Epilepsy Classification


Hans Lüders, Naoki Akamatsu, Shahram Amina, Christoph Baumgartner, Selim Benbadis, Adriana  Bermeo, Andrew Bleasel,  Alireza Bozorgi, Mar Carreño, Michael Devereaux, Rebeca Dyer, Lauren Ghanma, Guadalupe Fernandez-Baca Vaca, Stefano Francione, Naiara García Losarcos, Hajo Hamer, Hans Holthausen, Shirin Jamal Omidi, Giri Kalamangalam, Andrés Kanner, Susanne Knake, Patrick Landazuri, Shi Hui Lim, Luisa V Londoño, Jayanti Mani, Jonathan Miller, Soheyl Noachtar, André Palmini, Jun Park, Felix Rosenow,  Asim Shahid, Stephan Schuele, Bernhard Steinhoff, Charles Szabo, Nitin Tandon, Kiyohito Terada, Walter van Emde Boas, Peter Widdess-Walsh, Philippe Kahane.

We read with interest the comments of R. Fisher et al (Fisher et al., 2019) in response to our critique of the 2017 ILAE classification of seizures and epilepsy (Luders et al., 2019a). We answer Fisher et al commentaries below but we will first summarize the numerous attempts by the ILAE to develop a classification of epileptic seizures and epilepsies that is widely accepted and, most importantly, reflects the continuous advances in our understanding of epilepsy.

From 1964 to 1970, the ILAE under the auspices of the World Health Organization,  developed a clinical and electroencephalographic classification of epileptic seizures (Gastaut, 1970). This classification has been “updated, amended and improved” several times over the last 50 years (1981; 1985; 1989; Engel and International League Against, 2001; Fisher et al., 2017; Scheffer et al., 2017). There has been much discussion about terminology (example: “partial seizures with complex symptomatology and “generally impairment of consciousness” evolved into “complex partial seizures” and eventually into “focal unaware seizures.”) and definition of epileptological terms (example: differential diagnosis of partial vs generalized epilepsy).  On the other hand, there has been only a limited discussion about the basic framework of the proposed classifications. 

Since the initial proposal for an international classification of epileptic seizures and epilepsies in 1964 (Arnautova and Nesmeianova, 1964), there has been an exponential growth of our diagnostic armamentarium, treatment modalities and understanding of the different etiological factors leading to epileptic seizures.  In spite of these advances, the “framework” of the classifications proposed more than 50 years ago remains essentially unchanged. The pillars of all these proposals are specific electro-semiological seizure characteristics, seizure types, epileptic syndromes and etiologies. Many of these pillars partially or even completely overlap. For example, in the 2017 ILAE classification the epileptic seizure types overlaps near completely with the epilepsy types. Epileptic syndromes represent another essential pillar of the epilepsy classifications proposed by the ILAE since 1985. While some of the syndromes are still useful in the management of epileptic patients, it is evident that in many instances epileptic syndromes are just constellations of electro-semiological seizure types that tend to occur at a certain age and often respond to similar antiseizure drugs with a more or less defined prognosis. Therefore, it is likely that in the future a better understanding of the etiology (particularly genetics) will progressively decrease the importance of syndromic diagnosis. 

We feel that at this point the ILAE should try to identify a completely new framework to classify the epilepsies which will accommodate the extraordinary advances in the diagnosis, etiology and therapy of the epilepsies. An ideal framework for classification of the epilepsies should fulfill the following characteristics:

  1. The pillars of such a framework should specify as accurately as possible the main characteristics of the epilepsy phenotype.
  2. The different pillars should be independent and be specified by different tests.
  3. The terminology used to define each pillar should be “classical terminology” that is widely accepted by neurologists, epileptologists, other specialists, general practitioners as also patients with epilepsy.
  4. The same framework should allow classifying each pillar with varying degrees of precision depending on the expertise of the classifier and the information and diagnostic methods available to each classifier.
  5. The same framework should also be applicable to different epilepsy cohorts: newborns, infants, toddlers, children, elderly, status epilepticus, etc.
  6. The dynamic progression of epileptic seizures should be clearly reflected in the epilepsy classification by specification of the seizure evolution with different degrees of precision.

