African American (AA) women experience higher breast cancer mortality than white (W) women. These differences persist even among estrogen receptor (ER)-positive breast cancers. The 21-gene recurrence score (RS) predicts recurrence in patients with ER-positive/lymph node-negative breast cancer according to RS score-low risk (RS, 0-18), intermediate risk (RS, 19-31), and high risk (RS, >31). The high-risk group is most likely to benefit from chemotherapy, to achieve minimal benefit from hormonal therapy, and to exhibit lower ER levels (intrinsically luminal B cancers). In the current study, the authors investigated racial differences in RS testing, scores, treatment, and outcome.
Tumor registry data from 3 Atlanta hospitals identified women who were diagnosed with breast cancers during 2005 through 2009. Medical record abstraction provided information on RS and other tumor/treatment factors. Statistical analyses used chi-square/exact tests and logistic regression.
Of 2186 patients, including 1192 AA women and 992 W women, 853 women had stage I or II, ER-positive/lymph node-negative disease and, thus, were eligible for RS testing (AA = 372 [31.2%]; W = 481 [48.5%]; P < .0001); and 272 women (31.8%) received testing (AA = 76 [20.4%]; W = 196 [40.7%]; P < .0001). Tumors were distributed into the following groups according to risk: low risk (n = 133), medium risk (n = 113), and high risk (n = 26). The mean RS did not differ by race, but risk groups did (low-risk group: 46.1% vs 50% for AA women and W women, respectively; high-risk group: 15.8% vs 7.1%, respectively; P = .043). In multivariate analyses, AA race (odds ratio, 3.6) was associated independently with high risk scores.
AA women were half as likely as W women to receive 21-gene RS testing but were 2-fold more likely to be categorized as high risk. The current data suggested that testing guidelines are not applied equivalently, testing bias may attenuate racial differences in RS, and disparate outcomes may be explained in part by differences in RS, although compliance and pharmacogenomics also may play a role.