Non‑suicidal self‑injury (NSSI) is common in adolescence and is particularly prevalent among patients with major depressive disorder (MDD). The behavior—defined as deliberate self‑inflicted tissue damage without suicidal intent and often operationalized as at least five episodes in the past year—carries substantial risk to physical and mental health and increases subsequent suicide risk. Adolescence is a period of heightened neuroplasticity and emotional regulation development, which may contribute to vulnerability for NSSI. Prior resting‑state functional MRI (rs‑fMRI) studies have implicated regions such as the prefrontal cortex, limbic system, caudate, and precuneus, and networks including the default mode network (DMN) and visual network (VN), but findings remain inconsistent and most work has emphasized behavioral features rather than motivations for NSSI.
This study aimed to clarify neural mechanisms of NSSI in clinical adolescents with MDD by combining rs‑fMRI metrics—specifically low‑frequency amplitude fluctuations (ALFF) and seed‑based functional connectivity (FC)—with a machine learning–style cross‑validation framework to assess robustness and generalizability of brain–behavior associations. The investigators also examined associations with NSSI motivations (intrapersonal vs interpersonal), addressing a gap in the literature about motivational correlates of neural alterations.
A total of 133 adolescent participants were recruited and grouped as follows: 45 adolescents with MDD and NSSI, 46 adolescents with MDD without NSSI, and 42 age‑ and sex‑matched healthy controls. All participants underwent resting‑state fMRI. The analytic pipeline assessed whole‑brain ALFF to identify regions showing group differences; those regions were then used as seeds for subsequent seed‑based FC analyses. Associations between imaging metrics (ALFF and FC) and measures of NSSI behaviors and motivations were evaluated. To test stability and generalizability, the authors applied a 4‑fold cross‑validation procedure using linear regression models to predict behavioral measures from imaging features.
ALFF quantifies low‑frequency amplitude fluctuations in the resting BOLD signal and is interpreted as a measure of spontaneous regional brain activity. Seed‑based FC measures temporal correlation between the time series of a seed region and other brain regions, indexing interregional functional coupling. In this study, regions with significant group differences in ALFF served as seeds for FC analysis; significant seed–target pairs were then examined for behavioral associations.
The primary group difference reported was an increase in ALFF in the left precuneus among adolescents with MDD who engaged in NSSI compared with the other groups. Using the left precuneus as a seed, the NSSI group also showed increased functional connectivity between the left precuneus and left cuneus relative to comparison groups.
The study found that higher left precuneus ALFF was significantly associated with measures of NSSI behaviors. Increased left precuneus–left cuneus FC was significantly associated with both NSSI behaviors and intrapersonal negative reinforcement motivations (motivations related to internal emotion regulation or relief). The authors emphasize the relevance of DMN and adjacent posterior medial parietal regions—notably the precuneus—in self‑referential processing and emotion regulation, linking these functions to the observed behavioral and motivational associations.
To evaluate the robustness and generalizability of the associations, the investigators used a 4‑fold cross‑validation approach with linear regression to predict NSSI behaviors and motivations from imaging metrics. Prediction results reported by the authors were:
These cross‑validated correlations were interpreted as evidence that left precuneus ALFF and precuneus–cuneus FC reliably relate to NSSI behaviors and intrapersonal motivations across data partitions.
The authors interpret increased left precuneus activity and enhanced precuneus–cuneus coupling as neural markers associated with NSSI in adolescents with MDD. Given the precuneus's role within the default mode network and in self‑referential processes and emotion regulation, these findings are framed as potentially mechanistic links to intrapersonal motives for self‑injury. The cross‑validated predictive results support the potential utility of these rs‑fMRI markers for identifying individuals at risk for NSSI and for informing precision interventions that target disrupted posterior DMN and visual network interactions.
The article notes that prior literature shows inconsistent findings and that differences across studies may reflect sample characteristics, NSSI scales, and imaging indices. Specific limitations of this cohort or analytic approach beyond those general considerations were not detailed in the provided source excerpt. The authors made demographic data, scale scores, and ALFF and FC values available via the Open Science Framework repository cited in the paper.
In adolescents with MDD, NSSI was associated with increased spontaneous activity in the left precuneus and increased left precuneus–left cuneus functional connectivity. Both imaging metrics were linked to NSSI behaviors, and precuneus–cuneus FC was additionally associated with intrapersonal negative reinforcement motivations. Cross‑validated prediction analyses supported the robustness of these associations, suggesting these resting‑state markers may serve as candidate neural indicators for NSSI prediction, diagnosis, and targets for tailored interventions. The study’s data are publicly available on OSF as reported by the authors.