|Home | About | Journals | Submit | Contact Us | Français|
To determine if shared genetic or environmental vulnerabilities could underlie depression and migraine.
Depression and migraine headaches frequently coexist and their comorbidity may be due to shared etiologies.
Female twins in the University of Washington Twin Registry responded to a mailed survey regarding their health history. Depression and migraine were determined by self-report of a physician’s diagnosis. We used bivariate structural equation modeling to test for shared genetic, common environmental, and unique environmental components, and to estimate the magnitude of any shared component.
Among 758 monozygotic (MZ) and 306 dizygotic (DZ) female pairs, 23% reported depression and 20% reported migraine headaches. Heritability was estimated to be 58% (95% confidence interval: 48–67%) for depression and 44% (95% confidence interval: 32–56%) for migraine. Bivariate structural equation modeling estimated that 20% of the variability in depression and migraine headaches was due to shared genes and 4% was due to shared unique environmental factors.
The comorbidity of depression and migraine headache may be due in part to shared genetic risk factors. Research should focus attention on shared pathways, thereby making progress on 2 disease fronts simultaneously and perhaps providing clinicians with unified treatment strategies.
Depression and migraine headaches are common illnesses that exact high personal and economic costs.1, 2 These 2 disorders frequently arise in the same individuals,3–5 but the mechanism underlying this comorbidity remains poorly understood.6 Migraine headaches reportedly affect almost half (46%) of individuals with unipolar depression,7 and the prevalence of major depression among migraine headache sufferers is similar (41%).4 There is evidence supporting a bidirectional relationship between depression and migraine headache;8 this relationship is specific to migraine as opposed to other forms of severe headache.3 Further, both disorders are likely familial5, 9, 10 and may in part be influenced by genetic factors.9, 11 For these reasons, some investigators have proposed a shared underlying etiology or common pathway.3,12
Twin studies offer a unique opportunity to investigate shared disease etiology, by evaluating the relative contributions of genetic and environmental factors to two phenotypes. We investigated shared genetic and environmental vulnerability for depression and migraine in a sample of female twins derived from a community-based twin registry. Using bivariate structural equation modeling, we sought to determine: 1) the association between self-reported depression and migraine, 2) the magnitude of the additive genetic component for both conditions, and 3) whether a shared genetic or environmental vulnerability could underlie both conditions.
All twins were participants in the University of Washington Twin Registry, a community-based registry of twin pairs derived from applications for drivers’ licenses in Washington State. The construction and characteristics of the University of Washington Twin Registry and its sample population are described in full elsewhere, including information on response rates.13 Briefly, because drivers’ license numbers are assigned on the basis of an individual’s name and date of birth, the Department of Licensing asks each new applicant whether he or she is a member of a twin pair to avoid issuing duplicate license numbers. The University of Washington receives lists of applicants who are twins, and each member of the pair is invited to join the Registry and complete a brief health survey. The survey contains items on demographics, symptoms, physician-diagnosed health conditions, habits, healthcare use, and various abridged, standardized measures of physical and mental health. We studied female pairs enrolled in the Registry because depression and migraine both disproportionately affect women. All University of Washington Twin Registry procedures and the data collection involved in this study were approved by the University of Washington Institutional Review Board. Informed consent was obtained from all twins.
As part of the mailed questionnaire, all twins were asked questions about childhood similarity to assess zygosity. Studies in both U.S. and Scandinavian twin registries have repeatedly demonstrated that questions about childhood similarity in twin pairs can be used to classify zygosity with an accuracy of 95–98% compared with zygosity determined by biological indicators.14, 15 Responses to these similarity questions were used in a multi-step process to assign zygosity.
Sociodemographic factors collected in the survey included age, race, education, and marital status.
Lifetime depression and migraine were assessed by self-report of a physician diagnosis. Twins were given a list of conditions, including depression and migraine headaches, and asked: “Has your doctor ever told you that you have any of the following conditions?” Twins were also asked about recent depressive symptoms of sadness and anhedonia for 2 weeks in the past year. Finally, twins responded to a list of symptoms that accompany their typical headaches, including: feeling sick to the stomach or vomiting, more sensitive to light and/or noise, a throbbing feeling in the head, pain on only one side of the head, a decrease in normal daily activity, and a preceding warning such as problems with vision, sensation, or strength.
Descriptive statistics for demographic and health characteristics were calculated by using means and standard deviations for continuous variables and percentages for categorical variables. Logistic regression was used to examine the association of depression and headache symptoms with self-reported physician’s diagnoses of depression and migraine, respectively. We also examined the prevalence of depression and migraine according to the depression/migraine status of the co-twin. In this descriptive analysis of co-twin phenotypes, we categorized twins into 4 groups: 1) co-twin without depression or migraine, 2) co-twin with depression only, 3) co-twin with migraine only, and 4) co-twin with both depression and migraine. Logistic regression models were also used to calculate the odds ratio for depression or migraine in one twin according to her co-twin’s illness status. The referent group for all odds ratios consisted of twins whose co-twin had neither depression nor migraine. All regression models used generalized estimating equations and the robust sandwich variance estimator to adjust standard errors for correlation within twin pairs.
