Abstract
Background
Clinical guidelines often use predicted lifetime risk from birth to define criteria for making decisions regarding breast cancer screening rather than thresholds based on absolute 5-year risk from current age.
Methods
We used the Prospective Family Cohort Study of 14 657 women without breast cancer at baseline in which, during a median follow-up of 10 years, 482 women were diagnosed with invasive breast cancer. We examined the performances of the International Breast Cancer Intervention Study (IBIS) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) risk models when using the alternative thresholds by comparing predictions based on 5-year risk with those based on lifetime risk from birth and remaining lifetime risk. All statistical tests were 2-sided.
Results
Using IBIS, the areas under the receiver-operating characteristic curves were 0.66 (95% confidence interval = 0.63 to 0.68) and 0.56 (95% confidence interval = 0.54 to 0.59) for 5-year and lifetime risks, respectively (Pdiff < .001). For equivalent sensitivities, the 5-year incidence almost always had higher specificities than lifetime risk from birth. For women aged 20-39 years, 5-year risk performed better than lifetime risk from birth. For women aged 40 years or older, receiver-operating characteristic curves were similar for 5-year and lifetime IBIS risk from birth. Classifications based on remaining lifetime risk were inferior to 5-year risk estimates. Results were similar using BOADICEA.
Conclusions
Our analysis shows that risk stratification using clinical models will likely be more accurate when based on predicted 5-year risk compared with risks based on predicted lifetime and remaining lifetime, particularly for women aged 20-39 years.
Breast cancer clinical guidelines often use model-based lifetime risk estimates (from birth, not from current age) to define criteria for making decisions regarding breast cancer screening, chemoprevention, and/or risk-reducing surgery (1-6).
Five-year risk estimates can be prospectively validated, and their risk discrimination can also be validated over a decade or more using cohort studies (7-9). Lifetime risk estimates from birth, however, cannot be validated in practice, and their risk discrimination for short-term outcomes remains unknown (10). Other risk estimates, such as remaining lifetime risk and risk stratification based on the number of first-degree relatives (FDRs) diagnosed with breast cancer, have similar practical limitations. Here, we examined whether 5-year risk estimates could be more appropriate for making clinical recommendations.
Methods
Study Design and Participants
The Prospective Family Study Cohort (ProF-SC) comprises baseline and follow-up data from the Breast Cancer Family Registry Cohort, a collaboration by 6 centers in the United States, Canada, and Australia and the Australian Kathleen Cuningham Foundation Consortium for Research into Familial Breast Cancer (11). All participants provided written informed consent before enrollment, and the study protocols were approved by institutional review boards (11-14). ProF-SC includes 18 856 women who did not have breast cancer at enrollment (1992-2011). All studies in ProF-SC collected breast cancer risk factor data using identical baseline questionnaires, which captured demographic characteristics, height, weight, history of benign breast disease, breast surgeries, reproductive history, lifestyle factors, and family history of breast and other cancers (including age at diagnosis and vital status, with date or age of death where applicable, across multiple generations).
Information on breast cancer diagnoses was collected by personal or family reports or by cancer registry reports. In mid-2014, a systematic follow-up of ProF-SC was completed (11), and we verified reported invasive breast cancer diagnoses for 81% of women through pathology reports, cancer registries, medical records, death certificates, or pathologist review of tissue samples (11-13). Screening for germline pathogenic variants in BRCA1 and BRCA2 typically involved screening the youngest affected family member at baseline; if that person carried a mutation, other family members were also tested. Further details are described in the Supplementary Methods (available online) (15,16).
For this analysis, from the 18 856 women who did not have a personal history of breast cancer at baseline, we excluded women who were known to carry BRCA1 or BRCA2 pathogenic variants (n = 1075) because there are separate clinical guidelines for them. We also excluded women aged younger than 20 years or older than 70 years at baseline (n = 2082), women who had a bilateral prophylactic mastectomy (n = 113) or ovarian cancer (n = 316), and those for whom information about family history was not available from family pedigrees (n = 96), leaving 14 657 women. We limited eligibility to age 70 years or younger at baseline because Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) does not calculate risk beyond age 80 years.
