Will there be resistance to the RECIST (Response Evaluation Criteria in Solid Tumors)?

From the earliest days of the Cancer Chemotherapy Cooperative Group Clinical Trials program in the 1950s, there have been concerns about the definition of response in solid tumors and the reliance on quantitative measurements. In the first clinical trial in solid tumors published by Zubrod et al. (1), “Treatment was considered to give a positive response if either the total measured tumor mass decreased, with no lesions increasing in size and no new lesions appearing . . . or the group of voting physicians considered that the treatment had been of benefit to the patient as a whole. . . . .” On the initiative of the World Health Organization (WHO) and following two meetings on the standardization of reporting results of cancer treatment in 1977 and 1979, Miller et al. (2) proposed uniform criteria for reporting response, recurrence, and disease-free interval and the grading of acute and subacute toxicity. These criteria have received wide acceptance and have become known as the WHO criteria in reporting the results of cancer treatment. In 1992, the Southwest Oncology Group (SWOG), in cooperation with the National Cancer Institute and other major cooperative oncology groups, participated in meetings to develop new toxicity criteria, end point definitions, and response criteria. A set of new guidelines was proposed because of “uncertainties in clinical trials objectives, limitations in the resolution of imaging methods, and demands for greater rigor in response and endpoint definitions” (3). In this issue of the Journal, the criteria have been re-examined by a broad array of members from an international assortment of cooperative cancer study groups, and new guidelines for evaluating response have been proposed (4), which they refer to as RECIST (Response Evaluation Criteria in Solid Tumors). These guidelines are very similar in their definitions of response and progression to those proposed recently by James et al. (5) who have three co-authors in common with (4). What, then, are these new guidelines and how do they differ from the WHO criteria? Table 1 gives a comparison of the WHO and RECIST guidelines (with differences in bold) in the definitions of the measurability of lesions at baseline, objective response, overall response, and duration of response. The major proposed change is that RECIST uses unidimensional measurements of the sum of the longest diameters (LDs) of tumors instead of the conventional bidimensional WHO method of the product of the longest diameter and that perpendicular to it, summed over all measured tumors. Also, the criteria for progressive disease (PD) differ between RECIST and WHO guidelines. From Table 1, the definition of complete response (CR) is essentially the same between the guidelines; however, the definition of partial response (PR) differs. For PR, WHO requires a 50% decrease in the sum of the products of the perpendicular diameters from baseline, confirmed at 4 weeks, whereas RECIST requires at least a 30% decrease in the sum of LDs from baseline, confirmed at 4 weeks. These criteria are almost equivalent if one assumes spherical tumors and that the LD and the diameter perpendicular to the LD both decrease by at least 30% (although the latter was not measured by RECIST) because then the sum of the products of the diameters would decrease by approximately 50% or more. Therasse et al. (4) do not give a rationale for the choice of the 30% decrease criterion; however, James et al. (5) argue that measuring one dimension is simpler than two and that, for tumors of different volumes, the requirement for PR is more nearly linearly related to changes in LD than changes in the product of perpendicular diameters when tumors are spherical [Fig. 1 of (5)]. However, Hilsenbeck and Von Hoff (6) point out that measuring one dimension of a tumor is not really less laborious than measuring two, since usually multiple measurements are needed to make sure that one has the maximum diameter, especially for nonspherical tumors. There is general agreement in the literature that a spherical tumor 1 cm in diameter has approximately 10 cells. Patients will have varying tumor volumes present at the start of study and, ideally, one would define PR based on the percent reduction in tumor volume at some time after the start of treatment. However, since tumor volumes cannot be accurately measured, is it better to use the LD or the product of the perpendicular diameters as a substitute for tumor volume? Skipper et al. (7) tested chemotherapeutic agents in animal systems, including L1210, CA755, and S180. Skipper et al. stated that “Basic studies in model leukemia systems indicate: the approximate average number of leukemic cells surviving therapy may be estimated by assuming logarithmic proliferation and a critical lethal number and that the percent (not the absolute number) of a leukemic cell population killed by a given dose of a given drug is reasonably constant.” In some animal and human tumor systems, a first approximation to a dose–response curve is that the logarithm of percent of cells surviving after treatment changes approximately linearly with log dose. If the assumption is made that tumors are spherical and that responding patients have equivalent percentage reductions in the measures of length, width, and depth of the tumor, then there would be essentially no difference in defining PR based on changes in LD or the product of perpendicular diameters. However, if for some tumors percent changes in LD do not reflect changes in the other dimensions, then surely measuring two dimensions would be better than one in estimating tumor volume. RECIST (unidimensional) and WHO (bidimensional) criteria were applied to the same patients recruited in 14 different trials