OLINK proteomics identifies inflammatory protein signatures associated with vascular cognitive impairment in diabetes.

Introduction: Vascular cognitive impairment (VCI) is a syndrome of cognitive dysfunction attributable to vascular risk factors and cerebrovascular diseases, representing a major component of global dementia burden. Dysglycemia, encompassing diabetes mellitus and impaired glucose regulation, is increasingly recognized as a modifiable risk factor for cognitive decline, particularly VCI. Chronic low-grade inflammation mediates the association between metabolic dysfunction and neurodegeneration. The lack of validated diagnostic tools for early VCI detection in high-risk dysglycemic populations highlights the urgent need for robust diagnostic models based on molecular signatures. Methods: We analyzed serum samples from 94 participants, categorized into three groups: normal glucose with normal cognition (NG-NC), dysglycemia with normal cognition (Dys-NC), and dysglycemia with VCI (Dys-VCI). Using the Olink Target 96 Inflammation Panel, we quantified 92 inflammation-related proteins. Differentially expressed proteins (DEPs) were identified as potential biomarkers, followed by functional enrichment analysis to explore associated biological pathways. Logistic regression models, combined with ROC analysis, assessed the diagnostic utility of selected protein panels across groups. Results: Compared with NG-NC, the Dys-NC group exhibited upregulated pro-inflammatory mediators (CXCL1, CXCL5, OSM) and downregulated anti-inflammatory proteins (FGF-21, AXIN1). The Dys-VCI group showed significant increases in TNFB, IL-12B, TNF, and CSF-1. Key proteins, including AXIN1 and CX3CL1, displayed progressive changes across the metabolic-cognitive spectrum. Pathway analysis revealed enrichment in cytokine-cytokine receptor interaction, viral protein interaction with cytokine and cytokine receptor, TNF signaling, and chemokine signaling pathways. A four-protein panel (CCL3, CX3CL1, FGF-21, CXCL1) achieved an area under the curve (AUC) of 0.903 for distinguishing Dys-NC from NG-NC, while another panel (TNFB, IL10, IL-12B, CX3CL1) demonstrated an AUC of 0.799 for identifying Dys-VCI among Dys-NC. Discussion: Our study identified unique inflammatory protein profiles associated with different metabolic and cognitive states, providing insights into inflammatory mechanisms linking dysglycemia and VCI, and highlighting potential biomarker candidates for longitudinal validation.