I think with trials involving experimental life saving treatment for terminal illnesses, doing things like control groups and placebos isn't considered particularly ethical, since you'd be giving false hope to someone who is knocking on death's door.
Although not 100%, almost all terminal patients will die. If you select your test group well so they are very likely to die, then you improve the information you get from the test, favor the class of patients who are most likely to benefit from the treatment and least likely to be harmed by it overall.
Ethically it's also dodgy.
You'll end up with bad science.
Btw I think it might be just as cruel to give medications to people that sometimes might not help much or at all (after all it's testing!), and that makes them suffer lots of side effects.
The typical approach for phase I trials is instead to select patients who have somehow exhausted all standard available treatment options. However, there is often a requirement that trial subjects aren’t as close to death as you propose, as the trial needs to observe patients for safety events as much as possible, even in cases where they aren’t receiving much benefit.
The idea that you can't otherwise avoid death means all other factors won't have much, or any effect, on the outcome.
What's going to come next is a randomized case-control trial to show there's a benefit.