TCS Daily


From Peer Review to Fear Review

By Elizabeth M. Whelan - May 22, 2007 12:00 AM

For years, self-appointed "environmental advocates" have generated press releases claiming that "chemicals" in our air, water, food, and consumer products such as cosmetics pose a risk of cancer. Nearly twenty years ago, the Natural Resources Defense Council caused a national panic by asserting that the agricultural chemical Alar posed a cancer risk to children (a claim that was later determined to be false). Advocates on Long Island have long claimed -- with a paucity of evidence -- that the elevated breast cancer rate in that area is the result of exposure to environmental chemicals like PCBs and DDT. These assertions can be dismissed as pure scare tactics, as they are not based on scientific data which have survived the rigor of the peer review process and been published in a professional medical journal.

But data and conclusions reported in peer review journals are sound and trustworthy, right?

Apparently, we can no longer assume that peer review journals are free of "junk science." The peer-reviewed journal Cancer, a publication of the American Cancer Society (ACS), just published a special online supplement that clearly meets the definition of "junk"in every way.

This journal, which boasts a long and distinguished list of editorial advisors, featured an article by Dr. Julia Brody, the executive director of the Silent Spring Institute and researchers from the Roswell Park Cancer Institute. The journal supplement section was funded by the foundation Susan G. Komen for the Cure. Brody is the principal investigator of the Cape Cod Breast Cancer and Environment Study -- a study of exposures to "mammary carcinogens" from air and water pollutants, pesticides, detergents, plastics, and cosmetics.

Why is this study "junk"? Let me count the ways:

First, the conclusions drawn by the authors, namely that environmental pollutants cause breast cancer, are not based on human studies but instead on high-dose animal studies. The authors identify a series of synthetic chemicals that cause breast tumors in rodents and then leap to the assumption that these chemicals also cause breast cancer in women.

There is now a nearly-universal rejection by scientists of the use of laboratory rodent data to predict human cancer risks. Thus, it is astounding that this lengthy paper is predicated on the assumption that rodents are just "little women." Indeed, in a companion article in the same Cancer supplement -- this one written exclusively by staff members of the Silent Spring Institute -- the authors recklessly elevate the role of rodent tests by asserting that "identifying chemical carcinogens in animal studies is currently the primary means of anticipating cancer effects in humans." If indeed that statement were true, we would be classifying a whole host of natural foods as "cancer risks" because they naturally contain chemicals that cause cancer in rodents -- including mushrooms (hydrazines), table pepper (safrole), and bread (ethyl carbamate)

Second, the authors brazenly conclude that, while they cannot state with confidence how many breast cancer cases annually are due to exposure to trace levels of "chemicals" (including pesticides, ingredients in cosmetics and other "environmental pollutants"), they feel the evidence (what evidence?) is strong enough to warrant "strategies" to reduce exposures in an effort to reduce breast cancer risk.

Further, it is clear from the affiliation of the senior author that this "study" was in no way neutrally conducted. In its own description, the Silent Spring Institute claims to be "a non-profit scientific research organization dedicated to identifying the links between the environment and women's health, especially breast cancer." The study was commissioned by the Susan G. Komen foundation's "environmental factors and breast cancer" project. Thus from the get-go the assumption is that breast cancer is causally linked to "environmental" factors -- specifically chemicals in the environment. This article is not science -- it is environmental advocacy.

Apparently, the peer reviewers of this paper were unfazed by the authors' conclusion that trace chemicals cause breast cancer, despite the fact that this idea is completely at variance with our understanding of the epidemiology of breast cancer. No serious cancer causation expert believes exposure to trace levels of environmental chemicals contributes to the toll of breast cancer.

In the latest edition of the "bible" of cancer epidemiology (Cancer Epidemiology and Prevention, edited by Schottenfeld and Fraumeni), the authors reviewed the causation of breast cancer by assessing the role of "environmental factors" in the causation of breast cancer -- meaning factors other than heredity, age, gender, etc. They considered the influences of reproductive factors (for example, having a first child at an early age may offer protection), use of hormones, weight, and nutritional factors (obesity after middle age is a risk factor for breast cancer). They then evaluated the claim that "other environmental factors" contribute to breast cancer risk -- including exposure to trace levels of pesticides and industrial chemicals. They concluded that evidence from large pooled human studies found no association between exposure to trace chemicals like PCBs and DDT and breast cancer risk: "Overall, recent studies have not found evidence of increased risk of breast cancer, and [pollutants] appear unlikely to be major breast cancer risk factors."

