PCDH19 Clustering Epilepsy — Comprehensive Research Report

Prepared for dismech KB curation | Category: Mendelian (X-linked)

A quick word before we wade in, since this is going into a knowledge base: I've pulled PMIDs from the literature and from search, but per the project's own anti-hallucination SOP, treat every PMID and quote below as a lead to verify with just fetch-reference PMID:XXXX before it becomes an evidence snippet. I've flagged the handful where my confidence on the exact PMID digits is softer. The ontology IDs (MONDO/HGNC) I pulled straight from the local OAK adapters, so those are solid.

PCDH19 is one of biology's little contrarians. It sits on the X chromosome, and yet it breaks the classic X-linked rulebook completely backwards — the girls (heterozygous) get the disease, and the boys carrying the exact same broken copy (hemizygous) mostly walk away fine. The disease needs the mosaic. A brain that's uniformly broken is, paradoxically, a brain that works. That's the whole strange heart of this thing, and it drives basically every section below.

1. Disease Information

Overview. PCDH19 Clustering Epilepsy (PCDH19-CE) is an X-linked developmental and epileptic encephalopathy caused by loss-of-function variants in PCDH19 (protocadherin 19). Its signature is early-life seizures that come in tight clusters, are strongly fever-sensitive, and are frequently accompanied by intellectual disability, autism spectrum features, and other neuropsychiatric problems. It affects females and rare mosaic males, while "transmitting" hemizygous males are typically spared.

Key identifiers (verified against local MONDO/HGNC adapters):

Synonyms / alternative names (from MONDO): DEE9; EIEE9; EFMR (Epilepsy and Mental Retardation Limited to Females); Epilepsy, Female-restricted, with Mental Retardation; PCDH19-related female-limited epilepsy (PCDH19-FE / FLE); PCDH19-related infantile epileptic encephalopathy; Juberg-Hellman syndrome; "Girls Clustering Epilepsy." The field has largely converged on "PCDH19 Clustering Epilepsy" because the older names got the sex-limitation and the ID emphasis slightly wrong (mosaic males exist; many patients have normal intellect).

Data provenance. Disease-level aggregated resources (OMIM, Orphanet, MONDO) plus cohort/case-series literature. Not EHR-derived at the individual level; the large phenotyping series (e.g. Kolc et al., Transl Psychiatry 2020, PMID:32366910) are patient-registry/cohort aggregations.

MONDO places it as a subclass of X-linked intellectual disability–epilepsy syndrome (MONDO:0016160), genetic developmental and epileptic encephalopathy (MONDO:0100062), and X-linked complex neurodevelopmental disorder (MONDO:0100148).

2. Etiology

Primary cause — genetic. Heterozygous (female) or mosaic (male) loss-of-function variants in PCDH19. This is a monogenic disorder; there is no established environmental cause. Fever/illness is a potent trigger/precipitant of seizure clusters, not a cause of the disease itself.

Genetic risk factors. - The causal event is the PCDH19 variant. There is no separate "susceptibility locus" architecture in the GWAS sense — this is Mendelian. - Sex is the dominant effect modifier: being female (constitutively mosaic via random X-inactivation) is effectively the prerequisite for the classic phenotype. Genetic males require somatic mosaicism to be affected. - X-inactivation skewing plausibly modulates the mutant:wild-type cell ratio and hence severity, though a clean genotype–XCI–severity correlation has not been firmly nailed down.

Environmental / trigger factors. Febrile illness is the archetypal precipitant of clusters; vaccination-associated fever and intercurrent infection are commonly reported cluster triggers. These modulate expression/timing, not disease occurrence.

Protective factors. None genetically established. The most striking "protective" observation is the counterintuitive one: uniform hemizygous expression of mutant PCDH19 in males is protective against the epilepsy (the mosaic, not the mutation per se, is pathogenic) — the cellular-interference logic in §6.