It is interesting to analyze the historical evolution of epilepsy and epileptic seizure classifications. Initially the classifications were dominated by seizure semiology. With the availability of EEG the main emphasis shifted to electroclinical syndromes. Neuroimaging was the third major advancement allowing the in vivo discovery of brain abnormalities as cause of epileptic seizures and, together with EEG, increased greatly our ability to precisely localize the epileptogenic zone. In spite of all these advances, little attempts were made to classify epilepsy in a multidimensional framework in which each dimension is relatively independent of each other. Instead, just the contrary approach was taken, as “epileptic syndromes” became the gold standard defining constellations that linked the different dimensions together. At the same time, epileptic zone location, a dimension we can now define in many patients with great precision, has been more or less neglected. This is in contrast to other parts of neurology where a precise localization has continued to be the basis of diagnosis. In the latest 2017 ILAE classification only 3 locations can be defined: focal, generalized and unknown.

The 4 dimensional classification of epileptic seizures and epilepsies (Luders et al., 2019b) fulfills all the criteria for a suitable new framework. It is based on the widely accepted framework used to classify neurological diseases. Example: stroke and Parkinson’s disease:

Dimension Epilepsy Stroke Parkinson's Disease
Semiology Abdominal aura →
Automotor seizure
Right arm paralysis
Resting tremor
Localization Mesial temporal Left middle cerebral
artery stroke
Substantia nigra
Etiology Mesial temporal sclerosis Atrial clot embolism PARK2 mutation
Co-morbidities Verbal memory deficit Diabetes
Atrial fibrillation
Severe insomnia

In the following rebuttal, we will discuss the specific comments Fisher et al (Fisher et al., 2019) made in reply to our critique. At this point, however, we would encourage the ILAE to take a step backwards and consider alternative frameworks for classification of epileptic seizures and epilepsies and question if indeed the 2017 ILAE classification is “a step in the right direction.” It is clear that in the near future there will be a revolution in genetics that will lead to a completely different approach to diagnosis and treatment of epilepsy. These changes should lead us to a new way to classify etiology. Scheffer et al (Scheffer et al., 2017) took an excellent step forward by defining the main etiological categories seen in epilepsies. However, in the clinical setting where management remains the most crucial function of the physician, these broad categories often do not provide enough detail to advance patient care.


Fisher et al (Fisher et al., 2019) replied to our critique of the 2017 ILAE classification (Luders et al., 2019a) in 13 paragraphs. We will follow here the same order.

    We agree that a classification of epileptic seizures and epilepsies should “provide a common language to aid clinicians in diagnosis and management of individual patients and to be of value for research and scientific study.” However, we disagree with Fisher et al. assessment that the 4 dimensional classification was complicated, only suitable for use of specialists in epilepsy and required advanced diagnostic technology. Similar to the 2017 ILAE classification, the 4-dimensional classification offers simple versions for the non-specialist and can be applied without the need of any diagnostic tool except a detailed patient history. Still the level of knowledge expected for semiological description in the field of Neurology in general is much higher than what the 4 dimensional classification proposes. The 4 dimensional classification, however, uses terminology known to both specialists and non-specialists. It allows detailed classification of semiology that is extremely useful both in routine clinical care, epilepsy surgery centers and also as a teaching tool for epilepsy fellows and neurology residents. We also disagree with the position that “evolutionary advances…. avoiding unnecessary disruption” is the way to proceed. On the contrary, we feel that the major advances in diagnosis and epilepsy treatment calls for development of a classification with a completely different framework (see above).


    We agree with Fisher et al that nothing is universally accepted but that does not mean that ideally a classification should be universally accepted.


