Do Children with Benign Rolandic Epilepsy Have a Higher Prevalence of Migraine than Those with Other Partial Epilepsies or Nonepilepsy Controls?


Address correspondence and reprint requests to Dr. E. Wirrell at Division of Pediatric Neurology, Alberta Children's Hospital, 2888 Shaganappi Trail NW, Calgary, AB, Canada T3B 6A8. E-mail:


Summary: Purpose: Prior studies have given conflicting data concerning the association of benign rolandic epilepsy of childhood (BREC) and migraine but were limited by lack of sensitive, diagnostic criteria for childhood migraine. By using revised International Headache Society (IHS-R) criteria, we compared the prevalence of migraine in children with BREC with that of those (a) with cryptogenic/symptomatic partial epilepsy and (b) without epilepsy.

Methods: Three cohorts of children, gender and age matched (within 1 year) were identified: (a) BREC, (b) cryptogenic/symptomatic partial epilepsy, and (c) no history of seizures. Parents were queried in a standardized interview about migraine and migraine equivalents in their child, and in either biologic parent. Migraine was defined by using the IHS-R (for children) and IHS criteria (for parents). Children with headache were divided into definite (meeting IHS-R criteria), probable (recurrent, throbbing headaches with nausea, vomiting, photophobia or phonophobia, not meeting IHS-R criteria), possible (recurrent headaches with throbbing character or associated nausea/vomiting), or nonmigraine groups. χ2 analysis was used to determine whether the cohort with BREC had a higher prevalence of definite, definite or probable, or definite, probable, or possible migraines or migraine equivalents than the other two cohorts.

Results: Each cohort consisted of 53 children (mean age, 9.8–9.9 years, M/F ratio, 35:18). Those with BREC had higher rates of definite and probable (p = 0.05), of definite, probable, and possible migraine (p = 0.05), and of migraine equivalents excluding motion sickness (p < 0.005) than did those without seizures; however, they did not differ significantly from the cryptogenic/symptomatic partial epilepsy cohort.

Conclusions: Partial epilepsy, regardless of etiology, is associated with higher rates of migraine in children. The pathophysiologic link between epilepsy and migraine is unknown.

Benign rolandic epilepsy of childhood (BREC) is the most common focal epilepsy in childhood with an incidence of 6.2–21 per 100,000 children aged 15 and younger (Heijbel et al., 1975; Cavazzuti, 1980; Astradsson et al., 1998). It accounts for between 13 and 23% of all childhood epilepsies (Kriz and Gadzik, 1978; Cavazzuti, 1980).

Migraine is one of the most common causes of neurologic disability, affecting 4–10% of school-aged children (Sillanpaa, 1983; Mortimer et al., 1992; Abu-Arefeh and Russell, 1994; Raieli et al., 1995). Migraine increases with age, with a striking increase around puberty. Prevalence ranges from ∼3% between 3 and 8 years of age, to as high as 19% by early adolescence. Although the gender ratio is similar before puberty, significantly higher numbers of girls are affected after puberty. The inheritance of migraine both with and without aura is most likely multifactorial and controlled by more than one gene. Genetic and environmental factors play important roles in migraine pathogenesis, with genetic factors being more significant in migraine with aura than in migraine without aura. Congdon and Forsythe showed that 77% of children who had migraine with aura, and 67% who had migraine without aura had at least one first-degree relative with migraine (Congdon and Forsythe, 1979).

The impact of migraine on children and adolescents can be significant. Recurrent headaches may lead to school absenteeism, impaired school performance, and affected interpersonal development resulting from restricted sport, work, recreation, and family activities.

The diagnosis of migraine is based on clinical history and a physical examination that excludes signs of symptomatic causes of headache. The International Headache Society (IHS) has published criteria for this diagnosis (Headache Classification Committee of the IHS, 1988); however, the sensitivity of these “adult” criteria in children is low (Maytal et al., 1997). To address this low sensitivity, revised IHS (IHS-R) criteria were proposed and evaluated in a multicenter, prospective study by the Pediatric Headache Committee of the American Association for the Study of Headache in 1997 (Winner et al., 1997). Overall, the diagnosis of migraine with or without aura was established in 66% of cases by using IHS criteria, but improved to 93% by using IHS-R criteria. In children younger than 12 years, the results were even more striking, with diagnostic rates improving from 49 to 87%.

