Childhood Absence Epilepsy

1. Disease Information

Childhood absence epilepsy is one of those conditions where the metaphor writes itself: the kid is a radio that keeps dropping the signal for a few seconds, then comes right back mid-song like nothing happened. It's a genetic (idiopathic) generalized epilepsy of childhood, defined by frequent typical absence seizures — abrupt, brief lapses of awareness (a blank stare, arrest of activity, maybe a flutter of the eyelids), starting and stopping like a light switch, no aura, no post-ictal fog, often dozens to hundreds of times a day. On EEG it has a fingerprint: generalized, bilaterally synchronous, ~3 Hz (2.5–4 Hz) spike-and-wave discharges on a normal background, classically provoked by hyperventilation.

It's common as childhood epilepsies go — roughly 10–17% of all epilepsy diagnosed in school-age kids — and the prognosis is, refreshingly, mostly good.

Key identifiers: - MONDO: MONDO:0010826 (childhood absence epilepsy) — verified locally via OAK - OMIM: The disorder is genetically heterogeneous, split across susceptibility loci rather than one Mendelian entry: - %600131 ECA1 (mapped to 8q24) - #607681-linked ECA2 → GABRG2 (5q31.1) - ECA4 (#611136) → GABRA1 (5q34) - ECA5 (#612269) → GABRB3 (15q12) - ECA6 (#611942) → CACNA1H (16p13) - ICD-10: G40.A (Absence epilepsy syndromes, in current coding; older mapping G40.3, generalized idiopathic epilepsy). ICD-11: 8A62 (typical absence seizures) within the generalized epilepsy block. - MeSH: D004832 ("Epilepsy, Absence") - Orphanet: ORPHA:64280 (Childhood absence epilepsy)

Common synonyms: pyknolepsy / pyknoleptic epilepsy (the old term — "pyknos" = crowded/frequent, referring to the sheer number of seizures a day), petit mal epilepsy (obsolete, discouraged), CAE.

Data provenance: This entry draws almost entirely on aggregated disease-level resources — ILAE syndrome definitions, OMIM/Orphanet, randomized trials, and cohort studies — not individual EHR-derived patients.

2. Etiology

Causal factors — it's polygenic, not a single broken gene. CAE is the poster child for complex/polygenic inheritance: most cases don't come from one high-penetrance mutation but from a combination of common variants nudging thalamocortical circuits toward oscillation. Think of it less like a single snapped guitar string and more like a whole orchestra tuned slightly sharp. A minority of families show single-gene contributions (GABA-A receptor subunit genes, T-type calcium channel genes), but even those behave as susceptibility factors with incomplete penetrance rather than clean Mendelian causes.

Genetic risk factors: - GABA-A receptor subunit genes: GABRG2 (HGNC:4402), GABRA1 (HGNC:4075), GABRB3 (HGNC:4083) — variants impairing inhibitory GABAergic signaling. - T-type calcium channel genes: CACNA1H (Cav3.2, HGNC:1395) — >12 non-synonymous variants found preferentially in CAE patients (PMID:15888660, Vitko et al., J Neurosci 2005: functional variants that would "increase firing of neurons" in thalamocortical models). CACNA1G (Cav3.1) is the animal-model archetype. - Chloride channel: CLCN2 (HGNC:2020) — historically implicated in IGE, now largely disputed/downgraded as a monogenic cause; flag with caution. - SLC2A1 (GLUT1, HGNC:11005): a small but clinically pivotal fraction (~10% of early-onset, <4 yr absence epilepsy; ~1% of general IGE) are actually GLUT1 deficiency syndrome masquerading as CAE — see Diagnostics.

Environmental / demographic risk factors: age (the 4–10 yr window is itself the biggest "risk factor"), female predominance (~60–70% girls in most series), and family history of generalized epilepsy in first-degree relatives. Hyperventilation is a reliable provocateur (not a cause). No robust toxic, infectious, or occupational exposure is established.

