Breast Cancer Risk and Progressive Histology in Serial Benign Biopsies
Breast Cancer Risk and Progressive Histology in Serial Benign Biopsies
Daniel W. Visscher 0
Ryan D. Frank 0
Jodi M. Carter 0
Robert A. Vierkant 0
Stacey J. Winham 0
Ethan P. Heinzen 0
Brendan T. Broderick 0
Lori A. Denison 0
Teresa M. Allers 0
Joanne L. Johnson 0
Marlene H. Frost 0
Lynn C. Hartmann 0
Amy C. Degnim 0
Derek C. Radisky 0
0 Affiliations of authors: Department of Laboratory Medicine and Pathology (DWV and JMC), Department of Health Sciences Research (RAV, SJW, EPH, BTB), Department of Oncology (TMA, JLJ, MHF, LCH), Department of Information Technology (LAD), and Department of Surgery (ACD), Mayo Clinic Rochester, MN; Departments of Health Science Research (RDF) and Cancer Biology (DCR), Mayo Clinic Jacksonville , FL , USA
Background: More than 1 million women per year in the United States with benign breast biopsies are known to be at elevated risk for breast cancer (BC), with risk stratified on histologic categories of epithelial proliferation. Here we assessed women who had serial benign biopsies over time and how changes in the histologic classification affected BC risk. Methods: In the Mayo Clinic Benign Breast Disease Cohort of 13 466 women, 1414 women had multiple metachronous benign biopsies (10.5%). Both initial and subsequent biopsies were assessed histologically. BC risk for clinical and prognostic factors was assessed using subdistribution models to account for competing risks, and logistic regression/Wilcoxon/chi-square tests to assess covariates. All statistical tests were two-sided. Results: Breast cancer risk for women with serial biopsies, stratified by histologic category in the later biopsies, was similar to women with a single biopsy. We found that changes in histological category between initial and subsequent biopsy statistically significantly impacted BC risk. Women with nonproliferative initial findings and subsequent proliferative findings had an increased risk (hazard ratio [HR] ¼ 1.77, 95% confidence interval [CI] ¼ 1.06 to 2.94, P ¼ .03) compared with no change. Among women with proliferative disease without atypia at initial biopsy, risk decreased if later biopsy regressed to nonproliferative (HR ¼ 0.49, 95% CI ¼ 0.25 to 0.98) and increased if later biopsy showed progression to atypical hyperplasia (HR ¼ 1.49, 95% CI ¼ 0.73 to 3.05) compared with no change (P ¼ .04). Conclusions: We found that breast cancer risk increases in women with progressive epithelial proliferation over time and decreases in women whose biopsies show less proliferation. This finding has important implications for effective clinical management of the 100 000 women per year who have multiple benign breast biopsies.
One to 2 million women per year in the United States have a
breast biopsy with benign findings (
) and are thus classified
as having benign breast disease (BBD). As a group, women with
BBD are known to have a higher risk for subsequent
development of breast cancer (BC), and this risk can be stratified by
degree of histological abnormality in the benign biopsy (
Histological findings in BBD comprise a highly diverse spectrum
of changes involving both terminal duct lobular unit (TDLU)
epithelium and accompanying stromal/connective tissue
constituents. Over half of benign biopsies in clinical practice are
characterized by nonproliferative (NP) changes, consisting
mostly of simple cysts, fibrosis, or fibroadenoma; these women
are at only slightly increased risk for subsequent BC
development (4). Most of the remaining biopsies demonstrate one or
more forms of conventional (ie, nonatypical) epithelial
hyperplasia, collectively referred to as proliferative disease without
atypia (PDWA). PDWA conveys an approximately twofold
increased likelihood for subsequent BC compared with
agematched population controls (
). The remaining 10% of
benign breast biopsies show atypical hyperplasia (AH); for women
with AH, BC risk is elevated approximately fourfold (
Because histological subtype of BBD is strongly associated
with risk, this information is central to the assessment and
management of BC risk. Other factors that affect BC risk for women
with BBD include BRCA1/2 mutations, family history of BC,
extent of age-related lobular involution, mammographic breast
density, and parity (
), and models developed for
prediction of BC risk have identified complex interactions of these
factors for risk assessment (
). Analysis of cohorts of women
with BBD indicate that a substantial fraction have undergone a
prior benign breast biopsy (
), and two commonly used
models for BC risk assessment, the Gail/Breast Cancer Risk
Assessment Tool (BCRAT) and International Breast cancer
Intervention Study (IBIS) models, incorporate number of benign
biopsies as a risk prediction feature (
). However, there has
not previously been information regarding how specific
characteristics of the prior BBD impact BC risk at subsequent biopsy for
women with multiple serial benign biopsies. The fact that breast
cancers in women with any form of BBD typically develop on
average eight to 12 years after their benign biopsy (
that breast carcinogenesis is lengthy and complex. Risk
assessments based on pathologic findings observed at a single time
point may lack potentially relevant data.
