What the LOGISTIC procedure is saying is that you have an over-parametrized model. To fix that you either need more data (so all the combinations are represented) or fewer variables.
If Radical is defined at any one of Abdominal, Laproscopic, or Vaginal, then just drop Radical, as you can estimate it's effect via the other parameters.
ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.