Genome-based benchmarking of 177 different breast cancer polygenic risk scores
This study benchmarks 177 breast cancer polygenic risk scores in a hereditary breast and ovarian cancer cohort, demonstrating that top-performing scores converge on similar accuracy and variant composition, thereby offering a cost-effective tool to stratify risk for patients lacking monogenic findings and close a critical diagnostic gap.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Breast cancer is the most frequently diagnosed cancer in women worldwide, and for decades, doctors have relied on age and family history to guess who is most at risk. While we know that some families carry specific, powerful genetic mutations that drastically increase the chance of developing the disease, these mutations are rare. Most women who meet the strict criteria for high-risk screening do not carry these known mutations, leaving them in a difficult middle ground where their risk is unclear. To fill this gap, scientists have developed a tool called a polygenic risk score. Instead of looking for a single broken gene, this tool adds up the tiny effects of thousands of common genetic variations scattered across a person's DNA. By combining these small signals, the score creates a single number that estimates a person's overall genetic likelihood of developing breast cancer, offering a way to refine risk predictions for those who do not have a clear genetic cause.
A team of researchers at the University of Tübingen in Germany recently tackled a major hurdle in using these scores: there are now hundreds of different versions available, each built using different methods and sets of genetic markers, and no one knew which one was best for clinical use. To solve this, the scientists gathered a massive group of 775 women who had been diagnosed with breast cancer and met the criteria for hereditary testing, along with 1,313 women of similar background who had never had the disease. All participants had their entire genomes sequenced using high-quality clinical methods. The researchers then took 177 different polygenic risk scores from a public database and calculated a score for every single person in the study using the exact same computer pipeline. This allowed them to compare how well each of the 177 versions could distinguish between the women with cancer and the women without.
The results showed that the best-performing scores were remarkably similar in their ability to predict risk, creating a clear plateau of effectiveness. The top scores could correctly separate women with breast cancer from those without with a high degree of accuracy, far better than random chance. When the researchers looked specifically at the women who did not have any of the known high-risk mutations, the best scores identified a group where the odds of having breast cancer were more than five times higher than in the lowest-risk group. This finding is particularly important because it targets the patients who currently have the least guidance; for these women, the genetic test reveals a hidden layer of risk that family history alone often misses. The study also found that these scores worked best for the most common types of breast cancer, which are driven by hormones, but were less effective for a more aggressive type that lacks hormone receptors.
Perhaps the most surprising discovery was that the specific genetic markers used in the top scores were largely the same, even though the scores were created by different research teams using different mathematical approaches. The top 20 performers all relied on a shared core of genetic variations, suggesting that the science has converged on the most important signals. Interestingly, the researchers also found that a very compact score using only a few hundred markers performed nearly as well as scores using millions of markers. This means that for hospitals already performing full genome sequencing to look for rare mutations, adding a risk score calculation is a low-cost step that requires no extra lab work. The study concludes that the choice of which specific score to use is less critical than the decision to use one at all, providing a clear path for doctors to integrate these tools into routine care to better protect women who fall through the cracks of traditional testing.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.