Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Sep 1.
Published in final edited form as: Ann Intern Med. 2020 Dec 1;174(3):408–412. doi: 10.7326/M20-5874

Cases in Precision Medicine: The Role of Polygenic Risk Scores in Breast Cancer Risk Assessment

Nur Zeinomar 1, Wendy K Chung 2,3
PMCID: PMC7965355  NIHMSID: NIHMS1670979  PMID: 33253037

Abstract

Polygenic risk scores (PRS) have been consistently associated with elevated breast cancer risk in cohort studies and are associated with risk in both women with and without a family history of breast cancer. However, before clinical implementation we need to address several issues including understanding the potential clinical utility and the optimal method to communicate personalized screening recommendations that incorporate the PRS. Several trials are currently underway to answer some of these questions and facilitate clinical implementation. Since these PRSs have been developed in European ancestry women, an important question is to understand the limitations of the predictive ability of these PRSs to other ancestral groups. Finally, the value of PRS will lie in considering it along with other clinical, familial, and rare genetic factors that are currently used in personalized risk assessment of breast cancer.


The following hypothetical patient illustrates the potential use of emerging polygenic risk scores (PRSs) to improve breast cancer risk assessment. The patient is a 40-year old non-Hispanic White woman with a family history including her mother diagnosed with breast cancer at age 40 years, and her sister who was diagnosed at age 45. Her mother and sister have been tested and do not have any pathogenic mutations in the breast cancer genes BRCA1 and BRCA2, although their carrier status in other high-penetrance genes associated with breast cancer is unknown. The patient is not currently taking any medications although she has previously used hormonal oral contraceptives for 10 years. She gave birth to her only child when she was 28 years old. She is obese with a body mass index of 31 kg/m2 and reports not engaging in any physical activity. Additionally, she does not have a history of benign breast biopsies including non-proliferative lesions, atypical hyperplasia or lobular carcinoma in situ. She is concerned about her breast cancer risk given her family history and asks if you can test for additional genetic factors to assess her risk. The patient believes knowledge is power and believes she can deal with some ambiguity if a variant of uncertain significance (VUS) is identified. Additionally, she has recently learned about PRSs from direct-to-consumer advertising and wants your advice on whether these tests can inform her screening decisions and risk assessment.

What are polygenic risk scores?

For over a decade, genome wide association studies have examined the role of low-penetrance common genetic variants in disease risk including cancer risk. While genome wide association studies have identified associations of individual common variants, or single nucleotide polymorphisms (SNPs), these variants individually confer minimal risk for disease and therefore have limited clinical utility. Aggregating many of these common low-penetrance variants into a weighted PRS improves discrimination for many diseases including breast cancer, coronary artery disease, and diabetes (1). The predictive ability of PRSs is an area of active research with the potential to improve clinical management and decision making by improving disease risk assessment, guiding selection of therapeutic or preventative interventions, informing individual screening decisions, or by refining penetrance estimates for highly penetrant monogenic mutations.

One common approach for estimating the PRS is as an estimated sum of the number of alleles an individual carries weighted by the effect size for each given locus to provide one overall risk score for each individual. Several different methods to determine which variants to include in the PRS and their given weights are used to calculate the PRS. These methods generally account for essential factors including how to adjust the effect sizes which may be inflated for statistical reasons such as overfitting and controlling for linkage disequilibrium between SNPs (2)(for a detailed tutorial on performing PRS analyses, see(2)). Additionally, other approaches of estimating the PRS also take into account the allele frequencies from the populations under study. Regardless of the approach, PRSs have been consistently associated with diseases including breast cancer in large consortia studies (3). The areas under the receiver operator curves (AUC), which are used to measure the discrimination of the PRS, or the ability to correctly classify women with and without breast cancer, have been modest (AUCs: 0.60 – 0.64) (1, 4). However, despite modest AUCs, PRSs have potential to improve the identification of individuals who are at higher risk for cancer. For example, a recent evaluation of the clinical utility of PRS in identifying high-risk individuals from the UK biobank identified 40% of the study participants as having more than a two-fold elevated risk for at least one site-specific cancer, underscoring the potential clinical utility even with modest discrimination (5). Additionally, PRSs can define lifetime trajectories of disease greater than age alone, and these estimated trajectories can theoretically be calculated from birth (6). Although it is important to note that for many complex diseases, these trajectories are not deterministic and can in many cases be modified by lifestyle and other factors, throughout the life-course.

What role can polygenic risk scores play in breast cancer risk estimation?

