Improving classical breast cancer risk prediction with single nucleotide polymorphisms and mammographic density

Authors: Elke M. van Veen M.Sc.1, Adam R. Brentnall Ph.D. 2, Helen Byers B.Sc.1, Elaine F. Harkness Ph.D.3,4,5, Susan M. Astley Ph.D.3,4,5,8, Sarah Sampson B.Sc. 2, Anthony Howell MD.3,6,8, William G. Newman M.D., Ph.D.1,7,8, Jack Cuzick Ph.D,2, D. Gareth R. Evans M.D.1,3,6,7,8.

1 Division of Evolution and Genomic Sciences, School of Biological Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester Academic Health Science Centre, Manchester, UK.

2 Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Charterhouse Square, Barts and The London, Queen Mary University of London, London, UK.

3 Prevention Breast Cancer Centre and Nightingale Breast Screening Centre, University Hospital of South Manchester, Manchester, UK.

4 Division of Informatics, Imaging and Data Sciences, Faculty of Biology, Medicine and Health, University of Manchester, Manchester, UK.

5 Manchester Academic Health Science Centre, University of Manchester, Manchester, UK.

6 The Christie NHS Foundation Trust, Manchester, UK.

7 Manchester Centre for Genomic Medicine, Manchester University NHS Foundation Trust, Manchester, UK.

8 Manchester Breast Centre, Manchester Cancer Research Centre, University of Manchester, Manchester, UK.

Corresponding author:

Professor DGR Evans,

Manchester Centre for Genomic Medicine,

St. Mary’s Hospital,

Oxford Road,

Manchester, M13 9WL, UK.

Tel: 44 (0)161 276 6206,

Fax: 44 (0)161 276 6145,

Email: gareth.evans@

HYPERLINK "mailto:gareth.evans@cmft.nhs.uk"

HYPERLINK "mailto:gareth.evans@cmft.nhs.uk"mft.nhs.uk

Word count: 2912

Tweet: 106/117 characters

SNPs perform extremely well in predicting breast cancer and improve the accuracy of risk prediction models

?alternative? Key message is they can be combined?

Genetic test combined with breast density and questionnaire accurate for breast cancer risk stratification

Key points

Question: Can panels of single nucleotide polymorphisms be combined with mammographic density and classical risk factors to improve breast cancer risk assessment?

Findings: In a general-screening cohort a panel of 18 single nucleotide polymorphisms was similarly predictive when unadjusted or adjusted for mammographic density and classical factors (odds ratio per inter-quartile range respectively 1.56, 95%CI 1.38-1.77 and 1.53, 95%CI 1.35-1.74). Meaning: Single nucleotide polymorphism risk panels substantially improve the ability of breast cancer risk prediction models to accurately identify women who may benefit most from preventive therapy or additional screening modalities.Abstract

Importance: Single nucleotide polymorphisms (SNPs) have demonstrated an association with breast cancer susceptibility, but there is limited evidence on how to incorporate them into current breast cancer risk prediction models.

Objective: To determine whether a panel of 18 SNPs (SNP18) may be used to predict breast cancer in combination with classical risk factors and mammographic density.

Design: A case-cohort study within a prospective cohort, set up specifically to evaluate breast cancer risk assessment methods for women attending population-based screening.

Setting: Recruitment from multiple screening centres in Greater Manchester, UK.

Participants: Women aged 46-73 years attending the national program for breast screening, without a previous breast cancer diagnosis, were recruited between 10/2009-06/2015 with follow-up to 01/2017. 466 cases (prevalent=271; incident=195) were included, and a sub-cohort of 8897 women.

Exposures: Genotyping of 18 SNPs, visually-assessment percentage mammographic density and classical risk assessed by the Tyrer-Cuzick risk model from a self-completed questionnaire at cohort entry.

Main Outcome and Measure: The predictive ability of SNP18 for breast cancer diagnosis (invasive and ductal carcinoma in situ) was assessed using logistic regression odds ratios per inter-quartile range of the predicted risk.

