TAF1-Related Disorders - TAF1 Syndrome (MRXS33) and X-Linked Dystonia-Parkinsonism

Two allelic disorders of the same X-chromosome gene: a neurodevelopmental disorder of prenatal or neonatal onset caused mainly by missense variants and duplications, and a neurodegenerative disease of Filipino adults caused by a founder haplotype carrying an intronic retrotransposon insertion. Alazami-Yuan syndrome, an autosomal recessive disorder of another TFIID subunit, TAF6, with a phenotype close to Cornelia de Lange syndrome, is also covered.

Literature and regulatory information updated: 31 October 2026.

0. Initial Reference Table

1. Introduction

1.1. One gene, two diseases

The TAF1 gene encodes the largest subunit of general transcription factor IID (TFIID), the complex that recognises the promoter and nucleates assembly of the RNA polymerase II pre-initiation complex, probably at almost all protein-coding genes [2, 13]. Two diseases without clinical overlap result from alterations of the same locus. TAF1 syndrome, or X-linked syndromic intellectual developmental disorder 33 (MRXS33), is a neurodevelopmental disorder of boys, present from birth, caused by missense variants, by a start-codon variant, by rare splicing variants and by duplications that include the gene [1, 2, 4]. X-linked dystonia-parkinsonism (XDP) is a neurodegenerative disease of adult men with maternal ancestry from the island of Panay, in the Philippines, caused by a founder haplotype carrying an insertion of the SVA retrotransposon in intron 32, with no change in the coding sequence [28, 29]. The distinction has practical consequences: tests that detect one disease do not detect the other, the risks for the family follow different timelines and clinical management belongs to different specialties [11].

1.2. History of TAF1 syndrome

In 2015, O'Rawe and colleagues described 14 boys from 11 families with global developmental delay, intellectual disability, facial dysmorphism, generalised hypotonia and variable neurological signs; 12 had single-nucleotide variants in TAF1 and two had duplications of about 0.42 Mb at Xq13.1 that included the gene [1]. The study used a "genotype-first" strategy, brought together independent groups through DECIPHER and a bioRxiv preprint, and built on an earlier proposal of TAF1 as a candidate intellectual disability gene, based on segregation of missense variants in two families [1]. In 2020, Cheng and colleagues added 27 families, 22 of which were analysed in detail (24 individuals), broadened the phenotype to brain malformations, seizures, hearing impairment and heart defects, and showed the difficulty of classifying maternally inherited missense variants in an X-chromosome gene [2]. These were followed by a five-generation family studied in Sweden with the same variant already described in two other families [30], patients with cerebellar hypoplasia [31] and, in 2026, two boys with infantile epileptic spasms syndrome [32]. Also in 2026, Han and colleagues identified a TAF1 start-codon variant in the original Abidi syndrome family, described in 1999 with eight affected men over three generations and mapped to Xq13.2, which brings this entity into the TAF1 spectrum [4, 19].

1.3. Nomenclature of TAF1 syndrome

The literature refers to the entity as TAF1 syndrome, TAF1 intellectual disability syndrome or TAF1/MRXS33 syndrome [1, 2, 31]. OMIM uses intellectual developmental disorder, X-linked, syndromic 33 (300966), with X-linked recessive inheritance [16]. Orphanet chose a descriptive name that includes the sacral caudal remnant (ORPHA:480907) and classifies it as a malformation syndrome of prenatal onset [3]. Abidi syndrome retains the "%" symbol in OMIM (phenotype with unknown molecular basis), last edited in July 2025, before Han's publication [4, 19]. The term "TIMES syndrome" does not appear in the titles and abstracts indexed in Europe PMC or in the core sources consulted; the term "transcriptomopathy" was used by Cheng and colleagues in relation to the clinical overlap between TAF1, biallelic TAF6 and Cornelia de Lange syndrome [2].

1.4. History of XDP

Lee and colleagues described in 1976 an unusually high frequency of torsion dystonia on Panay: of 28 Filipino patients, 23 were from the island and 19 from the province of Capiz, all men, with mean onset at 31 years [6]. The initial designation was dystonia musculorum deformans; lubag means twisting or intermittent posturing in the local dialect, and the terms wa-eg (sustained postures) and sud-sud (shuffling gait) describe other signs [5]. Kupke and colleagues demonstrated X-linked inheritance in 1990 and localised the locus to Xq21; the interval was later narrowed to Xq13.1 [5, 6]. A Filipino family with X-linked parkinsonism published by Johnston and McKusick in 1963 corresponds to the same disease [6]. In 2003, Nolte and colleagues described five disease-specific nucleotide changes (DSC1, 2, 3, 10 and 12) and a 48 bp deletion that define the haplotype, and named DYT3 a transcript system sharing 3' exons of TAF1 [5, 6]. In 2007, Makino and colleagues found the SVA insertion in intron 32 of TAF1 with reduced TAF1 expression in the caudate nucleus [28], and in 2018 Aneichyk and colleagues restricted the cause to the TAF1 locus and showed that excision of the insertion by CRISPR/Cas9 normalises transcription in patient-derived cells [29]. The International Parkinson and Movement Disorder Society nomenclature designates the disease DYT/PARK-TAF1 [7].

1.5. Definitions used

TAF1 syndrome is defined here as the early-onset neurodevelopmental disorder associated with a TAF1 variant classified as pathogenic or likely pathogenic, or with an Xq13.1 duplication that includes TAF1, bearing in mind that duplications include other genes and are associated with a more severe and progressive picture [1]. XDP is defined by the presence of the founder haplotype with the SVA insertion in a patient with the dystonia and/or parkinsonism phenotype, or in a still asymptomatic carrier [5, 24].

1.6. Alazami-Yuan syndrome (TAF6): history and definition

The eponym combines two studies published in 2015. Alazami and colleagues sequenced the exomes of 143 multiplex consanguineous families with neurological disorders, in which known genes had been excluded by autozygosity mapping and candidate gene analysis, and described 33 recessive genes with no previous disease association, among them TAF6 [33]. Yuan and colleagues studied by exome 32 Turkish patients with a clinical diagnosis of Cornelia de Lange syndrome, 7 (21.9%) of whom were children of consanguineous parents, and found in a boy without a molecular diagnosis a homozygous TAF6 variant; they combined it with the Saudi family from the Alazami series, in which three siblings were homozygous for another variant of the same gene, and proposed TAF6 as the gene for an autosomal recessive form of Cornelia de Lange syndrome or an overlapping phenotype [8]. The same authors argued that Cornelia de Lange syndrome and neighbouring entities are better understood as "transcriptomopathies", global disorders of transcription, than solely as cohesinopathies [8].

A third family, Turkish, followed in 2020; a Chinese boy of non-consanguineous parents with compound heterozygous variants, in 2022; an American adolescent with a clinical diagnosis of Cornelia de Lange syndrome since early childhood, reported in 2025 as the seventh patient with a molecular diagnosis; and a Jordanian boy with epileptic encephalopathy and complete atrioventricular block, in 2026 [10, 18, 22, 34]. OMIM records the entity under number 617126 and Orphanet under code ORPHA:694946, classified as a malformation syndrome [9, 20]. Alazami-Yuan syndrome is defined here as the clinical picture associated with biallelic TAF6 variants classified as pathogenic or likely pathogenic; the Chinese patient, in whom one of the alleles is a variant of uncertain significance, is described with that caveat [18]. The entity shares with TAF1 syndrome the general mechanism, TFIID dysfunction, and the main differential diagnosis, Cornelia de Lange syndrome [2, 8, 10].

2. Molecular Genetics and Pathophysiology

2.1. Structure and function of TAF1

TFIID is formed by the TATA box-binding protein (TBP) and 12 to 14 TBP-associated factors (TAFs) [1]. TAF1 (TAFII250) has 1893 amino acids in the numbering used by the clinical literature and several activities: the N-terminal TAND domain binds TBP and inhibits its binding to DNA; a central region (DUF3591, with the histone acetyltransferase domain) binds TAF7; a winged-helix domain and a zinc knuckle domain bind DNA; two tandem C-terminal bromodomains recognise acetylated, butyrylated and crotonylated histones; and N- and C-terminal kinase domains phosphorylate the RAP74 protein (a TFIIF subunit) [2, 4, 13]. The acetyltransferase activity is inhibited by TAF7 binding [1]. The gene was first cloned as CCG1 (cell cycle gene 1), because it corrected a mutant hamster line arrested in G1 [13]. The gene has 38 conserved exons and five downstream exons (d1–d5), of recent evolutionary origin, that take part in the DYT3 transcript system; there is also a neuronal isoform (NTAF1, TA14-391) formed by inclusion of microexon 34 [2, 13].

TAF1 is among the most constrained genes in the genome: it ranks 53rd among 1003 genes with maximal constraint, has a pLI of 1.0 and, in gnomAD v2.1.1, a missense Z-score of 5.49, with 326 observed missense variants against 748 expected (observed/expected ratio 0.44) [1, 2, 30]. No hemizygous or homozygous loss-of-function variants in the canonical isoform are known in population databases [2]. Zebrafish with a complete taf1 knockout die at the embryonic stage and show dysregulation of neurodevelopmental genes [30], and male mice with a ubiquitous Taf1 knockout are also embryonic lethal; heterozygous females show no structural brain changes, which suggests X inactivation skewed against the null allele, but they gain weight and move less [35]. These data support the hypothesis that human variants compatible with life are hypomorphic.

2.2. Mechanism in TAF1 syndrome

In the 2015 series, four of the eight missense variants affected domains important for TAF7 binding: p.Cys807Arg lies at the centre of the triple barrel formed by TAF1 and TAF7; p.Asp976His alters the glycine-rich loop that contacts an arginine-rich motif of TAF7; p.Arg1246Trp and p.Arg1190Cys lie in the RAP74 interaction domain (all in legacy numbering) [1]. The variant published as p.Arg1431His, p.(Arg1432His) in MANE Select, affects a surface helix of the first bromodomain and may destabilise the acetyl-lysine binding pocket [1]. In the 2020 series, structural modelling predicted destabilisation for p.(Arg727Pro) (ΔΔG 1.56) and loss of contact at the TAF1-TAF7 interface for p.(Ser985Phe); in the bromodomain, p.(Asn1481Ile) hits the asparagine that anchors the acetyl group, and p.(Lys1576Glu) introduces a repulsive charge at the interface between the two bromodomains [2]. No disease variant has been described in the zinc knuckle domain (amino acids 1261–1300) [2].

Functional studies are limited. Blood RNA-seq from the family with p.Ile1337Thr identified 213 differentially expressed genes (179 down and 34 up), with enrichment for targets of E-box proteins (corrected p 0.0052) and for Parkinson, Alzheimer and Huntington disease pathways; the authors regard these results as preliminary, from a single family and potentially confounded by age, sex and cell composition [1]. In zebrafish, morpholino knockdown of taf1 reduced the optic tectum area by about 10% (133 embryos against 134 controls; p<0.0001), an effect rescued by normal human mRNA [1]. The p.Ser1600Gly variant (legacy numbering) altered, in SH-SY5Y cells, the expression of neuronal ion channels, including CACNA1G, differentiation and proliferation [31]. The p.(Thr79Asn) and p.(Asp1591Asn) variants (legacy numbering) reduced protein levels in HEK293T cells, and Taf1 knockdown in mouse cortical neurons reduced neurite and dendrite outgrowth and increased excitability [32]. Lymphoblasts from patients with p.(Met1?) and p.(Gln1429Pro) showed reduced TAF1-dependent phosphorylation of RAP74 [4].

