SU86. Multimodal Social Brain Network and Facial Affect Recognition in Schizophrenia

Abstract

Abstract Background: Facial Affect Recognition (FAR) is the capacity to understand and interpret facial expressions. It is previously shown that schizophrenia patients and siblings (albeit to a lesser extent) perform worse on FAR (Meijer et al, Psychol Med, 2012). The social brain network (SBN) is associated with emotion processing, and the interaction between grey and white matter within the network might explain poor FAR performance. Here, we investigate whether disturbed FAR can be explained by a differential association between the following modalities: cortical volume [CV] and white matter integrity (fractional anisotropy [FA], radial [RD], and axial diffusivity [AD]) within the SBN between patients, siblings, and controls. Methods: We acquired brain scans at 1.5 T of 103 patients, 125 nonpsychotic siblings, and 78 controls. Using FreeSurfer 5.1, volumes of the medial orbitofrontal, fusiform gyrus, insula, caudal/rostral cingulated, and superior/middle temporal cortex were extracted, standardized, and summed to represent cortical volume of the SBN. With TRACULA 5.3, an anatomy restricted, automated, probabilistic tractography package, FA, RD, and AD were extracted of the arcuate, superior longitudinal, and uncinate fasciculus, and the cingulum-cingulate and cingulum-angular pathway then standardized and summed. SBN areas and tracts were registered to the same space. SBN Modality-by-Modality interactions and Task-by-Modality interactions were compared between groups using linear mixed-effects modeling. FDR correction was applied. Partial-omegas-squared (ωp2) were calculated. Interpretation of this effect size is as follows: small: .01 .35. Results: Within the SBN, patients had significantly higher RD (P = .018), lower AD (P = .001), and lower CV (P = .007) than controls. Siblings had smaller CV than controls (P = .030). No group effect was found in the interaction between modalities (CV*AD, CV*FA, or CV*RD). The difference in FAR between groups was significantly explained by 2 interactions, ie, FAR*AD and FAR*(CV*AD). This implicates that in patients smaller AD as well as the association between lower CV and lower AD were related to poor FAR, which was not present in controls (P = .041, ωp2 = .29 and P = .038, ωp2 = .47, respectively). Conclusion: We showed that in patients, FAR is related to an interaction of reduced fiber-tract integrity connected to smaller cortical volume within the SBN. Poorer performance in siblings did not show interacting brain structure differences related to FAR. Possibly, the extent of the abnormality in FAR performance and brain structure were too subtle. Our findings suggest that with a multimodal network approach we can identify brain networks that are related to emotion processing deficits in schizophrenia patients. Assessing structural covariance within this network is the next step.