ZBTB18- and HNRNPU-Related Disorders and 1q43q44 Microdeletion Syndrome

Class of neurodevelopmental disorders of the 1q43q44 subtelomeric region: haploinsufficiency of ZBTB18 (intellectual developmental disorder, autosomal dominant 22), of HNRNPU (developmental and epileptic encephalopathy 54) and contiguous gene deletion that also includes AKT3.

Literature and regulatory information updated: 28 October 2026.

0. Reference Summary Table

1. Introduction

1.1. From the 1q subtelomeric deletion to the genes of the region

Monosomy of the distal end of the long arm of chromosome 1 was first recognised as a rare chromosomal anomaly, which Orphanet defines by the association of intellectual disability, progressive microcephaly, convulsions, growth retardation, facial dysmorphism and midline defects, including cardiac, corpus callosum, gastro-oesophageal and urogenital anomalies [4]. Johnson and colleagues proposed in 1985 distal 1q deletion as a definable syndrome, based on a female infant with a 1q42-qter deletion, microcephaly, psychomotor delay, hypotonia, poor weight gain and dysmorphism, and on 14 published cases; the common face included a round face with a marked Cupid's bow and downturned corners of the mouth, a thin vermilion, a long upper lip with a smooth philtrum, a short broad nose, epicanthus, low-set ears and micrognathia, and the deletion was de novo in 9 of the 15 cases [18]. In 2001, de Vries and colleagues described the first submicroscopic 1qter subtelomeric deletion, in a boy with ID, convulsions, short stature, microcephaly, partial agenesis of the corpus callosum, absent speech and delayed visual use [18]. Van Bon and colleagues gathered in 2008 eleven unrelated patients with 1q43-q44 deletion, with severe ID, marked speech deficit, hypotonia and motor delay (some learned to walk) and, frequently, low birth weight, microcephaly, short stature, convulsions and feeding difficulties; the face included a round face, deep-set eyes, a prominent metopic ridge, a short nose with a broad tip, a thin arched upper lip and widely spaced teeth, with low-set ears, hypertelorism, epicanthus, long philtrum, cleft palate and micrognathia being infrequent. The nine patients with corpus callosum anomalies shared a 360 kb region at 1q44 that did not include AKT3; two siblings with a deletion inherited from an unaffected mother and a different phenotype were left outside the definition of the syndrome [18]. The year before, Boland and colleagues had defined by rearrangement mapping a 3.5 Mb critical region for microcephaly and agenesis of the corpus callosum and located, in a patient with postnatal microcephaly and agenesis of the corpus callosum, the breakpoint of a balanced translocation t(1;13)(q44;q32) about 20 kb upstream of AKT3; sequencing of AKT3 in 45 patients with agenesis of the corpus callosum revealed no pathogenic variants, but expression of the orthologue in the central nervous system of the mouse embryo and the phenotype of the Akt3-null mouse made it the candidate for microcephaly [18, 22]. With the introduction of array comparative genomic hybridisation (array-CGH), deletions were no longer seen as a uniform block and began to be sized precisely, which made it possible to compare patients with overlapping deletions and to search for the critical regions of each feature [13, 26]. Caliebe and colleagues described in 2010 four patients with speech delay, convulsions and variable corpus callosum thickness who shared a 0.440 Mb deletion at 1q44 with FAM36A (now COX20), HNRNPU, EFCAB2 and KIF26B, retaining two copies of AKT3 and ZNF238, and proposed HNRNPU as a candidate for epilepsy [29].

In 2012, three studies established the model that still guides clinical interpretation. Ballif and colleagues studied 22 individuals with 1q43q44 microdeletions by high-resolution array-CGH and defined three critical regions: small or intragenic deletions of AKT3 were associated with microcephaly, the critical region for corpus callosum anomalies contained ZNF238 (ZBTB18) and the seizure region included FAM36A, C1ORF199 and HNRNPU; about 90% of cases fitted this model, and the rest were attributed to variable expressivity, incomplete penetrance, position effects or multigenic effects [26]. The deletions in this series measured from 120 kb to 6.0 Mb, all with different breakpoints; 13 of the 22 patients had microcephaly, and all but one of these had the deletion encompassing or disrupting AKT3, including a boy who inherited from his mother, also microcephalic, an intragenic deletion of AKT3; 7 had corpus callosum anomalies, six of whom had deletion of ZNF238, but three patients with deletion of this gene had normal MRI; 9 had epilepsy, and two patients without seizures also had deletion of the seizure critical region. Combining this series with five earlier studies, the model explained 42 of 45 cases with microcephaly (93%), 37 of 43 with agenesis of the corpus callosum (86%) and 40 of 46 with convulsions (87%) [18, 26]. The symbol C1ORF199 designated the antisense locus of HNRNPU (HNRNPU-AS1), now merged into the NCBI Gene record of HNRNPU [12]. Nagamani and colleagues, with seven patients, showed that the four with corpus callosum anomalies shared a minimal region containing only CEP170, SDCCAG8 and ZNF238 [30]. Thierry and colleagues described 11 patients with 1q44 microdeletion, all de novo when both parents were tested, with moderate to severe ID and convulsions; nine had neither microcephaly nor corpus callosum anomalies and had small deletions sparing AKT3 and ZNF238 [15]. Perlman and colleagues reinforced in 2013 the role of ZNF238 with a 1.47 Mb de novo deletion that included this gene, but not CEP170 or AKT3, in a child with dysgenesis of the corpus callosum [31].

1.2. Point variants and definition of the monogenic entities

The next step came from exome sequencing. Rauch and colleagues performed trio sequencing in 51 children with sporadic non-syndromic severe ID from the German Mental Retardation Network and found, among others, a de novo variant in ZBTB18 [13, 32]. In 2014, de Munnik and colleagues described the first de novo nonsense variant in ZBTB18, identified by exome in a child with features of the 1q43q44 microdeletion syndrome (short stature, microcephaly, global delay and pronounced language delay), but with a structurally normal corpus callosum [33]. She was a 34-month-old girl conceived by in vitro fertilisation, with arched eyebrows, telecanthus, short palpebral fissures, a long nose with a prominent tip, a short philtrum, a thin upper lip and micrognathia, and with suspected absence seizures; the variant was c.397G>T p.(Glu133Ter), no other variants of the gene were found in nine patients with a similar phenotype and the Exome Variant Server contained no truncating variants of ZBTB18 [18, 19]. The Portuguese cohort of Lopes and colleagues, with 19 patients with a Rett-like phenotype, added a de novo variant in ZNF238 [34]. Cohen and colleagues gathered in 2017 five new patients and ten from the literature, the DDD study (Deciphering Developmental Disorders) and ClinVar, and concluded that de novo missense and truncating variants in ZBTB18 cause ID with variable features and with corpus callosum anomalies in 4 of 7 patients with imaging [13]. In the same year, Depienne and colleagues dissected 54 1q43q44 deletions and the point variants of ZBTB18 and HNRNPU, attributing microcephaly mainly to AKT3, corpus callosum anomalies to ZBTB18 and epilepsy to HNRNPU, with additive effects between genes [6].

The first point variants of HNRNPU emerged from large-scale studies: in 2013, targeted resequencing of 500 patients with epileptic encephalopathy (Carvill) and the exome of 264 probands of the Epi4K consortium; in 2014, the exome of 41 trios with ID (Hamdan); and, in 2016, sequencing of 351 candidate genes in patients with epileptic encephalopathy (de Kovel), which led OMIM to create the DEE54 entry in 2017 [10, 20]. For HNRNPU, the 2017 series (Yates, Leduc, Bramswig), the Durkin cohort in 2020 and the Taylor cohort in 2022 consolidated a distinct developmental and epileptic encephalopathy, summarised in the 2022 GeneReviews with 83 individuals [3, 35–39]. In 2026, the ClinGen intellectual disability and autism curation group classified the association of ZBTB18 with autosomal dominant complex neurodevelopmental disorder as definitive [7].

1.3. Nomenclature and scope of the article

The OMIM designation “Mental retardation, autosomal dominant 22” (MRD22, OMIM 612337) corresponds, in current terminology, to intellectual developmental disorder, autosomal dominant 22; the same OMIM number is used for the 1q43-q44 chromosomal deletion syndrome [1, 13]. In the version of the entry edited on 12 December 2024, the main title is already intellectual developmental disorder, autosomal dominant 22 and the old form became an alternative title; the # prefix indicates that the molecular basis (heterozygous variants in ZBTB18) is established, with mapping key 3, and the entry includes as entities the 1q43-q44 deletion syndrome and the 1qter deletion syndrome [18]. The gene entry (*608433) records ZNF238, RP58 and C2H2-171 as former designations [19]. ZBTB18 (zinc finger and BTB domain containing 18) is the HGNC-approved designation; ZNF238, RP58 (repressor protein 58 kDa) and TAZ-1 are former designations, frequent in the experimental literature [7, 40]. For HNRNPU (heterogeneous nuclear ribonucleoprotein U, also called SAF-A), the OMIM phenotype is developmental and epileptic encephalopathy 54 (DEE54, OMIM 617391), and GeneReviews uses the designation HNRNPU-NDD [3]. OMIM retains as an alternative title the former designation, early infantile epileptic encephalopathy 54 (EIEE54), and defines DEE54 by psychomotor delay, early-onset refractory seizures, often febrile at onset and afebrile later, and severe ID, referring the genetic heterogeneity of the DEEs to the general phenotypic series [20].

This article treats the 1q43q44 region as a class: ZBTB18-related disorder, HNRNPU-related disorder (a member of the class with its own GeneReviews) and the contiguous gene deletion syndrome, in which AKT3 is the main determinant of microcephaly. The gain-of-function forms of AKT3 (megalencephaly and hemimegalencephaly) and the 1q43-q44 duplication with macrocephaly appear here only as phenotypic inverses [23]. Rauch-Steindl syndrome, sometimes associated with this region in earlier lists, is caused by variants in NSD2 (OMIM 619695). The separation has practical consequences, because the clinical picture differs from that of this class. Rauch-Steindl syndrome (#619695; NSD2, 4p16.3; autosomal dominant) was delineated in 2021 by Zanoni and colleagues with 18 patients from 16 families, identified by exome through GeneMatcher; the variants were mostly de novo, with two families with vertical transmission and incomplete penetrance described in another family [41]. The phenotype corresponds to an attenuated version of Wolf-Hirschhorn syndrome: prenatal and postnatal growth restriction, sometimes short stature, small head circumference, hypotonia, late walking and speech, ID of variable severity and often mild, behavioural changes and facial dysmorphism that is less marked and distinctive enough to justify nosological autonomy; unlike Wolf-Hirschhorn syndrome, epilepsy is exceptional (one patient in the series, with febrile and myoclonic seizures resolved by 8 years) [41]. The first observation was that of Rauch and colleagues in 2001: a 191.5 kb de novo interstitial deletion disrupting NSD2 (then WHSC1) in a boy with a partial Wolf-Hirschhorn phenotype, without seizures [41].

2. Molecular Genetics and Pathophysiology

2.1. The ZBTB18 gene and the RP58 protein

ZBTB18 lies at 1q44 (GRCh38 chr1:244,048,491–244,057,476) and has two exons; only 13 nucleotides of the first exon are coding, so almost the entire coding sequence, including the zinc fingers, is in the second and last exon [7, 13]. The MANE Select transcript is NM_205768.3, which encodes the longest isoform, NP_991331.1, of 531 amino acids; the clinical literature of 2016–2024 uses version NM_205768.2, and the hg19 genomic coordinates published by Depienne and by van der Schoot coincide with the VariantValidator mapping on NM_205768.3 for the confirmed variants [6, 8, 42]. The protein has an N-terminal BTB/POZ domain, which mediates protein–protein interactions, a central region without defined domains and four C-terminal C2H2 zinc fingers (ZnF1–4), which recognise regulatory sequences with an E-box motif (CANNTG) [2, 13, 43].

The gene was cloned twice independently. Becker and colleagues isolated in 1997, from a hippocampal library, the C2H2-171 cDNA, which encoded a protein of 484 amino acids (55 kDa) with an N-terminal POZ domain and four C2H2 zinc fingers; in the human brain, expression was highest in the cerebellum, followed by the occipital pole, the frontal and temporal lobes, the putamen, the amygdala and the hippocampus, and in the mouse it increased in the head from embryonic day 15 until birth, with much higher levels in cerebellar granule neurons than in astrocytes [19]. Aoki and colleagues cloned RP58 in 1998, with 522 amino acids (58 kDa) and a nuclear localisation signal, detected mainly in the brain, in a punctate nuclear pattern associated with electron-dense chromatin; they demonstrated high-affinity binding to the E-box and dose-dependent repression through this sequence [19]. Meng and colleagues suggested in 2000 that the two cDNAs were splice variants, mapped the gene to 1q44-qter by fluorescence in situ hybridisation and described it as intronless [19]. Current annotation distinguishes isoform 1 (NP_991331.1, 531 amino acids) from isoform 2 (NP_006343.2, 522 amino acids, with the length of the original RP58), and the nine-codon difference between them explains the 27-nucleotide offset between numberings on NM_205768 and NM_006352 [13, 44]. The current two-exon annotation differs from that description of an intronless gene; since the first exon of the MANE transcript contains only 13 coding nucleotides, almost the entire coding sequence is nevertheless contained in a single exon [7, 19].

