Abstract
Evidence of a familial risk factor in cluster headache is accumulating and studies of twin concordance may resolve family resemblance into genetic and environmental influences. The past literature on cluster headache in twins comprises a few case reports of concordant monozygotic pairs. Swedish twin pairs with a diagnosis of cluster headache were selected through a cross-match of national registers of twin births and hospitalizations. Seventeen discordant twin pairs were found, in which it was possible to verify cluster headache status in 11 complete pairs (two monozygotic, four dizygotic, and five unlike-sexed pairs). In both members of a female monozygotic pair, migraine without aura developed after birth of the first child and remitted by menopause, whereas post-menopausal development of chronic cluster headache occurred in only one of them. The importance of individual specific factors for cluster headache was demonstrated. However, to explain familial aggregation a larger sample of affected twin pairs is necessary.
Introduction
There is increasing evidence of familial aggregation in cluster headache (1–4). Twins are utilized to elucidate the cause(s) of family resemblance (5). A higher concordance risk in monozygotic (MZ) twins than dizygotic (DZ) twins may suggest that genetic effects are important, assuming that MZ twins share no more trait relevant environmental influences than DZ twins. Significant genetic variance has recently been found with respect to migraine headaches among adult twins (6–9) and recurrent headaches among prepubescent twins (10). Twin similarity can also be due to the shared family environment (11, 12).
Eadie and Sutherland reported en passant that one of their cluster headache patients was a MZ twin and that his twin brother also suffered from ‘almost certain migrainous neuralgia’ (13). Couturier and his colleagues noticed that both members of a male MZ pair had cluster headache and thought that their observation was more than coincidental when considering the rare occurrence of cluster headache (14). Three other male MZ pairs concordant for cluster headache have been published so far (15, 16). However, genetic inference cannot be made solely from observations of concordant MZ twins. Further, these selected observations are likely to be biased by the procedure of the publication itself, since discordant twin pairs and DZ twins are less likely to be reported on. We noted that within the 24 pedigrees described by Kudrow and Kudrow (2) there was actually one unlike-sexed twin pair, which was discordant for cluster headache. Here we report on a population-based sample of Swedish twin pairs affected by cluster headache and selected from national health registries.
Subjects and methods
Twin pairs with cluster headache were sampled by linkage of two medical data files, the Swedish Twin Registry and the Swedish Inpatient Registry.
The Swedish Twin Registry is a population-based register of twin pairs in three birth cohorts (1886–1925, 1926–1967, and 1968–1990) (17). Twin births were recorded from nationwide birth registries with little or no deviation of the numbers from the official statistics (18, 19). In 1961, same-sex twin pairs belonging to the oldest cohort and still living within the country as unbroken pairs were traced by means of a system of migration registries kept at the local parish office, county offices, and postal questionnaires (18). In 1971, twin pairs belonging to the middle cohort and still living within the country as unbroken pairs were traced via the national register of the total Swedish population (19). The actual twin register covers, respectively, 78% and 89% of these possible twin pairs. It has subsequently been expanded with all multiple births for the years 1968–1990. The twin register currently consists of 144 406 twins, of whom 49% are men, 51% women, and 82% alive in March 2000. Zygosity has been determined with at least 95% accuracy using the questionnaire method of similarity as children (17). The zygosity distribution of the middle cohort is 16.6% MZ, 26.1% DZ, 36.5% unlike-sexed, and for 20.8% of the twins the zygosity is unknown. Reasons for unknown zygosity are primarily no contact by the registry with twins (born 1958–1967) and inconsistent responses within pairs. The representativeness of twins has been addressed for the old twin cohort using comparison with a random sample of the general population (18). Defined as a population of unbroken pairs, the twin cohort was not representative of the total Swedish population with respect to age and sex. It was argued that the partnership per se could lead to some atypical behaviour in twins. However, only moderate and non-tendentious differences were found between twins and the control group with respect to various environmental, social, and medical variables.
The Swedish Inpatient Registry consists of information on hospital discharges since 1987 from all public hospitals in Sweden and since 1964 for most parts of Sweden. Diagnostic codes for cluster headache, 346,01 (ICD8) and 346C (ICD9), are available between 1969 and 1996. During this time period, a total of 1041 subjects had been hospitalized with these diagnoses, of which 736 subjects had cluster headache as a major diagnosis and 305 subjects cluster headache as a secondary diagnosis. In October 1998, the twin register was cross-matched to the inpatient register via the unique personal number, with 1 January 1969, as the first discharge date and 31 December 1996, as the last follow-up date. Index twins, with a hospital diagnosis of cluster headache, and their cotwins were identified.
