Centrotemporal spikes in families with rolandic epilepsy

Benign epilepsy of childhood with centrotemporal spikes (BECTS), or rolandic epilepsy, belongs to the idiopathic partial epilepsies. According to the international classification of seizures and epilepsies, the syndrome is defined by brief, simple, partial, hemifacial motor seizures, frequently having associated somatosensory symptoms. These partial seizures have a tendency to evolve into generalized tonic-clonic seizures. Both seizure types are often related to sleep. Onset occurs between the ages of 3 and 13 years, with recovery before the ages of 15 to 16 years.¹ Characteristics features are described as follows: 1) a somatosensory onset with unilateral paresthesias involving the tongue, lips, gums, and inner cheeks; and 2) unilateral, tonic, clonic, or tonic-clonic convulsions involving the face, lips, and tongue as well as the pharyngeal and laryngeal muscles causing 3) speech arrest and drooling because of sialorrhea and saliva pooling. At this stage the seizure may end or it may develop into a generalized major convulsion. Nocturnal seizures, the most frequent variant of this syndrome, frequently become generalized (reviewed in ^{reference 2}). Generalized tonic-clonic seizures are, by a wide margin, the most frequently observed seizure type. Nonfacial partial motor seizures have also been reported (reviewed in ^{reference 3}). The EEG hallmark of BECTS are blunt, high-voltage, characteristically shaped centrotemporal spikes (CTS) often followed by slow waves. The spikes are activated by sleep and tend to spread from side to side, frequently with the features of a horizontal dipole.^1,4 At a younger age children with BECTS also tend to show spikes in the posterior temporal-occipital regions that, with maturation, can be replaced by, or occur simultaneously with, spikes in the centrotemporal area (e.g., ^{reference 5}, reviewed in ^{reference 3}). This form of epilepsy accounts for one-sixth of all childhood epilepsies and has a prevalence of more than 1 per 1,000 (e.g., ^{reference 6}, reviewed in ^{reference 3}).

Landau-Kleffner syndrome and epilepsy with continuous spike and wave during slow-wave sleep show major similarities and can overlap with BECTS, but they have a much less favorable prognosis that can include mental deterioration and intractable epilepsy. One view is that they represent a spectrum, with BECTS forming the benign, common end of the spectrum and Landau-Kleffner syndrome and epilepsy with continuous spike and wave during slow-wave sleep the severe, less common end.^7-11 Determining a genetic marker for BECTS will help to resolve this question.

The genetic etiology of BECTS is unequivocal, and, like other common idiopathic epilepsies, a multifactorial mode of inheritance is likely.¹² However, it has been reported that, in families with BECTS, not the epilepsy but the associated EEG trait-CTS-follows an autosomal dominant mode of inheritance with high but incomplete penetrance and age dependency.^13-16 This conclusion was challenged for methodologic reasons.¹⁷ The high population prevalence of approximately 1.5% also favors a heterogeneous or complex mode of inheritance for the EEG trait.¹⁸

In autosomal dominant, nocturnal frontal lobe epilepsy, for the first time a genetic defect in an idiopathic partial epilepsy syndrome was identified. In two unrelated kindreds a mutation in the alpha 4 subunit of the neuronal nicotinic acetylcholine receptor (AChR) resulting in a defective receptor molecule was detected.^19-21 Linkage of BECTS/CTS to the alpha 4 subunit gene was excluded,²² rendering all other known neuronal expressed nicotinic AChR subunits encoded on chromosomes 1 (β₂), 8 (α₂, β₃), 15 (α₃, α₅, α₇, β₄), and 20 (α₄) first-order candidates.^23,24

Methods. Ethical approval for this study was obtained from the Committees on Ethics of the University of Kiel, medical faculty, and Uppsala University, medical faculty.

Pedigrees and diagnosis. Diagnosis of BECTS and CTS was performed according to the international classification of seizures and epilepsies as described.¹ Sleep activation, characteristic location, characteristic shape, and field distribution demonstrated on referential montages as well as classification by two independent individuals were further requirements for determination of the EEG trait. Only a minority of carriers of CTS develop seizures, making the EEG a valuable tool to identify trait carriers.³ Therefore families with index cases were asked for family EEG for scientific reasons only. In case of a positive finding they were entered into the Kiel family archive of epilepsy. Families 1 to 11 were taken from this archive. The remaining 11 families were ascertained by referral. The entire sample contains 22 families with 54 CTS trait carriers, 43 of these with BECTS and 11 without seizures (figure 1). Two individuals showed, in addition to BECTS symptomatology, astatic seizures (P 4, II:1) or atypical absences (P 12, II:1). These features have been described for atypical, benign partial epilepsy of childhood.^25,26

