| Domain | Best quantitative finding | Evidence type/source/date | Interpretation/limitation |
|---|---|---|---|
| Clinical and variant spectrum | **15 unrelated individuals; 14 different SMARCA4 variants; 12 missense variants, of which 11/12 mapped to residues 761–1,243 in or near the helicase/ATPase domains.** Ten of 12 missense variants were confirmed de novo; parental data were unavailable for two. | Human CSS4 registry cohort; Li et al., July 2020; DOI: [10.1002/ajmg.a.61732](https://doi.org/10.1002/ajmg.a.61732) (pqac-00000005, pqac-00000006) | Supports enrichment of pathogenic non-truncating variants in critical BRG1 domains and marked clinical variability. The small, clinically ascertained cohort cannot establish population penetrance or precise genotype–phenotype rules. |
| Development and classic morphology | In the **12-person SMARCA4 subset**, fifth-digit anomalies occurred in **7/12 (58%)**, sparse scalp hair in **4/12 (33%)**, hypertrichosis or hirsutism in **6/12 (50%)**, high or cleft palate in **6/12 (50%)**, and microcephaly in **5/12 (42%)**. | Parent-reported human CSS/BAF registry; Mannino et al., October 2018; DOI: [10.1002/ajmg.a.40471](https://doi.org/10.1002/ajmg.a.40471) (pqac-00000008) | Absence of fifth-digit hypoplasia or other classic findings does not exclude CSS4. Parent reporting, missing observations, age-dependent ascertainment, and the small denominator limit precision. |
| Developmental milestones | Median ages in the SMARCA4 subset were **rolling 8 months, sitting 12 months, crawling 15 months, walking 26 months, and first words 15.5 months**. | Parent-reported human registry, n=12; Mannino et al., October 2018 (pqac-00000008) | Demonstrates early developmental delay, particularly gross-motor delay. Medians do not describe the full range, intellectual-disability severity, or adult adaptive outcome. |
| Neurologic, musculoskeletal, and feeding findings | Seizures **1/12 (8%)**, hypotonia **2/12 (17%)**, corpus-callosum abnormalities **2/12 (17%)**, scoliosis **5/12 (42%)**, kyphosis **4/12 (33%)**, feeding-tube use **2/12 (17%)**, and constipation **5/12 (42%)**. | Parent-reported human registry, n=12; Mannino et al., October 2018 (pqac-00000008, pqac-00000009) | Indicates multisystem morbidity but may underestimate findings requiring formal examination, imaging, or age-dependent diagnosis. The reported autism or intellectual-delay value of 4/12 should not be interpreted as the true frequency of developmental impairment. |
| Cardiopulmonary involvement | Registry findings included heart murmur **2/12 (17%)**, PFO or PDA **1/12 (8%)**, VSD **1/12 (8%)**, valve stenosis **1/12 (8%)**, coarctation **1/12 (8%)**, airway malacia **4/12 (33%)**, and obstructive sleep apnea **3/12 (25%)**. A separate 15-person cohort reported congenital heart disease in approximately **50%**, ranging from PFO or PDA to tetralogy of Fallot. | Human registries and cohort; Mannino et al., October 2018; Li et al., July 2020 (pqac-00000009, pqac-00000020) | Cardiac disease may be clinically important and occasionally severe, supporting baseline cardiac assessment. Differences between estimates probably reflect cohort composition, definitions, and ascertainment. |
| Hearing and vision | Hearing loss **4/12 (33%)**, myopia **3/12 (25%)**, strabismus **3/12 (25%)**, astigmatism **2/12 (17%)**, and retinal abnormality, nystagmus, ptosis, or cortical visual impairment each **1/12 (8%)**. | Parent-reported human registry, n=12; Mannino et al., October 2018 (pqac-00000009) | Supports formal audiology and ophthalmology surveillance. Small counts and lack of standardized examination limit phenotype-frequency estimates. |
| Epigenomic biomarker | A peripheral-blood BAFopathy DNA-methylation classifier used probes with **at least 5% methylation difference**, retained probes with **AUC greater than 0.85**, used a radial-basis-function support-vector machine with **10-fold cross-validation**, and classified at a **0.5** probability threshold. | Human CSS/NCBRS methylation study; Aref-Eshghi et al., November 2018; DOI: [10.1038/s41467-018-07193-y](https://doi.org/10.1038/s41467-018-07193-y) (pqac-00000018, pqac-00000019) | Potential adjunct for ambiguous BAFopathy cases and SMARCA4 VUS assessment, not a stand-alone diagnostic test. The retrieved evidence does not provide SMARCA4-only sensitivity or specificity, and the signature overlaps other BAFopathies. |
| Current management and trials | **No cure** is available. Care is multidisciplinary and supportive: early developmental intervention, PT/OT, speech and feeding therapy, AAC when needed, and standard treatment of seizures and behavioral, cardiac, orthopedic, hearing, vision, sleep, and nutritional problems. No SMARCA4-CSS4 disease-modifying interventional trial was identified in the search. | Current expert synthesis; GeneReviews comprehensive update, **15 May 2025** (pqac-00000023, pqac-00000024, pqac-00000027) | Recommendations are largely general CSS guidance rather than evidence from SMARCA4-specific controlled trials. Treatment-response rates and genotype-directed pharmacotherapy are unavailable. |
| Allele context versus RTPS2 | Typical CSS4 alleles are predominantly **de novo missense or in-frame variants** affecting ATPase/helicase regions and are interpreted as dominant-negative or gain-of-function; RTPS2 is associated predominantly with **truncating loss-of-function variants**. Rare CSS-like phenotypes from nonsense variants or whole-gene deletions demonstrate biological overlap. | Human CSS4 cohort, deletion case, structural synthesis, and expert review; 2020–2025 (pqac-00000004, pqac-00000006, pqac-00000012, pqac-00000025) | CSS4 and RTPS2 must not be conflated. Cancer risk for canonical non-truncating CSS4 variants is unquantified; individualized cancer-genetics review is reasonable for truncating or deletion alleles, but routine RTPS2 surveillance cannot automatically be extrapolated to all CSS4 patients. |


*Table: Compact evidence summary of the principal human cohorts, phenotype frequencies, molecular diagnostic findings, management status, and allele-specific distinction between SMARCA4-related CSS4 and RTPS2. Small cohorts and overlapping BAFopathy signatures require cautious interpretation.*