For breast cancer, while several PRSs have been developed, the PRS with the highest discriminatory ability incorporates information on 313 genetic variants (PRS313) and has been consistently associated with elevated breast cancer risk (3). For example, compared with women in the middle quintile (at population-average risk), those in the highest 10% and 1% of PRS313 had a nearly two-fold and four- fold greater risk of developing breast cancer, respectively (3). The lifetime absolute risk of developing breast cancer by age 80 years ranged from approximately 2% for those in the lowest 1% of PRS313 to over 30% for those in the highest 1% of the PRS313 distribution. Although the observed association was attenuated in women with a family history, this association was present both in women with and without a family history of breast cancer, suggesting that the association of PRS with breast cancer risk is independent of family history and that both PRS and family history should be considered in risk estimation (4, 7). Further, several studies have demonstrated a potential role for PRSs developed using population-level data in refining risk estimates for BRCA1 and BRCA2 mutation carriers, with the PRSs showing large differences in absolute risk (810). For example, BRCA1 mutation carriers in the lowest and highest 5% of the PRS distribution had an estimated absolute risk of breast cancer to age 80 years of 59% and 83%, respectively, while the predicted absolute risks to age 50 years were 31% and 58%, respectively (9). These findings are also independent of family history and BRCA1 pathogenic variant, suggesting that PRS needs to be considered along with family history and BRCA1 pathogenic variant for a comprehensive risk assessment (9). Moreover, in addition to family history, the PRSs have been shown to be independent of classical breast cancer risk factors including reproductive factors (age at menarche, parity, age at first birth, breastfeeding), exogenous hormonal factors (use of oral contraceptives, use of menopausal hormonal therapy), and lifestyle factors (body mass index, alcohol consumption, cigarettes smoking) (11). These findings suggest that the absolute risk of breast cancer for these risk factors will be greater for women at the higher end of the underlying risk spectrum (either determined by family history or PRS) (11).

It is important to note that PRSs need to be combined with established non-genetic risk factors in order to be clinically meaningful. Considering all known risk factors together, the PRS along with lifestyle, hormonal, reproductive, and anthropomorphic risk factors have recently been incorporated into existing breast cancer risk prediction models such as the Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA), a pedigree-based model that includes family history information and rare and common genetic variants (12). However, this newer version of BOADICEA has not been widely validated. Additional validated breast cancer risk models commonly used in clinical settings are the Breast Cancer Risk Assessment Tool (BCRAT, also known as the Gail model) and International Breast Cancer Intervention Study Model (IBIS, also known as the Tyrer-Cuzick model) (1317). The Gail model is recommended for general population use for women ≥ 35 years of age (not for carriers of high-risk mutations in BRCA1 and BRCA2 genes or women with a strong family history) and uses several clinical factors (age, race, age at menarche, age at first live birth, number of first-degree relatives with breast cancer, number of breast biopsies, and diagnosis of atypical hyperplasia) to provide 5-year and lifetime risks of breast cancer. A five-year risk ≥ 1.67 % has been used to determine eligibility for use of chemoprevention. The Tyrer-Cuzick model uses a comprehensive definition of family history along with other clinical factors (age, reproductive history, BMI, exogenous hormone exposure, history of breast biopsies, breast density, and genetic test results of BRCA1 and BRCA2) to estimate both lifetime and 10-year breast cancer risks. Incorporating PRS into both of these models has been shown to improve the model’s discrimination for predicting breast cancer, but these findings need to be validated (1821).

Finally, it is important to note that a major limitation is that the majority of these PRSs have been developed and validated in women of European ancestry, and therefore their prediction accuracy in non-European ancestry populations is unclear. In fact, studies in Black and LatinX populations have reported mixed results (2225). Studies examining European ancestry–based PRSs in African ancestry women found the PRS did not perform well and did not have similar discriminatory ability for breast cancer risk prediction (22, 23).On the other hand, European-based PRSs have been found to perform well in Latinas (25) and women of Asian-ancestry (24). Notably, the study in Latinas examining PRS and breast cancer risk was mainly comprised of women from the western US, Central, and South America, and may not be generalizable to Caribbean Latinas and other Latinas who have a higher proportion of African ancestry (25). As such, it is not yet clear how generalizable these PRSs are to women of non-European ancestries, particularly women with a higher proportion of African ancestry, and therefore we must be cautious in the implementation of PRSs so as not to provide inaccurate information and exacerbate existing health disparities.

