The basic reply to this, I think, is that stars are not relevant examples, as they are, like planets, mere accidental unities. They are hence more analogous to, say, human societies, which do have tendencies and regularities and are hence identifiable for the purposes of certain queries, but are not substances.
Choosing a different analogy, Fr. William A. Wallace writes in his The Modeling of Nature:
"The unity of a star would seem to be analogous to the unity of the earth: largely a mass of different substances held together by natural forces of one type or the other. And if the evolutionary model of stellar development is correct, a star can have a history even though it has not a single nature like an oak or a chipmunk (p 69)".
I realize this wouldn't be too useful to someone critical of the general A-T approach to the subject, but this does seem to address the immediate worry.