RP58 is a transcriptional repressor of the POK (POZ-Krüppel) family, to which BCL6 and PLZF also belong; it represses its targets by recruiting chromatin remodelling and histone deacetylation complexes, associates with condensed chromatin and interacts through the BTB domain with DNMT3A and FOXG1 [40, 43, 45]. In the developing cortex it is expressed mainly in glutamatergic projection neurons; it controls cell-cycle exit of progenitors, the survival and maturation of cortical neurons and the multipolar–bipolar transition of migration, repressing Neurog2 and Rnd2 (Ngn2–Rnd2 pathway) and, through Neurog2, NeuroD1 and NeuroD2 [13, 40, 43]. In the cerebellum it is required for foliation, lamination and growth, and regulates the proliferation and differentiation of glutamatergic and GABAergic progenitors [13]. Outside the nervous system it promotes skeletal myogenesis by repressing the inhibitors ID2 and ID3 [1]. Yokoyama and colleagues showed in 2009 that MyoD induces Rp58 expression and that this protein binds directly to the Id2 and Id3 promoters; knockdown of Rp58 in C2C12 myoblasts severely disturbed myogenesis [19]. Expression is documented in brain, skeletal muscle, pancreas, testis and spleen [2].

Mouse models establish the effect of loss of the gene on the brain. Complete loss of RP58 is lethal at birth, with neocortical and hippocampal defects; conditional loss in the central nervous system produces microencephaly, agenesis of the corpus callosum, cerebellar hypoplasia and cortical dysplasia, with survival to about three weeks, reduced neuronal and increased glial differentiation, and loss of repression of Pax6, Ngn2 and NeuroD1, of which Ngn2 and NeuroD1 are direct targets [13, 46]. Rp58-null homozygous embryos are of normal size at day 18.5 but die shortly after birth and show marked reduction of the diaphragm and other skeletal muscles, with fewer myofibres, which links neonatal lethality also to the muscle defect [19]. In a 2026 study on the evolution of the mammalian neocortex, Li and colleagues identified a regulatory network centred on Zbtb18 in excitatory neurons; deletion of the gene in these neurons reduced molecular diversity and corticospinal and callosal projections and increased intrahemispheric associative projections to the prefrontal cortex, and Zbtb18 binding motifs were enriched in regulatory elements of neurons with callosal projection, which provides an additional experimental basis for the relationship between the gene and the corpus callosum [19]. The heterozygous mouse, a model of human haploinsufficiency, has no gross cytoarchitectural changes but shows corpus callosum dysplasia, deficits in motor learning, working memory and memory flexibility, and glutamatergic synaptic dysfunction, with lower expression of glutamate receptors, altered NMDA receptor-mediated responses, reduced saturation of long-term potentiation and abnormal dendritic spines [47]. The same model shows early decline of object location memory at 4–5 months, with accumulation of DNA damage, mitochondrial damage and microglial activation in the dentate gyrus [48]. Loss of Zbtb18 also disorganises the dentate gyrus and alters the neuronal composition of the embryonic hippocampus, with increased Prox1 and shorter neurites [49].

2.2. Disease mechanism: haploinsufficiency and altered DNA binding

The gene is intolerant to variation: pLI of 1 and LOEUF of 0.13 in gnomAD v4.0, %HI of 8.38 in DECIPHER, haploinsufficiency score 3 (sufficient evidence) and triplosensitivity score 0 in ClinGen [7]. As early as 2017, the RVIS index placed ZBTB18 among the 6.17% most intolerant human genes, with no loss-of-function variants in ExAC [13]. Deletions including the whole gene and truncating variants act through haploinsufficiency. Since all published truncating variants lie in the last exon, nonsense-mediated mRNA decay (NMD) is not expected; synthesis of truncated proteins lacking the zinc fingers is predicted, and most nonsense and frameshift variants lie before or within ZnF1 [13, 42]. Cohen and colleagues suggested that these truncated proteins, if stable, might have dominant-negative or gain-of-function effects, a hypothesis not demonstrated [13].

Pathogenic missense variants cluster in the zinc fingers (11 of 13 in the Yang 2024 review) and, in smaller numbers, in the BTB domain [25, 50]. In the DNA recognition helices of C2H2 fingers, positions −1, 2, 3 and 6 contact specific bases; Asn461 and Arg464 occupy positions 3 and 6 of ZnF3, and homology modelling shows Arg464 oriented towards the major groove of the DNA, whereas Tyr447 points towards the phosphate backbone [13, 42]. Functional studies have shown that the effect is not reducible to loss of function. Half of the disease missense variants affect residues essential for specific DNA contact (Asn461, Arg464, Gln486) and the remainder residues with a minor contribution (Leu434, Tyr447, Arg495) [51]. The p.(Asn461Ser) variant has low steady-state levels, restored by the proteasome inhibitor MG132, loses transcriptional repression and binds DNA promiscuously; p.(Arg495Gly) has normal stability, binds promiscuously and enhances transcription [43]. With in utero electroporation into the embryonic mouse cortex, knockdown of Zbtb18 impairs neuronal migration and is rescued by the normal protein; p.(Asn461Ser) increases migration and p.(Arg495Gly) worsens the defect [43, 51]. These data suggest altered transcriptional regulation in a different direction depending on the variant.

Interpretation of rare missense variants must take into account that altered in vitro function is not enough to prove pathogenicity: of 12 missense variants from the general population (gnomAD) tested, 7 (58.3%) altered DNA binding, 5 (41.6%) transcriptional activity, 4 (33.3%) both and 4 (33.3%) had negligible impact; population variants lie preferentially at residues without DNA contact [52]. Among five structural approaches compared to separate disease and population missense variants in the zinc fingers, thermodynamic integration and Residue Scanning performed best [53].

2.3. ZBTB18 variant spectrum and nomenclature normalisation

The Yang 2024 review gathered 32 patients with sequence variants: 13 missense (40.6%), 8 nonsense (25%) and 11 frameshift (34.3%) [25]; Wu 2026 counted 33 published pathogenic variants (13 missense, 8 nonsense, 12 frameshift) [2]. In October 2026, ClinVar had 502 records with ZBTB18 (including multigene 1q CNVs): 124 pathogenic, 43 likely pathogenic, 7 pathogenic/likely pathogenic, 180 of uncertain significance (VUS), 16 with conflicting classifications and 122 benign or likely benign. Pathogenic or likely pathogenic missense variants are concentrated at residues 428–498 (numbering of protein NP_991331.1, encoded by the MANE transcript NM_205768.3, as in ClinVar; for example Phe428Ser/Cys, Leu434Pro, Arg436Cys/His, His437Tyr, Cys449Arg, Cys452Arg/Tyr, His460Tyr, Asn461Ser, Arg464Cys/His/Pro/Leu, His469Arg, Gln486Arg, Arg492Gly, His493Arg, His498Pro/Arg) and some in the BTB domain (Glu11Val, His15Arg, Leu20Pro, Cys54Arg); residue Arg464 has four different substitutions classified as pathogenic or likely pathogenic [50]. The de novo variants catalogued in SFARI Gene add, among others, p.(Lys454Met) and p.(Ser506Leu) in autism cohorts and a truncating variant in a Kenyan cohort [54]. Two de novo synonymous variants listed in that database have no established significance [54].

Source: VariantValidator (GRCh38, MANE Select) and ClinVar, consulted on 28 October 2026 [8, 50]; clinical origin according to the publications cited and the OMIM allelic variants [2, 6, 13, 19, 33, 42].

Secondary reviews have propagated several discrepancies that the full text of Cohen 2017 resolves. Cohen's patient 4 (adult man, 34 years in the table and 37 in the text) has c.1390C>T p.(Arg464Cys), and not p.(Arg464His) as given in the van der Schoot 2018 table; SFARI Gene and the Wu 2026 table give the same notation as Cohen [2, 13, 42, 54]. Thus, the count of four p.(Arg464His) carriers in van der Schoot 2018 and in Yang 2024 corresponds, in the primary sources, to three p.(Arg464His) carriers (DDD, van der Schoot patient 4, Depienne Z2) and one p.(Arg464Cys) carrier [6, 13, 25, 42]. The variant of the Rauch 2012 patient is c.1483C>G p.(Arg495Gly) on NM_205768.2, originally described as c.1456C>G p.(Arg486Gly) on NM_006352.3; the designation p.(Arg495Cys) used by van der Schoot 2018 and by Wu 2026 does not correspond to c.1483C>G and designates another variant (c.1483C>T), present in ClinVar with conflicting classifications, whereas “R495G” is the variant studied functionally by Hemming 2019 and Heng 2022 [8, 13, 43, 50, 51]. The variant of the Portuguese patient of Lopes 2016 is c.583C>T p.(Arg195Ter) on NM_205768.2, originally described as c.556C>T p.(Arg186Ter) on NM_006352.3; it is the same variant numbered on another transcript, and it is this old notation that SFARI Gene reproduces [13, 54]. The frameshift variant of Cohen's patient 2 is c.943_944del p.(Arg315GlyfsTer4) on NM_205768.2; SFARI lists c.916_917del p.(Arg306GlyfsTer4), a difference of 27 nucleotides and 9 codons identical to that of Cohen's two notes on NM_006352.3, but the correspondence of this third variant was not confirmed in VariantValidator [13, 54]. For c.1046dup (DDD), Cohen 2017 describes p.(Glu350ArgfsTer151) and later tables p.(Glu350ArgfsTer15); the protein notation is therefore given with its source [2, 13]. The OMIM gene entry, edited on 2 June 2026, selects seven allelic variants (608433.0001–0007) and confirms, in the description of Cohen's patient 4 (37-year-old man), the variant c.1390C>T p.(Arg464Cys) and, in that of patient 2, c.943_944del p.(Arg315GlyfsTer4) on NM_205768.2; the Lopes 2016 variant appears with the old numbering, c.556C>T p.(Arg186Ter) on NM_006352 [19]. The same entry, however, reproduces in variant 608433.0007 van der Schoot's statement that p.(Arg464His) had previously been described in three patients (Cohen 2017 and Depienne 2017), which, added to van der Schoot's patient 4, gives four carriers and contradicts the attribution of p.(Arg464Cys) to Cohen's patient in variant 608433.0004 itself; the count of three primary p.(Arg464His) carriers obtained from the original publications prevails [13, 19, 42]. In the Wu 2026 variant, the text refers to an asparagine, but the notation and VariantValidator indicate aspartic acid: p.(Arg492AspfsTer11) [2, 8].

2.4. HNRNPU and the related disorder

The hnRNP U protein binds DNA and RNA and takes part in nuclear chromatin organisation, regulation of telomere length, alternative splicing and mRNA stability, Xist-mediated transcriptional silencing, the mitotic cycle, negative regulation of glucocorticoid transactivation and circadian regulation; it is expressed in at least six tissues, with highest expression in the cerebellum [3]. It was cloned in 1992, from HeLa cells, as the largest of the major hnRNPs: an abundant nuclear phosphoprotein of 806 amino acids (about 89 kDa), without homology to proteins then known, with an acidic N-terminal region, a nuclear localisation signal, a nucleoside triphosphate binding motif and a glycine-rich C-terminus, in which a 26-amino-acid segment with RGG repeats is sufficient for RNA binding [10]. In 1994 it was found to be the same protein as scaffold attachment factor A (SAF-A), which binds DNA attachment regions to the nuclear matrix, the origin of the synonym, and that the gene produces two mRNAs (3.9 and 3.1 kb) by alternative polyadenylation, present in similar amounts in all tissues studied [10]. The cloned protein has the length of isoform b (NP_004492.2, 806 amino acids), whereas the MANE Select transcript encodes isoform a (NP_114032.2, 825 amino acids); the numbering of the same variant therefore differs according to the reference used [8, 10, 11]. In a mouse model of preaxial polydactyly, hnRNP U bound preferentially to a mutated enhancer element located about 1 Mb upstream of Shh and to the 5′ UTR of this gene, which points to a role of the protein in long-range regulation of gene expression through the nuclear matrix [10]. The disease mechanism is loss of function through haploinsufficiency, supported by the predominance of de novo truncating variants and by the haploinsufficiency score of 3 [3, 9, 36]. Sequencing detects about 98% of pathogenic variants and intragenic deletions or duplications about 2% (deletion of the last three exons; deletion of exons 1–11), excluding contiguous gene deletions [3]. Taylor and colleagues described recurrent variants, such as c.1089G>A p.(Trp363Ter), also observed by Leduc, c.706_707del, c.847_857del and c.1681del, numbered on the transcript used by the authors, not stated in the abstract [36, 39]. No genotype–phenotype correlation has been identified [3]. Heterozygous pathogenic variants of HNRNPU have a reproducible DNA methylation episignature, incorporated into the EpiSign test, with predominant hypermethylation [55, 56]. In a child with isolated seizures and a 23 bp duplication in HNRNPU classified as pathogenic by one laboratory and as a VUS by another, the negative episignature and the presence of the variant in the asymptomatic father and grandmother led to reclassification as benign [57].

