Statistical analysis
The associations of prognostic factors with death were assessed using the chi-squared test for univariate analysis and analyzed with Cox proportional hazards regression for multivariate analysis (considering the time from chemotherapy initiation to the last follow-up time or death) to establish the hazard ratio (HR) of death in 298 GTN patients. Only positive variables in univariate analysis were enter into the multifactor model, except for FIGO stage and FIGO score. Descriptive analyses were performed for clinical characteristics, treatment, and prognosis information of 11 ultra-high-risk GTN patients with liver or brain metastases. The rates of CR after initial treatment and resistance to initial treatment in ultra-high-risk GTN patients were compared between those with and without liver or brain metastases using the chi-squared test. Survival time was compared in 298 GTN patients stratified according to liver or brain metastases using the log-rank test. Survival time was compared in 18 high-risk GTN patients stratified according to having/not having liver or brain metastases was compared by the log-rank test. A secondary analysis of longer-term survival was performed at six years (72 months), which was the median follow-up time for the entire cohort. There were no missing values. P -values <0.05 were considered to be statistically significant. All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).