CHD2-Related Developmental and Epileptic Encephalopathy — Comprehensive Research Report

Prepared: 2026-07-22 · Target KB: dismech kb/disorders/ · Category: Mendelian

Evidence provenance and verification status

Every PMID below is tagged with its verification status in this session. Only ✅ VERIFIED identifiers were confirmed against a live PubMed/publisher record during this research pass. Items marked ⚠️ UNVERIFIED PMID have a confirmed title/journal/year/DOI but the numeric PMID was not independently resolved — run just fetch-reference on these before using them as evidence, per the project's anti-hallucination SOP. Suggested ontology terms are candidates requiring OAK verification (just validate-terms-file); HPO in particular has renamed several seizure-semiology terms in recent releases, so labels must be checked rather than trusted.

1. Disease Information

Overview

CHD2-related developmental and epileptic encephalopathy (CHD2-DEE) is an autosomal dominant, almost always de novo, neurodevelopmental disorder caused by heterozygous loss-of-function variants in CHD2, which encodes chromodomain helicase DNA-binding protein 2 — an ATP-dependent chromatin remodeler. The disorder sits at the intersection of three clinical worlds: it is a developmental and epileptic encephalopathy, an intellectual disability / autism syndrome, and a photosensitive generalized epilepsy. Its most distinctive clinical signature is an unusually intense, sometimes self-induced, sensitivity to flickering light superimposed on a myoclonic-predominant, drug-resistant generalized epilepsy.

The canonical presentation is a child with normal or mildly delayed early development who, typically in the second or third year of life, develops an explosive onset of multiple daily myoclonic and absence seizures, followed by developmental plateau or frank regression and long-term intellectual disability. The phenotypic range is however much broader than the original DEE description, extending from mild intellectual disability with well-controlled epilepsy to, at the far end, adult-onset non-syndromic pharmacoresistant epilepsy.

Importantly, the disorder is now best conceptualized not as a single syndrome but as a spectrum of CHD2-related neurodevelopmental disorders, of which DEE94 is the severe pole. GeneReviews explicitly frames the entity this way ("CHD2-Related Neurodevelopmental Disorders"), and ClinGen's Epilepsy Gene Curation Expert Panel curated the gene-disease relationship against the broad entity "complex neurodevelopmental disorder" rather than a narrow DEE label.

Key identifiers

Note on the triplosensitivity score. The ClinGen TS score of 0 predates the 2024 CHASERR discovery. There is now direct human evidence that increased CHD2 dosage is also pathogenic (see §4), so the TS=0 assertion should be treated as historically accurate but biologically superseded. This is a good candidate for a discussions entry with kind: KNOWLEDGE_GAP.

Synonyms and alternative names

• CHD2 myoclonic encephalopathy (MedlinePlus / MONDO synonym)
• Developmental and epileptic encephalopathy 94; DEE94
• Epileptic encephalopathy, childhood-onset (EEOC) — legacy OMIM title
• CHD2-related neurodevelopmental disorders (GeneReviews preferred, broader)
• CHD2-related epilepsy
• CHD2 encephalopathy
• Childhood-onset epileptic encephalopathy with CHD2 mutation
• Myoclonic encephalopathy, CHD2-related

Information provenance

Knowledge of this disorder derives predominantly from aggregated, disease-level resources: multicenter genetic-epilepsy cohorts, systematic literature pooling, and curated databases (ClinVar, ClinGen, OMIM, GeneReviews). The largest single evidence base is a 2025 pooled analysis of 236 individuals assembled from 74 published articles plus the PERC (Pediatric Epilepsy Research Consortium) Genetics registry (PMID:40934838 ✅). There is also a genuine individual-patient/EHR layer: the Coalition to Cure CHD2 has run a digital natural-history study (62 participants, CLIRINX platform), a RARE-X caregiver-reported registry, and a Ciitizen medical-record aggregation project that codes real patient records into HPO terms for computational analysis (PMID:39391213 ✅). For dismech purposes, most evidence items will carry evidence_source: HUMAN_CLINICAL, with a substantial MODEL_ORGANISM and IN_VITRO layer for mechanism.

2. Etiology

Primary causal factor

CHD2-DEE is a monogenic, chromatin-mediated developmental disorder. The proximate cause is heterozygous loss of function of CHD2, producing haploinsufficiency of the CHD2 chromatin remodeler in the developing brain. There is no environmental cause, no infectious contribution, and no established multifactorial component to disease occurrence.

GeneReviews states the mechanism plainly: "CHD2-related neurodevelopmental disorders result from haploinsufficiency of CHD2." The variant-class distribution supports this — "The overwhelming majority of CHD2 pathogenic variants lead to either truncation of the protein or loss of gene expression by whole-gene deletion." A 2018 mechanistic review quantified the same point in an early cohort: "The vast majority (83%, 33/40) of patients carry truncating CHD2 variants, suggesting that the pathogenic mechanism that underpins CHD2-associated epilepsy is haploinsufficiency."

Genetic risk factors

Causal variants. De novo heterozygous LoF variants in CHD2 — nonsense, frameshift, canonical splice-site, and whole-gene/multi-exon deletions — are causal, not merely risk-conferring. In the 2025 pooled cohort, 95% (170/179) of variants with parental testing were confirmed de novo (PMID:40934838 ✅). De Maria et al. found 72.5% of patients carried truncating variants (Am J Med Genet A 2022;188(2):522–533, DOI:10.1002/ajmg.a.62548 ⚠️ UNVERIFIED PMID).

Susceptibility (non-DEE) role. This is a mechanistically important nuance for the KB: CHD2 variation is also a quantitative risk factor for photosensitivity in epilepsy generally, not only a cause of DEE. Galizia et al. compared 580 individuals with photosensitive seizures and/or an abnormal photoparoxysmal EEG response against 34,427 population controls and found "unique CHD2 variation was over-represented in cases overall (P = 2.17 × 10⁻⁵)", with 11 unique variants in cases versus 128 in the much larger control set (PMID:25783594 ✅). Within syndromes, the enrichment concentrated in eyelid myoclonia with absences (EMA): "there was over-representation of unique CHD2 variants (3/36 cases) in the archetypal photosensitive epilepsy syndrome, eyelid myoclonia with absences (P = 3.50 × 10⁻⁴)." This makes CHD2 one of the few genes with an established endophenotype-level genetic contribution to photosensitivity — a strong candidate for a dismech relationship_type: SUSCEPTIBILITY gene annotation alongside the causal one.

Modifier genes. No validated human modifier loci have been reported. However, genetic background is a powerful modifier in mouse, which is mechanistically suggestive: in a Chd2 frameshift-truncation model, "no phenotypes were observed on the pure C57BL/6J background," whereas "crossing these mice onto the 129X1/SvJ genetic background gradually uncovered neurodevelopmental phenotypes" (Mavashov et al., Epilepsia 2026, DOI:10.1002/epi.70073 ⚠️ UNVERIFIED PMID; and bioRxiv 2025.03.18.643778). This is a defensible basis for a HUMAN_MODEL_MISMATCH discussion entry: strain-dependent penetrance in mouse implies unidentified human modifiers, but none are mapped.

The CHASERR locus as a second genetic etiology. See §4 — de novo deletion of the adjacent lncRNA gene CHASERR causes a distinct, more severe disorder by the opposite dosage mechanism.

Environmental risk factors

None are established for disease causation. Advanced paternal age is a general risk factor for de novo mutation across the genome and is a plausible but unquantified-for-CHD2 contributor; do not assert it without a CHD2-specific source.

