Cornelia de Lange Syndrome 4

Mendelian MONDO:0013864 Pathograph 20 Show in embeddings browser Cornelia de Lange syndrome

Cornelia de Lange syndrome 4 (CdLS4) is the RAD21-related form of the cohesinopathy spectrum, and it is defined as much by what it is milder than as by what it is. RAD21 encodes the kleisin subunit that closes the cohesin ring between SMC1A and SMC3 and recruits STAG1/STAG2, so a RAD21 lesion sits inside the ring itself rather than in the machinery that loads it — which is where NIPBL, the gene behind classic CdLS, acts. The disease-causing variants cluster at the RAD21 interfaces with its partner subunits, impair the cellular DNA damage response, and disrupt transcription; the resulting phenotype is an attenuated CdLS, with the facial and limb features present but muted and, most strikingly, cognition and behaviour far less affected than in NIPBL- or SMC1A-related disease. Two further things separate it from classic CdLS in practice. Variants are frequently familial rather than de novo, with striking variability between and within families, so an affected parent may never have been recognised. And RAD21 is dose- and zygosity-split: two RAD21 alleles do not give a more severe CdLS but a different disease entirely, Mungan syndrome, an enteric neuromyopathy with chronic intestinal pseudo-obstruction.

Ask OpenScientist

Ask a research question about Cornelia de Lange Syndrome 4. OpenScientist will conduct autonomous deep research using the Disorder Mechanisms Knowledge Base and PubMed literature (typically 10-30 minutes).

Submitting...

Do not include personal health information in your question. Questions and results are cached in your browser's local storage.

1
Inheritance
5
Pathophys.
15
Phenotypes
3
Gaps
20
Pathograph
1
Genes
2
Medical Actions
3
Differentials
1
Models
1
References
👪

Inheritance

1
Autosomal Dominant HP:0000006
Heterozygous RAD21 variants, transmitted dominantly. Unlike classic NIPBL-related CdLS, which is overwhelmingly de novo, RAD21 variants are frequently inherited: 12 of 28 patients with a known transmission pattern in a systematic review carried a germline variant from a parent. That, together with the attenuated phenotype, means an affected parent may be recognised only after the child is diagnosed.
Autosomal dominant inheritance
Show evidence (2 references)
CGGV:assertion_30ad5023-6c68-42ab-b8ea-a601a134eaf6-2020-01-08T170000.000Z SUPPORT DIRECT Other
"RAD21 | HGNC:9811 | Cornelia de Lange syndrome | MONDO:0016033 | AD | Definitive"
ClinGen classifies the RAD21-CdLS gene-disease relationship as definitive with autosomal dominant inheritance. Note this assertion is keyed to the CdLS umbrella term MONDO:0016033 and makes no statement about the attenuated RAD21 phenotype; it is cited here only for inheritance and gene-disease validity.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Variants were frequently familial, and genotype-phenotype analyses demonstrated striking interfamilial and intrafamilial variability."
Establishes that RAD21 variants are often inherited rather than de novo, which is the counselling difference from classic CdLS.
?

Discussions and Knowledge Gaps

3
Why does a lesion in the cohesin ring itself produce a milder phenotype than reduced dosage of the factor that loads it?
KNOWLEDGE GAP cdls4_attenuation_mechanism
Attached to
pathophysiology#Attenuated Cornelia de Lange Phenotype
The attenuation is the best-replicated fact about CdLS4 — reported in the entity-defining paper and confirmed in the largest series — and it is counterintuitive. A structural subunit of the ring might be expected to matter at least as much as its loader. Two partial explanations are on the record and neither has been tested: that the missense-worse-than-LOF gradient means RAD21 haploinsufficiency is comparatively well tolerated while missense protein interferes actively, and that cohesin composition rather than dosage determines developmental outcome. No study has measured chromatin-bound cohesin or transcriptional output side by side in NIPBL and RAD21 patient cells, which is what would separate them. The RAD21 series offers one concrete lead, which this entry records but does not assert as mechanism: Deardorff found that RAD21 haploinsufficiency approximately halves RAD21 RNA in a patient cell line, whereas NIPBL haploinsufficiency is often accompanied by compensatory upregulation from the intact allele. RAD21 therefore appears simply more tolerant of reduced dosage than NIPBL is, despite having no compensatory mechanism -- which inverts the usual expectation and is the opposite of what a compensation argument would predict.
Why do two RAD21 alleles give an enteric neuromyopathy rather than a more severe Cornelia de Lange syndrome?
KNOWLEDGE GAP cdls4_zygosity_split
Attached to
genetic#RAD21
Dose-response is the usual expectation, and RAD21 does not follow it. Heterozygous variants give attenuated CdLS; biallelic variants give Mungan syndrome, whose reported mechanism runs through specific transcriptional targets — failure to bind the APOB promoter, reduced RUNX1 — and whose zebrafish phenotype is a deficit of enteric neurons. Whether the two presentations reflect different residual cohesin function, different tissue-specific target sets, or an allele-specific effect of the particular biallelic variant reported has not been addressed. No published work compares heterozygous and homozygous RAD21 patient material directly.
Does any model reproduce the attenuation that defines CdLS4?
HUMAN MODEL MISMATCH cdls4_no_model_of_attenuation
Attached to
animal_models#Zebrafish
The one in vivo model reported for CdLS4 is the zebrafish transcription assay from the defining paper, which tests whether patient variants disrupt transcription and nothing else. There is no mouse, no knock-in at the endogenous locus, and no model in which the RAD21 phenotype has been compared with a Nipbl model under the same conditions. The defining clinical claim of this entity — milder than NIPBL disease, especially for cognition — therefore has no experimental counterpart at all, and the mechanism discussion above cannot be resolved without one.
⚙

Pathophysiology

5
Heterozygous RAD21 Variant
A single altered RAD21 allele. The variant spectrum is dominated by frameshift alleles (36% of 36 published patients), followed by missense (19%) and nonsense (17%), with whole-gene and intragenic microdeletions also reported. Where variants have been mapped to protein domains, most fall in the RAD21-SMC1A interaction domain. The allelic gradient runs the opposite way to intuition for a structural subunit: dominant missense alleles cause worse structural and cognitive findings than loss-of-function alleles do, which argues that the missense protein interferes with the ring rather than simply failing to join it.
Genetic context RAD21 hgnc:9811 HUGO Gene Nomenclature Committee (hgnc) Relation: this genetic context concerns this gene This genetic context concerns RAD21 (hgnc:9811). hgnc:9811 is a gene from the HUGO Gene Nomenclature Committee. zygosity: HETEROZYGOUS
Downstream
Disruption of the Cohesin Ring at the RAD21 Subunit Interfaces RAD21 is the subunit that bridges SMC1A, SMC3 and STAG, so a lesion in it acts on those interfaces.
Show evidence (2 references)
PMID:22633399 SUPPORT DIRECT Human Clinical
"Our data suggest that, compared to loss-of-function mutations, dominant missense mutations result in more severe functional defects and cause worse structural and cognitive clinical findings."
The allelic gradient this node describes, in the paper that established RAD21 as a CdLS gene.
PMID:39286962 SUPPORT DIRECT Human Clinical
"Frameshift variants constituted the predominant variant type, representing 36% (13/36) of cases."
Quantifies the variant spectrum across all published CdLS4 patients.
Disruption of the Cohesin Ring at the RAD21 Subunit Interfaces
RAD21 is the kleisin that closes the tripartite cohesin ring: its N-terminus engages SMC3, its C-terminus engages SMC1, and its central STAG domain recruits STAG1 or STAG2. CdLS4 variants act at those interfaces. This is the step that distinguishes CdLS4 from classic CdLS at the level of mechanism — NIPBL is the cohesin loading factor and NIPBL-related disease is a dosage problem in getting cohesin onto chromatin, whereas CdLS4 is a lesion in the ring itself.
cohesin complex GO:0008278 Gene Ontology (GO) Relation: this pathophysiological event involves this cellular component This pathophysiological event involves cohesin complex (GO:0008278). GO:0008278 is a cellular component from the Gene Ontology.
Downstream
Cohesin-Dependent Transcriptional Dysregulation The developmental phenotype of the cohesinopathies is attributed primarily to cohesin's interphase role in transcription and genome organization.
Impaired DNA Damage Response Patient RAD21 variants impair the cellular response to DNA damage.
Show evidence (2 references)
PMID:22633399 SUPPORT DIRECT In Vitro
"Mechanistically, these mutations act at the RAD21 interface with the other cohesin proteins STAG2 and SMC1A"
Localizes the lesion to the RAD21 subunit interfaces. Only the interface clause of the source sentence is quoted here; the sentence's other two claims are quoted separately on the nodes they belong to, because the transcription result is a zebrafish experiment and the DNA damage result is a cell-based one.
PMID:39286962 SUPPORT DIRECT Human Clinical
"Among the patients, 41.9% (13 out of 31) had variants situated in the RAD21-SMC1A domain"
Shows that the variants patients actually carry concentrate in one of the subunit interaction domains.
Cohesin-Dependent Transcriptional Dysregulation
Beyond holding sister chromatids together, cohesin organizes the genome in interphase and regulates transcription, and it is this arm rather than mitotic cohesion failure that is held responsible for the developmental phenotype. Disrupted transcription was demonstrated directly for CdLS4 variants in a zebrafish model.
neuron CL:0000540 Cell Ontology (CL) Relation: this pathophysiological event involves this cell type This pathophysiological event involves neuron (CL:0000540). CL:0000540 is a cell type from the Cell Ontology.
regulation of DNA-templated transcription GO:0006355 Gene Ontology (GO) Relation: this pathophysiological event involves this biological process This pathophysiological event involves dysregulated regulation of DNA-templated transcription (GO:0006355). GO:0006355 is a biological process from the Gene Ontology. ↕ DYSREGULATED chromatin organization GO:0006325 Gene Ontology (GO) Relation: this pathophysiological event involves this biological process This pathophysiological event involves dysregulated chromatin organization (GO:0006325). GO:0006325 is a biological process from the Gene Ontology. ↕ DYSREGULATED
Animal models
Zebrafish expressing CdLS4 RAD21 variants Zebrafish recapitulates Patient RAD21 variants disrupt transcription in zebrafish, which is the arm of cohesin biology that the developmental phenotype is attributed to.
Downstream
Attenuated Cornelia de Lange Phenotype How transcriptional dysregulation produces the specific pattern of facial, limb, growth and cognitive features is not established for any cohesinopathy, let alone for RAD21 specifically.
Show evidence (2 references)
PMID:22633399 SUPPORT DIRECT Model Organism
"and disrupt transcription in a zebrafish model"
The transcriptional consequence, demonstrated in zebrafish expressing the patient variants. Quoted as the zebrafish clause alone so this item and the cell-based items from the same sentence each carry one evidence_source.
PMID:32687945 SUPPORT INDIRECT Other
"In interphase, cohesin also functions in the control of gene expression by binding to numerous sites within the genome."
Establishes the interphase transcriptional role this node depends on. Graded INDIRECT because it states cohesin biology generally rather than a CdLS4 result.
Impaired DNA Damage Response
RAD21 is also a double-strand-break repair protein, and CdLS4 variants impair the cellular DNA damage response. This arm is well demonstrated biochemically but is not connected to any specific clinical feature of CdLS4, so it is curated as a consequence of the ring lesion without a downstream edge into the phenotype.
DNA damage response GO:0006974 Gene Ontology (GO) Relation: this pathophysiological event involves this biological process This pathophysiological event involves decreased DNA damage response (GO:0006974). GO:0006974 is a biological process from the Gene Ontology. ↓ DECREASED
Show evidence (2 references)
PMID:22633399 SUPPORT DIRECT In Vitro
"impair cellular DNA damage response"
The DNA damage response defect measured in cells carrying patient variants.
PMID:32687945 SUPPORT INDIRECT Other
"RAD21 (also known as KIAA0078, NXP1, HR21, Mcd1, Scc1, and hereafter called RAD21), an essential gene, encodes a DNA double-strand break (DSB) repair protein that is evolutionarily conserved in all eukaryotes from budding yeast to humans."
Establishes the repair function whose impairment this node describes. Graded INDIRECT because it states the protein's role rather than a CdLS4 measurement.
Attenuated Cornelia de Lange Phenotype
The clinical result is recognisably CdLS but muted. Compared with NIPBL- and SMC1A-related disease the facial morphology and limb anomalies are milder, and cognition and behaviour are considerably less affected — 14 of 26 scored patients were mild and only one severe on the CdLS severity score. Because the consensus clinical score is built around the classic phenotype, CdLS4 patients score low on it, which is a diagnostic problem rather than a cosmetic difference.
Downstream
Intellectual Disability The cognitive component, attenuated relative to classic CdLS.
Behavioural Phenotype The behavioural component, and the sharpest attenuation in the entry.
Microcephaly Postnatal head growth restriction.
Ptosis One of the few features more frequent in RAD21 than in NIPBL disease.
Congenital Heart Disease Cardiac malformation, major or minor.
Gastroesophageal Reflux Disease The commonest gastrointestinal manifestation.
Hirsutism Excess body hair, markedly less frequent than in the comparison genes.
Clinodactyly of the 5th Finger The other digital feature alongside the short fifth finger.
Delayed Speech and Language Development Verbal developmental delay, near-universal in reported patients.
Characteristic Facial Morphology The CdLS facial gestalt, present but muted.
Short Stature Growth restriction.
Short 5th Finger The limb component; reduction defects are milder than in classic CdLS.
Hearing Impairment Hearing loss in a third of reported patients.
Show evidence (6 references)
PMID:22633399 SUPPORT DIRECT Human Clinical
"Notably, unlike children with mutations in NIPBL, SMC1A, or SMC3, these individuals have much milder cognitive impairment than those with classical CdLS."
The attenuation claim, in the paper that established the entity.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Their phenotype is an attenuated CdLS phenotype compared to that caused by variants in NIPBL or SMC1A for facial morphology, limb anomalies, and especially for cognition and behavior."
Independent confirmation of attenuation in the largest RAD21 series, and the source that specifies which domains are attenuated.
PMID:39286962 SUPPORT DIRECT Human Clinical
"it was found that among the 26 patients meeting the criteria, only 1 was classified as severe, 11 reached moderate levels, and 14 were considered mild"
Quantifies the severity distribution using the CdLS scoring system.
+ 3 more references
⬡

