Roanoke Times
Copyright (c) 1995, Landmark Communications, Inc.
DATE: SUNDAY, April 1, 1990 TAG: 9004010259
SECTION: HORIZON PAGE: F5 EDITION: METRO
SOURCE: MALCOLM GLADWELL THE WASHINGTON POST
DATELINE: LENGTH: Long
They know the effect of massive doses of each chemical on rats and mice. But public policy is about humans, not animals, and it must apply to chemicals often found in very small amounts. To draw any conclusions about the real world from what they have seen in the laboratory, scientists have had to speculate.
For 20 years, toxicologists have lived with this uncertainty in trying to measure the dangers posed by the chemicals we eat, drink and breathe. But when rising public concern about potentially dangerous chemicals is in conflict with rising costs of eliminating risk, society is faced with increasingly difficult choices. Many scientists are beginning to argue that the guesses made in the name of risk assessment are no longer good enough.
"Many of the leading people in the field are getting very suspicious about animal tests and what they mean," said Bruce Ames, director of the environmental health center at the University of California at Berkeley. "I think we may be being led down the wrong path.'
\ The problem arises from the fact that the risks of developing cancer from typical exposure to many toxic chemicals may be in the range of one in a million or even less. That means that for a laboratory animal experiment to have a chance of detecting a chemical's carcinogenicity at realistic doses, it would have to use millions of animals.
Because that is hardly practical, scientists do the next best thing. They feed a far smaller group of rats or mice a much higher than normal amount of the chemical. It is understood that this creates an unrealistically high incidence of cancer. Then the researchers extrapolate backward to estimate how many cancers would have been caused at a more realistic dose.
In extrapolating, both the Environmental Protection Agency and the Food and Drug Administration make what is called a linear assumption. For example, if 50 out of 100 rats got cancer from eating a bowl of a particularly nasty chemical every day, then federal health officials assume that at half a bowl a day, half as many would get cancer, and at quarter bowl, a quarter would get sick and so on down to the point where if the rats were nibbling on the chemical only once a month, just one or two would develop tumors.
This is called a linear assumption because a graph of it, called a dose-response curve, would be as straight as a yardstick.
Here is where the controversy begins. Although it seems logical that the risk of cancer should change in proportion to the chemical dose, that is only a guess - a guess that is now being called into question.
"When it comes right down to it, why should we expect nature to give us dose-response curves that follow any predictable mathematical formula?" said John Bailar, a toxicologist at Montreal's McGill University. "Why should they follow a straight line or some regular curve? I don't think you can make an iron law about these things."
What if, for example, a chemical is so potent that it poses a grave risk in even the smallest amounts? If low doses do about as much damage as is possible, then large doses could not make much difference. The dose-response curve in this case would be shaped like a kind of street lamp, rising very sharply in the beginning and leveling off at higher doses. If policy decisions are made on the assumption of a straight-line relationship, the risk of intermediate doses would be underestimated.
It may be that a bowlful of this chemical every day, like the hypothetical nasty chemical, still kills 50 out of 100 rats. But a nibble every month, which the linear assumption guessed was killing just one or two, might actually cause cancer in 15 or 20 rats.
Or consider the opposite. What if high doses of a chemical are needed to trigger the complex chain of events that cause cancer? This is a well-known phenomenon among some carcinogens. Formaldehyde is an example. In such cases the risk of cancer would look like a hockey stick, increasing only slightly until the critical mass of the chemical is reached and then rising sharply.
Here as well the typical extrapolations about the dangers of low doses might be wrong. A bowl of this chemical, like the hypothetical nasty chemical, might still kill 50 out of 100 rats. But a quarter dose might not even come close to killing a quarter of the rats. If that quantity fell below the cancer threshold, it wouldn't be any more dangerous than a monthly nibble.
The differences among various dose-response curves have implications for policymaking. A chemical the government thinks is safe for human consumption might not be safe at all if its curve resembles the street lamp. Hockey stick chemicals, on the other hand, could be pronounced unacceptable risks when in fact at realistic doses they are virtually harmless.
Large sums spent to reduce exposures to a hockey-stick chemical to extreme lows might buy very little added safety. The money might be better spent redoubling efforts to protect society against street-lamp chemicals.
\ How might chemicals deviate from the yardstick model? Vinyl chloride, a street lamp chemical, is dangerous because the body processes it into another substance that causes potentially cancerous mutations. But the body can only metabolize a limited amount of vinyl chloride at one time. While cancer risks from the chemical rise steeply at lower doses, they level off as the body can no longer process any more of it.
At the same time, recent research into how cancer begins lends some credence to the idea that some chemicals may have a hockey stick shaped-curve. For example, the greatest risk of cancer from chemical toxins appears to be when cells are dividing rapidly, replacing cells killed by the toxin.
But low levels of a chemical might not cause enough cell damage to kill cells and cause rapid cell proliferation. In many cases, wholesale cell killing - with its accompanying increased risks of cancer - occurs only when a toxin reaches a critical level in the body.
The problem, however, is that scientists understand very few chemicals well enough to know which model they fit, if any.
"Our position is that linear dose extrapolations are the only ones we can justify as a general principle," said David Gaylor, director of the biometry staff of the National Toxicology Program. "It's a default position unless there is convincing evidence to the contrary, which there rarely is. It would be very risky to assume hockey sticks as a rule."
"In most cases the government's procedure probably overestimates cancer risks," said John Graham, a professor at Harvard University's School of Public Health. "In some cases they are probably underestimating it, and sometimes they might be right on the mark. But what we have to keep in mind is that no one really knows the right answer."
by CNB