Epilepsy: A Comprehensive Textbook
2nd Edition

In 1994, autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) was first described as an inherited form of partial epilepsy. ADNFLE is characterized by clusters of brief motor seizures, which occur mostly during non–rapid eye movement (REM) sleep.¹⁰⁶ Nonspecific aura phenomena including epigastric, sensory, or psychic symptoms sometimes precede the seizures. The seizures start with gasps, grunts, or some vocalizations followed by thrashing hyperkinetic activity or tonic stiffening of the limbs, often with superimposed clonic jerking. The consciousness is only infrequently impaired. Seizure frequency can be highly variable even within the same family. In some of the patients seizures are frequent and tend to cluster, while in others seizure occurrence is more sporadic and long periods of spontaneous remission are reported. In general, seizure frequency tends to decrease in later adulthood. Physical examinations and brain imaging results are usually normal. Interictal electroencephalographic (EEG) abnormalities are rarely seen, and only some patients show ictal epileptiform activity. The onset of ADNFLE is usually during childhood or early adolescence, with a penetrance of 70% to 80%. Seizures are often, but not in all patients readily controlled by antiepileptic drugs, especially carbamazepine.

Mutations in two genes have so far been identified in different ADNFLE families.^46,121 CHRNA4 and CHRNB2 encode the α4- and β2-subunits of the neuronal nicotinic acetylcholine receptor (nAChR), respectively. The first CHRNA4 mutation, a S248F amino acid exchange within the second transmembrane domain, was identified in 1995.¹²¹ Descriptions of other nAChR mutations, either in CHRNA4 or in CHRNB2, followed.^62,73,98,122 With the exception of one mutation (CHRNB2-I312M in transmembrane domain III), all known ADNFLE mutations are located within the second transmembrane domain. Both the second and the third transmembrane domain contribute to the walls of the ion channel; thus, it seems that only mutations that have a direct effect on the ion pore are able to cause ADNFLE. In total, four CHRNA4 and three CHRNB2 mutations have been described. These include six amino acid exchanges and one small in frame insertion. Some of those mutations have been found again in unrelated families. They offer the opportunity to study the effects differences in genetic and ethical backgrounds might have on the penetrance of the disorder and the clinical features associated with these mutations (Fig. 1).

The nAChRs, like the glycine, GABA_A, and serotonin receptors, are members of the large family of ligand-gated ion channels. They are constituted by the assembly of five subunits arranged pseudosymmetrically around the central axis, forming a cation-selective ion pore. Pharmacologic and ligand-binding studies have shown that, depending on their subunit composition, the different subtypes of homo- or heteropentameric nAChRs not only vary with respect to their expression patterns in the brain, but also display different biopharmacologic channel properties.¹¹ Both α4- and β2-subunits are found in almost all brain regions, and the heteropentameric α4/β2 receptor is thought to be the major high-affinity nicotine nAChR subtype in mammalian brain. The surprising and so far not very well understood part is that an nAChR subtype ubiquitously present in brain is able to cause a partial type of epilepsy rather than a generalized one. In support of the recognized importance of the cholinergic system in brain functions, recent work has shown that nAChR mutations not only cause epilepsy, but can also be associated with additional neurologic and psychiatric features or cognitive deficits.^29,62,80 Examples are the CHRNA4-776ins3 mutation and the CHRNA4-S252L mutation. The 776ins3 mutation was found in 11 members of a large Norwegian ADNFLE family. Six of the 11 mutation carriers not only had epilepsy, but also had serious psychiatric problems, mostly schizophrenia, negative symptoms of schizophrenia, or recurrent unclassified psychosis. These observations suggest that the 776ins3 mutation may be a risk factor for psychiatric disorders.⁸⁰ The S252L mutation has been described in two unrelated families of Korean and Japanese origin. In both families the seizures not only tended to be resistant to carbamazepine treatment, but were also associated with a high rate of mental retardation and/or behavioral problems.^29,62 The similarities of the clinical phenotype in two families with different genetic background but the same S252L-ADNFLE mutation strongly suggest that the associated neurologic features are mutation specific rather than caused by modifying genes or environmental factors (Fig. 2).