Results: SNP18 was similarly predictive when unadjusted or adjusted for mammographic density and classical factors (odds ratio per inter-quartile range respectively 1.56, 95%CI 1.38-1.77 and 1.53, 95%CI 1.35-1.74), with observed risks being very close to expected (adjusted observed to expected odds ratio 0.98, 95%CI 0.69-1.28). A combined risk assessment indicated 18% of the sub-cohort to be at ≥5% 10-year risk, compared with 30% of all, 35% of interval-detected and 42% of stage 2+ cancers, respectively. In contrast, 33% of the sub-cohort were at <2% risk but accounted for only 18%, 17% and 15% of the total, interval and stage 2+ breast cancers, respectively.

Conclusions and Relevance: SNP18 adds substantial information to risk assessment based on the Tyrer-Cuzick model and mammographic density. A combined risk is likely to aid risk-stratified screening and prevention strategies.

Keywords: breast cancer, risk prediction, SNP, mammographic density, Tyrer-Cuzick

Introduction

Breast cancer is the most commonly diagnosed cancer among women worldwide. Approximately half of all breast cancers in women with a family history of the disease are explained by a known genetic component1,2. Pathogenic variants in BRCA1/BRCA2 and single nucleotide polymorphisms (SNPs) explain a large proportion of the risk in women with a strong family history. SNPs also contribute to the development of non-familial breast cancer in women, accounting for ~16 % of genetic risk2. On a population basis the polygenic risk conferred by these susceptibility SNPs is greater than the risk from single pathogenic variants in a single high or moderate risk gene1, especially for women without any family history of breast cancer3,4. Dependent on genotyping of susceptibility SNPs (i.e. 0 risk-alleles, 1 risk-allele or 2 risk-alleles) a risk estimate can be derived, which may be used for risk prediction by combining the risk estimates for each SNP in a polygenic risk score (PRS).

Breast cancer risk models mainly include classical risk factors including increased risk from family history, younger age at menarche, older age at first full-term pregnancy and later menopause, age, body mass index (BMI), benign breast disease and current use of hormone replacement therapy5,6. In addition, high mammographic density is also a well-delineated risk factor for breast cancer and several studies have found that mammographic density improves the accuracy of risk prediction models

ADDIN EN.CITE.DATA7,8. Recent studies have considered the value of including SNP data into risk prediction algorithms with promising results

HYPERLINK \l "_ENREF_9"

ADDIN EN.CITE.DATA9-11, but there is very little data on including both mammographic density and SNP data into risk prediction models12.

We have collected data on classical risk factors and mammographic density from 57,902 women participating in the Predicting Risk Of Cancer At Screening (PROCAS) study, which recruited women attending a national breast cancer screening program. A sub-cohort volunteered to provide saliva samples that were genotyped for 18 breast cancer susceptibility SNPs (SNP18)13. Here, we report on the predictive ability of SNP18 alone and when adjusted for classical risk factors (annotated by the Tyrer-Cuzick model5) and mammographic density, in a case-cohort study, and in a secondary analysis that only included women in the sub-cohort without breast cancer when they provided a saliva sample.

Methods

Patients

57,902 women aged between 46-73 years from the Greater Manchester area were recruited between 10/2009-06/2015. Women were recruited at the time of attendance for mammographic screening in the National Health Service Breast Screening Programme. Breast cancer risk factors were collected by self-completion of a two-page paper questionnaire. Participants were excluded from this study if they had been diagnosed with breast cancer before completing the questionnaire; cancers detected as a result of the screening test were included.

The PROCAS study was approved by the North Manchester Research Ethics Committee (ref. 09/H1008/81) and written informed consent was obtained from each participant.

Specimen characteristics

Saliva samples were collected from 9956 participants after their initial study mammogram at drop-in days in Greater Manchester, using Oragene saliva lysate tubes (DNA Genotek Inc., Ottawa, Ontario, Canada) and DNA extraction was performed using Gen-Probe extraction.

