Cornelia de Lange Syndrome and Cohesinopathies

Literature and regulatory information updated: 6 October 2026.

Cornelia de Lange syndrome (Brachmann-de Lange syndrome, typus degenerativus amstelodamensis), classical and non-classical spectrum associated with NIPBL, SMC1A, SMC3, RAD21, HDAC8, BRD4 and MAU2, overlap with KBG syndrome (ANKRD11), and cohesinopathies in the differential diagnosis

0. Initial Reference Table

1. Introduction

Cornelia de Lange syndrome is a multisystem developmental disorder with physical, cognitive and behavioural manifestations, whose clinical recognition rests on a distinctive facial pattern, prenatal-onset growth restriction, upper-limb abnormalities and intellectual disability of variable severity [3]. The first description of a patient with severe oligodactyly consistent with the condition is attributed to Vrolik, in 1849, and in 1916 Brachmann reported, from an autopsy, a child with symmetrical monodactyly, antecubital webs, dwarfism, cervical ribs and hirsutism [1, 19]. In 1933, the Dutch paediatrician Cornelia de Lange described two unrelated girls with similar characteristics and named the condition after the city where she worked, typus degenerativus amstelodamensis; the designation “Brachmann-de Lange syndrome” persists in some of the literature, but the eponym “Cornelia de Lange” has become the most widely used [1, 19]. The clinical delineation of the classical form was consolidated by the series of Ptacek and colleagues (1963) and the review of 310 individuals by Jackson and colleagues (1993), which already demonstrated a higher proportion of mild cases than was then accepted and a frequency of upper-limb deficiencies of only 27% [1, 20].

The molecular basis was identified in 2004, when two independent groups demonstrated pathogenic variants in NIPBL, the human homologue of the Nipped-B gene of Drosophila melanogaster and the fungal sister-chromatid cohesion protein Scc2; Krantz and colleagues identified the gene through linkage analysis in 12 families and a de novo translocation t(5;13)(p13.1;q12.1), and Tonkin and colleagues named the protein product delangin [21, 22]. Descriptions of variants in SMC1A (X-linked, 2006), SMC3 (2007), HDAC8 and RAD21 (2012), BRD4 (2018) and MAU2 (initial case in 2020; series of 18 individuals in 2026) followed, all in structural proteins or regulators of the cohesin complex [23–29]. The identification of these genes revealed individuals with milder or atypical phenotypes and transformed the entity into a spectrum, in which the mild form may account for most cases [1].

Expansion of the spectrum required criteria to be defined. In 2007, Kline and colleagues revised the diagnostic criteria, including milder forms, and proposed a severity scoring system based on experience with more than 500 patients [30]. In 2018, an international group of 43 participants from 30 institutions in 9 countries, representing the Scientific Council of the World Federation of CdLS Support Groups, published the first international consensus, developed through a modified Delphi process and voted on by 37 authors with greater than 90% agreement on all 68 recommendations [3]. That document distinguished cardinal and suggestive signs, defined thresholds for classical and non-classical forms, and established that not all cohesinopathies belong to the spectrum: Roberts syndrome and Nicolaides-Baraitser syndrome, for example, were excluded because of limited overlap with the classical phenotype [3]. In France, the 2022 national diagnosis and care protocol adopted this consensus as its methodological basis [7], and on 16 June 2026 GeneReviews published a comprehensive update that incorporated MAU2 among the causative genes and moved ANKRD11 to the differential diagnosis [1].

The ESCO2 spectrum has a distinct nosological history. In 1919, Roberts described a child with bilateral cleft lip and palate, protrusion of the premaxilla and imperfect development of the bones of all four extremities; Appelt and colleagues reported a second case in 1966, and Freeman and colleagues reviewed 17 previous patients in 1974 and added 5 [31]. In 1969, Herrmann and colleagues described two families with milder phocomelia and survival into adulthood, designated by the initials of their surnames (“SC syndrome” or “pseudothalidomide”), and in 1972 Hall and Greenberg described the “hypomelia–hypotrichosis–facial haemangioma syndrome” [16, 31, 32]. In 1993, Van Den Berg and Francke reviewed 100 cases, described the cytogenetic phenomenon of pericentromeric heterochromatin repulsion and proposed a severity scale [33]. In 2005, Vega and colleagues identified ESCO2 variants, the human homologue of the yeast ECO1 gene, in 15 families with Roberts syndrome, and Schüle and colleagues showed in the same year that SC phocomelia is caused by variants in the same gene [34, 35]. The series of 49 patients published in 2010 established the clinical criteria, and GeneReviews adopted the designation “ESCO2 spectrum”, with Roberts syndrome at the severe end and SC phocomelia at the mild end [16, 36].

This article addresses Cornelia de Lange syndrome as the central entity, with subsections by gene. The ESCO2 spectrum (Roberts syndrome and SC phocomelia), an autosomal recessive cohesinopathy excluded from the Cornelia de Lange spectrum by the 2018 consensus, is described in its own right in the sections on genetics, clinical features, diagnosis, differential diagnosis, management, counselling, epidemiology and follow-up [3, 16]. The other cohesinopathies outside the spectrum (DDX11-related cohesinopathy, STAG1-, STAG2- and WAPL-related disorders) and ANKRD11-related KBG syndrome are addressed insofar as they affect differential diagnosis and counselling [3, 7, 37–39].

2. Molecular Genetics and Pathophysiology

2.1. The cohesin complex and its role in gene regulation

Cohesin is a ring-shaped structure that regulates multiple aspects of chromatin biology, including chromosome segregation, genomic stability, three-dimensional genome organisation and transcriptional regulation [1]. The core proteins of the complex are SMC1A, SMC3 and RAD21, to which STAG1, STAG2 or STAG3 bind directly; NIPBL and MAU2 form a heterodimer (termed kollerin) required to load cohesin onto DNA, BRD4 interacts with NIPBL, and HDAC8 controls the release of cohesin from chromatin by deacetylating SMC3 [3]. Cohesin participates in chromatin loop extrusion and the delimitation of topologically associated domains (TAD, topologically associated domains), allowing distant regulatory sequences to act on promoters [1, 7]. GeneReviews emphasises that some of these functions, such as bringing enhancers close to promoters and promoting RNA polymerase II activity, are also regulated by ANKRD11, EP300, BRD4 and AFF4, explaining the convergence of Cornelia de Lange-like phenotypes when these genes are altered; for this reason, the condition is described as a “cohesinopathy” and a “transcriptomopathy” [1].

Available evidence points to transcriptional dysfunction during development, rather than a mitotic cohesion defect, as the main mechanism underlying the phenotype [7]. HDAC8 deficiency illustrates this conclusion: loss of enzyme activity increases SMC3 acetylation, prevents efficient dissolution of “used” cohesin released from chromatin in prophase and anaphase, reduces occupancy of cohesin-binding sites and produces a pattern of transcriptional alteration shared by cell lines with variants in NIPBL or HDAC8 [25]. BRD4 also binds acetylated histones (H3K27ac) in clusters of enhancers; its immunoprecipitate contains NIPBL, and a missense variant that retains NIPBL binding but loses affinity for acetylated histones was associated with a more typical phenotype, suggesting that NIPBL sequestration is part of the mechanism [3, 27]. In 2025, a structural study demonstrated that an N-terminal ANKRD11 peptide binds cohesin with high affinity and competes with the CTCF factor; the experimental Tyr347Ala substitution (site-directed mutagenesis, with no corresponding clinical variant) abolished this interaction, disrupted gene expression in mouse embryonic stem cells and reproduced, in mice, neural and craniofacial abnormalities resembling those of KBG syndrome, providing a mechanistic basis for the overlap between the two syndromes [40].

Table sources: proportions, loci and mechanisms according to GeneReviews; MANE Select transcripts confirmed in VariantValidator [1, 41, 42].

2.2. NIPBL: the main gene and denominator heterogeneity

NIPBL is the most frequent cause of the syndrome, but the attributed proportion varies with the denominator: approximately 80% of all cases in the 2026 GeneReviews, approximately 70% in the 2018 consensus and Orphanet summary, 65% point variants plus 5% deletions in Smith, 65% in the Brazilian reference text, approximately 60% in the Oxford Desk Reference, and “almost 60% of individuals with the classical form” in a 2012 study [1–3, 19, 25, 43, 44]. The cohort of 716 probands analysed by Kaur and colleagues in 2023 attributed more than 60% of cases to NIPBL and identified it as the only gene whose variants produce the severe or classical form [45]. The differences reflect inclusion criteria (classical or expanded spectrum), detection methods (blood or other tissues) and the inclusion of mosaic cases. Approximately 97% of pathogenic NIPBL variants are detected by sequencing and approximately 3% by intragenic deletion and duplication analysis [1]. In addition to coding variants, genome sequencing of previously “negative” individuals identified potentially causative non-coding variants in four people, clustering in an upstream open reading frame (uORF) and a small region of intron 21 [46].

Loss-of-function variants cause a more severe phenotype than missense variants, which tend, although not always, to be associated with a less pronounced presentation; missense variants and in-frame deletions are associated with less severe growth restriction, milder intellectual disability and fewer structural abnormalities [1, 3]. Truncating NIPBL variants are particularly associated with major limb defects [3]. The French protocol adds that nonsense variants and frameshift deletions are associated with greater cognitive severity [7], although the international consensus notes that the type or location of the NIPBL variant was not associated with the degree of intellectual disability in most studies, with one exception that found less severe disability with missense variants and in-frame deletions [3]. Even in mosaicism, missense substitutions preferentially occur in the HEAT-repeat domain of NIPBL [47].

2.3. SMC1A: two opposite allelic phenotypes

SMC1A encodes a core cohesin subunit, is located on the X chromosome and is not fully inactivated; variants causing the Cornelia de Lange phenotype are typically missense variants or in-frame deletions and account for approximately 5% of cases [1, 3]. The series establishing the gene described three boys from one family and one sporadic case [23], and structural analysis of the first 14 additional variants showed that all preserved the reading frame and probably produced functional complexes with altered binding dynamics, in a consistently mild phenotype without major structural abnormalities [24]. In an international study of 51 individuals with SMC1A variants compared with 67 with NIPBL variants, growth was less impaired, facial signs were less pronounced (except the periorbital signs and thin upper vermilion), there were no major limb abnormalities, and cognitive and adaptive levels were higher; in the Dutch subgroup, 5 of 13 individuals, all female, had a Rett-like presentation, with epileptic encephalopathy and severe or profound intellectual disability [48]. Truncating SMC1A variants do not cause Cornelia de Lange syndrome: among the 10 female patients with de novo truncation described by Symonds and colleagues, none had that clinical diagnosis, and all had epilepsy beginning between less than 4 weeks and 28 months, with a tendency towards seizure clusters associated with regression and moderate to severe developmental impairment; the authors consider that these variants are probably non-viable in males [49]. This entity corresponds to developmental and epileptic encephalopathy type 85, with or without midline brain defects (OMIM 301044) [1].

2.4. SMC3: dominant-negative effect and haploinsufficiency

SMC3 variants causing the Cornelia de Lange phenotype are typically missense; modelling of 16 patients suggested a dominant-negative effect on the multimeric complex, with postnatal microcephaly, less distinctive facies, milder prenatal growth restriction worsening in childhood, few cardiac defects and no limb deficiencies, and confirmed that de novo variants explain 1–2% of overlapping phenotypes [50]. Conversely, predicted loss-of-function variants and whole-gene deletions produce a milder presentation: in 14 individuals, findings converged on low growth parameters, developmental delay or intellectual disability and dysmorphism “reminiscent of the atypical form”, with some individuals lacking growth, developmental or dysmorphic abnormalities; the gene behaves as an archetypal haploinsufficient gene in population data, and the epigenetic profile of these carriers did not consistently cluster with that of the typical syndrome [51]. GeneReviews records this presentation as an allelic disorder with variable penetrance [1].

2.5. RAD21, HDAC8, BRD4 and MAU2

Children with RAD21 variants have growth restriction, minor skeletal abnormalities and overlapping facies, with much milder cognitive impairment than that observed with NIPBL, SMC1A or SMC3; initial functional studies showed that variants act at the interface with STAG2 and SMC1A, impair the DNA damage response and disrupt transcription in a zebrafish model, and suggested that dominant missense variants may have more severe functional effects than loss-of-function variants [26]. The series of 49 individuals from 33 families (24 intragenic variants and 7 microdeletions) described an attenuated phenotype in facial morphology, limbs, cognition and behaviour, with frequently familial variants and marked inter- and intrafamilial variability; patients with biallelic variants had Mungan syndrome, and others had holoprosencephaly with or without Cornelia de Lange features [52]. In that series, 10% had normal cognition, 45% mild impairment, and none severe or profound disability [1]. Approximately 90% of pathogenic RAD21 alterations are detected by sequencing and approximately 10% by deletion analysis; exon 14 lies within a segmental duplication, which has technical implications [1]. RAD21 (8q24.11) is also included in the contiguous gene deletion of Langer–Giedion syndrome, between TRPS1 and EXT1.

HDAC8 encodes the SMC3 deacetylase. In a cohort of 38 individuals, most variants were missense and de novo, all caused loss of enzymatic function, many heterozygous carriers had markedly skewed X-inactivation in blood, and the eight hemizygous males were more severely affected; Cornelia de Lange syndrome had not been considered before genomic testing in several cases [53]. The facies overlaps with the typical pattern but includes delayed closure of the anterior fontanelle, hypertelorism, hooded eyelids, a broader nose and dental abnormalities, with a personality described as cheerful or affable [1, 53]. Most heterozygous females show skewing of X-inactivation towards the normal allele, and their severity depends on the inactivation pattern; they may have mosaic skin pigmentation [1, 3].

BRD4 was implicated following a de novo deletion that included it in a patient with an atypical phenotype, followed by the identification of de novo intragenic variants; haploinsufficiency produces an overlapping phenotype, and a missense variant was associated with a more typical presentation [3, 27]. A 2025 review places BRD4 alterations within non-classical or milder forms [54]. MAU2 was initially associated with an in-frame deletion of seven amino acids that impairs interaction with the N-terminus of NIPBL [28]. In 2026, the description of 18 individuals with 15 heterozygous variants showed that in-frame variants mainly impair the NIPBL-MAU2 interaction, whereas truncating variants cause MAU2 haploinsufficiency and secondary reduction of NIPBL; most patients have a methylation profile consistent with the syndrome's episignature, two variant-type-dependent MAU2-specific episignatures were described, and a heterozygous mouse model reproduced short stature and microcephaly [29]. The phenotype ranges from the classical form to mild presentations in heterozygous parents, with short stature (90%), microcephaly (90%), Cornelia de Lange facies (70%) and fifth-finger clinodactyly, small feet, hirsutism and intellectual disability (50% each), without oligodactyly or severe malformations described [1].

2.6. ANKRD11 and the boundary with KBG syndrome

The 2018 consensus considered ANKRD11 the seventh causative gene, based on five de novo variants in individuals with a non-classical phenotype and unpublished observations by the authors [3]. Ansari and colleagues had already identified three de novo ANKRD11 variants in a cohort with Cornelia de Lange syndrome and similar phenotypes, demonstrating overlap with KBG syndrome [55], and Smith refers to five individuals, all with a non-classical phenotype [19]. In genome sequencing of 105 previously undiagnosed families, ANKRD11 was the second most frequent locus (6 probands), after NIPBL (10) [46]. The 2026 GeneReviews includes ANKRD11 (a pathogenic variant or a 16q24.3 deletion including it) in the differential diagnosis and recommends that the panel include it [1], and the French protocol and Orphanet summary also treat it as a differential diagnosis whose inclusion in the panel is necessary [2, 7]. In practical terms, an ANKRD11 variant leads to a molecular diagnosis of KBG syndrome, which has different surveillance requirements (section 6) [39].

