Preoperative factors and models predicting mortality in liver transplantation

We analyzed preoperative factors related to postoperative mortality after liver transplantation among a cohort of 268 consecutive liver transplant patients over 6 years. We studied the impact of 10 recipient variables, 14 donor features, and three operative aspects. We also studied the correlation with death and survival using various predictive scores (Child, Cordoba Score, MELD, and UCLA). Univariate analysis showed that the factors with a significant association with postoperative mortality were the use of noradrenaline in the donor, total ischemia time (>12 hours), and transplant indication (hepatitis C virus versus the rest). Multivariate analysis of mortality showed the impact of female donor sex, recipients over >60 years, recipient albumin less than 2.8, and total graft ischemia time more than 12 hours. Univariate analysis of 1-year survival showed a statistically significant relation with D/R gender similarity, as well as donor GOT (>170) and GPT (>140) values. Multivariate analysis of 1-year survival showed donor GOT (>170) and donor/recipient gender similarity to be significant. Concerning the prediction models, Child-Pugh (AB versus C) best determined postoperative mortality (P < .006), MELD was predictive of 1-year survival (P < .03). The most important variables related to postoperative mortality were total ischemia time over 12 hours, recipient albumin less than 2.8, and age above 60 years. The variable with most impact on 1-year survival was the degree of graft hepatocyte lesion as determined by GOT. The Child-Pugh system is still the best indicator of postoperative mortality, although MELD may also be a good predictor of survival.