Environmental precipitants (distinct from risk factors)

This is where CHD2 is unusual and where dismech should be careful to separate "what causes the disease" from "what triggers the seizures":

1. Photic stimulation — the dominant, near-defining trigger. Flickering light, television, sunlight through trees, video screens. GeneReviews: "Clinical photosensitivity (i.e., seizures triggered by photic stimulation) is a distinguishing feature reported in a total of 80% (20/25) of individuals where it was specifically queried." The pooled analysis reports 59% (80/136) (PMID:40934838 ✅), and the adult cohort 64% (PMID:39601014 ✅).
2. Self-induction — a striking behavioral feature where affected children deliberately seek out flicker. Thomas et al.: of 10 patients, "7 exhibited exquisite clinical photosensitivity; 6 self-induced with the television" (PMID:25672921 ✅). This is a behavioral phenotype that is simultaneously a seizure trigger — a genuine feedback loop worth modeling explicitly as a pathophysiology node.
3. Fever — 37% (33/90) fever-sensitive in the pooled cohort (PMID:40934838 ✅). Suls et al. specifically titled the phenotype a "fever-sensitive myoclonic epileptic encephalopathy sharing features with Dravet syndrome" (PMID:24207121 ✅). Note the discrepancy with GeneReviews, which observes that "febrile seizures, which are characteristic of Dravet syndrome, have only been reported in 11 individuals with CHD2 variants" — i.e. fever sensitivity of established seizures is common, but classic febrile seizures as a presenting event are not. Curate these as two distinct claims.

Protective factors

No genetic or environmental protective factors are established. The only actionable "protective" intervention is avoidance of photic triggers — GeneReviews advises to "Avoid flickering lights that may provoke seizures." This is best modeled as a treatment/management action (MAXO-annotated behavioral intervention), not a protective factor in the etiologic sense.

Gene–environment interactions

The single well-supported GxE interaction is CHD2 genotype × photic environment. The interaction is dose-graded rather than binary: complete LoF produces DEE with exquisite photosensitivity, while rarer/milder CHD2 variation raises photosensitivity risk in otherwise typical epilepsies (PMID:25783594 ✅). Critically, the interaction is conserved across species — zebrafish chd2 knockdown "markedly enhanced mild innate zebrafish larval photosensitivity" (PMID:25783594 ✅), providing model-organism corroboration of a human GxE effect. That cross-species conservation is unusually strong evidence for a genuine mechanistic interaction rather than an ascertainment artifact.

A second, weaker interaction is genotype × febrile illness (37% fever sensitivity), analogous to but milder than the SCN1A/Dravet paradigm.

3. Phenotypes

Frequency data — the two principal cohorts

Two independent aggregations give slightly different denominators. Both should be curated, with population distinguishing them.

HPO caution on photosensitivity. I could not verify a specific HPO ID for "photosensitive seizure" or "photoparoxysmal EEG response" in this session, and I will not guess one — HPO has multiple similarly-named terms in the seizure-precipitant and EEG-abnormality branches. Look these up with runoak -i sqlite:obo:hp search "photo" before curating. This is the single highest-risk term in the entry precisely because it is the disease's most characteristic feature.

Seizure semiology (the core phenotype cluster)

Onset. GeneReviews: "Seizure onset is typically between ages six months and four years" with median onset at 30 months. Thomas et al. report a mean of 26 months across 10 patients (PMID:25672921 ✅). The pooled analysis found a far wider true range: "Seizure onset ranged from 1 day to 22 years" (PMID:40934838 ✅), and Chen et al. observed onset from "3 months to 10 years 5 months" across 17 patients (PMID:31677157 ✅).

Onset character. Distinctively abrupt. GeneReviews: onset is "explosive in many children, characterized by multiple daily myoclonic and absence seizures." This "explosive" quality — going from no seizures to many daily seizures over days to weeks — is a useful discriminator in the differential.

Seizure types, in rough order of characteristic-ness:

• Myoclonic seizures — the signature type; present in all 10 of Thomas et al.'s patients (PMID:25672921 ✅). HP:0002123 Generalized myoclonic seizure (verify label).
• Atypical absence / absence seizures — HP:0007270 Atypical absence seizure; HP:0011147 Typical absence seizure (verify).
• Myoclonic-atonic seizures and drop attacks — GeneReviews lists "drop attacks, myoclonus" and a distinctive "atonic-myoclonic-absence" composite seizure. HP:0032794 Myoclonic-atonic seizure (verify).
• Atonic seizures — HP:0010819 Atonic seizure.
• Generalized tonic-clonic / bilateral tonic-clonic — HP:0002069 (label recently changed to "Bilateral tonic-clonic seizure" — verify).
• Eyelid myoclonia with absences — in roughly 10% per the Epilepsiome summary; CHD2 is "the first identified cause of the archetypal generalized photosensitive epilepsy syndrome, eyelid myoclonia with absences" (PMID:25783594 ✅).
• Focal seizures — a minority; De Maria et al. found generalized types accounted for 75.5% of all epilepsies, implying ~25% with focal features (DOI:10.1002/ajmg.a.62548 ⚠️).
• Epileptic spasms — rare; 2/17 in Chen et al. (PMID:31677157 ✅). HP:0011097 Epileptic spasm.
• Status epilepticus — reported, frequency not well quantified. HP:0002133 Status epilepticus.

Multiplicity. "More than half of affected individuals have multiple seizure types, with a predominance of myoclonic-atonic, myoclonic, and absence seizures." Evolution to multiple refractory types over months is characteristic and is part of what qualifies the disorder as an epileptic encephalopathy.

Cognitive, developmental, and behavioral phenotypes

Intellectual disability severity distribution. GeneReviews: "intellectual disability ranges from mild (in 7/15 individuals) to severe (8/15 individuals)" — roughly bimodal, split evenly. The Epilepsiome breakdown gives approximately 30% mild, 25% moderate, 15% severe-profound.

Regression. Developmental regression coinciding with seizure onset is a defining feature of the DEE framing (OMIM: "Affected individuals have cognitive regression and impaired intellectual development"). Pre-seizure development is inconsistently reported — GeneReviews notes psychomotor development prior to seizures "can be delayed…but is often not reported and this feature requires further evaluation." This is an honest, citable knowledge gap worth capturing as a discussions entry: whether CHD2 encephalopathy is truly seizure-driven regression or an underlying developmental trajectory unmasked by seizure onset is unresolved.

Language. Prominent language impairment is repeatedly emphasized, including as a relatively isolated feature in milder presentations (De Maria et al. ⚠️). HP:0000750 Delayed speech and language development; HP:0001344 Absent speech.

Behavioral and psychiatric. GeneReviews: "Challenging behaviors, most often aggression, have been described," plus ADHD, anxiety, and rarely psychosis or schizophrenia. The adult cohort is far more granular and more sobering (PMID:39601014 ✅): behavioral issues in 100%, internalizing features such as anxiety in 71%, self-injurious behaviors in 50%, ASD diagnosed in 71%.

Candidate HPO terms: HP:0000718 Aggressive behavior; HP:0100716 Self-injurious behavior; HP:0000739 Anxiety; HP:0007018 ADHD; HP:0000717 Autism; HP:0002360 Sleep disturbance.

Motor, gait, and systemic phenotypes

The adult study identified a cluster that is under-recognized in pediatric reports: "seizure severity is associated with worse comorbidities such as maladaptive behaviors, gait, gastrointestinal, sleep, and abnormal pain responsiveness" (PMID:39601014 ✅). Only 43% could ambulate independently in adulthood. Abnormal pain responsiveness is a notable and easily missed feature.

Cerebellar signs appear in some series — one imaging study reported inferior vermis hypoplasia (7/10) and mild cerebellar atrophy (4/10), suggesting ataxia may be under-reported.

Dysmorphism and growth

CHD2 haploinsufficiency is characteristically non-dysmorphic — GeneReviews describes a "brain-restricted phenotype." This is diagnostically useful and is the sharpest clinical discriminator from CHASERR deletion, which produces "shared facial dysmorphisms" (NEJM 2024). Absence of dysmorphism is worth curating explicitly as a negative finding.