    The EEG is not “ignored” or a “second level classifier” in the 4 dimensional epilepsy classification. Let us analyze the case discussed by Fisher et al as it would be classified in the 4 dimensional classification. We would first classify the seizure semiologically. However, to do that properly we need additional anamnestic information. Is the patient aware that she had a behavioral arrest? Does she remember what happened during the arrest? Was the patient tested for responsiveness during the behavioral arrest? There are many classification options depending on the response to these questions. Let us assume that the patient is totally amnestic of the behavioral arrest and is also unresponsive during the event. He would then be classified as having unspecified “paroxysmal events” and “dialeptic events.” We use here the expression paroxysmal event and dialeptic event because we do not have enough information to establish that the patient actually has epilepsy. Once we have “proof” of the epileptic nature of the events, the terms would change to “epileptic paroxysmal event” and “dialeptic seizures” (see below). Assuming this is all the information we have the patient would be classified as follows:

    ILAE classification 4 dimensional classification
    * Paroxysmal Event Paroxysmal Event
    Seizure Type Unknown Onset Behavioral Arrest Event semiology Dialeptic event
    Epilepsy Type Unknown **
    Epilepsy Syndrome Unknown ***
    Etiology Unknown Etiology Unknown
    Co-morbidities None

    *The ILAE does not classify the paroxysmal events.
    **In the 4-dimensional classification the dimension “Epileptogenic zone” is classified only if he patient has an Epileptic Paroxysmal Event.
    *** In the 4 dimensional classification epilepsy syndromes can be added in parenthesis after having specified the epileptogenic zone.

    We now obtain an EEG that showed, for example, left temporal sharp waves and electrographic seizures arising from the left temporal lobe. The patient also had an MRI demonstrating a left mesial temporal DNET. In addition the patient now reported that with some seizures just seconds before losing awareness she had an “unusual feeling in the stomach” and on one occasion she had whole body shaking. The patient also complaint of memory problems. The classifications now change to the following:

    ILAE classification 4 dimensional classification
    * Paroxysmal Event Epileptic Paroxysmal Event
    Seizure Type Focal to bilateral tonic-clonic seizure Seizure semiology Abdominal aura → Dialeptic → bilateral clonic seizure
    Epilepsy Type Focal Epileptogenic zone Left mesial temporal lobe
    Epilepsy Syndrome Not available **
    Etiology Structural Etiology Structural (left mesial temporal DNET)
    Co-morbidities Memory problems

    *The ILAE does not classify the paroxysmal events.
    ** In the 4 dimensional classification epilepsy syndromes can be added in parenthesis after having specified the epileptogenic zone.

    Analysis of the 2 classification systems demonstrates that both systems include EEG findings in their classification as a method to confirm that the seizures most likely originated from the left mesial temporal area. However, vital semiological information, including seizure evolution as well as an accurate localization of the epileptogenic zone are not part of the ILAE classification. Notice how little clinical information is described in the ILAE classification whereas the 4-dimensional classification provides a clear picture of the seizure semiology (including seizure evolution), an accurate localization of the epileptogenic zone as also details of the etiology. Fisher et al (Fisher et al., 2019) indicated that the Commission made “an intentional choice to mix semiological and electro-anatomical data”, but do not specify what advantage this approach has, particularly when compared to the 4-dimensional classification that keeps semiological data and electro-anatomical data clearly separated in 2 different dimensions (seizure semiology and epileptogenic zone).

    We agree with Fisher et al that there is insufficient evidence to classify seizures by pathophysiology.


    We agree with Fisher et al (Fisher et al., 2019) that the subdivision generalized vs focal is important. We, however, see no reason why a more specific localization of the epileptogenic zone should not be encouraged. Seizures arising from different brain regions show strikingly different ictal semiologies. The localization of the epileptogenic zone is essential when we consider all epilepsy treatments, particularly epilepsy surgery. The spatial relationship between the epileptogenic zone and structural lesions is important when we try to determine if a specific structural lesion is or is not causally related to the epileptogenic lesion. It is interesting to notice that seizure and epilepsy labels in the epilepsy literature in the last century and until today frequently contain an anatomical modifier specifying the location of the epileptogenic zone (example: frontal lobe seizures, temporal lobe epilepsy, posterior quadrant epilepsy, etc.) The fact that most focal seizures may have a similar pathophysiology or that all focal seizures tend to respond to the same antiseizure drugs is not an argument against providing a framework that permits detailed classification of the epileptogenic zone.


