Single-Gene Autosomal Recessive Intellectual Disability

Class of neurodevelopmental disorders in which intellectual disability results from biallelic variants in a single autosomal gene; it includes the non-syndromic forms of the OMIM phenotypic series “Intellectual developmental disorder, autosomal recessive” (MRT/IDDAR, PS249500), with PRSS12 (MRT1) as the historical prototype, and the related recessive forms with syndromic expression.

Literature and regulatory information updated: 28 October 2026.

0. Initial Reference Table

Single-gene autosomal recessive intellectual disability (ARID) is treated as a class. The common element is intellectual disability (ID), usually with global developmental delay (GDD) and disproportionate language delay, caused by biallelic variants (homozygous or compound heterozygous) in one of several hundred autosomal genes. The first table summarises the class; the second identifies the genes covered in detail, with the gene-disease validity assigned by the Clinical Genome Resource (ClinGen), and the related entities used as differentials. The term “deficiência intelectual”, current in Brazil, is synonymous.

Sources for the second table: OMIM series and cytobands according to GeneGuidance, checked against the OMIM entries for MRT1, MRT7, MRT38 and the CRBN gene [3, 6, 21–23]; Orphanet codes [1, 24–28]; ClinGen classifications consulted on 28 October 2026 [4]; RNU2-2 [5, 29].

1. Introduction

1.1. Definition and scope

Intellectual disability, with onset in the developmental period and operationally defined by an intelligence quotient (IQ) below 70, affects 1–2.5% of the population when the IQ threshold below 70 is used and 0.3–0.5% in its severe form [30]. This article deals with an aetiological subset: ID caused by biallelic variants in a single autosomal gene. Orphanet designates the non-syndromic form as an aetiological subtype (ORPHA:88616), within the group of rare non-syndromic IDs (ORPHA:101685), and classifies it under codes F70–F73 of the International Classification of Diseases, 10th revision (ICD-10) [1, 8]. Online Mendelian Inheritance in Man (OMIM) groups these entities in phenotypic series PS249500, numbered in order of description: MRT1, MRT2 and onwards, now labelled intellectual developmental disorder, autosomal recessive [3].

The expression “non-syndromic” designates ID without malformations, consistent dysmorphism or neurological and systemic signs that allow the cause to be recognised clinically; in practice, it means that the gene is identified only by sequencing and not by examination [3]. The distinction between non-syndromic and syndromic is, however, a continuum. Detailed phenotyping of additional patients has shown that several forms described as “pure” have accompanying signs, and even dominant causes involving chromatin remodelling produce, in some patients, poorly specific ID phenotypes [31]. For this reason, the article includes the recessive forms with syndromic expression that result from the same genes or that arise in the same clinical context (TRAPPC9, MAN1B1, ADAT3, ST3GAL3, TRMT1), and uses CNTNAP2, NRXN1, THOC6 and the recessive form of RNU2-2 as differentials and models.

1.2. History and milestones of molecular discovery

The hypothesis of a heterogeneous recessive component in ID without other signs predates molecular genetics. In the 1950s and 1960s, surveys of North American institutions frequently found several siblings affected by ID without an aetiological classification, and the first population genetics estimates attributed undifferentiated severe ID to homozygosity at any one of several dozen loci (69 in a 1960 estimate, 114 in a 1965 review), which anticipated the heterogeneity later confirmed by sequencing [6]. Until the beginning of the 21st century, the molecular genetics of ID concentrated on the X-linked forms, which were easier to map in families with several affected boys. In 2002, Molinari and colleagues, at the Necker Hospital in Paris, described the first gene cause of autosomal recessive non-syndromic ID: a four-base-pair deletion in the PRSS12 gene, which encodes neurotrypsin, a neuronal serine protease, in an Algerian family of eight siblings, children of first cousins, in which four (three girls and one boy) had moderate to severe ID, with IQ below 50, after apparently normal development in the first one to two years [6, 32]. In situ hybridisation in fetal brains showed high expression in structures linked to learning and memory, and immunoelectron microscopy localised the protein to presynaptic terminals, over the membrane lining the synaptic cleft; the authors proposed that neurotrypsin-mediated proteolysis is necessary for synaptic function [32, 33].

In 2004, Higgins and colleagues mapped in a large North American family a locus for mild ID (IQ 50–70) to a 4.2 Mb interval at 3p25-pter and identified a nonsense variant in a then new gene, with an ATP-dependent Lon protease domain, cereblon (CRBN, MRT2); the family was of German ancestry [23, 34]. In 2006, Basel-Vanagaite and colleagues, in Israel, noted that only two genes for autosomal recessive non-syndromic ID had been demonstrated, each in a single family; by homozygosity mapping at 19p13.12 they identified a truncating deletion of CC2D1A in nine consanguineous Israeli Arab families with severe ID [35]. In 2008, two groups published in the same issue of the American Journal of Human Genetics the involvement of TUSC3, a subunit of the oligosaccharyltransferase complex: a homozygous deletion in a large Iranian family and a one-base-pair insertion in two French siblings, children of first cousins; this was the first demonstration that a defect of N-glycosylation could cause ID without other signs [36, 37]. TRAPPC9 followed in 2009, identified in Pakistani, Iranian and Israeli Arab families [38, 39].

The decisive step was methodological. In 2011, Najmabadi and colleagues combined homozygosity mapping, exon enrichment and next-generation sequencing (NGS) in 136 consanguineous families, mostly Iranian, and found variants in 23 genes already implicated in ID or neurological diseases and probably causal variants in 50 new candidate genes, distributed across transcription and translation, cell cycle, energy metabolism and fatty acid synthesis pathways [40]. In their 2014 review, Musante and Ropers emphasised that, in populations with frequent parental consanguinity, autosomal recessive ID is the most common form of ID and that it is not rare even in outbred Western populations [2]. The large consanguineous cohorts from Pakistan, Iran, Saudi Arabia and the Middle East followed, with diagnostic yields between 36.8% and 58% and dozens of candidate genes per study [11–13, 41, 42]. In 2026, the description of the recessive form of RNU2-2 syndrome, in a non-coding small nuclear RNA (snRNA) gene, changed the perception of the relative frequency of the genes in the class: the authors consider it the most prevalent recessive neurodevelopmental disorder known [5, 29].

1.3. Nomenclature, classifications and reclassifications

Four classification systems coexist. The OMIM series PS249500 is cumulative and numbers each gene in order of description; it includes provisional entries and entries that, with new patients, have acquired a syndromic name (for example, MRT42, PGAP1, now described as a neurodevelopmental disorder with dysmorphism, spasticity and brain anomalies) [3]. It also contains locus entries without an identified gene: MRT4 (611107) corresponds to a 6.6 Mb interval at 1p21.1-p13.3, defined by linkage in a consanguineous Turkish family, in which no variants were found in the candidates ALX3 and RBM15 [7]. MRT4 is distinct from the ST3GAL3 locus (MRT12, 1p34.1), and the association of the two in the GeneGuidance curation has no support in OMIM [3, 7]. The numbering also has redundancies: the designation MRT22, assigned by the HUGO nomenclature committee to an 8p22 locus, coincides with the MRT7 locus of TUSC3 [21]. Orphanet separates the non-syndromic form (ORPHA:88616, 54 linked genes) from the syndromic entities, which receive their own codes, and maintains parallel entries for autosomal dominant non-syndromic ID (ORPHA:178469) and X-linked non-syndromic ID (ORPHA:777) [1]. The Clinical Genome Resource (ClinGen), through the ID and autism expert panels (Intellectual Disability and Autism Gene Curation Expert Panel, GCEP), classifies the validity of each gene-disease relationship as definitive, strong, moderate, limited, disputed or refuted, and uses generic MONDO terms (intellectual disability, complex neurodevelopmental disorder, syndromic intellectual disability) that do not always coincide with the OMIM title [4]. The SysID database groups ID genes by clinical manifestation and by biological modules (chromatin disorders, DNA repair and others) [43].

Three reclassifications illustrate the fragility of the “non-syndromic” label. MAN1B1 was identified in 2011 in families with recessive non-syndromic ID, with dysmorphism in only one of five [44]; in 2013, Rymen and colleagues showed in seven patients a congenital disorder of glycosylation type II, with mild dysmorphism, psychomotor delay and truncal obesity in all, and the entity is now MAN1B1-CDG, also called Rafiq syndrome [45, 46]. CC2D1A, described as a cause of isolated ID, later revealed ASD, aggressiveness and seizures in a substantial proportion of patients and, in 2024, ciliopathy features [47, 48]. MED23 is curated by ClinGen as a cause of syndromic ID, whereas Orphanet links it to non-syndromic ID and the OMIM title refers only to ID with or without epilepsy [1, 3, 4]. In the opposite direction, the authors of the Maltese TRAPPC9 case proposed abandoning the term mental retardation 13 and using “TRAPPC9 deficiency” [49].

1.4. Entities within scope

This article covers, in dedicated subsections, the genes of the series with sufficient data: PRSS12 (MRT1), CRBN (MRT2), CC2D1A (MRT3), NSUN2 (MRT5), GRIK2 (MRT6), TUSC3 (MRT7), ST3GAL3 (MRT12), TRAPPC9 (MRT13), TECR (MRT14), MAN1B1 (MRT15), MED23 (MRT18), ADAT3 (MRT36), METTL23 (MRT44), FMN2 (MRT47), ZC3H14 (MRT56) and TRMT1 (MRT68). The remaining genes of the series appear in the table of types (section 2.3). CNTNAP2 and NRXN1 (recessive Pitt-Hopkins-like forms), THOC6 (Beaulieu-Boycott-Innes syndrome) and the recessive form of RNU2-2 are treated as related entities and differentials. MAN1B1-CDG and the CRADD form (MRT34, lissencephaly variant) also enter the differential.

2. Molecular Genetics and Pathophysiology

2.1. Genetic architecture of the class

Locus heterogeneity is the dominant characteristic of the class. In 2016, the SysID database brought together 746 genes associated with ID [43]; in 2017, the number of identified ID genes exceeded 950 [30]. From the consanguineous cohorts, Harripaul and colleagues estimated that there are more than 2500 autosomal ID genes, mostly recessive [13], and Hu and colleagues, in 404 consanguineous families, mostly Iranian, found probably causal variants in 77 already known recessive genes and in 77 new candidate genes, alongside 21 X-linked genes and 9 genes previously associated with other diseases [12]. Each gene explains a small fraction of cases: in the large cohorts, most genes appear in a single family, and recurrence of a gene in several families often results from a regional founder variant [13, 14]. The most recent exception is the recessive form of RNU2-2 syndrome, estimated at about 10% of families with recessive neurodevelopmental disease currently diagnosable by sequencing [29].

The genes of the class encode proteins from very diverse pathways. Co-expression analysis of the candidate genes from the Pakistani families showed enrichment in the fetal frontal cortex and in the temporoparietal cortex and subcortex from infancy to adulthood, and twelve of the new genes interacted directly with known ID proteins in six pathways [11]. The genes described by Harripaul had prominent expression in the prenatal brain and overlapped with the gene sets carrying de novo variants in other neuropsychiatric disorders [13]. The table summarises the pathways represented by the genes covered in this article.

Sources for the table: [3, 10, 13, 29, 46, 50–52].

2.2. Pathophysiology: common principles

The dominant mechanism in the class is loss of function. In the Pakistani and Iranian cohorts, half of the candidate variants were loss-of-function variants and, in 43 families, the only candidate variant was a homozygous loss of function; none of the new loss-of-function genes was tolerant of loss of function in the exomes of 3222 British adults of Pakistani origin [13]. Heterozygotes are, as a rule, asymptomatic, which allows the alleles to persist in populations with consanguineous marriage. There are exceptions in which the same gene causes a dominant disorder by a different mechanism (gain or alteration of function, haploinsufficiency with incomplete penetrance), covered in section 2.8.

How loss of function of such diverse proteins converges on a cognitive phenotype is explained by three observations. First, many genes act at the synapse or in pathways that regulate it, and animal models show deficits in memory and synaptic plasticity (for example, impaired spaced long-term potentiation in the neurotrypsin-deficient mouse, lower spine density in the Fmn2-deficient mouse) [3]. Second, the neuron is particularly sensitive to defects in general cellular processes, such as tRNA modification or glycosylation, which are compensated in other cells; TUSC3 is the clearest example, because redundancy with the paralogue MAGT1 keeps N-glycosylation normal in fibroblasts and the defect manifests only in the brain [37, 46]. Third, detailed phenotyping shows that many of these defects also affect other organs, mildly or variably, which explains the shift of “non-syndromic” entities to syndromic ones.

2.3. Table of types in the MRT/IDDAR series

The table includes the entries of the series with data in the curation consulted, the TRMT1 form and the related entities. The complete OMIM series exceeds 70 entries. Most rows follow GeneGuidance, Orphanet and ClinGen; the rows for MRT1, MRT2, MRT4, MRT7 and MRT38 were checked against the corresponding OMIM entries [1, 3, 4, 6, 7, 21–23].

Sources for the table: [1, 3, 4, 7, 9, 10, 22].

2.4. Genes covered in depth

2.4.1. PRSS12 (MRT1): neurotrypsin

PRSS12 (4q26; OMIM gene 606709; HGNC:9477; MANE transcript, Matched Annotation from NCBI and EMBL-EBI, NM_003619.4) encodes neurotrypsin, also called motopsin, a secreted serine protease with several domains: a proline-rich region, a kringle domain, four scavenger receptor cysteine-rich (SRCR) domains and a trypsin-like catalytic domain [3]. The protein is released into the synaptic cleft, where it cleaves agrin; this cleavage participates in activity-dependent structural remodelling of the synapse, one of the cellular substrates of learning and memory [3, 32]. The founder allele described by Molinari, a four-base-pair deletion normalised as NM_003619.4:c.1355_1358del p.(Asp452AlafsTer50), removes the catalytic domain [32, 53, 54]. It was absent in 200 controls (100 from the Maghreb); in screening of another 17 consanguineous families and of 23 isolated children with non-syndromic ID, only one child, also Algerian and the child of first cousins, carried it homozygously, with the same haplotype, which points to distant consanguinity or a founder effect in Algeria [6].

The supporting evidence is mainly non-human: the neurotrypsin-deficient mouse has impaired spaced long-term potentiation and reduced modulation of dendritic spines, and the Drosophila orthologue (Tequila) is required for long-term memory [3]. In 2026, motopsin-deficient mice showed amyloid neuropathy, tauopathy, loss of cholinergic and dopaminergic neurons, long-term memory and motor deficits and altered sleep, and overexpression of the protein promoted adult neurogenesis in the dentate gyrus [55, 56]. The human evidence is scarce: in a linkage study of seven consanguineous Pakistani families with ID, only one was compatible with linkage to PRSS12, without identification of a variant [57], and a family association study in the Qinba region of China was published only as a letter [58].