The incidence of migraine may be increased in children with BREC; however, studies have been limited by small numbers of patients and lack of clearly stated diagnostic criteria for childhood migraine (Kinast et al., 1982; Bladin, 1987; Giroud et al., 1989; Septien et al., 1991; Andermann and Andermann, 1992; Ong et al., 1999; Andermann, 2000). Hence, not all pediatric neurologists are supportive of this association (Santucci et al., 1985; Andermann and Zifkin, 1998).

Determination of whether migraine is more common in children with BREC may assist in our understanding of the pathophysiology of both BREC and migraine and would stimulate research in this area. Better understanding of pathophysiology may lead to more effective treatments. Second, if migraine is significantly more common in BREC, we would be able to offer anticipatory guidance to these children and their families, allowing earlier identification and treatment, to minimize the impact that recurrent migraine may have on school performance and social functioning.

In this study, we compared the prevalence of migraine in children and their biologic parents in three cohorts: BREC, cryptogenic or symptomatic partial epilepsy, and healthy controls, to determine whether migraine was more prevalent in the BREC group.


This study was a matched cohort study of three groups of children, the first cohort with BREC, a second cohort with cryptogenic/symptomatic partial epilepsy, and a third cohort with no history of seizures. The three cohorts were age (within 1 year) and gender matched, as the prevalence of migraine is more common at specific ages and in girls (after puberty). The cohorts consisted of identical numbers; for every child with BREC, one child with cryptogenic/symptomatic partial epilepsy, and one nonepileptic best friend were identified.

Children in all cohorts were aged 5–18 years and had an estimated developmental quotient of ≥70. Children were not required to be receiving active treatment or to have ongoing seizures, at a minimum frequency, to be entered into the study.

BREC cohort

BREC was defined clinically, with all subjects meeting each of the following criteria:

  • a. Typical semiology: Seizures either in wakefulness (simple partial seizures consisting of unilateral paresthesias and/or clonic or tonic activity of the tongue or lower face, dysarthria and/or drooling) or sleep (either simple or complex partial seizures affecting the lower face or secondarily generalized convulsions).
  • b. Typical EEG: High-amplitude, centrotemporal spikes with a characteristic horizontal dipole, with marked activation in drowsiness and non–rapid eye movement sleep.
  • c. If neuroimaging has been performed, no structural lesion accounting for these symptoms.
  • d. Age at onset, 3–13 years.

Cases of BREC were identified through the Neurology Clinic at ACH, by two routes:

  • a. The Neurology Clinic Database was reviewed for children diagnosed with BREC. The diagnosis was then confirmed by review of the clinic chart, and families were then mailed a letter explaining this study.
  • b. Patients seen in clinic with a diagnosis of BREC over the study period were informed of the study at the time of their clinic visits.

The parent or caregiver of each potential subject was then contacted by phone to request participation in this study.

Cryptogenic/Symptomatic partial epilepsy cohort

Cryptogenic/symptomatic partial epilepsy was defined as follows:

  • 1Two or more simple or complex partial seizures, with or without secondary generalization
    • a. At least one EEG showing focal/multifocal epileptiform discharge.
    • b. Seizure semiology not consistent with BREC or benign occipital epilepsy of childhood.
    • c. An imaging study that was either unremarkable or showed a nonprogressive lesion that would not cause headache (i.e., cortical dysplasia, mesial temporal sclerosis, prior ischemia).

For each child with BREC, a child with cryptogenic/symptomatic partial seizures who was age (within 12 months) and gender matched was identified through the Neurology Clinic Database. The parent or caregiver was then mailed a letter explaining this study, which was followed up by a telephone call to request participation.

Nonseizure cohort

For each child with BREC, a child without seizures, who was age (within 12 months) and gender matched, and who was seen in the Orthopedic outpatient clinic at the Alberta Children's Hospital, was identified. The parent or caregiver was given a letter explaining the study at the time of the clinic visit, and either completed the study interview at the time of clinic, or by phone, based on parent preference and time constraints. Children with a history of seizures and central nervous system disorders such as cerebral palsy, spina bifida, or malformations of cortical development were excluded.

Study protocol

The mother or primary caregiver was asked to participate in a structured interview conducted in person or by telephone, depending on family preference, regarding migraine/migraine equivalents in their child, as well as in either of the child's biologic parents. The following information was collected.