Pathograph

Use the checkboxes to hide or show graph categories. Hover nodes for evidence and cross-linked metadata.
Pathograph: causal mechanism network for Cornelia de Lange Syndrome 4 Interactive directed graph showing how pathophysiology mechanisms, phenotypes, genetic factors and variants, experimental models, environmental triggers, and treatments relate through causal and linked edges.
●

Phenotypes

15
Cardiovascular 1
Congenital Heart Disease FREQUENT Abnormal heart morphology HP:0001627 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Congenital heart malformation, annotated with Abnormal heart morphology (HP:0001627). HP:0001627 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Heart (major and minor) HP:0001627 9/23 39 13/44 30 18/66 27"
39% in the RAD21 column falls in the FREQUENT band (30-79%).
Digestive 1
Gastroesophageal Reflux Disease FREQUENT HP:0002020 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Gastroesophageal reflux disease, annotated with Gastroesophageal reflux (HP:0002020). HP:0002020 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Gastroesophageal reflux disease HP:0002020 13/25 52 25/42 60 47/66 71"
52% in the RAD21 column falls in the FREQUENT band (30-79%).
Ear 1
Hearing Impairment FREQUENT HP:0000365 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Hearing impairment (HP:0000365). HP:0000365 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Vision is mostly normal; hearing loss is found in a third of individuals and may require hearing devices."
A third falls in the FREQUENT band (30-79%). The systematic review reports the same figure, 8 of 24 patients.
Eye 2
Visual Impairment VERY_RARE HP:0000505 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Visual impairment (HP:0000505). HP:0000505 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Visual impairment HP:0000505 0/24 0 20/38 53 29/66 44"
No visual impairment in 24 assessed RAD21 patients, against 53% for SMC1A and 44% for NIPBL. The systematic review reports 1 of 28, or 4%; both fall in the VERY_RARE band, and the band is set rather than the record dropped because the contrast with the comparison genes is the informative part.
Ptosis FREQUENT HP:0000508 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Ptosis (HP:0000508). HP:0000508 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Ptosis HP:0000508 11/26 42 4/40 10 8/42 19"
42% in the RAD21 column falls in the FREQUENT band (30-79%), against 19% for NIPBL.
Head and Neck 3
Characteristic Facial Morphology VERY_FREQUENT Long philtrum HP:0000343 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Cornelia de Lange facial gestalt (long or smooth philtrum, thick eyebrows, synophrys), annotated with Long philtrum (HP:0000343). HP:0000343 is a phenotype from the Human Phenotype Ontology.
Show evidence (4 references)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Long and/or smooth philtrum HP:0000343/HP:0000319 26/29 90 27/43 63 54/67 81"
A row of the RAD21 versus SMC1A versus NIPBL comparison table, whose columns are RAD21 (n = 29), SMC1A (n = 51) and NIPBL (n = 67). The RAD21 figure, 26/29 or 90%, sets the VERY_FREQUENT band.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Thick eyebrows HP:0000574 20/24 83 37/46 80 61/67 91"
Thick eyebrows in 83% of RAD21 patients, from the same comparison table.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Synophrys HP:0000664 19/28 68 37/46 80 61/67 91"
Synophrys, the single most recognisable CdLS facial sign, is present in 68% of RAD21 patients against 91% of NIPBL patients. This is the attenuation visible in one feature, and it is why the band for this record is set from the philtrum figure rather than from synophrys.
+ 1 more reference
Microcephaly FREQUENT HP:0000252 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Postnatal microcephaly, annotated with Microcephaly (HP:0000252). HP:0000252 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Postnatal head circumference < − 2SD HP:0000252 16/28 57 23/36 64 54/62 87"
57% in the RAD21 column falls in the FREQUENT band (30-79%). The row is the postnatal one; the same HPO term appears on a separate prenatal row with different figures.
Short Nose VERY_FREQUENT HP:0003196 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Short nose (HP:0003196). HP:0003196 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Short nose HP:0003196 23/26 88 26/46 57 58/67 87"
88% in the RAD21 column falls in the VERY_FREQUENT band (80-99%).
Integument 1
Hirsutism FREQUENT HP:0001007 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Hirsutism (HP:0001007). HP:0001007 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Hirsutism HP:0001007 10/26 38 37/47 79 37/43 86"
38% in the RAD21 column falls in the FREQUENT band (30-79%), against 86% for NIPBL.
Limbs 2
Short 5th Finger VERY_FREQUENT HP:0009237 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Short fifth finger, annotated with Short 5th finger (HP:0009237). HP:0009237 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Short 5th finger HP:0009237 23/28 82 21/45 47 42/63 67"
82% in the RAD21 column falls in the VERY_FREQUENT band (80-99%). Note it is higher than the NIPBL figure of 67%, so the mild limb involvement in CdLS4 is not simply a diluted version of the classic limb phenotype.
Clinodactyly of the 5th Finger FREQUENT HP:0004209 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Clinodactyly of the 5th finger (HP:0004209). HP:0004209 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Clinodactyly 5th finger HP:0004209 13/24 54 21/45 47 42/63 67"
54% in the RAD21 column falls in the FREQUENT band (30-79%).
Nervous System 3
Intellectual Disability VERY_FREQUENT HP:0001249 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Intellectual disability (HP:0001249). HP:0001249 is a phenotype from the Human Phenotype Ontology.
Show evidence (3 references)
PMID:39286962 SUPPORT DIRECT Human Clinical
"Clinical features included verbal developmental delay and intellectual disorder observed in 94% of patients."
94% falls in the VERY_FREQUENT band (80-99%).
PMID:32193685 SUPPORT DIRECT Human Clinical
"The majority of RAD21 patients (16/29, 55%) have normal or mildly impaired cognitive functioning (SMC1A group 32%; NIPBL group 7%)"
Quantifies the attenuation directly against the two comparison genes: a majority of RAD21 patients are cognitively normal or only mildly impaired, against 7% of NIPBL patients. This is the sharpest single number in the entry and is why the phenotype description says the degree, not the frequency, is what is attenuated.
PMID:22633399 SUPPORT DIRECT Human Clinical
"Notably, unlike children with mutations in NIPBL, SMC1A, or SMC3, these individuals have much milder cognitive impairment than those with classical CdLS."
Records that the severity, not the frequency, is what is attenuated.
Delayed Speech and Language Development VERY_FREQUENT HP:0000750 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Verbal developmental delay, annotated with Delayed speech and language development (HP:0000750). HP:0000750 is a phenotype from the Human Phenotype Ontology.
Show evidence (2 references)
PMID:39286962 SUPPORT DIRECT Human Clinical
"Clinical features included verbal developmental delay and intellectual disorder observed in 94% of patients."
94% falls in the VERY_FREQUENT band (80-99%). The same reported figure covers verbal delay and intellectual disability in this cohort.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Typical development is somewhat slow, mainly in speech development, and physical therapy or speech therapy may be indicated."
Characterizes the delay as speech-predominant and mild, which is the qualitative counterpart of the 94% frequency.
Behavioural Phenotype FREQUENT Anxiety HP:0000739 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Anxiety, attention deficit and obsessive-compulsive behaviour with little aggression or self-injury, annotated with Anxiety (HP:0000739). HP:0000739 is a phenotype from the Human Phenotype Ontology.
Show evidence (5 references)
PMID:32193685 SUPPORT DIRECT Human Clinical
"Obsessive–compulsive behavior 6/19 32 10/26h 38 Anxiety 10/19 53"
Two adjacent rows of the behavioural table, quoted together because the anxiety row on its own is too short to stand as a snippet. Obsessive-compulsive behaviour in 32% and anxiety in 53% of assessed RAD21 patients. The 53% falls in the FREQUENT band (30-79%) and is the figure this record's band is set from.
PMID:32193685 SUPPORT DIRECT Human Clinical
"Self-injurious behavior 1/18 6 11/31 35 47/61 77"
Self-injurious behaviour in 6% of RAD21 patients against 77% of NIPBL patients. This is the sharpest single attenuation contrast in the entry, sharper than the limb-malformation row, and it is the concrete content of the repeatedly quoted claim that the attenuation is "especially for cognition and behavior".
PMID:32193685 SUPPORT DIRECT Human Clinical
"Aggression 1/16 6 12/15h 80"
Aggression in 6% of RAD21 patients against 80% of SMC1A patients, the same contrast in a second behaviour.
+ 2 more references
Growth 1
Short Stature FREQUENT HP:0004322 Human Phenotype Ontology (HP) Relation: this clinical feature is this phenotype This clinical feature is Short stature (HP:0004322). HP:0004322 is a phenotype from the Human Phenotype Ontology.
Show evidence (1 reference)
PMID:39286962 SUPPORT DIRECT Human Clinical
"Among the 32 cases with height records at the time of diagnosis, 16 had a height of less than -2 SDS and could be diagnosed as "Short stature" (50%)."
50% falls in the FREQUENT band (30-79%).
🧬

