Phenotype‐guided etiologic workup in a prospective cohort of 144 adults with developmental and epileptic encephalopathy

Abstract

Objectives

Adults with developmental and epileptic encephalopathies (DEEs) often enter adult neurology care without etiologic clarification because of incomplete transition from pediatric services, outdated investigations, and attenuation of childhood electro‐clinical features over time. We aimed to assess the clinical utility and diagnostic yield of a structured, phenotype‐guided etiologic workup in adults meeting study criteria for DEE and to explore associations between electro‐clinical phenotype and etiologic category.

Methods

We prospectively enrolled consecutive adults (≥18 years) with drug‐resistant epilepsy and neurodevelopmental impairment temporally related to epileptic activity, consistent with the ILAE operational definition of DEE, referred to a tertiary epilepsy center between May 2022 and December 2025. Patients underwent a 4‐phase reassessment pathway including critical review of prior documentation, detailed clinical and semiologic phenotyping, prolonged video‐EEG monitoring, phenotype‐guided genetic testing (targeted resequencing, chromosomal microarray), high‐resolution neuroimaging, and multidisciplinary case review to support etiologic clarification and treatment planning.

Results

Among 144 patients (mean age 28.4 years; 57.6% male), a confirmed etiology was identified in 91 (63.2%). Genetic causes predominated (65 patients, 45.1%), including monogenic (53, 36.8%) and chromosomal (12, 8.3%) disorders, primarily resolved via targeted resequencing gene panels and via chromosomal microarray analysis respectively, while structural‐metabolic etiologies accounted for 26 cases (18.1%). Lennox–Gastaut syndrome was significantly associated with structural‐metabolic and chromosomal etiologies (69.2% and 58.3% vs. 9.4% in monogenic; p < 0.0001), whereas, excluding patients with a broader Lennox–Gastaut syndrome phenotype, absence seizures occurred exclusively in monogenic cases (p = 0.024). Etiologic clarification changed clinical management in 27.1% of the cohort, including antiseizure medication optimization/precision therapies (25%) and pre‐surgical referral (2.1%).

Significance

In adults with suspected DEE, the practical value of this approach lies not only in diagnostic yield, but in showing that structured adult reassessment remains clinically worthwhile. Electro‐clinical phenotype can help prioritize etiologic testing, and etiologic clarification can support more informed treatment planning and counseling even long after transition from pediatric care.

Plain Language Summary

Adults with developmental and epileptic encephalopathy (DEE) often reach adult epilepsy services without a clear diagnosis. In this study, we used a structured approach combining clinical review, video‐EEG, MRI, genetic testing, and multidisciplinary discussion in 144 adults with suspected DEE. We identified a confirmed cause in 63.2% of patients. Genetic causes were the most common, but structural and metabolic causes were also important. Some seizure patterns helped guide diagnosis: Lennox–Gastaut syndrome was more common in structural‐metabolic and chromosomal cases, while absence seizures occurred only in monogenic cases. Identifying the cause also helped guide treatment decisions, including medication changes, targeted therapies, surgery referral, and genetic counseling. This study shows that adults with suspected DEE can still benefit from careful diagnostic reassessment, even years after childhood.

Key points.

• In a prospective cohort of 144 adults with suspected DEE, a structured phenotype‐guided workup identified a confirmed etiology in nearly two‐thirds of patients.

• Genetic diagnoses predominated, but a substantial proportion of patients had structural‐metabolic etiologies, supporting the need for a broad multidisciplinary approach.

• Electro‐clinical phenotype provided useful diagnostic clues: LGS was enriched in structural‐metabolic and chromosomal etiologies, whereas absence seizures were restricted to monogenic cases.

• Etiologic clarification remained clinically actionable in adulthood, informing treatment optimization, targeted therapies, presurgical referral, and genetic counseling.

• Adult reassessment of longstanding, previously unresolved DEE should be considered worthwhile, particularly after incomplete transition from pediatric care or outdated prior investigations.

1. INTRODUCTION

Developmental and epileptic encephalopathies (DEEs) are a heterogeneous group of disorders in which developmental impairment arises from the combined effects of the underlying etiology and the epileptic process itself, including seizures and epileptiform activity. ¹ , ² Although increasingly recognized across the lifespan, adults with suspected DEE often reach adult neurology services without etiologic clarification because of incomplete transition from pediatric care, fragmented historical documentation, attenuation of childhood electro‐clinical features, and outdated investigations. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ Moreover, major advances in genetic testing, neuroimaging, and precision‐oriented therapeutics have substantially expanded the diagnostic and therapeutic landscape of DEEs. ¹ , ³ , ⁵ , ⁹ , ¹⁰ , ¹¹ , ¹² In this context, we prospectively applied a structured, phenotype‐guided reassessment pathway for adults meeting study criteria for DEE referred to a specialized tertiary epilepsy center. Our aim was to determine whether systematic adult re‐evaluation could reduce long‐standing diagnostic uncertainty and provide information relevant to treatment planning. We therefore assessed the diagnostic yield of this approach, characterized the etiologic spectrum identified, explored whether electro‐clinical features could help prioritize etiologic testing, and examined the clinical management implications of etiologic clarification in adult DEE care.

2. MATERIALS AND METHODS

2.1. Study setting and patient cohort

The protocol was applied prospectively to a cohort of adult patients consecutively referred to the outpatient clinic for rare and complex epilepsies within the Epilepsy Center–Neurology Unit of the Fondazione IRCCS Casa Sollievo della Sofferenza, San Giovanni Rotondo, Italy, between May 2022 and December 2025.

2.1.1. Inclusion criteria

Patients were included if they met all of the following criteria:

1. Age ≥ 18 years at the time of enrollment.

2. Drug‐resistant epilepsy, defined according to the International League Against Epilepsy (ILAE) criteria. ¹³

3. Neurodevelopmental impairment (intellectual disability, developmental delay, or regression) documented by clinical assessment and standardized neuropsychological testing.

4. Evidence of DEE, defined according to the ILAE operational definition. ²

5. Epilepsy onset and/or evidence of neurodevelopmental impairment documented before the age of 18 years, consistent with the childhood or early‐onset nature of DEE.

2.1.2. Exclusion criteria

Patients were excluded if:

1. Neurodevelopmental impairment was clearly independent of epileptic activity (i.e., epilepsy with intellectual disability [ID+E] without evidence of epileptic encephalopathy).

2. Epilepsy was acquired in adulthood (e.g., post‐stroke, post‐traumatic) without pre‐existing or epilepsy‐related developmental impairment.

3. Incomplete clinical data or insufficient follow‐up to establish the temporal relationship between epilepsy and developmental impairment.

2.1.3. Distinction between DEE and ID+E

The critical distinction between DEE and ID+E was established through comprehensive review of developmental history, including: ² , ³ , ⁴

• Temporal correlation: Documentation of developmental slowing, stagnation, or regression coinciding with or following epilepsy onset.

• EEG burden: Assessment of interictal epileptiform discharge frequency and correlation with cognitive/behavioral changes.

• Treatment response: Evidence of developmental improvement following seizure control or reduction of epileptiform activity.

