CRITICAL REVIEW AND INVITED COMMENTARY
The etiologic classification of epilepsy
Address correspondence to Simon Shorvon, UCL Institute of Neurology, National Hospital for Neurology and Neurosurgery, PO Box 5, Queen Square, London WC1N 3BG, United Kingdom. E-mail: email@example.com
The etiology of epilepsy is a major determinant of clinical course and prognosis, yet the current classifications of epilepsy do not list etiology in any detail. In this article, a classification (database) of the etiologies of epilepsy is proposed. In this scheme, the etiology of epilepsy is divided into four categories: idiopathic, symptomatic, provoked, and cryptogenic. These are defined and subcategories are proposed. A commentary addressing the following points is included: problems associated with assigning causation, symptomatic versus idiopathic epilepsy, focal versus generalized epilepsy, acquired epilepsy, acute symptomatic epilepsy, risk factor analysis, provoked epilepsy genetic and developmental epilepsy, and epilepsy as a disease not a symptom.
It is remarkable how few attempts have been made to develop a synoptical listing of causes of epilepsy, or indeed how little the etiology of epilepsy has been considered in the official classifications of epilepsy, which have largely focused on seizure semiology and electroencephalography. This is despite the oft-repeated mantra that “epilepsy is a symptom of underlying neurologic disease.” A major purpose of any classification scheme is to provide a framework for clinical practice, and an emphasis on etiology is one central aspect of this. The diagnostic process in clinical practice comprises two stages: (1) the classification of the seizure type/syndrome and (2) the assignment of cause. The classifications of epilepsy have focused largely on the former and not the latter, but etiology is as important and is a major determinant of treatment, prognosis, and clinical course. Etiology was mentioned only in passing in the 1969, 1981, and 1989 International League Against Epilepsy (ILAE) classifications of seizures and syndromes and later schemes (Gastaut, 1969, Commission on Classification and Terminology of the International League Against Epilepsy 1981, 1989, Engel, 2001; Luders et al., 2003). In their recent report, the ILAE Commission for Classification of the Epilepsies has addressed etiology and divided epilepsies into three categories (genetic, structural/metabolic, unknown cause), but they do not list etiologic categories any further (Berg et al., 2010). The Commission recommended that a future classification will be flexible and multidimensional, in essence a “database forming the basis of a diagnostic manual.” It is in response to this aspiration that the etiologic scheme presented herein is conceived, as a database of etiologies. The purpose of this commentary is to propose a framework for such a database of etiology, and also to suggest that a detailed consideration of etiology should be an important axis or dimension of a classification scheme. It is hoped that this will be prove a useful synoptical overview of all the etiologies underpinning epilepsy. It should be added, however, that this etiologic dimension is only one of the multidimensional aspects of a classification, and as a “database” does not replace other aspects of the ILAE classification systems.
Any consideration of cause should also pay attention to the historical development of the concept, and a brief historical survey is attempted in the accompanying article (Shorvon, 2011). In relation to classification, it can be noted that in the earlier ILAE classification schemes the emphasis on seizure semiology and electroencephalographic features was perhaps largely because electroencephalography (EEG) was the major investigatory modality available. With advances in imaging and molecular chemistry, it is now possible to identify an etiologic basis for many types of epilepsies, and one can muse on how different the ILAE classifications of epilepsy might have been for instance if magnetic resonance imaging (MRI) had been invented before EEG.
Suggested Scheme for a Classification of Epilepsy by Etiology
The classification (database) of etiologies of the epilepsies is divided into four main categories (Table 1).
