Abstract
We studied four members of a family suffering from typical attacks of familial hemiplegic migraine (FHM) caused by a new mutation, R548C, of ATP1A2 gene in exon 12. One individual had also childhood absence epilepsy and generalized tonic-clonic seizures (GTCS). GTCS were followed by a severe attack of hemiplegic migraine at four times. Sodium valproate enabled control of both the epileptic seizures and the most severe FHM attacks. This association of FHM and epileptic seizures and their control with the same treatment suggest similar pathophysiological mechanisms.
Introduction
Familial hemiplegic migraine (FHM; MIM: 141500) is a rare autosomal dominant variant of migraine with aura, which includes at least two reversible episodes of unilateral hemiparesis or hemiplegia, lasting for 30–60 min. Other aura symptoms include unilateral paraesthesia or numbness, hemianoptic blurring of vision or dysphasia. These symptoms are followed by a severe pulsatile headache lasting a few hours (1). FHM is genetically heterogeneous. Mutations in the CACNA1A gene (FHM1) encoding the α1A subunit of Cav2.1 neuronal voltage-dependent calcium channel have been identified in 50% of FHM families (2). Mutations in the ATP1A2 gene (FHM2), which encodes the Na+/K+-ATPase α2 subunit, are linked to 20% of FHM families. Mutations in the SNC1A gene (FHM3), which encodes a neuronal voltage-gated channel, have recently been described in FHM families (3).
We report a family with FHM phenotype associated with epileptic seizures in one individual due to a new missense mutation in the ATP1A2 gene, the R548C mutation.
Case report
The pedigree (Fig. 1) is four generations of a Caucasian family originating from northern France. Four individuals (II.2, III.1, III.2, IV.1) were examined and diagnoses of FHM were made according to International Headache Society criteria (1). One asymptomatic individual (IV.2) was also examined. Brain computed tomography (CT) scan (II.2, III.1, IV.1), magnetic resonance imaging (MRI) (IV.1) and electroencephalograms (EEG) (IV.1) were performed on individuals, as indicated here in parentheses. The two other affected living members (III.5 and III.6) were not examined, their clinical phenotype being obtained through II.2. After obtaining written informed consent, screening of ATP1A2 gene was performed as previously described (4).

Pedigree of the French family with familial hemiplegic migraine (FHM). Propositus is indicated by an arrow. Symbols represent the hemiplegic migraine (left) and epilepsy (right). Genetic status and current ages are indicated. Wt, wild-type; ND, not done. Numbers above indicate current age or age at death.
The propositus (IV.1) had had attacks of migraine since the age of 4 years. Migrainous aura was initially composed of short occurrences of left arm numbness lasting < 15 min. She had also had childhood epilepsy. Two seizure types were distinguished: brief generalized tonic–clonic seizures (GTCS) since 5 years of age and absence seizures since 6 years of age (Fig. 2).

Discharge of spikes and slow waves (3 Hz/s) linked to episodes of loss of consciousness, which is typical of childhood absence epilepsy at 6 years old (IV.1).
At the age of 5 years, she had had two brief GTCS (< 3 min), followed by a 6-h confusional state and headache. During the second episode, EEG had shown left hemispherical delta wave abnormalities lasting 4 days. Carbamazepine (400 mg/day) was introduced, but, because of the occurrence of typical absence seizures at 6 years of age, carbamazepine was switched to sodium valproate (750 mg/day), and later ethosuximide (250 mg/day) was added. At the age of 10 years, after the absence seizures had been completely controlled for 4 years, all treatments were stopped.
At the age of 12 years, the patient had a new brief GTCS followed by confusion and then agitated coma (stage 1) with fever (39°C). The brain CT scan, MRI, routine blood tests and cerebrospinal fluid were normal. EEG several hours after the GTCS (Fig. 3) was remarkable for diffuse slow activity (delta waves), mostly on posterior regions. She recovered fully within 7 days. One month later, after a minor head injury, she had a brief GTCS followed by a left 1-h hemiparesis. Unilateral pulsating headache accompanied by nausea, vomiting and phonophobia started immediately after resolution of the aura and lasted 5–6 h. EEG during the aura showed right hemispherical delta wave abnormalities. Sodium valproate was re-introduced (750 mg/day). Since this antiepileptic drug was re-introduced, the patient has noticed only mild episodes of headache once per year and she has had normal educational and cognitive development.

