Breast cancer and genome-wide association studies

The famous cloning of the BRCA1 and -2 susceptibility genes in the mid-1990s promised great hope for families predisposed to developing breast cancer and for our ability to predict risk. However, despite publication of the human genome sequence and massive efforts in breast cancer genetics worldwide since then, progress has been slow.

Now, a study published in Nature in May has reignited the field. A global consortium headquartered at Cambridge University identified five new and relatively unexpected genetic variants that increase a woman's susceptibility for breast cancer, and the promise is that this is just the start.

Dr Georgia Chenevix-Trench is head of the Cancer Genetics Laboratory at the Queensland Institute of Medical Research (QIMR) and a pivotal part of the genome-wide association study reported in Nature.

It involved scientists from 22 groups across 15 countries, including Australia, the UK, Denmark, France, Korea, Singapore and the US, who together comprise the Breast Cancer Association Consortium (BCAC). Talking to Chenevix-Trench filled in some of the history of this major and important endeavour.

The known susceptibility or 'risk' genes for breast cancer, namely BRCA1 and -2 plus a handful of more recent finds, account for less than 25 per cent of the familial risk (or one to two per cent of all breast cancer cases).

"Researchers have been trying to find more common polymorphisms associated with breast cancer risk for a decade," Chevenix-Trench says. "Plenty of circumstantial evidence was out there for multiple elements or polygenes of small effect contributing to risk in the general population and not just in high-risk families."

Hints came along the way with the discovery of variants in genes such as ATM and CHEK2; however, these genes were still not exactly what scientists were looking for - they did confer a moderate to low risk but were still relatively rare in the general population. Importantly for the field though, the most conclusive evidence in the earlier CHEK2 study came from a collaboration of 10 different groups worldwide that all showed the same effect in terms of associated risk.

"This proved to us that when you really find the right thing, it is clear that you have found it, as opposed to having different and variable risks among studies in different populations," she says. "It also showed that this sort of effort could be done quite quickly and efficiently and that the genotyping in different centres was compatible."

SNPs and drinks

The Cancer Research UK's group at Cambridge, headed by Professors Bruce Ponder and Doug Easton, set out about three years ago to identify risk-associated genes for breast cancer by doing a genome-wide study, the first of this kind undertaken for cancer.

Initially, they conducted a two-stage study in 4,398 breast cancer cases and 4,316 controls, testing 227,876 commonly occurring single nucleotide polymorphisms (SNPs). However, to obtain meaningful validation for the third and biggest stage numbers-wise, they knew that a huge group of different datasets would be required - a consortium was needed.

Not long after their study began, Ponder was sitting at a bar on South Stradbroke Island at the end of the annual scientific meeting of kConFab, the Kathleen Cunningham Consortium for Research into Familial Breast Cancer, strategising with Chenevix-Trench and Malcolm Pike, a cancer epidemiologist from the University of Southern California.

Over seaside drinks, the trio set out the blueprint for the type of international consortium required for the next stage of Ponder's study.

They planned for six-monthly meetings in the UK, with the first one held in April 2005 at Cambridge. In the event, three consortia were launched - BCAC, a similar one on ovarian cancer and a third looking at modifiers of BRCA1/2, driven primarily by Chenevix-Trench. According to her, all three really have been amazingly successful.

About 12 months ago, the 30 most statistically significant SNPs coming out of the Cambridge study second stage were handed out to BCAC.

Subsequently, 21860 breast cancer cases and 22,578 controls were screened for the polymorphisms, with all groups doing their own case and control sets.

The SNP sequences were anonymous, except to the few laboratories that needed to know for their specific technology platform.

The final analyses identified sequence variants in five novel, independent loci in the genome that exhibited strong and consistent association with breast cancer.

Four of these contain plausible causative genes, namely FGFR2, TNRC9, MAP3K1 and LSP1. Polymorphisms in these four genes are believed to increase breast cancer risk in the general population of women by between 20 and 60 per cent.

The gene with the strongest association was FGFR2, or fibroblast growth factor receptor 2; it is also the only gene with a reported link to breast cancer biology. The gene identities were unexpected since almost all previously identified breast cancer susceptibility genes are involved in DNA repair, whereas these latest candidates are largely related to cell growth and signalling.

