A predictive model based on radiomics, clinical features, and pathologic indicators for disease-free survival after liver transplantation for hepatocellular carcinoma: a 7-year retrospective study

Hao Xie; Bin Shi; Junzhen Fan; Shui Liu; Qiaozhi Ma; Junnan Dai; Siqing Dong; Ying Liu; Han Meng; Hui Liu; Ya Yang; Xuetao Mu

doi:10.21037/jgo-24-347

A predictive model based on radiomics, clinical features, and pathologic indicators for disease-free survival after liver transplantation for hepatocellular carcinoma: a 7-year retrospective study

J Gastrointest Oncol. 2024 Oct 31;15(5):2187-2200. doi: 10.21037/jgo-24-347. Epub 2024 Oct 29.

Authors

Hao Xie^{1

2}, Bin Shi³, Junzhen Fan⁴, Shui Liu⁵, Qiaozhi Ma⁶, Junnan Dai⁶, Siqing Dong⁷, Ying Liu⁸, Han Meng¹, Hui Liu², Ya Yang⁹, Xuetao Mu⁶

Affiliations

¹ Postgraduate Training Base of the Third Medical Center of Chinese PLA General Hospital, Jinzhou Medical University, Beijing, China.
² Department of Radiology, the Jintang First People's Hospital, Chengdu, China.
³ Department of Organ Transplantation, the Third Medical Center of Chinese PLA General Hospital, Beijing, China.
⁴ Department of Pathology, the Third Medical Center of Chinese PLA General Hospital, Beijing, China.
⁵ Department of Radiology, Aerospace Center Hospital, Beijing, China.
⁶ Department of Radiology, the Third Medical Center of Chinese PLA General Hospital, Beijing, China.
⁷ Department of Radiation Oncology, Jinzhou Central Hospital, Jinzhou, China.
⁸ School of Medical Imaging, Weifang Medical University, Weifang, China.
⁹ Department of Ultrasound Medicine, the Jintang First People's Hospital, Chengdu, China.

Abstract

Background: Disease-free survival (DFS) is an essential indicator for evaluating the prognosis of liver transplantation (LT) in hepatocellular carcinoma (HCC) patients. Despite progress in the prediction of DFS by radiomics, only preoperative clinical features have been combined in most studies. The aim of this study was to construct a nomogram model (NM) using preoperative clinical features, radiomics, and postoperative pathological indicators for more effective prediction of DFS.

Methods: This was a retrospective study of a single-center cohort comprising 139 HCC patients. Using the whole cohort, we constructed and assessed a clinical model (CM) based on alpha-fetoprotein (AFP) and alkaline phosphatase (ALP), a pathological model (PM) based on Ki-67 and tumor number, a radiomics model (RM) based on the radiomics score (Rad-score), and an NM based on the above five independent predictors.

Results: Significant correlations between the NM and DFS were observed in the training and validation cohorts. Among the four prediction models, the C-index of the NM was the highest [(training/validation cohort) CM: 0.664/0.676, PM: 0.737/0.691, RM: 0.706/0.697, NM: 0.817/0.760], and the areas under the receiver operating characteristic curves (AUCs) of the NM prediction of 1-year, 2-year, and 3-year DFS were also the highest [(training/validation cohort) 1-year, 2-year, and 3-year CM: 0.726/0.726, 0.685/0.744, 0.645/0.686, PM: 0.789/0.780, 0.801/0.748, 0.841/0.735, RM: 0.769/0.752, 0.717/0.805, 0.748/0.765, NM: 0.882/0.854, 0.867/0.849, 0.882/0.801]. The NM also exhibited the highest net clinical benefit.

Conclusions: Based on radiomics, clinical features, and pathological indicators, the NM could be used to effectively predict DFS after LT in HCC patients, guiding the follow-up and complementary treatment.

Keywords: Hepatocellular carcinoma (HCC); disease-free survival (DFS); liver transplantation (LT); nomogram model (NM); radiomics.