Xq13.1 duplications including TAF1 (0.423 Mb and 0.42 Mb in the original series) were associated with severe, progressive neurodegeneration and less overlap with the phenotype of point variants; one of them coexisted with a 0.63 Mb deletion at 17q21.31 that included KANSL1, and neither family had undergone exome sequencing, so the contribution of TAF1 and of neighbouring genes is unresolved [1]. More distal Xq13.2–q13.3 duplications that include RLIM cause another syndrome with mild intellectual disability and a distinctive face [36].

2.3. Mechanism in XDP

The XDP haplotype includes seven changes inherited en bloc: the antisense SVA insertion in intron 32 and six non-coding changes, among them DSC1 (in an Alu element), DSC2 (in a LINE2) and DSC10 (in intron 32), DSC12 (in intron 18), the 48 bp deletion (in intron 2) and DSC3, the only one in a transcribed exon (exon d4 of the DYT3 system), designated c.94C>T (p.Arg32Cys) in that transcript [5, 14]. These changes were present in 46 of 46 patients and absent in 178 Filipino controls (208 X chromosomes), in a region of unusually low heterozygosity [6]. The insertion is 2627 bp long, about 70% GC and has more than 150 CpG dinucleotides, and it is hypermethylated in the caudate of patients [13, 28]. In brain tissue, expression of TAF1 and of the dopamine D2 receptor gene (DRD2) was reduced in the caudate nucleus [28].

Integrated genome and transcriptome assembly showed a decrease in the canonical transcript (cTAF1) and alternative splicing with retention of intron 32 next to the SVA, anticorrelated with total TAF1 expression, in several neuronal lineages derived from pluripotent stem cells; excision of the SVA corrected the transcriptional signature [29]. In 2026, Capponi and colleagues showed that the SVA induces premature cleavage and polyadenylation within the locus and that BRD4 depletion corrects the molecular signature by a mechanism independent of its canonical coactivator function [37]. In a reporter model, the length of the hexanucleotide modulated transcriptional activity, and the tract has motifs with the potential to form G-quadruplexes [14]. Neuropathology shows mosaic neuronal loss and astrocytosis in the caudate and putamen; the dystonic phase has been associated with loss of striosomes with relative sparing of the matrix, and the parkinsonian phase with involvement of the matrix [5, 6]. A series of 18 brains published in 2026 described degeneration in a dorsoventral gradient, with greater vulnerability of the matrix than previously thought, more related to disease duration than to repeat length [38].

2.4. Hexanucleotide repeat, modifiers and penetrance in XDP

The (CCCTCT)n tract within the SVA (written (AGAGGG)n on the other strand) ranges from 35 to 52 repeats in patients and correlates inversely with age at onset (n=140) [14]. In a study of 355 patients, repeat number correlated inversely with age at onset (n=295) and with TAF1 expression in blood (n=31), and positively with severity (n=28) and with cognitive dysfunction (n=15); age at onset, and not repeat number, determined whether the disease began with dystonia or with parkinsonism; the repeat was unstable between generations, with a tendency to expand, and showed somatic mosaicism in the brain [39]. The repeat explains about 50% of the variability in age at onset; a genome-wide association study identified three loci, in or near MSH3 and PMS2, that explain 25.6% of the remaining variance (13.0% of the total), with protective alleles that delay onset by about seven years [40]. Nanopore sequencing of 202 patients found divergent mosaic interruptions at the start of the tract; together, the repeat and MSH3 and PMS2 variants explain about 65% of the variance in age at onset [41]. In 2026, Mejia Maza and colleagues showed that MSH3 modifies somatic expansion and contraction of the repeat in blood and brain, in parallel with age at onset [42]. In a cohort of 27 patients from Mindanao, the repeat explained about 68% of the variability in age at onset and a published predictive model did not differ from the observed age [43]. Anticipation was not described in the 2018 GeneReviews [5], despite the tendency for repeat expansion on transmission [39].

Kupke and colleagues observed that 64 of 120 sons of carrier mothers (52%) were affected, a proportion close to that expected for complete penetrance in men [6]. Women may manifest the disease through homozygosity, skewed X inactivation or aneuploidy [44].

2.5. Variant spectrum and HGVS nomenclature

The MANE Select transcript is NM_004606.5 (NP_004597.3). The clinical literature from 2015 to 2026 mostly uses NM_004606.3/.4 (NP_004597.2), whose start codon lies 60 nucleotides upstream: the MANE c. position is 60 units lower and the p. position is 20 units lower, a relationship confirmed variant by variant in VariantValidator [2, 12]. Hence the p.(Arg1190Cys) variant described in three families corresponds to NM_004606.5:c.3508C>T p.(Arg1170Cys). Han's 2026 paper already uses MANE numbering [4, 12]. Laboratory reports and databases may present the same variant with different numbers, and comparison should be made by genomic coordinates.

In the 2020 series, the ACMG/AMP (American College of Medical Genetics and Genomics/Association for Molecular Pathology) classifications assigned by the authors were: likely pathogenic for 11 variants (10 de novo, one maternal), uncertain significance (VUS) for 11 maternal variants and likely benign for six, one of them reclassified when a healthy maternal uncle was found to carry it [2]. None of the 22 variants analysed was present in men in gnomAD, but the p.(Asp1591Asn) variant (legacy numbering) occurs in a hemizygous man in gnomAD (frequency 0.0011%) [2, 32]. In ClinVar, the search TAF1[gene] returned, on 31 October 2026, 1200 records, 353 of them classified as pathogenic or likely pathogenic, including large structural variants spanning the gene; many coding variants of uncertain significance are linked by laboratories to the XDP condition, which is not caused by coding variants [45]. The SVA insertion is recorded as OMIM allele 313650.0001 (ClinVar RCV000010537) [13].

2.6. Genotype-phenotype correlations

In TAF1 syndrome, variants are distributed across the TBP-binding domain, the TAF7 interaction region, the RAP74 interaction domain, the bromodomains and the start codon, with no established correlation between domain and phenotype [2, 4, 32]. Cluster analysis using 51 Human Phenotype Ontology (HPO) terms separated a "classic" group, with global delay, speech delay, hypotonia, intellectual disability, feeding difficulties, thin upper lip, strabismus and sacral dimple, from groups with fewer signs; a variant that later proved benign fell in the most distant group [2]. Duplications are associated with a more severe and progressive picture [1]. In XDP, all patients carry the same haplotype, and variability depends on the repeat and on modifiers of the mismatch repair pathway (MSH3, PMS2) [5, 40].

2.7. Allelic disorders and related entities

XDP and TAF1 syndrome are allelic disorders, and Abidi syndrome falls within the TAF1 syndrome spectrum [4, 5]. Other TFIID subunits cause neurodevelopmental disorders: biallelic TAF2, TAF6 and TAF13 cause autosomal recessive intellectual disability with microcephaly, and biallelic TAF6 causes Alazami-Yuan syndrome, with a picture overlapping that of Cornelia de Lange syndrome (subsections 1.6, 2.8 and 3.13); CAG expansion in TBP causes spinocerebellar ataxia type 17 [2].

2.8. TAF6 and Alazami-Yuan syndrome: gene, mechanism and variants

TAF6 is located at 7q22.1 (OMIM gene 602955; HGNC:11540) and encodes a TFIID subunit with an N-terminal histone-fold domain, similar to histone H4, and a C-terminal HEAT repeat domain [15, 46]. In the TFIID core, a TAF6 homodimer links the modules of the complex into a rigid structure, so the protein has a structural as well as a regulatory role; TAF6 also belongs to the core TAF module shared by TFIID and the SAGA coactivator complex [46, 47]. In yeast, substitutions in the histone-fold domain and in the HEAT domain independently abolished promoter occupancy by TFIID and SAGA and transcriptional activation dependent on the Gcn4 and Gal4 activators [46]. The MANE Select transcript is NM_139315.3 (NP_647476.1, alpha isoform of 677 amino acids). The clinical literature uses NM_005641.3, which encodes the same 677-amino-acid isoform, and all published variants have identical numbering on both transcripts, verified in VariantValidator [12]. NM_001190415.1, cited in the 2025 description, shifts the numbering by 111 nucleotides: c.212T>C corresponds there to c.323T>C, and c.1863dup to c.1974dup [10].

The first two variants described, p.(Arg46Cys) and p.(Ile71Thr), lie in the histone-fold domain. In Drosophila S2 cells depleted of endogenous TAF6, the human N-terminal region with p.(Arg46Cys) partially lost interaction with most of the components tested (TAF1, TAF4, TAF9) and the region with p.(Ile71Thr) had reduced binding to all of them; both no longer bound the TATA box-binding protein (TBP), which suggests instability of the complex [8]. In a complementation assay in a temperature-sensitive yeast strain, growth decreased only slightly, which is consistent with hypomorphic alleles and with patient survival [8]. The link between a ubiquitously expressed general transcription factor and a restricted morphological phenotype is explained, according to the authors, by the convergence of cohesin, chromatin regulators and TFIID on the regulation of transcription by RNA polymerase II; TAF6 depletion in Drosophila produced a wing-restricted phenotype, suggestive of a morphogenetic function [8]. The last-exon duplication described in 2025 did not cause mRNA decay or alter splicing in the blood transcriptome; it replaces the last 56 codons with a new sequence whose stop codon lies in the 3' untranslated region, which no longer exists in the canonical form, and the presumed effect is at the protein level [10]. There is no biochemical assay of this dysfunction applicable to diagnosis [15].

With eight genotyped patients, there is no established genotype-phenotype correlation. p.(Ile71Thr) is the only recurrent variant, in families of Saudi, Turkish and American origin; it has an allele frequency of 1 in 13,000 in the Exome Sequencing Project (ESP) and was absent from 516 Saudi exomes [8, 10]. The only patient homozygous for an early frameshift variant, the Jordanian boy, had the most severe picture, with epileptic encephalopathy and complete atrioventricular block [22]; the Chinese patient, with normal intelligence, has as his second allele a relatively frequent missense variant of uncertain significance, which makes attribution of the phenotype provisional [18]. Support for the gene-disease association remains scant: ClinGen has not curated TAF6 and the two GenCC submissions diverge (strong in PanelApp Australia, limited at Labcorp/Invitae) [15].

In the wider family of TFIID disorders, besides TAF1 and TAF6, biallelic TAF2 and TAF13 variants cause autosomal recessive intellectual disability with microcephaly [2], and de novo predicted loss-of-function TAF4 variants have been described in eight individuals with an autosomal dominant neurodevelopmental disorder, with intellectual disability, behavioural changes and facial dysmorphism [49].

3. Clinical Spectrum and Manifestations by System

3.1. TAF1 syndrome: reference cohorts

The founding series included 14 boys aged 1 to 22 years, most of European ancestry, with families from Ecuador, Colombia, Spain, the United Kingdom, Albania and Greece [1]. The 2020 allelic series brought together 27 families from 15 institutions in seven countries, four of them identified in the British Deciphering Developmental Disorders study; frequencies were calculated over 24 individuals from 22 families, including VUS carriers, so the authors acknowledge errors in both directions [2]. The five-generation family studied in Sweden added six men aged 6 to 65 years, the longest published observation [30]. The frequencies from the two main series are shown in the following table (denominators with known information).

Sources: [1, 2, 16]. VSD, ventricular septal defect; ASD, atrial septal defect.

3.2. Neurodevelopment and behaviour

The typical presentation is hypotonia and developmental delay in the first year of life, followed by a diagnosis of intellectual disability, autism spectrum disorder or both; language delay is the rule [2]. The severity of intellectual disability was described as generally severe in the initial series and variable in the allelic series [1, 2]. Self-injury (3 of 24) and stereotyped hand movements (2 of 24) were recorded [2]. A 5-year-old boy with p.Ser1600Gly was sociable, with normal hearing, hand flapping and speech apraxia [31]. In the family studied in Sweden, all six men had intellectual disability and speech delay, with no autistic behaviours recorded in the two assessed for this item [30].

