Abstract A102: The role of socioeconomic status, pathology and genetic biomarkers in explaining racial disparities in breast cancer survival

Abstract

Abstract Background and Objective: The current body of evidence suggests that while a portion of the racial disparity in breast cancer survival can be attributed to socioeconomic factors an independent effect of race remains, which may be partially due to differences between African American and Caucasian women in phenotypes of gene expression for markers that either promote or prevent tumor growth. While these data suggest a strong genetic pathway through which race may influence breast cancer survival, no studies to date have been able to simultaneously study race, socioeconomic status (SES), pathology in addition to biomarker status. This analysis builds upon earlier work with a breast cancer registry by adding measures of SES to permit a more detailed assessment of the relationship between race and SES on breast cancer survival in light of pathological and biomarker status. Methods: With IRB approval, retrospective data were compiled from the Thomas Jefferson University Hospital (TJUH) tumor registry of African-American and Caucasian breast cancer patients including tumor stage, histologic grade, ER and PR positivity, Ki-67 proliferation index, and c-erb-B2 and p53 expression. Patient overall survival data were assessed through the end of 2007. The tumor registry data were then merged to the 2000 Census data, matching on zip codes. The federal poverty level (FPL) and median household income of each zip code were used as proxies for individual patient SES. A multiple imputation modeling approach was utilized to increase the power of the results and minimize potential bias due to missing data. Multivariable Cox proportional hazards models were used to model association of demographic, clinical, pathological, and socioeconomic measures with breast cancer patient survival. Results: Registry data from 2,230 patients diagnosed between 1994-2002 were available. Approximately 13% of the patients in the registry were African-American, with the remaining 87% Caucasian. Median income appears to explain a greater portion of the race effect than does the FPL, as the hazard ratio for race decreases from 1.40 to 1.21 in the presence of median income but only decreases to 1.31 in the presence of the FPL. When pathology factors are added to the median income model, the hazard ratio for race decreases again to 1.17. Finally, the addition of biomarkers to the median income model further reduces the hazard ratio for race to 1.1. Conclusions: The analyses presented here provide evidence that although a portion of the impact of race on breast cancer survival is explained by differences in SES between African-American and Caucasian women, the race effect is also partially explained by racial variations in breast cancer phenotypes. Furthermore, the results of this study indicate that the relationships among race, SES, pathology and biomarker status are all intertwined, rendering typical statistical methodologies for multivariate models inadequate for fully describing the relationships among these important variables. Understanding what proportion of racial disparities is attributed to SES factors versus biological factors has implications for determining the most efficient interventions for reducing the disparity. Citation Information: Cancer Epidemiol Biomarkers Prev 2010;19(10 Suppl):A102.