be employed to investigate sources of a sex difference in the prevalence of a particular disorder. The sex effect can be treated as a differential threshold, with the less commonly affected sex having the higher threshold (38). Accumulation of more genetic and/or environmental factors is required for the disorder to become manifest at the higher threshold. Because such individuals have increased genetic and/or environmental loading for the disorder, they have more factors to transmit to their relatives. Therefore, the relatives of probands with the higher threshold are more likely to be affected with the disorder (39). Thus, in the case of migraine, relatives of male probands with migraine would be expected to exhibit greater migraine rates compared with relatives of female probands with migraine—if the sex difference was attributable to genetic factors. Although a few prior family studies have presented data on the effect of sex of the probands on the rates of migraine among relatives (40), a systematic analysis of the familial explanations for the female preponderance of migraine has not been presented previously. In this study, we tested whether the female preponderance of migraine is attributable to genetic factors using data from a controlled family study.
Methods
Sample
Probands
Two hundred and sixty probands participated. Probands were originally selected with the goal of examining the familial aggregation of anxiety disorders; therefore, we did not specifically recruit from headache specialty clinics (41), but from out-patient clinics for anxiety disorders at the Connecticut Mental Health Center in New Haven, CT. All patients treated at the clinics over a 3-year period were screened for eligibility and those who met the appropriate diagnostic criteria were invited to participate. The control subjects were selected from the population of the greater New Haven area through a random digit dialling procedure. All probands were assigned to one of three lifetime diagnostic groups as follows: (i) migraine (N = 41); (ii) anxiety disorders without migraine (N = 165); and (iii) controls (no migraine or any Axis I psychiatric disorders) (N = 54). For this investigation, the latter two groups (anxiety disorders without migraine and controls) were combined to form the comparison group (N = 219) of the migraine probands (N = 41), and anxiety disorders were controlled for in the analyses.
Relatives
A total of 1232 relatives (111 migraine relatives, 1121 non-migraine relatives) were included in these analyses.
Measures
A lifetime history of migraine was obtained using the Diagnostic Interview for Headache Syndromes (DIHS), a structured instrument which was developed to collect the major subtypes of headache as defined according to the 2004 IHS Criteria for Headache Syndromes (42). Information was collected on the symptoms, frequency, duration, onset, course, treatment, impairment, triggers and associated features of migraine. The DIHS was developed at the Yale University Genetic Epidemiology Research Unit by a multidisciplinary team that included collaboration with three internationally renowned headache experts, a neuropsychiatrist, a clinical psychologist and an epidemiologist. The DIHS is a semistructured diagnostic interview comprising an open-ended interview regarding the lifetime history of headache to identify the headache subtypes requiring further assessment. The DIHS includes diagnostic criteria for the IHS headache syndromes of migraine with and without aura, tension-type headache, cluster headache and post traumatic headache. Direct interviews were administered to all interviewed relatives. Studies of the reliability and validity of this interview are now underway in a family study of migraine probands and controls. A family history of migraine based on the interview for migraine from the Australian Twin Study (43) was administered to probands regarding all first-degree relatives. In the current study, the subtypes of migraine were not distinguished because the family history interview could not reliably differentiate between migraine with and without aura.
A lifetime history of psychiatric disorders was obtained using the semi-structured Schedule for Affective Disorders and Schizophrenia (44), which was modified to incorporate Diagnostic and Statistical Manual of Mental Disorders (DSM-III and DSM-III-R) diagnostic criteria.
Family procedures
Complete pedigrees on first-degree relatives including offspring over 18 years old were obtained. Both direct (face-to-face or telephone) and indirect (family history by relatives) assessments of first-degree relatives including parents, siblings and offspring >18 years were conducted by clinically experienced interviewers who were blind to the diagnostic status of the proband. Because the intention was to include all family data in these analyses, including non-interviewed relatives, it was not possible to discriminate between specific subtypes of migraine (i.e. with and without aura). Information on relatives who were deceased, refused to participate, or were geographically distant, was obtained (blind to proband diagnoses) from multiple relatives. Best estimate diagnostic assessments were made by a panel of experienced clinicians based on all available information. Treatment records were requested for all probands and relatives who had received treatment.
Statistical methods
Generalized estimating equations (GEE) (45, 46) were used to investigate the effect of migraine among probands and the sex of probands and relatives on migraine among relatives, while controlling for relative age, relative interview status, comorbid anxiety in probands and relatives, and proband sampling source (community vs. clinic). The GENMOD procedure in SAS 9.1 (SAS Institute Inc., Cary, NC, USA) was used for GEE analysis.
Tetrachoric correlation r was also used to estimate the degree to which migraine clustered in families. The tetrachoric correlation assesses the correlation in the underlying liability to develop a disorder in probands and relatives by transforming the thresholds in relatives of controls to a normal Z distribution with mean 0 and SD of 1. It is used as a measure of the magnitude of between-family sources of variation in relation to the total variation (47).
Results
Table 1 describes the sociodemographic characteristics of the probands with and without migraine. There were more females in the migraine group (66% vs. 47%; P = 0.03).
Sociodemographic characteristics of probands
The lifetime prevalence of migraine in male and female relatives by male and female probands with migraine is shown in Table 2. Although female relatives had nearly three times the risk of migraine than male relatives (29% vs. 10%), the prevalence of migraine among the relatives of male and female probands was not statistically different (17% and 22%, respectively, P = NS).
Lifetime prevalence (%) of migraine in relatives (N = 1232) of migraine probands
χ2 P-value = not significant.