Table 1. Suggested scheme for an etiological classification of epilepsy
|Idiopathic epilepsy||Pure epilepsies due to single gene disorders||Benign familial neonatal convulsions; autosomal dominant nocturnal frontal lobe epilepsy; generalized epilepsy with febrile seizures plus; severe myoclonic epilepsy of childhood; benign adult familial myoclonic epilepsy|
|Pure epilepsies with complex inheritance||Idiopathic generalized epilepsy (and its subtypes); benign partial epilepsies of childhood|
|Symptomatic epilepsy |
Predominately genetic or developmental causation
|Childhood epilepsy syndromes||West syndrome; Lennox-Gastaut syndrome|
|Progressive myoclonic epilepsies||Unverricht-Lundborg disease; Dentato-rubro-pallido-luysian atrophy; Lafora body disease; mitochondrial cytopathy; sialidosis; neuronal ceroid lipofuscinosis; myoclonus renal failure syndrome|
|Neurocutaneous syndromes||Tuberous sclerosis; neurofibromatosis; Sturge-Weber syndrome|
|Other neurologic single gene disorders||Angelman syndrome; lysosomal disorders; neuroacanthocytosis; organic acidurias and peroxisomal disorders; prophyria; pyridoxine-dependent epilepsy; Rett syndrome; Urea cycle disorders; Wilson disease; disorders of cobalamin and folate metabolism|
|Disorders of chromosome function||Down syndrome; Fragile X syndrome; 4p-syndrome; isodicentric chromosome 15; ring chromosome 20|
|Developmental anomalies of cerebral structure||Hemimegalencephaly; focal cortical dysplasia; agyria-pachygyria-band spectrum; agenesis of corpus callosum; polymicrogyria; schizencephaly; periventricular nodular heterotopia; microcephaly; arachnoid cyst|
| Predominately acquired causation||Hippocampal sclerosis||Hippocampal sclerosis|
|Perinatal and infantile causes||Neonatal seizures; postneonatal seizures; cerebral palsy; vaccination and immunization|
|Cerebral trauma||Open head injury; closed head injury; neurosurgery; epilepsy after epilepsy surgery; nonaccidental head injury in infants|
|Cerebral tumor||Glioma; ganglioglioma and hamartoma; DNET; hypothalamic hamartoma; meningioma; secondary tumors|
|Cerebral infection||Viral meningitis and encephalitis; bacterial meningitis and abscess; malaria; neurocysticercosis, tuberculosis; HIV|
|Cerebrovascular disorders||Cerebral hemorrhage; cerebral infarction; degenerative vascular disease; arteriovenous malformation; cavernous hemangioma|
|Cerebral immunologic disorders||Rasmussen encephalitis; SLE and collagen vascular disorders; inflammatory and immunologic disorders|
|Degenerative and other neurologic conditions||Alzheimer disease and other dementing disorders; multiple sclerosis and demyelinating disorders; hydrocephalus and porencephaly|
|Provoked epilepsy||Provoking factors||Fever; menstrual cycle and catamenial epilepsy; sleep-wake cycle; metabolic and endocrine-induced seizures; drug-induced seizures; alcohol and toxin-induced seizures|
|Reflex epilepsies||Photosensitive epilepsies; startle-induced epilepsies; reading epilepsy; auditory-induced epilepsy; eating epilepsy; hot-water epilepsy|
|Cryptogenic epilepsiesb|| || |
- 1 Idiopathic epilepsy—defined here as an epilepsy of predominately genetic or presumed genetic origin and in which there is no gross neuroanatomic or neuropathologic abnormality. Included here are epilepsies of presumed multigenic or complex inheritance, but for which currently the genetic basis has not been elucidated.
- 2 Symptomatic epilepsy—defined here as an epilepsy of an acquired or genetic cause, associated with gross anatomic or pathologic abnormalities, and/or clinical features, indicative of underlying disease or condition. We thus include in this category developmental and congenital disorders where these are associated with cerebral pathologic changes, whether genetic or acquired (or indeed cryptogenic) in origin. Also included are single gene and other genetic disorders in which epilepsy is only one feature of a broader phenotype with other cerebral or systemic effects.
- 3 Provoked epilepsy—defined here as an epilepsy in which a specific systemic or environmental factor is the predominant cause of the seizures and in which there are no gross causative neuroanatomic or neuropathologic changes. Some “provoked epilepsies” will have a genetic basis and some an acquired basis, but in many no inherent cause can be identified. The reflex epilepsies are included in this category (which are usually genetic) as well as the epilepsies with a marked seizure precipitant.
- 4 Cryptogenic epilepsy—defined here as an epilepsy of presumed symptomatic nature in which the cause has not been identified. The number of such cases is diminishing, but currently this is still an important category, accounting for at least 40% of adult-onset cases of epilepsy.
