Abstract B68: Breast cancer mortality differences by race/ethnicity: Distinct differences by decade and hormone status.

Introduction: Our limited understanding of breast cancer mortality differences by race/ethnicity may be enhanced by examination of changes in those differences over time, and by further assessment of such differences by hormone receptor status, an important prognostic indicator. We sought to determine whether mortality disparities remained consistent with calendar time, and whether any differences were apparent between hormone receptor subgroups, in a large population-based study.

Methods: Using Surveillance Epidemiology End Results (SEER) data for invasive breast cancer cases in 9 registries diagnosed from 1975-2004 and followed through 2009, we utilized Cox regression to calculate hazard ratios (HR) for 5-year breast cancer-specific mortality among women of African American race or Hispanic ethnicity, compared with non-Hispanic white women. Women considered Hispanic due only to a Hispanic surname were omitted. HR were estimated separately for women diagnosed in each succeeding 5 years to examine trends with calendar time. Separate models were fit for cases diagnosed in 1990 or later by estrogen receptor (ER) status. The proportion of mortality disparity by race/ethnicity attributable to ER status was evaluated using the change in the HR estimate when ER status was included in Cox models.

Results: During the 5 years following diagnosis, 65300 deaths occurred in 418229 women. Among African-Americans, the HR for 5 year cause-specific mortality increased steadily every 5 calendar years, from 1.6 (95% Confidence Interval (CI) 1.5-1.7) among 1975-79 diagnoses to 2.4 (95% CI 2.2-2.5) in 2000-04 diagnoses. Among Hispanics, the corresponding HR increased from 1.0 (95% CI 0.9-1.2) to 1.5 (95% CI 1.4-1.7), with most change occurring in 1995 or later diagnoses. Mortality declined in all groups with calendar time, but white non-Hispanic women had a lower baseline mortality in 1975, and experienced a greater decline. Among 1990 or later diagnoses, the HR for ER-positive disease exceeded that for ER-negative disease (African-American: ER-positive HR =2.1 (95% CI 2.0-2.2) ; ER-negative HR=1.6 (95% CI 1.5-1.7); p-interaction <.0001 ; Hispanic: ER-positive HR=1.4 (95% CI 1.3-1.6); ER-negative HR=1.2 (95% CI 1.1-1.6);p-interaction=.02). The greater mortality among ER-positive cases accounted for 27% of the total mortality disparity in African-Americans in 1990-2009, and 30% in Hispanics.

Conclusions: Misclassification of Hispanic ethnicity in earlier decades may have attenuated HR estimates, thus improved categorization may be responsible in part for the Hispanic HR increase with calendar year. Our results pertain only to the 5 years following diagnosis, and not to subsequent years, during which HR decline in both groups and the proportional hazards assumption is violated. Our findings suggest increasing mortality differences during the initial years following diagnosis among African-American (and possibly Hispanic) women from 1975-2004 that cannot be attributable to genetics alone. Further, our results imply that such disparities are stronger among those that are ER-positive. Thus efforts to reduce mortality differences in these groups might be directed to improving survival in ER-positive women.

Citation Format: Deirdre Hill, Catherine Axtell, Huining Kang, Melanie Royce. Breast cancer mortality differences by race/ethnicity: Distinct differences by decade and hormone status. [abstract]. In: Proceedings of the Fifth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2012 Oct 27-30; San Diego, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2012;21(10 Suppl):Abstract nr B68.