I was thinking this originally, but as I read it, it seems the referenced paper is much more narrowly focused. The idea is that the studies involved are easier

I was thinking this originally, but as I read it, it seems the referenced paper is much more narrowly focused. The idea is that the studies involved are easier to do on late-stage patients because mortality is the benchmark and people who are in an advanced stage die sooner. You find out whether it decreases mortality much quicker when the patients are closer to the end. Decreased testing time increases the time when the patent is both active and FDA approved (i.e. years they can make money on it). The usual reasoning behind "there's no money in a cure" is that you can squeeze a lot more money out by treating patients for years than by selling them a cure once. That's still a reasonable cynicism, but it's not what this paper was looking into. Edit: another way of thinking about it is that if a company is maximally evil (economically rational), they would want to research early-stage cancer treatments because that would allow them to milk the patient much longer. Even with this incentive, they still don't do it, they research treatments that are faster to verify efficacy for.