The 2017 and 2022 ILAE epilepsy classification systems identify needs and opportunities in care: A paediatric hospital-based study

Objectives: There is a paucity of studies reporting the epilepsy spectrum using the 2017 and 2022 ILAE classification systems in everyday clinical practice. To identify gaps and opportunities in care we evaluated a hospital-based cohort applying these epilepsy classification systems, including aetiology and co-morbidity, and the utility of molecular genetic diagnosis to identify available precision therapies. Methods: Cross sectional retrospective study of all children with epilepsy ( ≤ 16 years) attending University Hospital Galway (2017 – 2022). Data collection and analysis of each case was standardised to ensure a systematic approach and application of the recent ILAE categorisation and terminology (2017 and 2022). Ethics approval was obtained. Results: Among 356 children, epilepsy was classified as focal (46.1 %), generalised (38.8 %), combined (6.2 %), and unknown (9 %). Epilepsy syndrome was determined in 145/356 (40.7 %), comprising 24 different syndromes, most commonly SeLECTS (9 %), CAE (7 %), JAE (6.2 %) and IESS (5.9 %). New aetiology-specific syndromes were identified (e.g. CDKL5 -DEE). Molecular diagnosis was confirmed in 19.9 % (n = 71) which encompassed monogenic (13.8 %) and chromosomopathy/CNV (6.2 %). There was an additional 35.7 % (n = 127) of patients who had a presumed genetic aetiology of epilepsy. Remaining aetiology included structural (18.8 %, n = 67), infectious (2 %, n = 7), metabolic (1.7 %, n = 6) and unknown (30.3 %, n = 108). En-cephalopathy categorisation was determined in 182 patients (DE in 38.8 %; DEE in a further 11.8 %) associated with a range of co-morbidities categorised as global delay (29.2 %, n = 104), severe neurological impairment (16.3 %, n = 58), and ASD (14.6 %, n = 52). Molecular-based “ precision therapy ” was deemed available in 21/ 356 (5.9 %) patients, with “ molecular precision ” approach utilised in 13/356 (3.7 %), and some benefit noted in 6/356 (1.7 %) of overall cohort or 6/71 (8.5 %) of the molecular cohort. Conclusion: Applying the latest ILAE epilepsy classification systems allow comparison across settings and identifies a major neuro-developmental co-morbidity rate and a large genetic aetiology. We identified very few meaningful molecular-based disease modifying “ precision therapies ” . There is a monumental gap between aetiological identification, and impact of meaningful therapies, thus the new 2017/2022 classification clearly identifies the major challenges in the provision of routine epilepsy care.


Introduction
The International League Against Epilepsy (ILAE) produced major updated classification systems for both seizures and the epilepsies in 2017 [1,2].The 2017 classification systems use clearer language, remove imprecise terminology, include previously excluded seizure types, and incorporate advances in our understanding of epilepsy aetiology.While "Epilepsy Syndromes" have been recognized for over 50 years, for the first time in 2022, the ILAE published consensus criteria, definitions, and categorisation of the Epilepsy Syndromes [3][4][5][6][7].In the current era of advanced genomics, the molecular mechanisms underpinning the epilepsies have gained major interest, with many recent publications reporting examples of precision care particularly in rare monogenic epilepsies [8][9][10][11][12][13][14].As aetiology, and particularly the explosion in genetic diagnosis for a large proportion of epilepsies, has become an important component of epilepsy classification, new aetiology-specific Epilepsy Syndromes have been introduced [6].Epilepsy-related comorbidities are an essential component of current epilepsy classification systems [2].
While the primary aim of epilepsy classification is to provide a common international language and terminology for clinical practice, application of the recent ILAE epilepsy classifications systems afford an opportunity to benchmark practice in the care of children with epilepsy and identify gaps in care, regardless of setting or resources.There is, however, a paucity of studies reporting the spectrum of epilepsy and precise aetiologies of epilepsy cohorts in clinical practice using the 2017 and 2022 ILAE classification systems [15][16][17].There are few studies examining the current availability of molecular precision therapies against the burden of disease, particularly outside of tertiary level centres [15,16].Our study applies the 2017 ILAE epilepsy classification, including aetiology and co-morbidities in routine clinical practice and the recently published ILAE 2022 Epilepsy Syndrome classification systems, to our paediatric cohort, while also reviewing the use of any precision therapies available for those patients with a molecular genetic aetiology.
The aims of this study were to (i) determine the spectrum of paediatric epilepsies among children attending a second level unit applying the ILAE 2017 (Epilepsy) and ILAE 2022 (Epilepsy Syndrome) classification systems and terminologies (ii) determine the proportions of various aetiologies, and degree of co-morbidities, and (iii) evaluate the frequency of genetic diagnosis and the proportion of patients eligible for molecular precision therapeutic approaches.Furthermore, we aimed to evaluate the differences in proportions of selected variables across various age groups of epilepsy-onset and within aetiology categories.