It must be emphasized that there are obviously cases for which categorization is difficult and to a significant extent arbitrary. Such difficulties are encountered particularly in relation to: (1) The differentiation of “presumed genetic” (i.e., idiopathic) and “cryptogenic,” for instance in relation to the benign focal epilepsies or even the idiopathic generalized epilepsies. This distinction is arbitrary, but the idiopathic category is kept separated here because of the general clinical features and/or age specificity of these conditions, which point strongly to a presumed genetic (including epigenetic and epistatic) etiology. (2) The categorization of some of the childhood syndromes. Some of these are included under the “idiopathic grouping,” but evidence of a genetic basis is only presumptive, for instance, the benign rolandic epilepsy. Others are included in the symptomatic epilepsy despite the fact that there is a presumption of a genetic cause in at least a proportion of cases, for instance in the West or Lennox-Gastaut syndromes. (3) The categorization as “symptomatic” those single gene disorders without clear neuropathologic abnormalities but which have systemic metabolic features and which cause a wide range of symptoms of which epilepsy is only one (e.g., Rett syndrome, Angelman syndrome, progressive myoclonic epilepsies). There are arguments for including these conditions within the category of idiopathic epilepsy.
Problems with assigning causation
There are a number of factors that complicate the assignment of cause, and these are often overlooked. These will have an impact on any attempt to classify epilepsies according to cause:
- 1 Multifactorial cause of epilepsy: As has been frequently reiterated, epilepsy is in the great majority of cases multifactorial. In any individual case, the epilepsy is often (perhaps almost always) the result of both genetic and acquired influences and also influenced by provoking factors (and many epilepsies clearly have both exciting and predisposing causes, to reintroduce an old concept). Assignment of such cases to any single etiology is, therefore, to an extent arbitrary. However, in most cases, there is a predominant cause to which the case is assigned. In clinical practice there are many obvious reasons why such assignment of predominant cause is necessary and valuable.
- 2 Cause versus mechanism: Hughlings Jackson urged that the cause of epilepsy be considered to be mechanism of ictogenesis or epileptogenesis, and not the more downstream underlying condition leading to the epilepsy (see Taylor, 1930). He viewed most seizures to have a similar final common pathway, regardless of causative lesion. In the genetic epilepsies, we are currently closer to defining cause by molecular mechanism, but in most symptomatic epilepsies, the mechanisms of ictogenesis are largely obscure. If knowledge was more complete, categorization of cause according to mechanism would be the optimal approach, and indeed should be at the heart of any classification scheme. This issue should be a major focus of future research but currently a list of downstream causes remains necessary.
- 3 Cause depends on degree of investigation: The identification of cause is of course highly dependent on investigatory technology. Prior to the advent of MRI for instance, most cases of cortical dysplasia and of hippocampal sclerosis could not be identified definitively. Similarly, prior to the advances in genomics, most genetic causes of epilepsy (both idiopathic and symptomatic) could not be diagnosed.
Symptomatic and idiopathic epilepsy
The concept of “symptomatic epilepsy” might, at first glance, seem fairly straightforward, but in fact this is not the case. For most of the 19th century, symptomatic epilepsy (often previously known as organic epilepsy) was not considered to be a “genuine epilepsy” at all, and was rather ignored. The pendulum swung, and by the middle of the 20th century, it had become axiomatic that all or almost all epilepsies were in fact symptomatic—in the sense that the epilepsy was a symptom of an underlying cause, even if the cause could not be identified. In more recent times, a new meaning has been assigned to the term, largely because the meaning of its opposite, idiopathic epilepsy, has changed. This latter term was previously used to denote any epilepsy in which there was no demonstrable cause, but now is used only for those epilepsies that are primarily genetic in origin and in which there is no gross neuroanatomic or neuropathologic abnormality.
In the framework proposed in this article, genetic conditions that result in either pathologic or anatomic change (for instance, the epilepsy due to tuberous sclerosis or neurofibromatosis), or more subtle changes at the molecular pathologic level (for examples the epilepsies due to Rett syndrome, CDKL5, Angelman syndrome) are included in the symptomatic category. Also included are epilepsies due to developmental abnormalities where there are neuropathologic changes, despite the fact that these are due to aberrant development (and are sometimes largely genetic) rather than to an external acquired cause. These developmental/congenital disorders are a gray area between core “idiopathic” and core “acquired” epilepsies.