Genetic Associations

1
RAD21 (Heterozygous Pathogenic Variants)
Gene: RAD21 hgnc:9811 HUGO Gene Nomenclature Committee (hgnc) Relation: this disease-associated gene is this gene This disease-associated gene is RAD21 (hgnc:9811). hgnc:9811 is a gene from the HUGO Gene Nomenclature Committee. relationship_type: CAUSATIVE
Show evidence (5 references)
CGGV:assertion_30ad5023-6c68-42ab-b8ea-a601a134eaf6-2020-01-08T170000.000Z SUPPORT DIRECT Other
"RAD21 | HGNC:9811 | Cornelia de Lange syndrome | MONDO:0016033 | AD | Definitive"
ClinGen gene-disease validity for RAD21 and CdLS. The assertion is keyed to the umbrella MONDO term, so it validates the gene-disease relationship and nothing about the RAD21-specific phenotype.
PMID:32193685 SUPPORT DIRECT Human Clinical
"In the 20 individuals with limited clinical information, additional phenotypes include Mungan syndrome (in patients with biallelic variants) and holoprosencephaly, with or without CdLS characteristics."
Establishes the zygosity split within one gene: biallelic variants give Mungan syndrome, not a more severe CdLS.
PMID:39286962 SUPPORT DIRECT Human Clinical
"Among the 28 patients with known genetic patterns, it was observed that 12 patients exhibited germline variants, while the remaining 16 patients displayed de novo variants."
Quantifies the inherited versus de novo split.
+ 2 more references
💊

Medical Actions

2
Multidisciplinary Supportive Care
Action: supportive careNCI Thesaurus (NCIT) Relation: this treatment is this clinical intervention This treatment is supportive care (NCIT:C15747). NCIT:C15747 is a clinical intervention from the NCI Thesaurus. Ontology label: Supportive Care NCIT:C15747
Platform: Behavioral / lifestyle
Management follows the CdLS management standard and is not RAD21-specific: therapy services to support development, aggressive treatment of gastro-oesophageal reflux, and surveillance covering growth and nutrition, development and education, behaviour, vision, hearing and dentition. No treatment study specific to CdLS4 exists.
Show evidence (2 references)
PMID:20301283 SUPPORT DIRECT Other
"Physical, occupational, and speech therapy to optimize psychomotor development and communication skills."
The therapy component of CdLS management, which is what most CdLS4 patients actually receive given the developmental rather than structural emphasis of their phenotype.
PMID:20301283 SUPPORT DIRECT Other
"behavioral assessment for anxiety, attention deficits, and aggressive or self-injurious behaviors"
The surveillance recommendation that maps onto the behavioural phenotype curated in this entry; anxiety and attention deficit are the two most frequent behavioural findings in the RAD21 series.
Genetic Counseling
Action: genetic counselingNCI Thesaurus (NCIT) Relation: this treatment is this clinical intervention This treatment is genetic counseling (NCIT:C15240). NCIT:C15240 is a clinical intervention from the NCI Thesaurus. Ontology label: Genetic Counseling NCIT:C15240
Autosomal dominant with a 50% recurrence risk to offspring. Counselling in CdLS4 differs from classic CdLS in three ways that follow from the data: variants are frequently inherited rather than de novo, expression varies strikingly within as well as between families, and a mildly affected parent may be identified only through the child.
Show evidence (2 references)
PMID:32193685 SUPPORT DIRECT Human Clinical
"The current study should be helpful when counseling families with a RAD21 variation."
The authors frame their genotype-phenotype findings as counselling material.
PMID:20301283 SUPPORT DIRECT Other
"Each child of an affected individual has a 50% chance of inheriting the CdLS-related pathogenic variant."
The source of the 50% recurrence figure this treatment asserts, which no other cited item states.
🔬

Diagnosis

1
Molecular diagnosis after CdLS clinical scoring
Diagnosis rests on identifying a heterozygous RAD21 variant. Whether the international CdLS consensus score reliably selects these patients for testing is the one place the two main sources disagree, and the disagreement is curated rather than resolved. The dedicated 49-individual RAD21 series found that every patient with full clinical data scored at least five, which is above the threshold of four that indicates molecular testing, and states outright that no RAD21 variant would have been missed by the consensus criteria. The later systematic review nonetheless had to exclude patients from its own statistics for scoring under four. Both authors also note that the clinical diagnosis is harder here than in classic CdLS. The practical reading is that the score is adequate as a gate to testing on present evidence, but that recognising CdLS clinically is not, and the attenuated presentation means it will more often be reached through broad sequencing than through a CdLS-shaped clinical suspicion.
Show evidence (4 references)
PMID:32193685 SUPPORT DIRECT Human Clinical
"No RAD21 variant would have been missed using the CdLS consensus criteria for molecular studies"
The authors' explicit finding that the consensus criteria do not under-select RAD21 patients for testing. This is the statement that prevents this entry from asserting the opposite, which the systematic review's exclusion rule would otherwise have suggested.
PMID:32193685 SUPPORT DIRECT Human Clinical
"The clinical diagnosis of CdLS may, therefore, be difficult."
The same authors on clinical recognition, which they distinguish from the performance of the scoring criteria once CdLS is being considered.
PMID:39286962 SUPPORT DIRECT Human Clinical
"Patients with a clinical feature score less than 4 are not diagnosed with CdLS and do not meet the molecular detection criteria (Kline et al. 2018)."
The scoring gate, and the rule under which this review excluded some RAD21 patients from its statistics.
+ 1 more reference
📊

Prevalence

1
Worldwide
Cases In Literature Ultra Rare
No population prevalence exists for CdLS4 specifically. Two overlapping published case counts bound it: 36 patients with confirmed RAD21 variants reported worldwide between May 2012 and March 2024, and 49 individuals from 33 families in a dedicated RAD21 series that included 24 previously unpublished cases. CdLS overall has a reported prevalence of 1:30,000 to 1:10,000 live births, and RAD21 accounts for a small minority of it; that umbrella figure is deliberately not converted into a rate for this entry, because the RAD21 share is a diagnostic-cohort proportion rather than a population one. Such a proportion does exist: a 716-proband cohort solved 422 molecularly, of which six (about 1%) were RAD21. That figure sits in the full text of PMID:37377026 and not in its cached abstract, so it is recorded here rather than quoted as evidence; it also measures ascertainment through a CdLS-shaped clinical suspicion, which an attenuated phenotype passes less often.
Show evidence (2 references)
PMID:39286962 SUPPORT DIRECT Human Clinical
"We reviewed 36 patients with CdLS related to RAD21 gene variants reported worldwide from May 2012 to March 2024."
The published case count on which the CASES_IN_LITERATURE measure rests.
PMID:32193685 SUPPORT DIRECT Human Clinical
"We gathered a series of 49 individuals from 33 families with RAD21 alterations"
The largest single RAD21 series, overlapping the 36-patient literature count rather than adding to it.
🔀

Differential Diagnoses

3

Conditions with similar clinical presentations that must be differentiated from Cornelia de Lange Syndrome 4:

Mungan Syndrome
Overlapping Features The other RAD21 disease, and the reason the gene cannot be reduced to a single CdLS subtype line. Biallelic RAD21 variants cause an enteric neuromyopathy with chronic intestinal pseudo-obstruction, megaduodenum, long-segment Barrett oesophagus and cardiac abnormalities. Its mechanism runs through RAD21's transcriptional targets rather than through a general cohesin dosage effect: the mutant protein fails to bind the APOB promoter, RUNX1 expression falls in patient cells, and rad21a-knockdown zebrafish show delayed intestinal transit with greatly reduced enteric neuron numbers. Mungan syndrome is not curated in dismech.
Show evidence (2 references)
PMID:25575569 SUPPORT DIRECT Human Clinical
"We identified a homozygous mutation (p.622, encodes Ala>Thr) in RAD21 in patients from a consanguineous family with CIPO."
Establishes the biallelic RAD21 disease as a distinct entity from CdLS4.
PMID:25575569 SUPPORT DIRECT Model Organism
"rad21a Morpholino zebrafish had delayed intestinal transit and greatly reduced numbers of enteric neurons, similar to patients with CIPO."
The enteric phenotype that separates the biallelic disease mechanistically from CdLS4.
Other Cohesinopathies
Overlapping Features dismech curates several gene-level cohesinopathies as standalone entries alongside the CdLS umbrella: Mullegama-Klein-Martinez syndrome (STAG2), WAPL-related developmental disorder, CHOPS syndrome (AFF4), CTCF-related neurodevelopmental disorder and Warsaw breakage syndrome. CdLS4 is placed the same way.
🐁