Gene–environment interaction. The central G×E axis is variant × febrile/inflammatory stress: the mutant mosaic brain has a lowered threshold such that fever/illness reliably pushes it into clustered seizures. There is emerging interest in a blood–brain-barrier / neuroinflammation contribution to that fever sensitivity (see §6).

3. Phenotypes

Frequencies below are anchored to the standardized phenotyping cohort (Kolc et al. 2020, PMID:32366910, n≈112) and corroborating series.

Seizures (core). - Seizure clusters — the defining feature: 94% (106/112) had clustered seizures. Clusters averaged ~4.6 days and ~15 seizures each (range 2–100). → HPO HP:0032807 Cluster of seizures (verify exact HP label); HP:0001250 Seizure. - Focal (focal-onset) seizures, often with a fearful/affective component ("fearful screaming") — highly characteristic. → HP:0007359 Focal-onset seizure. - Febrile / fever-sensitive seizures — clusters typically ignited by fever. → HP:0002373 Febrile seizure / HP:0011171 Generalized-onset seizure as applicable; consider temporality: RECURRENT. - Also tonic-clonic, tonic, absence, myoclonic, atonic seizures reported. → HP:0002069, HP:0032792, HP:0002121, HP:0002123, HP:0010819.

Neurodevelopmental / cognitive. - Intellectual disability — spectrum, and importantly ~56% have normal intellect in the standardized cohort; among the rest: mild 18%, moderate 8%, severe 12.5%, profound 2%. → HP:0001249 Intellectual disability (severity qualifier variable). - Autism spectrum disorder — ~62% met ASD criteria. → HP:0000717 Autism. - Executive dysfunction — ~63%. → HP:0002019-adjacent / behavioral abnormality terms. - ADHD / hyperactivity — frequently reported. → HP:0007018 Attention deficit hyperactivity disorder. - Obsessive-compulsive / psychiatric features — OCD ~21% of assessed; psychiatric risk (including psychotic/schizophrenia-spectrum disorders) rises in adulthood. → HP:0000722 Obsessive-compulsive behavior; HP:0100753 Schizophrenia. - Aggression, mood/behavioral disturbance common. → HP:0000718 Aggressive behavior.

Phenotype characteristics (summary). - Onset: infancy/early childhood; mean ~12 months, median ~10 months in females (range 1.5–60 mo); OMIM/Orphanet quote ~10 months average, range ~2 mo–3 yr. - Severity: variable — from normal-cognition/seizure-limited to severe DEE. - Course: episodic/clustered seizure pattern superimposed on a chronic disorder; many see seizure attenuation in later childhood/adolescence. - Genotype-independent severity driver: earlier onset + higher seizure burden → worse ID, ASD, executive dysfunction (Kolc 2020: association p=0.001 for ASD, p≈4.7×10⁻⁴ for executive dysfunction).

Quality-of-life impact. High: 75% scored "very high" on SDQ impact. Cluster unpredictability, fever-triggered emergencies, and neuropsychiatric comorbidity dominate family burden; adult psychiatric risk is a major long-horizon concern.

4. Genetic / Molecular Information

Causal gene. PCDH19 (protocadherin 19), Xq22.1, OMIM *300460, hgnc:14270. Encodes a δ2-protocadherin, a non-clustered member of the cadherin superfamily — a calcium-dependent cell–cell adhesion glycoprotein predominantly expressed in brain. Domain architecture: 6 extracellular cadherin (EC) repeats, a transmembrane domain, and a cytoplasmic tail with conserved motifs CM1/CM2 (per protocadherin structure review, PMID:34201522 — verify).

