Blood Proteins Reveal an Inflammatory Signature Behind Self-Harm in Depressed Teens

Non-suicidal self-injury, the deliberate harming of one’s own body without suicidal intent, is one of the most alarming behaviors clinicians encounter in adolescents with depression. Cutting, burning, and hitting oneself often begin in the teenage years and can persist into adulthood, yet the biological forces that push some depressed adolescents toward self-harm while sparing others have remained stubbornly opaque. A new study published in BMC Psychiatry by Hui Xu of Wuhu Fourth People’s Hospital and colleagues now offers a molecular foothold, using a high-throughput blood protein survey to show that depressed adolescents who self-injure carry a distinct immunoinflammatory fingerprint in their plasma.

The research team recruited three carefully matched groups of twenty participants each: adolescents diagnosed with depression but no history of self-injury, adolescents with depression who engaged in non-suicidal self-injury, and mentally healthy controls. Each participant provided a blood sample, and the researchers measured the levels of dozens of circulating proteins using the Olink proximity extension assay, a technology that pairs antibodies with DNA tags so that minute quantities of proteins can be quantified with high specificity from small plasma volumes. This approach allowed the team to scan the peripheral immune landscape without relying on invasive procedures such as lumbar puncture.

The first comparison established a baseline: depression alone, relative to healthy controls, was associated with eleven differentially expressed proteins. Among the most striking changes were significant elevations of IL-17A, IL-10, and CCL19, alongside reductions in IL-13 and CCL11. These proteins span the classical cytokine axis, including pro-inflammatory interleukins, regulatory anti-inflammatory signals, and chemokines that orchestrate immune cell trafficking. When the researchers ran pathway enrichment analyses on this set, the proteins clustered strongly in inflammatory response, cytokine-mediated signaling, immune response, and cytokine activity, reinforcing the long-standing hypothesis that adolescent depression involves measurable peripheral inflammation rather than being purely a disorder of mood and cognition.

The more revealing comparison, however, came from contrasting the two depressed groups against each other. Adolescents with depression and comorbid self-injury showed seven proteins that differed significantly from their depressed but non-self-injuring peers. Five were elevated: matrix metalloproteinase-1, oncostatin M, hepatocyte growth factor, vascular endothelial growth factor A, and caspase-8. Two were reduced: cystatin D and CD6. This pattern is biologically coherent. MMP-1 is an enzyme that degrades extracellular matrix components and is often upregulated during tissue remodeling under inflammatory stress. Oncostatin M is an inflammatory cytokine in the interleukin-6 family with well-documented effects on neuroinflammation and the blood-brain barrier. HGF and VEGFA are growth factors involved in vascular repair and permeability, while caspase-8 sits at the gateway of apoptotic and inflammatory cell-death pathways. Cystatin D and CD6, by contrast, are linked to protease inhibition and immune regulation respectively, and their reduction hints at a loss of counterbalancing control.

Pathway analysis of these seven proteins pointed to specific signaling circuits. Using the Kyoto Encyclopedia of Genes and Genomes framework, the team found significant enrichment in TNF signaling, IL-17 signaling, and JAK-STAT signaling. All three pathways are central conduits through which peripheral immune activation can communicate with the central nervous system. TNF and IL-17 signaling are hallmarks of chronic inflammation and have been implicated in stress-induced changes in brain circuits governing emotion and impulse control. The JAK-STAT pathway translates cytokine signals arriving at cell membranes into changes in gene expression, meaning that sustained elevation of oncostatin M or related cytokines could, in principle, reprogram stress-responsive neurons and glial cells. Because the blood-brain barrier becomes more permeable under inflammatory conditions, and because VEGFA and HGF themselves regulate vascular permeability, the authors suggest these proteins may represent a peripheral window into inflammatory crosstalk affecting the brain.