Protective factors: No validated genetic protective alleles. On the environmental side, the strongest "protective" lever is simply correct drug choice (ethosuximide/valproate over lamotrigine) and avoidance of seizure-aggravating drugs (carbamazepine, oxcarbazepine, phenytoin, vigabatrin, gabapentin can worsen absence).

Gene–environment interaction: Modest and poorly mapped for CAE specifically. The clearest example of a modifiable metabolic modifier is the GLUT1 subgroup, where the ketogenic diet bypasses the transporter defect — a genotype that dictates an environmental (dietary) intervention.

3. Phenotypes

The phenotype list is short, stereotyped, and where CAE earns its clinical elegance. Suggested HP terms verified locally via OAK.

Onset: childhood, 4–10 yr, peak 5–7 yr (HP:0011463 Childhood onset). Severity: individual seizures are mild and self-limited, but the disease severity is driven by seizure frequency (can impair schooling) and comorbidity burden. Progression: episodic seizures; not neurodegenerative — no developmental regression (explicitly absent; HP:0002376 would be a red flag arguing against CAE). Quality-of-life impact: disproportionate to the benign-looking seizures. Caplan et al. (Epilepsia 2008, PMID:18557780) reported "61% had a psychiatric diagnosis, particularly ADHD and anxiety disorders," and long-term cohorts describe "poor psychiatric, social, and vocational adult outcomes." The learning cost of blanking out 100×/day in a classroom is real even when the neurology looks tidy.

4. Genetic / Molecular Information

Causal / susceptibility genes (all susceptibility-weighted, not deterministic):

Variant classification & type: predominantly missense SNPs (especially CACNA1H, GABR subunits), with GLUT1DS additionally showing nonsense, frameshift, splice, and whole-gene deletions. ACMG interpretation is fraught here — many CACNA1H variants are best classified as risk alleles / VUS-to-low-penetrance rather than clean pathogenic calls, because they recur in a polygenic background and often appear at appreciable frequency in gnomAD. Contrast with SLC2A1 GLUT1DS variants, which are frequently de novo* and confidently pathogenic/likely-pathogenic.

Functional consequence — the unifying theme: either too little inhibition (GABA-A subunit LoF) or too much low-threshold burst excitability (T-type Ca²⁺ gain), both converging on the thalamocortical loop's tendency to oscillate at ~3 Hz. Vitko et al. (PMID:15888660): computer modeling predicted several CACNA1H variants "would increase firing of neurons, with three of them inducing oscillations at similar frequencies, as observed during absence seizures."

Modifier genes: poorly defined; the polygenic architecture means "modifier" and "susceptibility" blur together.

Epigenetics / chromosomal: GABRB3 sits in the imprinted 15q11–q13 (Angelman/Prader-Willi) region, making dosage/imprinting biologically interesting, but no consistent CAE-specific methylation or large-scale cytogenetic signature is established. CAE is not a copy-number/aneuploidy disorder.

5. Environmental Information

Thin section, honestly — CAE is a channel/circuit disease, not an exposure disease. - Environmental factors: no established toxin, radiation, or pollutant cause. Hyperventilation is the classic seizure trigger (and diagnostic provocation); photic stimulation triggers a minority. - Lifestyle factors: sleep deprivation and hyperventilation can precipitate events; no dietary or activity cause. (Ketogenic diet is therapeutic in the GLUT1 subset — an environmental modifier, not a cause.) - Infectious agents: none. CAE is not post-infectious or para-infectious.

6. Mechanism / Pathophysiology

Here's the good stuff — the causal chain, because CAE is arguably the best-understood circuit epilepsy we have.

The circuit: absence seizures are generated by the thalamocortical loop — a reciprocal three-way conversation between (1) cortical pyramidal neurons, (2) thalamic relay (thalamocortical) neurons, and (3) the GABAergic reticular thalamic nucleus (nRT), the loop's inhibitory gatekeeper. Normally this loop produces sleep spindles. In CAE it gets hijacked into pathological, hypersynchronous ~3 Hz spike-wave oscillations.

