The only thing such a state of affairs clearly indicates is broken is using mice as proto-human test subjects.
Which of course, when you state it like that, is obviously suboptimal. But no one knows what to do about it.
It's useful as a very early test. They used a shotgun approach, they tested 9 bacterias and 1 of them was suspenseful. At least it was suspenseful for 2 weeks, until the study ended. It's very difficult to extrapolate that to the 5 year survival rate in humans.
If that particular bacteria doesn't work in humans it may still trigger a search for a bacteria that does.
Still being optimistic about this :)