Robert G. Tardiff

Comprehensive Risk Assessment

The assessment of risks to public health is a highly complex undertaking. A single risk assessment comprises many discrete decisions—assumptions, interpretations, weighing of evidence—that are necessary if useful conclusions are to be reached concerning the existence and magnitude of risks to human health presented by a substance. This chapter describes those principles (some of which have been extracted from previous chapters) applicable to various facets of health risk assessment and integrates them into a functional and scientifically defensible approach to the interpretation of data bases comprised of in vivo and in vitro toxicological information and results from human studies.

Comment on “Occurrence and Potential Significance of Perfluorooctanoic Acid (PFOA) Detected in New Jersey Public Drinking Water Systems”

Post et al. (1) report on their commendable attempt to estimate a health-based drinking water concentration for PFOA. Their estimate is based on propositions and approaches that led to an overly and unnecessarily conservative recommendation of 0.04 μg PFOA/L (ppb). Based on our recent comprehensive risk and safety evaluation of PFOA in tap water (2), we found that lifetime consumption of PFOA in drinking water is safe for all forms of toxicity at levels in the range of 0.9-7.7 ppb. They have relied on internal dosimetry as the dose metric for PFOA. We concur that this is the most reliable metric for systemic toxicity. However, we find that the abbreviated means by which they did so (ratio of water (median):serum concentrations) undermines the value of internal dosimetry and distorts their recommended guideline for PFOA in tap water. Using an unsubstantiated inference, they rely on the unreplicated median ratio of 100, without including a sensitivity analysis for the range of variability of that ratio in the studied cohort. We submit that the more accurate and reliable approach to estimate serum levels of PFOA from ingestion of PFOAcontaining water is the application of an internal dosimetry (PBTK) model to calculate external from internal doses which are then converted to drinking water equivalent levels. The preferred model for interspecies comparisons is that of Clewell (3), which takes into account relevant physiological and kinetic parameters and the half-life (3.5 years) of PFOA in humans. That half-life has now been reported to be reduced to 2.3 years (4), which would correspondingly increase our estimated safe levels of exposure to PFOA in tap water from those noted above. Post et al. consider the cancer dose-response to be linear, a USEPA default, and estimate that a water concentration of 0.06 μg/L would equate to an upper-bound risk level of 10-6. Rather, the data support a nonlinear dose-response relationship: (a) PFOA is not genotoxic in assorted in vivo and in vitro assays, (b) Leydig cell (testis) tumors in rats are found to result largely from a cascade of biochemical changes (e.g., PPARR induction is unrelated to humans) leading to sustained cell proliferation, processes which are recognized as having a nonlinear dose-response; and (c) Pancreatic Acinar cell tumors in rats are also dependent on a mode of action (MoA) regulated by PPARR leading to sustained cell proliferation, a process with a nonlinear dose-response. Based on MoA, a nonlinear approach (Benchmark Dose and Uncertainty Factors [UFs]) might have been used preferentially for carcinogenicity by Post et al. Post et al. apply UFs to NOAELs to estimate their PFOA “health-based drinking water concentrations” for noncancer effects. Their UF of 10 for interspecies extrapolation is overly conservative. By relying on internal dosimetry, as they claim, the kinetics part of the UF should be 1 leaving the remaining toxicodynamics part as 2.5 (5). They also apply a default UF of 10 for reliance on a 6-month study in monkeys with a rationale that the duration of exposure was less than lifetime. Actually, that duration UF is justified as 1 on the basis that PFOA, while persistent, is not bioaccumulative; it reaches steady-state based on the level of exposure, meaning that the serum concentration will not increase further with increasing duration once steadystate has been achieved (<6 months in monkeys).

A Summary of the Workshop "Applying Biomarkers to Occupational Health Practice"

1National Institute for Occupational Safety and Health, Cincinnati, Ohio 2University of ErlangenNuremburg, Erlangen, Germany 3Occupational Safety and Health Administration, Washington, DC 4Johns Hopkins University, Baltimore, Md. 5Rohm and Haas, Co, Spring House, Pa. 6University of Maryland, Baltimore, Md. 7University of Washington, Seattle, Wash. 8Columbia University, New York, N.Y. 9University of Cincinnati, Cincinnati, Ohio 10Sapphire Group, Inc., Washington, DC 11University of Montreal, Montreal, Canada INTRODUCTION