Remember that a randomized trial with PFS or OS endpoint does not a prior require measurable disease.  ORR is a secondary endpoint in those settings, so measurable is not required.  If allowed as an eligibility factor, it should (must) be a stratification factor.  If permitted within the trial, more criteria including how to select (AGO/iMODEL) and sensitivity analysis as a function of CGR to see if it created a bias to outcome of the trial.