Characteristics of breast cancer cases and controls from Mexico are presented in . Mexican cases were significantly older at diagnosis, had fewer full-term pregnancies, were less likely to breast feed, were more likely to report a personal history of benign breast disease, a family history of breast cancer, or history of HRT use, and had higher alcohol intake and higher daily caloric intake. Cases also reported a significantly higher level of education and socioeconomic status and lower BMI than controls (the relationship between BMI and breast cancer risk was only observed among pre-menopausal women). Finally, women with breast cancer had more European and less Indigenous American ancestry than controls. There were no significant differences between cases and controls in age at menarche or African ancestry. The proportion of European and Indigenous American ancestry in Mexican controls differed by recruitment site. Monterrey had the largest proportion of European ancestry (40%, standard deviation (sd)=16) compared to Mexico City (28%, sd=19) and Veracruz (30%, sd=18). The proportion of Indigenous American ancestry was estimated to be 54% (sd=15) in Monterrey, 69% (sd=20) in Mexico City, and 64% (sd=20) in Veracruz. The proportion of African ancestry was estimated to be 6% (sd=5) in Monterrey, 3% (sd=4) in Mexico City, and 6% (sd=8) in Veracruz.
Characteristics of Mexican women, by case-control status, 2004-2007
We investigated the association between the different reproductive, demographic, lifestyle characteristics and genetic ancestry among the controls () and observed that SES, education, daily kilocalorie intake and family history of breast cancer significantly differ by ancestry category, with family history being more common among women with higher European ancestry and SES, education and daily kilocalorie intake being higher in women with higher European ancestry. We also explored the relationship between the different characteristics of the controls and SES (). These results show a very strong relationship between SES and genetic ancestry, all reproductive variables, education, alcohol consumption and daily kilocalorie intake. Women with higher SES tend to have less full term pregnancies, higher European ancestry, breast-feed less, consume more alcohol and kilocalories and have more years of education than women with lower SES.
Characteristics of Mexican controls, by European genetic ancestry, 2004-2007
Characteristics of Mexican controls, by SES, 2004-2007
The association between genetic ancestry and breast cancer risk in the sample of Mexican women is shown in . The unadjusted model showed a strong association with European genetic ancestry. Using 0-25% European ancestry as the referent, odds ratios were 1.16 (95% CI=0.94-1.43, p=0.171) for the 26-50% European ancestry, 1.80 (95% CI=1.35-2.39, p=<0.001) for the 51-75% category, and 2.22 (95% CI=1.46-7.11, p=0.004) for the 76-100% category. When known risk factors were adjusted for (), the association with European ancestry was attenuated but the trend remained statistically significant (26-50% OR=1.01, CI=0.78-1.30, p=0.962; 51-75% OR=1.35, 95% CI=0.96-1.91, p=0.087; 76-100% OR= 2.44, 95% CI=0.94-6.35, p=0.067; p for Trend: 0.044). We chose to represent ancestry as a categorical variable since it allowed us to compare the effect size for the extremes of the ancestry distribution. However, the effect is also seen when ancestry is entered as a continuous variable in the model. For the model that included genetic ancestry as a continuous variable, the OR for every 25% increase in European ancestry was 1.20 (95% CI=1.03-1.41, p=0.019). To ensure that there was no confounding due to regional differences between cases and controls, the unadjusted model was stratified by recruitment site (Mexico City, Monterrey, Veracruz) with all results showing the same trend as the global analysis () with the effect weakening as sample size decreases.
Association between genetic ancestry and breast cancer risk in women residing in Mexico, 2004-2007
In the adjusted model, the associations between breast cancer and alcohol consumption, parity, family history, age at menarche, benign breast disease, kilocalorie intake and moderate physical activity were in the expected direction (). When the model testing the association between genetic ancestry and breast cancer risk included SES and education as the only covariates, the two covariates showed an effect on breast cancer risk. However, when other covariates were added to the model, the effect of SES and education became non significant (). Number of full term pregnancies, daily kilocalorie intake and benign breast disease were the variables that when added to the model, absorbed most of the effect of SES and education.
We found no evidence of significant interaction between genetic ancestry and HRT use, BMI, parity, age at first full-term pregnancy, breastfeeding and menopausal status.