Measuring response in solid tumors: Unidimensional versus bidimensional measurement

Background: Tumor shrinkage is a common end point used in screening new cytotoxic agents. The standard World Health Organization criterion for partial response is a 50% or more decrease in the sum of the products of two measurements (the maximum diameter of a tumor and the largest diameter perpendicular to this maximum diameter) of individual tumors. However, theoretically, the simple sum of the maximum diameters of individual tumors is more linearly related to cell kill than is the sum of the bidimensional products. It has been hypothesized that the calculation of bidimensional products is unnecessary, and a 30% decrease in the sum of maximum diameters of individual tumors (assuming spherical shape and equivalence to a 50% reduction in the sum of the bidimensional products) was proposed as a new criterion. We have applied the standard response and the new response criteria to the same data to determine whether the same number of responses in the same patients would result. Methods: Data from 569 patients included in eight studies of a variety of cancers were reanalyzed. The two response criteria were separately applied, and the results were compared using the kappa statistic. The importance of confirmatory measurements and the frequency of nonspherical tumors were also examined. In addition, for a subset of 128 patients, a unidimensional criterion for disease progression (30% increase in the sum of maximum diameters) was applied and compared with the standard definition of a 25% increase in the sum of the bidimensional products. Results: Agreement between the unidimensional and bidimensional criteria was generally found to be good. The kappa statistic for concordance for overall response was 0.95. Conclusion: We conclude that one dimensional measurement of tumor maximum diameter may be sufficient to assess change in solid tumors.