2.7. Postzygotic mosaicism and haematopoietic selection

Mosaicism is frequent in Cornelia de Lange syndrome: the international consensus estimates that 15–20% of individuals with a classical phenotype have mosaic NIPBL variants not detected in lymphocytes, and GeneReviews reports somatic NIPBL mosaicism in approximately 10–15% [1, 3]. In 2013, a buccal swab study detected an NIPBL variant in 10 of 13 individuals without a variant in lymphocytes and estimated somatic mosaicism at 23% (10/44) of patients with a reliable clinical diagnosis, without clinical differences between constitutional and mosaic variants [56]. The 2021 review calculated a mosaicism prevalence of 13.1% among patients with a positive molecular diagnosis, ranging from 7.9% to 27% across cohort studies; in a Spanish cohort of 43 patients with a clinical diagnosis, 39 (90.7%) obtained a molecular diagnosis, and 4 of these (10.26%) had NIPBL variants detected only in saliva or fibroblasts [47]. In 29 of 38 published cases of NIPBL mosaicism with more than one tissue studied, the variant was absent or present at a very low allele frequency in blood, which the authors interpret as negative selection against the variant in haematopoietic cells; most mosaic individuals have a phenotype at least as severe as constitutional carriers [47]. Selection against haematopoietic cells expressing the variant has been described for HDAC8 [3], and in SMC1A mosaicism was described in two boys and, in 2025, in a woman with the classical form, with similar allele frequencies in blood and buccal mucosa (approximately 30%) (section 10) [57]. SMC3 and RAD21 mosaicism is rare [3].

2.8. Epigenetic signature

An epigenetic signature (a disease-specific DNA methylation profile in peripheral leucocytes) has been identified in the syndrome associated with NIPBL, RAD21, SMC3, SMC1A and MAU2 [1, 29, 58]. The 2020 study validating 34 robust episignatures in 42 syndromes demonstrated their usefulness in resolving ambiguous cases and the need for multiclass models to avoid misclassification between closely related signatures [58]. There is also a specific episignature for ANKRD11-related KBG syndrome, which GeneReviews for that syndrome proposes to clarify suggestive cases without an identified variant or with a variant of uncertain significance, giving it utility in the differential diagnosis [39].

2.9. ESCO2 spectrum (Roberts syndrome and SC phocomelia)

ESCO2 is located at 8p21.1 and encodes the N-acetyltransferase ESCO2, a 601-amino-acid protein of the Eco1/Ctf7 family [16, 59]. ESCO2 associates with the replicative MCM helicase and with PCNA, travels with the replication fork and acetylates lysine residues of the SMC3 subunit of cohesin, which stabilises the binding of the ring to DNA and holds newly replicated sister chromatids together [16, 60]. The difference from Cornelia de Lange syndrome is mechanistic: in the latter, sister chromatid cohesion is as a rule preserved and transcriptional dysregulation predominates, whereas in ESCO2 deficiency centromeric cohesion fails in a way that is visible under the microscope [3, 16]. Without ESCO2, metaphase chromosomes lose the primary constriction; the arms remain joined and allow alignment, but cells enter anaphase without centromeric cohesion, with inefficient division, aneuploidy, micronuclei and increased apoptosis of progenitor cells [16, 60]. Patient cells also have an altered gene expression profile, whose clinical significance has not been established [16].

In a mouse model with conditional inactivation of Esco2 in limb mesenchyme (Esco2fl/fl;Prrx1-Cre), limb reduction resulted from a pre-apoptotic mesenchymal population with activation of the p53 pathway from embryonic day 9.5, followed by vascular defects and limb haemorrhage at day 12.5; in utero treatment with pifithrin-α, a p53 inhibitor, corrected the haemorrhage [61]. The altered genes overlapped with those associated with thalidomide embryopathy and other genetic limb reductions, suggesting a common vascular pathway [61]. This is preclinical evidence, with no therapeutic application in humans.

The mechanism is loss of function. By the 2025 GeneReviews update, 35 different pathogenic variants had been described, mostly small insertions and deletions with a premature stop codon; missense variants are rare and no large rearrangements have been described [16]. Two missense variants in the acetyltransferase domain reduced acetylation activity in vitro [36]. The following table presents the recurrent and founder variants in current nomenclature, normalised in VariantValidator on the MANE transcript and with the ClinVar classification [16, 62, 63].

No genotype–phenotype correlation has been established. Schüle and colleagues found only truncating variants, regardless of clinical severity, and variability is great even among siblings [16, 35]. In the series of 49 patients, the severity of facial anomalies paralleled that of limb reduction, and corneal opacities appeared to be associated with specific variants [36]. In situ hybridisation in human embryos showed ESCO2 expression in the brain, face, limbs, kidney and gonads, corresponding to the affected structures [36].

As for allelic disorders, GeneReviews states that no other phenotype has been associated with germline ESCO2 variants [16]. In 2020 and 2021, however, Kantaputra and colleagues described the homozygous variant NM_001017420.3:c.1654C>T, p.(Arg552Ter), in one patient and in two siblings of the Lisu ethnic group, in Thailand, with the picture of Juberg–Hayward syndrome (orocraniodigital syndrome; OMIM 216100; ORPHA:2319): cleft lip and palate, ptosis, short stature, hypoplastic thumbs, humeroradial synostosis or radial head dislocation and normal intelligence; the karyotype showed premature centromere separation [64, 65]. The authors consider the two entities allelic and distinct; microcephaly was not constant [64, 65]. This association explains why curation databases relate ESCO2 to the code ORPHA:2319, which corresponds to Juberg–Hayward syndrome and not to Roberts syndrome (ORPHA:3103) [4, 18].

3. Clinical Spectrum and Manifestations by System

3.1. Classical form, non-classical form and previous classifications

The classical form is usually recognisable from birth by experienced paediatricians and geneticists, but presentation ranges from mild to severe forms, with differing degrees of facial and limb involvement [3]. The Brazilian reference text reproduces the classification of Van Allen and colleagues (1993), which distinguished type I or classical, with prenatal and postnatal growth restriction, moderate to severe cognitive impairment, characteristic dysmorphism and severe malformations already visible at birth; type II or moderate, with typical facies, mild to moderate impairment of growth and cognition and subtle or absent malformations, sometimes recognised only at 2–3 years; and type III, corresponding to phenocopies caused by environmental exposure (such as alcohol during pregnancy) or chromosomal alterations [43]. In practice, this historical classification has been replaced by the 2018 consensus distinction between classical and non-classical forms and molecular classification by gene [1, 3]. Expressivity tends to be relatively consistent within a family, according to most families with more than one affected member, although intrafamilial variability is marked in SMC1A, HDAC8 and RAD21 [1, 3].

3.2. Frequency of manifestations

ASD, atrial septal defect; VSD, ventricular septal defect. GeneReviews frequencies derive mainly from Jackson et al. (1993) and the 2018 consensus and refer predominantly to clinically diagnosed cohorts; Smith percentages refer to classical cases [1, 19].

3.3. Growth, puberty and body composition

Growth restriction has prenatal onset in most patients; slow, symmetrical growth produces proportionate short stature that becomes more evident from six months, and in the classical form mean height and weight remain below the 5th percentile throughout life [1]. Birth weight is usually below the 5th percentile, and height, weight and head circumference remain below values in the general population [3]. The French protocol reports birth length and weight frequently below the 3rd percentile [7], and the Brazilian text notes that approximately one-third of children are born prematurely and/or with intrauterine growth restriction [43]. In the EUROCAT registry, almost 70% of term newborns weighed ≤2500 g, and low weight was associated with a more severe phenotype [6]. Smith reports delayed bone maturation (100%) and mean adult heights of 156 cm in males and 131 cm in females [19]. Growth is less affected in mild forms, mosaic cases and SMC1A than in NIPBL, and secondary weight faltering due to reflux disease and feeding difficulties may coexist [1, 3]. Specific growth charts exist, constructed from clinically diagnosed patients without molecular stratification [1, 3].

Growth hormone secretion is normal in most children; a single patient with an NIPBL variant, low growth hormone levels and accelerated growth following supplementation has been described [3]. Puberty occurs in most, slightly delayed (mean onset at 15 years in boys and 13 in girls), with menarche on average one year later than in the general population, 5% of girls without menstruation, frequently irregular cycles, and breast development in approximately 80% [3]. Smith reports puberty at the normal age, sometimes incomplete, with a tendency towards obesity and some evidence of premature ageing in adults [19]. More than 30% of adults are overweight and at least 50% are considered obese, and 4% develop type 2 diabetes mellitus, although whether these proportions exceed those in people with the same cognitive and mobility levels is unclear [3].

3.4. Craniofacial features and skin

The facies includes microcephaly (mean head circumference below the 2nd percentile), synophrys with arched and/or thick eyebrows, long, thick eyelashes, a short nose with a depressed nasal bridge and anteverted nares, a long and/or smooth philtrum, thin upper vermilion, downturned mouth corners, a high palate with or without cleft, small, widely spaced teeth and micrognathia with or without mandibular spurs [1]. Smith adds a depressed nasal bridge (83%), anteverted nares (85%), high palate (86%), delayed eruption of widely spaced teeth (86%), thick, dysplastic, posteriorly rotated ears, low posterior hairline (92%), short neck (66%) and prominent mental symphysis (66%) [19]. Hypertrichosis involves thick hair extending to the temporal regions and sometimes the face, ears, back and arms [1]; cutis marmorata and hypertrichosis affect up to 60% and 80% of patients, respectively [7]. Hypoplastic nipples and umbilicus occur in 50% [1]. The facies is similar in adults and younger individuals, although some appear older than their chronological age [3]. The consensus describes the cardinal facial signs as synophrys and thick eyebrows, a short nose with a concave ridge and anteverted tip, a long, smooth philtrum, thin upper vermilion and downturned mouth corners; its link to consensus figure 3 allows these features to be visualised (https://pmc.ncbi.nlm.nih.gov/articles/PMC7136165/figure/Fig3/) [3].

3.5. Limbs and skeleton

Limb abnormalities, symmetrical or asymmetrical, range from severe deficiencies with complete absence of the elbow and forearm, through forms of oligodactyly mainly affecting the upper limbs, to small hands and phalangeal changes (fifth-finger clinodactyly and a short first metacarpal with a proximally placed thumb) [1]. Severe upper-limb abnormalities occur in 25% of patients, and deficiencies ranging from absent forearms to forms of oligodactyly are present in approximately one-third of individuals with classical features; in the absence of deficiency, micromelia, proximally placed thumbs and fifth-finger clinodactyly are present in almost all, and radioulnar synostosis is common and may cause elbow flexion contractures [1]. In patients with NIPBL variants, major upper-limb abnormalities are most frequent (25%), whereas they are rare in carriers of SMC3, HDAC8 and RAD21 variants and absent in SMC1A [3]. Major abnormalities are almost exclusive to the upper limbs, more frequent in males and asymmetrical in 65% of cases (with the right side more affected in 75% of these); underdevelopment and dislocation of the radial head are present in 79%, and minor abnormalities (proximal thumbs, clinodactyly) in 65–85% [3]. In the EUROCAT registry of classical cases, limb abnormalities were the most frequent associated malformation (73.1%), and severe abnormalities were significantly more frequent in males [6].

In the lower limbs, major malformations (absence of the tibia or fibula, split foot) are rare; minor length differences occur in 46%, Legg-Calvé-Perthes disease in 4%, and hip dislocation in 10% later in life, mainly in wheelchair users or bedridden people [3]. Cutaneous syndactyly between the second and third toes occurs in more than 80%, and bunions are reported by 75% of adults [1, 3]. Scoliosis, predominantly thoracic, develops in one-third, frequently by 10 years, and is more common in adults with reduced mobility; flexion contractures (knees, elbows, hip) occur in 18–25% [3]. Radiologically, Smith describes a mandibular spur by 3 months, a dislocated or hypoplastic radial head, hypoplastic first metacarpal and middle phalanx of the fifth finger, a short sternum with early fusion, 13 pairs of ribs, enlarged cerebral ventricles and white matter atrophy [19]. The presentation is illustrated in figures 1 to 3 of the Smith chapter (four affected individuals; clinodactyly and proximal thumb; mildly affected mother with severely affected children) [19].

3.6. Gastrointestinal system and feeding

Feeding difficulties are present in almost all newborns and infants and frequently persist into childhood and adulthood; cleft palate (including submucous cleft, 20%), micrognathia and dental problems contribute [3]. Gastro-oesophageal reflux disease is the most prevalent and severe gastrointestinal problem: in infants it may manifest as Sandifer-like dystonic episodes, feeding refusal, recurrent chemical pneumonitis, poor weight gain, agitation or disturbed sleep, may explain some self-injurious behaviour, and tends to persist or worsen [3]. Reported frequencies include 71% with NIPBL variants and 60% with SMC1A variants in a questionnaire, 55% in 38 individuals with NIPBL variants, endoscopic oesophagitis in 65% of 43 patients aged 6 to 32 years, and, in 49 adolescents and adults, confirmed reflux disease in 75% with Barrett's oesophagus in 10%, metaplasia in 9%, eosinophilia in 16% and strictures in 12% [3]. Approximately one-third have evidence of aspiration and 15% require a feeding tube [1]. Oesophageal adenocarcinomas have been described in young adults with longstanding reflux [3, 19].

Gastrointestinal malformations include duodenal atresia, annular pancreas, imperforate anus, Meckel's diverticulum and congenital diaphragmatic hernia; hypertrophic pyloric stenosis has been described in up to 7%, and inguinal hernia is common in childhood [3]. GeneReviews reports pyloric stenosis in 4% (the most frequent cause of persistent vomiting in the neonatal period), intestinal malrotation in 2%, and diaphragmatic hernia in 1%, the latter possibly underdiagnosed in newborns dying in the perinatal period [1]. The consensus reports malrotation in 5% of 73 and 10% of 49 individuals, usually manifesting as acute caecal volvulus, sometimes recurrent, in infancy, childhood or puberty, and difficult to diagnose because of atypical symptoms and communication difficulties [3]. Constipation and rumination occur in 10–15% of adults (32% with NIPBL, 43% with SMC1A and 22–46% in clinically diagnosed patients, depending on the study); diarrhoea (18%), flatulence (48%) and lactose intolerance (18%) are also common, without an evident association with coeliac disease [3].

3.7. Heart, kidneys and genital system

Approximately 30% have congenital heart disease; in descending order: pulmonary or peripheral pulmonary artery stenosis, ventricular septal defect, atrial septal defect, coarctation or hypoplasia of the aortic arch, aortic valve abnormality, tetralogy of Fallot, double-outlet right ventricle and atrioventricular septal defect [1]. The consensus reports 25% cardiac abnormalities [3], the French protocol 25–30%, possibly more frequent with HDAC8 and NIPBL and in moderate and severe forms [7], and the EUROCAT registry 45.6% among classical cases [6]. Cardiac defects associated with SMC3, generally mild, occur in approximately 56% of these patients and are less frequent with SMC1A [1]. Smith occasionally reports hypertrophic cardiomyopathy and late-onset dysplastic valvular disease [19]. NIPBL-related syndrome also appears in the list of genes and syndromes associated with left ventricular non-compaction in a cardiogenetics textbook [66].

Renal abnormalities affect approximately 10% (renal hypoplasia or dysplasia, corticomedullary dysplasia) [3, 7], and vesicoureteric reflux has been described in 12% [1]. The Brazilian text reports vesicoureteric reflux in 40%, dysplastic kidney and calyceal dilatation [43]. Cryptorchidism occurs in 73% of boys, genital hypoplasia in 57% and hypospadias in 9% [1]; the consensus reports 80% cryptorchidism, 37% small penis and 9% hypospadias, with no description of voice change at puberty [3]. Bicornuate uterus was observed in 5 of 20 patients in GeneReviews and in 19% of female patients in the consensus, potentially causing abdominal pain and a slightly increased risk of miscarriage and preterm delivery [1, 3, 7].

3.8. Eyes

In addition to synophrys (99%), long eyelashes (99%) and eyebrow hypertrichosis (96%), clinically relevant ocular manifestations include ptosis (44%), epiphora (22%), nasolacrimal duct obstruction (16%), blepharitis (25%), myopia (58%), peripapillary pigmentation (83%) and microcornea (21%) [1]. The consensus reports unilateral ptosis in 37% and bilateral ptosis in 44%, compensatory chin elevation in 57%, nasolacrimal obstruction in 24% with HDAC8 and 67–80% with NIPBL, visual impairment in 44–53%, spherical equivalent >5.0 dioptres in 38% and >10.0 dioptres in 9%, hypermetropia in 15%, nystagmus in 14–17% and strabismus in 16–26% (esotropia more frequent than exotropia) [3]. The peripapillary pigment ring, present in up to 83%, is a benign finding without clinical consequences, but retinal detachment may occur through high myopia or self-inflicted trauma [3]. In a Brazilian cohort of 16 patients with NIPBL variants (5 to 44 years), all had ocular abnormalities: myopia in 75% (high in 18.8%), astigmatism ≥2.5 dioptres in 50%, strabismus in 37.5%, nystagmus in 18.8%, iris atrophy in 56.2%, and tear film break-up time below 10 seconds in 7 of 8 assessable patients [67].

