A polygenic risk score (PRS) created using genetic information from more than 35,000 women of African ancestry more accurately predicts breast cancer risk in these women than standard models that typically use data from women of European descent.
“The developed PRS models for overall breast cancer and its subtypes are robust for predicting breast cancer risk in women of African ancestry,” said Dezheng Huo, PhD, professor of public health sciences at the University of Chicago and senior author of the study. “These models could support more precise risk assessment and tailored breast cancer screening, which may help reduce racial disparities in breast cancer mortality.”
Huo and co-investigators explain in Nature Genetics that although PRS models, which estimate inherited breast cancer risk based on the combined effects of many genetic variants across a person’s genome, have emerged as valuable tools in women of European-ancestry, they have had limited accuracy for women of African-ancestry.
Indeed, a 313-variant PRS developed for women of European ancestry demonstrated “considerable predictive performance,” with an area under receiver operating characteristic curve (AUC) of 0.60–0.64, on a scale of 0–1 with higher numbers indicating better performance. Yet, the same PRS only achieved AUCs of 0.56–0.59 when used in women of African ancestry.
The disparity is largely driven by differences in the distribution of common genetic variants between ancestries, say Huo et al., who therefore set out to develop an accurate PRS specifically for women of African ancestry, noting that these women are significantly more likely to die from breast cancer than women of European ancestry.
The researchers applied several different PRS methods to data from the African Ancestry Breast Cancer Genetics consortium, which included 17,391 women with breast cancer and 18,800 controls from the U.S., Caribbean, and sub-Saharan Africa.
The best models across all methods achieved an AUC of 0.612 for predicting overall breast cancer risk, 0.621 for estrogen receptor (ER)-positive breast cancer, 0.611 for ER-negative breast cancer, and 0.639 for triple-negative breast cancer (TNBC), with the performance maintained in external validation studies.
“By leveraging information across ancestries, subtypes and PRS methodologies, we introduced PRS models that currently have the highest reported predictive performance in women of African ancestry,” Huo et al. remark.
Huo told Inside Precision Medicine, that “AUC provides an overall measure of the ability to distinguish diseased and healthy individuals, but it does not indicate other aspects of clinical utility. An AUC in the range of 0.61–0.65 suggests that we cannot completely separate breast cancer cases from healthy individuals; however, individuals in the extreme high end of the PRS spectrum are likely to have a significantly increased chance of developing breast cancer.”
He added: “The remaining uncertainty (0.35–0.39) may be addressed by non-genetic risk factors, other genetic factors, known (e.g. pathogenic variants in high-penetrance genes) and unknown, and chance alone.”
“Regarding ‘chance alone,’ it is important to note that we do not expect an AUC of 1.0 for risk assessment, as it aims to predict future events. This is different from current diagnosis, which requires a very high AUC.”
The researchers showed that the risk for breast cancer increases significantly with in line with PRS. For example, individuals in the top 5th percentile of PRS scores had a significant 2.4-fold higher risk for breast cancer overall, and 2.4- 2.1-, and 2.4-fold increased risks for ER-positive, ER-negative, and TNBC, respectively, compared with those in the 40–60th percentile.
The data also suggest that women of African ancestry in the highest percentiles of PRS could benefit from earlier screening. The American Cancer Society currently recommends women to start regular annual mammography screenings at age 45 years, when their estimated average 10-year risk for breast cancer is around two percent. Yet women in the top percentile of PRS in the current study reached this two percent risk threshold at age 32 years.
“This suggests that our PRS models may provide better stratification for determining risk and optimal age for initiating intensive screening,” Huo et al. remark. “Given the lower survival rates associated with TNBC and ER-negative [breast cancer], more intensive screening for women of African-ancestry with high PRS scores could reduce the chance of missing aggressive cancers that may develop between traditionally recommended annual mammography screening intervals for women between the ages of 45 and 54 years and aid in the early detection of cancers that may develop before this recommended screening age.”
The researchers acknowledge that a “key limitation of our study is the scope of African ancestries that are represented within the AABCG dataset, which involves primarily participants who are African American women and of predominantly West African ancestry.” They therefore suggest that further research is needed to develop PRS models that are accurate in these populations.
Nonetheless, “these advanced testing models bring us closer to a future where everyone, no matter their ancestry, gets an equal chance at early detection, effective treatment and survival,” Huo said.