On 21 August 2026, the ClinGen ID and autism GCEP classified the PRSS12–non-syndromic ID relationship as limited, because the human evidence rests on a very small number of families over two decades, with little independent replication [3, 4]. In ClinVar, the founder variant (VCV000004101) has only the OMIM submission, without an aggregate classification; other truncating variants are classified as likely pathogenic by single submitters (c.441_442del p.(Trp148fs), c.1586G>A p.(Trp529Ter), c.1022del p.(Pro341fs)), and c.2155C>T p.(Arg719Ter), truncating in the last segment of the protein, is of uncertain significance [54]. In practice, a biallelic PRSS12 result should not be reported as an established cause without phenotypic concordance, familial segregation and multidisciplinary discussion, and an isolated heterozygous finding has no clinical significance [3]. Large 4q26-q28 deletions and duplications encompassing PRSS12 that appear as pathogenic in ClinVar correspond to CNV syndromes and not to this entity [54].

2.4.2. CRBN (MRT2): cereblon

CRBN (3p26.2; OMIM gene 609262; NM_016302.4) encodes cereblon, the substrate recognition subunit of the DDB1-CUL4 E3 ubiquitin ligase; in neurons, it regulates large-conductance calcium-activated potassium (BK) channels and maintains presynaptic glutamate release [3]. In rat brain, cereblon binds to the cytosolic C-terminal segment of the BK channel α subunit, encoded by KCNMA1, and reduces tetramer assembly and surface expression of the functional channel [23]. The founder variant of the North American family, NM_016302.4:c.1255C>T p.(Arg419Ter) (1274C>T in the original numbering), lies in the last exon and escapes nonsense-mediated mRNA decay (NMD), but the truncated protein has increased autoubiquitination and is degraded, which amounts to loss of function; Crbn-deficient mice have learning and memory deficits [3, 34]. The truncation spares the Lon domain but removes the C-terminal segment, which acts as a domain inhibiting autoubiquitination; in patient lymphoblastoid lines, the mutant protein was associated with postnatal persistence of a KCNMA1 isoform with the site 2 insert, normally repressed in adult life, which generates BK channels that are more calcium-sensitive, with faster activation and slower deactivation [23]. In ClinVar, this variant has conflicting classifications (VCV000001821), whereas other truncating variants (c.433C>T p.(Arg145Ter), c.124A>T p.(Lys42Ter), c.641C>G p.(Ser214Ter), c.835+1G>A) are classified as pathogenic or likely pathogenic [54].

The phenotype is mild non-syndromic ID, with IQ of 50–70, without dysmorphism or autistic traits and with slightly delayed motor milestones; in the founder family, IQ was lower in males [3, 34]. Since 2004 only two more families have been described: three Syrian siblings with mild ID and a homozygous splicing variant, in Reuter's cohort, and five affected members of a Saudi family with a homozygous missense variant and a more severe phenotype (severe ID, seizures, self-injury), which ClinGen did not score for lack of functional evidence [3, 42]. In the Saudi family, the variant NM_016302.4:c.1171T>C p.(Cys391Arg), at a conserved residue of the CULT domain, was homozygous in three siblings with severe ID, seizures and self-injurious behaviour from about four years of age, and in two nephews aged three and four years with GDD, attention deficit hyperactivity disorder and seizures; there were no functional studies [23, 53]. ClinGen re-evaluated the relationship on 4 September 2026 and kept it moderate [4]. Cereblon is the target of thalidomide, lenalidomide, pomalidomide and the “molecular glue” degraders; this oncological pharmacology has no therapeutic relationship with the disorder and none of these drugs is indicated or under investigation for it [3].

2.4.3. CC2D1A (MRT3)

CC2D1A (19p13.12; OMIM gene 610055) encodes a protein with C2 domains, for calcium-dependent phospholipid binding, and DM14 domains, unique to this protein family; it acts as a nuclear repressor of HTR1A, as a platform for the PI3K/PDK1/AKT pathway, in centriolar cohesion, in synaptic maturation and in homeostatic control of nuclear factor κB (NF-κB) [3, 35]. The Israeli Arab founder variant is a 3589-base-pair deletion, NM_017721.5:c.1223-294_1642-220del (VCV000001354, pathogenic) [54]. Manzini and colleagues described two founder variants with a cognitive phenotype of complete penetrance but variable expression with the same genotype: mild to severe ID, ASD and seizures [47]. Both gain and loss of function of Cc2d1a increase NF-κB activation, and reduced expression in neurons decreases dendritic complexity, reversed in vitro by NF-κB inhibition [47]. In 2024, four patients from two families with new variants had cardiac heterotaxy, cystic or dysplastic kidneys and abnormal cerebrospinal fluid (CSF) circulation, and the fibroblasts had defective ciliogenesis; this is an emerging phenotypic expansion, not yet established as a rule [48]. ClinGen classifies the relationship as definitive for complex neurodevelopmental disorder [4].

2.4.4. NSUN2 (MRT5)

NSUN2 (5p15.31; OMIM gene 610916) encodes a methyltransferase that forms 5-methylcytosine (m5C) in tRNA, mRNA and long non-coding RNA [3, 50]. Three consanguineous Iranian and Kurdish families had moderate to severe ID with a long face, characteristic eyebrows, long nose and small chin, a consistency that suggests a syndromic form [59]; a Pakistani family had ID with distal myopathy [60]; and a family from the United Arab Emirates had a Dubowitz-like phenotype (mild microcephaly, growth delay, eczema, characteristic facial appearance) with a homozygous splicing variant, absence of protein and loss of m5C in tRNA [61]. The recurrent pattern is delay, short stature, microcephaly and dysmorphism. In a cohort of 31 individuals with a clinical diagnosis of Dubowitz syndrome, biallelic NSUN2 explained one family among 13 molecular diagnoses, so the association of NSUN2 with Dubowitz syndrome in Orphanet should be read as a phenocopy and not as equivalence [3]. ClinGen classifies the relationship with syndromic ID as definitive (2023) [4].

2.4.5. GRIK2 (MRT6)

GRIK2 (6q16.3) encodes the GluK2 subunit of the kainate receptor. Biallelic loss of function causes mild to severe ID without other neurological problems, malformations or dysmorphism, with normal growth and head circumference; ClinGen classifies it as moderate [3, 4]. Heterozygous variants in the M3 transmembrane segment and the M3-S2 linker cause a more severe dominant disorder, through slowed channel deactivation: axial hypotonia, GDD, late walking, poor speech, ataxia and behavioural changes, and some patients have early epileptic encephalopathy, absence of walking and speech and hypomyelination; the recurrent variant p.Thr660Lys is associated with severe epilepsy [3]. The dominant relationship is definitive [4]. GRIK2 is thus an example of a gene with two inheritance patterns and opposite mechanisms (section 2.8).

2.4.6. TUSC3 (MRT7)

TUSC3 (8p22; OMIM gene 601385; NM_006765.4) encodes an accessory subunit of the ER oligosaccharyltransferase (OST) complex, associated with STT3B and redundant with the X-linked paralogue MAGT1; it facilitates glycosylation of acceptor sites close to cysteines [3, 37]. In the six families reviewed up to 2015, all patients had speech delay and moderate to severe ID, with mild and non-specific facial dysmorphism; about one third had microcephaly and another third short stature, and malformations were practically absent [62]. The original Iranian family had seven affected individuals with moderate to severe ID and normal MRI in the two examined; in a later Omani family, four children had IQ between 40 and 50, one had ASD, three had short stature and head circumference below the 3rd centile and three had camptodactyly or distal syndactyly, with mild facial dysmorphism that also occurred in an unaffected relative [21]. The alleles include nonsense variants (NM_006765.4:c.244C>T p.(Arg82Ter), VCV002428857), insertions, partial or whole-gene deletions (including a homozygous 1.16 Mb deletion detected by microarray in a Turkish child) and a homozygous intragenic duplication detected by high-resolution array comparative genomic hybridisation [36, 37, 53, 54, 62, 63]. In Harripaul's Pakistani families, a homozygous deletion of about 170 kb removed nine of the ten exons [13]. The deletion in the original Iranian family measured 120–150 kb, included the first exon, was heterozygous in all parents and was not found in 192 Iranian controls; the nonsense variant NM_006765.4:c.163C>T p.(Gln55Ter), in three Iranian siblings, followed, as did the single-nucleotide deletion described as c.222delA, normalised as NM_006765.4:c.225del p.(Lys75AsnfsTer3), in the Omani family [21, 53].

A laboratory point distinguishes TUSC3: screening for congenital disorders of glycosylation (CDG) is normal, because N-glycosylation in fibroblasts is maintained by redundancy with MAGT1; a normal screen does not exclude the disease [37, 62]. In 2025, Park and colleagues showed that TUSC3 forms with the protein ERMA an ER magnesium (Mg²⁺) entry complex; its loss depletes ER Mg²⁺, activates the PERK-eIF2α pathway and causes synaptic dysfunction, and patient fibroblasts reproduce the deficit; in the mouse, magnesium supplementation restored ER Mg²⁺ and rescued the cognitive deficits [64]. These data are exclusively preclinical. In ClinVar, the classification of many variants of uncertain significance in the gene is inflated by multigene 8p22 CNVs [3]. ClinGen classifies the relationship as definitive [4].

2.4.7. ST3GAL3 (MRT12 and DEE15)

ST3GAL3 (1p34.1) encodes the Golgi α2,3-sialyltransferase, the main producer of the sialyl-Lewis a epitope. OMIM maintains two recessive entries, MRT12 (611090) and developmental and epileptic encephalopathy 15 (DEE15, 615006), which correspond to a spectrum; Orphanet groups them as ST3GAL3-CDG (ORPHA:697734) [3, 28]. At the mild end, two Iranian families had non-syndromic ID with missense variants that retain the enzyme in the ER [65]. At the severe end, a review of 24 published cases identified epilepsy in 8 of the 12 patients with information (67%), with median onset at 5.5 months, epileptic spasms as the most frequent seizure type (67%), infantile spasms syndrome and Lennox-Gastaut syndrome in four and refractoriness in 6 of 8 (75%) [66]. Transferrin isoelectric focusing (IEF) has no consistent pattern in this disorder and a normal result does not exclude it [3]. ClinGen: moderate [4].

2.4.8. TRAPPC9 (MRT13)

TRAPPC9 (8q24.3; OMIM gene 611966; 23 exons; MANE transcript NM_001160372.4) encodes NIBP, a subunit of the TRAPP II transport particle complex, which tethers vesicles in ER-to-Golgi, intra-Golgi and endosome-to-Golgi trafficking; the protein also binds the kinases IKKβ and NIK and activates the NF-κB pathway [14, 38]. The gene belongs to the “TRAPPCopathies”, Golgipathies with microcephaly, structural brain changes and atrophy or a thin corpus callosum [52]. It is the gene of the class with the best clinical documentation and one of those that recur in the consanguineous cohorts, and is therefore described in more detail.

The systematic review by Calvo and colleagues, published in June 2026, harmonised 76 patients (75 published in 31 studies and one new case) [14]. The origin was the Middle East and North Africa (MENA) region in 47.2% (34/72), South Asia in 26.3% (19/72) and European ancestry in 18.1% (13/72), and consanguinity was documented in 81.7% (58/71). Genetically, 84.2% (64/76) were homozygous and 15.8% compound heterozygous, with no de novo variant; among 32 distinct sequence variants, 62.5% were loss-of-function and 37.5% missense; and 13.2% (10/76) had CNVs [14]. The most recurrent variant, NM_001160372.4:c.1129C>T p.(Arg377Ter), homozygous in 16 of the 76 patients (21%), corresponds to “R475X” in the older literature, described in transcript NM_031466.7 as c.1423C>T p.(Arg475Ter); the conversion was confirmed in VariantValidator and the variant is classified as pathogenic (VCV000000758) [14, 53, 54]. The variants are distributed across all exons, without a preferential region [67], and no statistically significant genotype-phenotype correlations were found (CNVs versus sequence variants, loss of function versus missense, p.(Arg377Ter) versus the others) [14]. In ClinVar in May 2026, the gene had 115 pathogenic and 36 likely pathogenic variants [14].

Three patients with biallelic missense variants had a serum N-glycosylation defect compatible with type I CDG, confirmed in fibroblasts and corrected by complementation with the normal protein; the authors proposed the designation TRAPPC9-CDG and the usefulness of the transferrin profile [68]. In Trappc9-deficient mice, obesity and hepatic steatosis were associated with fewer dopamine release structures in the striatum and were attenuated by chronic manipulation of dopaminergic transmission (preclinical data) [69].

2.4.9. TECR (MRT14)

TECR (19p13.12; NM_138501.6) encodes the ER trans-2,3-enoyl-CoA reductase, which catalyses the fourth step of the elongation of very long-chain fatty acids used in sphingolipid synthesis [51]. The only known variant, NM_138501.6:c.545C>T p.(Pro182Leu) (VCV000030810), was identified by exome sequencing in a Hutterite sibship: it was homozygous in the five affected siblings and in none of the eight unaffected; three of the five had intention tremor and slowing of rapid finger movements, without resting tremor or ataxia [51, 54, 70]. In lymphoblastoid cells, the variant reduces enzyme activity and stability and decreases sphingomyelin and C24 ceramide [71]. A second unrelated Hutterite individual, with dolichocephaly and corpus callosum dysgenesis, confirmed the founder character of the allele [72]. ClinGen classifies the relationship as limited and a new variant outside the Hutterite context has no precedent [3, 4].

2.4.10. MAN1B1 (MRT15): MAN1B1-CDG

MAN1B1 (9q34.3; NM_016219) encodes the class I α1,2-mannosidase of ER quality control, which removes mannoses from N-glycans and directs misfolded glycoproteins to ER-associated degradation [3]. In the series of 34 patients from 21 families, all had mild to severe ID or GDD, most had abnormal speech, hypotonia, facial dysmorphism and truncal obesity, and about half had behavioural changes (verbal and physical aggressiveness, autism, inappropriate sexual behaviour, hyperphagia); transferrin IEF shows a type 2 pattern and mass spectrometry identifies specific hybrid N-glycans [45, 46]. The dysmorphism includes prominent eyebrows with lateral sparseness, downslanting palpebral fissures, a bulbous nasal tip, large ears and a thin upper lip, and patient cells show Golgi dilatation and fragmentation [3, 45]. ClinGen: definitive [4].

2.4.11. MED23 (MRT18)

MED23 (6q23.2) encodes a tail subunit of the Mediator complex that links the transcription factors ELK1/TCF4 to RNA polymerase II in the activation of the immediate early genes FOS and JUN [73]. The founder family had non-syndromic ID with the homozygous missense variant p.Arg617Gln, and expression of the immediate early genes was dysregulated [73]. Later patients had profound ID, spasticity, congenital heart disease, brain anomalies and epilepsy; in a 2024 review, ID was present in about 87.5%, delay in 50% and seizures in about 18.75%, and MRI shows delayed myelination, a thin corpus callosum and pontine hypoplasia [3]. ClinGen classified the relationship with syndromic ID as moderate on 7 September 2026 [4]. The transcript of the founder variant is not given in the sources consulted, so the published protein designation is retained.

