Rodent Cancer Bioassays - Is Body Weight Depression Hormetic?

George M. Gray, Ph.D.

Harvard Center for Risk Analysis, Harvard School of Public Health, 718 Huntington Ave., Boston, MA 02115

Tel: 617-432-4341, Fax: 617-432-0190


It is clear that caloric restriction has many positive effects on the health and longevity of experimental animals. Turturro et al. present a clear and logical discussion about the relationship between caloric restriction and hormetic responses. I believe a point made by Turturro et al. that requires further exploration is the relationship between caloric restriction, body weight reductions, and possible hormesis in experimental settings. This is of great importance in the interpretation of rodent carcinogenesis bioassays, most of which demonstrate significant body weight decrements in chemically-treated animals. Although no one would consider the doses used in standard rodent cancer bioassays to be low, it is possible that the effects of chemical treatment on body weight could induce hormetic effects as defined by Turturro et al.. This might be a different type of response than those usually observed at lower doses (Calabrese et al., 1999)

Although sometimes used synonymously with caloric restriction, body weight reduction is only a biological marker of caloric restriction and can occur through other mechanisms than restricting animals' access to food. The question of whether body weight reduction caused by mechanisms other than caloric restriction could be hormetic is critical to understanding cancer bioassays. Turturro et al. note some mechanisms of chemically-induced body weight reduction (like increased physical activity or temperature dysregulation) that mimic caloric restriction and would be expected to be hormetic. An important question is whether all modes by which body weight is reduced would also be hormetic.

In thinking about other chemically-induced changes in body weight, one can imagine three broad types, each with different implications for hormetic potential. First is body weight reduction due to "pure caloric restriction" induced by changes in food intake due to reduced palatability of food. This would only occur when animals are dosed in feed, a relatively uncommon route of exposure in rodent cancer bioassays. This would certainly be hormetic as defined by Turturro et al..

A second type of body weight reduction would induce "quasi-caloric restriction" and would likely be hormetic by the Turturro et al. definition. In this case the beneficial effects of body weight depression would outweigh the adverse effects of chemical treatment.

The third type would be "toxic" body weight reduction in which any beneficial effects of decreased body weight are overwhelmed by toxicity. This balance would likely operate in a dose-related manner with benefits (hormesis) occurring until the dose level at which the toxicity becomes dominant. In this case, decreased body weight would not be expected to be associated with hormetic responses.

An interesting possibility that arises with chemically-induced body weight decreases is site-specific hormetic effects. If hormesis due to body weight decrements from chemical treatment is actually competing against toxicity from the chemical, it would be expected that different organs would have different sensitivities to the toxin and different points at which benefits are overcome by toxicity. Of course, there may also be different hormetic relationships with body weight reduction in different organs.

We have studied responses in rodent cancer bioassays in the CBDS database of studies conducted by the National Cancer Institute and National Toxicology Program up to 1983. Two recent publications focused on anticarcinogenicity, site-specific decreases in tumor rates following chemical treatment (Linkov et al., 1998a; 1998b; 1999). It is possible that these are hormesis-like responses, demonstrating lower rates of disease in treated animals than in controls.

Figure 1 is a cumulative distribution function of percent average weight decrease in all dosed groups versus control animals at one year of age for the 312 chemicals in our database. The analysis separates those chemicals that were anticarcinogenic in at least one site from those that demonstrated no anticarcinogenicity (at p< 0.025). There are several things to note. First, roughly half of all chemicals cause body weight decrements of 10% or greater in treated animals compared to controls. Clearly, weight decreases are common in these studies. Second, there are differences in weight decrease between anticarcinogens and non-anticarcinogens, but the magnitude is not large. There are chemicals that clearly reduce body weight but are not anticarcinogenic. In addition, there are a number of anticarcinogens that are not associated with weight decreases at one year.

Figure 1. Cumulative Distribution Function of Body Weight Depression for Chemicals Anticarcinogenic in at Least One Site of Mice or Rats.

These data do not allow us to cleanly evaluate the three types of body weight reduction described above. If hormetic responses are the cause of anticarcinogenicity then the category of non-anticarcinogens that induce weight decrements would be examples of "toxic" body weight reductions with no accompanying hormetic effect. The anticarcinogens would be made up of both "true caloric restriction" and "quasi-caloric restriction" and these data don't let us tell them apart. The CBDS database does contain some data on food consumption and further analysis could separate these two types of body weight reductions.

It is important to note that many chemicals classified as anticarcinogens also have carcinogenic effects. For mice, 49% of chemicals demonstrating anticarcinogenicity also had carcinogenic effects at another site (Table 1). In rats, 60% of anticarcinogenic chemicals were carcinogenic as well (Table 2). This could be a manifestation of site-specific body weight related hormesis along with site-specific tumor induction.

Table 1 and Table 2

The effects of changes in body weight on tumor responses are important for interpretation of rodent bioassays. For example, if anticarcinogenicity is a hormetic effect induced by body weight reductions, then tumor responses in treated animals might be lower than they would be in the absence of depression of body weight and resulting estimates of carcinogenic potency might be too low. Similarly, findings of anticarcinogenicity at high doses would not be relevant for lower doses that show no change in weight. Our interpretation of Figure 1, and other data in Linkov et al. (1998b) is that body weight changes are not the sole cause of, or even a large contributor to, anticarcinogenicity. Given the high rates and levels of weight reduction in the bioassays, it is possible that the high doses of chemical administered so alter normal physiology that few responses are relevant to lower doses.

Further work in this area is clearly needed. Comparisons of studies from ad libitum and restricted feeding would help sort out the effects of chemical administration and weight reductions. An interesting difficulty that would arise is animals on dietary restriction living longer and having a greater opportunity to develop tumors.

I applaud Turturro et al. for their work in this area and BELLE for encouraging an active exchange of ideas.


Calabrese, E.J., Baldwin, L.A., and Holland, C.D. (1999). Hormesis: A Highly Generalizable and Reproducible Phenomenon with Important Implications for Risk Assessment. Risk Analysis 19:261-281.

Linkov, I., Wilson, R., and Gray, G.M. (1998). Anticarcinogenic Responses in Rodent Bioassays are Not Explained by Random Effects. Toxicological Sciences 43:1-9.

Linkov, I., Wilson, R., and Gray, G.M. (1998). Weight and Survival Depression in Rodent Bioassays With and Without Tumor Decreases. Toxicological Sciences 43:10-18.

Linkov, I., Wilson, R., and Gray, G.M. (1999). Erratum to Toxicological Sciences 43:1-9 and Toxicological Sciences 43:10-18. Toxicological Sciences 48:141-142.

Turturro, A. Hass, B.S., and Hart, R.W. (1999). Does Caloric Restriction Induce Hormesis? BELLE Newsletter.