ATP6V1A-Related Disorders (DEE93 and ARCL2D Cutis Laxa)

Two distinct diseases of the same gene encoding subunit A of the vacuolar proton pump (V-ATPase): developmental and epileptic encephalopathy 93, autosomal dominant and almost always caused by de novo missense variants, and autosomal recessive cutis laxa type 2D, caused by biallelic variants; placed within the family of V-ATPase disorders.

Literature and regulatory information updated: 28 October 2026.

0. Reference Summary Table

0.1. Profile of the two entities

0.2. Family of V-ATPase disorders (summary)

Table sources: [1, 2, 5, 12, 18, 19, 21, 28]. The full differential diagnosis, with distinguishing features and discriminating tests, is in section 6.

1. Introduction

1.1. Scope and definition of the entities

The ATP6V1A gene encodes subunit A of the cytosolic V1 domain of the vacuolar proton-translocating adenosine triphosphatase (V-ATPase), the pump that acidifies endosomes, lysosomes, secretory vesicles, and synaptic vesicles [2, 8]. Variants in this gene cause two diseases that do not overlap clinically and have different inheritance: epileptic encephalopathy is not a feature of the recessive cutis laxa, and cutis laxa is not part of the dominant encephalopathy [8]. This article covers the two entities and the gene as a unit, with its core devoted to DEE93; ARCL2D is covered in summary, with the data that resolve discrepancies between sources, without developing the treatment and surveillance of the skin, lungs, and aorta. ATP6V1A-CDG appears in the list of congenital disorders of glycosylation, and the class of developmental and epileptic encephalopathies (DEE) and the general treatment of infantile epileptic spasms syndrome are not developed.

Developmental and epileptic encephalopathy 93 (DEE93; OMIM 618012) is defined in OMIM as an autosomal dominant neurological disorder with psychomotor delay, early-onset refractory seizures, and impaired intellectual development, of highly variable severity: some patients neither speak nor walk and have spastic tetraparesis and poor eye contact, whereas others have moderate intellectual disability (ID) [16]. Series published after 2022 extended the spectrum towards the milder end, with childhood focal epilepsy of favourable course and normal development, and with febrile seizures and mild ID without epilepsy [29, 30]. Autosomal recessive cutis laxa type IID (ARCL2D; OMIM 617403) is characterised by generalised wrinkling of the skin with scant subcutaneous fat, a dysmorphic progeroid face and, in most patients, severe hypotonia with cardiovascular and neurological involvement [17].

1.2. Historical overview and molecular milestones

The first disease associated with ATP6V1A was the recessive form. The first patient, a German boy, was published in 2014 by Van Asbeck and colleagues as a novel cutis laxa phenotype with abnormal fat distribution, cardiomyopathy, and cataract; analysis of 12 genes for cutis laxa and progeroid syndromes was negative [17]. In 2017, Van Damme and colleagues identified biallelic variants in ATP6V1E1 and ATP6V1A in families with autosomal recessive cutis laxa, glycosylation abnormalities, and cardiopulmonary and neurological involvement, which gave rise to the designations ARCL2C and ARCL2D [31]. For ATP6V1A, the series included three probands from unrelated families (the German boy of 2014, homozygous for p.Arg338Cys, and two infants, one Pakistani and one Turkish, homozygous for p.Gly72Asp), and the authors proposed integrating the form into the CDG classification as ATP6V1A-CDG [11, 17]. In 2021, Vogt and colleagues described three further patients from two families, including the first truncating allele, and reviewed the six cases then known [15].

The dominant form was described in 2018 by Fassio and colleagues, from exome sequencing of 1444 probands with developmental or epileptic encephalopathies from three cohorts (a European one, from the DESIRE project, a Japanese one, and a third) [2]. They found four de novo missense variants (p.Pro27Arg, p.Asp100Tyr, p.Asp349Asn, and p.Asp371Gly), three in trio studies and one in an isolated proband, all in the "unclassified epileptic encephalopathy" subgroup; the four patients, unrelated to one another and aged 8 to 14 years at publication, were of Italian (p.Asp100Tyr), Japanese (p.Asp349Asn), Turkish (p.Asp371Gly), and Mexican (p.Pro27Arg) ancestry [11, 16]; the enrichment of de novo missense variants in the gene reached genome-wide significance (4 observed versus 0.1 expected; P = 2.63×10⁻⁷, for a Bonferroni threshold of 2.55×10⁻⁶) [2]. The authors called it "ATP6V1A encephalopathy" and noted that, like PRRT2, TBC1D24, RELN, and SLC2A1, the gene has dominant phenotypes and a more severe recessive phenotype [2].

In 2022, Guerrini and colleagues published the reference cohort, with 26 patients carrying 20 distinct de novo missense variants, gathered from diagnostic and research series in Europe, the United States, and Japan, with the help of a family group created after the first report [12]. This work showed progressive cerebral atrophy, changes in endolysosomal pH, mild accumulation of lysosomal substrates, and ultrastructural inclusions in fibroblasts and in neurons derived from induced pluripotent stem cells (iPSC), and proposed ATP6V1A encephalopathy as a "new paradigm among lysosomal diseases" [12, 32]. Between 2023 and 2025 there followed a Chinese family with dominantly transmitted epilepsy, a series with childhood focal epilepsy of favourable course, a case of febrile seizures with mild ID without epilepsy, a review of 33 patients, and a paediatric series of 10 children including one case of mosaicism [9, 13, 14, 29, 30].

1.3. Nomenclature and classifications

The current OMIM name of the dominant form is "developmental and epileptic encephalopathy 93"; the literature also uses "ATP6V1A encephalopathy", and ClinVar groups submissions under "ATP6V1A-related disorder" [1, 4]. Orphanet has no dedicated entry: DEE93 is included in the group ORPHA:442835 ("non-specific early-onset epileptic encephalopathy", which aggregates about 60 OMIM entries), and the gene belongs to the category ORPHA:309778 ("defect in V-ATPase") [3, 8, 18]. The 2022 classification of the International League Against Epilepsy (ILAE) separates, among syndromes with onset before age 2 years, the self-limited forms from the DEE, defined as disorders in which developmental impairment results both from the underlying aetiology, independently of epileptiform activity, and from the epileptic encephalopathy itself; the same classification recognises aetiology-defined syndromes and designates infantile spasms as infantile epileptic spasms syndrome (IESS) [33, 34]. DEE93 meets this definition: the progression of cerebral atrophy and the lack of acquisitions in patients without seizures show that the impairment does not depend on the seizures alone [12].

The recessive form has three designations in use. In the clinical classification of the cutis laxas adopted by OMIM and GeneReviews it is ARCL2D, a recessive type 2 form with developmental and glycosylation involvement [20, 21]. In the CDG nomenclature, which since 2008 has used the official gene symbol followed by "-CDG", it is called ATP6V1A-CDG and belongs to the group of defects of Golgi pH and ion homeostasis, with the brain and eyes as affected organs [5]. In Orphanet, it is in the subtype ORPHA:357074, shared with ARCL2A (ATP6V0A2) and ARCL2C (ATP6V1E1) [7]. The GeneReviews entry on ELN groups these forms under the term "neurometabolic cutis laxa", whose first mechanism is aberrant glycosylation from reduced acidification of secretory vesicles due to defects in V-ATPase subunits (ATP6V0A2, ATP6V1E1, ATP6V1A, and ATP6AP1) [20].

2. Molecular Genetics and Pathophysiology

2.1. The gene and the protein

ATP6V1A maps to 3q13.31, has 15 exons and, unlike most V-ATPase subunits, has no isoforms encoded by paralogous genes [8, 9]. The MANE Select reference transcript is NM_001690.4, which encodes the protein NP_001681.2; the previous version NM_001690.3, used in the 2018 and 2021 articles, gives the same c. coordinates, so published variants carry over unchanged [10]. Other identifiers: HGNC:851, Entrez 523, UniProt P38606; historical aliases Vma1, VA68, VPP2, ATP6A1, and ATP6V1A1 [8]. The gene spans about 65 kb in GRCh38 (3:113 747 035–113 812 056) and encodes a protein of 617 amino acids and about 68 kDa, with 99% identity to bovine subunit A [11]. The initial cloning, in 1993, started from an osteoclastoma library, in which a 615-amino-acid variant (HO68) was also identified whose expression proved to be restricted to that tumour [11]. Expression of the gene is ubiquitous, with the highest levels in skin and cultured fibroblasts, which is consistent with the severity of skin involvement in the biallelic forms [11].

V-ATPase is a multimeric proton pump with a peripheral V1 domain, which hydrolyses adenosine triphosphate (ATP), and an integral membrane V0 domain, which translocates protons by a rotary mechanism [2]. The V1 domain has eight subunits (A to H) and V0 has five; 22 autosomal genes encode a redundant set of subunits, which allows complexes with different properties and tissue expression to be assembled, and 16 of these subunits had been associated with human disease by 2022 [9, 12]. Subunit A contains the nucleotide-binding site, with the P-loop motif (residues 250 to 257) and the magnesium ion coordinated by Asp349, and forms interfaces with subunits B and D that take part in catalysis and rotation [2, 12]. The structure is organised into three Pfam domains: the N-terminal β-barrel domain (ATP-synt_ab_N), an extension (ATP_synt_ab_Xtn), and the central nucleotide-binding domain of the ATP synthase α/β family (ATP-synt_ab, residues 230 to 455) [12].

Acidification of intracellular compartments underpins protein sorting, glycosylation in the Golgi, receptor-mediated endocytosis, lysosomal degradation, activation of the mTORC1 complex at the lysosomal surface, and the loading of neurotransmitters into synaptic vesicles, where V-ATPase creates the proton gradient used by vesicular transporters [2, 8]. In nutrient signalling, ATP hydrolysis by V-ATPase is required for amino acids in the lysosomal lumen to modulate its interaction with the Ragulator complex and promote translocation of mTORC1 to the lysosomal surface; in tumour cells with oncogenic RAS, relocation of V-ATPase to the plasma membrane is required for macropinocytosis, an observation with no clinical application in the diseases of this article [11]. Neuronal expression of V-ATPase is high, which explains the vulnerability of the developing brain [2]. In the recessive forms with cutis laxa, V-ATPase maintains the pH gradient along the secretory pathway and facilitates vesicle fusion; the exact mechanism by which its dysfunction alters glycosylation remains unclear [5].

2.2. Mechanism of DEE93: opposite effects on pH depending on the variant

DEE93 does not result from a single loss-of-function mechanism. In the founding work, structural modelling predicted that p.Pro27Arg disrupts the interaction with subunit B, p.Asp100Tyr destabilises folding through steric hindrance, p.Asp349Asn, near the nucleotide-binding site, affects magnesium coordination, and p.Asp371Gly, at the A/B/D interface, compromises rotation [2]. In cellular studies, p.Asp100Tyr reduced protein expression through increased degradation, with reduced LysoTracker staining and reduced expression of the lysosomal membrane protein LAMP1, behaving as a loss of function; p.Asp349Asn maintained expression, increased LysoTracker staining, and lowered endocytic pH in lymphoblasts, indicating increased proton pumping, that is, gain of function [2]. Both reduced recruitment of V-ATPase to autophagosomes, without major change in autophagic flux, and both caused, in rat hippocampal neurons, a comparable defect of neurite elongation and loss of excitatory synapses [2].

The 2022 cohort confirmed the dichotomy in fibroblasts from four patients. In the two most severely affected patients (p.Asp100Tyr and p.Glu356Asp), LAMP1 expression and LysoTracker staining decreased and organelle pH, measured with the ratiometric probe LysoSensor, rose; in the two less severely affected (p.Asp349Asn and p.Gly363Val), LysoTracker staining increased, pH fell, and LAMP1 was unchanged [12]. Reduced proton pumping may result from loss of expression through degradation (p.Asp100Tyr) or from loss of catalytic function (p.Glu356Asp, at the A/B interface on the catalytic side); the mechanism of hyperacidification in the milder forms is unclear, but excessive acidification also reduces the activity of lysosomal hydrolases [12]. Molecular dynamics simulations of p.Asp349Asn showed that the variant does not compromise magnesium-ATP binding, contrary to what the 2018 structural modelling suggested, which is consistent with the gain of function observed in cells [12].