2.4.12. ADAT3 (MRT36)

ADAT3 (19p13.3; OMIM gene 615302) encodes the non-catalytic subunit of the ADAT2/ADAT3 heterodimer, which deaminates adenosine to inosine at wobble position 34 of tRNA [50, 74]. The phenotype, designated in Orphanet as ID-strabismus syndrome (ORPHA:363528), includes GDD, ID of variable severity and speech delay; strabismus is universal in the Saudi cohorts; microcephaly and growth delay are present in most, and dysmorphism becomes more recognisable with age [26, 74, 75]. Hypotonia, hyperactivity and aggressiveness, recurrent otitis media, growth hormone (GH) deficiency, dental anomalies, distal contractures, spasticity and peripheral neuropathy have also been described; seizures occur in about 20% and respond well to treatment, and in an Egyptian cohort of 24 patients corpus callosum anomalies were present in about 92% [3, 76].

The founder variant NM_138422.4:c.430G>A p.(Val144Met) (VCV000183301, pathogenic or likely pathogenic) lies on a haplotype about 1600 years old; it was homozygous in the eight original Saudi families, in 15 additional Saudi and Emirati patients and in 70% of the Egyptian cohort, in which a second founder variant was identified [54, 74–76]. It is described as the most common cause of recessive ID in Saudi Arabia [3]. The early literature used the designation p.Val128Met (c.382G>A), from an earlier numbering; VariantValidator returned a reference error for this designation in the older versions of the transcript, so the formal correspondence between the two numberings could not be obtained and the variant should be reported only in the MANE form [53]. More than 80 patients have been published [3]. ClinGen: moderate (15 September 2025) [4].

2.4.13. METTL23 (MRT44)

METTL23 (17q25.2) encodes an arginine methyltransferase that forms H3R17me2a and acts as a transcriptional partner of GABPA [77]. About 20 individuals from six families have been described. Austrian and Pakistani families had mild non-syndromic ID [78]; a large consanguineous Arab family had seven affected individuals with moderate to severe ID, mapped with a maximum multipoint LOD (logarithm of the odds of linkage) of 6.01 [77]. Speech is disproportionately affected; seizures, attention deficit hyperactivity disorder (ADHD) and aggressiveness occur in a minority; MRI is often normal; and the face is consistent (short upturned nose with a depressed root, thin lips, flat occiput, high hairline, prognathism) [3]. A 2026 report described two Turkish sisters with a homozygous frameshift variant, severe ID with epilepsy and mitral valve prolapse in both [3]. ClinGen raised the relationship to definitive on 17 February 2026 [4]. Heterozygous variants were associated with normal-tension glaucoma in a Japanese family with dominant inheritance, but in a study of 331 cases and 362 controls the overall variant burden did not differ (1.5% versus 2.5%; P=0.27), so this association has mixed evidence [3].

2.4.14. FMN2 (MRT47)

FMN2 (1q43) encodes formin-2, which nucleates actin with SPIRE1/2 and regulates dendritic spine density. Two consanguineous families identified by homozygosity mapping and sequencing had truncating variants and non-syndromic ID without dysmorphism; the Fmn2-deficient mouse has fewer spines and neurons derived from patient induced pluripotent stem cells have lower synaptic density [79]. In Harripaul's Pakistani families two frameshift variants appeared, and in one of them the sequenced affected individual was a genocopy, with no explanation for the other four affected — an example of intrafamilial heterogeneity [13]. De novo heterozygous 1q43 deletions including the gene have been described in children with ID, but the dominant relationship has not been adopted by OMIM, GenCC or PanelApp, and heterozygotes should not be reported as causal [3].

2.4.15. ZC3H14 (MRT56)

ZC3H14 (14q31.3) encodes a poly(A)-binding protein with CCCH zinc fingers. Three siblings from a consanguineous family with mild to moderate ID had a homozygous nonsense variant, and loss of the Drosophila orthologue Nab2 impairs neural function [80]. A second family, with severe ID, had an intronic variant frequent in gnomAD, which ClinGen did not score. ClinGen classified the relationship as limited on 24 August 2026, for lack of replication [3, 4].

2.4.16. TRMT1 (MRT68)

TRMT1 (19p13.13) encodes the enzyme that dimethylates guanosine 26 (m2,2G) of nuclear and mitochondrial tRNAs [50, 81]. Identified in consanguineous families with recessive non-syndromic ID, it now has the largest cohort of the class outside TRAPPC9: 43 individuals from 31 families, all with GDD or ID, with variable behavioural changes, epilepsy and dysmorphism; earlier reports included congenital microcephaly, isolated hypoplasia of the inferior vermis and cystic leukomalacia [3, 81]. ClinGen classified the relationship as definitive on 4 March 2026 [4].

2.4.17. HERC2 (MRT38)

HERC2 (15q13.1; OMIM gene 605837) encodes a large E3 ubiquitin ligase. The recessive entity was defined in Amish and Mennonite patients by autozygosity mapping, which delimited a shared homozygous segment of 8.2 Mb at 15q13, followed by exome sequencing: all were homozygous for the missense variant NM_004667.6:c.1781C>T p.(Pro594Leu) [22, 53]. The mutant protein is unstable, forms cytosolic aggregates and is greatly reduced in patient fibroblasts; loss of ligase activity would reduce ARC degradation and postsynaptic AMPA receptor density, a mechanism close to that proposed for loss of UBE3A in Angelman syndrome, located in the same chromosomal region [22]. The first series (seven patients) and a second series of 15 Amish patients aged 11 months to 39 years described motor, language, adaptive and social GDD, hypotonia and weak suck in infancy, late and unsteady walking, absent or poor speech, hyperactivity and poor concentration, autistic traits, aggressiveness and self-injury; dysmorphism was subtle (plagiocephaly, prognathism, narrow palate, long hallux, gap between the first and second toes) and most had light blue irides. Four of the 15 patients in the second series had seizures and three of the six imaged had absence of the posterior half of the corpus callosum [22]. The variant was absent in 760 alleles from Amish and Mennonite controls in the first series, but was found in 2 of 158 control chromosomes from the Amish community in the second series, which is compatible with a founder effect [22]. In a patient with an Angelman-like picture and a normal 15q11-q13 methylation study, biallelic HERC2 is a diagnosis to consider, especially with Amish ancestry [22].

2.5. Related recessive syndromic forms

2.5.1. CNTNAP2

CNTNAP2 (7q35-q36.1) encodes CASPR2 [3]. The biallelic disorder (OMIM 610042; ORPHA:163681) is a severe developmental and epileptic encephalopathy: moderate to profound ID, disproportionately affected language, focal epilepsy in the first two years (median onset at 22.5 months in the largest series), often followed by language regression, normal or slightly delayed motor milestones, autism, stereotypies, hyperactivity, aggressiveness, hyporeflexia and relative macrocephaly [3, 82]. In the Old Order Amish, Strauss and colleagues described the founder form with symptomatic focal epilepsy, cortical dysplasia and neuronal migration abnormalities with astrogliosis [83]. In Zweier's series, recessive CNTNAP2 was found in at least 1% of 179 patients with severe ID, a yield in a selected population and not a population frequency [84]. The biallelic relationship is definitive and the dominant (heterozygous) relationship is disputed: monoallelic carriers have significantly less delay, epilepsy, autism and cognitive impairment [4, 82].

2.5.2. NRXN1

Heterozygous exonic deletions of NRXN1 (2p16.3) are a susceptibility factor for neurodevelopmental disorders: they were found in 0.21% of 19,263 clinical cases, with an odds ratio of 8.14 (95% confidence interval, 95% CI, 2.91–22.72) relative to controls, with penetrance of about 32% for deletions of exons 6–24 and 13% for exons 1–5, and without substantial risk for intronic deletions [3]. The biallelic disorder, Pitt-Hopkins-like syndrome 2 (OMIM 614325; ORPHA:600663), has severe early psychomotor delay, severe ID, absent speech, stereotypies, autism, epilepsy, episodic breathing abnormalities (hyperventilation, apnoea) and sleep-wake cycle disturbance, without the facial appearance of Pitt-Hopkins syndrome [9, 84]. In a child with an exonic NRXN1 deletion and a severe Pitt-Hopkins-like phenotype, the other allele should be sequenced [3].

2.5.3. THOC6 and the model of the Hutterite founder allele

Beaulieu-Boycott-Innes syndrome (OMIM 613680; ORPHA:363444), due to biallelic variants in THOC6, a member of the THO complex, which is part of the TREX complex for nuclear mRNA processing and export, presents with moderate to severe ID, non-specific dysmorphism (high forehead, deep-set eyes, short upslanting palpebral fissures, epicanthus, long nose with low columella), microcephaly of −2 to −3 standard deviations (SD) in 13 of 19 patients, dental anomalies (10/19), heart disease with atrial or ventricular septal defect (9/19), short stature (8/19), cryptorchidism (7/19 boys), renal malformation with unilateral agenesis (6/19), prenatal ventriculomegaly and hypergonadotrophic hypogonadism in women [10]. The variant NM_024339.5:c.136G>A p.(Gly46Arg) (VCV000064681) has a frequency of 3% in Dariusleut Hutterite controls and 2% in Lehrerleut, and in these groups it can be tested directly [10, 53, 54]. ClinGen classifies the relationship as definitive [4].

2.6. The recessive form of RNU2-2 syndrome

RNU2-2 is a non-coding gene that gives rise to a U2 snRNA involved in pre-mRNA splicing. In April 2026, Greene and colleagues demonstrated by genetic association (log Bayes factor of the recessive model of 18.2) a recessive form: 17 rare variants with a posterior probability of pathogenicity above 0.8 identified 18 probands and 5 affected siblings with two alleles in trans, plus 13 candidate probands at a threshold of 0.6 and 9 cases in replication collections [29]. The phenotype is ID, GDD and seizures. Pathogenic alleles reduce expression by more than 90% in whole-blood RNA, with compensation by the normal allele in carriers, which indicates loss of function [29]. The authors estimated that the form accounts for about 10% of families with recessive neurodevelopmental disease diagnosable by sequencing and about 60% of the number of families with ReNU syndrome (RNU4-2, dominant), and designated it the most prevalent recessive neurodevelopmental disorder known [29].

In the same issue of Nature Genetics, analysis of 200 snRNA genes in a French cohort of 34,329 people with rare diseases identified RNU2-2 variants in 141 individuals: 35 with recurrent dominant variants and 91 affected individuals from 73 families with biallelic variants; the recessive form was at least twice as frequent as the dominant one and often involved a de novo variant in trans with an inherited allele, reflecting the high mutability of these genes [5]. In a Swedish series of 14 individuals from 9 families, the phenotype was a severe developmental and epileptic encephalopathy, with severe to profound ID, absence of walking and communication, hyperkinesia, infantile spasms and refractory tonic seizures and frequent progression of the Lennox-Gastaut type; aberrant splicing was demonstrated by RNA sequencing (RNA-seq) in fibroblasts and was not detectable in blood [85]. For RNU4-2, a Japanese series described a recessive form, due to compound heterozygous variants outside the critical region of the dominant form, with similar but milder ID and without involvement of other organs [86]. ClinGen classifies the dominant form of RNU2-2 (DEE119) as strong; the recessive form had not yet been curated at the time of the query [4].

2.7. Variant spectrum and nomenclature normalisation

The single nucleotide variants (SNVs) and small insertions and deletions of the class are mostly loss-of-function (nonsense, frameshift, canonical splicing, exon or whole-gene deletions), with a minority of missense variants whose effect has been demonstrated by functional studies. In the consanguineous cohorts, almost all are homozygous; compound heterozygosity is more frequent in families without known consanguinity and in genes with frequent de novo variants (RNU2-2) [5, 14]. Homozygous intragenic CNVs are a relevant fraction in some genes: 13.2% of patients with TRAPPC9 and several families with TUSC3 [13, 14, 62]. The table brings together the variants cited in this article, described according to Human Genome Variation Society (HGVS) nomenclature and normalised on the versioned MANE transcript.

Sources for the table: [3, 14, 49, 53, 54].

The table illustrates three pitfalls. The first is the change of reference transcript: the most frequent TRAPPC9 variant has different numbers in the old and MANE transcripts, and a laboratory comparing old reports with current databases should convert using VariantValidator and not by hand [53]. The second is that a truncating variant is not automatically pathogenic: in PRSS12, a nonsense variant in the last segment of the protein is classified as of uncertain significance [54]. The third is the conflicting classification of old founder variants, described in a single family and never re-evaluated, such as that of CRBN [54].

2.8. Two inheritance patterns in the same gene

In several genes of the class, biallelic and monoallelic variants cause different disorders, by different mechanisms, and the two relationships are curated separately. In GRIK2, biallelic loss of function causes non-syndromic ID (moderate relationship) and heterozygous variants in M3 cause a severe dominant disorder through altered gating (definitive relationship) [3, 4]. In NRXN1, heterozygous deletion is a susceptibility factor with incomplete penetrance and biallelic loss causes Pitt-Hopkins-like syndrome 2 [3, 84]. In CNTNAP2, the recessive relationship is definitive and the dominant one disputed [4]. In METTL23, heterozygosity has been associated with normal-tension glaucoma with mixed evidence [3]. In FMN2, the dominant relationship proposed for 1q43 deletions has not been adopted [3]. In RNU2-2 and RNU4-2, dominant and recessive forms coexist, which the authors describe as a gradient of variant impact [5, 86]. For the laboratory report, this means stating the gene-disease relationship in question and the corresponding mode of inheritance, and not assuming that a heterozygote for a recessive gene is a carrier without consequences, nor the reverse.

2.9. Genotype-phenotype correlations, penetrance and modifiers

Penetrance of the cognitive phenotype is complete in the well-studied genes (CC2D1A, TRAPPC9), but expressivity varies within the same family and with the same genotype: in CC2D1A, the same founder variant produces mild or severe ID, with or without ASD and seizures [47]. In TRAPPC9, no comparison between variant classes showed a significant phenotypic difference [14]. The correlations described are of another kind: the ST3GAL3 spectrum, in which missense variants with ER retention are associated with isolated ID and others with encephalopathy with spasms, and the difference between recessive and dominant forms in the same gene [65, 66]. Genetic modifiers have not been characterised. In consanguineous families, a confounding factor is the coexistence of two recessive disorders in the same patient or of different causes between siblings: Reuter found a family with two causal homozygous variants in different genes, and Harripaul described genocopies and intrafamilial heterogeneity [13, 42].

3. Clinical Spectrum and Manifestations by System

3.1. Common pattern of the class

The common picture is developmental delay recognised in early childhood, with slightly or moderately delayed motor milestones and disproportionate delay of expressive language, which evolves into ID of variable severity. The course is non-progressive in the non-syndromic forms: PRSS12, CRBN, FMN2 and biallelic GRIK2 have no described progressive neurological signs [3]. Loss of acquired skills is not part of the pattern and should prompt reconsideration of recessive metabolic and neurodegenerative disorders, much more frequent in the context of consanguinity (section 4) [87]. The exception is language regression after seizure onset in CNTNAP2 and, in general, in the epileptic encephalopathies of the class [82].

