Multivariate coupling of EEG microstate class B hyperactivity with a distinct psychosocial profile in adolescents with major depressive disorder and self-harm

Abstract Background Self-harm (SH) frequently co-occurs with major depressive disorder (MDD) in adolescents and represents a critical risk factor for subsequent suicidal behavior. While resting-state EEG microstate alterations have been implicated in depression pathophysiology, the specific neurophysiological signatures distinguishing SH from non-self-harm (NSH) remain elusive. This study aimed to identify distinct microstate abnormalities in adolescents with MDD and SH and examine their multivariate associations with psychosocial characteristics. Methods A total of 84 adolescents with MDD (42 with SH, 42 with NSH) were enrolled. Psychosocial assessments included the Self-rating Depression Scale (SDS), Teenager Social Support Rating Scale (TSSRS), and Inventory of Parent and Peer Attachment (IPPA). Resting-state EEG was recorded, and four standard microstate classes (A–D) were extracted. Group comparisons were assessed using permutation analysis of covariance adjusting for demographic and clinical covariates. Regularized canonical correlation analysis (rCCA) was used to examine multivariate associations between psychological variables and EEG parameters, with leave-one-out cross-validation to optimize regularization. Results Compared to the NSH group, the SH group showed higher depression severity, higher subjective and objective social support, and lower father and total parental attachment (all FDR-q < 0.05). EEG analysis showed a longer Microstate Class B duration ( q = 0.038), whereas Class B contribution was not significant ( q = 0.182). Because visual imagery was not directly assessed, this alteration cannot be interpreted as direct evidence of sustained visual imagery processing. The SH group also showed increased bidirectional B↔C and decreased A↔D transition probabilities ( q = 0.014–0.038). Class B duration correlated positively with depression severity and social support, but negatively with father attachment ( r = − 0.506, q < 0.001). The rCCA identified a significant canonical correlation ( r = 0.538, P = 0.027) between a psychological profile of “high depression, high social support, low parental attachment” and Microstate B hyperactivity. Conclusions Adolescents with depression and self-harm exhibit a distinct neuro-psychosocial profile characterized by Microstate Class B hyperactivity coupled with high depression, high social support, and low parental attachment. These EEG features are candidate electrophysiological correlates of SH requiring longitudinal and multi-center validation.