Abstract
In order to better understand sporadic hemiplegic migraine (SHM) and particularly its relation to familial hemiplegic migraine (FHM), migraine without aura (MO) and typical migraine with aura (typical MA), we investigated the occurrence of MO and typical MA among probands with SHM and their first-degree relatives. The pattern of familial aggregation of MO and typical MA was assessed by population relative risk calculations. A total of 105 SHM probands and 483 first-degree relatives were identified in the Danish population. Compared with the general population, SHM probands had no increased risk of MO, but a highly increased risk of typical MA. First-degree relatives of all SHM probands had an increased risk of both MO and typical MA, whereas first-degree relatives of probands with exclusively SHM had no increased risk of MO but an increased risk of typical MA. Our data suggest that SHM is a genetically heterogeneous disorder.
Keywords
Introduction
Sporadic hemiplegic migraine (SHM) has clinical characteristics similar to familial hemiplegic migraine (FHM), but without other affected cases in the family (1). It has been described previously in the literature that patients with attacks of SHM also have attacks of migraine without aura (MO) and/or typical migraine with aura (typical MA) (2, 3). However, it is presently unknown whether co-occurrence of MO and typical MA among probands with SHM and their first-degree relatives is increased compared with the general population.
In order to better understand SHM and particularly its relation to FHM, MO and typical MA, we here analyse the risk of MO and typical MA among probands with SHM and their first-degree relatives compared with the general population.
Materials and methods
Data collection
A systematic search was performed employing three different search strategies of the Danish population of 5.2 million inhabitants. The search included a computer search of the National Patients Register, screening more than 27 000 case records from private practising neurologists and headache clinics and advertisements for patients with hemiplegic migraine. All probands with SHM from this study population were included. Probands and first-degree relatives were interviewed and diagnosed according to the operational diagnostic criteria of the International Headache Society (IHS criteria) (4) by using a validated semistructured telephone interview. However, for the purpose of the present study SHM and typical MA were analysed separately. SHM was defined as MA including motor weakness where there were no other relatives affected with MA including motor weakness, whereas typical MA was defined as MA that did not include motor weakness. All first-degree relatives >15 years were interviewed, whereas first-degree relatives of ≤ 15 years were only contacted for an interview, if they were suspected of having or previously having had headache or aura symptoms. A neurological research fellow experienced in headache diagnostics (L.L.T.) did all interviews of SHM probands. Furthermore, SHM probands had a face-to-face interview and a physical and neurological examination performed by a neurological resident (E.O.) (5). Three medical students thoroughly trained in headache diagnostics and interview techniques (M.K.E., S.F.R., I.A.) did all interviews of their relatives. To make sure that no relatives had had hemiplegic migraine (that the proband was a true ‘sporadic hemiplegic migraine case’), all second or third or more distant relatives, that were suspected of having had migraine and/or aura symptoms, were also interviewed. However, only probands and interviewed first-degree relatives were included in the analysis. A more detailed description of the methodology and the study population has been given elsewhere (5).
The ascertainment of relatives of probands with SHM and their response and participation pattern is shown in Fig. 1. The participation rate was high among those who it was possible to contact; 96% (339/354) of first-degree relatives were interviewed. The reasons for not participating were: first-degree relative declined to participate for unspecified reasons (n = 11) and contact was impossible to obtain (n = 4).
Ascertainment of first degree relatives of probands with sporadic hemiplegic migraine.
A total of 53 first-degree SHM relatives were not contactable and therefore not interviewed. The primary reason was death (n = 41). Other reasons were dementia (n = 1), unspecified disease (n = 4), unknown address (n = 2) and no contact with the family according to the proband (n = 5).
Among first-degree relatives, 76 children of ≤ 15 years did not have a history of migraine and were not contacted. The project was approved by the Danish ethical committees.
Statistical methods
All probands and exclusively first-degree relatives >15 years of age were included in the calculations. The observed prevalence of MO and typical MA among first-degree relatives was estimated by dividing the number of relatives with either condition by the number of first-degree relatives.
The risk of familial occurrence was assessed by estimating the population relative risk of the condition in specified groups of relatives (6). The risk was calculated according to the following equation:
[Prob (Relative is affected/Proband is affected)/Prob (Random member of the population is affected)]
A family aggregation is implied when this risk ratio significantly exceeds 1.
