Neuro-Electrophysiological Study Of Impulsivity Among Adolescent Depressive Patients With Non-Suicidal Self-Injury

Objectives:This study intends to explore the impulsivity characteristics on different dimensions and their neuro-electrophysiological characteristics among adolescent depressive patients with non-suicidal self-injury(NSSI)by scale measurement,behavioral task,resting electroencephalograms(EEG)and event-related potentials(ERPs).Meanwhile,we aim to establish a prediction model of adolescent depressive patients with NSSI based on the psychological and neuro-electrophysiological characteristics by machine learning method.Methods:Three groups of subjects were included in this case-control study:adolescent depressive patients with NSSI(NSSI group),adolescent depressive patients without NSSI(non-NSSI group)and healthy controls(HC group).Participants were interviewed with the Mini International Neuropsychiatric Interview for children and adolescents 5.0 edition to confirm the diagnosis of MDD and eliminate other mental diseases.All the subjects were asked to finish children’s depression inventory for the screening of depression symptom,the short version of the UPPS impulsive behavior scale(UPPS)and Borderline Personality Features Scale for children(BPFS)for evaluate impulsivity.Ottawa Self-injury Inventory and Columbia-suicide Severity Rating Scale were used to assess the severity of NSSI among the adolescent depressive patients with NSSI.Hamilton Rating Scale for depression was used to assess the severity of depression for all the adolescent depressive patients.1.The study 1 included 50 subjects in NSSI group,46 subjects in non-NSSI group and 50 subjects in HC group.Data of resting EEG with 5minutes ofthe three groups with eyes closed were collected,and the differences of resting EEG among the three groups were compared by EEG microstate analysis.2.The study 2 included 50 subjects in NSSI group,45 subjects in non-NSSI group and 50 subjects in HC group.Go/no-go behavior task and ERPs were conducted for impulsive behavior and neurological measurement.The impulsive differences of behavior task and ERPs were compared among the three groups.3.The study 3 was based on the psychological and neuro-electrophysiological characteristics of adolescent depressive patients with NSSI found in the first and second study,a predicting model for adolescent depressive patients with NSSI was established and tested by machine learning of support vector machine.Results:1.In study 1 and study 2,there was gender difference between NSSI group and HC group(p < 0.05),and there were not significant statistical differences in other general demographic informations among the three groups.There were not significant statistical differences in the severity of depression and other disease characteristics between NSSI group and non-NSSI group.2.In study 1 and study 2,NSSI group and non-NSSI group scored significantly higher than HC group in the total score and subscale scores of BPFS,as well as the negative urgency and positive urgency subscales of UPPS(p < 0.05),in both the general population or the first episode patients.The total score and subscale scores of BPFS in NSSI group were higher than that in non-NSSI group(p < 0.05),but there were not sigificant statistical differences in the negative urgency and positive urgency subscales of UPPS(p > 0.05).The total score and subscale scores of BPFS and positive urgency subscales of UPPS were positively correlated with the severity of NSSI in the past 1,6,and 12 months(p <0.001).3.The results of resting EEG microstate in study 1 showed that the occurrence of microstate A in NSSI group was higher than that in HC group(p < 0.05),and there were not significant statistical differences between non-NSSI group and NSSI group,HC group and non-NSSI group.The occurrences of microstate A were positively correlated with severity of NSSI in the past 1,6,and 12 months(r = 0.184—0.188,p <0.05).After excluding the potential effects of gender and drugs,the occurrence of microstate A was still significantly higher in NSSI group than in HC group(p < 0.05).In addition,the coverage of microstate A was higher than that in HC group(p < 0.05),and the coverage of microstate A was positively correlated with the severity of NSSI behavior in the past 1 and 12 months(r =0.191,0.203,p < 0.05).4.The results of go/no-go behavior task in study 2 showed that the accuracy of go task in NSSI group was significantly lower than that in HC group,and there were not significant statistical differences between NSSI group and non-NSSI group,non-NSSI and HC group.The accuracy of go task were negatively correlated with the severity of NSSI behavior in the past 1,6,and 12 months(r =-0.248—-0.190,p < 0.05).After excluding the potential effects of drugs,the accuracy of go task in NSSI group was still lower than that in HC group(p < 0.05),while the accuracy of go task was only negatively correlated with the severity of NSSI behavior in the past 6 months(r =-0.196,p < 0.05).5.In study 2,at Pz and Cz channels,the amplitude of P3 on the no-go stimulation were higher than that of the go stimulation both in the total population and the first episode patients(p < 0.001).There were not significant statistical differences in latency or amplitude of N2 or P3 among the three groups at different electrodes both in the total population and the first episode patients(p > 0.05).6.In study 3,the machine learning prediction model based on impulsivity traits and resting EEG microstate shown it has good efficiency for predicting adolescent depressive patients with NSSI,with accuracy of 83.33%,sensitivity of 83.33%,specificity of 88.89% and area under curve(AUC)of 0.93.Conclusions:1.Adolescent depressive patients with NSSI had higher self-rated and behavioral impulsivity,which were closely related to the severity of NSSI.2.Adolescent depressive patients with NSSI showed abnormal microstate A of resting EEG in neuro-electrophysiological which related to impulsivity.And the abnormal microstate A were closely related to the severity of NSSI.3.The machine learning prediction model based on resting EEG microstate combined with impulsive traits was expected to be precisely and sensitively classify adolescent depressive patients with NSSI.