Headaches in child or biologic parents

Where possible, we attempted to speak directly to persons with headache to obtain a more accurate headache history. Although the IHS-R criteria for the diagnosis of migraine in children lead to increased diagnostic sensitivity (Winner et al., 1997), they may still fail in 13–26% of those aged 12 years and younger (Winner et al., 1997; Hershey et al., 2005). As the mean age of our study population was only 9.8 years, we elected to include probable and possible categories for migraine. In children, headaches were classified into definite[meeting revised IHS–for children, criteria (Winner et al., 1997)], probable (recurrent headaches with both pulsating quality and nausea, vomiting, photophobia, or phonophobia, but which do not meet IHS-R criteria), possible (recurrent headaches with either a pulsating quality or usually associated with nausea or vomiting) or nonmigraine (recurrent headaches that are not pulsating in character and do not have associated nausea or vomiting), and subjects with definite, probable, or possible migraine were also asked about aura or complex features.

Subjects with epilepsy who described headaches were asked about the timing of their headaches related to seizures. Headaches that consistently occurred only immediately before or after a seizure were classified as periictal or postictal headaches, not migraine. However, further descriptions of the character of these headaches and associated features were obtained.

As the IHS criteria have high diagnostic sensitivity in adults, headaches in biologic parents were classified as definite if they met these criteria (Headache Classification Committee of the IHS, 1988), possible if they had at least two of the following migrainous features including unilateral location, throbbing character, or frequent association with nausea, vomiting, or photophobia, but did not meet IHS criteria, or nonmigraine if they described recurrent headaches that did not meet these criteria.

Migraine equivalents in the child and biologic parents

Migraine equivalents, including benign paroxysmal vertigo, cyclical vomiting, abdominal migraine, Alice in Wonderland, The Rushes, benign paroxysmal torticollis, motion sickness, and aura only were queried for children in each cohort. Biologic parents were also queried about motion sickness and aura only.

Seizure details for the seizure cohorts

The following data were collected for both seizure cohorts: age at onset and at last seizure, age at onset, and discontinuation of antiepileptic drug (AED) treatment (if treated), approximate total number of seizures, previous AEDs used and responses to these medications, current AED therapy (number and type), seizure timing (diurnal vs. nocturnal), prior seizure types (partial simple or complex, generalized tonic–clonic), prior status epilepticus, prior Todd paresis, and current developmental status (mild delay vs. normal). Status epilepticus was defined as a single seizure lasting ≥30 min, or recurrent serial seizures over a period of ≥30 min without regaining consciousness. Developmental status was estimated clinically, based on the child's developmental history and any current academic difficulties, and stratified either as normal (estimated developmental quotient ≥80 and not needing resource help at school) or delayed (estimated developmental quotient <80 and/or needing resource help at school). Medical charts were also reviewed for results of neuroimaging and EEG studies. In seizure cases, a history of seizures in either biologic parent, including description and epilepsy syndrome diagnosis, was obtained.

Statistical analysis

Data analysis was performed using SPSS Version 13.0.

  • 1Are children with BREC more likely to have migraine than age- and gender-matched controls with nonidiopathic partial epilepsy or without seizures? The definite, definite and probable, and definite, probable, and possible categories of migraine diagnosis were combined. χ2 analysis was used to examine whether children with BREC were more likely than their age- and gender-matched controls to have definite, definite or probable, and definite, probable, or possible migraine, with a p value of <0.05 indicating statistical significance.
  • 2Are children with BREC more likely to experience migraine equivalents than age- and gender-matched controls with nonidiopathic partial epilepsy or without seizures?χ2 analysis was used to determine whether children with BREC were more likely than their age- and gender-matched controls to have any migraine equivalent. A second analysis was performed after excluding children with only motion sickness from the migraine equivalent group.
  • 3Are biologic parents of children with BREC more likely than biologic parents of children in the age- and gender-matched control groups to have a diagnosis of migraine?χ2 analysis was used to examine whether biologic parents of children with BREC were more likely than biologic parents of age- and gender-matched controls to have (a) definite vs. possible, nonmigraine or no headache, or (b) definite or possible vs. nonmigraine or no headache. χ2 analysis was also used to determine whether aura-only or motion sickness were more common in parents of children with BREC than in the other two control groups.
  • 4Are children with BREC and migraine more likely to have a higher total number of seizures, history of Todd paresis, status epilepticus, mild delay/learning difficulties, or to have had at least one AED fail for lack of efficacy?

