ABSTRACT
Many new anti-cancer therapies target tumors with specific biological characteristics. It is important in these situations to not only establish efficacy of the new therapy but also to accurately identify the subgroup of patients who will benefit. The premise underlying molecularly targeted anti-cancer therapy development is that an understanding of the biological characteristics of these tumors (as well as germline genetic make-up) and of how an anti-cancer agent interacts with those characteristics should help to optimize therapy for patients. Often a new therapy will benefit only a subset of
patients or be effective against only a subset of tumor types, namely those exhibiting the biological characteristics that can be targeted by a therapy having a specific mechanism of action. This paradigm of biologically driven therapy selection presents challenges for the statistical design of clinical trials to establish clinical benefit of new therapies because the evaluation of a therapy cannot be separated from the evaluation of a biomarker test that will identify the patients having the characteristics predictive of benefit. In this chapter we will assume that the biological characterization is accomplished by use of biomarker-based tests, although the principles discussed here apply also to other methods of biological characterization such as imaging. The outline of this chapter is as follows. The chapter begins with a brief discussion of some considerations in the evaluation of analytical performance of a biomarker-based test; a biomarker assay should meet some minimal analytical performance criteria before being used in a clinical trial where it might influence the therapy a patient receives. Next, the distinction between prognostic and predictive biomarkers is explained in order to stress the importance of considering potential prognostic effects of biomarkers in clinical trial designs for evaluating use of those biomarkers for therapy selection. The focus shifts then to a discussion of designs for phase I, II, and III clinical trials incorporating biomarkers. Several design options for phase II and III studies are presented with particular attention to the increasingly important role of randomization when only a biomarker-defined subgroup of patients is eligible to receive the new therapy. 2.2 Analytical Performance of a Biomarker-
Ideally the development of the biomarker assay forming the basis for a biomarker-based test would be synchronized with the development of the targeted therapy, but this does not always happen. However, it is important that by the time the therapeutic development process reaches the stage of large phase II trials or phase III trials, the biomarker assay has undergone sufficient analytical validation. In particular the assay should be reproducible
and accurately reflect the biologic characteristic of interest. Table 2.1 presents considerations for determination of the readiness of a biomarker assay for use in a clinical trial. For the specific situation of a dichotomous biomarker (e.g., mutation present/absent) or a test based on a continuous biomarker measurement to which a cutpoint is applied to report a dichotomous result, a more in-depth discussion on analytical validation considerations is given elsewhere [1]. To achieve an acceptable level of analytical performance prior to use of a biomarker assay in a trial may require multiple iterations of assay refinement. The reader is referred to documents published by CLSI (Clinical and Laboratory Standards Institute) and the U.S. Food and Drug Administration (FDA) for further useful information on the topic of assay analytical validation [2-5].
