... Given that many patients included in our meta-analysis were undergoing medication at the time of the scan (Table S5), this variable is likely to have some impact on the COA-O network. However, there is some evidence that suggests that medication alone is unable to explain our results: first, morphological effects of medication are often found to be localized to restricted regions, such as medial temporal lobe and subgenual cortex with lithium ( Germaná et al., 2010 ;Hafeman et al., 2012 ) and BG with antipsychotics ( Navari and Dazzan, 2009 ), while our increase nodes are situated in many other brain areas; second, some studies report that medications could attenuate pathological decreases rather than increase GM volume in patients compared to controls ( Hibar et al., 2016 ;Sarrazin et al., 2019 ;Sheline et al., 2003 ;Wada et al., 2005 ;Zung et al., 2016 ); third, atypical antipsychotics, although found by some to be neurotrophic and induce neurogenesis ( Bai et al., 2003 ;Halim et al., 2004 ;Park et al., 2006 ;Wakade et al., 2002 ;Wang et al., 2004 ) produced mixed volumetric findings ( Massana et al., 2005 ;Navari and Dazzan, 2009 ) but often no increase effects were found in the BG ( Chakos et al., 1995 ;Frazier et al., 1996 ;Lang et al., 2004Lang et al., , 2001Scheepers et al., 2001 ;Westmoreland Corson et al., 1999 ). Moreover, a study reports an absence of increased volume of BG also for typical antipsychotics ( Kreczmanski et al., 2007 ); fourth, anticonvulsant drugs, used in the treatment of bipolar disorder but also for epilepsy (the only neurological disease with a non-negligible number of experiments in our database) showed to produce decreases or no effect ( Abé et al., 2016 ;Chang et al., 2009 ;Germaná et al., 2010 ;Hibar et al., 2018 ); fifth, the increased striatal volume of relatives of schizophrenic patients suggests a genetic factor ( Oertel-Knöchel et al., 2012 ). ...