Study design
We retrospectively evaluated a cross sectional cohort of all children with epilepsy (CWE) attending a second level epilepsy service from July 2017 to July 2022 (inclusive).Children, including neonates up to 16 years, were included from hospital admissions and outpatient clinics.Data were retrieved using patient records using a predetermined standardised proforma to ensure a systematic and standardised approach to data collection.This study was conducted in accordance with local institutional guidelines.Research and ethics committee (REC) approval was obtained prior to data collection.

Application of ILAE classifications and definitions
Epilepsy diagnoses were established in concordance with the ILAE operational criteria [18].The definitions used to classify seizure type, epilepsy type, and Epilepsy Syndrome were consistent with the operational definitions outlined by ILAE position papers [2][3][4][5][6][7], including newly defined aetiology-specific syndromes, and the application of developmental and/or epileptic encephalopathy (DEE) terminology where applicable.Classification was agreed among the authors including a paediatric neurologist, and paediatrician (with special interest in neurology and epilepsy).
Regarding aetiology classification, patients' clinical data and records were reviewed in detail applying the principles from the ILAE classification papers [2].Genetic investigation (karyotype, array CGH, gene panel, WES, etc) and onward referral to clinical genetics was conducted as part of routine practice as this was a retrospective study.Molecular diagnosis was assigned when a gene variant, or chromosomal disorder, was identified as pathogenic or predicted pathogenic in clinical practice.Similar to genetic aetiology, patients were classified as having a structural, metabolic, immunological, or infectious aetiology by relevant laboratory or neuroimaging investigations by the treating physicians.
The extent of investigations are summarised in the Supplementary Table .A presumed genetic aetiology was defined by the presence of (i) genetically generalised epilepsies (GGEs) including GGE/IGE syndromes (ii) self-limited focal epilepsy (SeLFE) syndromes, and (iii) Epilepsy Syndromes with clear autosomal dominant inheritance patterns [e.g.sleep related hypermotor (hyperkinetic) epilepsy (SHE)] (Supplementary material).Structural aetiology was further divided into (i) acquired, (ii) non-acquired with known cause (e.g.genetic), and (iii) nonacquired without a known cause.
The neuropsychiatric and developmental co-morbidities were only included if formally established in clinical practice by the shared caring specialist teams (e.g.Neurodevelopmental Paediatric or Child Psychiatry services using formal structured standardised assessments applying diagnostic criteria in an MDT setting).If patients displayed features or traits of a certain comorbidity but did not have a formal diagnosis at the time of data collection, the comorbidity was not included.
As all patients with epilepsy expectantly experience individualised or personalised medicine, the term 'precision therapy' is described only in relation to patients with a genetic molecular diagnosis, where knowledge of the underlying genetic variant lead to a precision therapy approach (Supplementary material).

Statistical analysis
SPSS v21.0 was used for statistical analysis.Categorical variables were grouped into contingency tables, and the percentage of cells with an expected count less than five was calculated.Fisher exact test was used when > 20 % of cells had an expected count of less than 5. Pearson Chi-square test was used when < 20 % of cells had an expected count of less than 5.A p-value < 0.05 was considered statistically significant.

