Background: We expanded an existing cohort of workers (n = 2,588) considered highly exposed to polychlorinated biphenyls (PCBs) at two capacitor manufacturing plants to include all workers with at least 90 days of potential PCB exposure during 1939-1977 (n = 14,458) . Causes of death of a priori interest included liver and rectal cancers, previously reported for the original cohort, and non-Hodgkin lymphoma (NHL) , melanoma, and breast, brain, intestine, stomach, and prostate cancers, based on other studies. Methods: We ascertained vital status of the workers through 1998, and cumulative PCB exposure was estimated using a new job exposure matrix. Analyses employed standardized mortality ratios (SMRs ; U.S., state, and county referents) and Poisson regression modeling. Results: Mortality from NHL, melanoma, and rectal, breast, and brain cancers were neither in excess nor associated with cumulative exposure. Mortality was not elevated for liver cancer [21 deaths ; SMR 0.89 ; 95% confidence interval (CI) , 0.55-1.36], but increased with cumulative exposure (trend p-value = 0.071) . Among men, stomach cancer mortality was elevated (24 deaths ; SMR 1.53 ; 95% CI, 0.98-2.28) and increased with cumulative exposure (trend p-value = 0.039) . Among women, intestinal cancer mortality was elevated (67 deaths ; SMR 1.31 ; 95% CI, 1.02-1.66) , especially in higher cumulative exposure categories, but without a clear trend. Prostate cancer mortality, which was not elevated (34 deaths ; SMR 1.04 ; 95% CI, 0.72-1.45) , increased with cumulative exposure (trend p-value = 0.0001) . Conclusions: This study corroborates previous studies showing increased liver cancer mortality, but we cannot clearly associate rectal, stomach, and intestinal cancers with PCB exposure. This is the first PCB cohort showing a strong exposure-response relationship for prostate cancer mortality.
A.M. Ruder, National Institute for Occupational Safety and Health, Mailstop R-16, 4676 Columbia Parkway, Cincinnati, OH 45226 USA