Functional analysis carried out with recombinant nAChR receptors in Xenopus oocytes or HEK cells, including patch clamp characterization of several known ADNFLE mutations, revealed different biopharmacologic profiles.¹¹ Co-expression of the S248F mutation and the wild-type CHRNA4 allele yielded acetylcholine-evoked currents of amplitudes comparable to wild-type receptors but with a higher sensitivity for the natural agonist. Receptors carrying either the 776ins3, S252L, or V287M mutation displayed increased acetylcholine sensitivity but to different degrees compared with the S248F mutation. A reduction of calcium permeability was observed for the mutants S248F and 776ins3 but not for the S252L mutation.¹¹ Increased acetylcholine sensitivity and therefore a gain-of-function effect is the only functional feature common to all known ADNFLE mutations. It is tempting to speculate that the particular functional signatures of each mutant contribute to the above-described observations of associated neurologic features in ADNFLE, while the gain-of-function effect might be responsible for the epilepsy phenotype. The latter part of this hypothesis is supported by the observation that, compared to wild-type litter mates, mice carrying gain-of-function mutations in the CHRNA4 gene are dramatically more sensitive to nicotine-induced seizures. Furthermore, nicotine application in these mice resulted in enhanced hippocampal theta rhythms, as demonstrated by in vivo electrophysiologic recordings. This observation would be consistent with a model in which gain-of-function effects in presynaptically located nAChRs activate inhibitory GABAergic interneurons in the neocortex and hippocampus. The enhanced GABA release could then increase network excitability by inhibiting inhibitory pathways. Alternative models for the epileptogenic effect of nAChR mutations could include a disturbance of hippocampal “peacemaker” theta activity due to increased input from septal cholinergic neurons.

Benign familial neonatal convulsions (BFNC) is a rare autosomal dominant seizure disorder of the newborn (see also Chapter 223). The disorder is characterized by an age of onset between the first day and, latest, the fourth month of life. The seizures are mostly unprovoked, generalized, or multifocal and of the tonic and/or clonic type. The course of the disorder is often benign and self-limiting, and, with or without pharmacotherapy, in most patients the seizures remit spontaneously within a few days or weeks.^103,137 Later in life seizures can reoccur in up to 15% of the patients, starting mostly at school age or in young adulthood. These so-called late-onset seizures tend to be infrequent and are often provoked, for example, by lack of sleep. Several untypical BFNC patients with a more severe course of the disorder have been described. These patients often show a higher frequency of seizures and seizures still occur after the age of 4 months. Follow-up studies often revealed moderate delays of psychomotor development, and in some patients have a family history positive for epilepsy and mental retardation. Two BFNC families have come to attention in which a mutation carrier developed drug-resistant seizures and/or epileptic encephalopathy shortly after birth, resulting in severe psychomotor retardation. In one of the two severely affected index patients a de novo mutation was found, while the second index patient inherited his mutation from a parent with typical “benign” BFNC.^36,109 It remains unclear if these two patients had a second yet unrecognized condition or if certain risk factors (e.g., perinatal hypoxia) in combination with a BFNC mutation increase the risk for such an unfavorable outcome.

The molecular basis of BFNC is a mutation in the voltage-gated potassium channel genes KCNQ2 or KCNQ3 (chromosome 20q13.3 and 8q24, respectively).^13,27,115 Both the KCNQ2 and KCNQ3 genes encode ion channel subunits that are composed of six transmembrane domains. The subunits have identical structures including a voltage sensor in transmembrane domain 4, a loop between transmembrane domains 5 and 6 that builds the ion channel pore, and a long C-terminal region of unknown function. The BFNC mutations found so far in KCNQ2 are either missense mutations located in one of the transmembrane domains or truncating mutations (including nonsense, insertion/deletions, and splice site mutations) located mostly in the C-terminal region. Most mutations are private, meaning that they have not been found in other BFNC families. So far, only three mutations have been found in KCNQ3. All known KCNQ3 mutations are missense mutations located within the vicinity of the pore region.

