Background: Breast cancer incidence is lower in foreign-born Latina women compared to U.S.-born Latinas, and the risk increases with younger age at migration. It is only partially known what factors contribute to these risk patterns. Several explanations have been proposed (e.g., changes in reproductive factors, diet, or environmental pollutants), but risk remains higher in U.S.-born Latinas after adjusting for established breast cancer risk factors. The goal of the present study was to find new risk factors that could explain how the migration experience, with its concomitant environmental and lifestyle-changes, affects the risk of breast cancer. Small reactive molecules have long been suspected of causing cancer by binding DNA and causing mutations. Protein adducts result from the presence of these reactive molecules but have the advantage of being easier to measure than DNA adducts.
Methods: We measured levels of a panel of 23 human serum albumin (HSA)-Cys34 adducts by high-resolution mass spectrometric methods in selected samples of Latina women from the San Francisco Bay Area Breast Cancer Study for whom serum was available (146 controls and 146 cases, 45% foreign-born). Genetic ancestry was estimated from genome-wide genotypes. To identify adducts that were predictive of foreign-born status and breast cancer diagnosis, we used bootstrapped regularized logistic regression and random forest to rank all adducts and additional covariates such as percent of Indigenous American (IA) ancestry in terms of prediction importance, and then selected variables with higher prediction importance. As a measure of the prediction accuracy of selected variables for the binary outcomes, we calculated a cross-validated area under the receiver operating curve (AUC) estimate using the set of selected adducts and covariates, as well as a corresponding 95% confidence intervals.
Results and Conclusions: In a model that included foreign-born status as the dependent variable, all adducts and demographic variables, two adducts were selected as predictors: mass to charge ratio (m/z)=849.07 (addition of S2O3H) and m/z=847.10 (S-methylsulfonylation). Adduct 849.07 had an associated random forest variable importance that was considerably greater than that of any other adduct. This adduct had lower levels in foreign-born Latinas and in Latinas with high IA ancestry. Additionally, adduct 849.07 was the top-ranking variable when breast cancer status was analyzed as the dependent variable. Other variables, including body mass index (BMI), African and IA ancestry, were also selected by the regularized regression as predictive, and together all five variables had a cross-validated AUC estimate of 0.61, 95% CI=(0.55,0.68) when predicting foreign-born status. Our results suggest that adduct 849.07 may be associated with the increased risk of breast cancer observed in U.S-born vs. foreign-born Latinas. More studies are needed to replicate our findings and identify the precursor molecule(s) for this adduct.
Citation Format: Silvia J. Serrano-Gómez, Courtney Schiffman, Hasmik Grigoryan, Henrik Carlsson, Sandrine Dudoit, Esther M. John, Stephen M. Rappaport, Laura Fejerman. Breast cancer risk in U.S-born Latina women: potential role of reactive electrophiles [abstract]. In: Proceedings of the Eleventh AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2018 Nov 2-5; New Orleans, LA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl):Abstract nr C065.