We calculated 5-year risk, lifetime risk from birth, and remaining lifetime risk of invasive breast cancer using BOADICEA v3 (https://pluto.srl.cam.ac.uk/cgi-bin/bd3/v3/bd.cgi) (to 80 years) (17) and International Breast Cancer Intervention Study (IBIS) v8b (http://www.ems-trials.org/riskevaluator/) (to 85 years) (18) using baseline pedigree and other risk factor information. Data were generally available for all items required by these models, except for mammographic density and polygenic risk score used in IBIS. Country- and age-specific breast cancer incidences for Australia, Canada, and US white (Hispanics and non-Hispanics combined) were used for BOADICEA, whereas United Kingdom incidences were used for IBIS. We also classified women in 2 ways based on the number of affected FDRs diagnosed with breast cancer (≥1 and ≥2).
Statistical Analysis
Time at risk of breast cancer started at 2 months after the baseline questionnaire and continued to the first of 10 years and 2 months later, date last known to be undiagnosed with breast cancer, date of diagnosis of invasive or in situ breast cancer, date of bilateral mastectomy, or date of 80th birthday. We censored women at the time of a diagnosis of in situ breast cancer because treatment for in situ cancers can affect the risk of future invasive diagnoses. Deaths from nonbreast cancer causes were considered as competing risks (not censored), but this assumption did not materially affect risk estimates (9).
We evaluated risk discrimination overall and by age group (20-39, 40-49, 50-70 years) between women who did and those who did not develop breast cancer within 10 years using C statistics and plotting the area under the receiver-operating characteristic curves (AUC), accounting for relatedness between family members and incomplete follow-up (19). The C statistic can range from 0.50 (no discriminative ability) to 1.00 (perfect discrimination). Specificities were calculated for 5-year risk thresholds that achieved the same sensitivity as lifetime risk from the AUC output. C statistics and their 95% confidence intervals (CIs) were calculated accounting for clustering within related family members using the somersd package in Stata (version 14.2) (20). The lincom command in Stata was used to derive P for differences in C statistics between risk estimates based on the χ2(df = 1) test (Pdiff). All statistical tests were 2-sided, and a P value of less than 0.05 was considered statistically significant. The data for the AUC curves were derived using the Risk Model Assessment Package (versions 0.03-01) in R.
Results
During follow-up (median = 10 years), 482 women were diagnosed with a first invasive breast cancer. Participants’ baseline characteristics are summarized in Table 1. Overall, the AUC was higher for 5-year IBIS risk (AUC = 0.66, 95% CI = 0.63 to 0.68) than for lifetime risk (AUC = 0.56, 95% CI = 0.54 to 0.59; Pdiff < .001; Figure 1). For example, specificity was 0.16 higher for a 1.6% 5-year IBIS risk compared with a 20% lifetime IBIS risk (sensitivity = 0.68, 95% CI = 0.64 to 0.73, for both measures).
Table 1.