KCNQ2 and KCNQ3 encode subunits of the M-channel, a very slowly opening and closing potassium channel that is ubiquitously found in brain.¹⁴¹ The M-current, first discovered some 20 years ago,²¹ is a powerful controller of neuronal repetitive firing. M-currents regulate the number of action potentials of individual neurons by opposing sustained membrane depolarization. M-channels activate at membrane potentials that are more negative than the action potential threshold and at which few other ion channels are active. Therefore, they can be assumed to have a pivotal role in the stabilization of membrane potentials and are likely to control excess neuronal excitability and prevent seizures. Expression studies in Xenopus oocytes and HEK cells have shown that BFNC mutations cause only modest reductions (20% to 30%) of potassium currents in reconstitution experiments. It seems, therefore, that even slight alterations of M-channel activity are sufficient to cause seizures.¹³

Exceptions to this haploinsufficiency concept are some rare KCNQ2 mutations that appear to have a dominant negative effect on channel function, as demonstrated by a >50% reduction in current magnitude. One of these mutations is KCNQ2/R207W, which causes the BFNC/myokymia syndrome. The mutation abolishes the third of six positive charges in transmembrane domain 4 that is thought to represent the voltage sensor in the cation channel superfamily. The R207W amino acid exchange was the first KCNQ2 mutation described that causes a disorder with symptoms outside the central nervous system; it is associated not only with episodic symptoms, but also with continuous symptoms. On the clinical level, heterozygosity for the R207W mutation causes both BFNC and myokymia, a spontaneous and repetitive involuntary contraction of muscle fiber groups.³⁷ In the BFNC/myokymia syndrome, multifocal or generalized tonic–clonic convulsions typically start around day 3 after birth and disappear spontaneously after a few weeks or months. Electromyographic (EMG) recordings in the patients are initially normal but,
starting in infancy, reveal spontaneous repetitively discharged normal motor unit potentials of 20 to 60 msec in duration. Generalized myokymia with spontaneous finger twitching occurs some years later. The association with both central and peripheral neurologic symptoms might be explained by the unusual electrophysiologic profile of the underlying mutation. R207W causes a loss of K⁺ current, whose magnitude depends strongly on the pattern and time course of depolarization. With short periods of depolarization the loss of current is more severe than with other known BFNC mutations. The dominant negative effect is not observed with longer periods of activation. Thus, at long (>1 second) trains of action potentials or during seizure activity, R207W mutant channels may be activated significantly and may even reach wild-type levels. It is possible that the time-dependent dominant negative effect of R207W establishes itself only in the peripheral nervous system and that this might be the reason for peripheral neurologic symptoms like myokymia. This hypothesis rests on the assumption that the neuronal activity of motoneurons is interrupted by longer quiescent periods than is the activity of central neurons whose hyperexcitability triggers seizures. Under such conditions, the dominant negative effect of the R207W mutation would have consequences for motoneurons but not for central neurons.³⁷

In mice, KCNQ2 knockout experiments resulted in early postnatal lethality in homozygous animals, while hemizygous knockout mice developed and behaved normal but showed an increased sensitivity to pentylenetetrazole, an epileptic inducer.¹⁴² Transgenic mice that conditionally express a dominant negative KCNQ2 pore mutation in the brain only during defined developmental periods demonstrated the critical role of M-channel activity in early postnatal brain development. Suppression of the M-current during the first postnatal weeks was associated with spontaneous seizures, behavioral hyperactivity, and morphologic changes in the hippocampus.⁹⁶ These results support the notion that M-currents are critical determinants of cellular and neuronal network excitability. The changes in brain morphology found in transgenic mice also raise the interesting question of whether BFNC caused by KCNQ2 mutations in humans might be (at least in part) due to submicroscopic structural brain changes (morphologic etiology) rather than exclusively to a reduction of M-currents (functional etiology).