Study design

Women who lived within the smaller defined Withington area (South Manchester) were invited to participate for subsequent risk-assessment including a saliva sample. All women with breast cancer diagnosed after completion of the questionnaire were invited to provide saliva samples and participate as cases. Breast cancer diagnosis (invasive or ductal carcinoma in situ (DCIS)) was at the entry screen or subsequently before 5th January 2017, and was ascertained through monthly updates from (National Health Service) Breast Screening Systems. Saliva samples were collected between 10/2009-12/2013, close to but after the time of the woman’s screening visit.

Assay methods

17 SNPs (Supplementary Table 1) were genotyped by a custom designed Sequenom MassARRAY iPLEX assay (Agena Bioscience GmbH, Hamburg, Germany)14, and one SNP (rs10931936) was genotyped using a TaqMan® SNP Genotyping Assay (Fisher Scientific-UK Ltd, Loughborough, United Kingdom). Two duplicate positive and two negative controls for quality assurance were genotyped for each 96 sample plate.

Statistical methods

A PRS for SNP18 was computed using published per-allele odds ratios (ORs) obtained from the iCOGS database and allele frequencies as described earlier10. Briefly, SNP18 was calculated by multiplying the per-allele odds ratio (OR) for each SNP, and normalising the risk by the average risk expected in the population using published minor allele frequencies10.

Mammographic density at entry to the cohort was estimated independently by two readers using a visual analogue scale (VAS) as previously described8.Briefly, each mammogram was scored on a linear scale ranging for 0-100 for the density of the breast. The derived percent density was adjusted for BMI and age and reported as a ‘density residual’ (DR) and was also expressed as an OR by calibrating and standardizing it to the wider cohort8. Women with bilateral cancer on prevalent study screen or with breast implants had no assessable VAS score and were given a pro-rata DR of 1.0. The Tyrer-Cuzick 10-year risk (v6) was based on classical risk factors from the questionnaire self-reported at entry.

Baseline characteristics were compared between cases and controls, and those included in the case-cohort study or not. Differences between the DR after adjustment for parity were assessed by a Wald test from a linear model

.

The predictive ability (discrimination) of SNP18 was assessed using logistic regression with Wald confidence intervals, and expressed as the OR per inter-quartile range in controls. Calibration of the observed (O) to expected (E) SNP18 odds ratio was estimated using the log score regression coefficient and 95% Wald confidence interval, so that O/E=1 would indicate perfect calibration, and further inspected by SNP18 decile, with confidence intervals following Wilson’s method for the binomial parameter. Adjusted analyses were used to assess SNP18 beyond mammographic density and the Tyrer-Cuzick model. Subgroups were considered for (i) all women without breast cancer at saliva sample (prospective sub-study); (ii) estrogen-receptor (ER) status using a Wilcoxon test, and (iii) DCIS and invasive cancer.

A combined 10-year risk was calculated assuming independence by multiplying the Tyrer-Cuzick 10-year absolute risk by DR and SNP18. It was stratified in 10-year risk groups: <2%; 2.00-3.49%; 3.50%-4.99%, 5.00-7.99% and 8.00%+ risk8, for which the frequency of cases and percentage of controls was determined, and by cancer stage, time of diagnosis, and only those cases diagnosed with breast cancer after sample collection. A sensitivity analysis using computer simulation assessed the predicted percentage of the wider PROCAS cohort with VAS measurements (50,588 participants) in each risk category, based on the results from the case-cohort study (Supplementary Methods).

Area under the receiver operating characteristic (AUC) statistics with 95% DeLong confidence intervals were calculated to assess discrimination.

The number of cancers expected was estimated from Tyrer-Cuzick 10-year risks censored at time of breast cancer diagnosis, death or 5th January 2017, whichever was earliest. Exact Poisson confidence intervals were given for rates. A two-sided P-value less than 0.05 was called significant.

