Abstract
Background: Survival benefit of surgical resection for hepatocellular carcinoma (HCC) patients with portal vein tumor thrombus (PVTT) has been approved recently. However, risk factors for in-hospital mortality in these patients remain unclear. We aimed to determine risk factors and reduce the mortality of these patients. Methods: We analyzed data for 521 of all 1531 HCC patients with PVTT underwent surgery. The primary outcome measure was in-hospital mortality after surgical resection. Univariate and Multivariate cox-regression were performed to identify independent predictors of in-hospital mortality. The methods of Kaplan-Meier, bootstrap and ten-fold-cross validation were applied to validate the risk factors. Results: 521 of 1531 patients in 2004-2012 occurred for the diagnosis of HCC associated with PVTT and underwent surgical resection as a training cohort. Other 325 patients in 2013-2016 were included as a validation cohort. Overall mortality of postoperative in-patients was 3.3% (17/521) and 2.8 % (9/325), respectively. Univariate analysis of mortality revealed that frequency of hospitalization, total albumin, different types of PVTT, bleeding volume, blood transfusion, resection volume, and tumor volume were related with mortality. Therefore, the bootstrap validation reflected that the risk factors of multivariate cox regression in model1(frequency of hospitalization, bleeding volume, and tumor volume) and model 2 (frequency of hospitalization, bleeding volume and total albumin) were stable with mortality in hospital. Ten-fold cross-validation of cox regression analysis showed that the mean C-statistic with 95%CI of model1 and model2 respectively were 0.887(0.779-0.976) and 0.867(0.789-0.966) for predicting in-hospital mortality. Consistency results of models were in the training cohort and validation cohort. Conclusion: Total albumin, tumor volume, intraoperative bleeding and frequency of hospitalization were independent predictive factors for in-hospital mortality in HCC patients with PVTT under surgery. Further study is warranted to utilize these factors to lower in-hospital mortality.