2.5. Architecture of the 1q43q44 deletion and contribution of each gene

The Depienne 2017 analysis gathered 54 deletions (17 new and 37 published, all <6 Mb) spanning 10 Mb with 83 genes, of which 39 are olfactory receptor genes and seven are predicted to be intolerant to haploinsufficiency (RGS7, AKT3, ZBTB18, HNRNPU, KIF26B, CNST, AHCTF1) [6]. Twelve deletions included the three main genes, six only AKT3, nine AKT3 and ZBTB18 without HNRNPU, two ZBTB18 and HNRNPU without AKT3 and 25 only HNRNPU [6]. The results by feature are given in the following table.

Source: [6].

COX20, included in the epilepsy critical region, tolerates haploinsufficiency and is associated with recessive disease, which reinforces HNRNPU as the epilepsy gene [6]. AKT3 and ZBTB18 are about 200 kb apart and ZBTB18 and HNRNPU about 792 kb apart; the additive effects observed between these pairs (microcephaly and agenesis of the corpus callosum) explain why patients with a deletion are, as a rule, more severely affected than carriers of an isolated ZBTB18 variant [1, 6]. Deletions limited to AKT3 showed normal head circumference in two patients and in a healthy father, and in a review of 13 individuals with pure AKT3 deletion, microcephaly and ID had incomplete penetrance [6, 16]. The presence of additional genes explains part of the features of large deletions: in the three patients of Domínguez-Quezada 2026, the common interval of 171 genes included GREM2 and NLRP3, and RYR2 was deleted only in the patient with a deletion of about 12.1 Mb [17].

2.6. Genotype–phenotype correlations, penetrance and expressivity in ZBTB18

Corpus callosum anomalies have incomplete penetrance: the corpus callosum was normal in 5 of 22 patients with haploinsufficiency due to deletion and in 4 of 11 with point variants in the Depienne series [6], and 10 of 17 patients (58.8%) with imaging had hypoplasia in the Yang 2024 review [25]. Agenesis associated with point variants or small deletions of ZBTB18 is partial, with a small splenium and absence of the genu and rostrum, whereas complete or subtotal agenesis occurs in larger 1q deletions; uniform thinning is the typical form with loss of HNRNPU [6]. Whether there is a correlation with variant type is controversial. Van der Schoot 2018 and Yang 2024 found no statistical difference between missense, nonsense and frameshift variants, and van der Schoot proposed that missense variants cause partial loss of function with a less distinctive phenotype [25, 42]. Zhang 2022, in a review of 28 children that according to Yang included patients with deletions, found motor delay, language delay and epilepsy more associated with truncating variants [25, 58]. Cohen 2017 observed the opposite in their cohort: the two patients with missense variants in ZnF3 (p.(Asn461Ser) and p.(Arg464Cys)) had severe to profound ID, Rauch's patient with p.(Arg495Gly) had mild ID and a patient with a nonsense variant had overall cognitive ability in the low-average range, which led the authors to suggest that position in the recognition helix matters more than variant type [13]. Ehmke 2017 described a child with a nonsense variant in ZBTB18 and a de novo 15q13.3 microdeletion (CHRNA7), with a more severe phenotype attributed to an additive effect [59]. With samples of fewer than 35 patients, none of these correlations makes it possible to predict the phenotype of an individual patient.

2.7. Allelic disorders and phenotypic inverses

Gain-of-function variants of AKT3, mostly somatic, cause hemimegalencephaly, focal cortical dysplasia and megalencephaly; 1q43-q44 duplications including AKT3 cause macrocephaly and ID; deletions with presumed loss of function cause microcephaly and ID [6, 23]. In the OMIM gene entry, the allelic phenotype recorded is megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome type 2 (MPPH2, #615937), with two de novo constitutional variants described by Rivière in 2012, c.1393C>T p.(Arg465Trp) and c.686A>G p.(Asn229Ser) on NM_005465.7, and the somatic variant c.49G>A p.(Glu17Lys), detected in brain tissue and absent from the blood of patients with hemimegalencephaly (allele fraction of 40.4% in one case) [8, 22]. p.(Glu17Lys) reduces the affinity of AKT3 for the ubiquitin ligase SIAH1 and escapes proteasomal degradation, which has been associated with dysmorphic neurons in models of neuronal development [22]. AKT3 is a serine/threonine kinase of 479 amino acids with a pleckstrin homology domain, whose activation by insulin depends on PI3K and on phosphorylation of Thr305; in the human perisylvian cortex at 9 weeks of gestation it is the most expressed AKT isoform [22]. Mouse models show both directions of dosage: the Akt3-null mouse has a selective reduction of about 20% in brain size, due to smaller and fewer cells, with attenuation of the mTOR pathway and a higher seizure threshold, whereas the activating variant p.Asp219Val (Nmf350 mutant) causes brain enlargement, ectopic neurons in the hippocampus and seizures with a low threshold [22]. Until 2017 no constitutional loss-of-function point variant of AKT3 was known [6]. Heng 2022 cites 1q43-q44 dosage changes, including gains, associated with macrocephaly [43]. For ZBTB18, ClinGen found no evidence of triplosensitivity [7], so an isolated duplication of the gene is not interpreted as pathogenic by dosage.

3. Clinical Spectrum and Manifestations by System

3.1. ZBTB18-related disorder: data sources and frequencies

The phenotype is described from case series whose patients overlap: the 15 of Cohen 2017 (five described in detail, three published previously, five from the DDD study and two from ClinVar), the 25 of van der Schoot 2018 (including those of Cohen and the four of Depienne 2017) and the 32 of Yang 2024 [13, 25, 42]. Denominators vary because many DDD and ClinVar patients have no detailed clinical data. The table brings together the frequencies from the two reviews with explicit denominators; this is not population natural history, but patients ascertained mainly by exome in the context of ID.

Sources: [13, 25].

In the intermediate review by van der Schoot 2018 (25 patients, four new and 21 published), all had developmental delay, and the remaining features varied: microcephaly in 7 of 17, corpus callosum anomalies in 9 of 15, facial dysmorphism in 10 of 13 and convulsions in 4 of 17; the authors found no genotype–phenotype correlation [18, 42]. Cohen's most severely affected patient, with profound ID, walked with a walker at 6 years, had no words and also had a bifid uvula, widely spaced nipples, cryptorchidism and axial hypotonia with lower limb spasticity; one of Cohen's five patients had normal motor development, and the four with imaging had hypoplasia of the corpus callosum to varying degrees [13, 18].

3.2. Development, language and cognition

All patients described have global developmental delay or ID, with expressive language as the most affected domain [25, 42]. In Depienne's four patients, independent walking emerged between 22 and 27 months (mean 24 months) and ID ranged from mild to severe; a boy with p.(His15Arg) had about 100 words at 11 years and short sentences at 12–14 years, with oral dyspraxia, and a girl with p.(Arg464His) said her first words at 3 years, had a verbal quotient of 58 and a performance quotient of 49 on the Wechsler Intelligence Scale for Children (WISC-IV) at 6 years and read syllables at 12 years [6]. In the literature reviewed by Depienne, patients with ZBTB18 variants walked on average at 28 months (n = 8) and 5 of 8, aged 3 to 37 years, did not form sentences; in contrast, 6 of 7 patients with a ZBTB18 variant or with a ZBTB18 deletion without HNRNPU came to form sentences, which none of the four with deletion of the three genes achieved [6]. The range is wide: Cohen's patient 1, with p.(Asn461Ser), did not speak at 7 years and walked with a walker at 6 years, whereas patient 5, with p.(Arg45Ter), walked at 14 months, made 2–3-word sentences at 3 years and had an overall quotient of 80 on the Differential Ability Scales-II at 5 years, with spatial reasoning well below verbal reasoning [13]. Gestural or sign communication is frequent and effective in some patients: Cohen's adult patient had four words but used more than 200 signs combined into short sentences [13].

3.3. Neuroimaging

Corpus callosum anomalies range from thinning to global or posterior hypoplasia and partial agenesis of the splenium; in Cohen's series, two patients had hypoplasia of the posterior part of the corpus callosum (“apple-core” pattern), one global hypoplasia with hypoplasia of the inferior vermis and one mild global hypoplasia with a small bump suggestive of dysplasia [13]. In Depienne's patients, magnetic resonance imaging (MRI) showed partial agenesis in two and a short, dysgenetic corpus callosum in a third; the patient with p.(Ser200Ter) had a thin corpus callosum with a hypoplastic splenium, enlarged cerebellar interfolial spaces, wide Virchow-Robin spaces and diffuse hypomyelination [6]. Figure 3 of Depienne 2017 illustrates partial agenesis and dysgenesis of the corpus callosum in three patients (figure) [6], and figure 1 of Cohen 2017 shows the sagittal sections of the four patients with MRI [13]. Other published findings include pituitary anomalies, small hippocampi, hypoplasia of the vermis and cerebellar tonsils and periventricular white matter changes [25]. A normal corpus callosum does not exclude the diagnosis: it was normal in the first patient with a nonsense variant, in three patients published before 2016 and in several later patients [13, 25, 33]. Imaging may also change with age: in the child of Wu 2026, thinning of the splenium at 12 months was no longer visible on follow-up MRI, which the authors interpreted as delayed myelination [2].

3.4. Epilepsy

Epilepsy is a minor and inconsistent component. It affected 6 of 17 patients with information (35.3%) in Yang's review, 4 of 17 in van der Schoot's and 3 of 11 patients with point variants in Depienne's [6, 25, 42]. The types described include tonic-clonic seizures and seizures with head turning and cyanosis (onset at 8 months and 9 years in Depienne's patients), focal motor seizures with and without impaired awareness from 3 years, generalised tonic-clonic seizures, infantile spasms with hypsarrhythmia, sleep-related focal motor seizures with atypical absences and febrile convulsions [2, 6, 13, 58, 60]. The electroencephalogram (EEG) may be abnormal without clinical seizures: Cohen's patient 1 had multifocal spike activity at 8 months, with a later normal EEG and no seizures [13]. The response to treatment described is limited to isolated reports (valproate, levetiracetam and perampanel), covered in section 7 [2, 6, 13, 58]. In 1q43q44 deletions epilepsy is much more frequent, because it depends on HNRNPU (see 3.9).

3.5. Behaviour and autism spectrum disorder

Most patients are described as sociable and with good eye contact [13]. A minority have attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD), motor and vocal stereotypies, screaming episodes, tics, obsessive-compulsive behaviours or, as in the patient with p.(Ser200Ter), oppositional behaviour, frustration intolerance, rigidity, tantrums and hyperphagia [6, 13, 25]. SFARI Gene classifies ZBTB18 in category “S” (syndromic), with two de novo loss-of-function variants in probands with autism and an excess of ultra-rare disruptive variants in a case–control burden analysis [54, 60]. The Portuguese patient of Lopes 2016 had a Rett-like phenotype, with regression at 8 months, hand stereotypies, intense eye communication, screaming episodes, bruxism, breathing abnormalities, dystonia, absent speech, severe ID and severe microcephaly, with normal MRI; prior exclusion of MECP2 variants had ruled out Rett syndrome [13, 18, 34].

3.6. Growth, dysmorphism and other systems

Head circumference is heterogeneous: absolute microcephaly in 40% and relative microcephaly in 13% of Cohen's 15 patients, but also head circumferences at the 90th–95th percentile (Rauch's patient, 57 cm at 18 years) and tall stature in some (192 cm at 23 years in Depienne's patient with p.(Leu434Pro)) [6, 13]. Dysmorphism is non-specific and inconsistent: round face, prominent forehead, flat nasal bridge, hypertelorism, epicanthic folds, low-set and malformed ears, slanting palpebral fissures, bulbous nasal tip, long or flat philtrum, thin upper lip and micrognathia [1, 13, 25, 42]. Wu 2026 described a “juvenile facies”, with little emotional expression, a fixed gaze and an immature facial gestalt [2]. Limb anomalies include cutaneous syndactyly between the 2nd and 3rd toes, 5th finger clinodactyly, short broad hands, broad fingertips and broad first digits; hair changes were recorded in 3 of 5 DDD patients [13, 42]. Ocular disease (myopia, strabismus, visual deficit) has been described in four cases and led Yang to recommend ophthalmological attention [25]. Gastrointestinal problems, sialorrhoea, cryptorchidism, bifid uvula and widely spaced nipples have also been described [13]. Heart disease is rare in the point-variant form: none in the patients with data in van der Schoot [42]. Thyroid dysfunction in a 6-year-old boy (Yang 2024) is an isolated observation, with no established causal relationship and a possible laboratory error acknowledged by the authors [25].

3.7. Prenatal phenotype

Birnbaum and colleagues described the first prenatal diagnosis: a fetus with intrauterine growth restriction, shortening of the long bones, a single umbilical artery and a short corpus callosum. At the second-trimester anomaly scan all parameters, including the corpus callosum, were small but within normal limits; only targeted follow-up in the third trimester, with neurosonography and fetal MRI, revealed the malformation, and amniocentesis identified a likely pathogenic de novo variant [61].

3.8. HNRNPU-related disorder (DEE54)

GeneReviews summarises 83 individuals published up to 2022, with the following frequencies [3].

Source: [3].