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Pedigree (P) structure and affection status of families 1 to 22. Most likely haplotypes for D15S1002, D15S1048, D15S165, D15S1043, D15S1010, D15S144, D15S1007, and D15S118 exemplified for D15S165, D15S1010, and D151007 were reconstructed by GENEHUNTER. A "0" indicates alleles that could not be typed unambiguously. Recombination events are indicated by asterisks. Individuals affected with benign epilepsy of childhood with centrotemporal spikes and the centrotemporal spikes (CTS) trait are represented by filled symbols and CTS trait carriers without epilepsy by half-filled symbols. All unaffected individuals were considered affection status unknown (affected-only analysis).

Genotyping. DNA was extracted from whole blood by standard techniques.²⁷ Genotyping for 65 polymorphic markers entirely covering chromosomes 1, 8, 15, and 20 was performed in families 1 to 12 using fluorescence-labeled primers and semiautomated techniques as described.²⁸ Twelve markers taken from the Genethon human linkage map²⁹ were added to cover candidate loci densely on 1p + 1q cen (β2), 8p21 (α2), 8p11.2 (β3), 15q24 (α3, α5, β4), and 15q14 (α7). PCRs were performed in 20-µL reaction volume containing 30 ng DNA, optimized for each primer over a range of annealing temperatures (50 to 60 °C) at 1.5 mMol MgCl₂ in 96-well microtiter plates (Costar, Cambridge, MA) by a Biometra Uno machine (Göttingen, FRG). Pooled amplified DNA was electrophoresed on 6% denaturing acrylamide gels using an automated DNA sequencer (377; PE-Applied Biosystems, Foster City, CA). Fragment sizing was performed using Genescan software version 2.02 (PE-Applied Biosystems). Genotyping was done by two individuals blinded to the affection status. Before introducing the affection status, all markers were collapsed to eight alleles and calculated against each other at individual intermarker distances. Recombination events were reexamined to identify typing errors.

Statistical analysis. The MLINK program (version 5.1) of the LINKAGE program package was used for parametric pairwise lod score calculations.³⁰ The GENEHUNTER (version 1.1) program was used for parametric and nonparametric two- and multipoint calculations.³¹ Nonparametric linkage analysis (by GENEHUNTER) was performed in the Z_all mode to include inheritance vector information of the parents in families 15 and 16. Two-point affected sibpair analysis was additionally performed by the SIBPAIR program, which includes a weighing function for multiple sibships and has proven sensitive in earlier studies on sibpair-type families with complex traits (features described in ^{references 32 through 34}). Genetic distances were taken from Dib et al.²⁹

The CTS trait is age dependent, resolving at puberty; therefore, the trait could not be investigated in parents (with the exception of P 15 and P 16). Also, the sensitivity of EEG recordings is unknown but certainly incomplete; therefore, "affected-only" analysis was performed. Trait prevalence was adopted at 1.5%.¹⁸ For parametric analysis penetrances of genetic cases used were 1.0. Penetrances of nongenetic cases were fixed at 0.001. Individuals with the CTS trait with and without epilepsy were considered affected. All individuals entered into the analysis were genotyped rendering allele frequencies unimportant.

Results. A linkage study was conducted in 22 families with BECTS and familial CTS defined by standard clinical and EEG criteria. In an initial screen, markers covering loci on chromosomes 1, 8, 15, and 20 were genotyped in families 1 to 12 only. Nonparametric analysis was employed from the outset because no mendelian mode of inheritance can be assigned unambiguously for the CTS trait.¹⁷ For parametric lod score analysis, an affected-only study rendering penetrances unimportant was applied.

All initially analyzed markers in families 1 to 12 gave lod scores and p values far from significant except for D15S165. This marker gave a (nonparametric) SIBPAIR lod score of 2.24. The alpha 7 subunit of the AChR gene has earlier been mapped to a 7-cM interval between D15S165 (cen) and D15S144 (qter) by Freedman et al.³⁵ Aligning their data to the Genethon map shows that D15S1010-2 cM distal (qter) to D15S165-is localized directly where their highest lod score was obtained.²⁹ D15S1010 and seven other neighboring markers were typed until multipoint information extraction of this 7-cM candidate region exceeded 90%.