Baseline characteristics
A total of 356 CWE were identified.Baseline characteristics are described in Table 1 and Sup.Table S1.There were more males, and more patients with epilepsy onset under age 6 years.The most common ethnicity was White Irish (n = 285, 80.1 %).Forty-one patients (11.5 %) had previous febrile seizures and 57 (16 %) had a family history of epilepsy.Levetiracetam was most the common first line therapy followed by sodium valproate.A total of 70 patients (19.7 %) had drug resistant epilepsy.During the study period, a total of 72 (20.2 %) patients were not on anti-seizure medication, mostly due to remission.

Spectrum of epilepsies
For seizure type, generalised-onset seizures were slightly more common than focal-onset seizures, and motor were more frequent than non-motor seizures (Table 1 (SeLECTS), childhood absence epilepsy (CAE), juvenile absence epilepsy (JAE), followed by infantile epileptic spasms syndrome (IESS) (Fig. 1).An Epilepsy Syndrome was present in 47.4 % of CWE onset greater than 6 years (n = 73/154) and in 39.3 % of CWE onset aged 2-5 years (n = 42/107), followed by those with epilepsy onset in first year of life (31.6 %, n = 30/95) (Fig. 2).The most common syndromes in those CWE onset aged 2-5 years and greater than 6 years were GGE and self-limited focal Epilepsy Syndromes, while IESS was most common in infants (<1 year).
Those with a clearly defined known mechanistic aetiology (i.e.including molecular genetic, structural, metabolic, immunological, metabolic, and infectious but excluding unknown and presumed genetic aetiologies) totalled 34 % (n = 121/356).There was a greater proportion of these known aetiologies in CWE onset within the first year of life compared to CWE onset aged 2-5 years (68.4 % infants < 1 year, 29.0 % of CWE onset aged 2-5 years), and CWE onset at six years or greater (16.2 %) (p < 0.001).
Table 2 represents the differences in aetiologies within ages of epilepsy onset, epilepsy type or selected neurodevelopmental comorbidity.In CWE onset during infancy, 48.4 % had a molecular genetic diagnosis which was the most frequent aetiology identified compared to other aetiologies (p < 0.001).In CWE onset at older ages, a presumed genetic aetiology was more common (41.1 % CWE onset aged 2-5 years, 51.3 % CWE onset 6 + years, p < 0.001), largely comprising the GGE and selflimited focal epilepsies.Children with an Epilepsy Syndrome (n = 145) mostly had a presumed genetic (71.7 %, n = 104) or molecular-related (19 %, n = 27) aetiology compared to other aetiologies, while the most common aetiology in those CWE without a recognised Epilepsy Syndrome (n = 211) was the unknown category (44.5 %, n = 94).

Molecular precision therapeutic approaches identified
All therapies used in our cohort are briefly summarised in Sup.Table S1.A precision therapy approach specific to molecular genetic diagnoses (Fig. 3) was deployed in 18.3 % (13/71) of the genetic molecular diagnosis cohort (Table 3).Benefit (seizures or self/physician reported developmental improvement) was noted in 6 of the 13 patients, but was often only transient (e.g., 4-aminopryidine use in GOF KCNA2-DEE) or inferred (e.g. by avoiding potential drug-induced exacerbation) in our cases.When disease mechanism was reviewed (Sup.Fig. S1), no treatment was identified to modify gene/protein expression.Taken together, potential benefit was noted in 6/71 (8.5 %) of the molecularly diagnosed cohort or 6/356 (1.7 %) of the entire cohort (Table 3).