A quick review of the Pub Med articles bank reveals numerous epidemiological articles looking for a possible link between breast cancer risk and PCBs, DDT, and other chemicals -- and they end with the conclusion that "the combined evidence does not support an association with breast cancer risk." ACS's own Detailed Guide: Breast Cancer What Are the Risk Factors for Breast Cancer? says "research does not show a clear link between breast cancer risk and exposure to environmental pollutants."

Yet the Cancer article made little mention of the fact that human epidemiological studies contradict the conclusions that are based on lab-rat "evidence."

Media, interested in sensationalizing environmental health risks, used this story uncritically. The LA Times headline was "Common Chemicals Linked to Breast Cancer." Readers were led to believe that "overall exposure to mammary gland carcinogens is widespread" -- when there is no legitimate science to back up such a frightening claim. And of course, such reporting predictably sparked outrage: an editorial in the San Francisco Chronicle blasted "cancerous chemicals" and went even further than the study authors did, asserting that these chemicals may be a more important factor in determining breast cancer risk than genetics. The editorial called for more laws, more regulations to protect women from "carcinogens."

The American Cancer Society has done the cause of breast cancer prevention a disservice by publishing this article. ACS has tarnished the much-touted image of the peer review process as something superior to the "science by press release" approach. The article diverted attention from scientifically-based ways of reducing breast cancer risks (including using FDA-approved medications as a means of chemo-prevention -- an approach that shows great promise).

Recent news documents that fewer women are getting mammograms over the past several years -- a decline that reversed the salutary trend of the past twenty years. This is important, and women need to be reminded to get their mammograms. Yet, instead of focusing women's attention on the critical importance of early detection -- and the good news about treatment and very low rates of recurrence -- the ACS has supported unfounded fears of inescapable, invisible, chemical agents causing cancer in helpless women.

Dr. Elizabeth M. Whelan is president of the American Council on Science and Health (ACSH.org, HealthFactsAndFears.com).

Categories:

13 Comments

Is this a matter of science, or belief?
Dr Whelan bases her homily on the error of using lab rats as human proxies in cancer studies. But let her suggest a better way to explore the link between environmental contaminants and human breast cancer. Should we use humans instead? Perhaps prison inmates, or illegal aliens?

When you find a "hot spot" of elevated cancer incidence, such as the area in Long Islamd, and you find in the same place elevated levels of known carcinogens, you have information strongly suggestive of a possible link. I do not think it advances the cause of science to say that no experimentation can perfectly match the conditions in Long Island, so therefore no science can be done.

Let the good doctor propose a superior method of exploring from known data to a valid conclusion. I think most of us would be unwilling to just let the matter drop, and let real estate values in the zip codes of elevated risk erode of their own accord. SOMETHING is causing this anomaly. We should do our best to find out what it is.

Finally, cancer occurs on a cellular level. A predisposition toward breast cancer is strongly causative, of course (although I am not sure it is a necessary precondition). But also some sort of insult to the cell appears to be required. What better way to speed up the time required to study the issue but to give large doses to lab animals similar to humans? The only thing the large doses do is to create more incidences where suspect chemicals come into contact with individual cells. The use of large doses serves to speed up the experiment, as opposed to waiting thirty years, in a low dose environment, for one cell to become malignant.

To suggest, as the author does, that "No serious cancer causation expert believes exposure to trace levels of environmental chemicals contributes to the toll of breast cancer" is first to say that all the scientists who perform exploratory science under that hypothesis are not "serious", and second to say that the instinctive feeling that belief is superior to the use of the Scientific Method is the way science should go.

I will strongly disagree with both theses.

Kudos to Dr. Whelan!
Put your glasses on before you try to read, Beanie Baby.