This analysis was restricted to children with BREC and therefore was exploratory, given the small numbers. χ2 was used to examine the relation between migraine diagnosis (definite, probable, or possible) and each of the following variables: total number of seizures, prior status epilepticus, prior Todd paresis, current developmental status (delayed vs. normal), and history of having one or more AEDs fail.

Sample-size calculation was based on comparing the prevalence of definite and probable migraine between children with BREC and nonepileptic controls. A sample size of 50 pairs would have 90% power to detect a difference of 35% between discordant pairs at the 2.5% level of significance.


Fifty-three children in each cohort were recruited. Gender ratio was 35M/18F, in each cohort, and mean ages (SD, range) were 9.8 years (2.6; 5.7–17.7) for the BREC cohort, 9.8 years (2.7; 5.2–17.0) for the cryptogenic/symptomatic partial epilepsy cohort, and 9.9 years (2.6; 5.8–18.2) for the nonseizure cohort (p = NS). Two of 55 cases of BREC approached for the study declined participation and were thus excluded from the study. Four of 57 controls with cryptogenic/symptomatic partial epilepsy did not consent to the interview. For these triads, another similar matched control with cryptogenic/symptomatic partial epilepsy was approached, and all agreed to participate. Seizure data for both the BREC and cryptogenic/symptomatic partial epilepsy are shown in Table 1. Compared with the cryptogenic/symptomatic partial epilepsy controls, cases with BREC were significantly more likely to have simple partial seizures (p < 0.01), to be taking fewer AEDs at present (p < 0.01) and ever (p < 0.03), and they tended to have fewer total seizures (p = 0.05). Of the cryptogenic/symptomatic partial epilepsy group, 15 (28%) had frontal foci, 12 (23%) had temporal foci, three (6%) had frontotemporal foci, three (6%) had occipital foci, one (2%) had parietal foci, eight (15%) had central foci, and 10 (19%) had foci in more than one region. Five (9.4%) of the cryptogenic/symptomatic partial epilepsy cohort had undergone prior epilepsy surgery.

Table 1. Characteristics of seizure cohorts
 BREC cohortCryptogenic/ symptomatic partial epilepsy cohort
Median age at seizure onset in yr (25, 75%ile)7.3 (5.5, 8.5)5.2 (3.0, 7.5)
Seizure type:
 -Simple partial only23/53 (43%)5/53 (9%)
 -Complex partial ± simple partial8/53 (15%)20/53 (38%)
 -2° generalized ± complex or simple partial22/53 (42%)28/53 (53%)
Median number of current AEDs (25, 75%ile)1 (0, 1)1 (1, 2)
Median number of AEDs ever used (25, 75%ile)1 (0, 1)2 (1, 3)
Median number of seizures in past (25, 75%ile) 5 (2, 16)  20 (7, 100)
Proportion seizure free in the past yr17/53 (32%)24/53 (45%)

Prevalence of headache and migraine equivalents in cohorts

The number of cases in each cohort with definite, probable, possible, and nonmigraine headaches, and the number of cases in the two seizure cohorts with exclusively postictal headaches are shown in Table 2. No significant differences were found in prevalence of migraine between cases with BREC and cryptogenic/symptomatic partial epilepsy controls, although the latter group had a higher rate of headache confined to the postictal period (p < 0.02). Compared with nonseizure controls, cases with BREC tended to have higher rates of definite and probable migraine (p = 0.05) and of definite, probable, and possible migraine (p = 0.05). Eight children in the cryptogenic/symptomatic partial epilepsy cohort reported their headaches to occur solely postictally versus none in the BREC cohort. Of these eight, three had headaches resembling definite migraine, one had headaches resembling possible migraine, and four had headaches with no migrainous features.

Table 2. Prevalence of headache in cohorts
 BREC n = 53 (%)Cryptogenic/ symptomatic partial epilepsy n = 53 (%)p ValueaNonseizure n = 53 (%)p Valueb
  1. aComparing cryptogenic/symptomatic partial epilepsy cohort with BREC.

  2. bComparing nonseizure cohort with BREC.

  3. cThree postictal headaches resembled definite migraine, one resembled possible migraine, and four had no associated migrainous features.

