Can we satisfactorily measure the clinical value of new classes of oncology agents with a single summary measure?