The association between depression and migraine in monozygotic (MZ) and dizygotic (DZ) pairs was assessed by 3 types of correlations: twin, phenotypic, and cross-twin cross-trait. Twin correlations assess the within-pair similarity—in other words, the likelihood that one twin will have the phenotype or disease when her co-twin is affected. Phenotypic correlations assess the co-occurrence of depression and migraine within individuals. Cross-twin cross-trait correlations describe the occurrence of migraine in one twin when her co-twin has depression, or vice versa.
Classical twin study analyses are based on comparing phenotypic similarity in MZ twins, who have identical genotypes, and DZ twins, who share, on average, half of their genes. Greater phenotypic similarity, indicated by a higher correlation in MZ twins than DZ twins, suggests a genetic component in the etiology of the condition. Structural equation modeling is a general statistical approach that is useful for estimating genetic and environmental effects in classical twin studies.16 In this approach, genetic and environmental effects are modeled as latent variables representing an underlying liability for one or more phenotypes, such as depression or migraine. Structural equation modeling is a highly flexible technique that can estimate the relative magnitude of the genetic, common environmental, and unique environmental effects. For multiple phenotypes, structural equation modeling can be used to estimate how much of the variability in 2 or more phenotypes is due to shared vulnerabilities—whether they are genetic, common environmental, or unique environmental effects.
We used univariate structural equation modeling to estimate the additive genetic (A), common environmental (C), and unique environmental (E) influences on depression and migraine individually.16 Models were fitted by assuming an additive genetic correlation of 1.0 for MZ and 0.5 for DZ twins and a shared environmental correlation of 1.0 for all twins. Modeling began by estimating parameters for the full model (ACE). Reduced models were constructed by removing a specific parameter, and we then compared the goodness-of-fit of each reduced model to the full model by using a likelihood ratio test. We present parameter estimates, 95% confidence intervals, and goodness-of-fit statistics for 3 models: the full model (ACE), a model in which all variance was attributable to genetic and unique environmental factors (AE), and a model in which all variance was produced by common and unique environmental factors (CE).
Parameters were removed from the model if their removal did not result in a significant degradation of model fit (P≤0.05). Models were also evaluated by using Akaike’s Information Criterion.17 The model with the lowest Akaike’s Information Criteria was judged to have the best fit. We then estimated the proportions of variance for additive genetics, common environment, and unique environment from the final best-fitting model. We also performed post-hoc univariate analysis stratifying twins by age (<50 and ≥50).
Following the univariate analysis, we used bivariate structural equation modeling to estimate shared genetic and environmental vulnerabilities to depression and migraine. We began modeling with a full Cholesky decomposition that specified a general multivariate covariance structure and allowed for both specific and shared influences on depression and migraine. Model fitting for the bivariate analysis followed a similar approach to that used in the univariate modeling. We identified the final best-fitting and most parsimonious model by removing parameters that did not significantly degrade the fit of the model based on likelihood ratio tests and Aikaike’s Information Criterion.17 We present goodness-of-fit statistics for the full and reduced bivariate models of depression and migraine, and trait-specific and shared variance components for the best-fitting model. Descriptive analyses and tetrachoric correlations were computed by using Stata 9.2 for Windows (StataCorp LP, 2006). Structural equation models were fit by using MxGUI version 1.4.06.18
The initial Registry sample consisted of 2172 female twins. Both members of 22 pairs were excluded because of missing depression or migraine data for one or both twins, leaving a total of 758 MZ and 306 DZ female pairs (n=2128). The twins’ mean age was 32±15 years, with a range of 18–90 years; 86% of the sample were white, reflecting the demographics of Washington State. The mean for years of education was 14±2 years, and 43% of the sample were married or cohabitating.
The overall prevalence of a doctor’s diagnosis of depression was 23%. Similarly, 20% of twins reported a doctor’s diagnosis of migraine headache, and 8% reported both depression and migraine. Twins reporting a doctor’s diagnosis of depression were significantly more likely than those without self-reported depression to endorse 2 weeks in the past year of depressed mood (75% vs. 55%, P<0.001; n=31 missing) or anhedonia (55% vs. 14%, P<0.001; n=41 missing). Among twins reporting a doctor’s diagnosis of migraine, 71% stated that their typical headache was accompanied either by nausea and vomiting or by phonophobia or photophobia in combination with 2 of the following characteristics: throbbing quality, unilateral headache, and decrease in usual activity. This symptom constellation was reported by significantly fewer of those without a self-reported migraine diagnosis (28%, P<0.001; n=34 missing). A preceding aura was also significantly more likely among twins who reported a migraine diagnosis (39% vs. 8%, P<0.001; n=36 missing).