Animal Models

1
Zebrafish expressing CdLS4 RAD21 variants
The zebrafish system used in the entity-defining study to test whether patient RAD21 variants disrupt transcription. It is the only in vivo model reported for CdLS4 specifically.
Species
Zebrafish
Genotype
rad21 variant expression modelling patient RAD21 missense alleles
Publication
PMID:22633399
Pathograph links
Cohesin-Dependent Transcriptional Dysregulation recapitulates fidelity: moderate Patient RAD21 variants disrupt transcription in zebrafish, which is the arm of cohesin biology that the developmental phenotype is attributed to.
Limitations: The abstract reports the transcriptional readout only; no facial, limb, growth or cognitive correlate is claimed, and no zebrafish line carrying a knock-in patient allele at the endogenous locus is described. The model therefore supports the transcriptional step and says nothing about the attenuation that defines the human entity.
Readouts
Transcription in zebrafish expressing patient RAD21 variants
Direction: ALTERED
Interpretation: Disrupted transcription, the functional consequence this node asserts.
Show evidence (1 reference)
PMID:22633399 SUPPORT DIRECT Model Organism
"and disrupt transcription in a zebrafish model"
Reports the transcriptional measurement.
Show evidence (1 reference)
PMID:22633399 SUPPORT DIRECT Model Organism
"Here we show that human mutations in the integral cohesin structural protein RAD21 result in a congenital phenotype consistent with a "cohesinopathy.""
The study in which the zebrafish work was done to support the human cohesinopathy claim.
{ }

