unfortunately, 50% in mice is rather low for cancer even. You usually want to start out at 70-80% in mice for cancer. If you get 50% in humans that's a great success rate, but not in mice. The reason for this is that the mouse model is called a xenograft model, you take a human cancer and implant it into the mouse. Typically, you are running the experiment on mice implanted with the cell line that gave the best hit in your in vitro model, so it better do darned well, relatively speaking.

I could be wrong about this though, usually the xenograft is done in immunocompromised mice (because otherwise the mice would reject the human cancer line)... So there could be complicating factors.