Predictive modeling of repetitive transcranial magnetic stimulation efficacy in treating anhedonia in adolescents using connectome-based approaches
10.3760/cma.j.cn113661-20250331-00153
- VernacularTitle:基于连接组预测建模的重复经颅磁刺激治疗青少年快感缺失疗效预测研究
- Author:
Jianghua NING
1
;
Runxin LYU
;
Yifei ZHANG
;
Yangchao LIU
;
Dongyu CHEN
;
Baojuan LI
;
Min CAI
;
Huaning WANG
Author Information
1. 空军军医大学第一附属医院心身科，西安 710032
- Publication Type:Journal Article
- Keywords:
Depressive disorder;
Adolescents;
Anhedonia;
Repetitive transcranial magnetic stimulation;
Connectome-based predictive modeling
- From:
Chinese Journal of Psychiatry
2025;58(12):912-924
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the characteristics of brain functional connectivity changes associated with repetitive transcranial magnetic stimulation (rTMS) in adolescents with anhedonia symptoms, and to develop a predictive model of treatment efficacy based on baseline functional connectivity.Methods:A total of 88 adolescents (aged 13-18 years) with major depressive disorder and comorbid anhedonia, diagnosed according to the Diagnostic And Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), were enrolled in a randomized, double-blind, block-design trial. Participants received either active rTMS ( n=44) or sham stimulation ( n=44) for 15 consecutive days with individualized targeting. Resting-state functional magnetic resonance imaging (fMRI) data and clinical assessments were collected before and after the intervention. Brain regions were parcellated using the Brainnetome Atlas to construct whole-brain functional connectivity matrices. Linear mixed-effects models were used to identify functional connections showing significant group×time interaction effects. The percentage change in Snaith-Hamilton Pleasure Scale (SHAPS) scores (ΔSHAPS) served as the dependent variable in multiple regression analyses to examine the explanatory power of connectivity changes for treatment response. A connectome-based predictive modeling (CPM) approach was employed to predict individual treatment responses based on baseline functional connectivity with permutation testing used to validate model robustness. Results:Thirty-one functional connections showing significant group×time interaction ( F=6.67-15.69, all P<0.01) were identified between the active and sham stimulation groups, primarily involving the subcortical network (SCN), dorsal attention network (DAN), limbic network (LN), and default mode network (DMN). Changes in these connections accounted for 53% of the variance in ΔSHAPS (adjusted R2=0.53, F=4.574, P=0.001). The CPM model based on baseline connectivity showed strong predictive performance (10-fold cross-validation: r=0.65, R2=0.40, MAE=0.095, permutation P<0.001; leave-one-out cross-validation: r=0.74, R2=0.52, MAE=0.013, permutation P<0.001). Among the 59 predictive features, those originating from the LN contributed most substantially, particularly cross-network connections with the DMN and SCN. Correlation analyses revealed widespread associations between baseline predictive features and rTMS-induced connectivity changes, including significant negative correlations between baseline LN-DMN connectivity and post-treatment changes in DAN and subcortical connectivity. Conclusion:rTMS significantly alleviates anhedonia symptoms in adolescents with depression and induces widespread reconfiguration of functional connectivity across multiple brain networks. The CPM model based on baseline connectivity features effectively predicts rTMS treatment efficacy for anhedonia, providing new insights for individualized treatment strategies in adolescent depression.
