Beyond the Lab Rat

By examining humans at the molecular level, researchers hope to pin down cancer's true causes.

By Mark Caldwell|Wednesday, May 01, 1996
RELATED TAGS: CANCER, GENES & HEALTH
Cancer is terrible enough--still resisting cure 25 years after President Richard Nixon declared war on it, often intractably painful, and lamentably frequent. In 1994 the United States recorded 538,000 deaths from cancer, more than a fifth of the nation’s total. Yet bad as it is clinically, the fear it inspires magnifies its agony. We can’t trace cancer to any single agent or fateful event. It claws its way into existence out of the billions of complex interactions that make up our cellular biochemistry; its causes, as a decades-old drumbeat of research demonstrates, are myriad. Some are inherited--like the gene called BRCA-1, whose malfunction appears to be a component in many cases of breast cancer. But many are environmental and hence presumably avoidable: asbestos, infamously; tobacco; air pollution; a diet low in vegetables and fruits; alcohol; the ultraviolet radiation in sunlight.

If we could count out the dangers on the fingers of one or two or even a dozen hands, we might consider ourselves forearmed. But as researchers screen more and more substances, a dizzying number emerge from the lab festooned with skull and crossbones. Peanut butter, mustard, mushrooms, all-natural root beer--they’ve all been found to contain at least trace amounts of known carcinogens. So long has the list of possible carcinogens become that a certain cynicism has set in. Are you really courting malignancy every time you crunch a slice of bacon or take in a deep draft of your mobile home’s formaldehyde-laden air? Is life itself the ur-carcinogen?

On the face of it, there’s no good reason to flout the warnings. The animal experiments traditionally used to measure toxicity are eminently sensible. Researchers dose a population of laboratory-bred rats or mice with a suspect substance for an extended period of time, then measure the number of tumors that appear against a control population of genetically similar rodents that haven’t been given the test chemical. If the exposed animals show a significant rise in malignancies, the implication seems clear: you’ve found a carcinogen.

Much of the virtue of this system lies in its simplicity. Lab animals are bred for genetic uniformity. Unlike people, they live uncomplicated lives. They haven’t eaten thousands of foods, come into contact with hundreds of chemicals, breathed dozens of atmospheres. You can subject them to carefully designed and tightly controlled experiments, dosing them with a suspected carcinogen while shielding them from exposure to other substances that might confuse your results. If they develop a telltale symptom in response to your experiment, you can be reasonably confident that it’s not the result of some long-ago exposure you can’t recover or even imagine. Trying to monitor disease in large human populations, by contrast, is far messier. Suppose a thousand people contract cancer in a city blanketed by carcinogen-laden smog. You might plausibly infer that their illnesses originated in the air--but you can never be absolutely sure. Ten, 100, or 950 of the cases might have come from some hidden cause you’ve overlooked: a quirk of local diet or a forgotten lump of plutonium in the bus station basement.

Nevertheless, when it comes to assessing cancer risk, some scientists aren’t happy relying exclusively on rodents. First of all, rats and mice, while they share with us humans our mammaldom, are far away from us genetically. Some of the strains often used in experiments are naturally more prone to tumors than people, and the malignancies they get are different from those that typically afflict us. Biochemist Bruce Ames, of the University of California at Berkeley, has long cited these differences in criticizing the way we use animal models to determine cancer risk. In one test, Ames and his colleagues surveyed 226 known carcinogens: 96, it turned out, caused cancer in mice but not in rats; 56 were carcinogenic in rats but harmless in mice. What, Ames asks, are you to expect when you make the Grand Canyon-size leap from rodents to humans?

The root of the problem, really, is that cancer is such a complex disease, ever ramifying as you burrow into the microscopic processes that cause it. A carcinogen may work by causing a mutation in an oncogene--a gene that, if it malfunctions, lays the groundwork for the uncontrolled proliferation of the cell where it resides. But a slew of other factors can come into play as well. For example, says Ames, whenever you poison healthy cells you increase the likelihood of tumors arising, because any kind of cell injury stimulates cell division. The more cells you poison, the more cell division you cause, and the higher the probability that a spontaneous cancerous mutation will occur.

This, as Ames has pointed out, raises a considerable difficulty for animal testing, which often subjects experimental animals to megahits of suspected carcinogens. The aim of those high doses is to smoke out even low-level carcinogens. But if you flood an animal with high levels of the test substance, it may produce cancer by poisoning great numbers of healthy cells, even though it’s incapable by itself of triggering a cancerous mutation. Of course, that makes it a carcinogen for the mouse or rat inundated with it. But what if it’s a substance no human would ever be exposed to in quantities sufficient to kill healthy cells?