    We agree with Fisher et al (Fisher et al., 2019) that the few seizure evolutions included in the 2017 ILAE classification (or any of the previous ILAE classifications) do not really permit adequate classifications of seizure evolutions. However, we feel that identifying semiological seizure components and to link them by arrows to express the seizure evolution is an effective way to describe and to classify “the thousands of possible propagation patterns.” Expressing the semiological evolution of epileptic seizures has also an important pedagogic value. Fisher et al (Fisher et al., 2019) are correct that the 4 dimensional classification is a pure semiological system and, therefore, only expresses the sequential activation of different symptomatogenic zones. This stands in contrast to the epileptogenic which is constant and by definition cannot “propagate.”

    We do not feel that focusing only on “high lightening the initial symptom of a seizure” is sufficient. We agree that in many cases the initial component provides the most accurate information as long as we are aware of what seizure component(s) follow. However, instances where the initial symptom is not the most informative are not infrequent. Examples in which the second or third symptom is the most informative (essential component marked in bold):

    A. Aura → Hypermotor seizure

    B. Aura Dialeptic → Left arm tonic → Generalized clonic seizure

    C. Bilateral asymmetric tonic → Generalized Clonic Seizure
        Lateralizing sign: Left Todd’s paralysis

    D. Dialeptic → Left versive → Generalized tonic-clonic seizure


    We agree with Fisher et al (Fisher et al., 2019) that “seizures require a full description, not just a classification.” Fisher et al (Fisher et al., 2019) feel that “classifying” all the semiological data would result in many hundreds of seizure types.” They suggest, therefore, to replace a classification by a free text description using “descriptors”. This approach, however, is not a classification and in clinical practice would lead separate observers to write different descriptions for the same seizure.

    A classification is just ordering “something in classes or categories.” To accomplish this in the 4 dimensional classification we use a hierarchal classification with a library of clearly defined, distinct seizure types (components). All possible semiological seizure manifestations are classified initially in broad semiological classes: auras, autonomic seizures, dyscognitive seizures, motor seizures, special seizures and asymptomatic EEG seizures. These are the broad categories that can be used easily to classify seizures by non-experts. Each one of these broad categories is subsequently classified into more specific subgroups. These groups can be subdivided further in more detailed groups and so on. This approach is similar to the approach Linnaeus took when classifying animal and plants. In Linnaeus classification each Kingdom is subdivided in progressively more detailed subgroups labelled sequentially Phylum, Class, Order, Family, Genus and Species. This approach permits classification of any animal, plant or in our case any epileptic seizures with the detail the observer feels is appropriate. This approach allowed Linnaeus to classify millions of species and should allow us to do the same with the different semiological seizure types.

    Fisher et al (Fisher et al., 2019) question the importance to differentiate between pilomotor and vasomotor seizures. However, pilomotor seizures are relatively specific seizures seen almost exclusively in patients with temporo-sylvian lobe epilepsy, most frequently occurring in left sided (language dominant) epilepsies and tend to be ipsilateral with respect to the epileptogenic zone (Stefan et al., 2002). Vasomotor seizures on the other hand are semiologically distinct, relatively non-specific and of no known localizing or lateralizing value.


    We feel that the classification of epileptic seizures and epilepsies, neonatal seizures and status epilepticus should all be based on the same framework. The 4 dimensional classification can accommodate all these seizures within the same basic framework.

    It is clear the semiology of seizures tend to vary in different age groups. This is discussed in detail in the articles of Fernandez Baca-Vaca et al (Fernandez-Baca Vaca et al., 2018) and Pressler et al (Pressler et al., 2018). These differences, however, do not call for a new classification. The 4 dimensional classification incudes all the semiological seizure types, including those that are seen only in adults and children (example: auras, generalized tonic-clonic seizures) or are seen almost only in infants and toddlers (example: epileptic spasms). The inclusion of all possible semiological seizures types that occur at different ages permits using a unified classification. It also provides better understanding of how seizure semiology changes with age.

