There's a theory that it doesn't pay our genes to invest resources in longer lives -- presumably resources like better DNA repair mechanisms, etc., in the case of cancer -- so we end up with our parts mutually adapted to some optimal lifespan for our niche. I'm surprised too that the same basic plans can scale across such a wide variation in lifespans -- but then elephants are way different from shrews in other ways, too...

They do, sometimes. Usually the ones that do don't go on to reproduce, so genes that protect you from getting cancer that early predominate. Dogs and cats have already had their offspring by their teens, so it really doesn't matter, in terms of evolutionary fitness, whether they get cancer at that age.

Well, pets and humans have totally different life cycles. 20 years for a dog is probably like 100 for a human.