The designation of a dose-response as hormetic may be counter-intuitive because the term is not widely known and may be erroneously linked to hormone-like or homeopathic activities, as well as to a subjective conclusion as to the potential benefits of the effect. The concept of hormesis is also not compatible with the governmental regulatory choice of linearity of response and is therefore politically incorrect. Sagan⁴ considered hormesis as 'any physiological effect that occurs at low doses and which cannot be anticipated by extrapolating from toxic effects noted at high doses'. Another recent definition is a dose-response characterized by a beta-curve, e.g., low doses stimulate while higher doses are inhibitory.⁵ Boxenbaum⁶ proposed a more rigorous definition that 'hormesis is a biological phenomenon in which a beneficial or stimulatory effect is obtained through the application of a nonessential agent generally considered to be detrimental or toxic to the system under scrutiny'. Furthermore, for longevity he distinguished between caloric restriction that is associated with an irreversible longevity enhancement and a reversible effect that is dependent on continuing exposure. Thus, by his definition, caloric restriction is excluded. Turturro et al. have further broadened the debate by suggesting the use of the term 'co-hormesis', i.e., the presence of hormesis in the presence of other modifying factors.

2. What is the base-line for caloric intake?

In clinical medicine and in animal bioassays the 'normal range' or controls reflect the characteristics of the contemporary untreated population rather than those of the population exposed to the environmental conditions that were present during their evolution. Crawford and Marsh⁷ discussed the recent radical changes in human diets away from those associated with our evolution. We now choose diets that are already known to induce a wide range of chronic disease. They described how in 1925 McCarrison compared the effects on rats of the then current Sikh diet with a diet based on the 'unsophisticated foods of nature'. The former was associated with a wide range of chronic diseases and a mortality after two and a half years of 34 percent, compared with a mortality below one per cent in the latter group. This suggests that calories are only part of the problem and that the quality of food is also important. There has been very little research into the food sources that were associated with the evolution of rats and mice or even how the current food sources of wild rat and mouse populations would alter the health and longevity of animals in the laboratory. Current diets were derived from diets developed to ensure growth and reproductive success and fresh plants and vegetables were withdrawn from rats and mice around World War I because of the difficulties of standardization, supply and economics. Thus, the norm should relate to the dietary conditions associated with evolution rather than those dictated by the arbitrary conditions that are imposed on animals in the laboratory. If this is so, then the over-fed, under-exercised animal is itself an animal under experiment rather than a normal control.⁹ Caloric restriction may be returning animals towards this real normal state, so that any consideration of hormesis as a result of dietary restriction in overfed, under-nourished animals would be the antithesis of the natural situation.¹⁰

Another aspect of diet quality that it is usually neglected is that diet content may alter very dramatically the tumor type that results from carcinogen exposure.¹¹

3. What are the parameters for dose- response in dietary restriction experiments?

The majority of caloric restriction experiments rely on a single level of restriction rather than multiple levels or 'doses' of restriction. Linearity is often assumed between the controls and restricted groups, which seems unlikely considering the benefits associated with even a minimal restriction of 10-15 per cent.¹² Thus, there is a dearth of precise data over the whole range of restrictions and other environmental interventions in the same experiment. In an experiment with multiple doses it might be argued that hormesis occurred if a small degree of caloric restriction resulted in a one directional effect and a larger degree produced an effect in the other direction, but this is apparently not the case unless a very high degree of restriction is considered, analogous to a supra- maximum tolerated dose in toxicology.

A surrogate for degrees of dietary restriction that is commonly used is a parameter such as body weight (or Body Mass Index, BMI in humans) even though there is a distribution around the relationship between them and other factors are also important such as genetics and physical activity. Figure 2 of Turturro et al. indicates a U-shaped relationship between body weight and mortality in mice. A similar relationship may exist in humans, although the causes of mortality may vary between the two ends of the curve.¹³

U-shaped- and beta- curves are obviously both non-linear, but suggest differing relationships between the parameters, although hormesis may be present in both cases. Caution is necessary in using an indirect measure of restriction for dose-response modeling. Another important parameter which is often neglected, but was addressed by Turturro et al., is time. Early caloric restriction may leave a life-long imprint, so that differing windows of restriction must also be considered for the dose-response relationship.

4. Modeling of mortality information.

Turturro et al. presented graphical information on a normal-normal scale. It is also useful to model such data using a Gompertzian scale because age-specific mortality increases exponentially.^6,14,15 It may be difficult to estimate changes in mortality rates until relatively late in life unless such models are used, due to the relatively small numbers of animals that are used in experiments and the low background rates that may be present. Boxenbaum⁶ assumed that the Makeham or non-age related parameter was not necessary under laboratory conditions. However, Stevenson et al.¹⁵ showed that even in the ED01 study, there was evidence that early deaths could result in some non-linearity. Given the large changes in background mortality rates during the course of a life-time, it is important to use models that can reflect changes at various time-points. A Gompertz-Makeham power model has been proposed for this purpose, coupled with a residual analysis to define where data departs from the model predictions.¹⁵ Gavrilov and Gavrilova¹⁴ concluded that the Gomertz-Makeham plots for different human populations tend to converge at approximately 95 years of age. However, the plots of calorie unrestricted or restricted rats apparently increasingly diverge with age², emphasizing that the uniqueness of dietary restriction in animals may not be reflected fully in the human experience.

