Statistical Analysis
All statistical analyses were performed using R software (version 3.6.3; The R Foundation for Statistical Computing), SPSS (version 23, Chicago) and STATA/MP(version 14, USA). Patients’ continuous characteristics were summarized as median±quartile for non-normally distributed variables and mean standard± deviations (SDs) for normally distributed variables. Categorical variables were presented as numbers and percentages. Differences between VD groups were analyzed using wilcox and chi-square tests, where appropriate. We conducted Kaplan-Meier method, univariable Cox proportional hazards models and multivariable Cox model (variables with P <0.05) to estimate hazard ratios (HRs) and 95% confidence interval to discover independent prognostic variables on OS. All subgroup analyses were presented with a COX test for interaction. A two-sided P < 0.05 was considered statistically significant.