Electroencephalogram recorded during the coma episode at 12 years old (IV.1).
Patient III.1 had had stereotypical attacks since the age of 13 years that fulfilled the criteria of HM. Attacks began with blurred vision, aphasia, hemiparesis and hypoaesthesia, mainly on the right side. These symptoms developed within a few minutes and lasted 15 min. Unilateral pulsatile headache accompanied by nausea, vomiting and phono/photophobia started after resolution of aura and lasted < 48 h. The first HM attack was associated with fever and a confusional state, which lasted for approximately 4 days. EEG showed left delta waves during this period, and neurological examination was normal in the interictal period. Since the age of 30, attacks of HM have occurred less than once per year.
Patient III.2 had had stereotypical attacks since the age of 7 years. Auras were characterized by scintillating scotomas affecting the whole visual field and left hemiparaesthesias and hemiparesis, which usually lasted 1 h. Auras were usually followed by bilateral pulsatile headache, nausea, vomiting and photo/phonophobia, which could last 4 days. During adulthood, HM attacks occurred about three times per year. They were treated by paracetamol.
Patient III.5 had his first attack of HM at 20 years old. Attacks occurred about four times per year and could last up to 2 days. He had left- and right-sided hemiparesis. Headaches were treated with paracetamol.
Patient III.6 had had rare attacks of HM after the age of 30 years, which were treated with paracetamol.
Patient II.2 had had her first attack of HM, which was associated with confusion, at the age of 13 years. Subsequent attacks started with brief kaleidoscopic vision, unilateral paraesthesias or hemiplegia over several minutes, and then bilateral pulsatile headache accompanied by nausea and vomiting for 2 days. She had about four attacks of HM per year, treated with paracetamol.
The genetic analysis enabled detection of a missense point mutation within the ATP1A2 gene in all four affected members, which was not detected in the asymptomatic (IV.2) individual (Fig. 1). This C→T transition at nucleotide 1642, within exon 12, led to the switch of an arginine amino acid residue to a cysteine at position 548 (R548C), within the M4-M5 intracytoplasmic loop. This mutation was not detected in a panel of 200 control chromosomes.
Discussion
Here, we report on a novel, R548C, missense mutation segregating in a FHM family. Although functional studies are not yet available, several lines of evidence strongly suggest that this R548C amino acid substitution is disease causing, including: (i) its co-segregation with the affected phenotype within the family, (ii) its absence in a panel of 200 ethnically matched control chromosomes, and (iii) the strong conservation of this R548C amino acid in other species. In addition, we have now identified this mutation in one additional unrelated FHM patient (Riant et al., unpublished data). Interestingly, another mutation affecting this amino acid, R548H, has been detected in several individuals with basilar migraine (5) and in a child fulfilling hemiplegic migraine criteria (Riant et al., unpublished data). Taken together, these data suggest that, depending on the amino acid which is substituted for the arginine at position 548, either a basilar migraine or a hemiplegic migraine phenotype may occur.
One individual, propositus (IV.1), had a unique phenotype: four brief GTCS were followed by confusional states that did not match with usual post-ictal states because (i) the durations were far more prolonged than would have been expected after a brief epileptic seizure, and (ii) they were followed by unexpected clinical symptoms such as pulsating headaches, nausea, vomiting, photophobia, hemiparesis and coma. We considered these episodes as attacks of FHM, because of (i) the family history of hemiplegic migraine, (ii) the previous clinical reports of FHM in which prolonged confusional states and coma were reported beside common symptoms of FHM (6), (iii) the mutation R548H of the ATP1A2 gene that is likely to be pathogenic, and (iv) the re-occurrence of the attacks.
Although the occurrence of a migrainous attack after a seizure was not always the case, each severe attack of hemiplegic migraine of propositus (IV.1) was triggered by an epileptic seizure. Thus, this sequence clearly shows a close temporal connection between epileptic seizure and FHM. Such close relationships between epileptic phenomena and FHM attacks have been occasionally reported. In FHM1 families, seizures (partial and generalized) were reported during FHM attacks in three of 104 subjects with a mutation of the CACN1A1 gene (6). In FHM2 families, two members of two different families who experienced epileptic seizures during their first migraine attacks have been reported (7). One of these people was an 8-year-old boy who experienced a secondary generalized tonic-clonic seizure 2 h after the onset of his first migraine attack. The other was a 5-year-old boy who had several epileptic seizures during his first attack of hemiplegic migraine. This child had clonic movements of the limbs predominately on the right side. These movements were associated with a left adversion of the head and eyes. A similar case has been reported in a 6-year-old child who had a seizure during the course of his first hemiplegic migraine attack (8).
These relationships between FHM and epileptic seizures are far different from those reported in rare FHM families in which an epilepsy syndrome co-segregated. In a Dutch-Canadian family, among patients who carried the same R689E ATP1A2 mutation, 11 had FHM, seven had infantile seizures and four had both FHM and infantile convulsions (9). In two FHM2 families (10), seizure susceptibility was associated with the D718N and the P979L ATP1A2 mutations. GTCS occurred between the ages of 4 and 7 years in the first family and between 8 months and 2 years in the second family, but they did not occur during the course of FHM attacks.
If there is indeed a cause-and-effect link between the occurrence of a seizure and a severe attack of FHM in propositus (IV.1), then antiepileptic drugs should treat both. Of note, sodium valproate controlled both GTCS and severe attacks of FHM in propositus (IV.1). We propose that, for propositus (IV.1), epileptic seizures could facilitate the occurrence of a severe FHM attack. In patients with both FHM attacks and epileptic seizures, seizures may trigger severe FHM attacks. As shown here, reducing the frequency of epileptic seizures may reduce the frequency and severity of FHM attacks. Thus, antiepileptic drugs have to be tried in patients with this association of episodic disorders.
Migraine and epilepsy are two paroxysmal disorders that seem to have some overlapping pathways related to dysfunction of ion exchange. The recent identification of mutations in the SCN1A gene has brought new evidence for a possible link between migraine and epilepsy (3). This gene was found mutated in FHM families and also in a large number of children who had severe myoclonic epileptic syndrome of infancy (Dravet's syndrome). This observation of intricate seizures and FHM is an additional argument for a link between these two episodic brain disorders.
Footnotes
Acknowledgements
The authors thank Larry Koreen, MD for his assistance in the preparation of the manuscript.