To verify the hospital diagnosis in index twins and evaluate cluster headache in cotwins, alive and traceable twins were invited to participate in a telephone interview regarding headaches. These follow-up interviews were conducted in 2001 by a neurologist with extensive experience in headache diagnosis. Physicians were blind to zygosity while headaches were diagnosed using the International Headache Society criteria (20). In addition, medical records regarding cluster headache diagnosis in the index twin were collected. The discordance time was computed as the year at follow-up (2001) minus the year at cluster headache onset in the affected index twin.
The data were collected with permission of the Swedish Data Inspection authority, and the Ethics Committee of Karolinska Institutet and Huddinge University Hospital, respectively.
Results
Seventeen twins (15 men and two women), who belonged to different pairs, had been discharged from Swedish hospitals between 1981 and 1995 with a diagnosis of cluster headache. In 14 twins, cluster headache was the main reason for hospitalization, whereas in three twins cluster headache was stated as a secondary diagnosis. The twins were born between 1924 and 1962, with the 1950s as the most common decade at birth (8/17), and they were evenly recruited from all parts of Sweden. Two twins were MZ (12%), five DZ (29%), seven unlike-sexed (41%), and three of unknown zygosity (18%).
We verified the hospital diagnosis of cluster headache in 14 of 17 index twins (13 men and one woman). It was not possible to confirm the hospital diagnosis of cluster headache in three index twins because they refused to participate in our telephone interview or did not respond to the invitation letter. Medical records for two of these index twins stated a diagnosis of cluster headache but without further details, whereas for one twin the medical record was not retrieved.
Onset of cluster headache in men occurred in the second (2/13), third (10/13), and fifth (1/13) decades of life, respectively. The single female case had her onset at 54 years of age. Four of the index twins suffered from chronic cluster headache, and two of them had the secondary chronic variety. One twin had had two cluster headache periods, and each of them lasted 1 week. All other index twins had had many cluster headache periods of about 1 month in length, though in one twin the cluster period usually lasted 5–6 months. The predominant location of pain was the right side in 11 twins, left side in two twins, and left or right side in one twin. Thirteen of the index twins had a history of smoking, two had experienced an accidental trauma involving head, neck, and back, at some time prior to onset of cluster headache, and three had used narcotic drugs before, or in close proximity to, onset of cluster headache. Nevertheless, we were not able to establish the exact temporal relationship between these possible risk factors and onset of cluster headache.
Three of the 17 cotwins had died prior to our investigation. Cluster headache did not occur in 11 of the 14 remaining cotwin subjects. Three cotwins were not contacted for the telephone interview due to refusal or inability to trace the person. Thus, out of 17 pairs, discordance status of cluster headache was established in a total of 11 pairs; two MZ, four DZ, and five unlike-sexed pairs (Table 1). The discordance time of these twin pairs ranged from 10 years to 31 years.
Cluster headache in 11 Swedish twin pairs
∗Diagnosis of other types of headache between parentheses.
†Sex of the affected cotwin is placed first; sex of the unaffected cotwin is placed second.
‡Discordance time: the year at neurologist interview (2001) minus the year at onset of cluster headache in the index twin.
MZ, Monozygotic twins; DZ, dizygotic twins; US, unlike-sexed twins; F, female; M, male; MR, medical record; NI, neurologist interviews on the telephone; MO, migraine without aura; CTTH, chronic te nsion-type headache; ETTH, episodic tension-type headache.
In the female MZ pair, both the index twin with cluster headache and the non-affected cotwin developed migraine without aura soon after their first childbirth and had remission of migraine attacks after menopause. Episodic tension-type headache was diagnosed in one index twin and two cotwins, whereas chronic tension-type headache was diagnosed in three cotwins.
Discussion
We performed a nationwide search for twins suffering from cluster headache in national health registries, resulting in 17 discordant twin pairs, in which cluster headache status was verified in 11 complete pairs (two MZ, four DZ, and five unlike-sexed pairs). A higher concordance risk in MZ twins than DZ twins would have indicated the importance of genetic variance, but none of the 11 cotwins had cluster headache at follow-up. With the present approach it was not possible to resolve previous observations of increased family risk and MZ concordance into causes due to inherited alleles, shared family environments, or both. However, evidence of a rare major susceptibility allele (21) favours the genetic hypothesis. Two discordant MZ pairs of the present study compared with previous case reports of five concordant MZ pairs could reflect publication bias. This discrepancy in concordance risk may also reflect population differences in prevalence due to variation in environmental exposures and allele frequencies across nations. Further, cluster headache may be a heterogeneous disorder where genetic, or shared environmental, liability is more pronounced among concordant pairs than discordant pairs (22).