Because the prevalence of the conditions depends on sex and age (Table 1) the value of the denominator was adjusted according to the distribution of sex and age in the group of relatives studied (7). Hence, this standardized population relative risk was estimated by dividing the observed number of affected first-degree relatives by the expected number according to the prevalence rates in the population. The expected number was calculated by adding the products of the current sex- and age-specific rates (8) and the number of relatives within each corresponding sex/age category. The adjusted population relative risks of MO and typical MA were estimated separately. The χ2 test was used. A 5% level of significance was used. The 95% confidence limits of the prevalence ratios and the population relative risk were estimated according to the assumption that the numbers of those affected followed a binomial distribution.
Sex- and age-specific prevalence of typical migraine with aura and migraine without aura per 1000 inhabitants (8)
∗Prevalence estimated to be 10% higher than in the age group 30–39.
Results
A total of 105 SHM probands and their 339 first-degree relatives were identified from the Danish population. Of the latter, 316 first-degree relatives were above age 15 and 23 first-degree relatives were age 15 or below.
The prevalence of MO and typical MA among probands with SHM and their first-degree relatives
Table 2 shows the occurrence of MO and/or typical MA among SHM probands and their first-degree relatives. SHM and typical MA co-occurred in 59% of probands and typical MA occurred in 20% of their first-degree relatives. SHM and MO co-occurred in 29% of probands and MO occurred in 23% of their first-degree relatives.
Migraine without aura (MO) and typical migraine with aura (typical MA) among probands with sporadic hemiplegic migraine (SHM) and among their first-degree relatives
Population relative risk of MO and typical MA among probands with SHM
Because the prevalence of the conditions (MO and typical MA) depends on sex and age (Table 1), a sex and age standardized population relative risk was estimated for SHM probands (Table 3) and first-degree relatives of different groups of SHM probands (Table 4).
Sex- and age-standardized risk of migraine without aura (MO) and typical migraine with aura (MA) among probands with sporadic hemiplegic migraine (SHM)
Sex- and age-standardized risk of migraine without aura (MO) and typical migraine with aura (MA) among first-degree relatives in different groups of probands with sporadic hemiplegic migraine (SHM)
BM, Basilar migraine symptoms during SHM attacks, fulfilling the IHS criteria for basilar migraine.
SHM probands had a numerically but non-significant 1.7-fold [95% confidence interval (CI) 1.0, 2.4] increased risk of MO compared with the general population (Table 3). This did not differ from the expected, when considering the sex and age distribution of our material and the prevalence rates of MO in the general population (P = 0.07, Fisher's exact test). On the other hand, SHM probands had a highly significant 7.7-fold (95% CI 3.6, 11.8) increased risk of typical MA compared with the general population (Table 3). This did differ from the expected, when considering the sex and age distribution of our material and the prevalence rates of typical MA in the general population (P < 0.001; χ2 test).
Population relative risk of MO among first-degree relatives of probands with SHM
Compared with the general population, the total group of first-degree relatives of probands with SHM had a 1.5 times (95% CI 1.2, 1.8) significantly increased risk of MO compared with the general population.
However, first-degree relatives of probands exclusively with SHM (no attacks of MO or typical MA) had no significantly increased risk of MO (Table 4). A similar pattern was seen among first-degree relatives of probands with SHM and co-occurrence of MO, who also had no increased risk of MO compared with the general population (Table 4).
Population relative risk of typical MA among first-degree relatives of probands with SHM
Compared with the general population, first-degree relatives of probands with exclusively SHM had a 2.0 times significantly increased risk of typical MA (Table 4). Also first-degree relatives of probands with SHM and co-occurrence of typical MA had a 3.1 times significantly increased risk of typical MA (Table 4). A similar pattern was seen among the total group of first-degree relatives of probands with SHM, who had a 2.5 times (95% CI 2.0, 3.0) significantly increased risk of typical MA compared with the general population.
Table 5 shows the effect of selected variables on the population relative risk of MO and typical MA among first-degree relatives of probands with SHM. Sex or age of proband at onset of SHM attacks did not influence the population relative risk.
Effects of sex and age on population relative risk of migraine without aura (MO) and typical migraine with aura (MA) among first-degree relatives of probands with sporadic hemiplegic migraine
Population relative risk of MO and typical MA in relation to basilar migraine symptoms
Compared with the general population, the risk of MO among first-degree relatives of SHM probands with basilar migraine symptoms fulfilling the criteria of basilar migraine (BM) (4) was increased 1.6 times (95% CI 1.3, 1.9), which was significant. Whereas the risk of MO among first-degree relatives of SHM probands without BM symptoms was increased 1.1 times (95% CI 0.6, 1.6), which was not significant.
