Maximizing Diagnostic Yield in Intellectual Disability Through Exome Sequencing: Genotype–Phenotype Insights in a Vietnamese Cohort

Background: Intellectual disability (ID) is a heterogeneous condition caused by diverse genetic factors, including single-nucleotide variants (SNVs) and copy number variants (CNVs). Whole-exome sequencing (WES) and clinical exome sequencing (CES) have become essential tools for identifying pathogenic variants; however, their relative diagnostic performance in ID has not been fully characterized. Methods: Children diagnosed with ID or related neurodevelopmental disorders underwent WES or CES. Identified variants were classified according to ACMG/AMP and ClinGen guidelines, with segregation analysis performed when parental samples were available. Diagnostic yields were compared across demographic, prenatal, and phenotypic subgroups. A multidimensional semi-quantitative scoring system encompassing 15 clinical domains (e.g., age at onset, neuro-motor function, seizures, MRI findings, vision, and dysmorphic features) was developed. Z-scores were calculated for each parameter, followed by hierarchical cluster analysis (HCA) and correlation modeling to define genotype–phenotype associations and pathway-level clustering. Results: A broad spectrum of pathogenic and likely pathogenic variants across multiple genes and biological pathways was identified in our study. CNV-associated cases frequently exhibited prenatal anomalies or multisystem phenotypes associated with large chromosomal rearrangements. Monogenic variants and their corresponding phenotypic profiles were identified through clinical exome sequencing (CES) and whole-exome sequencing (WES). Phenotypic HCA based on Z-scores revealed three major biological groups of patients with coherent genotype–phenotype relationships: Group 1, severe multisystem neurodevelopmental disorders dominated by transcriptional and RNA-processing genes (POLR1C, TCF4, HNRNPU, NIPBL, ACTG1); Group 2, intermediate epileptic and metabolic forms associated with ion-channel and excitability-related genes (SCN2A, PAH, IQSEC2, GNPAT); and Group 3, milder or focal neurodevelopmental phenotypes involving myelination and signaling-related genes (NKX6-2, PLP1, PGAP3, SMAD6, ATP1A3). Gene distribution significantly differed among these biological categories (χ2 = 54.566, df = 34, p = 0.0141), confirming non-random, biologically consistent grouping. Higher Z-scores correlated with earlier onset and greater neurological severity, underscoring the clinical relevance of the multidimensional analytical framework. Conclusions: This study highlights the genetic complexity and clinical heterogeneity of intellectual disability and demonstrates the superior diagnostic resolution of WES and CES. Integrating multidimensional phenotypic profiling with genomic analysis enhances genotype–phenotype integration and enables data-driven phenotype stratification and pathway-based re-analysis. This combined diagnostic and analytical framework offers a more comprehensive approach to diagnosing monogenic ID and provides a foundation for future predictive and functional studies.