The lysosomal consequences were documented by three approaches. The activities of α-L-iduronidase, α-galactosidase, and acid β-glucocerebrosidase measured in vitro at optimal pH did not differ from those of controls, but the natural substrates were increased in patient cells (heparan sulfate +69 to +240% in all but one, glucosylsphingosine +21 to +223%, globotriaosylsphingosine +30 to +80%), indicating that the abnormal pH creates a suboptimal catalytic environment in vivo without a primary enzyme defect [12]. Electron microscopy of fibroblasts from four patients showed single-membrane vacuoles with osmiophilic material, phospholipid-like lamellar membrane structures, and lipid droplets; in glutamatergic neurons derived from iPSC of two patients with DEE, lysosomes were dark lamellar bodies or irregular structures, with a significantly smaller area than in controls (P < 0.001) and similar density [12]. The authors frame the disease as a lysosomal disorder with a particular mechanism: unlike the classic storage diseases, which are recessive or X-linked, a single dominant allele suffices because it simultaneously compromises several pH-dependent enzymes [12].

In mouse hippocampal neurons, depletion of Atp6v1a reduced overall expression of the V1 subunits, altered the regulation of lysosomal pH and the progression of autophagy, with aberrant lysosomes in the soma and enlarged vacuoles in synaptic boutons, and prevented the elongation and stabilisation of neurites, the function of excitatory synapses, and synaptic rearrangement after induction of plasticity [35]. This is preclinical evidence, consistent with the hypothesis that V-ATPase monitors synaptic integrity and plasticity [30, 35]. Along the same lines of research, the TBC1D24 protein, whose gene causes dominant and recessive epilepsies, interacts with V-ATPase in the brain and regulates intra-organelle pH in neurons, and the ROGDI protein, mutated in Kohlschütter-Tönz syndrome (epilepsy with amelogenesis imperfecta), proved to be a subunit of the Rabconnectin-3 complex, which takes part in V1–V0 assembly [36, 37]. These data are preclinical and link DEE93 mechanistically to differentials with epilepsy and enamel defects.

Glycosylation is also affected in the dominant form, to a lesser degree than in the recessive form. In two severely affected patients from the 2022 cohort, mass spectrometry of transferrin and of total serum N-glycans showed a mild increase in hyposialylated glycoforms (mainly monosialylated biantennary) and an approximately threefold increase in antennary fucosylation; O-glycosylation of apolipoprotein C-III was normal in the patient tested [12]. The authors attribute these changes to reduced activity or abnormal localisation of pH-dependent sialyltransferases and fucosyltransferases in the Golgi [12].

A 2024 review summarised the proposed correlation: severely affected patients with reduced LAMP1 and increased endolysosomal pH (loss of function or degradation) and mildly to moderately affected patients with normal LAMP1 and decreased pH (gain of function) [9]. The correlation rests on four cell lines and remains speculative for variants not studied [9, 12]. A 2024 case report states that dominant de novo variants are gain of function and recessive ones loss of function [30]; the primary functional data contradict this simplification, because p.Asp100Tyr and p.Glu356Asp, dominant variants from severely affected patients, reduce proton pumping [2, 12].

2.3. Mechanism of ARCL2D

In ARCL2D, all known missense variants destabilise the V1 sector, and the first truncating allele, c.1513_1514del p.(Asp505Ter), undergoes complete degradation; the authors speculate that the combination of two null alleles is incompatible with life [15]. Patient fibroblasts show a fragmented Golgi apparatus, delayed brefeldin A-induced retrograde transport, and reduced sialylation in the Golgi, and the dermis has reduced and fragmented elastic fibres [8, 15]. In the 2017 patients, complexome analysis in fibroblasts showed a marked reduction of the assembled V1 domain and of the complete V1V0 complex, with normal gene expression [11]. In fibroblasts from the patient homozygous for p.Arg338Cys, a substitution in an α-helical region, ICAM1-positive cells (a marker whose reduction is typical of the CDG) were reduced, the Golgi was dilated and fragmented, and retrograde transport between the Golgi and the endoplasmic reticulum was delayed, with two to three times more cells retaining Golgi remnants than controls; electron microscopy of the dermis showed elastic fibres of normal quantity and structure, but loosely aggregated collagen fibres of variable diameter [11]. In the dermis of a patient homozygous for p.Gly72Asp, a substitution at the interface between subunits A and B, the elastic fibres were irregular, sparse, and fragmented, with the same collagen abnormality [11]. Elastic involvement of the dermis therefore varies between patients. Serum glycosylation is variable: transferrin isoelectric focusing was normal in one of the patients of the Hungarian family, and the 2017 patients had only mild hyposialylation of transferrin through lack of a single sialic acid [12, 15, 17]. In the German patient, the apolipoprotein C-III glycosylation pattern, initially abnormal, normalised by age 6 years [17]. Glycosylation abnormalities in the ATP6V1A forms are less marked than in the ATP6V0A2 and ATP6V1E1 forms, probably owing to differences in tissue expression of the subunits [12].

2.4. Genetic constraint, spectrum, and location of variants

The gene is intolerant of missense variation: pLI of 0.99 and missense Z of 3.03 in ExAC, according to the 2018 work; missense Z of 3.37 in gnomAD v2.1.1 and RVIS of −1.041 (11th percentile), according to the 2022 work, with a probability of 0.911 of harbouring dominant variants by the DOMINO tool [2, 12]. In the 2022 cohort, all 20 variants were de novo missense, absent from population databases (ExAC, gnomAD, HGVD, BRAVO), and predicted to be deleterious; 17 of the 20 (85%) were in the ATP-synt_ab domain, which shows significant genome-wide constraint, and 19 of the 20 residues had a missense tolerance ratio (MTR) below the neutrality threshold, Pro27 being the exception [12]. Five substitutions recurred in unrelated patients (p.Pro375Thr, p.Ser316Phe, p.Pro249Arg, p.Arg381Pro, and p.Arg359Gly), one occurred in monozygotic twins (p.Gly276Arg), and two residues had two different substitutions (Asn314 and Pro412) [12]. Mapped onto the cryo-electron microscopy structure of rat V-ATPase (98% identity with the human protein), 13 of the 20 variants lay near the P-loop or at the A/B interface on the catalytic side, three at the A/D or non-catalytic A/B interface, and four in less specific regions, with a probable effect on stability [12].

The 2024 review counted 31 published pathogenic missense variants: 21 in epileptic encephalopathy, 5 in cutis laxa (homozygous or compound heterozygous), and 5 in cohorts with autism spectrum disorder (ASD) or encephalopathy, clustered in the central domain [9]. Other dominant phenotypes have been attributed to the gene in large cohorts, such as ASD and cerebral visual impairment, and ClinVar has p.Asn314Asp submitted with "cerebral visual impairment and ID" [4, 30]. Two de novo missense variants from the denovo-db database without validation (p.Asp11Asn in a boy with ASD and p.Pro249Arg in a boy with a severe developmental disorder) were later partly integrated: p.Pro249Arg is patient 17 of the 2022 cohort [2, 12].

The following table presents the variants with a published phenotype, in Human Genome Variation Society (HGVS) nomenclature, normalised in VariantValidator on the MANE transcript (GRCh38) [10].

In GRCh37 (NC_000003.11), the coordinates of the DEE93 variants are, for example, g.113497680 for c.80C>G, g.113503157 for c.298G>T, and g.113513775 for c.1045G>A [10]. Other missense variants from the 2022 cohort, described only at the protein level, include p.Gln103Arg, p.Gly276Arg, p.Asn314Ser, p.Asn314Asp, p.Ser316Phe, p.Arg323Gly, p.Arg359Gly, p.Pro375Thr, p.Arg381Pro, p.Pro412Leu, p.Pro412Ser, and p.Leu435Phe, as well as substitutions at residues Cys254 and Leu345 [12]. The two cohort variants studied in fibroblasts and not published in 2018, p.Glu356Asp and p.Gly363Val, have the c. notation in OMIM (c.1068G>T and c.1088G>T), confirmed in VariantValidator (GRCh37: g.113513798G>T and g.113513818G>T) [10, 11].

As of 28 October 2026, ClinVar had 56 pathogenic or likely pathogenic records for the gene, including copy number variants; missense variants with DEE93 or a non-specific condition cluster between codons 312 and 378, among them p.Val312Ala, p.Asn314Ser, p.Met318Ile, p.Pro319Ser, p.Val320Ala, p.Val320Ile, p.Arg323Gly, p.Asp349Glu, p.Glu356Asp, p.Ala357Pro, p.Gly363Val, and p.Pro375Thr, as well as p.Asp100Val, p.Arg212Cys, p.Ser264Pro, p.Pro412Ser, p.Phe432Ile, and p.Met550Thr; p.Gly281Arg is submitted with ARCL2D [4]. The raw count of submissions in GeneGuidance (44 pathogenic, 14 likely pathogenic, and 228 of uncertain significance) includes multigene records and is not a frequency [8]. There are 3q13.2-q13.31 deletions of about 3 to 5 Mb classified as pathogenic with the condition "3q13.31 deletion syndrome" that include ATP6V1A; the contribution of the gene to that phenotype is not established, and there is no evidence that haploinsufficiency causes DEE93, whose mechanism is missense [4, 12].

2.5. Inheritance, penetrance, and expressivity

DEE93 is autosomal dominant. In the series with parental testing, the variants arose de novo in 26 of 26 patients of the 2022 cohort and in 10 of 10 of the 2025 Chinese series, one of them mosaic in the proband; the 2024 review counted 29 sporadic cases and one pair of monozygotic twins [9, 12, 13]. An extended Chinese family with epilepsy without other neurodevelopmental abnormalities, carrying p.Leu378Phe and with good control on levetiracetam, shows that variants with a milder effect can be transmitted; the report was read in abstract only, and the variant was classified as of uncertain significance by ACMG/AMP criteria, despite functional studies showing reduced RNA and protein [14, 30]. Expressivity is highly variable and depends mainly on the variant: the two unrelated patients with p.Ser316Phe had a severe neonatal phenotype and early death, the two substitutions at Pro412 gave a rapidly progressive encephalopathy with profound atrophy, and p.Asn314Ser gave a milder phenotype than p.Asn314Asp, which is more disruptive of the structure [12]. Variants affecting folding, the catalytic site, or the vicinity of the P-loop were more often associated with severe phenotypes, although the number of patients does not allow statistical validation [12].

ARCL2D is autosomal recessive. Of the six patients published up to 2021, all male, three were children of consanguineous parents (two from different Turkish families and one from a Pakistani family) and homozygous; the two Hungarian brothers, of non-consanguineous parents, were compound heterozygous for p.(Leu106Ser), inherited from the mother, and the truncating allele p.(Asp505Ter), inherited from the father [15]. The German patient was homozygous for p.Arg338Cys, with no consanguinity reported in the 2021 review, and in each 2017 family the variant segregated with the disease [15, 17]. Heterozygotes for ARCL2D variants have no described phenotype, and the dominant encephalopathy does not occur in the parents of patients with cutis laxa, which is consistent with different mechanisms: biallelic destabilisation of the V1 sector in cutis laxa, a dominant effect of missense variants at catalytic or interface residues in DEE93 [8, 15].