Results

57,902 participants were recruited to the PROCAS cohort, of which 907 were diagnosed with breast cancer before entry. SNP18 was available for 9899 women. After excluding 536 women with breast cancer before entering PROCAS 9363 women were included in the cohort of whom 466 were diagnosed with breast cancer (including 89 with DCIS) at the baseline mammogram or during follow up (Supplementary Figure 1).

The quality of the Sequenom MassARRAY iPLEX and TaqMan® assays was assured as there was 100% concordance of genotyping between duplicate samples for all SNPs.

The baseline characteristics in Supplementary Table 2 show that most women were overweight (BMI>25) and over the age of 56 years. Compared with those in the cohort that were not included, controls were significantly older, less overweight, more likely to have had children when older or not at all, with a family history of breast cancer, and a previous breast biopsy (Supplementary Table 2). Cases included were also slightly older, less overweight and less likely to have children. Because some selection bias was reflected in questionnaire risk factor differences between women who volunteered to donate saliva and those who did not, we only adjusted for the Tyrer-Cuzick model in the main case-cohort analysis, and did not directly assess its predictive ability. There was a significant difference in mammographic density between the controls included and excluded, with a higher average density for those included. However, the difference was mostly explained by BMI and parity (DR adjusted for parity P=0.057).

A non-significant correlation was observed between the SNP18 and DR (Spearman 0.019, P=0.068), but a significant small correlation between SNP18 and the 10-year Tyrer-Cuzick risk was seen (0.031, P=0.003), indicating these risk factors have very small correlations.

SNP18 was almost perfectly calibrated across the spectrum of predicted relative risk sub-groups (unadjusted O/E OR=1.03, 95%CI 0.74-1.32, Figure 1) indicating that SNP18 is a very good predictor across the continuum of risk, and had an unadjusted IQ-OR=1.56 (95%CI 1.38-1.77). Results were very similar after adjustment for the Tyrer-Cuzick model (IQ-OR=1.54, 95% 1.36-1.75; O/E OR=1.00, 95%CI 0.71-1.30) and showed comparable discrimination to mammographic density (adjusted IQ-OR=1.50, 95%CI 1.33-1.70, see also8). Furthermore, additional adjustment for mammographic density also did not substantially affect predictive power of SNP18 (IQ-OR=1.53, 95%CI 1.35-1.74; O/E OR=0.98, 95%CI 0.69-1.28). Similar results were obtained for the sub group of 169 prospective cancers (adjusted O/E OR=1.08, 95%CI 0.60-1.56, Supplementary Table 3). As expected, there was little difference in the performance of SNP18 when ER-negative cancers were excluded (409 ER+ and 14 unknown, Supplementary Table 3) and SNP18 was much less predictive for the 43 ER- cancers (heterogeneity P=0.081). There was also very little difference between SNP18 as a predictor of invasive breast cancer or DCIS (Supplementary Table 3).

When combining the risk from the Tyrer-Cuzick model, mammographic density and SNP18 assuming independence we observed that 16% (76/466) of cases and 9.5% (841/8897) of controls moved into the increased risk category (≥5% 10-year risk) compared with using the Tyrer-Cuzick model alone, but only 5% (22/466) of cases and 4% (353/8897) of controls moved out of this category. Thus the number of cases in this group increased by an absolute 11%, while the number of controls only by 5.5% (Supplementary Table 4).

The ability of a combination of SNP18, mammographic density and the Tyrer-Cuzick model 10-year risk to improve risk stratification is further illustrated in Supplementary Table 5. Table 1 shows that individuals in the highest-risk (>8%) group were >4 times as likely to develop cancer, both on the prevalent mammogram and prospectively than the low-risk (<2%) group. Additionally 14% of the cancers occurred in this group which comprised only 6% of the population (Tables 2, 3). Stage 2 or higher cancers were also more likely to develop in the moderate/high-risk population as 42% of high stage cancers and 35% of all interval cancers were identified in the 19% who were at ≥5% 10-year risk. The moderate/high-risk group were 5-fold more likely to develop a high stage cancer than the low-risk group (P<0.0001). In the NICE defined 8%+ 10-year risk group 22% of the stage 2+ cancers and 16.7% of the interval cancers were identified in just 6% of the population representing a 5-fold risk for interval and an 8-fold risk for high stage (Table 1) compared with the low-risk group. A sensitivity analysis (Supplementary Table 5) to model risk stratification in the wider PROCAS cohort showed a similar trend to the controls (Supplementary Table 5), with 37% in the low-risk category, but proportionally fewer women were assigned to the higher risk groups due to the increased risk of study controls relative to the wider cohort (Supplementary Table 2).

