BackgroundNeuroanatomic studies have not yet addressed how subtle phenotypic distinctions in psychosis alter the underlying brain changes, and whether there is evidence for psychosis as a dimensional construct. We explored the relationship of cortical GM thickness to psychotic phenotypes in children. MethodsCross-sectional comparison of anatomic brain imaging between patients referred as childhood-onset schizophrenia (COS) but ruled out after a drug free inpatient observation. Groups included: patients with no evidence of psychosis (n=22) after drug free observation, patients with psychosis not otherwise specified (PNOS; total n=29) further divided into those without other axis I diagnoses (n=13) and those with other axis I comorbidities (n=16), age/sex matched COS patients (n=48), and 51 matched healthy controls. GM cortical thickness was compared between the groups, and regressed on patients' SAPS, SANS and GAS scores. ResultsPatients with no evidence of psychosis showed no cortical GM deficits. Presence of psychosis (PNOS with or without co-morbidities) showed some areas of temporal and prefrontal deficits, more subtle compared to the extensive bilateral cortical deficits seen for COS. GAS SAPS and SANS scores showed a relationship with cortical GM thickness although it did not survive adjustment for multiple comparisons. ConclusionsThese results highlight the need for careful phenotypic characterization, as subtle diagnostic distinctions appear to reflect distinct underlying patterns of brain deficits. The incremental nature of cortical deficits from no psychosis to PNOS to COS may further support dimensional model for psychosis.