3.3. Nervous system: neuroimaging, epilepsy and motor function

Hypoplasia of the corpus callosum was the most frequent imaging finding in the initial series; the 2020 series added ventriculomegaly and other brain malformations [1, 2]. Cerebellar hypoplasia was described in one boy, with loss of cerebellar volume at 21 months, enlargement of the fourth ventricle, a thin corpus callosum and delayed myelination, without later progression [31]. Seizures occurred in 25% of the 24 patients in the 2020 series, with focal and generalised tonic-clonic, tonic, myoclonic and atonic seizures and absences [2, 32]. In 2026, two boys with infantile epileptic spasms syndrome and hypsarrhythmia were described, with onset at 7 and 3 months, one of whom became seizure-free and the other drug-resistant with severe intellectual disability [32]. A 2018 Chinese West syndrome cohort had already described maternal TAF1 missense variants, including one present in a hemizygous individual in gnomAD, and overlap of patients between the two Chinese studies is not stated [2, 32]. Motor signs include spasticity (spastic diplegia; lower-limb spasticity treated with botulinum toxin in one patient), dystonic movements, tremor, possible ataxia, postural instability and gait abnormality in about one third [1, 2, 31].

3.4. Growth, feeding and gastrointestinal system

Intrauterine growth restriction and postnatal growth delay are frequent, and oropharyngeal dysphagia was recorded in nine of ten patients with information in the initial series; feeding difficulties affected 62.5% and gastro-oesophageal reflux 20.8% in the 2020 series [1, 2]. The family studied in Sweden had no growth delay, and the men had large hands and feet [30].

3.5. Dysmorphism, spine and skeleton

The facial phenotype described includes prominent supraorbital ridges, downslanting palpebral fissures, deep-set eyes, sagging cheeks, long philtrum, low-set prominent ears, long face, high palate, thin upper lip, pointed chin and anteverted nares with a bulbous nasal tip [1, 16]. In the 2020 series, facial features were more variable and the authors concluded that the gestalt is not specific enough to recognise the syndrome without testing [2]. An abnormal gluteal crease with sacral caudal remnant and sacral dimple was an almost constant sign in the initial series (12 of 13), with no spinal cord defect on magnetic resonance imaging in two patients, and was present in about one third in the allelic series [1, 2]. Absent ribs, scoliosis, kyphosis, pectus, hip dysplasia, flat feet, toe syndactyly, brachydactyly and preaxial polydactyly were recorded in minorities [2, 30]. Two patients received a clinical diagnosis of Cornelia de Lange syndrome before the molecular diagnosis, one with a score of 10 on the consensus criteria for the non-classic form [2].

3.6. Sense organs, heart and genitourinary system

Hearing impairment (sensorineural or mixed), chronic otitis media and strabismus are frequent; nasolacrimal duct obstruction, myopia, supernumerary nipples and aplasia cutis congenita have also been described [1, 2, 31]. The congenital heart defects were ventricular septal defect (25%), atrial septal defect, coarctation of the aorta and tetralogy of Fallot; cryptorchidism and hypospadias were the genitourinary anomalies recorded [2].

3.7. Xq13.1 duplications and course in adulthood

The two boys with duplications that included TAF1 had severe, progressive neurodegeneration, and one of them died at 8 years of cardiopulmonary failure attributed to infection [1]. Three patients with point variants also had progressive symptoms [1]. In the family studied in Sweden, the men aged 52, 54 and 65 years still had intellectual disability, with no description of neurodegeneration; three men in the family had died at 6, 25 and 65 years, and the clinical table of the paper does not give the cause of death [30]. Natural history data in adults are therefore scarce.

3.8. Female carriers and affected girls

Carrier mothers are usually asymptomatic and show X inactivation skewed in favour of the normal allele (99:1, 97:3, 100:0, 92:8, 90:10 or 96:4 in the families studied) [1, 2, 31]. In the family studied in Sweden, five carriers had 100% inactivation of the mutated allele and three non-carriers had a random pattern [30]. Two girls with de novo variants had a TAF1 syndrome phenotype, and the one tested had X inactivation >90:10 [2].

3.9. Abidi syndrome

The original Abidi syndrome family had eight men with intellectual disability over three generations and variable, mild somatic signs: short stature in 4 of 8, small head circumference in 5 of 8, sloping forehead in 5 of 8, hearing impairment in 3 of 8, cupped ears in 2 of 8 and small testes in 4 of 6 [19]. A second family with three affected men and the p.(Gln1429Pro) variant also had intellectual disability with variable somatic signs [4].

3.10. XDP: presentation and course in men

Onset in men occurs at a mean of 39 years (12–64) according to GeneReviews; other cohorts gave means of 31 years (28 patients, 1976), 34.8±8.1 years (42 patients from 21 families, 1990) and 38.6 years (12–56) [5, 6]. Although dystonia was described as the initial sign in older series, longitudinal follow-up of confirmed carriers showed that parkinsonism is almost always the first sign, with early, sometimes asymmetric, impairment of rapid alternating movements [5]. In a series of 42 patients, the first manifestation was located in the head and neck in 39%, the lower limbs in 33%, the upper limbs in 24% and the trunk in 9% [6]. Parkinsonism includes rest tremor, bradykinesia, rigidity, postural instability and shuffling gait with falls; some patients have pure parkinsonism for years, asymmetric and levodopa-responsive, and are mistaken for Parkinson disease [5].

Dystonia begins focally, mainly in the jaw, neck, trunk and eyes, and less often in the limbs, tongue, pharynx and larynx [5]. The most characteristic form is jaw-opening dystonia that progresses to the neck, with retrocollis more frequent than torticollis, sometimes with neck extension of more than 90 degrees and trunk hyperextension; dystonic head tremor and sensory tricks are frequent; foot dystonia is rare, unlike DYT-TOR1A [5]. Pharyngeal dystonia, which accompanies oromandibular-lingual dystonia, causes weight loss, aspiration pneumonia and early death, and laryngeal dystonia with stridor may cause sudden death [5]. The interval between onset of dystonia and generalisation was 1 to 23 years, with a mean of 3.8 years; a 2010 review cited in a 2026 scoping review recorded generalisation within five years in 84% [5, 50]. Other signs are essential-type tremor, a slow 1 to 3 Hz tremor resembling myorhythmia, distal chorea with athetosis and action myoclonus of cortical origin [5].

Forms with pure parkinsonism and little dystonia have a better prognosis, with mildly disabling, slowly progressive symptoms; those combining parkinsonism with oro-bucco-lingual and cervical dystonia in the first one to two years generalise between the second and fifth year, lead to confinement to bed and premature death from aspiration pneumonia, laryngeal stridor or infections [5]. Global cognition is relatively preserved, with frontal executive deficits; depression is frequent, and pathological gambling and other impulse control disorders have been described [5]. A 2023 review emphasises the frequency of cognitive and mood symptoms [51], and cognitive dysfunction correlated with repeat number [39]. Scoping reviews describe stigma, loss of identity, caregiver burden, financial hardship and late diagnosis [50].

3.11. XDP in women

The Filipino registry of January 2010 included 505 cases, 500 men and 5 women; the authors considered that both sexes followed a similar progressive course [17]. GeneReviews describes symptomatic women with non-progressive focal dystonia, chorea (sometimes hemichorea), focal tremor or mild parkinsonism, and rarely generalised dystonia [5]. In a study of 17 carriers from five families, nine (53%) were symptomatic or had an abnormal examination; of the eight symptomatic women, seven were heterozygous and one homozygous for DSC3, with mean onset at 52 years (26–75), parkinsonism or tremor in four, chorea in three and dystonia in one [6]. In 2025 the first deep brain stimulation in a woman with manifest XDP was published [52].

3.12. XDP: imaging and biomarkers

Computed tomography and magnetic resonance imaging were normal in 20 symptomatic patients in a clinical series, with mild cerebral atrophy in some and caudate atrophy in advanced disease; diffusion imaging showed involvement of the white matter and putamen; fluorodopa PET showed pre- and postsynaptic involvement, and fluorodeoxyglucose PET putaminal changes in early disease with normal MRI; dopamine transporter SPECT may resemble that of Parkinson disease, even in early pure parkinsonism [5]. Transcranial sonography showed hyperechogenicity of the substantia nigra in 79% and of the lenticular nuclei in 81% of patients [5]. In a 2026 study with 138 participants (61 patients, 19 non-manifesting carriers and 58 controls), lentiform nucleus hyperechogenicity was found in 81%, 47% and 20%, respectively, and was associated with an earlier estimated age at onset and more repeats in carriers [53]. Olfaction is reduced: on the culturally adapted 25-item odour identification test (ccUPSIT), 20 patients scored 18±3.19 points against 20.5±3.02 in 20 controls (p=0.003), with no correlation with phenotype, severity or duration [5]. Nerve conduction studies, evoked potentials, electroencephalography and the blink reflex were normal in ten patients [5].

3.13. Alazami-Yuan syndrome: published patients and clinical spectrum

By October 2026, eight genotyped patients had been published in six families, and three untested siblings with a picture described as similar died in infancy or adolescence [8, 10, 18, 22, 34]. All data come from case reports and one family, with no cohort or systematic follow-up, and the denominators of the frequencies vary because not every feature was assessed in each patient; the 2025 synthesis grouped the three Saudi siblings in a single column for lack of individual data [10].

In the comparison compiled in 2025 with the first seven patients, arched eyebrows (7/7) and long eyelashes (5/5) were the only universal features; they were followed by prominent nasal bridge (6/7), developmental delay and intellectual disability (6/7), synophrys, low hairline and hypertrichosis (5/7 each), thin upper lip (5/6), microcephaly (3/4), prenatal (2/3) and postnatal (2/4) growth restriction, scoliosis (2/2) and deafness (1/3); consanguinity was present in 5 of 7 [10]. None of the patients assessed had epilepsy (0/3), congenital heart disease (0/3), congenital diaphragmatic hernia (0/4) or oligodactyly (0/6) [10]. The Jordanian patient, described after that synthesis, added to the spectrum developmental and epileptic encephalopathy, complete atrioventricular block and bilateral congenital cataracts, previously seen only, unilaterally, in the American patient [10, 22].

The resemblance to Cornelia de Lange syndrome is real but incomplete. A prominent nasal bridge and a low hairline were the features of TAF6 patients that the 2025 comparison did not find in the Cornelia de Lange syndrome reference series; conversely, no TAF6 patient had the short nose with concave bridge, the hand anomalies or the diaphragmatic hernia, which are more specific to that syndrome [10]. Using the clinical score of the international consensus, all TAF6 patients scored between 4 and 10 points, enough to justify molecular testing for Cornelia de Lange syndrome, but none reached the 11 points of the classic form, and only the Turkish boy CdLS-4, with 10 points and three cardinal signs, met the criteria for the non-classic form [10, 25].

Natural history is poorly known. The American patient spent two weeks in neonatal intensive care because of feeding difficulty, needed a nasogastric tube at 6 months after aspiration pneumonia and a gastrostomy with Nissen fundoplication at 11 months; at 17 years she fed mainly orally, with intermittent laryngeal penetration of thin liquids without tracheal aspiration on videofluoroscopy [10]. From the age of 9 onwards she had three to four episodes a year of painful interphalangeal arthritis resolving spontaneously within 4 days, an erythrocyte sedimentation rate of 44 mm/h and a diagnosis of palindromic rheumatism at 16 years; the authors do not know whether the autoimmunity is part of the syndrome or a comorbidity [10]. The Chinese boy, with normal intelligence, was assessed for testicular enlargement and rapid growth at 11 years, with bone age advanced by 2 years, normal pituitary MRI, gonadotrophins and testosterone at early pubertal levels, vitamin D deficiency and fasting hyperinsulinaemia (23.1 mU/L) [18]. The early mortality of the untested siblings in the Turkish and Jordanian families suggests that the spectrum includes forms lethal in childhood, but the cause of death was not documented [8, 22].