We finally considered a subset of those who were unaffected at the time of DNA sampling, consisting of 9064 women who had a total of 44419 years of follow up to diagnosis or censoring on 5/1/2017. In total 167 breast cancers occurred (including 28 DCIS (16.7%) after a mean 4.9 years follow up at an annual rate of 3.7 per 1000 women (3.2 for invasive cancer only) and 155 cancers were expected from Tyrer-Cuzick, density and SNP18 combined (O/E=1.07; 0.91 for invasive cancer). The combined assessment was similarly predictive to the complete case-cohort study (Supplementary Table 3). Breast cancer rates in each group were within the predicted range from the combined assessment. The observed risk from Tyrer-Cuzick had an AUC of 0.58 (95%CI 0.52-0.62); 0.64 (95%CI 0.60–0.68) when mammographic density was also incorporated, and 0.67 (95%CI 0.62–0.71) when SNP18 was further incorporated.

Discussion

This study found that SNP18 stratifies breast cancer risk beyond classical factors and mammographic density. The results underline the additional value of incorporating polygenic risk scores with mammographic density and classical risk factors to evaluate risk in women participating in a non-selective national screening program. The addition of SNP18 gave a better risk stratification with more women in both the lower and higher risk groups.

Risk-adapted screening strategies that might be evaluated include to start screening later in lower-risk women, increase the screening interval or potentially to not invite them for screening at all. We found relatively few interval or higher stage cancers occurred in the lowest risk group in our UK cohort undertaking three-yearly screening. In countries with screening programmes on a 2-yearly cycle there might be scope to save health resources by reducing screening to 3-yearly in this low-risk group, constituting 33% of the study control population and possibly up to 37% of the whole PROCAS cohort. In contrast, UK NICE guidelines would permit women in the high-risk group to receive annual screening until 60 years and to be offered preventive therapy. This group (6% at ≥8% 10-year risk) could even be considered for risk reducing surgery and a further 11% at moderate-risk to consider chemoprevention. The rationale for this is that the data suggest that approximately 1 in 6 of the population at moderate/high-risk (≥5% 10-year risk) are likely to develop approximately 35% of the total interval cancers and 42% of the stage 2 and higher cancers in a 3-yearly screening programme. There was also no excess of cases of DCIS in the moderate/high-risk groups (11% vs 16% for lower risks) indeed all three cases of DCIS in the high-risk group were high grade. There is therefore great scope for down-staging of potentially lethal cancers with more frequent screening (perhaps employing strategies such as tomosynthesis mammography, automated breast ultrasound and MRI particularly in those where masking from density is an issue). Breast cancer risk reduction with tamoxifen, raloxifene or an aromatase inhibitor may prevent between 30-50% of these cancers15,16.

The participant characteristics indicate that women who are aware of their elevated risk were more likely to agree to further investigation related to their risk. Although there might be a bias towards including women that are at higher risk based on classical risk factors, this is very unlikely to have influenced the SNP18 genotypic signature in controls because it was close to that expected (Supplementary Table 2) and only slightly correlated with classical risk factors.

The weak correlation between SNP18 and the other risk factors, DR and Tyrer-Cuzick risk, justifies the use of all three measures in combination. Indeed, there was effectively no change in OR when adjusted for Tyrer-Cuzick. This is consistent with Vachon et al.12 who also did not find an association between PRS and mammographic density.

