Problems with Response Criteria

Summary

During the past 20 years of oncologic research, objective tumor response has become an important study endpoint for phase II and phase III studies involving advanced solid tumors. A variety of factors are changing the basic assumptions behind the response criteria. With the development of new methods for disease visualization, patients formally classified as having unevaluable or nonmeasurable disease are now considered to have measurable disease. Disease can also be visualized more accurately than ever before. Similarly, new methods for the delivery of disease-directed therapy have given new importance to the concept of stable disease. These changing perspectives require sensitivity to the biologic principles of tumor growth and the therapeutic mechanism of action and, in turn, the application of new analytic methodologies to exploit more precise measurements of disease and to model serial measurements. These needs will provide significant opportunities to the oncologist and biostatistician in the formulation of more sophisticated models and analyses.

A variety of issues must be addressed as both disease evaluation methods and therapeutic modalities evolve and they pose the following questions: 1) What are the flaws with the current objective response criteria? 2) What biologic principles should be included in models of tumor growth? 3) What data analysis alternatives are there to the presentation of response rates? How can the actual tumor measurements be utilized to reduce sample size needs? Suggested answers are presented for each of these questions. These proposed remedies will promote more meaningful evaluations of new drugs and modalities, while taking advantage of the increased precision with which the disease can be visualized.