Compared with the general population, the risk of typical MA among first-degree relatives of SHM probands with BM symptoms was increased 2.7 times (95% CI 2.1, 3.3), which is significant, whereas the risk of typical MA among first-degree relatives of SHM probands without BM symptoms was increased 1.9 times (95% CI 0.95, 2.8), which was not significant. Thus, the risk of MO and typical MA among first-degree relatives was influenced by whether the proband had SHM with BM symptoms or SHM without BM symptoms (Table 4).
Discussion
Methodological considerations
Our study is based on a representative group of thoroughly characterized sporadic hemiplegic migraine (SHM) probands identified from a systematic search of the Danish population (1). The sex- and age-standardized population relative risk among the first-degree relatives was calculated using Danish prevalence rates from the general population (8, 9). However, since only all first-degree relatives above age 15 had been interviewed, the population relative risk was calculated among first-degree relatives above age 15.
Since migraine is a subjective complaint where the diagnosis relies exclusively on the information provided by the patient, it is important to minimize interviewer bias. A low diagnostic variability was secured by having one neurological resident with experience in headache diagnosis to conduct all interviews of probands. Three medical students interviewed the relatives. To minimize diagnostic variability they were simultaneously trained and thoroughly supervised in interview techniques and headache diagnosis by the neurological resident who interviewed the probands. Furthermore, the interviewers were blinded with respect to proband status, i.e. co-occurrence of MO and/or typical MA. The patients were diagnosed according to the universally accepted diagnostic criteria of the IHS using a validated semistructured telephone interview (8).
SHM, typical MA and MO were analysed separately because epidemiological, clinical, pathophysiological and familial data have indicated that they are different disorders (7–;12). Thus, the present study met high standards for a genetic epidemiological study of SHM. A more detailed description of the methodology has been reported elsewhere (5).
The FHM gene and its possible involvement in SHM
The calcium channel gene, CaCNA1A, causing FHM in about 50% of the FHM families, was the first identified migraine gene. Recently, a second gene, ATP1A2, was identified on chromosome 1 in FHM (13). Since there are many clinical similarities between FHM and SHM, previous studies have examined whether the CACNA1A gene could also be involved in SHM (3, 14–;17). So far three different missense mutations have been identified in the CACNA1A gene, causing SHM (3, 14–;16), the Y1385C mutation in just a single sporadic case (3, 16), the R583Q mutation in both a single sporadic case (15) and also in several familial cases (3), whereas the T666M mutation has been identified both in two sporadic cases (3, 14, 15) and in several familial cases (3). Thus, so far only two missense mutations in the CACNA1A gene have been demonstrated to cause both SHM and FHM. This suggests that only a small fraction of SHM patients are unrecognized FHM cases, whereas the majority have a different genetic background.
The prevalence of MO and typical MA in probands with SHM
Probands with SHM had no significantly increased risk of MO, but a significantly increased risk of typical MA. This could be due to a shared genetic mechanism between SHM and typical MA, but speaks against a shared genetic mechanism between SHM and MO. Alternatively, the genetic mechanism causing SHM may also be expressed as attacks of typical MA but less likely as attacks of MO.
The population relative risk of MO among first-degree relatives of probands with SHM
First-degree relatives of all SHM probands had a slightly increased risk of MO (1.5; 95% CI 1.2, 1.8), which was comparable to the increased risk of MO previously found among first-degree relatives of probands having exclusively attacks of MO (1.86; 95% CI 1.56, 2.16) (7). However, we found no increased risk of MO either in first-degree relatives of probands having exclusively SHM (no attacks of MO or typical MA) or in first-degree relatives of probands with SHM and co-occurrence of MO.
This could indicate that MO and SHM in its pure form (without co-occurrence of MO or MA) are two different entities, and furthermore, that MO and SHM with co-occurrence of MO are two different entities. However, since first-degree relatives of all SHM probands have a significantly increased risk of MO, which is comparable to the increased risk of MO previously found among relatives of probands having exclusively attacks of MO (7), our data suggest a relation between SHM and MO and this relation seems confined to families where the proband also has attacks of MO and typical MA.
The population relative risk of typical MA among first-degree relatives of probands with SHM
In first-degree relatives of probands having exclusively attacks of SHM (no attacks of MO or typical MA), we found a two-times (95% CI 1.1, 2.9) significantly increased risk of typical MA, whereas in first-degree relatives of SHM probands with co-occurrence of typical MA, we found a 3.1 times (95% CI 2.1, 4.1) significantly increased risk of typical MA compared with the general population. Thus, SHM patients have a significantly increased risk of typical MA among their first-degree relatives and the increased risk has a tendency to increase further when SHM co-occurs with MA (P > 0.05).
