Reply to L.A. Newman.

Abstract

We thank Newman for her comments on the potential for descriptive analyses such as those in Park et al to shed light on racial disparities in breast cancer outcomes. Overall, US agestandardized breast cancer mortality rates have been higher for blacks than whites (non-Hispanic white women), but rates have been decreasing in both blacks and whites since about 2000. The mortality rates are influenced by factors that act both before and after diagnosis. Breast cancer incidence has increased more in black than white women since 1990, and black women have less favorable stage and molecular features at diagnosis. The less favorable stage distribution may reflect differences in rates of screening mammography and/or tumor biology. After diagnosis, access to treatment and responsiveness to treatment can affect breast cancer mortality. More data are needed on the comparative effectiveness of treatment of black and white women, but the best data from clinical trials suggest that black and white patients with similar stage, cancer subtypes, and treatments have similar breast cancer–free survival after treatment. However, the fact that treatments are effective in black women in clinical trials does not guarantee that similar women will receive adequate treatment in the general population. Menashe et al concluded that factors affecting breast cancer mortality after diagnosis, such as treatment access, explained more of the disparity in US breast cancer mortality rates than estrogen receptor status at diagnosis. Thus, numerous factors, including incidence rates, tumor characteristics, socioeconomic status, health care delivery, and response to treatment, affect US breast cancer mortality rates and warrant the type of investigation recommended by Newman. Park et al focused on breast cancer mortality after diagnosis in the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program. Here, we extend those analyses to compare black and white patients. In Table 1, we present the cumulative pure risk of dying of breast cancer within 5 years of diagnosis in 1973 to 1979 and in 2005 to 2010, separately for black and white patients. This 5-year risk is calculated as 1 minus the Kaplan-Meier estimate of not dying of breast cancer within 5 years, treating competing risks as independent censoring. Both in 1973 to 1979 and in 2005 to 2010, black patients had a higher risk of breast cancer death than white patients within each age group and stage. There have been large decreases in the risk of breast cancer death for both black and white patients, but the percentage decreases were larger in white patients. If, as suggested by Dignam, treatments are equally effective for blacks and whites with comparable disease and treatment in randomized clinical trials, these data suggest either that black patients have had less access to proper treatment or that black patients at a given age and stage have breast cancers that are more difficult to treat than white patients in the same age and stage group in the general population. More detailed analyses taking into account tumor size within stage, estrogen receptor status, and human epidermal growth factor receptor 2 (HER2)/neu status are needed to investigate the latter possibility, but the results in Menashe et al on estrogen receptor status suggest a key role for access to proper treatment. As expected from treatment patterns, the percent decreases in breast cancer mortality were greater in women younger than 70 years of age with local or regional disease, regardless of race (Table 1). To evaluate concerns raised by Newman, we used methods of Park et al to calculate the proportion of the decrease in ratios of breast cancer mortality hazard rate within 5 years (comparing 2005 to 2010 with 1973 to 1979) that could be explained by stratification on tumor size within stage and age group (Fig 1). Underlying this approach is the tacit assumption that the percentage explained by stratification on tumor size is related to interventions that occur before diagnosis, such as screening, rather than to factors that occur after diagnosis, such as treatment. For black and white women younger than the age of 70 years, only a small proportion of the improvement can be explained by changes in tumor size, but changes in tumor size distribution explain more of the improvement in women 70 years of age or older at diagnosis. This