Construction And Evaluation Of A Cross-scale Prediction Model For Liver Failure After Hepatectomy In Patients With Hepatocellular Carcinoma

Background: Hepatectomy is an effective treatment for patients with hepatocellular carcinoma(HCC),but posthepatectomy liver failure(PHLF)may result in increased postoperative mortality.Therefore,it is important to assess the safety of surgery and to identify patients who are likely to develop PHLF before surgery.Various functional tests and scoring systems are available clinically,such as Child-Pugh score,model for end-stage liver disease score(MELD)score,Albumin-Bilirubin score(ALBI)score,Fibrosis-4 score(FIB-4)score,aspartate aminotransferase(AST)and platelet(PLT)Ratio Index(APRI)Index(APRI)scores have attempted to address such challenges,but both have some degree of drawbacks.Therefore,the ideal model for predicting the safety of hepatectomy should consider different scales of influencing factors simultaneously to establish and improve the prediction for PHLF and provide clinical guidance and basis for diagnosis and treatment.Objective: To investigate the risk factors for liver failure after laparoscopic hepatectomy for hepatocellular carcinoma and to construct a cross-scale prediction model.Methods: Retrospective collection Retrospective collection of clinical case data of 164 patients with surgically resected HCC admitted to Department of Hepatobiliary Surgery,Provincial Hospital of Anhui Medical University from March 2019 to March 2022.Among them,132 cases were male and 32 cases were female.The patients were divided into 31 cases in the post-hepatectomy liver failure group(PHLF group)and 133 cases in the non-hepatectomy liver failure group(non-PHLF group)according to whether PHLF outcome occurred after surgery.The risk factors for PHLF risk were analyzed by using the least absolute shrinkage and selection operator(LASSO)regression,and the prediction model was constructed by incorporating multifactorial logistic regression to construct the nomogram prediction model.The accuracy of the nomogram was verified using the receiver operating characteristic curve(ROC),the calibration ability of the prediction model was verified using the calibration curve analysis,and the clinical usefulness of the model was verified using the decision curve analysis.The area under the ROC curve(AUC)of the nomogram,Child-Pugh score,MELD score,ALBI score,FIB-4 score,and APRI score were compared to determine whether the nomogram was better than the existing clinical scoring methods in predicting liver failure after hepatectomy.Result:(1)For the preoperative liver failure group and non-liver failure group,the analysis of preoperative general data showed that: there were statistically significant differences between the two groups in TBIL,ALT,AST,PAB,ALB,PT,INR,portal hypertension,PS score,Child-pugh classification,APRI score,MELD score and FIB-4 score(P < 0.05).(2)Four predictors(TBIL,PT,portal hypertension,MELD score)were screened using LASSO regression,included in logistic regression and constructed as line graphs.The column line graph model was validated with column line graph C-index=0.967,and ROC curve analysis was performed for the prediction probability of the model,with AUC of 0.967(95% CI:0.944-0.991),sensitivity 93.55%,and specificity 89.47%.The calibration curve of the model fitted well with the ideal curve and could better respond to the risk factors of PHLF.The decision curve suggested the clinical applicability of the model.Hosmer Lemeshow’s test level P= 0.942(P > 0.05)indicates that the scoring model works well.(3)Comparison of the predictive efficacy of the nomogram prediction model and the associated composite liver function scores.The ROC curves of the nomogram prediction model,Child-Pugh score,MELD score,ALBI score,FIB-4score,and APRI score were plotted,and the AUC values were calculated,which were 0.967 for the prediction model The AUC of this prediction model was 0.967,the AUC of MELD score was 0.914,the AUC of Child-pugh score was 0.670,the AUC of FIB-4 score was 0.644,the AUC of FIB-4 score was 0.598,the AUC of ALBI score was 0.819,and the AUC of PS score was 0.666,indicating that this nomogram prediction model was better than other scoring systems for the subjects in our center.The prediction effect of this nomogram model is better than other scoring systems for the study subjects.Conclusion: For patients with hepatocellular carcinoma,TBIL,PT,portal hypertension,and MELD score are independent risk factors affecting the development of liver failure after hepatectomy in patients with HCC.The cross-scale nomogram based on the above independent risk factors has good accuracy and predictive ability,which can help clinical workers to preoperatively determine patients at high risk of PHLF.