Pathogenic variants. - Variant types: the full loss-of-function spectrum — missense, nonsense, frameshift (small indels), splice-site, and whole/partial-gene deletions; a triplication mechanism has also been described (MDPI Genes 2024, PMC11506946). Missense variants cluster in the extracellular cadherin domains (EC1–EC6), consistent with disrupted homophilic adhesion. - >120 distinct variants reported; most are family-specific/private, with a handful of recurrent ones. - De novo rate: ~48–70% across cohorts (Kolc 2020: 48% de novo; epilepsiome/clinical reviews cite ~70%). - Classification: per ACMG/AMP, LoF is an established mechanism, so null variants are readily P/LP; missense VUS burden exists — check ClinVar/ClinGen. - Allele frequency: pathogenic variants are essentially absent from population controls (gnomAD) given severity/de-novo nature. - Somatic vs germline: germline in females; affected males are typically postzygotic somatic mosaics (variant allele fraction correlates loosely with phenotype severity — PMC9669318). - Functional consequence: loss of function / haploinsufficiency at the cellular level, but the pathogenic unit is the mosaic tissue, not the single null cell.

Modifier genes / downstream expression. PCDH19 dysfunction perturbs steroidogenic gene expression — notably genes in the allopregnanolone (neurosteroid) synthesis pathway (e.g. AKR1C1/2/3, CYP steroidogenic enzymes), yielding an "allopregnanolone deficiency" signature in patients (Tan et al. 2015 — verify PMID). This is both a candidate modifier axis and a therapeutic hook (§12).

Epigenetic. No established disease-defining methylation signature (episignature) as of current literature; X-inactivation is the dominant epigenetic determinant because it sets the mosaic ratio. Worth flagging as a knowledge gap.

Chromosomal abnormalities. Large PCDH19 deletions and the reported triplication are detectable by CMA; otherwise this is a single-gene disorder, not an aneuploidy syndrome.

Suggested annotations: gene hgnc:14270; GO cellular component GO:0005911 cell-cell junction / GO:0005912 adherens junction; molecular function GO:0005509 calcium ion binding, GO:0098631 cell adhesion mediator activity.

5. Environmental Information

• Environmental factors: no toxin/pollutant/radiation etiology. The relevant "environmental" input is febrile/inflammatory stress as a cluster precipitant.
• Lifestyle factors: not a driver of disease occurrence; fever management and illness avoidance matter for cluster prevention.
• Infectious agents: not causal, but intercurrent infection (via fever/inflammation) is the leading cluster trigger. This is a candidate node for a BBB/neuroinflammation mechanism rather than a pathogen-specific one.

6. Mechanism / Pathophysiology

This is the meaty part, and it's genuinely one of the more elegant puzzles in epilepsy genetics. Think of it like a tissue that only misbehaves when you mix two paint colors — a wall painted all one color (all-mutant hemizygous male) looks fine; a wall with random patches of two colors (mosaic female) develops the crack.

The protein's day job. PCDH19 is a calcium-dependent adhesion molecule. It does homophilic binding (PCDH19-cell to PCDH19-cell) and partners with N-cadherin (CDH2) at adherens junctions to form a strongly adhesive complex. Through this it governs neural progenitor division balance, neuronal migration, axon outgrowth, and synaptogenesis — the scaffolding of a properly wired cortex.

Causal chain (upstream → downstream):

