Common gene variants

Image: DNA by Heino Boekhout, CC licenseThe Journal of the National Cancer Institute (JNCI) is one of the most important journals in cancer research, and the material it publishes are a useful indicator of the direction of cancer research in general. The May issue of JNCI provides an illuminating insight into how large population based genetic studies are helping the way we think about what causes cancer and how we might prevent it. First up was an effort led by University of Cambridge scientists that compared genetic profiles of over 30,000 women with breast cancer (i.e. cases) and a comparable group of women without (i.e. controls).

Their focus was not on the rare high risk gene mutations, such as those in BRCA1 and BRCA2, but on common gene variants that individually have a very small effect, but together may have a substantial effect. In particular, they used data from 77 common genetic variants that had previously been found to have an effect on breast cancer risk. The upshot was that women in the top 1% of genetic risk based on these variants were at over a threefold risk of developing breast cancer than the average. This translates to a lifetime risk of 25% for the high risk women as opposed to an average of about 8%. The authors go on to argue that this level of risk stratification could be important enough to inform prevention strategies such as screening at a younger age or more intensive screening for particularly susceptible women. While appealing, an editorial in the same issue by Steven Narod raises important questions about the feasibility of such genetic models having an important impact on reducing breast cancer mortality, including who will pay for such tests, and whether a more population based approach for all will have a greater impact. Giving the emphasis on ‘Precision medicine’, including the recent initiative by President Obama in his State of the Union speech, it is an issue that we will need to consider in more detail in future posts.

A second paper, again led by scientists from University of Cambridge, investigated the extent to which common gene variants may influence survival after diagnosis. It is revealing that while genome-wide studies have identified over 300 genetic loci associated with onset of cancer, none of these appear to be relevant for clinical outcome. Their study included a genetic analysis of 40,000 women with breast cancer of whom almost 3,000 had died, and with up to 9 million genotypes from each woman. The authors identified one variant on the 11th chromosome that was associated with a twofold increased risk of dying with another potential variant on the second chromosome. While underlying genes responsible for these findings were not immediately evident the results do open the door to further understanding why some cancers are more aggressive, and will help to provide enthusiasm for additional large genetic studies of survival for other cancers.

A third paper came from within the Genetics section of IARC (where this blogger spends his days) and involved an analysis of a rare variant in BRCA2. While BRCA2 mutations are more commonly linked with high risks of breast cancer, this particular variant (rs11571833) has only a very modest effect on breast cancer (in the region of a 20% increase in risk), but an important 250% increase in risk of lung cancer. The authors extended their work to show that the variant had a substantial similar increase for cancers of the head and neck, again about 250%. Even though only about 1% of the European population have this variant, it does suggest that much of what we have learnt about BRCA2 positive breast cancers may also illuminate our understanding of tobacco related cancers. It also highlights the unpredictable nature of new genetic findings from these very large studies.

Expect much more to come.

Paul Brennan

Paul Brennan

Paul Brennan is head of the Genetics Section at the International Agency for Research on Cancer, and also one of the EPIC principal investigators. Most of his work focuses on identifying new biomarkers and genes that are involved in some common cancers and how this knowledge can inform cancer prevention, either through identification of new risk factors or early detection.

Research Interests: Cancer aetiology and prevention.
Paul Brennan

Share this article Facebooktwittergoogle_pluslinkedinmail
Follow us on Twitter twitter

1 comment to Common gene variants

  • Petra H Peeters

    Nature and/or Nurture
    Another paper in the same May issue of the JNCI is also of great interest.(1) Mitchell Gail was asked to comment on 25 years of Breast Cancer risk models. Mitchell Gail developed a simple model: only age and answers to five questions about reproductive, family, and medical history are needed. The model (sometimes called the “Gail model”) is available at as the National Cancer Institute’s (NCI’s) Breast Cancer Risk Assessment Tool (BCRAT). The model predicts based on the individual answers to few simple questions the absolute risk of breast cancer in the next 5 years or at age 90 (cumulative lifetime risk).
    In the past, calibration of this model was improved by designing special models for different ethnic groups. Discrimination, expressed in Area Under the Curve (AUC), of the BCRAT is modest: near 0,60 for women of same age. In women who have breast biopsies, potentially useful prognostic pathological features may improve discrimination. Moreover, mammographic density, and biopsy features suggest that AUC values near 0.7 will be achievable in women with biopsies. For women without biopsies, mammographic density and SNPs may be helpful. (1) Recently, the usefulness was assessed of 76 such SNPs that have been identified from very large genome-wide association studies (2). Garcia-Closas et al. estimated that elaborate questionnaire data plus mammographic density data plus these SNPs would yield an AUC of 0.68.(2)
    In the paper of Mavaddat et al., a 77-SNP polygenic risk score (PRS) was constructed and the discriminatory power –without any lifestyle data- was 0.62 and thus comparable to the BCRAT –including only lifestyle data. (3) The authors of this paper conclude that assuming that the multiplicative model is correct, the C-statistic would increase to 0.66 with the addition of lifestyle risk factors to the PRS.
    After 25 years of research the two fields of research seem to collapse, one that studies the lifestyle (environment) in relation to breast cancer and the other field with a focus on genes (predisposition of disease). Predicting future breast cancer risk based on 5-7 lifestyle characteristics is as good (or bad) as predicting future risk based on 77 SNPs. Combining lifestyle and genetic information adds little predictive capacity.
    This is an example of what we might call interaction. Lifestyle seems to cause disease in case genetic make-up does not vary; and genes seem to cause disease in case lifestyle does not differ. Mavaddat et al. fitted a multiplicative model and conclude that if modifiable risk factors and the PRS also act multiplicatively, targeting public health interventions to women at higher genetic risk should result in a larger absolute risk reduction. (3). (Considering that no interaction on a multiplicative scale implies additive interaction).
    Another explanation could be that in fact the 5-7 lifestyle characteristics are determined largely by the 77 SNPs. In fact, ‘age, menarcheal age, age at first live birth, relatives with breast cancer, and biopsy results’, representing the lifestyle factors of BCRAT are not so modifiable after all.
    In any case, in order to increase discriminatory accuracy, other strong risk factors need to be found.
    I am not so sure this will happen in the near future.

    (1) Mitchell H. Gail. 75th anniversary commentary. Twenty-five Years of Breast Cancer Risk Models and Their Applications. J Natl Cancer Inst (2015) 107(5).
    (2) Garcia-Closas M, Gunsoy NB, Chatterjee N. Combined associations of genetic and environmental risk factors: implications for prevention of breast cancer. J Natl Cancer Inst. 2014;106(11): dju305 doi:10.1093/jnci/dju305.
    (3) Mavaddat N, Pharoah PD, Michailidou K, et al. Prediction of breast cancer risk based on profiling with common genetic variants. J Natl Cancer Inst (2015); 107(5):djv036 doi:10.1093/jnci/djv036.

Leave a comment