INTRODUÇÃO: Enquanto a análise de sobrevida tradicional estima inadequadamente o prognóstico futuro dada alguma sobrevida inicial, a sobrevida condicional ajusta a sobrevida futura pela já observada, permitindo a compreensão da distribuição temporal do impacto dos preditores. OBJETIVOS: Estimar e analisar as sobrevidas global e livre de doença até o décimo ano pós-operatório; identificar preditores independentes destes desfechos; estimar e analisar as sobrevidas condicionais global e livre de doença de cinco anos dada a sobrevida até o quinto ano pós-operatório; analisar o comportamento dos preditores dos desfechos ao longo do tempo. MÉTODOS: Estudo retrospectivo envolvendo 13 centros brasileiros. Dados clínicos, radiológicos e anatomopatológicos foram considerados. Utilizou-se o método de Kaplan-Meier com o teste log-rank para comparar fatores e a regressão de Cox obteve a razão de riscos. A sobrevida condicional foi calculada a partir das tábuas de sobrevida e a diferença padronizada reavaliou as variáveis consideradas significativas. RESULTADOS: 1157 pacientes foram incluídos. A sobrevida global de 1, 3, 5, 7 e 10 anos foi 78,6%, 72,3%, 66,0%, 61,3% e 59,4%, respectivamente. Foram preditoras de sobrevida global: idade [HR 1,04 (IC 95% 1,02-1,06), p 0.000], sexo feminino [HR 1,35 (IC 95% 1,02-1,79), p 0.038], recidiva pós-operatória do CHC [HR 1,35 (IC 95% 1,08-1,79), p 0.003], diâmetro do maior nódulo viável no explante [HR 1,01 (IC95% 1,01-1,02), p 0.006], invasão vascular não discriminada [HR 3,18 (IC95% 1,48-6,85), p 0.004], invasão micro [HR 1,65 (IC 95% 1,27-2,15), p 0.001] e macrovascular [HR 2,25 (IC 95% 1,30-3,89), p 0.000]. A sobrevida condicional global de 5 anos ao final do 1°, 3° e 5° anos foi 79,5%, 82,2% e 90,0%, respectivamente. As variáveis preditoras na análise univariada tiveram comportamento errático ao longo do tempo. A sobrevida atuarial livre de doença em 1, 3, 5, 7 e 10 anos foi 94,2%, 90,1%, 89,8%, 87,5% e 87,5%, respectivamente. Foram preditoras de sobrevida livre de doença: nível sérico de alfa-fetoproteína no diagnóstico [HR 1,0 (IC 95% 1,01-1,02), p 0.000], CHC dentro do critério de Milão no diagnóstico [HR 0,42 (IC 95% 0,22-0,80), p 0.008], explante dentro do critério de Milão [HR 0,34 (IC 95% 0,17-0,68), p 0.002], explante com neoplasia pouco diferenciada ou hepatocolangiocarcinoma [HR 3,04 (IC 95% 1,75-5,30), p 0.000], invasão vascular não discriminada [HR 15,72 (IC 95% 3,44-71,83), p 0.000], invasão micro [HR 3,40 (IC 95% 1,83-6,28), p 0.000] e macrovascular [HR 11,96 (IC 95% 5,20-27,47), p 0.000]. A sobrevida condicional livre de doença de 5 anos ao final do 1°, 3° e 5° anos foi 94,1%, 97,1% e 97,4%, respectivamente. Variáveis preditoras na análise univariada em geral tem maior impacto no primeiro ou segundo ano. CONCLUSÕES: Os resultados do transplante no Brasil foram comparáveis àqueles observados nos EUA e Europa. Considerando-se as perdas precoces, as curvas de sobrevida pelo método Kaplan-Meier foram pessimistas e a análise de sobrevida condicional fornece outra perspectiva para estes dados. O comportamento das variáveis determinantes de prognóstico não é uniforme ao longo do tempo

INTRODUCTION: Traditional survival analysis provides inadequate estimates of the future prognosis for patients with accrued survival. Conversely, conditional survival adjusts future survival by the already accrued survival. It provides insights into the temporal distribution of the effect of predictors. OBJECTIVES: To estimate and to analyse overall and disease free survival until the 10th post-operative year; to identify independent predictors of these outcomes; to estimate and to analyse 5-year overall and disease free conditional survival until the 5th post-operative year; to analyse the behaviour of the predictors of outcomes during follow-up. METHODS: Retrospective cohort from 13 Brazilian transplantation centers. Clinical, radiological, and anatomopathological data were considered. The Kaplan-Meier method with the longrank test for the comparison of factors was applied and the Cox proportional hazards model provided the hazard ratios. Conditional survival was calculated through life tables, while differences between significative variables were reassessed by the standardized difference. RESULTS: 1157 patients were included. Overall survival in 1, 3, 5, 7 and 10 years was 78.6%, 72.3%, 66.0%, 61.3%, and 59.4%, respectively. 350 (30.3%) deaths were observed, 240 (68.6%) in the 1st year. Overall survival was independently predicted by age [HR 1.04 (95% CI 1.02-1.06), p 0.000], female sex [HR 1.35 (95% CI 1.02-1.79), p 0.038], post-operative HCC recurrence [HR 1.35 (95% CI 1.08-1.79), p 0.003], diameter of the largest viable nodule on the explant [HR 1.01 (95% CI 1.01-1.02), p 0.006], non-discriminated vascular invasion [HR 3.18 (95% CI 1.48-6.85), p 0.004], micro [HR 1.65 (95% CI 1.27-2.15), p 0.001] and macrovascular invasion [HR 2.25 (95% CI 1.30-3.89), p 0.000]. 5-year overall conditional survival at the end of the 1st, 3rd and 5th post-operative years was 79.5%, 82.2%, and 90.0%, respectively. Predictors of overall survival identified on univariate analysis presented an erratic behaviour over time. Disease free survival in 1, 3, 5, 7 and 10 years was 94.2%, 90.1%, 89.8%, 87.5%, and 87.5%, respectively. 97 (8.4%) reccurrences occurred. Disease free survival was independently predicted by serum alpha-fetoprotein upon diagnosis [HR 1.0 (95% CI 1.01-1.02), p 0.000], HCC within the Milan criteria upon diagnosis [HR 0.42 (95% CI 0.22-0.80), p 0.008], explant within the Milan criteria [HR 0.34 (95% CI 0.17-0.68), p 0.002], undifferentiated tumor or hepatocholangiocarcinoma on the explant [HR 3.04 (95% CI 1.75-5.30), p 0.000], non-discriminated vascular invasion [HR 15.72 (95% CI 3.44-71.83), p 0.000], micro [HR 3.40 (95% CI 1.83-6.28), p 0.000], and macrovascular invasion [HR 11.96 (95% CI 5.20-27.47), p 0.000]. 5-year disease free conditional survival at the end of the 1st, 3rd and 5th post-operative years was 94.1%, 97.1%, and 97.4%, respectively. Predictors of recurrence on the univariate analysis usually presented with greater impact during the 1st or 2nd post-operative year. CONCLUSIONS: Outcomes of liver transplantation in Brazil were comparable to those from the US and Europe. Survival estimates through the Kaplan-Meier method were pessimistic due to greater early losses. Conditional survival offers a different perspective for the same data. The behaviour of predictive values varies over time