Joint effects of nine polychlorinated biphenyl (PCB) congeners on breast cancer risk

Background Polychlorinated biphenyls (PCB) have been a major environmental health concern because of their wide distribution and persistence in the en vironment. Estimating joint effects of all congeners in a single analysis i s complicated by correlation among exposure levels, and the resulting colli nearity makes the results difficult to interpret. Methods Patients with breast-related surgery at Yale-New Haven Hospital wer e interviewed using a standardized questionnaire, and breast adipose tissue samples were analysed for nine PCB congeners (74, 118, 138, 153, 156, 170, 180, 183, 187). The study recruited 490 women (304 cases and 186 controls) between 1994 and 1997. Logistic ridge regression was used to analyse the i nstability caused by collinearity. Results Although total PCB did not appear to be associated with breast canc er risk, significant differences in effect were observed among the nine con geners. Logistic ridge regression demonstrated a protective effect on breas t cancer risk for a potentially anti-oestrogenic and dioxin-like congener, 156, while two phenobarbital, CYP1A and CYP2B inducers had an adverse effec t, 180 and 183. This analysis also suggested that a protective effect for a nother phenobarbital congener, 153, was largely explained by instability ca used by collinearity. Conclusions These results indicate that studies of PCB congeners and health require an in-depth statistical analysis in order to better understand the complex issues related to their collinearity.