Definite migraine  3 (5.7%) 4 (7.5%) 0.701 (1.9%)0.31
Definite or probable migraine 6 (11.3) 6 (11.3)1.01 (1.9)0.05
Definite, probable, or possible migraine11 (20.8) 9 (17.0) 0.624 (7.5)0.05
Nonmigraine24 (45.3)24 (45.3)1.027 (50.9)0.56
No headache18 (34.0)12 (22.6) 0.2022 (41.5)0.42
Postictal headache only08c (15.1)<0.02N/A

No significant differences were found between cohorts for migraine with aura, but numbers of children with this diagnosis were small (BREC, two of 53; cryptogenic/symptomatic partial epilepsy, one of 53; nonseizure, one of 53).

The prevalence of migraine equivalents for each cohort are shown in Table 3. Again, no significant differences were seen between the two epilepsy cohorts. However, compared with the nonseizure cohort, children with BREC had significantly higher rates of benign paroxysmal vertigo (p < 0.03) and of any migraine equivalent, excluding motion sickness (p < 0.005).

Table 3. Prevalence of migraine equivalents in cohorts
 BREC (%)Cryptogenic/ symptomatic partial epilepsy (%)p ValueaNonseizure (%)p Valueb
  1. aComparing cryptogenic/symptomatic partial epilepsy cohort with BREC.

  2. bComparing nonseizure cohort with BREC.

Benign paroxysmal vertigo 5 (9.4%)   7 (13.2%)0.540 (0%)<0.03  
Cyclical vomiting 2 (3.8)5 (9.4)0.240 (0)  0.15
Abdominal migraine 7 (13.2)11 (20.8)0.302 (3.8)0.08
Alice in Wonderland 3 (5.7)3 (5.7)1.0 1 (1.9)0.31
The Rushes 3 (5.7) 6 (11.3)0.300 (0)  0.08
Benign paroxysmal torticollis 1 (1.9)2 (3.8)0.560 (0)  0.32
Aura only 5 (9.4) 8 (15.1)0.371 (1.9)0.09
Motion sickness12 (22.6) 9 (17.0)0.4720 (37.7)0.09
Any of the above equivalents24 (45.3)25 (47.2)0.8522 (41.5)0.70
Any migraine equivalent excluding motion sickness18 (34.0)23 (43.4)0.424 (7.5) <0.005 

Prevalence of headache, aura only, and motion sickness in biologic parents of cohorts

We were able to speak directly to 91% of mothers (BREC cohort, 84%; cryptogenic/symptomatic partial epilepsy cohort, 92%; and nonseizure cohort, 93%) but only to 53% of fathers (BREC cohort, 58%; cryptogenic/symptomatic partial epilepsy cohort, 51%; and nonseizure cohort, 50%) who were reported to have headache. The prevalence of headache, aura without headache, and motion sickness in the biologic mothers and fathers of children in each of the three cohorts are shown in Table 4. No significant differences in either prevalence of migraine, aura only, or motion sickness were found between cohorts, with the exception that mothers of children with cryptogenic/symptomatic partial epilepsy had a significantly higher rate of aura only (p < 0.005).

Table 4. Prevalence of headache, aura only, and motion sickness in biologic parents
 BREC (%)Cryptogenic/ symptomatic partial epilepsy (%)p ValueaNonseizure (%)p Valueb
  1. aComparing cryptogenic/symptomatic partial epilepsy cohort with BREC.

  2. bComparing nonseizure cohort with BREC.

Mother     (n = 53)     (n = 53)      (n = 53) 
 Definite migraine  12 (22.6%)  12 (22.6%)1.0   13 (24.5%)0.82
 Definite or possible migraine17 (32.1)14 (26.4)0.5216 (30.2)0.83
 Nonmigraine30 (56.6)35 (66.0)0.3230 (56.6)1.0 
 Motion sickness25 (47.2)26 (49.1)0.8526 (49.1)0.85
 Aura only3 (5.7)14 (26.4) <0.005  9 (17.0)0.07
Father     (n = 50)      (n = 47)       (n = 52)  
 Definite migraine 5 (10.0) 8 (17.0)0.315 (9.6)0.95
 Definite or possible migraine 7 (14.0) 9 (19.1)0.50 7 (13.2)0.94
 Nonmigraine19 (38.0)26 (55.3)0.0929 (55.8)0.07
 Motion sickness12 (24.0)11 (23.4)0.9515 (28.8)0.58
 Aura only2 (4.0)5 (9.4)0.212 (3.8)0.97

Epilepsy characteristics of children with BREC, stratified for presence of migraine

Of the 53 children with BREC, eight had a history of Todd paresis, two had at least one prior bout of status epilepticus, eight had developmental delay, and 12 had tried more than one AED. The presence of definite, probable, or possible migraine did not predict higher seizure number (p = 0.61), history of Todd paresis (p = 0.69), history of status epilepticus (p = 0.46), developmental delay (p = 0.21), or history of AED failure (p = 0.69).