The 18 SNPs used in this study have previously been found to be most strongly associated with ER positive disease17-19. Although there was no significant heterogeneity in performance of SNP18 by ER status, this was to be expected due to lack of power, and the point estimates for the IQ-OR of SNP18 was substantially lower for ER-negative cancers (1.09 vs 1.61).

A limitation of the main case-cohort analysis is that saliva was not provided at recruitment to the wider cohort. However, this appeared to have minimal impact on the results, because they were similar in our secondary analysis of cancers diagnosed after sample donation.

Of note we used a different density measure (VAS instead of BI-RADS density) compared to some previous studies12,20; and a different model to combine classical risk factors (Tyrer-Cuzick model instead of the Breast Cancer Surveillance Consortium risk prediction model

HYPERLINK \l "_ENREF_20"

ADDIN EN.CITE.DATA20, which already incorporates mammographic density). Using a continuous scale measure like VAS may well be a better discriminator than leveraging women into a particular one of four BI-RADs categories. However, both are perhaps not ideal due to inter and intra-reader variation in determining mammographic density21.

Although there are now more than 100 SNPs linked to breast cancer risk, it remains to be determined in prospective cohort studies whether additional SNPs will substantially improve risk prediction beyond classical factors and mammographic density 22,23. SNP18 explains a large proportion of the current known familial component derived from SNPs, and the ORs used are likely more robust than many of the more recently identified SNPs that have very small ORs. Most SNPs identified to date have been more strongly associated with ER+ disease, as here. Further analysis within the breast cancer association consortium case-control studies may help to identify SNPs that are more predictive for other sub-types of breast cancer.

In summary, the increased sensitivity (proportion of cancers in a group identified over a given period) for a given positive predictive value (the risk in the group over that period) is an important aim for risk-stratified screening and prevention strategies. SNP18 substantially improved the accuracy of risk prediction when combined with Tyrer-Cuzick estimates and mammographic density. Routinely incorporating SNP18 into risk prediction models will provide women attending routine screening a more informative risk prediction that could be used in more personalized prevention and early detection strategies.

Acknowledgements

This work was supported by Prevent Breast Cancer (GA09-002 and GA11-002) and the National Institute for Health Research (NF-SI-0513-10076 to DGE). None of the funding bodies were involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Many thanks to Paula Stavrinos B.Sc., Jake Southworth, Lynne Fox, Jill Fox, Louise Donnelly Ph.D., Sarah Sahin, Donna Watterson B.Sc., Faiza Idries B.Sc., Helen Ruane and Sarah Ingham Ph.D. for administrative support and data management.

We thank Prof Antonis Antoniou for providing the allele frequencies and univariate odds ratios associated with each SNP from the iCOGS database.

DGE designed the study and obtained funding. Data collection was done by HB, SS, EH, DGE, AB, EvV, EH and SA. DGE, AB, EvV, EH and SA carried out the analysis of the data. EvV, AB and DGE drafted the manuscript. All authors commented and approved the final version.

DGE and AB had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis

Drs Cuzick and Brentnall report royalty payments through Cancer Research UK for commercial use of the Tyrer-Cuzick algorithm. All other authors report no conflicts of interest.

References

1.Eccles SA, Aboagye EO, Ali S, et al. Critical research gaps and translational priorities for the successful prevention and treatment of breast cancer. Breast cancer research : BCR. 2013;15(5):R92.

2.Michailidou K, Beesley J, Lindstrom S, et al. Genome-wide association analysis of more than 120,000 individuals identifies 15 new susceptibility loci for breast cancer. Nat Genet. Apr 2015;47(4):373-380.

3.Kapoor NS, Curcio LD, Blakemore CA, et al. Multigene Panel Testing Detects Equal Rates of Pathogenic BRCA1/2 Mutations and has a Higher Diagnostic Yield Compared to Limited BRCA1/2 Analysis Alone in Patients at Risk for Hereditary Breast Cancer. Annals of surgical oncology. Oct 2015;22(10):3282-3288.

