Paper: Occupational exposures and postmenopausal breast cancer: Impact of sensitivity analysis

Author(s) and Affiliation(s):
France Labrèche, Institut de recherche Robert-Sauvé en santé et en sécurité du travail du Québec
Mark S. Goldberg, McGill University
Marie-France Valois, McGill University
Louise Nadon, INRS-Institut Armand-Frappier
Download Presentation PDF:
Day/Time: Friday at 13:30
Room: St. Patrick Room, 3rd Floor

Over the last 40 years, incidence rates for breast cancer have increased steadily in Canadian postmenopausal women and, although the increase has stabilized since 1999, it has not decreased as in other jurisdictions. We conducted a case-control study to determine whether organic solvents and other occupational agents, such as polycyclic aromatic hydrocarbons (PAHs), confer a higher risk of developing postmenopausal breast cancer.


This study was conducted in Montreal between 1996 and 1997 and included 556 incident histologically-confirmed cases of malignant breast cancer among postmenopausal women age 50-75 years. 613 controls were matched for age, date of diagnosis, and hospital, and selected from other sites of cancer, excluding sites possibly associated with solvent exposure (liver; pancreas; lung, bronchus, trachea; brain, central nervous system; leukemia; lymphoma). Face-to-face or telephone interviews elicited information on non-occupational risk factors and on lifetime job history; industrial hygienists then attributed exposure to about 300 chemical and physical agents for each job. Unconditional logistic regression was used to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CI). A first analysis, including all controls, showed elevated risks with exposures to mono-aromatic hydrocarbons (MAHs), to PAHs, and to a few synthetic fibres. Sensitivity analyses were performed by excluding certain control cancer sites based on mechanistic considerations and by adjusting for co-exposures.


Based on the hypothesis that some of these substances behave as xenoestrogens, we compared risks obtained with all cancer controls to risks obtained after excluding hormonal cancers from the control series. We found that after exclusion the ORs increased slightly with exposures to MAHs, and, for certain hormonal phenotypes, to acrylic fibres, to organic solvents with reactive metabolites and to synthetic fibres as a group, without changing the level of association from the first analyses. For the other substances, risks remained similar or decreased slightly, with decreased statistical precision for nylon fibers and PAHs from petroleum sources.


The associations did not change substantially after excluding hormonal cancers from the control series (despite the reduced sample size) for 3 occupational exposures: acrylic fibres (all ages, and exposures before age 36 years); MAHs (all ages); and organic solvents with reactive metabolites (exposures before age 36). This sensitivity analysis adds some support to the finding of increased risks of postmenopausal breast cancer associated with these occupational exposures. Adjustment for co-exposures and exploration of dose-response relationships will complement our analysis. Additional studies are warranted to further our understanding of the role of chemicals in the development of breast cancer.