Abstract B85: Racial differences in false-positive mammogram rates: Results from the ACRIN Digital Mammographic Imaging Screening Trial (DMIST)

Abstract

Abstract Introduction: Though screening mammography reduces breast cancer mortality, it carries a significant burden of false-positives, which may lead to unnecessary procedures, anxiety, and increased costs. Because of the relatively low risk of cancer and high level of false-positives in younger women, the U.S. Preventative Services Task Force recommends against routine screening for women younger than 50. This guideline is controversial, and some have raised concerns about its application to black women, who tend to be diagnosed with breast cancer at younger ages and with more advanced disease than whites. However, few studies have quantified the burden of false-positive mammograms among black women. Methods: The ACRIN DMIST trial compared the diagnostic accuracy of digital versus screen-film mammography among 49,528 asymptomatic women enrolled from 2001-2003 at 33 sites in the U.S. and Canada. The current study includes a subset of participants who self-identified as white (N=26,446) or black/African American (N=3176) with no prior history of breast cancer. We compared screening outcomes for white and black women based on the results of digital mammography. False-positive mammograms were defined as those with BIRADS 0, 4, or 5 designation with no subsequent cancer diagnosis in the 15-month follow-up period. Logistic regression was used to estimate the odds of false-positive mammogram by race, breast cancer risk factors, and prior films. A generalized linear mixed model was also fit to account for study site as a random effect. Results: Black participants were slightly younger than white participants (mean 53.7 vs. 55.0). As expected, black and white women differed in terms of reproductive history, with black women having higher parity, lower age at first birth, and lower prevalence of breastfeeding. Black women were also less likely to have prior films available at mammogram interpretation than whites (86.8% vs. 91.3%). The cancer detection rate did not differ significantly by race (whites 0.95%, blacks 0.79% p=0.359). The false-positive rate was higher among blacks (9.2%) than whites (7.8%, p=0.009). Higher false-positive rates were also associated with lack of prior films, higher breast density, and being seen at a study site that enrolled a larger proportion of black participants (r=0.34, p=0.05). After adjusting for age, black women had 17% increased odds of false-positive mammograms compared to whites (OR=1.17, p=0.017). We sequentially added breast cancer risk factors (menopause status, age at menarche, age at first birth, breastfeeding, birth control use, estrogen replacement therapy, prior biopsy, family history of breast cancer), breast density, prior films, and study site to the model. Breast cancer risk factors, prior films, and study site each attenuated the coefficient for race by more than 20%. In the fully adjusted model including study site as a random effect, there was no significant difference in the odds of false-positive mammogram between black and white women (OR=1.04, p=0.561). There were no significant interactions between race and prior films, breast density, age, or menopause status with respect to false-positives. Conclusions: Black women enrolled in the ACRIN DMIST trial had a 1.4% higher false-positive rate than whites. This higher burden of false-positives for black women is clinically significant, given the large numbers of women who undergo screening mammography in the U.S. The higher rate of false-positives among black women was explained by differences in breast cancer risk factors, lower availability of prior mammogram films, and differences in mammogram facility false-positive rates between black and white women in the ACRIN DMIST trial. Citation Format: Anne Marie McCarthy, Jianing Yang, Mirar Bristol, Emily Conant, Katrina Armstrong. Racial differences in false-positive mammogram rates: Results from the ACRIN Digital Mammographic Imaging Screening Trial (DMIST). [abstract]. In: Proceedings of the Sixth AACR Conference: The Science of Cancer Health Disparities; Dec 6–9, 2013; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2014;23(11 Suppl):Abstract nr B85. doi:10.1158/1538-7755.DISP13-B85