*2.5. Statistical Analysis*

Time-dependent receiver operating characteristic (ROC) curve analysis was performed to determine the optimal cut-off values for the prediction of death. However, the optimal values selected by this method did not give a clear cut-off as the cut-off of choice gave unbalanced data. As a result, a median point was used to create bivariate data for the PET parameters. Numerical data are expressed as median (interquartile range (IQR)). Since the data was non-normal, the Mann–Whitney test was used to compare the median values of the SUVmax, MTV, TLG, whole-body MTV and whole-body TLG between the survivors and the non-survivors. Categorical data were summarized using frequencies and percentages. Survival curves for OS were constructed using the Kaplan–Meier method and the log-rank test was used to determine as mortality rates significant differences between groups of the PET bivariate variables. Univariate and multivariate Cox proportional hazards models were fitted for OS to determine independent demographic and clinicopathologic prognostic factors for OS. Cox models results are expressed as a hazard ratio (HR) with corresponding 95% confidence intervals (CI). Statistical significance was set at 5%. Statistical analyses were performed using the STATA package (version 16).