5. Implications of caloric restriction for risk assessment.

In spite of the reservations expressed above for the possible description of hormesis as applied to caloric restriction, the phenomenon described by Turturro et al. has very important implications for health risk assessment. First, dietary restriction may have the beneficial effect of partly or completely blocking the carcinogenic response to chemicals.¹⁶ It has been argued that the overfed undernourished animal equates to the U.S. population. On the other hand, control of diet intake and quality provides a global way of improving health that could not ever be achieved by limiting low level exposures to specific chemicals one by one.

Second, the response of organisms to changes in their environment is a demonstration that organisms are not inanimate, but rather self-organizing dynamical systems with feedback loops that are activated by both external and internal forces.^17,18 Such systems are inherently non-linear, although their responses within certain ranges may be linear as a result of the interactions between opposing forces. Examples of such 'invading' and 'defending' forces with time were modeled by Sielken and Stevenson¹⁹, demonstrating that a wide range of dose-response relationships is possible. A fundamental assumption in the deterministic calculus of Newton is that a body remains internally unaltered when acted on by an external force. This is clearly not the case for biological systems, where the outcome is probabilistic and is not predetermined. Thus, the policy choice of low-dose linearity does not reflect our increasing understanding of the behavior of biological systems.²⁰ The term 'co-hormetic' may be a partial description of the behavior of biological dynamical systems.

Conclusions

The caloric restriction model coupled with other information on nutritional requirements provides valuable insights into the behavior of biological systems, as well as ways that could be used to improve public health if coupled with more attention to the states of nutrition that prevailed during our evolution. However, it is concluded that caloric restriction under the current conditions used to maintain laboratory animals moves the animals towards their 'normal state' rather than departing from it. Using the evolutionary base-line, the response to caloric restriction may not be considered hormetic. However, in considering responses from current controls it may be useful to consider the implications of hormesis. Current knowledge suggests that while humans also benefit from caloric restriction, it is unlikely that this intervention will substantially extend life-span beyond the currently accepted maximum life-span, although it will increase the period of healthy life within the allotted span of life.

REFERENCES

1. Cutler RG. Antioxidants, aging and longevity. Free Radicals in Biology 1984; 6:371-428.

2. Sacher GA. Life Table Modification and Life Prolongation. Handbook of the Biology of Aging. C.E. Finch and L Hayflick (Eds.), pp. 582-638. van Nostrand Reinhold Company, New York, 1977.

3. Brody S. Bioenergetics and Growth: With Special Reference to the Efficiency Complex in Domestic Animals. Hafner Press, New York, 1974.

4. Sagan LA. What is hormesis and why haven't we heard about it before? Health Physics 1987; 52:521-535.

5. Calabrese EJ, Baldwin LA. Chemical hormesis: Its historical foundations as a biological hypothesis. Toxicologic Pathology 1999; 27:195-216.

6. Boxenbaum H. Hypotheses on Mammalian Aging, Toxicity, and Longevity Hormesis: Explication by a Generalized Gompertz Function. Biological Effects of Low Level Exposures to Chemicals and Radiation, E. Calabrese (Ed.). Lewis Publishers, Boca Raton, FL, 1992.

7. Crawford M, Marsh, D. Nutrition and Evolution. Keats Publishing Inc. New Canaan, Connecticut, 1995.

8. Holehan AM, Merry BJ. Experimental manipulation of aging by diet. Biological Reviews 1986; 61:329-368.

9. Keenan KP. The uncontrolled variable in risk assessment: Ad Libitum overfed rodents- fat, facts and fiction. Toxicologic Pathology 1996; 24:376-383.

10. Cutler RG. Life Span Extension in Aging: Biology and Behaviour. McGaugh JL, Kiesler SB. (Eds.). Academic Press, New York, 1981.

11. Schmahl D, Habs M, Wolter S, Kuenstler K. Experimental investigation of the influence upon chemical carcinogenesis: 4^th communication - influence of different diets on colon carcinogenesis by 1,2-diemethylhydrazine in Sprague-Dawley rats. Journal of Cancer Research in Clinical Oncology 1979; 93:57-66.

12. Roe FJC. Are nutritionists worried about the epidemic of tumours in laboratory animals? Proceedings of Nutritional Society 1981; 40:57-65.

13. Spirduso WW. Physical Dimensions of Aging. Human Kinetics, Champaign, IL, 1995.

14. Gavrilov LA, Gavrilova NS. The Biology of Life Span: A Quantitative Approach. Harwood Academic Publishers, New York, 1991.

15. Stevenson DE, Brezlaff RS, Sielken RL, Jr, MacDonald RL. Dose-response characterization of life, death and hormesis. Comments Toxicology 1994; 5:151-180.

16. U.S. Department of Health and Human Services. National Toxicology Program Report on the Effects of Dietary Restriction on Toxicology and Carcinogenesis Studies in F344/N Rats and B6C3F1 mice. NTP TR 460; NIH Publication No. 95-3376, 1995.

17. Kauffman S. At Home in the Universe: The Search of the Laws of Self-Organization and Complexity. Oxford University Press Inc. New York, 1995.

18. Capra F. The Web of Life. Anchor Books Doubleday, New York, 1996.

19. Sielken RL, Jr, Stevenson DE. Modeling to incorporate defense mechanisms into the estimation of dose responses. Environmental Health Perspectives 1998; 106:341-348.

20. Sielken RL, Stevenson DE. Some implications for quantitative risk assessment if hormesis exists. Human & Experimental Toxicology 1998; 17:259-262.