In its extensive comment on EPA’s draft toxicological review of perchloroethylene, HSIA argues that the draft fails to conduct an objective and comprehensive assessment of the relevant studies in humans and laboratory animals, including several more recent publications specifically designed to answer questions raised in earlier studies. HSIA summarized its concerns as follows -

Epidemiology
EPA’s assessment of the cancer epidemiological evidence for perchloroethylene contrasts sharply with the assessment recently completed by the European Union which considered the same data. The primary difference is that the European report recognizes the limitations of the mixed solvent community drinking water studies on which EPA appears to place equal weight to the cohort studies. It is particularly disturbing that EPA has discounted a study (Lynge et al. 2006) designed expressly to answer the key questions posed by the extensive epidemiologic data base: why were increased esophageal cancers observed in perchloroethylene-only subcohorts studied by the National Institute for Occupational Safety and Heath and the National Cancer Institute, but not in other studies?

Neurotoxicity
While thorough, EPA’s assessment of human neurotoxicity studies fails to translate the limitations of the reviewed studies into an evaluation of the potential bias that could explain the findings. This is particularly true for its discussion of the residential studies. While an objective review of the available human data might conclude that higher-level exposures (>50 parts per million, or ppm) are probably associated with subtle changes in neurosensory perception, it would note that the available evidence does not support the firm conclusions regarding lower-level exposures that EPA reaches in the draft assessment. The proposed congruency for cognitive effects in animals and human beings suggested by EPA, moreover, is not based on the available data and is not defensible

PBPK Modeling
EPA has erroneously concluded that there is no basis for preferring one physiologically based pharmacokinetic (PBPK) model over another. To the contrary, there are important differences among the available models. EPA fails to use the available data sets to evaluate and compare these models to assist in identifying those models that are inappropriate. By including an inappropriate model EPA inflates its unit risk estimate by a factor of 10 relative to the other models EPA used. Furthermore, EPA does not consider a more comprehensive model developed in 2007 -- the only PBPK model that takes into account data on urinary excretion of the major metabolite trichloroacetic acid (TCA) in humans at a wide range of exposures including levels below the point at which the primary metabolic pathway becomes saturated.

EPA also has failed to assess the bioavailability of TCA in mice in drinking water bioassays. As a consequence, the Agency has dramatically overestimated the amount of TCA available and, therefore, underestimated the potency of TCA as a mouse liver carcinogen. This, in turn, has led to EPA’s erroneous opinion that TCA formed from perchloroethylene cannot explain the number of liver tumors observed in mice. This is a fundamental factor in considering the role of TCA in the mode of action for perchloroethylene induction of mouse liver tumors.

Mode of Action for Mouse Liver Tumors
Existing data on perchloroethylene and its metabolites suggest that activation of the receptor, PPARα, is the most plausible mode of action (MOA) for the formation of mouse liver tumors. Perchloroethylene, and/or its major metabolite TCA, have been shown to bind and activate the PPARα, which induces peroxisome proliferation, cell proliferation and oxidative stress, resulting in selective clonal expansion. In its draft review, EPA discredits this MOA by relying on inaccurate and irrelevant data, including information that is not consistent with the biological changes that have been observed in the liver of exposed mice, and omitting information on the key events that support an MOA involving the PPARα.

Having discounted the most plausible MOA, EPA defaults to a mutagenic MOA to explain the mouse liver tumors, even though perchloroethylene and its major metabolite TCA do not appear to be genotoxic at the doses at which liver tumors were observed. This is not a scientifically supportable position.

Mononuclear Cell Leukemia
In its decision to develop cancer potency values on the basis of the occurrence of mononuclear cell leukemia in a single rat strain (F344), EPA ignores the findings from studies in other strains of rats and strong evidence that the relatively high level of susceptibility in the F344 rat make it a very poor model for human response. EPA’s speculative mode of action for a leukemogenic response in humans, moreover, is directly refuted by the available evidence.

In support of its decision to use mononuclear cell leukemia data for calculating cancer potency, EPA inappropriately gives greater weight to community-based drinking water studies suggesting an association between perchloroethylene and leukemia than to cohort studies that do not.

Subjective Review of Studies
There is no indication in the draft assessment how EPA assigned weight and relevance to the many studies that it reviewed as part of the assessment. The draft document makes little reference to current EPA guidelines for interpretation of data and there is no evidence of a rigorous and systematic application of the criteria established in those guidelines. Rather, the assessment seems to take at face value studies that show an association between perchloroethylene and adverse effects and to disregard studies that show no such effects.

Uncertainty Analysis
EPA is to be congratulated for undertaking an uncertainty analysis as part of its assessment of perchloroethylene. We recognize the difficulty of applying formal uncertainty analysis to the complex toxicological and epidemiological databases for perchloroethylene. However, the uncertainty analysis does little to inform the reader about how the uncertainties identified should be taken into account. The presentation is confused and unstructured and, in some cases, amounts to little more than an opportunity for the authors to engage in wild speculation. The discussion of uncertainty lacks the careful application of scientific judgment and synthesis to the available data.