2.6. Clinical validity and allelic disorders

ClinGen classified the ATP6V1A–ARCL2D relationship as limited (3 April 2024, CDG expert panel) and, as of 28 October 2026, had no curation of the relationship with DEE93 [19]. The lack of curation does not reflect weakness of the evidence for DEE93, which includes genome-wide significant enrichment of de novo missense variants and more than 40 patients with de novo variants and concordant functional studies [2, 9, 12]. There is no ClinGen actionability classification for the gene [8]. The two allelic disorders are those of this article; the associations with ASD and cerebral visual impairment come from cohorts with limited clinical information and, in the 2022 cohort, the patients with p.Asn314Asp and p.Pro249Arg, previously listed in those cohorts, had a full encephalopathy picture [12, 30].

3. Clinical Spectrum and Manifestations by System

3.1. DEE93: overall spectrum and sources of the figures

DEE93 forms a continuum of severity with four presentations: early lethal encephalopathy with massive and rapidly progressive cerebral atrophy; severe DEE without acquisitions; static moderate ID with epilepsy; and, in the most recent series, childhood epilepsy of favourable course or febrile seizures with mild ID without epilepsy [12, 29, 30]. The frequencies in this article come from the full text of the 2022 cohort (26 patients from series in Europe, the United States, and Japan, mean age at last follow-up 7 years, range 6 weeks to 22 years), which is the only series with explicit denominators for each item [12]. The abstract of the same article and the review that cites it use different denominators, which produces divergent percentages, and section 3.6 shows the correspondence. All patients described had access to antiseizure treatment; there are no data on untreated natural history, and the percentages describe treated disease in referral centres, with the selection bias towards severe cases inherent to exome cohorts in encephalopathies.

3.2. Early lethal form

Four of the 26 patients (15%) had profound and rapidly progressive encephalopathy from birth, with severe hypotonia and no acquisitions; three died, at 6 weeks, 5 months, and 2 years [12]. Two died without ever having had seizures, one died shortly after seizure onset, and the fourth had no seizures at 25 months; magnetic resonance imaging (MRI) showed evident atrophy from the third month of life, worsening rapidly [12]. The authors suggest that severe atrophy may reduce synchronised electrical activity and, paradoxically, epileptogenesis [12]. The two patients with p.Ser316Phe required neonatal intensive care and died early, and the two with substitutions at Pro412 had a similar picture, with early profound atrophy [12].

3.3. Severe DEE

Sixteen patients surviving beyond the first year were classified as DEE, that is, 16 of 25 with clinical information (64%) and 16 of 21 with epilepsy (76%); onset was in the first year in 13 and in childhood in three [12]. In infant-onset DEE, infantile spasms were the dominant seizure type (11 of 13, 85%), with a mean age at onset of 7 months (2 to 13 months), followed later by myoclonic, tonic, tonic-clonic, atonic, absence, and focal seizures [12]. In one patient with childhood-onset DEE, intractable generalised tonic-clonic seizures appeared at 7 years, followed by loss of motor acquisitions [12]. All patients with DEE had failed several antiseizure medications [12]. Ten of the 16 had profound delay from early on, with hypotonic tetraplegia and no motor, communicative, or developmental acquisitions (10 of 16, 63%); the other six acquired motor skills, which two lost, one with worsening of epilepsy at 4 years and the other after seizure onset at 7 years [12]. Only the two patients with childhood-onset epilepsy acquired some language, always poor; the mean age at last follow-up was 6 years and 5 months (10 months to 14 years) [12].

3.4. Static ID with epilepsy

Six patients (6 of 26, 23%) had moderate (five) or mild to moderate (one) ID with epilepsy, without the seizure-related progression typical of the DEE, although epilepsy was severe in two [12]. Seizure types were generalised tonic-clonic (3 of 6), focal (2 of 6), myoclonic (1 of 6), and absence (1 of 6), and one patient had only rare febrile seizures; the mean age at seizure onset was 27 months (15 months to 3 years) [12]. All walked unaided, four acquired simplified language, three had mild autistic traits and one had traits of attention deficit hyperactivity disorder (ADHD); the mean age at last follow-up was 10 years and 6 months (4 to 22 years), and the authors warn that long-term worsening remains possible [12].

3.5. Mild and familial forms described after 2022

The China Epilepsy Gene Project identified, by trio exome in patients with epilepsy without an acquired cause, three de novo variants (p.Gly250Ala, p.Gln261Arg, and p.Met368Thr) in three unrelated patients with childhood focal epilepsy, good response to antiseizure medications, and normal psychomotor development; the authors propose childhood focal epilepsy of favourable course as a phenotype of the gene [29]. In an extended Chinese family, a heterozygous p.Leu378Phe variant segregated with epilepsy without other neurodevelopmental abnormalities, with seizures controlled by levetiracetam; the index case, a 5-year-old boy with seizures since 9 months, caught up developmentally after seizure control [14, 30]. A boy with de novo p.Val28Met had febrile seizures and mild non-progressive ID, without epilepsy and with normal MRI over five years (section 10) [30].

The Peking University Hospital series (January 2019 to December 2024) included 10 children, four boys and six girls, all with de novo heterozygous missense variants (nine distinct variants, seven novel; one mosaic) [13]. Epilepsy began at a median of 28 months (interquartile range 9 to 48), from the neonatal period to childhood, and the first seizure was febrile in five; there were focal seizures in six, epileptic spasms in five, and tonic-spasm and atonic seizures in one patient each [13]. Eight had global delay, two with ASD, and two had normal development; five met criteria for DEE, four of them with IESS [13]. With a median follow-up of 78 months, seizures were controlled in six and refractory to at least three antiseizure medications in four [13]. In an Indonesian series of 10 children with drug-resistant epilepsy studied by exome, one had a variant of uncertain significance in ATP6V1A, without details in the abstract [38].

3.6. Correspondence of the 2022 cohort figures

Sources: [12, 30]. The small internal differences (20 or 21 patients with epilepsy; 10 or 11 with spasms) exist in the article itself, and the 2024 review transcribes percentages with different denominators; the table uses the values from the full text.

3.7. Manifestations by system in DEE93

Epilepsy. In the 2022 cohort, febrile convulsive seizures were the initial manifestation in seven of the patients with epilepsy, and only one of these did not progress to epilepsy; seizures were an initial symptom in 9 of 25, never isolated, but in the context of early hypotonia in most and of delay in all [12]. In the 2024 review of 33 patients, 28 had seizures, 23 in the first 3 years, with the earliest onset at 2 months; among the 28, there were febrile seizures in 10, infantile spasms in 9, tonic in 11, clonic in 6, occipital focal in 6, generalised tonic-clonic in 7, and myoclonic in 6 [9]. All four patients of the founding work began with febrile seizures, which progressed to spasms and tonic, clonic, myoclonic, and occipital focal seizures in the two severely affected patients, and to occipital focal or generalised tonic-clonic seizures in the two with moderate ID [2]. The first febrile seizures appeared at 11 months in the two severely affected patients and at about 3 years in the two with moderate ID [16].

Electroencephalogram. In the 2022 cohort, with an electroencephalogram (EEG) in 24 patients (19 with epilepsy), there was hypsarrhythmia in 2 (10% of patients with epilepsy), multifocal epileptiform discharges in 11 (58%), focal abnormalities in 7 (37%), and slow background activity in 13 of 24 (54%) [12]. In the 2024 review (26 EEGs), the most common pattern was diffuse slow background activity with multifocal discharges (13), followed by hypsarrhythmia (4) and centrotemporal spike-wave (2) [9]. In the two severely affected patients of the founding work, the EEG also showed continuous slow spike-wave during sleep, in addition to hypsarrhythmia and diffuse and multifocal discharges [16]. Hypsarrhythmia is uncommon despite the frequency of spasms, which is compatible with the definition of IESS, which does not require hypsarrhythmia [24].

Development and language. Of the 19 patients older than 2 years at last follow-up, 13 did not speak and 6 had poor language; no patient in the 2022 cohort developed communicative language [12]. In the 2024 review, two-thirds of patients did not speak and one-third had poor language, in relation to the frequency and severity of seizures [9]. The review stresses that seizure control was not accompanied by improvement of the delay in severely affected patients [9]. In the mild forms described after 2022, development may be normal [13, 29].

Motor function. Hypotonic tetraplegia is the rule in severe forms, with dyskinetic movements in 4 of 26 (15%), cerebellar signs in 3 of 26 (12%), and congenital arthrogryposis in 3 of 26 (12%) [12]. The two severely affected patients of the founding work had hypotonic and dyskinetic tetraparesis and did not walk; one of the patients with moderate ID had a broad-based gait and hypotonia [2]. The OMIM definition mentions spastic tetraparesis in some patients, although the clinical description in the same entry, based on the 2018 patients, refers to dyskinetic tetraparesis [16].

Head growth. Microcephaly (head circumference below −2 SD) was present in 9 of 24 (38%) and was progressive in six; head circumference at birth was negative in 10 of 12 patients with a measurement, and 14 patients had values below −1 SD, eight of them already in the first year [12]. In one patient of the founding work, head circumference went from −1.9 SD at birth to −7 SD at 12 years [2]. The other severely affected patient in that series had microcephaly of −3.2 SD, both with global growth retardation, and the two with moderate ID had normal MRI at 7 years [16].

Teeth. Amelogenesis imperfecta or enamel dysplasia, with eruption defects, deficient enamel mineralisation, excessive caries, and friable teeth, was described in 11 of 26 patients (42%), although dental information was not available for all [12]. V-ATPase is upregulated in ameloblasts at the maturation stage, and recessive osteopetrosis due to TCIRG1 (subunit a3) has the same enamel defect, which supports the causal link [12]. This finding is a diagnostic pointer (section 4).

Vision. Lack of eye contact was recorded in 17 of 26 (65%), optic atrophy in 4 of 26 (15%), and iris coloboma in one patient [12]. In the 2018 patients, hypermetropia, optic atrophy, and iris coloboma were also recorded, as well as cryptorchidism and mild facial dysmorphism [16].

Feeding and other features. Five patients (19%) were gastrostomy-dependent [12]. One patient had precocious puberty at 9 years and another headaches [2]. Dysmorphic features are rare and mild (broad forehead, deep-set eyes, beaked nose in one patient; long face, malar hypoplasia, and hypotelorism in another) [2, 30]. There were no signs of osteopetrosis or renal tubular acidosis, manifestations of other V-ATPase subunits [12].

Neuroimaging. MRI was abnormal in 20 of 25 patients (80%), examined between 1 week and 11 years of age, with atrophy of the cerebral cortex, brainstem, and cerebellum, enlargement of the ventricles and subarachnoid spaces, thinning of the corpus callosum, and hypomyelination (the latter in 17 patients) [12]. In the eight patients with repeat MRI, atrophy always worsened, at variable speed: in two, atrophy progressed dramatically in the first months and paralleled clinical deterioration until death; in others, mild atrophy only became apparent between 11 months and 5 years; the white matter hyperintensity attributed to hypomyelination did not progress [12]. In the 2024 review (22 MRIs), there was hypomyelination in 13, mild cerebral and cerebellar atrophy in 13, a thin corpus callosum in 4, and ventricular enlargement in 1 [9]. Five patients in the 2022 cohort had normal MRI, and normal MRI is characteristic of the mild forms [12, 30]. The authors interpret the overall picture as prenatal compromise of the central nervous system followed by a neurodegenerative course [12].