Of course this definition is open to criticisms on a number of fronts. The idiopathic epilepsies may have subtle anatomic abnormalities, or synaptic, membrane, neurotransmitter, or network changes. The distinction from symptomatic epilepsy, based as it is on the absence of a “gross lesion” (defined above as any identifiable pathologic or anatomic abnormality that can be detected in normal clinical investigation, including clinical microscopy, histology, and neurochemistry) is to an extent, therefore, arbitrary.
In the most recent ILAE Classification Commission report, it was suggested that the terms idiopathic, symptomatic, and cryptogenic are replaced by the terms genetic, structural/metabolic, and unknown. This suggestion has not been followed here, for a number of reasons. Firstly, the term idiopathic has been honored in history and should be replaced only if there are major advantages to doing so, and there is widespread disagreement about the need for this change (see for instance Ferrie, 2010; Guerrini, 2010; Wolf, 2010). Furthermore, the idiopathic epilepsies are due in all likelihood to a combination of genetic and environmental influences (epistatic and epigenetic influences, particularly in development), and although genetic influences probably predominate, these have proved largely conjectural. There are also many genetic causes of “symptomatic epilepsy.” For these reasons, the term idiopathic seems worthy of retention. Similarly, replacing the term “symptomatic” by “structural/metabolic” also seems largely unnecessary, not the least because many of the symptomatic conditions are neither structural nor metabolic, in the normal sense of these words. Replacing the term “cryptogenic epilepsy” with “epilepsy of unknown cause” seems also simply to anglicize and remove the venerable Greek origin to the word, rather than to change the conceptual basis in any meaningful way.
This category is included in this “database,” and it is not found in the recent commission report. In any etiologic listing, however, this category is, in the author’s opinion, valuable and important. In the 19th century, it was fully recognized that most cases of epilepsy had “exciting” as well as “predisposing” causes (the spark and the gunpowder; see accompanying article, Shorvon, 2011a). Both are “causes,” and it is surely ingenuous to somehow claim that the provoking factor (including “stress,” lack of sleep, and so on) is in some way not a “cause.” The exciting causes were environmental or systemic, and in recent years have been largely ignored. This has occurred despite the fact that one estimate found a provoking factor to be the predominant cause of epilepsy in 17% of 500 drug-resistant cases, and that manipulating these factors in these cases could greatly improve seizure control in such patients (Aird, 1983). It is clear that seizure provocation can influence genetic and acquired epilepsies and focal or generalized epilepsies, and does not map easily across conventional seizure-type or syndromic classifications, but how such precipitants produce seizures is largely obscure. It is for this reason that the category has been included here, to incorporate the causes of reflex epilepsies (those epilepsies in which the predominant cause is a highly specific provocation) and also the more common seizure precipitants (stress, lack of sleep, and so on).
Focal versus generalized epilepsy
It should be emphasized that an etiologic categorization often does not divide the epilepsies into clear-cut focal or generalized subdivisions, and this distinction (problematic as it is) does not map across the idiopathic versus symptomatic categorization. Some symptomatic epilepsies are generalized and some idiopathic epilepsies are focal. Furthermore, both generalized and focal seizures may be “provoked,” and provoked seizures can be either genetic or acquired. This dichotomy really has little utility when it comes to etiologic classification.
The term “acquired” is used to refer to “symptomatic” epilepsies excluding the predominately genetic or developmental causes. The term includes those epilepsies due external or environmental causes as well as internal pathologic processes, which have no known major environmental component (e.g., tumor, neurodegenerative disorders, autoimmune disorders). Excluded also are the epilepsies due to systemic nonneurologic diseases (e.g., fever, metabolic change, reflex epilepsy) without neuropathologic findings, and these are categorised under the term “provoked epilepsy” (a distinction based on the 19th century dichotomy of exciting/predisposing causes).
Acute symptomatic epilepsy
One term that should probably be dropped is “acute symptomatic epilepsy.” Currently this is used to include: (1) “causes” that are better included as “provoking factors” such as fever, metabolic disturbance, alcohol, and (2) acute brain injury, which are best included as “symptomatic” causes. It is senseless to include both types of “causes” within one category, as they are wholly different in terms of physiology and clinical features. If the term is to survive it should be restricted to the “early seizures” in acute brain injury (trauma, stroke, etc.), which clearly have a different pathophysiology and prognosis compared to the later seizures induced by these conditions (Shorvon & Guerrini, 2010).