Ten additional BECTS families were ascertained during the ongoing study. These families were typed at all eight markers and were included in the analysis. Family structure and affection status for all 22 families are depicted in figure 1. Results are summarized in tables 1 and 2 and in figure 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Multipoint lod score with heterogeneity (HLOD) and nonparametric multipoint linkage score (Z_all score) for families 1 to 22. Genetic loci and the scale in cM are indicated along the x-axis. Inverted arrowheads point at the putative location of the alpha 7 acetylcholine receptor gene as mapped by Freedman et al.³⁵ HLOD-AD = parametric lod score with heterogeneity under an autosomal dominant transmission is represented by a solid line. HLOD-AR = parametric lod score with heterogeneity under an autosomal recessive transmission is represented by a dotted line. Z_all score calculated by GENEHUNTER³¹ is represented by a dashed line.

Analyzing these 22 families jointly, an 8-loci multipoint nonparametric linkage score of 3.33 (p = 0.000494) was obtained by GENEHUNTER at D15S1010. At this position GENEHUNTER calculated a multipoint information extraction of 97%. Parametric two- and multipoint results for autosomal dominant inheritance fell short of the results obtained by nonparametric calculations (see tables 1 and 2 and figure 2). For autosomal dominant inheritance a maximum multipoint lod score with heterogeneity of 2.15 with 66% of the families linked to the locus was obtained. However, analyzing the data under an autosomal recessive mode of inheritance yielded results equivalent (and under heterogeneity even surpassing) those obtained by nonparametric methods. The highest (multipoint) lod score of 3.56 was calculated under an autosomal recessive mode of inheritance with heterogeneity (70% of the families linked to the locus) between D15S165 and D15S1043. Both affected children in pedigree 11 (P 11 in figure 1) do not share a single haplotype and are therefore unlinked to this locus regardless of the mode of inheritance.

Discussion. These data show that the chromosomal region 15q14 harbors a susceptibility locus for the CTS trait in families with BECTS. Peak scores were found in an 8-cM interval spanning D15S165, D15S1010, and D15S1007 encompassing the 7-cM region where the alpha 7 subunit of the AChR was localized by others.³⁵ Highest evidence for linkage was obtained under an autosomal recessive mode of inheritance with heterogeneity in the form of an affected-only study (multipoint maximum lod score 3.56). In one family (P 11) the affected siblings shared no haplotype in common, providing strong proof of heterogeneity. The nonparametric linkage score calculated by GENEHUNTER, which does not allow for heterogeneity, produced a Z_all score of 3.33 (nominal p value < 0.0005).

Prompted by the identical candidate locus approach, linkage and genetic heterogeneity in a large sample of families with juvenile myoclonic epilepsy were reported for the same chromosomal region. Significant lod scores and p values were obtained under an autosomal recessive mode of inheritance with 50% penetrance by Elmslie et al.³⁶ Applying this model to our dataset produced results comparable with those reported for the affected-only study. The maximum multipoint lod score for autosomal recessive transmission with heterogeneity changed from 3.56 (affected only) to 3.16 (50% penetrance; affected/nonaffected).

Maximum lod score curves obtained in both studies overlap but clearly differ. Peak values were reached in an interval corresponding to the area ranging from D15S1007 to D15S118 (see figure 2) by Elmslie et al.³⁶ Data provided by both studies do not allow a precise localization of the linked locus. The presence of heterogeneity makes gene localization more difficult, and therefore it cannot be determined whether both studies point toward the same gene.

Amazingly, the alpha 7 AChR subunit was mapped convincingly in individuals with a familial reduced inhibition of cortical responses to repeated auditory stimuli (P50 auditory evoked response).³⁵ In knock-out mice deficient for the alpha 7 AChR subunit, a novel hypersynchronous neocortical EEG phenotype was recorded.³⁷

Although these results are currently difficult to interpret, it is remarkable that the same candidate approach results in the detection of linkage at the same chromosomal area in a generalized as well as in a partial common idiopathic epilepsy syndrome. Several possibilities are conceivable. Closely neighboring but different genes could be responsible for either phenotype. Different disease-specific mutations in one gene might account for different phenotypes. Defects in the same gene not associated with a specific phenotype but influencing cortical excitability in a general sense seems a third possibility. Mutational analysis of the alpha 7 AChR subunit gene in all three phenotypes is now required to clarify these questions. This might be of potential importance because this receptor type can be influenced by pharmacologic agents.³⁸