Cluster analysis of autistic patients based on principal pathogenetic components.

ABSTRACT

We have recently described four principal pathogenetic components in autism (I) circadian and sensory dysfunction, (II) immune abnormalities, (III) neurodevelopmental delay, and (IV) stereotypic behaviors. Using hierarchical and k-means clustering, the same 245 patients assessed in our principal component analysis can be partitioned into four clusters (a) 43 (17.6%) have prominent immune abnormalities accompanied by some circadian and sensory issues; (b) 44 (18.0%) display major circadian and sensory dysfunction, with little or no immune symptoms; (c) stereotypies predominate in 75 (31.0%); and (d) 83 (33.9%) show a mixture of all four components, with greater disruptive behaviors and mental retardation. The "immune" component provides the largest contributions to phenotypic variance (P = 2.7 x 10(-45)), followed by "stereotypic behaviors." These patient clusters may likely differ in genetic and immune underpinnings, developmental trajectories, and response to treatment.