Zhang, Y, JP Wise, TR Holford, H Xie, P Boyle, ZH Zahm, J Rusiecki, K Zou, B Zhang, Y Zhu, P Owens and T Zheng. 2004. Serum polychlorinated biphenyls, cytochrome P-450 1A1 Polymorphisms and risk of breast cancer in Connecticut women. American Journal of Epidemiology 160: 1177-1183.

Epidemiological studies have reported conflicting results on the risks of breast cancer associated with exposure to PCBs. This study sheds light on one possible cause of that inconsistency, revealing that the risk is elevated in women carrying a particular form of a gene involved in metabolizing toxic compounds.By including people without that genetic characteristic in a study, the impacts on a susceptible subpopulation can be missed.

This study was carried out by drawing subjects from a study population in Connecticut that had been analyzed before (1,2). The earlier analyses found no relationship between PCB level and breast cancer risk. By adding information about genotype, this new analysis was able to detect an elevation in risk associated with one subset of the population, women possessing a particular variant of a gene important to detoxifying contaminants. PCBs are known to alter the behavior of the gene in question.

This finding, important because of the insights it provides into the source of conflicting results over PCBs and breast cancer, has implications far beyond that specific issue. Increasingly sophisticated work on genetic variation in susceptibility is suggesting that vulnerable subpopulations are the rule, not the exception. The epidemiological literature, which only recently began including detailed data on genetic characteristics of subjects, may thus be laced with false negatives.

Zhang et al. conclude that women with a variant of the CYP1A1 gene called m2 are at greater risk to breast cancer when exposed to PCBs. They estimate that risk is elevated over four-fold in post-menopausal women exposed to comparatively high levels of PCBs. Their results are consistent with recent studies that have suggested that variations in this gene may affect susceptibility to PCB exposure.   The protein produced by the CYP1A1 gene converts PCBs to other chemicals, called metabolites. These metabolites can be more, or less, toxic than the original PCB. Prior studies had suggested that one specific form of CYP1A1, the m2 allele, converts PCBs to a highly reactive metabolite, potentially causing DNA damage and cancer.

What did they do? Zhang et al. built their case-control study working with women that had participated in two earlier studies (1,2) conducted in Connecticut to examine the effects of exposures to organochlorine contaminants on breast cancer risk. These earlier studies, which as noted above found no relationship between PCB exposure and breast cancer risk, had stopped short of acquiring data on the genetic characteristics of participating women.

Alleles and genotypes

Each of a person's 20,000+ genes can have multiple forms, determined by the sequence of nucleotides making up each gene's DNA. Different forms are called alleles. Differences in nucleotide sequence are translated during gene expression to differences in molecular structure of the proteins made by that gene.

The protein products of different alleles can behave very differently. Hence allele differences can be extremely important in determining whether the protein can do its job, in this case, converting PCBs into metabolites.

 

Subjects were women between the ages of 30 and 80 with no previous diagnosis of cancer (except malignant melanoma). Out of the original study group, 374 women were selected as cases. They were matched with 406 controls, selected to be similar in age and who had undergone breast-related surgery but were confirmed to be without breast cancer.

Each participant was interviewed about medical history and dietary habits. A blood sample was taken to determine PCB levels and to allow for genotyping. The genotypic analysis was limited to Caucasians (374 cases and 406 controls) because of known differences in the frequency of alleles for the gene in question (CYP1A1) among races. Total PCB level was determined by summing the levels of 9 measured PCB congeners.

 
Zhang et al. used PCR techniques to determine which CYP1A1 alleles each subject carried. Alleles themselves can have minor variations in DNA sequence. These are called allele variants. This study focused on several alleles of the CYP1A1 gene... m1, m2, and m4. Each of these alleles has a common form, called by geneticists the 'wild type'... and an uncommon form, called the 'variant.' So the uncommon form of the m2 allele is called the "CYP1A1 m2 variant."

What did they find? First, several demographic factors increased breast cancer risk: post-menopausal age; later age of menopause; later age of first full-term pregnancy. These results are consistent with many earlier studies.

Second, independent of PCB contamination, women with at least one of the CYP1A1 m2 variant alleles had a 2-fold increase of breast cancer risk ( OR=2.1; 95% CI 1.1 to 5.0 ) compared to women with the normal m2 allele. Variants of other CYP1A1 alleles (m1 and m4) did not have increased risk, even after correcting for menopausal status.

What about PCBs? For women with a normal form of the m2 allele, Zhang et al. found a borderline significant increase of breast cancer if they also had higher PCB levels (OR=1.5; 95% CI 1.0 to 2.4). But if they carried the m2 variant allele, their risk increased by over three-fold compared to women with low PCBs carrying the normal m2 allele. (OR=3.6; 95% CI 1.5 to 8.2). And specifically among post-menopausal women with the m2 variant allele, the increase in risk was over four-fold (OR=4.3; 95% CI 1.6 to 12.0), again, compared to women with low PCBs and the normal m2 allele.

They found no significant association between PCB exposure in women with either the m1 or m4 alleles, neither for the normal version of these alleles or the variant form.

Zhang et al. also analyzed different types of PCBs separately, using PCB measurements in adipose tissue. Some PCB congeners mimic estrogen's activities, others oppose it; a third category of PCBs is genotoxic. Comparing women with the normal m2 allele and low PCBs, with women carrying the variant m2 and different PCB types, they found the following:

  • For estrogenic compounds, an elevation in risk of 3.5 (95% CI= 0.4 to 30.9)
  • For anti-estrogenic, OR= 3.5 (95% CI 0.4 - 30.8)
  • For genotoxic, OR= 4.9 (95% CI 0.6-43.3)

The wide confidence intervals of these estimates make it difficult to assess the significance of the association.

What does it mean? There are two take-home messages from this study:

  • The wrong gene combined with exposure to PCBs leads to a large increase in breast cancer risk.
  • Without information on the genetic makeup of the women in the study, this large increase in risk would not have been discovered.

Because PCBs are known to affect the expression of the CYP1A1 gene via endocrine disruption, links between PCBs and breast cancer risk are biologically plausible. It is also known that PCBs turn on the m2 genotype more readily than other alleles. That too makes sense.

Because approximately 7-17% of Caucasian women possess the m2 allele, according to Zhang et al., "this information has important public health implications."

Additional factors complicate the study of what causes breast cancer. Some of these involve the high probability that breast cancer is initiated long before it is detected, yet most studies sample exposure levels only after detection. Because of metabolic conversions of contaminants to other compounds, those measurements after disease detection may be very poor estimates of the actual exposures at the time the disease started. That weakens the ability of epidemiology to detect an effect. The fact that all exposures involve mixtures also contributes.