Doubts Raised on Testing for Carcinogens

A growing body of evidence is raising doubts among some scientists about the ability of animal experiments to predict whether certain chemicals such as drugs, food additives and pesticides will cause cancer in humans.

The controversy centers on a critical litmus test used by regulatory agencies in deciding whether chemicals are carcinogenic. Critics of the test say there is increasing evidence that the test is overpriced, inefficient and unreliable.

The test--in which rodents are exposed to high doses of chemicals, then dissected and examined for tumors--has helped identify hundreds of suspected carcinogens. As a result, many have been banned from use or permitted only under strict controls.

May Be Misleading

But critics say there is reason to doubt that some of those chemicals actually cause cancer in people, even if they do cause tumors in rodents. Some say the tests, which can cost $1 million per chemical and take years to complete, may be misleading in a third of all cases.

Those critics want broader use of alternative tests that simply examine the effects of chemicals on individual cells in the laboratory. Those, too, are fallible. But proponents say they would make it possible to screen more chemicals, and at a lower cost.

“Everybody was hoping that there would be some way of testing that you could say, ‘positive’ or ‘negative’ and that’s it,” said Dr. Gary Williams of the American Health Foundation in New York City. “Well, nature just isn’t that simple.”

The debate comes at a time when the Environmental Protection Agency is updating its guidelines for the assessment of environmental health hazards in light of scientific advances. The controversy reflects both the complexity of cancer and scientists’ limited understanding of how the disease develops.

The testing issue affects products used in daily life, such as artificial sweeteners, pharmaceuticals and pesticides. Carcinogenicity testing also has significant financial implications for individual firms and society at large.

“This is a controversial area because it involves not just science but economics,” said Raymond W. Tennant, a researcher in the National Toxicology Program at the National Institute of Environmental Health Sciences. “There are gaps in scientific knowledge; and when you have those sorts of gaps, they’re often filled by opinions.”

Thousands of Drugs

At least 50,000 chemicals are now manufactured or processed in the United States. About 1,000 new ones are produced in commercial quantities each year. In addition, there are 600 active ingredients registered as pesticides with the federal government, and thousands of drugs.

Many countries now regulate chemical production and use, largely in response to concern over the health effects of toxic exposures. In the United States, most drugs, food additives and pesticides, and some new chemicals, must be tested for carcinogenicity.

However, officials estimate that less than 10% of the chemicals available have been subjected to adequate testing--a figure they say underlines the pressing need for a quick, reliable and affordable method of screening.

‘Gold Standard’

The lifetime rodent test is now considered by regulators the “gold standard” of carcinogenicity tests. Although short-term laboratory tests are used widely in industry and even in regulation, regulators require rodent studies as the final word.

“The rodent bioassays (are) probably the most reliable indicator, short of human data, for identifying chemicals that may pose a cancer risk for man, and for trying to express (their) potency,” said Jack Moore, acting deputy administrator of the EPA.

The tests entail exposing two species, usually rats and mice, to high doses of the chemical in question. Animals of both sexes must be studied. For purposes of comparison, control animals must also be kept, under identical conditions but without the exposure.

Tissue Scrutinized

After about two years--the bulk of a rodent’s lifetime--the animals are killed. Their tissue is then scrutinized for tumors. If there are significantly more tumors in the exposed animals than in the control group, the chemical is often suspected to be the cause.

Because many carcinogens produce cancer by altering genetic material in cells, and because the mammalian system is similar from one species to another, scientists believe that substances that cause tumors in rodents may well cause cancer in people.

“With very, very few exceptions, if a substance causes tumors in a human being, it will cause tumors--not necessarily of the same kind or in the same place--in another species,” said Jerry M. Rice, a top biochemist at the National Cancer Institute.

But does it work the other way? Do chemicals that cause tumors in animals at high doses necessarily produce cancer in humans? Critics of the federal reliance on rodent tests say there is increasing evidence that, in some cases, the answer may be no.

A few examples:

- Researchers investigating the effects of exposure to gasoline vapors found an increase in kidney tumors in male rats. Now even regulators say the increase may be traceable to a protein unique to male rats, which apparently played a critical role in tumor development.

