Solving the etiology of Developmental and Epileptic Encephalopathy with Spike-Wave activation in Sleep (D/EE-SWAS)

Viswanathan S, Oliver KL, Regan BM, et al. (senior authors Ingrid E. Scheffer and Heather C. Mefford) — Annals of Neurology 96(5):932–943, November 2024. DOI 10.1002/ana.27041. NIH PMC author manuscript PMC11534527.

Findings summary

Largest published etiological cohort to date of patients meeting the 2022 ILAE definitions of DEE-SWAS and EE-SWAS (the umbrella diagnosis that replaces the older epilepsy-aphasia spectrum: E-CSWS, ABFE, Landau-Kleffner, and unclassified EAS).

Cohort structure

• Core cohort (Melbourne Epilepsy Genetics database): n=91 (44 DEE-SWAS, 47 EE-SWAS).
• Expanded cohort (adds 10 collaborator-referred solved cases): n=101.
• Median age 17.8 years (range 5–33); 53% male; one death at age 12 (Rett/MECP2).

Epileptology (n=101)

• 92% had seizures (mean onset 4 years). 8% had SWAS without seizures.
• Mean SWAS onset 6.3 years, 28 months after seizure onset. Mean SWAS duration 3.8 years.
• Convulsive SE in 9; non-convulsive SE in 16.
• 3/101 evolved from DEE-SWAS to Lennox-Gastaut syndrome (tonic seizures and paroxysmal fast activity arising 3–19.5 years after SWAS detection).

Developmental trajectory

• All 101 regressed or plateaued in at least one domain; 14 regressed across all four (speech/language, cognition, behavior, motor).
• Speech and language was the most commonly affected domain (77%), then cognition (60%).
• 50/101 had behavioral regression, 13 of whom had no speech/language regression.

DEE-SWAS vs. EE-SWAS

• Similar seizure rates (90% vs. 94%) and similar regression patterns.
• DEE-SWAS: earlier seizure onset (3.3 vs. 4.4 y), longer epilepsy duration (median 10.0 vs. 5.2 y; p=0.00013), longer SWAS duration (median 8.9 vs. 5.8 y; p=0.071), and worse intellect (49% moderate-to-profound ID vs. 8%; adjusted p=6.1e-5).
• Higher etiological yield in DEE-SWAS: 29/44 (66%) vs. 13/47 (28%) in EE-SWAS (adjusted p=0.007). Genetic yield: 55% vs. 15%.
• LKS reclassification: of 10 original-LKS patients, 3 became DEE-SWAS (pre-existing speech delay) and 7 remained EE-SWAS.

Genetic findings (Expanded cohort, 40/101 with a pathogenic variant)

• 20 single genes in 32 patients; 6 CNVs in 7 patients; 1 chromosomal abnormality.
• De novo in 24/40; inherited in 12/40 (high intrafamilial penetrance: 26/28 carrier relatives affected).
• GRIN2A is the single most common cause (10 probands; 23% of solved Core cases). Most GRIN2A variants were inherited (7/10). 5/6 GRIN2A-positive patients of normal intellect had protein-truncating variants; all 3 with moderate-to-severe ID had missense variants — but the genotype–phenotype relationship is imperfect (one p.Thr531Met patient had severe ID despite prior LoF characterization).
• 10 novel D/EE-SWAS genes: ATP1A2, CACNA1A, FOXP1, GRIN1, KCNMA1, KCNQ3, PPFIA3, PUF60, SETD1B, ZBTB18.
• 10 previously described genes: ARID1B, CNKSR2 (X-linked recessive; 3 probands), CUL4B (X-linked recessive), GRIN2A, GRIN2B, KCNH5, MECP2 (X-linked dominant), NPRL2, SCN1A, SCN2A (2 probands).
• 2 novel CNVs: 17p11.2 duplication not involving RAI1 (new CNV cause distinct from Potocki-Lupski syndrome), and 5q22 deletion (14.5 Mb including APC and KCNN2).
• Previously associated CNVs in the series: 15q duplication (Dup15q), 16p11.2 duplication, Xp11.23-p11.22 duplication, 4p16 deletion, 1q32 duplication, isodicentric chromosome 15.