The severity of ID partially distinguishes the genes. The forms described as mild include CRBN (IQ 50–70) and the Austrian and Pakistani METTL23 families; the moderate to severe forms include TUSC3, NSUN2 and TRAPPC9, the latter severe in 66.7% (35/48) of patients with information [14, 34, 59, 62, 78]. Within the same gene and the same family, severity may vary from mild to severe [47]. Speech is the most affected domain in almost all genes: in TRAPPC9, absent or severely delayed speech was present in all 42 patients with information in the 2020 series, and the Maltese case never acquired language despite about fifteen years of speech therapy [49].

3.2. From non-syndromic to syndromic

Detailed phenotyping has converted several “non-syndromic” forms into syndromic ones: MAN1B1 revealed truncal obesity and dysmorphism; CC2D1A, ASD, seizures and ciliopathy features; METTL23, a consistent facial appearance; NSUN2, dysmorphism, microcephaly and short stature; and MED23, spasticity and congenital heart disease [3, 45, 47, 48]. The PRSS12, CRBN, biallelic GRIK2 and FMN2 forms remain essentially isolated; TECR adds intention tremor in three of five siblings and TUSC3 mild dysmorphism with microcephaly and short stature in about one third [3, 51, 62]. Cohort inclusion criteria contribute to the label: Harripaul's series, for example, included only families without evident dysmorphism or comorbidities [13], whereas Reuter recorded other clinical signs in 92.1% of the 152 families [42].

3.3. Manifestations by system

Central nervous system. Microcephaly, usually postnatal, is present in 82.3% of patients with TRAPPC9, in most patients with ADAT3, in 13 of 19 with THOC6 and in about one third with TUSC3 [10, 14, 62, 75]; KPTN and CRADD are associated, by contrast, with macrocephaly or megalencephaly [3]. MRI is abnormal in 95.9% of patients with TRAPPC9, mainly due to white matter changes and thinning of the corpus callosum, with cerebellar hypoplasia and reduced brain volume [14]; the corpus callosum is abnormal in about 92% of the Egyptian ADAT3 cohort [76]; delayed myelination and pontine hypoplasia occur in MED23, and MRI is often normal in METTL23 [3]. In ST3GAL3, three individuals had T2 hyperintensity and restricted diffusion in the brainstem and middle cerebellar peduncle [3].

Epilepsy. The frequency varies between genes: 15.9% in TRAPPC9, about 18.75% in MED23, about 20% in ADAT3 (with a good response), a minority in METTL23 and almost all patients with biallelic CNTNAP2 [3, 14, 82]. In ST3GAL3, 67% of patients with information had epilepsy, with epileptic spasms, infantile spasms syndrome and Lennox-Gastaut syndrome, refractory in 75% [66]. The recessive form of RNU2-2 is an epileptic encephalopathy with infantile spasms and refractory tonic seizures [85].

Behaviour and ASD. ASD was present in 43.8% of patients with TRAPPC9 and behavioural changes (aggressiveness, self-injury, hyperactivity) in 49.2% [14]. CC2D1A combines ASD and aggressiveness [47]; MAN1B1 combines verbal and physical aggressiveness, autism, inappropriate sexual behaviour and hyperphagia in about half [46]; TUSC3 combines, in isolated reports, aggressiveness, impulsivity, hyperactivity, emotional dysregulation and sleep and appetite changes [88]. TRAPPC9 deficiency may present as severe ASD and be diagnosed as such for decades [49].

Growth and metabolism. Obesity is present in 52.8% of patients with TRAPPC9, with a picture reminiscent of Prader-Willi syndrome (obesity, hypotonia, ID) [14, 89]; truncal obesity is constant in MAN1B1-CDG [45]. Short stature and growth delay characterise NSUN2, ADAT3 (with GH deficiency in some patients), TTI2 and THOC6 [3, 10, 59, 75].

Eyes, ears and teeth. Strabismus is universal in the Saudi ADAT3 cohorts; recurrent otitis media and dental anomalies complete the picture; PUS3 combines grey sclerae and strabismus [3, 75]. In THOC6, dental anomalies are present in 10 of 19 patients [10].

Other organs. Congenital heart disease occurs in MED23 and in 9 of 19 patients with THOC6; renal malformation in THOC6 and in the CC2D1A patients with ciliopathy features, who also had heterotaxy; mitral valve prolapse has been described in a family with METTL23; and transaminase and coagulation abnormalities occur in a few patients with MAN1B1-CDG [3, 10, 46, 48]. Distal myopathy has been described in a family with NSUN2 and peripheral neuropathy in ADAT3 [3, 60].

3.4. Frequency of manifestations in published cohorts

The cohorts in the table are series of published cases, subject to ascertainment bias towards the more severe phenotypes and towards consanguineous families from the Middle East and South Asia; they describe patients with heterogeneous care and do not separate natural history from the effect of interventions. There are no prospective natural history studies for any gene of the class.

3.5. Adulthood

Information on adults is scarce. The oldest patient with MAN1B1-CDG described was 35 years old [46]; the first four patients with THOC6 are young adults, the oldest in his early 40s, and the disorder is probably under-recognised in adults [10]; the Maltese man with TRAPPC9 was 27 years old and had increasing agitation and aggression towards others when faced with new stimuli [49]. In a sample of adults with ID recruited from community services in the United States, exome sequencing identified pathogenic or likely pathogenic variants in three of five families, one of them in TUSC3, which shows that adults never investigated still have accessible diagnoses [90].

4. Clinical Recognition and Red Flags

4.1. When to suspect

Suspicion of recessive ID rests on the family context more than on the phenotype. The elements that support it are parental consanguinity (parents who are second cousins or closer) or origin in a group with traditional consanguineous marriage, even without known kinship; the same phenotype in siblings of different sexes or in two sisters, which makes X-linked inheritance unlikely; and clinically normal parents [87, 91]. A practical classification distinguishes three levels: known recessive disorder (metabolic disorders, including milder late-onset forms, and recognisable recessive syndromes, each ethnic group with its own); probable recessive disorder (affected siblings of different sexes, regression); and possible recessive disorder (features similar to those of a known recessive disorder, or severe ID without environmental factors) [87].

Absence of consanguinity does not exclude the class. In the TRAPPC9 review, 18.3% of patients with information (13/71) had no documented consanguinity; the Maltese case and two Italian sisters had parents with no known kinship; and, in the recessive form of RNU2-2, one of the alleles is often de novo [5, 14, 49, 89]. Conversely, about 5% of children with a developmental disorder born to consanguineous parents have a de novo dominant variant, and consanguinity should not restrict analysis to recessive genes [30].

4.2. Red flags and emergency signs

Sources for the table: [3, 14, 46, 66, 85, 87].

4.3. Newborn screening

There is no newborn screening for the genes of the class. In children from consanguineous communities, especially migrants, it should be confirmed that newborn screening was performed and was normal, in particular for phenylketonuria and medium-chain acyl-CoA dehydrogenase deficiency, treatable causes of delay that are frequent in these communities; children born in countries without screening should have plasma amino acids measured [87]. Expanded newborn screening by sequencing is under study (for example, the Early Check project) [19].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Criteria

There are no formal clinical criteria for the class or for any non-syndromic gene. Diagnosis is molecular: ID or GDD with biallelic pathogenic or likely pathogenic variants in a gene with an established recessive gene-disease relationship, with phase (trans) confirmed by parental testing or long-read sequencing, and with compatible segregation in affected and unaffected siblings [50, 92]. The strength of the conclusion depends on the validity of the gene-disease relationship: in genes with definitive or strong evidence (for example, TUSC3, TRAPPC9, CC2D1A, METTL23, TRMT1), a biallelic pathogenic variant establishes the diagnosis; in genes with moderate evidence (CRBN, MED23, ADAT3, ST3GAL3), it supports it; in genes with limited evidence (PRSS12, TECR outside the Hutterite context, ZC3H14), it should be reported with reservation and accompanied by segregation and multidisciplinary review [3, 4].

5.2. Guidelines on genetic investigation

Sources for the table: [15–18, 93–99].

The recommendations converge on placing genomic sequencing at the start of the investigation and diverge on the place of the microarray. The 2021 ACMG guideline places ES/GS as first or second tier on the basis of a yield superior to standard tests, clinical utility for management, for the family and for reproduction, and a possibly better cost-effectiveness when requested early [15]. The 2019 meta-analysis estimated an exome yield of 36% across neurodevelopmental disorders as a whole, 31% in isolated forms and 53% when other associated conditions were present, versus 15–20% for the microarray [16]. The 2025 AAP report replaces that of 2014, which placed the microarray as first tier, and guides the choice of test by the phenotype and, in the absence of a hypothesis, by yield and practicability [93, 94]; the full text was not consulted and the exact hierarchy between ES/GS and microarray in the new algorithm is not described here. In the search carried out, no European guideline (European Society of Human Genetics or ERN ITHACA) specifically on first-tier ES/GS in ID was found. None of these guidelines is specific to consanguineous families; the cohorts from these populations recommend the exome as first tier [41, 42].

5.3. Clinical history and pedigree in consanguineous families

The consultation should have an interpreter when necessary, because consanguineous marriage is frequent in migrant communities [87, 100]. The three-generation pedigree should ask directly about consanguinity and define relationships precisely; it should extend to cousins and nephews and nieces, go back to the common ancestors and, when there is multiple consanguinity, record the declared relationships in writing [91]. School difficulties in other children in the family, pregnancies and deaths in childhood or early adulthood are indirect markers of severe recessive disease [87]. The examination should record growth (height, arm span, weight, head circumference and proportions), the neurological examination (tone, spasticity, ataxia, reflexes), the face (coarse in storage disorders and congenital hypothyroidism), organomegaly, the ocular fundus, the skin (pigmentation, photosensitivity) and the skeleton [87].

5.4. First-tier metabolic investigation

The prior probability of metabolic disease is higher in children of consanguineous parents, and metabolic investigation runs in parallel with genomics and not after it [87]. The two-tier algorithm of the TIDE network proposes, in the first tier, accessible and minimally invasive urine and blood tests that identify about 60% of treatable inborn errors of metabolism (IEM) with prominent ID; in the second tier, investigation is guided by clinical signs [98]. The 2021 review counted 116 treatable IDs, caused by 139 genes, and the 2024 Metabolic Treatabolome identified 275 treatable IEM, 18% of the 1564 IEM in the international classification [18, 99]. The Treatable ID app supports the choice of tests [101].

The tests to consider in consanguineous families include basic biochemistry with fasting lactate and pyruvate, plasma amino acids, urinary amino acids and organic acids, urinary mucopolysaccharides and, with a coarse face, oligosaccharides, a search for vacuolated lymphocytes, very long-chain fatty acids, 7-dehydrocholesterol and creatine kinase [87]. Leucocyte enzymes and the study of neuronal ceroid lipofuscinoses are not first tier, except in the presence of regression or other signs [87]. The transferrin glycosylation profile is of specific interest in the class, with results that differ by gene (section 5.7).

5.5. Sequencing, microarray and runs of homozygosity

In consanguineous families, the first-tier test is ES or GS, preferably with CNV and ROH analysis included in the same examination [15, 17, 42]. The SNP microarray detects ROH: when they are confined to one chromosome they suggest uniparental isodisomy; when they are distributed across several chromosomes they indicate parental relatedness; true uniparental disomies are rare (less than 0.5%) [17, 30]. ROH, which reflect homozygosity by descent (HBD), allow prioritisation of variants located in autozygous regions; this is how many families with METTL23, FMN2, NSUN2, TECR and TRAPPC9 were solved [3, 13, 38]. The ACMG draws attention to the factors that confound inference of a specific parental relationship and to the ethical and legal implications: detection of ROH may reveal undeclared parental relatedness, and laboratories should have a defined reporting policy with the ethics committee and legal advice [17, 96].

Autozygosity mapping remains useful in three situations: to prioritise candidates when the exome finds several rare homozygous variants; to exclude a candidate gene when affected siblings do not share the ROH containing it; and to detect intrafamilial heterogeneity, such as the genocopies described by Harripaul [13, 87]. In Reuter's cohort, the yield was higher in families with remotely related parents (44.1%) than overall (36.8%) [42].

5.6. Diagnostic yield by cohort

The yields are not directly comparable between studies. They differ in design (selected multiplex families versus consecutive individuals), inclusion criteria (non-syndromic ID versus any ID), definition of diagnosis (established genes versus candidates), method and period. The common conclusion is that, in consanguineous populations, most diagnoses are recessive, in contrast with the de novo dominant predominance in outbred populations [2, 41, 102]. In the Saudi cohort, clinical assessment alone suggested a diagnosis in 16% of patients and only 70% of those suspicions were confirmed, which justifies genomics as first tier [41].

5.7. Limits of the tests

Intragenic CNVs and structural variants. Homozygous intragenic deletions and duplications (TUSC3, TRAPPC9) escape the exome without CNV calling and low-resolution microarrays [14, 62]. In the Italian TRAPPC9 case, the 60,000-probe array and trio exome were negative; the 400,000-probe array detected a 31–39 kb microdeletion, below the limit of the 60K and 180K platforms, and trio optical genome mapping, with 30× genome, demonstrated a breakpoint of a paternal balanced translocation disrupting the gene, in trans with the maternal deletion [14]. When clinical suspicion is strong and tests are negative, the genome, optical mapping or long-read sequencing can resolve cryptic structural variants.

Non-coding genes. The snRNA genes (RNU2-2, RNU4-2) are non-coding, and the cohorts that described their recessive forms mainly used genome sequencing data [5, 29, 85]. Their assessment should not be assumed in a conventional exome; in a patient with a negative exome and a recessive epileptic encephalopathy phenotype, the laboratory should state whether these genes were analysed.

Expansions and methylation. The exome does not detect repeat expansions or methylation changes; fragile X and imprinting disorders require specific tests or long reads [30, 92].

Phase. Two heterozygous variants in the same gene constitute recessive disease only if they are in trans. Phase is determined by parental testing or by long-read sequencing, which resolves it without parental samples [92]. Parental testing also reveals de novo variants on one allele and deletions on the other, which manifest as false homozygosity [10].

Glycosylation. The transferrin profile detects only some CDG and has false positives with some methods [46]. In the class, it gives a type 2 pattern in MAN1B1-CDG, a type I pattern in three patients with missense TRAPPC9 variants, a normal result in TUSC3 and inconsistent results in ST3GAL3 [3, 37, 46, 68]. Mass spectrometry of intact transferrin gives a characteristic profile in MAN1B1-CDG [3].

Functional validation. For RNU2-2, aberrant splicing was demonstrated in fibroblasts and not in blood, and functional validation may require a skin biopsy [85].

A negative result does not exclude. Even in the best cohorts, about half of families remain without a diagnosis, and many of the rest are left with candidate genes not yet confirmed [12, 42].

