Clinical Spectrum and Treatment Outcomes in Korean Pediatric Patients with CHD2-Related Disorders: Limited Genotype–Phenotype Correlation

Article information

Ann Child Neurol. 2026;34(2):126-135

Publication date (electronic) : 2026 March 31

doi : https://doi.org/10.26815/acn.2025.01249

You Min Kang, MD^,*, Se Hee Kim, PhD, Joon Soo Lee, PhD, Ara Ko, PhD, Hoon-Chul Kang, PhD

Department of Pediatric Neurology, Severance Children’s Hospital, Yonsei University College of Medicine, Seoul, Korea

Corresponding authors: Ara Ko, PhD Department of Pediatric Neurology, Severance Children’s Hospital, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea Tel: +82-2-2228-1004 E-mail: arako@yuhs.ac

Hoon-Chul Kang, PhD Department of Pediatric Neurology, Severance Children’s Hospital, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, Korea Tel: +82-2-2228-1004 Fax +82-2-393-9118 E-mail: hipo0207@yuhs.ac

*Current affiliation: Department of Pediatrics, Chonnam National University Children's Hospital, Chonnam National University Medical School, Gwangju, Korea

Received 2025 November 24; Revised 2026 January 18; Accepted 2026 February 5.

Abstract

Introduction

Chromatin remodeling is a functional process that regulates gene expression by altering chromatin structure and, consequently, RNA transcription [1]. Chromatin remodeling protein complexes use ATPase activity to modify histone–DNA interactions; by influencing the accessibility of RNA polymerase and transcription factors, they ultimately regulate gene expression [2]. The chromodomain helicase DNA-binding (CHD) family is one such group of chromatin remodelers and comprises nine members. Each member differs slightly in the type and arrangement of domains and subunits [2,3]. The CHD2 gene (Online Mendelian Inheritance in Man [OMIM]: 602119), which encodes a member of the CHD protein family, is located on chromosome 15q26.1, contains 39 exons, and comprises four functional domains [4].

In 2008, a case report described a patient with short stature, developmental delay, and mild dysmorphic features who carried a balanced translocation, t(15;22)(q26.1;q11.2) [5]. Subsequent case reports and studies of genomic alterations involving 15q26.1 identified CHD2 within this region, suggesting a potential association between CHD2 and the observed clinical features; these conditions were later collectively referred to as CHD2-related disorders [6,7].

CHD2-related disorders have been associated with neurologic phenotypes in humans, including seizures, developmental delay, intellectual disability, and neuropsychiatric features such as autism spectrum disorder (ASD) [8-10]. More recent studies suggest that pathogenic variants in CHD2 may contribute to developmental epileptic encephalopathy (DEE) [11-15]. Despite increasing recognition of CHD2 as a causative gene for DEE, its clinical and genetic characteristics remain insufficiently understood [16,17] and have not been systematically investigated in Korean patients.

In this study, we aimed to delineate the clinical and genetic spectrum of CHD2-related disorders in Korean pediatric patients and to explore potential genotype–phenotype relationships and population-specific characteristics that may contribute to the understanding and diagnosis of these rare neurodevelopmental disorders. 

Materials and Methods

We screened patients who visited the pediatric neurology center of Severance Hospital, a tertiary medical institution in South Korea, and performed a customized gene panel study containing 172 genes related to epilepsy and DEE from March 2015 to December 2022. Twenty-two patients with pathogenic/likely pathogenic variants in the CHD2 gene were identified. Three patients with insufficient clinical data were excluded; thus, 19 participants were ultimately enrolled in this study.

Demographic, clinical, and radiologic characteristics were collected through a retrospective review of medical records. Seizure and epilepsy classifications were determined according to the 2024 guidelines of the International League Against Epilepsy. Seizure freedom was defined as the absence of clinical seizures for more than 1 year at the most recent follow-up. Given the retrospective design and heterogeneous follow-up durations, outcome assessment was descriptive and not adjusted for follow-up length. Psychiatric comorbidities (e.g., ASD and other behavioral disorders) were determined based on documented clinical diagnoses by pediatric neurologists/psychiatrists in the medical records; when available, diagnoses were supported by standardized assessments and Diagnostic and Statistical Manual of Mental Disorders-based clinical criteria.

Genomic data were extracted from leukocytes in venous blood samples using the QIAamp Blood DNA Mini Kit (Qiagen, Hilden, Germany). Sequencing data were analyzed using Sequencher version 5.3 software (Gene Codes Corp., Ann Arbor, MI, USA), and large-scale deletions and duplications were detected using the Multiplex Ligation-dependent Probe Amplification (MLPA) kit (MRC Holland, Amsterdam, The Netherlands). Targeted gene panel sequencing, alignment, variant calling, and interpretation pipeline were performed as previously described [13]. Genetic testing for patients’ family members was also performed in the same manner, if needed. Pathogenic or likely pathogenic variants were determined according to the American College of Medical Genetics and Genomics guidelines [18].

