Prospective evaluation of a breast-cancer risk model integrating classical risk factors and polygenic risk in 15 cohorts from six countries

Abstract

Background: Rigorous evaluation of the calibration and discrimination of breast-cancer risk-prediction models in prospective cohorts is critical for applications under clinical guidelines. We comprehensively evaluated an integrated model incorporating classical risk factors and a 313-variant polygenic risk score (PRS) to predict breast-cancer risk.

Methods: Fifteen prospective cohorts from six countries with 239 340 women (7646 incident breast-cancer cases) of European ancestry aged 19-75 years were included. Calibration of 5-year risk was assessed by comparing expected and observed proportions of cases overall and within risk categories. Risk stratification for women of European ancestry aged 50-70 years in those countries was evaluated by the proportion of women and future cases crossing clinically relevant risk thresholds.

Results: Among women <50 years old, the median (range) expected-to-observed ratio for the integrated model across 15 cohorts was 0.9 (0.7-1.0) overall and 0.9 (0.7-1.4) at the highest-risk decile; among women ≥50 years old, these were 1.0 (0.7-1.3) and 1.2 (0.7-1.6), respectively. The proportion of women identified above a 3% 5-year risk threshold (used for recommending risk-reducing medications in the USA) ranged from 7.0% in Germany (∼841 000 of 12 million) to 17.7% in the USA (∼5.3 of 30 million). At this threshold, 14.7% of US women were reclassified by adding the PRS to classical risk factors, with identification of 12.2% of additional future cases.

Conclusion: Integrating a 313-variant PRS with classical risk factors can improve the identification of European-ancestry women at elevated risk who could benefit from targeted risk-reducing strategies under current clinical guidelines.

Keywords: Breast cancer; iCARE; model validation; polygenic risk score; risk prediction; risk stratification.

Figure 1.

Conceptual diagram of the model building and validation of the Individualized Coherent Absolute Risk Estimator (iCARE) for breast cancer. iCARE-BPC3, Individualized Coherent Absolute Risk Estimation model based on Breast and Prostate Cancer Cohort Consortium; iCARE-Lit, iCARE model based on literature review.

Figure 2.

Relative-risk calibration of integrated breast-cancer risk models [with classical risk factors and polygenic risk score (PRS)] based on meta-analysis across validation studies. Classical risk factors include age at menarche, age at first live birth, parity, oral-contraceptive use, age at menopause, hormone-replacement-therapy use, type of hormone replacement therapy, alcohol intake, height, BMI, breast-cancer family history (i.e. presence or absence of breast cancer in at least one first-degree relative) and benign breast disease. Meta-analysis is based on a reduced set of risk factors that were available in the majority of the validation cohorts. History of benign breast disease and type of hormone replacement therapy (iCARE-Lit model for women ≥50 years old) was set to missing for all subjects. Meta-analysis of the iCARE-Lit model for women <50 years old included GS, NHS II and UK Biobank. Meta-analysis of the iCARE-Lit model for women ≥50 years old additionally included CPS-II, EPIC NL, EPIC UK, KARMA, MMHS, NHS, PLCO and WGHS. The AUC estimates were adjusted for age at enrolment. AUC, area under the curve; χ², chi-square goodness-of-fit test statistic; BMI, body mass index; CPS, Cancer Prevention Study; EPIC, European Prospective Investigation into Cancer and Nutrition; GS, Generations Study; iCARE-Lit, iCARE model based on literature review; KARMA, KARolinska MAmmography Project; MMHS, Mayo Mammography Health Study; NHS, Nurses’ Health Study; PLCO, Prostate, Lung, Colorectal, Ovarian Cancer Screening Trial; PRS, polygenic risk score; UK, United Kingdom; WGHS, Women’s Genome Health Study.

Figure 3.

Risk discrimination measured by the model area under the curve (AUC) of the iCARE-Lit models based on a meta-analysis across studies for the risk-factor combinations: classical risk factors only; classical risk factors and polygenic risk score (PRS); and classical risk factors and PRS (including the effect of age). See Figure 2 and Supplementary Figures 11 and 12 (available as Supplementary data at IJE online) for more details. Coloured dots were used to denote estimates and coloured horizontal lines denote the 95% confidence intervals. iCARE-Lit, iCARE model based on literature review; PRS, polygenic risk score.

