M61. Data-Driven Approach Identified Functionally and Physiologically Distinct Psychosis Subtypes

Abstract

Background: Unlike diagnoses in other branches of medicine, current classification of psychiatric disorders relies on phenomenological observations, despite the increasing evidence pointing to superior characterization by physiological-based markers. Common risk genes, symptoms, and medications had been found between patients traditionally diagnosed with schizophrenia, schizoaffective, and bipolar disorder with psychosis. To address this issue, and to ultimately achieve customizable treatment plans for illnesses with psychotic features, identification of biomarkers for psychosis and their criteria is essential. Methods: Part of Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP), the present study was conducted in continuation of previous work that identified psychosis Biotypes (Clementz et al., 2016), and it included 709 probands with psychotic features and 342 control subjects. In addition to the biomarker panel used in Clementz et al. (2016), this iteration included additional biomarkers capturing variance in ocular motor performance, inhibition, and cognitive control. Similar multivariate analysis procedures were used. Birchwood Social Functioning Scale and structural MRI characteristics from Freesurfer analyses were used as validators of cluster membership. Results: On average, probands showed lower cognitive performance, lower accuracy to motor tasks, and lower EEG power to auditory stimuli than controls. Interestingly, these anomalies were not ubiquitous among the probands. From within-task principle component analysis (PCA) and k-means clustering, 4 (rather than 3 as previously reported by Clementz et al.) phenotypically distinct proband groups emerged: Group 4 had hyperactive motor response and low cognitive control; Group 3 showed normal cognition and motor control but abnormally low EEG amplitude in response to auditory stimuli; Group 2 was slow in motor tasks combined with poor cognitive control, and Group 1 performed more similarly to controls across tasks. Social functioning scale confirmed with these findings: a significant difference in total score was found between groups, with Group 1 scoring similarly to control subjects, higher than the other 3 (F(3, 560) = 3.882, P = .009). Various cortical regions involved in motor and executive functions, including middle temporal and superior frontal areas, showed drastic differences in volume and thickness across groups (all Ps < .001). Conclusion: Four clusters within probands showing non-unified structural differences, increments in cognitive deficits, and low reactivity to auditory inputs in only one group suggested a high degree of heterogeneity among psychotic populations that does not necessarily conform to traditional diagnoses. Investigation to psychosis phenotypes at network and circuit level will help further discern disease subtypes and refine diagnoses and prescription guidelines.