A brief model evaluated outcomes after liver transplantation based on the matching of donor graft and recipient.

Abstract

A precise model for predicting outcomes is needed to guide perioperative management. With the developments of liver transplantation (LT) discipline, previous models may become inappropriate or noncomprehensive. Thus, we aimed to develop a novel model integrating variables from donors and recipients for quick assessment of transplant outcomes. The risk model was based on Cox regression in a randomly selected derivation cohort and verified in a validation cohort. Perioperative data and overall survival were compared between stratifications grouped by X-tile. Receiver operating characteristic curve and decision curve analysis were used to compare the models. Violin and raincloud plots were generated to present post-LT complications distributed in different stratifications. Overall, 528 patients receiving LT from 2 centers were included with 2/3 in the derivation cohort and 1/3 in the validation cohort. Cox regression analysis showed that cold ischemia time (CIT) (P=0.012) and the Model for End-Stage Liver Disease (MELD) (P=0.007) score were predictors of survival. After comparison with the logarithmic models, the primitive algorithms of CIT and MELD were defined as the CIT-MELD Index (CMI). CMI was stratified by X-tile (grade 1 ≤1.06, 1.06< grade 2 ≤1.87, grade 3 >1.87). In both cohorts, CMI performed better in calculating transplant outcomes than the balance of risk score, including perioperative incidents and prevalence of complications. Model integrating variables from graft and recipient made the prediction more accurate and available. CMI provided new sight in outcome evaluation and risk factor management of LT.