On the carcinogenicity risk assessment of chromium compounds

The evaluations of the potential carcinogenicity ofmetallic chromium and trivalent chromium by severalinternational and national agencies [International Agencyfor Research on Cancer (IARC), 1990; Agency for ToxicSubstances and Disease Registry (ATSDR), 1993; CanadianEnvironmental Protection Act (CEPA), 1994; United StatesEnvironmental Protection Agency (US EPA), 1998; UnitedStates National Toxicology Program (US NTP), 2002] andindividual scientists [Langa˚rd, 1990; Lees, 1991; Hayes,1998] are unanimous that the evidence is inadequate inhumans. For some occupational sources of chromiumexposure (e.g., ferrochrome industry and manufacture ofchrome pigments) and for some occupations (e.g., leathertannery workers, painters, and chromium platers) there areincreased risks, but almost invariably in the epidemiologicstudies the available data do not permit discriminationbetween simultaneous exposure to trivalent chromium andhexavalent chromium. Although the chromium compoundthatincreasestheriskoflungcancerandsinonasalcancerhasyet to be identied, there is fairly general agreement thathexavalentspeciesareresponsibleforthesediseasesandthatthe trivalent and metallic species are not.Gibbetal.[2000]claimedthattheirstudyonlungcanceramong workers in chromium production offers the bestquantitative evidence to date of the relations betweenhexavalent chromium exposure and lung cancer. This maywell be the case, but their conclusion that ‘‘cumulativehexavalent chromium exposure was associated with anincreased lung cancer risk; cumulative trivalent chromiumexposure was not’’ warrants a critical comment. In an efforttorefutetheconclusiondrawnbyMancuso[1997]accordingtowhichallformsofchromiumarecarcinogenic,Gibbetal.modeled for the separate effects of cumulative hexavalentchromium, Cr(VI), exposure and cumulative trivalentchromium, Cr(III), exposure on the risk of death from lungcancer.Proportional hazards models using age as the timevariate,cumulativeexposureasatime-varyingcovariate,andsmoking as a confounding covariate were applied to assessthe relation between cumulative chromium exposure andlung cancer mortality risk. When one exposure variate at atimewasincludedinthemodel,cumulativeCr(VI)exposureand cumulative Cr(III) exposure were found to haveseparately almost equivalent, statistically signicant risk(or hazard) ratios for each tenfold increase in cumulativeexposure:1.38(95%condenceinterval1.20–1.63)and1.32(95% condence interval, 1.15–1.51), respectively. Despitethe very strong correlation between the log of cumulativeCr(VI) exposure and the log Cr(III) exposure (correlationcoefcient