Breast Cancer : What we know

Breast cancer is the most common cancer in women in the world, in both industrialized and developing countries. In 1999, 467,000 women's deaths were attributed to breast cancer (1.7% of all female deaths).

And as the graph indicates, breast cancer incidence rates are increasing in many countries, although mortality rates are stable or slightly declining in some.

For the vast majority of cases of breast cancer, however, we cannot explain the causes.

We know that a small percentage of cases are linked to an inherited gene. Women who have inherited that gene within families that carry it are more far likely to experience breast cancer. But we also know that many more women without that gene type develop breast cancer also.

As this work has developed, additional complications have arisen. While they limit what we can say based on current evidence, they provide strong guidance for how future research can and must be conducted.

• First, some of the compounds that had been thought to be estrogenic—to bind with the estrogen receptor and activate genes that estrogen itself would—are instead anti-androgenic or even anti-estrogenic (they block action of androgens and estrogens, respectively). This is the case for DDE (an especially persistent metabolite of DDT) and some very persistent forms of PCBs. Thus a generation of epidemiological work that intended to test the role of estrogenic organochlorines because of the observation that estrogen itself elevates breast cancer risk was misguided. Studies making this mistake continue to be done, nonetheless, for example the Long Island Breast Cancer Research Project. Based on this more recent understanding of DDE's hormonal activity, we should not expect DDE to be associated with an increase in breast cancer risk. DDT itself is another matter, because DDT is a true estrogen.
• This raises the second complication. Most of these studies have measured breast cancer at the time of breast cancer diagnosis, or even later. [That is one reason why DDE has been a focus: DDT is converted in the body after exposure to several metabolites, of which one form of DDE, p,p'-DDE, is the most persistent.] Yet research in the laboratory with animals and epidemiological studies of women increasingly indicate that the cellular events sowing the seeds for breast cancer take place decades before breast cancer can be detected. It is very unlikely that chemical measurements decades later accurately, or perhaps even remotely, reflect conditions at the time when those cellular events began. This is especially the case because while some estrogenic compounds like dieldrin are persistent, others like bisphenol A are not.
• The third complication is that all these chemicals come in mixtures. The presence of mixtures weakens the power of epidemiology to find links between exposure and disease quite dramatically. Real links can be there but be masked by the complications caused by the presence of tens, if not hundreds, of compounds.

What do these complications mean for future research? First, a strong emphasis should be placed on studies that can assess chemical exposures during crucial periods of mammary gland development (especially in the womb and around puberty) and examine the links between those exposures and breast cancer risk much later in life. Second, at least for the "environmental estrogen hypothesis," research should concentrate on compounds that actually are estrogens. Third, epidemiological studies links between environmental estrogens and breast cancer risk need to broaden their scope of inquiries beyond the traditionally-studied persistent organochlorine compounds and include nonpersistent estrogens like bisphenol A in their analyses. And finally, new methods in epidemiology are urgently needed to cope with the obstacles that mixtures pose for research.

Where does that leave us now? Current scientific evidence does not prove definitively that contaminants are involved in the causation of breast cancer. They are implicated, nonetheless, and their involvement is highly plausible, based on what we know from animal experiments, from basic mechanisms in biology, and from well-designed epidemiological studies.

Stronger evidence of links is emerging as the studies become more sophisticated and explicitly incorporate methods that circumvent some of the complications noted above. Less weight should be given to research that, by ignoring those complications, has increased its vulnerability to what statisticians call "false negatives" ... finding no statistical association when there really was one.

Last modified 20 October 2002