Clinical phenotypes, bone-age delay, and exploratory coding-variant architecture in idiopathic short stature: a single-center integrative analysis

Abstract Background. Idiopathic short stature (ISS) is a diagnosis of exclusion applied to children with height more than two standard deviations below the population mean in whom no causative disorder is identified. ISS is clinically and genetically heterogeneous, and integrating routinely collected clinical data with sequencing data may help to generate testable hypotheses about its architecture. We performed a single-center, exploratory analysis combining structured clinical variables, phenotypes extracted from free-text notes, and sample-level coding-variant matrices in children with ISS and normal-height clinic controls. Methods. Children evaluated at a single pediatric clinic were classified as ISS (height SDS < − 2; n = 25) or normal height (height SDS > − 1; n = 22); children with intermediate height SDS were excluded. Sequencing matrices were available for 22 ISS and 7 normal-height children. Clinical variables (IGF-1, IGFBP-3, bone-age difference [BAD]), free-text–derived phenotype categories, mutation burden, ISS-enriched genes (nominal Fisher’s exact P < 0.05), phenotype subgroups, and a seed-gene protein–protein interaction network were analyzed with Wilcoxon, Fisher’s exact, and Spearman tests, followed by exploratory Gene Ontology over-representation analysis. Results. ISS children had lower IGF-1 (median 101.4ng/ml vs 351.2ng/ml), lower IGFBP-3 (3.19ug/ml vs 5.20ug/ml), and more delayed bone age (BAD − 2.1 vs 0.1 years) than normal-height children. Facial, skeletal/limb, and neurodevelopmental phenotype categories extracted from clinical notes were enriched in ISS. Within ISS, height SDS correlated with BAD (r_s = 0.51), and familial short stature was associated with greater bone-age delay. Total and high-impact coding-variant burdens were lower in ISS than in the small normal-height sequenced set. Thirty-five unique genes were nominally enriched for variants in ISS; exploratory Gene Ontology terms included chromatin-remodeling complexes involving EP400 and BPTF. A seed-gene STRING network placed SRCAP adjacent to EP400 and BPTF; SRCAP missense variants were absent in sequenced ISS (0/22) but present in 4/7 sequenced normal-height children, a contrast that cannot be interpreted as pathogenic or protective because variant matrices were not filtered for rarity or predicted pathogenicity. Conclusions. This hypothesis-generating, single-center study links clinical severity of ISS to bone-age delay and nominates chromatin-remodeling genes as candidates for follow-up in larger, stringently filtered cohorts. Findings require independent replication and should not be used for clinical decision-making.