Vast quantities of human DNA have been sequenced for biomedical research and clinicians, researchers and innovators are now beginning to focus on the next phase of the genomic revolution: translating these data into tools that aim to make a difference in the real world.
New insights into breast cancer risk
Using genetics to understand cancer risk is not a new idea. Take, for example, breast cancer, the most common cancer amongst Western women. We’ve known since the 1990s that mutations in the BRCA genes influence a woman’s chances of getting breast or ovarian cancer. Whilst this has led to a range of important screening initiatives that have saved lives, women with mutations in these genes only account for a small proportion of all breast cancer cases. It’s also the case that not all people with a BRCA mutation go on to develop breast cancer.
What’s more, tests for BRCA mutations are only really prescribed (and reimbursed by medical insurers) for the small subset of the population where prior information, such as a family history of early onset breast or ovarian cancer, points towards these gene variants having a putative role in the disease. (Angelina Jolie, who famously had a mastectomy following a positive test for a BRCA mutation, knew beforehand that her mother had died from ovarian cancer caused by a BRCA mutation.) Many others with BRCA mutations but who don’t have a family history of disease will not have the opportunity to be tested. So, whilst BRCA genes has been transformative for defining risk in families, as a general assay for disease risk, such tests have limited broad-scale utility.
We’ve written before about the growing realisation that most human traits and diseases are polygenic. This includes breast cancer, where bigger studies and diverse datasets are showing that breast cancer risk is about more than the BRCA genes. A 2017 study, published in Nature, found almost 6000 different genetic variants that influence breast cancer risk, based on an analysis over a quarter of a million individuals with and without breast cancer. These variants were found in 100 different regions of the genome, two thirds of which were previously unknown. This means that these 65 new regions of the genome associated with breast cancer were all found outside of the BRCA genes that carry the infamous mutation.
A separate study focused on understanding the polygenic nature of so called estrogen-receptor (ER) negative breast cancer. This is a breast cancer that, unlike the much more common ER positive version, does not respond to hormone therapies. Again, multiple variants, 125 in this case, were found to be associated with breast cancer, underlining the polygenic nature of breast cancer risk.
Towards a greater personalisation of breast cancer risk
A third study aimed to go beyond single gene tests based on BRCA genes to generate a risk prediction methodology that is both applicable to the general population and can define risk for different subtypes of breast cancer. Using lists of genomic variants identified in almost 80 Genome Wide Association Studies (GWAS) on breast cancer, the researchers built a polygenic risk prediction algorithm to estimate an individual’s genetic liability for breast cancer, based only on their genetics. This polygenic risk score (PRS) summarises the predicted contribution of an individual’s genome to disease risk as a single number.
The researchers explored different methodologies for generating PRS using different numbers of variants and validated their scores on independent datasets to produce a PRS using 313 independent variants. Whilst this used a larger number of variants than previous attempts for a breast cancer PRS, it didn’t contain the thousands mentioned above because not all of these previously identified variants provided predictive power. The researchers found that individuals with the highest 1% of scores had a four times greater lifetime risk of developing breast cancer than those in the middle of the PRS distribution, equating to an absolute lifetime risk for these individuals of about one in three.
Combining monogenic and polygenic risk
So how do we marry what we know about monogenic disease risk through the BRCA genes with this new polygenic paradigm? This was addressed in a fascinating new preprint (preprints are studies that have not yet undergone any peer review, but are nevertheless publicly published on the internet). The researchers show that an understanding of your polygenic risk can modulate the risk of disease from monogenic mutations for three major diseases: breast cancer; coronary artery disease and colorectal cancer.
For breast cancer, using polygenic information provides further discrimination of risk for BRCA mutation carriers. Those carriers with the highest polygenic risk had an almost 80% risk of getting breast cancer by age 75, compared to around 30% risk for those with the highest polygenic scores who weren’t also carrying a BRCA mutation. This has real world impact. Knowing that your have a four out of five lifetime chance of getting breast cancer may lead you to have annual mammograms, or even choose — like Angelina Jolie — to have an elective mastectomy.
Don’t forget the environment
Whilst genetics clearly influences breast cancer risk, environment still plays a leading role. In the case of breast and ovarian cancer, regardless of genetics, menopausal women who use combined hormone replacement therapy for more than 5 years have twice the risk of getting breast cancer as non-users, irrespective of genetic predisposition. In an intriguing new analysis of UK Biobank data, individuals who followed healthy lifestyles had lower risk of breast cancer even in high genetic risk groups. This suggests that we can go some way to counteracting genetic risk of disease by modifying our environment.
At Allelica we are building risk prediction models incorporating information on non-genetic factors. Our PRS for breast cancer implements an improved algorithm that outperforms the predictive power of the PRS outlined above. Users can request a demo here of the software we used to build and compute the new breast cancer PRS.