3.8. ARCL2D

ARCL2D is congenital. In the six male patients published up to 2021 (two Hungarian brothers, two Turkish patients from different families, one German, and one Pakistani), all had generalised cutis laxa, redundant skin, facial dysmorphism with a mask-like face, and hypotonia; the cutis laxa improved with age in the three with sufficient follow-up [15]. There was chest deformity in four, contractures in three, entropion in two, hip dysplasia in one, cardiac anomalies in four (heart failure in one), aortic dilatation in one, structural brain abnormalities in five, seizures in two, and early death in two (at 8 days and at 3 months) [15]. Myopathy is constant in the patients studied: creatine kinase of 13,000 to 15,000 U/L in the newborn who died, 392 to 1082 U/L in his brother, and up to 1980 U/L in the 2021 Turkish patient, with muscle biopsy showing vacuolar myopathy with rimmed vacuoles, accumulation of LAMP2, and of mitochondria [15]. Neonatal cardiorespiratory decompensation may require ventilation (8 days and 3 days in the Hungarian brothers), and after that period the phenotype tends to improve: the surviving brother sat with support at 24 months and walked with support at 28 months, but at 17 years had normal cognition and attended secondary school; the Turkish patient walked at 14 months, had normal cognitive development, and developed a restrictive ventilatory syndrome and severe scoliosis operated on at 14 years [15]. According to the 2022 summary, the patients of the 2021 series additionally had a progeroid appearance with a triangular face, cataracts, limb contractures and, in some, low plasma proline, arginine, or ornithine [12]. With the exception of the plasma amino acids, these features are not exclusive to the 2021 series: the three 2017 patients had a progeroid face, and the German patient had bilateral cataract from birth [17].

The three patients of 2014 and 2017. All had generalised skin wrinkling with scant subcutaneous fat and a progeroid face with recurring features: triangular mask-like face, short forehead, hypertelorism, entropion, low-set ears with poorly folded helices, beaked nose with a broad base and narrow nostrils, and a short, pointed chin [17]. The German boy was born hypotonic, with generalised cutis laxa, fat accumulation in the genital region and hips, sloping forehead, small palpebral fissures, redundant supraorbital skin, prominent ears, bilateral cataract, strabismus, and retrognathia; he had growth retardation in the first year (from the 97th centile at birth to the 25th centile at 3 years), cardiomyopathy diagnosed at 8 months (at 4 years, mild hypertrophy of the interventricular septum with normal ejection fraction, long QT interval, and incomplete right bundle branch block), microcephaly from 1 year, late closure of the fontanelle, and a marfanoid habitus at 14 years [17]. MRI showed enlarged ventricles with white matter involvement and parieto-occipital periventricular gliosis; speech delay was severe, independent walking appeared only at 5 years, and he had a generalised tonic-clonic seizure at 14 years; the skin improved early, and by 8 months the redundancy was confined to the neck and axillae [17]. The Pakistani infant had severe bilateral clubfeet, dislocated hips, flexion contractures of all joints, camptodactyly, micropenis, cryptorchidism, and right inguinal hernia, a small atrial septal defect with a tortuous aortic arch, generalised and complex focal seizures, and MRI suggestive of polymicrogyria with a thin corpus callosum; he died after progressive heart failure, pneumonia, and sepsis, following prolonged intensive care [17]. The Turkish infant had a ventricular septal defect and an anatomical variant of the cavum septum pellucidum [17]. Mass spectrometry of intact transferrin showed, in all three, an increase in the glycoform lacking one sialic acid [17]. These descriptions are the origin of most of the items in the Brazilian textbook cited below.

The Brazilian reference textbook describes, for this form, inguinal hernia, abnormal fat distribution, atrial septal defect, dilatation of the ascending aorta, tortuous aortic arch, heart failure, hypertrophic cardiomyopathy, QT interval abnormality, incomplete right bundle branch block, marfanoid habitus, flexion contractures, dislocated hips, camptodactyly, seizures, speech and motor delay, enlarged ventricles with white matter involvement, parieto-occipital periventricular gliosis, thickened cortex, thin corpus callosum, bilateral cataract, micropenis, cryptorchidism, and abnormal serum glycosylation [6]. GeneReviews describes the form as similar to ARCL2A in the face, with myopathy, lipodystrophy, marfanoid habitus, and potentially lethal respiratory problems in infancy [20, 21].

Discrepancies between sources and their resolution. The differential tables of the two GeneReviews entries state "no seizures; often no ID" for ARCL2D, whereas the Brazilian textbook and the Ghent review attribute seizures, delay, and brain abnormalities to it [6, 20, 21]. The primary 2021 review resolves the discrepancy: seizures occurred in two of six patients, both from the 2017 series (German and Pakistani), and the survivors of the 2021 series never had seizures and had normal cognition, despite a thin corpus callosum or cortical atrophy on MRI [15]. In short, seizures and delay occur in a minority and structural brain abnormalities in the majority. As for aneurysms, the GeneReviews entry on ELN (2022) lists them as present and the entry on ATP6V0A2 (2023 revision) as absent; in the six patients, there was only aortic dilatation in one, with no aneurysms described, which supports the more restrictive classification [15, 20, 21].

3.9. Treated disease and natural history

There are no studies comparing the course with and without treatment. In DEE93, seizure control was achieved in the mild forms and in some of the forms with spasms, but all patients with DEE in the 2022 cohort failed several antiseizure medications, and progression of cerebral atrophy occurred independently of seizures, including in patients who never had any [12, 13]. In ARCL2D, neonatal survival depends on intensive care, and improvement of the skin and development after the first year is part of the natural history [15].

4. Clinical Recognition and Red Flags

4.1. When to suspect DEE93

In the newborn and young infant, suspicion arises in the presence of severe hypotonia from birth without acquisitions, with microcephaly or a head circumference falling away, and MRI with hypomyelination and cerebral, cerebellar, or brainstem atrophy that worsens between scans a few months apart; seizures may be absent in this form [12]. Between 2 and 13 months, the typical picture is a hypotonic and delayed infant who develops epileptic spasms, often preceded by febrile convulsive seizures, with an EEG showing a slow background and multifocal discharges more often than hypsarrhythmia [2, 9, 12]. In the older child, the pattern is moderate ID with epilepsy beginning between 15 months and 3 years, absent or poor language, and acquired walking [12]. In the mild forms, febrile seizures, childhood focal epilepsy, or mild ID with language delay do not distinguish DEE93 from other causes, and the diagnosis is almost always made by panel, exome, or genome [29, 30].

The sign that most points to the gene, in any of these pictures, is the enamel defect (amelogenesis imperfecta, enamel hypoplasia, excessive caries, friable teeth), present in 42% of the 2022 cohort [12]. Dental examination is part of the assessment of any child with epilepsy and delay, because dental or enamel hypoplasia also points to SLC13A5 and to Kohlschütter-Tönz syndrome [1, 39]. Progressive cerebral atrophy on serial MRI, particularly when associated with non-progressive hypomyelination, is the second pointer [12].

4.2. When to suspect ARCL2D

ARCL2D should be considered in the newborn with generalised cutis laxa, a triangular mask-like progeroid face, hypotonia, and raised creatine kinase, especially if there is neonatal respiratory or heart failure, chest deformity, and contractures [15]. A large anterior fontanelle, scant fat, and association with a type 2 CDG are common to the ARCL2 forms; transferrin isoelectric focusing should be requested when there is cutis laxa or neurological involvement, bearing in mind that it may be normal [8, 40].

4.3. Red flags and emergency signs

Epileptic spasms in an infant are an emergency: the NICE NG217 guideline recommends seeking advice and referring to a tertiary paediatric neurologist within 24 hours of suspicion, for a sleep EEG and prompt start of treatment [23]. Loss of acquired skills (described at 4 and 7 years in patients with DEE), neurological deterioration with progressive atrophy in the first months, food refusal requiring gastrostomy and, in ARCL2D, neonatal respiratory distress and heart failure also require prompt assessment [12, 15]. In the assessment of regression, pseudoregression due to high-dose antiseizure medications should be excluded [41].

4.4. Newborn screening

There is no newborn screening for either disease; no blood biomarker identifies them [8].

5. Diagnosis: Criteria, Testing, and Algorithms

5.1. Criteria

There are no formal clinical criteria for DEE93 or ARCL2D; the diagnosis requires the identification of pathogenic or likely pathogenic variants in ATP6V1A, because no biochemical or functional test confirms it [8]. DEE93 is diagnosed with a heterozygous variant, ideally confirmed de novo, in a patient with a compatible phenotype; ARCL2D with two variants in trans in a patient with congenital cutis laxa [8, 15]. Classification of the epilepsy syndrome follows the 2022 ILAE criteria, including those for IESS; the full text of these criteria was not consulted [33].

5.2. Indication and choice of genetic test

The National Society of Genetic Counselors (NSGC) recommends, with strong strength, genetic testing for all people with epilepsy of unknown cause, with no age limit, with exome or genome and/or a multigene panel of more than 25 genes as first line, followed by chromosomal microarray; it conditionally suggests exome or genome rather than a panel; and it recommends, also with strong strength, that tests be ordered and interpreted by qualified professionals, with genetic counselling before and after [22]. The NICE NG217 guideline recommends considering genome sequencing in epilepsy of unknown cause with onset before age 2 years, with features suggestive of a genetic syndrome, or with ID, ASD, a structural anomaly, or unexplained cognitive decline, and between 2 and 3 years if agreed by a specialist multidisciplinary team; in the English National Health Service, the indication corresponds to test R59 of the national directory [23]. The two guidelines diverge: the NICE committee concluded that the evidence does not justify testing all people with epilepsy, whereas the NSGC recommends testing all [22, 23]. Either covers patients with DEE93, who almost always present before age 3 years and have delay.

The yield of epilepsy panels is highest when seizures begin before 2 months, at around 40%, and the heterogeneity of the DEE justifies early genomic testing [39]. In the founding cohort, DEE93 accounted for 4 of 1444 probands (0.28%) with encephalopathy of presumed genetic cause; this is not an estimate of frequency in the population with epilepsy (section 9) [2]. In a resource-limited Indonesian series, exome sequencing provided a diagnosis in 3 of 10 children with drug-resistant epilepsy and detected a variant of uncertain significance in ATP6V1A [38].

5.3. Ancillary investigations

The work-up of a child with seizures and delay includes a three-generation family history and enquiry about consanguinity, serial head circumference, dysmorphic features, eyes, skin, teeth, and signs in other organs; MRI (an MRI performed before 15 months may need to be repeated); EEG or video-EEG with sleep; genomic analysis; and basic biochemical screening, in which transferrin isoelectric focusing is part of the urine and serum screen for CDG [39, 41]. When IESS is suspected, the EEG should be performed within a few days, and non-epileptic paroxysmal events and other seizure types can mimic spasms, so confirmation is by video-EEG [24]. In DEE93, EEG and MRI provide orientation (hypomyelination and progressive atrophy) and dental assessment identifies enamel dysplasia [8].

In ARCL2D, the work-up includes skin biopsy with light and electron microscopy (reduced and fragmented elastic fibres, irregular collagen fibres), creatine kinase, echocardiography, respiratory assessment, and glycosylation screening [8, 15]. In ARCL2 due to ATP6V0A2, transferrin isoelectric focusing shows a type 2 pattern and that of apolipoprotein C-III is also abnormal, owing to a combined defect of N- and O-glycosylation [5]. In ARCL2D, however, glycosylation was abnormal in only two of six patients, and a normal transferrin glycoform profile does not exclude the diagnosis [8, 15]. In DEE93, the glycosylation abnormalities detected by mass spectrometry were mild (mild hyposialylation and increased antennary fucosylation) and have no diagnostic use [12].

Fibroblast assays (LysoTracker and LysoSensor staining, LAMP1 immunoblot, quantification of lysosomal enzyme substrates, electron microscopy) and, in ARCL2D, studies of Golgi morphology and trafficking are research tools that support variant interpretation; they are not validated as clinical tests [8, 12].

5.4. Algorithm

Algorithm sources: [8, 12, 15, 22–24].

