Perhaps none. It might be that the loss didn't result in a significant enough disadvantage to hinder the growth of the population without it. There are plenty of genetic traits that are not particularly adaptive: chance, too plays a role. The assumption that the genetically youngest populations are optimal is a kind of teleology, no?
Perhaps a large metabolic cost for larger animals- hard to maintain that energy cost when you've just had your limb cut off + trying to fight off infection, run away from predators, etc.
Having had a herpetologist roommate at one time, A large part of being a reptile, snakes in particular, involves sitting around and doing nothing.
Mammals seem to have a different take, always moving and scurrying. Perhaps it was not tenable?
Maybe I misread the article, but it sounded like having the gene could raise one's risk of cancer.
It could (particularly since p21's suppression is related to general anti-cancer processes), though they explicitly said they did not observe it in the mice trials (they were actively looking for cancer).
Mice have short lives. I am not sure how frequent cancer is in mice compared to humans.
If it increases the risk of cancer it might be infeasible for species with a long generation cycle. With a short generation cycle the cancers might not kill off too many individuals before they regenerate?