Progression, severity, and quality-of-life impact

• Progression: Seizures are generally most severe in early childhood, but do not reliably remit — 79% of adults still have ongoing seizures (PMID:39601014 ✅). Some individuals show progressive brain atrophy: "MRI has shown atrophy that tends to be more posterior and can be progressive," with 3 of 4 individuals having sequential imaging showing progressive atrophy.
• A progressive myoclonic epilepsy mimic: CHD2-DEE can phenocopy PME — worsening myoclonus, cognitive decline, and ataxia — and has been formally reported as such (Chityala et al., Epileptic Disorders, DOI:10.1002/epd2.70196 ⚠️ UNVERIFIED PMID). This matters clinically because it can send workups down an expensive PME/storage-disease path.
• Quality of life: No disease-specific validated QoL instrument exists. The Coalition to Cure CHD2 collects HRQoL questionnaires via RARE-X (PMID:39391213 ✅). Caregiver-prioritized burdens, from a 75-participant community poll, ranked in order: seizure control > behavior > regression > intellectual disability > phenotypic severity > medication side effects > speech. That ranking is genuinely useful for a KB — it is patient-derived rather than clinician-assumed.

4. Genetic / Molecular Information

Causal gene

CHD2 (chromodomain helicase DNA-binding protein 2), 15q26.1, hgnc:1917, OMIM *602119, UniProt O14647, canonical transcript NM_001271.4. The protein is ~1,828 aa (~211 kDa).

Constraint

CHD2 is among the most LoF-intolerant genes in the genome: pLI = 1, LOEUF ≈ 0.07–0.17 (gnomAD). This is exactly the constraint profile expected of a haploinsufficient dominant developmental gene and provides strong supporting evidence (ACMG PVS1 applicability) for truncating-variant interpretation.

Variant spectrum

Detection rates by method (GeneReviews): sequence analysis detects 93% (129/139) of pathogenic variants; chromosomal microarray detects 7% (10/139) — i.e. whole-gene and multi-exon deletions are a small but non-trivial slice, meaning a negative gene panel does not fully exclude the diagnosis.

Variant classes: - Truncating (nonsense, frameshift): the majority — 72.5% (De Maria ⚠️), 83% in an earlier 40-patient series. - Splice-site: canonical and, increasingly, deep-intronic (e.g. ClinVar RCV003740922, c.4593-14A>G). - Missense: a minority, and mechanistically informative — they "cluster in the functional domains" and disrupt DNA binding or chromatin remodeling capacity. Missense variants outside functional domains are frequently VUS; an "interdomain" missense variant was associated with the unusual adult-onset pharmacoresistant epilepsy presentation (De Maria ⚠️). - Whole-gene deletions / CNVs: ~7%. - Synonymous variants: numerous benign/likely-benign ClinVar entries exist (e.g. c.3036T>C p.Ser1012=, c.4260A>G p.Ser1420=, c.2098A>C p.Arg700=) — a reminder that ClinVar CHD2 records include many non-pathogenic classifications.

Origin: Overwhelmingly germline de novo (95%, 170/179). Somatic CHD2 mutation occurs in cancers (notably CLL) but is mechanistically and clinically unrelated to the neurodevelopmental disorder; do not conflate them in the KB.

Functional consequence: Loss of function. GeneReviews: "Loss-of-function predominates; no gain-of-function mechanisms reported" for germline disease. No dominant-negative mechanism has been established.

The CHASERR locus — bidirectional dosage sensitivity

This is the most important recent development in CHD2 genetics and deserves a dedicated section in any KB entry.

CHASERR (CHD2 Adjacent, Suppressive Regulatory RNA; OMIM 620993) is a conserved long noncoding RNA gene transcribed immediately upstream of CHD2 that acts as a negative cis-regulator of CHD2 expression. In mouse, "Chaserr inhibits expression of Chd2 in cis and is required for postnatal mouse development"* (Rom et al., Nat Commun 2019, DOI:10.1038/s41467-019-13075-8 ⚠️ UNVERIFIED PMID).

In 2024, three unrelated children were reported with de novo deletions of the CHASERR promoter and first three exons that spared CHD2 and its promoter entirely. They had "severe encephalopathy, shared facial dysmorphisms, cortical atrophy, and cerebral hypomyelination — a phenotype that is distinct from the phenotypes of patients with CHD2 haploinsufficiency." Mechanistically, "The CHASERR deletion results in increased CHD2 protein abundance in patient-derived cell lines and increased expression of the CHD2 transcript in cis" (NEJM 2024;DOI:10.1056/NEJMoa2400718; the medRxiv preprint is PMID:38496558 ✅ — the NEJM PMID was not resolved ⚠️).

The authors' conclusion is the headline: "these findings indicate that CHD2 has bidirectional dosage sensitivity in human disease." This is also the first demonstration that haploinsufficiency of a lncRNA causes a Mendelian disease. OMIM has assigned the resulting phenotype its own entry: #621012, NEDFSAB (neurodevelopmental disorder with dysmorphic facies, absent speech and ambulation, and brain abnormalities).

KB implication: CHD2 too little → DEE94, non-dysmorphic, brain-restricted. CHD2 too much (via CHASERR loss) → NEDFSAB, dysmorphic, hypomyelinating, more severe. These are two distinct dismech entries linked by an inverse-dosage relationship — an excellent candidate for a Grouping with grouping_basis: SHARED_GENE_FAMILY / SHARED_MECHANISM, or for a mechanistic hypothesis capturing the dosage-window model.

Epigenetic information

CHD2 is an epigenetic regulator, and its loss leaves a measurable epigenomic fingerprint. A validated DNA methylation episignature exists for CHD2 haploinsufficiency: "CHD2 haploinsufficiency is one of several genetic conditions with a distinct episignature" (PMID:39391213 ✅). GeneReviews confirms clinical utility — DNA methylation episignatures "can help resolve variants of uncertain significance in CHD2." A refinement of the CHD2 episignature specifically in genetically unsolved DEEs has been published (medRxiv 2023.10.11.23296741).

This is functionally significant: the episignature is simultaneously (a) a diagnostic test, (b) a functional-evidence source for ACMG variant classification, and (c) "a strong biomarker candidate" for future clinical trials (PMID:39391213 ✅).

Chromosomal abnormalities

15q26.1 deletions encompassing CHD2 — ranging from intragenic multi-exon deletions to larger contiguous-gene deletions — account for ~7% of cases and are detected by CMA. Larger 15q26 deletions may add non-neurological features from neighboring genes.

5. Environmental Information

• Environmental factors: None causally implicated. No toxin, radiation, pollutant, or occupational exposure is associated with CHD2 de novo mutation beyond generic de novo mutational processes.
• Lifestyle factors: Not applicable to causation. Relevant only to seizure management (sleep deprivation and screen/flicker exposure as generic and CHD2-specific precipitants respectively).
• Infectious agents: None. Febrile illness acts as a non-specific seizure precipitant (37% fever sensitivity), not as an etiologic agent — any pathogen producing fever can do this. Curate as a precipitant, never as an infectious cause.

The one environmental exposure that genuinely matters is photic: television, video screens, sunlight flicker, strobe lighting. It is a seizure trigger of unusual potency in this disorder and the target of the only broadly applicable non-pharmacologic intervention.

6. Mechanism / Pathophysiology

Protein architecture

CHD2 contains, N→C: tandem chromodomains, an SNF2-family ATPase/helicase domain, and a C-terminal DNA-binding domain. Notably, "The N-terminal chromodomains serve an autoinhibitory function, while deletion of this region increases both DNA-binding and ATPase activities." Additionally, "The C-terminus of CHD2 also associates with a poly ADP-ribose (PAR) binding domain that is involved in DNA damage repair" — implying a secondary genome-maintenance role distinct from its transcriptional one.