Characteristics of women who were unaffected at baseline and were not known to carry BRCA1 or BRCA2 pathogenic variants, the Prospective Family Study Cohort (ProF-SC)
| Risk factors | Unaffected after 10 years | Follow-up <10 years | Died within 10 years | Invasive breast cancer diagnosis within 10 years |
|---|---|---|---|---|
| No. (%) (n = 8307) | No. (%) (n = 5483) | No. (%) (n = 385) | No. (%) (n = 482)a | |
| Age at baseline questionnaire, y | ||||
| 20-29 | 1121 (13.5) | 884 (16.1) | 5 (1.3) | 16 (3.3) |
| 30-39 | 1911 (23.0) | 1202 (21.9) | 22 (5.7) | 70 (14.5) |
| 40-49 | 2157 (26.0) | 1369 (25.0) | 48 (12.5) | 132 (27.4) |
| 50-59 | 1832 (22.1) | 1216 (22.2) | 105 (27.3) | 142 (29.5) |
| 60-70 | 1286 (15.5) | 812 (14.8) | 205 (53.2) | 122 (25.3) |
| Age at menarche, y | ||||
| ≤11 | 1424 (17.1) | 1012 (18.5) | 80 (20.8) | 76 (15.8) |
| 12-13 | 4369 (52.6) | 2761 (50.4) | 170 (44.2) | 260 (53.9) |
| ≥14 | 2435 (29.3) | 1626 (29.7) | 131 (34) | 139 (28.8) |
| Unknown | 79 (1.0) | 84 (1.5) | 4 (1.0) | 7 (1.5) |
| Body mass index, kg/m2 | ||||
| <25 | 4607 (55.5) | 2581 (47.1) | 159 (41.3) | 230 (47.7) |
| 25 - <30 | 2171 (26.1) | 1476 (26.9) | 115 (29.9) | 147 (30.5) |
| ≥30 | 1360 (16.4) | 1310 (23.9) | 99 (25.7) | 101 (21.0) |
| Unknown | 169 (2.0) | 116 (2.1) | 12 (3.1) | 4 (0.8) |
| Age at first live birth, y | ||||
| <20 | 954 (11.5) | 850 (15.5) | 94 (24.4) | 56 (11.6) |
| 20-24 | 2445 (29.4) | 1501 (27.4) | 150 (39.0) | 165 (34.2) |
| 25-29 | 1886 (22.7) | 1069 (19.5) | 58 (15.1) | 96 (19.9) |
| ≥30 | 901 (10.8) | 586 (10.7) | 28 (7.3) | 71 (14.7) |
| Nulliparous | 2121 (25.5) | 1477 (26.9) | 55 (14.3) | 94 (19.5) |
| Menopausal hormone therapy use | ||||
| Ever | 1869 (22.5) | 1062 (19.4) | 153 (39.7) | 145 (30.1) |
| Never | 6196 (74.6) | 4324 (78.9) | 221 (57.4) | 323 (67.0) |
| Unknown | 242 (2.9) | 97 (1.8) | 11 (2.9) | 14 (2.9) |
| Menopausal status | ||||
| Premenopausal | 4732 (57.0) | 3293 (60.1) | 58 (15.1) | 202 (41.9) |
| Postmenopausal | 2776 (33.4) | 1866 (34.0) | 305 (79.2) | 236 (49.0) |
| Unknown | 799 (9.6) | 324 (5.9) | 22 (5.7) | 44 (9.1) |
| Personal history of benign breast disease | ||||
| Yes | 2279 (27.4) | 1490 (27.2) | 99 (25.7) | 172 (35.7) |
| No | 5746 (69.2) | 3882 (70.8) | 278 (72.2) | 296 (61.4) |
| Unknown | 282 (3.4) | 111 (2.0) | 8 (2.1) | 14 (2.9) |
| No. of first-degree relatives with breast cancer | ||||
| 0 | 1382 (16.6) | 964 (17.6) | 78 (20.3) | 56 (11.6) |
| 1 | 5464 (65.8) | 3581 (65.3) | 220 (57.1) | 273 (56.6) |
| ≥2 | 1461 (17.6) | 938 (17.1) | 87 (22.6) | 153 (31.7) |
81% of the breast cancer cases were histologically confirmed.
Figure 1.
Performance in terms of the sensitivity and specificity for International Breast Cancer Intervention Study (IBIS) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA). Receiver-operating characteristic curves of 5-year risk and lifetime risk from birth models. Black line denotes the 5-year breast cancer risk, grey line the lifetime breast cancer risk from birth calculated using the IBIS model or the BOADICEA model. Black percentages show the difference in specificity when using 5-year risk estimates compared with lifetime risk from birth. The dashed line represents the line of no discrimination. The χ2(df = 1) test was used to calculate 2-sided P values. 5R = 5-year risk; AUC = area under the receiver–operating characteristic curve; CI = confidence interval; LR = lifetime risk from birth.
