Multivariate analyses of immune markers reveal increases in plasma EN-RAGE in first-episode psychosis patients

The study investigates whether a broad panel of immune-related proteins in plasma differs between drug-naive, first-episode psychosis (FEP) patients and healthy controls, addressing the hypothesis that acute psychosis is associated with dysregulated immune signaling. Prior work links infections, immune genetics, and inflammatory markers to psychotic disorders, but findings vary due to confounding factors and the focus on limited protein sets. By profiling 92 inflammatory proteins with multiplex PEA technology and applying multivariate analyses, the authors aim to identify discriminative cytokine patterns, particularly assessing cytokines like IL-6 and the neutrophil-derived mediator EN-RAGE, to inform biomarkers and mechanisms relevant to early psychosis.

Existing literature indicates immune dysregulation in psychosis, including associations with prenatal infections, immune-related genetic loci, and elevated inflammatory markers in blood and CSF. Meta-analyses have reported increases in cytokines such as IL-6 and TNF-α during acute phases of schizophrenia spectrum disorders. However, variability across studies arises from differences in sample characteristics, confounders (age, BMI, smoking), and limited analyte panels. Recent evidence suggests neutrophil involvement, including increased neutrophil extracellular traps (NETs) in early schizophrenia, and elevated EN-RAGE in other inflammatory and cardiovascular conditions. Studies using different sample types (serum vs plasma) and platforms (biochip arrays vs Olink PEA) also report differing detectability for cytokines like IL-2 and IL-4.

Design and cohorts: Cross-sectional comparison of 60 antipsychotic-naive FEP patients (ICD-10 F20–F29; onset within past 3 years; age 18–45) recruited at the Psychiatric Clinic of Tartu University Hospital, Estonia (2009–2013), and 50 age-matched healthy controls (HC) recruited via advertisement. Exclusions: diabetes, neurological or immune-related disorders; for HC, any psychopathology or family history of psychosis. Substance use history was not an exclusion criterion. Clinical assessments: Demographics and lifestyle (age, sex, BMI, waist circumference, smoking, alcohol, cannabis). FEP severity: BPRS, GAF, CGI. Sample collection and processing: Fasting blood (0900–1100 h) collected in heparin tubes, centrifuged (15 min, 2000 rpm, 4 °C). Plasma aliquoted, stored at −20 °C up to 2 weeks, then at −80 °C until analysis. Proteomic assays: Olink Proteomics Inflammation panel (92 proteins) using proximity extension assay (PEA). Two assay runs (Run1: FEP n=30, HC n=30; Run2: FEP n=30, HC n=20). Eight bridging samples included in Run2 to enable merging. NPX units (log2 scale) reported. Bridge normalization used OlinkAnalyze (R), with Run1 as reference and median pairwise differences applied to Run2. Intra-assay CV across bridging samples: 0.82–30.77%. Proteins with >40% values below LOD were excluded (20 proteins), leaving 72 proteins for analysis. Values below LOD were kept as provided NPX. Statistics: Demographics compared with Shapiro–Wilk normality tests, followed by Wilcoxon rank-sum and chi-square tests; for sparse cells, chi-square p-values simulated with 2000 replicates. Significance threshold p≤0.05. Multivariate analyses: PCA (all proteins, all participants) to explore data; two strong outliers identified but retained. OPLS-DA (SIMCA v17) with mean-centering and unit variance scaling modeled group discrimination (FEP vs HC). Model significance via CV-ANOVA; performance via R² (fit) and Q² (prediction). Discriminative proteins identified by VIPpred ≥1.0 and p(corr) ≥0.3. Protein network analysis via STRING v11.5 (Homo sapiens; only query proteins; medium confidence 0.40) to assess protein-protein associations.

- Cohort characteristics: FEP and HC did not differ significantly in age, sex, BMI, or waist circumference. FEP had higher cannabis use (χ²=12.04, p=.005) and different smoking behavior (χ²=6.19, p=.04). - Multivariate model: OPLS-DA yielded a significant model with one predictive and two orthogonal components (F(6,103)=15.16, p=2.25e-12), R²=0.76, Q²=0.47, showing clear separation between FEP and HC. - Discriminative proteins (11/72 met VIPpred ≥1.0 and p(corr) ≥0.3): • Increased in FEP: EN-RAGE/S100A12 (VIPpred 2.49; FEP mean 3.89 NPX vs HC 2.93), OSM (2.21; 3.33 vs 2.31), TGF-α (1.91; 2.94 vs 2.70), HGF (1.80; 7.44 vs 7.21), IL-6 (1.72; 2.50 vs 1.92), CCL23 (1.64; 9.97 vs 9.66), 4E-BP1 (1.59; 6.58 vs 5.95), FGF-21 (1.52; 5.07 vs 4.16), TNFSF14 (1.40; 3.74 vs 3.49). • Decreased in FEP: Flt3L (2.63; 8.14 vs 8.63), TNFβ/LTA (1.91; 4.11 vs 4.61). - Network analysis: STRING revealed a significantly enriched interaction network centered on IL-6 (PPI enrichment p=1.94e-10; average local clustering coefficient 0.78). EN-RAGE and OSM are linked to IL-6 induction; CCL23 relates to neutrophil/monocyte chemotaxis; TNFβ and TNFSF14 reflect stress/inflammatory signaling. - Overall, 9 proteins were elevated and 2 were reduced in FEP plasma, with EN-RAGE among the strongest discriminators.

Findings support a broad immune activation profile in antipsychotic-naive FEP, featuring elevated IL-6 and other pro-inflammatory mediators alongside reduced Flt3L and TNFβ. The prominence of EN-RAGE (S100A12), a neutrophil-derived protein that signals via RAGE to upregulate adhesion molecules and cytokines, suggests neutrophil-driven mechanisms during acute psychosis. This aligns with reports of increased NETs in early schizophrenia and prior observations of EN-RAGE elevation in inflammatory and cardiovascular conditions. Elevated OSM and IL-6 fit an acute-phase response framework, while increased CCL23, HGF, FGF-21, and TNFSF14 indicate broader inflammatory and metabolic involvement. Differences from prior cytokine studies (e.g., detectability of IL-2/IL-4) likely reflect platform and sample-type differences (plasma Olink vs serum biochip). The enriched IL-6-centered network underscores interconnected immune signaling in FEP. The data suggest that EN-RAGE and neutrophil-related pathways may orchestrate early immune responses impacting the vasculature and possibly the brain during acute psychosis.

The study identifies a distinct plasma immune signature in drug-naive first-episode psychosis, characterized by elevations in EN-RAGE, IL-6, OSM, and other pro-inflammatory markers, and reductions in Flt3L and TNFβ. EN-RAGE emerges as a strong discriminator and a plausible blood-based biomarker candidate for acute psychosis, implicating neutrophil/RAGE signaling in pathophysiology. Future work should replicate these findings in independent, larger, and ethnically diverse cohorts, incorporate longitudinal designs to assess biomarker dynamics and prognostic value, and explore mechanistic studies of RAGE-ligand interactions and vascular effects in neuropsychiatric disorders.

- Modest sample sizes for both FEP (n=60) and HC (n=50). - Potential residual confounding: while age, sex, and BMI were comparable, group differences in cannabis use and smoking, as well as unmeasured clinical or lifestyle factors, may influence immune markers. - Platform-related limitations: 20 proteins excluded due to >40% values below LOD (e.g., IL-2, IL-4), potentially biasing the profiled network. - Cross-sectional design precludes causal inferences and temporal dynamics of biomarkers. - Two assay runs required bridging normalization and showed variable intra-assay CVs, which may introduce technical variability.