Core molecular function

CHD2 is an ATP-dependent chromatin remodeler that "use[s] the energy from ATP hydrolysis to remodel chromatin into periodic nucleosome arrays." Its most mechanistically specific activity is histone variant deposition: "CHD2 interacts with H3.3, a histone variant incorporated into the nucleosome at transcriptionally active genes," and "the chromodomain of CHD2 facilitates H3.3 incorporation into the nucleosome, poising genes necessary for differentiation for expression."

This gives a clean, causally-legible mechanism: CHD2 pre-loads differentiation genes with H3.3-containing nucleosomes so they are poised to fire at the right developmental moment. Halve the CHD2 and you do not abolish those genes — you mistime them. That framing explains why the disorder is developmental-window-dependent rather than a simple constitutive deficiency.

Recruitment and interaction partners

CHD2 does not choose its own targets; it is recruited by cell-type-specific transcription factors: - NKX2-1 in interneuron lineages — "CHD2 ChIP-qPCR revealed an overlap with NKX2-1 binding at three candidate genes important for interneuron development." CHD2 is itself an NKX2-1 direct target, and the two "could coregulate cIN gene expression by cobinding shared genomic regulatory regions" (Meganathan et al., PNAS 2017, DOI:10.1073/pnas.1712365115 ⚠️; and Sci Rep 2022, PMID:36115870 ✅). - REST (RE1-silencing transcription factor) in progenitors — "candidate ChIP-seq revealed CHD2 binding at REST," and loss of CHD2 reduces REST expression. Since REST is the master repressor keeping neuronal genes off in non-neuronal and progenitor cells, this is a plausible upstream node for premature neuronal differentiation.

The causal chain (upstream → downstream)

Node 1 — MOLECULAR: CHD2 haploinsufficiency. De novo truncating variant / whole-gene deletion → ~50% reduction in functional CHD2 protein. biological_scale: MOLECULAR.

Node 2 — MOLECULAR: Impaired H3.3 deposition and defective chromatin remodeling at differentiation loci. Reduced ATP-dependent nucleosome remodeling; loss of poising at developmental genes. GO:0043044 ATP-dependent chromatin remodeling; GO:0043486 histone exchange (verify).

Node 3 — CELLULAR: Loss of neural progenitor self-renewal / premature neuronal differentiation. "Chd2 is predominantly expressed in Pax6+ radial glia in the VZ/SVZ from E12–E18"; "Chd2 knockdown promotes premature neuronal differentiation during embryonic mouse cortical development due to a decrease in Pax6+ neural progenitor cells and an increase in Tbr2+ intermediate progenitor cells." The interpretation: "Chd2 deficiency suppresses the self-renewal capacity of the radial glia and instead promotes premature neuronal differentiation." Cell types: CL:0000681 radial glial cell; CL:0000031 neuroblast.

Node 4 — CELLULAR: Deficient cortical GABAergic interneuron generation and maturation. This is likely the key seizure-relevant node. "CHD2 gene expression levels gradually increase during the differentiation of human embryonic stem cells (hESCs) to cortical interneurons," and "CRISPR-Cas9 mediated biallelic knockout of CHD2 resulted in fewer TUBB3+ neurons with shorter neurites." GeneReviews: "Complete CHD2 loss in a human stem cell model resulted in defects in the development of inhibitory interneurons and altered expression of genes important in neurotransmission." Impaired interneuron differentiation is described as a "well established… pathogenic mechanism in epilepsy." Cell types: CL:0000617 GABAergic neuron; CL:0011005 GABAergic interneuron (verify).

Node 5 — CELLULAR/TISSUE: Reduced neuron number and altered excitatory/inhibitory balance. "A heterozygous Chd2 loss mouse showed deficits in neuronal development including reduced number of both excitatory and inhibitory neurons and severe impairments in long-term memory" (Kim et al., Neuron 2018;100:1180–1193.e6 ⚠️ UNVERIFIED PMID). "Loss of a single Chd2 copy leads to deficits in neuron proliferation and a shift in neuronal excitability."

Node 6 — TISSUE: Cortical network hyperexcitability with altered background oscillations. Mouse ECoG shows "a global reduction in the total power of background activity" plus "increased susceptibility to seizures induced by acute administration of 4-aminopyridine" (Mavashov et al. ⚠️). Notably, the background-EEG change is a state abnormality independent of ictal events — a candidate translational biomarker.

Node 7 — ORGANISM: Photosensitive, myoclonic-predominant generalized epilepsy + developmental encephalopathy. Seizures, regression, ID, ASD.

A parallel/feedback loop worth modeling separately: Node 4/6 (cortical hyperexcitability with occipital-predominant photic drive) → photosensitivity → self-induction behavior → increased seizure burden → further encephalopathy. The self-induction arm is behavioral, not purely neurophysiological, and is unusual enough to merit its own node.

Suggested GO terms (verify all with OAK)

Biological process: GO:0006338 chromatin remodeling; GO:0043044 ATP-dependent chromatin remodeling; GO:0006325 chromatin organization; GO:0034728 nucleosome organization; GO:0006357 regulation of transcription by RNA polymerase II; GO:0030182 neuron differentiation; GO:0050767 regulation of neurogenesis; GO:0021895 cerebral cortex neuron differentiation; GO:0007399 nervous system development; GO:0006281 DNA repair (for the PAR-binding arm). Molecular function: GO:0004386 helicase activity; GO:0003682 chromatin binding; GO:0016887 ATP hydrolysis activity; GO:0140658 ATP-dependent chromatin remodeler activity (verify). Cellular component: GO:0005634 nucleus; GO:0000785 chromatin.

Other mechanism domains

• Metabolic changes: None established. No metabolic biomarker, no storage material, no enzyme deficiency. This is a negative finding worth recording because it is what distinguishes CHD2 from the true progressive myoclonic epilepsies it can mimic.
• Immune involvement: None established.
• Tissue damage mechanisms: Progressive posterior-predominant cerebral atrophy is documented, but its driver — seizure-related excitotoxic injury versus a primary progressive neurodevelopmental/degenerative process — is unresolved. Another good KNOWLEDGE_GAP entry.
• Protein dysfunction: Loss of function via truncation/degradation or, for missense variants in functional domains, impaired DNA binding and remodeling. No aggregation, no misfolding pathology.
• Single-cell / cell-type specificity: An important, explicitly-flagged gap — CHD2 is expressed in oligodendrocytes "almost to the same extent as neurons, but…did not study this cell type in their mouse model" (PMID:39391213 ✅). Given that CHASERR deletion (excess CHD2) produces cerebral hypomyelination, an oligodendrocyte arm of CHD2 biology is strongly suspected and essentially unexplored. CL:0000128 oligodendrocyte.

7. Anatomical Structures Affected

Organ level. The phenotype is brain-restricted — GeneReviews emphasizes that affected individuals show a "brain-restricted phenotype," implying "a unique role for CHD2" in human brain despite ubiquitous expression. No primary cardiac, renal, hepatic, or skeletal involvement. Body system: nervous system only. Secondary involvement is downstream of neurological disability — gastrointestinal dysfunction (constipation, feeding difficulty) and sleep disturbance, both prominent in adults (PMID:39601014 ✅), and injury risk from drop attacks.