Source YAML

click to show 
name: Cornelia de Lange Syndrome 4
creation_date: "2026-09-01T00:00:00Z"
category: Mendelian
description: >-
  Cornelia de Lange syndrome 4 (CdLS4) is the RAD21-related form of the
  cohesinopathy spectrum, and it is defined as much by what it is milder than as
  by what it is. RAD21 encodes the kleisin subunit that closes the cohesin ring
  between SMC1A and SMC3 and recruits STAG1/STAG2, so a RAD21 lesion sits inside
  the ring itself rather than in the machinery that loads it — which is where
  NIPBL, the gene behind classic CdLS, acts. The disease-causing variants
  cluster at the RAD21 interfaces with its partner subunits, impair the cellular
  DNA damage response, and disrupt transcription; the resulting phenotype is an
  attenuated CdLS, with the facial and limb features present but muted and, most
  strikingly, cognition and behaviour far less affected than in NIPBL- or
  SMC1A-related disease. Two further things separate it from classic CdLS in
  practice. Variants are frequently familial rather than de novo, with striking
  variability between and within families, so an affected parent may never have
  been recognised. And RAD21 is dose- and zygosity-split: two RAD21 alleles do
  not give a more severe CdLS but a different disease entirely, Mungan syndrome,
  an enteric neuromyopathy with chronic intestinal pseudo-obstruction.
disease_term:
  preferred_term: Cornelia de Lange syndrome 4
  term:
    id: MONDO:0013864
    label: Cornelia de Lange syndrome 4
synonyms:
- CDLS4
- Cornelia de Lange syndrome type 4
- Cornelia de Lange syndrome caused by mutation in RAD21
- RAD21 Cornelia de Lange syndrome
parents:
- Cornelia de Lange syndrome
references:
- reference: PMID:20301283
  title: Cornelia de Lange Syndrome.
  tags:
  - GeneReviews
  findings: []
inheritance:
- name: Autosomal Dominant
  description: >-
    Heterozygous RAD21 variants, transmitted dominantly. Unlike classic
    NIPBL-related CdLS, which is overwhelmingly de novo, RAD21 variants are
    frequently inherited: 12 of 28 patients with a known transmission pattern in
    a systematic review carried a germline variant from a parent. That, together
    with the attenuated phenotype, means an affected parent may be recognised
    only after the child is diagnosed.
  inheritance_term:
    preferred_term: Autosomal dominant inheritance
    term:
      id: HP:0000006
      label: Autosomal dominant inheritance
  evidence:
  - reference: CGGV:assertion_30ad5023-6c68-42ab-b8ea-a601a134eaf6-2020-01-08T170000.000Z
    reference_title: "RAD21 / Cornelia de Lange syndrome (Definitive)"
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: "RAD21 | HGNC:9811 | Cornelia de Lange syndrome | MONDO:0016033 | AD | Definitive"
    explanation: >-
      ClinGen classifies the RAD21-CdLS gene-disease relationship as definitive
      with autosomal dominant inheritance. Note this assertion is keyed to the
      CdLS umbrella term MONDO:0016033 and makes no statement about the
      attenuated RAD21 phenotype; it is cited here only for inheritance and
      gene-disease validity.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Variants were frequently familial, and genotype-phenotype analyses
      demonstrated striking interfamilial and intrafamilial variability.
    explanation: >-
      Establishes that RAD21 variants are often inherited rather than de novo,
      which is the counselling difference from classic CdLS.
pathophysiology:
- name: Heterozygous RAD21 Variant
  biological_scale: MOLECULAR
  description: >-
    A single altered RAD21 allele. The variant spectrum is dominated by
    frameshift alleles (36% of 36 published patients), followed by missense
    (19%) and nonsense (17%), with whole-gene and intragenic microdeletions also
    reported. Where variants have been mapped to protein domains, most fall in
    the RAD21-SMC1A interaction domain. The allelic gradient runs the opposite
    way to intuition for a structural subunit: dominant missense alleles cause
    worse structural and cognitive findings than loss-of-function alleles do,
    which argues that the missense protein interferes with the ring rather than
    simply failing to join it.
  genetic_context:
    gene:
      preferred_term: RAD21
      term:
        id: hgnc:9811
        label: RAD21
    zygosity: HETEROZYGOUS
  evidence:
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Our data suggest that, compared to loss-of-function mutations, dominant
      missense mutations result in more severe functional defects and cause
      worse structural and cognitive clinical findings.
    explanation: >-
      The allelic gradient this node describes, in the paper that established
      RAD21 as a CdLS gene.
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Frameshift variants constituted the predominant variant type, representing
      36% (13/36) of cases.
    explanation: >-
      Quantifies the variant spectrum across all published CdLS4 patients.
  downstream:
  - target: Disruption of the Cohesin Ring at the RAD21 Subunit Interfaces
    causal_link_type: DIRECT
    description: >-
      RAD21 is the subunit that bridges SMC1A, SMC3 and STAG, so a lesion in it
      acts on those interfaces.
- name: Disruption of the Cohesin Ring at the RAD21 Subunit Interfaces
  biological_scale: MOLECULAR
  description: >-
    RAD21 is the kleisin that closes the tripartite cohesin ring: its N-terminus
    engages SMC3, its C-terminus engages SMC1, and its central STAG domain
    recruits STAG1 or STAG2. CdLS4 variants act at those interfaces. This is the
    step that distinguishes CdLS4 from classic CdLS at the level of mechanism —
    NIPBL is the cohesin loading factor and NIPBL-related disease is a dosage
    problem in getting cohesin onto chromatin, whereas CdLS4 is a lesion in the
    ring itself.
  cellular_components:
  - preferred_term: cohesin complex
    term:
      id: GO:0008278
      label: cohesin complex
  evidence:
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: IN_VITRO
    directness: DIRECT
    snippet: >-
      Mechanistically, these mutations act at the RAD21 interface with the other
      cohesin proteins STAG2 and SMC1A
    explanation: >-
      Localizes the lesion to the RAD21 subunit interfaces. Only the interface
      clause of the source sentence is quoted here; the sentence's other two
      claims are quoted separately on the nodes they belong to, because the
      transcription result is a zebrafish experiment and the DNA damage result
      is a cell-based one.
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Among the patients, 41.9% (13 out of 31) had variants situated in the
      RAD21-SMC1A domain
    explanation: >-
      Shows that the variants patients actually carry concentrate in one of the
      subunit interaction domains.
  downstream:
  - target: Cohesin-Dependent Transcriptional Dysregulation
    causal_link_type: DIRECT
    description: >-
      The developmental phenotype of the cohesinopathies is attributed primarily
      to cohesin's interphase role in transcription and genome organization.
  - target: Impaired DNA Damage Response
    causal_link_type: DIRECT
    description: >-
      Patient RAD21 variants impair the cellular response to DNA damage.
- name: Cohesin-Dependent Transcriptional Dysregulation
  biological_scale: CELLULAR
  description: >-
    Beyond holding sister chromatids together, cohesin organizes the genome in
    interphase and regulates transcription, and it is this arm rather than
    mitotic cohesion failure that is held responsible for the developmental
    phenotype. Disrupted transcription was demonstrated directly for CdLS4
    variants in a zebrafish model.
  cell_types:
  - preferred_term: neuron
    term:
      id: CL:0000540
      label: neuron
  biological_processes:
  - preferred_term: regulation of DNA-templated transcription
    term:
      id: GO:0006355
      label: regulation of DNA-templated transcription
    modifier: DYSREGULATED
  - preferred_term: chromatin organization
    term:
      id: GO:0006325
      label: chromatin organization
    modifier: DYSREGULATED
  evidence:
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: MODEL_ORGANISM
    directness: DIRECT
    snippet: >-
      and disrupt transcription in a zebrafish model
    explanation: >-
      The transcriptional consequence, demonstrated in zebrafish expressing the
      patient variants. Quoted as the zebrafish clause alone so this item and the
      cell-based items from the same sentence each carry one evidence_source.
  - reference: PMID:32687945
    reference_title: "Cohesin subunit RAD21: From biology to disease."
    supports: SUPPORT
    evidence_source: OTHER
    directness: INDIRECT
    snippet: >-
      In interphase, cohesin also functions in the control of gene expression by
      binding to numerous sites within the genome.
    explanation: >-
      Establishes the interphase transcriptional role this node depends on.
      Graded INDIRECT because it states cohesin biology generally rather than a
      CdLS4 result.
  downstream:
  - target: Attenuated Cornelia de Lange Phenotype
    causal_link_type: INDIRECT_UNKNOWN_INTERMEDIATES
    description: >-
      How transcriptional dysregulation produces the specific pattern of facial,
      limb, growth and cognitive features is not established for any
      cohesinopathy, let alone for RAD21 specifically.
- name: Impaired DNA Damage Response
  biological_scale: CELLULAR
  description: >-
    RAD21 is also a double-strand-break repair protein, and CdLS4 variants
    impair the cellular DNA damage response. This arm is well demonstrated
    biochemically but is not connected to any specific clinical feature of
    CdLS4, so it is curated as a consequence of the ring lesion without a
    downstream edge into the phenotype.
  biological_processes:
  - preferred_term: DNA damage response
    term:
      id: GO:0006974
      label: DNA damage response
    modifier: DECREASED
  evidence:
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: IN_VITRO
    directness: DIRECT
    snippet: >-
      impair cellular DNA damage response
    explanation: >-
      The DNA damage response defect measured in cells carrying patient
      variants.
  - reference: PMID:32687945
    reference_title: "Cohesin subunit RAD21: From biology to disease."
    supports: SUPPORT
    evidence_source: OTHER
    directness: INDIRECT
    snippet: >-
      RAD21 (also known as KIAA0078, NXP1, HR21, Mcd1, Scc1, and hereafter
      called RAD21), an essential gene, encodes a DNA double-strand break (DSB)
      repair protein that is evolutionarily conserved in all eukaryotes from
      budding yeast to humans.
    explanation: >-
      Establishes the repair function whose impairment this node describes.
      Graded INDIRECT because it states the protein's role rather than a CdLS4
      measurement.
- name: Attenuated Cornelia de Lange Phenotype
  biological_scale: ORGANISM
  description: >-
    The clinical result is recognisably CdLS but muted. Compared with NIPBL- and
    SMC1A-related disease the facial morphology and limb anomalies are milder,
    and cognition and behaviour are considerably less affected — 14 of 26 scored
    patients were mild and only one severe on the CdLS severity score. Because
    the consensus clinical score is built around the classic phenotype, CdLS4
    patients score low on it, which is a diagnostic problem rather than a
    cosmetic difference.
  evidence:
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Notably, unlike children with mutations in NIPBL, SMC1A, or SMC3, these
      individuals have much milder cognitive impairment than those with
      classical CdLS.
    explanation: >-
      The attenuation claim, in the paper that established the entity.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Their phenotype is an attenuated CdLS phenotype compared to that caused by
      variants in NIPBL or SMC1A for facial morphology, limb anomalies, and
      especially for cognition and behavior.
    explanation: >-
      Independent confirmation of attenuation in the largest RAD21 series, and
      the source that specifies which domains are attenuated.
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      it was found that among the 26 patients meeting the criteria, only 1 was
      classified as severe, 11 reached moderate levels, and 14 were considered
      mild
    explanation: >-
      Quantifies the severity distribution using the CdLS scoring system.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Major limb malformation HP:0001180/HP:0009776 0/29 0 0/49 0 17/67 25
    explanation: >-
      The starkest single contrast in the comparison table: no major limb
      malformation in any of 29 RAD21 patients, against 25% of NIPBL patients.
      Limb reduction is one of the cardinal features of classic CdLS and it is
      absent here.
  - reference: PMID:20301283
    reference_title: Cornelia de Lange Syndrome.
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: >-
      Individuals with a milder phenotype have less severe growth, cognitive,
      and limb involvement, but often have facial features consistent with CdLS.
    explanation: >-
      GeneReviews' description of the mild end of the CdLS spectrum, which is
      the position CdLS4 occupies. It matches the RAD21 series independently:
      growth, cognition and limbs attenuated, facial gestalt retained.
  - reference: PMID:37377026
    reference_title: "Genomic analyses in Cornelia de Lange Syndrome and related diagnoses: Novel candidate genes, genotype-phenotype correlations and common mechanisms."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: INDIRECT
    snippet: >-
      Pathogenic variants in cohesin genes other than NIPBL tend to result in a
      less severe phenotype.
    explanation: >-
      A 716-proband cohort makes the same attenuation point at the level of the
      whole non-NIPBL group. Graded INDIRECT because it is a statement about
      cohesin genes other than NIPBL collectively rather than about RAD21
      specifically.
  downstream:
  - target: Intellectual Disability
    causal_link_type: DIRECT
    description: The cognitive component, attenuated relative to classic CdLS.
  - target: Behavioural Phenotype
    causal_link_type: DIRECT
    description: >-
      The behavioural component, and the sharpest attenuation in the entry.
  - target: Microcephaly
    causal_link_type: DIRECT
    description: Postnatal head growth restriction.
  - target: Ptosis
    causal_link_type: DIRECT
    description: >-
      One of the few features more frequent in RAD21 than in NIPBL disease.
  - target: Congenital Heart Disease
    causal_link_type: DIRECT
    description: Cardiac malformation, major or minor.
  - target: Gastroesophageal Reflux Disease
    causal_link_type: DIRECT
    description: The commonest gastrointestinal manifestation.
  - target: Hirsutism
    causal_link_type: DIRECT
    description: >-
      Excess body hair, markedly less frequent than in the comparison genes.
  - target: Clinodactyly of the 5th Finger
    causal_link_type: DIRECT
    description: The other digital feature alongside the short fifth finger.
  - target: Delayed Speech and Language Development
    causal_link_type: DIRECT
    description: Verbal developmental delay, near-universal in reported patients.
  - target: Characteristic Facial Morphology
    causal_link_type: DIRECT
    description: >-
      The CdLS facial gestalt, present but muted.
  - target: Short Stature
    causal_link_type: DIRECT
    description: Growth restriction.
  - target: Short 5th Finger
    causal_link_type: DIRECT
    description: >-
      The limb component; reduction defects are milder than in classic CdLS.
  - target: Hearing Impairment
    causal_link_type: DIRECT
    description: Hearing loss in a third of reported patients.
phenotypes:
- category: Neurological
  name: Intellectual Disability
  description: >-
    Intellectual disability of varying degree in 94% (32/34) of published CdLS4
    patients. The frequency is near-universal but the degree is the point: it is
    much milder than in NIPBL-, SMC1A- or SMC3-related CdLS, and one reported
    patient had uncompromised intelligence with above-average school
    performance.
  frequency: VERY_FREQUENT
  phenotype_term:
    preferred_term: Intellectual disability
    term:
      id: HP:0001249
      label: Intellectual disability
  evidence:
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Clinical features included verbal developmental delay and intellectual
      disorder observed in 94% of patients.
    explanation: >-
      94% falls in the VERY_FREQUENT band (80-99%).
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      The majority of RAD21 patients (16/29, 55%) have normal or mildly impaired
      cognitive functioning (SMC1A group 32%; NIPBL group 7%)
    explanation: >-
      Quantifies the attenuation directly against the two comparison genes: a
      majority of RAD21 patients are cognitively normal or only mildly impaired,
      against 7% of NIPBL patients. This is the sharpest single number in the
      entry and is why the phenotype description says the degree, not the
      frequency, is what is attenuated.
  - reference: PMID:22633399
    reference_title: RAD21 mutations cause a human cohesinopathy.
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Notably, unlike children with mutations in NIPBL, SMC1A, or SMC3, these
      individuals have much milder cognitive impairment than those with
      classical CdLS.
    explanation: >-
      Records that the severity, not the frequency, is what is attenuated.
- category: Neurological
  name: Delayed Speech and Language Development
  description: >-
    Verbal developmental delay in 94% (30/32) of published patients, generally
    requiring language proficiency training.
  frequency: VERY_FREQUENT
  phenotype_term:
    preferred_term: Verbal developmental delay
    term:
      id: HP:0000750
      label: Delayed speech and language development
  evidence:
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Clinical features included verbal developmental delay and intellectual
      disorder observed in 94% of patients.
    explanation: >-
      94% falls in the VERY_FREQUENT band (80-99%). The same reported figure
      covers verbal delay and intellectual disability in this cohort.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Typical development is somewhat slow, mainly in speech development, and
      physical therapy or speech therapy may be indicated.
    explanation: >-
      Characterizes the delay as speech-predominant and mild, which is the
      qualitative counterpart of the 94% frequency.
- category: Craniofacial
  name: Characteristic Facial Morphology
  description: >-
    The CdLS facial gestalt, curated as one record rather than as several
    near-identical ones. In the dedicated RAD21 series the components are a long
    or smooth philtrum in 90%, thick eyebrows in 83% and synophrys in 68%; the
    same three features are the ones the evidence below quotes, and the band is
    set from the philtrum figure the bound term names. Short nose, at 88% the
    highest-frequency facial feature and the one that is not attenuated at all,
    is curated separately so that observation is not buried. Synophrys is the
    component that shows the attenuation most clearly, at 68% against 91% in
    NIPBL disease.
  frequency: VERY_FREQUENT
  phenotype_term:
    preferred_term: Cornelia de Lange facial gestalt (long or smooth philtrum, thick eyebrows, synophrys)
    term:
      id: HP:0000343
      label: Long philtrum
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Long and/or smooth philtrum HP:0000343/HP:0000319 26/29 90 27/43 63 54/67 81
    explanation: >-
      A row of the RAD21 versus SMC1A versus NIPBL comparison table, whose
      columns are RAD21 (n = 29), SMC1A (n = 51) and NIPBL (n = 67). The
      RAD21 figure, 26/29 or 90%, sets the VERY_FREQUENT band.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Thick eyebrows HP:0000574 20/24 83 37/46 80 61/67 91
    explanation: >-
      Thick eyebrows in 83% of RAD21 patients, from the same comparison table.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Synophrys HP:0000664 19/28 68 37/46 80 61/67 91
    explanation: >-
      Synophrys, the single most recognisable CdLS facial sign, is present in
      68% of RAD21 patients against 91% of NIPBL patients. This is the
      attenuation visible in one feature, and it is why the band for this record
      is set from the philtrum figure rather than from synophrys.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Their phenotype is an attenuated CdLS phenotype compared to that caused by
      variants in NIPBL or SMC1A for facial morphology, limb anomalies, and
      especially for cognition and behavior.
    explanation: >-
      Records that the facial morphology, though frequent, is muted relative to
      classic CdLS.
- category: Musculoskeletal
  name: Short 5th Finger
  description: >-
    Brachydactyly of the fifth finger in 83% (25/30) of published patients. The
    limb involvement in CdLS4 is at this mild end of the CdLS limb spectrum
    rather than the severe reduction defects of classic disease.
  frequency: VERY_FREQUENT
  phenotype_term:
    preferred_term: Short fifth finger
    term:
      id: HP:0009237
      label: Short 5th finger
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Short 5th finger HP:0009237 23/28 82 21/45 47 42/63 67
    explanation: >-
      82% in the RAD21 column falls in the VERY_FREQUENT band (80-99%). Note it
      is higher than the NIPBL figure of 67%, so the mild limb involvement in
      CdLS4 is not simply a diluted version of the classic limb phenotype.
- category: Growth
  name: Short Stature
  description: >-
    Height below -2 SDS in 50% (16/32) of published patients with recorded
    heights at diagnosis.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Short stature
    term:
      id: HP:0004322
      label: Short stature
  evidence:
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Among the 32 cases with height records at the time of diagnosis, 16 had a
      height of less than -2 SDS and could be diagnosed as "Short stature" (50%).
    explanation: >-
      50% falls in the FREQUENT band (30-79%).
- category: Neurological
  name: Hearing Impairment
  description: >-
    Hearing loss requiring hearing equipment in 33% (8/24) of published
    patients.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Hearing impairment
    term:
      id: HP:0000365
      label: Hearing impairment
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Vision is mostly normal; hearing loss is found in a third of individuals
      and may require hearing devices.
    explanation: >-
      A third falls in the FREQUENT band (30-79%). The systematic review reports
      the same figure, 8 of 24 patients.
- category: Ophthalmological
  name: Visual Impairment
  description: >-
    Visual impairment is essentially absent in CdLS4: none of 24 assessed
    patients in the dedicated RAD21 series, and 1 of 28 in the systematic
    review. The contrast is with SMC1A (53%) and NIPBL (44%), where
    ophthalmological involvement is common.
  frequency: VERY_RARE
  phenotype_term:
    preferred_term: Visual impairment
    term:
      id: HP:0000505
      label: Visual impairment
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Visual impairment HP:0000505 0/24 0 20/38 53 29/66 44
    explanation: >-
      No visual impairment in 24 assessed RAD21 patients, against 53% for SMC1A
      and 44% for NIPBL. The systematic review reports 1 of 28, or 4%; both fall
      in the VERY_RARE band, and the band is set rather than the record dropped
      because the contrast with the comparison genes is the informative part.
- category: Behavioral
