Posttransplant mortality risk assessment for adult-to-adult right-lobe living donor liver recipients with benign end-stage liver disease.

OBJECTIVE

A model for living donor liver transplantation (LDLT) outcomes, in concert with pretransplant disease severity assessment, would facilitate informed decision-making on both sides considering donation and transplantation. So far, however, few of studies have focused on models specifically for adult-to-adult right-lobe LDLT recipients with benign end-stage liver diseases. Therefore, we aimed to develop such a prognostic model based on easily obtainable and objective pretransplant characteristics.

METHODS

With data retrospectively collected on 120 recipients, we used Cox proportional-hazards regression to analyze six donor characteristics and 33 pretransplant recipient variables for correlation with posttransplant mortality. In both a modeling set and a prospective validation set with 30 recipients, the performances of the new Cox model, MELD, and MELD-Na+ were assessed by measuring both calibration ability and discriminative power with the Hosmer-Lemeshow test and receiver operating characteristic analysis, respectively.

RESULTS

By univariate and multivariate analysis, donor age, serum total bilirubin, creatinine, and HBV-DNA level were significantly associated with posttransplant mortality. The Cox model, employing these four variables, yielded good calibration ability in the modeling set χ² = 2.465, p = 0.653) and the validation set χ² = 2.836, p = 0.586), and high discriminative power in the modeling set (c-statistic = 0.826, p = 0.001) and validation set (c-statistic = 0.816, p = 0.028). The calibration ability and discriminative power of MELD and MELD-Na+ in both sets were poor.

CONCLUSIONS

The newly derived Cox model was valuable in posttransplant mortality risk assessment for adult-to-adult right-lobe LDLT recipients with benign end-stage liver diseases.