Xtandi: You can't look at median data like that in this case because of heavy censors. Look at patients at risk at median: for OS, around 2 for Xtandi, 27 for placebo; for time to chemo, around 78 for Xtandi, 415 for placebo. Note initial at risk patients were 872 for Xtandi, 845 for placebo. Thus only median time to chemo for placebo arm is close to true median, all other median numbers are going to change dramatically as data mature. HR is more accurate measure of relative risk here.