  name: Behavioural Phenotype
  description: >-
    Behaviour is where CdLS4 is most clearly attenuated, and the contrast is
    sharper than for any physical feature. Anxiety is the commonest problem, in
    53% of assessed RAD21 patients, with attention deficit disorder in 35%,
    obsessive-compulsive behaviour in 32% and autistic-like features in 35%. The
    severe behaviours that characterise classic CdLS are largely absent:
    self-injurious behaviour in 6% of RAD21 patients against 77% of NIPBL
    patients, and aggression in 6% against 80%. The band is set from anxiety, the
    most frequent component.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Anxiety, attention deficit and obsessive-compulsive behaviour with little aggression or self-injury
    term:
      id: HP:0000739
      label: Anxiety
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Obsessive–compulsive behavior 6/19 32 10/26h 38 Anxiety 10/19 53
    explanation: >-
      Two adjacent rows of the behavioural table, quoted together because the
      anxiety row on its own is too short to stand as a snippet.
      Obsessive-compulsive behaviour in 32% and anxiety in 53% of assessed RAD21
      patients. The 53% falls in the FREQUENT band (30-79%) and is the figure
      this record's band is set from.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Self-injurious behavior 1/18 6 11/31 35 47/61 77
    explanation: >-
      Self-injurious behaviour in 6% of RAD21 patients against 77% of NIPBL
      patients. This is the sharpest single attenuation contrast in the entry,
      sharper than the limb-malformation row, and it is the concrete content of
      the repeatedly quoted claim that the attenuation is "especially for
      cognition and behavior".
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Aggression 1/16 6 12/15h 80
    explanation: >-
      Aggression in 6% of RAD21 patients against 80% of SMC1A patients, the same
      contrast in a second behaviour.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Attention deficit disorder ± hyperactivity 8/23 35
    explanation: >-
      Attention deficit disorder with or without hyperactivity in 35% of
      assessed RAD21 patients. No comparison figure is given for the other genes
      in this row.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Autistic-like features 7/20 35 18/31h 56 9/13h 69
    explanation: >-
      Autistic-like features in 35% of RAD21 patients against 69% of NIPBL
      patients. Curated inside this record rather than as its own phenotype
      because the source reports it as one row of a behavioural assessment
      rather than as a diagnosis.
- category: Neurological
  name: Microcephaly
  description: >-
    Postnatal head circumference below minus two standard deviations in 57% of
    assessed RAD21 patients, against 87% of NIPBL patients. The RAD21 series
    separately reports no correlation between microcephaly and the severity of
    cognitive impairment in these patients.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Postnatal microcephaly
    term:
      id: HP:0000252
      label: Microcephaly
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Postnatal head circumference < − 2SD HP:0000252 16/28 57 23/36 64 54/62 87
    explanation: >-
      57% in the RAD21 column falls in the FREQUENT band (30-79%). The row is
      the postnatal one; the same HPO term appears on a separate prenatal row
      with different figures.
- category: Craniofacial
  name: Ptosis
  description: >-
    Ptosis in 42% of assessed RAD21 patients. Along with the short fifth finger
    it is one of the few features that is more frequent here than in classic
    NIPBL-related disease, where it is 19%.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Ptosis
    term:
      id: HP:0000508
      label: Ptosis
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Ptosis HP:0000508 11/26 42 4/40 10 8/42 19
    explanation: >-
      42% in the RAD21 column falls in the FREQUENT band (30-79%), against 19%
      for NIPBL.
- category: Craniofacial
  name: Short Nose
  description: >-
    Short nose in 88% of assessed RAD21 patients, the highest-frequency single
    facial feature in the series and slightly above the NIPBL figure of 87%.
    Curated separately from the facial-gestalt record because it is the one
    facial feature that is not attenuated at all.
  frequency: VERY_FREQUENT
  phenotype_term:
    preferred_term: Short nose
    term:
      id: HP:0003196
      label: Short nose
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Short nose HP:0003196 23/26 88 26/46 57 58/67 87
    explanation: >-
      88% in the RAD21 column falls in the VERY_FREQUENT band (80-99%).
- category: Cardiovascular
  name: Congenital Heart Disease
  description: >-
    Cardiac malformation, major or minor, in 39% of assessed RAD21 patients —
    slightly higher than in the SMC1A and NIPBL comparison groups.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Congenital heart malformation
    term:
      id: HP:0001627
      label: Abnormal heart morphology
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Heart (major and minor) HP:0001627 9/23 39 13/44 30 18/66 27
    explanation: >-
      39% in the RAD21 column falls in the FREQUENT band (30-79%).
- category: Gastrointestinal
  name: Gastroesophageal Reflux Disease
  description: >-
    Gastro-oesophageal reflux in 52% of assessed RAD21 patients. Less frequent
    than in NIPBL disease (71%) but common enough that the CdLS management
    standard's emphasis on aggressive reflux treatment applies here too.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Gastroesophageal reflux disease
    term:
      id: HP:0002020
      label: Gastroesophageal reflux
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Gastroesophageal reflux disease HP:0002020 13/25 52 25/42 60 47/66 71
    explanation: >-
      52% in the RAD21 column falls in the FREQUENT band (30-79%).
- category: Dermatological
  name: Hirsutism
  description: >-
    Excess body hair in 38% of assessed RAD21 patients, against 86% in NIPBL
    disease — one of the larger attenuations among the physical features.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Hirsutism
    term:
      id: HP:0001007
      label: Hirsutism
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Hirsutism HP:0001007 10/26 38 37/47 79 37/43 86
    explanation: >-
      38% in the RAD21 column falls in the FREQUENT band (30-79%), against 86%
      for NIPBL.
- category: Musculoskeletal
  name: Clinodactyly of the 5th Finger
  description: >-
    Clinodactyly of the fifth finger in 54% of assessed RAD21 patients, the
    second digital feature alongside the short fifth finger.
  frequency: FREQUENT
  phenotype_term:
    preferred_term: Clinodactyly of the 5th finger
    term:
      id: HP:0004209
      label: Clinodactyly of the 5th finger
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Clinodactyly 5th finger HP:0004209 13/24 54 21/45 47 42/63 67
    explanation: >-
      54% in the RAD21 column falls in the FREQUENT band (30-79%).
prevalence:
- population: Worldwide
  measure_type: CASES_IN_LITERATURE
  prevalence_class: ULTRA_RARE
  notes: >-
    No population prevalence exists for CdLS4 specifically. Two overlapping
    published case counts bound it: 36 patients with confirmed RAD21 variants
    reported worldwide between May 2012 and March 2024, and 49 individuals from
    33 families in a dedicated RAD21 series that included 24 previously
    unpublished cases. CdLS overall has a reported prevalence of 1:30,000 to
    1:10,000 live births, and RAD21 accounts for a small minority of it; that
    umbrella figure is deliberately not converted into a rate for this entry,
    because the RAD21 share is a diagnostic-cohort proportion rather than a
    population one. Such a proportion does exist: a 716-proband cohort solved
    422 molecularly, of which six (about 1%) were RAD21. That figure sits in the
    full text of PMID:37377026 and not in its cached abstract, so it is recorded
    here rather than quoted as evidence; it also measures ascertainment through
    a CdLS-shaped clinical suspicion, which an attenuated phenotype passes less
    often.
  evidence:
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      We reviewed 36 patients with CdLS related to RAD21 gene variants reported
      worldwide from May 2012 to March 2024.
    explanation: >-
      The published case count on which the CASES_IN_LITERATURE measure rests.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      We gathered a series of 49 individuals from 33 families with RAD21
      alterations
    explanation: >-
      The largest single RAD21 series, overlapping the 36-patient literature
      count rather than adding to it.
genetic:
- name: RAD21
  gene_term:
    preferred_term: RAD21
    term:
      id: hgnc:9811
      label: RAD21
  relationship_type: CAUSATIVE
  association: Heterozygous Pathogenic Variants
  notes: >-
    RAD21 is the only gene that causes CdLS4. The dosage axis matters clinically
    and is the reason this gene needs an entry rather than a subtype line:
    heterozygous variants give attenuated CdLS, whereas biallelic variants give
    Mungan syndrome, an enteric neuromyopathy with chronic intestinal
    pseudo-obstruction, which is not a severe CdLS and is not curated in dismech.
    Variants concentrate in exon 14 (30.6% of published cases) and in the
    RAD21-SMC1A interaction domain. Twelve of 28 patients with a known
    transmission pattern inherited the variant; sixteen were de novo.
  evidence:
  - reference: CGGV:assertion_30ad5023-6c68-42ab-b8ea-a601a134eaf6-2020-01-08T170000.000Z
    reference_title: "RAD21 / Cornelia de Lange syndrome (Definitive)"
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: "RAD21 | HGNC:9811 | Cornelia de Lange syndrome | MONDO:0016033 | AD | Definitive"
    explanation: >-
      ClinGen gene-disease validity for RAD21 and CdLS. The assertion is keyed to
      the umbrella MONDO term, so it validates the gene-disease relationship and
      nothing about the RAD21-specific phenotype.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      In the 20 individuals with limited clinical information, additional
      phenotypes include Mungan syndrome (in patients with biallelic variants)
      and holoprosencephaly, with or without CdLS characteristics.
    explanation: >-
      Establishes the zygosity split within one gene: biallelic variants give
      Mungan syndrome, not a more severe CdLS.
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Among the 28 patients with known genetic patterns, it was observed that 12
      patients exhibited germline variants, while the remaining 16 patients
      displayed de novo variants.
    explanation: >-
      Quantifies the inherited versus de novo split.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Familial mutation 5/12 42 4/47 9 n/a n/a
    explanation: >-
      From the RAD21 versus SMC1A versus NIPBL comparison table: 42% of RAD21
      variants were familial against 9% of SMC1A variants. This is the
      counselling difference from classic CdLS stated as a number.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: INDIRECT
    snippet: >-
      Deardorff et al. found haploinsufficiency for RAD21 led to approximately
      halved RAD21 RNA in a cell line from a patient with classical CdLS
    explanation: >-
      Establishes that RAD21 haploinsufficiency is not buffered by compensatory
      upregulation, which the authors use to argue RAD21 is simply less
      dosage-sensitive than the other CdLS genes. Graded INDIRECT because it is
      one cell line from a patient with classical rather than attenuated
      disease, cited for the dosage argument rather than as a CdLS4 result.
diagnosis:
- name: Molecular diagnosis after CdLS clinical scoring
  description: >-
    Diagnosis rests on identifying a heterozygous RAD21 variant. Whether the
    international CdLS consensus score reliably selects these patients for
    testing is the one place the two main sources disagree, and the disagreement
    is curated rather than resolved. The dedicated 49-individual RAD21 series
    found that every patient with full clinical data scored at least five, which
    is above the threshold of four that indicates molecular testing, and states
    outright that no RAD21 variant would have been missed by the consensus
    criteria. The later systematic review nonetheless had to exclude patients
    from its own statistics for scoring under four. Both authors also note that
    the clinical diagnosis is harder here than in classic CdLS. The practical
    reading is that the score is adequate as a gate to testing on present
    evidence, but that recognising CdLS clinically is not, and the attenuated
    presentation means it will more often be reached through broad sequencing
    than through a CdLS-shaped clinical suspicion.
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      No RAD21 variant would have been missed using the CdLS consensus criteria
      for molecular studies
    explanation: >-
      The authors' explicit finding that the consensus criteria do not
      under-select RAD21 patients for testing. This is the statement that
      prevents this entry from asserting the opposite, which the systematic
      review's exclusion rule would otherwise have suggested.
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      The clinical diagnosis of CdLS may, therefore, be difficult.
    explanation: >-
      The same authors on clinical recognition, which they distinguish from the
      performance of the scoring criteria once CdLS is being considered.
  - reference: PMID:39286962
    reference_title: "Clinical Characteristics, Genetic Analysis, and Literature Review of Cornelia de Lange Syndrome Type 4 Associated With a RAD21 Variant."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Patients with a clinical feature score less than 4 are not diagnosed with
      CdLS and do not meet the molecular detection criteria (Kline et al. 2018).
    explanation: >-
      The scoring gate, and the rule under which this review excluded some
      RAD21 patients from its statistics.
  - reference: PMID:20301283
    reference_title: Cornelia de Lange Syndrome.
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: >-
      The diagnosis of CdLS is established in a proband with suggestive clinical
      features and/or a heterozygous pathogenic variant in BRD4, MAU2, NIPBL,
      RAD21, or SMC3
    explanation: >-
      GeneReviews' diagnostic statement, which names RAD21 among the autosomal
      dominant CdLS genes and allows the molecular finding alone to establish
      the diagnosis -- the route by which an attenuated patient is most likely
      to reach it.
animal_models:
- name: Zebrafish expressing CdLS4 RAD21 variants
  species: Zebrafish
  genotype: rad21 variant expression modelling patient RAD21 missense alleles
  description: >-
    The zebrafish system used in the entity-defining study to test whether
    patient RAD21 variants disrupt transcription. It is the only in vivo model
    reported for CdLS4 specifically.
  publication: PMID:22633399
  modeled_mechanisms:
  - target: Cohesin-Dependent Transcriptional Dysregulation
    relationship: RECAPITULATES
    fidelity: MODERATE
    description: >-
      Patient RAD21 variants disrupt transcription in zebrafish, which is the
      arm of cohesin biology that the developmental phenotype is attributed to.
    limitations: >-
      The abstract reports the transcriptional readout only; no facial, limb,
      growth or cognitive correlate is claimed, and no zebrafish line carrying a
      knock-in patient allele at the endogenous locus is described. The model
      therefore supports the transcriptional step and says nothing about the
      attenuation that defines the human entity.
    readouts:
    - name: Transcription in zebrafish expressing patient RAD21 variants
      target: Cohesin-Dependent Transcriptional Dysregulation
      direction: ALTERED
      interpretation: >-
        Disrupted transcription, the functional consequence this node asserts.
      evidence:
      - reference: PMID:22633399
        reference_title: RAD21 mutations cause a human cohesinopathy.
        supports: SUPPORT
        evidence_source: MODEL_ORGANISM
        directness: DIRECT
        snippet: >-
          and disrupt transcription in a zebrafish model
        explanation: Reports the transcriptional measurement.
    evidence:
    - reference: PMID:22633399
      reference_title: RAD21 mutations cause a human cohesinopathy.
      supports: SUPPORT
      evidence_source: MODEL_ORGANISM
      directness: DIRECT
      snippet: >-
        Here we show that human mutations in the integral cohesin structural
        protein RAD21 result in a congenital phenotype consistent with a
        "cohesinopathy."
      explanation: >-
        The study in which the zebrafish work was done to support the human
        cohesinopathy claim.
treatments:
- name: Multidisciplinary Supportive Care
  description: >-
    Management follows the CdLS management standard and is not RAD21-specific:
    therapy services to support development, aggressive treatment of
    gastro-oesophageal reflux, and surveillance covering growth and nutrition,
    development and education, behaviour, vision, hearing and dentition. No
    treatment study specific to CdLS4 exists.
  therapeutic_modality: BEHAVIORAL
  treatment_term:
    preferred_term: supportive care
    term:
      id: NCIT:C15747
      label: Supportive Care
  evidence:
  - reference: PMID:20301283
    reference_title: Cornelia de Lange Syndrome.
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: >-
      Physical, occupational, and speech therapy to optimize psychomotor
      development and communication skills.
    explanation: >-
      The therapy component of CdLS management, which is what most CdLS4
      patients actually receive given the developmental rather than structural
      emphasis of their phenotype.
  - reference: PMID:20301283
    reference_title: Cornelia de Lange Syndrome.
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: >-
      behavioral assessment for anxiety, attention deficits, and aggressive or
      self-injurious behaviors
    explanation: >-
      The surveillance recommendation that maps onto the behavioural phenotype
      curated in this entry; anxiety and attention deficit are the two most
      frequent behavioural findings in the RAD21 series.
- name: Genetic Counseling
  description: >-
    Autosomal dominant with a 50% recurrence risk to offspring. Counselling in
    CdLS4 differs from classic CdLS in three ways that follow from the data:
    variants are frequently inherited rather than de novo, expression varies
    strikingly within as well as between families, and a mildly affected parent
    may be identified only through the child.
  treatment_term:
    preferred_term: genetic counseling
    term:
      id: NCIT:C15240
      label: Genetic Counseling
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      The current study should be helpful when counseling families with a RAD21
      variation.
    explanation: >-
      The authors frame their genotype-phenotype findings as counselling
      material.
  - reference: PMID:20301283
    reference_title: Cornelia de Lange Syndrome.
    supports: SUPPORT
    evidence_source: OTHER
    directness: DIRECT
    snippet: >-
      Each child of an affected individual has a 50% chance of inheriting the
      CdLS-related pathogenic variant.
    explanation: >-
      The source of the 50% recurrence figure this treatment asserts, which no
      other cited item states.
differential_diagnoses:
- name: Classic NIPBL-Related Cornelia de Lange Syndrome
  description: >-
    The principal differential and the comparison that defines CdLS4. NIPBL acts
    as the cohesin loading factor and its disease is a dosage problem; RAD21 sits
    inside the ring. Clinically, NIPBL-related disease has more severe facial
    morphology, limb anomalies and, especially, cognitive and behavioural
    involvement, and is overwhelmingly de novo where RAD21 variants are often
    familial. dismech curates the umbrella as Cornelia de Lange syndrome.
  evidence:
  - reference: PMID:32193685
    reference_title: "Delineation of phenotypes and genotypes related to cohesin structural protein RAD21."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      Their phenotype is an attenuated CdLS phenotype compared to that caused by
      variants in NIPBL or SMC1A for facial morphology, limb anomalies, and
      especially for cognition and behavior.
    explanation: >-
      The contrast that separates CdLS4 from the classic form.
- name: Mungan Syndrome
  description: >-
    The other RAD21 disease, and the reason the gene cannot be reduced to a
    single CdLS subtype line. Biallelic RAD21 variants cause an enteric
    neuromyopathy with chronic intestinal pseudo-obstruction, megaduodenum,
    long-segment Barrett oesophagus and cardiac abnormalities. Its mechanism runs
    through RAD21's transcriptional targets rather than through a general cohesin
    dosage effect: the mutant protein fails to bind the APOB promoter, RUNX1
    expression falls in patient cells, and rad21a-knockdown zebrafish show
    delayed intestinal transit with greatly reduced enteric neuron numbers.
    Mungan syndrome is not curated in dismech.
  evidence:
  - reference: PMID:25575569
    reference_title: "Mutations in RAD21 disrupt regulation of APOB in patients with chronic intestinal pseudo-obstruction."
    supports: SUPPORT
    evidence_source: HUMAN_CLINICAL
    directness: DIRECT
    snippet: >-
      We identified a homozygous mutation (p.622, encodes Ala>Thr) in RAD21 in
      patients from a consanguineous family with CIPO.
    explanation: >-
      Establishes the biallelic RAD21 disease as a distinct entity from CdLS4.
  - reference: PMID:25575569
    reference_title: "Mutations in RAD21 disrupt regulation of APOB in patients with chronic intestinal pseudo-obstruction."
    supports: SUPPORT
    evidence_source: MODEL_ORGANISM
    directness: DIRECT
    snippet: >-
      rad21a Morpholino zebrafish had delayed intestinal transit and greatly
      reduced numbers of enteric neurons, similar to patients with CIPO.
    explanation: >-
      The enteric phenotype that separates the biallelic disease mechanistically
      from CdLS4.
- name: Other Cohesinopathies
  description: >-
    dismech curates several gene-level cohesinopathies as standalone entries
    alongside the CdLS umbrella: Mullegama-Klein-Martinez syndrome (STAG2),
    WAPL-related developmental disorder, CHOPS syndrome (AFF4),
    CTCF-related neurodevelopmental disorder and Warsaw breakage syndrome. CdLS4
    is placed the same way.
discussions:
- discussion_id: cdls4_attenuation_mechanism
  kind: KNOWLEDGE_GAP
  attaches_to:
  - pathophysiology#Attenuated Cornelia de Lange Phenotype
  prompt: >-
    Why does a lesion in the cohesin ring itself produce a milder phenotype than
    reduced dosage of the factor that loads it?
  rationale: >-
    The attenuation is the best-replicated fact about CdLS4 — reported in the
    entity-defining paper and confirmed in the largest series — and it is
    counterintuitive. A structural subunit of the ring might be expected to
    matter at least as much as its loader. Two partial explanations are on the
    record and neither has been tested: that the missense-worse-than-LOF gradient
    means RAD21 haploinsufficiency is comparatively well tolerated while missense
    protein interferes actively, and that cohesin composition rather than dosage
    determines developmental outcome. No study has measured chromatin-bound
    cohesin or transcriptional output side by side in NIPBL and RAD21 patient
    cells, which is what would separate them.