Epilepsy is the manifestation that distinguishes this entity. Seizures begin in early childhood, often in a febrile context; in Depienne's six constitutional patients they began between 2.5 months and 4 years (five in the first year), with fever as the trigger in five and status epilepticus in two [6, 35]. The mean age at onset was 13.5 months (median 8.5; n = 9) in carriers of an HNRNPU point variant [6]. Isolated cases of West, Lennox-Gastaut and Doose syndromes have been described [3]. Seizures generally respond to the usual antiepileptic drugs, but some patients need more than one drug, the ketogenic diet or recent drugs [3]. ID is moderate to severe, most patients do not speak and adults live mostly in assisted settings with some degree of independence [3]. Hypotonia predominates in the neonatal period and infancy and may progress to hypertonia or spasticity; Leduc described variable neurological regression [3, 36].

The dysmorphism includes prominent, thin and arched eyebrows, long or slanting palpebral fissures, epicanthus, a low-hanging columella, a thin upper lip and widely spaced teeth [3, 35, 38]. About half of the patients with data have proportionate short stature; heart disease (atrial more than ventricular septal defect and patent ductus arteriosus, with isolated cases of tricuspid atresia, tetralogy of Fallot, aortic dilatation and transposition of the great arteries) was recorded in 19 individuals; strabismus in about 30%, cryptorchidism in about 20% of boys and renal anomalies in about 8% (renal agenesis, multicystic kidney, pelvic ectasia) [3]. In Bramswig's series, cardiac anomalies affected 4 of 6 and renal anomalies 3 of 4 [37]. MRI was abnormal in 38 of 62 patients (61%), with ventriculomegaly as the most frequent finding, followed by thinning of the corpus callosum [3]. Sleep disturbances are common and some patients have sleep apnoea; at least two had episodes of hyperventilation and apnoea, in one case with syncope and a pattern similar to that of Rett and Pitt-Hopkins syndromes [3, 62]. The behaviours described include autistic traits, aggression, anxiety and obsessive-compulsive behaviours [39].

The four patients on whom the OMIM DEE54 entry was based illustrate the breadth of the phenotype [20]. The oldest, a 33-year-old man from the Carvill 2013 cohort with the nonsense variant p.(Tyr805Ter) (absent in the mother; father not studied), had developmental delay before seizures and severe ID with regression; seizures began at 2 years as atonic seizures, followed by atypical absences, myoclonus, non-convulsive status epilepticus and tonic and tonic-clonic seizures, and the EEG showed generalised spikes and polyspike-waves, diffuse slowing, slow spike-wave and generalised paroxysmal fast activity. A 3.5-year-old boy from the Hamdan 2014 series, with de novo p.(Gln171Ter), had severe ID, early-onset borderline microcephaly, autistic traits, hypotonia and seizures, did not speak or walk, and MRI showed enlarged ventricles and delayed myelination. A girl from the de Kovel 2016 series, with a de novo frameshift variant, had developmental delay from the first months and febrile seizures at 8 months, did not speak at 2 years and had hypotonia, hyperlaxity, deep-set eyes with epicanthus, a narrow palate, greyish sclerae and a short second finger on both hands, with delayed myelination on MRI. An 11-year-old girl from the Epi4K 2013 study, with a small de novo insertion-deletion at a splice acceptor site, had delay before febrile seizures at 9 months, followed by several types of refractory afebrile seizures; at 11 years she did not speak and had severe ID, autistic traits, aggression and incontinence, and she also carried a de novo variant in SMAD1 previously described by Need 2012 [20]. The phenotype entry gives the variants only in protein notation; the gene entry adds, for two of these patients, the notation on NM_031844.2 [10, 20]. Normalised to the MANE Select transcript NM_031844.3, the three variants selected by OMIM are c.511C>T p.(Gln171Ter) (Hamdan 2014; ClinVar 393465); c.1812dup p.(Val605CysfsTer7) (de Kovel 2016; ClinVar 393466), published as c.1810_1811insT p.(Val604fsTer24), a notation that VariantValidator right-aligns as a duplication of c.1812 and whose predicted protein consequence differs from that published; and c.2471_2472delinsGA p.(Tyr824Ter) (Carvill 2013; ClinVar 393464), designated p.(Tyr805Ter) in the numbering of the 806-amino-acid isoform, in which residue 805 is a tyrosine; all three are classified as pathogenic in ClinVar [8, 10, 11, 63].

3.9. 1q43q44 microdeletion syndrome

In Depienne's cohort of 54 deletions, microcephaly affected 26 of 49 patients with known head circumference, corpus callosum anomalies 20 of 47 with imaging (10 with agenesis, 5 with dysgenesis and 5 with thinning) and epilepsy 36 of 54; epilepsy was present in 90% of deletions including HNRNPU and absent in 14 of 15 without this gene [6]. The mean age at seizure onset was 12.3 months (n = 13) in deletions of HNRNPU without ZBTB18 and 12.9 months (n = 7) in those including both [6]. All patients with a deletion had developmental delay or ID, except one with a borderline quotient and a deletion limited to AKT3 inherited from the father [6, 16]. Independent walking was achieved by 18 of 22 patients over 2 years of age (mean 35 months; median 24), at 18 months (17–19; n = 3) in deletions of AKT3 and ZBTB18 without HNRNPU, at 32.6 months (21–59; n = 13) in deletions of HNRNPU only and at 6.5 years (n = 2) when the three genes were deleted; four patients did not walk [6]. Ballif also described limited or absent expressive language, a characteristic facies and anomalies of the hands and feet [26]. In larger 1q44 deletions, renal anomalies, strabismus and hypotonia are more frequent than with HNRNPU point variants, and the face shows upslanting palpebral fissures, widely spaced eyes, telecanthus, a thin vermilion and an accentuated Cupid's bow [3].

Cardiac involvement deserves specific attention. Clinical Cardiogenetics includes interstitial 1q43 deletion among the chromosomal defects associated with left ventricular non-compaction (LVNC), based on a report by Kanemoto 2006, and cites an infant with isolated LVNC, distal 4q trisomy and distal 1q monosomy [64]. The manifestations listed in that table include microcephaly, upslanting palpebral fissures, epicanthus, a broad nasal root, micrognathia, low-set ears, an arched upper lip, widely spaced teeth, a short webbed neck, congenital heart disease, severe ID, speech deficit, convulsions and agenesis of the corpus callosum [64]. The three patients with deletions of 9.5 and 12.1 Mb described in 2026 all had congenital heart disease, in addition to microcephaly, corpus callosum anomalies, ID, language deficit, hypotonia and micro- or retrognathia; of the two presumably monozygotic twins, only one developed convulsions [17].

3.10. Course with age and adult patients

Longitudinal data are scarce and describe patients treated with rehabilitation and, when necessary, antiepileptic drugs. In Cohen's adult with p.(Arg464Cys), severe ID coexisted with semi-independent living in his own flat, support from carers for daily tasks, an adapted workplace, sign communication and the ability to ride a bicycle and go cross-country skiing [13]. In the patient of Yang 2024, clumsy gait and difficulty forming sentences persisted at 6 years, with mainly gestural communication [25]. In the child of Wu 2026, at 6 years there were sentences with disorganised thinking, a full-scale intelligence quotient of 49, ankle hypertonia and microcephaly [2]. For HNRNPU, life expectancy does not appear to be greatly reduced, with several adults described, and the behavioural and neurological course is still being studied [3].

4. Clinical Recognition and Red Flags

4.1. When to suspect

None of the entities has a phenotype recognisable at sight; clinical suspicion is at the level of the class and the diagnosis is almost always molecular, obtained by microarray or by exome requested for ID or global delay [1, 3]. In the prenatal period, the combination of a short corpus callosum, growth restriction and short long bones, especially if progressive between the second and third trimesters, should lead to targeted fetal neurosonography and MRI and to genetic testing [61]. In the infant, the association of hypotonia, motor delay, microcephaly and a corpus callosum anomaly on MRI suggests the 1q43q44 deletion or ZBTB18; early epilepsy, often triggered by fever, with global delay and non-specific dysmorphism, suggests HNRNPU [3, 6]. In the preschool child, the most consistent sign of ZBTB18 is expressive language far below motor development and comprehension, sometimes with good social contact and gestural communication [13, 25]. A Rett-like presentation (regression, hand stereotypies, breathing abnormalities) has also been described with ZBTB18, and episodes of hyperventilation and apnoea with HNRNPU [34, 62].

4.2. Red flags and emergency signs

The signs requiring rapid assessment are prolonged seizures or clusters (status epilepticus has been described in HNRNPU carriers), fever-triggered seizures, episodes of apnoea, cyanosis or syncope associated with hyperventilation, dysphagia with recurrent respiratory infections and, in large deletions, signs of heart failure or arrhythmia, given the association described with congenital heart disease and non-compaction cardiomyopathy [3, 6, 17, 62, 64]. Starting perampanel, levetiracetam or aripiprazole requires monitoring for aggression and psychiatric changes, and valproate carries an increased risk of hepatotoxicity below 2 years and in patients with ID and organic brain disease (see 7.2) [65, 66].

4.3. Newborn screening

There is no newborn screening. The applicable guidelines are those for the aetiological assessment of ID and global delay: chromosomal microarray as the first-line cytogenetic test (ISCA consensus, International Standard Cytogenomic Array, 2010, and American Academy of Pediatrics, 2014) and exome or genome as a first- or second-line test in paediatric patients with a congenital anomaly with onset before 1 year or with developmental delay or ID with onset before 18 years (ACMG 2021, strong recommendation) [27, 28, 67].

5. Diagnosis: Criteria, Testing and Algorithms

5.1. Criteria

There are no consensus clinical criteria for any of the entities [1, 3]. GeneReviews considers suggestive of HNRNPU a developmental delay or moderate to severe ID with at least one of the following: generalised hypotonia in infancy, feeding difficulties, absent or delayed speech, ASD or autistic traits, non-specific dysmorphism, epilepsy (tonic-clonic, absences), short stature or strabismus; MRI may show ventriculomegaly and thinning of the corpus callosum [3]. The diagnosis of any of the three entities is established by identification of a heterozygous pathogenic or likely pathogenic variant in ZBTB18 or HNRNPU, or of a deletion including the gene; a VUS neither confirms nor excludes the diagnosis [1, 3]. The “pathogenic” and “likely pathogenic” classes of the ACMG and Association for Molecular Pathology (ACMG/AMP) standards are equivalent for clinical decision-making [3, 68].

5.2. Non-molecular examinations

Brain MRI is the most informative non-molecular examination and should assess the corpus callosum (thinning, hypoplasia, partial agenesis, dysgenesis), the cerebellar vermis, the ventricles, the hippocampi and myelination; it neither confirms nor excludes the diagnosis, given incomplete penetrance and the possibility of normalisation with myelination [1, 2, 6]. EEG is indicated in the presence of paroxysmal events and may be abnormal without clinical seizures [13]. Serial head circumference, growth and formal assessment of development and language complete the characterisation; no biochemical marker is useful [1].

5.3. Genetic tests in order and algorithm

The exome has a yield similar to that of panels and includes recently described genes; the genome is an alternative [3, 27]. In the Kennedy Krieger cohort, two of the roughly 300 patients with neurodevelopmental disorders studied by clinical exome over four years had ZBTB18 variants [13]. In a Tunisian series of 107 patients with corpus callosum anomalies, array-CGH found pathogenic or likely pathogenic variants in 10 of 54 tested (18.5%), including 1q43-q44 deletions, and the exome in six unresolved cases identified no diagnostic variants [69].

5.4. Test limitations

A negative result does not exclude the diagnosis. For HNRNPU, sequencing detects about 98% and testing for intragenic deletions and duplications about 2% of known pathogenic variants; these values are proportions of detectable variants in a database, not the clinical sensitivity of the test [3]. Targeted methods (quantitative and long-range polymerase chain reaction, multiplex ligation-dependent probe amplification or MLPA, gene-targeted microarray) may fail to detect the breakpoints of large deletions or the extension to adjacent genes, so CMA remains necessary when a contiguous gene deletion is suspected [9, 23]. CMA does not detect balanced rearrangements or low-level mosaicism [28]. Parental leucocyte DNA may not reveal low-level somatic mosaicism or pure germline mosaicism; one of Depienne's patients with an HNRNPU variant inherited it from a mosaic father [3, 6]. Truncating ZBTB18 variants lie in the last exon and do not trigger NMD; application of the PVS1 criterion should take this into account, although haploinsufficiency of the gene is established and the truncations remove the zinc fingers [7, 42, 68]. CNVs are classified by the quantitative five-class ACMG/ClinGen system, and the inclusion of ZBTB18 or HNRNPU, both with a haploinsufficiency score of 3, weighs in the classification of a deletion [7, 9, 24].

5.5. Report content and variants of uncertain significance

The report should state the gene, the transcript with version (NM_205768.3 for ZBTB18), the Human Genome Variation Society (HGVS) nomenclature at cDNA and protein level, the genomic coordinate with the reference genome, zygosity, origin (de novo with confirmed parentage, inherited or unknown) and the ACMG/AMP class with the criteria used; for CNVs, the size, breakpoints, genes included and the ACMG/ClinGen class [8, 24, 68]. The published ZBTB18 reports applied, for example, PVS1_strong, PS2 and PM2_strong for one frameshift variant [2] and PS2, PVS1_moderate and PM2_supporting for another [58]. A missense VUS should be reassessed with segregation, phenotype, position in the zinc fingers and population data; demonstration of altered DNA binding in vitro is not enough, because half of the population missense variants tested also alter it [52]. For HNRNPU, the methylation episignature helps classify VUS and made it possible to reclassify one variant as benign [55–57]. No published episignature for ZBTB18 was found in the October 2026 search, which does not demonstrate that none is in development.

