Read the peer-reviewed article that challenges NWBO's trial conclusions. They are neurooncologists and frame the cherry-picking better than I -- read it yourself as the Regulators will.
The protocol was for a PFS primary endpoint -- cross-over was intended as salvage medication for those that don't respond. It caused a confounding of the nOS by repeated dosing of cherry-picked patients (read the article). The external comparators didn't cherry-pick like NWBO did -- that's how you could improve you OS -- healthier patients to start.