Speech and language disorders are known to have a substantial genetic contribution. Although frequently examined as components of other conditions, research on speech differences as separate phenotypic subgroups has been limited so far. In this study, we performed an in-depth characterization of speech disorders in 52,143 individuals, reconstructing clinical histories using a large-scale data mining approach of the Electronic Medical Records (EMR) from Children's Hospital of Philadelphia.

The frequency of speech disorders was the highest between 2 and 5 years old. We found a significant overlap with neurodevelopmental disorders and to a lesser degree with movement disorders and epilepsies with STXBP1, PTEN, and CACNA1A as the most common single gene disorders retrievable through EMR analysis. When assessing genotype-phenotype correlations, we observed associations of STXBP1 and aphasia as well as MYO7A with speech and language development delay due to hearing loss. In order to assess the reliability of EMR analysis to reflect speech diagnoses, we developed an EMR Visibility Index to assess how diagnosis codes reflect patient note content analyzed through Natural Language Processing. Using this metric, we found that aphasia and speech apraxia could easily be retrieved through diagnosis codes, while stuttering was EMR-invisible with only one in ten individuals having appropriate ICD10 codes. Finally, in a subcohort of 522 individuals, we analyzed raw whole exome sequencing data and identified rare variants associated with speech phenotypes. We found that missense variants in GRM2, a gene encoding a metabotropic glutamate receptor 2 without an established phenotype, are associated with aphasia.

In summary, this study outlined the landscape of pediatric speech disorders, demonstrating an unexpected phenotypic complexity and novel genotype-phenotype associations. Subgroups of pediatric speech disorders differ significantly with respect to the composition of monogenic etiologies. Speech apraxia and aphasia are frequently associated with single gene disorders, while stuttering is not associated with any single genetic etiology.