Perhaps the most clinically tantalizing result emerged when the researchers tested whether the protein signatures could distinguish self-injuring adolescents from depressed adolescents without self-injury. A panel combining all seven differentially expressed proteins achieved an area under the receiver operating characteristic curve of 0.825, a level of discrimination that, if replicated, could support the development of a blood-based screening tool. A leaner four-protein panel consisting of CD6, cystatin D, VEGFA, and HGF performed nearly as well, with an area under the curve of 0.800. In a field where risk assessment currently depends almost entirely on clinical interviews and self-report questionnaires, a measurable biomarker panel would represent a meaningful shift toward objective risk stratification.

The team went further by connecting protein levels to the fine-grained structure of self-injurious behavior. Using validated NSSI scales that break the behavior into specific actions, motivations, and frequencies, they computed correlations between each protein and individual scale items. MMP-1, oncostatin M, and VEGFA levels correlated positively with several behavioral indices, including items measuring the frequency and form of self-injury. MMP-1 in particular tracked motivational dimensions of the behavior, correlating with items assessing why adolescents engage in self-harm, while oncostatin M and VEGFA tracked the behavioral items themselves. Linear regression analysis confirmed that these associations held after accounting for other variables, suggesting the protein signals are not merely proxies for overall depression severity but are specifically tied to the self-injury phenotype.

Statistical rigor was a feature of the analysis throughout. The researchers applied Benjamini-Hochberg correction to control the false discovery rate across the many protein comparisons, and they used both Pearson and Mantel tests to verify the correlations between protein levels and behavioral scores. The study also received ethics approval from two independent hospital committees and was conducted under the Declaration of Helsinki, with written informed consent obtained from every participant or their legal guardian. These methodological safeguards matter because small-sample proteomic studies are vulnerable to spurious findings, and the field of psychiatric biomarkers has been repeatedly criticized for underpowered and unreplicated claims.

That caveat leads directly to the study’s principal limitation, which the authors state plainly: the findings require independent validation before any clinical use. With twenty participants per group, the results should be regarded as hypothesis-generating rather than definitive. Self-injury in adolescents is influenced by a web of psychological, social, and familial factors, and no single protein panel will capture that complexity. Moreover, the cross-sectional design cannot determine whether the observed protein changes precede the onset of self-injury, result from it, or reflect a shared upstream cause such as chronic stress. Longitudinal studies following at-risk adolescents over time, and replication in larger and more diverse cohorts, will be essential next steps.

Even with those qualifications, the study adds a significant piece to a rapidly growing body of work linking immune dysregulation to self-directed harm. It moves the conversation beyond the question of whether inflammation matters in adolescent depression toward the more precise question of which inflammatory pathways distinguish the subgroup of depressed adolescents at highest risk of hurting themselves. If the seven-protein signature, or the leaner four-protein version, survives validation, clinicians could one day pair a simple blood draw with existing psychological assessments to identify vulnerable patients earlier, intervene sooner, and potentially prevent the escalation of self-injurious behavior during one of the most treatable windows of psychiatric development. For now, the study stands as a compelling demonstration that the biology of self-harm leaves traces in the blood that modern proteomics can finally read.

Subject of Research: Immuno-inflammatory plasma protein changes associated with non-suicidal self-injury in adolescents with depression

Article Title: Olink proteomics reveals immuno-inflammatory profile changes and potential regulatory mechanisms of non-suicidal self-injury behavior associated with depression in adolescent

Article References: Xu, H., Liang, X., Xia, Y., Fang, X.-L., Xu, C.-H., Wu, P.-P., Wang, D.-C., Song, C.-F., & Yang, F. (2026). Olink proteomics reveals immuno-inflammatory profile changes and potential regulatory mechanisms of non-suicidal self-injury behavior associated with depression in adolescent. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08624-7

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08624-7

Keywords: adolescent depression, non-suicidal self-injury, Olink proteomics, biomarkers, TNF signaling, IL-17 signaling, JAK-STAT pathway, MMP-1, VEGFA, oncostatin M, inflammation, BMC Psychiatry