The literature reports six concordant and two discordant monozygotic twin pairs and one concordant (unlike sex) and nine discordant (four same and five unlike sex) dizygotic twin pairs (3,7,25,33,37,38,45). The concordant twin pairs were all case reports, except for two twin pairs from clinic populations, while the discordant twin pairs were from a population-based twin study. Interpretation of the twin data should take into account the limited sample size and publication bias (21).

Probands from 12 clinic population-based studies from 1947 to 1985 reported that 47 first-degree relatives were affected in 1182 families (31). This suggests an increased family risk, even though the diagnosis may not have been confirmed in all families. More recently, an Italian and a Dutch survey with physician-confirmed diagnoses reported familial occurrence in 2.3% (5/222) and 4.1% (70/1720) of the families, respectively (20,51).

Four genetic epidemiologic surveys provide more complete information about the relatives (10,14,15,28,29). Table 86-1 illustrates the risk to first- and second-degree relatives based upon a population prevalence of cluster headache of 1 per 500 inhabitants. The first-degree relatives had a five- to 18-fold increased risk of cluster headache compared to the general population. Second-degree relatives had a one- to threefold increased risk. Varying observations may, at least in part, be explained by methodologic differences. The French survey acquired a more comprehensive dataset, as all first-degree relatives were directly interviewed by a physician (10). A physician interviewed possibly affected relatives in the Danish survey (28,29). This causes the risk of cluster headache to be a minimum figure, as probands may fail to report about affected relatives. The American survey was based on probands reports only (14). This can cause either under-or overestimation of the risk of cluster headache depending on whether the probands misclassify their relatives more or less than they fail to report about the affected.
The Danish probands reports were inaccurate, as the diagnosis of cluster headache was confirmed in only 57% while the remaining 43% had migraine (29). The Italian survey was based on probands reports, and possibly affected relatives were interviewed by a physician (15). Eleven of the 57 affected relatives had probable cluster headache. This is either an under- or overestimation of the risk of cluster headache, depending on whether the number of those with probable cluster headache is more or less than those affected not reported by the probands. The significantly increased familial risk strongly suggests that cluster headache has a genetic cause. Theoretically, a shared environment can produce relative risks of the magnitude observed for cluster headache only under extreme conditions (13).