This study was approved by the Institutional Review Board of Yonsei University Health System Severance Hospital (4-2020-0331), and written informed consent for genetic testing was obtained from the parents or guardians of all participants. 

Results

1. Demographics and clinical manifestations

An overview of the patients’ demographic and clinical features is provided in Table 1. Of the 19 patients, 11 (57.9%) were male. Four patients (21.1%) had documented perinatal events, including three premature births and one case of congenital hypothyroidism. All patients manifested seizures, with a median age at onset of 2.33 years (range, 0.25 to 8.00). Six patients (31.6%) had a family history of epilepsy or seizures based on medical record review, and 10 (52.6%) had a history of fever-provoked seizures before the diagnosis of epilepsy. Developmental delay involving at least one domain was observed in 17 of 19 patients (89.5%). Language delay was the most frequent feature (17/19, 89.5%), followed by intellectual disability (15/19, 82.0%) and motor delay (11/19, 57.9%). A diagnosis of global developmental delay (GDD) was established in 17 patients (17/19, 89.5%). Comorbid psychiatric disorders were present in nine patients (47.4%); attention-deficit/hyperactivity disorder (ADHD) was the most common (7/19, 36.8%), followed by ASD (1/19, 5.3%) and behavioral disorder with impaired impulse control (1/19, 5.3%).

Table 1.

Patient characteristics

3. Characteristics of epilepsy and electroencephalogram findings

As shown in Table 2, seizure types were diverse. Generalized-onset seizures were the most common, occurring in eight of 19 patients (42.1%), followed by focal-onset seizures (6/19, 31.6%) and both focal- and generalized-onset seizures (5/19, 26.3%).

Table 2.

Epilepsy characteristics

When classified by semiology, generalized tonic or generalized tonic-clonic seizures were observed in 14 patients (73.6%). Other seizure types included myoclonic seizures in seven patients (36.8%), myoclonic-atonic seizures in five (26.3%), eyelid myoclonia with absences in three (15.8%), focal impaired-awareness seizures in two patients (10.5%), and absence seizures and focal motor seizures in one patient each (5.3%).

Based on semiology, clinical features, electroencephalogram (EEG) patterns, and MRI findings, six patients were diagnosed with specific epilepsy syndromes: four with epilepsy with myoclonic-atonic seizures (EMAtS) and one each with infantile epileptic spasm syndrome (IESS) and Lennox–Gastaut syndrome (LGS). The patient initially diagnosed with IESS and one patient with EMAtS eventually progressed to LGS.

EEG was performed in all patients. At the initial EEG evaluation, abnormal findings were present in all but three patients. Generalized spike-wave or polyspike-wave discharges were the most frequent findings, observed in nine patients (47.4%), followed by multifocal sharp waves (7/19, 36.8%). At the most recent follow-up, six patients demonstrated normal EEG findings. One of these (patient 14) had normal findings at the initial evaluation, while among the remaining five patients (patients 3, 5, 9, 13, and 16) whose EEGs normalized over time, three were among the four individuals diagnosed with EMAtS.

5. Genetic analysis

Eighteen pathogenic/likely pathogenic variants were identified in 19 individuals, including eight novel variants. These comprised nine nonsense variants (50.0%), five splice-site variants (27.8%), and two missense variants (11.1%). Two individuals carried large deletions involving one or more exons. Notably, one recurrent truncating variant was identified in two unrelated patients. Despite harboring the same variant, these two patients (patients 1 and 11) exhibited distinctly different clinical features, including differences in age at seizure onset, seizure semiology, and EEG and MRI findings. Inheritance analysis revealed that six patients had de novo variants, while one patient (patient 7) inherited an exon 5 deletion from her asymptomatic father. Detailed clinical, electroencephalographic, neuroimaging, and genetic information for each patient, as outlined above, is provided in Table 3.

Table 3.

Clinical and genetic characteristics of patients with CHD2 variants

Discussion

In this study, seizures were the most common clinical manifestation, observed in all patients. Seizures were accompanied by heterogeneous seizure types and multiple epilepsy syndromes, followed by GDD (17/19, 89.5%). Overall, the clinical spectrum observed in this study aligns closely with previously reported CHD2-related disorders, particularly with respect to developmental delay and phenotypic heterogeneity. The largely comparable distribution of seizure types and epilepsy syndromes across studies supports the consistency of CHD2-associated epilepsy phenotypes despite differences in cohort size and study design [10,17,19].