4. Clinical Recognition and Warning Signs

4.1. TAF1 syndrome

In a male newborn or infant, the combination of hypotonia, feeding difficulties or oropharyngeal dysphagia, growth restriction, microcephaly and an abnormal gluteal crease with sacral caudal remnant or sacral dimple should lead to inclusion of TAF1 in molecular diagnosis [1, 3]. In the child, the pattern is global delay with marked language delay, progressing to intellectual disability and/or autism, with hearing impairment, recurrent otitis, strabismus, hypoplasia of the corpus callosum or cerebellar hypoplasia and, in about a quarter, epilepsy [2, 31]. A family history of affected boys through the maternal line, with healthy women, points to X-linked disease; the absence of a family history does not exclude it, because about half of the variants are de novo [1, 2]. The face is not specific and the syndrome may be clinically confused with Cornelia de Lange syndrome [2].

Signs requiring urgent assessment are epileptic spasms (clusters of contractions, with developmental arrest or regression), which require electroencephalography to look for hypsarrhythmia [32], dysphagia with aspiration, weight loss and recurrent respiratory infections [1], and neurological regression, which should prompt a search for an Xq13.1 duplication or a second diagnosis [1, 2]. In the 2020 series, four patients had a second molecular diagnosis (1p36 deletion syndrome, de novo variants in RAC1 and HNRNPU, and mucopolysaccharidosis IIIA) [2].

4.2. XDP

XDP should be considered in men with adult-onset dystonia, especially jaw, cervical (retrocollis) or trunk dystonia, with parkinsonism, and in men with asymmetric, levodopa-responsive parkinsonism, when there is Filipino maternal ancestry, particularly from Panay [5, 54]. Ethnic origin is part of the history in the approach to any dystonia: early-onset autosomal dominant torsion dystonia is 5 to 10 times more frequent in Ashkenazi Jews, and X-linked dystonia-parkinsonism occurs in Filipinos [54]. About 94% of patients have a family history [5], and patients have been described outside Panay, in Mindanao, as well as Filipino emigrant men treated in Germany and Ireland [43, 55, 56]. In women, chorea, focal tremor or mild parkinsonism in a family with XDP should raise suspicion of the disease [5].

Warning signs in XDP are oromandibular-lingual or pharyngeal dystonia with dysphagia and weight loss, stridor (laryngeal dystonia, with risk of sudden death), rapid generalisation in the first years and aspiration pneumonia [5]. A patient with rapidly worsening generalised dystonia and nasogastric tube feeding was referred for deep brain stimulation 12 months after presentation [56].

4.3. Newborn screening

There is no newborn screening for either entity in the sources consulted.

4.4. Alazami-Yuan syndrome

Suspicion arises in a child with a facial gestalt close to that of Cornelia de Lange syndrome (arched eyebrows, long eyelashes, synophrys, thin upper lip, hypertrichosis), developmental delay and growth restriction, without a variant in the cohesin genes, and especially when the parents are consanguineous or there are affected siblings of both sexes, which points to autosomal recessive inheritance [8, 10]. The absence of limb reduction defects, oligodactyly and diaphragmatic hernia, with a prominent nasal bridge instead of a short nose with concave bridge, shifts the hypothesis from the classic form to an overlapping entity [10]. A long-standing clinical diagnosis of Cornelia de Lange syndrome without molecular confirmation justifies reassessment at any age: the American patient had that diagnosis from 6 months of age, with negative NIPBL and SMC1A tests, and only received the TAF6 diagnosis at 16 years [10]. Congenital cataract, sensorineural hearing loss, reflux with aspiration and epileptic seizures are warning signs requiring targeted assessment; bradycardia or syncope requires an electrocardiogram, because one patient had complete atrioventricular block at 18 months [10, 22]. There is no newborn screening for this disease.

5. Diagnosis: Criteria, Tests and Algorithms

5.1. TAF1 syndrome: criteria and strategy

No formal clinical criteria have been published. Diagnosis rests on identification of a pathogenic or likely pathogenic TAF1 variant (or a duplication including it) in a patient with a compatible phenotype [2, 3]. The usual sequence starts with clinical assessment and chromosomal microarray (which detects Xq13.1 duplications), followed by exome or genome sequencing, preferably as a trio; fragile X testing and karyotype were normal in most patients in the 2020 series who had already undergone them [1, 2, 11]. Brain magnetic resonance imaging, electroencephalography, audiological assessment and echocardiography are guided by the phenotype [11].

5.2. TAF1 syndrome: variant interpretation and limitations

Classification of maternally inherited missense variants is the main difficulty. By March 2019, one commercial laboratory had sequenced 18,256 probands with neurodevelopmental delay and reported 71 with a TAF1 variant (58 VUS, 12 likely pathogenic, one likely benign, none pathogenic); another laboratory, with about 8100 probands, reported 15 variants (one likely pathogenic and 14 VUS) [2]. These numbers measure reported variants and not the frequency of the syndrome. The elements that help classify a variant are de novo occurrence, segregation with the affected men in the family and absence in healthy men (a healthy maternal uncle carrying a variant led to its reclassification as likely benign), absence in hemizygous individuals in gnomAD, skewed X inactivation in the carrier mother (a random pattern weighs against pathogenicity) and concordance with the core phenotype [2]. The presence of an alternative diagnosis should be considered a criterion against pathogenicity [2]. The proportion of variants in the gene in databases is not equivalent to the clinical sensitivity of the test.

Maternal X inactivation was used as a reanalysis tool in an Italian cohort: skewed patterns (>90%) were found in 16 of 186 mothers of boys with neurodevelopmental disorders (8.6%) and in 12 of 90 affected girls (13.3%), against 3.6% in the general population, and exome reanalysis solved 7 of 28 cases with skewing (25%), one of them with a TAF1 variant [23]. In patients with a variant of uncertain significance, RNA expression studies (as for the c.3708A>G splicing variant in legacy numbering) and RAP74 phosphorylation studies in lymphoblasts have been used in research, and are not validated as clinical tests [1, 4].

The laboratory report should state the gene, the transcript with version, the HGVS nomenclature, hemizygosity or heterozygosity, the ACMG/AMP class and the parental origin when studied; given the 60-nucleotide difference between NM_004606.5 and earlier versions, genomic coordinates avoid errors of correspondence with the literature [12]. A negative exome does not exclude deep intronic variants or undetected rearrangements.

5.3. XDP: criteria and tests

The suspected clinical diagnosis rests on adult-onset dystonia and/or parkinsonism, a family history consistent with X-linked inheritance and maternal ancestry from Panay; molecular testing is required when there is no family history, when symptoms are very early or when the phenotype is pure parkinsonism, pure tremor or chorea without dystonia, and in symptomatic women [5]. The 2018 GeneReviews describes targeted analysis of the DSC3 marker variant, c.94C>T (p.Arg32Cys) in the DYT3 transcript, detected in 100% of patients of Filipino ancestry, as an indirect test; variants at the locus not associated with XDP cannot be interpreted as causal [5]. Later literature regards detection of the SVA insertion as confirmatory [24, 29]. The sequence and the repeat are technically difficult to analyse; the tract can be characterised by amplicon PCR with nanopore sequencing, repeat-primed PCR and Cas9 enrichment [5, 41].

The most important practical point is that XDP is not caused by a coding variant: exome, dystonia gene panels based on sequencing of the coding region and copy number analysis do not detect the haplotype, and a result "negative for TAF1" does not exclude it [11]. The ERN-RND recommendation to perform genetic testing before any deep brain stimulation for dystonia applies here through a targeted test [7, 11]. Conversely, many TAF1 coding variants of uncertain significance appear in ClinVar linked to the XDP condition, which has no mechanistic basis [45]. Olfaction, dopamine transporter SPECT and transcranial sonography support the diagnosis, and GeneReviews accepts the use of the olfactory test as support in rural areas without access to genetic testing, without it replacing molecular confirmation [5, 53].

5.4. Practical algorithm

5.5. Prenatal and preimplantation diagnosis

In TAF1 syndrome, with the familial variant known and classified as pathogenic or likely pathogenic, prenatal diagnosis and preimplantation genetic testing are technically possible; they should not be based on VUS [2]. In XDP, prenatal testing and preimplantation diagnosis are possible once the haplotype has been identified in an affected relative [5]. The GeneReviews caveat about the indirect nature of the DSC3 marker has lost weight with identification of the SVA mechanism, but targeted testing still requires a laboratory with experience of this locus [5, 29].

5.6. Alazami-Yuan syndrome: molecular diagnosis and limitations

There are no specific clinical criteria; diagnosis is molecular and requires two TAF6 variants in trans, both classified as pathogenic or likely pathogenic [10, 18]. The clinical score of the Cornelia de Lange syndrome consensus serves as a starting point: 4 or more points with at least one cardinal sign justify molecular testing, and all TAF6 patients reached that threshold [10, 25]. The first-line test is the exome, preferably as a trio, or a broad panel including TAF6, as well as the genes for Cornelia de Lange syndrome and overlapping disorders: the American patient was diagnosed with a 2775-gene panel for autism spectrum disorder, developmental delay and intellectual disability analysed as a trio, after karyotype, microarray, NIPBL and SMC1A sequencing and testing for NIPBL deletions and duplications without a result [10]. In consanguineous families, mapping of regions of absence of heterozygosity from the exome or from a SNP microarray guides filtering: in the two 2015 families, the TAF6 variant lay within one of these regions [8].

The limitations are those of any exome: deep intronic variants and structural variation may be missed, and a negative result does not exclude the disease. In compound heterozygosity, parental testing is essential to confirm that the variants are in trans [10, 18]. Novel or relatively frequent missense variants remain variants of uncertain significance until segregation or functional data are available, such as the c.76A>T in the Chinese patient [18]; blood transcriptome analysis helped to interpret the effect of c.1863dup by excluding mRNA decay and altered splicing [10]. The report should state the gene, the MANE transcript with version (NM_139315.3), the HGVS nomenclature, the zygosity, the confirmed phase and the ACMG/AMP class [12]. In a patient with a classic Cornelia de Lange syndrome phenotype and a negative blood study, the most likely hypothesis remains NIPBL mosaicism, present in 15% to 20% of patients with the classic form and undetectable in lymphocytes; the consensus recommends studying uncultured fibroblasts or another tissue before looking for recessive causes [25]. Prenatal diagnosis and preimplantation genetic testing are possible when both familial variants are classified as pathogenic or likely pathogenic, and are not based on variants of uncertain significance [18].

6. Differential Diagnosis

6.1. Differential diagnosis of TAF1 syndrome

The differential diagnosis includes the other forms of X-linked intellectual disability with variable facial and somatic features, chromatin and transcription syndromes with overlapping gestalt and brain malformations with hypotonia [2]. Cornelia de Lange syndrome deserves emphasis, because two patients had that clinical diagnosis before the TAF1 diagnosis [2]. Xq13 duplications including RLIM have a distinctive face, mild intellectual disability and mothers with skewed X inactivation, and are distinguished by microarray [36]. Abidi syndrome is part of the TAF1 spectrum [4, 19].