5.5. Test limitations

Panels and exome sequencing detect well the coding missense variants that cause DEE93, but a negative result does not exclude ATP6V1A-related disease when there are variants in poorly covered regions, deep intronic variants, or structural rearrangements, which are general limitations of exome and panel sequencing. Mosaicism is relevant in both directions: a child in the 2025 series had the variant in mosaic form, which can be missed at a low allele fraction in blood, and parental mosaicism, if present, may not be detected in a parental blood sample [13, 39]. The proportion of variants detectable by sequencing is not equivalent to the clinical sensitivity of the test for the disease, which has not been established for ATP6V1A. Microarray detects the 3q13.31 deletions that include the gene, but these do not reproduce the missense mechanism of DEE93 [4]. For ARCL2D, confirmation of the phase of the two variants requires parental testing: in the Hungarian patient, the exome did not show whether the two variants were in trans, which only segregation demonstrated [15].

5.6. Variant classification and report content

The report should state the gene, the transcript NM_001690.4 with its version, the variant in HGVS (c. and p.), zygosity, origin (confirmed de novo, inherited, or unknown), the ACMG/AMP class with the criteria applied, and the associated condition (DEE93 or ARCL2D) [9, 10]. The criteria most used for DEE93 missense variants are PS2 (confirmed de novo) or PM6 (assumed de novo), PM2 (absence from population databases, at supporting weight), PP2 (gene with a low rate of benign missense variation and missense as the mechanism), PP3 (concordant computational predictions), and PM5 (another pathogenic missense variant at the same residue, for example p.Pro249Leu relative to p.Pro249Arg) [9, 30]. A 2024 report applied PVS1 to a missense variant (p.Val28Met); the criterion applies to null variants in a gene in which loss of function causes disease and is not appropriate here, but the pathogenic classification is supported by the remaining criteria [30]. Variants of uncertain significance, such as p.Leu378Phe in the Chinese family and the variant in the Indonesian series, should not guide clinical decisions without reclassification through segregation, new cases, or functional studies [14, 38].

5.7. Prenatal and preimplantation diagnosis

When the familial variant is known, prenatal diagnosis by chorionic villus sampling or amniocentesis is possible; for de novo dominant variants, non-invasive prenatal diagnosis may be appropriate [39]. Preimplantation genetic testing is an option in families with ARCL2D or with a familial dominant form, under the general conditions applicable to monogenic diseases (section 8). The cerebral atrophy and hypomyelination of DEE93 develop mainly after birth, although the authors propose a prenatal onset of the impairment; prenatal ultrasound has not been described as a means of detection [12]. In ARCL2D, reduced fetal movements were described in one patient [15].

6. Differential Diagnosis

6.1. Differential diagnosis of DEE93

The differential diagnosis of DEE93 is that of infant-onset DEE and of encephalopathies with progressive atrophy, and in practice it is resolved by genomic testing. Some features, however, allow ATP6V1A to be placed at the top of the list or a variant to be given weight: the enamel defect, hypomyelination with progressive atrophy, acquired microcephaly, and onset with febrile seizures followed by spasms [12]. The combination of epilepsy and enamel defect has its own differential: DEE25 due to SLC13A5, with neonatal epilepsy and dental or enamel hypoplasia, and Kohlschütter-Tönz syndrome (amelo-cerebro-hypohidrotic syndrome) due to ROGDI, whose protein takes part in V-ATPase assembly [1, 37, 39]. Isolated amelogenesis imperfecta (autosomal dominant due to enamelin, X-linked due to amelogenin, and recessive forms) has no neurological involvement [42]. The combination of hypomyelination and dental anomalies is reminiscent of the POLR3 leukodystrophies (delayed dentition) and Heimler syndrome (PEX1, PEX6; enamel hypoplasia with deafness) [43].

The other DEE with prominent spasms (CDKL5, ARX, STXBP1, FOXG1, GNAO1, and tuberous sclerosis complex, responsible for about 25% of spasms) should be excluded before counselling the family [39]. Within the V-ATPase family, DEE104 and the neurodevelopmental disorder with epilepsy and atrophy due to ATP6V0A1, early-onset epilepsy 3 due to ATP6V0C, DEE81 due to DMXL2, and the ATP6AP2 disorders share the mechanism of abnormal endolysosomal acidification; in cells with ATP6V0A1 missense variants, reduced LysoTracker staining was also observed [1, 12]. Diseases with early-onset progressive cerebral atrophy include cellular trafficking defects (ATP6AP2, AP4E1, TBCD, TBCK, TRAPPC-opathies), tubulinopathies with hypomyelination, CDG, aminoacyl-tRNA synthetase disorders, neuronal ceroid lipofuscinoses, PEHO syndrome (progressive cerebellar atrophy), and mitochondrial DNA depletion disorders such as POLG [28, 39, 41, 43]. Transferrin isoelectric focusing and alkaline phosphatase help to recognise CDG and glycosylphosphatidylinositol (GPI) defects among the causes of epilepsy with delay [41].

Table sources: [1, 12, 28, 39, 41–44].

6.2. Differential diagnosis of ARCL2D

The differential diagnosis of ARCL2D is that of the neurometabolic cutis laxas and the neonatal progeroid syndromes [20, 21]. ARCL2A (ATP6V0A2) has a similar face, a large fontanelle, abnormal type 2 glycosylation and, in some patients, seizures and ID; ARCL2C (ATP6V1E1) has cutis laxa, severe aortic root dilatation, biventricular hypertrophy, pneumothorax, and recurrent dislocations [6, 21]. Myopathy with raised creatine kinase and the mask-like face are prominent in ARCL2D [15, 20]. ARCL3A or De Barsy syndrome (ALDH18A1) and the PYCR1 forms have a progeroid appearance, cataracts, and amino acid abnormalities (proline, ornithine, citrulline, arginine), and some patients with ARCL2D in the 2021 series also had low plasma proline, arginine, or ornithine [12, 20]. Occipital horn syndrome and Menkes disease (ATP7A) are distinguished by low copper and caeruloplasmin [43]. The neonatal progeroid syndromes (Wiedemann-Rautenstrauch) have intrauterine growth restriction, relative macrocephaly, and a large fontanelle, and overlap with ARCL with CDG and with the progeroid variant of Ehlers-Danlos syndrome [40]. Among the V-ATPase defects, ATP6AP1-CDG (X-linked, with liver disease and immunodeficiency), TMEM199-CDG and CCDC115-CDG (liver disease, raised cholesterol, and low caeruloplasmin), and X-linked myopathy with excessive autophagy (VMA21) complete the group [5, 18].

Table sources: [1, 5, 15, 20, 21, 40, 43].

7. Treatment and Risk Management Strategies

7.1. Principles and sources of recommendations

There is no disease-modifying treatment or therapy targeted at ATP6V1A for either disease, no guideline mentions the gene, there is no ClinGen actionability assertion, and searches of ClinicalTrials.gov for "ATP6V1A", "developmental and epileptic encephalopathy 93", and "V-ATPase" on 28 October 2026 found no targeted trials [8, 45]. The only result with the term DEE93 was an observational study restricted to Dravet syndrome with an SCN1A variant (ENVISION, NCT04537832), outside the scope [45]. Treatment of DEE93 is that of the epilepsy syndrome the patient presents, above all infantile epileptic spasms syndrome, and of the neurodevelopmental, motor, feeding, visual, and dental impairment; that of ARCL2D is multidisciplinary supportive care (skin, myopathy, heart, lungs) [8]. The following recommendations extrapolate guidelines for IESS and paediatric epilepsy to DEE93, on the basis of the frequency of spasms and of focal and generalised seizures in the series; this extrapolation is declared at each point. Reference doses are those of the guidelines; the labels of the United States Food and Drug Administration (FDA) and the Portuguese summary of product characteristics (SmPC) are cited alongside for comparison.

7.2. Infantile epileptic spasms syndrome

Two recent guidelines were consulted: NICE NG217 (United Kingdom, published on 27 April 2022 and updated on 5 August 2026) and the interdisciplinary guideline coordinated by the Society for Neuropaediatrics of German-speaking countries (GNP, 2022), developed according to the SIGN methodology, with levels of evidence and grades of consensus [23, 24]. The two converge on the goal, the urgency, and the first-line drugs, and diverge on the choice between hormonal monotherapy and combination with vigabatrin.

The therapeutic goal is rapid cessation of spasms and resolution of hypsarrhythmia in wakefulness and sleep (GNP, strong recommendation, high evidence) [24]. The EEG should be performed within a few days and treatment started as early as possible with a fast-acting drug; response is assessed clinically and electroencephalographically about 14 days later, with clinical success defined as absence of spasms for 48 hours and electroencephalographic response as disappearance of hypsarrhythmia; a mere reduction in spasms is not a sufficient response [24]. NICE recommends reviewing the child at least weekly during treatment, repeating the sleep EEG at 2 weeks and, after cessation, reviewing monthly and repeating the EEG if there is recurrence [23].

As first line, NICE recommends offering the combination of high-dose oral prednisolone with vigabatrin for spasms not due to tuberous sclerosis, unless the child is at high risk of corticosteroid adverse effects, in which case vigabatrin alone is considered; in tuberous sclerosis, vigabatrin alone, adding prednisolone if there is no response after 1 week [23]. The GNP recommends hormones (adrenocorticotrophic hormone, ACTH, or prednisolone) alone or the combination of hormones with vigabatrin (strong recommendation, high evidence, consensus >75%), and vigabatrin as first line in tuberous sclerosis or when hormones are contraindicated [24]. The basis of the combination recommendation is the International Collaborative Infantile Spasms Study (ICISS), in which vigabatrin added to ACTH or prednisolone stopped spasms at 2 weeks more often than hormone alone, without superiority in developmental and epilepsy outcomes at 18 months; the authors attributed this lack of difference to the fact that non-responders to hormone alone received vigabatrin after 2 weeks [24]. The combination has reports of movement disorders, encephalopathy, and dysautonomia, particularly in children with trisomy 21 [24]. NICE notes that vigabatrin in combination is off-label use [23].

For DEE93, there are no data that allow one strategy to be preferred. Patients with DEE in the 2022 cohort failed several antiseizure medications; in the 2024 review, a girl with spasms at 17 months did not respond to topiramate or to intramuscular ACTH, had a 50% reduction in frequency with levetiracetam and a ketogenic diet and, after vigabatrin was added at a dose of 108.7 mg/kg/day, had 1 to 2 seizures per day, with improvement of the interictal EEG but no cessation at 3 months [9, 12]. An Indian boy with spasms treated successively with ACTH, valproate, levetiracetam, and zonisamide had fewer seizures with profound ID and delay; this report was read in abstract and indirectly [9, 46]. The relevance of rapid cessation of spasms remains, but the family should be informed that, in DEE93, the developmental prognosis depends mainly on the underlying disease and the progressive atrophy [9, 12].

When first line fails, NICE recommends discussion with a tertiary specialist, and the GNP suggests ketogenic diet, sulthiame, topiramate, valproate, zonisamide, or benzodiazepines (strong recommendation, low evidence) [23, 24]. The GNP recommends early evaluation for epilepsy surgery when there is a focal lesion, with the absence of degenerative or metabolic disease as a prerequisite; in DEE93, the lysosomal nature and diffuse progressive atrophy make resective surgery inappropriate in principle [12, 24]. Unvalidated therapies should be used only in clinical trials [24]. The GNP also recommends repeated information for parents and support from a psychologist and a social worker (strong consensus) [24].

7.3. Dosing in infantile epileptic spasms syndrome

Vigabatrin carries an FDA boxed warning for permanent vision loss, with bilateral concentric constriction of the visual field in 30% or more of adults studied, no dose free of risk, risk increasing with dose and cumulative exposure, and frequency poorly characterised in infants owing to difficulty of assessment [48]. The Portuguese SmPC reports visual field defects in about one-third of treated patients [49]. Visual monitoring intervals differ between labels (section 11), and the GNP notes that perimetry is reliable only from a developmental age of 8 years and that electroretinography in infants requires sedation, which in DEE93, with lack of eye contact in 65% and optic atrophy in 15%, makes baseline assessment and the interpretation of visual changes particularly difficult [12, 24]. In the FDA setting, vigabatrin is dispensed under a risk evaluation and mitigation strategy (REMS) programme [48]. Vigabatrin-induced MRI signal changes in infants should be considered when reading serial MRI of patients with DEE93, who already have hypomyelination [12, 48].

