I n this issue of Cancer, Dr. Roberti reviews the role of histologic grade in the prognosis of breast carcinoma and wonders why, because it is available, it has not been widely used in predicting outcome. The position of this editorial is that there must be some fundamental reason, after 100 years of progress on histologic grade, that confusion persists regarding its prognostic value.
The systematic use of morphologic variation at the cellular level of analysis as a prognostic factor in cancer has been fraught with controversy. Currently, there is no universally agreed on set of necessary and sufficient conditions for the definition of histologic grade in breast carcinoma. There has been uncertainty regarding the identification of what variation was important, how the variation should be organized, and whether it should be integrated into a staging or index system.
An additional issue is that grading system criteria have been selected based on their ability to create subgroups of patients using histologic distinctions to produce significant differences in outcome. There are two problems with this approach. First, there are many possible criteria that can create significant differences between subgroups and there is no analytic method for finding the best criteria. 2 Second, statistical significance is not necessarily accuracy. Significance is the chance that two or more distributions of variables, as represented by their parameter estimates, for example, means and variances, are really the same. Accuracy assesses the strength of asso-Supported in part by a research grant from the ciation between two or more variables. 4 In general, accuracy quanti-U.S. Army Medical Research and Development fies how good a variable is at predicting another variable. Specifically, Command Breast Cancer Research Program
we are interested in the strength of association between grade and (DAMD 17-94-J-4383).
survival, i.e., how good is grade at predicting survival.