Anatomical sites (UBERON candidates, verify): - UBERON:0000955 brain — primary - UBERON:0000956 cerebral cortex — the principal site of dysgenesis and hyperexcitability - Occipital/posterior cortex — implicated by both the posterior-predominant atrophy pattern and by photosensitivity physiology (visual cortex as the photic entry point). "Atrophy that tends to be more posterior and can be progressive." UBERON:0016530 occipital cortex (verify). - UBERON:0002037 cerebellum, UBERON:0004720 cerebellar vermis — inferior vermis hypoplasia 7/10 and mild cerebellar atrophy 4/10 in one imaging series - UBERON:0002421 hippocampal formation — hippocampal signal alterations 4/10, volume loss 2/10; memory deficits in mouse - Ventricular/subventricular zone — the developmental site of the progenitor defect (verify the correct UBERON ID) - Medial ganglionic eminence — origin of the NKX2-1⁺ interneurons most affected (verify)

Lateralization: Bilateral and generally symmetric. Generalized epilepsy with bilateral EEG discharges; atrophy is diffuse-to-posterior rather than focal or asymmetric.

Cell populations (CL candidates, verify): CL:0000617 GABAergic neuron and CL:0011005 GABAergic interneuron (primary); CL:0000681 radial glial cell (developmental origin); CL:0000679 glutamatergic neuron and CL:0000598 pyramidal neuron (also reduced in number); CL:0000031 neuroblast; CL:0000128 oligodendrocyte (suspected, understudied).

Subcellular: GO:0005634 nucleus; GO:0000785 chromatin. Interestingly, Xenopus work found CHD2 "localizes to microtubules of the mitotic spindle" (PMID:39391213 ✅) — a non-canonical localization that, if confirmed, could link CHD2 to progenitor division mechanics rather than transcription alone.

8. Temporal Development

Onset. Typical seizure onset 6 months to 4 years, median 30 months (GeneReviews), mean 26 months (PMID:25672921 ✅). Full observed range 1 day to 22 years (PMID:40934838 ✅). Onset pattern is characteristically acute to subacute and "explosive" — multiple daily seizures appearing over a short interval — rather than insidious. Developmental delay may precede seizures but is inconsistently documented.

Stages (a reasonable natural-history model, not a formal staging system — none exists): 1. Pre-seizure (birth–~2y): normal to mildly delayed development. 2. Explosive onset / encephalopathic phase (~1–4y): abrupt multiple daily myoclonic and absence seizures; developmental plateau or regression; photosensitivity emerges; self-induction may begin. 3. Established refractory phase (childhood): multiple seizure types, drug resistance, ID consolidates, ASD and behavioral features become prominent. 4. Adult phase: seizures persist in 79%; photosensitivity persists in 64%; behavioral and psychiatric burden dominates; ambulation independent in only 43% (PMID:39601014 ✅).

Progression rate and course. Variable; broadly chronic and lifelong with an early-childhood peak in seizure burden. The course is not classically progressive-degenerative, but progressive posterior atrophy in a subset and PME-mimicking trajectories in others mean "stable after early childhood" would be an overstatement. Curate as chronic with variable progression rather than as a single course descriptor.

Remission. Uncommon and treatment-mediated when it occurs. GeneReviews: "Only 13 of 33 affected individuals have been reported to be seizure free on ASM treatment for two to five years" (≈39%). Spontaneous remission is not established.

Critical periods. Two matter therapeutically. (1) The prenatal/early-postnatal progenitor window (mouse E12–E18 equivalent), when CHD2 governs radial glial self-renewal and interneuron specification — likely already passed by the time of diagnosis, which is a hard constraint on any restorative therapy. (2) The early-childhood seizure-onset window, when seizure control might plausibly protect development. Whether intervening in window 2 alters cognitive outcome is unknown and is arguably the central open clinical question in the disease.

9. Inheritance and Population

Prevalence. "The prevalence of CHD2-related neurodevelopmental disorders is not known" (GeneReviews) — this should be recorded as prevalence_class: NOT_YET_DOCUMENTED or UNKNOWN rather than fabricated. Available proxies (all diagnostic yields, not population prevalence — do not convert them): - 1.2% of epileptic encephalopathies in the original targeted-resequencing cohort (PMID:23708187 ✅). - ~1% of individuals in DEE cohorts carry a CHD2 variant (GeneReviews). - 0.25% of individuals in broad comprehensive-testing cohorts. - CHD2 was the fourth most highly implicated gene in one neurodevelopmental-disorder-plus-epilepsy cohort.

A defensible order-of-magnitude inference — clearly flagged as inference, not data — is a birth prevalence in the low single digits per 100,000, but I would not curate a number without a primary source.

Inheritance. Autosomal dominant, de novo in ~95%. GeneReviews: "CHD2-related neurodevelopmental disorders are autosomal dominant disorders typically caused by a de novo pathogenic variant."

Suggested Inheritance block: HP:0000006 Autosomal dominant inheritance. This is not a digenic/oligogenic disorder.

Penetrance. "Penetrance for CHD2-related neurodevelopmental disorders is unknown but assumed to be complete." Caveat: "a small number of instances" of inheritance from mildly affected parents are documented — so either penetrance is incomplete or expressivity is wide enough that mild carriers go unrecognized. Both readings are live.

Expressivity. Markedly variable — from severe DEE with profound ID to adult-onset non-syndromic epilepsy with preserved cognition. De Maria et al. describe "a wide spectrum of conditions" (⚠️).

Genetic anticipation. Not applicable — not a repeat-expansion disorder.

Germline mosaicism. Documented. GeneReviews: "Presumed parental germline mosaicism was reported in a family with unaffected parents and sib recurrence." This is the basis for counseling that recurrence risk after a de novo case is "slightly greater than that of the general population."

Recurrence risk. ~1% or slightly higher for siblings when the variant is confirmed de novo in the proband (germline mosaicism); 50% per pregnancy if a parent carries the variant.

Founder effects / carrier frequency / consanguinity. None, none, and no role respectively — a dominant de novo disorder in a gene under maximal LoF constraint (pLI=1) will not accumulate carriers or founder haplotypes.

Demographics. No ethnic or geographic predilection reported. Cohorts span Europe, North America, China, and Australia. Sex ratio approximately equal with a slight male excess: 53% male (108/205) (PMID:40934838 ✅) — consistent with chance and with the general male excess in ascertained neurodevelopmental cohorts; do not over-interpret. Age distribution is lifelong from early childhood; the adult population is real but historically under-described (PMID:39601014 ✅).

10. Diagnostics

Genetic testing — the definitive route

GeneReviews: "The diagnosis of a CHD2-related neurodevelopmental disorder is established in a proband with suggestive findings and a heterozygous pathogenic variant in CHD2 identified by molecular genetic testing."

Recommended approach. "An epilepsy or intellectual disability multigene panel that includes CHD2 and other genes of interest…is most likely to identify the genetic cause." Exome sequencing is advantaged because it "includes genes recently identified as causing intellectual disability whereas some multigene panels may not."

A practical warning worth curating: if a child has the classic CHD2 clinical picture but panel/exome is negative, escalate to genome sequencing rather than stopping. The CHASERR phenotype is caused by a non-coding deletion adjacent to CHD2 that standard coding-focused pipelines will not report.

Omics-based diagnostics

DNA methylation episignature is the standout, and is clinically actionable today: a validated CHD2 episignature "can help resolve variants of uncertain significance in CHD2" (GeneReviews; PMID:39391213 ✅). It functions as orthogonal functional evidence for ACMG classification (PS3-type) and can also exclude a provisional diagnosis when negative. Transcriptomics, proteomics, and metabolomics have no established diagnostic role.

Electrophysiology

EEG is the key functional test. Epileptiform abnormalities in 88% (122/138) (PMID:40934838 ✅). Characteristic findings: - Generalized spike-wave discharges (GeneReviews). - Photoparoxysmal response (PPR) on intermittent photic stimulation — but here is an important and clinically counterintuitive point: GeneReviews notes that a PPR "has only been recorded in two affected individuals," despite clinical photosensitivity in ~60–80%. Clinical photosensitivity in CHD2 markedly exceeds laboratory-demonstrable photoparoxysmal response. Do not use a negative IPS study to exclude photosensitivity or to lift precautions. Curate these as two distinct phenotypes with very different frequencies. - Background slowing and, in mouse, reduced total background power — a candidate quantitative biomarker.

