Early-response Prediction Of Antidepressant Treatment In Major Depressive Disorder Via Magnetic Source Imaging

Currently,treatment selection for major depressive disorder(MDD)is mainly based on behavioral symptoms of patients and experience of psychiatrists.There is a lack of reliable predictors to reduce unsuccessful treatment trials for MDD patients,so exploring the response-related biomarkers to achieving individualized prediction at baseline is important.Neuroimaging has provided some potential predictors of antidepressant response,but there is still a lack of research on biomarkers in specific time period and frequency band.Based on the advantage of magnetoencephalography(MEG)in temporal resolution,neural activation under sad face stimulation associated with the early-response of antidepressant was explored from the sensor level and the source level in this study.The main contents are as follows:1.Neural signals of the whole brain channels were analyzed to find the regional features related to the early-response of antidepressant.After dividing the time period and frequency band,it was found that the left occipital cortex in beta band of 0-200 ms and the left parietal cortex in gamma band of 0-200 ms showed significant difference among the three group(responders,non-responders and healthy controls)by cluster-based permutation test(beta band:p=0.013,gamma band: p=0.048).It may reflect that the antidepressant response is related to the degree of dysfunction in negative emotion processing of MDD patients at baseline.The statistical analysis of the mean activation in significant clusters showed that the regional features obtained from the sensor level can not achieve the baseline prediction of antidepressant response.2.Based on the dual-routed model of emotion processing,regions of interest(ROIs)were selected to explore the biomarkers from the source level.The power spectrum of ROIs and functional connectivity between any two ROIs in subcortical pathway and cortical pathway were calculated.After statistical test among three groups,no significant feature was obtained.It may be caused by the noise disturbance in signal analysis,or the power spectrum and the functional connectivity can not reflect the neural activation pattern related to antidepressant response.3.Cross frequency coupling between ROIs was analyzed to explore the biomarkers of antidepressant early-response based on the dual-routed model.Utilizing the phase-amplitude cross frequency coupling analysis,the cross frequency coupling matrices of ROIs were calculated for all participants.The statistical results after cluster-based permutation test revealed that the right thalamus,the right orbitofrontal cortex and the right amygdala showed a significant modulation pattern based on the dual-routed model and this signal interaction pattern was closely related to the antidepressant early-response(subcortical pathway:p=0.0086,cortical pathway: p=0.027).The dysfunction of region-to-region modulation in the emotion processing circuit may lead to more serious negative emotion bias in the non-responder group at baseline,and then affect their responses to the subsequent medication.By analyzing the mean values of the significant clusters,it was found that the features obtained from the coupling matrices were related to the improvement of clinical symptoms after two-week antidepressant treatment(significant cluster in subcortical pathway: r=0.46,significant cluster in cortical pathway: r=0.45),and these features may be used as biomarkers for baseline prediction of antidepressant early-response(sensitivity=0.82,specificity=0.69).In conclusion,utilizing MEG data under sad face stimulation,we found some features in specific bands of 0-200 ms from the sensor level,which are related to the antidepressant early-response.By further exploration of cross frequency modulation pattern based on negative emotion processing from the source level,we found some potential biomarkers for baseline prediction of antidepressant early-response.