We review our studies on the role of polycyclic aromatic hydrocarbon (PAH)-DNA adducts in breast cancer. Additionally we report on analyses of the reliability of the scoring procedures used with immunohistochemical assay for PAH-DNA adducts and of potential bias arising from the use of benign breast disease (BBD) controls. We conducted a case-control study utilizing two control groups: BBD controls who donated tissue and blood samples, and healthy controls who donated blood samples. In comparisons of tumor tissue from cases and benign tissue from BBD controls, increasing adduct levels were significantly associated with case-control status [odds ratio (OR) = 2.40, 95% confidence interval (CI) 1.18-4.92], whereas in comparisons of nontumor tissue from cases and benign tissue from BBD controls the association was nonsignificant (OR = 1.97, 95% CI 0.94-4.17). We also show among cases, but not among BBD controls, that the GSTM1 null genotype is associated with increased adduct levels in breast tissue. Our reliability study found the scoring procedures used with the immunohistochemical assay to have high reliability, 0.93 in nontumor, 0.82 in tumor, and 0.74 in benign tissues. However, we found that the technician significantly contributed to the total variability of a series of data. Finally, we did not find a consistent bias to the null associated with the use of BBD controls; however, BBD controls may overestimate the prevalence of family history of breast cancer compared to that of healthy controls (18% vs.14%). We hypothesize that the higher prevalence results from a referral bias and discuss how this may influence our results.
Copyright 2002 Wiley-Liss, Inc.