The genes identified account for a further four per cent of familial relative risk of breast cancer. Additional, as yet unidentified polygenes, and environmental factors, are believed to account for the rest.

The gene variants are also common in the general population, although the cancer risk conferred by each is relatively low. In contrast, BRCA1/2 mutations are relatively rare, but women who have them run a high risk of developing the disease.

This sort of result was exactly what researchers like Chenevix-Trench had always thought they would find eventually, and proved the importance of power in such association studies.

Genome-wide studies

Two other genome-wide association studies on breast cancer susceptibility were completed around the same time and published back to back in Nature Genetics.

One study, led by Australian scientist David Hunter, of the Department of Medicine, Brigham and Women's Hospital and Harvard Medical School in the US, also found four variations in the FGFR2 gene as their top 'hit' in around 2,000 women with breast cancer compared to the same number of control cases.

The other study, led by Simon Stacey of deCODE genetics in Iceland, didn't have FGFR2 in its top ten, but instead identified TNRC9, as well as another important gene polymorphism associated with an increased risk of estrogen receptor-positive breast cancer.

Chenevix-Trench says the most exciting aspect of these studies was that "researchers now understood how to find these genes and the data already collected will help them know where to look.

"This is likely to represent only the tip of the iceberg, and delving deeper into the existing mass of data from these three studies alone will likely reveal many other mutations with similar significance to breast cancer risk. These findings show conclusively that susceptibility genes do not have to be high risk."

The breast cancer consortium was the most advanced from its inception and has published other papers leading up to this year's big one.

The first, published in the Journal of the National Cancer Institute, constituted a stocktake of all published and unpublished data from the 22 groups on 15 gene polymorphisms associated with breast cancer.

"Negative data used to get published as well but now that we can do so much so quickly (five SNPs per week instead of per year), it is just too hard to put it all out, and we are drowning in negative data. Chevenix-Trench says.

She stressed the importance of this article for highlighting how much bias is introduced when only positive association data is considered, particularly for variants of low risk, and for proving the landmark nature of such a large, collaborative effort.

Where to from here?

Three Australian groups contributed to the BCAC study published in Nature - the Australian Breast Cancer Family Study, led by John Hopper, based in Melbourne and genotyped at QIMR; the Kathleen Cunningham Foundation Consortium for Research into Familial Breast Cancer (kConFab - currently chaired by Chenevix-Trench), and the Melbourne Cohort Study headed by Graham Giles. Thus, about one-tenth of the genotypes in this paper were Australian.

Research-wise, this study will generate many further projects. "One of the nice things about these consortia is that because of the meetings are set up back-to-back every six months, things move very rapidly, and can cross over," Chevenix-Trench says.

All three consortia are meeting in Queensland in August 2008. The validated breast cancer SNPs are already being tested in ovarian cancer and in the search for modifiers of BRCA1/2. The data are also being studied with respect to different aspects of breast cancer such as survival, cancer biology, lifestyle factors and possible gene-gene interaction effects.

Chenevix-Trench predicts several more genome-wide studies with significant validation in the near future. "There could be between 50 and 200 gene-susceptibility loci for breast cancer, and we will probably be able to find them all within the next few years, and probably with the available data from the Cambridge genome wide study and other similar efforts."

The short-term clinical value of the Nature study is most likely in genetic screening. Chenevix-Trench envisions, for the first time, that "within five to 10 years, we may be able to screen large populations of women for SNPs in 20, 50 or 100 of these genes in one go, quickly and at low cost."

Women tested in this way could potentially be banded according to risk of developing breast cancer regardless of family history or anything else. This could help to target further screening more appropriately and effectively, potentially not only saving lives but also reducing health costs.

Chenevix-Trench cautions, however, that more such studies need to be done and more risk factors revealed before any genetic testing is considered. "A woman would need to carry multiple bad variants to really increase their cancer risk."

In the longer term, these large-scale genomics approaches will help researchers unravel the biology of cancer, as well as a range of common diseases. In fact, similar studies are underway to find genes for other cancers, including melanoma, bowel and lung cancer.

"Ultimately, knowing more about the biology of the genes involved in diseases like breast cancer will lead to better treatments or prevention."