• Multidisciplinary assessment: Integration of neurological, neuropsychological, and electrophysiological data to determine whether epileptic activity contributed to developmental impairment (see Supporting Information).

2.2. Patient screening and enrollment flow

Between May 2022 and December 2025, a total of 1456 adult patients were referred to the rare and complex epilepsy outpatient clinic and assessed for eligibility. To be included in the final cohort, patients had to be completely undiagnosed in terms of etiology; any individual with an already established genetic, metabolic, or definitive structural cause at baseline was strictly excluded. Consequently, all 144 sequentially enrolled cases entered the study with an unknown etiology.

Of these, 1312 were excluded after initial screening: 427 did not meet the ILAE criteria for drug‐resistant epilepsy; 330 had neurodevelopmental impairment clearly independent of epileptic activity and were therefore classified as epilepsy with intellectual disability (ID+E) rather than DEE; 199 had epilepsy acquired in adulthood without pre‐existing epilepsy‐related developmental impairment; and 356 had incomplete clinical data precluding adequate temporal reconstruction of the relationship between epilepsy onset and developmental trajectory. The remaining 144 patients fulfilled all inclusion criteria and constituted the study cohort. The DEE/ID+E classification was established by consensus of the multidisciplinary team described above. In cases of diagnostic uncertainty, a third independent epileptologist was consulted and consensus was reached in all instances.

Final Cohort: A total of 144 patients (M/F: 83/61; ratio 1.4:1) with a mean age of 28.4 ± 10.3 years (median 26 years, range 18–58 years) met all inclusion criteria and were enrolled. All patients presented with drug‐resistant epilepsy, having failed a median of 4 ASMs prior to enrollment (range: 2–11 ASMs).

The Epilepsy Center operates with a fully integrated multidisciplinary team (MDT) comprising epileptologists, a clinical neurophysiology technician, a neuropsychologist, a medical geneticist, molecular biologists, and neuroradiologists. Epileptologists oversaw clinical phenotyping (Phase I), electro‐clinical assessment (Phase II), and treatment design (Phase IV). The clinical neurophysiology technician was responsible for prolonged video‐EEG acquisition and analysis (Phase II). The neuropsychologist performed standardized cognitive and behavioral assessments (Phase III). The genetics team, including a medical geneticist and molecular biologists, managed phenotype‐guided genetic testing and variant interpretation (Phases III and IV). Neuroradiologists performed critical re‐evaluation of high‐resolution MRI (Phases I and III). This coordinated approach, facilitated by co‐location within a single institution, allowed for prompt validation of findings and integrated data interpretation. In particular, the refinement of our diagnostic workflow relies on bi‐weekly multidisciplinary consensus meetings where raw electrographic data (including long‐term polygraphic video‐EEG) and high‐resolution neuroimaging sequences are reviewed side‐by‐side with molecular genetics findings. This specific synergy allows the unit to reconcile borderline electro‐clinical phenotypes with genomic variants, maximizing diagnostic accuracy in a population prone to atypical adult clinical presentations (Figure 1).

FIGURE 1.

Adult DEE roadmap. Four‐phase protocol for systematic etiological diagnosis in adults with developmental and epileptic encephalopathy, integrating clinical phenotyping (Phase I), video‐EEG monitoring (Phase II), targeted investigations (Phase III), and multidisciplinary treatment planning (Phase IV).

2.3. Protocol description

2.3.1. Phase I: Clinical and Semiologic phenotyping

• Critical Review of Prior Documentation: The Roadmap begins with the mandatory critical re‐evaluation of all pre‐existing electro‐clinical, radiological, neuropsychological, and genetic data. This involves reviewing the raw images (EEGs, MRIs) rather than simply accepting written reports and re‐discussing prior genetic testing results with our dedicated neuroradiologists and geneticists, along with re‐evaluating neuropsychological assessment data.

• Detailed History and Development: We collect an exhaustive patient history, including perinatal events, early developmental milestones, family history, and the evolution of non‐epileptic comorbidities (e.g., intellectual disability, psychiatric disorders, and movement disorders).

• Seizure Semiologic Analysis: This involves meticulous documentation of seizure types, frequency, severity, and response to previous antiseizure medications (ASMs).

• Differential Diagnosis: A rigorous evaluation is conducted to differentiate epileptic seizures from paroxysmal non‐epileptic events and to assess whether the developmental impairment is independent of or significantly exacerbated by the epileptic activity.

2.3.2. Phase II: The electro‐clinical assessment

• Prolonged EEG/Video‐EEG Monitoring: Long‐term video‐EEG/polygraphic monitoring is performed to capture both ictal and interictal epileptiform activity, assess the background rhythm, document paroxysmal episodes, and confirm specific electro‐clinical syndromes such as Lennox–Gastaut syndrome (LGS) or support the clinical suspicion of disorders like Dravet syndrome.

• Electro‐Clinical Correlation: We correlate the patient's clinical symptoms with the video‐EEG/polygraphic findings. The electro‐clinical pattern is used to stratify the patient into categories of suspicion (e.g., suspicion of genetic etiology, structural lesion), informing the subsequent request for specialized tests.

2.3.3. Phase III: Targeted etiological investigations

• Neuropsychological assessment: A standardized neuropsychological assessment is administered to evaluate the pattern and severity of cognitive impairment, developmental delay, or regression. This standardized assessment includes comprehensive measures of Intelligence Quotient (IQ) (e.g., WISC‐V) and key domains of Executive Performance Index (EPI). Furthermore, for conditions like DEEs, it integrates tools such as the Adaptive Behavior Assessment System (ABAS‐3) and the Behavior Rating Inventory of Executive Function (BRIEF2) to accurately capture daily functional and behavioral impacts. For patients with severe intellectual disability precluding standardized IQ testing, adaptive behavior scales (ABAS‐3) and clinical assessment were used to characterize functional level.

• Neuroimaging: High‐resolution Brain MRI was performed on a 3 T scanner using a dedicated epilepsy protocol including 3D T1‐weighted volumetric sequences (1 mm isotropic), T2‐weighted, FLAIR, and susceptibility‐weighted imaging (SWI). For patients exhibiting severe intellectual disability, severe claustrophobia, or involuntary movements precluding high‐quality acquisition, imaging was performed under deep intravenous sedation monitored by an anesthesiologist.

• Metabolic and Immune Workup: Testing for underlying metabolic disorders or autoimmune etiologies was performed where clinically indicated by specific symptoms (e.g., rapid regression, new‐onset focal deficits). Particularly, a baseline metabolic screening—including plasma/urinary amino acids, urinary organic acids, acylcarnitines, lactate, ammonia, and homocysteine—was performed in all 144 patients during their baseline evaluation. Targeted second‐tier investigations were reserved for 10 patients presenting with specific clinical or neuroradiological red flags.

