Issue Date: April 5, 2004
The concept that small amounts of a toxic chemical may be good for health is the center of a controversy in the field of toxicology. This debate has implications for the health of people and wildlife, the cost of pollution controls and cleanup, and the regulation of chemicals.
The issue has its genesis in the way toxicity tests are conducted. To determine whether a chemical can cause health effects such as cancer, laboratory animals are given specific doses of the substance over time. These laboratory studies commonly employ high doses of chemicals so that the study will produce measurable effects that have statistical significance within a small group of animals. Running tests on lower doses would require many more animals, which would raise serious issues about animal welfare, costs, and feasibility.
However, humans and wildlife are rarely exposed to the high levels of chemicals used in toxicity studies. So regulators use the data from high-dose experiments to derive what the effects of chemicals are likely to be at levels actually found in the air, water, or soil.
One of two methods are generally used for extrapolating health effects from high-dose studies. In one, called the linear dose-response model, a line is fitted through the data from the high-dose experiments so that the response to the dose becomes zero only when the dose equals zero. This model, which is used for nearly all carcinogens, assumes that no dose of a chemical is safe--even tiny amounts of a chemical can cause tumors.
A second method for extrapolating from high doses to lower ones involves tests on laboratory animals to determine the level of chemical exposure below which a substance produces no observable adverse effect. This is called the threshold model, because it assumes that there is an exposure concentration, or threshold, for the chemical that is considered to be safe.
Extrapolations for adverse health effects other than cancer are often calculated using the threshold method. However, industry has been pressuring the Environmental Protection Agency to use the threshold model for some chemicals that can cause cancer--especially those that act indirectly rather than damaging DNA and causing mutations. For example, after industry filed a lawsuit, EPA in 2001 used the threshold model when setting a goal for the maximum contaminant level of chloroform, a carcinogen, in drinking water.
Now, interest is growing over a third model for extrapolating from the effects of high doses of a chemical to low doses. And that method is sparking debate among toxicologists in industry, regulatory agencies, and academe.
THIS MODEL would use the concept of hormesis. Edward J. Calabrese, a professor of toxicology at the University of Massachusetts, Amherst, School of Public Health, is the primary proponent of hormesis. This concept holds that low doses of some substances, such as those found in the environment, could actually be beneficial to health even though these chemicals are demonstrably toxic at higher doses. If this theory were adopted by federal agencies, it could significantly change the way trace amounts of chemicals are regulated.
According to Calabrese, hormesis involves the stimulation of an effect--such as growth or cognitive function--at low doses of a chemical and an inhibition of that same effect at high doses. The hormesis theory simply describes how the effect changes with dose level, he says. It does not make a judgment on whether the stimulation at low doses actually is beneficial, Calabrese stresses.
The idea of hormesis has been around for more than a century, Calabrese says. But the idea became discredited because of its association with homeopathy, an alternative medical practice that treats conditions with minute doses of a substance that would cause similar symptoms in a healthy individual. An example would be using dilute syrup of ipecac, a substance that can induce vomiting, to treat vomiting.
Calabrese distances himself from homeopathy but says hormesis is a real, observable phenomenon.
The low-dose stimulation of hormesis is fairly subtle, Calabrese says. Studies show that the observed low-dose effects, such as the increased growth of plants, are only about 10 to 50% greater than if the experimental organism gets none of the chemical under scrutiny, he says. Yet factors such as plant growth can show a variability up to 40% in experimental control groups.
"Trying to distinguish a real 40% increase [due to hormesis] from background variability places great demands on study design, on numbers of doses, dose spacing, replication of the findings, statistical significance, things of that nature" to demonstrate a real biological effect, Calabrese says.
Calabrese has compiled two databases containing about 6,000 published toxicology studies that he says demonstrate hormesis. He says he has yet to find any chemical class that does not demonstrate hormetic effects. Determining whether these low-level stimulation effects are beneficial or harmful would require scientific follow-up, he adds.
Not all chemicals exhibit hormesis, but it happens often enough, he says. "We've observed it to occur about 40% of the time."
THE CONCEPT of hormesis challenges long-held beliefs in the field of toxicology and risk assessment practices of federal regulatory agencies, such as EPA and the Food & Drug Administration, Calabrese says. Acceptance of hormesis for low-dose extrapolation in risk assessments could markedly loosen the exposure standards for contaminants in air, water, food, and soil, and could slash the cost of cleanups and of compliance with environmental regulations, Calabrese says.