Causal chain (upstream → downstream):

1. Trigger (upstream): A genetic tilt in excitability — either reduced GABA-A inhibition (GABRG2/A1/B3) or enhanced T-type (low-voltage-activated) Ca²⁺ current in nRT and relay neurons (CACNA1H/CACNA1G). Reduced tonic GABA-A inhibition combined with excessive tonic GABA-B receptor activation on relay neurons is a recurring finding across models.
2. Cellular mechanism: T-type Ca²⁺ channels (Cav3.x) mediate low-threshold calcium spikes that let neurons fire in rhythmic burst mode rather than tonic mode. When this burst tendency is amplified, nRT and thalamocortical cells lock into synchronized oscillation. Thalamocortical circuit reviews describe "synchronous reciprocal excitation between the neocortex and thalamus, with inhibitory neurons in the reticular thalamic nucleus and excitatory thalamocortical neurons being key players in generating" spike-wave discharges.
3. Network mechanism: The oscillation propagates bilaterally and synchronously across cortex → generalized 3 Hz spike-wave on EEG → behavioral absence (impaired consciousness) for its duration.
4. Clinical manifestation (downstream): the brief blank-out, ending as abruptly as it began when the oscillation terminates.

Why the drugs work — mechanism confirms the model: ethosuximide blocks T-type Ca²⁺ channels (and reduces persistent Na⁺ current), directly damping the low-threshold burst engine — which is exactly why it's first-line and why the T-type story is more than correlation. Valproate has broad action (↑GABA, Na⁺/T-type modulation). Conversely, drugs that enhance GABA-B tone or block Na⁺ channels selectively (carbamazepine, phenytoin, vigabatrin, tiagabine) can paradoxically worsen absence — the flip side of the same circuit logic.

Suggested ontology terms: - Biological processes (GO): GO:0051899 membrane depolarization; GO:0070588 calcium ion transmembrane transport; GO:0007268 chemical synaptic transmission; GO:1902476 chloride transmembrane transport; GO:0060080 inhibitory postsynaptic potential; GO:0001508 action potential. - Cell types (CL): CL:0000679 glutamatergic neuron (thalamocortical relay / cortical pyramidal); CL:0000617 GABAergic neuron (reticular thalamic nucleus); CL:0000598 pyramidal neuron. - Cellular components (GO CC): GO:0005891 voltage-gated calcium channel complex; GO:1902711 GABA-A receptor complex; GO:0045211 postsynaptic membrane.

Molecular profiling: No robust human transcriptomic/proteomic/metabolomic signature for CAE specifically — the disease is defined electroclinically, and molecular insight comes overwhelmingly from rodent models (see §15), not human -omics. This is a genuine knowledge gap worth flagging in the entry.

7. Anatomical Structures Affected

• Organ / system level: central nervous system, specifically the thalamocortical network. No systemic organ involvement — this is a functional circuit disorder in a structurally normal brain (normal MRI is expected).
• Primary structures (UBERON):
• UBERON:0001897 dorsal thalamus / thalamus (relay neurons)
• Reticular thalamic nucleus (the GABAergic pacemaker) — UBERON:0002733 reticular nucleus of thalamus
• UBERON:0000956 cerebral cortex (neocortex; frontal/perirolandic onset emphasis)
• Tissue/cell level: neuronal (glutamatergic relay + cortical pyramidal; GABAergic nRT interneurons) — see CL terms above. No gliosis, no neuronal loss, no fibrosis.
• Subcellular: the plasma membrane / voltage-gated ion channel complexes (T-type Ca²⁺ channels, GABA-A receptors) at the synapse — GO:0005886 plasma membrane, GO:0045202 synapse.
• Localization / lateralization: bilateral and synchronous by definition (generalized, not focal). Any consistently focal or lateralized feature argues against CAE.

