Latent class analysis identifies distinctive behavioral subtypes in children with fragile X syndrome

Abstract

AbstractFragile X syndrome (FXS) is associated with a characteristic profile of physical and neurobehavioral abnormalities. These phenotypical features are highly variable among affected individuals, which leads to difficulties in developing and evaluating treatments as well as in determining accurate prognosis. The current investigation employed data from FORWARD, a clinic-based natural history study of FXS, to identify subtypes by applying latent class analysis (LCA).A pediatric cross-sectional sample of 1,072 males and 338 females was subjected to LCA to identify neurobehavioral classes (groups). Input consisted of multiple categorical and continuous cognitive and behavioral variables, including co-occurring behavioral conditions, sleep and sensory problems, measures of autistic behavior (SCQ, SRS-2), and scores on the Aberrant Behavior Checklist revised for FXS (ABCFX). Clinically relevant class solutions were further delineated by identifying predictors using stepwise logistic regressions and pairwise comparisons. Following this, classes were characterized in terms of key demographic, genetic, and clinical parameters.LCA fit parameters supported 2- to 6-class models, which showed good correspondence between patterns of co-occurring conditions and scores on standardized measures. The 5-class solution yielded the most clinically meaningful characterization of groups with unique cognitive and behavioral profiles. The “Mild” class (31%) included patients with attention problems and anxiety but few other major behavioral challenges as reflected by scale scores. Most individuals in the “Severe” class (9%) exhibited multiple co-occurring conditions and high mean scale scores on behavioral measures. Three “Moderate” classes were identified: a “Moderate Behavior” class (32%), a “Social Impairment” class (7%), and a “Disruptive Behavior” class (20%). All classes displayed distinctive SRS-2, SCQ, and ABCFX profiles, which reflected their degree of non-overlap as estimated by pairwise effect sizes. Groups differed with regard to sex, intellectual disability, autism spectrum disorder diagnosis, and medication use.These findings support the notion that, it is possible to identify behavioral subtypes in children with FXS, reflecting both overall level of severity and specific areas of impairment. These subtypes have implications for clinical management and therapeutic development and assessment. Future studies are needed to determine the stability of these group profiles and their relationship with other aspects of the FXS phenotype.