Table 3 examines the association of migraine between probands and relatives, and the sex of probands and relatives, while controlling for age of relatives (by increments of 10 years because migraine rates can fluctuate per decade), interview status of relatives (direct interview vs. family history), proband sampling source (community vs. clinic) and comorbid anxiety in probands and relatives. The odds ratio (OR) for proband migraine was 3.0 [95% confidence interval (CI) 1.7, 5.0], which indicates relatives of migraine probands had a threefold increased risk of migraine compared with that of relatives of probands with no migraine (controls). There was no effect of the sex of the proband on migraine among relatives (OR 1.5, 95% CI 0.9, 2.5); however, female relatives had double the risk for migraine than male relatives (OR 2.0, 95% CI 1.3, 3.1).
Association of migraine between probands and relatives∗
Adjusted for sampling source (community vs. clinic) of proband and comorbid anxiety in probands and relatives. Proband anxiety was not significant. Sex of proband by sex of relative interaction was not significant. Relative interview status indicates the relative was directly interviewed vs. migraine diagnoses made by family history method.
The tetrachoric correlations for migraine between male and female relatives and their same-sex and opposite-sex relatives are presented in Table 4. The correlations between male probands and male relatives (r = 0.25) and between female probands and female relatives (r = 0.37) did not differ statistically. The correlations between male and female probands and their opposite-sex relatives were similar to each other and not different from the correlations obtained for same-sex relatives.
Tetrachoric correlations for migraine between probands and relatives
Discussion
The results of this study report that females do not have an increased genetic loading for migraine, i.e. there is no sex of proband effect of migraine to relatives. Among the previous controlled family studies (40, 48–50), three (40, 48, 50) examined sex of proband effects with similar findings to this study. When sex differences in the rates of migraine in relatives are reported, they are consistently greater among females.
Our findings confirm the familial aggregation of migraine reported in prior studies (48–51), which found a two- to threefold increased risk of migraine in the relatives of probands with migraine. Even greater migraine risk among previous studies was conferred in migraine probands with the aura subtype, earlier age of onset and more severe migraine (40, 48–50). In this study, additionally, comorbid anxiety disorders in probands, a potentially important confounding factor, did not affect the association between migraine in probands and relatives, despite the well-known comorbidity of migraine and anxiety disorder (see review (52)). Likewise, three of the four large, community-based twin studies of migraine found no significant sex difference in the heritability of migraine (53–56), whereas one reported that there was greater heritability among female twins (57). When taken together with other family studies of migraine (49, 58–62), in general, twin and family studies suggest the same genetic factors underlie migraine in men and in women.
If genetic factors underlying migraine do not differ in men and women, it suggests that the increased risk of migraine in females is likely to result from increased exposure to non-familial endogenous or exogenous risk factors for migraine, which lowers the threshold for expression of migraine in women. The most widely studied risk factor affecting expression of migraine in women is the fluctuation of reproductive hormones because of the well-established association between migraine and the reproductive system landmarks. The prevalence of migraine between the sexes diverges at puberty, when females experience migraine strongly associated with the menstrual cycle, and the first trimester of pregnancy, but with a reduction following menopause (63). Changes in migraine rate, frequency, intensity and severity also coincide with exogenous manipulation of the oestrogen and progesterone milieu, such as is the case with oral contraceptive use, surgical hysterectomy and oophorectomy and hormone replacement therapy.
Migraine has been shown to be triggered by acute oestrogen withdrawal following days of high-dose oestrogen ‘priming’—a hormonal context which occurs naturally during the late-luteal phase of the menstrual cycle and during the placebo week of the combined oral contraceptives (64–67). By varying the timing and dosing of oestrogen given in the luteal phase, either migraine onset can be delayed or the days with headache and headache severity can be reduced (68–70). In contrast, providing progesterone during the same phase does not alter the headache in any way (71, 72). Thus, the increasing oestrogen levels in the latter two trimesters of pregnancy explain the improvement of migraine, and oestrogen's acute fall postpartum predicts its recurrence. The precise direct and indirect mechanisms through which oestrogen exerts its influence on migraine still await full elucidation and are the topic of much investigation (73, 74).
Another possible explanation for the female excess is greater exposure or sensitivity to environmental triggers; however, although much research has been dedicated to the search for specific environmental precipitants of migraine (e.g. sensory stimuli, food, wine, caffeine, weather changes, physiological changes, etc.), to date there is no compelling evidence for systematic differences between the sexes. Likewise, women have not been found to have increased sensitivity to stress through perturbations in the hypothalamic–pituitary–adrenal axis (75, 76). Further investigation is warranted to delineate the mechanisms of the observed female sex preponderance in migraine.
These results should be considered in light of several study limitations. Due to our inclusion of family data on non-interviewed relatives, analyses of migraine subtypes could not be presented. The relatively small size of the sample yielded insufficient power to detect subgroups who may exhibit different modes of transmission. Also, given the small number of male probands with migraine, the power to detect a sex difference among their relatives was diminished.
Strengths of this study include the recruitment of probands from both clinic and community settings to minimize clinical ascertainment bias (i.e. Berkson's bias). Direct interviews with relatives were obtained using a structured diagnostic interview administered by experienced clinical interviewers. Additionally, the interview status of the relatives was controlled for in the analyses, and a control group was included. Finally, anxiety disorders in probands and relatives were examined in the analyses, as they may mediate transmission in at least a subset of migraine families.
In summary, our findings do not suggest that there are differential genetic mechanisms in men and women. However, gene–environment interaction may be more potent in women if they are more susceptible to migraine attacks when exposed to endogenous and/or exogenous triggers. Future studies should identify mechanisms for increased expression in women, particularly those that may be amenable to early intervention or prevention of this widespread and disabling disorder.