4.Thompson ER, Rowley SM, Li N, et al. Panel Testing for Familial Breast Cancer: Calibrating the Tension Between Research and Clinical Care. J Clin Oncol. Jan 19 2016.

5.Tyrer J, Duffy SW, Cuzick J. A breast cancer prediction model incorporating familial and personal risk factors. Statistics in medicine. Apr 15 2004;23(7):1111-1130.

6.Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. Dec 20 1989;81(24):1879-1886.

7.Warwick J, Birke H, Stone J, et al. Mammographic breast density refines Tyrer-Cuzick estimates of breast cancer risk in high-risk women: findings from the placebo arm of the International Breast Cancer Intervention Study I. Breast cancer research : BCR. 2014;16(5):451.

8.Brentnall AR, Harkness EF, Astley SM, et al. Mammographic density adds accuracy to both the Tyrer-Cuzick and Gail breast cancer risk models in a prospective UK screening cohort. Breast cancer research : BCR. 2015;17(1):147.

9.Brentnall AR, Evans DG, Cuzick J. Distribution of breast cancer risk from SNPs and classical risk factors in women of routine screening age in the UK. Br J Cancer. Feb 4 2014;110(3):827-828.

10.Evans DG, Warwick J, Astley SM, et al. Assessing individual breast cancer risk within the U.K. National Health Service Breast Screening Program: a new paradigm for cancer prevention. Cancer prevention research. Jul 2012;5(7):943-951.

11.Dite GS, MacInnis RJ, Bickerstaffe A, 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 : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. Dec 16 2015.

12.Vachon CM, Pankratz VS, Scott CG, et al. The contributions of breast density and common genetic variation to breast cancer risk. J Natl Cancer Inst. May 2015;107(5).

13.Turnbull C, Ahmed S, Morrison J, et al. Genome-wide association study identifies five new breast cancer susceptibility loci. Nat Genet. Jun 2010;42(6):504-507.

14.Gabriel S, Ziaugra L, Tabbaa D. SNP genotyping using the Sequenom MassARRAY iPLEX platform. Current protocols in human genetics. Jan 2009;Chapter 2:Unit 2 12.

15.Cuzick J, Sestak I, Forbes JF, et al. Anastrozole for prevention of breast cancer in high-risk postmenopausal women (IBIS-II): an international, double-blind, randomised placebo-controlled trial. Lancet. Mar 22 2014;383(9922):1041-1048.

16.McIntosh A SC, Evans G, Turnbull N, Bahar N, Barclay M, Easton D, Emery J, Gray J, Halpin J, Hopwood P, McKay J, Sheppard C, Sibbering M, Watson W, Wailoo A, Hutchinson A. Clinical Guidelines and Evidence Review for The Classification and Care of Women at Risk of Familial Breast Cancer. 2004 2006; https://www.nice.org.uk/Guidance/CG164. Accessed 15/02/2017, 2017.

17.Michailidou K, Hall P, Gonzalez-Neira A, et al. Large-scale genotyping identifies 41 new loci associated with breast cancer risk. Nat Genet. Apr 2013;45(4):353-361, 361e351-352.

18.Garcia-Closas M, Couch FJ, Lindstrom S, et al. Genome-wide association studies identify four ER negative-specific breast cancer risk loci. Nat Genet. Apr 2013;45(4):392-398, 398e391-392.

19.Horne HN, Chung CC, Zhang H, et al. Fine-Mapping of the 1p11.2 Breast Cancer Susceptibility Locus. PLoS One. 2016;11(8):e0160316.

20.Shieh Y, Hu D, Ma L, et al. Breast cancer risk prediction using a clinical risk model and polygenic risk score. Breast Cancer Res Treat. Oct 2016;159(3):513-525.

21.Irshad A, Leddy R, Ackerman S, et al. Effects of Changes in BI-RADS Density Assessment Guidelines (Fourth Versus Fifth Edition) on Breast Density Assessment: Intra- and Interreader Agreements and Density Distribution. AJR. American journal of roentgenology. Dec 2016;207(6):1366-1371.