5.8. Interpretation and report content

The report should state the gene, the versioned reference transcript (preferably MANE), the variant in Human Genome Variation Society (HGVS) nomenclature at DNA and protein level, the zygosity, the phase when known, the class according to the ACMG/AMP criteria (American College of Medical Genetics and Genomics/Association for Molecular Pathology), the gene-disease relationship and its validity, and the limitations of the method (coverage, CNVs, non-coding genes) [4, 53]. In the Maltese case, the variant was classified as likely pathogenic by the criteria of loss of function in a gene in which this is the mechanism (PVS1), absence from the population (PM2) and specific phenotype (PP4) [49]. A homozygous variant of uncertain significance in an autozygous region, in a candidate gene or one with limited evidence, requires segregation in affected and unaffected siblings and, when possible, functional study [3, 13]. ClinGen recommends particular caution with PRSS12, TECR and ZC3H14 [4]. Findings of parental consanguinity should be reported according to laboratory policy [17].

5.9. Practical algorithm

5.10. Prenatal and preimplantation diagnosis

With both familial variants identified and classified as pathogenic or likely pathogenic, prenatal diagnosis is performed by chorionic villus sampling (CVS) or amniocentesis and preimplantation genetic diagnosis (PGD) is possible [10, 87]. Fetal ultrasound detects only a small proportion of severe recessive disorders (polydactyly, cystic kidneys, heart disease, brain anomalies); most neurodevelopmental disorders have no ultrasound correlate [91]. Details are given in section 8.

6. Differential Diagnosis

The differential diagnosis of recessive ID has two levels. At the first, non-recessive causes that also occur in consanguineous families are distinguished: environmental causes, chromosomal anomalies, de novo dominant variants (about 5% of children with a developmental disorder born to related parents) and X-linked forms [30, 87]. A non-progressive condition with malformations should be a reminder that chromosomal anomalies and non-recessive dysmorphic syndromes are frequent in any population [87]. At the second level, recessive disorders with treatment or with their own clinical consequences are distinguished, which should not be included without investigation in the group of supportive-care recessive IDs.

Sources for the table: [3, 9, 10, 13, 22, 30, 41, 42, 46, 49, 50, 87].

6.1. Treatable causes in the same context

The consanguineous cohorts show that treatable IEM appear among the diagnoses of recessive ID. In Reuter's cohort, five of the 152 families had potentially treatable disorders due to variants in PAH, CBS, MTHFR, CYP27A1 and HIBCH [42]; Anazi identified a potentially treatable form of hypermanganesaemia with neurodegeneration due to SLC39A14 [41]; and Harripaul identified a variant in NAGS, whose deficiency is treatable [13]. For this reason, first-tier metabolic screening and review of the treatable ID genes in the exome are mandatory before concluding that a form requires supportive care only [18]. Among the CDG, MPI-CDG is treatable with oral mannose, in contrast with the CDG of the class, which have no targeted treatment [46].

6.2. Pitt-Hopkins-like forms

Recessive CNTNAP2 and NRXN1 share with Pitt-Hopkins syndrome (TCF4, dominant) severe global delay, absent speech, stereotypies and episodic breathing abnormalities, but not the characteristic facial appearance [9]. In CNTNAP2, focal epilepsy with language regression and relatively preserved motor milestones distinguish it from Pitt-Hopkins syndrome; in NRXN1, sleep-wake cycle disturbances are more frequent than in Pitt-Hopkins syndrome [9, 82]. The remaining differentials in the same table (EHMT1, FOXG1, MECP2, MEF2C, RHOBTB2, SOX11) are not recessive [9].

6.3. Recessive ID and ASD

TRAPPC9 deficiency may be confused with severe ASD: the Maltese man was diagnosed with ASD at 18 months and only at 27 years did the exome reveal the cause [49]. The authors recommend extensive genetic testing in children with severe ID and absent speech of Mediterranean or Middle Eastern origin, especially with consanguinity [49]. In CC2D1A, MAN1B1 and TRAPPC9, ASD is a manifestation of the disorder and not an alternative diagnosis [14, 46, 47].

6.4. Genocopies and dual diagnoses

In consanguineous families with several affected members, different patients may have different causes or one patient may have two recessive disorders. Reuter described a family with two causal homozygous variants in different genes, and Harripaul described families in which one of the affected members was a genocopy [13, 42]. A discordant phenotype between siblings with the same genotype should prompt a search for a second cause.

7. Treatments and Risk Management Strategies

7.1. General principles

No non-syndromic gene of the class has a targeted treatment approved by the Food and Drug Administration (FDA) or the European Medicines Agency (EMA), and a ClinicalTrials.gov search on 28 October 2026 found no trials targeting PRSS12, CC2D1A, CRBN, TUSC3 or TRAPPC9 [19]. Management is supportive and rests on four components: developmental and communication intervention, treatment of comorbidities (epilepsy, behaviour, sleep, feeding, motor function), surveillance of the manifestations specific to each gene and family counselling. The molecular diagnosis nonetheless has clinical utility: it ends the diagnostic investigation, identifies the specific risks of each gene, enables reproductive counselling and carrier testing in the extended family and, in a minority, reveals a treatable disorder [15, 42].

7.2. Developmental intervention, education and rehabilitation

The GeneReviews model of care for GDD and ID, developed for THOC6 syndrome and applicable to the class, organises intervention by age: from 0 to 3 years, early intervention with occupational therapy, physiotherapy, speech and feeding therapy; from 3 to 5 years, preschool education with an individualised education plan; at all ages, follow-up by developmental paediatrics, an education plan reviewed annually and transition planning in adolescence [10]. Physiotherapy prevents contractures, scoliosis and hip dislocation; occupational therapy works on fine motor skills; oromotor dysfunction is assessed at each visit, with swallowing studies, adaptation of food texture and, when necessary, nasogastric tube or gastrostomy [10]. Spasticity and dystonia are treated with baclofen, tizanidine, botulinum toxin, antiparkinsonian drugs or orthopaedic surgery [10].

Augmentative and alternative communication (AAC) should be introduced early and does not hinder speech acquisition [10]. In genes in which absent speech is the rule, such as TRAPPC9, the authors of the Maltese case suggest that, when speech therapy produces no results, visual and augmentative communication and motor training for activities of daily living should be prioritised; this is an authors' opinion based on one case [49]. Behavioural changes benefit from the interventions used in ASD, including applied behaviour analysis, and severe aggressiveness requires child and adolescent psychiatry [10].

7.3. Treatable disorders in the same context

The therapeutic priority of the investigation is not to miss a treatable cause. The consanguineous cohorts found deficiencies of PAH, CBS, MTHFR, CYP27A1, HIBCH, NAGS and SLC39A14 among the diagnoses of recessive ID [13, 41, 42]. The 2021 review describes 116 treatable IDs, with interventions that are mainly nutritional, pharmacological and supplementation with vitamins and trace elements; the evidence is level 1–3 in 19% and level 4–5 in 81% of treatments, and the effects include improvement or stabilisation of clinical deterioration (62%), of neurological manifestations (47%) and of development (37%) [18]. The 2024 Treatabolome extends the list to 275 treatable IEM, with pharmacological (34%), nutritional (34%) and vitamin or trace element (12%) strategies [99].

7.4. Epilepsy

Epilepsy is treated with antiseizure medications chosen according to seizure type and the electroclinical syndrome; there is no specific drug for any gene of the class. The data by gene are observational. In biallelic CNTNAP2, antiseizure medications control about two thirds of patients (22 own patients and 50 published), and resective surgery was repeatedly ineffective, both in the Amish cohort and in a series of four children, so the surgical indication should be weighed with this information [3, 82, 83]. In ST3GAL3, 75% of patients with epilepsy are refractory; spasms follow the approach to infantile spasms syndrome, and two siblings with tonic seizures with onset between 2.5 and 5 years had long seizure-free periods with carbamazepine [66]. In MED23, a child with refractory epilepsy became seizure-free from the first day of the ketogenic diet (single case) [3]. In ADAT3, seizures occur in about 20% and respond well to treatment [3]. In the recessive form of RNU2-2, seizures are predominantly spasms and refractory tonic seizures, with frequent progression to a Lennox-Gastaut-type picture [85].

7.5. Behaviour: aggressiveness, irritability and hyperactivity

There are no trials of psychotropic drugs in patients with recessive ID of defined genetic cause. The doses used are those of the official labels for the symptomatic indication (aggressiveness in conduct disorder with ID, irritability in autistic disorder, ADHD), which constitutes extrapolation to the genetic aetiology and should be declared. Before prescribing, physical and social causes of aggressiveness, such as pain or inappropriate environmental demands, should be excluded, and pharmacological treatment should form part of a psychosocial and educational programme [103]. No guideline with specific doses for this population was found in the sources consulted; the labels are therefore the dosing reference.

The only published experience in a gene of the class is a report of two sisters from Qatar with TUSC3, with aggressiveness, impulsivity, hyperactivity, emotional dysregulation and sleep and appetite changes, treated with behavioural intervention and psychotropic drugs (aripiprazole 3.75 mg, clonidine 0.025 mg and guanfacine 1 mg); guanfacine at 2 mg/day reduced aggressiveness, hyperactivity, impulsivity and insomnia [88]. This is a case report with off-label use. Risperidone in the EU is authorised for aggression in conduct disorder with ID, but the Committee for Medicinal Products for Human Use (CHMP) did not accept the indication in autistic disorder, whereas in the United States the paediatric indication is irritability in autistic disorder; this difference determines whether use in a child with recessive ID and aggressiveness is within or outside the indication in each jurisdiction [103, 104].

7.6. Comparative regulatory status

Sources for the table: [19, 103–106]. The INFARMED summaries of product characteristics (SmPC) and the Brazilian Health Regulatory Agency (ANVISA) package inserts for these drugs were not consulted in this edition.

7.7. Clinical trials and limits of the evidence

The therapeutic evidence for the class comes down to trials of psychotropic drugs in populations defined by the symptom and not by the cause, and to case reports. The risperidone trials did not include patients with a molecular diagnosis and their result applies to the symptom; the effects of the different drugs cannot be compared indirectly between trials with different populations, scales and durations. The case reports describe individual responses, without controls, and do not establish efficacy. No negative, suspended or withdrawn trials targeting the genes of the class were identified, because no targeted trials of any kind were identified; the ClinicalTrials.gov search returned only diagnostic and natural history studies (for example, NCT02340871, on gene identification by sequencing, with unknown status; NCT06147414, on non-invasive prenatal diagnosis of monogenic diseases, recruiting; NCT07400913, on implementation of long-read sequencing in rare diseases, not yet recruiting) [19].

7.8. Developments in the last twelve months

The preclinical results do not translate into recommendations. Magnesium supplementation in TUSC3 deficiency has no human trial, dose or safety data, and the dopaminergic manipulation that attenuated obesity and steatosis in the Trappc9-deficient mouse (D2 agonist with D1 antagonist) has not been studied in humans [64, 69]. No regulatory applications or authorisations for therapies targeted at the genes of the class were found; a negative search does not prove that none exist [19].

7.9. Drugs and interventions to avoid or to weigh carefully

Cereblon is the target of thalidomide and its derivatives; these drugs have no role whatsoever in CRBN deficiency and their use on that basis is unfounded [3]. Epilepsy surgery in biallelic CNTNAP2 was repeatedly ineffective and the indication requires careful consideration [3, 83]. Magnesium supplementation in TUSC3 deficiency should not be prescribed on the basis of the mouse data [64]. The sedation caused by risperidone may impair learning and should be monitored [103].

7.10. Pregnancy, lactation, perioperative care and transition

The sources consulted do not describe pregnancies in women with single-gene recessive ID, and the known individuals with THOC6 have not reproduced [10]. For risperidone, there are no adequate data in pregnancy; reversible extrapyramidal symptoms have been observed in the newborn after exposure in the last trimester, the drug should be used only if strictly necessary and should not be stopped abruptly; it passes in small amounts into breast milk [103]. The sources consulted do not describe specific perioperative recommendations; in MAN1B1-CDG coagulation and transaminase abnormalities have been described in some patients [46]. Transition to adult care should be planned in adolescence and integrated into the education plan; in the United States, school services extend up to 21 years of age [10].

7.11. Recommendations and divergences between sources

There are no treatment guidelines for the class. The applicable management recommendations derive from the GeneReviews model of care for syndromic ID, from the investigation guidelines (which aim mainly to identify treatable causes) and from the labels of the psychotropic drugs [10, 18, 103, 104]. The material divergence is regulatory: the EU authorises risperidone for aggression in conduct disorder with ID and refuses the indication in autistic disorder, and the United States authorises it for irritability in autistic disorder, with different dosing and duration regimens [103, 104].

8. Reproductive Options and Family Counselling

8.1. Recurrence risks with a molecular diagnosis

With both variants identified and the parents heterozygous, each pregnancy has a 25% probability of an affected child, 50% of an asymptomatic carrier child and 25% of a child with neither variant [10]. Clinically unaffected siblings of a patient have a two-thirds probability of being carriers [91]. Carriers are asymptomatic in strictly recessive genes, but not necessarily in genes with two inheritance patterns: heterozygotes for exonic NRXN1 deletions have an increased risk of neurodevelopmental disorders, and METTL23 heterozygosity has been associated, with mixed evidence, with normal-tension glaucoma [3].

There are three situations in which the simple model does not apply. If the child is homozygous and only one parent is a carrier, the explanations are a deletion of the other allele (which mimics homozygosity) or uniparental isodisomy, already described on chromosome 16 in THOC6 syndrome; the recurrence risk is then much lower [10, 30]. If the child is compound heterozygous and only one parent has one of the variants, the other may have arisen de novo; the rate is low in recessive disorders in general, but in the recessive form of RNU2-2 a de novo variant in trans with an inherited allele is frequent, and the recurrence risk depends on germline mosaicism in the parent in whom the variant arose [5, 10]. Finally, in families with several affected members, intrafamilial heterogeneity requires confirmation of the genotype in each patient before counselling the branches of the family [13].

8.2. Risks without a molecular diagnosis

When the parents are consanguineous and the cause of non-specific ID has not been found, a recurrence risk of 25% should be assumed unless proven otherwise [30]. Without consanguinity, the empirical risk is lower, because the proportion of de novo dominant causes is higher. Classification as probable or possible recessive (section 4.1) guides counselling [87].

8.3. Risk for consanguineous couples without known disease

First cousins share, on average, one eighth of their genes; siblings and parents and children share half, and second-degree relatives (half-siblings, uncle and niece) one quarter [91]. In birth studies, the prevalence of severe congenital or genetic disease diagnosed in the first year of life rises from about 2–2.5% in unrelated parents to about 4–4.5% in first cousins, and studies with longer follow-up, which include late-diagnosed neurological disorders, estimate about 4% versus about 8% [91]. The theoretical increase, assuming that each common grandparent carries one deleterious recessive allele, is about 3% (1/32). For second cousins or first cousins once removed, the empirical increase is about one percentage point and the theoretical one less than 1% (1/128) [91]. In closer unions (uncle and niece, double first cousins) the probability of homozygosity by descent is greater, and in multiple consanguinity the contribution of each common ancestor is calculated and all are summed [91]. In communities with a long tradition of consanguineous marriage, the actual risks may exceed those estimated from a three-generation pedigree [87].

