If you're trying to prove a positive benefit, then leaping from mice to humans is risky. If you're concerned about possible negative effects of something, then mice is a good place to start.

Yes, you are right, but ideally a team should test several genetic backgrounds of mice. Almost all cancer treatments have some negative effects. It is crucial to know what genetic and exposure variables to avoid to maximize therapeutic benefits.

Cadmium in some strains of mice is highly toxic to male testes. But if, as in the C57BL/6J strain, you have a “lucky” transporter mutation, then no problems at all. This kind of variability has been known since the turn A. Garrod in the early 1900s. And ignored by many.

Here is the data on the cadmium example I just mentioned:

https://genenetwork.org/show_trait?trait_id=13035&dataset=BX...