8. Temporal Development

• Onset: childhood, 4–10 yr, peak 5–7 yr; onset before 4 yr should trigger a GLUT1 (SLC2A1) workup. Onset pattern is subacute/insidious — often first noticed as "daydreaming" or inattention at school before anyone realizes they're seizures.
• Course: frequent daily absences during the active period; episodic seizures on a stable, non-progressive baseline. No stages in the oncologic sense.
• Duration & remission: typically self-limiting over childhood/adolescence. Remission rates across cohorts span ~56–84%, with roughly 65% in long-term remission; many are successfully weaned off medication after a few seizure-free years (one cohort: treatment ceased in 79.2% after mean 3.2 yr).
• Progression risk: ~10–15% evolve to juvenile myoclonic epilepsy (JME) or develop GTCS, a lifelong-epilepsy trajectory (classic natural-history finding, Wirrell et al., Neurology 1996 (verify PMID:8857720)).
• Critical window: the school-age years are both the vulnerability window and the intervention window — controlling seizures early matters for the cognitive/academic trajectory.

Poor-prognosis predictors (from cohort follow-up): cognitive difficulty at diagnosis, absence status epilepticus, emergence of GTCS/myoclonic seizures after treatment onset, abnormal EEG background, and family history of generalized seizures in first-degree relatives.

9. Inheritance and Population

• Epidemiology: CAE accounts for ~10–17% of school-age epilepsy and 2–8 per 100,000 children as an incidence estimate; it's among the more common pediatric epilepsy syndromes. (Prevalence figures vary by ascertainment; treat as RARE-to-common band and cite the specific cohort when curating.)
• Inheritance pattern: complex / polygenic / multifactorial — this is the headline. Monogenic families exist (GABA-A subunit, T-type Ca²⁺) but are the exception. Use HP:0000007 cautiously; the honest MOI is multifactorial. For the GLUT1 phenocopy subset, SLC2A1 is typically autosomal dominant / de novo.
• Penetrance / expressivity: incomplete penetrance, variable expressivity are the rule; the same GABRG2/CACNA1H variant can produce CAE, another IGE subtype, febrile seizures, or nothing.
• Anticipation / mosaicism / founder effects: not features of CAE (no repeat-expansion mechanism).
• Consanguinity: not a notable driver (polygenic, not recessive).
• Demographics: female predominance (~60–70%); no strong ethnic enrichment; onset-age distribution tightly clustered 4–10 yr.

10. Diagnostics

The diagnosis is fundamentally electroclinical — a characteristic child + a characteristic EEG.

• Electrophysiology (the linchpin): EEG showing generalized, bilaterally synchronous 2.5–4 Hz spike-and-wave on normal background, classically elicited by 3–5 min hyperventilation (which reliably provokes an absence in the office — dramatic and diagnostic). Per ILAE 2022: "An ictal EEG is not required for diagnosis, provided the interictal study shows paroxysms of 2.5–4-Hz generalized spike-wave discharge during wakefulness." HP:0010848 / HP:0011182 (Interictal epileptiform activity).
• Clinical diagnostic criteria (ILAE 2022, Hirsch et al., Epilepsia 2022 — PMID:35503716 (verify)): CAE is defined among the four IGE syndromes (CAE, juvenile absence epilepsy, JME, GTCS-alone). Mandatory features: onset 4–10 yr, typical absences, characteristic EEG; exclusionary "alerts" include developmental regression, focal features, prominent myoclonus, or an abnormal background — any of which push you off the CAE diagnosis.
• Imaging: MRI is normal and is used to exclude structural mimics, not to confirm CAE.
• Neuropsychological testing: recommended given the attention/language comorbidity load, even when seizures are controlled.
• Genetic testing — the one that changes management: not required for routine CAE, BUT test SLC2A1 (GLUT1) in atypical or early-onset (<4 yr) absence, drug-resistant absence, or absence + movement disorder. GLUT1DS has an actual biomarker: low CSF glucose with CSF:serum glucose ratio typically <0.5 (hypoglycorrhachia), confirmed by SLC2A1 sequencing. As one source put it: genetic testing has "a pre-test probability of ~10% for early-onset absence epilepsy" for GLUT1. Broader gene panels / WES catch the GABA-A and T-type contributors but rarely change management outside GLUT1.
• Differential diagnosis: juvenile absence epilepsy (later onset, sparser absences, more GTCS), atypical absence (slower <2.5 Hz spike-wave, abnormal background → suggests Lennox-Gastaut/developmental epileptic encephalopathy), focal impaired-awareness seizures with automatisms (focal EEG, post-ictal confusion — CAE has none), daydreaming/inattention (no EEG correlate), and the crucial GLUT1DS phenocopy.

