Enhancing the BOADICEA cancer risk prediction model to incorporate new data on RAD51C, RAD51D, BARD1 updates to tumour pathology and cancer incidence

Abstract

Background: BOADICEA (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) for breast cancer and the epithelial tubo-ovarian cancer (EOC) models included in the CanRisk tool (www.canrisk.org) provide future cancer risks based on pathogenic variants in cancer-susceptibility genes, polygenic risk scores, breast density, questionnaire-based risk factors and family history. Here, we extend the models to include the effects of pathogenic variants in recently established breast cancer and EOC susceptibility genes, up-to-date age-specific pathology distributions and continuous risk factors.

Methods: BOADICEA was extended to further incorporate the associations of pathogenic variants in BARD1, RAD51C and RAD51D with breast cancer risk. The EOC model was extended to include the association of PALB2 pathogenic variants with EOC risk. Age-specific distributions of oestrogen-receptor-negative and triple-negative breast cancer status for pathogenic variant carriers in these genes and CHEK2 and ATM were also incorporated. A novel method to include continuous risk factors was developed, exemplified by including adult height as continuous.

Results: BARD1, RAD51C and RAD51D explain 0.31% of the breast cancer polygenic variance. When incorporated into the multifactorial model, 34%-44% of these carriers would be reclassified to the near-population and 15%-22% to the high-risk categories based on the UK National Institute for Health and Care Excellence guidelines. Under the EOC multifactorial model, 62%, 35% and 3% of PALB2 carriers have lifetime EOC risks of <5%, 5%-10% and >10%, respectively. Including height as continuous, increased the breast cancer relative risk variance from 0.002 to 0.010.

Conclusions: These extensions will allow for better personalised risks for BARD1, RAD51C, RAD51D and PALB2 pathogenic variant carriers and more informed choices on screening, prevention, risk factor modification or other risk-reducing options.

Keywords: genetic counseling.

Figure 1

Predicted risks by age for a female born in 1985 with an unknown family history based on pathogenic variant carrier status for the new genes in the model. Figure (A) shows the breast cancer risk for carriers of pathogenic variants in BARD1, RAD51C and RAD51D along with the population risk. Figure (B) shows the ovarian cancer risk for carriers of pathogenic variants in PALB2 along with the population risk. Predictions are based on UK cancer incidences.

Figure 2

Predicted lifetime cancer risks (from age 20–80 years) for a female born in 1985 with a pathogenic variant in BARD1 (breast cancer risk), RAD51C (breast cancer risk), RAD51D (breast cancer risk) and PALB2 (ovarian cancer risk) on the basis of the different predictors of risk (pathogenic variant (PV) status, questionnaire-based risk factors (QRFs), mammographic density (MD) and polygenic risk score (PRS)). All figures show the probability density against the absolute risk. Figures (A), (C), (E) and (G) show risks for a female with unknown family history, while Figures (B), (D), (F) and (H) show risks where the individual’s mother has had cancer at age 50. The backgrounds of the graphs are shaded to indicate the risk categories. For breast cancer, these are the categories defined by the National Institute for Health and Care Excellence familial breast cancer guidelines: (1) near-population risk shaded in pink (<17%), (2) moderate risk shaded in yellow (≥17% and<30%) and (3) high risk shaded in blue (≥30%). For ovarian cancer, the categories are: (1) near-population risk shaded in pink (<5%), (2) moderate risk shaded in yellow (≥5% and <10%) and (3) high risk shaded in blue (≥10%). Predictions were based on UK cancer incidences. The line-labelled population denotes the average population risk in the absence of knowledge of family history, PV status, RFs or a PRS. All figures assume the population distributions of QRFs and MD.

Figure 3

The tumour pathology proportions in the general population and among carriers of pathogenic variants (PVs) in the breast cancer (BC) susceptibility genes included in the BOADICEA (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) model. Figure (A) shows the proportion of oestrogen-receptor-negative (ER−) tumours among all tumours, and figure (B) shows the proportion of triple-negative (TN) (ER−, progesterone receptor-negative and human epidermal growth factor receptor 2) tumours among ER− tumours. The general population, BRCA1 PV and BRCA2 PV values are the same as previously used in the model, while those for the other genes are updated using recent BRIDGES data.

Figure 4

The probabilities of carrying a pathogenic variant estimated by BOADICEA (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) model in the genes PALB2, CHEK2, ATM, BARD1, RAD51C and RAD51D for an affected female born in 1985 as a function of her age at diagnosis based on different tumour pathology. Figures (A), (C), (E) and (G) show the probabilities based on the updated proportions (current model), while figures (B), (D), (F) and (H) are based on the previously assumed tumour pathology proportions (previous model version) and where proportions for BARD1, RAD51C and RAD51D, which were not in the previous model, are assumed to be the same as in the general population. In figures (A) and (B), the woman has had an oestrogen receptor-positive (ER+) tumour; in figures (C) and (D), the female has had an oestrogen receptor-negative (ER−) tumour, but the triple-negative (TN) status is unknown; in figures (E) and (F), the woman has had an ER− tumour that is not TN and in figures (G) and (H), the woman has had a TN tumour. Predictions are based on UK cancer incidences. BC, breast cancer.

Figure 5

Predicted lifetime breast and ovarian cancer risks as a function of height for a female born in 1985 with unknown family history, comparing the updated model, where height is treated as continuous, to the previous model, where height was treated as categorical. Figures (A), (C) and (E) show breast cancer, while figures (B), (D) and (F) show ovarian cancer risks. Figures (A) and (B) show the predicted risk as a function of height, while figures (C) and (D) show the probability density/mass of risk as a function of height. Predictions are based on UK cancer incidences. Figures (E) and (F) show the log (base 10) of the root-mean-squared relative discretisation error as a function of the number of bins. The error was taken to be the absolute difference between the value and the asymptotic extrapolation of the measurements as a function of the number of bins. The average is taken over 100 heights that are spaced 1% apart, from 0.5% to 99.5%.