Abstract

Patients registered to a Phase II study often are heterogeneous and may not be expected to respond equally to a new treatment. Tumors can consist of multiple sub-types diagnosed either histologically or by molecular methods. For example, soft tissue sarcomas or non-Hodgkin lymphomas have multiple subtypes that can be identified by microscopic inspection of the tumor sample. Similarly, newer genomic measurements such as gene expression or comparative genomic hybridization on these tumors can now lead to refinement or regrouping of patients for clinical studies. For instance, expression of measures of HER2, EGFR, VEGF may influence the expectation of response by certain drugs in some tumors. While we are specifically interested in Phase II studies in oncology, our ideas relate more broadly to the use of biomarker or subtype information in the design of clinical experiments.