The literature to date has given conflicting data concerning the association of BREC and migraine (Kinast et al., 1982; Santucci et al., 1985; Bladin, 1987; Giroud et al., 1989; Septien et al., 1991; Andermann and Andermann, 1992; Andermann and Zifkin, 1998; Ong et al., 1999; Andermann, 2000), but all previously published studies were done before publication of the proposed IHS-R criteria. These criteria have improved diagnostic sensitivity, particularly in children aged 12 years and younger (Winner et al., 1997).

In 1989, Giroud et al. compared the incidence of migraine in four groups of children: (1) 28 children with absence epilepsy, (2) 42 children with BREC, (3) 38 children with partial epilepsy, and (4) 30 children with head trauma (Giroud et al., 1989). The incidence of migraine in BREC was 62%, compared with 34% in children with absence epilepsy, 8% in those with partial epilepsy, and 6% in those with a history of head trauma. In 1987, Bladin (Bladin, 1987) followed up 30 cases of BREC and noted that 20 (67%) patients had recurrent headaches during the evolution of seizures, and 24 (80%) developed typical migraine after remission of BREC. Two studies of the Cleveland Clinic suggested that children with migraine are more likely to show benign focal epileptiform discharges; however, none of these children studied had clinical seizures (Kinast et al., 1982; Ong et al., 1999). In contradistinction to these results, Santucci et al. (1985) found no significant difference in migraine prevalence between children with BREC and normal controls.

Although we did find that migraine tended to be more common in children with epilepsy compared with those without seizures, the prevalence did not differ between those with BREC and with cryptogenic/symptomatic partial epilepsy. Unlike the study by Bladin (1987), however, the majority of our children were surveyed during the “active” phase of their BREC, with 68% having had seizures within the preceding year. We therefore cannot comment on whether they will be more likely to develop migraine after remission of BREC. In addition, we found that migraine equivalents (excluding motion sickness) were also significantly more common in the cohorts with epilepsy.

The coexistence of migraine and epilepsy has been well documented in the adult literature. Ottman and Lipton (1994) found that adults with epilepsy had a twofold increased rate of migraine compared with their first-degree relatives. In children, only a single published study exists on this topic. Yamane et al. (2004) compared the prevalence of headache in 50 children with epilepsy with that of their nonepilepsy siblings and found that headache was significantly more common in the epilepsy group (46% vs. 2.5%; p < 0.01). Nearly half of children with headache had migraine, and more than one third described a temporal relation between their headaches and seizures. Some authors have also noted frequent postictal headache in children with occipital epilepsy (Andermann and Zifkin, 1998).

Several recent reports also found that migraine-like headaches occur in approximately one fourth of adults with partial epilepsy in the postictal period (Ito et al., 2004; Yankovsky et al., 2005). In our study, postictal headaches occurred much more commonly in those with cryptogenic/symptomatic partial epilepsy than in the BREC group (21 of 53 vs. four of 53; p < 0.001). However, migraine-like features were present in approximately half of postictal headaches in both cohorts; meaning that only two of 53 children with BREC but 10 of 53 children with cryptogenic/symptomatic partial epilepsy had postictal headaches with clear migrainous features. No child with BREC had exclusively postictal headaches, whereas this pattern was seen in eight of 53 cases of cryptogenic/symptomatic partial epilepsy.

The pathophysiologic link between epilepsy and migraine is unknown. Does an underlying susceptibility predisposes to both conditions, or do seizures predispose to migraine, or vice versa?

As previously noted, migraine has a strong genetic predisposition. However, in our study, we found no difference in the prevalence of migraine in the biologic parents of any of the cohorts, making it less likely that the seizure groups had a higher genetic predisposition to migraine. The rates of migraine found in parents in all three cohorts were consistent with those reported in epidemiologic studies (Sheffield, 1998). Similarly, Ottman and Lipton (Ottman and Lipton, 1996) also found no evidence of a shared genetic susceptibility for epilepsy and migraine.

