Possible implications of device-specific variability in post-EVAR sac regression and endoleaks for surveillance categorization.