Discussion
Classification of epilepsy can be a complex task in children, due to the extreme variability in clinical presentation, multiplicity of seizure types, Epilepsy Syndromes, thousands of aetiological mechanisms, and many co-morbidities which often require access to advanced diagnostics Abbreviations: SNI, severe neurologic impairment; ID, intellectual disability; ASD, autism spectrum disorder; ADHD, attention deficit hyperactivity disorder; OCD, obsessive-compulsive disorder; DCD, developmental co-ordination disorder; DEE, developmental epileptic encephalopathy; DE, developmental encephalopathy.a Some patients had more than one aetiology.and service infrastructures [16].This study provides data regarding the spectrum of childhood epilepsies, applying the recent 2017 ILAE epilepsy and 2022 ILAE Epilepsy Syndrome classification, in a second level centre, at the dawn of the molecular precision medicine era in epilepsy.
Applying the 2022 ILAE position papers [3][4][5][6][7], an Epilepsy Syndrome was identified in approximately 41 % of our cohort, encompassing 24 different syndromes.Similar rates of Epilepsy Syndrome were seen in our study as compared to historical literature prior to 2022 [36].Aetiology classification is particularly emphasised in the 2017 ILAE classification given the advances in genomic investigations and neuroimaging, and reinforced in the 2022 ILAE Epilepsy Syndrome publications exemplified by the inclusion of aetiology-specific    Percentages are reported as a proportion of the row characteristic.Some patients had more than one aetiology thus a using first aetiology.Metabolic aetiologies were listed as second aetiology in all cases.b The aetiologies of immune, metabolic and infectious were considered together in the category of "other" due to small sample size.c The probabilities were calculated using Pearson Chi-square test (two-tailed) across each aetiology, when sample size was appropriate (not done for combined epilepsy/unknown epilepsy/DEE/ASD).P-value < 0.05 is significant.syndromes and the acknowledgement of genetic mechanisms as a dedicated subsection for each Epilepsy syndrome.The presence of aetiology specific syndromes, mainly restricted currently to the infant age group, did not impact the overall Epilepsy Syndrome rate in our study, as these are individually very rare (1.1 % in our cohort).The frequency of reported Epilepsy Syndromes is not expected to dramatically change, given the consideration already afforded to Epilepsy Syndrome diagnosis in paediatric practice and prior literature on their importance.However, the ILAE 2022 consensus helps standardise criteria for Epilepsy Syndromes and facilitates comparisons across different settings.In the future, as more aetiology-specific syndromes are identified, they will present as, co-exist with, and evolve into more traditional Epilepsy Syndromes given the spectrum of phenotypes associated with the many molecular-and other specific aetiologies.Among our cohort of 356 children, we found a higher rate of focal epilepsies (46.1 %) in CWE akin to two recent studies, although conducted in different settings [15,16] (Sup.Table S1).In our cohort, focal epilepsies were more common with younger onset, and generalised epilepsies slightly more frequent in older children although these trends were not statistically significant (Fig. 2).In the Norwegian study by Aaberg et al. [15], the new combined focal and generalised category made up 18 % of CWE (with less patients classified as generalised epilepsies using the ILAE 1989 Classification), while our study reports 6.2 % combined focal and generalised.We noted that Epilepsy Syndromes were more prevalent in CWE under age 6 years and were more frequently associated with DE/DEE than those syndromes in older children.Applying the 2017 "developmental and epileptic encephalopathy (DEE)" descriptions, we found the rate (11.8 %) similar to the rate of 11 % of CWE found by the population-based study conducted by Aaberg et al., and lower than the identified DEE (37 %) described by Symonds et al. following their description of a younger paediatric cohort [15,17].In addition, we also applied the term "developmental encephalopathy (DE)" and identified an additional 138 (38.8 %) CWE, which parallels and underscores the massive developmental comorbidity rate in current clinical practice.
In our cohort it was possible to assign aetiology in a high proportion of CWE (70.7 %), although 35.7 % were of presumed genetic aetiology.When compared to the study by Aaberg et al [15], we found a higher overall aetiology rate, and twice as many in genetic aetiology (55.6 % of CWE, when using both the presumed and molecular genetic subgroups), and less in the unknown, and structural aetiologies.The higher genetic aetiology noted here likely reflects the inclusion of criteria for presumed genetic aetiology, and older children as part of our cohort compared to Aaberg et al, [15], as well as increasing use of NGS diagnostics (see Sup. Table S1).Compared to historical data, the percentage of patients of the overall cohort with a specific molecular aetiology is high (20 % of 365 CWE) [36][37][38].As such, the 2017 classification adds precision to epilepsy diagnostic categorisation compared to previous systems.It could also be argued that a proportion of patients in our study not included following non-diagnostic imaging, metabolic and genetic investigations (e.g.non-acquired focal epilepsy or a DEE or DE) remaining unexplained, are also presumed genetic or will go on to have a genetic cause in future [39,40].
When we include only mechanism-known aetiologies (molecular genetic, structural, immune, metabolic, and infection, but excluding presumed genetic) a noteworthy fall in (identified) aetiology proportion to approximately 34 % occurred.While these known aetiology rates are still reportedly higher than studies of the pre-genomics era (18-26 %) [36][37][38] it also reveals that the advances in diagnostics have not entirely uncovered aetiology for many, and its application is predominantly for