For women aged 20-39 years, the AUC was equivalent to more than double the risk gradient using 5-year IBIS risk (AUC = 0.70, 95% CI = 0.64 to 0.75) compared with lifetime IBIS risk (AUC = 0.59, 95% CI = 0.52 to 0.65; Pdiff < .001) (see Figure 2 and Table 2). For women aged 40 years or more, AUCs were similar for 5-year and lifetime IBIS risks. The correlation between 5-year and lifetime IBIS risk increased with age (Figure 2). Results were similar when using the BOADICEA model (Table 2). Analyses restricted to women with no more than 1 affected relative (to reduce the likelihood of including undetected BRCA1 or BRCA2 mutation carriers) gave similar results (Figure 3). For neither model were the AUCs for remaining lifetime risk better than chance (Figure 4). For both models, classifications based on 2 or more affected FDRs had similar efficiency to using equivalent thresholds based on 5-year risk, but classifications based on 1 or more affected FDRs performed more poorly than 5-year risk.
Figure 2.
Performance in terms of the sensitivity and specificity for International Breast Cancer Intervention Study (IBIS) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) models. Receiver operating characteristic curves of 5-year risk and lifetime risk from birth, by age group. Black line denotes the 5-year breast cancer risk, grey line the lifetime breast cancer risk from birth calculated using the IBIS model or the BOADICEA model. The dashed line represents the line of no discrimination. The χ2(df = 1) test was used to calculate 2-sided P values. 5R = 5-year risk; AUC = area under the receiver–operating characteristic curve; CI = confidence interval; LR = lifetime risk from birth.
Table 2.
Predicted absolute risk based on IBIS and BOADICEA for women not known to carry BRCA1 or BRCA2 pathogenic variants, Prospective Family Study Cohort
| Model and age group | Risk based on lifetime risk from birth |
5-year riska |
Differenceb |
|||||
|---|---|---|---|---|---|---|---|---|
| Risk estimate, % | Sensitivity (95% CI) | Specificity (95% CI) | Threshold, % | Specificity (95% CI) | Specificity | P diff c | ||
| IBIS | ||||||||
| 20-70 y | 20 | 0.68 (0.64 to 0.73) | 0.37 (0.37 to 0.38) | 1.60 | 0.54 (0.53 to 0.55) | +0.16 | <.001 | |
| 25 | 0.49 (0.45 to 0.54) | 0.58 (0.57 to 0.59) | 2.30 | 0.71 (0.70 to 0.72) | +0.13 | <.001 | ||
| 30 | 0.32 (0.28 to 0.37) | 0.79 (0.78 to 0.79) | 3.10 | 0.85 (0.85 to 0.86) | +0.07 | <.001 | ||
| 20-39 y | 20 | 0.82 (0.73 to 0.90) | 0.18 (0.17 to 0.19) | 0.40 | 0.47 (0.46 to 0.48) | +0.29 | <.001 | |
| 25 | 0.66 (0.55 to 0.76) | 0.41 (0.40 to 0.42) | 0.60 | 0.60 (0.58 to 0.61) | +0.19 | <.001 | ||
| 30 | 0.47 (0.36 to 0.58) | 0.70 (0.69 to 0.71) | 0.80 | 0.77 (0.76 to 0.78) | +0.07 | <.001 | ||
| 40-49 y | 20 | 0.70 (0.61 to 0.77) | 0.36 (0.35 to 0.38) | 1.50 | 0.42 (0.40 to 0.44) | +0.06 | <.001 | |
