[Problems in evaluating the effects of cancer chemotherapy]

A clinical response of a solid tumor can obviously be defined as some measurable reduction in the patients total tumor burden lasting for at least one month. However, this is a variable definition. Many pitfalls exist in interpreting improvement in a measurable criterion of response as evidence of an anticancer effect. Measurement error on palpation has been widely mentioned. Furthermore, on the x-ray examination, it is difficult to be sure that we are measuring even accurately across the same planes from one examination to the next. Patients with only non-measurable (but evaluable) lesions should be excluded from those clinical trials that are designed to assess response. Because of measurement error, minor response and stable disease categories should be abandoned unless they are of long duration. A period of regression lasting for only one month usually conveys little or no benefit in terms of quality or quantity of life to patients. To search for antitumor activity in a modality by using survival as an endpoint is a far too complex and time-consuming effort. In some rare tumors biochemical measurements performed on peripheral blood give an indication of the total tumor burden. However, in the most common tumors, no such reliable markers exist. One more major deficiency in assessing response to treatment is the lack of methods for reliably assessing the quality of life of the treated patients. Furthermore, large differences in rates of response are probably due to variable and inappropriate exclusion of patients from analysis of results. Differences in policies for excluding patients from analysis are a source of variation in reporting the results. We believe that some measure of change in tumor size will continue to be viewed as an acceptable endpoint until better methods of assessing response become available. On the other hand, at present, since we cannot eliminate the measurement error, it would seem advisable to make an effort to understand it, to learn how great an effect it can have on the validity of reporting results, and consider how to defend against it. Finally, more stringent criteria that relate to the true clinical benefit be required as measures of response to treatment.