Preoperative prediction of hepatocellular carcinoma with portal vein tumor thrombus based on conventional data

// Pengpeng Zhu 1, * , Yan Liao 1, 2, * , Jiyuan Fan 1, * , Xin Li 3 , Lili Su 4 , Jun Li 1 , Shengguang Yuan 5 , Junxiong Yu 6 and Weijia Liao 1 1 Laboratory of Hepatobiliary and Pancreatic Surgery, Affiliated Hospital of Guilin Medical University, Guilin, Guangxi, P.R. China 2 Disease Prevention and Control Center of Guilin, Guilin, Guangxi, P.R. China 3 Department of Pathology and Pathophysiology, Xiangya hospital, Central South University, Changsha, P.R. China 4 Department of Clinical Laboratory, Nanxishan Hospital of Guangxi Zhuang Autonomous Region, Guilin, Guangxi, P.R. China 5 Department of Hepatobiliary and Pancreatic Surgery, Affiliated Hospital of Guilin Medical University, Guilin, Guangxi, P.R. China 6 Department of Anesthesiology, The Second Affiliated Hospital of Guilin Medical University, Guilin, Guangxi, P.R. China * These authors have contributed equally to this work Correspondence to: Weijia Liao, email: liaoweijia288@163.com Junxiong Yu, email: yujunxiong2897320@163.com Shengguang Yuan, email: ysg518@qq.com Keywords: hepatocellular carcinoma; portal vein tumor thrombosis; prediction; clinical index; survival Received: July 14, 2017     Accepted: September 20, 2017     Published: October 31, 2017 ABSTRACT Hepatocellular carcinoma (HCC) has a high predilection with portal vein tumor thrombosis (PVTT). However, part of the PVTT type can be found only under the microscopy, which was namely as type I 0 . The objective of this study was to establish a simple and inexpensive non-invasive model to predict the presentation of PVTT at HCC patients. A total of 815 HCC patients were retrospectively evaluated and randomly assigned into 2 groups: the training group (n = 408) and validation group (n = 407). A new index model, namely WγAL, was built to predict the presence of PVTT in the training subjects and was further validated in the validation subjects. At the optimal cutoff of 8.90, WγAL identified patients with a hazard ratio (HR) of 7.139 for the presence of PVTT. The area under receiver operating characteristic (AUROC) of WγAL was 0.795 (sensitivity: 71.9%; specificity: 78.6%) for differentiation between PVTT and non-PVTT patients in the training group. The AUROC of WγAL in differentiating patients with PVTT type I 0 from non-PVTT patients was 0.748 (sensitivity: 72.1%; specificity: 68.4%) with an HR of 5.355. In addition, WγAL > 8.90 was significantly associated with large tumors, multiple tumor numbers, TNM stage III-IV, metastasis, and overall survival in HCC patients. The novel predictive model represents a simple and inexpensive model that can identify the presence of PVTT in HCC patients with a high degree of accuracy, with important clinical significance in the future therapeutic management of HCC patients.