Hormesis, Regulation, Toxicity and Risk Assessment

Karl K. Rozman, Ph.D.

University of Kansas Medical Center


3901 Rainbow Blvd.

Kansas City, KS 66160-7417

Phone: 913-588-7717

Fax: 913-588-7501

Email: krozman@kumc.edu

John Doull, Ph.D., M.D.

Dept. of Pharmacology, Toxicology and Therapeutics

University of Kansas Medical Center

Kansas City KS 66160-7417

Phone: 913-588-7140

Fax: 913-588-7501

Email: jdoull@kumc.edu

This article was written in response to an invitation to address the question of how regulatory and public health agencies should deal with the biological effects of low level exposures. Our response focuses on the interactions between toxicity and hormesis and on the role of hormesis in risk assessment and in regulation.


The potential impact of hormesis on regulation has been considered in many of the articles that have appeared in the Belle newsletters over the past six years and in two issues (1,2) that were devoted entirely to this topic. One of the most frequent observations from these articles is that because thresholds are inherent in U-shaped dose response curves, the linear no-threshold extrapolation method is not an appropriate approach for regulating hormetic agents. In addition, the collateral argument that the threshold or reference dose (RfD) approach should be used in any situation (including carcinogensis) where hormesis occurs provides yet another argument which supports recent suggestions (3,4) that it is time to do away with or at least re-evaluate the use of the linear low-dose approach to regulate carcinogens.

There is, however, an additional and important conclusion regarding public health benefits which can be made for agents exhibiting hormesis. This is that approaches which rely on banning or attempts to establish zero dose exposure limits should not be used to regulate toxic agents which exhibit hormesis or beneficial effects at sub-toxic doses. Such agents should be regulated by setting tolerances or standards which will preserve the beneficial public health effects while avoiding the toxic effects of the agents. It would be preferable in fact, if we had adequate benefit data, to base all regulatory actions on such a risk/benefit analysis rather than on the current one-sided approach of using risk assessment. Hormesis and U-shaped response curves could be pivotal in moving regulatory agencies in this direction and they may also be powerful tools for enhancing public understanding of dose response as a toxicologic and regulatory principle. We could also enhance public understanding of the importance of dose response in risk assessment by presenting adverse effect dose information in terms of the number of molecules required to produce the effects (Figures 1, 2) since such figures clearly illustrate the relationship between the exposure dose and a zero dose. The dose intercept in these figures is actually at one molecule which is the smallest unit that can have biological activity rather than at zero.

Figure 1. Molecular scale for the presentation of EPA's 10-6 risk, various human exposure scenarios, and animal toxicity data (acute toxicity, cancer) of tetra-CDD. (Ref. 16)

Figure 2. Dose responses on the molecular scale for the acute toxicity of tetra-and octa-CCD, and for the carcinogenicity of tetra-CDD, aflatoxin, benzo[a]pyrene, and furan. (Ref. 16)


The practical significance of hormesis both in regulation and in science generally will depend in part on whether it is an isolated/unusual phenomenon or a general rule of biology. Using the definition that hormesis is the stimulatory action of sub-inhibitory amounts of a toxin, Calabrese and Baldwin (5) have compiled an extensive data base which demonstrates that hormesis occurs in a variety of species (microbes, plants and animals including humans), with a broad range of chemical classes and that it can involve many different endpoints such as growth, physiologic or metabolic change, longevity, reproductive activity etc. Davis and Svendsgaard (6) have also compiled a list of examples of hormesis but they suggested that the effects of drugs, vitamins, nutrients, essential trace elements etc. not be included as examples of hormesis in spite of the fact that they are clearly examples of the beneficial effects of sub-toxic doses of chemicals. These authors also suggested that Paracelsus's dictum on dose response "provides no specific guidance for toxicological risk assessment". As indicated above and elsewhere (7) it is our opinion that Paracelsus's principles of dose response provide scientific and logical guidance for assessing risk and the purpose of this paper is to suggest that these principles apply equally well to hormesis.