7.4. Other seizures and chronic epilepsy

Outside spasms, the seizures of DEE93 (focal, tonic, tonic-clonic, myoclonic, atonic, absence) are treated according to the general principles of paediatric epilepsy; there are no comparative studies for the gene [9, 12]. Reports of response focus on levetiracetam: seizures controlled in the index case and in the family with p.Leu378Phe, partial improvement in combination in other cases [9, 14, 46]. The three patients with childhood focal epilepsy described in 2024 responded well to antiseizure medications, without the abstract stating which [29]. In the 2025 Chinese series, seizures were controlled in six of ten children and four continued to have seizures despite three or more antiseizure medications [13]. The authors of the 2024 review propose that presynaptically acting antiseizure medications (levetiracetam, brivaracetam, gabapentin, pregabalin) may be more effective, given the role of V-ATPase in the loading of synaptic vesicles; this is an untested hypothesis [9].

Valproate warrants particular caution in DEE93, because the label identifies children under 2 years, with a congenital metabolic disorder, with severe epilepsy and ID, and with organic brain disease as the group at highest risk of fatal hepatotoxicity, and severely affected patients with DEE93 combine several of these characteristics [12, 44]. The supplements (L-carnitine, vitamin B12, calcium) used in a seizure-free patient in the 2024 review, in combination with rehabilitation, coincided with improvement on developmental scales, but without a control group the improvement cannot be attributed to the supplements [9].

7.5. Multidisciplinary care in DEE93

Long-term follow-up should be provided by paediatric neurology, with developmental support (physiotherapy, occupational and speech therapy, augmentative communication, given that no patient in the 2022 cohort developed communicative language), feeding assessment with gastrostomy when necessary (19% in the 2022 cohort), orthopaedic surveillance of hypotonic tetraplegia, contractures, and arthrogryposis, and ophthalmological assessment of optic atrophy and eye contact [8, 12, 39]. Regular dental review is justified by enamel dysplasia in 42%, with eruption defects, excessive caries, and friable teeth, and should include preventive measures and rehabilitation according to severity [8, 12]. Seizure-free patients should be monitored, because seizures may appear later [9]. The associated disorders in the mild forms (autistic traits, ADHD) benefit from neuropsychological assessment [12].

7.6. Care in ARCL2D

The neonatal period is the most dangerous: respiratory distress and heart failure may require ventilation and neonatal intensive care, as in the Hungarian brothers (8 and 3 days of ventilation; one died at 8 days) [15]. After the first year, follow-up includes cardiology (cardiac anomalies in four of six, aortic dilatation in one), respiratory function (one patient developed a restrictive syndrome and was monitored twice a year), neuromuscular care (myopathy with raised creatine kinase), orthopaedics (severe scoliosis operated on at 14 years in one patient; chest deformities, contractures, hip dysplasia), ophthalmology (entropion, cataracts), and development [12, 15]. General measures for cutis laxa include skin protection, plastic surgery, and surveillance for hernias and of the aorta.

7.7. Regulatory status

No drug is authorised for DEE93 or ARCL2D in any jurisdiction consulted [8, 45]. Registration and package leaflet data from the Brazilian Health Regulatory Agency (ANVISA, Brazil) were not consulted.

7.8. Clinical trials

None of these trials is specific to ATP6V1A. DEEp OCEAN is testing bexicaserin (LP352), registered by Longboard Pharmaceuticals, and EMERALD relutrigine, from Praxis Precision Medicines, described by the patient association as a sodium channel modulator; both accept DEE of mixed aetiology and are publicised by the v-ATPase Alliance as accessible to patients with V-ATPase disorders [26, 51, 52]. The results, when published, cannot be read as proof of efficacy in DEE93 without subgroup analysis, and relative reductions in seizure frequency should be interpreted against the absolute baseline frequency. Indirect comparison between trials with different populations and outcomes is not valid. No specific negative trials, suspensions, or withdrawals were found, but a negative search does not prove that none exist [45].

7.9. Developments in the last 12 months

Between November 2025 and October 2026, the literature on the gene added a clinical series of 10 Chinese children including one case of mosaicism (December 2025; observational study) and the detection of a variant of uncertain significance in an Indonesian exome series (August 2026; observational study) [13, 38]. At the preclinical level, the identification of ROGDI as a subunit of the Rabconnectin-3 complex involved in V-ATPase assembly and the interaction of TBC1D24 with V-ATPase in the regulation of neuronal pH were published in 2025 [36, 37]. Among regulatory and guidance documents, NICE NG217 was updated on 5 August 2026 and the FDA label for Acthar® Gel was republished on 16 October 2026 [23, 47]. The two phase 3 trials in mixed DEE still have no results [51, 52]. There is no authorisation, opinion, or trial targeted at ATP6V1A.

7.10. Agents to avoid, special populations, perioperative care, and transition

Corticosteroids and ACTH. Before hormonal treatment, the GNP recommends MRI, examination, blood pressure, full blood count, blood glucose, electrolytes, liver and renal function, and echocardiography; live vaccines are contraindicated from 4 weeks (minimum 2) before until 8 weeks after, and inactivated vaccines should be at least 1 week before and 1 week after [24]. During treatment: gastric protection, early antibiotic therapy for infections, prevention of varicella (serology and passive immunisation within 72 hours of an exposure), blood pressure weekly and then twice weekly after 4 weeks, weekly blood glucose, and echocardiography if there is hypertension [24]. NICE recommends discussing adverse effects, testing for varicella-zoster virus antibodies, and issuing a steroid card before prednisolone, monitoring blood pressure and glycosuria weekly, and tapering according to guidance on adrenal insufficiency [23]. The hypothalamic-pituitary-adrenal axis may remain suppressed for up to 1 year, with a steroid card during this period and the physiological prednisolone dose of 2.5–3 mg/m²/day as the reference for stress cover [24].

Perioperative care. Adrenal suppression after corticosteroids or ACTH in the previous 12 months should be considered for any surgery (for example, gastrostomy or orthopaedic surgery) [24]. In ARCL2D, neonatal respiratory failure, myopathy, and the scoliosis operated on in one patient indicate pre-anaesthetic respiratory and neuromuscular assessment; the sources consulted do not describe specific anaesthetic complications in either disease [15].

Pregnancy and lactation. There are no reports of pregnancy in patients with DEE93 or ARCL2D in the sources consulted. Valproate carries a fetal risk of major malformations (including neural tube defects), decreased IQ, and neurodevelopmental disorders, and should not be used in women of childbearing potential unless other options have failed, with effective contraception [44]. This restriction is relevant for girls with mild forms who reach adolescence.

Puberty and transition. One patient had precocious puberty at 9 years [2]. The oldest patient in the 2022 cohort was 22 years old, and there are no data on adult course, transition to adult care, or reproduction; the authors warn that the mild form may worsen in the long term [12].

8. Reproductive Options and Family Counselling

8.1. DEE93: de novo variant, mosaicism, and familial forms

In DEE93, the variant arises de novo in almost all patients: 26 of 26 in the 2022 cohort and 10 of 10 in the 2025 series [12, 13]. When the variant is not detected in the blood of either parent, the recurrence risk for siblings is low, of the order of 1.3% according to the general estimate for de novo DEE, but not zero, because germline or somatic mosaicism in a parent may exist without being detected in blood; if parental mosaicism is identified, the risk increases and should be estimated case by case [39]. The sources consulted do not describe parental mosaicism in ATP6V1A; the mosaicism documented in the 2025 series is somatic in the proband [13]. The monozygotic twins in the 2022 cohort shared the same de novo variant, which does not change the risk for other siblings [12].

The mild forms change counselling. The Chinese family with epilepsy and p.Leu378Phe shows that a mildly affected carrier can transmit the variant, with a 50% risk in each pregnancy and variable expressivity in the offspring [14, 39]. Therefore, when an ATP6V1A variant is found in a child with a mild phenotype, the parents should be tested even if they appear healthy or have only a history of febrile seizures or controlled epilepsy. An adult with a mild form has a 50% risk of transmitting the variant, and severity in the children cannot be predicted from the adult's own phenotype [39].

8.2. ARCL2D: recessive inheritance

The parents of a child with ARCL2D are, as a rule, heterozygous, and each pregnancy has a 25% risk of an affected child, 50% of a carrier child, and 25% of a child with neither variant; consanguinity was present in three of the six published patients [15]. Confirmation that the two variants are in trans requires parental testing, as was done in the Hungarian family [15]. Healthy siblings can have carrier testing in adulthood, when the result has reproductive consequences. Carriers of ARCL2D variants have no described phenotype, and the sources do not describe DEE in parents of patients with ARCL2D [8, 15].

8.3. Reproductive options and testing

When the familial variant is known, invasive prenatal diagnosis is available and, for de novo variants with a residual risk from mosaicism, non-invasive prenatal diagnosis [39]. Preimplantation genetic testing applies to families with ARCL2D and to transmitted dominant forms; gamete donation is an alternative in ARCL2D. Diagnosis should be preceded and followed by genetic counselling by qualified professionals, as recommended by the NSGC and NICE for genetic testing in epilepsy [22, 23].

8.4. Counselling of minors and cascade testing

Testing of asymptomatic minor siblings for DEE93 is not justified when the parents do not carry the variant, because the risk is that of parental mosaicism and there is no preventive intervention; a sibling with symptoms should be tested as a patient. In ARCL2D, the phenotype is congenital and evident, so testing minor siblings without signs serves mainly to determine carrier status, which is deferred until the age at which the person can decide for themselves. In counselling, parents should receive information on the spectrum of severity, including the possibility of a progressive course and of mild forms, on the absence of specific treatment, and on support resources (section 9.4) [12, 26].

9. Special Considerations: Population Diversity and Epidemiology

9.1. What is known about frequency

There is no population prevalence estimate for either disease [8]. The available figures are case counts and frequencies in selected cohorts, which cannot be converted into prevalence. DEE93 accounted for 4 of 1444 probands (0.28%) with developmental or epileptic encephalopathies of presumed genetic cause studied by exome in cohorts from Europe, Japan, and a third group; this value measures the proportion in a research cohort enriched for severe cases, not the frequency in the population with epilepsy or in the general population [2]. The 2022 cohort brought together 26 patients from Europe, the United States, and Japan; the 2024 review counted 33 published patients; and a single-centre series from Beijing identified 10 children between 2019 and 2024 [9, 12, 13]. In ARCL2D, six patients had been published up to 2021, of Hungarian, Turkish, German, and Pakistani origin [15]. For the cutis laxas as a whole, the only estimate available in GeneReviews is 1:4 000 000, based on a French regional registry (Eurocat Rhône-Alpes) and a personal observation, a weak figure that does not apply to ARCL2D on its own [21].

In the context of the class, epilepsy affects about 0.64% of the population according to the NSGC guideline [22]. The clinical genetics reference textbook consulted gives values for the incidence of childhood epilepsy and the prevalence of infantile spasms whose order of magnitude is not compatible with other sources; these figures are not reproduced, and the sources consulted do not provide a primary estimate of the incidence of IESS [39, 41].

The data collection programme of the patient association v-ATPase Alliance with RARE-X, started on 19 March 2024, indicates that ATP6V1A accounts for 61% of those enrolled with V-ATPase disorders and that 79.7% have or have had epilepsy; the data are self-reported and preliminary, the page does not give the absolute number enrolled, and the composition reflects who enrols, not the real distribution of the diseases [26].

