Grey matter volume differentiates psychosis patients according to cognitive ability

Background: Brain-based intermediate phenotypes may be common to schizophrenia (SZ) and bipolar I disorder (BD-I) in association with shared genetic vulnerability. We used voxel-based morphometry (VBM) to identify grey matter volume (GMV) related to working memory performance regardless of diagnosis. Methods: Using VBM8 toolbox, GMV in SZ (n = 37) and BD-I (n = 21) were assessed relative to healthy controls (HC; n = 39) and each other. Focal analyses were conducted for groups defined by performance on a working memory task: patients achieving less than 50% accuracy were treated as ‘poor performers’ (PP, n=26); those above 50% were grouped as ‘good performers’ (GP, n = 31). 31 HCs performed above 50%. Results: After controlling for age, sex, years of education and medication dosage, analyses of cases by performance revealed that, relative to HC, PP patients showed GMV reductions in medial frontal cortex, and the GP patients showed reductions in precuneus; additionally, reduced inferior frontal gyrus differentiated PP from GP. Traditional group analyses demonstrated GMV reductions in anterior cingulate cortex for Sz relative to HC (p < 0.0001 uncorr), and in precentral gyrus and precuneus (p < 0.05 FWE) for BD-I relative to HC and SZ. Conclusions: In the context of discrete grey matter reductions in SZ and BD-I, GMV successfully differentiated patients with poor working memory (regardless of diagnosis) in regions relevant for successful task performance. These findings point toward the existence of shared brain-based intermediate phenotypes among SZ and BD-I cases with more severe cognitive deficits.