    It is of interest also that Pressler et al (Pressler et al., 2018) stress the frequency of neonatal EEG seizures without any clinical correlate. In the last decade or so, continuous EEG monitoring of children and adult patients with altered level of consciousness has become extremely popular. We now realize that many patients with PLEDS have also EEG seizures without any clinical correlate. Again, this observation does not require a new classification. In the 4D classification these EEG seizures are called “subclinical EEG seizures”, one of six categories in the basic version of the 4-dimensinal Epilepsy Classification.

    Fisher et al (Fisher et al., 2019) make no mention of the Report of the ILAE Task Force on Classification of Status Epilepticus. This Task Force in their proposal suggests a completely new framework with four axes or “dimensions” ( Semiology, etiology, EEG and age) that are completely different of the pillars of the classification of Fisher et al (Fisher et al., 2019). It is reasonable to ask, if we should use another classification framework in a single epilepsy patient because a seizure lasts longer than 5 minutes.

  8. AURA

    We feel that aura is an extremely popular, well recognized, and useful term. Patients usually are aware that the occurrence of an aura is a warning sign that a major seizure may be coming. None of the co-authors of this manuscript recalls having any problem with patients assuming that they can neglect auras because they are of no importance.

    In the 19th century and in the first part of the 20th century aura referred to the “minor seizure” that preceded the generalized tonic-clonic seizure. In other words, it was equivalent to the expression “simple partial seizure”. More recently, it is used frequently as the subjective part of the seizure that usually occurs at the beginning of a seizure. With an appropriate modifier, it is an important part of the seizure frequently pointing to the location and lateralization of the epileptogenic zone (Example: left visual aura, right hand somatosensory aura, etc.). We feel that there is no need to replace these “classical” terms with “focal aware seizure with visual symptomatology” or “focal aware seizure with somatosensory symptomatology.” Notice also that the terminology proposed by the ILAE does not include any lateralizing or somatotopic information.


    We support Fisher et al. (Fisher et al., 2019) efforts to eliminate what they called “obscure” terms. The same is true in the 4 dimensional classification. The 4 dimensional classification utilized, as much as possible, established and widely accepted terminology. However, there are ictal semiological features that had no known appropriate label. One of those exceptions is “dialepsis”. This is a condition in which the patient remains relatively motionless without loss of tone, may or not stare, and is unresponsive to external stimuli. After the episode, the patient is completely amnestic of the episode. These dialepsis may occur in patients with focal epilepsy or generalized epilepsy (Baykan et al., 2011). The ideal expression should be “absence seizure.” However, epileptologists always relate absence seizures with 3 Hz generalized spike-and-wave EEG seizures. Therefore, to have a pure semiological term that identifies this condition independent of EEG findings, we had to coin a new term, i.e. dialeptic seizures.


    Fisher et al (Fisher et al., 2019) argue there would be little consensus on how specific seizures should be localized. We do not agree. Like the other dimensions the “location dimension” should initially consists of a broad subdivision, namely generalized vs focal and then allow the user to apply progressively more precise subdivisions, depending on need and available information.

    We indicated before that in the 2017 ILAE classification Seizure type and Epilepsy type are nearly redundant. We do not understand Fisher et al’s response to this criticism.

    We agree that in the near future etiology will occupy a dominant role in epilepsy classification. It is also correct that in the near future some “classical” epileptic syndromes will still play a role in assisting in the management of patients with epilepsy and also in the elucidating some epilepsy etiologies. That is the reason the 4-dimensional epilepsy classification permits syndrome specification (Luders et al., 2019b). However, as we define better genetic alterations which either determine or influence the epileptic phenotype, it will become even more necessary to design a classification that more specifically details the phenotype determined by these genetic abnormalities.


    We agree all terms in a classification should be clearly defined. However, requiring mapping from one classification to another is frequently of little practical value. For example, let us assume we study the spreading of epileptic seizures types arising from different brain lobes. What can we then actually map? The 2017 ILAE classification only recognizes focal, generalized or unknown epilepsies. Therefore, all the seizures would be mapped to focal epilepsies. Furthermore, the 2017 ILAE classification includes only a handful of spreading patterns defined. Given that extremely restricted choice, what is then the value of mapping?