- Some sulfa drugs produce an increase in thyroid tumors in rats. But the tumors may result less from the chemical than from the high doses. Those doses may produce hormonal changes, leading to tumors, that might not occur at human doses.

- Other chemicals trigger bladder tumors in rodents. But regulators say those tumors may result from bladder stones--which, when analyzed, turn out to be composed almost entirely of the chemical being tested, crystallized in the rodent’s bladder.

“There are two big questions,” said Lester B. Lave, co-author of a paper on rodent tests published this month in the British scientific journal Nature. “How do you go from rats to people? And how do you go from extremely high doses to low doses?”

Lave and his co-authors noted in the paper that half the chemicals tested by the U.S. National Toxicology Program have produced tumors in rats or mice. In 70% of the cases, the results were the same in both species. But in 30%, they were not.

‘Fraught With Uncertainty’

Since mice and rats are more similar to each other than they are to people, Lave suggests that extrapolating to humans is “fraught with uncertainty.” At best, rodent experiments might predict human experience accurately 70% of the time, the authors calculate.

Lave, an economist at Carnegie-Mellon University, and his co-authors believe that the rodent test leads to misclassifications: Some carcinogens are probably labeled non-carcinogens, and some non-carcinogens are falsely labeled carcinogens, they say.

As an example, Lave cites the 1969 ban on the sweetener cyclamate because of links to bladder tumors in rats. Now scientists say cyclamate is unlikely to cause cancer by itself; instead, it may be a promoter that enhances the effects of true carcinogens.

Also often cited are certain pesticides such as DDT, identified in the 1970s as suspected human carcinogens. Because those chemicals produced tumors almost exclusively in mouse livers, some scientists are uncertain how well the finding applies to humans. Tumors in multiple organs and two species would be more convincing, scientists say.

Improved Understanding

Nevertheless, those pesticides remain listed as suspected carcinogens, said Moore of the EPA. They are simply viewed as perhaps less potent than initially thought, in light of improved understanding of the forces underlying the development of tumors.

“It depends on where you’re coming from,” said Tennant, the federal carcinogenicity researcher. “If you’re responsible for public health policy, I think the public would demand that you view the potential hazard of these substances in a fairly conservative manner.”

In their paper, Lave’s group calculated the cost of misclassifications--the cost of additional cancer cases or of withholding beneficial chemicals from use. They then concluded that society would be better served by relying on several short-term tests.

Reduce Need

They and others suggest that the tests be selected from about 100 now available. The tests are less expensive, less time-consuming and require no animals. They would not eliminate, but would reduce, the need for rodent bioassays, proponents say.

They also offer insight into how cancer develops.

The tests involve exposing cells in culture to a chemical, then examining the cells for any genetic effects. Did the chemical react with the cell’s genetic material? Did it produce mutations in the cells or were chromosomes altered?

“We have known for years that most chemical carcinogens produce cancer by attacking genetic material--genes and DNA,” said Williams of the American Health Foundation. “We can identify those kinds of carcinogens readily by using the short-term tests.”

‘An Open Question’

The tests do not, however, detect chemicals that produce tumors indirectly--say, through a toxic effect of high doses in experimental animals. But Williams says, “Whether effects under those conditions have any relevance to human hazard is an open question.”

Others, however, have less confidence in short-term tests.

Tennant recently published a study of four such tests. He found that no single test could anticipate the diverse mechanisms that lead to cancer, and the advantages of using several tests at once were unclear.

In his study, Tennant found that the three most potent carcinogens tested in his study were not detected in any of the four short-term tests. Three weaker carcinogens produced no response; and three non-carcinogens tested positive as carcinogens.

Identifying Chemicals

“It’s not to say the tests aren’t useful,” Tennant said in a telephone interview. He said they are highly reliable in identifying chemicals that change genetic material directly. But they are not useful, he said, in identifying chemicals that cause cancer indirectly.

How such chemicals work is a subject of much speculation, Tennant said. Some may create conditions that in turn damage genetic material. Others may not cause mutations at all but instead influence the way cells grow and divide.