NONLINEAR LOW-DOSE EXTRAPOLATIONS

Traditional cancer risk assessment has taken a different approach. As a defanlt assumption, carcinogens have been considered to act withont a threshold of toxicity 

Cancer Risk Assessment, edited by Ching-Hung Hsu and Todd Stedeford Copyright 2010 John Wiley Sons, Inc. 

The low-dose linearity assmnption has largely been considered a health-protective policy decision reflecting an upper bound on cancer risk estimates. One 

---

In considering the shape of the tumor dose-4 esponse curve, it is important to recognize that the mere observation of a tumor response that appears different from linearity is not in itself an adequate justification for the determination that nonlinear low-dose extrapolation should be used for that chemical. For example. Lutz et al. 

The second step of the dose-response assessment is an extrapolation to lower dose levels, i.e., below the observable range. The purpose of low-dose extrapolation is to provide as much information as possible about risk in the range of doses below the observed data. The most versatile forms of low-dose extrapolation are dose-response models that characterize risk as a probability over a range of environmental exposure levels. Otherwise, default approaches for extrapolation below the observed data range should take into account considerations about the agent s mode of action at each tumor site. Mode-of-action information can suggest the likely shape of the dose-response curve at these lower doses. Both linear and nonlinear approaches are available. 

Most attempts at describing CWA PK and PD have used classical kinetic models that often fit one set of animal experimental data, at lethal doses, with extrapolation to low-dose or repeated exposure scenarios having limited confidence. This is due to the inherent nonlinearity in high-dose to low-dose extrapolations. Also, the classical approach is less adept at addressing multidose and multiroute exposure scenarios, as occurs with agents like VX, where there is pulmonary absorption of agent, as well as dermal absorption. PBPK models of chemical warfare nerve agents (CWNAs) provide an analytical approach to address many of these limitations. 

High-low dose extrapolation. The model predicts nonlinearity due to saturated elimination at concentrations well above 100 ppm, even in combination with physical exercise and ethanol. The comparison between the simulated arterial blood levels and experimental arterial blood levels obtained after exposure to 20 ppm 2-butoxyethanol for 2 hours indicates linear kinetics at this concentration, which is lower than the PEL in the United States (PEL = 50 ppm, OSHA 1974). Thus, linear kinetics would be expected in ordinary occupational inhalation exposure, although low-dose extrapolation was not specifically addressed in this model. 

Reitz RH, Quast JF, Schumann AM, et al. 1980. Nonlinear pharmacokinetic parameters need to be considered in high-dose/low-dose extrapolation. Arch Toxicol Suppl 3 79-94. 

There is considerable controversy over the shape of the dose-response curve at the chronic low dose levels important for environmental contamination. Proposed models include linear models, nonlinear (quadratic) models, and threshold models. Because risks at low dose must be extrapolated from available data at high doses, the shape of the dose-response curve has important implications for the environmental regulations used to protect the general public. Detailed description of dosimetry models can be found in Cember (1996), BEIR IV (1988), and Harley (2001). 

It is important to distinguish between factors leading to linearity as dose approaches zero (low-dose linearity) and considerations of Unearity or nonlinearity of the overall dose-response curve. Thus, while consideration of the kinetics of fundamental biological processes (e.g., absorption, DNA reactivity, DNA repair) indicates that these processes tend to be linear as the dose approaches zero, that linear slope may be very different from the slope derived by extrapolation from the animal tumor data. Several authors have also reported nonlinearities in DNA mutation, even for direct-acting mutagens, although issues of assay sensitivity make it difficult to distinguish nonlinearity from a true threshold. 

The EPA does not list MO As with probable low-dose nonlinearity. While several MOAs have some acceptance as being nonlinear or having a threshold (e.g., cytotoxicity followed by regeneration, and receptor-mediated processes), the establishment of nonlinear dose-responses for specific carcinogenic MOAs is still controversial (Andersen et al. 2003). More discussion on implementation of nonlinear extrapolation by regulatory agencies for nonlinear carcinogens is presented later in this chapter. 

Despite the challenges that nonlinear extrapolation for carcinogens offers in terms of regulatory implementation, it is important that the best available science guide risk assessment practices. Default assumptions designed to be health-protective, such as low-dose linear assumption, are useful when no other alternatives are available, but public health initiatives will be best served when grounded in biologically based approaches. 

A critical problem in the application of pharmacokinetic principles to risk extrapolation is the potential change in metabolism or other biochemical reactions as external exposure levels of the toxic agent decrease. Linear pharmacokinetic models are often used. However, there are numerous examples of nonlinear behavior in the dose range studied, and these nonlinear kinetics pose significant problems for quantitative extrapolation from "high" to "low" doses if the kinetic parameters are not measured (27-29). 

Pharmacokinetic models involving nonlinear kinetics of the Michaelis-Menten form have the important extrapolation characteristic of being linear at low dose levels. This low dose linearity contrasts with the low dose nonlinearity of the multihit and Weibull models. Each model, pharmacokinetic, multihit, and Ifeibull, has the desirable ability to describe either convex (upward curvature) or concave (downward curvature) dose-response relationships. Other models, stich as the log normal or multistage, are not consistent with concave relationships. However, the pharmacokinetic model differs from the multihit and Heibull in that it does not assume the nonlinear behavior observed at high dose levels will necessarily correspond to the sane nonlinear behavior at low dose levels. 

---