1. Trigger — PCDH19 LoF variant in a mosaic context. Random X-inactivation in females (or postzygotic mutation in males) yields a salt-and-pepper mixture of PCDH19-positive and PCDH19-negative neurons.
2. Cellular interference / abnormal cell sorting. The two populations mis-sort and mis-communicate at their boundaries — the mixed population is what's pathogenic, uniform populations are not. Pederick et al. (Neuron 2018, PMID:29429936 — verify) showed abnormal cell sorting segregating wild-type from mutant cells in mosaic mouse cortex, the mechanistic cornerstone of the "why females, not hemizygous males" riddle.
3. Altered neurogenesis (parallel/contributing arm). Patient-iPSC and cortical-organoid models (PMC8268119; PMC11024992) show accelerated/precocious neurogenesis, a shift toward asymmetric progenitor divisions, premature differentiation, longer neurites, up-regulated neurogenic markers (NCAD, MAP2, TUBB3), mitotic-spindle/centrosome abnormalities (PCDH19 colocalizes with γ-tubulin; ~17% multipolar metaphases vs 4% control), and smaller organoids. Notably, a Xenopus model (PNAS 2024, PMC on pnas.org) argues some phenotypes are mosaicism-independent, so cellular interference isn't the whole story.
4. Neurosteroid / GABAergic arm. PCDH19 dysfunction down-regulates allopregnanolone synthesis (a potent positive allosteric modulator of GABA_A receptors). Reduced allopregnanolone → decreased GABA_A-mediated inhibition → tilted excitation/inhibition balance. Emerging interneuron data (2025 mouse preprint, biorxiv 688097) show focal reductions in parvalbumin-expressing cortical interneurons and hyperthermic seizure susceptibility — a concrete substrate for both the E/I imbalance and the fever sensitivity.
5. E/I imbalance → hyperexcitability → seizures. The convergent endpoint: network hyperexcitability and hypersynchrony → clustered focal seizures, fever-facilitated.
6. Neurodevelopmental output. The same adhesion/migration/neurogenesis disruption yields the ID/ASD/behavioral phenotype in parallel with the epilepsy.

Additional proposed contributors (Trivisano et al., Epilepsy & Behavior 2024, S1525-5050(24)00111-2): asymmetric cell division/heterochrony, altered steroid-gene expression, blood–brain-barrier dysfunction (a candidate mediator of fever-triggered clusters), and reduced GABA_A function.

Cell types / processes for annotation: - Cell types: CL:0000617 GABAergic neuron; CL:0000561 — use CL:0000499 stromal? no — use CL:0011005 GABAergic interneuron / CL:0000598 pyramidal neuron; CL:0000031 neuroblast (neural progenitor); parvalbumin interneuron (CL for cortical interneuron). - Biological processes: GO:0007156 homophilic cell adhesion via plasma-membrane adhesion molecules; GO:0007399 nervous system development; GO:0001764 neuron migration; GO:0022008 neurogenesis; GO:0051301 cell division / GO:0000278 mitotic cell cycle; GO:0006874 cellular calcium ion homeostasis; neurosteroid-relevant GO:0006702 androgen biosynthetic process / allopregnanolone synthesis; GO:0060078 regulation of postsynaptic membrane potential (GABA_A).

This is a place to lean on the project's epilepsy_excitation_inhibition_imbalance module — the "excitation-inhibition imbalance → neuronal hyperexcitability/hypersynchrony → seizure generation" chain (conformance target epilepsy_excitation_inhibition_imbalance#Excitation-Inhibition Imbalance) fits PCDH19-CE's downstream endpoint cleanly, with the PCDH19-specific upstream (mosaic cellular interference + allopregnanolone/GABA_A) substituted in.

7. Anatomical Structures Affected

• Organ / system: central nervous system (UBERON:0000955 brain), specifically cerebral cortex (UBERON:0000956) and limbic/frontal networks (fear/affective seizure semiology implicates temporo-limbic circuitry). Body system: nervous system (UBERON:0001016).
• Tissue / cell: cortical neural progenitors (ventricular/subventricular zone), GABAergic interneurons (incl. parvalbumin-expressing), excitatory pyramidal neurons. → CL terms in §6.
• Subcellular: adherens junctions / cell–cell junctions (GO:0005912), plasma membrane (GO:0005886), centrosome / mitotic spindle (GO:0005813, GO:0005819), and GABA_A-receptor-bearing postsynaptic membrane (GO:0045211).
• Localization / lateralization: brain imaging (MRI) is usually normal/nonspecific; no consistent focal lesion. Seizures are focal but multifocal/bilateral-capable; no fixed lateralization.