3.9. Ear, nose and throat

Hearing loss is very frequent (85–90%), usually bilateral and present from childhood, with severity ranging from mild to severe (40–50%); it is sensorineural in 25% and conductive in 75% according to the consensus, and in adults sensorineural hearing loss is reported in 45% [3]. GeneReviews cites a series with sensorineural hearing loss in 40% and conductive hearing loss in 60% [1]. Conductive hearing loss frequently results from persistent otitis media with effusion (80–85%), and the auditory canals are small and stenotic in approximately one-third; middle- and inner-ear abnormalities include malformed ossicles, small mastoids, cochlear and vestibular abnormalities and soft-tissue opacification of the tympanomastoid cavity, which correlates with audiometric findings [3]. Hearing improves over time in approximately 50% of adults, including some with severe hearing loss, justifying longitudinal assessments [1, 3]. One-third have recurrent sinusitis, and nasal polyps have been described [3]. Intubation may be difficult because of a small mouth, micrognathia, short neck, temporomandibular joint stiffness and cleft palate [3].

3.10. Nervous system, sleep and autonomic function

Approximately 25% have seizures; focal epilepsy is most common, usually begins before 2 years, responds to standard therapy, and treatment can frequently be discontinued after a few years [1]. Frequency varies by gene (45% with SMC1A, 15% with NIPBL and 20–26% without molecular confirmation); 35 of 39 individuals with complete data responded well to standard therapy [3]. In a study of 47 patients, 35% had asymmetrical sympathetic responses in the lower limbs and abnormal sudomotor tests, which may contribute to thermoregulation difficulties [1]; in a questionnaire of 65 patients aged ≥8 years, autonomic function was mildly impaired in 81% and markedly in 26% [3]. Structural brain abnormalities (cortical malformations, small corpus callosum, white matter, cerebellar and brainstem changes) occur mainly with NIPBL variants, and tethered cord is rare and usually asymptomatic [3]. Sleep disturbances have highly variable frequencies (12–72%), including nocturnal apnoea, insomnia, daytime sleepiness and frequent naps; difficulties initiating and maintaining sleep affect 55–65% from the first year, and 30% of 74 clinically diagnosed patients had consecutive days without sleep [3]. Smith describes early hypertonia, a low-pitched, weak infant cry (74%), high pain tolerance and a broad-based gait [19]; the characteristic low-pitched cry occurs in 75% and is associated with a more severe phenotype [1].

3.11. Cognition, communication and behaviour

Intelligence quotient ranges from less than 30 to 102, with a mean of 53, and individuals with classical features are more likely to have severe to profound disability [1]. The consensus supplementary tables show heterogeneity across series: in a study involving treating physicians, patients with NIPBL variants (n=58) were distributed across profound (19%), severe (47%), moderate (28%) and mild (7%) disability, and patients with SMC1A (n=20) across profound (5%), severe (25%), moderate (40%), mild (20%) and normal (10%) [3]. Expressive language is more impaired than receptive language, and the latter more than cognition [1]. Half of children walk at 24 months and 95% by 10 years; half feed themselves at 3 years and 95% by 10 years [1]. In a series of 51 children, those with SMC1A variants achieved some milestones earlier than those with NIPBL, and of the latter, at 5 years, 99% sat, 63% walked unsupported and 38% had begun to speak [3]. Specific executive function deficits (mental flexibility, short-term visual memory) may occur beyond those expected from cognitive level, and adaptive behaviour is impaired throughout life, more with NIPBL than with other causes [3].

Clinically significant self-injurious behaviour occurs in 56%, predominantly with hand-directed injuries; risk markers include more severe cognitive, communication and adaptive impairment, NIPBL variants, impulsivity, repetitive behaviours and autistic traits [1, 3]. The prevalence of autism spectrum symptomatology is 43%, and differences from idiopathic autism include social anxiety, extreme shyness and selective mutism, which intensify with age and social demands [3]. Anxiety is frequent as a primary condition (generalised anxiety, separation anxiety, selective mutism) and as an exacerbating factor for repetitive behaviours, mood disorders and aggressive or self-injurious behaviours; repetitive behaviours include stereotyped movements, insistence on sameness and rituals, with lining up and arranging objects being characteristic [3]. Cross-sectional studies suggest age-related worsening of anxiety, depressed mood, autistic traits, self-injurious behaviour and impulsivity, with adolescence and early adulthood being periods of marked change; the same was not observed for aggression, hyperactivity or sleep [3]. Smith also reports hyperactivity, attention deficit, aggression, depression, obsessive-compulsive behaviours, perseveration and circadian rhythm disturbances [19]. Sensory processing difficulties (hypo- and hypersensitivity) occur at all levels of intellectual disability [3].

3.12. Immunology and haematology

Antibody deficiency has been described in several patients, justifying immunological screening and management in the presence of severe or recurrent infections (chronic otitis, chronic viral respiratory infections, pneumonia); T-cell dysfunction may accompany antibody deficiencies [1]. Recurrent respiratory infections are mainly attributed to anatomy, hypotonia and incoordination of swallowing and coughing, and immunological abnormalities are occasional [3]. Thrombocytopenia frequently resolves after the first year and rarely progresses to persistent idiopathic thrombocytopenic purpura; anaemia is usually transient [1]. The consensus considers thrombocytopenia generally non-progressive and asymptomatic [3].

3.13. Profile by gene

++++, ≥90%; +++, 70–89%; ++, 50–69%; +, 20–49%; −, <20%; NR, not reported. Selected rows from table 2 of the 2018 consensus, also reproduced in appendix 1 of the French protocol; MAU2 was not included [3, 7].

3.14. ESCO2 spectrum: clinical manifestations and natural history

The picture combines growth restriction of prenatal onset, limb malformations (bilateral symmetrical tetraphocomelia or hypomelia due to mesomelic shortening), hand anomalies, joint contractures, craniofacial anomalies and, frequently, ocular abnormalities; intellectual disability is common but not constant [16]. Published frequencies depend on the series and the denominator, as the table shows.

Sources: [16, 31].

The limbs follow a cephalocaudal and anteroposterior pattern: the upper limbs are more affected than the lower, and the frequency and degree of involvement decrease from the radius to the ulna and humerus, and from the fibula to the tibia and femur [16, 36]. In the hands, oligodactyly with aplasia or hypoplasia of the thumb predominates, followed by clinodactyly and hypoplasia or agenesis of the fifth finger; in severe forms there are only three fingers and, rarely, one [16]. In mild forms, synostoses of the knees and ankles and humeroradial synostosis are more frequent [16]. Radiology shows findings ranging from tetra-amelia, tetraphocomelia and tetrameromelia to mild hypoplasia of the long bones, humeroradial, humeroulnar and femorotibial synostoses, oligodactyly, syndactyly, equinovarus or calcaneovalgus feet and, in some cases, Wormian bones [32].

The face is characterised by cleft lip and/or palate with a prominent premaxilla, micrognathia (77%), brachycephaly (63%), downslanting palpebral fissures (57%), exophthalmos due to shallow orbits (59%), hypoplasia of the malar region (88%) and of the alae nasi (77%), bluish sclerae and malformed ears with hypoplastic lobules; the hair is sparse, sometimes silvery-blond [16, 31]. Patients with mild limb involvement as a rule have mild facial malformations, without a cleft or with only a high-arched palate, and the most severely affected may have a fronto-ethmoido-naso-maxillary encephalocele [16]. Ocular abnormalities include microphthalmia, nystagmus, glaucoma and corneal opacities, sometimes with lens involvement [16].

Intellectual disability segregates by family and appears to be associated with corneal opacities; there are families with severe phocomelia and bilateral cleft lip and palate, without corneal opacities and without intellectual disability, and normal intelligence has been documented in several patients [16]. The combination of intellectual disability and corneal opacities is associated with an increased risk of intracranial aneurysms [16]. Moyamoya disease (2%) and stroke secondary to intracranial aneurysms (4%) have been described in late adolescence and young adulthood, and two patients had third cranial nerve palsy due to a cavernous angioma [16]. In 2023, a Chinese adolescent girl with biallelic variants NM_001017420.3:c.1220A>T, p.(His407Leu), and NM_001017420.3:c.1562del, p.(Ala521ValfsTer15), was described; she had an ischaemic stroke at 6 years of age and died at 13 years from a cerebellar haemorrhage due to aneurysm rupture; imaging showed the triad of leukoencephalopathy, calcifications and cysts, and histology revealed angiomatous changes and perivascular cysts suggestive of chronic small vessel vasculopathy [62, 68].

Severely affected patients have increased susceptibility to infection and mortality from sepsis, and the risk may extend to the end of the first year of life [16]. Extracranial neoplasms have been reported (one malignant melanoma and one rhabdomyosarcoma) [16]. In adults, the ten cases then published showed mainly the typical limb defects, craniofacial anomalies, growth delay and intellectual disability, with aortic stenosis, ocular abnormalities (cavernous haemangioma, paracentral scotoma, tilted and atrophic optic nerve) and a possible increased risk of veno-occlusive disease [31]. A 15-year-old Malay adolescent girl, homozygous for the splice variant NM_001017420.3:c.955+2_955+5del, developed type 2 diabetes mellitus, premature ovarian insufficiency and increased optic disc cupping with tortuous vessels and hypopigmented patches, without learning difficulties [59, 62].

4. Clinical Recognition and Warning Signs

4.1. When to suspect the condition

In the neonatal period, the combination of markedly low weight for gestational age, microbrachycephaly, synophrys, long eyelashes, a short nose with anteverted nares, a long philtrum, a thin upper lip, hypertrichosis and upper-limb abnormalities should raise suspicion [1, 44, 69]. In infants and children without limb defects, hand radiographs showing a short first metacarpal (proximally placed thumb) and a short fifth finger with clinodactyly are useful [1]. In the approach to hypertrichosis, the Oxford Desk Reference emphasises that babies with the syndrome are usually very small for gestational age [44]; in the approach to radial defects, it identifies very low birth weight and cleft palate as clues, and includes the syndrome in the differential diagnosis of fetal or neonatal death with limb defects [70]. Among limb reduction defects, upper-limb defects are much more frequent than lower-limb defects, and minor hand abnormalities include proximal thumbs, a single palmar crease, clinodactyly and occasionally fusion of digits or polydactyly [71]. Mild forms may only be recognised at 2–3 years [43], and some phenotypes (HDAC8, SMC1A, RAD21, BRD4, MAU2) are only identified through genomic sequencing, without prior clinical suspicion [1, 53].

Clues suggesting a specific gene include a large anterior fontanelle, hypertelorism, hooded eyelids, a cheerful personality and mosaic pigmentation in a female (HDAC8); fuller, less arched eyebrows, a rounder face, absence of major limb defects and more frequent epilepsy (SMC1A); subtle or absent synophrys, a bulbous nose and a long but well-formed philtrum (SMC3); and relatively preserved cognition with overlapping facies (RAD21) [1, 3, 53].

4.2. Warning signs of complications

The consensus and its accompanying emergency card identify potentially fatal acute complications: intestinal obstruction and volvulus, aspiration pneumonia due to reflux or dysphagia, seizures, cardiac problems, urinary tract infections, retinal detachment and a small airway with anaesthetic risk [3]. Any acute abdominal presentation, at any age, should prompt consideration of intestinal malrotation [3]. The French protocol adds that any acute or subacute behavioural change in a person with intellectual disability should raise suspicion of a somatic emergency [7]. Recurrent respiratory infections should prompt exclusion of reflux and aspiration, and reflux disease should always be considered in challenging behaviour, given its variable presentation [3]. Self-injurious behaviour may be a sign of or response to pain associated with reflux, otitis media, constipation, dental disease or hip problems, and pain may go unrecognised because of communication limitations; specific pain assessment tools for people with intellectual disability are recommended, such as the FLACC scale (face, legs, activity, cry, consolability) [3]. Smith identifies aspiration in infancy, apnoea, complications of intestinal obstruction, diaphragmatic hernia and cardiac defects as the main survival risks [19].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Clinical criteria of the 2018 consensus

The international consensus classified classical features into cardinal signs, considered the most characteristic, and suggestive signs, which support diagnosis but are less specific (recommendation R1), and defined a score based on these signs (R2) [3]. A score ≥11 confirms a diagnosis of the classical form, regardless of whether a pathogenic variant in a known gene is found [3]. The French protocol and Smith reproduce the same system [7, 19].

SD, standard deviation. Source: box 1 and recommendation R2 of the international consensus [3].

The criteria were tested in 75 individuals with an NIPBL variant (62 previously classified as classical and 13 as non-classical): classical cases scored 12–16 (mean 13.5; 3–5 cardinal signs) and non-classical cases 9–11 (mean 10.0) [3]. In 46 individuals with an SMC1A variant, the 40 with a Cornelia de Lange phenotype scored 2–13 (mean 8.0) and the 6 with a Rett-like phenotype scored a mean of 3.5; no individual with the syndrome phenotype and molecular confirmation in NIPBL or SMC1A would have been missed [3]. In specificity assessment in patients with Coffin-Siris, Rubinstein-Taybi or Nicolaides-Baraitser syndromes, none scored as classical and 4–10% scored as non-classical; the authors emphasise that true specificity can only be determined in a dedicated study [3]. Existing severity scales should be used cautiously because they do not reflect severity as experienced by families or involvement of all organs (R3) [3]. Before 2018, the consensus criteria reproduced in the Brazilian text required typical facies with synophrys and at least three of eight facial signs, associated with criteria in six domains (growth, development, behaviour, musculoskeletal, neurosensory/skin and other major systems) [30, 43].

5.2. Molecular algorithm

The consensus recommends, where available, next-generation sequencing targeting all spectrum genes as the first test (R9), acknowledging that medicolegal, technical and funding practices may require other methods, such as Sanger sequencing of individual genes [3]. The algorithm in consensus figure 5 (consulted: https://pmc.ncbi.nlm.nih.gov/articles/PMC7136165/figure/Fig5/) follows two branches: in the classical form, screening of known genes by next-generation sequencing (or individual NIPBL sequencing if this is unavailable), followed, if negative, by investigation of mosaicism in non-lymphocyte tissues and then investigation of NIPBL deletions and duplications by multiplex ligation-dependent probe amplification (MLPA) or chromosomal microarray; in the non-classical form, sequencing of known genes or testing targeted to the gene suggested by the phenotype, with mosaicism investigation if negative [3]. If first-line testing is exome or genome sequencing, deletions and duplications are detectable in the same data and analysis can be extended to other genes [3].

GeneReviews proposes two options: a multigene panel including the syndrome genes and phenocopy genes such as AFF4, ANKRD11 and EP300 when the phenotype suggests the diagnosis, or exome or genome sequencing when the phenotype is atypical; for both it recommends considering a technique capable of detecting mosaicism, such as next-generation sequencing in buccal cells, and notes that the American College of Medical Genetics and Genomics (ACMG) and American Academy of Pediatrics recommend exome or genome sequencing as a first- or second-line test for children with developmental delay, intellectual disability or multiple congenital abnormalities [1]. The French protocol recommends, as the initial investigation in typical forms, analysis of a panel of syndrome and differential diagnosis genes in a saliva sample or uncultured skin biopsy, at one of the two national reference laboratories, followed by genome-wide analysis if the panel is negative [7]. The Orphanet summary considers saliva or a buccal swab the preferred sample [2].

Sources: [1, 3, 7].

5.3. Mosaicism: tissue selection and interpretation

When the variant is not detected in lymphocytes, mosaicism should be considered and other tissues investigated, such as fibroblasts, buccal cells or bladder epithelial cells (R5); the gold standard is DNA from uncultured fibroblasts, and analysis of buccal swabs, cultured fibroblasts, urothelial cells, uncultured skin biopsies or surgical specimens has improved the detection rate [3]. Obtaining a buccal swab at the time of blood collection facilitates analysis; in the 2013 study, sufficient DNA was obtained in 21 of 22 individuals using instructions and a single swab sent to families [1, 56]. The Brazilian ophthalmological cohort illustrates the practice: 11 patients were diagnosed by blood exome sequencing, 2 by RNA sequencing, and 3 (18.8%) had mosaic variants identifiable only by exome sequencing in buccal cells [67]. The degree of mosaicism may vary within the same tissue over time: in a boy with an SMC1A variant, the allele frequency in saliva was 53% at 14.3 years and 10% at 18.3 years [57]. Absence of the variant in a parent's blood does not exclude somatic or gonadal mosaicism, and investigation of other parental tissues using sensitive techniques may be considered [1, 47].