Because most women will undergo only one benign biopsy
in clinical practice, the natural history of the microscopic
lesions that comprise BBD is obscure. It is unknown, for example,
whether and how often proliferative or atypical lesions persist,
“regress” (to nonproliferative), or “progress” (ie, from
proliferative to atypical). Knowledge of histological changes in women
with BBD over time might be useful not only for improved BC
risk assessment, but also for substantially improved
understanding of the natural history of BC development overall.
This study reports the pathologic findings and breast cancer
occurrence in a single institutional cohort of women with BBD
who underwent metachronous benign biopsies. We hypothesized
that histological category would change in at least some
followup biopsies (ie, compared with the initial biopsy) and that risk
status might change in a clinically significant manner for groups
defined by different pathologic findings in second biopsies.
The Mayo Benign Breast Disease Cohort is an ongoing internal
review board–approved initiative that presently consists of 13
466 women age 18 to 85 years who underwent a benign breast
biopsy for clinical indication at Mayo Clinic Rochester during
the years 1967 to 2001, as detailed previously (
). Within this
cohort, women were identified who had undergone more than
one benign biopsy prior to any diagnosis of breast cancer (n ¼
1649). Biopsies occurring within 60 days of the index biopsy
were excluded (n ¼ 168), as well as 67 subjects without complete
histology information, resulting in a final sample size of 1414.
Outcomes and demographic data were obtained via medical
records, tumor registries, and serial questionnaires. Family
history was categorized as none, weak, or strong. Women were
categorized as strong family history if they had at least one
first-degree relative with breast cancer before the age of 50
years or two or more relatives with breast cancer, with at least
one being a first-degree relative. Any lesser degree of family
history of breast cancer was categorized as weak. Appropriate
documentation of research authorization as per Mayo Clinic
Institutional Review Board policies was obtained.
Original archival hematoxylin and eosin–stained tissue sections
from all biopsies were reviewed by a single study pathologist
(DWV) who was blinded to initial pathologic diagnoses and all
clinical data, including outcome. Overall histological
classifications were assigned to all biopsies as described previously
(Supplementary Table 1, available online) (
Among all women, we compared those in the multiple biopsy
cohort (MBC) with those with just one biopsy using chi-square
tests for categorical variables and Wilcoxon rank sum tests for
continuous variables. We calculated age-adjusted P values
using logistic regression. Similar comparisons were made
across development of a subsequent breast cancer (yes vs no)
for women in the MBC. The duration of follow-up was
calculated as days from benign biopsy to breast cancer diagnosis,
death, prophylactic mastectomy, reduction mammoplasty,
LCIS, or last contact. We examined comparisons of time from
initial and follow-up biopsy to breast cancer using Fine and
Gray’s proportional subdistribution hazards model, accounting
for death as a competing risk (
). Bowker’s test of symmetry
was used to compare impression at index and second biopsy.
We examined associations of clinical and pathological
characteristics at index biopsy with change in overall impression
using logistic regression, stratified by histologic impression.
Each variable was initially examined univariately; subsequent
analyses adjusted for all variables found to be statistically
significant in univariate models.
We fit Fine and Gray models stratified by impression at index
biopsy as NP women are not able to regress and AH women are
not able to progress. Analyses adjusted for age at index biopsy,
year of index biopsy, extent of lobular involution, family history
of breast cancer, and time between index and second biopsy. One
final model was fit including women of all index impressions.
Cumulative incidence curves accounting for death as a competing
risk were created by impression at second biopsy for illustrative
purposes. Breast cancer risk at follow-up biopsy in the MBC was
compared with risk at index biopsy for women not in the MBC
using 1 degree of freedom Wald tests on hazard ratios (HRs)
stratified by presence in the MBC. All analyses were performed
using SAS version 9.4. A P value of less than .05 was considered
statistically significant, and all statistical tests were two-sided.