Benign familial infantile convulsions (BFIC) can be distinguished from BFNCs by a later age at onset¹³⁵ (Fig. 3). The autosomal dominant disorder is characterized by the onset of seizures around 6 months of age (range 3 to 9 months). Seizures occur in clusters and respond well to antiepileptic drug treatment. The seizures are partial with secondary generalization and in most patients spontaneous remission has occurred by the age of 12 months. Usually no subsequent neurologic abnormalities develop. Two putative loci were reported, BFIC1 (19q) and BFIC2 (16p12-q12), but no genes have been identified yet.^23,54 Interestingly, three other disorders with overlapping neurologic features map to the same region as BFIC2: The infantile convulsions and choreoathetosis syndrome (ICCA; OMIM 602066),¹²⁴ paroxysmal kinesigenic choreoathetosis (PKC; OMIM 128200),¹²⁹ and a syndrome comprising rolandic epilepsy, paroxysmal exercise-induced dystonia, and writer’s cramp (EPRPDC; OMIM 608105).⁵³ Therefore, the possibility exists that the four disorders are allelic.

A rare seizure disorder with an age of onset intermediate between BFNC and BFIC is benign familial neonatal/infantile convulsions (BFNIC). BFNIC has been shown to be caused by mutations in the voltage-gated sodium channel subunit gene SCN2A^9,61 (Fig. 3).

Febrile seizures (FSs) affect 5% to 10% of children under the age of 6 years. Twin and family studies demonstrated the involvement of genetic factors in the etiology of febrile seizures. In most patients an oligo- or polygenic background rather than a monogenic one can be assumed, and a substantial degree of heterogeneity is likely. However, rare families with an apparent autosomal dominant mode of inheritance have been described, and linkage studies identified several putative gene loci, including FEB1 on chromosome 8q13-q21, FEB2 on 19p, FEB3 on 2q23-q24, FEB4 on 5q14-q15, FEB5 on 6q22-q24, and FEB6 on 18p11.2.^{64,81,88,90,92,138} In one family with febrile and afebrile seizures, a nonsense mutation (S2652X) was identified in the MASS1 gene. MASS1 is part of the large G-protein coupled receptor gene VLGR1.⁹⁰ The mutation causes a deletion of the C-terminal 126 amino acid residues in the predicted MASS1 protein. If confirmed in independent families, MASS1 can be regarded as a rare cause for familial febrile convulsions. In another family a missense mutation in the SCN1A gene chromosome 2q23-24 has been found to cosegregate febrile seizures, indicating that familial febrile seizures should also be considered a channelopathy.⁸¹

Febrile seizures are the most common seizure type in families with the recently described syndrome of generalized epilepsy with febrile seizures plus (GEFS+) (see also Chapter 256), followed by febrile seizures plus (FS+; seizures with fever may persist beyond the age of 6 years and/or may be associated with variable afebrile seizures). Afebrile seizure types in affected GEFS+ individuals include generalized tonic–clonic, myoclonic, absence, and atonic seizures or, at least in some families, partial seizures¹⁰⁵ (Fig. 4). Given the highly variable clinical phenotype, it is not surprising that the mode of inheritance underlying GEFS+ is still a matter of debate. Although in some families the trait is likely to be autosomal dominant, in others it is probably better described as oligogenic or as a major gene effect. A genetic concept involving more than one gene would also better fit the clinical variability observed within and between GEFS+ families. The first GEFS+ mutation (C121W) was identified in the SCN1B gene on chromosome 19q13.1.
SCN1B encodes a small accessory subunit of the voltage-gated sodium channel. The mutation C121W can be predicted to disrupt a disulphide bridge and to interfere with the ability of the β-subunit protein to modulate the function of the much larger, pore-forming α-subunits.¹⁴⁰ Subsequent studies showed that most GEFS+ mutations are located in the gene coding for the α1-subunit of the voltage-gated sodium channel, SCN1A, on chromosome 2q24.⁴¹ SCN1A codes for a large ion channel protein composed of four domains, each containing six transmembrane segments (S1 to S6). Minor GEFS+ genes are SCN2A, coding for the γ2-subunit gene of the voltage-gated sodium channel, or the γ2-subunit gene of the GABA_A receptor (GABRG2).^{8,56,66,82,123,139}