6.2. Differential diagnosis of XDP

GeneReviews refers to the general review of dystonias and emphasises the initial confusion with Parkinson disease, Parkinson-plus syndromes or essential tremor when dystonia is absent or subtle [5]. Caudate atrophy in advanced disease resembles Huntington disease [5]. Foot dystonia, rare in XDP, is frequent in DYT-TOR1A, which is more common in Ashkenazi Jews [5, 54]. Orphanet catalogues other entities with dystonia-parkinsonism, such as rapid-onset dystonia-parkinsonism (ATP1A3), dystonia 16 (PRKRA), adult-onset PLA2G6-associated neurodegeneration and infantile dystonia-parkinsonism [57].

6.3. Differential diagnosis of Alazami-Yuan syndrome

The central differential diagnosis is Cornelia de Lange syndrome, caused by variants in NIPBL, SMC1A, SMC3, RAD21, HDAC8, BRD4 and ANKRD11 according to the international consensus, with autosomal dominant or X-linked inheritance and, in the great majority, de novo variants [25]. Recessive inheritance, consanguinity and absence of the most specific signs (short nose with concave bridge, limb reduction defects, diaphragmatic hernia) favour TAF6 [10]. The other disorders of the transcription and chromatin machinery with overlapping gestalt form the second group: in patients with Coffin-Siris, Rubinstein-Taybi or Nicolaides-Baraitser syndromes, the consensus score never indicated the classic form and indicated the non-classic form in 4% to 10% [8, 25].

7. Treatments and Risk Management Strategies

7.1. TAF1 syndrome: principles

There is no targeted treatment and no published management guideline; care is supportive and directed at the manifestations: early intervention and developmental therapies (physiotherapy, occupational therapy and speech and language therapy), feeding support, treatment of epilepsy, auditory rehabilitation and repair of heart defects when present [2, 11]. The guidance in the following table is based on the published phenotype and not on trials.

Of the two boys with infantile spasms, the first received adrenocorticotrophic hormone (ACTH) 2 U/kg/day for 28 days, valproate 30 mg/kg/day and topiramate, with resolution of hypsarrhythmia and freedom from seizures from 2 years of age; the second did not respond to ACTH at the same doses or to levetiracetam, valproate and lamotrigine, and had six seizure-free months on a ketogenic diet started at 18 months, followed by relapse [32]. These doses are those of the case report and do not constitute a specific recommendation for TAF1 syndrome; treatment follows the general guidance for infantile spasms.

7.2. XDP: pharmacological treatment

No drug is authorised for XDP, and there are no randomised trials [11, 24]. GeneReviews describes a phased strategy, supported by clinical experience and case series: in early focal or segmental dystonia, anticholinergics (trihexyphenidyl, Artane™, with a more consistent effect than biperiden, Akineton™, in the moderate to advanced phases) and benzodiazepines (clonazepam) are used, with a greater effect when combined; in multifocal or generalised dystonia, polypharmacy gives partial relief, and zolpidem and tetrabenazine benefit mainly patients with phasic dystonia without contractures [5]. With zolpidem, some patients have almost complete improvement of dystonia for hours; the effect lasts 6 to 8 hours per 10 mg dose in the first weeks and then falls to 2 to 3 hours, and drowsiness is tolerated or overcome [5]. In parkinsonism, levodopa is effective mainly in pure parkinsonism, loses effect as dystonia progresses and does not cause induced dyskinesias with prolonged use; dopamine agonists control tremor, are less effective for bradykinesia and gait and rarely worsen dystonia [5]. A scoping review of 20 therapeutic studies considered oral drugs marginally effective [24].

7.3. XDP: botulinum toxin and other injections

Botulinum toxin improves focal dystonia, especially cervical dystonia, blepharospasm, and lingual and jaw dystonia; in a series of 109 patients, botulinum toxin type A gave marked improvement of oromandibular and lingual dystonia, moderate improvement of axial and truncal dystonia and a significant reduction in pain [5]. Injections into the neck or tongue may markedly worsen swallowing in patients with pre-existing dysphagia, and cost limits access in rural areas [5, 24]. Botulinum toxins have an FDA label for cervical dystonia and blepharospasm (for example Botox®, Dysport®, Xeomin® and Daxxify®), and not for XDP; the dose is that of the labelled indications and of focal dystonia practice, adjusted to the muscle [11]. Ethanol and lidocaine injections for afferent block, which are cheaper, give benefit for one to two weeks, with intense pain on injection and fibrosis and contractures with repetition [5].

7.4. XDP: deep brain stimulation and lesional surgery

The first patient treated with bilateral globus pallidus internus stimulation (GPi-DBS), with generalised dystonia, severe jaw-opening dystonia, sialorrhoea, dysphagia and unintelligible speech despite levodopa, piribedil, trihexyphenidyl and zolpidem, improved in the first week and maintained benefit at 13 years [5]. In a prospective study of 16 Filipino men (mean age 40.9±7.3 years, duration 1 to 6 years, predominant dystonia) operated on in Lübeck between 2013 and 2014 and followed in the Philippines for up to 46 months, dystonia severity rated on video by blinded observers fell by 55% at the first week (range −94% to +59%; p<0.01) and by 59% at 6 months (−100% to +22%; p<0.001); the Parkinson motor scale (UPDRS-III) fell by 19% and 27%; activities of daily living, pain, weight and quality of life improved; caudate atrophy predicted less benefit (r=0.817; p=0.004), and the authors proposed considering surgery early [55]. The ERN-RND systematic review (search up to June 2024) gathered 15 patients with individual pre- and postoperative scores: 12 (80%) were good responders, with a mean improvement of 64.2% (standard deviation 23.5) on the Burke-Fahn-Marsden motor scale (BFMDRS-M), three with more than 80% and one non-responder [7]. Parkinsonism responds less and may progress despite the benefit in dystonia [5, 7]. An impulse control disorder after GPi-DBS has been described [5].

In Ireland, two Filipino men were treated early with GPi-DBS with recording of local field potentials: the first (32 years) went from 17 to 6.5 points on the BFMDRS (62%) and from 49 to 16 on the XDP-specific scale at one year, returning to work; the second (47 years, predominantly bulbar dystonia) improved by 28% on the BFMDRS and 41% on the specific scale at 9 months, without improvement of lingual dystonia, which worsened intolerably when stimulation was switched off [56]. Pallidothalamic tractotomy by magnetic resonance-guided focused ultrasound (tcMRgFUS) has been described in three patients, with an overall improvement of 36.2% on the XDP-MDSP scale at one year, moderate to severe arm pain between 2 and 7 months in two patients and worsening of a non-motor subscale [58]; there is an expanded access programme in the Philippines (NCT05592028) [59]. Thalamotomies, pallidotomies and a cerebellar implant failed in earlier patients [5].

None of these studies has a placebo or sham surgery control group; the percentages depend on the scale, the disease phase and selection (predominant dystonia), and should not be compared across studies.

7.5. XDP: recommendations and divergences

There is no XDP-specific guideline. The ERN-RND systematic review formulated, by expert consensus, a recommendation for GPi-DBS in generalised or segmental dystonia due to DYT/PARK-TAF1 after failure of non-invasive therapies, for genetic testing before any DBS for dystonia, and for systematic assessment before and after surgery with a dystonia scale, a scale for each other movement disorder and a patient-reported outcome [7]. The 2018 GeneReviews reserves DBS for advanced, refractory disease [5], whereas the 2019 study and ERN-RND point to an earlier indication, given the relationship between caudate atrophy and worse outcome [7, 55]. The 2024 scoping review regards pallidal DBS as the best current option and tcMRgFUS as promising but with limited data [24]. DBS was approved by the FDA for dystonia in 2003 under a humanitarian device exemption (H020007) [7].

7.6. XDP: prevention of complications and supportive care

Swallowing assessment guides diet modification and swallowing techniques that reduce aspiration, and physiotherapy, combined with medical and surgical treatment, may delay confinement to bed [5]. Weight loss is frequent; an open controlled trial with a local food supplement in 50 patients showed no significant difference in body mass index at 12 weeks (p=0.084), with one death in the control group and no hospitalisation for infection [60]. After diagnosis, GeneReviews recommends neurological examination, speech, swallowing and nutritional assessment, surface electromyography and a genetics clinic consultation [5]. Scoping reviews recommend integrating psychological support, stigma reduction, caregiver support and community networks [50].

7.7. Developments in the last 12 months

Between November 2025 and October 2026 the following were published: the demonstration, in cells, that the SVA causes premature cleavage and polyadenylation and that BRD4 depletion corrects the molecular signature (preclinical) [37]; the role of MSH3 in somatic expansion and contraction of the repeat, with the proposal of strategies that reduce or inhibit MSH3 (human genetic study, without intervention) [42]; lentiform nucleus hyperechogenicity as a prodromal marker (observational study) [53]; the neuropathology of 18 brains [38]; the ERN-RND consensus recommendation on DBS [7]; and, in TAF1 syndrome, the integration of Abidi syndrome [4] and the description of infantile spasms [32]. On 31 October 2026 there were no pharmacological trials registered in ClinicalTrials.gov for XDP or TAF1 syndrome; registered studies were expanded access to tcMRgFUS (NCT05592028), a German observational study of sensorimotor integration (NCT05713721, status unknown) and the completed nutritional trial (NCT03019458) [59, 60]. In Alazami-Yuan syndrome, the only new data from the period is the description, in August 2026, of a patient with epileptic encephalopathy, complete atrioventricular block and bilateral cataracts, treated with transcutaneous auricular vagus nerve stimulation (case report, subsection 7.10); there were no registered trials [22, 61].

7.8. Preclinical research

Excision of the SVA by CRISPR/Cas9 normalised TAF1 expression in patient-derived cells [29]. A natural history study of 87 men (29 symptomatic positives, 7 asymptomatic positives and 51 negatives) followed for 18 months identified a minimal battery of 21 measures (questionnaires, scales and objective measures of bulbar and nutritional function) that explains 67.3% of the variance over the study, to serve as an outcome in trials [62]. In TAF1 syndrome, the T-type calcium channel enhancer SAK3, given orally at 0.25 mg/kg to rats with CRISPR/Cas9 editing of Taf1, corrected behavioural changes and protected Purkinje cells and cortical neurons, normalising the BDNF/AKT/GSK3β pathway [63, 64]. The mouse with a conditional Taf1 allele has been proposed as a therapeutic platform [35]. None of these approaches has reached human trials in the sources consulted.

7.9. Special populations and situations

In TAF1 syndrome, transition to adult care has few data, with men described up to 65 years of age, and the sources consulted do not describe specific perioperative care [30]. In XDP, dysphagia and laryngeal dystonia increase the risk of aspiration; symptomatic female carriers generally have mild disease and have been treated with levodopa and, in one case, with DBS [5, 6, 52]. The sources consulted do not describe pregnancy in women with manifest XDP or in carrier mothers of TAF1 syndrome.

7.10. Alazami-Yuan syndrome: management

There is no targeted treatment, nor any registered trial, guideline, consensus or case series on the treatment of Alazami-Yuan syndrome; a ClinicalTrials.gov search for TAF6 and for the name of the syndrome returned no studies in October 2026, and no gene therapies in development are described in the sources consulted [15, 61]. Management is symptomatic and, because of the phenotypic overlap, is guided by extrapolation from the international consensus on Cornelia de Lange syndrome, adjusted to each patient's findings [25]. From that consensus, the applicable points are speech and language therapy started in the first 18 months, with use of non-verbal communication; follow-up by a dietitian; treatment of gastro-oesophageal reflux disease with modification of enteral feeding and proton pump inhibitors as first line, endoscopy if there is no response and fundoplication reserved for failure of medical and nutritional treatment; initial hearing assessment with audiometry and otoacoustic emissions and/or brainstem auditory evoked potentials; regular vision assessment; scoliosis surveillance; an anaesthetic consultation before any surgery; and an emergency card with the main clinical data [25]. In the American patient, severe reflux led to gastrostomy with Nissen fundoplication at 11 months, deafness was corrected with hearing aids, the progressive congenital cataract had surgical extraction scheduled and the 15° scoliosis was kept under surveillance without intervention; she receives speech and language, occupational and physical therapy at school [10].