8. Temporal Development

• Onset: infantile/early-childhood, mean ~10–12 months (range ~2 months to 3–5 years). Onset is typically subacute, heralded by a fever-triggered seizure cluster.
• Progression / course: episodic-clustering superimposed on a chronic neurodevelopmental disorder. Clusters recur (often every few weeks to months, illness-linked) then, characteristically, seizures tend to attenuate with age.
• Remission: at age ≥11, ~28% seizure-free; mean seizure "offset" age ~17.6 years (range 11–38); ~14% reach 10-year resolution (Kolc 2020). Seizures frequently ease in adolescence — but neuropsychiatric burden persists and adult psychiatric risk rises, so "seizure remission" ≠ "recovery."
• Critical windows: the infantile/early-childhood window (active neurogenesis + first fever exposures) is when both epileptogenesis and developmental impact are set — the rationale for early diagnosis and cluster-abortive strategies.

9. Inheritance and Population

• Inheritance pattern: X-linked with an unusual sex-limited expression — heterozygous females and mosaic males affected; hemizygous ("transmitting") males spared. This is not standard X-linked recessive or dominant; model it explicitly. The mechanism is cellular interference / mosaic-dependent (§6). → HPO mode-of-inheritance: HP:0001417 X-linked inheritance (with a note on the female-limited, mosaic-dependent expression — this is a genuine curation nuance worth a discussion/knowledge-gap block).
• Penetrance: high in females — ~97% of pathogenic-variant-carrying females are affected (~3% asymptomatic carriers).
• Expressivity: highly variable (normal cognition + limited seizures → severe DEE), even within families.
• Anticipation: not a repeat-expansion disorder; no anticipation.
• Germline/somatic mosaicism: central — affected males are somatic mosaics; unaffected transmitting fathers pass the variant to affected daughters; germline/gonadal mosaicism in a parent can produce recurrence.
• Founder effects / consanguinity: not relevant (mostly de novo, private variants; not consanguinity-driven).
• Carrier frequency: unaffected hemizygous-male "carriers" transmit; population carrier frequency is not meaningfully tabulated given the de-novo-dominated architecture.
• Epidemiology: no precise prevalence figure is established; PCDH19-CE is described as "one of the most common monogenic epilepsies" and the second most clinically relevant epilepsy gene after SCN1A. In SCN1A-negative Dravet-like females, PCDH19 variants are found in ~25% (Depienne 2009). For the KB prevalence block, the honest fill is a qualitative prevalence_class: RARE / Orphanet "rare" with measure_type: UNKNOWN and notes citing the "second most common epilepsy gene" framing — a hard cases-per-100,000 number isn't well sourced.
• Demographics: overwhelmingly female; no strong ethnic/geographic clustering (private de novo variants worldwide). Age distribution: pediatric onset, lifelong condition.

10. Diagnostics

• Genetic testing is the definitive diagnostic.
• Gene panels (epilepsy/DEE panels) and WES/WGS are first-line, especially in females with fever-sensitive clustered seizures who are SCN1A-negative (the classic "looks like Dravet but isn't, and is a girl" scenario). → MAXO MAXO:0000004-adjacent; genetic testing action.
• Single-gene PCDH19 sequencing + deletion/duplication analysis (MLPA/CMA) captures point variants and CNVs (deletions, the reported triplication).
• Mosaicism-aware testing (deep sequencing) is important in males — standard sequencing can miss low-VAF somatic variants. → repeat/mosaic detection considerations.
• Chromosomal microarray for large deletions/duplications.
• EEG: interictal often normal early; ictal recordings show focal onsets (frontotemporal), sometimes multifocal. No pathognomonic pattern. → electrophysiology (EEG).
• MRI: typically normal or nonspecific — helps exclude structural mimics rather than confirm.
• Biomarkers: reduced serum/plasma allopregnanolone (allopregnanolone sulfate, Allo-S) is a mechanistic biomarker and was used to stratify patients in the ganaxolone trial — a candidate but not yet a validated clinical diagnostic. → CHEBI: allopregnanolone CHEBI:50169.
• Clinical criteria / differential diagnosis: no formal consensus criteria; diagnosis is gene-confirmed pattern recognition. Differentials: Dravet syndrome (SCN1A) — the top mimic; SCN1B, GABRG2, STXBP1, CDKL5 and other DEEs; febrile seizures plus (GEFS+). The female sex + clustered fever-seizures + SCN1A-negative triad is the key clinical flag.
• Screening: not on newborn-screening panels; cascade/carrier testing in families (identifying transmitting males and at-risk female relatives) and prenatal/PGT for known familial variants are the relevant screening applications.