Figure 1 presents the proportion of twins reporting a doctor’s diagnosis of depression, migraine, both, or neither, according to the disease status of the co-twin. Thus, among women whose co-twin did not report depression or migraine (Figure 1A), the prevalence of depression was 10% and the prevalence of migraine was 12%. Conversely, twins whose co-twin reported depression (Figure 1B) or migraine (Figure 1C) were at greater risk for these conditions. For example, the odds of depression when a co-twin reported depression were 3.7 times higher than when the co-twin reported neither depression nor migraine (95% confidence interval 2.5–5.5). The odds of migraine were 3.1 times higher when a co-twin reported migraine (95% confidence interval 2.0–4.9). Similarly, when a co-twin reported both disorders (Figure 1D), the proportion of twins affected by comorbid depression and migraine was much larger than when co-twins reported neither disorder (odds ratio 9.9, 95% confidence interval 5.5–17.6). The concordance in disorders between twins is potentially genetic, based on the higher twin correlations in MZ pairs than in DZ pairs observed in the tetrachoric correlations for depression and migraine presented in Table 1.
Looking at Figures 1A–C, we also observed that the proportion of twins reporting migraine was 14% when a co-twin reported no depression or migraine, but was 21% (with or without comorbid depression) when a co-twin reported depression. Similarly, 16% of twins reported depression when their co-twin reported no depression or migraine, and the proportion of twins reporting depression (with or without comorbid migraine) was 19% when a co-twin reported migraine. While these are not dramatic increases, they do suggest shared vulnerabilities to both disorders as the frequency of one diagnosis is higher when a co-twin reports the other diagnosis. This is further borne out by the much higher proportion of twins with depression, migraine, or both when their co-twin reports both diagnoses (Figure 1D). Higher cross-twin cross-trait correlations in MZ twin pairs than in DZ twin pairs also suggest that there are shared genetic influences on both traits, although these data are not entirely consistent (Table 1).
We next conducted structural equation twin modeling to separately estimate the genetic architecture of the 2 traits. Results of twin models are depicted in Table 2. The best-fitting and most parsimonious model for both conditions included only additive genetic and unique environmental effects (AE) to the overall variance. The heritability estimates for depression and migraine were 58% (95% confidence interval: 48–67%) and 44% (95% confidence interval: 32–56%), respectively. For both conditions, the unique environmental (E) component (encompassing environmental effects specific to the individual, random error, and the contribution of gene-environment interactions) accounted for the remainder of the variance. The post-hoc analyses stratifying twins by age found that the AE model was the best-fitting model for each phenotype regardless of age group (results not shown).
The purpose of bivariate analyses was to evaluate shared genetic and/or environmental vulnerabilities for depression and migraine. The most parsimonious model included only additive genetic and unique environmental effects (AE) (Table 3). We estimated that 20% of the variance in depression and migraine is due to shared genetics, and 4% of the unique environmental component is shared (Table 4).
We found that depression and migraine headache were frequently comorbid in female twins. Both conditions had a significant genetic basis. Results from the bivariate structural equation modeling suggest that the association between depression and migraine may be due in part to shared genetic risk factors. Genetic pleiotropy, which refers to the effect of a single gene or set of genes on multiple phenotypes, has been demonstrated in monogenic diseases in humans and may result from the disruption of multiple functions of the gene.19 Genetic pleiotropy has also been suggested as a contributing factor in polygenic diseases, including conditions that frequently co-exist, such as asthma and obesity,20 or osteoporosis and obesity.21 We have provided evidence for the presence and magnitude of shared genetic influences in depression and migraine headache.