Small wonder, then, that researchers have been casting about for a surer link between the microscopic realm of carcinogenesis and the sometimes expensive and disruptive practical decisions individuals and governments have to make about the threat of cancer. What’s safe to eat; what poses a real cancer risk? What substances can we allow in water and air; what’s dangerous and should--whatever the cost--be eliminated?

There’s a developing scientific field designed to address such difficult questions with a new degree of confidence and accuracy. It’s called molecular epidemiology: molecular because it enlists techniques pioneered in the 1970s and 1980s to peer into the submicroscopic arena where cells interact with (and occasionally battle) foreign chemicals; epidemiology because it’s concerned with how a disease spreads through the human community and how it can be controlled.

The field may ultimately offer definitive answers as to whether a suspect chemical is really a human carcinogen and not just a rodent one. More important, once something has indeed been identified as a carcinogen, molecular epidemiology may be able to distinguish between an insignificant dose of the substance and a potentially harmful one. It may even be able to establish the different risks a single dose of a substance poses to different individuals. And ultimately, it may produce a simple blood test that shows whether you’re harboring a clinically dangerous amount of a cancer-causing substance.

Molecular epidemiology won’t do away with lab rats. As Frederica Perera, of Columbia University, one of the field’s pioneers, notes, Animal experiments have held up pretty well in their ability to predict. Molecular epidemiology, though, aims to achieve a greater accuracy and reliability by spying on the biochemical interplay between a suspected substance and human cells, in the hope of either catching a potential carcinogen red-handed as it inaugurates the cancer process or absolving it of suspicion. Then, with their submicroscopic evidence in hand, molecular epidemiologists want to make clear the connections between their microbiological data and human disease, for both individuals and society.

Perera and her colleagues in the field begin their work not with rodents but with humans. They amass tissue and body fluid samples from human volunteers who either have cancer or have been exposed to suspect chemicals. They then comb through these tissues for telltale biomarkers-- chemical products that reveal interaction between a suspected carcinogen and human cells.

Molecular epidemiologists pursue a wide range of such biomarkers, since they want to monitor every aspect of cancer’s progression, from the first infiltration of troublemaking chemicals to the subsequent complex array of bodily responses that end in a full-scale malignancy. Often the smoking gun they’re hunting for--an activated oncogene, for example-- indicates that your body has reacted to a carcinogen with a fateful cellular move. But other biomarkers can reveal an earlier stage--the first arrival, say, of a potential cancer-causing substance in your system. In this case investigators most often hunt for an adduct--a suspicious chemical bond between the substance and human DNA.

DNA-carcinogen adducts can be the first stage in a process that disrupts DNA replication during cell division. That can cause mutations. So when an adduct happens to form along an oncogene, the machinery of cellular replication can run haywire, setting the stage for malignancy. Over the last decade and a half, researchers have devised several highly sensitive assays for adducts between human DNA and various carcinogens. Finding a high level of them is a danger signal, suggesting that some of the conditions for cancer have been met.

Eventually, Perera thinks, molecular epidemiologists may establish a link between the quantity of adducts and other biomarkers in somebody’s tissue and that person’s likelihood of developing cancer. Ideally, you would be able to calculate your personal risk exactly by means of a simple blood test. Should it turn out to be high, you could take precautions. Adducts aren’t disease, Perera notes, but if somebody had a high level of them, we might reduce or eliminate his exposure, or give him supplements of certain micronutrients--antioxidant agents like vitamins A, C, or E, and carotenoids. They all inhibit the formation of adducts.

As yet, such a linkage between biomarkers and risk is not practical. Any realistic assessment of individual risk will have to wait for the completion of extensive studies. According to Perera, the method most likely to yield a useful payoff is the prospective study. In it, you assemble a group of volunteers, collect and store blood and tissue samples, and then monitor their health over the long term. Whenever a member of the group contracts cancer, you can perform what’s known as a nested case control study by going back and looking at the history of the patient’s fluctuations in whatever biomarker you’re trying to evaluate. You can then find controls from the same group--people who match the sick volunteer in every respect (age, sex, smoking history, ethnic group, and history of exposure to a suspected carcinogen) except that they haven’t gotten sick. If there’s a significant and consistent difference in levels of DNA- carcinogen adducts between the people who get sick and those who don’t, you’re on the road to a potentially revealing test.

Amass enough data of this sort and it should be possible to set parameters--to decide what level of a given DNA-carcinogen adduct in your blood establishes that you’ve been exposed to a serious danger. It may also be possible to calculate whether a given ethnic group is unusually susceptible to a given kind of cancer. If researchers can assemble a sufficiently large library of adducts and other biomarkers, each peaking at a different twist in the long path that leads from initial exposure to tumor, Perera says, we can pick up intermediate stages that lead toward cancer, instead of knowing somebody’s been exposed, then having to wait blindly for 20 years before we know whether or not he’s going to get sick.