Mutations in the SCN1A gene have also been identified in the majority of patients with the syndrome of severe myoclonic epilepsy of infancy (SMEI = Dravet syndrome; also Chapter 230) and in some cases of phenotypically overlapping syndromes like borderline SMEI (SMEB) or intractable childhood epilepsy with generalized tonic–clonic seizures (ICEGTC).^31,45 The clinical phenotype of SMEI is characterized by an initially normal psychomotor development, onset of febrile seizures within the first year of life, and subsequent manifestation of various afebrile seizure types including absence, myoclonic, and partial seizures. During the second year of life psychomotor delay becomes evident.

Although caused by the same gene, SCN1A, differences in the mutation detection rates and the mutational spectrum present in GEFS+ and SMEI patients are obvious.⁸⁷ SCN1A mutations are only found in 5% to 10% of GEFS+ patients but are present in 33% to 100% of SMEI patients (the variation is likely to be due to different mutation detection methods and/or patient inclusion criteria). All known GEFS+ mutations are missense mutations that cause amino acid exchanges in the encoded protein, while SMEI mutations are composed of truncating (nonsense or frame shift) mutations (47%), missense mutations (43%), splice site mutations (7%), and deletions (3%). Nearly all reported SMEI mutations (76 of 80) are de novo; only a few of them (4 of 80) were inherited from a parent with a less severe type of epilepsy. Mutations in SMEI tend to be either truncating and/or located in functionally critical parts of the gene. GEFS+ mutations are mostly found in less highly conserved parts of SCN1A such as the distal parts of S1 to S4 segments or in the short loops connecting those segments (Fig. 5). It is therefore likely that SMEI and GEFS+ mutations differ with respect to their predicted impact on ion channel function. SMEI mutations can be expected to cause a more severe disturbance of protein function, consistent with the more severe phenotype.

The exact functional effects of SMEI and GEFS+ mutations and their correlations to clinical phenotypes are still unclear. Expression studies of SCN1A channels with different SMEI and GEFS+ mutations in Xenopus oocytes or HEK293 cells showed a variety of biophysical aberrations including complete loss of function, altered gating properties, and even gain-of-function effects.^77,78,101 These conflicting results can probably partially be explained by different experimental setups, since the biophysical properties of voltage-gated sodium channels strongly depend on the expression system used in the experimental setup. SCN1A channels expressed in HEK293 cells are known to differ from those expressed in Xenopus oocytes with respect to their kinetic properties and their sensitivity to β-accessory subunits. Furthermore, to be able to compare the different functional studies, the question needs to be addressed whether SMEI and/or GEFS+ mutations alter channel surface expression by affecting gene transcription, mRNA stability, protein folding, or trafficking.

Another syndrome of early childhood epileptic encepha- lopathy that is phenotypically and etiologically related to both SMEI and GEFS+ is ICEGTC. Both GEFS+ and ICEGTC are severe types of early childhood epilepsy that are characterized by fever sensitivity, intractable seizures, and developmental decline after seizure onset. Contrary to GEFS+, myoclonic or absence seizures usually do not occur in ICEGTC patients. SCN1A mutations were found in 7 of 10 ICEGTC patients. All mutations were missense mutations that were, similar to SMEI mutations, either located in highly conserved (S4 to S6, pore loop) or, as in GEFS+, less conserved parts of the gene (S1 to S4). Two of the seven published ICEGTC patients inherited their SCN1A mutations from a parent with GEFS+-like epilepsy, raising the possibility that additional genetic and/or environmental factors contribute to the more severe phenotype in ICEGTC.⁴⁵