Other electrophysiology (EMG, nerve conduction, ECG) has no established role.

Neuroimaging

MRI is normal in the majority — abnormal in only 19% (PMID:40934838 ✅); one series found no abnormality in 14 of 17. When abnormal: cerebral atrophy (posterior-predominant, sometimes progressive on serial imaging — 3 of 4 with sequential studies), hippocampal signal change (4/10) with volume loss (2/10), inferior vermis hypoplasia (7/10), mild cerebellar atrophy (4/10). PET, CT, and ultrasound have no specific role.

Laboratory tests and biomarkers

There is no diagnostic blood, urine, or CSF biomarker. No enzyme assay, no metabolite. Routine metabolic workup is normal — which is itself diagnostically useful when excluding PME/storage disease. Biopsy and histopathology have no role; there is no characteristic pathology.

Clinical criteria and differential diagnosis

No consensus clinical diagnostic criteria exist — diagnosis is genetic. GeneReviews notes the differential must consider "all genes known to be associated with epileptic encephalopathy (~90 have been identified)."

Key differentials: - Dravet syndrome (SCN1A) — the closest mimic; Suls et al. framed CHD2 as "sharing features with Dravet syndrome." Discriminators: CHD2 has later onset (median 30 mo vs. <12 mo), true febrile seizures are uncommon, and photosensitivity/self-induction is far more prominent. - Myoclonic-atonic epilepsy (Doose syndrome) — overlapping semiology; genetic testing distinguishes. - Lennox-Gastaut syndrome — CHD2 variants are found in LGS cohorts. - Eyelid myoclonia with absences (Jeavons syndrome) — CHD2 is the first identified genetic cause (PMID:25783594 ✅). - Progressive myoclonic epilepsies (Unverricht-Lundborg, Lafora, NCL, MERRF) — CHD2 can phenocopy PME (DOI:10.1002/epd2.70196 ⚠️); normal metabolic/storage workup and absence of a defining biopsy finding point away from true PME. - Other chromatin-remodeler NDDs — CHD8 (autism/macrocephaly), CHD7 (CHARGE), CHD1/CHD3/CHD4. - CHASERR-deletion disorder (NEDFSAB, OMIM #621012) — dysmorphic, hypomyelinating, more severe.

Screening

There is no newborn screening, no carrier screening, and no population screening for CHD2 — appropriately, given a de novo dominant disorder with no presymptomatic intervention. Cascade testing in families is limited to confirming de novo status in parents (which informs recurrence risk) and testing at-risk relatives when a parent is found to carry the variant.

11. Outcome / Prognosis

Survival and mortality. No published disease-specific survival curve, life-expectancy figure, or mortality rate. Adults into their mid-forties are reported (age range 18–45, median 21 in the adult cohort, PMID:39601014 ✅), so the condition is compatible with survival into at least middle adulthood. SUDEP risk should be presumed elevated given refractory generalized epilepsy with tonic-clonic seizures, but I found no CHD2-specific SUDEP data — do not assert a rate.

Seizure outcome. Poor. GeneReviews: "Most individuals remain refractory to treatment and require multiple anti-seizure medications. Only 13 of 33 affected individuals have been reported to be seizure free on ASM treatment for two to five years" (≈39%). In adulthood, 79% still have ongoing seizures (PMID:39601014 ✅).

Cognitive and functional outcome. Intellectual disability in ~86–95%, ranging mild to severe with a roughly even split at the extremes. In adults: independent ambulation in only 43%; ASD in 71%; behavioral issues in 100%; self-injury in 50%; anxiety/internalizing features in 71% (PMID:39601014 ✅). Independent living is not the expected outcome.

Complications. Injury from drop attacks and myoclonic-atonic seizures; status epilepticus; aspiration and feeding difficulty; gastrointestinal dysmotility/constipation; sleep disturbance; ASM adverse effects (a top-seven caregiver concern); self-injurious behavior; caregiver burden.

Recovery potential. Developmental regression is generally not recovered. There is no evidence that seizure control reverses established cognitive impairment, though the possibility that early control protects trajectory remains an open and important question.

Prognostic factors. The single most useful published prognostic signal comes from the adult cohort: "seizure severity is associated with worse comorbidities such as maladaptive behaviors, gait, gastrointestinal, sleep, and abnormal pain responsiveness" (PMID:39601014 ✅). A relative genotype-phenotype association has been suggested — the Chinese cohort reported that "The phenotypes, especially seizure control and fever sensitivity, and genotypes had a relative association" — but no robust, replicated genotype-based prognostic rule exists. Earlier onset and truncating variants trend toward greater severity, and interdomain missense variants toward milder/later-onset presentations, but these are trends, not predictors.

Prognostic biomarkers. None validated. Quantitative background-EEG measures are the most promising candidate, supported by both the mouse background-power finding and the human EEG abnormality rate.

12. Treatment

The honest headline

There is no disease-specific or targeted therapy, and no evidence-based ASM algorithm. GeneReviews: "At this time, no specific guidelines regarding choice of specific anti-seizure medications exist, as the best regimen for CHD2-related neurodevelopmental disorders is not yet established." The Epilepsiome concurs: "Thus far, there is no specific recommended treatment regimen for patients with CHD2 mutations." And the 2024 roadmap: "there are currently no targeted therapies available for CHD2-related disorders" (PMID:39391213 ✅).

I searched specifically for CHD2 ASM-response cohort data and did not find a published head-to-head comparison. This absence should be curated as an explicit knowledge gap rather than papered over with generalized-epilepsy inferences.

Pharmacotherapy — practice-based, not evidence-based

Treatment follows generalized/myoclonic-epilepsy principles: broad-spectrum agents (valproate, levetiracetam, clobazam, lamotrigine with caution, topiramate, zonisamide, ethosuximide for absences, clonazepam for myoclonus). Sodium-channel blockers (carbamazepine, oxcarbazepine, phenytoin, lamotrigine in some patients) can aggravate myoclonic and absence seizures in generalized epilepsies — a standard and important caution, but note that I found no CHD2-specific aggravation data, so this should be curated as general myoclonic-epilepsy practice rather than a CHD2-specific claim.

Suggested annotation pattern:

treatment_term:
  preferred_term: Pharmacotherapy
  term: {id: NCIT:C15986, label: Pharmacotherapy}
  therapeutic_agent:
  - preferred_term: valproic acid
    term: {id: CHEBI:39867, label: valproic acid}   # verify with OAK

CHEBI candidates requiring verification: valproic acid, levetiracetam, clobazam, clonazepam, lamotrigine, topiramate, ethosuximide, zonisamide, cannabidiol, fenfluramine. Per prior project experience, prefer CHEBI over NCIT for therapeutic_agent — NCIT drug terms frequently fail dynamic-enum validation.

Pharmacogenomics: No CHD2-specific pharmacogenomic guidance exists. Standard CPIC guidance (e.g. HLA-B*15:02 and carbamazepine) applies as it would to any patient.

Ketogenic diet: Reported ineffective in the small number tried — "Ketogenic diet was not effective in three affected individuals" (GeneReviews). n=3 is very weak evidence; curate the number, not a conclusion. MAXO:0000088 dietary intervention.

Fenfluramine: I searched specifically and found no CHD2 case series. It is used in Dravet and LGS and is mentioned in the roadmap as an example of successful drug repurposing in a related disorder, but there is no CHD2 evidence. Do not curate it as a CHD2 treatment.