Fourteen patients experienced more than one seizure type, and seizures were well controlled in approximately two-thirds of the cohort. Among the six patients with intractable seizures, four (66.7%) had abnormal brain MRI findings. Both patients with cerebellar atrophy on MRI (patients 10 and 14) belonged to the intractable epilepsy group; cerebellar atrophy has been previously described in CHD2-related disorders [17], although its relationship to seizure severity has not been established. In contrast, other structural abnormalities did not show a consistent association with treatment response. Two patients exhibited pituitary hypoplasia (patients 4 and 7), yet only one progressed to LGS while the other achieved good seizure control. Additionally, one patient (patient 15) demonstrated focal seizures in the context of left hippocampal signal abnormality and incomplete rotation, but her seizures remained well controlled. Although cerebellar or posterior-predominant atrophy has been reported in a subset of patients with CHD2-related disorders, including cases with longitudinal progression, prior studies likewise have not demonstrated a consistent or predictive association between MRI abnormalities and epilepsy severity or treatment response [15,20]. Taken together, while cerebellar atrophy was observed only in intractable cases in our cohort, there is currently no evidence supporting a causal or predictive relationship between MRI abnormalities and seizure refractoriness in CHD2-related DEE, underscoring the need for further investigation in larger cohorts.

Psychiatric symptoms were observed in nine patients in our study, with ADHD being the most frequent (7/19, 36.8%), followed by ASD (1/19, 5.3%). While CHD2 has historically been implicated in ASD and was initially highlighted in genetic analyses of individuals with ASD [21], our findings demonstrate a higher prevalence of ADHD than ASD among patients with CHD2-related disorders. Previous studies have also reported ADHD or attentional and behavioral dysregulation in individuals with CHD2 variants, suggesting that the neuropsychiatric impact of this gene extends beyond its original association with autism [9,17,22]. CHD2 is an adenosine triphosphate-dependent chromatin remodeler critical for cortical circuit organization and gamma-aminobutyric acid (GABA)-ergic interneuron development, and it regulates activity-dependent transcriptional programs and neuronal differentiation [22,23]. Furthermore, epigenetic dysregulation has increasingly been implicated in cortical network dysfunction and ADHD-like behavioral phenotypes [24]. Consequently, it is biologically plausible that CHD2 haploinsufficiency contributes to attentional deficits and ADHD-like behavioral manifestations. These findings suggest that the phenotypic spectrum of CHD2-related disorders is broader than previously recognized and includes attentional and behavioral domains, indicating the need to clarify the contribution of this gene to ADHD.

In terms of genetic variation, truncating variants accounted for the majority (84.2%), and mutations were distributed across a wide range of gene regions, without clustering in a specific domain or showing associations with particular phenotypes. This pattern supports CHD2 haploinsufficiency as the principal pathogenic mechanism and suggests that no clear mutational hotspot has been identified, consistent with previous reports [8]. Given this lack of regional clustering and the predominance of loss-of-function variants, genotype–phenotype correlations in CHD2-related disorders are expected to be limited. Nevertheless, our cohort revealed three notable observations that further illustrate the extent and nature of phenotypic variability associated with CHD2 haploinsufficiency.

First, two unrelated male patients carrying the same truncating CHD2 variant exhibited markedly different clinical trajectories despite sharing an identical predicted loss-of-function mechanism. Both patients ultimately developed epilepsy with GDD and intellectual disability, but differed substantially in age at seizure onset, seizure semiology, developmental profiles, and EEG/MRI findings. One patient (patient 1) had later-onset seizures, pre-existing motor and speech delay with ADHD, focal interictal epileptiform discharges on EEG, and structural MRI abnormalities, including asymmetric lateral ventricles (left > right) and a left middle cranial fossa arachnoid cyst. In contrast, the other (patient 11) had earlier-onset seizures, isolated language delay before epilepsy onset, subsequent developmental regression, focal slowing on EEG, and normal MRI findings. The loss of CHD2 function, although a well-established molecular mechanism, does not fully explain the considerable phenotypic divergence of these individuals. The variability in clinical presentation is therefore likely influenced by additional modifying factors, including interindividual genetic background, differences in epigenetic regulation, and environmental influences, each of which may contribute to differences in seizure susceptibility and neurodevelopmental outcomes [4,10,23-25].