| 25 | 0.51 (0.43 to 0.60) | 0.60 (0.59 to 0.62) | 2.00 | 0.63 (0.61 to 0.65) | +0.03 | .02 | ||
| 30 | 0.37 (0.28 to 0.45) | 0.78 (0.77 to 0.79) | 2.20 | 0.74 (0.72 to 0.75) | −0.04 | <.001 | ||
| 50-70 y | 20 | 0.63 (0.57 to 0.69) | 0.59 (0.58 to 0.60) | 2.50 | 0.59 (0.58 to 0.60) | 0.00 | .80 | |
| 25 | 0.43 (0.37 to 0.49) | 0.73 (0.71 to 0.74) | 3.30 | 0.73 (0.72 to 0.74) | 0.00 | .90 | ||
| 30 | 0.26 (0.21 to 0.31) | 0.86 (0.85 to 0.86) | 4.00 | 0.85 (0.84 to 0.86) | −0.01 | .42 | ||
| BOADICEA | ||||||||
| 20-70 y | 20 | 0.45 (0.40 to 0.49) | 0.67 (0.67 to 0.68) | 2.40 | 0.72 (0.72 to 0.73) | +0.05 | <.001 | |
| 25 | 0.23 (0.19 to 0.27) | 0.85 (0.85 to 0.86) | 3.30 | 0.87 (0.87 to 0.88) | +0.02 | <.001 | ||
| 30 | 0.10 (0.08 to 0.13) | 0.93 (0.93 to 0.94) | 4.30 | 0.96 (0.95 to 0.96) | +0.02 | <.001 | ||
| 20-39 y | 20 | 0.59 (0.48 to 0.70) | 0.68 (0.67 to 0.69) | 0.80 | 0.70 (0.69 to 0.71) | +0.02 | .009 | |
| 25 | 0.33 (0.23 to 0.44) | 0.85 (0.84 to 0.86) | 1.30 | 0.89 (0.88 to 0.89) | +0.03 | <.001 | ||
| 30 | 0.21 (0.13 to 0.31) | 0.93 (0.92 to 0.93) | 1.70 | 0.92 (0.91 to 0.93) | −0.01 | .10 | ||
| 40-49 y | 20 | 0.50 (0.41 to 0.59) | 0.62 (0.61 to 0.64) | 1.90 | 0.60 (0.58 to 0.62) | −0.02 | .03 | |
| 25 | 0.26 (0.19 to 0.34) | 0.82 (0.81 to 0.84) | 2.80 | 0.85 (0.84 to 0.86) | +0.02 | .005 | ||
| 30 | 0.12 (0.07 to 0.18) | 0.93 (0.92 to 0.93) | 3.40 | 0.93 (0.92 to 0.94) | 0.00 | .62 | ||
| 50-70 y | 20 | 0.37 (0.31 to 0.43) | 0.77 (0.76 to 0.78) | 3.10 | 0.74 (0.73 to 0.75) | −0.03 | .002 | |
| 25 | 0.18 (0.13 to 0.23) | 0.91 (0.90 to 0.92) | 4.00 | 0.90 (0.89 to 0.91) | −0.01 | .06 | ||
| 30 | 0.07 (0.04 to 0.10) | 0.96 (0.95 to 0.96) | 4.70 | 0.96 (0.95 to 0.97) | 0.00 | .27 | ||
For each row, 5-year risk specificity was calculated for the same sensitivity as lifetime risk from birth. BOADICEA = Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm; CI = confidence interval; IBIS = International Breast Cancer Intervention Study.
Difference in specificity between 5-year risk and lifetime risk from birth.
P for the test of differences based on χ2 (df = 1) test.
Figure 3.
Performance in terms of the sensitivity and specificity for International Breast Cancer Intervention Study (IBIS) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) models. Receiver operating characteristic curves of 5-year risk and lifetime risk from birth for women with ≤1 affected first-degree relative. Black line denotes the 5-year breast cancer risk, grey line the lifetime breast cancer risk from birth calculated using the IBIS model or the BOADICEA model. The dashed line represents the line of no discrimination. The χ2(df = 1) test was used to calculate 2-sided P values. 5R = 5-year risk; AUC = area under the receiver–operating characteristic curve; CI = confidence interval; LR = lifetime risk from birth.