Non-pharmacologic and supportive management

• Photic-trigger avoidance — the only CHD2-specific intervention with a clear rationale. GeneReviews: counsel that "exposure to intensely flickering lights may provoke seizures including eyelid myoclonias, absence seizures, and generalized tonic-clonic seizures." Practically: blue/Z1 tinted lenses, screen management, avoidance of strobe environments, and — given self-induction — behavioral strategies to interrupt the seeking behavior.
• Early intervention and education — ages 0–3 early intervention; 3–5 developmental preschool; ongoing special education, speech, occupational, and physical therapy. MAXO:0000011 physical therapy; speech therapy term (verify).
• Behavioral intervention — "Children may qualify for and benefit from interventions used in treatment of autism spectrum disorder, including applied behavior analysis (ABA)."
• Genetic counseling — MAXO:0000079 genetic counseling.
• Supportive care — MAXO:0000950 supportive care.
• Surgery/VNS — no CHD2-specific data; resective surgery is not rational for a generalized genetic epilepsy. VNS/corpus callosotomy for drop attacks would follow generic drug-resistant-generalized-epilepsy practice.

Advanced therapeutics in development

This is the most dynamic part of the field (all from PMID:39391213 ✅ unless noted):

• CHD2-upregulating ASOs — the lead strategy. "One of the most promising approaches for CHD2-RD is using antisense oligonucleotides…to increase CHD2 expression to overcome haploinsufficiency."
• CHASERR-targeting ASOs — the most elegant strategy. Because CHASERR is a negative cis-regulator, knocking it down raises CHD2: in mice, "targeting Chaserr…leads to an increase of CHD2 messenger RNA." The 2024 human CHASERR-deletion data both validate the target and define its therapeutic window — too much CHD2 causes a worse disease. Any CHASERR-directed therapy must therefore titrate into a narrow dosage band, which is a real and specific development risk. A practical obstacle: "CHASERR was presented to the n-Lorem Foundation but was not accepted due in part to the lack of a pharmacodynamic biomarker."
• Fusion-transcript induction — a newer approach reported to upregulate haploinsufficient CHD2 (bioRxiv 2025.05.28.656657).
• miRNA modulation, stop-codon read-through (nonsense variants only), and targeted drug repurposing screens in animal models, "currently underway."
• Gene replacement is largely off the table: AAV vectors cannot carry the >5 kb CHD2 coding sequence, and "the cis-acting feedback loop between CHD and CHASERR would likely introduce additional challenges."

Delivery constraint: all candidates must "cross the blood–brain barrier or be injected directly into the cerebrospinal fluid" — i.e. intrathecal ASO delivery on the nusinersen/tofersen model.

If the KB entry uses the antisense_oligonucleotide_therapy module, note that the CHD2 approach is a fourth mechanism class not currently in that module — neither RNase H knockdown of a pathogenic transcript, nor splice modulation, nor steric translation blockade, but upregulation of a haploinsufficient gene via knockdown of its cis-repressive lncRNA. That is a genuine gap in the existing module and worth flagging.

Clinical trials: I found no registered interventional trial for a CHD2-targeted therapy. Do not populate clinical_trials with an NCT unless one is verified on ClinicalTrials.gov at curation time.

13. Prevention

Primary prevention: Not possible. De novo mutations in a constrained gene are not preventable by any known modifiable exposure.

Secondary prevention: Early genetic diagnosis is the actionable lever. Rapid panel/exome testing in early-onset DEE enables trigger counseling, avoidance of aggravating ASMs, early developmental intervention, accurate recurrence counseling, and research/trial eligibility. There is no population screening program and none is warranted.

Tertiary prevention (preventing complications) — the substantive category here: - Photic-trigger avoidance to reduce seizure burden. - Injury prevention — protective headgear for drop attacks; supervision. - Avoiding seizure-aggravating ASMs. - Seizure action plans and rescue medication for status epilepticus. - Proactive management of the adult-emergent comorbidity cluster — gait, GI, sleep, behavior, pain responsiveness (PMID:39601014 ✅). The adult data argue for surveillance of these domains rather than reactive management.

Genetic counseling and reproductive prevention. Counseling should cover: ~95% de novo origin; ~1% or slightly higher sibling recurrence risk from germline mosaicism; 50% transmission risk if a parent carries the variant; testing of both parents to establish de novo status. Prenatal diagnosis and preimplantation genetic testing are technically available once the familial variant is known — most relevant for the germline-mosaicism scenario and for affected/mildly-affected parents.

Immunization: No disease-specific vaccine strategy. Routine childhood immunization should proceed normally; given 37% fever sensitivity, prophylactic antipyretics around vaccination are a reasonable practice-level consideration (extrapolated from Dravet practice — not CHD2-evidenced, so flag as such).

Public health / environmental interventions: Broad flicker-safety standards for broadcast and gaming content (the "Pokémon shock" regulatory lineage) benefit photosensitive individuals generally, including CHD2 patients. This is a real but non-specific intervention.

14. Other Species / Natural Disease

Taxonomy of orthologs: | Species | NCBI Taxon | Gene | Notes | |---|---|---|---| | Homo sapiens | NCBITaxon:9606 | CHD2 (NCBI Gene 1106) | | | Mus musculus | NCBITaxon:10090 | Chd2 | Principal model | | Danio rerio | NCBITaxon:7955 | chd2 | Photosensitivity model | | Xenopus tropicalis/laevis | NCBITaxon:8364 / 8355 | chd2 | Spindle-localization finding | | Drosophila melanogaster | NCBITaxon:7227 | kis (kismet, related CHD) | Distant ortholog |

Naturally occurring disease in other species: None reported. I found no OMIA entry, no veterinary case series, and no naturally occurring CHD2-related epilepsy in companion animals or wildlife. All animal disease is experimentally induced. Record this as a clear negative.

Breed (VBO): Not applicable — no breed-associated natural disease.

Comparative biology. CHD2 is deeply conserved across vertebrates in both sequence and function, and — importantly — the photosensitivity phenotype itself is conserved: chd2 knockdown "markedly enhanced mild innate zebrafish larval photosensitivity" (PMID:25783594 ✅). Conservation of a specific, unusual clinical feature across ~400 million years of divergence is strong evidence that the human photosensitivity is a direct consequence of CHD2 loss rather than a downstream epiphenomenon. The Chaserr lncRNA is likewise conserved and functionally equivalent in mouse.

Zoonotic potential / cross-species transmission: Not applicable — a germline genetic disorder.

15. Model Organisms

Zebrafish (Danio rerio, NCBITaxon:7955)

The best model for the seizure and photosensitivity phenotypes. Suls et al. established it alongside the human genetics: "They knocked down chd2 in zebrafish, and chd2-knockdown larvae exhibited altered locomotor activity, with field-potential recordings revealing epileptiform discharges similar to seizures in affected persons" (PMID:24207121 ✅). Galizia et al. extended it to the defining feature: "Knockdown of chd2 in zebrafish with targeted morpholino results in larvae displaying seizure-like behavior and photosensitivity, recapitulating the phenotype seen in humans" (PMID:25783594 ✅). Morphant larvae also show "pericardial edema, microcephaly, body curvature, absent swim bladder, stunted growth, and epileptiform discharges" (Baraban lab, UCSF).

Recapitulation: Excellent for seizures and photosensitivity — arguably the only model that captures the disease's signature feature. Limitations: morpholino knockdown carries well-known off-target and toxicity concerns; some phenotypes (edema, curvature) are non-specific morphant artifacts; larval models cannot address cognition, ASD-like behavior, or long-term progression. Primary application: high-throughput drug repurposing screens.

Mouse (Mus musculus, NCBITaxon:10090)

The best model for development, circuits, and cognition; the weakest for seizures.

