Brain Inflammation and Network Changes Linked to Depression After Herpes Simplex Encephalitis

Herpes simplex encephalitis is one of the most feared infections of the human brain. Prompt antiviral therapy has dramatically improved survival, but surviving the virus is not the same as recovering from it. Many people who live through the illness are left with persistent memory deficits, fatigue, seizures, and mood disturbances. Among these lingering problems, depression has received comparatively little systematic attention from clinicians and researchers, who have traditionally focused follow-up care on survival, seizure control, and cognitive outcomes. A new cross-sectional observational study published in Frontiers in Neurology now offers a detailed biological picture of what may accompany depressive symptoms in these patients, linking elevated levels of an inflammatory molecule in the cerebrospinal fluid to measurable alterations in the brain’s functional network organization.

The research, conducted at two centers in Zhejiang Province, China, between June 2023 and June 2025, recruited 74 participants: 24 healthy controls and 50 adults with polymerase chain reaction-confirmed herpes simplex encephalitis. Within the patient group, trained clinicians used structured interviews based on DSM-5 criteria to identify 27 survivors without depression and 23 with depression. Depressive symptom severity was quantified with the 17-item Hamilton Depression Rating Scale, a continuous measure of current symptom burden rather than a standalone diagnostic tool. The researchers also assessed anxiety with the Hamilton Anxiety Rating Scale and cognition with the Mini-Mental State Examination and the Montreal Cognitive Assessment, allowing them to place depressive symptoms within each patient’s broader neuropsychiatric profile.

The central biological hypothesis concerned tumor necrosis factor-alpha, or TNF-alpha, a proinflammatory cytokine that participates in innate immune signaling among immune cells, endothelial cells, microglia, and astrocytes. Cerebrospinal fluid was obtained by lumbar puncture during the assessment visit, and TNF-alpha concentrations were measured in duplicate using a sandwich enzyme-linked immunosorbent assay. Before any analysis, the team prespecified TNF-alpha as the primary cytokine for the imaging, regression, and mediation analyses, while interleukin-6, IL-1beta, IL-10, and IL-8 were retained for secondary group comparisons and symptom correlations. This prespecification matters scientifically because it reduces the risk of cherry-picking among multiple measured biomarkers, a common pitfall in small observational studies.

The results showed a clear inflammatory gradient across the groups. Median CSF TNF-alpha concentrations were 14.9 picograms per milliliter in the depressed HSE group, 10.5 in the non-depressed HSE group, and 7.8 in healthy controls, a difference that reached statistical significance. In the pooled HSE cohort, CSF TNF-alpha correlated positively with Hamilton depression scores, with a Spearman rho of 0.345. CSF IL-6 also differed across groups and correlated with symptom severity, whereas IL-1beta, IL-10, and IL-8 showed no overall group differences. Notably, no measured cytokine correlated with cognitive scores, disease duration, or hospitalization duration, suggesting that the inflammatory signal was specifically tied to the affective phenotype rather than to general illness chronicity.

To capture the brain-side of the story, the researchers performed resting-state functional MRI on a 3.0-Tesla scanner and analyzed the data with a voxel-wise graph measure called degree centrality, or DC. Degree centrality quantifies how strongly each voxel’s spontaneous blood oxygen level-dependent signal correlates with the rest of the gray matter, without requiring a predefined seed region. The team computed Pearson correlations between every pair of voxels, retained connections above a threshold of r = 0.25, and counted each voxel’s positive connections with all other gray-matter voxels. Rigorous preprocessing included slice-timing correction, rigid-body realignment, nuisance regression with a 24-parameter motion model, and strict motion exclusion criteria, with mean framewise displacement and censored-volume counts included as covariates to reduce motion-related bias.

The primary whole-brain contrast comparing depressed with non-depressed HSE survivors, corrected for multiple comparisons using Gaussian random field methods, identified four clusters with significantly lower degree centrality in the depressed group: the right amygdala, the right insula, the left fusiform gyrus, and the left postcentral gyrus. These regions are functionally meaningful. The amygdala is central to emotional processing, and its resting-state connectivity is consistently altered in major depressive disorder. The insula supports salience detection and interoception, the integration of internal bodily signals with external information. The fusiform gyrus contributes to higher-order visual processing, while the postcentral gyrus handles somatosensory processing and bodily representation. In secondary regional comparisons, the differences between depressed and non-depressed survivors remained significant in the right amygdala and left postcentral gyrus but not in the fusiform or insula, which the authors therefore treated as exploratory candidate regions rather than depression-specific abnormalities.

