We appreciate the interest in our recent article on tumor response measurements.1 Before this study was performed, we undertook an extensive review of the published literature and found no other similar reports. Unfortunately, the referenced Journal of Clinical Oncology abstract was not yet available and, to date, has not been published as a full article. Although our data may be contrary to that abstract and current practice, a correlation between early changes in tumor size as an accurate surrogate for survival has not been established. This concept has been propagated in clinical practice but not justified by scientific studies. Thus, although Drs. Oxnard and Schwartz suggest that our unexpected results should prompt reconsideration of our dataset and analyses, we believe the findings should prompt a reassessment of the underlying tumor biology and mechanism of therapeutic response.
We recognize that this was not the ideal study and that there were limitations. A larger, prospective, multi-institutional trial would have been better to control for variables, including clinical characteristics. However, we do believe that the study results are sufficient to raise concerns and to re-examine current tumor response criteria and measurement strategies.
In our study, we decided to use a simple measurement approach, because the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines were not designed as the optimal method to evaluate response. The number of measurements and response criteria used in RECIST were selected arbitrarily so that clinical trials could be standardized. Prior reports may suggest that measuring more lesions will be better than measuring fewer lesions, but the referenced study only correlated measurements to response and not to long-term survival. It remains unclear that measuring more lesions will be more accurate in predicting outcomes.
We agree with Drs. Oxnard and Schwartz that measurements for response need to be reconsidered. Although sophisticated imaging tools to monitor lesions have been developed, a better understanding of tumor biology is essential if we are to design response criteria as a substitute for survival. Ultimately, we can envision an accurate system that phenotypes tumors at the time of diagnosis, which would then determine the appropriate therapy. When such predictive markers are fully developed and validated, response guidelines will not need to rely on strict tumor measurements and conventional imaging criteria.