Figure 4.
Performance in terms of the sensitivity and specificity for International Breast Cancer Intervention Study (IBIS) and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (BOADICEA) models and classifications based on number of affected first-degree relatives (FDRs). Receiver operating characteristic curves of 5-year risk, lifetime risk from birth, and remaining lifetime risk. Black line denotes the 5-year breast cancer risk, grey line the lifetime breast cancer risk from birth, light grey line the remaining lifetime risk calculated using the IBIS model or the BOADICEA model. The dashed line represents the line of no discrimination. ≥1 affected FDRs = 1 or more first-degree relatives diagnosed with breast cancer, ≥2 affected FDRs = 2 or more first-degree relatives diagnosed with breast cancer. 5R = 5-year risk; AUC = area under the receiver–operating characteristic curve; CI = confidence interval; LR = lifetime risk from birth; RLR = remaining lifetime risk.
Discussion
This study provides empirical evidence that for predicting breast cancer risk with a given sensitivity, a 5-year risk estimate has a higher specificity at most thresholds (and therefore gives fewer false-positives) than does lifetime risk from birth for younger women (age 20-39 years) but is comparable to lifetime risk from birth for older women. Remaining lifetime risk and classifications based on the number of affected FDRs were also generally inferior to 5-year risk estimates.
Most cohorts have had sufficient follow-up time to independently validate 5-year risk, but not longer time thresholds, particularly lifetime (7,9,21). Even though the ProF-SC cohort has longer follow-up than most other cohorts (median 10 years), it is still insufficient for validating lifetime risk estimates. Therefore, classifying women using clinical guidelines and counseling based on lifetime risk thresholds is suboptimal (10).
There are several challenges in implementing a shift away from using a lifetime risk from birth, including the need to update women regularly about changing risk as well as the selection of the risk threshold as it changes with age. For example, having a 1.6% 5-year risk would likely result in different recommendations for a 45-year-old woman than for a 65-year-old woman. Five-year risk is very important for young women who would not typically be screened and might want to evaluate decisions regarding risk-reducing surgeries and chemoprevention at shorter time intervals in connection with other life decisions regarding pregnancy and breastfeeding. Additionally, screening and risk-reducing options are changing, and some women want to wait for additional choices and need accurate short-term risk estimates to assist their decisions, including risk-reducing medications and/or intensified screening rather than an irreversible outcome like prophylactic mastectomy.
Shifting to a 5-year risk should improve classification and reflect the dynamic nature of many components of current breast cancer risk models, and their outcomes change as women’s age and risk factors change and as competing health risks and priorities change over the life course. Therefore, updating absolute risk predictions regularly (eg, every 5 years) will also be beneficial for more precise risk estimates and better informed decision making. Younger women would benefit most from being able to better guide decisions regarding magnetic resonance imaging and initiating mammography screening. Increasingly, younger women with a family history of breast cancer and those in genomic health screening programs are undergoing genetic assessment to estimate their cancer risk. Depending on a woman’s underlying risk, it might be preferable to update risk more often, because a previous study showed that the risk percentage recommended for initiation of high-risk screening increases 1.5-fold between ages 30 and 50 years (22). Our findings need to be replicated using other cohorts; it is noteworthy that the Women Informed to Screen Depending On Measures of risk (WISDOM) study is evaluating screening guidelines based on 5-year breast cancer risk, but results are not yet available (23).
As risk models improve, the specificities should increase, but the general advantages of using estimates of 5-year risk compared with the current practice of using estimates of lifetime risk from birth, particularly for women aged 20-39 years, will likely remain.
Funding
This work was primarily supported by a grant from the US National Institute of Health (RO1CA159868). Additionally, the cohorts were supported through additional resources. The Australian Breast Cancer Family Registry was supported by the Australian National Health and Medical Research Council, the New South Wales Cancer Council, the Victorian Health Promotion Foundation, the Victorian Breast Cancer Research Consortium, Cancer Australia, and the National Breast Cancer Foundation. The 6 sites of the Breast Cancer Family Registry were supported by grants from the US National Cancer Institute (UM1 CA164920 and U01CA164920).