The integration of the inflammatory and imaging data was the study’s most technically ambitious component. CSF TNF-alpha was inversely associated with degree centrality in the right amygdala, right insula, and left postcentral gyrus, with partial correlation coefficients of -0.355, -0.411, and -0.299 respectively. Regional degree centrality values, in turn, correlated negatively with depressive symptom severity across all four regions after adjustment for TNF-alpha. Because anxiety scores differed between the patient groups, the team added the Hamilton anxiety score to their models as a sensitivity analysis. The associations remained directionally consistent but were attenuated, indicating that anxiety symptoms partially overlap with, but do not fully explain, the observed relationships. The positive correlation between CSF TNF-alpha and depression scores also remained significant after adjustment for anxiety.

Hierarchical regression models built on these findings incrementally. A base model including age, sex, education, cognitive score, and disease duration explained 19.4 percent of the variance in depression scores. Adding log-transformed CSF TNF-alpha increased the explained variance to 30.2 percent, and further adding right amygdala and right insula degree centrality raised it to 42.1 percent. Within the final model, right amygdala centrality remained independently associated with symptom severity, but the TNF-alpha coefficient was attenuated and no longer significant, suggesting overlapping statistical information between the inflammatory marker and the imaging measure. Exploratory single-region mediation models, using 5,000 bias-corrected bootstrap resamples, yielded positive standardized indirect-effect estimates of 0.149 for the right amygdala, 0.081 for the left fusiform gyrus, 0.163 for the right insula, and 0.091 for the left postcentral gyrus. The authors emphasized that these estimates describe statistical associations and do not establish causal mediation.

The study’s limitations are substantial and honestly acknowledged. The cross-sectional design precludes any conclusion about temporal sequence, so it remains unknown whether inflammation drives network changes, whether network changes produce inflammation, or whether both reflect a third process. The sample of 74 participants is modest, limiting precision in multivariable and mediation models. The depressed group included heterogeneous DSM-5 categories, with only three cases of primary major depressive disorder, three of depressive disorder due to another medical condition, four of adjustment disorder, and thirteen other or unspecified depressive disorders. Longer hospitalization and poorer cognitive scores in the depressed group suggest greater neurological burden, raising the possibility of residual confounding by lesion location, acute disease severity, medication exposure, sleep disturbance, and shared symptom variance. Because all inflammatory biomarkers were measured in cerebrospinal fluid, the findings pertain specifically to the central compartment; no peripheral blood measurements were included, and prior systematic reviews indicate that blood and CSF cytokine relationships are dynamic and context-dependent.

Despite these caveats, the study points toward a multimodal framework for understanding post-encephalitic depression, in which central inflammatory signaling and altered functional network organization jointly accompany depressive symptoms. The authors suggest that neuropsychiatric follow-up after herpes simplex encephalitis should extend beyond seizure surveillance and cognitive testing to include systematic assessment of depression and anxiety, particularly after complicated acute courses. However, neither CSF TNF-alpha nor degree centrality is a validated diagnostic or prognostic biomarker, and clinical translation would require reproducibility across scanners, independent validation, and evidence that testing improves management or outcomes. Longitudinal studies with repeated CSF and peripheral sampling, prespecified or externally defined imaging regions, and dedicated trials testing whether modifying inflammatory pathways improves neuropsychiatric recovery are the necessary next steps. For now, the data do not support TNF-targeted treatment, but they do provide a hypothesis-generating biological scaffold for a population of survivors whose affective recovery has long been overshadowed by the fight to save their lives.

Subject of Research: Cerebrospinal fluid TNF-α, resting-state brain network centrality, and depressive symptoms in survivors of herpes simplex encephalitis

Article Title: Cerebrospinal fluid TNF-α, regional degree centrality, and depressive symptoms after herpes simplex encephalitis: a cross-sectional observational study

Article References: Ma, J., Li, Y., Zhang, Y., Yu, Y., Bai, Y., Chen, Y., Liao, Y., Zheng, C., Yan, J., Feng, Q., & Chen, C. (2026). Cerebrospinal fluid TNF-α, regional degree centrality, and depressive symptoms after herpes simplex encephalitis: a cross-sectional observational study. Frontiers in Neurology, 17, Article 1942369. https://doi.org/10.3389/fneur.2026.1942369

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1942369

Keywords: herpes simplex encephalitis, TNF-alpha, cerebrospinal fluid, degree centrality, resting-state fMRI, post-encephalitic depression, Cerebrospinal, fluid, TNF-, regional, degree, centrality