Hundreds of genetic variants can nudge someone’s risk of breast cancer up or down or towards a particular subtype. The studies identifying those gene variants, though, have largely involved people with European ancestry and thus give a less accurate picture of breast cancer risk for people who are not white.
That’s beginning to change. Last week, researchers published a genome-wide association study on breast cancer in roughly 40,000 people of African descent in Nature Genetics, marking a leap forward in scientists’ knowledge of breast cancer genetics in people of African ancestry.
“Before we started this study in 2016, there were just several thousand cases for Black Americans. It was a very small number,” said Wei Zheng, the study’s senior investigator and a cancer epidemiologist at Vanderbilt University. This study combined data from dozens of other studies and included genetic data for thousands of new participants, making it the largest combined breast cancer genetics study done with people with African ancestry.
Specifically, the study compiled data from about 30 different studies investigating breast cancer in African or African American people. About 18,000 of them had breast cancer, while the other 22,000 were healthy controls, and investigators were able to scour their genetic data for specific variations that seemed closely related to breast cancer. The statistical power that comes with such numbers enabled the team to make two key advances.
First, the team found 12 loci, or locations in the genome, that showed a significant association with breast cancer. Of those, the team identified variants of three genes that appear to increase the risk of triple negative breast cancer, one of the most aggressive subtypes. Since everyone has two copies or alleles of each gene, that means someone could have anywhere between one and six risk-related alleles of these three genes. Those who had all six risk-related alleles had roughly double the chance of getting triple negative breast cancer than those who only had three.
That could provide a foothold for scientists to begin predicting who might get this aggressive form of breast cancer, and it might offer an opportunity to better understand the biology of triple negative breast cancer by highlighting genes that seem to be important. “Finally, we have enough data to drill down to estrogen negative and triple negative breast cancer, which are twice as common in the African American population as any other population, said Julie Palmer, an author of the study and a cancer researcher at Boston University.
The other advance came when the researchers used the data to build a breast cancer risk prediction model for people with African ancestry. Such models take into account hundreds of different genetic variants that can slightly push breast cancer risk up, adding them all up into a polygenic risk score.
In the past, these scores always performed better for white people than Black people, mainly because there’s so much more research done in people with European ancestry — a combined total of more than 100,000 participants for breast cancer. Polygenic risk scores have had an AUC, a measure of the model’s performance, of about 0.63 for people with European ancestry compared to 0.58 for the African ancestry population. When researchers combined the data from this study into their new model, however, that figure rose to 0.60. That equates to the model being able to correctly distinguish between someone who’s likely to get breast cancer and someone who isn’t about 60% of the time.
Even if this work is validated in other studies, as it still needs to be, that figure is not too useful for most individuals. An astute observer might note an AUC of 0.63 is only passably better than a coin toss. That’s an indication polygenic risk scores don’t perform as well overall as scientists would like even at their best. When polygenic risk scores are combined with other breast cancer risk factors, like age of first childbirth or breast density, “we’re still not very good at predicting breast cancer,” Palmer said.
But research is continually improving on that. The hope is, one day, these scores will help scientists build tools that can reliably distinguish people who are more likely to get breast cancer — and thus might have more to gain by beginning screening earlier or more frequently. Or, conversely, they could help weed out people who aren’t likely to get breast cancer and could thus screen less. “If you don’t need it, then why do it?” said Laura Fejerman, a cancer researcher and epidemiologist at the University of California, Davis.
Polygenic risk scores might already be able to provide some of that context for a small minority of people, Fejerman added. For the 1% of people with the highest polygenic risk, “their lifetime risk was a little bit above 30%,” Fejerman said. That could be an argument for them to screen more often, even if they had no other risk factors. “If you learn that, you might be more on top of your screening than most people who maybe let three years pass. So, if you could get the highest-risk women to screen every year, that would be a big benefit.”
Without datasets in non-European ancestry populations, other racial demographic groups could be left out of that progress. In that sense, this new paper “is definitely a big step forward for achieving racial equity,” said Swati Biswas, a statistician and cancer researcher at the University of Texas at Dallas who did not work on the study.
In particular, the data are needed if scientists ever hope to create a “unified” polygenic risk score that works for everyone. At the moment, many models rely on racial categorization — Black people use an African ancestry model; white people use a European model. But using such models in clinical practice isn’t optimal, pointed out Jennifer James, a sociologist who studies breast cancer and bioethics at the University of California, San Francisco.
Imagine someone whose ancestry is 5% African and 95% European, but who also happened to inherit breast cancer risk alleles that were only found in the African ancestry population. That would mean the African ancestry polygenic risk model might work better for them, even if they didn’t identify as Black themselves. “You could be 1% Black, but the one thing you got was that allele,” James said. “We need to move towards a unified polygenic risk score.”
That still won’t be enough to end the breast cancer mortality gap between Black and white people, even if scientists created a perfectly accurate polygenic risk model, James added. That’s because part of the reason for the disparity has to do with the health care system writ large, not subtle biological differences across populations.
“We know Black women have a longer time to diagnosis, longer time from diagnosis to treatment,” James said. “If everyone had equal access to healthcare, that would do more to close gaps in mortality than tweaking prediction models. I want when someone finds a lump in their breast or needs a mammogram, they have equal access to care.”