Racial disparities in breast cancer age-specific incidence rates suggest risk factors interact to cause elevated rates of high grade estrogen receptor negative disease, especially in African American women Free

A17

We hypothesized that age-specific incidence rates for breast cancers categorized by aggressiveness would provide important insights into the etiology of racial disparities in incidence. Incidence data are independent of population structure, and facilitate identification of trends not possible with case-case analyses. Using state cancer registry and population data for 1996-2004, restricted to African-American (AA) and European-American (EA) women, we compared age-specific incidence patterns of invasive female breast cancer in South Carolina (a southern, mostly rural population) and Ohio (a northern, more urbanized, more racially and ethnically diverse population). Each case was categorized by estrogen receptor expression status (ER negative (ER-) versus positive (ER+)) and histologic grade (G1 = low grade, G2 = intermediate, G3 = high grade), with ER-/G3 disease the most aggressive. Missing ER and grade data were imputed using state, age and race-specific distributions of known grade and ER. Age-specific incidence patterns were modeled using Poisson regression, where predictors were age and stage, plus race, ER status and grade, and, as appropriate, interaction variables. Rates for each ER/Grade combination were age-adjusted to the 2000 US standard population. Distinct age-specific incidence patterns were found for each ER/Grade group, with several striking differences by race. EA women had a substantial excess incidence of both ER+/G1 and ER+/G2 cancers (combined, 75% higher than AA women in SC, 44% in Ohio), primarily in post-menopausal years. Conversely, AA women had a substantially higher incidence of ER-/G3 cancer (63% in SC, and 67% in Ohio). Racial differences in rates by ER/Grade were fairly constant between ages 30 and 65 years. In both races, regardless of grade, the incidence ratio of ER- cancer (versus ER+) peaked at about age 35. Comparing Ohio to SC, rates in EA women were extremely similar for each ER/Grade group. Rates in AA women also varied little between States, notably except ER+/G1 and ER+/G2 cancers which combined had a 28% higher rate in Ohio; State differences in AA ER-/G3 rates were < 4%. Based on Poisson regression modeling, risk of breast cancer increased strongly when both ER- and G3 characteristics were present(interaction odds ratio 8.85, p < 0.00001). The three-way interaction between ER-, G3 and AA race was significant but modest: odds ratio 1.68, p = 0.047. These results suggest that (a) ER negativity and high histologic grade result from different risk factors; (b) these risk factors synergize, not only to produce a more aggressive disease but also to dramatically increase the risk of breast cancer; (c) prevalence of the co-occurrence of risk factors for ER negativity and high grade disease is greater in AA than EA women; and (d) incidence rates of ER positive cancers in AA women may rise as populations become more urbanized and/or with racial mixing, especially in post-menopausal ages, but rates of ER-/G3 cancer, the most aggressive form, may remain relatively constant.