Most chemicals exhibit more than one type of effect and these effects are dose related and usually have different mechanisms of action. With dioxin (TCDD), for example, increasing dose and time produce sequentially; enzyme stimulation, endocrine effects, wasting/hemorrhage, anemia, pulmonary tumors and liver tumors. Designation of low dose effects as stimulatory or beneficial and high dose effects as inhibitory or adverse is somewhat arbitrary and does not provide a consistent definition for hormesis. Sagan's definition of hormesis as a paradoxical or unanticipated effect at low doses (8) avoids this problem and may be helpful in distinguishing between dose related effects of the agent and different responses of the host. In either case, the outcome would fulfill Sagan's definition of hormesis and also satisfy the Arndt-Schultz Law of hormesis which states that "poisons in low doses stimulate biological processes while being inhibitory at high doses" which is identical in concept with the dictum of Paracelsus.

Graphically U-shaped response curves can be considered to consist of decending and ascending dose response curves. Similarly the hormesis beta curve (Figure 3) can be divided into three parts; a stimulatory phase (A), antagonism of the stimulatory effect (phase B) and an inhibitory phase (C) which may be an extension of phase B. Each of these phases can be represented by a separate dose response curve (Figure 4) having different mechanisms of action although phases B and C could have the same mechanism of action. The no-effect-level or NOEL in a U-shaped response curve is at the bottom of the trough and thus the NOEL in the beta curve of Figure 3 should be at the maximum response of phase A (beginning of the antagonism phase B) rather than at the point where the response returns to the control level (the beginning of phase C). Using this type of dose-response interpretation for hormesis establishes the link between hormesis and toxicity and by assuming different mechanisms of action for the stimulatory and antagonism phases of the response, it avoids the need to assume that increasing doses of any agent could produce both an increase and a decrease in the same biological response which is mechanistically unlikely. Using multiple dose responses to describe hormesis broadens the role of hormesis in biology, results in the inclusion of more types of agents (drugs, nutrients, essential metals, oxygen etc.), extends the concept of hormesis to other disciplines and will encourage the use of testing methodology appropriate for detecting hormesis. Selecting dose ranges that are capable of detecting all of the effects of agents rather than ranges which focus only on the high dose adverse effects will provide a more complete characterization of the toxicity of an agent which in turn will improve our ability to carry out an informed and appropriate risk assessment.

Figure 3. Dose response curve depicting characteristics of the chemical hormesis zone. (Ref. 1)

Figure 4. Dose response curves for typical hormesis beta curve. (Fig. 3)


Toxicity is dependent on the chemical or AGENT, the SUBJECT or target and on the conditions of EXPOSURE. Because of its relationship to toxicology, risk assessment is also dependent on the type and severity of adverse effects produced by the chemical or AGENT, the susceptibility of the SUBJECT or population and on the EXPOSURE scenario. To characterize the toxicity of an agent, we need (a) information about all the adverse effects that the agent can produce in any species, by any route or rate of exposure and (b) dose response information for each of these effects. To obtain this information we use toxicokinetics which is the study of the effects of the subject on the agent or chemical and toxicodynamics which is the study of the effects of the chemical or agent on the subject. These two approaches are linked by the exposure which includes both the dose or concentration of the agent and the time of exposure (duration, frequency, rate etc.). Traditionally, risk assessment has used this information to predict a NOEAL (no observable adverse effect level) which was then divided by appropriate safety factors to arrive at an ADI (acceptable daily intake) or RfD (reference dose). This approach was not used for cancer since the clonal nature of malignant tumors suggested that a single hit on the DNA could lead to a heritable transformation of a cell which in turn could lead to cancer. Studies by Druckery (9) and others (10) indicated that the latency period of cancer sometimes obeyed the rule that ct = constant but more often ctx = constant. From this, it was concluded that cancer has no threshold since even the smallest dose would cumulate (11).Use of the no threshold approach for carcinogens gained momentum during the early 70s and culminated in the linearized multistage cancer risk assessment (12) which made the additional assumption that time was not a variable at low doses. Although not always explicitly stated, time has always been implicit in toxicology and there is currently a growing effort to re-establish time as one of two major variables in toxicology (13,14).

