To have done what you are suggesting would have required two trials being run concurrently with the same primary endpoint: one in which all patients were on Crestor, and a second in which all were on Lipitor.  As greater patient numbers and funding would have been needed, it might not have been feasible. 

In the absence of scientific evidence (eg, animal or cellular studies) that apabetalone might be more effective in the presence of one statin than the other, presumably the FDA thought it was not justifiable to exclude Lipitor merely on the basis of a post hoc subgroup analysis of the earlier trials (which can produce misleading results  owing to such factors as poor matching and small numbers).

Hopefully, in the prespecified Crestor vs Lipitor subgroup comparison in BETonMACE the two groups will have been well matched as the randomization to placebo or apabetalone was carried out after assignment to one or other statin treatment group.