11. Outcome / Prognosis

• Survival/mortality: generally not life-limiting for most; lifespan is broadly preserved. SUDEP risk exists as with other DEEs but PCDH19-CE is not characterized by high early mortality. (No robust disease-specific mortality rate published — flag as gap.)
• Seizure prognosis: frequently improves with age — ~28% seizure-free by ≥11 yr, offset mean ~17.6 yr (§8).
• Neurodevelopmental/psychiatric prognosis is the real story: ID (in the ~44% who have it), ASD (~62%), executive dysfunction (~63%) persist, and adult-onset psychiatric illness (including psychotic/schizophrenia-spectrum disorders) is an increasingly recognized late outcome. So the trajectory is often "seizures fade, the neurodevelopmental/psychiatric load stays."
• Prognostic factors: earlier onset and higher seizure frequency predict worse cognitive/behavioral outcome (Kolc 2020, robust associations). Genotype–phenotype correlation is otherwise weak.
• QoL/morbidity: high family/behavioral burden (§3).

12. Treatment

There's no cure and no single reliably effective drug — the honest framing from the 2025 systematic review title is literally "Tough to treat." Management is cluster-abortive + broad-spectrum ASM + emerging neurosteroid.

• Antiseizure medications (broad-spectrum, polytherapy typical):
• Clobazam / benzodiazepines — among the more consistently reported effective agents, both for maintenance and acute clusters. → MAXO pharmacotherapy; CHEBI clobazam CHEBI:31413.
• Bromide (potassium bromide) — repeatedly noted as notably effective for PCDH19 clusters (an old drug finding a niche). → CHEBI bromide CHEBI:15858.
• Corticosteroids / ACTH — used for acute cluster interruption (case series suggest benefit during severe clusters). → MAXO; CHEBI corticosteroid class.
• Levetiracetam, valproate, topiramate, stiripentol (single-case benefit), and the ketogenic diet are used with variable/individual response. → MAXO:0000088 dietary intervention (ketogenic diet).
• Sodium-channel blockers are less consistently helpful here than in some epilepsies (and unlike in Dravet, are not strictly contraindicated — an important differential-management point).
• Neurosteroid / targeted (mechanism-driven):
• Ganaxolone — synthetic allopregnanolone analogue, positive allosteric GABA_A modulator, directly targeting the allopregnanolone-deficiency arm. The VIOLET phase-2 RCT (NCT03865732) showed a larger seizure reduction vs placebo (−61.5% vs −24.0%) that did not reach significance (p=0.17) (Epilepsy Res 2023, PMID:36870093). Somnolence was the main AE. → CHEBI ganaxolone; therapeutic_modality: SMALL_MOLECULE; a good candidate for a target_mechanisms link back to the E/I-imbalance / GABA_A node.
• Acute cluster / rescue: benzodiazepine rescue protocols; aggressive fever management as prophylaxis against triggers.
• Advanced/experimental: no approved gene/RNA therapy; the mosaic mechanism makes gene-replacement conceptually tricky (you can't just flood every cell with wild-type without recreating a "uniform" state). Neurosteroid pharmacology remains the most active targeted avenue.
• Supportive/rehabilitative: developmental therapies, ASD/behavioral management, psychiatric surveillance into adulthood, genetic counseling (MAXO:0000079).