9.2. Populations

The published patients with DEE93 come from Europe, the United States, Japan, China, India and a family of North African origin [2, 12, 13, 30, 46]; the founding work included patients of Italian, Turkish, Japanese and Mexican ancestry [16], and an Indonesian series recorded a variant of uncertain significance [38]. Because the variants are de novo, no founder effects or differences in frequency by ancestry are expected, and no recurrent variant is associated with a population [12]. In ARCL2D, consanguinity explains homozygosity in Turkish and Pakistani families [15]. The sources consulted include no data from Portugal or Brazil for either disease; the Brazilian reference textbook describes ARCL2D on the basis of the international literature [6]. The absence of data in the sources consulted does not demonstrate an absence of patients in these countries.

9.3. Special populations

Infants are the group in which the highest-risk therapeutic decisions (hormones, vigabatrin) are concentrated, with the precautions in section 7.10 [24, 48]. Patients with tetraplegia, gastrostomy, and lack of eye contact require adapted assessment of vision, nutrition, and pain. Girls with mild forms who reach adolescence should have their antiseizure treatment reviewed in view of reproductive potential, especially if they are taking valproate [44]. Adults are almost absent from the literature (maximum of 22 years in the 2022 cohort) [12].

9.4. European Reference Networks and patient associations

The European Reference Network for rare and complex epilepsies, EpiCARE (https://epi-care.eu), brings together specialist centres, a registry, and working groups and is the relevant network for DEE93 [25]. The v-ATPase Alliance (https://vatpasealliance.org) is a non-profit association in the United States, founded by families with ATP6V1A and other V-ATPase disorders, which maintains a family group, meetings, "genetic passports", the patient registry with RARE-X (https://rare-x.org/v-atpase/), a clinical data collection partnership with Citizen Health, and biobank collections with partners of the COMBINEDBrain consortium, and publicises the trials in mixed DEE [26]. The 2022 cohort was gathered in part with the help of a family group created after the first report [12]. General epilepsy associations, such as Epilepsy Action (United Kingdom) and the Epilepsy Foundation (United States), are listed in the reference textbook [39]. For ARCL2D, the resources and networks are those of the hereditary cutis laxas.

10. Illustrative Clinical Case

10.1. Mild form: febrile seizures and mild ID without epilepsy (p.Val28Met)

Sirchia and colleagues described in 2024 a 6-year-old boy seen in January 2018 in the paediatric emergency department for a 5-minute febrile bilateral tonic-clonic seizure during an influenza A virus infection, which resolved spontaneously; the postictal period included confusion, transient hypertonia with hyperreflexia, and complete amnesia [30]. He had been born at term, by elective caesarean section, to consanguineous parents of North African origin, after an uneventful pregnancy, with normal weight, length, and head circumference (36 cm, +1 SD) [30]. Development was considered normal until 5 years, when expressive and receptive language disorder and mild ID were diagnosed (full-scale IQ 64; verbal 55; performance 76); two years earlier, the parents had noticed a febrile "absence" episode lasting less than 1 minute, which was not investigated [30].

Examination showed microcephaly (head circumference 48 cm, −2.2 SD), a long face, mild malar hypoplasia, slightly downslanting palpebral fissures, mild hypotelorism, high-set ears, and enamel hypoplasia, as well as gross and fine motor dyspraxia [30]. Cerebrospinal fluid was normal. Prolonged video-EEG showed slow high-voltage theta background activity with bilateral multifocal spikes predominating on the left; brain MRI, audiometry, and abdominal ultrasound were normal, and the single nucleotide polymorphism array showed no microdeletions or microduplications [30]. A multigene epilepsy panel identified the de novo heterozygous variant NM_001690.4:c.82G>A p.(Val28Met), absent from gnomAD, which the authors classified as pathogenic [10, 30]. The classification included the PVS1 criterion, which does not apply to a missense variant, but the de novo status, population absence, and computational predictions support it (section 5.6) [30].

After five years of follow-up, the patient had had no further seizures, retained mild non-progressive ID, language delay, and minimal dyspraxia, and had developed no MRI changes (repeated at 6 and 11 years) or ADHD or oppositional disorder; at 8 years, head circumference was 49 cm (−2.4 SD), and the final EEG at 11 years was normal, with an 8 Hz occipital alpha rhythm [30]. The article does not mention maintenance antiseizure treatment [30]. The figures of the article (https://pmc.ncbi.nlm.nih.gov/articles/PMC11431710/), consulted through their legends in the full text, show the facial phenotype and enamel dysplasia (Figure 1), the sleep EEG at 5 years and the normal waking EEG at 11 years (Figure 2), and the MRI without hypomyelination or atrophy (Figure 3) [30]. The legend of Figure 2 describes the first sleep EEG as having a normal background, with some frontotemporal delta waves, whereas the text describes the prolonged video-EEG with a slow background and multifocal spikes; the two descriptions refer to different recordings or diverge within the article itself [30].

Clinical reading. The case shows the mild end of the spectrum: seizures only with fever, static mild ID, normal MRI, mild microcephaly, and dysplastic enamel, the latter being the finding that brings the phenotype closer to typical DEE93 [12, 30]. The p.Val28Met variant lies in the N-terminal domain, next to Pro27, which is the only residue in the 2022 cohort outside the central domain, but p.Pro27Arg caused severe DEE, so position does not predict severity [2, 12]. The absence of maintenance treatment after a febrile seizure with a prolonged postictal state was an individual decision of the report and does not constitute a recommendation.

10.2. Severe form with spasms (p.Trp354Leu)

Ma and colleagues described in 2024 a girl with spasms at 17 months, hypsarrhythmia, and MRI with hypomyelination and ventriculomegaly (Figure 1a of the article, https://pmc.ncbi.nlm.nih.gov/articles/PMC11919110/), carrying the de novo variant c.1061G>T p.(Trp354Leu) (Figure 2) [9]. Oral topiramate and intramuscular ACTH were ineffective; levetiracetam combined with a ketogenic diet for 6 months reduced frequency by 50%, and vigabatrin at a dose of 108.7 mg/kg/day reduced seizures to 1 to 2 per day, with improvement of the interictal EEG, without complete cessation at 3 months [9]. The therapeutic sequence in the report (topiramate and ACTH alone before vigabatrin) differs from the first line of current guidelines, which favour hormone and/or vigabatrin from the outset (section 7.2) [23, 24].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

In DEE93, prognosis depends on the presentation and, in part, on the variant. In the 2022 cohort, 3 of the 4 patients with early lethal encephalopathy died (at 6 weeks, 5 months, and 2 years); 10 of the 16 patients with DEE (63%) acquired no motor, communicative, or developmental skills, and 2 of the 6 who acquired motor skills lost them; the 6 patients with ID and epilepsy (23%) walked unaided, but none of the 26 developed communicative language [12]. Cerebral atrophy worsened in all 8 patients with repeat MRI, and even in the mild group the authors consider late worsening possible [12]. Later series show a better prognosis at the mild end: controlled focal epilepsy with normal development in 3 of 3 patients, seizures controlled in 6 of 10 with normal development in 2 of 10, and five years without seizures and without progression in the case in section 10 [13, 29, 30]. Seizure control was not accompanied by improvement of the delay in severely affected patients [9]. Life expectancy in the non-lethal forms is unknown; the oldest patient described was 22 years old [12].

In ARCL2D, 2 of the 6 patients died early (at 8 days and at 3 months), from heart and respiratory failure in the documented case; the 3 patients with follow-up into adolescence (15 to 17 years) had improvement of the cutis laxa and normal cognition or mild ID, one of them attending secondary school at 17 years and another maintaining normal strength and endurance at 15 [8, 15].

11.2. Surveillance

There is no published surveillance protocol for either disease; the following items and intervals come from the IESS guidelines, the drug labels and, when no published interval exists, the natural history, this gap being indicated in the table [8].

12. Summary and Final Considerations

ATP6V1A encodes the catalytic subunit A of V-ATPase, the proton pump that acidifies endolysosomal, secretory, and synaptic compartments, and causes two distinct diseases. DEE93, autosomal dominant, results from missense variants that are almost always de novo, clustered in the central nucleotide-binding domain, and ranges from early lethal encephalopathy with massive atrophy to severe DEE without acquisitions, moderate ID with epilepsy and, in recent series, childhood focal epilepsy with normal development. In the reference cohort of 26 patients, early hypotonia was the first manifestation in 75% of patients with information, epilepsy appeared in 80%, infantile spasms dominated infant-onset DEE, no patient acquired communicative language, acquired microcephaly occurred in 38%, enamel dysplasia in 42%, and cerebral atrophy progressed on all repeat MRIs. The variants have opposite effects on endolysosomal pH, loss of proton pumping in the severe forms and hyperacidification in the milder ones, and both disturb lysosomal function, with mild substrate accumulation and inclusions, which makes DEE93 a dominant lysosomal disease.

Diagnosis is by exome, genome, or panel, which guidelines indicate in epilepsy of early onset or with delay, and dysplastic enamel with hypomyelination and progressive atrophy is the most useful clinical clue. There is no targeted treatment; seizures are treated according to the syndrome, with IESS managed according to current guidelines (hormones and/or vigabatrin, with the dosing and monitoring of the guidelines and the labels as comparison), and developmental prognosis depends mainly on the underlying disease. Counselling rests on the de novo nature, with a residual risk from mosaicism, and on the possibility of transmission of mild forms. ARCL2D, autosomal recessive and known in six patients up to 2021, combines congenital cutis laxa, a progeroid face, myopathy, and neonatal cardiorespiratory risk, with improvement in survivors; seizures occurred in a minority, which resolves the discrepancy between sources, and management follows that of the hereditary cutis laxas. Prevalence estimates, adult data, surveillance protocols, and any trial targeted at the gene are lacking.