Toxicity is a function of exposure and exposure is a function of dose and time and this concept provides the basis for a decision tree type of approach for assessing risk which does not depend on defaults or assumptions (15). The first step in this approach is to define the limits of the response. If there is no exposure, there is no toxicity and conversely with continuous exposure, the worst case situation occurs when the product of the dose and time is a constant (ct=k). This is a true worst case since it is defined by the actual exposure conditions required to produce the adverse effect rather than by hypothetical conditions or statistical limits. Thus for a specific type of toxicity or adverse effect, the worst case occurs when the threshold for injury is exceeded and there is no recovery (from repair, reversibility or adaptation) or when there is absorption but no elimination (from distribution, metabolism or excretion). Toxicologists will recognize ct=k as Haber's Law but may fail to recognize that steady state conditions for toxicodynamics (recovery equals injury) or toxicokinetics (elimination equals absorption) are needed for this expression to correctly predict toxicity. Since Haber's Law predicts toxicity rather than exposure, the prefix toxico is used for the kinetic and dynamic processes in this description rather than pharmaco which is more appropriate for the kinetic and dynamic processes related to exposure. Constants determined from Haber's Law for specific adverse effects or types of toxicity, can be used to predict the steady state conditions of dose and time that will produce different levels of toxicity or severity of the adverse effect in an individual or population and a plot of the log dose versus log time provides a linear transformation of ct=k with a slope of 1. A more general expression of this concept is provided by the following;

dT = dT . dD . dK

dE = dD dK dE

D=toxicodynamic processes
K=toxicokinetic processes

Exposure scenarios that deviate from the continuous exposure-worst case situation will be more complex (ctx =k) but can be used for predicting levels of toxicity by identifying the rate- determining and/or rate -limiting steps in either the toxicokinetic or toxicodynamic processes. For example, if the recovery half life of the response is greater than the elimination half life of the agent, then toxicodynamics becomes rate-determining (one-compartment model) or rate-limiting (multi-compartment model). Conversely when the elimination half life of the agent is longer than the recovery half life of the effect, then toxicokinetics becomes rate-
determining or rate- limiting and the toxicokinetic AUC (area under the curve) and toxicodynamic AUC will be identical after both are at steady state. Interactions at the molecular and cellular levels and higher will propagate through the toxicokinetic or toxicodynamic causality chains all the way to the manifestations of toxicity in the subject. By determining which of these are rate-determining and/or rate-limiting this information can be used to predict toxicity in a more precise and less uncertain manner than is provided by any of the current cancer or non-cancer risk assessment procedures. This approach can also be used for other types of risk evaluation. The margin of exposure for a specific adverse effect, for example, could be expressed in terms of the actual exposure (ctx =k) divided by the worst case exposure (ct=k). Since exposure can occur without toxicity (when the lifetime exposure does not exceed the threshold of injury) we need to define margins of toxicity or safety in terms of the specific type of toxicity or adverse effect and the dose/time threshold values for these effects.

The interpretation of beneficial effects at sub-toxic doses or hormesis in risk assessment depends both on the type of effect and on the mechanisms involved. Hormesis could result from toxicokinetic (less absorption or more elimination) or toxicodynamic (less injury or enhanced recovery) effects at low levels of exposure which could antagonize the toxic effects of higher dosage levels of the agent. Another hypothesis which has been suggested for radiation and some other toxic agents is that low levels of exposure act by stimulating defense mechanisms such as enhanced cell turnover, DNA repair, apoptosis, enzyme induction, immune stimulation etc. which permit the host to withstand subsequent toxic exposures. Similar arguments can be made for physiological hormesis such as occurs with exercise, pharmacologic effects such as allergy, sensitization, anaphylaxis etc. and other types of adaptation. As described previously, adaptation is one of the toxicodynamic recovery processes for reducing injury and its effects can be either therapeutic or prophylactic. In either case, however, the hormetic mechanism results in changes in the susceptibility of the subject rather than in changes in the potency of the agent.


