Statistical analysis
Statistical analysis was conducted using the statistical software SPSS 25 (IBM Corp., Armonk, NY). Chi-square test was used to analyze the associations between pathological response and patient characteristics. Complete remission was considered as normal serum tumor markers (<10 mg/L or below the age-related reference value) and absence of all imaging abnormalities in computed tomography or magnetic nuclear resonance imaging. Partial remission was defined as AFP decline and shrinkage of tumor size by at least 50%. Relapse was defined as detection of new lesions in local or distant sites in terms of complete remission. The RFS were generated according to the Kaplan and Meier method. The influences of suspected prognostic factors associated with RFS were analyzed with the Breslow test due to most relapse occurring in short time. Log-rank test was performed to determine the survival difference in long term. RFS was calculated as the time from diagnosis to the first relapse or death (death related with disease or therapy-oriented complication) or from the first relapse to the next relapse. All statistical tests were two-sided and the P values < 0.05 were considered statistically significant.