Differences in Risk Factors for Breast Cancer Molecular Subtypes in A Population-Based Study
Xiaohong R. Yang,Mark E. Sherman,David L. Rimm,Jolanta Lissowska,Louise A. Brinton,Beata Peplonska,Stephen M. Hewitt,William F. Anderson,Neonila Szeszenia-Dabrowska,Alicja Bardin-Mikolajczak,Witold Zatonski,Richard Cartun,Daniza Mandich,Grzegorz Rymkiewicz,Marcin Ligaj,Stanislaw Lukaszek,Radzisaw Kordek,Montserrat Garcia-Closas
DOI: https://doi.org/10.1158/1055-9965.epi-06-0806
2007-01-01
Cancer Epidemiology Biomarkers & Prevention
Abstract:Analysis of gene expression data suggests that breast cancers are divisible into molecular subtypes which have distinct clinical features. This study evaluates whether pathologic features and etiologic associations differ among molecular subtypes. We evaluated 804 women with invasive breast cancers and 2,502 controls participating in a Polish Breast Cancer Study. Immunohistochemical stains for estrogen receptor α, progesterone receptor, human epidermal growth factor receptors (HER2 and HER1), and cytokeratin 5 were used to classify cases into five molecular subtypes: luminal A, luminal B, HER2-expresing, basal-like, and unclassified. Relative risks were estimated using adjusted odds ratios and 95% confidence intervals. We observed that compared with the predominant luminal A tumors (69%), other subtypes were associated with unfavorable clinical features at diagnosis, especially HER2-expressing (8%) and basal-like (12%) tumors. Increasing body mass index significantly reduced the risk of luminal A tumors among premenopausal women (odds ratios, 0.71; 95% confidence intervals, 0.57-0.88 per five-unit increase), whereas it did not reduce risk for basal-like tumors (1.18; 0.86-1.64; Pheterogeneity = 0.003). On the other hand, reduced risk associated with increasing age at menarche was stronger for basal-like (0.78; 0.68-0.89 per 2-year increase) than luminal A tumors (0.90; 0.95-1.08; Pheterogeneity = 0.0009). Although family history increased risk for all subtypes (except for unclassified tumors), the magnitude of the relative risk was highest for basal-like tumors. Results from this study have shown that breast cancer risk factors may vary by molecular subtypes identified in expression studies, suggesting etiologic, in addition to clinical, heterogeneity of breast cancer. (Cancer Epidemiol Biomarkers Prev 2007;16(3):439–43)