In the Born in Bradford birth cohort study (United Kingdom, 13,776 babies recruited between 2007 and 2011), 18% of babies were children of first cousins, mostly of Pakistani origin (37% in this group); consanguinity was associated with a doubling of the risk of congenital anomaly (multivariate relative risk 2.19; 95% CI 1.67–2.85), with no association with deprivation, and 31% of anomalies in children of Pakistani origin were attributed to consanguinity; the outcome measured was congenital anomalies and not ID [107]. In a Birmingham birth study, the prevalence of recessive disorders was 0.28% in children of northern European origin (0.4% with related parents) and 3.0–3.3% in British children of Pakistani origin (69% with related parents) [91]. In populations with high traditional consanguinity, no measurable increase in miscarriages or infertility has been found [91].

The 2021 focused revision of the NSGC practice resource on counselling and screening of consanguineous couples acknowledges that there are no evidence-based guidelines for this counselling and rests on expert opinion and the literature since 2002 [97, 108]. The 2010 Geneva international workshop emphasised the need for evidence-based counselling that recognises both the risks and the social benefits attributed to these unions by the communities that practise them [100]. Counselling should be sensitive to the communities at risk, and information should be offered to the extended family, in bilingual format when necessary [87, 107].

8.4. Carrier testing and the extended family

With the familial variants known, carrier testing can be offered to the parents' siblings, to the patient's siblings and to their partners when these share ancestry or are related [3, 10]. The risk for the offspring of a relative is calculated by multiplying the probability of each member of the couple being a carrier by one quarter. For example, an unaffected sibling of a patient (two-thirds probability of being a carrier) who marries a first cousin (one quarter, as niece of an obligate carrier) has a risk of about 1/24 per pregnancy; an uncle of a patient who marries a first cousin once removed has a risk of about 1/128 [91]. When the variants are not known, testing of the spouse is not possible and the risk calculated from the pedigree is the basis of counselling [87, 91]. In consanguineous couples without known disease, carrier screening may be guided by ancestry (for example, haemoglobinopathies and cystic fibrosis in several populations, Tay-Sachs disease in Ashkenazi Jews) or performed with expanded panels [91].

8.5. Prenatal and preimplantation diagnosis

With both pathogenic or likely pathogenic variants identified, prenatal diagnosis is performed by CVS or amniocentesis, and PGD is possible; the laboratory should be consulted about the type of sample [10, 87]. Detailed fetal ultrasound should be offered to couples who are first cousins or closer, but it detects only a small proportion of severe recessive disorders (polydactyly, cystic kidneys, heart disease, brain anomalies) and almost no neurodevelopmental disorders; for more distant relationships routine obstetric ultrasound is sufficient [91]. When a variant in a gene with limited evidence (PRSS12, TECR, ZC3H14) or a variant of uncertain significance is the only finding, its use in reproductive decisions should be discussed explicitly with the couple, because the result may not explain the disease [3, 4]. Non-invasive prenatal diagnosis of monogenic diseases is under investigation [19].

8.6. Counselling of minors and the extended family

The sources consulted contain no class-specific recommendations on carrier testing in minors; information, education and the offer of counselling should extend to the extended family [87]. Laboratory detection of undeclared parental consanguinity raises ethical and legal issues, and its reporting follows laboratory policy [17]. Parents and family may be referred to patient associations (section 9.5).

9. Special Considerations: Population Diversity and Epidemiology

9.1. What is known about frequency

There is no validated estimate of the prevalence of the class; Orphanet records no numerical value for ORPHA:88616 and classifies the prevalence of the group ORPHA:101685 as unknown [1, 3]. The available data measure different things. The prevalence of ID in the general population lies at 1–3%, according to different authors [11, 30, 98]. The recessive fraction varies with consanguinity: in the cohorts from the Middle East and South Asia, most diagnoses are recessive (81% of point mutations in Saudi Arabia), whereas in a mostly non-consanguineous Chinese cohort recessive variants accounted for 12.31% of diagnostic SNVs [41, 102]. In outbred populations, the fraction of ID with a recessive cause has been estimated at 13–24% [2, 13]. The counts of published patients per gene (for example, 76 for TRAPPC9, more than 80 for ADAT3, 43 for TRMT1, 24 for ST3GAL3, three families for CRBN) reflect research activity and not the actual frequency [3, 14, 66, 81].

The recessive form of RNU2-2 is an exception to the rarity of each gene: it accounts for about 10% of families with recessive neurodevelopmental disease diagnosable by sequencing in Greene's cohorts and was at least twice as frequent as the dominant form in the French cohort of 34,329 people with rare diseases [5, 29]. These figures measure the frequency among sequenced patients and not the prevalence in the population.

9.2. Consanguinity worldwide

About 1.1 billion people live in countries where consanguineous marriage is customary, and in these about one in three marriages is between cousins; the preference persists in migrant communities in Western countries [100]. The countries with consanguinity above 10%, mainly in Africa, the Middle East and South Asia, account for about 1.3 billion people; in Pakistan, about 62.7% of marriages are consanguineous, of which about 80.4% are between first cousins, and in Iran about 40% [13]. Traditional consanguinity also exists in some Jewish communities and among Irish travellers and is, in general, independent of religion [91]. The countries with frequent parental consanguinity represent about one seventh of the world population [12].

9.3. Founder populations and recurrent variants

European and North American data on the class come mainly from migrant families and isolates (Hutterites, Amish), from families of unrelated parents from the same locality (Sicily, Italy) and from rare disease cohort studies, such as the French genomic sequencing plan that described the recessive form of RNU2-2 [3, 5, 89]. In the TRAPPC9 review, 18.1% of patients had European ancestry [14], and the 2020 review placed all the families then known in the Mediterranean region, the Middle East, Southern Europe and South Asia [49].

9.4. Portugal and Brazil

The sources consulted contain no prevalence data or series of Portuguese or Brazilian patients with single-gene recessive ID. The French RNU2-2 study included a Brazilian collaboration (Ribeirão Preto), without separate frequency data for Brazil [5]. The lack of data in the sources consulted does not demonstrate that there are no publications on these populations.

9.5. Special populations, European networks and patient associations

Adults with ID who have never been investigated form a population with accessible diagnoses and with less access to the exome than children [90]. The median age at death of people with ID in the United Kingdom is 13 years lower than that of the general population in men and 20 years lower in women, according to a study of avoidable mortality cited by Harripaul and colleagues [13]. Migrant families with consanguinity need an interpreter, bilingual material and culturally sensitive counselling [87, 107].

The relevant European reference network is ERN ITHACA (rare malformation syndromes and ID) [20]. Patient and family associations include Unique (United Kingdom; rare chromosome and gene disorders), Climb (United Kingdom), the National Organization for Rare Disorders (NORD, United States) and the SWAN UK and SWAN USA networks, aimed at families without a diagnosis [87, 109]. The Treatable ID app supports clinicians in the search for treatable causes [101].

10. Illustrative Clinical Case

10.1. TRAPPC9 deficiency in a 27-year-old man from Malta

Wilton and colleagues described in 2020 a 27-year-old man, born in Malta by uncomplicated vaginal delivery, the son of parents with no known consanguinity and with no family history of ID or ASD (pedigree in figure 1a) [49]. He sat at 6 months and walked at 14 months, and in early childhood had a social smile and eye contact. At 19 months he did not speak, did not imitate tongue or lip movements and clapped his hands repetitively; the karyotype was normal and, at around 18 months, he was diagnosed with ASD. He received intensive therapy, including speech therapy up to 16 years of age, for about fifteen years, without ever acquiring language. At 27 years he followed simple commands, was dependent in all activities of daily living and remained ambulant; he had restricted and repetitive behaviours and worsening agitation, with episodes of aggression towards others when faced with new stimuli or environments [49].

On examination he made no eye contact and had a sloping forehead, prominent supraorbital ridges and fleshy ears with retrolobular creases (figure 1b); he measured 171.8 cm and had a head circumference of 54.5 cm, close to the 3rd centile, a microcephaly relative to height; the hands were small, 17 cm long with a 7 cm middle finger (figure 1c) [49]. Two brain MRIs more than ten years apart were similar: global loss of brain volume, cerebellar atrophy, thinning of the corpus callosum and a lacunar infarct in the right anterior basal ganglia, including the head of the caudate nucleus (figures 1d to 1i). Cytochrome P450 pharmacogenomic testing did not explain the agitation with psychotropic drugs [49].

Exome sequencing (Mayo Medical Laboratories) identified a homozygous seven-base-pair deletion in TRAPPC9, described on transcript NM_031466.7 as c.568_574del p.(Trp190ArgfsTer95), which corresponds on the MANE transcript to NM_001160372.4:c.274_280del p.(Trp92ArgfsTer95), absent from gnomAD and classified as likely pathogenic (criteria PVS1, PM2 and PP4); the variant was confirmed by Sanger sequencing and both parents were heterozygous [49, 53]. The authors reviewed 48 patients from 19 families then published: ID in 100%, absent or delayed speech in 42/42, microcephaly in 41/45, abnormal MRI in 22/22 (thin corpus callosum in 20) and behavioural changes in only 9/48; 84% of individuals and 69% of families had some degree of consanguinity, and all families were of Mediterranean, Middle Eastern, Southern European or South Asian origin (figure 2c) [49].

The figures of the article are available in PubMed Central (figure 1: pedigree, facial appearance, hands and MRI, https://pmc.ncbi.nlm.nih.gov/articles/PMC7216808/figure/mgg31211-fig-0001/; figure 2: location of the variants in the protein and geographical distribution, https://pmc.ncbi.nlm.nih.gov/articles/PMC7216808/figure/mgg31211-fig-0002/); the description is based on the full text and the figure legends [49].

10.2. Lessons from the case

The case brings together several points of this article. First, the absence of known consanguinity does not exclude a recessive disorder, especially in island populations or those of Mediterranean origin [14, 49]. Second, TRAPPC9 deficiency may be labelled for decades as severe ASD; the authors recommend extensive genetic testing in children of Mediterranean or Middle Eastern origin with severe ID and absent speech, especially with consanguinity [49]. Third, the late diagnosis had practical consequences: the authors suggest prioritising visual and augmentative communication and motor training for activities of daily living when speech therapy does not work, an authors' opinion and not a guideline recommendation [49]. Fourth, the variant was described on an old transcript, and comparison with current databases requires conversion to the MANE transcript [53]. The management in the report (pharmacogenomics, psychotropic drugs) is that of the individual case and does not constitute routine practice.

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

Functional prognosis depends on the severity of ID and on comorbidities, which vary between genes. The non-syndromic forms are non-progressive and compatible with survival into adulthood: CRBN presents with mild ID (IQ 50–70) without deterioration [3, 34]. In the syndromic forms, prognosis is conditioned by epilepsy (refractory in 75% of patients with epilepsy due to ST3GAL3; controlled in about two thirds in CNTNAP2, with language regression after seizure onset), by absent speech (constant in TRAPPC9) and by obesity (52.8% in TRAPPC9) [14, 66, 82]. Life expectancy has not been established for any gene: the oldest patient with MAN1B1-CDG was 35 years old and the first patients with THOC6 reached their early fifth decade [10, 46]. In people with ID in general, the median age at death in the United Kingdom is 13 years lower than that of the population in men and 20 years lower in women [13]. The recessive form of RNU2-2 has the prognosis of a severe epileptic encephalopathy, with severe to profound ID and absence of walking and communication in many patients [85].

11.2. Surveillance

There are no published surveillance protocols for the genes of the class. The table adapts the GeneReviews model for recessive syndromic ID and adds the points specific to each gene and to the drugs used; the intervals are those of the sources, and where these say “at each visit” they do not set the frequency of visits.

11.3. Adult follow-up

Transition should be prepared in adolescence and integrated into the education plan [10]. Adults with undiagnosed ID should have access to sequencing, because in a community study the exome found pathogenic or likely pathogenic variants in three of five families, and adults' access to the exome is lower than that of children [90]. Agitation and aggression towards others may worsen in adulthood, as in the Maltese case, and require regular review of psychotropic drugs [49].

12. Summary and Final Considerations

Single-gene autosomal recessive ID brings together hundreds of genes, each responsible for a small fraction of patients, whose common element is non-progressive ID with disproportionate language delay, caused by biallelic loss of function. The distinction between non-syndromic and syndromic forms is a continuum: phenotyping of new patients has converted several “pure” genes into syndromes with microcephaly, epilepsy, obesity, ASD or dysmorphism, such as TRAPPC9, MAN1B1, CC2D1A and NSUN2. PRSS12, the first gene described in 2002, now has only limited evidence, and interpretation of biallelic variants in genes with limited or moderate evidence requires familial segregation and caution in the report.

The relative frequency of the class depends on the population. In countries with frequent parental consanguinity, which together account for about one seventh of the world population, recessive causes are the main genetic cause of ID; in outbred populations they have been estimated at 13–24% of cases and occur in families without known consanguinity, in isolates such as the Hutterites and the Amish and, in the recessive form of RNU2-2, with a de novo variant on one of the alleles. This form, described in 2026, is the most prevalent recessive neurodevelopmental disorder known and requires the analysis to include non-coding genes.

The investigation rests on a careful pedigree, on first-tier metabolic screening so as not to miss treatable causes and on exome or genome sequencing, preferably as a trio and with analysis of CNVs and runs of homozygosity. The limits of the tests are concrete: intragenic CNVs, cryptic structural variants, non-coding genes, a normal transferrin profile in TUSC3 and the possibility of de novo dominant variants even in children of consanguineous parents. There is no targeted treatment; management is supportive, with early intervention, augmentative communication, treatment of epilepsy and behaviour at label doses and surveillance of the manifestations of each gene. Counselling starts from a recurrence risk of 25% and extends to the extended family, with risk calculation from the pedigree, carrier testing when the variants are known and prenatal or preimplantation diagnosis.