5.4. Epigenetic signature

Episignature analysis in peripheral blood, or DNA banked from blood, may be considered to clarify diagnosis in two situations: suggestive findings without a pathogenic variant identified by sequencing or genomic testing of a buccal sample, and suggestive findings with a variant of uncertain significance [1]. The signature has been described for NIPBL, RAD21, SMC3, SMC1A and MAU2, but carriers of loss-of-function SMC3 variants did not cluster strongly with typical signatures, limiting interpretation in that subgroup [1, 51]. Haematopoietic selection against mutated cells described in NIPBL mosaics suggests caution when interpreting a negative blood episignature when mosaicism is probable; this is an inference from mosaicism data, not an explicit guideline recommendation [47].

5.5. Diagnostic yield

In a study of 44 individuals analysed for the five genes then known, including mosaicism investigation, a causative variant was detected in 84% [3]. According to Latorre-Pellicer and colleagues, approximately 85% of clinically diagnosed patients now obtain a molecular diagnosis, largely thanks to high-depth sequencing panels, and in the Spanish cohort of 43 patients the rate reached 90.7%, with 31 variants detected in blood (22 NIPBL, 4 SMC1A, 2 HDAC8, 1 SMC3, 1 RAD21 and 1 ANKRD11), four alterations by microarray or MLPA, and four mosaic variants in saliva or fibroblasts [47]. In a British series of 163 individuals with the syndrome phenotype or a similar phenotype (expanded inclusion criteria), pathogenic variants were distributed across NIPBL (28.2%), SMC1A (3.1%), SMC3 (3.1%), HDAC8 (3.6%) and RAD21 (0.6%), with three de novo variants in ANKRD11, a 1p36.3 deletion and a 12q13.13 duplication including ESPL1 (separase); at least 18% of individuals without a variant were classified as “NIPBL-like”, suggesting undetected mosaic cases [55]. In 2025, genome sequencing of 108 affected individuals from 105 previously negative families identified likely causative variants in 32 unrelated probands (30.5%), with NIPBL (10), ANKRD11 (6), EP300 (3) and EHMT1 (2) as recurrent loci and isolated variants in genes encoding chromatin-associated proteins without a defined role in cohesin [46].

5.6. Test limitations and report content

The proportion of variants detectable by a method (for example, 97% of NIPBL variants by sequencing) is not equivalent to the clinical sensitivity of the test, which depends on the tissue analysed, depth and proportion of mosaic or non-coding cases [1, 46, 47]. A variant of uncertain significance neither confirms nor excludes diagnosis, and likely pathogenic and pathogenic variants are equivalent for clinical decision-making under ACMG/AMP standards [1]. RAD21 exon 14 lies within a segmental duplication, which may affect short-read alignment [1].

The report should state the gene, reference transcript with version, HGVS nomenclature (Human Genome Variation Society), zygosity or hemizygosity, allele fraction when mosaicism is suspected, tissue analysed, ACMG/AMP classification and technical limitations. The MANE Select transcripts confirmed in VariantValidator on 6 October 2026 are NM_133433.4 (NIPBL), NM_006306.4 (SMC1A), NM_005445.4 (SMC3), NM_006265.3 (RAD21), NM_018486.3 (HDAC8), NM_001379291.1 (BRD4), NM_015329.4 (MAU2) and NM_013275.6 (ANKRD11) [41, 42]. The literature sometimes uses earlier versions: the case in section 10 was published with NM_006306.3:c.2369G>A, p.(Arg790Gln), which VariantValidator maps unchanged to NM_006306.4:c.2369G>A, NP_006297.2:p.(Arg790Gln), NC_000023.11:g.53403617C>T (GRCh38), with an outdated transcript version warning [41, 57]. This variant is classified as pathogenic in ClinVar (VCV000212267, multiple submitters without conflict, evaluated in October 2025) [74].

A variant listed as “notable” in a review is not equivalent to a current pathogenicity classification. GeneReviews highlights two recurrent SMC1A variants, NM_006306.4:c.1553G>A, p.(Arg518His), and c.1904G>A, p.(Arg635His), with the caveat that classifications were supplied by the authors and not independently verified by the editorial team [1]. VariantValidator confirms the nomenclature (NC_000023.11:g.53405949C>T and g.53405500C>T, respectively), but in ClinVar the first has a single submission as uncertain significance (VCV001397513) and the second has conflicting classifications (VCV001678779) [41, 75, 76]. For SMC1A and SMC3, variant type determines interpretation: missense variants or in-frame deletions suggest the Cornelia de Lange phenotype, whereas truncating SMC1A variants suggest epileptic encephalopathy (DEE85) and loss-of-function SMC3 variants a milder phenotype with variable penetrance [1, 49, 51].

5.7. Prenatal diagnosis

The main indications for prenatal diagnosis are a previous affected child, a known familial variant or, more frequently, suggestive ultrasound findings without a family history [3]. In 73 published cases with prenatal findings, symmetrical intrauterine growth restriction beginning in the second trimester was the most common finding (80%); limb abnormalities were seen in 66% (probable selection bias), approximately 50% had an abnormal facial profile (micrognathia and prominent maxilla), 51% increased nuchal translucency, 28% diaphragmatic hernia and 15% cardiac malformation [3]. The review of 53 pregnancies described a recognisable ultrasound profile and possible reduction in pregnancy-associated plasma protein A (PAPP-A) in maternal serum, also mentioned by GeneReviews and the French protocol, the latter reporting symmetrical growth restriction generally between 20 and 25 weeks of amenorrhoea [1, 7, 77]. In the EUROCAT registry, 68% of cases with major malformations were not detected by routine ultrasound during the last 11 years of the study period [6]. The Oxford Desk Reference includes the syndrome in the differential diagnosis of congenital diaphragmatic hernia (with growth restriction, sometimes in the third trimester, and upper-limb defects) and growth restriction with asymmetrical radial defects when the karyotype is normal [78, 79].

Molecular prenatal testing can be performed in chorionic villi, amniocytes or embryonic cells from in vitro fertilisation (R8) [3]. Non-invasive screening using cell-free fetal DNA with a multigene panel may identify de novo variants, but interpretation requires samples from both parents, and the possibility of undetectable mosaicism prevents its use for exclusion; the consensus authors therefore consider that this screening has no meaningful use in current practice [3]. When prenatal investigations are considered, their advantages and disadvantages should be discussed with parents (R7) [3].

5.8. Diagnosis of the ESCO2 spectrum

Suspicion rests on the clinical criteria derived from the series of 49 patients: growth restriction; symmetrical mesomelic limb shortening, predominantly of the upper limbs; and characteristic facies with microcephaly [36]. The family history may show affected siblings or consanguinity, but its absence does not exclude the diagnosis [16]. The diagnosis is established in a proband with suggestive findings and biallelic pathogenic or likely pathogenic ESCO2 variants, or with premature centromere separation on cytogenetic study; biallelic variants of uncertain significance, or one pathogenic variant with one of uncertain significance, neither confirm nor exclude the diagnosis [16].

The conventional karyotype, stained with Giemsa or by C-banding, shows in all metaphases premature centromere separation and heterochromatin repulsion (abbreviated together as PCS/HR, from premature centromere separation and heterochromatin repulsion) in most chromosomes: absence of the primary constriction, a “railroad track” appearance and “puffing” of the heterochromatic regions next to the centromeres and of the nucleolar organisers, most evident on chromosomes 1, 9 and 16, on the short arms of the acrocentrics and in the Yq heterochromatin [16, 33]. Cultures also show aneuploidy, micronuclei and multilobed nuclei [16]. This pattern is distinguished from the premature sister chromatid separation described in Cornelia de Lange syndrome and from the premature centromere division of mosaic variegated aneuploidy syndrome, in which separation involves the whole chromatids [16]. C-banding is the technique that best demonstrates it [31, 32]. A routine karyotype may nevertheless be read as normal: in the Malay patient described in 2023, the basic karyotype was normal and the chromosome breakage study was equivocal, which led to a diagnosis of mosaic Fanconi anaemia for 11 years, until exome sequencing identified the ESCO2 variant [59].

In molecular testing, ESCO2 sequencing detects all known pathogenic variants; targeted deletion and duplication analysis is technically possible, but no causal rearrangements have been described [16]. Single-gene testing, a limb malformation or cleft palate panel, or exome or genome sequencing may be used; some panels do not include ESCO2, so its presence should be confirmed [16]. If the proband appears homozygous, an undetected deletion on the other allele and uniparental isodisomy should be considered [16].

In the prenatal period, the findings described include growth restriction, fetal death, polyhydramnios, short long bones, limb reduction defects, oligodactyly, heart disease, increased nuchal translucency, cystic hygroma, cleft lip and palate, polycystic kidneys and frontal encephalocele [31]. The sensitivity of prenatal ultrasound for the spectrum is 41–65% [16]. In a low-risk pregnancy with suggestive findings, the diagnosis can be confirmed by biallelic variants or by PCS/HR in amniocytes or chorionic villi [16].

6. Differential Diagnosis

6.1. Principles

Overlap with Cornelia de Lange syndrome occurs in monogenic disorders, chromosomal abnormalities and fetal alcohol syndrome [1]. Many monogenic disorders are chromatinopathies with almost constant neurodevelopmental impairment, explaining phenotypic convergence and the need to include their genes in panels [3, 7]. The 2018 consensus notes that a variant in a gene with cohesin function is insufficient for diagnosis: individuals with a presumably pathogenic variant in one of these genes, but little or no resemblance to the classical phenotype, do not belong to the spectrum [3].

6.2. Differential diagnosis table

AD, autosomal dominant; AR, autosomal recessive; XL, X-linked.

6.3. Overlap with KBG syndrome

ANKRD11-related KBG syndrome results from an intragenic pathogenic variant (approximately 80%) or a 16q24.3 deletion including the gene (approximately 20%), so sequencing alone is insufficient and copy-number analysis is required [39]. The phenotype combines macrodontia (particularly of the upper central incisors), a triangular face, widely spaced eyes, a long philtrum, a prominent or bulbous nose, thick eyebrows and synophrys, postnatal short stature, skeletal abnormalities (brachydactyly, scoliosis, coccygeal abnormalities, cervical ribs) and electroencephalographic abnormalities with or without seizures; more than 90% have developmental delay, particularly speech delay, with intelligence ranging from normal to moderate disability, and approximately one-quarter have a movement disorder [39]. Smith records mild synophrys, low hairlines and 2-3 toe syndactyly, features that contribute to confusion with Cornelia de Lange syndrome [80]. Intrauterine growth restriction occurs in approximately 20% of people with KBG syndrome [39], whereas the consensus table places it at 70–89% of NIPBL variant carriers and below 20% of ANKRD11 variant carriers [3]. The demonstration that ANKRD11 binds cohesin provides a mechanistic explanation for this boundary [40], and KBG syndrome is addressed here only as a differential diagnosis.

6.4. Differential diagnosis of the ESCO2 spectrum

For the ESCO2 spectrum, physical examination and radiological study of the skeleton as a rule allow similar genetic and acquired conditions to be distinguished, and chromosomal study is considered indispensable for a sound diagnosis [16, 32]. The presence or absence of cleft lip and palate and of intellectual disability helps to separate the syndromes of complex limb defects [32].

AD, autosomal dominant; AR, autosomal recessive; XL, X-linked; PCS/HR, premature centromere separation with heterochromatin repulsion. Sources: [16, 32, 59, 64, 65].

The distinction from Fanconi anaemia has direct consequences, because the latter requires its own haematological and oncological surveillance; the Malay case shows that an equivocal breakage study does not replace the search for PCS/HR in metaphases or ESCO2 sequencing [16, 59]. With regard to Cornelia de Lange syndrome, the 2018 consensus kept Roberts syndrome outside the spectrum, because the overlap is limited to growth restriction, intellectual disability and limb reduction, and because the cytogenetic pattern is different [3, 16].

7. Treatments and Risk Management Strategies

7.1. General principles

There is no disease-modifying therapy; treatment is supportive, aims to improve quality of life, maximise function and reduce complications, and is ideally provided by multidisciplinary teams [1]. Diagnosis entails lifelong medical, multidisciplinary and social care, regardless of molecular aetiology, and the consensus recommends identifying a clinician responsible for coordinating each patient's care, training healthcare and social service professionals in the syndrome (R66), and offering specific, personalised care plans through shared decision-making (R67) [3]. People with the syndrome are more likely to experience delayed treatment, hospitalisation, complications and longer admissions because caregivers lack knowledge of the syndrome, reliable history is difficult to obtain and stigmatisation may occur; individualised plans, advance preparation of admissions and discharges, and leaflets using simple language and photographs are recommended [3]. Individualised emergency cards, already used in several countries, summarise clinical information and potentially fatal complications (R27) [3]. In France, follow-up is coordinated by reference or specialist centres, with the treating physician responsible for growth surveillance, complication screening and psychological support [7].

7.2. Initial evaluation after diagnosis

7.3. Feeding, reflux and digestive complications

Oral feeding is preferred if safe, stress-free and meal time does not exceed 3 hours per day; otherwise enteral nutrition is recommended, with dietitian involvement, and gastrostomy is preferred when prolonged tube feeding is required (R12) [3]. GeneReviews recommends a low threshold for clinical feeding assessment or radiological swallowing study in the presence of dysphagia signs, feeding therapy, thickening or cooling food for safety, and a nasogastric tube or gastrostomy when dysfunction is severe [1].

For reflux disease, no therapeutic outcome data from large series exist; nutritional modification and proton pump inhibitors are first line, antireflux medication should be used up to the maximum dose, and antireflux surgery is restricted to patients in whom nutritional and medical treatment has failed or airway safety is at risk (R33) [3]. Based on the authors' experience and analogy with people with neurological impairment, the consensus reports response to omeprazole at doses of 0.7–3.5 mg/kg/day, with approximately half the dose for maintenance [3]. This range derives from expert experience rather than the label: for the delayed-release oral suspension formulation Prilosec® (NDA022056), the FDA approves, for symptomatic reflux disease and erosive oesophagitis between 1 and 16 years, fixed doses by weight band (5 mg from 5 to <10 kg, 10 mg from 10 to <20 kg and 20 mg from 20 kg, once daily, for up to 4 weeks for symptomatic reflux and 4–8 weeks for oesophagitis), and for erosive oesophagitis between 1 month and <1 year doses of 2.5 to 10 mg by weight band for up to 6 weeks [84]. Doses above label weight bands or prolonged continuous treatment constitute use outside the US label conditions. The label identifies, among other risks, acute tubulointerstitial nephritis, Clostridioides difficile-associated diarrhoea, fractures with prolonged treatment, vitamin B12 deficiency with daily use exceeding approximately 3 years, hypomagnesaemia and fundic gland polyps; in children aged 1 to 16 years the safety profile resembles that in adults, with respiratory events and fever predominating [84]. In the European Union, paediatric oral omeprazole is not included in the EMA list of centrally authorised medicines consulted, so conditions of use depend on national authorisations [9]. The French protocol recommends proton pump inhibitors almost systematically, given reflux prevalence, and allows Nissen fundoplication after medical treatment failure [7]; GeneReviews and the Brazilian text describe an “aggressive” approach to reflux with fundoplication when severe [1, 43].

If reflux symptoms persist, endoscopy should be considered while the patient remains in paediatric care (R34) [3]. Barrett's oesophagus surveillance requires repeated endoscopies under anaesthesia, and the consensus rejects a general rule: the decision is shared with the family, weighing benefit and burden (R35) [3]. Similarly, investigation of malrotation in asymptomatic patients should be discussed and decided with the family (R30), whereas the French protocol reserves it for patients with persistent digestive symptoms [3, 7]; GeneReviews suggests considering an upper gastrointestinal contrast study at initial evaluation [1]. Pyloric stenosis, malrotation and diaphragmatic hernia are corrected surgically, and constipation is treated as in the general population (R31) [1, 3].