• Genetic testing (see Supporting Information):

  All 144 patients carried a clinical diagnosis of DEE and underwent first‐tier genetic testing with targeted resequencing (TRS) and chromosomal microarray analysis (CMA), making the execution of these initial molecular investigations independent of the clinical phenotype. However, the electro‐clinical phenotype identified in Phase II served as a crucial ‘diagnostic compass’ to guide variant interpretation and prioritize candidate genes. Specifically, detailed phenotyping allowed the multidisciplinary team to immediately target specific genes within the large sequencing data (e.g., prioritizing SCN1A analysis in patients with a clinical phenotype suggestive of Dravet syndrome) and provided the essential framework for “reverse phenotyping” during the evaluation of variants of uncertain significance (VUS).

  CMA was particularly informative in patients with dysmorphic features, growth retardation, or a complex syndromic phenotype suggestive of chromosomal imbalance, whereas TRS was especially relevant when the electro‐clinical pattern strongly suggested a specific gene or syndrome (e.g., clinical Dravet syndrome, prompting priority assessment of SCN1A). TRS, CMA, and clinical interpretation of identified variants were performed as previously described. ¹³ , ¹⁴

2.3.4. Phase IV: The multidisciplinary review and treatment strategy

• Integrated Case Discussion: All cases are reviewed by the MDT, allowing for prompt validation of findings and rapid consensus on VUS interpretation through reverse phenotyping (checking if the genetic finding aligns with any subtle clinical feature).

• Precision Treatment Design: Based on the confirmed etiology (genetic, structural, or metabolic), the treatment plan is designed to be personalized and precision‐guided. This includes identifying specific ASMs, targeted therapies (e.g., ketogenic diet), or considering new pharmacological strategies (e.g., gene‐specific medications) while actively avoiding drugs that could worsen the patient's condition, or evaluating the suitability of advanced interventions, including epilepsy surgery or neuro‐modulation techniques.

2.4. Statistical analysis

Descriptive statistics were used to summarize demographic and clinical characteristics. Continuous variables are presented as mean ± standard deviation (SD) and median with range; categorical variables are presented as frequencies and percentages. Potential associations between categorical variables (etiological group and seizure phenotype) were assessed using Chi‐square (χ ²) test or Fisher's exact test, as appropriate. A two‐sided p‐value <0.05 was considered statistically significant. All analyses were performed using R software (version 4.3.2; R Foundation for Statistical Computing, Vienna, Austria). The etiological categories (genetic, structural, and metabolic) are not mutually exclusive. Conditions such as Tuberous Sclerosis Complex inherently represent a dual genetic‐structural pathology. For categorical and statistical distribution purposes within the tables, patients exhibiting overlapping etiologies were assigned to a single classification group based on the primary diagnostic pipeline that established the management‐changing findings (e.g., tuberous sclerosis complex was categorized within the structural malformations subgroup, although its monogenic molecular nature is fully detailed in the results).

2.5. Ethics statements

This study was approved by the Casa Sollievo della Sofferenza Hospital Ethics Committee (protocol no. 90/CE/2024). Written informed consent was obtained from all patients or, when applicable, from their legal guardians. Also, formal pre‐test counseling was systematically provided to all patients or their legal guardians prior to blood sampling. This clinical consultation explicitly covered the potential identification of Variants of Uncertain Significance (VUS), the clinical management implications of positive results, and the risk of secondary or coincidental findings unrelated to the epilepsy phenotype, ensuring fully informed written consent. All relevant data generated or analyzed during this study are included in this article and its Supporting Information.

This study was designed to evaluate the feasibility and overall diagnostic yield of the structured pathway.

Detailed electro‐clinical characterization, genotype–phenotype correlations, and long‐term treatment outcomes will be reported in subsequent publications.

3. RESULTS

3.1. Overall diagnostic yield of the multi‐level pathway

The overall diagnostic yield achieved through our prospective, multi‐level phenotype‐guided pathway was 63.2%, successfully identifying a definitive etiology in 91 out of the 144 initially undiagnosed adult patients.

This cumulative diagnostic rate is distributed across two main etiological macro‐categories:

• Genetic causes accounted for the largest share, providing a diagnosis in 45.1% (n = 65/144) of the total cohort. Within this genetic group, 36.8% (n = 53/144) of the total cohort had a monogenic etiology, primarily resolved via targeted NGS gene panels or sequencing techniques, while 8.3% (n = 12/144) were identified with a chromosomal etiology via chromosomal microarray analysis (CMA).

• Structural‐metabolic etiologies were established in 18.1% (n = 26/144) of the total cohort. This group was predominantly driven by cortical malformations (n = 15), followed by acquired vascular (n = 6), infectious (n = 2), neoplastic (n = 1), and specific metabolic lesions (n = 2). Notably, 9 of these structural cases (34.6% of the 26 structural‐metabolic successes) were exclusively clarified due to the clinical execution of high‐resolution 3 T MRI protocols under deep sedation, which effectively eliminated severe motion artifacts.

The remaining 36.8% (n = 53/144) of the cohort could not be etiologically resolved with the current protocol. Importantly, the identification of a specific underlying etiology led to a direct change in clinical management or targeted therapy—including tailored ASM adjustments, avoidance of contraindicated drugs, or surgical evaluation—in 31.3% of these 91 solved cases.

3.1.1. Patient demographics

The final cohort comprised 144 patients (M/F: 83/61; ratio 1.4:1) with a mean age of 28.4 ± 10.3 years (median 26 years, range 18–58 years). All patients presented with drug‐resistant epilepsy, defined as failure of ≥2 appropriately chosen ASMs, having failed a median of 4 ASMs prior to enrollment (range: 2–11). Notably, a documented history of at least one episode of status epilepticus (SE)—including convulsive, non‐convulsive, or prolonged focal motor status—was present in 55 out of 144 patients (38.2%). Convulsive SE episodes occurring predominantly during infancy or childhood were highly represented in patients with underlying monogenic etiologies (such as SCN1A variants) and those with acquired structural lesions. Neurodevelopmental impairment (intellectual disability, developmental delay, or regression) temporally related to epilepsy onset or epileptic activity was documented in all patients, consistent with the inclusion criteria for DEE.

Upon enrollment, all 144 patients completely lacked a confirmed etiological diagnosis, having reached adult care with non‐specific diagnostic labels such as “idiopathic” or “cryptogenic” drug‐resistant epilepsy. Regarding prior syndromic frameworks, the historical documentation was highly fragmented: 86 patients (59.7%) had never received a specific syndromic classification in childhood. The remaining 58 patients (40.3%) carried tentative or historical pediatric syndromic labels. Our structured Phase I and II electro‐clinical reassessment allowed us to systematically re‐evaluate these baseline assumptions. For instance, among the 30 patients ultimately classified with LGS in this study, the syndromic label was historically suspected and firmly validated in adulthood in 18 cases, whereas it was established de novo in 12 cases due to incomplete historical records or attenuation of childhood features over time.

3.1.2. Protocol completion

All 144 patients completed the structured diagnostic pathway, with phase‐specific investigations applied according to clinical indication. Of the 68 patients who underwent prolonged video‐EEG monitoring (Phase II), 44 experienced a total of 440 ictal events over 2232 recording hours. High‐resolution brain MRI was repeated or performed de novo in 55 patients, 29 of whom required sedation. All 55 patients had undergone prior lower‐resolution neuroimaging (CT or 1.5 T MRI) in childhood or adolescence, meaning none were entirely imaging‐naive. The repeated high‐resolution 3 T MRI protocol provided significant added diagnostic value in 18 out of the 55 patients (32.7%). Within the specific subgroup of 29 patients scanned under sedation, the repeat 3 T MRI provided critical added value in 9 cases (31.0%) by successfully identifying subtle malformations of cortical development or structural details that had been obscured by motion artifacts or low resolution on historical tracks.

