Toward a Phase-Based Biological Stratification of Autism Spectrum Disorder: A Neuroinflammatory Developmental Framework.

Abstract

Autism spectrum disorder (ASD) is a clinically and biologically heterogeneous neurodevelopmental condition arising from diverse genetic, epigenetic, and environmental influences. Increasing evidence suggests that immune dysregulation and sustained neuroinflammatory mechanisms contribute to the pathophysiology of a biologically defined subgroup of individuals with ASD. However, the temporal evolution of these processes and their relationship to clinical heterogeneity remain poorly understood. We propose a phase-based neuroinflammatory developmental framework in which systemic immune activation in genetically or epigenetically susceptible individuals initiates sustained activation of microglia and astrocytes, cytokine-mediated synaptic dysfunction, and altered neurodevelopmental trajectories. Within this framework, neuroinflammatory activity may precede or accompany developmental regression during an early active phase, followed by progressive synaptic dysfunction and persistent neurobiological sequelae. In later stages, individuals may continue to exhibit core autistic features despite reduced or absent detectable inflammatory biomarkers, providing a potential explanation for the inconsistent biomarker findings reported across the literature. Unlike previous hypotheses that primarily emphasized neuroinflammation as one contributor to ASD, the present framework introduces four conceptual advances: (i) a temporal phase-based model of disease evolution; (ii) a biologically informed stratification of a neuroinflammatory subgroup within ASD; (iii) a mechanistic explanation for temporal variability in inflammatory biomarker detection; and (iv) a series of testable predictions linking biological stage to clinical manifestations and therapeutic responsiveness. Supporting evidence derives from studies demonstrating persistent microglial and astrocytic activation, elevated inflammatory cytokines, synaptic abnormalities, immune dysregulation, developmental regression, and selected cases of autoimmune encephalitis presenting with autism-like phenotypes. Additional translational support arises from mechanistic overlap in neuroimmune signaling and preliminary therapeutic observations involving immunomodulatory and neuromodulatory interventions. The proposed framework generates specific, falsifiable predictions that can be examined using longitudinal biomarker studies, molecular neuroimaging, immune profiling, and stage-oriented clinical trials. If validated, this model could complement existing genetic and neurodevelopmental theories by supporting biological stratification of ASD, facilitating biomarker-guided patient selection, and informing individualized therapeutic strategies alongside established behavioral and rehabilitative interventions.