In contrast to previous case reports of cluster headache concordant twin pairs, cluster headache discordant pairs of the present study, especially MZ pairs, reflect the importance of the individual's specific environment. Such within-pair differences include endogenous and exogenous environmental factors. Although not sufficient causes, genetic factors may still be necessary causes for disease development. Both members of a female MZ pair carried an autosomal dominant gene, while familial hemiplegic migraine developed in only one of the twins (23). In the present study both members of a female MZ pair had had a history of migraine without aura associated with reproductive events, i.e. onset after birth of the first child and remission after menopause, while post-menopausal development of chronic cluster headache occurred in only one of the twins. We have previously reported that about 10% of female cluster headache patients had onset of cluster headache after 50 years of age, and almost one-third of these women developed the chronic type, whereas the male predominance is attenuated for such late onset cases (24). Further, in patients with a history of both migraine and cluster headache, migraine usually develops first and is then replaced, or continued, with development of cluster headache (25, 26). In a few of our female hospital patients we have also noted a change in the expression of headache type over time, from migraine during the reproductive period, to cluster headache after menopause (unpublished observations). Our identification of an unusual MZ twin pair provides additional support for further investigation into the relationship(s) between migraine, cluster headache, and the female life cycle, where elements such as genes, non-heritable factors, gene–environment interaction, and phenotype–phenotype interaction may be involved. In addition, chronic tension-type headache in the unaffected cotwins also warrants further attention. Moreover, smoking, head trauma, and narcotic drug use were frequently found among affected index twins. Although their presence was not systematically evaluated among the unaffected cotwins, these exogenous environmental exposures could be risk factors for cluster headache and contribute to disease discordance (27).
It is worthy of note that in all unlike-sexed twin pairs of the present study, cluster headache was expressed in the male member and not in the female cotwin, despite the fact that these siblings on average share half of their genomes, have a common intrauterine history, and grew up in the same family environment. In contrast to our findings, cluster headache developed in the female twin but not in the male twin of an unlike-sexed pair belonging to a family with cluster headache in three generations (2).
A non-random sampling strategy is necessary for twin studies of a rare disease such as cluster headache. The frequency of doctor consultation and referral to neurology clinics is high among cluster headache sufferers (3). However, it is not possible to survey twins among patients attending out-patient clinics in Sweden because there is no national register of complete health records or organized nationwide system of clinics for the treatment of cluster headache. We were therefore confined to the in-patient register. Thus, our present results are limited to those cluster headache sufferers that are refractory and intractable cases and likely to be hospitalized.
According to our results the twin–singleton ratio for cluster headache hospitalization in Sweden is 1:60, which is a somewhat higher figure than the expected ratio of 1:80 (28). However, the zygosity distribution of our sample was representative of the twin register, and the 17 twins came evenly from all parts of Sweden. In the present study, the time interval from onset of cluster headache in the index twin to follow-up in the cotwin was at least 10 years, and in most pairs the discordance time exceeded 20 years. In four case reports of concordant MZ twins, the within-pair difference of the age at cluster headache onset was < 10 years (14–16). Therefore, it seems unlikely that our cotwins will develop cluster headache, although they have not passed the age at risk (24).
In the present nationwide selection of twin pairs affected by cluster headache among hospitalized cases, there was no clear evidence that twins are atypical with regard to cluster headache. Nevertheless, population differences and disease heterogeneity cannot be excluded. Despite a large study population, resolution of family resemblance in cluster headache into genetic and shared environmental influences requires improvement of the present ascertainment approach. Administration of a screening protocol (29) would probably overcome most problems related to reduced power and selection bias. Further, pooling of twin samples from several nations might be required to obtain a sufficiently large sample of affected twin pairs. However, our sample of discordant twin pairs demonstrated the importance of individual specific factors in cluster headache, and included one female MZ pair of particular interest.
Footnotes
Acknowledgements
Supported in part by grants from by Merck, Sharp & Dohme (MSD). The Swedish Twin Registry is supported by grants from the Department of Higher Education, the Swedish Scientific Council, and AstraZeneca.