All 144 patients underwent genetic testing. Targeted resequencing (TRS) and chromosomal microarray analysis (CMA) were performed in all patients as first‐tier investigations. Patients who tested negative on these analyses and had high clinical suspicion for a monogenic etiology were referred for whole‐exome sequencing (WES), which is still ongoing.

3.1.3. Diagnostic yield and etiological distribution

The structured phenotype‐guided workup led to the identification of a confirmed etiology in 91 patients (63.2%). The remaining 53 patients (36.8%) remained without a confirmed etiology and were classified as etiology unknown (Table 1).

TABLE 1.

Etiological distribution in 144 adults with developmental and epileptic encephalopathy.

Etiological category	Subcategory detail	Number of patients	Percentage of Total (N = 144)
Confirmed Etiology		91	63.2%
Genetic	Total	65	45.1%
	Monogenic	53	36.8%
	Chromosomal	12	8.3%
Structural‐Metabolic	Total	26	18.1%
	Cortical Malformations	15	10.4%
	Vascular	6	4.2%
	Infectious	2	1.4%
	Neoplastic	1	0.7%
	Metabolic	2	1.4%
Etiology Unknown (EU)		53	36.8%
Total Cohort		144	100.0%

Note: A confirmed etiology was identified in 91 patients (63.2%) through the structured phenotype‐guided workup. Genetic etiologies predominated (65 patients, 45.1%), including monogenic disorders and pathogenic chromosomal copy number variants. Structural‐metabolic etiologies accounted for 26 cases (18.1%), with cortical malformations representing the largest subcategory. Fifty‐three patients (36.8%) remained without a confirmed etiology despite comprehensive evaluation. Cortical malformations include tuberous sclerosis complex, lissencephaly/pachygyria, subcortical band heterotopia, and polymicrogyria. Metabolic etiologies include mitochondrial complex I deficiency (n = 1) and urea cycle disorder (n = 1).

Abbreviation: n, number of patients.

The etiological distribution was as follows:

• Genetic etiologies: 65 patients (45.1%)

  • ⚬
    Monogenic: 53 patients (36.8%)
  • ⚬
    Chromosomal: 12 patients (8.3%)
• Structural‐metabolic etiologies: 26 patients (18.1%)

  • ⚬
    Structural: 24 patients (16.7%)
  • ⚬
    Metabolic: 2 patients (1.4%)

3.1.4. Detailed etiological findings

Genetic group

The genetic cohort (n = 65) was further categorized into monogenic and chromosomal subgroups.

The Monogenic subgroup (n = 53, 36.8% of total cohort) revealed pathogenic or likely pathogenic variants in established DEE‐associated genes (Table 2). Genes with recurrent mutations (identified in more than one patient) included SCN1A, MECP2, CACNA1A, GRIN2A, SMC1A, TSC1, and SCN8A.

TABLE 2.

Summary of genetic findings in the monogenic DEE group.

Gene	Patient ID	Gender	Age (y)	Nomenclature	Type	Zygosity	Inheritance	Parental origin	ACMG classification
ANKRD11(NM_013275.6)	P1	M	25	c.2716C>T, p.(Arg906*)	Nonsense	Het	AD	dn	P
ADGRV1 (NM_032119.4)	P2	M	18	c.1701delC, p.(Leu568fs*8)	Frameshift	Het	AD	pat	LP
ANKRD17(NM_032217.5)	P3	M	33	c.2243delC, p.(Pro748Glnfs*61)	Frameshift	Het	AD	dn	P
ARID1B(NM_001374828.1)	P4	M	27	c.4072_4073del, p.(Ser1358Cysfs*10)	Frameshift	Het	AD	dn	P
ATP11A(NM_015205.3)	P5	F	45	c.1105G>C, p.(Glu369Gln)	Missense	Het	AD	dn	LP
CACNA1A(NM_001127222.2)	P6	M	40	c.3787G>A, p.(Glu1263Lys)	Missense	Het	AD	dn	LP
	P7*	M	26	c.4174G>A, p.(Val1392Met)	Missense	Het	AD	dn	P
	P8*	M	27	c.4174G>A, p.(Val1392Met)	Missense	Het	AD	dn	P
CAMK2B(NM_001220.5)	P9	M	18	c.1881delC, p.(Asp627Glufs*115)	Frameshift	Het	AD	dn	LP
CDKL5(NM_001323289.2)	P10	M	18	c.1247_1248del, p.(Glu416Valfs*2)	Frameshift	Hem	XL	NA	P
CHD2(NM_001271.4)	P11	M	28	c.2105A>G, p.(Lys702Arg)	Missense	Het	AD	dn	LP
CTCF(NM_006565.4)	P12	M	20	c.409dupG, p.(Ala137Glyfs*3)	Frameshift	Het	AD	dn	P
DEAF1(NM_021008.3)	P13	F	25	c.835 T>G, p.(Cys279Gly)	Missense	Het	AD/AR	dn	LP
DYNC1H1(NM_001376.5)	P14	M	46	c.10031G>A, p.(Arg3344Gln)	Missense	Het	AD	dn	P
GABRA1(NM_001127644.2)	P15	F	21	c.839C>A, p.(Pro280Gln)	Missense	Het	AD	dn	P
GABRA5(NM_000810.4)	P16	F	20	c.880G>T, p.(Val294Phe)	Missense	Het	AD	dn	P
GRIN2A(NM_001134407.3)	P17*	M	22	c.1783del, p.(His595Metfs*59)	Frameshift	Het	AD	dn	LP
	P18	F	20	c.2356G>A, p.(Gly786Ser)	Missense	Het	AD	NA	LP
	P19*	M	27	c.1783del, p.(His595Metfs*59)	Frameshift	Het	AD	pat	LP
MECP2(NM_001110792.2)	P20	F	22	c.504C>G, p.(Asp168Glu)	Missense	Het	XL	dn	P
	P21	F	32	c.491C>G, p.(Pro164Arg)	Missense	Het	XL	dn	P
	P22	F	19	c.916C>T, p.(Arg306*)	Stopgain	Het	XL	dn	P
	P23	F	21	c.491C>G, p.(Pro164Arg)	Missense	Het	XL	dn	P
	P24	F	21	c.961C>T, p.(Arg321Trp)	Missense	Het	XL	dn	P
NEXMIF(NM_001008537.3)	P25	F	26	arr[GRCh37] Xq13.2q13.3(73855585_74202615)x1	CNV Loss	Het	XL	dn	LP
PCDH19(NM_001184880.2)	P26	F	23	c.1456G>C, p.(Gly486Arg)	Missense	Het	XL	NA	LP
PIGN(NM_176787.5)	P27	F	19	c.1A>G, p.(Met1?) c.2354G>A, p.(Arg785His)	Start‐loss Missense	Compound Het	AR	pat mat	LP LP
SCN1A(NM_001165963.4)	P28*	F	23	c.3824G>A, p.(Gly1275Asp)	Missense	Het	AD	mat	LP
	P29*	F	18	c.3824G>A, p.(Gly1275Asp)	Missense	Het	AD	mat	LP
	P30	F	26	c.5632G>A, (p.Glu1878Lys)	Missense	Het	AD	pat	LP
	P31	F	29	c.335C>T, p.(Thr113Ile)	Missense	Het	AD	dn	P
	P32	F	23	c.5347G>A, p.(Ala1783Thr)	Missense	Het	AD	dn	P
	P33	F	24	c.602 + 1G>A, p.?	Splicing	Het	AD	NA	P
	P34	M	31	c.1511_1515del, p.(Arg504Thrfs*12)	Frameshift	Het	AD	NA	P
	P35	F	20	c.2415 + 1G>C, p.?	Splicing	Het	AD	NA	P
	P36	F	44	c.2803A>T, p.(Asn935Tyr) c.4179 T>A, p.(His1393Gln)	Missense Missense	Het Het	AD AD	pat mat	LP LP
	P37	M	38	c.2803A>T, p.(Asn935Tyr) c.4179 T>A, p.(His1393Gln)	Missense Missense	Het Het	AD AD	pat mat	LP LP
SCN2A(NM_001040142.2)	P38	M	23	c.668G>A, p.(Arg223Gln)	Missense	Het	AD	NA	P
SCN8A(NM_001330260.2)	P39#	F	35	c.5615G>A, p.(Arg1872Gln)	Missense	Het	AD	dn	P
	P40#	F	35	c.5615G>A, p.(Arg1872Gln)	Missense	Het	AD	dn	P
SHANK2(NM_012309.5)	P41	M	40	c.4102dupA, p.(Thr1368Asnfs*36)	Frameshift	Het	Unk	dn	LP
SHANK3(NM_001372044.2)	P42	M	18	c.1121_1123del, p.(Ala374_Arg375delinsGly)	Inframe deletion	Het	AD	dn	LP
SLC6A1(NM_003042.4)	P43	M	58	c.364 T>A, p.(Phe122Ile)	Missense	Het	AD	dn	LP
SLC6A8(NM_005629.4)	P44	F	35	c.395‐2A>C, p.?	Splicing	Het	XL	NA	LP
SMC1A(NM_006306.4)	P45	F	23	c.685 T>A, p.(Tyr229Asn)	Missense	Het	XL	dn	LP
	P46	F	52	c.1113 + 2 T>A, p.?	Splicing	Het	XL	dn	P
SPTAN1(NM_001130438.3)	P47	F	29	c.6946_6948del, p.(Gln2316del)	Inframe deletion	Het	AD	dn	LP
SPTBN1(NM_003128.3)	P48	M	49	c.3858 + 1G>A, p.?	Splicing	Het	AD	NA	LP
SYNGAP1(NM_006772.3)	P49	M	21	c.2024delA, p.(Asn675Thrfs*35)	Frameshift	Het	AD	dn	P
TAOK1(NM_020791.4)	P50	M	33	c.2149‐1G>C, p.?	Splicing	Het	AD	NA	LP
TSC1 (NM_000368.5)	P51	M	53	c.1525C>T, p.(Arg509*)	Stopgain	Het	AD	dn	P
TSC1 (NM_000368.5)	P52	M	53	c.1525C>T, p.(Arg509*)	Stopgain	Het	AD	dn	P
TSC2 (NM_000548.5)	P53	F	52	c.4723delC, p.(Leu1575*)	Stopgain	Het	AD	dn	P