11. Outcome / Prognosis

• Survival/mortality: essentially normal life expectancy; CAE is not a mortality-driving epilepsy (SUDEP risk is low relative to other epilepsies, though not zero if GTCS emerge).
• Seizure outcome: favorable — ~65% long-term remission (range 56–84%), most weaned off medication.
• Morbidity: the durable burden is neuropsychiatric/cognitive, not seizure-related mortality — ADHD/inattention (~a quarter with subtle cognitive deficits), anxiety, language difficulty, and, in long-term follow-up, "poor psychiatric, social, and vocational adult outcomes" (Caplan et al., PMID:18557780). This is the part clinicians historically under-treated (one cohort: only 23% receiving comorbidity intervention).
• Prognostic factors: good — pure typical absences, normal cognition, prompt response to ethosuximide/valproate, normal EEG background. Poor — early cognitive difficulty, absence status, emergence of GTCS/myoclonus, JME evolution, family history of generalized seizures.

12. Treatment

This is CAE's greatest hit, because it's backed by the single best trial in the field.

The evidence base — Glauser et al., NEJM 2010 (PMID:20200383): the NIH-funded, double-blind RCT of 446 children, the "first randomized controlled trial meeting ILAE criteria for class I evidence" in absence epilepsy. Result: ethosuximide and valproate were equally effective and superior to lamotrigine (freedom-from-failure ~53% ethosuximide, 58% valproate, 29% lamotrigine), and crucially ethosuximide caused fewer attentional side effects than valproate. 12-month follow-up (Glauser 2013, PMID:23167925) confirmed the durability. Bottom line clinicians actually use: ethosuximide is first-line for pure absence (best efficacy and best cognitive profile); valproate is reserved for kids who also have GTCS (ethosuximide doesn't cover tonic-clonic); lamotrigine is third-line.

Pharmacotherapy (with MAXO/CHEBI suggestions): | Drug | Class / MoA | Role | Ontology | |---|---|---|---| | Ethosuximide | T-type Ca²⁺ channel blocker | First-line (pure absence) | CHEBI:4887 ethosuximide; treatment_term MAXO:0000058 pharmacotherapy (or NCIT:C15986) | | Valproic acid / valproate | Broad (↑GABA, Na⁺/T-type) | First-line if GTCS co-occur | CHEBI:39867 valproic acid | | Lamotrigine | Na⁺ channel; broad-spectrum | Third-line / add-on | CHEBI:6367 lamotrigine | | Ketogenic diet | Metabolic | First-line/curative in the GLUT1DS subset; option in refractory CAE | MAXO:0000089 dietary therapy / ketogenic diet | | Levetiracetam, zonisamide | adjuncts | Refractory add-on | — |

Drugs to AVOID (can worsen absence): carbamazepine, oxcarbazepine, phenytoin, vigabatrin, tiagabine, gabapentin, pregabalin. Worth a hard callout in the entry — a well-meaning wrong prescription makes it worse.

Pharmacogenomics: valproate carries POLG-related hepatotoxicity and general teratogenicity concerns (avoid in adolescent girls where possible); no CAE-specific CPIC guideline for ethosuximide/lamotrigine beyond general HLA-B*15:02/lamotrigine SCAR caution.

Advanced/experimental: no gene or cell therapy in practice. T-type Ca²⁺ channel selective blockers are the rational next frontier — e.g., CX-8998 (MK-8998) evaluated for absence seizures (T-CALM trial, NCT03406702) — directly targeting the mechanistic linchpin.

