Identifying optimal candidates for autologous peripheral blood stem cell therapy in patients with decompensated liver cirrhosis: a prognostic scoring system

Abstract

Abstract Background Stem cell transplantation shows great potential to improve the long-term survival of cirrhosis patients. However, therapeutic effects may not be homogeneous across the whole study population. This study constructed an easy-to-use nomogram to improve prognostic prediction and aid in treatment decision making for cirrhotic patients. Methods From August 2005 to April 2019, 315 patients with decompensated cirrhosis receiving autologous peripheral blood stem cell (PBSC) transplantation were enrolled in this study. They were randomly classified into training (2/3) and validation (1/3) groups. A predictive model was developed using Cox proportional hazard models and subsequently validated. The predictive performance of the model was evaluated and also compared with other prognostic models. Results Age, creatinine, neutrophil-to-lymphocyte ratio, and Child–Turcotte–Pugh class were included in the nomogram as prognostic variables. The nomogram showed high discrimination power concerning the area under receiver operating characteristic curves (3/5-year AUC: 0.742/0.698) and good consistency suggested by calibration plots. Patients could be accurately stratified into poor- and good-outcome groups regarding liver-transplantation free survival after receiving PBSC therapy (P < 0.001). Compared with poor-outcome group, the liver function of patients listed for liver transplantation in the good-outcome group was significantly improved (P < 0.001). Besides, our nomogram achieved a higher C-index (0.685, 95% CI 0.633–0.738) and better clinical utility compared with other conventional prognostic models. Conclusions The proposed nomogram facilitated an accurate prognostic prediction for patients with decompensated cirrhosis receiving PBSC transplantation. Moreover, it also held the promise to stratify patients in clinical trials or practice to implement optimal treatment regimens for individuals.