13. References

1. National Center for Biotechnology Information. MedGen: Developmental and epileptic encephalopathy 93 (C4693934; OMIM 618012) and Autosomal recessive cutis laxa type 2D (C4479409; OMIM 617403), with OMIM definitions [Internet]; accessed 2026 Oct 28. Available from: https://www.ncbi.nlm.nih.gov/medgen/C4693934
2. Fassio A, Esposito A, Kato M, Saitsu H, Mei D, Marini C, et al. De novo mutations of the ATP6V1A gene cause developmental encephalopathy with epilepsy. Brain. 2018;141(6):1703-1718. doi:10.1093/brain/awy092. PMID: 29668857. PMCID: PMC5972584.
3. Orphanet. Non-specific early-onset epileptic encephalopathy (ORPHA:442835). Available from: https://www.orpha.net/en/disease/detail/442835
4. National Center for Biotechnology Information. ClinVar: ATP6V1A[gene], pathogenic/likely pathogenic records [Internet]. Bethesda (MD): NLM; accessed 2026 Oct 28. Available from: https://www.ncbi.nlm.nih.gov/clinvar/?term=ATP6V1A%5Bgene%5D
5. Jaeken J, Morava E. Congenital Disorders of Glycosylation, Dolichol and Glycosylphosphatidylinositol Metabolism. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 43 (introduction, Table 43.1; 43.4.5; 43.4.7). doi:10.1007/978-3-662-63123-2_43
6. Diniz MAN, Pinto CX, Ribeiro EM, Carvalho EDF. Cutis laxa. In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 273–277.
7. Orphanet. Autosomal recessive cutis laxa type 2, classic type (ORPHA:357074). Available from: https://www.orpha.net/en/disease/detail/357074
8. GeneGuidance. ATP6V1A — clinical curation by gene (DEE93, OMIM 618012; ARCL2D, OMIM 617403), including ClinGen and Orphadata data. Local database, version of 26 September 2026.
9. Ma J, Zhang H, Lv Y, Gao M, Gai Z, Liu Y. Clinical and Genetic Characteristics of Two Cases With Developmental and Epileptic Encephalopathy 93 Caused by Novel ATP6V1A Mutations and Literature Review. Hum Mutat. 2024;2024:4678670. doi:10.1155/2024/4678670. PMID: 40225911. PMCID: PMC11919110.
10. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: Accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166.
11. OMIM® Online Mendelian Inheritance in Man. *607027 ATPase, H+ TRANSPORTING, LYSOSOMAL, 70-KD, V1 SUBUNIT A; ATP6V1A. Johns Hopkins University; last edited 04/11/2022. Available from: https://omim.org/entry/607027 [accessed 28 October 2026].
12. Guerrini R, Mei D, Kerti-Szigeti K, Pepe S, Koenig MK, Von Allmen G, et al. Phenotypic and genetic spectrum of ATP6V1A encephalopathy: a disorder of lysosomal homeostasis. Brain. 2022;145(8):2687-2703. doi:10.1093/brain/awac145. PMID: 35675510. PMCID: PMC10893886.
13. Ouyang SJ, Wang T, Tan QZ, Li Y, Dong ZY, Liu CH, et al. [Genetic and clinical characteristics in epilepsy patients with ATP6V1A gene variants]. Zhonghua Er Ke Za Zhi. 2025;63(12):1354-1359. Chinese. doi:10.3760/cma.j.cn112140-20250716-00639. PMID: 41233134.
14. Xiaoquan C, Yuting L, Pu M, Haiying C, Zheng W, Ye W, et al. A heterozygous pathogenic variant in the ATP6V1A gene triggering epilepsy in a large Chinese pedigree. Clin Neurol Neurosurg. 2023;233:107956. doi:10.1016/j.clineuro.2023.107956. PMID: 37729800.
15. Vogt G, El Choubassi N, Herczegfalvi Á, Kölbel H, Lekaj A, Schara U, et al. Expanding the clinical and molecular spectrum of ATP6V1A related metabolic cutis laxa. J Inherit Metab Dis. 2021;44(4):972-986. doi:10.1002/jimd.12341. PMID: 33320377. PMCID: PMC8638669.
16. OMIM® Online Mendelian Inheritance in Man. #618012 DEVELOPMENTAL AND EPILEPTIC ENCEPHALOPATHY 93; DEE93. Johns Hopkins University; last edited 03/11/2022. Available from: https://omim.org/entry/618012 [accessed 28 October 2026].
17. OMIM® Online Mendelian Inheritance in Man. #617403 CUTIS LAXA, AUTOSOMAL RECESSIVE, TYPE IID; ARCL2D. Johns Hopkins University; last edited 20/10/2022. Available from: https://omim.org/entry/617403 [accessed 28 October 2026].
18. Orphanet. Defect in V-ATPase (ORPHA:309778). Available from: https://www.orpha.net/en/disease/detail/309778
19. Clinical Genome Resource. Gene-disease validity curations (download) [Internet]; accessed 2026 Oct 28. ATP6V1A – autosomal recessive cutis laxa type 2D, Limited (2024-04-03). Available from: https://search.clinicalgenome.org/kb/gene-validity/CGGV:assertion_f9b7d66a-3d77-4515-aa6b-9f33e59f2f09-2024-04-03T160000.000Z
20. Callewaert BL, Urban Z. ELN-Related Cutis Laxa. 2022 Sep 29. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK584550/
21. Van Maldergem L, Dobyns W, Kornak U. ATP6V0A2-Related Cutis Laxa. 2009 Mar 19 [updated 2021 Jan 28; revised 2023 Mar 16]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK5200/
22. Smith L, Malinowski J, Ceulemans S, Peck K, Walton N, Sheidley BR, et al. Genetic testing and counseling for the unexplained epilepsies: An evidence-based practice guideline of the National Society of Genetic Counselors. J Genet Couns. 2023;32(2):266-280. doi:10.1002/jgc4.1646. PMID: 36281494.
23. National Institute for Health and Care Excellence. Epilepsies in children, young people and adults (NG217). London: NICE; 2022 Apr 27 [last updated 2026 Aug 5]. Recommendations 1.4 (genetic testing) and 6.3 (infantile spasms syndrome). Available from: https://www.nice.org.uk/guidance/ng217
24. Ramantani G, Bölsterli BK, Alber M, Klepper J, Korinthenberg R, Kurlemann G, et al. Treatment of Infantile Spasm Syndrome: Update from the Interdisciplinary Guideline Committee Coordinated by the German-Speaking Society of Neuropediatrics. Neuropediatrics. 2022;53(6):389-401. doi:10.1055/a-1909-2977. PMID: 35882373. PMCID: PMC9643068.
25. European Reference Network for rare and complex epilepsies (EpiCARE) [Internet]; accessed 2026 Oct 28. Available from: https://epi-care.eu
26. v-ATPase Alliance [Internet]. Houston (TX): v-ATPase Alliance; 2026 [accessed 2026 Oct 28]. Available from: https://vatpasealliance.org
27. ERN-Skin. MCTD — Mendelian Connective Tissue Disorders thematic group. Available from: https://ern-skin.eu/mctd/ [accessed 2026-10-27].
28. Disorders of cellular trafficking. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 44 (44.1.3; 44.2.1).
29. Li B, Lan S, Liu XR, Ji JJ, He YY, Zhang DM, et al. ATP6V1A variants are associated with childhood epilepsy with favorable outcome. Seizure. 2024;116:81-86. doi:10.1016/j.seizure.2023.08.004. PMID: 37574426.
30. Sirchia F, Taietti I, Donesana M, Bassanese F, Clemente AM, Barbato E, et al. Expanding the Spectrum of Autosomal Dominant ATP6V1A-Related Disease: Case Report and Literature Review. Genes (Basel). 2024;15(9):1219. doi:10.3390/genes15091219. PMID: 39336810. PMCID: PMC11431710.
31. Van Damme T, Gardeitchik T, Mohamed M, Guerrero-Castillo S, Freisinger P, Guillemyn B, et al. Mutations in ATP6V1E1 or ATP6V1A Cause Autosomal-Recessive Cutis Laxa. Am J Hum Genet. 2017;100(2):216-227. doi:10.1016/j.ajhg.2016.12.010. PMID: 28065471. PMCID: PMC5294668.
32. Groszer M. Rare human ATP6V1A variants provide unique insights into V-ATPase functions. Brain. 2022;145(8):2626-2628. doi:10.1093/brain/awac255. PMID: 35857819.
33. Zuberi SM, Wirrell E, Yozawitz E, Wilmshurst JM, Specchio N, Riney K, et al. ILAE classification and definition of epilepsy syndromes with onset in neonates and infants: Position statement by the ILAE Task Force on Nosology and Definitions. Epilepsia. 2022;63(6):1349-1397. doi:10.1111/epi.17239. PMID: 35503712.
34. Specchio N, Wirrell EC, Scheffer IE, Nabbout R, Riney K, Samia P, et al. International League Against Epilepsy classification and definition of epilepsy syndromes with onset in childhood: Position paper by the ILAE Task Force on Nosology and Definitions. Epilepsia. 2022;63(6):1398-1442. doi:10.1111/epi.17241. PMID: 35503717.
35. Esposito A, Pepe S, Cerullo MS, Cortese K, Semini HT, Giovedì S, et al. ATP6V1A is required for synaptic rearrangements and plasticity in murine hippocampal neurons. Acta Physiol (Oxf). 2024;240(8):e14186. doi:10.1111/apha.14186. PMID: 38837572.
36. Pepe S, Aprile D, Castroflorio E, Marte A, Giubbolini S, Hopestone S, et al. TBC1D24 interacts with the v-ATPase and regulates intraorganellar pH in neurons. iScience. 2025;28(1):111515. doi:10.1016/j.isci.2024.111515. PMID: 39758816. PMCID: PMC11699390.
37. Winkley SR, Kane PM. The ROGDI protein mutated in Kohlschutter-Tonz syndrome is a novel subunit of the Rabconnectin-3 complex implicated in V-ATPase assembly. J Biol Chem. 2025;301(4):108381. doi:10.1016/j.jbc.2025.108381. PMID: 40049412. PMCID: PMC11997317.
38. Herini ES, Triono A, Iskandar K, Nugrahanto AP, Damroni RA, Mooiindie KH, et al. Whole-exome sequencing in pediatric drug-resistant epilepsy: diagnostic yield and clinical impact in a resource-limited setting. Neurogenetics. 2026;27(1):61. doi:10.1007/s10048-026-00932-0. PMID: 42671607.
39. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 410–414 (Epilepsy in infants and children).
40. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 204–205 (Large fontanelle).
41. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 308–310 (Seizures with developmental delay/intellectual disability).
42. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 331–332 (Unusual hair, teeth, nails and skin).
43. Clinical approach to inborn errors of metabolism in paediatrics. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 1 (1.5; 1.6.2, Table 1.34; 1.6.16).
44. Depakote (divalproex sodium) delayed-release tablets. Prescribing information. NDA 018723. U.S. Food and Drug Administration; label effective 2026-03-31 (openFDA set_id 08a65cf4-7749-4ceb-6895-8f4805e2b01f). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=08a65cf4-7749-4ceb-6895-8f4805e2b01f
45. ClinicalTrials.gov. Searches “ATP6V1A”, “developmental and epileptic encephalopathy 93”, “V-ATPase” [Internet]; accessed 2026 Oct 28. NCT04537832 (ENVISION, Encoded Therapeutics). Available from: https://clinicaltrials.gov/study/NCT04537832
46. Kadwa RA. Novel Mutation in ATP6V1A Gene with Infantile Spasms in an Indian Boy. Neuropediatrics. 2020;51(4):292-294. doi:10.1055/s-0040-1701657. PMID: 32045939.
47. Acthar Gel (repository corticotropin injection). Prescribing information. DailyMed set_id 7b48ddec-e815-45f4-9ca0-5c0daaf56f30 [Keenova ARD LLC]; published 2026 Oct 16. Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=7b48ddec-e815-45f4-9ca0-5c0daaf56f30
48. Lundbeck Pharmaceuticals LLC. Sabril (vigabatrin) tablets and powder for oral solution. Prescribing information. NDA 020427. U.S. Food and Drug Administration; label effective 2021-10-20 (openFDA set_id a5d389d2-d0e1-4395-a2a2-b552808e7f98). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=a5d389d2-d0e1-4395-a2a2-b552808e7f98
49. Sanofi – Produtos Farmacêuticos, Lda. Sabril 500 mg comprimidos revestidos por película (vigabatrina). Resumo das Características do Medicamento. Lisbon: INFARMED (Infomed); marketing authorisation 1991-01-04, renewal 2005-11-22. Available from: https://extranet.infarmed.pt/INFOMED-fo/
50. Keppra (levetiracetam) tablets and oral solution. Prescribing information. NDA 021035/021505. U.S. Food and Drug Administration; label effective 2026-01-28 (openFDA set_id 3ca9df05-a506-4ec8-a4fe-320f1219ab21). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=3ca9df05-a506-4ec8-a4fe-320f1219ab21
51. Longboard Pharmaceuticals. A Phase 3, Randomized, Double-Blind, Placebo-Controlled, Multicenter Study to Investigate the Efficacy, Safety, and Tolerability of LP352 in the Treatment of Seizures in Children and Adults With Developmental and Epileptic Encephalopathies (DEEp OCEAN). ClinicalTrials.gov ID NCT06719141; last update 2026 Sep 15. Available from: https://clinicaltrials.gov/study/NCT06719141
52. Praxis Precision Medicines. A Phase 3, Randomized, Multi-Center, Double-Blind, Placebo-Controlled Clinical Trial to Evaluate the Efficacy, Safety, Tolerability, and Pharmacokinetics of Relutrigine in Participants With DEE Followed by an Open-Label Extension (EMERALD). ClinicalTrials.gov ID NCT07010471; last update 2026 Aug 4. Available from: https://clinicaltrials.gov/study/NCT07010471