Inheritance-dependent genotype–phenotype architecture in DEAF1-associated neurodevelopmental disorders: five unrelated patients with de novo DEAF1 variants and an updated synthesis of 67 individuals

Background Pathogenic variants in DEAF1 cause DEAF1-associated neurodevelopmental disorders (DAND) with autosomal-dominant (typically de novo ) and autosomal-recessive (biallelic) presentations. We sought to characterize five unrelated children with de novo DEAF1 variants, define epilepsy phenotypes, and refine inheritance-specific genotype–phenotype architecture with developmental context. Methods Trio-based whole-exome sequencing with Sanger confirmation was performed in five unrelated children. Missense variants were evaluated using structural modeling and in silico stability/physicochemical assessments. DEAF1 expression was examined using GTEx, the Human Protein Atlas, BrainSpan, Evo-devo mammalian organ transcriptomes, and single-cell brain organoid datasets. A systematic literature review (inception–March 2026) curated eligible cases for intersection-based UpSet analyses and system-level stacked summaries. Results Five unrelated children carried heterozygous de novo DEAF1 missense variants; four mapped to the SAND domain (p.Leu203Pro, p.Gly212Ser, p.Ile228Met, p.Leu272Ser) and one was N-terminal (p.Ala29Pro). p.Ala29Pro and p.Ile228Met were not recorded in ClinVar, HGMD Professional, or the reviewed literature. All five patients had intellectual disability/developmental delay with prominent expressive language impairment and motor delay/hypotonia. Epilepsy was observed in all patients, with onset ranging from 7 months to 2 years 8 months, and focal impaired consciousness seizures (FIC) predominated. EEG abnormalities were observed in 4 of the 5 patients, and all five were seizure-free on anti-seizure medications at the last follow-up. Structural modeling suggested variant-dependent perturbations, particularly among SAND-domain substitutions, but these findings remain computational predictions. Expression analyses supported brain-enriched, developmentally regulated DEAF1 expression and broad neuronal/progenitor expression in human brain organoids. A literature review identified 62 individuals from 15 studies; combined with our cohort, 67 individuals were analyzed ( de novo n = 52; biallelic n = 15). De novo variants clustered within/around the SAND domain, whereas biallelic variants were more broadly distributed. UpSet and stacked-bar analyses revealed a shared neurodevelopmental core but distinct co-occurrence patterns, with autism/epilepsy shaping dominant intersections in de novo cases and microcephaly/MRI abnormalities contributing more prominently in biallelic disease. Conclusions Our study refines the de novo clinical and allelic spectrum of DAND and delineates inheritance-specific genotype–phenotype architecture with developmental and cellular context.