22.Evans DM, Visscher PM, Wray NR. Harnessing the information contained within genome-wide association studies to improve individual prediction of complex disease risk. Hum Mol Genet. 2009;18(18):7.

23.Ripatti S, Tikkanen E, Orho-Melander M, et al. A multilocus genetic risk score for coronary heart disease: case-control and prospective cohort analyses. Lancet. 2010;376(9750):8.

Figure legends.

Figure 1. Unadjusted observed vs expected odds ratios from SNP18 by decile.

Legend: The points show the value by decile with 95% confidence intervals extending horizontally (grey ---); The line of best fit from a logistic regression (---) is shown in comparison with the theoretical line for perfect calibration (- - -). The data and cut points for this plot are in Supplemental Table 6.

Table 1. Distribution of breast cancers over 10-year risk groups, when SNP18, density and the Tyrer-Cuzick model are combined.

 

Low

Average

Intermediate

Moderate

High

Total

 

(<2%)

(2-3.49%)

(3.5-4.99%)

(5-7.99%)

(>8%)

 

Case-cohort analysis

No breast cancer

2919 (33%)

2906 (33%)

1479 (17%)

1055 (12%)

538 (6%)

8897

Breast Cancer cases (all)

84 (18%)

133 (29%)

108 (23%)

74 (16%)

67 (14%)

466

Odds Ratio (95%CI)

1 (ref)

1.6 (1.2 – 2.1)

2.5 (1.9-3.4)

2.4 (1-8 – 3.4)

4.3 (3.1 - 6.0)

Breast Cancer diagnosis by type

Prevalent [DCIS]

50 [12] (19%)

79 [19] (29%)

67 [11] (25%)

44 [11] (16%)

31 [8] (11%)

271 (61)

Incident

20 (18%)

36 (32%)

19 (17.0%)

15 (13%)

22 (20%)

112

Interval

14 (17%)

18 (22%)

22 (27%)

15 (18%)

14 (17%)

83

Odds Ratio

1 (ref)

1.3 (0.6 – 2.6)

3.1 (1.6 – 6.1)

3.0 (1.4 - 6.2)

5.4 (2.6 – 11.4)

Post prevalent stage 2+

10 (14%)

15 (22%)

15 (22%)

14 (20%)

15 (22%)

69

Odds Ratio

1 (ref)

1.5 (0.7 – 3.4)

3.0 (1.3 – 6.6)

3.9 (1.7 – 8.7)

8.1 (3.6 – 18.2)

Post prevalent DCIS [high grade]

6 [3] (18%)+

8 [3] (15%)+

7 [5] (17%)+

4 [2] (13%)+

3 [3] (8%)+

28 (14.8%)

Breast cancer diagnosed after DNA sample*

Number breast-cancer free

2946

2951

1511

1084

572

9064

Total follow up (years)

14559

14453

7363

5265

2779

44419

Breast cancers (DCIS)

27 (5)

45 (8)

32 (6)

29 (4)

34 (2)

167 (26)

Rate (95%CI) per 1000 women/ year

DCIS + invasive

1.9 (1.2 - 2.7)

3.1 (2.3 – 4.2)

4.3 (3.0 – 6.1)

5.5 (3.7 – 7.9)

12.2 (8.5 – 17.1)

3.8 (3.2 – 4.4)

Rate 95%CI per 1000 women/ year

invasive only

1.5 (0.9 – 2.3)

2.6 (1.8 – 3.5)

3.5 (2.3 – 5.2)

4.7 (3.1 – 7.0)

11.5 (7.9 – 16.3)

3.2 (2.7 – 3.8)

*167 prospective breast cancers occurred in 9064 women unaffected with breast cancer at time of DNA sampling. In a mean follow up of 4.90 years 155 breast cancers were expected from Tyrer-Cuzick, density and SNP18 combined Observed Expected ratio = 1.07 or 0.91 for invasive only. +proportion of post prevalent cancers that were DCIS. Note: percentages may not sum to 100% due to errors introduced by rounding.

2