7.4. Eyes, ears, mouth and teeth

Ptosis should be surgically corrected if it significantly affects vision or if compensatory chin elevation interferes with walking (R36) [3]. Blepharitis is treated with eyelid hygiene (baby shampoo or dedicated cleansing products), and nasolacrimal obstruction should be suspected if symptoms do not improve (R37); probing and irrigation are considered only after this treatment fails, and severe obstructions may require nasolacrimal intubation or dacryocystorhinostomy [3]. GeneReviews recommends “aggressive” treatment of nasolacrimal obstruction because massage frequently fails in narrow, malformed ducts [1]. Refractive errors should be corrected early to prevent amblyopia (R38), although children tolerate spectacles, occlusion and contact lenses poorly, and refractive surgery may improve visual function [3].

In chronic or recurrent otitis media with effusion, myringotomy with ventilation tubes is first-line treatment and mastoidectomy may be considered when soft tissue is present in the middle ear and mastoid (R41); if tubes are ineffective, conventional or bone-anchored hearing aids are safe alternatives, although poorly tolerated by some, and cochlear implantation has produced variable functional gains [3]. Severe sensorineural hearing loss should prompt assessment for auditory neuropathy (R39), and sinusitis is treated according to general-population standards, with more intensive treatment (immunoglobulins, antibiotic prophylaxis) if immunodeficiency is present [3]. The palate should be inspected and palpated at diagnosis, with specialist referral if there are symptoms of submucous cleft (R14), and dental assessment and cleaning should be regular, potentially requiring anaesthesia (R15) [3]. A healthy diet, topical fluoride and periodic reviews by an interdisciplinary team are components of dental care [3].

7.5. Orthopaedics, rehabilitation and mobility

For major limb abnormalities, physiotherapy or surgery is rarely indicated because function is often good (R43); complete prostheses are rarely accepted, whereas devices with a specific function, such as those enabling independent eating, may be well tolerated (R44) [3]. Developmental and mobility prognosis should inform orthopaedic decisions (R45), and scoliosis and limb-length differences require attention at adult consultations (R46) [3]. Adaptive equipment (orthoses, walkers, wheelchairs) and safety equipment (helmets, door alarms, belt harnesses) should be considered; for contractures and hypertonia, physiotherapy, orthoses, botulinum toxin, baclofen, tizanidine or surgery may be indicated, as decided by specialists [1, 3].

7.6. Neurology and sleep

Seizures are treated according to general regimens (R47); sodium valproate was considered effective in a series of 14 clinically diagnosed patients, and many antiepileptic drugs may be effective without any having demonstrated syndrome-specific efficacy [1, 3, 7]. Caregiver education about seizure presentations is recommended [1]. For sleep disturbances, a medical cause (pain, reflux) should first be excluded, the context assessed and behavioural management instituted (R49); melatonin may be considered if behavioural measures fail [3, 7]. In the European Union, prolonged-release melatonin Slenyto® (1 mg and 5 mg tablets) was authorised on 20 September 2018 for insomnia in children and adolescents aged 2 to 18 years with autism spectrum disorder and/or Smith–Magenis syndrome; the Commission decision of 26 August 2024 extended the indication to neurogenetic disorders with abnormal daytime melatonin secretion and/or nocturnal awakenings, and that of 17 March 2025 added insomnia in those aged 6 to 17 years with ADHD, whenever sleep hygiene is insufficient; in autism spectrum disorder and neurogenetic disorders, the recommended starting dose is 2 mg, increasing to 5 mg if response is inadequate, up to a maximum of 10 mg, once daily, 0.5–1 hour before bedtime, with or after food, with reassessment at least every 6 months and consideration of discontinuation if there is no effect after 3 months [85]. The most frequent adverse effects (1–10%) include somnolence, fatigue, mood swings, headache, irritability, aggression and a hangover feeling [85]. Application of this label to a child with the syndrome depends on meeting the indication criteria (autism, or a neurogenetic disorder with the described sleep pattern); outside those criteria, use is off-label.

7.7. Behaviour, communication and mental health

Effective verbal and non-verbal communication improves quality of life, and speech and language therapy should begin within the first 18 months, with early augmentative and alternative communication interventions (gestures, icons, pictures, written language) tailored to each person (R64) [3]. Communication assessment should consider visual and hearing problems, speech abnormalities, intellectual disability, interaction difficulties and social anxiety, and video observation and responsive milieu teaching help identify subtle communicative signals (R63) [3]. GeneReviews adds that augmentative communication devices do not impair speech development, and that children may benefit from interventions used in autism, such as ABA (applied behavior analysis) [1].

In self-injurious behaviour, medical assessment directed at pain sources should occur first, then behavioural assessment of self-restraint, and finally functional analysis (R54), with treatment combining medical and behavioural strategies tailored to the behaviour's function (R55); restraints have been used first line to prevent permanent tissue loss [3]. Evidence for risperidone in self-injurious behaviour is limited, and the consensus requires monitoring of metabolic syndrome, weight and cognitive effects; there are no data on mood stabilisers [3]. In the USA, the Risperdal® label approves risperidone for irritability associated with autistic disorder in children and adolescents aged 5 to 17 years (including aggression, deliberate self-injury, temper tantrums and quickly changing moods), starting at 0.25 mg/day below 20 kg or 0.5 mg/day from 20 kg, increasing after at least 4 days to 0.5 mg/day or 1.0 mg/day, respectively, with an effective range of 0.5 to 3 mg/day; there are no dosing data below 15 kg [86]. The same label includes warnings about neuroleptic malignant syndrome, tardive dyskinesia, metabolic changes (hyperglycaemia, dyslipidaemia, weight gain) and hyperprolactinaemia [86]. Use in a person with the syndrome without a diagnosis of autistic disorder is off-label; oral risperidone does not appear in the EMA list of centralised authorisations consulted [9]. National authorisations, however, follow the summary of product characteristics harmonised after the 2008 referral procedure: in the EU, risperidone is authorised from 5 years of age only for the short-term symptomatic treatment (up to 6 weeks) of persistent aggression in conduct disorder, in children and adolescents with below-average intellectual functioning or intellectual disability, with a starting dose of 0.25 mg/day below 50 kg (optimum 0.5 mg/day; maximum 0.75 mg/day) and 0.5 mg/day from 50 kg (optimum 1 mg/day; maximum 1.5 mg/day); autism spectrum disorder is not a European indication [87].

Selective serotonin reuptake inhibitors are standard treatment for obsessive-compulsive disorder, but the consensus recommends caution because of possible behavioural activation and worsening agitation; second-generation antipsychotics may help with rigidity and insistence on sameness [3]. Autism spectrum disorder should be considered throughout life, with detailed observation beyond standardised tools and specific interventions integrated into the syndrome's social profile (R56–R58); interventions for repetitive behaviours should address anxiety, sensory problems, social demands and the environment (R59); atypical presentation of anxiety and mood disorders should be considered when behaviour changes (R60), and periods of change should be prepared for (R61), with psychosocial and pharmacological interventions for anxiety and mood (R62) [3]. Sensory difficulties should be assessed and supported throughout life (R50), and developmental and educational support should aim to maximise potential, with personalised adaptive-skills goals (R51–R53) [3]. The French protocol emphasises the risk of depression in late adolescence and early adulthood, potentially requiring intensified child and adolescent psychiatric follow-up [7].

7.8. Growth, endocrinology, gynaecology and urology

Growth should be monitored on specific charts (R11); if growth velocity is lower than expected, gastrointestinal problems, thyroid dysfunction and growth hormone abnormalities should be considered [3]. The benefits of possible growth hormone treatment should be weighed against the burden of daily subcutaneous injections and the absence of a positive impact of greater adult height on quality of life for most people with the syndrome [3]; the French protocol recommends endocrinology referral for growth deceleration, height below −2 SD on general charts or deviation of 1.5 SD from target height, and considers growth hormone only for the same indications as in the general population (proven deficiency) [7]. Overweight requires attention to diet and encouragement of activity (R22) [3].

Cryptorchidism should be surgically corrected to reduce testicular cancer risk, as in the general population [3]. Sexual education should suit the level of understanding, and contraception follows general-population standards (R20); menstrual suppression is preferred by some families, and hysterectomy is not recommended as primary contraception, being reserved for severe menorrhagia refractory to medical treatment (R21) [3]. Premenstrual syndrome and dysmenorrhoea may be associated with behavioural changes and are treated as in the general population [3, 7].

7.9. Immunology, haematology and vaccination

Vaccines should be administered according to the national programme (R17), and GeneReviews recommends routine immunisation unless there are specific concerns [1, 3]. Unusually frequent or severe infections justify immunological investigation and, if immunodeficiency is present, treatment by an immunologist [1, 3]. Severe thrombocytopenia may require intravenous immunoglobulin and/or corticosteroid therapy, as decided by a haematologist [1].

7.10. Anaesthesia and surgery

In 50% of children with the syndrome who underwent intubation, the procedure was difficult, and an adverse allergic reaction to midazolam may occur, although complications due to anaesthetic drugs are rare; the anaesthetist should be aware of potential intubation difficulty (R42) [3]. GeneReviews recommends preoperative assessment for thrombocytopenia and heart disease and procedures with anaesthetists experienced in the syndrome's difficult airway [1]. OrphanAnesthesia recommendations (2016) advise always anticipating a difficult airway, without a single reliable predictor, with difficult direct laryngoscopy described in up to 50% and success described with fibreoptic intubation through a laryngeal mask, a McCoy laryngoscope, Airwayscope® or blind nasotracheal intubation; awake fibreoptic intubation may fail because of lack of cooperation [12]. The same document recommends a preoperative full blood count (immunodeficiency and thrombocytopenia), cardiac assessment (pulmonary stenosis in approximately 39% in one series), aspiration prevention (the main cause of death and main anaesthetic risk), careful individual drug titration (oversedation described with standard doses of metoclopramide and midazolam), premedication only if necessary and in a monitored area, and states that the syndrome is not associated with malignant hyperthermia, that suxamethonium and volatile anaesthetics have been used safely, and that general and neuraxial anaesthesia have been performed successfully [12]. After difficult intubation, it suggests careful extubation (possibly over an exchange catheter), postoperative intensive care admission and advises against ambulatory anaesthesia; the document was reviewed by two disease experts and not by anaesthetists [12].

7.11. Regulatory status

ADHD, attention deficit hyperactivity disorder. Sources: [8–10, 84–87]. Reimbursement status in Portugal (INFARMED) and Brazil (ANVISA/CONITEC) was not verified.

7.12. Clinical trials and research

Sources: [10, 11, 88]. A search of the EU Clinical Trials Information System (CTIS) for “Cornelia de Lange” returned no trials [89].

These records do not allow efficacy conclusions: none has published results, the lithium trial is single-arm without a control group, the N-acetylcysteine trial has a planned size of 10 participants and has not yet begun recruitment, and the main behavioural outcome is an informant-report scale susceptible to expectancy effects. The rationale for lithium rests on preclinical evidence: in 2026, two- and three-dimensional models derived from human induced pluripotent stem cells (neural precursors, neurons and brain organoids) mimicking HDAC8 haploinsufficiency showed altered proliferation and differentiation and downregulation of the canonical WNT pathway, with improved neuronal differentiation following lithium treatment, an activator of that pathway [90]. These findings are preclinical and do not demonstrate clinical benefit. Lithium carbonate has a narrow therapeutic index, and the dose in the trial record is not a recommendation. N-acetylcysteine is being studied for its antioxidant effect and partial response in compulsive behaviours in other conditions, according to the registry itself [11].

7.13. Recent developments (2025 to October 2026)

In published, peer-reviewed evidence, the period from 2025 to October 2026 brought, among other contributions: definition of MAU2 as a causative gene with its own episignatures (March 2026) [29]; delineation of a disorder associated with WAPL, a cohesin release factor, as a new gene associated with cohesinopathies (August 2026) [38]; the first description of maternal gonadal mosaicism for a complete RAD21 deletion in two siblings [73]; successful preimplantation genetic diagnosis using haplotype analysis in a family with low-level maternal gonadal mosaicism for a microdeletion truncating HDAC8, with a healthy birth in February 2023 [72]; genome sequencing of “negative” individuals with a yield of 30.5% [46]; the first female case of the classical form due to SMC1A mosaicism [57]; Brazilian ophthalmology and family quality-of-life data [67, 91]; and an Italian cohort of 54 patients without increased cancer prevalence [92]. In preclinical work, the study of lithium in brain organoids [90]. The comprehensive GeneReviews update was published on 16 June 2026 [1]. No sponsor announcements, authorisation applications, Committee for Medicinal Products for Human Use (CHMP) opinions or authorisations for the syndrome were identified in the regulatory sources consulted [8, 9].

7.14. Management of the ESCO2 spectrum

There are no published clinical practice guidelines for the ESCO2 spectrum; the GeneReviews recommendations are based on the literature and the authors' experience [16]. There is no targeted pharmacological therapy, and the sources consulted indicate no drugs with a specific dose for the condition; a ClinicalTrials.gov search (1 November 2026, terms “Roberts syndrome”, “ESCO2” and “SC phocomelia”) returned no registered studies [16, 93]. p53 inhibition with pifithrin-α remains a result in a mouse model [61].

After diagnosis, it is recommended to measure weight, length and head circumference; assess the limbs in a multidisciplinary clinic (orthopaedics, physical medicine, occupational therapy and physiotherapy); assess cleft lip and palate in a craniofacial team; request paediatric ophthalmological assessment (corneal opacities in the newborn, cavernous angioma of the third nerve in adolescence and young adulthood), developmental, urological and cardiological assessment and renal ultrasound; measure serum immunoglobulins with immunological assessment; and request neurological assessment with brain imaging for vascular malformations and moyamoya, a complete skin examination by dermatology and assessment of the cranial nerves and the musculoskeletal system [16].

Nutritional support may require high-calorie foods or formulas and nasogastric or enteral tube feeding [16]. The limbs are managed with physiotherapy and occupational therapy, adaptive devices, stretching, night splints, serial casting, prostheses and surgery when indicated, avoiding prolonged periods of immobilisation, for example after surgery; hand surgery aims to achieve functional grasp early [16]. For cleft lip and palate, special feeding bottles are used; according to GeneReviews, the first lip repair operations are performed at around 10 weeks and palatal repair usually between 6 and 12 months, although the same text gives 18–24 months as the optimal interval for palatoplasty, so the schedule is decided by the craniofacial team, with postoperative speech and language therapy and assessment for Pierre Robin sequence if there is micrognathia [16]. Craniosynostosis and micrognathia may require surgical correction [16]. A Chinese publication on two Yemeni siblings operated on in the first three years of life for severe deformities described prolonged survival and functional benefit from corrective surgery [94].

Anaesthesia and surgery require preparation: patients often have fragile health, sometimes with cardiac and renal dysfunction, and GeneReviews considers that each anaesthetic may be life-threatening [16]. Infections should be treated promptly, with infectious disease support when necessary; stroke and neoplasms follow the usual treatment of the respective specialties [16]. Developmental support includes early intervention, special education, speech and language therapy and augmentative and alternative communication when expressive speech is limited; social support, home nursing, housing adaptations and palliative care are assessed on an ongoing basis [16].

8. Reproductive Options and Family Counselling

8.1. Modes of transmission by gene

Forms associated with BRD4, MAU2, NIPBL, RAD21 and SMC3 are autosomal dominant and those associated with HDAC8 and SMC1A are X-linked; in both groups, most patients have a de novo variant [1, 7]. Genetic counselling should be offered to all families and explain that recurrence risk depends on the gene involved (R6) [3]. Autosomal dominant transmission of non-classical forms occurs, although infrequently [3]; the Oxford Desk Reference already noted that a milder form may follow a dominant pattern, illustrated in figure 3 of the Smith chapter (mildly affected mother with severely affected children) [19, 71]. In the MAU2 series, transmission from an affected parent to an affected child was described in five families [1]. In RAD21, variants are frequently familial, with marked intrafamilial variability [52]. Counselling is particularly difficult for SMC1A, HDAC8 and RAD21 because of phenotypic variability, even within the family [3].