Abbrevaitions: *, sibling; #, twins; AD, autosomal dominant; AR, autosomal recessive; CNV, copy number variation; dn, de novo; F, Female; Hem, hemizygous; Het, heterozygous; LP, likely pathogenic; M, Male; mat, maternal; NA, not available; P, pathogenic; pat, paternal; Unk, unknown; XL, X‐linked.

The Chromosomal subgroup (n = 12, 8.3% of total cohort) included pathogenic copy number variations (CNVs) (Table 3): deletions at 2q24.3q31.1, 2q24.1q24.2, and 22q11.21; duplications at 15q11.2q13.1, Xp11.23p11.22, 7q11.23, and Xq28; and complex rearrangements involving Xq28.

TABLE 3.

Summary of genetic findings in the chromosomal DEE group.

Patient ID	Gender	Age	ISCN nomenclature (2020)	Type	Size (mb)	Genes	Parental origin	ACMG classification
P54	M	24	arr[GRCh37] 2q24.3q31.1(166385112_171074419)x1	Loss	4,7	35	NA	P
P55	F	23	arr[GRCh37] 2q24.1q24.2(155818224_163410336)x1	Loss	7,5	31	dn	P
P56	F	19	arr[GRCh37] 15q11.2q13.1(22770421_28545601)x3	Gain	5,8	20	dn	P
P57	M	33	arr[GRCh37] 15q11.2q13.1(22770421_28545601)x3	Gain	5,8	20	dn	P
P58	M	35	arr[GRCh37] 15q11.2q13.1(22770422_29023368)x3	Gain	6,3	25	dn	P
P59	F	21	arr[GRCh37] Xp11.23p11.22(48886249_52655422)x3	Gain	3,8	67	NA	P
P60	M	22	arr[GRCh37] 22q11.21(18916843_20311858)x1	Loss	1,4	29	mat	P
P61	M	24	arr[GRCh37] 7q11.23(72576873_74256860)x3	Gain	1,7	35	NA	P
P62	M	29	arr[GRCh37] Xq28(151140221_154110306)x2 arr[GRCh37] Xq28(154739500_155233731)x0	Gain Loss	3,0 0,50	74 5	mat mat	P
P63	M	33	arr[GRCh37] Xq28(151537595_153731581)x2	Gain	2,2	76	mat	P
P64	M	19	arr[GRCh37] Xq28(153149405_153624564)x2	Gain	0,47	15	mat	P
P65	M	26	arr[GRCh37] Xq28(153071371_153438781)x2	Gain	0,37	16	mat	P

Abbrevaitions: dn, de novo; F, female; M, male; mat, inherited from the mother; NA, not available; P, pathogenetic.

Structural‐metabolic group

The Structural subgroup (n = 24, 16.7% of total cohort) was characterized by a high prevalence of malformations of cortical development (MCD):

• Cortical malformations: 15 patients (10.4%), including tuberous sclerosis complex, lissencephaly/pachygyria, subcortical band heterotopia (double cortex), and polymicrogyria.

• Vascular lesions: 6 patients (4.2%).

• Infectious sequelae: 2 patients (1.4%).

• Neoplastic: 1 patient (0.7%).

The Metabolic subgroup (n = 2, 1.4% of total cohort) comprised one patient with mitochondrial complex I deficiency and one patient with a urea cycle disorder (UCD) presenting with hyperargininemia, hyperammonemia, and orotic aciduria.

