A normative model representing autistic individuals amidst Autism Spectrum Disorders phenotypic heterogeneity

Approaches to deal and understand Autism Spectrum Disorder (ASD) phenotypic heterogeneity, quantitatively and multidimensionally, are in need. Being able to access a specific individual relative to a normative reference ASD sample would provide a severity estimate that takes into account the spectrum variance. We propose such an approach analyzing the principal components of variance observable in a clinical reference sample. Using phenotypic data available in a comprehensive reference sample, the Simons Simplex Collection (n=2744 individuals), we performed Principal Component Analysis (PCA). The PCA considered ASD core-symptoms (accessed by ADI-R), important clinical features (accessed by VABS and CBCL) and IQ. PCA-projected dimensions supported a normative modeling where a multivariate normal distribution was used to calculate percentiles. An additional phenotypically homogeneous sample (ASD, IQ<75, 6-7yr, n=60) is presented as a case study to illustrate the phenotypic heterogeneity assessment and individual placement under the normative modeling approach. Three PCs embedded 72% of the normative sample variance, interpreted based on correlations (>0.50) with clinical features as: Social Functionality (39%), Behavioral Disturbance (18%) and Communication Problems (15%). A Multidimensional Severity Score (MSS) to evaluate new prospective single subjects was developed based on percentiles. Additionally, the disequilibrium among PCA-projected dimensions gave rise to an individualized Imbalance Score (ImS). The approach, named TEAplot, is implemented in user-friendly free software and was illustrated in a homogenous independent sample. Our approach proposes a basis for patient monitoring in clinical practice, guides research sample selection and pushes the field towards personalized precision medicine.