8.2. Recurrence risk in autosomal dominant forms

If a parent of the proband is affected and/or has the variant, the risk for siblings is 50% [1]. When the proband's constitutional variant is not detected in either parent's blood, recurrence results from parental gonadal or somatic and gonadal mosaicism, and estimates vary by source:

The French protocol summarises germline mosaicism risk as ranging from 0.89% to 5.4% depending on the study and uses 1% in its prenatal diagnosis chapter [7]. If the proband has a postzygotic mosaic variant, the sibling risk is presumed equal to that of the general population, and it is very unlikely that a parent has the variant [1]. Parental blood testing may not detect all somatic mosaicism and does not detect variants confined to germ cells; techniques capable of detecting parental mosaicism, such as next-generation sequencing in buccal cells, may be considered [1]. In 2026, the first sibling pair with a complete RAD21 deletion (8q23.3–q24.11) due to probable maternal gonadal mosaicism was described, with both parents' blood and paternal sperm negative by gap PCR and digital PCR [73]. Each child of a person with an autosomal dominant form has a 50% probability of inheriting the variant; the French protocol acknowledges that the risk may be lower if the variant is mosaic [1, 7].

8.3. X-linked forms (SMC1A and HDAC8)

The father of an affected boy is not a carrier and does not need testing [1]. In a family with more than one affected individual, the mother is probably heterozygous; if she has several affected children without other affected relatives and the variant is not detected in her blood, she probably has gonadal mosaicism [1]. Because SMC1A is not fully inactivated, a heterozygous mother tends to have milder features than affected children, and mothers with minimal signs who were heterozygous for HDAC8 variants have been identified [1]. Females heterozygous for HDAC8 may be affected or entirely healthy, presumably depending on which X chromosome is inactivated [3]. If the mother has the variant, transmission risk is 50% per pregnancy: boys inheriting it are affected and heterozygous girls may have some signs (SMC1A) or severity dependent on X-inactivation (HDAC8); a father with the variant transmits it to all daughters and no sons [1]. In a simplex case where the variant is not detected in maternal blood (boy) or either parent's blood (girl), sibling risk is approximately 1% because of possible gonadal mosaicism [1]. Gonadal SMC1A mosaicism has been described in a father with several affected daughters [1], and the 2025 review compiled three familial cases of parental SMC1A mosaicism associated with the syndrome, including an asymptomatic mother with a blood allele frequency of 9% [57].

8.4. Prenatal and preimplantation diagnosis

If the causative variant has been identified in an affected relative, prenatal testing and preimplantation genetic diagnosis are possible [1], and the consensus recommends discussing reliable prenatal diagnosis in chorionic villi or amniocytes when there is a causative variant in a child or previous pregnancy (R8) [3]. The French protocol distinguishes two situations: after the birth of a child with a variant not found in the parents, prenatal diagnosis may be discussed to exclude residual risk, but preimplantation diagnosis is not indicated because that risk is low; when one member of the couple has the syndrome with a known variant, the couple may request prenatal or preimplantation diagnosis, to be discussed at a multidisciplinary prenatal diagnosis centre [7]. Outside France, preimplantation diagnosis has been applied in proven gonadal mosaicism: in a Chinese family with two pregnancies affected by a 172.3 kb Xq13.1q13.2 microdeletion truncating HDAC8, low-level maternal gonadal mosaicism was confirmed by nanopore sequencing and digital PCR in ovarian samples, and haplotype-based testing resulted in a healthy birth [72]. In families without an identified variant, high-resolution ultrasound can assess growth, limbs, heart, diaphragm and palate [1]. The ideal time to determine risk and discuss options is before pregnancy, and counselling should be offered to affected or at-risk young adults [1]. When the molecular cause is not identified, DNA banking from the proband should be considered [1].

8.5. Fertility and pregnancy in people with the syndrome

People with severe forms do not usually reproduce, but mildly affected individuals may have children [1]. A small number of women with the syndrome have had children, often with maternal diagnosis made only after the child's; some men have fathered children, but there are no reliable data on male fertility [3]. The French protocol does not report difficulty conceiving in couples where one member has the syndrome, does not recommend specific pregnancy follow-up beyond that dictated by manifestations (for example, preconception or early cardiology consultation in women with heart disease), and notes a slightly increased risk of miscarriage and preterm delivery with a bicornuate uterus; the couple should be aware of transmission risk and possible ultrasound signs [7]. At the time of the anaesthetic recommendations, no cases of obstetric anaesthesia in this syndrome had been described [12].

8.6. Counselling in the ESCO2 spectrum

The ESCO2 spectrum is inherited in an autosomal recessive manner [16]. The parents of a proband are presumed to be heterozygous and should be tested to confirm carrier status and calculate the recurrence risk [16]. If only one parent is a carrier, with confirmed paternity and maternity, one of the variants may have arisen de novo in the proband or in a mosaic parent; if the proband appears homozygous, an undetected deletion and uniparental isodisomy should be excluded [16]. When both parents are heterozygous, each sibling has, at conception, a 25% chance of being affected, 50% of being an asymptomatic carrier and 25% of inheriting neither variant; carriers have no manifestations [16]. Each sibling of a proband's parent has a 50% chance of being a carrier [16].

Pregnancies in affected people are rare. GeneReviews reports two, one resulting in the birth of an unaffected girl and the other in a second-trimester miscarriage, and Smith's Recognizable Patterns reports that only one of three adult women who became pregnant had a term child [16, 31]. The children of an affected person are obligate carriers [16].

Carrier testing in relatives requires prior identification of the familial variants, and should be considered for the partner of a carrier, especially if there is consanguinity or shared ancestry, as in the Colombian population with founder variants [16]. Once the variants are known, prenatal diagnosis and preimplantation genetic testing are possible; alternatively, cytogenetic study of amniocytes or chorionic villi for PCS/HR, combined with serial ultrasound of the limbs, heart, palate and other organs, allows prenatal diagnosis [16]. DNA banking is recommended for probands without a molecular diagnosis [16]. In the Brazilian case in section 10, the couple's next pregnancy was studied prenatally and the fetus was homozygous for the normal allele [96].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Frequency and the denominator problem

The order-of-magnitude discrepancy between estimates results from what is measured: congenital anomaly registries mainly capture the classical form with major malformations, whereas higher estimates seek to include mild forms, probably the majority, which may go unrecognised [1, 6]. None of the sources consulted provides a validated population estimate encompassing the entire molecular spectrum. In the EUROCAT registry, all cases were sporadic, maternal and paternal age did not appear to be risk factors, 91.5% (97/106) were live births, 2.8% fetal deaths and 5.7% pregnancy terminations after prenatal diagnosis, and the most frequent associated malformations involved limbs (73.1%), heart (45.6%), central nervous system (40.2%) and cleft palate (21.7%) [6]. The Brazilian text reports a slight female predominance (1.3:1) [43], and the 1993 US series included 176 females and 134 males [20]; conversely, severe limb abnormalities are more frequent in males [3, 6].

9.2. Populations, cohorts and founder effect

Because most cases result from private de novo variants, the sources consulted describe no founder effects or recurrent variants associated with a population; recurrent SMC1A variants are rare [1]. The clinical literature rests mainly on US cohorts (310 individuals examined with the parent support group in 1993; 295 deaths with a known cause over 41 years; 716 probands with genomic analysis), European cohorts (EUROCAT registry; Spanish, Dutch, Italian and British cohorts) and, more recently, Chinese cohorts [6, 20, 45, 47, 48, 55, 72, 73, 92, 97]. In Brazil, a cohort of 16 patients with NIPBL variants, studied with approval from the Universidade Federal de São Paulo, characterised the ophthalmological phenotype, and a study of 70 families assessed family quality of life: the mean Barthel index was 41.8 ± 31.6 (severe dependence), overall family quality of life was 4.04 ± 0.47 on a 5-point scale, with lower values for emotional well-being (3.47) and physical and material well-being (3.83), and parental cohabitation, the father's higher educational attainment and congenital heart disease explained 31.3% of the variance; the authors recommend psychological support for all family members and investment in social policies, transport and services [67, 91]. The Brazilian reference manual devotes a chapter to the syndrome, emphasising the multiprofessional team [43]. No epidemiological data or cohorts from Portugal were identified in the sources consulted, which does not demonstrate their non-existence.

9.3. European networks and associations

The relevant European reference network is ERN ITHACA, dedicated to rare malformation syndromes and intellectual disability, whose experts reviewed the syndrome's Orphanet summary [2, 13]. In France, follow-up is organised through reference centres within the AnDDI-Rares and OSCAR networks [7]. Patient associations include CdLS World (World Federation of CdLS Support Groups), whose scientific council promoted the 2018 consensus, the Cornelia de Lange Syndrome Foundation in the USA, which provides syndrome-specific growth charts, and the Association Française du Syndrome de Cornelia de Lange [1, 3, 7, 14, 15]. The consensus emphasises that support groups and social networks have proved useful for families and that guardianship rules vary between countries and should be established before adulthood [3]. Specific Portuguese or Brazilian associations were not identified in the sources consulted.

9.4. Special populations

In adults, care coordination is necessary because many specialties are involved; people with the syndrome aged 50 or older have been described [3]. Females heterozygous for HDAC8 or SMC1A may go unrecognised until a child's diagnosis [1, 3]. Mosaic individuals generally have a phenotype similar to constitutional carriers, requiring that the diagnosis not be excluded on the basis of a negative blood result [47, 56].

9.5. Epidemiology and resources in the ESCO2 spectrum

There are no reliable prevalence estimates; GeneReviews counts about 173 individuals of diverse ethnic origins, Smith's Recognizable Patterns reports more than 200 identified patients, and a 2023 case review cites about 150, counts that reflect different dates and inclusion criteria [16, 31, 59]. Parental consanguinity is frequent; in the 1993 review of 100 cases it was present in 40.8% [16, 96]. Founder variants have been described in Colombia (NM_001017420.3:c.505C>T and NM_001017420.3:c.751dup), as has a recurrent variant in patients of Turkish and Pakistani origin (NM_001017420.3:c.879_880del) [16].

In Brazil, the case published in 2020 in Porto Alegre was described by the authors as the first with molecular confirmation in the country, after an earlier case with ultrasound diagnosis and autopsy confirmation, without genetic study [96]. A Thai family, a Malay patient and a Chinese patient have also been described, extending the description to Asian populations [59, 65, 68]. The sources consulted do not include data from Portugal. The relevant European network is ERN ITHACA [13]. GeneReviews lists as patient resources the Children's Craniofacial Association (USA), Face Equality International and REACH (United Kingdom, for children with upper limb differences) [16].

10. Illustrative Clinical Case

In 2025, Gil-Salvador and colleagues published the first case of a woman with the classical form of the syndrome caused by postzygotic SMC1A mosaicism [57]. She was the first daughter of a healthy, non-consanguineous couple, born at term by caesarean section for breech presentation after an uncomplicated pregnancy, with birth length, weight and head circumference within normal ranges [57]. At one and a half months she was admitted for vomiting and diagnosed with a multiple-malformation syndrome, with congenital heart disease that did not require surgery and hypertrophic pyloric stenosis operated on at 3 months [57]. At two and a half years, physicians suggested the clinical diagnosis of Cornelia de Lange syndrome because of coarse facies, synophrys, hirsutism, a single palmar crease and marked intellectual disability, with no language at that age [57].

At 16 years she developed bipolar disorder with obsessive traits, rituals, atypical gestures and constantly repeated questions, with disruptive, uncooperative behaviour, sometimes self-injurious; later, obsessive-compulsive symptoms, hyperphagia, anxiety, depression and hypomania emerged, requiring psychopharmacological adjustments [57]. At 42 years, still without a genetic diagnosis, she was reassessed by a clinical geneticist at the Zaragoza reference centre: she had thick eyebrows, a short nose with a slightly concave ridge, a smooth philtrum, thin upper vermilion and downturned mouth corners (four cardinal signs), global delay and intellectual disability, small hands and hirsutism, without oligodactyly, diaphragmatic hernia, microcephaly or postnatal growth restriction, totalling 11 points on the consensus scale, corresponding to the classical form [57]. Computer-assisted facial recognition analysis of a portrait suggested the syndrome as the leading diagnosis [57]. Article figure 1 (consulted: https://pmc.ncbi.nlm.nih.gov/articles/PMC12216401/figure/Fig1/) shows the facies, hands, facial analysis heat map, and variant chromatograms and alignments in blood and buccal mucosa [57].

A customised 35-gene panel in blood DNA identified the SMC1A variant NM_006306.3:c.2369G>A, p.(Arg790Gln), with an allele fraction of 32% (coverage 338×); in buccal mucosa the variant was present in 27% of reads (3342×), and Sanger sequencing confirmed it in both tissues and its absence in the parents [57]. The variant was absent from gnomAD, in silico predictors (REVEL 0.806) were concordant, it had been described constitutionally in two women with moderate and mild forms, and a different substitution at the same codon, p.(Arg790Trp), was known; it was classified as pathogenic according to ACMG criteria [57]. In current nomenclature (MANE Select), it corresponds to NM_006306.4:c.2369G>A, NP_006297.2:p.(Arg790Gln), and ClinVar classifies it as pathogenic without conflicts [41, 74]. Molecular dynamics simulations suggested local disorganisation of the “elbow” domain of the SMC1A C-terminal coiled coil [57].

The case illustrates three points. First, the classical form can result from a gene usually associated with non-classical forms, confirming that phenotype does not identify the gene [3, 57]. Second, unlike NIPBL mosaics and the two boys with SMC1A mosaicism (variant absent or at 2% in blood and 60% in buccal mucosa in one), in this woman the variant was present at similar frequencies in blood and mucosa; the authors propose that, in females, inactivation of the X chromosome containing the variant allows cells to escape haematopoietic selection, although this has not been demonstrated [47, 57]. Third, molecular diagnosis was only obtained at 41 years when a panel with very high sequencing depth was used, showing that exome sequencing followed by parental Sanger validation may be insufficient for low-level mosaics [57]. The psychiatric management described relates to an individual case and is not a recommendation; guidance for behaviour and mental health is provided in section 7.7 [3].

The second case illustrates the severe end of the ESCO2 spectrum. Da Costa Almeida and colleagues described in 2020 the first child of a young, healthy Caucasian couple from Porto Alegre, with no known consanguinity and no similar cases in the family [96]. The ultrasound scan in the fifth month of gestation suggested cleft lip and palate and showed short long bones. He was born at term by caesarean section weighing 1,570 g, with a length of 34 cm and a head circumference of 26 cm, and Apgar scores of 9 and 10 [96]. On the second day of life he had a very wide anterior fontanelle, hypertelorism, blepharophimosis, shallow orbits with exophthalmos, a central facial cleft involving the nose, lip and palate with a nasal rudiment, and small, low-set, posteriorly rotated ears; the humeri were very short, the forearms apparently absent, the hands small, with absence of the left thumb and an appendicular thumb on the right, and there was shortening of the thighs and legs with limitation of the knees and syndactyly of the fourth and fifth toes [96].

Radiology showed cranial deformity with reduced anteroposterior diameter, 11 pairs of ribs, short and deformed humeri, absent radii, bowed femora and bilateral absence of the tibia and fibula (figure 2 of the article, consulted) [96]. Echocardiography revealed a ventricular septal defect and tricuspid regurgitation; cerebral and abdominal ultrasound scans were normal [96]. High-resolution lymphocyte karyotyping (46,XY) showed absence of centromeric constriction in some chromosomes, premature centromere separation in others and heterochromatin repulsion, most evident on chromosomes 1, 9, 16 and Y, in 100% of the 32 cells analysed, with 8 chromosomes with PCS/HR per cell (figure 3 of the article, consulted) [96]. Sequencing of the ten coding exons of ESCO2 identified in homozygosity the deletion published as c.875_878delACAG, which in current nomenclature corresponds to NM_001017420.3:c.876_879del, p.(Asp292GlufsTer48), classified as pathogenic in ClinVar; both parents were heterozygous [62, 63, 96]. The newborn deteriorated and died at 20 days. In the next pregnancy, prenatal diagnosis showed a fetus homozygous for the normal allele [96].

The case shows the value of the karyotype as a rapid and almost pathognomonic test when there is tetraphocomelia with a facial cleft, the correlation between severity of malformations and early mortality, and the usefulness of molecular diagnosis for prenatal diagnosis in a subsequent pregnancy [16, 96]. The authors interpreted the similarity of the picture to the literature as a sign of little clinical variability; GeneReviews and the larger series describe, on the contrary, great variability, even among siblings [16, 35, 96].

