p53 Alterations and Other Tumor CharacteristicsJ ones et al. recently investigated p53 expression in cases of breast carcinoma obtained from 145 African-American (AA) and 177 white (W) patients. 1 Using immunohistochemistry, they found that breast carcinomas that develop in AA patients were more likely to be p53 positive compared with those in W patients (25% in AA patients vs. 7% in W patients). They also reported the following characteristics among the AA cases: later disease stage, higher tumor grade, and loss of estrogen receptor (ER) expression. We conducted a similar study involving breast carcinoma cases from 100 AA and 100 W patients. 2 We did not find any significant differences with regard to tumor grade or p53 expression between the two cohorts, and it appeared initially that our findings contradicted those in the study by Jones et al. However, on further review, the majority of these differences can be explained.Our cases were matched on age at diagnosis, stage of disease, and ER status. Matching was performed to minimize the confounding effects of demographic and clinicopathologic variables on the study endpoints, including prognosis. Therefore, the proportion of ER-positive cases was similar in both groups (73% in AA patients and 69% in W patients). It is well known that ER-positive breast carcinomas are more likely to be p53-negative, well differentiated tumors. 3 In fact, this inverse association between ER positivity and p53 positivity was noted in both our AA (odds ratio [OR] of 0.38) and W (OR of 0.13) cases. Therefore, grade distribution and p53 positivity (36% in AA patients and 38% in W patients) were similar among our AA and W cases. However, in the study by Jones et al., 1 the cases were not matched on ER status, disease stage, or patient age, and the proportion of ER-negative cases was found to be significantly higher among the AA cases (54% compared with 39% in the W cases), a finding that is consistent with previous reports. 4 Therefore, the findings of Jones et al. 1 of higher tumor grade and p53 expression may be explained in part by the higher prevalence of ER-negative tumors. However, the authors did include a multivariate analysis in which p53 positivity remained significantly more common among AA cases after adjusting for several variables, including ER status. It is not clear whether the contrasting results reported by our group and by Jones et al. are the result of actual differences in the two datasets or of errors that occurred during either matching or multivariate analysis. We hope that the types of studies performed by Jones et al. and