This work was also supported by grants to the Kathleen Cuningham Foundation Consortium for Research into Familial Breast Cancer (kConFab) and the kConFab Follow-Up Study from Cancer Australia (809195 and 1100868), the Australian National Breast Cancer Foundation (IF 17 kConFab), the National Health and Medical Research Council (454508, 288704, and 145684), the Queensland Cancer Fund, the Cancer Councils of New South Wales, Victoria, Tasmania, and South Australia, and the Cancer Foundation of Western Australia. K-AP is a Practitioner Fellow of the Australian National Breast Cancer Foundation (PRAC-17-004). JLH and MCS are Senior Principal and Senior Research Fellows of the National Health and Medical Research Council, Australia, respectively. MBT also thanks the Breast Cancer Research Foundation for support.
Notes
Role of the funders: The funding sources had no role 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.
Disclosures: GSD reports grants from Genetic Technologies Pty Ltd outside the submitted work. K-AP has a patent System and Process of Cancer Risk Estimation (Australian Innovation Patent) issued. The other authors declared no conflicts of interest during the conduct of this story outside the grant funding listed in the Funding section.
Disclaimer: The content of this manuscript does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the Breast Cancer Family Registry (BCFR), nor does mention of trade names, commercial products, or organizations imply endorsement by the US government or the Breast Cancer Family Registry.
Acknowledgements: We thank all the participants in this study, the entire team of BCFR past and current investigators as well as the kConFab investigators and all of the BCFR and kConFab coordinators, research nurses, interviewers, and data management staff, and the heads and staff of the participating Family Cancer Clinics.
Author contributions: RJM and MBT conceived the project, performed statistical analyses and co-wrote the original draft of the Article; RB and YL coordinated the data collection; ILA, EMJ, MBD, SSB, KAP, JLH and MBT obtained funding; JAK, WKC, RLM, ASW, NZ, GSD, MCS, DG, GGG, AWK, ILA, EMJ, MBD, SSB, KAP, JLH and MBT provided resources for the study; all authors contributed to the writing of the Article. RJM and MBT had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Data Availability
For information on how to collaborate with the ProF-SC cohort in making further use of the data and resources, and with the BCFR, please see http://www.bcfamilyregistry.org/. For access to kConFab resources, see www.kconfab.org.
Supplementary Material
References
- 1. Murphy CD, Lee JM, Drohan B, et al. The American Cancer Society guidelines for breast screening with magnetic resonance imaging: an argument for genetic testing. Cancer. 2008;113(11):3116–3120. [DOI] [PubMed] [Google Scholar]
- 2.National Institute for Health and Care Excellence. Familial breast cancer: classification and care of people at risk of familial breast cancer and management of breast cancer and related risks in people with a family history of breast cancer (CG164). 2013. http://www.nice.org.uk/CG164. [PubMed]
- 3. Oeffinger KC, Fontham ET, Etzioni R, et al. Breast cancer screening for women at average risk: 2015 Guideline update from the American Cancer Society. JAMA. 2015;314(15):1599–1614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.The Royal Australian College of General Practitioners. Guidelines for Preventive Activities in General Practice. 9th ed., updated ed. East Melbourne, Vic: RACGP; 2018. [Google Scholar]
- 5. Bevers TB, Helvie M, Bonaccio E, et al. Breast Cancer Screening and Diagnosis, Version 3.2018, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2018;16(11):1362–1389. [DOI] [PubMed] [Google Scholar]
- 6. Saslow D, Boetes C, Burke W, et al. ; for the American Cancer Society Breast Cancer Advisory Group. American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA Cancer J Clin. 2007;57(2):75–89. [DOI] [PubMed] [Google Scholar]