11. Prognosis and Long-Term Follow-Up

11.1. Survival and causes of death

In the absence of severe congenital internal-organ abnormalities, life expectancy is not significantly reduced [1]; the Orphanet summary considers it usually unchanged with appropriate care, but emphasises the lack of follow-up data and reduced life expectancy with severe or untreated malformations [2]. Most people reach adulthood thanks to improved care, particularly in the first year [3]. In a study of 48 patients born between 1917 and 1982, mortality was moderately higher than in the general population, with the greatest difference in younger age groups, and two patients died at 54 and 61 years [1]. In the EUROCAT registry, first-week survival among live births was 91.4% [6].

In a review of 426 clinically diagnosed probands who died between 1966 and 2007, 295 had an identified cause of death (ages from 21-week fetuses to 61 years): respiratory causes, including aspiration/reflux and pneumonia, were most frequent (31%), followed by gastrointestinal disease including obstruction and volvulus (19%), congenital abnormalities such as diaphragmatic hernia and heart defects (15%), neurological causes and accidents (8% each), sepsis (4%), acquired heart disease (3%), cancer (2%), renal disease (1.7%) and other causes (9%) [97]. The consensus disaggregates these data by age: diaphragmatic hernia (17%) and respiratory problems (13%) predominated in infants, and sequelae of heart defects (10%) and respiratory (32%) and gastrointestinal (18%) causes in children; in adults, deaths relate to the gastrointestinal, pulmonary and cardiac systems, infections or anaesthesia [3]. Caecal volvulus is associated with significant mortality [3]. Reliable data on the risk of death in childhood are lacking [3]. The Brazilian text identifies brain injury (hypoxic-ischaemic or dysgenesis) as a factor associated with poorer prognosis and indicates that satisfactory quality of life is possible in the mild phenotype [43]; the Orphanet summary identifies severe limb defects and behavioural and psychiatric disorders as the main determinants of quality of life [2].

11.2. Course in adulthood

Facial features and organ involvement in adults resemble those in paediatric ages; congenital heart defects detected in childhood rarely cause unexpected complications, hypertension and congestive heart failure have been described in 4–8% and 2–4%, respectively, and in a retrospective cohort of 97 adults there were 2 myocardial infarctions and 2 strokes [3]. Renal failure occurred in 1%, rarely fatal, and 30% of adults had structural renal malformations, of whom 24% had abnormal creatinine clearance [3]. Prostate enlargement was found in 10% of men by 41 years [3]. Three cases of oesophageal cancer with Barrett's oesophagus have been described; there is no increased cancer risk at young ages, but data for middle and older ages are lacking [3]. In an Italian cohort of 54 patients (median 13 years), the hypothesis of higher cancer prevalence was not confirmed, and RAD21 and STAG1/2 variants mainly appeared as somatic variants in the neoplasms studied, whereas germline NIPBL variants predominated in the Cornelia de Lange spectrum [92]. Hearing loss may improve over time and blepharitis tends to improve with age, but chronic sinusitis may persist [3, 19]. Adaptive skills may decline with age, but cross-sectional data may reflect cohort effects, and longitudinal studies are required [3]. Few adults have reported sleep problems, suggesting spontaneous improvement [3].

11.3. Surveillance

Follow-up should preferably be provided by a paediatrician experienced in the syndrome, with high frequency in infancy and early childhood and annually to every 3–5 years in adolescence and adulthood, adjusted to problems (R19) [3].

Transition from paediatric to adult care should begin early, with appropriate transfer of medical history and knowledge of personal characteristics, ideally with joint assessment by both teams (R68), and adolescence and early adulthood require additional person-centred support (R65) [3].

11.4. Prognosis and surveillance in the ESCO2 spectrum

Prognosis depends on the malformations: intrauterine, neonatal and first-year mortality is high in severe forms, and early death is generally due to respiratory complications and infections [16]. A birth length greater than 37 cm, less severe limb defects, absence of cleft palate and thin nostrils are associated with a better prognosis [31]. Mild forms are more likely to reach adulthood [16]. In most patients the cause of death has not been published; among those reported are infection (5 patients), aneurysm or haemorrhage (3) and neoplasm (3) [16]. Natural history in adulthood rests on isolated cases, so there are no validated survival estimates.

12. Summary and Final Considerations

Cornelia de Lange syndrome is a multisystem developmental disorder, described by Cornelia de Lange in 1933 and now defined as a spectrum of classical and non-classical forms caused by variants in cohesin complex genes, predominantly NIPBL and, less frequently, SMC1A, HDAC8, SMC3, RAD21, BRD4 and MAU2 [1, 3]. The predominant mechanism is transcriptional dysregulation during development rather than failure of sister-chromatid cohesion, bringing the syndrome close to other chromatinopathies such as ANKRD11-related KBG syndrome, which the 2018 consensus included in the spectrum and the 2026 GeneReviews treats as a differential diagnosis [1, 3, 7].

Clinical diagnosis uses the 2018 consensus score, in which 11 or more points with at least three cardinal signs define the classical form even without molecular confirmation [3]. Molecular testing should include all spectrum genes and differential diagnoses, and postzygotic mosaicism, present in approximately 10–20% of molecularly diagnosed patients and frequently absent from blood because of haematopoietic selection, justifies investigation of buccal mucosa, saliva or uncultured fibroblasts [1, 3, 47]. The methylation episignature and genome sequencing resolve some persistently negative cases [1, 46].

Treatment is supportive and multidisciplinary: safe feeding with early recourse to gastrostomy, proactive reflux treatment with proton pump inhibitors, vigilance for malrotation and volvulus in any acute abdomen, regular auditory and ophthalmological assessment, early augmentative communication, an approach to self-injury beginning with a search for pain, and anaesthetic preparation for a difficult airway [1, 3, 12]. No specific approved therapy exists; lithium and N-acetylcysteine trials are ongoing or yet to start and have no published results [10, 11]. Recurrence risk for non-carrier parents lies between approximately 0.9% and 1.5% in autosomal dominant forms and at approximately 1% in X-linked forms, with higher estimates in series enriched for familial cases, and parental testing should consider mosaicism beyond blood [1, 3, 95].

The ESCO2 spectrum (Roberts syndrome and SC phocomelia) is an autosomal recessive cohesinopathy outside the Cornelia de Lange spectrum, in which loss of function of the ESCO2 acetyltransferase impairs centromeric cohesion and produces a characteristic cytogenetic pattern of premature centromere separation with heterochromatin repulsion [3, 16]. The picture combines prenatal growth restriction, symmetrical phocomelia predominantly of the upper limbs, cleft lip and palate and microcephaly, with intellectual disability in some patients and a risk of severe infections in the first year, of aneurysms and moyamoya in adolescence and young adulthood, and of neoplasms [16]. The diagnosis is confirmed by biallelic ESCO2 variants or by karyotype, counselling follows the recessive model and management is multidisciplinary, without targeted therapy; Juberg–Hayward syndrome has been described as an allelic entity [16, 64].