Two commercial labs are currently returning PRS results (26, 27). Both of the currently available PRSs for clinical use analyze the genetic polymorphisms along with estimates from risk prediction models (specifically the Tyrer-Cuzick model) that incorporate both clinical and familial risk factors. Both scores are limited to women under age 85 years, with no personal history of atypical hyperplasia or lobular carcinoma in situ, who do not carry a mutation in a highly penetrant breast cancer risk gene and are of European ancestry. Additionally, both scores can only be added to multigene panels and are not available as a standalone test. There is no clinical standard on how to calculate the PRS such as which variants to include or how many variants should be included; one laboratory currently includes 86 SNPs while a second laboratory includes 100 SNPs. As such, the same person could and has received different scores and different estimated risks for developing breast cancer which would only obfuscate clinical recommendations.

Can PRS information be used to aid clinical practice now?

A key point for implementation of the PRS into clinical practice is understanding how it will impact a single individual, including communicating and translating relative risks into absolute risks that can be used to aid decision making. Several clinical trials are currently underway to examine the potential clinical utility of PRSs in providing personalized screening recommendations based on individual risk assessments including the Women Informed to Screen Depending on Measures of Risk study (WISDOM; https://clinicaltrials.gov/ct2/show/results/NCT02620852 ), My Personalized Breast Screening study (MyPeBS; https://clinicaltrials.gov/ct2/show/NCT03672331) as well as a trial examining the utility of a PRS along with multigene panel testing in breast cancer risk management recommendations (https://clinicaltrials.gov/ct2/show/NCT03688204 ). Findings from WISDOM and MyPeBS will facilitate clinical implementation of PRSs and answer some fundamental questions related to the incremental benefit and risks of personalized screening (that incorporates a PRS) over current routine screening and whether inclusion of a PRS improves breast cancer prevention. While these trials have focused on including women in age groups for which screening is recommended, they exclude young women (<40 years of age) and will not be generalizable to younger women who may qualify for enhanced screening. Additionally, MyPeBs is a European-based trial that excludes high-risk women, including women with a prior history of atypical hyperplasia or lobular carcinoma in situ, so will not be generalizable to these populations. Moreover, we need to determine the optimal method to communicate risk information from the PRS and if women will accept and prefer individualized risk assessments that include a PRS to standard routine screening. These trials are designed to answer these important questions, but until they are complete we have limited data to inform clinical implementation.

What are the next steps for your patient?

Given the high suspicion of hereditary cancer given the early onset breast cancer in two first-degree relatives of the patient, she meets the National Comprehensive Cancer Network (NCCN) guidelines for a consideration of genetic testing (28). Based on the patient’s family history, preference for maximum information, willingness to deal with ambiguity in the event of identifying a variant of uncertain significance,, you order a multigene panel test that includes other high and moderate risk genes for breast cancer including ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, MRE11A, MUTYH, NBN, NF1, PALB2, PTEN, RAD50, RAD51C, RAD51D, and TP53. PRSs, by contrast, are focused on estimating a composite risk score from hundreds to millions of low-penetrance polymorphisms across the genome that is independent of the high/moderate risk variants identified from gene panels.

The panel result finds that your patient does not have a mutation in any of these genes on the panel. You then review the detailed personal history, clinical and family history information to compute a lifetime breast cancer risk score using the Tyrer-Cuzick model (14). You find that her 10-year risk score for breast cancer is 4.0% and her lifetime risk score of 25.9% are above what is typically considered high risk (> 3.4% for 10-year risk and > 20% for lifetime risk). You discuss the benefits and risks of incorporating a PRS into this risk assessment and the lack of data from large trials or a validated standard set of markers and weighting function. You emphasize that at this time, we do not have enough information on how to accurately interpret the PRS in a clinical setting. You also cite concerns about whether this test will be reimbursed by her insurance and whether the findings from this test will be accepted by the insurance company to cover earlier or more expensive screening methods like breast MRI. You discuss strategies for risk reduction given her elevated risk based on the Tyrer-Cuzick model, including chemoprevention and enhanced screening including MRI. After a full discussion of the uncertainties and limited data, she still advocates for PRS testing to see whether this additional information can guide her decision to pursue risk-reduction options, so you order the PRS as an add on to the panel gene testing at the clinical lab. The results from the lab returns a lifetime risk of 16.1% to age 85. This result puts her below what is considered high risk (> 20% lifetime risk) for MRI screening. You explain that your earlier recommendation is not dramatically changed by the addition of the PRS and that given her increased risk based on the earlier assessment she may still benefit from enhanced breast cancer screening such as MRI screening. You stress that as more standardized and robust PRS scores are validated and that as her risk estimates change with changes in family history and additional life course data such as her age at menopause and changes in BMI, smoking, and drinking, you will continue to reassess her breast cancer risk and adjust your recommendations.

