The Value Of FibroScan And Serological Indicators To Predict Esophageal Varices In Cirrhosis

Objective:To investigate the construction of a new model based on liver stiffness values(LSM)and serological indicators detected by FibroScan,and to evaluate their diagnostic value for esophageal varices in cirrhosis compared with the non-invasive fibrosis models APRI,FIB-4 and King’s Score.Methods:The medical records of 193 patients who attended The Second Affiliated Hospital of Nanchang University from 06/2020 to 06/2022 and were clinically diagnosed with cirrhosis and met the inclusion and exclusion criteria were retrospectively screened,including:general data(e.g.age,gender,presence of ascites,presence of hepatic encephalopathy,etc.),laboratory data(e.g.PLT,ALT,AST,TBIL,ALB,APL,GGT,PT,INR,etc.)as well as FibroScan and gastroscopy findings.Based on the endoscopic findings,patients were divided into groups with or without esophageal varices(EV)and non-high-risk versus high-risk esophageal varices(HREV).Univariate and binary logistic regression analyses were performed to find non-invasive indicators that could independently predict EV and HREV,and to construct new models for predicting EV and HREV.The ROC curves of LSM,new models and noninvasive fibrosis models(APRI,FIB-4,King’s Score)were plotted separately using gastroscopy as the gold standard,and statistical indicators such as AUC,SE,SP,PPV,NPV and AC were calculated to evaluate the predictive efficacy of each model for EV and HREVResults:1.LSM,PLT and TBIL were identified as independent predictors of EV in the group with and without EV.A new model was constructed based on these three indicators.a The new model a had the highest predictive efficacy with an AUC of 0.798 and an optimal cut-off value of 0.565.The model SP and SE were 85.70%、65.30%,and 81.57%,71.81%and 75.66%for PPV,NPV and AC,respectively.This was followed by the LSM,which also had better predictive value in predicting EV alone,with an ACU of 0.715 and a best cut-off value of 21.85.This model had slightly lower SE(72.60%)and SP(64.30%)than the new model,and 66.35%,70.77%and 68.39%for PPV,NPV and AC respectively.the FIB-4,APRI,King’s indices had lower predictive efficacy.2.In the HREV and non-HREV groups,LSM,PLT,ALB,gender and presence of ascites were identified as independent predictors of HREV.A new model was constructed on this basic The new model b had an AUC of 0.907 and an optimal cut-off value of 0.377,which is higher for SP(87.90%)and a lower for SE(77.00%),and its PPV,NPV and AC were 72.73%,90.12%and 84.68%,respectively.LSM also showed superior predictive value in predicting HREV alone,with FIB-4,APRI and King’s index continued to have poor predictive performance.Conclusions:1.PLT,TBIL and LSM were independent predictors of the presence of EV in cirrhosis in the group with and without EV;the new model b had a higher predictive value for EV and was higher than the LSM and non-invasive fibrosis models.2.PLT,ALB,LSM and gender,presence or absence of ascites were independent predictors of HREV in the HREV and non-HREV groups;the new model b also had higher predictive value for HREV than the LSM and non-invasive fibrosis models.