• Kim et al. (Neuron 2018;100:1180–1193.e6 ⚠️ UNVERIFIED PMID) — "Chd2 Is Necessary for Neural Circuit Development and Long-Term Memory." Chd2 haploinsufficiency "compromises cortical development, synaptic function, and memory in mice"; heterozygotes show "reduced number of both excitatory and inhibitory neurons and severe impairments in long-term memory."
• Mavashov et al. (Epilepsia 2026, DOI:10.1002/epi.70073 ⚠️; bioRxiv 2025.03.18.643778) — a frameshift-truncation model. "Heterozygous and homozygous Chd2 mutant mice demonstrate reduced CHD2 expression, alteration in background electrocorticographic (ECoG) oscillations, behavioral deficits, and an increased susceptibility to seizures." Specifically: a "global reduction in the total power of background activity" and increased susceptibility to 4-aminopyridine-induced seizures; on the 129X1/SvJ background, motor deficits including clasping and rotarod impairment, and growth retardation in homozygotes.
• C-terminal deletion model: "Mice with homozygous deletions of the C-terminus of Chd2 exhibit perinatal lethality." Heterozygotes show systemic abnormalities but "no neurological defects were reported."
• Chaserr knockout mouse (Ulitsky lab): elevated Chd2; "early lethality" — validating the dosage-window concern for CHD2-raising therapy.
• Uncoupling study (Mol Psychiatry 2026, DOI:10.1038/s41380-026-03539-x ⚠️) — "Uncoupling memory impairments from autism-associated behaviors in Chd2 deficient mice," suggesting the cognitive and social phenotypes are mechanistically dissociable.

The central mouse limitation — and this is important enough to curate as a formal HUMAN_MODEL_MISMATCH: "No CHD2 mouse model exhibits spontaneous seizures," and models "did not appear to have a 50% CHD2 protein reduction" (PMID:39391213 ✅). The Ulitsky model showed "unusual EEG activity, but also did not have clinical seizures." So the field's best mechanistic model does not reproduce the disease's cardinal clinical feature, while the model that does reproduce it (zebrafish larvae) cannot address the developmental and cognitive core. Any mouse-derived efficacy claim about seizure control must be treated as extrapolation.

Xenopus

Willsey lab work found "a stronger phenotype seen in morpholinos than in CRISPR editing" and, notably, that CHD2 "localizes to microtubules of the mitotic spindle" — a non-chromatin localization with implications for progenitor division.

Human iPSC, hESC, and organoid models

The most human-relevant system, and the source of the interneuron mechanism. - hESC-derived cortical interneuron differentiation identified CHD2 as an NKX2-1 target and showed "CHD2 deficiency impaired cIN development and altered later cIN function" (PNAS 2017, DOI:10.1073/pnas.1712365115 ⚠️; Sci Rep 2022, PMID:36115870 ✅). - "CRISPR-Cas9 mediated biallelic knockout of CHD2 resulted in fewer TUBB3+ neurons with shorter neurites." - "Several induced CHD2 pluripotent stem cells (iPSC) in both human and CRISPR cells induced around the world," with an NIH grant supporting expansion into cortical organoids (PMID:39391213 ✅). - Patient-derived iPSCs from CHASERR-deletion individuals showed increased CHD2 protein, confirming the cis-repression mechanism in human cells (NEJM 2024).

Limitations: iPSC/organoid systems model early development, not mature circuits or seizures; and a specific technical caveat flagged in the roadmap — "CHD2 is lowly expressed in certain cell types," producing "irregular results including CHD2+/– models that do not appear to achieve 50% protein reduction." Protein quantification is a genuine unsolved methods problem in this field.

Model databases: MGI (mouse), ZFIN (zebrafish — Suls et al. is indexed as ZDB-PUB-131218-5), Xenbase, IMSR/IMPC/KOMP, Alliance of Genome Resources, Cellosaurus.

Curator notes for the dismech entry

A few things I'd flag before this gets turned into YAML, since some of them are the kind of thing that quietly breaks validation or, worse, quietly encodes something wrong:

Verify before citing. Seven references here have confirmed titles/journals/DOIs but PMIDs I could not resolve in this session: De Maria 2022 (AJMG A), Kim 2018 (Neuron), Meganathan 2017 (PNAS), Rom 2019 (Nat Commun), the CHASERR NEJM paper, Chityala (Epileptic Disorders), and Mavashov (Epilepsia). Run just fetch-reference on each and confirm the snippet is an exact substring before use. The verified set — 23708187, 24207121, 25783594, 25672921, 31677157, 40934838, 39601014, 39391213, 36115870, 38496558 — is safe to build the backbone on.

The riskiest ontology term is the most important one. Photosensitivity is this disease's signature, and I deliberately did not guess an HPO ID for it. Look it up properly. Same for "eyelid myoclonia," "myoclonic-atonic seizure," and the photoparoxysmal-EEG term — HPO has renamed several seizure-semiology terms recently and stale memory is exactly how a bad term.label sneaks past.

Two frequencies for the same phenotype, and that's correct. GeneReviews says 80% photosensitivity (20/25, when specifically asked); the pooled analysis says 59% (80/136). These aren't in conflict — they're different ascertainment. Curate both with distinct population values rather than picking a winner.

Don't collapse "photosensitive" and "photoparoxysmal response." Clinical photosensitivity runs 59–80%; a recorded PPR on EEG has been documented in a handful of patients. Two phenotypes, wildly different frequencies, opposite clinical implications for whether a normal IPS study means anything.

The CHASERR story wants its own entry. Too little CHD2 gives you DEE94; too much gives you NEDFSAB (OMIM #621012) — dysmorphic, hypomyelinating, more severe. Same gene, opposite direction, different disease. That's a separate dismech entry plus a grouping, and it also means the existing ClinGen TS=0 dosage call is historically true but biologically stale. Worth a discussions note rather than silently inheriting the old score.

Three honest knowledge gaps worth encoding rather than smoothing over: (1) no CHD2-specific ASM comparative data exists, so any treatment ranking is borrowed from generalized-epilepsy practice; (2) whether the progressive posterior atrophy is seizure-driven injury or primary neurodegeneration is unresolved; (3) the mouse models don't seize, which is a textbook HUMAN_MODEL_MISMATCH — the model that captures the mechanism can't show you the disease, and the model that shows you the disease can't tell you the mechanism.

Sources: - GeneReviews: CHD2-Related Neurodevelopmental Disorders - OMIM #615369 DEE94 · OMIM *602119 CHD2 · OMIM *620993 CHASERR · OMIM #621012 NEDFSAB - Carvill 2013, Nat Genet (PMID:23708187) - Suls 2013, AJHG (PMID:24207121) - Galizia 2015, Brain (PMID:25783594) - Thomas 2015, Neurology (PMID:25672921) - Chen 2020, DMCN (PMID:31677157) - Puri 2025, Seizure (PMID:40934838) - Adult Phenotype of CHD2-Associated Disorders (PMID:39601014) - Prince 2024, roadmap to cure CHD2-RD (PMID:39391213) - Meganathan 2022, Sci Rep (PMID:36115870) · Meganathan 2017, PNAS - Wilson 2018, Front Mol Neurosci — chromatin remodelers in epilepsy - CHASERR deletion, NEJM 2024 · preprint PMID:38496558 - Rom 2019, Nat Commun — Chaserr regulates Chd2 - Kim 2018, Neuron — Chd2 and long-term memory - Mavashov, Epilepsia 2026 · bioRxiv preprint - De Maria 2022, Am J Med Genet A - Chityala, Epileptic Disorders — PME mimic - ClinGen CHD2 gene curation · ClinGen CHD2 dosage - Epilepsiome: CHD2 — what you need to know (2023) - MedlinePlus: CHD2 myoclonic encephalopathy