1. Belle Newsletter, 1998,Vol 6 No 3, Jan.

2. Belle Newsletter, 1998, Vol 7, No 1, May.

3. Clayson, D. B.,1998, Is the Use of the Linear Low-Dose Extrapolation Still Justified for Carcinogens?, Reg. Tox. & Pharm., 28; 69-70

4. Sielken, R. L., Bretzlaff, R. S. and Stevenson, D. E., 1995, Challenges to Default Assumptions Stimulate Comprehensive Realism as a New Tier in Quantitative Risk Assessment, Reg. Tox. And Pharm., 21: 270-280

5. Calabrese. E. J., and Baldwin, L. A.,1998, Chemical Hormesis: Scientific Foundations, Final Report. School of Public Health, Univ. Of Massachusetts, Amherst, MA

6. Davis, J. M. and Svendsgaard, D. J.,1990, U-Shaped Dose Response Curves, J. Tox. and Environ. Health, 30: 71-83

7. Rozman, K. K. and Doull, J.,1998, General Principles of Toxicology, Chapter 1 in Environmental Toxicology. Current Developments, (J. Rose, Ed.), p 1-11, Gordon and Breach Sci. Publ. Amsterdam.

8. Sagan, L., 1993,A Brief History and Critique of the Low Dose Effects Paradigm, Bell Newsletter, Vol 2, No 2 Dec.

9. Druckery, H.,1967, Quantitative Aspects in Chemical Carcinogenesis In Potential Carcinogenic Hazards from Drugs, Evaluation of Risk. (R. Truhaut Ed.), Vol 7, pp 60-78, UICC Monograph Series, Springer Verlag, Berlin.

10. Peto, R., Gray, R.,. Branton, P and Grasso, P., 1991, Effects on 4080 Rats of Chronic Ingestion of N-nitrosodimethylamine or N-nitrosodiethylamine, a Detailed Dose Response Study. Cancer Research, 51, 6415-6451

11. Suss, R., Kinzel, V. and Scribner, J. D., 1973, A Closer Look at Chemical Carcinogenesis: Quantitative Aspects, in Cancer Experiments and Concepts, Springer-Verlag, New York, NY.

12. Crump, K. S., Hoel, D. G., Langley, C. H., and Peto, R., 1976, Fundemental Carcinogenic process and their Implications for Low-Dose Risk Assessment, Cancer Research, 36, 2973-2979

13. Rozman, K. K., 1998, Quantitative Definition of Toxicity: A Mathematical Description of Life and Death with Dose and Time as Variables, Med. Hypothesis; 51 175-178

14. Rozman, K. K., 1999, Delayed Acute Toxicity of 1,2,3,4,6,7,8-Heptachlorodibenzo-p-dioxin (HpCDD), After Oral Administration, Obeys Haber's Rule of Inhalation Toxicology, Toxicolog. Sciences: 46, 102-109

15. Rozman, K., 1999, Dose and Time as Variables of Toxicity, Working paper for NAS/NRC Project on Strategies to Protect the Health of Deployed U. S. Forces, Task 2.1 Assessing Health Risks During Deployments in Hostile Environments, In preparation

16. Rozman, K.K., Kerecsen, L., Viluksela, M.K., Osterle, D., Deml, E., Viluksela, M., Stahl, B.U., Greim, H., and Doull, J. 1996, A Toxicologist's View of Cancer Risk Assessment. Drug Metabolism Reviews, 28(1,2):29-52.