Abstract
OBJECTIVES: Sepsis remains a major global health challenge, yet specific diagnostic biomarkers are still lacking. This study aims to investigate the causal relationship between B7 homologue 3 (B7-H3) and sepsis susceptibility, severity, and clinical outcomes using Mendelian randomization (MR) analysis, in order to evaluate its potential as a biomarker. METHODS: Genetic data related to sepsis (including overall sepsis, sepsis-related mortality with 28 days, severe sepsis, and severe sepsis with 28-day mortality) were extracted from genome-wide association study (GWAS) datasets. Single nucleotide polymorphisms (SNPs) associated with B7-H3 were selected as instrumental variables. The inverse-variance weighted (IVW) was used as the primary approach for causal effect estimation, while weighted median (WME) and MR-Egger regression served as supplementary methods. Additionally, a constrained maximum likelihood-model average (cML-MA) approach was employed to enhance the reliability of causal effect estimation. Cochran's Q test was conducted to assess heterogeneity, and MR-PRESSO along with the MR-Egger intercept method were used to detect horizontal pleiotropy. Sensitivity analyses were performed using the leave-one-out method. A reverse MR analysis was performed with sepsis as the exposure and B7-H3 as the outcome to exclude potential reverse causation. RESULTS: IVW analysis indicated a significant positive causal association between B7-H3 and sepsis susceptibility, severity, and clinical outcomes. A genetically predicted 1-standard deviation (SD) increase in B7-H3 levels was associated with a 10.4% increased risk of sepsis (OR=1.104, 95% CI 1.021 to 1.194, P=0.013), a 26.2% increased risk of sepsis-related 28-day mortality (OR=1.262, 95% CI 1.078 to 1.476, P=0.004), a 22.3% increased risk of severe sepsis (OR=1.223, 95% CI 1.023 to 1.463, P=0.027), and a 60.2% increased risk of severe sepsis with 28-day mortality (OR=1.602, 95% CI 1.119 to 2.294, P=0.010). The causal effect direction remained consistent across IVW, WME, MR-Egger, and cML-MA analyses, reinforcing the robustness and reliability of the results. Cochran's Q test showed no heterogeneity (P>0.05), while MR-PRESSO and MR-Egger intercept tests indicated no evidence of horizontal pleiotropy (both P>0.05). The leave-one-out analysis showed that removing individual SNPs did not significantly alter the causal estimates. Reverse MR analysis showed no causal association between sepsis and B7-H3. CONCLUSIONS: B7-H3 may serve as an important biomarker for sepsis, as it is closely associated with sepsis susceptibility, severity, and clinical outcomes.