13. References

1. Orphanet. Autosomal recessive non-syndromic intellectual disability (ORPHA:88616) [Internet]. Paris: INSERM; 2026 [record of 2026-07-02; accessed 2026-10-28]. Available from: https://www.orpha.net/en/disease/detail/88616
2. Musante L, Ropers HH. Genetics of recessive cognitive disorders. Trends Genet. 2014;30(1):32-9. doi:10.1016/j.tig.2013.09.008. PMID: 24176302.
3. GeneGuidance. Curation of genes and conditions: PRSS12, CRBN, CC2D1A, NSUN2, GRIK2, TUSC3, ST3GAL3, TRAPPC9, TECR, MAN1B1, MED23, ADAT3, METTL23, FMN2, ZC3H14, TRMT1, CNTNAP2, NRXN1, THOC6 and MRT phenotypic series (OMIM PS249500) [database]. Version of 26 September 2026; accessed 28 Oct 2026.
4. Clinical Genome Resource (ClinGen). Gene-Disease Validity curations. Available from: https://search.clinicalgenome.org/kb/gene-validity (API /api/validity accessed 2026-10-28)
5. Leitão E, Santini A, Cogne B, Essid M, Athanasiadou M, LaFlamme CW, et al. Systematic analysis of snRNA genes reveals frequent RNU2-2 variants in dominant and recessive developmental and epileptic encephalopathies. Nat Genet. 2026;58(4):782-797. doi:10.1038/s41588-026-02547-5. PMID: 41912934. PMC13083260.
6. OMIM® Online Mendelian Inheritance in Man. #249500 INTELLECTUAL DEVELOPMENTAL DISORDER, AUTOSOMAL RECESSIVE 1; MRT1. Johns Hopkins University; last edited 2024-03-29. https://omim.org/entry/249500. Accessed 28 October 2026.
7. OMIM® Online Mendelian Inheritance in Man. %611107 INTELLECTUAL DEVELOPMENTAL DISORDER, AUTOSOMAL RECESSIVE 4; MRT4. Johns Hopkins University; last edited 2022-04-05. https://omim.org/entry/611107. Accessed 28 October 2026.
8. Orphanet. Rare non-syndromic intellectual disability (ORPHA:101685). Available from: https://www.orpha.net/en/disease/detail/101685
9. Sweetser DA, Gipson KS, Zar-Kessler C. Pitt-Hopkins Syndrome. 2012 Aug 30 [updated 2025 May 22]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK100240/
10. Lemire G, Innes AM, Boycott KM. THOC6 Intellectual Disability Syndrome. 2020 Aug 13. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK560442/ PMID: 32790266.
11. Riazuddin S, Hussain M, Razzaq A, Iqbal Z, Shahzad M, Polla DL, et al. Exome sequencing of Pakistani consanguineous families identifies 30 novel candidate genes for recessive intellectual disability. Mol Psychiatry. 2017;22(11):1604-1614. doi:10.1038/mp.2016.109. PMID: 27457812. PMC5658665.
12. Hu H, Kahrizi K, Musante L, Fattahi Z, Herwig R, Hosseini M, et al. Genetics of intellectual disability in consanguineous families. Mol Psychiatry. 2019;24(7):1027-1039. doi:10.1038/s41380-017-0012-2. PMID: 29302074.
13. Harripaul R, Vasli N, Mikhailov A, Rafiq MA, Mittal K, Windpassinger C, et al. Mapping autosomal recessive intellectual disability: combined microarray and exome sequencing identifies 26 novel candidate genes in 192 consanguineous families. Mol Psychiatry. 2018;23(4):973-984. doi:10.1038/mp.2017.60. PMID: 28397838.
14. Calvo M, Reynolds G, Luca M, Di Gregorio E, Cardaropoli S, Salvo E, et al. TRAPPC9-Related Intellectual Developmental Disorder: A Systematic Review and a Novel Case of a Complex Structural Variant. Genes (Basel). 2026;17(6):658. doi:10.3390/genes17060658. PMID: 42353817. PMC13299287.
15. 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.
16. Srivastava S, Love-Nichols JA, Dies KA, Ledbetter DH, Martin CL, Chung WK, et al. Meta-analysis and multidisciplinary consensus statement: exome sequencing is a first-tier clinical diagnostic test for individuals with neurodevelopmental disorders. Genet Med. 2019;21(11):2413-2421. doi:10.1038/s41436-019-0554-6. PMID: 31182824. PMC6831729.
17. Gonzales PR, Andersen EF, Brown TR, Horner VL, Horwitz J, Rehder CW, et al. Interpretation and reporting of large regions of homozygosity and suspected consanguinity/uniparental disomy, 2021 revision: A technical standard of the American College of Medical Genetics and Genomics (ACMG). Genet Med. 2022;24(2):255-261. doi:10.1016/j.gim.2021.10.004. PMID: 34906464.
18. Hoytema van Konijnenburg EMM, Wortmann SB, Koelewijn MJ, Tseng LA, Houben R, Stöckler-Ipsiroglu S, et al. Treatable inherited metabolic disorders causing intellectual disability: 2021 review and digital app. Orphanet J Rare Dis. 2021;16(1):170. doi:10.1186/s13023-021-01727-2. PMID: 33845862. PMC8042729.
19. ClinicalTrials.gov API v2. Searches “PRSS12”, “neurotrypsin”, “CC2D1A”, “TRAPPC9”, “TUSC3”, “autosomal recessive intellectual disability”. Available from: https://clinicaltrials.gov (accessed 2026-10-28)
20. ERN ITHACA — European Reference Network. Available from: https://ern-ithaca.eu/
21. OMIM® Online Mendelian Inheritance in Man. #611093 INTELLECTUAL DEVELOPMENTAL DISORDER, AUTOSOMAL RECESSIVE 7; MRT7. Johns Hopkins University; last edited 2025-09-26. https://omim.org/entry/611093. Accessed 28 October 2026.
22. OMIM® Online Mendelian Inheritance in Man. #615516 INTELLECTUAL DEVELOPMENTAL DISORDER, AUTOSOMAL RECESSIVE 38; MRT38. Johns Hopkins University; last edited 2021-12-15. https://omim.org/entry/615516. Accessed 28 October 2026.
23. OMIM® Online Mendelian Inheritance in Man. *609262 CEREBLON; CRBN. Johns Hopkins University; last edited 2026-09-04. https://omim.org/entry/609262. Accessed 28 October 2026.
24. Orphanet. Intellectual disability-obesity-brain malformations-facial dysmorphism syndrome (ORPHA:352530). Available from: https://www.orpha.net/en/disease/detail/352530
25. Orphanet. CNTNAP2-related developmental and epileptic encephalopathy (ORPHA:163681). Available from: https://www.orpha.net/en/disease/detail/163681
26. Orphanet. Intellectual disability-strabismus syndrome (ORPHA:363528). Available from: https://www.orpha.net/en/disease/detail/363528
27. Orphanet. MAN1B1-CDG (ORPHA:397941). Available from: https://www.orpha.net/en/disease/detail/397941
28. Orphanet. ST3GAL3-CDG (ORPHA:697734). Available from: https://www.orpha.net/en/disease/detail/697734
29. Greene D, Mendez R, Lees J, Barbosa M, Bruselles A, Chiriatti L, et al. Biallelic variants in RNU2-2 cause the most prevalent known recessive neurodevelopmental disorder. Nat Genet. 2026;58(4):774-781. doi:10.1038/s41588-026-02539-5. PMID: 41912932. PMC13083235.
30. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 188–193 (Intellectual disability).
31. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 816–818 (Coffin-Siris syndrome).
32. Molinari F, Rio M, Meskenaite V, Encha-Razavi F, Augé J, Bacq D, et al. Truncating neurotrypsin mutation in autosomal recessive nonsyndromic mental retardation. Science. 2002;298(5599):1779-81. doi:10.1126/science.1076521. PMID: 12459588.
33. Molinari F, Meskanaite V, Munnich A, Sonderegger P, Colleaux L. Extracellular proteases and their inhibitors in genetic diseases of the central nervous system. Hum Mol Genet. 2003;12 Spec No 2:R195-200. doi:10.1093/hmg/ddg276. PMID: 12925575.
34. Higgins JJ, Pucilowska J, Lombardi RQ, Rooney JP. A mutation in a novel ATP-dependent Lon protease gene in a kindred with mild mental retardation. Neurology. 2004;63(10):1927-31. doi:10.1212/01.wnl.0000146196.01316.a2. PMID: 15557513. PMC1201536.
35. Basel-Vanagaite L, Attia R, Yahav M, Ferland RJ, Anteki L, Walsh CA, et al. The CC2D1A, a member of a new gene family with C2 domains, is involved in autosomal recessive non-syndromic mental retardation. J Med Genet. 2006;43(3):203-10. doi:10.1136/jmg.2005.035709. PMID: 16033914. PMC2563235.
36. Garshasbi M, Hadavi V, Habibi H, Kahrizi K, Kariminejad R, Behjati F, et al. A defect in the TUSC3 gene is associated with autosomal recessive mental retardation. Am J Hum Genet. 2008;82(5):1158-64. doi:10.1016/j.ajhg.2008.03.018. PMID: 18452889. PMC2651624.
37. Molinari F, Foulquier F, Tarpey PS, Morelle W, Boissel S, Teague J, et al. Oligosaccharyltransferase-subunit mutations in nonsyndromic mental retardation. Am J Hum Genet. 2008;82(5):1150-7. doi:10.1016/j.ajhg.2008.03.021. PMID: 18455129. PMC2427205.
38. Mir A, Kaufman L, Noor A, Motazacker MM, Jamil T, Azam M, et al. Identification of mutations in TRAPPC9, which encodes the NIK- and IKK-beta-binding protein, in nonsyndromic autosomal-recessive mental retardation. Am J Hum Genet. 2009;85(6):909-15. doi:10.1016/j.ajhg.2009.11.009. PMID: 20004765. PMC2790571.
39. Mochida GH, Mahajnah M, Hill AD, Basel-Vanagaite L, Gleason D, Hill RS, et al. A truncating mutation of TRAPPC9 is associated with autosomal-recessive intellectual disability and postnatal microcephaly. Am J Hum Genet. 2009;85(6):897-902. doi:10.1016/j.ajhg.2009.10.027. PMID: 20004763. PMC2790576.
40. Najmabadi H, Hu H, Garshasbi M, Zemojtel T, Abedini SS, Chen W, et al. Deep sequencing reveals 50 novel genes for recessive cognitive disorders. Nature. 2011;478(7367):57-63. doi:10.1038/nature10423. PMID: 21937992.
41. Anazi S, Maddirevula S, Faqeih E, Alsedairy H, Alzahrani F, Shamseldin HE, et al. Clinical genomics expands the morbid genome of intellectual disability and offers a high diagnostic yield. Mol Psychiatry. 2017;22(4):615-624. doi:10.1038/mp.2016.113. PMID: 27431290.
42. Reuter MS, Tawamie H, Buchert R, Hosny Gebril O, Froukh T, Thiel C, et al. Diagnostic Yield and Novel Candidate Genes by Exome Sequencing in 152 Consanguineous Families With Neurodevelopmental Disorders. JAMA Psychiatry. 2017;74(3):293-299. doi:10.1001/jamapsychiatry.2016.3798. PMID: 28097321.
43. Kochinke K, Zweier C, Nijhof B, Fenckova M, Cizek P, Honti F, et al. Systematic Phenomics Analysis Deconvolutes Genes Mutated in Intellectual Disability into Biologically Coherent Modules. Am J Hum Genet. 2016;98(1):149-64. doi:10.1016/j.ajhg.2015.11.024. PMID: 26748517. PMC4716705.
44. Rafiq MA, Kuss AW, Puettmann L, Noor A, Ramiah A, Ali G, et al. Mutations in the alpha 1,2-mannosidase gene, MAN1B1, cause autosomal-recessive intellectual disability. Am J Hum Genet. 2011;89(1):176-82. doi:10.1016/j.ajhg.2011.06.006. PMID: 21763484.
45. Rymen D, Peanne R, Millón MB, Race V, Sturiale L, Garozzo D, et al. MAN1B1 deficiency: an unexpected CDG-II. PLoS Genet. 2013;9(12):e1003989. doi:10.1371/journal.pgen.1003989. PMID: 24348268.
46. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 43, Congenital Disorders of Glycosylation, Dolichol and Glycosylphosphatidylinositol Metabolism (43.1, quadro; 43.1.10 MAN1B1-CDG).
47. Manzini MC, Xiong L, Shaheen R, Tambunan DE, Di Costanzo S, Mitisalis V, et al. CC2D1A regulates human intellectual and social function as well as NF-κB signaling homeostasis. Cell Rep. 2014;8(3):647-55. doi:10.1016/j.celrep.2014.06.039. PMID: 25066123. PMC4334362.
48. Kim AH, Sakin I, Viviano S, Tuncel G, Aguilera SM, Goles G, et al. CC2D1A causes ciliopathy, intellectual disability, heterotaxy, renal dysplasia, and abnormal CSF flow. Life Sci Alliance. 2024;7(10). doi:10.26508/lsa.202402708. PMID: 39168639. PMC11339347.
49. Wilton KM, Gunderson LB, Hasadsri L, Wood CP, Schimmenti LA. Profound intellectual disability caused by homozygous TRAPPC9 pathogenic variant in a man from Malta. Mol Genet Genomic Med. 2020;8(5):e1211. doi:10.1002/mgg3.1211. PMID: 32162493. PMC7216808.
50. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 39, Disorders of Nucleic Acid Metabolism, tRNA Metabolism and Ribosomal Biogenesis (39.1 quadro clínico; 39.2 tRNA Processing Disorders).
51. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 42, 42.4.11 Trans-2,3-Enoyl-CoA Reductase Deficiency.
52. Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 44, Disorders of Cellular Trafficking (44.1.3).
53. 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. PMC5765404.
54. National Center for Biotechnology Information. ClinVar. Available from: https://www.ncbi.nlm.nih.gov/clinvar/ (VCV records cited in the text; accessed 2026-10-28)
55. Yokoi F, Tanaka Y, Sakai A, Isogai N, Miyata S, Mitsui S. Amyloid neuropathy, tauopathy, decreased cholinergic and dopaminergic neurons, long-term memory and motor deficits, and sleep disturbance in motopsin deficient mice. Geroscience. 2026;48(1):1009-1035. doi:10.1007/s11357-025-01694-7. PMID: 40407976. PMC12972478.
56. Miyata S, Tsuda M, Mitsui S. Overexpression of Motopsin, an Extracellular Serine Protease Related to Intellectual Disability, Promotes Adult Neurogenesis and Neuronal Responsiveness in the Dentate Gyrus. Mol Neurobiol. 2024;61(8):4929-4948. doi:10.1007/s12035-023-03890-y. PMID: 38153682.
57. Ali Z, Babar ME, Ahmad J, Yousaf MZ, Asif M, Shah SA. Molecular investigation of mental retardation locus gene PRSS12 by linkage analysis. Indian J Hum Genet. 2011;17(2):65-9. doi:10.4103/0971-6866.86178. PMID: 22090715. PMC3214320.
58. Xing L, Zhang K, Dai L, Li F, Guo Y, Huang S, et al. A family-based association study of PRSS12 and mental retardation in the Qinba mountain region of China. Psychiatr Genet. 2013;23(3):137. doi:10.1097/YPG.0b013e32835d6fed. PMID: 23344636.
