Exploratory mediation analysis of associations between serum biomarkers of neuronal injury, astrocyte reactivity, neuroinflammation, and neurotrophic support and fatigue in Parkinson's disease

Introduction and aim. Fatigue is a debilitating non-motor symptom in Parkinson's disease (PD). This exploratory cross-sectional mediation analysis investigated whether specific serum biomarkers mediate the relationship between PD and fatigue severity. Material and methods. Ninety PD patients and 45 matched controls were enrolled in the study. Fatigue was assessed using the Fibro-Fatigue scale (FF). Serum levels of ten biomarkers (neuron‑specific enolase (NSE), glial fibrillary acidic protein (GFAP), brain-derived neurotrophic factor (BDNF), β-amyloid-42, α-synuclein, ubiquitin carboxy‑terminal hydrolase L1 (UCHL1), high‑mobility group box 1 (HMGB1), R‑spondin 1 (RSPO1), Dickkopf‑1 (DKK1), and sclerostin) were quantified by enzyme-linked immunosorbent assay. Mediation analysis (PROCESS Macro, Version 4.2, Model 4) tested whether these biomarkers mediated the relationship between PD status and FF score. Results. The overall model predicting FF scores was highly significant (F(26,108)=54.08, p<0.001, R²=0.929). Mediation analysis revealed significant indirect effects from PD status to fatigue via GFAP, HMGB1, BDNF, and NSE. Moderation analysis indicated that the biomarker fatigue relationship was significantly modified by PD status only for HMGB1 and BDNF. For GFAP and NSE, the interactions were non‑significant. Bootstrapped analyses confirmed significant indirect effects of PD on fatigue through elevated levels of NSE (effect=1.743), HMGB1 (effect=1.207), GFAP (effect=1.101), and BDNF (effect=0.921). Conclusion. PD-related fatigue is significantly mediated by neuroinflammation (HMGB1), astroglial activation (GFAP), neuronal injury (NSE), and a paradoxical BDNF response. Disease‑specific moderation was confirmed only for HMGB1 and BDNF.