Treatment algorithm: confirm CAE electroclinically → screen for GLUT1 if atypical/early → ethosuximide first (or valproate if GTCS) → lamotrigine or dual therapy if refractory → reconsider diagnosis (GLUT1DS? JME? atypical absence?) if truly drug-resistant.

13. Prevention

CAE isn't a preventable disease in the primary sense (no vaccine, no exposure to avoid), so "prevention" here is really about early detection and complication-prevention: - Primary prevention: none available — it's a genetic circuit predisposition. - Secondary prevention: prompt EEG recognition of the "daydreaming child" so seizures (and their academic toll) are controlled early. The single most impactful "screen" is having a low threshold for GLUT1DS testing, because that subset has a disease-modifying intervention (ketogenic diet) that must start early to protect brain development. - Tertiary prevention: avoid absence-aggravating drugs; proactively screen for and treat the ADHD/anxiety/learning comorbidities (routinely under-addressed); monitor for JME/GTCS emergence. - Counseling: genetic counseling is generally reassuring given the polygenic architecture and good prognosis — recurrence risk in siblings is elevated but modest, and there's no clean single-gene test to offer most families (GLUT1DS being the AD exception).

14. Other Species / Natural Disease

• Taxonomy: the disease-as-such is human; the mechanism is studied in rat (NCBITaxon:10116) and mouse (NCBITaxon:10090) models (see §15).
• Natural disease in other species: no well-characterized spontaneous "childhood absence epilepsy" in companion animals (dogs/cats have idiopathic epilepsies, but absence-with-3Hz-spike-wave is not a defined veterinary syndrome the way generalized/focal canine epilepsy is). The rodent models are selectively bred, not naturally occurring disease in the OMIA sense.
• Comparative biology / conservation: the thalamocortical loop, T-type Ca²⁺ channels (Cacna1g/Cacna1h orthologs), and GABA-A receptor subunits are deeply evolutionarily conserved — which is exactly why rodent spike-wave discharges recapitulate the human 3 Hz mechanism so faithfully.
• Zoonosis: not applicable (non-transmissible genetic circuit disorder).

15. Model Organisms

CAE has an unusually rich and mechanistically faithful model menagerie — the reason we understand the circuit so well.

Rat models (genetic, polygenic — the best face-validity models): - GAERS (Genetic Absence Epilepsy Rats from Strasbourg) — spontaneous spike-wave discharges; carries a Cacna1h (Cav3.2) gain-of-function variant that "enhances T-type Ca²⁺ currents by altering calnexin-dependent trafficking of Cav3.2 channels" (Powell et al., Sci Rep 2017). SWDs "7–11/s… lasting 0.5–40 s, occurring hundreds of times a day, persisting throughout life." - WAG/Rij (Wistar Albino Glaxo from Rijswijk) — the most-used absence model; spontaneous SWDs, well-characterized comorbid depression-like phenotype.

Mouse models (monogenic Ca²⁺-channel-subunit mutants — great construct validity): - tottering (Cacna1a, P/Q-type α1A) · lethargic (Cacnb4, β4 subunit) · stargazer (Cacng2, stargazin/γ2) · ducky (Cacna2d2, α2δ2) · mocha · slow-wave-epilepsy (swe). The through-line: "in most cases the mutation affects a Ca²⁺ channel subunit… T-type Ca²⁺ current augmented in nRT." - Engineered α1G (Cacna1g) overexpression: elevating Cav3.1 low-voltage-activated current "induces pure absence epilepsy" (Ernst et al., J Neurosci 2009) — a clean causal demonstration that too much T-type current alone is sufficient.

Utility & limitations: these models nail the electrophysiology (SWDs, thalamocortical mechanism, drug pharmacology — ethosuximide suppresses SWDs in all of them, valproate too), which is why they're the workhorses for testing T-type blockers. Limitations: rodent SWDs run faster (7–11 Hz vs. human 3 Hz), the monogenic mouse mutants also carry ataxia/motor phenotypes not seen in human CAE, and no model fully captures the human polygenic architecture or the neuropsychiatric comorbidity profile. Per this repo's conventions, that mismatch (robust model electrophysiology vs. uncertain fidelity to the human polygenic/cognitive picture) is a candidate HUMAN_MODEL_MISMATCH discussion rather than a plain knowledge gap — evidence exists in models, but the translational validity of details (comorbidity, oscillation frequency, genetic complexity) is the open question.