Summary

PRSs have the potential to be powerful precision medicine tools in further refining risk stratification and allowing us to deliver the most appropriate clinical recommendations and care. While PRSs have been shown to stratify breast cancer risk in women of European ancestry, we have more foundational work to do before clinical implementation. Current PRSs that are clinically available cannot replace existing risk assessment methods that include clinical and family history information, and will ultimately add to these existing methods. Further research is needed to improve PRS prediction in non-European ancestries and understand the optimal method for communication and presentation of this information.

Key Summary Points:

  • Polygenic risk scores (PRS) have been associated with up to a 4-fold increase in breast cancer risk for those in the highest 1% of the PRS distribution compared to those at average risk.

  • PRSs have been able to stratify risk even in women who carry mutations in high-penetrance breast cancer genes such as BRCA1 and BRCA2.

  • The majority of PRSs have been developed in women of European ancestry and the performance and clinical utility of the PRS in non-European ancestry women is unclear and will be a major limitation to the clinical implementation.

  • The PRS is not yet ready for clinical implementation and PRSs are currently being evaluated in large trials that will examine their clinical utility and how to effectively communicate this risk to women, which will be an important step toward implementation.

  • Clinicians should continue to consider family history, presence of pathogenic mutations in high-penetrance genes, and classical risk factors in their personalized approach to breast cancer risk assessment in women. Even when the PRS is ready for clinical implementation, it will be a complement to existing models and strategies.

Acknowledgments

Funding: Dr. Zeinomar is supported by the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health grant TL1TR001875. Dr. Chung is supported by UL1TR001873. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Footnotes

NOTE: This is the prepublication, author-produced version of a manuscript accepted for publication in Annals of Internal Medicine. This version does not include post-acceptance editing and formatting. The American College of Physicians, the publisher of Annals of Internal Medicine, is not responsible for the content or presentation of the author-produced accepted version of the manuscript or any version that a third party derives from it. Readers who wish to access the definitive published version of this manuscript and any ancillary material related to this manuscript (e.g., correspondence, corrections, editorials, linked articles) should go to Annals.org or to the print issue in which the article appears. Those who cite this manuscript should cite the published version, as it is the official version of record.

