I don't know much about oncology, but it strikes me that they probably don't know with certainty which of the 20% it will be. They have a limited number of interventions that put the patient through hell, and it sometimes works.

There's probably extreme cases with bad metastasis where it's very unlikely that they can be saved, and they can probably tell some of those. But I think not all the time.

Must be a very depressing job to observe this over the years.