5.6. Prenatal and preimplantation diagnosis

Prenatal diagnosis is made by amniocentesis or chorionic villus sampling with CMA or targeted sequencing when the familial variant is known, or with fetal exome in the presence of ultrasound anomalies [3, 61]. Because the fetal phenotype may evolve between the second and third trimesters, a normal second-trimester anomaly scan does not exclude the diagnosis [61]. Preimplantation genetic testing is possible once the familial variant has been identified [3].

6. Differential Diagnosis

6.1. Differential within the 1q43q44 class and with the phenotypic inverses

AD: autosomal dominant; NGS: next-generation sequencing.

6.2. Considerations for practice

Within the class, the distinction is made by the molecular test and the gene content of the deletion, and the clinical profile only guides expectations: early and frequent epilepsy points to inclusion of HNRNPU, marked microcephaly to AKT3 and corpus callosum anomaly to ZBTB18 [6, 26]. The combination of microcephaly and corpus callosum anomaly without epilepsy, in a deletion without HNRNPU, is compatible with loss of AKT3 and ZBTB18 [6]. The coexistence of two alterations may explain more severe phenotypes, as in the child with a nonsense variant in ZBTB18 and a 15q13.3 microdeletion [59]. In the assessment of a child with a corpus callosum anomaly, CMA has a relevant yield and includes 1q43-q44 deletions [69]; the remaining corpus callosum anomaly genes form part of the genetic approach to agenesis of the corpus callosum. In patients with LVNC and ID, the hypothesis of a chromosomal anomaly, including the 1q43 deletion, justifies CMA [64]. Rauch-Steindl syndrome, sometimes confused with this class, is an NSD2 disorder and does not belong to the 1q43q44 class.

7. Treatment and Risk Management Strategies

7.1. General principles

There is no disease-modifying treatment or approved drug for any of the entities, and no specific clinical guidelines have been published for ZBTB18, for HNRNPU or for the 1q43q44 deletion [1, 3, 7]. The recommendations of the HNRNPU GeneReviews derive from case series and not from trials, and what is known about ZBTB18 is limited to reports of response to antiepileptic drugs in isolated patients [2, 3, 6, 13, 58]. Management is symptomatic, multidisciplinary and guided by each patient's profile, which depends heavily on the gene content of the alteration. The following table adapts the table of treatment of manifestations of the HNRNPU GeneReviews and applies it to the class, as it is the only published synthesis.

Source: table 4 of the HNRNPU GeneReviews, adapted [3]; echocardiogram and LVNC [64]. CPAP and BiPAP: continuous and bilevel positive airway pressure.

7.2. Epilepsy

7.2.1. Specific evidence

The ZBTB18 data are limited to four reports. Depienne's patient with p.(Ser200Ter), with seizures since 8 months, responded rapidly to valproate, with no dose published [6]. Cohen's patient 5, with focal motor seizures with and without impaired awareness since 3 years, was reasonably controlled with levetiracetam [13]. A Chinese child with generalised tonic-clonic seizures and p.(Phe428LeufsTer72) was controlled with levetiracetam and rehabilitation for almost six months, with psychomotor and language improvement, with no dose in the abstract [58]. In the child of Wu 2026, with sleep-related focal motor seizures and atypical absences, lacosamide (7.5 mg/kg/day) with valproic acid (18 mg/kg/day) failed, zonisamide up to 11 mg/kg/day was ineffective, a first trial of perampanel (0.5 mg/day) was stopped by the parents after two weeks because of increased seizures, and a second titration of perampanel from 0.5 mg/day to 2 mg at night, with withdrawal of zonisamide and lacosamide, achieved sustained control [2]. The authors cite a Chinese retrospective study that proposes, in children under 20 kg or under 4 years, starting at 0.5 mg/day and maintenance of 2–8 mg/day, and hypothesise that the response to an AMPA receptor antagonist reflects altered glutamatergic signalling, consistent with the glutamatergic synaptic dysfunction of the heterozygous mouse [2, 47]. None of these reports makes it possible to compare drugs or to recommend one in particular.

For HNRNPU, seizures generally respond to the usual antiepileptic drugs, sodium valproate is the most used drug and described as the most effective, and the ketogenic diet and newer drugs are reserved for refractory epilepsy [3]. GeneReviews gives no doses or ketogenic diet protocol [3].

7.2.2. Drugs and doses

No antiepileptic drug has an established dose for these entities, so the following doses are those of the official labels for epilepsy in general, applied by extrapolation; use below the authorised age is off-label. The brand names given are those of the labels consulted.

Sources: FDA labels and EMA SmPCs cited in each row. The European SmPCs of levetiracetam and lacosamide were not read for doses, and the doses in those rows are those of the FDA label. GABA: gamma-aminobutyric acid; AMPA: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; SV2A: synaptic vesicle protein 2A; ECG: electrocardiogram.

In the child of Wu 2026, perampanel was started at 2 years and 10 months, below the authorised age in the USA and the EU, with a starting dose of 0.5 mg/day (half the European starting dose for under 20 kg) and maintenance of 2 mg at night (at the lower limit of the European range of 2–4 mg for that weight), as monotherapy, whereas the European SmPC only authorises adjunctive therapy [2, 71]. Lacosamide at 7.5 mg/kg/day at about 3 years is within the FDA indication and outside the European one by age [2, 72, 73]. The aggression warning for perampanel and the behavioural changes with levetiracetam are relevant in these entities, in which some patients have behavioural disorders [6, 65].

7.2.3. Regulatory status

7.3. Ketogenic diet

The HNRNPU GeneReviews mentions the ketogenic diet as an option in refractory epilepsy, without a specific protocol [3]. There are no published data on the diet in ZBTB18 in the sources read; its application follows the general ketogenic diet protocols for refractory childhood epilepsies, with a specialised nutrition team.

7.4. Hypertonia, spasticity and dystonia

Some patients with HNRNPU progress from hypotonia to hypertonia or spasticity, and in ZBTB18 spastic diplegia, dystonia and ankle hypertonia have been described [2, 3, 13]. The basis of treatment is physiotherapy, occupational therapy, orthoses and positioning; in severe cases, GeneReviews refers to specialists for baclofen, tizanidine, botulinum toxin and orthopaedic procedures [3]. None of these drugs has an established dose for these entities, and the following doses come from labels with a different indication (extrapolation).

Source: FDA labels cited in each row.

The botulinum toxin warning weighs more in patients with dysphagia or axial hypotonia, which are frequent in this class [3, 77].

7.5. Behaviour, attention and sleep

The behavioural approach begins with non-pharmacological interventions: applied behaviour analysis (ABA) in patients with ASD, follow-up by a developmental paediatrician and child psychiatry referral for severe aggression; ADHD medication is used when necessary [3]. The sources describe no trials of psychotropic drugs in these entities; the following doses are those of the labels for the general indications (extrapolation).

Source: FDA labels cited in each row.

Sleep disturbances are frequent in HNRNPU, and suspected sleep apnoea justifies polysomnography; treatment, with oxygen, CPAP or BiPAP, is decided with a sleep specialist, reviewing opioids and antiepileptic drugs as contributing factors [3].

7.6. Breathing pattern abnormalities in HNRNPU

Spagnoli and colleagues described a patient with DEE54 and severe episodes of hyperventilation and apnoea with syncope, who improved significantly with the combination of acetazolamide, alprazolam and aripiprazole [62]. The doses used in that report were not available in the abstract read. By extrapolation, the acetazolamide label for epilepsy indicates 8–30 mg/kg/day in divided doses, with an optimal range of 375–1000 mg/day, and notes that paediatric safety and efficacy are not established and that prolonged treatment may slow growth through chronic acidosis [80]. The alprazolam label does not establish paediatric safety or efficacy and has no paediatric dose [81]. Aripiprazole has the doses given in 7.5 [79]. This is a single report, so the combination does not constitute routine treatment.

7.7. Development, communication, feeding and rehabilitation

Early intervention should begin as soon as delay is identified and includes occupational therapy, physiotherapy, speech and feeding therapy and special educational support [3]. Augmentative and alternative communication (AAC), with signs, pictograms or devices, does not hinder speech development and is particularly relevant in ZBTB18, in which expressive language is the most affected domain: Cohen's adult patient communicated with more than 200 signs and patient 1 was learning to use a picture-choice device [3, 13]. Physiotherapy aims at mobility and prevention of contractures, scoliosis and hip dislocation, with wheelchairs, walkers and orthoses when necessary; occupational therapy works on fine motor skills [3]. Oromotor dysfunction should be assessed at each visit; choking, poor weight gain, recurrent respiratory infections or food refusal justify a swallowing study, and measures range from thickening of foods to a nasogastric tube or gastrostomy [3]. GeneReviews describes the US model of support (early intervention up to 3 years, individualised education programme reviewed annually, transition plan in adolescence and school services up to 21 years) [3]; in Portugal and Brazil the equivalent national structures for early intervention and inclusive education apply.

7.8. Available therapeutic evidence

There are no clinical trials in any of the entities: the ClinicalTrials.gov search in October 2026 found no studies for ZBTB18, 1q43q44, 1q44 deletion or RP58, and the European trials register was not searched. The table brings together the existing evidence by level.

Case reports do not allow efficacy to be estimated: they are isolated patients, without a comparison group, with concomitant treatments and with publication bias in favour of positive responses. The response to perampanel in a single child, after a first attempt with worsening seizures, cannot be generalised. The preclinical studies with minocycline and forskolin do not constitute a therapeutic indication and there is no study in humans; the dose in the murine study has no clinical translation [48, 49].

7.9. Developments in the past 12 months

Between November 2025 and October 2026, the developments were as follows, by level of evidence. At the regulatory and curation level, ClinGen classified the validity of ZBTB18 for neurodevelopmental disorder as definitive on 22 May 2026 [7]. At the clinical level, the report of seizure control with perampanel (Wu 2026) was published, along with a report of three patients with 1q43q44 deletion, including twins discordant for epilepsy and a deletion resulting from a maternal translocation t(1;21)(q43;q22.3) [2, 17], a Tunisian series of corpus callosum anomalies with 1q43-q44 deletions detected by array [69], and a qualitative study with mothers of children with HNRNPU [83]. At the preclinical level, a study of the hippocampus in Zbtb18-deficient mice with partial in vitro correction by forskolin [49] and a compound screen in cells derived from patients with HNRNPU (preprint) [82]. A 2026 review discussed gene-targeted therapies for HNRNPU (viral vector supplementation, antisense oligonucleotides, CRISPR (clustered regularly interspaced short palindromic repeats) activation, editing) and their limits: a narrow dosage window, with cytotoxicity from overexpression, chromatin dynamics, cell-type sensitivity and short developmental windows; none has reached clinical trials [84]. The 2022 GeneReviews already mentioned single-dose gene therapy (mRNA or adeno-associated virus serotype 9 vector, AAV9) under exploration for haploinsufficiency epilepsies, including HNRNPU [3]. No marketing authorisation applications, regulatory opinions or registered trials were identified.

7.10. Agents to avoid, pregnancy, perioperative care and transition

GeneReviews recommends avoiding activities and agents that may trigger seizures [3]. Valproate should be used with extreme caution below 2 years and in polytherapy, is contraindicated in POLG mitochondrial disease and, in girls and women of childbearing potential, should only be used if alternatives fail and with effective contraception, because of the risk of major malformations, neural tube defects and neurodevelopmental effects after in utero exposure [66]. Perampanel is not recommended in pregnancy or in women of childbearing potential not using contraception, may reduce the efficacy of progestogen-containing hormonal contraceptives (an additional non-hormonal method is recommended) and, during breastfeeding, the decision weighs the benefit to the infant, because the risk cannot be excluded [71]. Opioids and sedating antiepileptic drugs may worsen sleep apnoea [3]. No patients with documented reproduction are known in HNRNPU [3], and the sources consulted describe no pregnancies, anaesthesia or surgery in patients with these entities; in the absence of data, pre-anaesthetic assessment should consider dysphagia, hypotonia, sleep apnoea, heart disease and antiepileptic medication. Transition to adulthood requires an educational and clinical transition plan, and most adults with HNRNPU live in assisted settings [3].

7.11. Guidelines and divergences

There are no specific guidelines, and the applicable recommendations are the general ones for the aetiological diagnosis of ID (ISCA 2010, AAP 2014, ACMG 2021) and for the interpretation of variants and CNVs (ACMG/AMP 2015, ACMG/ClinGen 2020) [24, 27, 28, 67, 68]. The divergence between these guidelines is one of period: the ISCA consensus and the AAP place CMA first-line, whereas ACMG 2021 places the exome or genome first- or second-line, which, for this class, means that a 1q43q44 deletion may be found by CMA or by CNV analysis of the exome or genome, and a point variant only by sequencing [27, 28, 67]. For epilepsy, the HNRNPU GeneReviews highlights valproate, whereas the ZBTB18 reports describe responses to valproate, levetiracetam and perampanel; the choice should follow seizure type, age, sex and adverse effect profile [2, 3, 6, 13].