    The RAD21 series offers one concrete lead, which this entry records but does
    not assert as mechanism: Deardorff found that RAD21 haploinsufficiency
    approximately halves RAD21 RNA in a patient cell line, whereas NIPBL
    haploinsufficiency is often accompanied by compensatory upregulation from
    the intact allele. RAD21 therefore appears simply more tolerant of reduced
    dosage than NIPBL is, despite having no compensatory mechanism -- which
    inverts the usual expectation and is the opposite of what a compensation
    argument would predict.
- discussion_id: cdls4_zygosity_split
  kind: KNOWLEDGE_GAP
  attaches_to:
  - genetic#RAD21
  prompt: >-
    Why do two RAD21 alleles give an enteric neuromyopathy rather than a more
    severe Cornelia de Lange syndrome?
  rationale: >-
    Dose-response is the usual expectation, and RAD21 does not follow it.
    Heterozygous variants give attenuated CdLS; biallelic variants give Mungan
    syndrome, whose reported mechanism runs through specific transcriptional
    targets — failure to bind the APOB promoter, reduced RUNX1 — and whose
    zebrafish phenotype is a deficit of enteric neurons. Whether the two
    presentations reflect different residual cohesin function, different
    tissue-specific target sets, or an allele-specific effect of the particular
    biallelic variant reported has not been addressed. No published work compares
    heterozygous and homozygous RAD21 patient material directly.
- discussion_id: cdls4_no_model_of_attenuation
  kind: HUMAN_MODEL_MISMATCH
  attaches_to:
  - animal_models#Zebrafish
  prompt: >-
    Does any model reproduce the attenuation that defines CdLS4?
  rationale: >-
    The one in vivo model reported for CdLS4 is the zebrafish transcription assay
    from the defining paper, which tests whether patient variants disrupt
    transcription and nothing else. There is no mouse, no knock-in at the
    endogenous locus, and no model in which the RAD21 phenotype has been compared
    with a Nipbl model under the same conditions. The defining clinical claim of
    this entity — milder than NIPBL disease, especially for cognition — therefore
    has no experimental counterpart at all, and the mechanism discussion above
    cannot be resolved without one.
notes: >-
  Entry type. Curated as a standalone Disease. The umbrella,
  kb/disorders/Cornelia_de_Lange_Syndrome.yaml, does bind RAD21 properly — the
  gene term hgnc:9811 appears both on its "Cohesin pathway gene disruption"
  pathophysiology node and on its own genetic record with the ClinGen Definitive
  assertion — so this is not the cblA situation of a gene named only in prose.
  Three things nonetheless argued for a separate entry rather than a subtype
  record.