The syndrome of autosomal dominant lateral temporal lobe epilepsy (ADLTE, also named autosomal dominant partial epilepsy with auditory features [ADPEAF]) is characterized by simple partial seizures with mainly acoustic and sometimes even visual hallucinations.¹⁴⁴ The age of onset is usually in the second or third decade of life. In some families the seizures tend to start with a brief sensory aphasia without reduced consciousness.⁵¹ In 2002, positional cloning efforts let to the identification of the LGI1 gene (leucine-rich glioma inactivated gene 1) on chromosome 10q24 as the gene responsible for ADLTE/ADPEAF.^{44,51,65,85,100} The mutational spectrum includes both missense mutations and truncating mutations, and no mutation hotspot has emerged yet. There are no obvious clinical differences between patients with truncating mutations and those with amino acid exchanges in the LGI1 gene.

So far the function of the LGI1 gene, which shows a strong expression in neurons within the temporal lobe, is mostly unknown. The LGI1 protein has a distinctive leucine-rich repeat (LRR) motif in its N-terminal end. This motif might be indicative of either receptor function or an interaction with the extracellular matrix. The LRR motif consists of repeated β-strands and α-helices connected by loops, building a domain that usually serves as a framework for the formation of protein–protein interactions and is present in a large number of proteins with diverse functions. The C-terminal half of the LGI1 protein contains seven epilepsy-associated repeats (EARs). EARs are characterized by tandem repeats with a core of about 50 residues that probably folds into a β-propeller structure.¹²⁰ The EARs were also identified in another epilepsy-relevant protein, MASS1/VLGR1, which is mutated in audiogenic epilepsy in mice and in one family with febrile seizures (see above).^91,116 Within the MASS1/VLGR1 protein, the epitempin repeat is part of the ligand-binding ectodomain, suggesting that this repeat, like the LRR domain, might be involved in protein–protein interactions.

LGI1 was first found to be interrupted by a translocation breakpoint in a glioblastoma cell line and was therefore labeled as a candidate tumor suppressor gene for gliomas.²⁸ This theory seemed to be supported by the observation that LGI1 expression is low or absent in high-grade gliomas.¹²
However, it has been demonstrated that LGI1 is mainly expressed in neurons rather than in glia cells.⁵² Therefore, a gradual loss or displacement of neurons during progression toward higher malignancy would be a plausible explanation for the low or absent LGI1 expression in high-grade gliomas. The absence of LGI1 expression would then represent a secondary effect rather than a causative event. This interpretation is supported by studies that found fewer LGI1-expressing cells in high-grade gliomas compared with low-grade tumors.¹² Furthermore, no evidence was found that the LGI1 mutations present in ADLTE families increase the rate of brain tumors or other malignancies. It is therefore unlikely that LGI1 acts as a first step or high penetrance tumor suppressor gene.²⁰ Nevertheless, experimental evidence exists that LGI1 plays a role in the suppression of cell migration. For example, experimentally forced expression of LGI1 significantly reduced the proliferation and migration ability of glioma cells. Invasion assays showed that glioma cells transfected with LGI1 lost their ability to migrate freely. These results suggest that LGI1 is involved in cellular control mechanisms concerning migration and invasiveness. Thus, the possibility remains that, although not a bona fide tumor suppressor gene, LGI1 plays a role in determining the malignancy grade of an already existing tumor.⁷⁰ It will be interesting to know if LGI1 also plays a role in the migration of neuronal cells during embryogenesis. ADLTE/ADPEAF mutations that interfere with such a function could theoretically cause subtle changes in the brain’s cytoarchitecture that might cause seizures through abnormal neuronal networks.

Seizures and myoclonus are common to many neurodegenerative and metabolic disorders (see also Chapter 261). In the following paragraphs some of the disorders in which epilepsy and/or myoclonus constitute a major feature are exemplarily described.