13. Bibliographic References

1. Deardorff MA, Raible SE, Krantz ID. Cornelia de Lange syndrome. 2005 Sep 16 [updated 2026 Jun 16]. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available at: https://www.ncbi.nlm.nih.gov/books/NBK1104/
2. Orphanet. Cornelia de Lange syndrome (ORPHA:199). Summary reviewed by Goldenberg A, Vera G (ERN ITHACA); last updated September 2021 [accessed 6 Oct 2026]. Available at: https://www.orpha.net/en/disease/detail/199
3. 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. PMID: 29995837. PMC7136165.
4. GeneGuidance. Clinical curation database by gene and condition (export of 26 September 2026): NIPBL, SMC1A, SMC3, RAD21, HDAC8, BRD4, MAU2, ANKRD11 [local project database; accessed 6 Oct 2026].
5. Clinical Genome Resource (ClinGen). Gene-disease validity curations: NIPBL, SMC1A, SMC3, RAD21, HDAC8, BRD4, ANKRD11 [export file accessed 6 Oct 2026]. Available at: https://search.clinicalgenome.org/kb/gene-validity
6. Barisic I, Tokic V, Loane M, Bianchi F, Calzolari E, Garne E, et al. Descriptive epidemiology of Cornelia de Lange syndrome in Europe. Am J Med Genet A. 2008;146A(1):51-9. doi:10.1002/ajmg.a.32016. PMID: 18074387.
7. Centre de Référence Maladies Osseuses Constitutionnelles; Centre de Référence Anomalies du développement et syndromes malformatifs. Protocole National de Diagnostic et de Soins (PNDS): Syndrome de Cornelia de Lange. Saint-Denis: Haute Autorité de Santé; November 2022. Available at: https://www.has-sante.fr/jcms/p_3385210/fr/syndrome-de-cornelia-de-lange
8. U.S. Food and Drug Administration. Search Orphan Drug Designations and Approvals [designation search for «Cornelia», «Brachmann», «CdLS» and “cohesin”, 1983 to 6 Oct 2026; accessed 6 Oct 2026]. Available at: https://www.accessdata.fda.gov/scripts/opdlisting/oopd/
9. European Medicines Agency. Medicines data: authorised medicines and orphan designations [accessed 6 Oct 2026]. Available at: https://www.ema.europa.eu/en/medicines/download-medicine-data
10. ClinicalTrials.gov. Cornelia De Lange Syndrome: Assessing Positive Effects of Lithium Treatment (CLoSER). NCT06789783. University of Milan [last updated 28 Jan 2025; accessed 6 Oct 2026]. Available at: https://clinicaltrials.gov/study/NCT06789783
11. ClinicalTrials.gov. Use of N-Acetylcysteine (NAC) in the Treatment of Repetitive Behaviors and Self-Injurious Behaviors in Cornelia de Lange Syndrome: a randomized double-blind placebo-controlled pilot study. NCT04381897. Johns Hopkins University [last updated 14 Aug 2026; accessed 6 Oct 2026]. Available at: https://clinicaltrials.gov/study/NCT04381897
12. Erker C. Anesthesia recommendations for patients with Cornelia de Lange syndrome. OrphanAnesthesia; November 2016. Available at: https://www.orphananesthesia.eu/en/rare-diseases/published-guidelines/cornelia-de-lange-syndrome.html
13. ERN ITHACA. European Network for Rare Malformation Syndromes [accessed 6 Oct 2026]. Available at: https://ern-ithaca.eu/
14. CdLS World — World Federation of CdLS Support Groups [accessed 6 Oct 2026]. Available at: https://www.cdlsworld.org/
15. Cornelia de Lange Syndrome Foundation [listed in GeneReviews; direct access blocked by automated verification on 6 Oct 2026]. Available at: https://www.cdlsusa.org/
16. Vega H, Gordillo M, Jabs EW. ESCO2 spectrum disorder. 2006 Apr 18 [updated 2025 Aug 14]. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available at: https://www.ncbi.nlm.nih.gov/books/NBK1153/
17. OMIM. #268300 Roberts syndrome; RBS [entry 269000, SC phocomelia syndrome, flagged as “moved/removed” in the mim2gene.txt file generated on 15 May 2026; gene ESCO2, MIM 609353]. Baltimore (MD): Johns Hopkins University. Available at: https://omim.org/entry/268300
18. Orphanet. Roberts syndrome (ORPHA:3103) [accessed 1 Nov 2026]. Available at: https://www.orpha.net/en/disease/detail/3103
19. Jones KL, Jones MC, del Campo M. Cornelia de Lange syndrome (Brachmann–de Lange syndrome). In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 100-5.
20. Jackson L, Kline AD, Barr MA, Koch S. de Lange syndrome: a clinical review of 310 individuals. Am J Med Genet. 1993;47(7):940-6. doi:10.1002/ajmg.1320470703. PMID: 8291537.
21. Krantz ID, McCallum J, DeScipio C, Kaur M, Gillis LA, Yaeger D, et al. Cornelia de Lange syndrome is caused by mutations in NIPBL, the human homolog of Drosophila melanogaster Nipped-B. Nat Genet. 2004;36(6):631-5. doi:10.1038/ng1364. PMID: 15146186.
22. Tonkin ET, Wang TJ, Lisgo S, Bamshad MJ, Strachan T. NIPBL, encoding a homolog of fungal Scc2-type sister chromatid cohesion proteins and fly Nipped-B, is mutated in Cornelia de Lange syndrome. Nat Genet. 2004;36(6):636-41. doi:10.1038/ng1363. PMID: 15146185.
23. Musio A, Selicorni A, Focarelli ML, Gervasini C, Milani D, Russo S, et al. X-linked Cornelia de Lange syndrome owing to SMC1L1 mutations. Nat Genet. 2006;38(5):528-30. doi:10.1038/ng1779. PMID: 16604071.
24. Deardorff MA, Kaur M, Yaeger D, Rampuria A, Korolev S, Pie J, et al. Mutations in cohesin complex members SMC3 and SMC1A cause a mild variant of Cornelia de Lange syndrome with predominant mental retardation. Am J Hum Genet. 2007;80(3):485-94. doi:10.1086/511888. PMID: 17273969.
25. Deardorff MA, Bando M, Nakato R, Watrin E, Itoh T, Minamino M, et al. HDAC8 mutations in Cornelia de Lange syndrome affect the cohesin acetylation cycle. Nature. 2012;489(7415):313-7. doi:10.1038/nature11316. PMID: 22885700.
26. Deardorff MA, Wilde JJ, Albrecht M, Dickinson E, Tennstedt S, Braunholz D, et al. RAD21 mutations cause a human cohesinopathy. Am J Hum Genet. 2012;90(6):1014-27. doi:10.1016/j.ajhg.2012.04.019. PMID: 22633399.
27. Olley G, Ansari M, Bengani H, Grimes GR, Rhodes J, von Kriegsheim A, et al. BRD4 interacts with NIPBL and BRD4 is mutated in a Cornelia de Lange-like syndrome. Nat Genet. 2018;50(3):329-32. doi:10.1038/s41588-018-0042-y. PMID: 29379197.
28. Parenti I, Diab F, Gil SR, Mulugeta E, Casa V, Berutti R, et al. MAU2 and NIPBL variants impair the heterodimerization of the cohesin loader subunits and cause Cornelia de Lange syndrome. Cell Rep. 2020;31(7):107647. doi:10.1016/j.celrep.2020.107647. PMID: 32433956.
29. Parenti I, Hesters A, Gil-Salvador M, Duffy L, Kanber D, Beygo J, et al. Pathogenic variants in the cohesin loader subunit MAU2 underlie a distinct Cornelia de Lange Syndrome subtype. Nat Commun. 2026;17(1):3036. doi:10.1038/s41467-026-71177-6. PMID: 41912533.
30. Kline AD, Krantz ID, Sommer A, Kliewer M, Jackson LG, FitzPatrick DR, et al. Cornelia de Lange syndrome: clinical review, diagnostic and scoring systems, and anticipatory guidance. Am J Med Genet A. 2007;143A(12):1287-96. doi:10.1002/ajmg.a.31757. PMID: 17508425.
31. Jones KL, Jones MC, del Campo M. Roberts syndrome (Roberts–SC phocomelia syndrome, SC phocomelia syndrome). In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 424-7.
32. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. Roberts/SC phocomelia syndrome (MIM 268300, 269000). In: Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. p. 866-8.
33. Van Den Berg DJ, Francke U. Roberts syndrome: a review of 100 cases and a new rating system for severity. Am J Med Genet. 1993;47(7):1104-23. doi:10.1002/ajmg.1320470735. PMID: 8291532.
34. Vega H, Waisfisz Q, Gordillo M, Sakai N, Yanagihara I, Yamada M, et al. Roberts syndrome is caused by mutations in ESCO2, a human homolog of yeast ECO1 that is essential for the establishment of sister chromatid cohesion. Nat Genet. 2005;37(5):468-70. doi:10.1038/ng1548. PMID: 15821733.
35. Schüle B, Oviedo A, Johnston K, Pai S, Francke U. Inactivating mutations in ESCO2 cause SC phocomelia and Roberts syndrome: no phenotype-genotype correlation. Am J Hum Genet. 2005;77(6):1117-28. doi:10.1086/498695. PMID: 16380922.
36. Vega H, Trainer AH, Gordillo M, Crosier M, Kayserili H, Skovby F, et al. Phenotypic variability in 49 cases of ESCO2 mutations, including novel missense and codon deletion in the acetyltransferase domain, correlates with ESCO2 expression and establishes the clinical criteria for Roberts syndrome. J Med Genet. 2010;47(1):30-7. doi:10.1136/jmg.2009.068395. PMID: 19574259.
37. Alkhunaizi E, Brosh RM, Chitayat D. DDX11-related cohesinopathy. 2019 Jun 6 [updated 2025 Jun 5]. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available at: https://www.ncbi.nlm.nih.gov/books/NBK541972/
38. Boone PM, Erdin S, Mohamed A, Haghshenas S, Faour KNW, Kao E, et al. Clinical, in vitro, and in vivo evidence of WAPL as a cohesinopathy-associated gene and phenotypic driver of 10q22.3q23.2 genomic disorder. Am J Hum Genet. 2026;113(8):1691-718. doi:10.1016/j.ajhg.2026.06.012. PMID: 42431198.
39. Morel Swols D, Tekin M. ANKRD11-related KBG syndrome. 2018 Mar 22 [updated 2026 Jun 11]. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available at: https://www.ncbi.nlm.nih.gov/books/NBK487886/
40. Liu H, Li H, Cai Q, Zhang J, Zhong H, Hu G, et al. ANKRD11 binding to cohesin suggests a connection between KBG syndrome and Cornelia de Lange syndrome. Proc Natl Acad Sci U S A. 2025;122(4):e2417346122. doi:10.1073/pnas.2417346122. PMID: 39847329.
41. VariantValidator. MANE Select transcripts (GRCh38) and variant validation (NM_006306.4:c.2369G>A, NM_006306.3:c.2369G>A, NM_006306.4:c.1553G>A, NM_006306.4:c.1904G>A) [accessed 6 Oct 2026]. Available at: https://variantvalidator.org/
42. 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. PMID: 28967166.
43. Feijó CM, Mororó GP, Ribeiro EM. Síndrome de Cornelia de Lange. In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 228-32.
44. Firth HV, Hurst JA. Hypertrichosis. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 178-81.
45. Kaur M, Blair J, Devkota B, Fortunato S, Clark D, Lawrence A, et al. Genomic analyses in Cornelia de Lange syndrome and related diagnoses: novel candidate genes, genotype-phenotype correlations and common mechanisms. Am J Med Genet A. 2023;191(8):2113-31. doi:10.1002/ajmg.a.63247. PMID: 37377026.
46. Ansari M, Halachev M, Parry D, Campos JL, D'Souza EN, Barnett C, et al. Whole genome sequencing of "mutation-negative" individuals with Cornelia de Lange syndrome. Hum Mutat. 2025;2025:4711663. doi:10.1155/humu/4711663. PMID: 40677927.
47. Latorre-Pellicer A, Gil-Salvador M, Parenti I, Lucia-Campos C, Trujillano L, Marcos-Alcalde I, et al. Clinical relevance of postzygotic mosaicism in Cornelia de Lange syndrome and purifying selection of NIPBL variants in blood. Sci Rep. 2021;11(1):15459. doi:10.1038/s41598-021-94958-z. PMID: 34326454.
48. Huisman S, Mulder PA, Redeker E, Bader I, Bisgaard AM, Brooks A, et al. Phenotypes and genotypes in individuals with SMC1A variants. Am J Med Genet A. 2017;173(8):2108-25. doi:10.1002/ajmg.a.38279. PMID: 28548707.
49. Symonds JD, Joss S, Metcalfe KA, Somarathi S, Cruden J, Devlin AM, et al. Heterozygous truncation mutations of the SMC1A gene cause a severe early onset epilepsy with cluster seizures in females: detailed phenotyping of 10 new cases. Epilepsia. 2017;58(4):565-75. doi:10.1111/epi.13669. PMID: 28166369.
50. Gil-Rodríguez MC, Deardorff MA, Ansari M, Tan CA, Parenti I, Baquero-Montoya C, et al. De novo heterozygous mutations in SMC3 cause a range of Cornelia de Lange syndrome-overlapping phenotypes. Hum Mutat. 2015;36(4):454-62. doi:10.1002/humu.22761. PMID: 25655089.
51. Ansari M, Faour KNW, Shimamura A, Grimes G, Kao EM, Denhoff ER, et al. Heterozygous loss-of-function SMC3 variants are associated with variable growth and developmental features. HGG Adv. 2024;5(2):100273. doi:10.1016/j.xhgg.2024.100273. PMID: 38297832.
52. Krab LC, Marcos-Alcalde I, Assaf M, Balasubramanian M, Andersen JB, Bisgaard AM, et al. Delineation of phenotypes and genotypes related to cohesin structural protein RAD21. Hum Genet. 2020;139(5):575-92. doi:10.1007/s00439-020-02138-2. PMID: 32193685.
53. Kaiser FJ, Ansari M, Braunholz D, Gil-Rodríguez MC, Decroos C, Wilde JJ, et al. Loss-of-function HDAC8 mutations cause a phenotypic spectrum of Cornelia de Lange syndrome-like features, ocular hypertelorism, large fontanelle and X-linked inheritance. Hum Mol Genet. 2014;23(11):2888-900. doi:10.1093/hmg/ddu002. PMID: 24403048.
54. Lonardo F, Falco M, Costabile C, Fontana P. Non-classic Cornelia de Lange syndrome due to BRD4 gene alterations: a literature review. Children (Basel). 2025;12(11):1440. doi:10.3390/children12111440. PMID: 41300558.
55. Ansari M, Poke G, Ferry Q, Williamson K, Aldridge R, Meynert AM, et al. Genetic heterogeneity in Cornelia de Lange syndrome (CdLS) and CdLS-like phenotypes with observed and predicted levels of mosaicism. J Med Genet. 2014;51(10):659-68. doi:10.1136/jmedgenet-2014-102573. PMID: 25125236.
56. Huisman SA, Redeker EJ, Maas SM, Mannens MM, Hennekam RC. High rate of mosaicism in individuals with Cornelia de Lange syndrome. J Med Genet. 2013;50(5):339-44. doi:10.1136/jmedgenet-2012-101477. PMID: 23505322.
57. Gil-Salvador M, Trujillano L, Lucia-Campos C, Del Rincón J, Pamplona P, Arnedo M, et al. Postzygotic mosaicism in SMC1A and the first reported case of a female with Cornelia de Lange syndrome. Sci Rep. 2025;15(1):20772. doi:10.1038/s41598-025-07268-z. PMID: 40593079.
58. Aref-Eshghi E, Kerkhof J, Pedro VP, Groupe DI France, Barat-Houari M, Ruiz-Pallares N, et al. Evaluation of DNA methylation episignatures for diagnosis and phenotype correlations in 42 Mendelian neurodevelopmental disorders. Am J Hum Genet. 2020;106(3):356-70. doi:10.1016/j.ajhg.2020.01.019. PMID: 32109418.
59. Tae SK, Mazlan RA, Thong MK. Case report: The evolving phenotype of ESCO2 spectrum disorder in a 15-year-old Malaysian child. Front Genet. 2023;14:1286489. doi:10.3389/fgene.2023.1286489. PMID: 38288163.
60. Gordillo M, Vega H, Trainer AH, Hou F, Sakai N, Luque R, et al. The molecular mechanism underlying Roberts syndrome involves loss of ESCO2 acetyltransferase activity. Hum Mol Genet. 2008;17(14):2172-80. doi:10.1093/hmg/ddn116. PMID: 18411254.
61. Strasser AS, Gonzalez-Reiche AS, Zhou X, Valdebenito-Maturana B, Ye X, Zhang B, et al. Limb reduction in an Esco2 cohesinopathy mouse model is mediated by p53-dependent apoptosis and vascular disruption. Nat Commun. 2024;15(1):7154. doi:10.1038/s41467-024-51328-3. PMID: 39168984.
62. VariantValidator. Normalisation of ESCO2 variants on NM_001017420.3 (MANE Select) and GRCh38 [accessed 1 Nov 2026]. Available at: https://variantvalidator.org/
63. ClinVar. ESCO2: VCV000001736, VCV000001738, VCV000021233, VCV000021250, VCV000021251, VCV001027671 [accessed 1 Nov 2026]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/
64. Kantaputra PN, Dejkhamron P, Intachai W, Ngamphiw C, Kawasaki K, Ohazama A, et al. Juberg-Hayward syndrome is a cohesinopathy, caused by mutation in ESCO2. Eur J Orthod. 2021;43(1):45-50. doi:10.1093/ejo/cjaa023. PMID: 32255174.
65. Kantaputra PN, Dejkhamron P, Tongsima S, Ngamphiw C, Intachai W, Ngiwsara L, et al. Juberg-Hayward syndrome and Roberts syndrome are allelic, caused by mutations in ESCO2. Arch Oral Biol. 2020;119:104918. doi:10.1016/j.archoralbio.2020.104918. PMID: 32977150.
66. Hoedemaekers YM, Germans T. Left ventricular noncompaction. In: Baars HF, van der Smagt JJ, Doevendans PAFM, van Tintelen JP, editors. Clinical Cardiogenetics. 3rd ed. Cham: Springer; 2020. p. 125 (Table 7.5).
67. Vilella T, Nunes BC, Del Valle GS, Pinheiro IF, Aoi H, Mizuguchi T, et al. Ophthalmological findings in Brazilian Cornelia de Lange syndrome patients with NIPBL variants. BMC Ophthalmol. 2025;25(1):567. doi:10.1186/s12886-025-04401-4. PMID: 41088053.
68. He S, Chen S, Li SJ, Zhang JW, Liang XL. Complex cerebrovascular diseases in Roberts syndrome caused by novel biallelic ESCO2 variations. Mol Genet Genomic Med. 2023;11(6):e2177. doi:10.1002/mgg3.2177. PMID: 37002187.
69. Firth HV, Hurst JA. Short stature. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 312-5.
70. Firth HV, Hurst JA. Radial ray defects and thumb hypoplasia. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 296-9.
71. Firth HV, Hurst JA. Limb reduction defects. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 212-5.
72. Meng L, Wang Y, Zhang J, Zhou R, Wang X, Qiao F, et al. Preimplantation genetic testing for Cornelia de Lange syndrome with low-level maternal gonadal mosaicism for a sub-megabase deletion in China. Appl Clin Genet. 2026;19:1-14. doi:10.2147/TACG.S603242. PMID: 42502680.
73. Xie Y, Li D, Li W, Tao Y, Tang L, Li L, et al. First Cornelia de Lange syndrome type 4 caused by the gonadal mosaicism of RAD21 deletion. Hum Mutat. 2026;2026:5591607. doi:10.1155/humu/5591607. PMID: 42719089.
74. ClinVar. VCV000212267: NM_006306.4(SMC1A):c.2369G>A (p.Arg790Gln) [accessed 6 Oct 2026]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/variation/212267/
75. ClinVar. VCV001397513: NM_006306.4(SMC1A):c.1553G>A (p.Arg518His) [accessed 6 Oct 2026]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/variation/1397513/
76. ClinVar. VCV001678779: NM_006306.4(SMC1A):c.1904G>A (p.Arg635His) [accessed 6 Oct 2026]. Available at: https://www.ncbi.nlm.nih.gov/clinvar/variation/1678779/
77. Clark DM, Sherer I, Deardorff MA, Byrne JL, Loomes KM, Nowaczyk MJ, et al. Identification of a prenatal profile of Cornelia de Lange syndrome (CdLS): a review of 53 CdLS pregnancies. Am J Med Genet A. 2012;158A(8):1848-56. doi:10.1002/ajmg.a.35410. PMID: 22740382.
78. Firth HV, Hurst JA. Congenital diaphragmatic hernia. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 714-5.
79. Firth HV, Hurst JA. Short limbs. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 784-7.
80. Jones KL, Jones MC, del Campo M. KBG syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 162-5.
81. Izumi K, Raible S, Krantz I. AFF4-related CHOPS syndrome. 2026 Feb 19. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available at: https://www.ncbi.nlm.nih.gov/books/NBK621037/
82. Jones KL, Jones MC, del Campo M. Duplication 3q syndrome. In: Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 30-1.
83. Firth HV, Hurst JA. DNA repair disorders. In: Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 398-401.
84. U.S. Food and Drug Administration. PRILOSEC (omeprazole magnesium) for delayed-release oral suspension. Prescribing information, NDA022056, version of 19 Mar 2024 [via openFDA; accessed 6 Oct 2026]. Available at: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=b6761f84-53ac-4745-a8c8-1e5427d7e179
85. European Medicines Agency. Slenyto (melatonin) prolonged-release tablets: EPAR — product information (summary of product characteristics), sections 4.1 and 4.2. EMEA/H/C/004425; first authorisation 20 Sep 2018; extension to neurogenetic disorders (II/0025) 26 Aug 2024; ADHD indication (II/0028) 17 Mar 2025; product information updated on 10 Jun 2026 [accessed 7 Nov 2026]. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/slenyto
86. U.S. Food and Drug Administration. RISPERDAL (risperidone) tablets, oral solution; RISPERDAL M-TAB. Prescribing information, NDA020272/NDA020588/NDA021444, version of 28 May 2026 [via openFDA; accessed 6 Oct 2026]. Available at: https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=7e117c7e-02fc-4343-92a1-230061dfc5e0
87. European Medicines Agency. Risperdal — Article 30 referral: Annex I–IV (harmonised summary of product characteristics, sections 4.1 and 4.2); CHMP opinion of 24 Jul 2008 and European Commission decision of 7 Oct 2008. Available at: https://www.ema.europa.eu/en/medicines/human/referrals/risperdal
88. ClinicalTrials.gov. Behavioral Assessment and Treatment of Problem Behavior in Children With Cornelia de Lange Syndrome. NCT05829668. Hugo W. Moser Research Institute at Kennedy Krieger [last updated 29 Oct 2025; accessed 6 Oct 2026]. Available at: https://clinicaltrials.gov/study/NCT05829668
89. European Medicines Agency. Clinical Trials Information System (CTIS), public search “Cornelia de Lange” [accessed 6 Oct 2026]. Available at: https://euclinicaltrials.eu/
90. Parodi C, Lettieri A, Grazioli P, Di Fede E, Grassi S, Taci E, et al. Lithium ameliorates neural differentiation restoring cell death balance in Cornelia de Lange syndrome 2D and 3D models. Cell Death Discov. 2026;12(1):203. doi:10.1038/s41420-026-03085-z. PMID: 41904182.
91. Apis A, Nunes BC, Vilella T, Pereira RCM, Pilotto RF, de Avó LRDS, et al. Quality of life of families who have children with Cornelia de Lange syndrome in Brazil: opportunities for improvement. Am J Med Genet A. 2025;197(10):e64120. doi:10.1002/ajmg.a.64120. PMID: 40395212.
92. Rigotti L, Rebellato S, Lettieri A, Castiglioni S, Mariani M, Totaro S, et al. Cohesins: crossroad between Cornelia de Lange spectrum and cancer predisposition. Am J Med Genet A. 2025;197(8):e64076. doi:10.1002/ajmg.a.64076. PMID: 40186517.
93. ClinicalTrials.gov. Search by condition “Roberts syndrome OR ESCO2 OR SC phocomelia” (API v2) [accessed 1 Nov 2026]. Available at: https://clinicaltrials.gov/
94. Zhou J, Yang X, Jin X, Jia Z, Lu H, Qi Z. Long-term survival after corrective surgeries in two patients with severe deformities due to Roberts syndrome: a case report and review of the literature. Exp Ther Med. 2018;15(2):1702-11. doi:10.3892/etm.2017.5592. PMID: 29434756.
95. Slavin TP, Lazebnik N, Clark DM, Vengoechea J, Cohen L, Kaur M, et al. Germline mosaicism in Cornelia de Lange syndrome. Am J Med Genet A. 2012;158A(6):1481-5. doi:10.1002/ajmg.a.35381. PMID: 22581668.
96. da Costa Almeida CB, Welter AT, Abech GD, Brandão GR, Flores JAM, Schüle B, et al. Report of the phenotype of a patient with Roberts syndrome and a rare ESCO2 variant. J Pediatr Genet. 2020;9(1):58-62. doi:10.1055/s-0039-1696636. PMID: 31976146.
97. Schrier SA, Sherer I, Deardorff MA, Clark D, Audette L, Gillis L, et al. Causes of death and autopsy findings in a large study cohort of individuals with Cornelia de Lange syndrome and review of the literature. Am J Med Genet A. 2011;155A(12):3007-24. doi:10.1002/ajmg.a.34329. PMID: 22069164.