Challenging Dose-Response Dogma

The central pillar upon which toxicological assessments are built is the dose-response relationship.

Edward Calabrese(edwardc@schoolph.umass.edu)
Feb 13, 2005
<p>FIGURE 1.</p>

Courtesy of Edward J. Calabrese

Stylized curves that illustrate the linear, threshold, and hormetic dose-response models for carcinogens.

The central pillar upon which toxicological assessments are built is the dose-response relationship. But reliance on theoretical prediction models for dose response creates a wolf-crying atmosphere that generates fear and superfluous costs to the public. A better alternative exists, however. A rich vein of data supports hormetic dose-response modeling, which allows for the observation that some toxins, in small amounts, confer benefits rather than harm.

Having been relegated for years to the toxicological waste heap by misconception and inertia, hormesis is regaining respect, and I believe that such modeling will replace the outmoded standards in toxicology and may ultimately influence most areas of biological research.1

Since the consolidation of toxicological and pharmacological conceptual thinking in the 1930's, the overwhelmingly accepted dose-response model has been the threshold model, which assumes...

THE HORMETIC CHALLENGE

<p>Edward J. Calabrese</p>

Courtesy of Edward J. Calabrese

The hormetic dose response challenges such long-standing toxicological dose-response model mindsets. Hormetic dose responses are biphasic, displaying either an inverted U- or J-shape depending on the endpoint measured (Figure 1). It's generally recognized, for example, that adults who consume a glass of wine most days have reduced risk to cardiovascular disease compared to nondrinkers, while excessive consumption increases such risks. This type of J-shaped dose response is now known to be quite common in toxicology and pharmacology, being seen with many dozens of chemicals, and for hundreds of important endpoints such as cancer risks, longevity, growth, performance on various types of intelligence tests, and more.3

Comprehensive assessments of the literature have shown the hormetic model to be biologically more fundamental than either major dose-response rival, more common in valid head-to-head comparisons, and generalizable across biological model, endpoint measured, and the chemical class or physical agents studied. Based on these conclusions I have argued that the hormetic dose-response model should replace both the threshold and linear models as the default model in risk assessments for noncarcinogens and carcinogens.2 Further, the hormetic dose response should be considered not just the dominant model in toxicology but also in the broader domain of the biomedical sciences including immunology, cancer cell biology, neuroscience, and all other fields that rely upon dose-response relationships.

Ultimately, the challenge that the hormetic model presents to the biomedical communities, including toxicology, is nothing less than a scientific revolution. It changes the understanding of how biological systems deal with low levels of chemical/physical agents. It should alter how studies assessing the dose response should be designed with respect to the number, size, and spacing of such doses, and the distribution of subjects within these and other frameworks. It should cause regulators to reevaluate how risk assessments are conducted, and how medical dosing should be optimized.

The hormesis concept also changes the way society might think about contaminants and drugs. A toxicant that enhances tumor formation at high doses may affect a reduction in tumor incidence at lower doses. For example, in studies with the rat model used by the US National Toxicology Program, DDT has been reliably demonstrated to reduce tumor incidence significantly below that of the control group at low doses while being a carcinogen at higher doses (Figure 2).4

This information is common within the Hormesis Database that we've developed at the University of Massachusetts.5 Such accumulated data from the peer-reviewed biomedical literature provide a challenge to the scientific and regulatory communities concerning how these data should be integrated into political, regulatory, and educational activities. The challenge extends to pharmaceuticals such as antitumor drugs. At high concentrations they inhibit tumor growth, while at lower concentrations stimulation of tumor growth may occur. Such possibilities have important implications dealing with not only drug design and testing but also for clinical management of cancer.

TOXICOLOGICAL BLUNDERS

<p>FIGURE 2.</p>

Courtesy of Edward J. Calabrese

Effect of DDT on number of GST P-positive foci in F344 rat livers in two bioassays assessing different but slightly overlapping doses of carcinogen. As the dose decreases the J-shaped dose-response becomes evident. Also note difference in scale between the two graphs. (Based on T. Sukata et al., Int J Cancer, 99:112–8, 2002).

Why the hormetic dose response has been either unknown or ignored within the biomedical and toxicological communities is a significant issue. Its absence from the toxicological literature during the past century resulted from a complex set of interacting factors. On the scientific side, the principal reason is that hormesis can be difficult to detect because the magnitude of the stimulation is modest, being only about 30%-60% greater than control at maximum. This modest response could readily be dismissed as normal variation unless the experimental design has a sufficient number of properly spaced doses. But since toxicology has long been a high-dose/few-doses discipline, based on the overriding belief in the threshold model, it does not usually explore possible hormetic responses.

Now institutionalized within the EPA and FDA, industrial and academic researchers have little incentive to explore beyond the high-dose/few-doses paradigm. In addition, researchers seek out animal models that display low background disease incidence for statistical reasons (i.e., to keep sample size low). But hormetic levels cannot be observed within a negligible background.

On the political side, the hormesis concept, immediately upon its discovery in the 1880s, became closely but incorrectly associated with the medical practice of homeopathy, becoming a victim of collateral damage in a long-standing and intensely bitter confrontation with traditional medicine. While hormetic effects are generally seen in the 10-4 to 10-9molar concentration range, homeopathy is often practiced at concentrations far below 10-18 M. Nonetheless, intellectual field leaders in pharmacology, such as the eminent scholar and researcher Alfred J. Clark of the University of Edinburgh, delivered powerful, convincing, and unrelenting criticisms of the hormesis concept (then called the Arndt-Schulz law) just as the dose-response concept was being consolidated in scientific, biostatistical, and governmental circles, leaving the hormesis concept in a marginalized status at best (see his Handbook of Pharmacology, 1937).

Since toxicology had its origins in pharmacology, and pharmacology was central to traditional medicine, it was only natural that the rejection of the Arndt-Schulz law (and the hormesis concept) would become part of the toxicological mantra that led the field.

HORMESIS REEMERGES

Despite the broad array of obstacles it confronted, hormesis has emerged from its dormant, and at times ridiculed past, to claim a place at the toxicological table of dose-response mechanisms – as seen with its inclusion in major texts including Hayes' Principles and Methods of Toxicology and Casarett and Doull's Essentials of Toxicology. Furthermore it challenges for that pristine seat, the default model upon which most decisions fall back in risk-assessment.

While most interest in the hormetic model has focused on its application to environmental risk assessment, especially for carcinogens, it will have enormous implications for the biomedical sciences and clinical medicine. In the world of clinical medicine the hormetic zone may be that component of the dose response to either avoid, such as in tumor, microbe, or viral stimulation, or to seek out, such as in enhancing cognition, sexual performance, hair growth, or cardiovascular health. In an ironic twist, the increased recognition, acceptance, and use of hormesis within the biomedical research and clinical medicine domains may prove to be the equivalent of a toxicological Trojan Horse, which will lead to its eventual acceptance in environmental risk assessment.

Edward J. Calabrese is a professor of toxicology at the University of Massachusetts, School of Public Health, Amherst. He has researched extensively in the area of host factors affecting susceptibility to pollutants, and he has authored more than 10 books, including Principles of Animal Extrapolation, Ecogenetics, and Nutrition and Environmental Health, vols. I and II.

He can be contacted at edwardc@schoolph.umass.edu.