Impact of African ancestry on the relationship between BMI and survival in early stage breast cancer: Retrospective analysis from E5103.

Abstract

1010 Background: Both Black race and obesity are associated with worse survival in early stage breast cancer. Obesity disproportionately affects Black women; however, the degree this contributes to racial disparities in breast cancer remains unclear. Prior work evaluated heterogeneous populations or used self- reported race, rather than genetic ancestry. African ancestry is associated with higher BMI and worse survival in breast cancer; however, the intersection between genetic ancestry and obesity on survival outcomes remains unknown. Methods: We analyzed data from the adjuvant trial E5103. Patients with high risk, HER2 negative breast cancer received doxorubicin/cyclophosphamide x 4, followed by weekly paclitaxel x 12, with or without bevacizumab. Genetic ancestry was determined on the 2,854 patients with available germline DNA, BMI, and outcome data using principal components from a genome-wide array. The primary objective assessed impact of BMI on DFS and OS by ancestry. Multivariate Cox proportional hazard models evaluated correlation between continuous or binary BMI and survival in African (AA) and European (EA) Americans. Results: 13.4% of patients were genetically classified as AA and 86.6% as EA. Higher continuous BMI was significantly associated with worse DFS and OS only in AAs (DFS: HR = 1.25 95% CI 1.07-1.46, p = 0.004; OS: HR = 1.38 95% CI 1.10-1.73, p = 0.005); not in EAs (DFS HR = 0.97 95% CI 0.90-1.05, p = 0.50; OS HR = 1.03 95% CI 0.93-1.14, p = 0.52). By disease subtype, BMI was associated with worse outcomes only in AAs with ER+, and not TNBC. By categorical BMI, WHO class III obesity (³ 40) significantly associated with worse DFS and OS only in AAs (DFS HR = 1.98, p = 0.010; OS HR = 2.07, p = 0.064), not in EAs (DFS HR = 0.97, p = 0.86; OS HR = 1.28, p = 0.30). Proportion of African ancestry (proAA) was associated with higher BMI and worse outcomes in the total population; however, within AAs there was no significant interaction between proAA and BMI on DFS (HR = 0.36, p = 0.06) or OS (HR = 0.38, p = 0.24). In AAs, BMI remained associated with DFS (HR = 2.78, p = 0.019), suggesting higher BMI is associated with worse DFS regardless of proAA. Coefficients for the interaction term indicate that as proAA increases the impact of BMI on outcome is lessened. Conclusions: Higher BMI is significantly associated with worse breast cancer outcomes in women of African ancestry in E5103, but not in those of European ancestry. Categorically, this association was significant only for severe obesity, indicating the relationship may depend on the degree of obesity. As proAA increased in AAs, the impact of BMI on outcome was lessened, suggesting other host factors may contribute more to obesity’s influence on outcome than genetics. Determination of the optimal populations for weight loss interventions will advance precision medicine efforts to impact racial disparities and outcomes in early stage breast cancer.