LAUSR

Recently, genetic analyses have identified highly shared overlaps in polymorphisms across threemajor psychiatric disorders: schizophrenia (SCZ), bipolar disorder (BD) and major depressive disorder (MDD) (Cross-Disorder-Group, 2013). Also, clinical analyses find similarities in some psychotic symptoms across three disorders and high comorbidity between diagnoses (Borsboom et al., 2011). Collectively, this evidence possibly suggest there exists a common neurobiological mechanism across three related disorders. In our recent meta-analysis of comparing the fractional anisotropy (FA) of white matter between SCZ and BD, we found shared impairments of FA at left genu of corpus callosum (GCC) and left posterior cingulum fibers (Dong et al., 2017). Meanwhile, Wise et al. (2016) found MDD and BD are both characterized by FA abnormalities in left GCC, and BD showedmore reducedwhite matter integrity in the left posterior cingulum when comparing to MDD. Combining with the two metaanalysis studies, we hypothesized that a) the GCC will represent common alterations in FA across the three disorders and b) that SCZ and BP will show significantly reduced FA in the left posterior cingulum relative to MDD. However, these presumptions lack direct comparison among three conditions. As an extension and integration,we performed the voxel-basedmeta-analytic comparison of whole-brain FA evaluated bydiffusion approach to identify the commonor disorder-specific structural abnormalities among three illnesses. For whole-brain Diffusion tensor imaging literature in SCZ and BD, we used our previous datasets, i.e., 24 datasets for SCZ; 23 datasets for BD (Dong et al., 2017). In accordance with our previous inclusion and exclusion criteria for study selection, thirty studies ofMDDwere included in following analyses (Table S1). Descriptive Information for each Sample was summarized in supplementary material. All analyses were performed using anisotropic effect-size-based algorithms (AES-SDM) software (http://www.sdmproject.com) in a standard process. First, separated analyses were conducted to investigate the impairments of FA within each disorder group (P b 0.005, shown in Fig. S1 and Table S2). We also did some additional analyses to guarantee the stability and replicability with each patient group, see Supplementary material for details. Secondly, we conducted three voxel-wise quantitative comparisons between each pair with controlling for age, gender (P b 0.0005). Thirdly, three separate conjunction analyses for each pair were conducted (P b 0.0025). Finally, another conjunction analysis of three separate conjunction maps were conducted to identify common FA alterations (see Dong et al. (2017) for more details).