Reply

William J. Waddell

14300 Rose Wycombe Lane Prospect, KY 40059 Fax: 502-228-6779 E-mail: bwaddell{at}louisville.edu

To the Editor:

Crump and Clewell (2003)Go plot some hypothetical data over a narrow range and contend that it appears to have an intercept of zero tumors with zero dose when dose is on a linear scale. Contrary to their contention, the example proves absolutely nothing; instead, it demonstrates why the linear plot for dose, with a forced fit through the zero, zero origin, is highly flawed. This is an artifact commonly seen in the literature from plotting dose on a linear scale; it serves to illustrate why the linear scale has been deceptive for so long to the toxicology community. By choosing a linear scale that truncates and compresses lower doses, the shape of the curve at these lower doses cannot be evaluated. Note in their linear graph that only doses from 1018 to 1019 are represented by the line. Furthermore, all doses 18 orders of magnitude below this (which is where our interest in extrapolation lies) are compressed into about 1/4 inch of abscissa. The compression of low doses can be so severe that the shape of the curve and its possible intercept with the abscissa are impossible to evaluate. A side-by-side comparison of a wider spectrum of data on a linear scale versus a logarithmic scale can be seen in Figure 4 of the Rozman et al.(1996)Go paper. That figure also shows a curvature of response, and difficulty of evaluation, at low doses when evaluation is done on a linear scale. The log-linear plot does not, in and of itself, always show a threshold; however, when it does, a threshold is the correct interpretation.

The criticism by Crump and Clewell of using the entire logarithmic spectrum from one molecule to 1023 molecules is that it makes it look like there is a steep slope. I am sure they are fully aware that the regression statistics will not change one iota whether the scale displayed is over this wide range or merely includes one or two orders of magnitude. It is the regression statistics that matter and not the visual inspection of steepness. Use of the entire spectrum, down to one molecule, however, places all doses in perspective and allows comparison with the doses to which humans are exposed.

The main point of my paper appears to have escaped the attention of Crump and Clewell. I chose a log-linear plot because that is the only plot that is consistent with fundamental principles of chemistry, namely the thermodynamic concept of chemical potential and the law of mass action (Waddell and Bates, 1969Go). This is at least an attempt to base the selection on sound science; its successful application is also currently being reported elsewhere (Waddell, 2002Go, 2003aGo, 2003bGo, 2003cGo, in pressGo). Extrapolations and risk assessments based merely on whims and unsupported assumptions have delayed our understanding of chemical carcinogenesis for decades, even though gross inconsistencies previously have been noted between linear extrapolations and the epidemiology of cancer (e.g., Rozman et al., 2001;GoWaddell, 1993Go, 1996Go, in pressGo) and many other authors. It is now time for a complete, objective reassessment of how animal experiments are evaluated and interpreted.

REFERENCES

Crump, K. S., and Clewell, H. J. (2003). Letter to the Editor. Toxicol Sci. 74, 485.[Free Full Text]

Rozman, K. K., Doull, J., and Hayes, Jr., W. J. (2001). Dose, time and other factors influencing toxicity. In Handbook of Pesticide Toxicology, Vol. 1 (R. Krieger, Ed.), pp. 1–93. Academic Press, San Diego.

Rozman, K. K., Kerecsen, L., Viluksela, M. K., Österle, D., Deml, E., Viluksela, M., Stahl, B. U., Greim, H., and Doull, J. (1996). A toxicologist’s view of cancer risk assessment. Drug Metab. Rev. 28, 29–52.[ISI][Medline]

Waddell, W. J. (1993). Human risk factors to naturally occurring carcinogens: Thresholds, inhibitors and inducers. J. Toxicol. Sci. 18, 73–82.[Medline]

Waddell, W. J. (1996). Reality versus extrapolation: An academic perspective of cancer risk regulation. Drug Metab. Rev. 28, 181–195.[ISI][Medline]

Waddell, W. J. (2002). Thresholds of carcinogenicity of flavors. Toxicol. Sci. 68, 275–279.[Abstract/Free Full Text]

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

Waddell, W. J. (2003b). Thresholds in chemical carcinogenesis: What are animal experiments telling us? Toxicol. Pathol. 31, 1–3.[ISI][Medline]

Waddell, W. J. (2003c). Threshold of carcinogenicity of nitrosodiethylamine for esophageal tumors in rats. Food Chem. Toxicol. 41, 739–741.[CrossRef][ISI][Medline]

Waddell, W. J. (in press). Comparison of human exposures to selected chemicals with thresholds from NTP carcinogenicity studies in rodents. Hum. Exp. Toxicol.

Waddell, W. J., and Bates, R. G. (1969). Intracellular pH. Physiol. Rev. 49, 285–329.[Free Full Text]