The published pharmacological approaches are from isolated patients and do not constitute a recommendation. In the Chinese boy, at 146.1 cm at 11 years, with a bone age of 13 years and normal IGF-1, the authors combined triptorelin (Decapeptyl®/Diphereline®) 3.75 mg injected every 28 days and somatropin 10 IU subcutaneously every night to improve final height, with a gain of 3.6 cm in 4 months, and vitamin D 2000 IU/day and calcium carbonate 500 mg/day for 3 months for vitamin D deficiency, with no adverse effects recorded [18]. Somatropin without documented growth hormone deficiency is off-label use, and the combination has not been evaluated beyond that case [18].

In the Jordanian boy, the epileptic encephalopathy, with 2 to 4 prolonged seizures a day, was refractory to perampanel, levetiracetam, phenobarbital, topiramate and clonazepam, and implanted vagus nerve stimulation was avoided because of the unipolar VVIR epicardial pacemaker placed at 18 months for complete atrioventricular block [22]. He was treated with transcutaneous stimulation of the auricular branch of the vagus nerve (taVNS) in the cymba conchae, at 25 Hz, 250 μs, intensity titrated up to 1.7 mA, 4 hours a day, started in paediatric intensive care with continuous electrocardiographic monitoring for 5 days and followed prospectively for 9 months with a seizure diary, electroencephalograms, electrocardiograms and pacemaker interrogations [22]. At the first month, seizures fell to 2 to 3 a week, mainly brief generalised tonic seizures, and remained at about 2 a week at 9 months; phenobarbital was stopped at month 3 and levetiracetam at month 7, with reduction of topiramate and clonazepam; the PedsQL-FIM score (family impact) rose from 10.4 to 84.0; the electroencephalogram remained markedly abnormal; there were no arrhythmias or device changes, and the only adverse effect was transient local skin irritation [22]. It is a single-patient, uncontrolled human study, published in August 2026, that shows the feasibility of the technique in a pacemaker-dependent patient, not its efficacy in the syndrome [22].

8. Reproductive Options and Family Counselling

8.1. TAF1 syndrome

In the published series, about half of the variants arose de novo and the other half were inherited from healthy carrier mothers: 11 families in the initial series (point variants: 6 de novo and 3 maternal; duplications: one de novo and one maternal) and 22 families in the allelic series (10 de novo and 12 maternal) [1, 2]. When the mother is a carrier, each son has a 50% probability of inheriting the variant and, if he inherits it, of being affected, and each daughter has a 50% probability of being a carrier, as a rule asymptomatic with skewed X inactivation [2, 30]. In the family studied in Sweden, the disease passed through five generations via healthy women [30]. When the variant is de novo in the proband and is not detected in maternal blood, the recurrence risk is low but not zero, because of the possibility of germline mosaicism; the sources consulted give no numerical estimate. Girls with de novo variants may be affected [2].

Testing of maternal sisters and aunts of reproductive age is only of value when the variant is classified as pathogenic or likely pathogenic; with a maternal VUS, segregation in the men of the family (affected and healthy) is the first step and may reclassify the variant [2]. Prenatal diagnosis and preimplantation genetic testing require a variant classified as pathogenic or likely pathogenic.

8.2. XDP

Inheritance is X-linked: the father of a patient is not a carrier; the mother is an obligate carrier when there is more than one affected individual in the family, and in an isolated case is probably a carrier, because de novo occurrence of the haplotype has not been observed [5]. Each son of a carrier has a 50% probability of inheriting the haplotype and becoming affected, and each daughter a 50% probability of being a carrier, as a rule without symptoms; an affected man transmits the haplotype to all his daughters and to none of his sons; maternal aunts may be carriers [5]. Kupke and colleagues observed 52% affected sons among 120 sons of carrier mothers [6]. The repeat tends to expand on transmission, and its length and MSH3 and PMS2 variants modulate age at onset, which allows estimates of age at onset in research but not a validated individual prediction for clinical practice [39, 40, 43].

Predictive testing in asymptomatic at-risk men is possible once the haplotype has been identified in the family, with prior counselling about the social and economic consequences and the need for follow-up [5]. In asymptomatic minors, GeneReviews considers predictive testing inappropriate, because the disease begins in adulthood and there is no early treatment with demonstrated benefit, based on the positions of the National Society of Genetic Counselors and the American Academy of Pediatrics/ACMG; symptomatic minors may be tested at any age [5]. Carrier testing, prenatal diagnosis and preimplantation diagnosis are possible once the haplotype has been identified, and the best time to discuss them is before pregnancy; DNA banking is useful in patients without molecular confirmation [5].

8.3. Alazami-Yuan syndrome

Inheritance is autosomal recessive: the parents of a patient are, as a rule, healthy heterozygotes, and each subsequent pregnancy has a 25% probability of an affected child, 50% of a healthy carrier and 25% of a non-carrier, regardless of sex [8, 18]. In the Turkish family, the parents and a sister were carriers and the healthy younger brother did not have the variant; in the Saudi family, the parents were carriers and the unaffected siblings were homozygous for the reference allele [8]. The risk that a healthy sibling is a carrier is two thirds. In consanguineous families, testing of relatives of reproductive age and of the consanguineous spouse is more useful, because the probability of a carrier couple within the same family is higher. Confirmation of the phase of the variants in the parents is mandatory before calculating risk, and prenatal diagnosis or preimplantation genetic testing are only offered with both variants classified as pathogenic or likely pathogenic [10, 18]. None of the sources describes heterozygous carriers with signs of the disease; in the Turkish family, a 17-year-old heterozygous sister had epilepsy and intellectual disability, which the authors did not attribute to the variant [8]. There are no data on reproduction in patients.

9. Special Considerations: Population Diversity and Epidemiology

9.1. TAF1 syndrome

There is no prevalence estimate. Orphanet classifies it as <1/1,000,000 and records only a case count, which is not a prevalence [3, 11]. The series describe patients of European ancestry (including the United Kingdom, Spain, Albania, Greece and Sweden), from Ecuador, Colombia, the United States, Japan and China [1, 2, 30, 32]. Data from American commercial laboratories (71 variants in 18,256 probands and 15 in about 8100) measure reported variants, mostly of uncertain significance, and not the frequency of the disease [2]. The sources consulted include no data from Portugal or Brazil.

9.2. XDP

XDP is a Filipino founder-effect disease with an estimated origin about 50 meiotic generations ago (about 1000 years) on the island of Panay [5]. Prevalence has been estimated at 5.24/100,000 in Panay, 18.9/100,000 in the province of Capiz and 0.34/100,000 in the Philippines as a whole [5]; Orphanet gives 0.31/100,000 in the Philippines [11, 21], and Lee and colleagues calculated about 1 in 4000 men in Capiz [17]. These estimates come from registries and regional surveys and not from population screening. The Filipino registry had 505 cases in January 2010 [17]. All known patients have Filipino ancestry, although not all have maternal ancestry traceable to Panay, and there are cohorts in Mindanao [5, 43]. Outside the Philippines, patients are emigrants or their descendants, such as those treated in Germany and Ireland [55, 56]; Orphanet does not assign a figure for Europe [11]. The sources consulted include no data from Portugal or Brazil. Inequalities in access to genetic testing, botulinum toxin and surgery in rural Filipino areas, with distances of more than 100 km, shape the treated disease [5, 50].

9.3. European networks and associations

ERN ITHACA (congenital malformations and intellectual disability) is the reference network for TAF1 syndrome [26], and ERN-RND (rare neurological diseases), which promoted the review on DBS in monogenic dystonia, is the network for XDP [7, 27]. For XDP, GeneReviews lists the Sunshine Care Foundation (Philippines), the Dystonia Medical Research Foundation, the American Parkinson Disease Association and the Parkinson's Foundation [5]. The sources consulted did not identify a patient association specific to TAF1 syndrome.

9.4. Alazami-Yuan syndrome

There is no prevalence estimate; the eight patients published up to October 2026 are a case count [10, 15, 22]. The families come from Saudi Arabia, Turkey (two), China, the United States and Jordan, and three of the six families are consanguineous (the two Turkish families and the Saudi one); the p.(Ile71Thr) variant was found in three families of different origins, with no haplotype study indicating a founder effect [8, 10, 18, 22, 34]. The sources consulted describe no European, Portuguese or Brazilian patients. ERN ITHACA is the relevant European network [26]; the sources consulted did not identify a specific patient association; the international consensus on Cornelia de Lange syndrome, applicable to management by extrapolation, was promoted by the scientific advisory council of the World Federation of CdLS Support Groups [25].

10. Illustrative Clinical Case

Xiao and colleagues described in 2026 a Chinese boy, with no relevant perinatal or family history, with appropriate development up to 6 months (head control at 3 months, sitting with support at 6) [32]. At 7 months, epileptic spasms appeared in clusters of 5 to 8, two to three clusters a day, with arrest of developmental progress, and the electroencephalogram showed typical hypsarrhythmia [32]. Brain magnetic resonance imaging and metabolic screening of blood and urine were normal [32]. Trio exome sequencing identified a hemizygous TAF1 missense variant, published as c.236C>A (p.T79N) and corresponding to NM_004606.5:c.176C>A p.(Thr59Asn), inherited from the asymptomatic mother; the older brother and the parents were healthy, and no other pathogenic variants associated with intellectual disability or epilepsy were found [12, 32]. The variant affects a conserved residue of the TBP-binding domain and reduced protein levels in transfected cells [32]. Treatment with ACTH 2 U/kg/day for 28 days, valproate 30 mg/kg/day and topiramate resolved the hypsarrhythmia at 9 months, with occasional persisting spasms; after medication adjustments, he was seizure-free from 2 years of age [32]. At 5 years he had moderate intellectual disability on the Gesell scales and, at 11 years, walked unaided but did not run and spoke in sentences of 7 to 8 words [32]. Figure 1 of the paper (legend consulted in the PMC full text) shows the pedigrees, the Sanger chromatograms, residue conservation and the TAF1 domain diagram (https://pmc.ncbi.nlm.nih.gov/articles/PMC13583436/figure/fig-0001/) [32]. The case shows a maternal variant in a patient with a healthy mother, the expansion of the epileptic phenotype and the favourable seizure outcome with standard infantile spasms treatment; the doses are those of the report and not a syndrome-specific protocol. The second patient in the same publication, with a bromodomain variant, had drug-resistant spasms and severe impairment, which illustrates the variability [32].

As a counterpoint in XDP, a 32-year-old Filipino man treated in Ireland presented with six months of pain and stiffness of the lower limbs, abdominal and hip flexor spasms and weight loss; testing confirmed XDP, the disease progressed to nasogastric tube feeding despite medication, and bilateral GPi-DBS, 12 months after presentation, reduced the BFMDRS from 17 to 6.5 points and allowed a return to full-time work, with sustained suppression of beta bursts in the local field potentials [56]. Figure 1 of that paper (legend consulted) shows the local field potentials and the clinical scales over follow-up (https://pmc.ncbi.nlm.nih.gov/articles/PMC12275003/figure/mdc370044-fig-0001/) [56].

