Circulating microRNAs and Plasma Gelsolin as Biomarkers of Sepsis: Molecular Insights and Prospects for Precision Medicine

Sepsis is a major medical emergency, characterized by a dysfunctional immune response to infection, which often progresses to multiple organ failure and death. Early diagnosis and prognostic evaluation present significant challenges due to limitations in the specificity and sensitivity of traditional biomarkers. This narrative review summarizes recent evidence on the potential of circulating microRNAs (<i>miRNAs</i>) such as <i>miR-150</i>, <i>miR-146a</i>, <i>miR-223</i>, <i>miR-155</i>, <i>miR-122</i>, and <i>miR-4772-5p</i> and plasma gelsolin (pGSN) as diagnostic and prognostic markers in sepsis. We discuss mechanisms involved and their potential for integration with artificial intelligence (AI) in personalized medicine. PubMed, Embase, and Web of Science databases were searched for relevant literature. Original research, systematic reviews, and meta-analyses focused on the diagnostic or prognostic value of circulating <i>miRNAs</i> or pGSN in sepsis were included; opinion papers and case reports were excluded. Altered expression of certain circulating microRNAs correlates with disease severity and mortality. Among circulating microRNAs (<i>miRNAs</i>), miR-122 and miR-150 have become the most consistently validated biomarkers in clinical studies, associated with sepsis severity and death rates. Additionally, other <i>miRNAs</i> such as <i>miR-146a</i>, <i>miR-155</i>, and <i>miR-223</i> play roles in modulating immune and endothelial responses, highlighting the complex regulation of sepsis pathophysiology. Low pGSN concentrations at admission are associated with severe sepsis and acute respiratory distress syndrome, and serve as an independent predictor of mortality. Preclinical studies suggest that supplementation with exogenous pGSN could increase survival. AI algorithms show promising results for early sepsis detection and optimization of therapeutic decisions. However, combining circulating <i>miRNAs</i> and plasma gelsolin (pGSN) into AI-based models is still an exploratory idea that needs prospective validation, assay standardization, and multicenter studies before it can be used clinically.