CPCA of the N-back working memory task in schizophrenia: a multivariate approach to the identification of intermediate phenotypes in schizophrenia

Wednesday, October 26, 2011 — Poster Session III
10:00 a.m. – Noon	Natcher Conference Center	NIMH	IMAG-4

Authors

• DAA Baranger
• J Paiement
• Y Tong
• VS Mattay
• TS Woodward
• DR Weinberger
• JH Callicott

Abstract

Constrained principal component analysis (CPCA) is a multivariate analysis technique for BOLD fMRI that identifies neural systems operating in tandem during tasks. CPCA is optimal for analyses of whole-brain connectivity as it is exploratory and data-driven. We compared network-based activity in an N-back working memory (WM) paradigm at 3-Tesla in 271 healthy controls, 44 patients with schizophrenia, and 99 unaffected siblings. CPCA identified 3 components of interest. The first component was a task-positive WM-like network (bilateral dorsal and ventral PFC, right parietal cortex, and anterior cingulate). The second component was a task-negative functional network (posterior cingulate and the medial frontal gyrus).The third component was a task-positive motor response network (left primary and supplemental motor cortices). We found that the task-positive and task-negative component predictor weights differentiated patients and their siblings from controls – suggesting such system distinctions as familial trait markers of schizophrenia risk. Patients differed from both siblings and controls on the third component, suggesting that this difference is secondary to confounding state variables (e.g., antipsychotics). In conclusion, component predictor weights from CPCA could be interesting candidates for neuroimaging intermediate phenotypes, as a single metric can reflect the efficiency of an entire functional network.

back to top