    Fisher et al (Fisher et al., 2019) quote studies that showed that the 2017 ILAE Classification can be applied successfully to a large number of patients with epilepsy. It is important, however, to stress that a classification is not necessarily better or more useful if it can be applied to a larger percentage of patients. For example, we could classify epilepsies just by the sex of the patient. This would allow us to classify almost 100% of the patient but it does not necessarily imply that the classification is particularly useful. Finally, experiences from tertiary referral epilepsy centers did not always confirm a high acceptance of the 2017 ILAE classification (Stockinger et al., 2018).


    We feel that the recent advances in epileptology call for a completely revised epileptic seizure and epilepsy classification that accommodates these striking advances. It is time to break with the traditional small “steps in the right direction” and to look for a new approach that embraces modern and foreseeable future epileptology. The 4 dimensional epilepsy classification is one possible alternative that should be considered by the ILAE.


Proposal for revised clinical and electroencephalographic classification of epileptic seizures. From the Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia 1981; 22(4): 489-501.

Proposal for classification of epilepsies and epileptic syndromes. Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia 1985; 26(3): 268-78.

Proposal for revised classification of epilepsies and epileptic syndromes. Commission on Classification and Terminology of the International League Against Epilepsy. Epilepsia 1989; 30(4): 389-99.

Arnautova EN, Nesmeianova TN. A Proposed International Classification of Epileptic Seizures. Epilepsia 1964; 5: 297-306.

Baykan B, Altindag E, Feddersen B, Ozel S, Noachtar S. Does semiology tell us the origin of seizures consisting mainly of an alteration in consciousness? Epilepsia 2011; 52(8): 1459-66.

Engel J, Jr., International League Against E. A proposed diagnostic scheme for people with epileptic seizures and with epilepsy: report of the ILAE Task Force on Classification and Terminology. Epilepsia 2001; 42(6): 796-803.

Fernandez-Baca Vaca G, Mayor CL, Losarcos NG, Park JT, Luders HO. Epileptic seizure semiology in different age groups. Epileptic Disord 2018; 20(3): 179-88.

Fisher RS, Cross H, D'Souza C, French JA, Haut S, Higurashi N, et al. 2017 International League Against Epilepsy classifications of seizures and epilepsy are steps in the right direction. Epilepsia 2019; 60(6): 1040-4.

Fisher RS, Cross JH, French JA, Higurashi N, Hirsch E, Jansen FE, et al. Operational classification of seizure types by the International League Against Epilepsy: Position Paper of the ILAE Commission for Classification and Terminology. Epilepsia 2017; 58(4): 522-30.

Gastaut H. Clinical and electroencephalographical classification of epileptic seizures. Epilepsia 1970; 11(1): 102-13.

Luders H, Akamatsu N, Amina S, Baumgartner C, Benbadis S, Bermeo-Ovalle A, et al. Critique of the 2017 epileptic seizure and epilepsy classifications. Epilepsia 2019a; 60(6): 1032-9.

Luders H, Vaca GF, Akamatsu N, Amina S, Arzimanoglou A, Baumgartner C, et al. Classification of paroxysmal events and the four-dimensional epilepsy classification system. Epileptic Disord 2019b; 21(1): 1-29.

Pressler RO, Ciio MR, Mizrahi EM, Moshe SL, Nunes ML, Plouin P, et al. The ILAE Classification of Seizures & the Epilepsies: Modification for Seizures in the Neonate. Proposal from the ILAE Task Force on Neonatal Seizures. Epilepsia 2018; In Press.

Scheffer IE, Berkovic S, Capovilla G, Connolly MB, French J, Guilhoto L, et al. ILAE classification of the epilepsies: Position paper of the ILAE Commission for Classification and Terminology. Epilepsia 2017; 58(4): 512-21.

Stefan H, Feichtinger M, Genow A, Kerling F. [Goose flesh and cold sensation. Symptoms of visceral epilepsy]. Nervenarzt 2002; 73(2): 188-93.

Stockinger J, Bacher M, Bernedo Paredes V, Intravooth T, Kornmeier R, Kursth R, et al. The new seizure and epilepsy classification in clinical practice. Z Epileptol 2018; 31: 307 - 10.

Back to Top