Suggested MAXO: MAXO:0000058-type pharmacotherapy / NCIT:C15986 Pharmacotherapy with CHEBI therapeutic_agent per the repo's therapeutic-agent pattern; MAXO:0000088 (ketogenic/dietary); MAXO:0000079 (genetic counseling); MAXO:0000950 (supportive care).

13. Prevention

• Primary prevention: none for occurrence (largely de novo). The actionable analog is trigger avoidance — proactive fever/illness management to reduce clusters (a form of tertiary prevention of seizure events).
• Secondary prevention: early genetic diagnosis (panel/WES in SCN1A-negative girls with clustered febrile seizures) to enable prompt cluster-management strategies.
• Tertiary prevention: aggressive cluster abortion (benzodiazepines/steroids), developmental and psychiatric surveillance to catch adult-onset psychiatric decompensation early.
• Genetic prevention: cascade testing, prenatal diagnosis and preimplantation genetic testing for known familial variants; counseling around transmitting-male fathers (all daughters at risk) and parental gonadal mosaicism recurrence risk.
• Genetic counseling is central (MAXO:0000079). No immunization or public-health/environmental prevention applies.

14. Other Species / Natural Disease

• Taxonomy: Homo sapiens (NCBITaxon:9606) is the disease species. No recognized naturally occurring PCDH19 clustering epilepsy in companion animals or wildlife (OMIA has no established natural-disease entry) — this is essentially a human-defined disorder.
• Orthologs: Pcdh19 is evolutionarily conserved across vertebrates — mouse (Pcdh19, NCBI Gene), rat, and zebrafish (pcdh19, a well-studied ortholog with roles in neuronal columnar organization and retinal/neural development). Conservation of the adhesion function is high.
• Comparative biology: the conserved PCDH19–N-cadherin adhesion and neurodevelopmental roles make cross-species modeling informative; the mosaic-dependence is the feature that must be engineered (it doesn't arise "naturally" in uniform-genotype animals).
• Transmission: not applicable (non-infectious, non-zoonotic).

15. Model Organisms

• Mouse (Mus musculus, MGI): Pcdh19 knockout / mosaic models are the workhorses. The landmark finding — abnormal cell sorting segregating wild-type vs mutant cortical cells in mosaic (heterozygous female) mice — recapitulates the human sex-specificity logic (Pederick et al., Neuron 2018, PMID:29429936 — verify). A 2025 mouse model shows hyperthermic (fever) seizure susceptibility and focal loss of parvalbumin interneurons (biorxiv 688097), directly modeling fever sensitivity + E/I substrate. Conditional/mosaic-inducible designs are needed because uniform nulls (like hemizygous males) don't show the phenotype — the model must be mosaic.
• Rat: a focal-mosaic Pcdh19 rat reproduces brain developmental abnormalities and behavioral phenotypes (PMC9070467).
• Zebrafish (Danio rerio, ZFIN): pcdh19 mutants show neurodevelopmental/columnar-organization defects — good for adhesion/neurogenesis biology.
• Xenopus: a Pcdh19 frog model reproduces epilepsy-like and repetitive behaviors and argues for mosaicism-independent contributions (PNAS 2024) — a useful counterweight to a pure cellular-interference view.
• Cellular / in vitro: patient-derived iPSCs, 2D neural cultures, cortical/brain organoids (PMC8268119, PMC11024992, PMC8998847) — these captured accelerated neurogenesis, asymmetric-division shift, mitotic-spindle/centrosome defects, altered calcium signaling, and abnormal cell sorting in mixed populations. A Neurogenin-2-induction iPSC model (Alaverdian et al., Epileptic Disord 2023, epd2.20065) is a recent addition. These are the natural home for evidence_source: IN_VITRO; the mouse/rat/zebrafish/Xenopus data are MODEL_ORGANISM.
• Model characteristics / limitations: models recapitulate cell-sorting, neurogenesis, fever-susceptibility, and behavior, but no single model reproduces the full human syndrome (seizure semiology + variable ID/ASD + adult psychiatric risk). The human-specific neurodevelopmental context (and the difficulty of matching human XCI mosaic ratios) is a recognized translational gap — a good candidate for a HUMAN_MODEL_MISMATCH discussion block rather than a plain knowledge gap.

