Clinical and Genetic Characterization of Preschool Girls with Autism Spectrum Disorder: Insights from Copy Number Variants and X-Chromosome Inactivation

Background/Objectives: Females remain underrepresented in autism spectrum disorder (ASD) research. We characterized clinical and genetic profiles of preschool girls with ASD, focusing on copy number variants (CNVs) and X-chromosome inactivation (XCI). Methods: Thirty-two girls with ASD (mean age 3.18 ± 0.85 years) underwent standardized assessment, chromosomal microarray analysis, FMR1 testing, and XCI evaluation using a methylation-sensitive androgen receptor assay. Clinical features were compared between participants with at least one reportable CNV and those with no reportable CNV detected. Identified CNVs were retrospectively reassessed according to ACMG/ClinGen standards. Multiple-testing correction and a sensitivity analysis comparing pathogenic/VUS CNV carriers with participants showing no reportable CNV or only likely benign CNVs were performed. Results: Developmental impairment was frequent, with 24/32 participants (75.0%) showing a GMDS-III developmental quotient below 70. At least one reportable CNV was identified in 14/32 participants (43.8%), and 18/32 (56.3%) had none. Following expert reassessment according to ACMG/ClinGen standards, 17 CNVs were classified as pathogenic (n = 2), variants of uncertain significance (n = 7), or likely benign (n = 8). Pathogenic CNVs were identified in two participants (6.3% of the cohort). Among 25 participants informative for XCI, 24 (96.0%) showed non-markedly skewed XCI and one (4.0%) showed highly skewed XCI (7:93). In primary analyses based on original reportable-CNV status, nominal differences were observed for ADOS-2 symptom severity and sleep difficulties, but these did not remain significant after false discovery rate correction. Continuous GMDS-III, VABS-II, and DP-3 scores did not differ significantly between groups. Primary cross-sectional phenotype–CNV associations did not remain significant after multiple-testing correction. In the exploratory CBCL 6–18 follow-up analyses, differences in Externalizing Problems, Total Problems, and DSM-oriented Attention-Deficit/Hyperactivity Problems remained significant after FDR correction under the original reportable-CNV grouping, but none remained significant in the sensitivity analysis based on CNV reclassification. Conclusions: Highly skewed XCI was uncommon in this cohort, whereas pathogenic CNVs were identified in a minority of participants despite the relatively high frequency of reportable genomic findings. Primary cross-sectional clinical differences did not survive multiple-testing correction, and phenotype–CNV associations were further attenuated after CNV reclassification. These findings underscore the heterogeneity of female ASD and support a cautious interpretation of genotype–phenotype associations based solely on the presence of reportable CNVs.