Functional grouping (brain gene co-expression analysis)

• 18/20 D/EE-SWAS genes fall into two significantly co-expressed clusters in BrainSpan data (Monte Carlo p=0.0002 and p=0.04).
• Cluster 1 (ion channels + scaffolding + adhesion): GRIN2A, GRIN2B, KCNH5, KCNQ3, CACNA1A, GRIN1, PPFIA3, SCN2A, CNKSR2, SCN1A, KCNMA1.
• Cluster 2 (transcriptional regulators / chromatin modifiers): FOXP1, PUF60, CUL4B, MECP2, ARID1B, SETD1B, ZBTB18.
• ATP1A2 and NPRL2 are outliers to both clusters (NPRL2 is in the GATOR1 complex and regulates mTOR).

MRI findings (Core cohort, 88 with MRI)

• 82% (72/88) had normal MRI. 16 were abnormal: 5 malformations of cortical development (3 bilateral perisylvian PMG, 2 unilateral PMG), 7 likely-acquired (5 unilateral non-progressive thalamic lesions; 2 post-hemorrhagic hydrocephalus with VP shunts), 4 incidental (3 DVAs, 1 minor peritrigonal WM signal).
• EEG lateralization was concordant with MRI findings in 9/12 structurally abnormal cases; in unilateral lesions, discharges were ipsilateral in wakefulness and bilaterally synchronous in sleep.

Nine recognized syndromes in the cohort

• Novel for D/EE-SWAS: 5q22 deletion, FOXP1-DEE, Verheij syndrome (PUF60), Potocki-Lupski syndrome (17p11.2 duplication with RAI1).
• Previously described: Dup15q, 16p11.2 duplication, Xp11.23-p11.22 microduplication, Rett (MECP2), Coffin-Siris (ARID1B).

Relevance to Levi

This paper is the closest published match on the specific question of "what etiologies account for DEE-SWAS, and how likely is each?" It is directly downstream of the ILAE 2022 reclassification that defines Levi's diagnosis. Several observations matter for Levi's current differential:

1. Overall etiological yield is informative for expectation-setting. In the Core cohort, 46% of patients were etiologically solved. In the DEE-SWAS subset (Levi's diagnosis), 66% were solved, with 55% having an identified genetic cause. This is higher than Levi's current standing — three adequate germline sequencing runs have returned negative. The paper's cohort is heavily enriched for referred, workup-able cases and Melbourne's Epilepsy Genetics research platform, which probably overestimates the yield for a typical clinical trio. Still, the gap (55% genetic yield in this cohort vs. 0 solved in Levi after three negative runs) reinforces that Levi is etiologically in the harder-to-solve minority and supports continuing the workup beyond standard germline testing (mosaic-sensitive tissue sequencing, methylation / episignature, somatic re-calling of existing BAMs).

2. Modest negative signal for the mTOR-axis theories. Of the 20 genes implicated in this 101-patient cohort, only one — NPRL2 (GATOR1 complex, an mTOR regulator) — sits in the PI3K-AKT-mTOR axis, and that patient also had polymicrogyria. None of the other classical PROS / mTORopathy genes (PIK3CA, PTEN, MTOR, AKT1/3, TSC1, TSC2, DEPDC5, NPRL3) appeared. The paper's overgrowth phenotype is not separately characterized, so Levi's overgrowth + DEE-SWAS may be under-represented. Still, the underrepresentation of the mTOR axis in the largest solved DEE-SWAS cohort to date is a mild negative signal for the germline-mTOR and mosaic-mTOR theories as the specific driver of Levi's DEE-SWAS, distinct from the overgrowth phenotype. The mosaic branch remains defensible because this cohort largely (82%) had normal MRIs, and high-VAF mosaic mTORopathies typically have structural MRI signatures that would have been detected.

