The predictive value of a polygene expression risk model for the prognosis of patients with sepsis

WANG Hai-man; WANG Ya-miao; LI Xiao-hong

doi:10.13820/j.cnki.gdyx.20232749

Abstract：Objective To explore the construction of a prognostic polygene expression risk model in patients with sepsis and to evaluate its predictive performance.Methods A total of 140 patients with sepsis from January 2019 to June 2022 were included.Patients were divided into the death group and survival group based on their 28-day mortality after hospitalization.Clinical data and laboratory indicators were compared between the two groups.Reverse transcription-quantitative polymerase chain reaction(RT-qPCR)was used to determine the gene expression levels of peripheral blood microRNA103(miR103),Toll-like Receptor 2(TLR2),Signal Transducer and Activator of Transcription 3(STAT3),and autophagy-related protein Beclin1 in all patients.Multivariate COX regression analysis was employed to identify rele-vant factors for the 28-day mortality risk in sepsis patients.The Ggrisk software was used for risk model construction,and the ROC curve was drawn to assess the model's performance.Results The death group showed statistically signifi-cant differences in SOFA score,age,and APACHE Ⅱ score compared to the survival group(P＜0.05).The relative ex-pression levels of miR103,TLR2,STAT3,and Beclin1 genes in the death group were significantly different from those in the survival group(P＜0.05).Multivariate COX regression analysis revealed that miR103,TLR2,STAT3,and Beclin1 genes were independent factors for the 28-day mortality risk in sepsis patients.ROC curve analysis demonstrated that the predictive risk model constructed from these genes had excellent predictive value for early warning of 28-day mortality(AUC=0.964,P=0.000),with a sensitivity of 97.50％and specificity of 91.00％.Comparative analysis showed that the predictive efficacy of the risk model constructed from these genes was significantly higher than that of the APACHE Ⅱscore and SOFA score(both P＜0.001).Conclusion The expression levels of miR103,TLR2,STAT3,and Beclin1 genes are correlated with the 28-day mortality risk in sepsis patients.The constructed multigene expression risk model can effectively predict the 28-day mortality risk in sepsis,demonstrating superior performance compared to the APACHEⅡ score and SOFA score.