Weight of evidence evaluation of potential human cancer risks from exposure to polychlorinated biphenyls: an update based on studies published since 2003.

Drawing on all data available in 2003, the WoE of the human epidemiological data for polychlorinated biphenyls (PCBs) demonstrates that exposure to a mixture of PCBs (i.e. Aroclors) did not pose a cancer risk to humans ( Golden et al. (2003) . This evaluation was based on criteria established by the US Environmental Protection Agency (EPA) as well as on a different methodology used by the Agency for Toxic Substances and Disease Registry (ATSDR). Subsequently, at least 15 more studies on the potential cancer risks (both incidence and mortality) of PCBs have been published. All studies published since 2003 are critically reviewed using the criteria established by the EPA (2005) and ATSDR (2000) . None of the studies published since 2003 change the conclusions drawn by Golden et al. (2003) : "that the weight of evidence does not support a causal association for PCBs and human cancer". This conclusion pertains to all cancers combined, as well as to the various cancers that have been sporadically reported in the occupational cohort mortality studies. With respect to breast cancer risk, the WoE is compelling that environmental exposure to PCBs is not etiologically implicated in breast-cancer risk. This conclusion is supported by the consistently negative findings for increased breast-cancer mortality in occupational studies, which now involve almost 9,000 women occupationally exposed to PCBs. Similarly, the incidence studies in which PCB background levels are reported to be associated with increased risk of non-Hodgkin's lymphoma or prostate, testicular, and intestinal cancer are not corroborated by occupational cohort studies with PCB exposures far in excess of environmental exposures. The most likely explanation for these discordant findings is discussed in this review. Finally, the recent elucidation of the mode of action by which PCBs promote liver tumors in rats, combined with the demonstration that none of the key events in the mode of action occurred until substantial tissue accumulation of total PCBs had occurred, casts further doubt that PCB exposure at environmental or occupational levels poses a carcinogenic risk to humans. The dramatic differences between rodents and humans in sensitivity to PCB-mediated induction of CYP1A1 suggests that even occupational exposures to PCBs have never resulted in PCB body burdens approaching the levels required to initiate the sequence of events involved in the promotion of liver tumors in rodents.