Cerebral hyperexcitability has been implicated in both disorders. Although the underlying neurobiologic mechanisms for hyperexcitability are not well defined for migraine, dysfunction of mitochondria and low magnesium levels may play a role (Welch, 2005). Additionally, as ion channels directly control membrane excitability of neurons, they may play a pivotal role in both migraine and epilepsy. Also intriguing is the fact that many AEDs, which affect neuronal excitability, also have efficacy in the treatment of migraine (Welch, 2005). In migraine, however, evidence for abnormal channels exists only for one uncommon syndrome, familial hemiplegic migraine (FHM). FHM type 1 is due to a mutant calcium channel, which allows influx of calcium in response to small depolarizations insufficient to open wild-type channels (Bolay et al., 2002; Moskowitz et al., 2004), whereas type 2 results from an abnormal catalytic subunit of a sodium–potassium ATPase (De Fusco et al., 2003). Although various channelopathies have been demonstrated as causes for rare monogenic idiopathic epilepsy syndromes (Mulley et al., 2003), their etiologic role in the majority of epilepsies still must be defined.

Is it possible that recurrent seizures make the brain more susceptible to migraine or vice versa?Moskowitz and colleagues (2004) have shown that cortical spreading depression can activate the meningeal trigeminovascular afferents, resulting in changes in the meninges and brainstem in animals, consistent with activation of trigeminal nociceptive pathways and development of head pain (Bolay et al., 2002; Moskowitz et al., 2004). Although cortical spreading depression is more difficult to elicit in the human cortex, it may occur much more readily in injured human brain (Strong et al., 2002; Parsons et al., 2004). Recurrent seizures might also predispose to this phenomenon, and the particularly high prevalence of migraine-type headaches occurring periictally in both our study and those of others (Ito et al., 2004; Yankovsky et al., 2005) is of interest. Cortical spreading depression leads to significant changes in the composition of extracellular fluid in rats, including increases in glutamate and K+ ions (Somjen, 2001; 2002). These changes will, in turn, result in hyperexcitability and possibly lowering of the seizure threshold. Rare cases of “migralepsy” have been reported in which a classic migraine attack evolves into an epileptic seizure (Mulligan and Bromfield, 2005). Theoretically, repeated bouts of migraine in combination with genetic or environmental factors may lead to focal cortical injury and reorganization.

This study has number of strengths. Children were age and gender matched to control variations in migraine prevalence resulting from these factors. Our study also used well-defined, accepted IHS-R criteria to diagnose migraine in children. However, as most of our study population was younger than 12 years, an age at which diagnostic sensitivity can still be problematic, we included both probable and possible categories of migraine to capture children with probable migraine who did not clearly meet IHS-R criteria. We also specifically queried the timing of headaches in relation to seizures, allowing differentiation of headaches occurring solely during the periictal period, from those occurring at other times.

One possible weakness is that many of our patients in both epilepsy cohorts were taking AEDs, which may also be efficacious in the treatment of migraine (Welch, 2005). However, this potential confounder would lead to an underestimation of the true migraine prevalence in the epilepsy cohorts, meaning that children with partial epilepsy are even more likely to develop migraine than our study shows. Additionally, those with cryptogenic/symptomatic partial epilepsy were more likely to be taking AEDs at the time of study than were those with BREC, leading to greater potential to mask their migraine tendency. Therefore children with nonidiopathic partial epilepsy may be at highest risk of migraine.

A second weakness is that, although we attempted to speak to each parent with a history of headaches to increase the accuracy of our data, this was not possible for nearly half of fathers. As such, the prevalence data for migraine in fathers may be inaccurate.

This study confirms the increased prevalence of migraine in children with epilepsy; however, it does not support the concept that those with BREC have a substantially higher rate of migraine than do other epilepsy types. Further investigation of the underlying pathogenesis of this link may lead to improved understanding of the role of ion channels in focal epilepsies and migraine.


Acknowledgment:  This study was funded by the Alberta Children's Hospital Foundation. We are indebted to Susan Graves, RN, for assistance with the interviews and to Dr. Jean Mah for allowing us to study her patients.