In Alazami-Yuan syndrome, Pappas and colleagues described in 2025 an American girl born at 36 weeks by caesarean section, because of preterm labour and breech presentation, small for gestational age (1.79 kg, below the 3rd centile), with no relevant family history or consanguinity [10]. She spent two weeks in neonatal intensive care because of feeding difficulty and a left congenital cataract; at 6 months she had aspiration pneumonia and was fed by nasogastric tube, and at that time the thick, arched eyebrows, long eyelashes, widely spaced teeth and microcephaly led to a clinical diagnosis of Cornelia de Lange syndrome [10]. Karyotype, microarray, NIPBL and SMC1A sequencing and testing for NIPBL deletions and duplications were normal; at 11 months she underwent gastrostomy with Nissen fundoplication for severe reflux, at 12 months she had global delay and mild to moderate bilateral sensorineural hearing loss, and magnetic resonance imaging at 23 months showed only myelination at the lower limit for age [10]. From the age of 9 she had episodes of interphalangeal arthritis and transient discolouration of the fingers; at 16 years she received diagnoses of palindromic rheumatism, generalised joint hypermobility and 15° lumbar scoliosis, with positive anti-SS-B antibodies [10]. At genetic reassessment at 16 years, a 2775-gene panel analysed as a trio identified NM_139315.3:c.212T>C p.(Ile71Thr), paternal, and c.1863dup p.(Thr622HisfsTer116), maternal, both likely pathogenic, with no variants in the genes for Cornelia de Lange syndrome or the overlapping entities included in the panel, among them TAF1 [10, 12]. Her score on the international consensus was 8, below the threshold for the non-classic form [10, 25]. At 17 years she weighed 41.7 kg, measured 152.4 cm and had a head circumference of 48.8 cm; she remained non-verbal but responded to speech by sign language with hearing aids, had no regression, fed mainly orally and was awaiting cataract extraction [10]. The case shows that a clinical diagnosis of Cornelia de Lange syndrome without molecular confirmation may conceal a recessive disease with a 25% recurrence risk, and that reassessment with a broad panel or exome has diagnostic yield even in adolescence.

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

In TAF1 syndrome, the prognosis is that of a permanent neurodevelopmental disorder, with intellectual disability and language delay in most; survival to advanced adulthood is documented in one family, with men aged 52, 54 and 65 years [2, 30]. Duplications including TAF1 and some patients with point variants had a progressive course, and one boy with a duplication died at 8 years [1]. Epilepsy, when present, may be drug-resistant [32]. There are no natural history studies of the syndrome.

In XDP, prognosis depends on the initial phenotype: pure parkinsonism progresses slowly, whereas the early combination of oro-bucco-lingual and cervical dystonia with parkinsonism generalises within 2 to 5 years and is associated with premature death from aspiration, stridor or infections [5]. Age at onset depends on the repeat and the modifiers [39, 40]. GPi-DBS improves dystonia in a sustained way for at least 46 months in the prospective study and 13 years in the first case, without an equivalent effect on parkinsonism [5, 55].

In Alazami-Yuan syndrome, the prognosis is that of a permanent neurodevelopmental disorder in most patients, ranging from the normal intelligence of the Chinese patient to the severe delay with epileptic encephalopathy of the Jordanian patient [18, 22]. The patients described were aged between 4 and 18 years, with no data on older adults; the American patient had no regression up to 17 years [10, 34]. Three untested siblings with a similar picture died at 5 and 9 months (Turkish family) and in adolescence (Jordanian family), which suggests early mortality in some of the families [8, 22].

11.2. Surveillance

The sources consulted do not define surveillance intervals for TAF1 syndrome; the intervals given for this entity derive from the phenotype and are clinical, not guideline-based. For Alazami-Yuan syndrome there are also no published surveillance intervals; the corresponding rows apply by extrapolation the Cornelia de Lange syndrome consensus, which recommends echocardiography and renal ultrasound for all patients with that syndrome at diagnosis, and add the electrocardiogram because of the atrioventricular block described in one patient [22, 25].

12. Summary and Final Considerations

TAF1 encodes the largest subunit of TFIID and is one of the most constrained genes in the genome; complete loss is lethal, and human diseases result from hypomorphic or expression changes. TAF1 syndrome (MRXS33) is a neurodevelopmental disorder of boys, with global delay, intellectual disability and/or autism, hypotonia, dysphagia, non-specific facial dysmorphism, gluteal crease with sacral caudal remnant, hearing impairment, hypoplasia of the corpus callosum and epilepsy in about a quarter, including infantile spasms. Half of the variants are inherited from healthy mothers with skewed X inactivation, which makes classification of missense variants dependent on segregation, X inactivation, population data and phenotype. Abidi syndrome falls within the spectrum, and Xq13.1 duplications including TAF1 are associated with a more severe picture. The literature uses a numbering 60 nucleotides above the MANE transcript NM_004606.5, which requires rigorous conversion. Treatment is supportive.

XDP is a neurodegenerative disease of adults of Filipino ancestry, caused by a founder haplotype with an intronic SVA insertion that reduces the canonical TAF1 transcript through aberrant splicing, intron retention and premature termination. The SVA hexanucleotide repeat, unstable and somatically mosaic, and MSH3 and PMS2 variants explain most of the variation in age at onset. The exome does not detect this cause, so diagnosis requires a targeted test. Treatment is symptomatic: oral drugs with limited effect, botulinum toxin used with caution in dysphagia and GPi-DBS, recommended by ERN-RND for segmental or generalised dystonia after failure of non-invasive therapies, with a better result before caudate atrophy develops. Annual surveillance of carriers from 30 years of age, six-monthly follow-up of patients and swallowing assessment complete management. The targets under investigation (BRD4, MSH3, excision of the SVA) remain at the preclinical stage.

Alazami-Yuan syndrome, caused by biallelic variants in TAF6, another TFIID subunit, is an autosomal recessive disease with a phenotype close to Cornelia de Lange syndrome, described in eight patients from six families. Arched eyebrows and long eyelashes are constant, the nasal bridge is prominent, and limb anomalies and diaphragmatic hernia are absent; no patient reached the score of the classic form of that syndrome. The spectrum includes developmental delay and intellectual disability in most, growth restriction, deafness, congenital cataract, reflux with aspiration and, in one patient, epileptic encephalopathy and complete atrioventricular block. Diagnosis is made by exome or broad panel with confirmation of phase in the parents, and should be sought in patients with a clinical diagnosis of Cornelia de Lange syndrome without a molecular cause, especially with consanguinity. Management follows, by extrapolation, the Cornelia de Lange syndrome consensus, and counselling rests on a 25% recurrence risk.