A good illustration of molecular epidemiology at work is offered by studies connected with polycyclic aromatic hydrocarbons (called PAHs for short). By-products of coal combustion, and ubiquitous in heavily industrialized areas, PAHs rank among the world’s best-known and nastiest environmental carcinogens. Once they make their way into a human body, they readily form adducts with cellular DNA. Perera, working with Regina Santella of Columbia’s School of Public Health, has been using such adducts as indexes of a population’s exposure to PAHs. In 1990, for example, they monitored the occurrence of PAH-DNA adducts in the population of Gliwice, an almost apocalyptically polluted town in Poland’s highly cancer-prone Silesia region. The air gets worse every winter, as residents spew the by- products of coal heating into Gliwice’s already choking ambient industrial smog. Perera found that PAH-DNA adducts in volunteers’ blood samples did indeed rise and fall in tandem with the seasonal fluctuations in the town’s pall of hydrocarbons.

Studies of Finnish foundry workers over the past ten years have produced results consistent with those of the Poles. The Finns make a good research cohort because the factory closes down every July for a four-week vacation, when everybody simultaneously enjoys a sharp drop in exposure to pollutants. Their levels of PAH-DNA adducts rose sharply in the weeks after their return to work. Similarly, a group of smokers, whose nicotine habit also exposes them to PAHs, showed dramatic drops in PAH-DNA adduct levels within months of stopping smoking. Such results, of course, strongly suggest that PAH-DNA adducts appear in the bloodstream in response to carcinogenic pollutants.

Still, you may well ask if such adducts are really connected with cancer. Apparently the answer here is also affirmative. In a recent study at New York’s Columbia-Presbyterian Medical Center, Perera’s group found significantly higher levels of PAH-DNA adducts in blood samples from lung cancer patients than in people without disease, even when they adjusted their results to account for whether the subjects smoked. Their results suggest that cancer patients are particularly susceptible to genetic damage from tobacco smoke.

All these studies, of course, illuminate only small corners of a vast and intricate picture. The work in Poland and Finland suggests that PAH-DNA adducts mount quickly in your blood when you’re exposed to airborne PAHs, and such adducts correlate with precancerous mutations in oncogenes. But no one yet knows how many of the Polish and Finnish workers with elevated PAH-DNA adduct levels will actually go on to develop disease. Nobody knows what level of these adducts in an early blood test constitutes an ignore-it-at-your-peril warning. Adducts, Perera says, are just a fingerprint.

These uncertainties underline the vast amount of work yet to be done in molecular epidemiology. Perera and her colleagues are now busy expanding their arsenal of biomarkers. They’ve already explored adducts between human DNA and other carcinogens besides PAH. They’ve also sampled other sorts of indicators that might serve as surrogates, like adducts between carcinogens and the proteins produced by DNA (which are useful because they’re easy to find even in very small blood samples).

For a complete molecular epidemiology, of course, you’d need a repertory of markers that reflected every known carcinogen and accurately recorded every stage from first exposure to tumor. They’d need to be consistent, so that a given level in a blood sample would indicate reliably what your chances were of full progression at each stage. And they’d have to be easy to harvest too, available, for example, from a blood sample--you wouldn’t want to part with a chunk of your liver every time you submitted to a shopping mall health screening. Then, in the kind of painstaking and exhaustive statistical research that forms the bedrock of epidemiology, the markers would have to be correlated over time with changes in the cancer status of groups that are big enough to yield significant results.

Perera and her colleagues believe that this effort will ultimately lead to a twofold reward. The earliest probable payoff will be a general one: accurate knowledge about which potential carcinogens most threaten the human population at large. We’ll be able to make better use of animal models; we’ll know better what adjustments we have to make in our animal results to apply them to humans. And if we can document the range of genetic or other preclinical damage to humans, we can fine-tune risks to the population at large. We’ll have better information on how to set environmental standards.

After that, the rewards may become more personal. You and I can be exposed to exactly the same amount of a chemical, Perera observes, and our responses will differ because we metabolize carcinogens differently, because we have different rates of DNA repair, or because of acquired factors like diet. In fact, we’ve estimated there may be fiftyfold to a hundredfold differences among individuals in the way they respond to a carcinogen.

Molecular epidemiology, in other words, is an enterprise that promises to unite the macro and the micro, the theoretical and the practical. Someday, the researchers dream, their work will give us a simple test that will finally tell you, if you’ve been exposed to a carcinogen, precisely what you want to know most of all: Are you one of the lucky individuals who can safely ignore it? Or are you the one at risk?
Comment on this article
ADVERTISEMENT

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

ADVERTISEMENT
ADVERTISEMENT
Collapse bottom bar
DSCOctCover
+

Log in to your account

X
Email address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it emailed to you.

Not registered yet?

Register now for FREE. It takes only a few seconds to complete. Register now »