Article Query
Forgive me if I'm misreading Brody et al.'s (2007) article, but after the authors summarize animal research in their rationale:

"In an emerging area of research into developmental toxicity, animal studies show that maternal exposure during pregnancy to atrazine or bisphenol A affects differentiation of the mammary glands in the offspring, which remain in a less differentiated state that is more susceptible to carcinogen exposure.[5-7]"

They appear to rather clearly identify the assumption underlying animal -> human inference:

"If these mechanisms similarly affect humans, reducing or eliminating chemical exposures could have substantial public health benefits"

Additionally, the study by Brody et al. actually focuses on human studies:

"We searched the PubMed database for articles in English published in peer-reviewed journals through June 2006 for human studies of breast cancer and environmental pollutants."

Is there something I'm misunderstanding here about Dr. Whelan's comments?

speaking of putting belief before fact, here comes roy
There is not a single study that suggests that rodents are a good model for human cancer response.

There are many studies that have found they aren't.

The fact that you would rather waste time on a method that has been proven to be flawed says a lot about your respect for science.

As is your claim that the mere fact that one area gets slightly more cancers than do other areas proves that something must be going on.

Try studying a little statistics for once.

Rat Racing
Rabid, do you remember perhaps 10 years back when researchers scared the **** out of half the U.S. population (among others) when they announced that coffee was probably a carcinogen? Well, they reached that conclusion, not by studying human coffee consumption habits, but by stuffing lab rats with so much caffiene that they couldn't sleep for a week. The rats got cancer.

No human being has ever been found to have contracted cancer from coffee, nor from any intake of caffiene. The study was put forth as serious research, just like this latest cancer study, but since then it has been completely discredited.

When researchers find that, for example, a set of chemicals probably causes cancer in animals, all that means is that researchers are justified in taking further steps (what that means is: asking for more money) to take the research to the next level and to come up with ways to study the effects of those chemicals on human beings, to see IF there MIGHT be an effect (and to measure what that effect is, if one is found). It does NOT mean that they are justified in making what amounts to a leap of faith by way of analogy.

When researchers do this, they betray a preconceived notion. Dr. Whelan knows that, and that is why she mentions, correctly, that scientists no longer take mouse-to-people deductive extrapolations seriously.

The Article
Although I suspect I was too young to have been too aware of your coffee scare (I'd have been 10), I'm familiar with the gap between animal and human testing; I do not see the claims Dr. Whelan claims were made actually being made in the article, however, which is why I posted the quotes I did. Rather, Dr. Brody and colleagues seemed to make clear they understood the leap from animal to human research, and their research study was actually focused on reviewing work with human participants, not on work involving animals.

Is there a part of the article I overlooked where they base projections of cancer incidence in humans or causal inference with to humans on research with animals?

Cancer studies
It's better to know something, if imperfectly, and improve on it, than it is to just throw your hands into the air and know nothing.

Mice, genetically, are your basic stripped-down mammal. They are the Model T of mammals, with few genes and fewer fancy improvements. They do not mimic the complexity of something like a human... they mimic the basic similarities between humans and other mammals.

Therefore they are an acceptable vehicle for research. Not the perfect vehicle, but an acceptable one. The perfect research subject would be, of course, a human. But we don't do those kinds of things any more. That went out of practise at the end of World War Two.

Then you say something I would like you to expand upon and clarify:

"As is your claim that the mere fact that one area gets slightly more cancers than do other areas proves that something must be going on.

"Try studying a little statistics for once."

You imply it means nothing for one place to engender more cancers than another. Why?

You can do better, I hope
Certainly you must realize what a dumb, incomplete statement you've made. I offered a goodly number of points in my comment. And supported them with what logic I could bring to bear.

If you want to refute anything I've said, please do so. I look forward to examining your substantive arguments-- but have little patience for ignorant, unsupported sniping.

Biological models...
I was in the business. I actually worked for a time at Huntingdon Farms in West Conshohocken in the early 1970's. We called it the Rat Factory and I was on the rat side of the house. (There was also a mouse side, of course.) I was very good. We bred and sold mice and rats for chronic studies and acute screening protocols.

Cancer studies are conducted with strains of mice that are genetically prone to develop tumors. In this way the researchers can count the rate such tumors occur in test groups when compared with the rate in control groups kept under the same conditions (but without the perturbation).

They all (all such groups) develop tumors during these studies. But the statistically significant difference in the occurrence of tumors (if there is any difference) is assigned to the variable under investigation.

