Disentangling autism heterogeneity by the distinction of disability versus difference over development

Heterogeneity is ubiquitous amongst autistic individuals at multiple scales. A multitude of phenotypes coexists under the autism diagnostic label. Nowadays, the concept of personalized medicine is changing how research and clinical areas operate by emphasizing individual psychobiological characteristics to complete the diagnostic assessment and to design intervention/therapeutic programs. Therefore, given its phenotypic variability, treating autism as a singular entity would not be the best option to adhere to precision medicine principles. This thesis focuses on the controversial topic of autism heterogeneity; it tries to provide a novel framework to study and interpret it. I am going to examine the hypothesis that, within noncore yet developmentally important domains (such as language, motor, cognition, and adaptive functioning), autism is a collection of discrete subtypes, each of which has distinctive functional characteristics and consequently unique needs and priorities. The first chapter traces the history of autism, focusing on how heterogeneity emerged, from the very first Kanner’s description up to the current DSM-5 diagnostic algorithm. Here, I also describe a new theoretical model designed to parse autism heterogeneity: AUTISMS-3D. This model sustains that autism can be divided at least into 2 subtypes that diverge during the development characterized by the presence (Type I; disability) or the absence (Type II; difference) of profound disability in non-core autism domains (e.g., language, motor, cognition, and adaptive functioning). The following empirical chapters utilized unsupervised learning techniques to detect subtypes within non-core autism domains to support the AUTISMS-3D model. Chapter 2 provides a practical example of how to distinguish autism Type I and Type II subtypes in a cohort of young autistic children (<68 months). Chapter 3 identifies autism subtypes within the domain of adaptive functioning and then assesses their prognostic validity in verbal, non-verbal cognitive, and motor skills. Chapter 4 describes autism subtypes based on the motor domain and then tests them for difference in kinematic motor noise during the execution of a repeated motor task. Results from the empirical chapters provide evidence that discrete subtypes can be detected within autism when non-core individual characteristics are taken into account. In the last chapter, which serves as a general conclusion, I will discuss how these findings provide insight about how to parse autism heterogeneity within the AUTISMS-3D theoretical framework.