8. Reproductive Options and Family Counselling

8.1. Inheritance pattern and recurrence risks

The three entities are transmitted in an autosomal dominant manner, but the great majority of cases result from a de novo alteration: all ZBTB18 variants with tested parents in the review tables were de novo, as were all 1q44 deletions with both parents tested in Thierry's series (10 of 11) and all HNRNPU loss-of-function variants in Durkin's cohort [2, 15, 38, 42]. There are exceptions relevant to counselling. In ZBTB18, Li and colleagues documented in 2023 the first vertical transmission, with the variant p.(His441Gln) in the 2nd zinc finger cosegregating with non-syndromic ID in a Chinese family, absent in unaffected relatives; the Wu 2026 table, which classifies it as de novo, contradicts the primary source, which prevails [2, 14]. Brea-Fernández described affected monozygotic twins with a missense variant in the BTB domain [45]. In HNRNPU, two affected siblings born to non-carrier parents indicate germline mosaicism, and one of Depienne's patients inherited the variant from a mosaic father [3, 6]. In 1q43q44 deletions, deletions limited to AKT3 are frequently inherited from a mildly affected or unaffected parent, and terminal deletions may result from balanced parental rearrangements, such as the maternal translocation t(1;21)(q43;q22.3), whose unbalanced product also includes a 21q22.3 duplication [16, 17].

8.2. Cascade testing and relatives

Parents should be tested with the method that identified the alteration (targeted sequencing for the point variant; CMA for the deletion), adding karyotype or FISH in the presence of a terminal deletion or a rearrangement phenotype [3, 28]. Leucocyte DNA may not detect low-level mosaicism or pure germline mosaicism [3]. If a parent has the alteration, that parent's siblings and other relatives should receive counselling; if the parents are negative, the risk for other relatives is presumed to be low [3]. Testing of asymptomatic minor siblings is of no use when the parents are negative; in the presence of developmental delay in a sibling, testing is diagnostic and not predictive.

8.3. Prenatal diagnosis and preimplantation genetic testing

Once the alteration has been identified in the proband, prenatal diagnosis and preimplantation genetic testing are possible and, given the possibility of germline mosaicism, may be offered even when the parents are negative [3]. Ultrasound is not sufficient to exclude the disease, because the fetal signs (short corpus callosum, growth restriction, short long bones) may only become evident in the third trimester [61]. Opinions on the use of prenatal diagnosis vary, especially when the decision is termination of pregnancy, and counselling should respect the parents' decision [3].

8.4. Content of counselling

Counselling should explain the variability of expression (from mild to profound ID in ZBTB18), the absence of a reliable genotype–phenotype correlation, the incomplete penetrance of corpus callosum anomalies and the dependence of the deletion phenotype on the genes included [6, 13, 42]. The experience of families with HNRNPU describes hypervigilance, a change of identity to carer and care manager, grief and rebuilding of hope, and social isolation, which justifies emotional and relational support beyond symptom control [83]. Referral to patient associations and registries (see 9.4) is part of counselling.

9. Special Considerations: Population Diversity and Epidemiology

9.1. Frequency and what each figure measures

There is no prevalence estimate for ZBTB18-related disorder or for HNRNPU, and the figures available are counts of published cases [1, 3]. For ZBTB18, the Yang 2024 review gathered 32 patients and Wu 2026 counted 33 published pathogenic variants, to which are added de novo variants in autism and neurodevelopmental disorder cohorts listed in SFARI Gene [2, 25, 54]. For HNRNPU, GeneReviews counted about 83 individuals published up to 2022 and acknowledges underdiagnosis in adults and ascertainment bias towards the severe end [3]. For the deletion, Orphanet assigns the 1q44 microdeletion syndrome a prevalence below 1/1,000,000, based on about 100 published cases and with the classification “Class only”, and has no prevalence data for the distal 1q deletion syndrome [4, 5]; Cohen 2017 refers to more than 100 published individuals with 1q43q44 deletions [13]. These values measure published patients and not frequency at birth.

Sequencing cohorts give an order of magnitude of the contribution of these genes. In the DDD study, with 4,293 families, five patients had ZBTB18 variants, and the estimated mean prevalence of developmental disorders caused by de novo variants (of all genes) was 1/213 to 1/448 births, depending on parental age, an overall value that does not apply to ZBTB18 [13, 85]. At the Kennedy Krieger Institute clinic (USA), two of about 300 patients with neurodevelopmental disorders studied by exome over four years had ZBTB18 variants [13]. Targeted sequencing of more than 16,000 cases of neurodevelopmental disorders in an international network (Europe, USA, Australia and China) identified several patients with variants in ZBTB18 and in HNRNPU [60]. An analysis of 188 patients with ID by targeted sequencing identified one patient with a truncating variant [13].

9.2. Geographical distribution and populations

The patients described come mainly from Europe (Germany and Switzerland, the Netherlands, France, the United Kingdom, Norway, Spain and Portugal) and North America, with recent series from China, India and Tunisia and one patient in a Kenyan cohort [6, 13, 25, 32, 34, 42, 45, 54, 69, 85, 86]. In Cohen's series, the patients were of European, Mexican and African American ancestry [13]; Depienne's four patients were of European ancestry [6]. No founder variants have been described, which is consistent with the predominance of de novo variants. In Portugal, the cohort of 19 patients with a Rett-like phenotype studied by Lopes and colleagues included one patient with the variant p.(Arg195Ter) in ZBTB18; array and exome identified pathogenic genomic imbalances in 2 patients (10.5%) and variants in known genes in 6 (32%) [13, 34]. The sources consulted include no data from Brazil. The Indian cohort of Pande 2024, with 54 families with syndromic ID and de novo variants, showed the usefulness of singleton exome in a resource-limited setting [86]. In a Polish cohort of 177 children with 17 recurrent CNVs, epilepsy affected 28.2%, but the abstract does not single out the 1q43q44 deletion [87]. Absence of data from a region reflects the literature consulted and does not demonstrate absence of patients.

9.3. Special populations

Adults are probably underdiagnosed, because many adults with ID have never had CMA or exome; late diagnosis remains useful for family counselling and surveillance [3]. In resource-limited settings, singleton exome has proved effective [86]. In the prenatal period, diagnosis depends on suspicion and targeted follow-up [61].

9.4. Reference networks, associations and registries

The relevant European Reference Networks are ERN ITHACA (rare congenital anomalies and intellectual disability; https://ern-ithaca.eu/) and ERN EpiCARE (rare epilepsies; https://epi-care.eu/). The Unique organisation (Understanding Chromosome and Gene Disorders, United Kingdom, rarechromo.org) publishes information for families, including a leaflet on HNRNPU; the Simons Searchlight registry includes patients with genetic neurodevelopmental disorders; the American Epilepsy Society, the Epilepsy Foundation and the National Institute of Neurological Disorders and Stroke (NINDS) provide information on epilepsy [3]. ClinGen GenomeConnect had seven participants with ZBTB18 variants in October 2026 [7]. No patient association dedicated to ZBTB18 was identified.

10. Illustrative Clinical Case

Wu and colleagues described in 2026 a Chinese girl, the first child of healthy parents, born at term by caesarean section owing to oligohydramnios, weighing 3.3 kg and without fetal distress [2]. She sat at 7 months, crawled at 12 and walked unsupported at 18 months, the age at which she said her first words; at 3 years and 10 months she had only a few words. At 1 year she was diagnosed with global developmental delay and malnutrition, and MRI showed thinning of the splenium of the corpus callosum; on a later MRI the splenium appeared normal, which the authors attributed to delayed myelination and not to a permanent structural defect. Figure 1 of the article shows the two MRIs and the EEG (figure 1); the legend places the second MRI at 2 years and the text at 2.5 years [2].

At 2.5 years, focal motor seizures appeared during sleep, with clonic jerks of the right upper limb, bilateral strabismus and impaired awareness, lasting about 20 seconds, progressing to the right limbs with sialorrhoea; the waking EEG showed left mid-temporal and parietal epileptiform discharges and slow activity [2]. At 2 years and 10 months the nocturnal seizures became recurrent. Lacosamide (7.5 mg/kg/day) combined with valproic acid (18 mg/kg/day) did not control them; a first trial of perampanel at 0.5 mg/day, combined with lacosamide, was stopped by the parents after two weeks because of increased seizures; video-EEG showed multifocal sharp waves, sharp-and-slow-wave complexes and slow activity with left predominance, with a discharge index of 80% in non-rapid eye movement (NREM) sleep and atypical absences; zonisamide, up to 11 mg/kg/day, was ineffective. Perampanel was reintroduced and titrated from 0.5 mg/day to 2 mg at night and, after withdrawal of zonisamide and lacosamide, seizure control was maintained on monotherapy [2].

Trio exome identified the de novo variant NM_205768.3:c.1474del p.(Arg492AspfsTer11) in ZBTB18, absent from gnomAD, ExAC and ClinVar, confirmed by Sanger and classified as pathogenic (PVS1_strong, PS2, PM2_strong) [2]. Figure 2 shows the pedigree, the Sanger sequencing and the location of the variant in the protein (figure 2) [2]. At 6 years, the child climbed stairs with a clumsy gait and formed sentences, but with a marked deficit in logical reasoning and disorganised thinking; she had a facies the authors called juvenile (little emotional expression, fixed gaze, immature gestalt), microcephaly, ankle hypertonia with grade 4 strength and, on the WISC, a verbal quotient of 53, a performance quotient of 56 and a full-scale quotient of 49 [2].

The case illustrates several points of the class: the disproportionate delay of expressive language, the corpus callosum anomaly that may not persist on later imaging, epilepsy as a possible but inconsistent component, and the importance of interpreting the position of the variant (last exon, no NMD expected) [2, 42]. The antiepileptic management described is that of a single patient: perampanel was used below the authorised age in the USA and the EU and as monotherapy, which the European SmPC does not cover, and the control obtained does not demonstrate superiority over other drugs (section 7.2) [65, 71].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

Functional prognosis depends on the gene and, in deletions, on the genes included. In ZBTB18-related disorder, ID ranges from mild to profound; walking usually appears between 14 months and 3 years, later in some patients, some acquire sentences and others remain without functional oral language; there are adults living semi-independently [6, 13, 25]. Epilepsy, when present, was controlled in the four reports with described treatment, sometimes after several drugs; the child with an additional 15q13.3 microdeletion had refractory seizures [2, 6, 13, 58, 59]. In HNRNPU-related disorder, ID is moderate to severe, most do not speak, some patients do not walk after 6 years, and most adults live in assisted settings; life expectancy does not appear to be greatly reduced [3, 6]. In 1q43q44 deletions, patients with the three genes deleted have the worst motor and language prognosis (walking at 6.5 years, no sentences), and those not including HNRNPU have, as a rule, less epilepsy [6]. There are no survival or quality-of-life studies for any of the entities.

11.2. Initial assessment after diagnosis

Sources: table 3 of the HNRNPU GeneReviews, applied to the class [3]; echocardiogram and LVNC [17, 64]; eyes and growth in ZBTB18 [25].

11.3. Surveillance

The sources define no specific surveillance for ZBTB18-related disorder or for the 1q43q44 deletion; the table applies the surveillance of the HNRNPU GeneReviews to the class and adds the items suggested by the authors of the ZBTB18 series and by the cardiac data of the deletions [3, 25, 64]. Thyroid function is not included, because the only published observation did not establish a causal relationship [25].

12. Summary and Final Considerations

The 1q43q44 region brings together three dosage-sensitive neurodevelopmental genes whose effects add up in the contiguous gene deletion: AKT3, the main determinant of microcephaly, ZBTB18, the determinant of corpus callosum anomalies with incomplete penetrance, and HNRNPU, the determinant of epilepsy and of much of the severity of ID. Each also causes its own monogenic disorder, and the dissection of the deletions by Ballif, Nagamani, Thierry and Depienne, completed by the point variants identified by exome, now makes it possible to read a patient's phenotype from the gene content of the alteration. ZBTB18-related disorder (MRD22) is characterised by ID of variable severity with disproportionate delay of expressive language, corpus callosum anomalies in about 60% of patients with imaging, epilepsy and microcephaly in about one third, and non-specific dysmorphism; it results from haploinsufficiency and, in the missense variants of the zinc fingers, from alterations of DNA binding and transcriptional regulation. HNRNPU-related disorder (DEE54) is a developmental and epileptic encephalopathy with seizures in 95% of patients, of early onset and often febrile, moderate to severe ID, and heart disease and renal anomalies in a minority.

The diagnosis is molecular. Microarray detects the deletion and defines the genes included, and the exome or genome detects point variants; a negative result does not exclude the diagnosis, missense VUS require segregation and caution in functional interpretation, and nomenclature should use the MANE transcript NM_205768.3, bearing in mind that several secondary reviews have propagated incorrect designations; in the primary sources, Cohen's adult patient has p.(Arg464Cys) and Rauch's patient p.(Arg495Gly). Treatment is symptomatic and multidisciplinary: rehabilitation, augmentative and alternative communication, feeding management, antiepileptic drugs chosen by seizure type and attention to behaviour, sleep, eyes, hearing, heart and kidneys; the reports of response to valproate, levetiracetam and perampanel are isolated cases, and the doses used in practice come from the labels for epilepsy in general. Research remains at the preclinical stage (minocycline and forskolin in ZBTB18 models; gene-targeted therapies in HNRNPU), with no trials in humans. Most cases are de novo, with a low recurrence risk for siblings, but vertical transmission in ZBTB18, mosaicism in HNRNPU, inherited AKT3 deletions and balanced parental translocations make it essential to test the parents with the appropriate method before quantifying the risk.