Figure 4.

Absolute-risk calibration for the integrated iCARE-Lit model with classical risk factors and polygenic risk score for women <50 years old (four cohorts). Risk categories were defined based on deciles of predicted 5-year absolute risk. Classical risk factors include age at menarche, age at first live birth, parity, oral-contraceptive use, age at menopause, hormone-replacement-therapy use, type of hormone replacement therapy, alcohol intake, height, BMI, breast-cancer family history (i.e. presence or absence of breast cancer in at least one first-degree relative) and benign breast disease. BMI, body mass index; E, average of predicted 5-year risk in the highest decile of predicted 5-year risk; GS, Generations Study; NHS, Nurses’ Health Study; O, observed proportion of subjects developing breast cancer within 5 years in the highest decile of predicted 5-year risk; UK, United Kingdom.

Figure 5.

Absolute-risk calibration for the integrated iCARE-Lit model with classical risk factors and polygenic risk score for women ≥50 years old (15 cohorts). Risk categories were defined based on deciles of predicted 5-year absolute risk. Classical risk factors include age at menarche, age at first live birth, parity, oral-contraceptive use, age at menopause, hormone-replacement-therapy use, type of hormone replacement therapy, alcohol intake, height, BMI, breast-cancer family history (i.e. presence or absence of breast cancer in at least one first-degree relative) and benign breast disease. BMI, body mass index; CPS, Cancer Prevention Study; DE, Germany; E, average of predicted 5-year risk in the highest decile of predicted 5-year risk; EPIC, European Prospective Investigation into Cancer and Nutrition; GS, Generations Study; KARMA, KARolinska MAmmography Project; MCCS, Melbourne Collaborative Cohort Study; MMHS, Mayo Mammography Health Study; NHS, Nurses’ Health Study; NL, the Netherlands; O, observed proportion of subjects developing breast cancer within 5 years in the highest decile of predicted 5-year risk; PLCO, Prostate, Lung, Colorectal, Ovarian Cancer Screening Trial; PROCAS, Predicting Risk Of Breast CAncer at Screening; UK, United Kingdom; WGHS, Women’s Genome Health Study.

Figure 6.

Women of European ancestry aged 50–70 years in the general populations of the six countries (Australia, Germany, the Netherlands, Sweden, the UK, the USA) expected to be identified at low and high risk of breast cancer according to two risk thresholds and the incident cases of breast cancer expected to occur in these groups within a 5-year interval. The expected number of women is calculated using 2017 population estimates (N = 2 960 506) from the Australian Bureau of Statistics for Australia, 2016 population estimates (N = 12 024 487) from the Federal Statistical Office for Germany, 2016 population estimates (N = 2 356 691) from the Central Agency for Statistics for the Netherlands, 2016 population estimates (N = 1 249 695) from Statistics Sweden for Sweden, mid-2016 population estimates (N = 8 275 453) from the Office of National Statistics for the UK and mid-2016 population estimates (N = 30 030 821) from the US Census Bureau for the USA. The expected numbers of cases are estimated using the average predicted 5-year risk in each population, calculated using the country-specific breast-cancer-incidence rates and risk-factor distributions (Supplementary Table 4, available as Supplementary data at IJE online). The 1.13% risk threshold corresponds to the average 5-year risk for US women aged 50 years. The 3% threshold is used by the US Preventive Services Task Force for recommending risk-reducing medications. Classical risk factors correspond to the iCARE-Lit model and include age at menarche, age at first live birth, parity, oral-contraceptive use, age at menopause, hormone-replacement-therapy use, type of hormone replacement therapy, alcohol intake, height, BMI, breast-cancer family history (i.e. presence or absence of breast cancer in at least one first-degree relative) and benign breast disease. AR, absolute risk; BMI, body mass index; PRS, polygenic risk score; UK, United Kingdom; US, United States.