3.1.5. Correlation of etiology and electro‐clinical phenotype

The distribution of seizure types and electro‐clinical syndromes was analyzed separately to ensure methodological rigor and avoid category overlap. Seizure types were classified as mutually exclusive categories based on the predominant seizure semiology (Table 4A), while the presence of LGS was analyzed as an independent binary variable (Table 4B), as LGS encompasses multiple seizure types by definition. Given the presence of zero or low‐frequency counts in several categories, Fisher's exact test was employed for all comparisons.

TABLE 4.

Correlation between etiological category and electro‐clinical phenotype in 91 adults with confirmed etiology.

(A)
Type of seizure/syndrome	Structural‐metabolic (26 patients)	Monogenic (53 patients)	Chromosomal (12 patients)
Focal Motor Seizures	7	16	1
Generalized Motor Seizures	10	21	7
Absence Seizures	0	7	0
Focal Non‐Motor Seizures	1	2	2
Other/Combined	8	7	2

(B)
Lennox–Gastaut syndrome	Structural‐metabolic (26 patients)	Monogenic (53 patients)	Chromosomal (12 patients)
	18	5	7

Note: (A) Distribution of predominant seizure types across etiological groups. Each patient was assigned to a single seizure category based on the predominant seizure semiology. (B) Prevalence of Lennox–Gastaut syndrome (LGS) by etiological group. LGS was analyzed as an independent binary variable, as it encompasses multiple seizure types by definition; patients with LGS are also classified by their predominant seizure type in Table 4A. Absence seizures were observed exclusively in the monogenic group (p = 0.024). LGS was significantly associated with structural‐metabolic and chromosomal etiologies compared to monogenic cases (69.2% and 58.3% vs. 9.4%, respectively; p < 0.0001).

Abbreviations: LGS, Lennox–Gastaut syndrome; n, number of patients; NS, not significant.

3.2. Lennox–Gastaut syndrome

The global association between etiological group and LGS was highly significant (Fisher's Exact Test, p < 0.0001). LGS was present in 69.2% (18/26) of structural‐metabolic cases, 58.3% (7/12) of chromosomal cases, and only 9.4% (5/53) of monogenic cases. A focused pairwise comparison between the structural‐metabolic and monogenic groups confirmed this association (Fisher's Exact Test, p < 0.0001), with an odds ratio of 21.6 (95% CI: 6.3–81.4), indicating that patients with structural‐metabolic etiologies had markedly higher odds of presenting with the LGS phenotype compared to those with monogenic etiologies. The high prevalence of LGS in the chromosomal group (58.3%) suggests that large‐scale genomic rearrangements may also predispose to this severe electro‐clinical phenotype.

3.3. Absence seizures

Excluding patients characterized by a broader LGS phenotype (who inherently experience atypical absences as part of their syndrome), absence seizures were found exclusively in the monogenic group (7/53, 13.2%) and were entirely absent in both the structural‐metabolic (0/26) and chromosomal (0/12) groups. This distribution reached statistical significance (Fisher's Exact Test, p = 0.024), supporting the hypothesis that absence seizures in DEEs are primarily linked to monogenic variants affecting thalamocortical circuit function (e.g., CACNA1A, GABRA1, SLC6A1). ¹⁵

3.4. Generalized motor seizures

Generalized motor seizures were the most common seizure type across all etiological groups, observed in 38.5% (10/26) of structural‐metabolic, 39.6% (21/53) of monogenic, and 58.3% (7/12) of chromosomal cases. However, this distribution did not reach statistical significance (Fisher's Exact Test, p = 0.378), indicating that generalized motor seizures are a common phenotypic endpoint across diverse etiologies in adult DEEs.

3.5. Focal motor seizures

Focal motor seizures were distributed across all three etiological groups: 26.9% (7/26) in structural‐metabolic, 30.2% (16/53) in monogenic, and 8.3% (1/12) in chromosomal cases. No significant association was found between focal motor seizures and etiological group (Fisher's Exact Test, p = 0.297). This clinical overlap underscores that genetic and structural drivers are frequently not mutually exclusive, but rather tightly intertwined. In these overlapping scenarios—such as Tuberous Sclerosis Complex or specific malformations of cortical development driven by somatic mutations—detailed electro‐clinical analysis (Phase II) is essential not to separate these mechanisms into mutually exclusive bins, but rather to determine the predominant network driver, identify potentially resectable epileptogenic zones, and characterize the electrographic syndromic presentation to guide targeted pharmacological or surgical decisions.

3.6. Focal non‐motor and other/combined seizures

Focal non‐motor seizures were infrequent overall (5.5% of the cohort) and showed a trend toward higher prevalence in the chromosomal group (16.7%, 2/12) compared to structural‐metabolic (3.8%, 1/26) and monogenic (3.8%, 2/53) groups, though this did not reach statistical significance. The “Other/Combined” category was most prevalent in the structural‐metabolic group (30.8%, 8/26), potentially reflecting the phenotypic complexity associated with acquired or multifactorial etiologies.

The presence of a structural‐metabolic or chromosomal etiology was strongly associated with the LGS phenotype in this adult cohort, while absence seizures were exclusive to monogenic etiologies. These findings validate the use of the electro‐clinical phenotype as a diagnostic compass: The identification of LGS should prompt targeted neuroimaging and chromosomal analysis, whereas the presence of absence seizures should prioritize monogenic testing targeting ion channel and neurotransmitter receptor genes.

3.6.1. Clinical management implications

Etiologic clarification led to clinically actionable management changes in 39 patients. These included:

• ASM optimization: Twenty‐six patients, including avoidance of contraindicated medications (e.g., sodium channel blockers in SCN1A‐related cases) and selection of etiology‐appropriate ASMs

• Targeted therapies: Ten patients, including fenfluramine (SCN1A), everolimus (TSC‐related), ketogenic diet (metabolic etiologies, SCN1A), and L‐serine (GRIN2A)

• Surgical evaluation: Three patients with focal structural lesions were referred for presurgical workup

Genetic counseling was provided to all patients with confirmed genetic etiologies (i.e., monogenic or chromosomal) and their families regarding disease implications, patient management, and familial recurrence risk.