Our findings support previous research on the genetic basis of depression and migraine, considered separately. Family studies have reproducibly shown that major depression is familial.5, 9 A meta-analysis of available twin studies suggests that the heritability of major depression is approximately 37%, with little or no common environmental influences on the phenotype.9 Migraine also appears to be a familial disorder5, 10, 22 with a multifactorial inheritance pattern.22 Twin studies consistently support a genetic component in migraine,23–25 with heritability estimates ranging from 34%24 to 61%.23 Genetic effects may be stronger in women for both depression26 and migraine.25
Despite good evidence of genetic inheritance for depression and migraine, progress has been slow in identifying the genes that underlie these disorders. In a whole-genome linkage study of families that include multiple members with recurrent early-onset major depression, results suggesting linkage were found on regions of chromosome 8p, 15q, and 17p.27 Using microsatellite markers, the chromosome 15q area was finely mapped to the region of 15q25–q26.28 Another study of major depression found evidence for linkage on chromosome 12q.29 Promising loci have been slow to emerge from linkage analyses for reasons that include the use of community samples and the probability that multiple genes make small contributions to the ultimate occurrence of depression.30 In addition, twin studies have demonstrated consistently that individual-specific effects, which encompass gene-environment interactions, are important in disease etiology.9 Hence, obtaining consistent genetic evidence in depression might be made more difficult if genes become relevant only in the presence of environmental triggers. Targeted candidate gene approaches have focused on genes involved in the serotonergic pathway in the brain.31 Polymorphisms in the promoter region of the 5-hydroxylase-tryptamine transporter (5-HTT) protein have been associated with personality traits that may predispose individuals to depression32 or interact with environmental factors such as life stress to result in depression.33 Studies of other candidate genes in affective disorders have yielded mixed results.31
Migraine genetics have been more successful on one front: familial hemiplegic migraine, a rare variant of migraine headache, was successfully traced to mutations in 3 genes encoding neural ion transporters.34 However, these genes have not been clearly linked to more common forms of migraine.11, 34 Evidence for linkage was found for chromosomal regions 4q, 5q, 6p, 11q, 14q, and 18p, but chromosome 4q is the only replicated finding.11, 35 Numerous studies have found significant associations between migraine and targeted candidate genes, but replication of findings and proof of mechanism are absent.11 Among the genes investigated are those involved in neurotransmitter, vascular, and hormonal function.35
No overlap exists in chromosomal locations identified by linkage studies of depression and migraine headache. Etiological theories of depression and migraine have focused on serotonergic function in the brain, because both conditions are therapeutically responsive to pharmacologic modulators of serotonin transmission, including selective serotonin reuptake inhibitors and triptans, respectively. The 5-HTT gene is a promising area of potential shared genetic vulnerability, as it may play a role in the etiology of both depression32, 33 and migraine.36, 37 The increasing availability of whole-genome scans will facilitate investigations into other potential regions where shared genetic vulnerabilities might be pursued. As both depression and migraine appear to be polygenic, it is likely that multiple genes make small contributions to the development of each condition. By focusing attention on shared pathways, researchers can make progress on 2 disease fronts simultaneously.
Our findings also reinforce the possibility of a role for environmental factors or gene-environment interactions in the etiology of depression and/or migraine. Prior research has documented that childhood maltreatment predisposes individuals to both depression and migraine.38 In addition, comorbidities to both disorders, such as obesity, could modify any genetic risk for migraine and depression.39 Obesity in particular may have physiologic40 or psychosocial consequences that directly influence the occurrence of these disorders. Finally, gene-environment interactions such as those demonstrated in depression,33 may also play a role in migraine.
This study has several potential limitations. First, the use of self-reported physician diagnoses could have resulted in misclassification of both depression and migraine. This approach also excludes individuals whose symptoms have not been formally diagnosed or who have limited access to health care. Although the rates that we obtained for each condition are consistent with those of other population-based studies of depression41 and migraine,42 we cannot be certain how our self-report measure would compare to a clinical assessment. We may have underestimated the prevalence of these disorders in our population because of undiagnosed disease or overestimated because of misdiagnosis or misperception of a physician’s diagnosis. In general, however, twins who reported depression or migraine diagnoses were more likely to report characteristic symptom profiles for each condition than were those who did not report a prior diagnosis. Moreover, random misclassification would tend to decrease the strength of associations and would not be expected to differ between MZ and DZ twin pairs. Second, zygosity may have been misclassified, with similar effects on our findings. Third, a wide range of ages was included in the sample. Diagnosis and treatment of these conditions has changed markedly over the last few decades, and therefore the accuracy of our measure might differ among the various age cohorts. While our post-hoc analysis did not show age-related differences in the relative influence of genetic and environmental factors, a larger cohort would be required to formally test hypotheses regarding age. Finally, our sample consisted of female twins only and was predominantly white, thereby limiting the generalizability of our findings.
In sum, using a community-based sample of twins, we provide evidence for a shared genetic vulnerability to depression and migraine headache. Our study goes beyond genetic approaches that focus on rare disorders, such as familial hemiplegic migraine, to suggest potential avenues for research into the more common forms of these two disorders. Further research should confirm our results by using standardized clinical criteria to establish the diagnosis of depression and migraine. Should consistent evidence be found for shared genetics, investigators could conduct association studies that use candidate gene or whole-genome approaches in samples selected for the presence of both depression and migraine. Such research may eventually yield new insights into the pathophysiology of and treatment options for these common conditions. Progress toward effective unified treatment strategies for individuals with depression and migraine headache would be welcomed by clinicians.
The authors report no conflicts of interest. We thank the twins who are taking part in the University of Washington Twin Registry for their time and enthusiasm.