A second example of variable expressivity was observed in a patient (patient 7) with an inherited exon 5 deletion. Although the daughter developed multiple generalized seizure types, GDD, and multifocal epileptiform discharges, her father—who carried the same variant—had only a few febrile seizures in childhood and no persistent neurologic or developmental abnormalities. This marked intrafamilial divergence highlights incomplete penetrance and variable expressivity in CHD2-related disorders. Exon-level deletions involving coding regions of CHD2 are predicted to result in loss-of-function and are therefore consistent with a pathogenic interpretation, given the established haploinsufficiency mechanism of CHD2. Supporting this mechanism, Chenier et al. [9] reported de novo deletions affecting CHD2 exonic sequences, with deletions confirmed and parental follow-up performed using orthogonal methods such as multiplex ligation-dependent probe amplification, array comparative genomic hybridization, and/or fluorescence in situ hybridization. In our study, interpretation of the inherited exon 5 deletion is limited by the retrospective nature of paternal phenotyping, as well as by the inability to definitively confirm or exclude somatic or germline mosaicism using routine clinical assays; accordingly, segregation-based inference regarding penetrance is constrained. Independent of these interpretive limitations, additional confirmatory analyses were performed to exclude the possibility that the identified CHD2 exon 5 deletion represented a technical artifact. Validation using long-range polymerase chain reaction verified the presence of the exon-level deletion, and concordant exon 5 copy number loss was consistently observed in both the proband and the father (Supplementary Figs. 1 and 2), supporting the authenticity of this structural variant. In parallel, a comprehensive review of all variants detected by the targeted epilepsy gene panel revealed no additional pathogenic or likely pathogenic variants that could plausibly account for the proband’s clinical phenotype. Several variants of uncertain significance, including those in alanyl-tRNA synthetase (AARS), seizure threshold 2 (SZT2), and carnitine palmitoyltransferase 1B (CPT1B), were identified; however, these were considered unlikely contributors based on inheritance pattern, known disease mechanisms, population frequency, and lack of clinical correlation. Taken together, these additional evaluations support the CHD2 exon 5 deletion as the most relevant genetic finding in this patient. The marked phenotypic discordance within this family may be compatible with a two-hit model, in which CHD2 haploinsufficiency establishes a baseline vulnerability, while additional perinatal or developmental factors—such as the patient’s preterm birth and prior germinal matrix hemorrhage—act as secondary hits that may amplify epileptogenesis and neurodevelopmental impairment [17,26,27].

Further evidence of phenotypic variability was observed in a patient (patient 17) with a large deletion involving exons 17–29 who exhibited only mild clinical features, including well-controlled seizures, moderate intellectual disability, ADHD, normal MRI findings, and no interictal epileptiform discharges on EEG. The relatively mild phenotype despite extensive genomic loss suggests that deletion size alone does not determine clinical severity and that the pathogenic impact of CHD2 variants may differ across functional domains. Together with the absence of a mutational hotspot in our study, these findings indicate that domain-specific contributions to CHD2 function remain incompletely understood, highlighting the need for future studies examining regional functional effects and modifier interactions that shape phenotypic outcomes.

Collectively, our findings reinforce that CHD2-related disorders arise predominantly from haploinsufficiency, but that clinical expression is strongly modulated by additional factors, resulting in substantial variability in severity, seizure type, developmental impact, and neurobehavioral outcomes. Future studies delineating the structural and regulatory functions of each CHD2 domain, as well as identifying genetic or environmental modifiers, will be essential to understanding the mechanisms underlying this heterogeneity.

This study has several limitations. First, as a single-center retrospective study with a relatively small sample size, the generalizability of our findings is limited. Second, the cohort was ascertained through a targeted epilepsy/DEE gene panel workflow at a tertiary pediatric neurology center, which may have introduced selection bias toward individuals with pediatric-onset and epilepsy-predominant phenotypes. Therefore, milder CHD2-related presentations (e.g., minimal epilepsy, preserved cognition) or adult-onset phenotypes are likely underrepresented, and our findings should be interpreted as reflecting the clinical spectrum of panel-tested pediatric patients rather than the full spectrum of CHD2-related disorders.

Epilepsy was the consistent and dominant clinical manifestation in our cohort of patients with CHD2-related disorders, accompanied by diverse developmental, cognitive, and neurobehavioral features. The discordance between genotype and phenotype observed across identical variants, inherited variants with divergent presentations, and large deletions with a mild phenotype reflects the substantial heterogeneity characteristic of CHD2-related disorders.

Accordingly, although CHD2-related disorders appear to be primarily driven by haploinsufficiency, clinical manifestations—including seizure type and severity, developmental trajectories, and neurobehavioral profiles—show considerable variability and cannot be explained by genotype alone. Comprehensive studies examining genotype–phenotype correlations, along with investigations into developmental, genetic, and environmental modifiers, will be essential to elucidate the mechanisms underlying this heterogeneity and to increase diagnostic precision and clinical prognostic assessment.

Supplementary material

Article information Continued