Curation notes / suggested KB scaffolding

• Identifiers to seed: disease_term → MONDO:0010246; causal gene → hgnc:14270; OMIM *300460 / #300088; ORPHA:101039.
• Module conformance: strong fit for epilepsy_excitation_inhibition_imbalance at #Excitation-Inhibition Imbalance, with PCDH19-specific upstream nodes (mosaic cellular interference → abnormal cell sorting → allopregnanolone/GABA_A deficit).
• Inheritance block: bind HP:0001417 X-linked inheritance but add a discussions KNOWLEDGE_GAP noting the female-limited, mosaic-dependent, cellular-interference expression that standard X-linked terms don't capture. The germline/gonadal-mosaicism recurrence route is worth an explicit note.
• Prevalence: qualitative RARE / measure_type: UNKNOWN, notes = "second most common monogenic epilepsy gene after SCN1A; ~25% of SCN1A-negative Dravet-like females."
• Ganaxolone treatment: clinical-trial block citing NCT03865732 + target_mechanisms to the GABA_A/E-I node; therapeutic_agent allopregnanolone-analogue (CHEBI).
• Human-model-mismatch discussion: the mosaicism-independent Xenopus findings vs the cell-sorting mouse model make a genuine open question about which mechanism dominates in human disease.

PMIDs to verify before use (leads, per project SOP): 32366910 (Kolc phenotype spectrum — fetched, solid), 36870093 (ganaxolone VIOLET — solid), 19752159 (Depienne de novo — from PubMed link), 34201522 (protocadherin structure review — from source, verify), 29429936 (Pederick cell-sorting — verify digits), 18469813 (Dibbens 2008 EFMR discovery — verify), 19282977 (Depienne 2009 PLoS Genet — verify), 18216183 (Scheffer 2008 Brain EFMR — verify), 30542067 (Kolc 2019 meta-analysis — verify), Tan et al. 2015 allopregnanolone (find exact PMID). Run each through just fetch-reference and confirm the snippet is an exact abstract substring.

Sources: - OMIM #300088 (DEE9) · OMIM *300460 (PCDH19) · Orphanet: PCDH19 - MalaCards: DEE9 - Kolc et al. 2020, standardized phenotype spectrum (PMC7198503) · Transl Psychiatry version - Dissecting PCDH19 in clustering epilepsy — iPSC neurogenesis (PMC8268119) · Cortical organoid cell-sorting model (PMC11024992) · 2D→3D organoid modeling (PMC8998847) - Trivisano et al., pathophysiology & clinical significance, Epilepsy & Behavior 2024 - Ganaxolone VIOLET phase-2 trial (PMID:36870093) · Neurosteroids & PCDH19 (IJMS 2021) - "Tough to treat" systematic review, Seizure 2025 - Mosaicism-independent mechanisms, Xenopus (PNAS 2024) · Focal-mosaic rat model (PMC9070467) · Parvalbumin interneuron mouse preprint 2025 - Depienne et al., de novo/single-generation EFMR (PMID:19752159) · Sporadic IEE resembling Dravet (PMC2633044) · Mutations/deletions in PCDH19 (PMC3033517) - Mosaic males: genotype & VAF (PMC9669318) · PCDH19 triplication mechanism (PMC11506946) · Epilepsiome: PCDH19 primer

One honest caveat before you build the entry: I couldn't get OMIM's and MedLink's pages to open directly (they threw 403s at the fetcher), so the OMIM-derived facts here lean on the MONDO record and the search summaries rather than the primary OMIM text — worth a manual glance when you're pinning down the original-reference PMIDs.