References

  • 1.Khera AV, Chaffin M, Aragam KG, Haas ME, Roselli C, Choi SH, et al. Genome-wide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations. Nat Genet. 2018;50(9):1219–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Choi SW, Mak TS, O’Reilly PF. Tutorial: a guide to performing polygenic risk score analyses. Nat Protoc. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Mavaddat N, Michailidou K, Dennis J, Lush M, Fachal L, Lee A, et al. Polygenic Risk Scores for Prediction of Breast Cancer and Breast Cancer Subtypes. Am J Hum Genet. 2019;104(1):21–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mavaddat N, Michailidou K, Dennis J, Lush M, Fachal L, Lee A, et al. Polygenic Risk Scores for Prediction of Breast Cancer and Breast Cancer Subtypes. Am J Hum Genet. 2019;104(1):21–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jia G, Lu Y, Wen W, Long J, Liu Y, Tao R, et al. Evaluating the Utility of Polygenic Risk Scores in Identifying High-Risk Individuals for Eight Common Cancers. JNCI cancer spectrum. 2020;4(3):pkaa021–pkaa. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Lambert SA, Abraham G, Inouye M. Towards clinical utility of polygenic risk scores. Hum Mol Genet. 2019;28(R2):R133–R42. [DOI] [PubMed] [Google Scholar]
  • 7.Mavaddat N, Pharoah PD, Michailidou K, Tyrer J, Brook MN, Bolla MK, et al. Prediction of breast cancer risk based on profiling with common genetic variants. J Natl Cancer Inst. 2015;107(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Gallagher S, Hughes E, Wagner S, Tshiaba P, Rosenthal E, Roa BB, et al. Association of a Polygenic Risk Score With Breast Cancer Among Women Carriers of High- and Moderate-Risk Breast Cancer Genes. JAMA Network Open. 2020;3(7):e208501–e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Barnes DR, Rookus MA, McGuffog L, Leslie G, Mooij TM, Dennis J, et al. Polygenic risk scores and breast and epithelial ovarian cancer risks for carriers of BRCA1 and BRCA2 pathogenic variants. Genet Med. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kuchenbaecker KB, McGuffog L, Barrowdale D, Lee A, Soucy P, Dennis J, et al. Evaluation of Polygenic Risk Scores for Breast and Ovarian Cancer Risk Prediction in BRCA1 and BRCA2 Mutation Carriers. J Natl Cancer Inst. 2017;109(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kapoor PM, Mavaddat N, Choudhury PP, Wilcox AN, Lindström S, Behrens S, et al. Combined associations of a polygenic risk score and classical risk factors with breast cancer risk. J Natl Cancer Inst. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Lee A, Mavaddat N, Wilcox AN, Cunningham AP, Carver T, Hartley S, et al. BOADICEA: a comprehensive breast cancer risk prediction model incorporating genetic and nongenetic risk factors. Genet Med. 2019;21(8):1708–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81(24):1879–86. [DOI] [PubMed] [Google Scholar]
  • 14.Tyrer J, Duffy SW, Cuzick J. A breast cancer prediction model incorporating familial and personal risk factors. Stat Med. 2004;23(7):1111–30. [DOI] [PubMed] [Google Scholar]
  • 15.Costantino JP, Gail MH, Pee D, Anderson S, Redmond CK, Benichou J, et al. Validation studies for models projecting the risk of invasive and total breast cancer incidence. J Natl Cancer Inst. 1999;91(18):1541–8. [DOI] [PubMed] [Google Scholar]
  • 16.Tice JA, Cummings SR, Smith-Bindman R, Ichikawa L, Barlow WE, Kerlikowske K. Using clinical factors and mammographic breast density to estimate breast cancer risk: development and validation of a new predictive model. Ann Intern Med. 2008;148(5):337–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Tice JA, Miglioretti DL, Li CS, Vachon CM, Gard CC, Kerlikowske K. Breast Density and Benign Breast Disease: Risk Assessment to Identify Women at High Risk of Breast Cancer. J Clin Oncol. 2015;33(28):3137–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Dite GS, MacInnis RJ, Bickerstaffe A, Dowty JG, Allman R, Apicella C, et al. Breast Cancer Risk Prediction Using Clinical Models and 77 Independent Risk-Associated SNPs for Women Aged Under 50 Years: Australian Breast Cancer Family Registry. Cancer Epidemiology Biomarkers & Prevention. 2016;25(2):359–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Fung SM, Wong XY, Lee SX, Miao H, Hartman M, Wee H-L. Performance of Single-Nucleotide Polymorphisms in Breast Cancer Risk Prediction Models: A Systematic Review and Meta-analysis. Cancer Epidemiology Biomarkers & Prevention. 2019;28(3):506–21. [DOI] [PubMed] [Google Scholar]
  • 20.Cuzick J, Brentnall AR, Segal C, Byers H, Reuter C, Detre S, et al. Impact of a Panel of 88 Single Nucleotide Polymorphisms on the Risk of Breast Cancer in High-Risk Women: Results From Two Randomized Tamoxifen Prevention Trials. Journal of Clinical Oncology. 2017;35(7):743–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Vachon CM, Pankratz VS, Scott CG, Haeberle L, Ziv E, Jensen MR, et al. The Contributions of Breast Density and Common Genetic Variation to Breast Cancer Risk. JNCI: Journal of the National Cancer Institute. 2015;107(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Allman R, Dite GS, Hopper JL, Gordon O, Starlard-Davenport A, Chlebowski R, et al. SNPs and breast cancer risk prediction for African American and Hispanic women. Breast Cancer Res Treat. 2015;154(3):583–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Wang S, Qian F, Zheng Y, Ogundiran T, Ojengbede O, Zheng W, et al. Genetic variants demonstrating flip-flop phenomenon and breast cancer risk prediction among women of African ancestry. Breast Cancer Res Treat. 2018;168(3):703–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Ho WK, Tan MM, Mavaddat N, Tai MC, Mariapun S, Li J, et al. European polygenic risk score for prediction of breast cancer shows similar performance in Asian women. Nat Commun. 2020;11(1):3833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Shieh Y, Fejerman L, Lott PC, Marker K, Sawyer SD, Hu D, et al. A Polygenic Risk Score for Breast Cancer in US Latinas and Latin American Women. J Natl Cancer Inst. 2020;112(6):590–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Black MH, Li S, Laduca H, Chen J, Hoiness R, Gutierrez S, et al. 2018;https://www.ambrygen.com/file/material/view/704/AmbryScore_0318_final.pdf accessed on October 1, 2020.
  • 27.https://myriadmyrisk.com/riskscore/#riskscore accessed on October 1, 2020.
  • 28.National Comprehensive Cancer Network. Genetic/Familial High-Risk Assessment: Breast, Ovarian, and Pancreatic (Version 1.2021) https://www.nccn.org/professionals/physician_gls/pdf/genetics_bop.pdf accessed on October 11, 2020. [DOI] [PubMed]

RESOURCES