4. DISCUSSION

In this prospective single‐center cohort of 144 adults meeting study criteria for DEE, a structured phenotype‐guided etiologic workup established a confirmed diagnosis in 91 patients (63.2%). Genetic causes represented the largest diagnostic category, accounting for 65 patients (45.1% of the total cohort), including both monogenic and chromosomal disorders, while structural‐metabolic etiologies were identified in 26 cases (18.1%). In addition, selected electro‐clinical features were distributed across etiologic groups, with LGS enriched in structural‐metabolic and chromosomal etiologies and absence seizures (outside of the LGS framework) observed only in the monogenic subgroup. Taken together, these findings support the clinical value of systematic etiologic reassessment in adults with persistent, drug‐resistant epilepsies and developmental impairment compatible with DEE. For the remaining 36.8% of unresolved cases, alternative genetic mechanisms must be considered, including deep intronic variants, repeat expansions, and low‐level somatic or germline mosaicism (frequently involved in mTORopathies like Tuberous Sclerosis Complex or focal dysplasias), which escape standard TRS or CMA filtering. The implementation of next‐generation Whole‐Genome Sequencing (WGS) represents the ideal future direction to capture both single‐nucleotide variants and complex pathogenic CNVs within a single tool. Furthermore, other classical techniques such as karyotyping remain essential for specific syndromic configurations like Ring Chromosome 20. The diagnostic yield of 63.2% observed in this cohort warrants contextualization against existing data from comparable adult populations. Studies specifically addressing genetic diagnostic yield in adults with epilepsy and comorbid intellectual disability or neurodevelopmental impairment have generally reported lower yields, ranging from approximately 20% to 45%, depending on the stringency of inclusion criteria, the testing strategy employed, and whether systematic electro‐clinical reassessment was incorporated into the diagnostic workflow. Zacher et al. reported a yield of 34% in 150 adults with intellectual disability and epilepsy using a systematic genetic workup, while Borlot et al. found diagnostic variants in 28% of adults with epilepsy and intellectual disability using multigene panel testing, and Johannesen et al. demonstrated the utility of genetic testing for therapeutic decision‐making in adults with epilepsy. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ Minardi et al., applying whole‐exome sequencing specifically to adults with developmental and epileptic encephalopathy in a retrospective single‐center cohort, achieved a yield of 47%. ¹² Notably, all patients in our cohort underwent first‐tier genetic testing (TRS and CMA), with WES, which is still ongoing, reserved for undiagnosed cases with high clinical suspicion.

The genetic diagnostic yield of 45.1% among all tested patients underscores that comprehensive genetic testing was the primary and indispensable tool for achieving an etiological diagnosis in this cohort. Both targeted resequencing and chromosomal microarray analysis demonstrated robust independent yields (36.8% monogenic and 8.3% chromosomal), reinforcing that both modalities should be considered essential, complementary first‐tier investigations in this patient population, as pathogenic CNVs would be missed by sequencing alone. Crucially, while the execution of these genetic tests was uniform across the cohort, the emphasis on a “phenotype‐guided” approach reflects how the detailed electro‐clinical characterization directly influenced the filtering and downstream clinical interpretation of genetic data, rather than the test selection itself. Importantly, the population enrolled here represents a carefully adjudicated tertiary‐care DEE cohort rather than an unselected sample of adults with epilepsy and intellectual disability, a distinction that likely enriches the proportion of etiologically informative cases. These considerations suggest that the yield reported here should not be interpreted as a universal benchmark for adult epilepsy practice, but rather as an indication of what may be achievable when structured, phenotype‐guided reassessment is applied to a well‐defined adult DEE population in a specialized setting. ⁹ , ¹⁰ , ¹¹ , ¹²

Dissecting the individual contributions of our protocol highlights a critical clinical lesson: while genetic testing understandably yields the highest diagnostic rate in DEEs (45.1% in our cohort), neuroimaging integrated within a multi‐level workflow remains an indispensable, non‐overlapping diagnostic pillar rather than a mere preliminary screening. A diagnostic yield of 18.1% for structural‐metabolic causes in an adult cohort is remarkably informative, considering these patients had undergone numerous scans during their lives. This success underscores that older, low‐field MRI scans or scans degraded by motion artifacts frequently miss subtle structural targets like focal cortical dysplasias or malformations. By combining high‐field (3 T) technology with targeted anesthesiological support for deep sedation, we unlocked definitive structural diagnoses in a substantial subset of these cases, which immediately shifted clinical management, including direct referral for pre‐surgical evaluation.

Moving beyond pure diagnostic clarification, the most transformative finding of our prospective workflow is that etiologic identification led to direct, clinically actionable management changes in 27.1% of the entire enrolled cohort (39/144 patients), representing 42.9% of the 91 solved cases. Achieving a concrete clinical optimization in over a quarter of these complex, historical patients strongly dismantles the traditional therapeutic nihilism surrounding adult neurodevelopmental epilepsy. These management adaptations were highly individualized, spanning from targeted ASM optimization (26 patients)—such as the crucial avoidance of sodium channel blockers in SCN1A‐related cases—to the initiation of etiology‐specific precision therapies (10 patients, including fenfluramine, everolimus, ketogenic diet, and L‐serine), and direct referral for structural pre‐surgical evaluation (3 patients).

Furthermore, a definitive molecular definition serves as an indispensable prerequisite for patient inclusion in international collaborative registries and upcoming clinical trials evaluating next‐generation pipeline therapies, such as antisense oligonucleotides (ASOs) or gene‐selective modulators. In our cohort, achieving genetic clarification has already allowed us to actively screen and prioritize candidates with specific monogenic profiles (e.g., SCN1A and SCN8A variants) for potential enrollment in ongoing translational studies. In an era transitioning toward precision neurology, keeping adult DEE patients etiologically unresolved systematically excludes them from accessing novel orphan drugs and compassionate‐use protocols. This 27.1% therapeutic translation demonstrates that resolving diagnostic uncertainty in adulthood is not an unessential retrospective exercise; rather, it is a highly impactful clinical intervention capable of correcting lifelong diagnostic inertia and unlocking a modern gateway to disease‐modifying care plans.

While the prospective application of a structured and integrated adult‐oriented workflow is a key methodological asset of this study, the core contribution of this work lies in demonstrating the high diagnostic yield (63.2%) achievable through detailed reassessment, alongside its immediate, actionable implications for clinical management, treatment optimization, and genetic counseling. In a population that remains insufficiently characterized in routine adult neurology practice, these findings shift the focus from a purely descriptive workflow to a highly impactful clinical intervention. Adults with suspected DEE are frequently referred with fragmented histories, incomplete pediatric documentation, outdated genetic testing, and neuroimaging or EEG interpretations that were never re‐evaluated in light of current syndromic frameworks. ³ , ⁴ , ⁶ , ⁷ In this context, our findings suggest that a systematic approach based on careful clinical reconstruction, expert re‐review of prior EEG and MRI data, targeted video‐EEG reassessment, and phenotype‐guided etiologic testing can still yield clinically meaningful diagnoses long after the pediatric period.

Moreover, a major challenge in adult cohorts is the distinction between DEE and the broader category of epilepsy associated with intellectual disability or developmental impairment. For this reason, inclusion in the present study required not only drug‐resistant epilepsy and neurodevelopmental impairment, but also evidence that epileptic seizures and/or epileptiform activity contributed to developmental slowing, regression, or persistent impairment beyond the underlying disorder, in line with current ILAE concepts. ¹ , ² In practice, this distinction was based on the integration of developmental history, electro‐clinical data, neuropsychological assessment, and multidisciplinary review. Nevertheless, this remains one of the most difficult aspects of adult DEE research. Childhood developmental trajectories are often reconstructed retrospectively, historical EEGs may be incomplete, and classical pediatric electro‐clinical signatures may become attenuated over time.