These results suggest that SHM, both in its pure form and when co-occurring with typical MA, could be caused by genetic factors that might also be expressed as typical MA. However, it is presently unknown whether typical MA in families of probands with SHM is similar to typical MA when occurring in families without hemiplegic attacks.
Population relative risk of MO and typical MA in relation to BM symptoms
First-degree relatives of probands with SHM with BM had a significantly increased risk of both MO and typical MA compared with the general population, whereas first-degree relatives of SHM probands without BM did not have a significantly increased risk of either MO or typical MA. These data could support the hypothesis that SHM probands with BM are more probably ‘pure’ SHM cases compared with SHM probands without BM, who could belong to the more heterogeneous group of SHM patients. However, this hypothesis awaits further genetic studies.
The results in the light of the migraine spectrum model
It has previously been suggested, that typical MA and FHM might be part of a so-called ‘genetic migraine spectrum’ (18, 19). If using this model (the multifactorial model), the individual becomes sick only when the liability exceeds a certain threshold level (20). In the model FHM represents the most severe attacks (with the highest genetic influence/lowest threshold), SHM represents the medium severe attacks and typical MA represents mild attacks. An FHM-affected patient would consequently have on average a higher personal liability than an SHM-affected individual, who again would have a higher liability than a typical MA-affected individual. This higher liability would manifest itself as a higher frequency of affected relatives. Thus, assuming that FHM, SHM and MA belong to the same disease entity, with different thresholds for showing the disease phenotypically, we would expect to see a higher risk of typical MA among first-degree relatives of probands with FHM and SHM compared with probands with MA.
As probands with MA previously have been shown to have an increased risk of MA of 3.79 (95% CI 3.21, 4.38) among their first-degree relatives (7), we would expect the increased risk among first-degree SHM relatives to be above the increased risk in MA relatives; however, we found the opposite and therefore conclude that the ‘genetic migraine spectrum hypothesis’ is highly unlikely in view of the present data.
The present SHM data suggest that the attacks of typical MA seen in probands and among first-degree relatives of probands with SHM could be a different kind of MA attack compared with attacks of MA in families without SHM. Future studies will show if ‘typical MA’ attacks in SHM probands and among first-degree relatives of probands with SHM could perhaps be abortive SHM attacks.
The significance of the present results
Despite the use of specific and strict diagnostic criteria when identifying our material of SHM patients, the present results clearly demonstrate that SHM, aetiologically, is a heterogeneous disorder.
Among SHM cases (non-familial hemiplegic migraine cases), several groups of patients probably exist: first, patients in whom the disease is due to a single mutation in one of the FHM genes. In such patients, the disease appears as non-familial because the patient has a de novo mutation of one of the FHM genes and his/her parents do not have the mutation, or because there is an incomplete penetrance and one of his/her parents has a mutation but no affected phenotype; or finally because there is a false paternity and the ‘real’ biological father of the patient has FHM. As estimated previously, possibly about 20–30% of SHM patients may have had an affected relative who for various reasons could not be detected and are thus undiagnosed FHM (5).
Second, SHM patients exist in whom the disease is not caused by a mutation in any FHM gene. In such patients, the disease is poorly understood and is aetiologically heterogeneous. Some of these patients could share common mechanisms with other varieties of migraine and other patients could have a completely different genetic background. Thus, some might have typical MA that just happened to have a couple of severe attacks including motor weakness. Such patients will contribute to an increased risk of typical MA but not MO among first-degree relatives, since this previously has been found among first-degree relatives of probands having exclusively attacks of MA (7). Others might be patients with MO including motor weakness of a non-migrainous nature, contributing to an increased risk of MO but also of typical MA among first-degree relatives, as previously has been described in probands having exclusively attacks of MO (7). The rest could be patients without migraine but with motor weakness of a non-migrainous nature erroneously diagnosed as hemiplegic migraine.
As more genes are found for FHM and possibly typical MA, it will become possible better to seed out the different groups of patients with the SHM phenotype.
Footnotes
Acknowledgements
We thank the many sufferers of sporadic hemiplegic migraine and their unaffected relatives, who agreed to participate. We also thank our many colleagues for their excellent collaboration. The work was supported by grants from the Cool Sorption Foundation of 1988, the Foundation for Research in Neurology, the Danish Headache Society, The A.P. M⊘ller Foundation for advancement of medical science, the Novo Nordisk Foundation, the IMK-Almene Foundation, The GlaxoWellcome Research price and Ms Else Torp Foundation.