TSC2
Used vigabatrin [26] Yes, seizure free at follow up younger CWE with the most severe phenotypes.While no patients with immunological or infectious causation were detected in the study by Aaberg et al. [15], in our study a small percentage were noted.It is noteworthy that in comparison to our data, the prospective epidemiological study by Symonds et al. [17], which evaluated all children under the age 3 years in Scotland, reported classification of aetiology in 54 % of cases (no presumed genetic aetiology included), while identifying aetiology as a key determinant of drug resistant epilepsy and global delay.They also describe structural aetiology found in 18 % of CWE (similar to the 18.8 % reported in our cohort).A higher rate of infectious and acquired structural aetiologies in resource poorer settings was recently described by Sharma et al [16].In their study, 18 % aetiology were infectious, and prior perinatal insult occurrence in 37 %.In our study structural aetiology was more common in infants (Table 2), more frequently due to acquired insults as well as malformations of cortical development as expected for resource richer settings.The significance of comorbidity for CWE is highlighted in the ILAE 2017 classification, encouraging early identification (optimising care and informing prognosis), and again in the ILAE 2022 positional papers where they are now included as criteria for Epilepsy Syndrome diagnosis [2,6,7].The integrated approach to epilepsy classification (2017) prompted us to capture the rates of the neurodevelopmental co-morbidities (identified approximately 53 % of CWE), uncovering a major burden of care above and beyond seizure management.These neurodevelopmental co-morbidities were higher among younger-onset CWE (Fig. 2).We found that 38.8 % of CWE met criteria for DE, and DEE in a further 11.8 %, associated with a range of co-morbidities categorised as global delay (29 %), severe neurological impairment (SNI) (16 %), isolated delay (15 %) and neuropsychiatric (e.g.ASD) (24 %).While children with DEEs were more straight forward to define (due to the coexisting developmental and epileptic processes), it was slightly more challenging to define precise lists of co-morbidities for milder developmental impairments, and therefore delineate CWE with DE.Given the increased occurrence of seizure disorders in the setting of neurodevelopmental co-morbidities/ phenotypes such as ASD, ID and language disorders [41,42] these particular entities could be considered for inclusion as part of the DE category.Co-morbidity rates were not examined in the Norwegian cohort study, but in the hospital-based resource poor setting evaluated by Sharma et al, 48 % of patients were described as having developmental delay/intellectual disability, while 23 % had CP, which was unsurprising given the high rate of structural causes (59 %, many of which were acquired) in this resource poor setting [16].The rate of DE was not described in either study [15,16].Furthermore, Symonds et al, 2021 found GDD in 193/390 (49 %) patients, similar rates to ours for that age group.Our cohort of presumed genetic were less affected by global delay and/or neurodevelopmental co-morbidities, except ASD which was slightly higher (27 %, Table 2).
The accurate classification of seizures and epilepsies in clinical practice has allowed for individualised care approaches for many decades (Sup.Fig. S1).By including aetiology and co-morbidities, complementary pieces of the diagnostic puzzle and decision trees, inform optimal individualised or personalised care in epilepsy.The confirmed molecular aetiology cohort, which represents approximately 20 % of our patients (more commonly seen in younger CWE and DE) and is higher than that seen in two recent studies [15,16]), are potentially targetable by precision therapy to their specific molecular diagnosis, if one was available.However, specific molecular-based precision treatments were deemed available (and with limited evidence base) for 21/356 (5.9 %) and utilised in 13/356 (3.7 %) CWE.The reason that such therapies were not used in all CWE is because the seizures, which existed as part of their overall neurodevelopmental disorder, were either remitted or "stable" for some patients.Benefit was noted in 6/356 (1.7 %) of overall cohort [or 6/71 (8.5 %) of the molecular cohort], described as mostly transient or inferred effects (or for e.g. in SCN1A-DEE by drug avoidance), and did not target modification of the fundamental pathophysiologicalmechanism causing the neurodevelopmental disease i.e. the DEE/DE.
It is fortunate that we can sometimes predict self-limiting epilepsies or drug responsiveness affecting many CWE, such as many presumed genetic GGE syndromes or SeLECTS group, however, collectively this group also experience some degree of neurodevelopmental comorbidity (28.3 % of presumed genetic aetiology patients, 24.6 % of GGE patients).Without precise understanding of the molecular mechanisms in this group, it is difficult to prevent the associated developmental comorbidities.
Our study describes our experience in a mainly second level centre documenting the paucity of molecular precision therapies available in current practice.This is not surprising given that most of the recent literature on precision therapy consists of papers derived from case reports or multicentre multinational series of single rare monogenic conditions, or data from disease specific cohorts in advanced tertiary or quaternary centres.Nevertheless, there are plenty of examples, in rare and ultra-rare monogenic disorders, where much better outcomes can be achieved with variously defined precision therapies that can modify the phenotype independent of actual correction or modification of the underlying molecular defect [43].We are limited because hospital-based and population-based studies cannot capture all the hundreds of rare monogenic forms of molecular genetic-associated epilepsy described.It is well established that the benefits of genetic diagnosis go above and beyond seizure control [44,45].The prospective study by Symonds et al. highlighted the importance of genetic testing in children younger than 3 years and described treatment implications in 64/80 (80 %) of those with a single gene diagnosis [46].Despite the paucity of molecular precision therapy available, it is important to remain optimistic as the dawn of the molecular era in development of disease modifying molecular therapies is now on the horizon [47].
Limitations of this study are like that of all retrospective studies, in that data collection may be prone to certain biases.We commenced the review to gain perspective of our practice and spectrum of CWE in the adjacent region (shared patients also attending tertiary centres were included) with a view to becoming a larger referral centre for wider region.Detailed chart review of each patient's notes, and prospective application of the recent ILAE systems to the data was performed.Our centre represents a resource rich setting, and we did not examine any ethnic minorities and other epidemiological factors, apart from the Irish Traveller population (5.6 % of our cohort).
In conclusion, there is a paucity of studies describing 2017/2022 ILAE seizure and epilepsy classification systems in paediatric practice.The recent 2017 and 2022 ILAE systems allow a multidimensional approach to classification and includes co-morbidities.Focal epilepsies were more common than generalised, and a high rate of Epilepsy Syndromes in the paediatric population (~40 %) was identified, including new aetiology-specific syndromes.Neuro-developmental comorbidities are prevalent (~53 %).Many CWE had genetic aetiology (~20 % molecular), and a smaller proportion had a known-mechanistic aetiology, more common in younger children.Lastly, there is a monumental gap between aetiological identification, molecular genetic diagnosis and impact of meaningful molecular based "precision therapies", reflecting major challenges in the provision of routine epilepsy care in current clinical practice.

