I'd argue that minimally positive results in these experiments should not be taken as any indication that the same therapy will work in humans unless they can c

I'd argue that minimally positive results in these experiments should not be taken as any indication that the same therapy will work in humans unless they can convince an _extremely_ skeptical and thorough committee of experts that this result is likely translatable. That would immediately disqualify all xenograft models, and probably most native tumor models as well. So what should we do then? If I have solved it I would be running a successful company selling that secret sauce. But if I were to bet money then I'd say the best approach would be to come up with well thought out in vitro (and some mouse model) experiments that will clearly test all the known ways this drug will interact with every system in the human body; especially the immune system if it's a monoclonal. The emphasis should be on complete declaration of bankruptcy on any hope of a model system showing therapeutic efficacy that's correlated with expected human efficacy.