Letter

Melvin E. Andersen1, Rory B. Conolly1 and David W. Gaylor2

1 CIIT Centers for Health Research 6 Davis Drive Research Triangle Park, NC 27709 2 13815 Abinger Court Little Rock, AR 72212

To the Editor:

We are responding to Waddell‘s article, "Thresholds of Carcinogenicity in the ED01 Study" (2003). This article evaluates dose-response relationships for 2-acetylaminofluorene (2-AAF) carcinogenicity. While the paper claims to provide evidence for thresholds, it uses an inappropriate dose-response model and draws unsubstantiated conclusions about the shapes of these curves.

Tumor response data are plotted against the logarithm of the daily dose (as molecules/kg/day) and fit to a linear function after discarding low dose data that do not fit a line based on the high dose data. The point on the dose axis where this function intersects zero-response is regarded by the author as a measure of a "threshold".

This response model is linear with log (dose),

When response is 0.0,

Thus,

In this straight-line model, response is unbounded, varying between infinity and minus infinity. This behavior is unrealistic.

Pharmacology and toxicology have many well-established approaches for assessing dose-response relationships. These include (1) probit analysis based on a distribution of sensitivities toward toxicity in a population, (2) exponential forms including multistage cancer models, and (3) maximal response models, such as Michealis-Menten or Hill-relationships, bounded to 100% as the upper limit of response. Any of these can be used to describe the 2-AAF cancer response relationships and the models could be evaluated to assess goodness of fit of the models to the data. All of these models predict ranges between 0 and 100 % response. The "straight line through the points" approach lacks a biological foundation and the analysis of fit of the model to data would have to include the deviations in regions where this model predicts response much lower than 0 and much greater than 100%.

Many toxicologists believe that there are dose thresholds for most adverse responses. We share this belief. However, thresholds will not be proven by empirical modeling of dose-response data or by developing data sets with larger numbers of animals per group. For example, even with no tumors present in 1,000 animals, one can only state with confidence that the true tumor level is less than 0.45%. Support for the existence of thresholds will depend on increasing appreciation of the biological mechanisms by which chemicals cause adverse responses, including the roles of homeostatic and adaptive determinants of low-dose dose-response. With 2-acetylaminofluorene, carcinogenic responses are believed to arise from promutagenic DNA-adducts derived from 2-AAF-metabolites and the possibility of high dose cellular toxicity of reactive metabolites. The interactions of these two modes of action—direct genotoxicity and cell toxicity and proliferation—are expected to determine the overall shape of the incidence dose curves for bladder and liver cancer with this compound. At low doses of 2-AAF, cellular adaptive processes such as cell cycle checkpoint control may also affect the shape of the curve, leading to thresholds.

The author also argues that chemical potential is a universal dose metric for assessing toxic responses. In the past three decades, increasing emphasis has been placed on use of pharmacokinetic models to estimate target tissue doses for dose-response analyses. In most instances, the preferred dose measures at target tissues are related to intensity of interactions at the target tissues, i.e., concentration, area under tissue curves for parent chemical or metabolite, receptor binding, or tissue reactivity. Modes of action determine the preferred metric for assessing target tissue toxicity. There is always room for discussion about the best measure of target tissue dose. There is, however, no reason to think that chemical potential of the administered compound is a universally correct representation of tissue dose in toxicology studies.

It is always important to separate what we believe, i.e., the existence of thresholds, from what we can say with confidence based on careful analysis of specific data sets using appropriate dose-response analyses. In this regard, the paper does not provide a meaningful contribution to the debate on the presence or absence of thresholds for toxicological responses or to further development and refinement of tools for dose-response evaluations.

REFERENCE

Waddell, W. J. (2003). Thresholds of carcinogenicity in the ED01 study. Toxicol. Sci. 72, 158–163.[Abstract/Free Full Text]