  First, the umbrella has no has_subtypes block at all, so folding CdLS4 in would
  mean creating a subtype axis from scratch rather than filling an existing slot.
  Second, and more substantively, the umbrella's pathograph is a single generic
  node listing six cohesin genes together; it has no place to record that RAD21
  is inside the ring while NIPBL loads it, that missense alleles are worse than
  loss of function, or that the resulting phenotype is attenuated — and the
  attenuation is the whole content of this entity. Third, RAD21 carries a
  zygosity split: biallelic variants give Mungan syndrome, a different disease
  that is not a severe CdLS, and a one-paragraph subtype line has nowhere to put
  that. dismech already keeps gene-level cohesinopathies standalone —
  Mullegama-Klein-Martinez syndrome (STAG2), WAPL-related developmental disorder,
  CHOPS syndrome, CTCF-related neurodevelopmental disorder, Warsaw breakage
  syndrome — and this follows that pattern. The umbrella was not modified.

  The ClinGen assertion is cited for what it says and no more. CGGV
  assertion_30ad5023 is keyed to MONDO:0016033, the CdLS umbrella term, and
  contains no statement about the RAD21 phenotype. It is cited twice, for
  inheritance and for gene-disease validity, and both explanations say
  explicitly that it does not support the attenuation claim. The attenuation
  rests on PMID:22633399 and PMID:32193685.

  A claim this entry does not make. An earlier draft asserted that the CdLS
  consensus clinical score under-detects CdLS4, reasoning from the systematic
  review's exclusion of patients scoring under four. The dedicated RAD21 series
  says the opposite in as many words -- every cohort A patient scored at least
  five and no RAD21 variant would have been missed by the consensus criteria --
  so the claim was removed rather than kept as a plausible-sounding inference.
  What both sources do support is that clinical recognition of CdLS is harder in
  this group, and that is what the diagnosis node now says.

  Named entity confusion. preflight-dr returns PASS with RAD21 mentioned 73
  times and the next gene, NIPBL, at 8 — but that reflects the report, not the
  literature, and RAD21 has a large off-target body that was excluded by hand:

  1. Somatic RAD21 in cancer. RAD21 is a recurrent somatic mutation in AML and
  MDS and is amplified in solid tumours. None of that literature is cited. The
  one review used here, PMID:32687945, spans both the developmental and the
  cancer roles; only its statements about RAD21's normal function are quoted,
  and both are graded INDIRECT because they are protein biology rather than
  CdLS4 results.

  2. RAD21L, the meiotic paralog, is a different gene and nothing about it is
  used.

  3. The 8q23-q24 contiguous-gene deletions spanning RAD21 and EXT1 overlap
  Langer-Giedion syndrome. Those are not monogenic RAD21 disease and none of
  their phenotypes were imported.

  4. Mungan syndrome is RAD21 but is not this disease. It is curated only as a
  differential and in the genetic notes, with its own evidence, and none of its
  features appear in the phenotypes block.

  Facial features are one phenotype, not five. Long philtrum, thick eyebrows,
  long eyelashes, depressed nasal bridge and thin upper vermilion all fall in the
  same VERY_FREQUENT band from the same sentence of the same cohort. They are
  curated as a single "Characteristic Facial Morphology" record bound to the
  long philtrum term, with the full list and every individual frequency in the
  description and the quoted snippet, rather than as five records that would
  quote one sentence five times.

  MONDO placement worth reporting upstream. MONDO:0013864 has two parents:
  Cornelia de Lange syndrome, which is right, and non-syndromic limb reduction
  defect, which is not — CdLS4 is a syndromic multiple-anomaly condition and its
  limb involvement is mild brachydactyly rather than a reduction defect. Recorded
  here rather than propagated into the entry, in the same way #10260 recorded the
  cblA/cblB synonym bug. Not fixed here.

  Tooling. `just clingen-refresh` exits non-zero on a drifted sha256 against the
  pinned snapshot but downloads gene_validity.csv before failing, which is enough
  to confirm that RAD21 carries exactly one ClinGen assertion and that it is the
  umbrella-keyed one already committed to references_cache. The manifest was not
  repinned and no new ClinGen cache file was generated.

  Trials. No clinical_trials block. The deep-research report surfaces trials
  recruiting across the CdLS spectrum, none of them RAD21-specific or reporting
  a RAD21 subgroup. Curating a pan-CdLS trial on a subtype entry would assert a
  link its registration record does not make, which is the same test applied to
  the named-entity exclusions above. There is no interventional trial in CdLS4.

  Not curated. No progression block: no natural history study of CdLS4 exists,
  and the age-at-diagnosis distribution in the systematic review describes
  ascertainment rather than disease course. No biochemical or imaging block. No
  datasets block. Growth hormone response is tabulated in PMID:39286962 for the
  reviewed patients, but the cached text reports it as a summary column rather
  than as an analysed outcome, so no treatment claim is made from it. A
  Cohesinopathies grouping does not exist in kb/groupings/ and would be the
  natural home for this entry alongside the other five cohesinopathy entries;
  that is a separate piece of work.
📚

References & Deep Research

References

1
PMID:20301283
Cornelia de Lange Syndrome.
No top-level findings curated for this source.