Risk factor analysis
The most precise approach to the assignment of etiology is to take a statistical route and to compare the frequency of an etiologic factor in the epilepsy population (preferably at the time of diagnosis) with that in a control population (of the same demographic and geographic constitution). This has seldom been attempted, although the reverse study—a case control study to define the frequency of epilepsy within a defined etiology—has been more commonly performed, for instance in head injury, stroke, and some infections. In a case control study, the following conditions should be met (as noted by Beghi, 2004): (1) temporal association—exposure to the risk factor should precede epilepsy; (2) strength of the association—the greater the difference in incidence between exposed and unexposed populations, the more likely is this to be a true association; (3) consistency—the association should be reproducible; (4) biologic gradient—evidence of a “dose–response” effect; and (5) biologic plausibility.
Genetic and developmental mechanisms
The recent advances in genetics have made limited inroads into understanding the genetic basis of epilepsy. The most impressive findings have been made in relation to the symptomatic epilepsies of metabolic origin, and the defective gene causing almost all of the single-gene metabolic–neurologic disorders has now been identified. Fifteen genes have also been identified coding for “pure” epilepsies, but despite intensive efforts, the genetic bases of the great majority of idiopathic epilepsies remain largely obscure. The reason is likely to be that idiopathic epilepsy is caused by more complex genetic or developmental processes, and large epistatic and epigenetic influences will be present. The current emphasis on finding causal single nucleotide polymorphisms (SNPs) seems naive, and untangling the epigenetic and epistatic mechanisms will pose a formidable challenge, yet these mechanisms probably hold the key to the “missing heritability” of epilepsy. Other genetic approaches may also help, and these include studies of such mechanisms as copy number variation, genomic imprinting, chromosomal imbalance, X inactivation, and mitochondrial mechanisms. How the category of “idiopathic” epilepsy will appear in the future after these further research efforts have come to fruition is unclear. It should also be noted that many genetic influences are not “all or none” but confer susceptibility. Where the line is drawn in these cases between a genetic or cryptogenic categorisation is arbitrary.
Epilepsy as a disease not a symptom
As mentioned earlier, epilepsy is, like headache or anemia, often considered a “symptom” of a neurologic condition and not a condition per se, and the main reason for considering it thus is because there are so many different potential causes. This of course raises a question about the definition of disease in general, a topic beyond the scope of this article (and as mentioned earlier, in the 19th century epilepsy was often considered only to apply to the idiopathic condition). Despite this, and perhaps rather contradictorily, there is also an increasing tendency to define “syndromes” of epilepsy, and to consider these entities in their own right. In relation to etiology, there is a danger of mixing up “disease” with “cause of disease” and this is at its most difficult in consideration of some syndromes. Some syndromes are genetic (e.g., Severe Myoclonic Epilepsy of Infancy) or presumed genetic (e.g., the Idiopathic Generalized Epilepsies), and others have mixed etiologies (e.g., West syndrome). The definition of syndrome is frequently sufficiently vague to allow inclusion of many different groups and subgroups, and there is a continuing debate among the splitters and lumpers of the epilepsy world about what to include. Furthermore, some conditions are so well defined from the etiologic point of view (severe myoclonic epilepsy of infancy is an example) that it is not really a “syndrome” but an epilepsy etiologic type, in the same sense that posttraumatic epilepsy is not considered a syndrome but a specific etiologic type of epilepsy. Therefore, whether epilepsy is considered a symptom or a disease, and to what extent epilepsies are grouped into syndromes or not, will depend to an extent on the importance placed on etiology. Certainly a definition and classification of epilepsy on etiologic grounds has very different results from those using semiologic or electrographic criteria.
The classification scheme proposed here is based on the framework proposed in the Causes of Epilepsy (Shorvon et al., 2011b), and the commentary on several chapters from this book (Shorvon, 2011b,c; Shorvon et al., 2011a).
I have no conflicts of interest to declare. This work was undertaken at UCL and received a proportion of funding from the Department of Health’s NIHR Biomedical Research Centre’s funding scheme. I confirm that I have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.