The etiologic spectrum observed in our cohort further supports the importance of maintaining a broad diagnostic perspective in adults with suspected DEE. The predominance of genetic diagnoses (45.1% of the total cohort) reinforces the continuing relevance of genetic testing beyond childhood, especially in adults who were never investigated with contemporary sequencing approaches or whose previous results require reinterpretation according to current standards. ⁸ , ⁹ , ¹⁰ At the same time, the identification of structural and metabolic causes in 18.1% of patients argues against an exclusively genetics‐centered model and supports the need for parallel neuroradiological and metabolic assessment when clinically indicated. In the chromosomal subgroup, in particular, pathogenic copy number variants were interpreted as etiologic at the level of genomic disorder, even when a single causative gene could not be confidently isolated. This point is important, because in adults with DEE, clinically meaningful diagnoses may arise from both single‐gene defects and larger genomic rearrangements with composite phenotypic effects.

The association between etiologic category and electro‐clinical phenotype is one of the most clinically relevant observations of the present study, although it should be interpreted cautiously. LGS was significantly more common in the structural‐metabolic and chromosomal groups than in the monogenic group (69.2% and 58.3% vs. 9.4%, respectively; p < 0.0001), whereas when excluding the LGS cohort to avoid syndromic overlap, absence seizures were restricted to monogenic cases (13.2%; p = 0.024). These findings support the concept that, even in adulthood, electro‐clinical characterization can function as a useful diagnostic guide. ¹² In fact, the identification of an LGS phenotype in an adult should prompt careful re‐evaluation for structural abnormalities and pathogenic chromosomal rearrangements, while the presence of absence seizures may increase suspicion for monogenic disorders affecting ion channels, synaptic transmission, or thalamocortical network function.

Expanding upon this electro‐clinical signal, a detailed retrospective review of our monogenic cohort revealed that a significant proportion of patients with genetic absence seizures also experienced prolonged absence status epilepticus (ASE) during their clinical course. Notably, this phenomenon was strictly associated with specific genetic subcategories and clustered around distinct chronological windows, while demonstrating a tendency for lifelong persistence. In patients harboring CACNA1A variants, recurrent and long‐lasting ASE—characterized by fluctuating stupor and subtle perioral or limb myoclonus—predominantly onset during late childhood and adolescence (between 11 and 16 years of age) but crucially persisted as a recurrent phenotypic trait well into adulthood. ¹⁵ Similarly, in our patient with a NEXMIF deletion, severe and prolonged ASE accompanied by continuous generalized spike–wave discharges and transient behavioral regression appeared much earlier, during early childhood (around 6 to 8 years of age); however, this non‐convulsive susceptibility was not outgrown and remained a prominent clinical feature throughout adulthood, further reinforcing the key role of this X‐linked synaptic regulator in sustaining lifelong early‐onset non‐convulsive configurations. ²⁰ This finding underscores that in adult DEE care, a history of non‐convulsive status epilepticus or prolonged ASE should not be overlooked as generic diagnostic noise; rather, when framed within specific age‐at‐onset windows and considering their potential for lifelong recurrence, they represent a core heuristic fingerprint that can highly prioritize targeted genetic screening.

Moreover, another crucial element is the prevalence of focal non‐motor seizures, a semiology highly susceptible to underreporting in adult DEE. Due to cognitive and communication barriers, these patients cannot self‐report subjective auras, forcing reliance on third‐party observations. Caregivers frequently misinterpret brief staring spells or transient autonomic shifts as baseline behavioral traits or apathy rather than epileptic events. While our prospective workflow mitigated this bias through expert‐led reviews of longitudinal video‐EEG and structured interviews, the true burden of focal non‐motor seizures likely remains underestimated. Recognizing this blind spot is vital: Unrecognized non‐motor events can cause subclinical fluctuations that impair baseline cognition and lead to inappropriate medication titration, warranting a low threshold for prolonged video‐EEG monitoring.

Beyond etiology, long‐term Video‐EEG refined the syndromic characterization in over a third of cases, clarifying semiology by resolving diagnostic ambiguities—such as differentiating generalized absence status from prolonged focal events—and capturing subtle myoclonic or tonic features essential for formal reclassification (e.g., LGS). Crucially, Video‐EEG identified co‐existing functional (non‐epileptic) seizures in three DEE patients, documenting atypical motor behaviors without ictal EEG correlates. This dual‐diagnosis identification directly prevented inappropriate anti‐seizure medication escalation and guided tailored care plans. These findings underscore that Video‐EEG is a major driver of clinical optimization in adult DEE, essential for overcoming diagnostic inertia and identifying functional comorbidities.

Phenotypic overlap remained substantial across groups, subgroup sizes were modest, and adult seizure phenotypes likely reflect not only the original etiology but also the cumulative effects of long‐standing epileptic activity on distributed neural networks. Therefore, we interpret these electro‐clinical patterns as heuristic signals that help prioritize testing, rather than as direct surrogates of etiology. The proportion of unresolved cases also remains highly informative. Despite a comprehensive workup, 53 patients (36.8%) remained without a confirmed etiology. This residual group likely reflects several factors, including pathogenic mechanisms not detectable by the methods used here, such as non‐coding variants, mosaic variants, repeat expansions, complex structural variation, or etiologies that are no longer identifiable because of incomplete historical data or subtle acquired lesions. ¹¹ It is also possible that some adult phenotypes classified within the DEE spectrum represent particularly complex phenocopies in whom the relative contribution of etiology and epileptic activity cannot be fully disentangled retrospectively. These unresolved cases highlight both the current limits of adult DEE diagnostics and the need for iterative re‐analysis over time as technologies and disease‐gene knowledge continue to evolve.

The study is limited by its single‐center tertiary setting, possible referral and selection bias, the absence of a comparator pathway, and the intrinsic difficulty of retrospectively distinguishing adult DEE from epilepsy with coexisting developmental impairment. In addition, some subgroup analyses were based on small numbers and should be considered exploratory.

Future multicenter studies should assess the reproducibility of this approach across different care settings, define the incremental contribution of individual diagnostic steps, and determine how etiological clarification affects longitudinal treatment choices and clinical outcomes in adult DEE populations.

AUTHOR CONTRIBUTIONS

Giuseppe d'Orsi: conceptualization; investigation; writing—original draft preparation; review and editing. Umberto Costantino, Carmela Pia Ferro, Maria Teresa Di Claudio, Mario Benvenuto, Pietro Palumbo, Leonardo D'Agruma: investigation. Massimiliano Copetti: statistical analysis. Marco Castori: investigation; review and editing. Massimo Carella: investigation; review and editing. Orazio Palumbo: conceptualization; investigation; writing—original draft preparation; supervision.

FUNDING INFORMATION

This research was funded by the European Union—Next‐Generation EU‐PNRR M6/C2‐Investment 2.1, Enhancement and strengthening of biomedical research in the NHS, CUP I23C24000250007 (Project no. PNRR‐MR1‐2023‐12 377 843) to Massimo Carella, and by the Italian Ministry of Health Ricerca Finalizzata 2021 (Project no. RF_2021_12373524) to Marco Castori.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflicts of interest.

ETHICS STATEMENT

This study was approved by the Casa Sollievo della Sofferenza Hospital Ethics Committee (protocol no. 90/CE/2024). Written informed consent was obtained from all patients or, when applicable, from their legal guardians. All relevant data generated or analyzed during this study are included in this article and its Supporting Information. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Supporting information

Data S1. Supporting Information.