Model databases: MGI (mouse mutants), RGD (GAERS/WAG-Rij rat strains), Alliance of Genome Resources for orthologs.

Key References (verify every PMID with just fetch-reference before curation)

• PMID:20200383 — Glauser TA et al. Ethosuximide, valproic acid, and lamotrigine in childhood absence epilepsy. NEJM 2010;362(9):790–799. (landmark RCT; from live search)
• PMID:23167925 — Glauser TA et al. …initial monotherapy outcomes at 12 months. Epilepsia 2013. (from live search)
• PMID:15888660 — Vitko I et al. Functional characterization and neuronal modeling of the effects of childhood absence epilepsy variants of CACNA1H, a T-type calcium channel. J Neurosci 2005;25(19):4844–4855. (from live search)
• PMID:18557780 — Caplan R et al. Childhood absence epilepsy: behavioral, cognitive, and linguistic comorbidities. Epilepsia 2008. (from live search)
• PMID:19015658 — Neuropsychiatric comorbidities in childhood absence epilepsy. (from live search)
• PMID:11994752 (verify) — Crunelli V, Leresche N. Childhood absence epilepsy: genes, channels, neurons and networks. Nat Rev Neurosci 2002. (classic mechanism review)
• PMID:35503716 (verify) — Hirsch E et al. ILAE definition of the Idiopathic Generalized Epilepsy Syndromes. Epilepsia 2022;63:1475–1499.
• PMID:8857720 (verify) — Wirrell EC et al. Long-term prognosis of typical childhood absence epilepsy (JME progression). Neurology 1996.
• Powell KL et al. The Cacna1h mutation in the GAERS model…Cav3.2 trafficking. Sci Rep 2017 (PMC5599688). (from live search)
• Ernst WL et al. Genetic enhancement of thalamocortical network activity by elevating α1G-mediated LVA Ca²⁺ current induces pure absence epilepsy. J Neurosci 2009;29(6):1615. (from live search)
• GLUT1DS/SLC2A1 early-onset absence — Suls A et al. (verify PMID:19578037); Arsov T et al., Epilepsia 2012 (verify).

Sources consulted: - OMIM 600131 – ECA1, OMIM 611942 – ECA6/CACNA1H, OMIM 612269 – ECA5/GABRB3 - Glauser NEJM 2010 - Vitko et al. J Neurosci 2005 (CACNA1H) · Ernst et al. J Neurosci 2009 (α1G) · GAERS Cav3.2, Sci Rep 2017 - Hirsch et al. ILAE IGE definition, Epilepsia 2022 - Caplan et al. Epilepsia 2008 (comorbidities) - Long-term prognosis of CAE, PMC10943932 · CAE follow-up/prognostic factors, PMC9600757 - Experimental Models of Absence Epilepsy, PMC8019851 · Thalamocortical circuits in generalized epilepsy - GLUT1 Deficiency (SLC2A1), Epilepsy Foundation · Ketogenic diet in SLC2A1, PMC8000344

That's the whole organism dissected, Harry — from the 3 Hz oscillation in the thalamocortical loop all the way out to the classroom desk where it actually costs something. A few honest gaps I'd want you to keep in mind when this becomes a KB entry: (1) the inheritance is genuinely polygenic, so resist the urge to over-annotate any single gene as "causal" — they're susceptibility loci; (2) the GLUT1/SLC2A1 phenocopy deserves its own callout because it's the one subtype where the diagnosis flips the treatment to a ketogenic diet; and (3) every (verify)-tagged PMID needs a real just fetch-reference pass before any snippet lands in YAML — I flagged those specifically because I'm citing them from memory, not from a live abstract, and this repo has zero tolerance for a hallucinated quote sneaking through.