Despite having had cancer for more than two years now, I never look at survival curves for my type of cancer. Why not? Because they are not an appropriate tool for telling me how long I am going to live.
They are useful to scientists and doctors in planning treatment and in recognizing what factors influence outcome, but for the individual patient, this means next to nothing. For example, the latest German CLL 8 trial clearly told us that FCR is a better regimen than FC and that if you have the del 11q adverse marker, then adding rituximab puts your prognosis back with main body of patients with CLL. But nothing tells an individual how long he is going to live.
Indeed, quite a lot of patients who received FC lived longer than quite a lot of patients who received FCR. This was obviously true for a number of reasons - e.g. age, stage, other illnesses, FCR toxicity, IgVH status, del 17p - and many of those reasons were quite unknown to doctor or patient.
Median survivals, simply say that at this time point 50% of the cohort were alive and 50% dead. Although randomizing evens out the differences between whole populations in either arm of a trial, it tells us nothing about one individual in a trial. When CLL patients in a trial are followed until everybody is dead - and that will usually take decades of follow-up - there will almost be a few individuals who got the less preferred treatment who form a small tail to the curve; i.e. are long survivors. Who knows, you or I might be one of them.