Immune-programmed cell death crosstalk implicates plasma sTNFR1 as a candidate biomarker for sepsis-associated encephalopathy

Sepsis-associated encephalopathy (SAE) is a frequent neurological manifestation of sepsis characterized by acute and diffuse brain dysfunction. Despite increasing evidence of molecular alterations in the septic brain, clinically applicable circulating biomarkers for SAE remain limited. Immune dysregulation and programmed cell death (PCD) pathways have been implicated in SAE pathogenesis, but their translational relevance for peripheral biomarker development remains unclear. This study aimed to identify immune-PCD crosstalk-related biomarkers for SAE through an integrated transcriptomic and clinical evaluation strategy.

Human brain transcriptomic data from GSE135838 were analyzed using differential expression analysis and gene set enrichment analysis. Immune-related pathways, PCD signatures, and extracellular protein annotations were integrated to prioritize candidate biomarkers with peripheral detection potential. An exploratory cross-species comparison was performed using mouse SAE microglial RNA-sequencing data (GSE198862). Plasma soluble TNFR1 (sTNFR1) was subsequently evaluated in a prospective cohort of 60 septic patients (26 SAE and 34 non-SAE). SAE was defined as acute diffuse cerebral dysfunction during sepsis, identified by reduced GCS and/or a positive CAM-ICU assessment after exclusion of alternative causes, including persistent deep sedation that precluded reliable neurological assessment. Plasma samples for sTNFR1 measurement were collected within 24 h after enrollment, while SAE status was determined during the baseline neurological assessment after ICU admission. Logistic regression, ROC analysis, and bootstrap validation assessed the incremental value of sTNFR1, with 28-day mortality examined exploratorily.

Transcriptomic analysis revealed activation of immune inflammatory responses and multiple PCD-related pathways in SAE brain tissue. Integration of immune, PCD, and differential expression signatures identified 15 immune-PCD crosstalk genes. Extracellular prioritization further identified FAS, IL6, and TNFRSF1A as candidate biomarkers. The immune-PCD signature showed substantial directional concordance in mouse SAE microglial transcriptomes, including concordant increases in Fas and Tnfrsf1a. In the clinical cohort, plasma sTNFR1 levels were significantly elevated in SAE patients compared with non-SAE patients (p < 0.001). After adjustment for age and SOFA score, sTNFR1 remained independently associated with SAE. Adding sTNFR1 increased the observed AUC from 0.791 to 0.891; however, the difference did not reach statistical significance by the paired DeLong test (p = 0.086). Continuous NRI nevertheless indicated improved reclassification (NRI = 0.950, 95% CI 0.411-1.536; p = 0.001). Exploratory analyses also associated SAE and higher sTNFR1 levels with increased 28-day mortality.

This study integrates sepsis-associated brain transcriptomic discovery with prospective peripheral clinical evaluation and identifies circulating sTNFR1 as a candidate biomarker associated with SAE. sTNFR1 may provide additional information beyond conventional severity assessment for SAE identification and merits further validation.