13. References

1. GeneGuidance. ZBTB18 — gene and condition curation (Mental retardation, autosomal dominant 22). Version of 2026-09-26.
2. Wu Y, Li B, Liu L, Wang Y, Li D. Identification of de novo ZBTB18 variant in a patient with global developmental delay, seizures, and juvenile facies. Int J Pediatr. 2026;2026:7226289. doi:10.1155/ijpe/7226289. PMID: 42183389. PMC13197796.
3. Balasubramanian M. HNRNPU-Related Neurodevelopmental Disorder. 2022 Mar 10. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK578573/
4. Orphanet. Distal deletion 1q syndrome (ORPHA:36367) [Internet]. Paris: INSERM; updated 2026 Jul 2. Available from: https://www.orpha.net/en/disease/detail/36367
5. Orphanet. 1q44 microdeletion syndrome (ORPHA:238769) [Internet]. Paris: INSERM; updated 2026 Jul 2. Available from: https://www.orpha.net/en/disease/detail/238769
6. Depienne C, Nava C, Keren B, Heide S, Rastetter A, Passemard S, et al. Genetic and phenotypic dissection of 1q43q44 microdeletion syndrome and neurodevelopmental phenotypes associated with mutations in ZBTB18 and HNRNPU. Hum Genet. 2017;136(4):463-479. doi:10.1007/s00439-017-1772-0. PMID: 28283832. PMC5360844.
7. Clinical Genome Resource (ClinGen). ZBTB18 (HGNC:13030) curation results [Internet]. [accessed 2026 Oct 28]. Available from: https://search.clinicalgenome.org/kb/genes/HGNC:13030
8. 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. https://variantvalidator.org (GRCh38, MANE Select transcripts), accessed 2026-10-28.
9. GeneGuidance. HNRNPU — gene and condition curation. Version of 2026-09-26.
10. OMIM® Online Mendelian Inheritance in Man. *602869 HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEIN U; HNRNPU. Johns Hopkins University; last edited 11/11/2020. https://omim.org/entry/602869. Accessed 28 October 2026.
11. National Center for Biotechnology Information. Protein: NP_114032.2 (heterogeneous nuclear ribonucleoprotein U isoform a, 825 aa) and NP_004492.2 (isoform b, 806 aa) [Homo sapiens] [Internet]. Bethesda (MD): NLM; [accessed 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/protein/NP_114032.2
12. National Center for Biotechnology Information. Gene: HNRNPU heterogeneous nuclear ribonucleoprotein U [Homo sapiens] (Gene ID 3192; discontinued record 284702 HNRNPU-AS1/C1orf199 merged into 3192) [Internet]. Bethesda (MD): NLM; [accessed 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/gene/3192
13. Cohen JS, Srivastava S, Farwell Hagman KD, Shinde DN, Huether R, Darcy D, et al. Further evidence that de novo missense and truncating variants in ZBTB18 cause intellectual disability with variable features. Clin Genet 2017;91(5):697-707. doi:10.1111/cge.12861. PMID: 27598823.
14. Li N, Kang H, Zou Y, Liu Z, Deng Y, Wang M, et al. A novel heterozygous ZBTB18 missense mutation in a family with non-syndromic intellectual disability. Neurogenetics 2023;24(4):251-262. doi:10.1007/s10048-023-00727-7. PMID: 37525067.
15. Thierry G, Bénéteau C, Pichon O, Flori E, Isidor B, Popelard F, et al. Molecular characterization of 1q44 microdeletion in 11 patients reveals three candidate genes for intellectual disability and seizures. Am J Med Genet A 2012;158A(7):1633-1640. doi:10.1002/ajmg.a.35423. PMID: 22678713.
16. Gai D, Haan E, Scholar M, Nicholl J, Yu S. Phenotypes of AKT3 deletion: a case report and literature review. Am J Med Genet A 2015;167A(1):174-179. doi:10.1002/ajmg.a.36710. PMID: 25424989.
17. Domínguez-Quezada MG, Rivera H, Figuera LE, Barros-Núñez P, Castrillo JL, Córdova-Fletes C, et al. Clinical and Cytogenomic Characterization of Three Patients With Distal 1q43q44 Deletion: Twin Sisters With a de novo Deletion and a Patient With der(1)t(1;21)(q43;q22.3)mat. J Clin Lab Anal 2026;40(19):e70338. doi:10.1002/jcla.70338. PMID: 42605264. PMC13478828.
18. OMIM® Online Mendelian Inheritance in Man. #612337 INTELLECTUAL DEVELOPMENTAL DISORDER, AUTOSOMAL DOMINANT 22; MRD22. Johns Hopkins University; last edited 12/12/2024. https://omim.org/entry/612337. Accessed 28 October 2026.
19. OMIM® Online Mendelian Inheritance in Man. *608433 ZINC FINGER- AND BTB DOMAIN-CONTAINING PROTEIN 18; ZBTB18. Johns Hopkins University; last edited 02/06/2026. https://omim.org/entry/608433. Accessed 28 October 2026.
20. OMIM® Online Mendelian Inheritance in Man. #617391 DEVELOPMENTAL AND EPILEPTIC ENCEPHALOPATHY 54; DEE54. Johns Hopkins University; last edited 03/12/2020. https://omim.org/entry/617391. Accessed 28 October 2026.
21. GeneGuidance. AKT3 — gene and condition curation (MPPH2). Version of 2026-09-26.
22. OMIM® Online Mendelian Inheritance in Man. *611223 AKT SERINE/THREONINE KINASE 3; AKT3. Johns Hopkins University; last edited 11/06/2021. https://omim.org/entry/611223. Accessed 28 October 2026.
23. Mirzaa G. MPPH Syndrome. 2016 Nov 17 [updated 2022 Jul 28]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–. Available from: https://www.ncbi.nlm.nih.gov/books/NBK396098/
24. Riggs ER, Andersen EF, Cherry AM, Kantarci S, Kearney H, Patel A, et al. Technical standards for the interpretation and reporting of constitutional copy-number variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics (ACMG) and the Clinical Genome Resource (ClinGen). Genet Med 2020;22(2):245-257. doi:10.1038/s41436-019-0686-8. PMID: 31690835. PMC7313390.
25. Yang F, Ding Y, Wang Y, Zhang Q, Li H, Yu T, et al. A de novo variant in ZBTB18 gene caused autosomal dominant non-syndromic intellectual disability 22 syndrome: A case report and literature review. Medicine (Baltimore) 2024;103(2):e35908. doi:10.1097/md.0000000000035908. PMID: 38215144. PMC10783315.
26. Ballif BC, Rosenfeld JA, Traylor R, Theisen A, Bader PI, Ladda RL, et al. High-resolution array CGH defines critical regions and candidate genes for microcephaly, abnormalities of the corpus callosum, and seizure phenotypes in patients with microdeletions of 1q43q44. Hum Genet 2012;131(1):145-156. doi:10.1007/s00439-011-1073-y. PMID: 21800092.
27. Manickam K, McClain MR, Demmer LA, Biswas S, Kearney HM, Malinowski J, et al. Exome and genome sequencing for pediatric patients with congenital anomalies or intellectual disability: an evidence-based clinical guideline of the American College of Medical Genetics and Genomics (ACMG). Genet Med 2021;23(11):2029-2037. doi:10.1038/s41436-021-01242-6. PMID: 34211152.
28. Miller DT, Adam MP, Aradhya S, Biesecker LG, Brothman AR, Carter NP, et al. Consensus statement: chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies. Am J Hum Genet 2010;86(5):749-764. doi:10.1016/j.ajhg.2010.04.006. PMID: 20466091. PMC2869000.
29. Caliebe A, Kroes HY, van der Smagt JJ, Martin-Subero JI, Tönnies H, van 't Slot R, et al. Four patients with speech delay, seizures and variable corpus callosum thickness sharing a 0.440 Mb deletion in region 1q44 containing the HNRPU gene. Eur J Med Genet 2010;53(4):179-185. doi:10.1016/j.ejmg.2010.04.001. PMID: 20382278.
30. Nagamani SC, Erez A, Bay C, Pettigrew A, Lalani SR, Herman K, et al. Delineation of a deletion region critical for corpus callosal abnormalities in chromosome 1q43-q44. Eur J Hum Genet 2012;20(2):176-179. doi:10.1038/ejhg.2011.171. PMID: 21934713. PMC3260920.
31. Perlman SJ, Kulkarni S, Manwaring L, Shinawi M. Haploinsufficiency of ZNF238 is associated with corpus callosum abnormalities in 1q44 deletions. Am J Med Genet A 2013;161A(4):711-716. doi:10.1002/ajmg.a.35779. PMID: 23494996.
32. Rauch A, Wieczorek D, Graf E, Wieland T, Endele S, Schwarzmayr T, et al. Range of genetic mutations associated with severe non-syndromic sporadic intellectual disability: an exome sequencing study. Lancet 2012;380(9854):1674-1682. doi:10.1016/s0140-6736(12)61480-9. PMID: 23020937.
33. de Munnik SA, García-Miñaúr S, Hoischen A, van Bon BW, Boycott KM, Schoots J, et al. A de novo non-sense mutation in ZBTB18 in a patient with features of the 1q43q44 microdeletion syndrome. Eur J Hum Genet 2014;22(6):844-846. doi:10.1038/ejhg.2013.249. PMID: 24193349. PMC4023223.
34. Lopes F, Barbosa M, Ameur A, Soares G, de Sá J, Dias AI, et al. Identification of novel genetic causes of Rett syndrome-like phenotypes. J Med Genet 2016;53(3):190-199. doi:10.1136/jmedgenet-2015-103568. PMID: 26740508.
35. Yates TM, Vasudevan PC, Chandler KE, Donnelly DE, Stark Z, Sadedin S, et al. De novo mutations in HNRNPU result in a neurodevelopmental syndrome. Am J Med Genet A 2017;173(11):3003-3012. doi:10.1002/ajmg.a.38492. PMID: 28944577. PMC6555908.
36. Leduc MS, Chao HT, Qu C, Walkiewicz M, Xiao R, Magoulas P, et al. Clinical and molecular characterization of de novo loss of function variants in HNRNPU. Am J Med Genet A 2017;173(10):2680-2689. doi:10.1002/ajmg.a.38388. PMID: 28815871.
37. Bramswig NC, Lüdecke HJ, Hamdan FF, Altmüller J, Beleggia F, Elcioglu NH, et al. Heterozygous HNRNPU variants cause early onset epilepsy and severe intellectual disability. Hum Genet 2017;136(7):821-834. doi:10.1007/s00439-017-1795-6. PMID: 28393272.
38. Durkin A, Albaba S, Fry AE, Morton JE, Douglas A, Beleza A, et al. Clinical findings of 21 previously unreported probands with HNRNPU-related syndrome and comprehensive literature review. Am J Med Genet A 2020;182(7):1637-1654. doi:10.1002/ajmg.a.61599. PMID: 32319732.
39. Taylor J, Spiller M, Ranguin K, Vitobello A, Philippe C, Bruel AL, et al. Expanding the phenotype of HNRNPU-related neurodevelopmental disorder with emphasis on seizure phenotype and review of literature. Am J Med Genet A 2022;188(5):1497-1514. doi:10.1002/ajmg.a.62677. PMID: 35138025. PMC9305207.
40. Okado H. Nervous system regulated by POZ domain Krüppel-like zinc finger (POK) family transcription repressor RP58. Br J Pharmacol 2021;178(4):813-826. doi:10.1111/bph.15265. PMID: 32959890.
41. OMIM® Online Mendelian Inheritance in Man. #619695 RAUCH-STEINDL SYNDROME; RAUST. Johns Hopkins University; last edited 11/01/2022. https://omim.org/entry/619695. Accessed 28 October 2026.
42. van der Schoot V, de Munnik S, Venselaar H, Elting M, Mancini GMS, Ravenswaaij-Arts CMA, et al. Toward clinical and molecular understanding of pathogenic variants in the ZBTB18 gene. Mol Genet Genomic Med. 2018;6(3):393-400. doi:10.1002/mgg3.387. PMID: 29573576. PMC6014478.
43. Heng JI, Viti L, Pugh K, Marshall OJ, Agostino M. Understanding the impact of ZBTB18 missense variation on transcription factor function in neurodevelopment and disease. J Neurochem. 2022;161(3):219-235. doi:10.1111/jnc.15572. PMID: 35083747. PMC9302683.
44. National Center for Biotechnology Information. Protein: NP_991331.1 (zinc finger and BTB domain-containing protein 18 isoform 1, 531 aa) and NP_006343.2 (isoform 2, 522 aa) [Homo sapiens] [Internet]. Bethesda (MD): NLM; [accessed 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/protein/NP_006343.2
45. Brea-Fernández AJ, Souto-Trinei FA, Iglesias E, Caamaño P, Rodríguez Sánchez B, Gómez Lado C, et al. Expanding the Clinical and Molecular Spectrum of FOXG1- and ZBTB18-Associated Neurodevelopmental Disorders. Cytogenet Genome Res 2023;163(5-6):301-306. doi:10.1159/000535660. PMID: 38056433.