Fig. 2 .
Fig. 2. Epilepsy type and selected co-morbidities in patients with different age of epilepsy onset.The probability values were calculated using Pearson Chi-square testing (two sided) where appropriate.**is used to denote a significant difference and p value of < 0.05.Abbreviations: y, years of age; Focal Epil, focal epilepsy; Gen Epil., generalised epilepsy; BP or DD, behavioural-psychiatric co-morbidity or developmental delay/impairment; DEE, developmental and epileptic encephalopathy; DE, developmental encephalopathy; GDD, global developmental delay; SNI, severe neurological impairment; DRE, drug resistant epilepsy; ASD, autism spectrum disorder.

as part of a DE/DEE
d IGE/GGE syndromes and DEE/DE syndromes are not mutually exclusive (e.g.EMAtS is a GGE and DEE syndrome).e GGE with no syndrome are not included in total number of Epilepsy Syndrome.

Table 3
Molecular-based precision therapy approach deployed in 13/71 patients molecularly confirmed genetic aetiologies In individual cases, the benefit (seizures or developmental improvement) was mostly transient (e.g., 4-aminopryidine use in GOF KCNA2-DEE) or inferred by avoiding potential drug-induced exacerbation e.g. in SCN1A-DEE.See Supplementary Figure for precision therapy criteria.