Long-term rodent studies are one of the major tools for determining potential carcinogenicity of environmental chemicals. Rodent studies are costly and time consuming however, their strengths and weaknesses are fairly well understood (Eastin et al., 1998
; Haseman, 2000
). Cost, time and sensitivity are major limitations in evaluations of potential carcinogenicity of chemicals but this is especially true for the hundreds of disinfection byproducts (DBPs)that occur at low concentrations in the drinking water. This prompted the NTP and U.S. EPA to suggest that genetically modified mice might be more useful in providing toxicity and carcinogenicity data on drinking water chemicals and DBPs for setting research priorities and guiding regulatory policies (Boorman et al., 1999
The Tg.AC hemizygous and the p53 haploinsufficient mice were reported to be more rapid, less expensive and more sensitive in detecting potential carcinogens than standard rodent models (Tennant et al., 1995
). Because Tg.AC transcript expression is found in both the skin and forestomach (Cannon et al., 1997
), it was possible that a good correlation might be found in papilloma response at the two sites between drinking water and dermal exposures. Should this correlation prove accurate, it would be possible to screen and rank numerous DBPs for relative carcinogenic potency using dermal exposures at a fraction of the cost of 2-year studies (Boorman et al., 1999
The NTP selected DCA, bromodichloromethane, and bromate as chemicals that are frequently found in the drinking water and for which considerable rodent and mechanistic data was available. DCA is especially well studied with eight mouse studies showing increased liver tumor incidence with drinking water exposure (IARC, 2004
) (also see and ).
In addition to the 39- to 41-week exposures, we included a 26-week exposure to determine whether the shorter exposure would be sufficient to detect a carcinogenic effect in genetically modified mice. DCA exposure by the dermal route in Tg.AC hemizygous mice and by the drinking water route in both Tg.AC and p53 haploinsufficient mice failed to cause a significant increase in tumors in any of the groups of mice at 26 weeks. Exposures longer than we conducted have diminishing utility because in both Tg.AC and P53 mice increasing background tumor incidences and increasing mortality begin to decrease sensitivity to detect an effect.
The top doses in these studies were at or near the maximum feasible concentrations. The highest dermal dose (500 mg/kg) is approximately 150 g DCA/l of dosing solution and the higher drinking water concentrations (1 or 2 g/l) resulted in both decreased water consumption and body weights in the Tg.AC and p53 mice. The decreases were more severe in the p53 mice where mean body weights in the top two dose groups were at least 15% lower than controls in the females and more than 20% lower in the males.
In the 39-week dermal study in Tg.AC mice, there was a clear papilloma response at the site of application in the 500 mg/kg dose group. However, at the next lowest dose (125 mg/kg), dermal papillomas were not significantly increased, suggesting that dermal exposure of water mixtures is unlikely to be a sensitive indicator of potential carcinogenicity for the family of haloacetic acids that might be found in the drinking water.
The p53 haploinsufficient mice failed to demonstrate a tumor response to exposures up to 2000 mg/l for 26 or 41 weeks. This may not be surprising because the p53 model is considered more sensitive to mutagenic carcinogens (Tennant et al., 1995
). A recent workshop concluded that the p53 haploinsufficient mouse is a useful model for risk assessment of genotoxic chemicals (MacDonald et al., 2004
). DCA is considered to be genotoxic in vivo
and in vitro
, but the results for mutagenesis in bacteria and mouse lymphoma cell lines are inconsistent (IARC, 2004
The consistent positive liver tumor response to DCA in drinking water in ten B6C3F1
mouse studies (IARC, 2004
; ) and in one rat study (DeAngelo et al., 1996
) suggested that we might find a liver tumor response in these genetically modified mouse models. A review of all of the pathology data in the current studies revealed only three benign (adenoma) hepatocellular tumors in the 300 mice of both strains and both sexes that were exposed to 500–2000 mg/l DCA in the drinking water. The negative results in the p53 haploinsufficient mouse is perhaps not surprising because it has previously been shown that this model generally does not respond to other chemicals that induce predominantly mouse liver tumors (Spalding et al., 2000
). What received less attention was that the Tg.AC mouse also failed to detect five of the seven nonmutagenic rodent carcinogens (Spalding et al., 2000
The drinking water exposure in the Tg.AC hemizygous mouse appeared to cause an increase in pulmonary tumors. In the 41-week study there was a significant increase in pulmonary adenomas (7/10) in the male mice exposed to 1000 mg/l DCA. Pulmonary adenomas were found in only three males and two females exposed to 2000 mg/l. In the 26-week DCA drinking water exposure study one dosed male and two dosed females were diagnosed with pulmonary alveolar/bronchiolar carcinomas. Although these carcinomas did not appear to be dose-related, they may have been related to DCA exposure.
This study, suggests that routine use of the Tg.AC hemizygous and p53 haploinsufficient mouse may underestimate the number of chemicals that would demonstrate carcinogenic activity if evaluated in standard rodent studies. Genetically modified mouse models continue to be an excellent research model to address mechanistic questions for a variety of disease processes including carcinogenicity. Their use in routine screening of unknown chemicals should be done with caution.