Abstract
INTRODUCTION: Dementia encompasses distinct subtypes characterized by different underlying mechanisms-most notably, Alzheimer's disease (AD), driven by neurodegeneration, and vascular dementia (VaD), stemming from cerebrovascular pathology. Growing evidence emphasizes the critical role of neuroinflammatory processes in both conditions, highlighting inflammatory biomarkers as potential tools for differential diagnosis. This study assesses the diagnostic accuracy of inflammatory biomarkers in distinguishing AD from VaD. METHOD: A systematic search of PubMed, EMBASE, Web of Science, and Cochrane Library databases was conducted in November 2025 to identify eligible studies. Standardized mean differences (SMDs) with 95% confidence intervals (CIs) were calculated to compare inflammatory marker levels between AD and VaD groups. Random-effects models were applied for all meta-analyses. RESULTS: Fifteen observational studies involving 1728 participants were included. Pooled analyses showed no significant differences in IL-6 (SMD: -0.13; 95% CI: -0.44 to 0.19; p = 0.433), TNF-α (SMD: -0.22; 95% CI: -0.71 to 0.28; p = 0.388), or CRP (SMD: 0.73; 95% CI: -0.17 to 1.62; p = 0.111) between AD and VaD overall. However, subgroup analyses indicated context-dependent variations: IL-6 and TNF-α levels were lower in patients with AD in studies conducted in Eastern regions, with larger sample sizes (≥100 participants), older populations (≥70 years), or higher methodological quality. CRP showed similar patterns in larger or higher-quality studies. Importantly, IL-1β levels were significantly higher in patients with AD compared to patients with VaD (SMD: 0.48; 95% CI: 0.18 to 0.79; p = 0.002). CONCLUSIONS: Exploratory analyses suggest IL-1β as a promising candidate for differentiating AD from VaD, warranting validation in larger, prospective studies. The preliminary, context-dependent signals for IL-1β, IL-6, and TNF-α indicate that inflammatory pathways differ between these dementias but are substantially influenced by methodological and population factors. SYSTEMATIC REVIEW REGISTRATION: INPLASY platform (number: INPLASY202570068).