The chromosomes located in the nucleus are not the only source of coding DNA in our cells. Mitochondria, the reminders of an ancient endosymbiosis between early eucaryotic cells and a proteobacteriumlike ancestor over 1.5 billion years ago, possess their own circular DNA molecule (mtDNA). The original bacterial genome lost most of its genes during evolution, and the few genes that survived on the mtDNA until today can be separated into two groups. Mitochondrial genes either code for proteins that take part in the process of aerobic respiration or code for the mitochondria’s transcription/translation system. Mitochondrial inheritance differs from mendelian inheritance in several aspects. The inheritance is strictly maternal, since the sperm mitochondria are selectively eliminated at fertilization. Mutations in mtDNA that seriously harm or abolish the ability of mitochondria to perform their various cellular processes are always heteroplasmic. Heteroplasmic disorders are characterized by a mixed intracellular population of normal and mutated mitochondria. With each cell division it is a matter of chance how many mutated and how many wild-type mitochondria each daughter cells inherits. Hence, heteroplasmy is one of the reasons for the great clinical variability typical for most mitochondrial disorders (Fig. 6).

Cytogenetically visible chromosomal rearrangements often cause the loss or gain of several or many genes. Since approximately 50% of our genes are, at least temporarily, expressed in brain, the resulting cross-genetic imbalance can be expected to cause major structural or functional disturbances in the central nervous system. It is therefore not surprising that seizures are a common finding in these patients.¹⁰⁷ Most chromosomal rearrangements occur de novo in one of the parental germ cells. Thus, even taken into account the rare possibility of a parental germ cell mosaicism, the recurrence risk for siblings is small in these families. Nevertheless, parents need to be karyotyped to exclude the possibility of familial chromosomal rearrangements. Such familial chromosomal rearrangements are mostly balanced translocations where two or more chromosomes exchanged fragments or whole chromosomes are fused together. The carrier of such a balanced translocation is phenotypically normal since he or she has the correct gene copy number, but can have germ cells with an unbalanced chromosomal status. This increases the risk for spontaneous abortions and/or the birth of a severely affected child. Given the number of chromosomes in our cells and size of our genome, the possibilities for individual chromosomal rearrangements are countless. Most of them have been described only in a single or few patients and no typical patterns of clinical features are known. Exceptions are structural rearrangements of certain chromosomal regions that, often due to the presence of repeated DNA elements, have an above-average chance to occur. The underlying mechanism is often a mispairing and incorrect exchange between homologous chromosomes in meiosis. An example that is frequently associated with epilepsy is the inv dup(15) marker chromosome.³⁰ Additional examples for clinically well-defined syndromes caused by structural chromosomal aberrations are the ring chromosome 20 syndrome, Miller-Dieker syndrome (see also Neuronal Migration Disorders), most cases of Angelman syndrome, Wolf-Hirschhorn syndrome, and the 1p36 deletion syndrome.

In the human cortex, neuronal migration starts at approximately 7 weeks of gestation. Neurons originating from the proliferative ventricular zone migrate radially along
specialized glial processes to their final locations. The correct migration depends on several factors, one of the most important being the contact between the migrating neuron and the radial glial fiber. This requires the interaction of adhesion molecules, trophic factors, and guidance molecules, all of them possible candidates for the various disorders of neuronal migration. A second mode of migration has been recently described, in which cells born in the ventricular zone migrate perpendicular to radial glial processes. This nonradial pathway is used by most future inhibitory interneurons. Disturbances of this complicated pattern of cortical development and migration can cause severe neurologic disorders that are often accompanied or characterized by seizures (Tables 2 and 3). Neuronal migration disorders can be noninherited or genetic with distinct inheritance patterns. The malformations found in nongenetic forms are often (but not always) focal, while in the genetic ones the pattern of abnormal neuronal migration is usually generalized. The group of genetic neuronal migration disorders includes different lissencephaly syndromes, disorders with neuronal dysplasia, abnormal neuronal proliferation, and several conditions with heterotopia. Two of the five genes underlying major syndromes with lissencephaly as well as the tuberous sclerosis complex will be exemplarily described here. (See Chapter 259 for further discussion).