Construction and Validation of a Risk Warning Model for Massive Bleeding during Liver Resection in Patients with Hepatocellular Carcinoma

ZHONG Yajun; CHE Hanyang; WU Manqian; LI Hong; CHEN Hui

doi:10.3969/j.issn.1671-315x.2025.11.004

Abstract：Objective To explore the risk factors of massive bleeding during hepatectomy in patients with hepatocellular carcinoma,and to construct and verify a risk early warning model based on this.Methods A total of 186 patients with hepatocellular carcinoma who underwent hepatectomy in a hospital from March 2021 to October 2023 were selected as the model group.They were divided into the bleeding group(n=33)and the non-bleeding group(n=153)according to whether massive hemorrhage occurred during the operation.Univariate and multivariate Logistic regression analyses were used to identify influencing factors and construct a predictive model.Another 79 hepatocellular carcinoma patients undergoing liver resection were included as the validation group.Results The incidence of massive hemorrhage during hepatectomy in patients with hepatocellular carcinoma in the model group was 17.74%(33/186).The results of multivariate analysis showed that tumor diameter>5.0 cm,preoperative PLT<100×109/L,operation time≥5 hours,tumor location in the inferior vena cava of the right posterior lobe of the liver,and a large extent of liver resection were all independent risk factors for massive hemorrhage during liver resection in patients with hepatocellular carcinoma(P<0.05).The area under the ROC curve of the model construction group was 0.800,the sensitivity was 0.697,and the specificity was 0.791.The goodness-fit test results of H-L showed that χ2=7.486,P=0.485.The fitting of the predictive model was good,the consistency was good,and the clinical application value was high.Conclusion The incidence of massive bleeding during liver resection in hepatocellular carcinoma patients warrants attention.The predictive model constructed based on five independent risk factors has strong clinical practical value and can assist healthcare professionals in identifying high-risk patients preoperatively and formulating targeted intervention measures.