59. Abbasi-Moheb L, Mertel S, Gonsior M, Nouri-Vahid L, Kahrizi K, Cirak S, et al. Mutations in NSUN2 cause autosomal-recessive intellectual disability. Am J Hum Genet. 2012;90(5):847-55. doi:10.1016/j.ajhg.2012.03.021. PMID: 22541559.
60. Khan MA, Rafiq MA, Noor A, Hussain S, Flores JV, Rupp V, et al. Mutation in NSUN2, which encodes an RNA methyltransferase, causes autosomal-recessive intellectual disability. Am J Hum Genet. 2012;90(5):856-63. doi:10.1016/j.ajhg.2012.03.023. PMID: 22541562.
61. Martinez FJ, Lee JH, Lee JE, Blanco S, Nickerson E, Gabriel S, et al. Whole exome sequencing identifies a splicing mutation in NSUN2 as a cause of a Dubowitz-like syndrome. J Med Genet. 2012;49(6):380-5. doi:10.1136/jmedgenet-2011-100686. PMID: 22577224.
62. El Chehadeh S, Bonnet C, Callier P, Béri M, Dupré T, Payet M, et al. Homozygous Truncating Intragenic Duplication in TUSC3 Responsible for Rare Autosomal Recessive Nonsyndromic Intellectual Disability with No Clinical or Biochemical Metabolic Markers. JIMD Rep. 2015;20:45-55. doi:10.1007/8904_2014_390. PMID: 25626710.
63. Gumus E. A Homozygous 1.16 Megabases Microdeletion at 8p22 Including The Whole TUSC3 in A Three Years Old Girl with Intellectual Disability and Speech Delay. Cell J. 2020;22(1):128-132. doi:10.22074/cellj.2020.6606. PMID: 31606977. PMC6791072.
64. Park G, Kim N, Kim SY, Lee H, Li CM, Seol JH, et al. TUSC3 regulates ERMA-mediated Mg(2+) uptake for synaptic function and neurodevelopment. Nat Commun. 2025;16(1):9752. doi:10.1038/s41467-025-65668-1. PMID: 41203647. PMC12594762.
65. Hu H, Eggers K, Chen W, Garshasbi M, Motazacker MM, Wrogemann K, et al. ST3GAL3 mutations impair the development of higher cognitive functions. Am J Hum Genet. 2011;89(3):407-14. doi:10.1016/j.ajhg.2011.08.008. PMID: 21907012.
66. Whitney R, Jain P, RamachandranNair R, Jones KC, Kiani H, Tarnopolsky M, et al. The epilepsy phenotype of ST3GAL3-related developmental and epileptic encephalopathy. Epilepsia Open. 2023;8(2):623-632. doi:10.1002/epi4.12747. PMID: 37067065.
67. Uctepe E, Yesilyurt A, Esen FN, Tumer S, Mancilar H, Sonmez FM. TRAPPC9-Related Intellectual Disability: Report of Two New Cases and Review of the Literature. Mol Syndromol. 2023;14(6):485-492. doi:10.1159/000531439. PMID: 38058760. PMC10697769.
68. Radenkovic S, Martinelli D, Zhang Y, Preston GJ, Maiorana A, Terracciano A, et al. TRAPPC9-CDG: A novel congenital disorder of glycosylation with dysmorphic features and intellectual disability. Genet Med. 2022;24(4):894-904. doi:10.1016/j.gim.2021.12.012. PMID: 35042660.
69. Li Y, Usman M, Sapp E, Ke Y, Wang Z, Boudi A, et al. Chronic pharmacologic manipulation of dopamine transmission ameliorates metabolic disturbance in Trappc9-linked brain developmental syndrome. JCI Insight. 2024;9(15). doi:10.1172/jci.insight.181339. PMID: 38889014. PMC11383600.
70. Çalışkan M, Chong JX, Uricchio L, Anderson R, Chen P, Sougnez C, et al. Exome sequencing reveals a novel mutation for autosomal recessive non-syndromic mental retardation in the TECR gene on chromosome 19p13. Hum Mol Genet. 2011;20(7):1285-9. doi:10.1093/hmg/ddq569. PMID: 21212097.
71. Abe K, Ohno Y, Sassa T, Taguchi R, Çalışkan M, Ober C, et al. Mutation for nonsyndromic mental retardation in the trans-2-enoyl-CoA reductase TER gene involved in fatty acid elongation impairs the enzyme activity and stability, leading to change in sphingolipid profile. J Biol Chem. 2013;288(51):36741-9. doi:10.1074/jbc.M113.493221. PMID: 24220030.
72. Platt D, Davis-Keppen L. Individual with Homozygous TECR Variant Expands Upon the Existing Phenotype for a Hutterite Founder Mutation. S D Med. 2022;75(4):150-153. PMID: 35709344.
73. Hashimoto S, Boissel S, Zarhrate M, Rio M, Munnich A, Egly JM, et al. MED23 mutation links intellectual disability to dysregulation of immediate early gene expression. Science. 2011;333(6046):1161-3. doi:10.1126/science.1206638. PMID: 21868677.
74. Alazami AM, Hijazi H, Al-Dosari MS, Shaheen R, Hashem A, Aldahmesh MA, et al. Mutation in ADAT3, encoding adenosine deaminase acting on transfer RNA, causes intellectual disability and strabismus. J Med Genet. 2013;50(7):425-30. doi:10.1136/jmedgenet-2012-101378. PMID: 23620220.
75. El-Hattab AW, Saleh MA, Hashem A, Al-Owain M, Asmari AA, Rabei H, et al. ADAT3-related intellectual disability: Further delineation of the phenotype. Am J Med Genet A. 2016;170A(5):1142-7. doi:10.1002/ajmg.a.37578. PMID: 26842963.
76. Rafat K, Abdel-Aleem AF, Elbendary HM, Issa MY, Essawi ML, Abdel-Ghafar SF, et al. ADAT3-related neurodevelopmental disorder in 24 new patients with a high frequency of the p.Val144Met and a new founder variant. Sci Rep. 2025;15(1):20329. doi:10.1038/s41598-025-06857-2. PMID: 40579404.
77. Reiff RE, Ali BR, Baron B, Yu TW, Ben-Salem S, Coulter ME, et al. METTL23, a transcriptional partner of GABPA, is essential for human cognition. Hum Mol Genet. 2014;23(13):3456-66. doi:10.1093/hmg/ddu054. PMID: 24501276.
78. Bernkopf M, Webersinke G, Tongsook C, Koyani CN, Rafiq MA, Ayaz M, et al. Disruption of the methyltransferase-like 23 gene METTL23 causes mild autosomal recessive intellectual disability. Hum Mol Genet. 2014;23(15):4015-23. doi:10.1093/hmg/ddu115. PMID: 24626631.
79. Law R, Dixon-Salazar T, Jerber J, Cai N, Abbasi AA, Zaki MS, et al. Biallelic truncating mutations in FMN2, encoding the actin-regulatory protein Formin 2, cause nonsyndromic autosomal-recessive intellectual disability. Am J Hum Genet. 2014;95(6):721-8. doi:10.1016/j.ajhg.2014.10.016. PMID: 25480035.
80. Pak C, Garshasbi M, Kahrizi K, Gross C, Apponi LH, Noto JJ, et al. Mutation of the conserved polyadenosine RNA binding protein, ZC3H14/dNab2, impairs neural function in Drosophila and humans. Proc Natl Acad Sci U S A. 2011;108(30):12390-5. doi:10.1073/pnas.1107103108. PMID: 21734151.
81. Efthymiou S, Leo CP, Deng C, Lin SJ, Maroofian R, Lin R, et al. Bi-allelic pathogenic variants in TRMT1 disrupt tRNA modification and induce a neurodevelopmental disorder. Am J Hum Genet. 2025;112(5):1117-1138. doi:10.1016/j.ajhg.2025.03.015. PMID: 40245862.
82. D'Onofrio G, Accogli A, Severino M, Caliskan H, Kokotović T, Blazekovic A, et al. Genotype-phenotype correlation in contactin-associated protein-like 2 (CNTNAP-2) developmental disorder. Hum Genet. 2023;142(7):909-925. doi:10.1007/s00439-023-02552-2. PMID: 37183190.
83. Strauss KA, Puffenberger EG, Huentelman MJ, Gottlieb S, Dobrin SE, Parod JM, et al. Recessive symptomatic focal epilepsy and mutant contactin-associated protein-like 2. N Engl J Med. 2006;354(13):1370-7. doi:10.1056/NEJMoa052773. PMID: 16571880.
84. Zweier C, de Jong EK, Zweier M, Orrico A, Ousager LB, Collins AL, et al. CNTNAP2 and NRXN1 are mutated in autosomal-recessive Pitt-Hopkins-like mental retardation and determine the level of a common synaptic protein in Drosophila. Am J Hum Genet. 2009;85(5):655-66. doi:10.1016/j.ajhg.2009.10.004. PMID: 19896112.
85. Henry OJ, Pekkola Pacheco N, Duba I, Burstedt M, Carlberg D, Delgado-Vega AM, et al. Phenotypic and transcriptomic characterization of biallelic RNU2-2 developmental and epileptic encephalopathy. Epilepsia. 2026;. doi:10.1002/epi.70473. PMID: 42678714.
86. Hayashi Y, Kajiwara K, Mizuno S, Okamoto N, Chan MY, Goto T, et al. Monoallelic and biallelic RNU4-2 variants in neurodevelopmental disorders. J Hum Genet. 2026;71(6):327-334. doi:10.1038/s10038-025-01442-x. PMID: 41408479.
87. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 126–127 (Developmental delay in the child with consanguineous parents).
88. Albobali Y, Shahwan MY, Madi MY, Al-Nuaimi S. Two Sisters From Qatar With TUSC3 Genetic Mutation: Psychiatric Considerations. Cureus. 2021;13(8):e17616. doi:10.7759/cureus.17616. PMID: 34646667. PMC8483409.
89. Marangi G, Leuzzi V, Manti F, Lattante S, Orteschi D, Pecile V, et al. TRAPPC9-related autosomal recessive intellectual disability: report of a new mutation and clinical phenotype. Eur J Hum Genet. 2013;21(2):229-32. doi:10.1038/ejhg.2012.79. PMID: 22549410. PMC3548258.
90. Sabo A, Murdock D, Dugan S, Meng Q, Gingras MC, Hu J, et al. Community-based recruitment and exome sequencing indicates high diagnostic yield in adults with intellectual disability. Mol Genet Genomic Med. 2020;8(10):e1439. doi:10.1002/mgg3.1439. PMID: 32767738. PMC7549560.
91. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 374–376 (Consanguinity).
92. Ek M, Kvarnung M, Ten Berk de Boer E, La Fleur L, Ljöstad L, Lyander A, et al. Long-read genome sequencing enhances diagnostics of pediatric neurological disorders. Genome Med. 2026;18(1):12. doi:10.1186/s13073-025-01596-5. PMID: 41514368. PMC12838436.
93. Rodan LH, Stoler J, Chen E, Geleske T. Genetic Evaluation of the Child With Intellectual Disability or Global Developmental Delay: Clinical Report. Pediatrics. 2025;156(1):e2025072219. doi:10.1542/peds.2025-072219. PMID: 40545261.
94. Moeschler JB, Shevell M. 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.
95. Michelson DJ, Shevell MI, Sherr EH, Moeschler JB, Gropman AL, Ashwal S. Evidence report: Genetic and metabolic testing on children with global developmental delay [RETIRED]: report of the Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society. Neurology. 2011;77(17):1629-35. doi:10.1212/WNL.0b013e3182345896. PMID: 21956720.
96. Rehder CW, David KL, Hirsch B, Toriello HV, Wilson CM, Kearney HM. American College of Medical Genetics and Genomics: standards and guidelines for documenting suspected consanguinity as an incidental finding of genomic testing. Genet Med. 2013;15(2):150-2. doi:10.1038/gim.2012.169. PMID: 23328890.
97. Bennett RL, Malleda NR, Byers PH, Steiner RD, Barr KM. Genetic counseling and screening of consanguineous couples and their offspring practice resource: Focused Revision. J Genet Couns. 2021;30(5):1354-1357. doi:10.1002/jgc4.1477. PMID: 34309119.
98. van Karnebeek CD, Shevell M, Zschocke J, Moeschler JB, Stockler S. The metabolic evaluation of the child with an intellectual developmental disorder: diagnostic algorithm for identification of treatable causes and new digital resource. Mol Genet Metab. 2014;111(4):428-38. doi:10.1016/j.ymgme.2014.01.011. PMID: 24518794.
99. den Hollander B, Hoytema van Konijnenburg EMM, Hewitson B, van der Meijden JC, Balfoort BM, Winter B, et al. The Metabolic Treatabolome and Inborn Errors of Metabolism Knowledgebase therapy tool: Do not miss the opportunity to treat!. J Inherit Metab Dis. 2025;48(1):e12835. doi:10.1002/jimd.12835. PMID: 39777714. PMC11707409.
100. Hamamy H, Antonarakis SE, Cavalli-Sforza LL, Temtamy S, Romeo G, Kate LP, et al. Consanguineous marriages, pearls and perils: Geneva International Consanguinity Workshop Report. Genet Med. 2011;13(9):841-7. doi:10.1097/GIM.0b013e318217477f. PMID: 21555946.
101. Treatable ID app. Available from: https://www.treatable-id.org/
102. Pan X, Qian G, Liu L, Zhang X, Yang Y, Zhang Q, et al. Genetic analysis and reporting from whole-exome sequencing data in 1052 patients with intellectual disability. J Mol Med (Berl). 2026;104(1):48. doi:10.1007/s00109-026-02652-2. PMID: 41764152.
103. European Medicines Agency. Risperdal and associated names (risperidone) — Article 30 referral: Annex I, II, III, IV (CHMP opinion 24 July 2008; Commission decision 7 October 2008). EMA/CHMP/384877/2008. Available from: https://www.ema.europa.eu/en/documents/referral/risperdal-article-30-referral-annex-i-ii-iii-iv_en.pdf
104. Risperidone tablets — prescribing information (generic, ANDA201003; text of the section “Irritability Associated with Autistic Disorder” identical to that of the reference RISPERDAL). Revised 2026-06-25. Available from: https://dailymed.nlm.nih.gov (search “risperidone tablets”)
105. Otsuka America Pharmaceutical. ABILIFY (aripiprazole) tablets — prescribing information. NDA021436. Revised 2025-01-29. DailyMed set id c040bd1d-45b7-49f2-93ea-aed7220b30ac. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=c040bd1d-45b7-49f2-93ea-aed7220b30ac
106. Takeda Pharmaceuticals America. INTUNIV (guanfacine) extended-release tablets — prescribing information. NDA022037. Revised 2025-06-23. DailyMed set id b972af81-3a37-40be-9fe1-3ddf59852528. Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=b972af81-3a37-40be-9fe1-3ddf59852528
107. Sheridan E, Wright J, Small N, Corry PC, Oddie S, Whibley C, et al. Risk factors for congenital anomaly in a multiethnic birth cohort: an analysis of the Born in Bradford study. Lancet. 2013;382(9901):1350-9. doi:10.1016/S0140-6736(13)61132-0. PMID: 23830354.
108. Bennett RL, Motulsky AG, Bittles A, Hudgins L, Uhrich S, Doyle DL, et al. Genetic Counseling and Screening of Consanguineous Couples and Their Offspring: Recommendations of the National Society of Genetic Counselors. J Genet Couns. 2002;11(2):97-119. doi:10.1023/A:1014593404915. PMID: 26141656.
109. Unique — Understanding Rare Chromosome and Gene Disorders. Available from: https://rarechromo.org/