EPA scientists for years have wrestled with the issue of low-dose extrapolations. But they do not share Calabrese's zeal for using a hormetic model.
"I think the whole toxicology community has moved" from an emphasis on high doses to looking at a wide range of doses, including the low levels of chemicals found in the environment, says Linda S. Birnbaum, director of the Experimental Toxicology Division at EPA's National Health & Environmental Effects Research Laboratory.
But regulators must consider all the health effects from exposure to low doses of a chemical--not just possible stimulation responses that could be beneficial, she says. For instance, research suggests that laboratory animals fed low doses of dioxins have fewer liver tumors than those with no exposure. But at those same low doses of dioxins, experimental animals have suppression of their immune systems and show developmental and reproductive effects, Birnbaum says.
"We've done a good job of regulating overt effects," such as death or birth defects, from exposure to toxic substances, Birnbaum says. Over the next decade, regulators need to grapple with the complexity of a population's responses to the toxicity of a chemical. EPA needs to regulate pollutants to protect the most susceptible members of the population from harm, she notes. Depending on the chemical and the effect, the most susceptible segments of the population are often developing fetuses and babies or the elderly.
"The less extrapolation we have to do, the better," says Birnbaum. Eventually, genomics and other "omic" technologies will help decipher low-dose effects, she adds.
EXTRAPOLATION to low-dose effects was the focus of a session at the Society of Toxicology's 2004 annual meeting in Baltimore last month. It highlighted some of the scientific data gaps and disagreements among toxicologists about this issue.
James S. Bus, director of external technology at Dow Chemical's Toxicology Research Labs in Midland, Mich., told the conference that extrapolating the effects of low doses of chemicals is becoming increasingly important as the monitoring of chemicals in people's bodies increases and as analytic methods improve to detect ever lower concentrations of substances. Bus told the meeting two of the questions that the field of toxicology will be called on to help answer: "What do these low levels really mean?" and "What should we do about them?"
Rory B. Conolly noted that responses to low doses are determined by the nature of the substance--for example, its chemical structure--and by the specific biology of the organisms under study. Understanding the biological mechanisms of how a chemical causes a health effect is key to determining the substance's effects at low doses, he said. Conolly is director of the Center for Computational Biology & Extrapolation Modeling at the CIIT Centers for Health Research, a research organization in Research Triangle Park, N.C., funded by large chemical companies.
Michael L. Dourson, director and founder of Toxicology Excellence for Risk Assessment, a nonprofit research and education firm in Cincinnati, defended the use of linear dose-response extrapolation when no data exist to suggest the use of another extrapolation model. "I think hormesis must be approached on a case-by-case basis," at least in the short term, Dourson said. He added that the field of toxicology needs to determine how much supporting data are necessary to use an extrapolation method other than the linear one.
Jean R. Sevin, of the Bethesda, Md.-based consulting firm Technical Resources International Inc., took the opposite view. She said linear low-dose extrapolation probably should not be used as a default. Instead, it should only be used when data about biological mechanisms support its use, she said.
Dennis J. Paustenbach, vice president of Exponent Inc., a consulting firm based in Menlo Park, Calif., proposed that the Society of Toxicology develop a protocol--an adjunct to classic toxicology studies--to test for hormetic effects.
"Right now, it continues to be a bit of theology," Paustenbach said of the concept of hormesis. "No one's going to spend money on this unless there's a protocol" to follow for experiments that would identify hormetic effects, he said. This protocol needs the blessing of a recognized scientific group so data generated through these studies would be accepted as sound, he said.
FOR NOW, the debate continues to evolve as toxicologists deliberate over the best way or ways to use high-dose data to extrapolate effects from low doses. While answers to this complex issue are not likely to emerge soon, high-level help is on the way.
The National Academy of Sciences recently created the Committee on the Future of Toxicity Testing & Assessment, James J. Reisa, director of the NAS Board on Environmental Studies & Toxicology, told C&EN at the meeting. This new panel was formed in response to an EPA request, he said, and its mission is to assess and advance current approaches to testing the toxicity of chemicals and how data from these studies are assessed. The committee is expected to hold its first meeting this summer.
Reports by this new NAS panel may spell out the kind of research needed to determine if and when hormetic effects occur at low doses.
Meanwhile, EPA continues to use the linear extrapolation and threshold models for its regulatory decision-making. And Calabrese said he scans the scientific literature every Friday seeking more studies to add to the database on studies that he believes demonstrate hormesis.
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society