- 7. Brentnall AR, Cuzick J, Buist DSM, et al. Long-term accuracy of breast cancer risk assessment combining classic risk factors and breast density. JAMA Oncol. 2018;4(9):e180174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Louro J, Posso M, Hilton Boon M, et al. A systematic review and quality assessment of individualised breast cancer risk prediction models. Br J Cancer. 2019;121(1):76–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Terry MB, Liao Y, Whittemore AS, et al. 10-year performance of four models of breast cancer risk: a validation study. Lancet Oncol. 2019;20(4):504–517. [DOI] [PubMed] [Google Scholar]
- 10. Quante AS, Whittemore AS, Shriver T, et al. Practical problems with clinical guidelines for breast cancer prevention based on remaining lifetime risk. J Natl Cancer Inst. 2015;107(7):djv124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Terry MB, Phillips KA, Daly MB, et al. Cohort profile: the Breast Cancer Prospective Family Study Cohort (ProF-SC). Int J Epidemiol. 2016;45(3):683–692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Mann GJ, Thorne H, Balleine RL, et al. ; the Kathleen Cuningham Consortium for Research in Familial Breast Cancer. Analysis of cancer risk and BRCA1 and BRCA2 mutation prevalence in the kConFab familial breast cancer resource. Breast Cancer Res. 2006;8(1):R12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. John EM, Hopper JL, Beck JC, et al. ; the Breast Cancer Family Registry. The Breast Cancer Family Registry: an infrastructure for cooperative multinational, interdisciplinary and translational studies of the genetic epidemiology of breast cancer. Breast Cancer Res. 2004;6(4):R375–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Phillips KA, Butow PN, Stewart AE, et al. Predictors of participation in clinical and psychosocial follow-up of the kConFab breast cancer family cohort. Fam Cancer. 2005;4(2):105–113. [DOI] [PubMed] [Google Scholar]
- 15. MacInnis RJ, Liao Y, Knight JA, et al. Considerations when using breast cancer risk models for women with negative BRCA1/BRCA2 mutation results. J Natl Cancer Inst. 2019; doi: 10.1093/jnci/djz194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Neuhausen SL, Ozcelik H, Southey MC, et al. ; Breast Cancer Family Registry. BRCA1 and BRCA2 mutation carriers in the Breast Cancer Family Registry: an open resource for collaborative research. Breast Cancer Res Treat. 2009;116(2):379–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Lee AJ, Cunningham AP, Kuchenbaecker KB, et al. ; Consortium of Investigators of Modifiers of BRCA1/2. BOADICEA breast cancer risk prediction model: updates to cancer incidences, tumour pathology and web interface. Br J Cancer. 2014;110(2):535–545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Tyrer J, Duffy SW, Cuzick J. A breast cancer prediction model incorporating familial and personal risk factors . Stat Med. 2004;23(7):1111–1130. [DOI] [PubMed] [Google Scholar]
- 19. Gong G, Quante AS, Terry MB, et al. Assessing the goodness of fit of personal risk models. Stat Med. 2014;33(18):3179–3190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Newson RB. Comparing the predictive powers of survival models using Harrell’s C or Somers’ D. Stata J. 2010;10(3):1–19. [Google Scholar]
- 21. Phillips KA, Liao Y, Milne RL, et al. ; kConFab Investigators. Accuracy of risk estimates from the iPrevent breast cancer risk assessment and management tool. JNCI Cancer Spectr. 2019;3(4):pkz066. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Ziogas A, Horick NK, Kinney AY, et al. Clinically relevant changes in family history of cancer over time. JAMA. 2011;306(2):172–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Esserman LJ,; the WISDOM Study and Athena Investigators. The WISDOM Study: breaking the deadlock in the breast cancer screening debate. NPJ Breast Cancer. 2017;3(1):34. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
For information on how to collaborate with the ProF-SC cohort in making further use of the data and resources, and with the BCFR, please see http://www.bcfamilyregistry.org/. For access to kConFab resources, see www.kconfab.org.