13. References

1. O'Rawe JA, Wu Y, Dörfel MJ, Rope AF, Au PYB, Parboosingh JS, et al. TAF1 variants are associated with dysmorphic features, intellectual disability, and neurological manifestations. Am J Hum Genet. 2015;97(6):922-32. doi:10.1016/j.ajhg.2015.11.005
2. Cheng H, Capponi S, Wakeling E, Marchi E, Li Q, Zhao M, et al. Missense variants in TAF1 and developmental phenotypes: challenges of determining pathogenicity. Hum Mutat. 2020;41(2):449-64. doi:10.1002/humu.23936
3. Orphanet. X-linked intellectual disability-global development delay-facial dysmorphism-sacral caudal remnant syndrome (ORPHA:480907). Available from: https://www.orpha.net/en/disease/detail/480907
4. Han M, Abidi F, Skinner C, Srikanth S, Wang R, Niranjan T, et al. Abidi syndrome and another family with X-linked intellectual disability have TAF1 variants and reduced phosphorylation of RNA polymerase II associated protein 74. Am J Med Genet A. 2026. doi:10.1002/ajmg.a.70268
5. Evidente VGH. X-Linked Dystonia-Parkinsonism. 2005 Dec 13 [updated 2018 Feb 15]. 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/NBK1489/
6. Online Mendelian Inheritance in Man, OMIM®. #314250 Dystonia 3, torsion, X-linked; DYT3 (dystonia-parkinsonism, X-linked; XDP). Baltimore (MD): Johns Hopkins University. https://omim.org/entry/314250
7. Indelicato E, Carmona-Hidalgo B, Quintero J, Escobar JDO, Koy A, Salamon A, et al; European Reference Network for Rare Neurological Diseases (ERN-RND). Efficacy of deep brain stimulation for the treatment of monogenic dystonia symptoms: a systematic review. Eur J Neurol. 2026;33(1):e70490. doi:10.1111/ene.70490
8. Yuan B, Pehlivan D, Karaca E, Patel N, Charng WL, Gambin T, et al. Global transcriptional disturbances underlie Cornelia de Lange syndrome and related phenotypes. J Clin Invest. 2015;125(2):636-51. doi:10.1172/JCI77435
9. Orphanet. Alazami-Yuan syndrome (ORPHA:694946). Available from: https://www.orpha.net/en/disease/detail/694946
10. Pappas A, Mooney M, Kohnen K, Owens JW, Zhang W, Hopkin RJ, et al. The phenotypic spectrum of the Cornelia de Lange-like "Alazami-Yuan syndrome": a case report of the 7th diagnosed individual and review of the literature. Clin Case Rep. 2025;13(5):e9208. doi:10.1002/ccr3.9208
11. GeneGuidance. TAF1 — clinical curation of the gene and OMIM conditions 300966/314250. GeneGuidance database, version of 2026-09-26.
12. 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):61-8. doi:10.1002/humu.23348
13. Online Mendelian Inheritance in Man, OMIM®. *313650 TAF1 RNA polymerase II, TATA box-binding protein-associated factor, 250-kD; TAF1. Baltimore (MD): Johns Hopkins University. https://omim.org/entry/313650
14. Bragg DC, Mangkalaphiban K, Vaine CA, Kulkarni NJ, Shin D, Yadav R, et al. Disease onset in X-linked dystonia-parkinsonism correlates with expansion of a hexameric repeat within an SVA retrotransposon in TAF1. Proc Natl Acad Sci U S A. 2017;114(51):E11020-8. doi:10.1073/pnas.1712526114
15. GeneGuidance. TAF6 — clinical curation of the gene and OMIM condition 617126. GeneGuidance database, version of 2026-09-26.
16. Online Mendelian Inheritance in Man, OMIM®. #300966 Intellectual developmental disorder, X-linked, syndromic 33; MRXS33. Baltimore (MD): Johns Hopkins University; last edited 2021-08-20. https://omim.org/entry/300966
17. Lee LV, Rivera C, Teleg RA, Dantes MB, Pasco PM, Jamora RD, et al. The unique phenomenology of sex-linked dystonia parkinsonism (XDP, DYT3, "Lubag"). Int J Neurosci. 2011;121 Suppl 1:3-11. doi:10.3109/00207454.2010.526728
18. Lin SZ, Feng JH, Sun LP, Ma HW, Wang WQ, Li JY. Novel compound heterozygous variants in the TAF6 gene in a patient with Alazami-Yuan syndrome: a case report. World J Clin Cases. 2022;10(6):1889-95. doi:10.12998/wjcc.v10.i6.1889
19. Online Mendelian Inheritance in Man, OMIM®. %300262 Intellectual developmental disorder, X-linked, syndromic, Abidi type; MRXSAB. Baltimore (MD): Johns Hopkins University; last edited 2025-07-15. https://omim.org/entry/300262
20. Online Mendelian Inheritance in Man, OMIM®. #617126 Alazami-Yuan syndrome. Baltimore (MD): Johns Hopkins University. https://omim.org/entry/617126
21. Orphanet. X-linked dystonia-parkinsonism (ORPHA:53351). Available from: https://www.orpha.net/en/disease/detail/53351
22. Al-Ghawanmeh R, Al-Momani H, Abed Alrahman R, Badaineh Y, Alarameen D. Transcutaneous auricular vagus nerve stimulation in a pacemaker-dependent child with syndromic drug-resistant epilepsy: a case report with prospective 9-month follow-up. Case Rep Neurol. 2026;18(1):580-8. doi:10.1159/crn/abyag011
23. Giovenino C, Trajkova S, Pavinato L, Cardaropoli S, Pullano V, Ferrero E, et al. Skewed X-chromosome inactivation in unsolved neurodevelopmental disease cases can guide re-evaluation for X-linked genes. Eur J Hum Genet. 2023;31(11):1228-36. doi:10.1038/s41431-023-01324-w
24. Alonto AHD, Jamora RDG. A scoping review on the diagnosis and treatment of X-linked dystonia-parkinsonism. Parkinsonism Relat Disord. 2024;119:105949. doi:10.1016/j.parkreldis.2023.105949
25. Kline AD, Moss JF, Selicorni A, Bisgaard AM, Deardorff MA, Gillett PM, et al. Diagnosis and management of Cornelia de Lange syndrome: first international consensus statement. Nat Rev Genet. 2018;19(10):649-66. doi:10.1038/s41576-018-0031-0
26. European Reference Network ITHACA. https://ern-ithaca.eu/
27. European Reference Network for Rare Neurological Diseases (ERN-RND). https://www.ern-rnd.eu/
28. Makino S, Kaji R, Ando S, Tomizawa M, Yasuno K, Goto S, et al. Reduced neuron-specific expression of the TAF1 gene is associated with X-linked dystonia-parkinsonism. Am J Hum Genet. 2007;80(3):393-406. doi:10.1086/512129
29. Aneichyk T, Hendriks WT, Yadav R, Shin D, Gao D, Vaine CA, et al. Dissecting the causal mechanism of X-linked dystonia-parkinsonism by integrating genome and transcriptome assembly. Cell. 2018;172(5):897-909.e21. doi:10.1016/j.cell.2018.02.011
30. Gudmundsson S, Wilbe M, Filipek-Górniok B, Molin AM, Ekvall S, Johansson J, et al. TAF1, associated with intellectual disability in humans, is essential for embryogenesis and regulates neurodevelopmental processes in zebrafish. Sci Rep. 2019;9(1):10730. doi:10.1038/s41598-019-46632-8
31. Hurst SE, Liktor-Busa E, Moutal A, Parker S, Rice S, Szelinger S, et al. A novel variant in TAF1 affects gene expression and is associated with X-linked TAF1 intellectual disability syndrome. Neuronal Signal. 2018;2(3):NS20180141. doi:10.1042/NS20180141
32. Xiao H, Xie C, Peng P, Ni X, Mao L, Peng J, et al. Infantile epileptic spasm syndrome caused by TAF1 missense variants: expanding the epileptic phenotype and revealing underlying impairments. Hum Mutat. 2026;2026:4104648. doi:10.1155/humu/4104648
33. Alazami AM, Patel N, Shamseldin HE, Anazi S, Al-Dosari MS, Alzahrani F, et al. Accelerating novel candidate gene discovery in neurogenetic disorders via whole-exome sequencing of prescreened multiplex consanguineous families. Cell Rep. 2015;10(2):148-61. doi:10.1016/j.celrep.2014.12.015
34. Tuc E, Bengur FB, Aykut A, Sahin O, Alanay Y. The third family with TAF6-related phenotype: Alazami-Yuan syndrome. Clin Genet. 2020;97(5):795-6. doi:10.1111/cge.13711
35. Crombie EM, Korecki AJ, Cleverley K, Adair BA, Cunningham TJ, Lee WC, et al. Taf1 knockout is lethal in embryonic male mice and heterozygous females show weight and movement disorders. Dis Model Mech. 2024;17(7):dmm050741. doi:10.1242/dmm.050741
36. Palmer EE, Carroll R, Shaw M, Kumar R, Minoche AE, Leffler M, et al. RLIM is a candidate dosage-sensitive gene for individuals with varying duplications of Xq13, intellectual disability, and distinct facial features. Am J Hum Genet. 2020;107(6):1157-69. doi:10.1016/j.ajhg.2020.10.005
37. Capponi S, Ehret S, Camgöz Z, Gather F, Vaine CA, Özyerli-Göknar E, et al. Correction of the molecular phenotype of X-linked dystonia-parkinsonism reveals a non-canonical function of BRD4. Nat Commun. 2026;17(1):4062. doi:10.1038/s41467-026-72319-6
38. Tan AYS, Thai NSSS, Arasaratnam CJ, Han SA, Penney EB, Acuña-Sunshine G, et al. Regional and compartmental changes drive progressive patterns of striatal degeneration in X-linked dystonia parkinsonism. Ann Neurol. 2026;100(4):751-67. doi:10.1002/ana.78306
39. Westenberger A, Reyes CJ, Saranza G, Dobricic V, Hanssen H, Domingo A, et al. A hexanucleotide repeat modifies expressivity of X-linked dystonia parkinsonism. Ann Neurol. 2019;85(6):812-22. doi:10.1002/ana.25488
40. Laabs BH, Klein C, Pozojevic J, Domingo A, Brüggemann N, Grütz K, et al. Identifying genetic modifiers of age-associated penetrance in X-linked dystonia-parkinsonism. Nat Commun. 2021;12(1):3216. doi:10.1038/s41467-021-23491-4
41. Trinh J, Lüth T, Schaake S, Laabs BH, Schlüter K, Laß J, et al. Mosaic divergent repeat interruptions in XDP influence repeat stability and disease onset. Brain. 2023;146(3):1075-82. doi:10.1093/brain/awac160
42. Mejia Maza A, Hincher M, Correia K, Gillis T, Nishiyama A, Penney EB, et al. MSH3 is a genetic modifier of somatic repeat instability in X-linked dystonia parkinsonism. Am J Hum Genet. 2026;113(1):83-99. doi:10.1016/j.ajhg.2025.12.002
43. Doquenia MLM, Dy Closas AMF, Algodon SM, Suarez-Uy R, Ng A, Laabs BH, et al. Clinicodemographic and genetic modifier correlation in an X-linked dystonia-parkinsonism cohort from Mindanao. Mov Disord Clin Pract. 2024;11(12):1604-8. doi:10.1002/mdc3.14193
44. Pozojevic J, von Holt BH, Westenberger A. Factors influencing reduced penetrance and variable expressivity in X-linked dystonia-parkinsonism. Med Genet. 2022;34(2):97-102. doi:10.1515/medgen-2022-2135
45. National Center for Biotechnology Information. ClinVar: TAF1[gene]. https://www.ncbi.nlm.nih.gov/clinvar/?term=TAF1%5Bgene%5D
46. Dahiya R, Natarajan K. Mutational analysis of TAF6 revealed the essential requirement of the histone-fold domain and the HEAT repeat domain for transcriptional activation. FEBS J. 2018;285(8):1491-510. doi:10.1111/febs.14423
47. Bieniossek C, Papai G, Schaffitzel C, Garzoni F, Chaillet M, Scheer E, et al. The architecture of human general transcription factor TFIID core complex. Nature. 2013;493(7434):699-702. doi:10.1038/nature11791
48. National Center for Biotechnology Information. ClinVar: TAF6 — VCV000180203 (c.136C>T) and VCV000180204 (c.212T>C). https://www.ncbi.nlm.nih.gov/clinvar/?term=TAF6%5Bgene%5D
49. Janssen BDE, van den Boogaard MH, Lichtenbelt K, Seaby EG, Stals K, Ellard S, et al. De novo putative loss-of-function variants in TAF4 are associated with a neuro-developmental disorder. Hum Mutat. 2022;43(12):1844-51. doi:10.1002/humu.24444
50. Ramos MD, Manning J, Harris J, Adorno M, Marye R, Evangelista L. The clinical, genetic, psychosocial, and care landscape of X-linked dystonia-parkinsonism in Filipino men: a scoping review. Am J Mens Health. 2026;20(4):15579883261467125. doi:10.1177/15579883261467125
51. Chin HL, Lin CY, Chou OH. X-linked dystonia parkinsonism: epidemiology, genetics, clinical features, diagnosis, and treatment. Acta Neurol Belg. 2023;123(1):45-55. doi:10.1007/s13760-022-02144-3
52. Bautista AND, Rosales RL, Diesta CCE, Oropilla JQ, Westenberger A, Brüggemann N, et al. X-linked dystonia-parkinsonism in a manifesting female: first case report of deep brain stimulation. Mov Disord Clin Pract. 2025;12(8):1204-7. doi:10.1002/mdc3.70082
53. Pauly MG, Diesta CCE, Cataniag P, Borsche M, Hanssen H, Oropilla JQL, et al. Striatal hyperechogenicity as an ultrasound imaging marker for prodromal X-linked dystonia-parkinsonism. NPJ Parkinsons Dis. 2026;12(1):133. doi:10.1038/s41531-026-01418-4
54. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 138–141.
55. Brüggemann N, Domingo A, Rasche D, Moll CKE, Rosales RL, Jamora RDG, et al. Association of pallidal neurostimulation and outcome predictors with X-linked dystonia parkinsonism. JAMA Neurol. 2019;76(2):211-6. doi:10.1001/jamaneurol.2018.3777
56. Donlon E, O'Keeffe C, Horan J, Ruggieri F, Fitzpatrick J, O'Neill M, et al. Clinical outcomes with prospective brain sensing data following bilateral globus pallidus deep brain stimulation in X-linked dystonia parkinsonism. Mov Disord Clin Pract. 2025;12(7):1024-7. doi:10.1002/mdc3.70044
57. Orphanet. Rare disease nomenclature (Orphanet nomenclature pack), consulted in a local project database; entities ORPHA:71517, ORPHA:199351, ORPHA:210571 and ORPHA:238455. https://www.orpha.net
58. Jamora RDG, Chang WC, Taira T. Transcranial magnetic resonance-guided focused ultrasound in X-linked dystonia-parkinsonism. Life (Basel). 2021;11(5):392. doi:10.3390/life11050392
59. University of the Philippines Manila – Philippine General Hospital. High intensity focused ultrasound for X-linked dystonia-parkinsonism. ClinicalTrials.gov NCT05592028. https://clinicaltrials.gov/study/NCT05592028
60. Sunshine Care Foundation. MINGO supplemental trial in X-linked dystonia-parkinsonism patients. ClinicalTrials.gov NCT03019458. https://clinicaltrials.gov/study/NCT03019458
61. ClinicalTrials.gov. Search “TAF6 OR Alazami-Yuan” (API v2), accessed 1 November 2026. https://clinicaltrials.gov/search?term=TAF6%20OR%20Alazami-Yuan
62. Acuna P, Supnet-Wells ML, Spencer NA, de Guzman JK, Russo M, Hunt A, et al. Establishing a natural history of X-linked dystonia parkinsonism. Brain Commun. 2023;5(3):fcad106. doi:10.1093/braincomms/fcad106
63. Janakiraman U, Dhanalakshmi C, Yu J, Moutal A, Boinon L, Fukunaga K, et al. The investigation of the T-type calcium channel enhancer SAK3 in an animal model of TAF1 intellectual disability syndrome. Neurobiol Dis. 2020;143:105006. doi:10.1016/j.nbd.2020.105006
64. Dhanalakshmi C, Janakiraman U, Moutal A, Fukunaga K, Khanna R, Nelson MA. Evaluation of the effects of the T-type calcium channel enhancer SAK3 in a rat model of TAF1 deficiency. Neurobiol Dis. 2021;149:105224. doi:10.1016/j.nbd.2020.105224