46. Xiang C, Baubet V, Pal S, Holderbaum L, Tatard V, Jiang P, et al. RP58/ZNF238 directly modulates proneurogenic gene levels and is required for neuronal differentiation and brain expansion. Cell Death Differ 2012;19(4):692-702. doi:10.1038/cdd.2011.144. PMID: 22095278. PMC3307985.
47. Hirai S, Miwa H, Shimbo H, Nakajima K, Kondo M, Tanaka T, et al. The mouse model of intellectual disability by ZBTB18/RP58 haploinsufficiency shows cognitive dysfunction with synaptic impairment. Mol Psychiatry 2023;28(6):2370-2381. doi:10.1038/s41380-023-01941-3. PMID: 36721027.
48. Tanaka T, Hirai S, Manabe H, Endo K, Shimbo H, Nishito Y, et al. Minocycline prevents early age-related cognitive decline in a mouse model of intellectual disability caused by ZBTB18/RP58 haploinsufficiency. J Neuroinflammation 2024;21(1):260. doi:10.1186/s12974-024-03217-1. PMID: 39396010. PMC11471036.
49. Zhou Y, Tang Y, Peng C. Zbtb18 Deficiency Impairs the Hippocampal Development and Neurite Growth. Mol Neurobiol 2026;63(1):957. doi:10.1007/s12035-026-06254-4. PMID: 42848297.
50. National Center for Biotechnology Information. ClinVar: ZBTB18[gene] [Internet]. Bethesda (MD): NLM; [accessed 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/?term=ZBTB18%5Bgene%5D
51. Hemming IA, Clément O, Gladwyn-Ng IE, Cullen HD, Ng HL, See HB, et al. Disease-associated missense variants in ZBTB18 disrupt DNA binding and impair the development of neurons within the embryonic cerebral cortex. Hum Mutat 2019;40(10):1841-1855. doi:10.1002/humu.23803. PMID: 31112317.
52. Hemming IA, Blake S, Agostino M, Heng JI. General population ZBTB18 missense variants influence DNA binding and transcriptional regulation. Hum Mutat 2020;41(9):1629-1644. doi:10.1002/humu.24069. PMID: 32598555.
53. Blake S, Hemming I, Heng JI, Agostino M. Structure-Based Approaches to Classify the Functional Impact of ZBTB18 Missense Variants in Health and Disease. ACS Chem Neurosci 2021;12(6):979-989. doi:10.1021/acschemneuro.0c00758. PMID: 33621064.
54. SFARI Gene. ZBTB18 — Human Gene Module [Internet]. Simons Foundation; [accessed 2026 Oct 28]. Available from: https://gene.sfari.org/database/human-gene/ZBTB18
55. Rooney K, van der Laan L, Trajkova S, Haghshenas S, Relator R, Lauffer P, et al. DNA methylation episignature and comparative epigenomic profiling of HNRNPU-related neurodevelopmental disorder. Genet Med 2023;25(8):100871. doi:10.1016/j.gim.2023.100871. PMID: 37120726.
56. Lee S, Ochoa E, Badura-Stronka M, Donnelly D, Lederer D, Lynch SA, et al. Germline pathogenic variants in HNRNPU are associated with alterations in blood methylome. Eur J Hum Genet 2023;31(9):1040-1047. doi:10.1038/s41431-023-01422-9. PMID: 37407733. PMC10474128.
57. Zuelke A, Li R, Taylor C, Gilkerson C, Platt D. EpiSignature Utility for Variant of Uncertain Significance Reclassification in an Apparently Atypical Presentation of HNRNPU-Related Neurodevelopmental Disorder: A Case Report. Mol Syndromol 2025;16(6):550-558. doi:10.1159/000543302. PMID: 41378238. PMC12688365.
58. Zhang J, Li Y, Luo H, Shen Y, Yuan M, Yang Z, et al. [Analysis of clinical features and ZBTB18 gene variant in a child with autosomal dominant mental disorder type 22]. Zhonghua Yi Xue Yi Chuan Xue Za Zhi 2022;39(3):293-296. doi:10.3760/cma.j.cn511374-20210630-00556. PMID: 35315038.
59. Ehmke N, Karge S, Buchmann J, Korinth D, Horn D, Reis O, et al. A de novo nonsense mutation in ZBTB18 plus a de novo 15q13.3 microdeletion in a 6-year-old female. Am J Med Genet A 2017;173(5):1251-1256. doi:10.1002/ajmg.a.38145. PMID: 28345786.
60. Wang T, Hoekzema K, Vecchio D, Wu H, Sulovari A, Coe BP, et al. Large-scale targeted sequencing identifies risk genes for neurodevelopmental disorders. Nat Commun 2020;11(1):4932. doi:10.1038/s41467-020-18723-y. PMID: 33004838. PMC7530681.
61. Birnbaum R, Markovitch O, Biron-Shental T, Kidron D, Ben-Sira L, Litz Philipsborn S, et al. Prenatal diagnosis of a likely pathogenic variant in ZBTB18: Natural evolution of fetal phenotype including the long bones and corpus callosum. Am J Med Genet A 2022;188(3):978-983. doi:10.1002/ajmg.a.62599. PMID: 34907638.
62. Spagnoli C, Rizzi S, Salerno GG, Frattini D, Koskenvuo J, Fusco C. Pharmacological Treatment of Severe Breathing Abnormalities in a Case of HNRNPU Epileptic Encephalopathy. Mol Syndromol 2021;12(2):101-105. doi:10.1159/000512566. PMID: 34012379. PMC8114048.
63. National Center for Biotechnology Information. ClinVar: variations 393464, 393465 and 393466 (HNRNPU, NM_031844.3) [Internet]. Bethesda (MD): NLM; [accessed 2026 Oct 28]. Available from: https://www.ncbi.nlm.nih.gov/clinvar/variation/393464/
64. Hoedemaekers YM, Germans T. Left ventricular noncompaction. In: Baars HF, van der Smagt JJ, Doevendans PAFM, van Tintelen JP, editors. Clinical Cardiogenetics. 3rd ed. Cham: Springer; 2020. p. 115–38. doi:10.1007/978-3-030-45457-9_7
65. Eisai Inc. FYCOMPA (perampanel) tablets and oral suspension — prescribing information [Internet]. NDA202834/NDA208277. SPL version effective 2022-12-28 (set id 71cf3309-e182-473c-8b0b-280cabd0e122). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=71cf3309-e182-473c-8b0b-280cabd0e122
66. AbbVie Inc. DEPAKOTE (divalproex sodium) delayed-release tablets — prescribing information [Internet]. NDA018723. SPL version effective 2026-03-31 (set id 08a65cf4-7749-4ceb-6895-8f4805e2b01f). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=08a65cf4-7749-4ceb-6895-8f4805e2b01f
67. Moeschler JB, Shevell M, Committee on Genetics. Comprehensive evaluation of the child with intellectual disability or global developmental delays. Pediatrics 2014;134(3):e903-18. doi:10.1542/peds.2014-1839. PMID: 25157020. PMC9923626.
68. Richards S, Aziz N, Bale S, Bick D, Das S, Gastier-Foster J, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med 2015;17(5):405-424. doi:10.1038/gim.2015.30. PMID: 25741868. PMC4544753.
69. Khadija B, Abdallah HH, Slimani W, Bennour A, Dimassi S, Soyah N, et al. Clinical and genomic characterization of corpus callosum abnormalities (CCA) in 107 Tunisian patients using a stepwise diagnostic approach. Front Genet 2026;17:1815268. doi:10.3389/fgene.2026.1815268. PMID: 42807679. PMC13617407.
70. UCB, Inc. KEPPRA (levetiracetam) tablets and oral solution — prescribing information [Internet]. NDA021035/NDA021505. SPL version effective 2026-01-28 (set id 3ca9df05-a506-4ec8-a4fe-320f1219ab21). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=3ca9df05-a506-4ec8-a4fe-320f1219ab21
71. European Medicines Agency. Fycompa (perampanel): EPAR — product information [Internet]. Amsterdam: EMA; authorisation 23/07/2012, latest EC decision 21/05/2026. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/fycompa
72. UCB, Inc. VIMPAT (lacosamide) — prescribing information [Internet]. NDA022253/NDA022254/NDA022255. SPL version effective 2026-05-12 (set id 9e79b42c-38a3-4b2c-a196-a5a1948250e2). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=9e79b42c-38a3-4b2c-a196-a5a1948250e2
73. European Medicines Agency. Vimpat (lacosamide): EPAR [Internet]. Authorisation 29/08/2008; EC decision 23/07/2026. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/vimpat
74. European Medicines Agency. Keppra (levetiracetam): EPAR [Internet]. Authorisation 29/09/2000; EC decision 01/04/2026. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/keppra
75. Direct_Rx. Baclofen tablets, USP — prescribing information [Internet]. ANDA209102. SPL version effective 2025-01-28 (set id 020632f4-5e8a-f4a0-e063-6294a90a6301). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=020632f4-5e8a-f4a0-e063-6294a90a6301
76. Covis Pharma US, Inc. ZANAFLEX (tizanidine) capsules and tablets — prescribing information [Internet]. NDA020397/NDA021447. SPL version effective 2025-05-05 (set id 043d9e51-bfa2-4add-9058-5ece332f7e99). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=043d9e51-bfa2-4add-9058-5ece332f7e99
77. Allergan, Inc. BOTOX (onabotulinumtoxinA) for injection — prescribing information [Internet]. BLA103000. SPL version effective 2023-11-18 (set id 33d066a9-34ff-4a1a-b38b-d10983df3300). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=33d066a9-34ff-4a1a-b38b-d10983df3300
78. Sandoz Inc. RITALIN (methylphenidate hydrochloride) tablets — prescribing information [Internet]. NDA010187. SPL version effective 2025-02-05 (set id d6fb2750-cdab-4749-ba0d-7534840a5892). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=d6fb2750-cdab-4749-ba0d-7534840a5892
79. Otsuka America Pharmaceutical, Inc. ABILIFY (aripiprazole) — prescribing information [Internet]. NDA021436. SPL version effective 2025-01-29 (set id c040bd1d-45b7-49f2-93ea-aed7220b30ac). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=c040bd1d-45b7-49f2-93ea-aed7220b30ac
80. Avet Pharmaceuticals Inc. Acetazolamide tablets — prescribing information [Internet]. ANDA205530. SPL version effective 2026-05-12 (set id 0700a910-79aa-4431-aecf-637fe14fcae8). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=0700a910-79aa-4431-aecf-637fe14fcae8
81. Pharmacia & Upjohn Company LLC. XANAX (alprazolam) — prescribing information [Internet]. NDA018276. SPL version effective 2023-01-18 (set id 388e249d-b9b6-44c3-9f8f-880eced0239f). Available from: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=388e249d-b9b6-44c3-9f8f-880eced0239f
82. Ye X, Tikhomirova D, Oksanen M, Gaetani M, Gharibi H, Mastropasqua F, et al. Transcriptomic-guided compound prioritization and proteomics validation for HNRNPU deficiency identify signalling correction. bioRxiv 2026. doi:10.64898/2026.05.04.722615.
83. Overton I, Docherty F, Balasubramanian M. Using Interpretative Phenomenological Analysis to Explore the Experiences of Mothers Who Have a Child With the Rare Genetic Disorder, HNRNPU-Related Neurodevelopmental Disorder: "Some Days, I Wear a Cape and I'm Super Mom and Some Days, I Have to Hide Under My Cape and I Can't Get Out for a Few Days". J Genet Couns 2026;35(5):e70310. doi:10.1002/jgc4.70310. PMID: 42922377.
84. Aiyedun D, Rogers SF, Taha DM, Balasubramanian M. Challenges in Gene-Directed Therapies for Rare Neurodevelopmental Disorders Using HNRNPU as an Exemplar. Clin Genet 2026. doi:10.1111/cge.70267. PMID: 42914159.
85. Deciphering Developmental Disorders Study. Prevalence and architecture of de novo mutations in developmental disorders. Nature 2017;542(7642):433-438. doi:10.1038/nature21062. PMID: 28135719. PMC6016744.
86. Pande S, Majethia P, Nair K, Rao LP, Mascarenhas S, Kaur N, et al. De novo variants underlying monogenic syndromes with intellectual disability in a neurodevelopmental cohort from India. Eur J Hum Genet 2024;32(10):1291-1298. doi:10.1038/s41431-023-01513-7. PMID: 38114583. PMC7616498.
87. Młynek M, Wicher D, Cieślikowska A, Urbańska K, Przywoźna-Zduńczyk K, Zawadzka-Więch U, et al. Epilepsy as a Component of the Dysmorphic-Neurodevelopmental Phenotype in Pediatric Patients with Recurrent Copy Number Variants. Genes (Basel) 2026;17(3):256. doi:10.3390/genes17030256. PMID: 41898790. PMC13026247.