3. Strengthens the chromatinopathy / transcriptional-regulator branch. Seven of the 20 D/EE-SWAS genes are transcriptional regulators (FOXP1, PUF60, CUL4B, MECP2, ARID1B, SETD1B, ZBTB18) and they form a significantly co-expressed brain cluster. Three of these (MECP2/Rett, ARID1B/Coffin-Siris, and — via SETD1B — the broader chromatin-writer axis) sit adjacent to the DNMT3A / TBRS mechanism already carved out as a dedicated theory in Levi's differential. SETD1B in particular is now explicitly on the EpiSign panel. This is a genuine positive for the existing "order an EpiSign / chromatinopathy episignature panel" diagnostic (rank 2).

4. GRIN2A is the single highest-yield DEE-SWAS gene and deserves explicit confirmation. GRIN2A accounted for 23% of genetically solved Core cases. It is mechanistically central to DEE-SWAS (speech dyspraxia, epilepsy-aphasia spectrum, loss-of-function channelopathy of the NMDA alpha-2 subunit). Levi's three negative germline workups (Stanford trio exome 2025-05-29, GeneDx trio WGS 2026-01-29, GeneDx reanalysis 2026-04-09) should already have covered GRIN2A at standard exome depth, and the reanalysis was HPO-guided. Explicit confirmation in the Stanford / GeneDx reports that GRIN2A coding and splice-site regions had adequate coverage is worth a small follow-up question for the family — this is the only gene in the 2024 Viswanathan cohort where a missed call would materially change Levi's working differential.

5. GRIN-related disorders have a candidate targeted therapy (L-serine) worth noting. Viswanathan 2024 cites Krey 2022 on L-serine treatment in GRIN-null variants producing improvements in behavior, EEG, and seizure frequency. This is a contingent therapeutic lever if GRIN2A, GRIN1, or GRIN2B is ever identified in Levi.

6. LGS evolution is a real longitudinal risk. 3/101 patients in the cohort evolved to Lennox-Gastaut syndrome 3–19.5 years after SWAS detection, with tonic seizures and paroxysmal fast activity arising late. All three became drug-resistant with moderate-to-profound ID. This doesn't change acute management but sharpens the prognostic-counseling caveat in Levi's case.

7. MRI-normal DEE-SWAS is the majority. 82% of the Core cohort had a normal MRI. Levi's structurally unremarkable MRI is typical and does not by itself argue for or against a genetic etiology.

8. Structural etiologies in D/EE-SWAS cluster on the thalamocortical network. Bilateral perisylvian polymicrogyria (3), unilateral polymicrogyria (2), unilateral non-progressive thalamic lesions (5), and post-hemorrhagic hydrocephalus with VP shunts (2) account for all 12 structural cases. The thalamus is called out as central to the SWAS-generating network. Levi's MRI has none of these, which continues to down-weight the structural-vascular-perinatal theory as currently ranked (3%).

9. Direct outreach candidates. Ingrid Scheffer (senior author; Austin Health / University of Melbourne) and Heather Mefford (co-senior author; St Jude Children's Research Hospital) are the two highest-authority DEE-SWAS etiology researchers globally. Mefford has an overlapping interest in mosaic mTORopathies via the Seattle / University of Washington Division of Genetic Medicine axis (which also includes Ghayda Mirzaa, already in Levi's people workspace). This warrants adding Mefford and/or Scheffer as outreach targets if the family is open to it.

Provenance

• Ingested 2026-04-18 from a Slack-uploaded PDF: raw/uploads/slack/2026-04-18/1776499409858-nihms-2009856.pdf.
• Paper is the author-manuscript version (NIHMS-2009856) of Viswanathan et al., Ann Neurol 96(5):932–943 (November 2024), DOI 10.1002/ana.27041, published online as PMC11534527.
• Companion memo: content/research/notes/2026-04-18-viswanathan-dee-swas-etiology.md.