Cause and effect hypothesis...you know.

Why would we use mice that are very inclined to develop tumors? Because if we used regular mice then our numbers would be too small to be measured and compared for statistical validation. Otherwise, we would need thousands and thousands of subjects in chronic studies that lasted months and months.

Furthermore, normal mice are very tough. They will die of old age before you can get very much out of them. Of course, you could actually poison them with some such compound or another. But if it was actually toxic then they might simply die rather than grow a tumor for you.

The enduring concern was that almost any sort of stress would cause these sensitive little mice to develop tumors at statistically significant, increased rates. Sterile water injected under their skin would do it.

As scare
Remember the As issue with Clinton?

Millions had to be spent to remove As from drinking water around the country.

"In choosing to lower the standard. the Clinton-EPA relied on animal studies and a high-dose exposure incident in Taiwan (more than 250 ppb), and chose to assume that you could do a straight extrapolation from high exposures to low, as though there was no safe dose for arsenic, even one molecule.

But that’s not necessarily a good assumption. For one thing, the entire lining of your digestive system is “sloughed off” every few days, taking all of the cells exposed to anything toxic right out of your body. That not only protects you from arsenic, but from cancer causing chemicals in your coffee, peanut butter, steak, and veggies. And that’s only the obvious defense. Your body also has mechanisms to pull toxins out of your blood if they get that far; barriers to prevent toxins from getting into cells; barriers to prevent them from getting into the nucleus of the cell; mechanisms to purge proteins that become damaged by toxins; mechanisms to fix your DNA if it is damaged, and the list goes on. The human body is quite able to handle low doses of most natural toxins."

http://www.rppi.org/032701.html

"Description
Whether the public or the environment is at risk is a commonly discussed question in numerous areas of public life, most recently and publicly with regard to issues like BSE, passive smoking and the dangers from pesticides in food production. It is therefore of great importance for everyone concerned with these issues - both policy makers and the public who may be subject to their decisions - to understand the basis on which 'risk' policy is made. The principle objective of this book is to highlight the uncertainties inherent in 'scientific' estimates of risk to the public and the environment resulting from exposure to certain hazards. Numerous examples of potential and real hazards are given. They all show that injury to personal health or the environment is a function not only of the toxicity (i.e. the lethality of a particular hazard) but of the level of exposure to the hazard concerned - in the words of the old maxim, the dose makes the poison. Existing regulation is criticized for being based on a flawed application of a poor epidemiological methodology, where toxicity is the basis of regulation and dose tends to be ignored. Furthermore, some authors conclude that risk is a subjective phenomenon that cannot be eliminated through regulation."

http://www.elsevier.com/wps/find/bookdescription.cws_home/677002/description#description

As with the CO2 AGW scare, correlation does not equal causation.

The worst part of all of these 'scares' is soon no scientist will be trusted.

roy would rather know something that's wrong, than work to know something right
Bad knowledge leads you to make bad decisions.
Acting on knowledge that you know is bad is the height of folly.

No mice are not stipped down humans.
Their biologies are not just simpler versions of what humans have, they are vastly different. Their reactions to many chemicals are vastly different from the way humans react.

Heck, there are chemicals that cause cancers in rats, but not in mice. But you want to pretend that we can use mice as an analogy for humans.

So much for roy's claim that it's better to know something that is wrong, then to know nothing.
...

new study shows that models underestimate increased rainfall by a factor of 3 - so much for peer rev
http://www.usatoday.com/tech/science/environment/2007-05-31-climate-models_N.htm

Rainfall comes from clouds. Thus, more evidence that the models underestimate the impace of clouds.

Low level clouds, (the type rain comes from) are very efficient at reflecting sunlight and cooling the planet.

So this is yet more evidence that the models over estimate how much the earth will warm from enhanced CO2.

Rainfall is a very efficient method for transporting heat from the surfact to the upper atmosphere, where it's easier to radiate into space.

Yet more evidence that the models over estimate how much the earth will warm from enhanced CO2.

Rainfall is a very efficient method for drying out the upper atmosphere, making it easier for infrared radiation to escape to space.

Yet more evidence that the models over estimate how much the earth will warm from enhanced CO2.

TCS Daily Archives