Demographic, Histological, and Outcome Characteristics of the Multiple Biopsy Cohort
Overall, women in the multiple biopsy cohort (MBC) comprised
10.5% of the Mayo BBD cohort (1414/13 466) (Supplementary
Table 2, available online). Age at time of (initial) biopsy was
statistically significantly younger for patients in the MBC compared
with the overall BBD cohort (48% vs 31% who were younger than
age 45 years, P < .001). Women in the MBC were also statistically
significantly more likely to have a family history of breast
cancer (49.3% vs 37.9%, P ¼ .001). At initial biopsy, 61.0% of MBC
women were classified histologically as NP, 34.4% as PDWAs,
and 4.6% as AH. This distribution was not statistically
significantly different from the Mayo BBD cohort overall.
The median time between biopsies was 5.6 years (Table 1).
About half of the second biopsies occurred within five years of
the initial biopsy (21.4% were < 2 years, 25.3% were 2–5 years)
but a substantial proportion (30%) occurred more than 10 years
later (indications for performance of subsequent biopsy in
Supplementary Table 3, available online). Second biopsies were
statistically significantly more likely than initial biopsies to
harbor proliferative and atypical histological findings (Table 2): 41.4
% of second biopsies were PDWA or AH compared with 34.4% of
initial biopsies, and indication for subsequent biopsy differed by
histologic impression of subsequent biopsy (Supplementary
Table 4, available online). Similarly, AH classification was more
common among follow-up samples (10.3% vs 4.6%, P .001).
Although most patients (56.1%) retained the same overall
histological classification in both samples, 29.9% progressed from NP
or PDWA at second biopsy, and 14.0% regressed from AH or
Breast cancers developed (ie, after second biopsy) in 140 of
the 1414 women (9.9%). Women who developed breast cancer
were more likely to have AH at either the first or second biopsy
compared with women who did not develop cancer;
furthermore, the highest frequencies of PDWA and AH (49.3% and
17.9%, respectively) were present among the follow-up biopsies
of patients who subsequently developed breast cancer (Table 1).
Breast cancer risk at initial and subsequent biopsies was not
affected by age of biopsy, year of biopsy, or lobular involution.
Histologic impression in adjacent benign tissue was available
for 124 of the 140 breast cancer cases (Supplementary Table 5,
available online). In these, the presence of atypical findings was
far more common than what was found in the previous benign
biopsies: 80 (64.5%) had either atypical hyperplasia or lobular
carcinoma in situ adjacent to the cancerous tissue, another 35
(28.2%) had proliferative disease without atypia, and nine (7.3%)
had nonproliferative findings.
Women with AH had increased risk of breast cancer,
whether present at the initial (HR ¼ 4.60, 95% confidence
interval [CI] ¼ 2.41 to 8.79, P < .001) or follow-up biopsy (HR ¼ 3.40,
95% CI ¼ 2.08 to 5.55, P < .001) (Table 1). Long-term risk for
women based on histologic classification in the second biopsy
was stratified by histologic category (Figure 1), similar to prior
findings in the overall BBD cohort (
Joint Effects of Histologic Impression at Initial and
Multivariable survival analyses examining the independent
effects of histologic classification at initial biopsy and at second
biopsy for those in the MBC are provided in Supplementary
Table 6 (available online). After adjustment for relevant
demographic and clinical characteristics and for histologic
impression at second biopsy, impression at first biopsy remained a
statistically significant predictor of breast cancer risk (PDWA
HR ¼ 1.56, 95% CI ¼ 1.04 to 2.35; AH HR ¼ 3.24, 95% CI ¼ 1.61 to 6.
52; P ¼ .003). Histologic impression at second biopsy was
similarly independently associated with risk after adjustment
for demographic and clinical variables and impression at initial
biopsy (PDWA HR ¼ 1.60, 95% CI ¼ 1.06 to 2.42; AH HR ¼ 2.42, 95%
CI ¼ 1.37 to 4.29; P ¼ .007).
Analysis of Factors Associated With Changed
Histological Impression in Second Biopsy
For women with NP at index biopsy, age at biopsy (<45 years
less likely and 45–55 years more likely), time between biopsies
(5þ years more likely, <5 years less likely), presence of
columnar alteration, and contralaterality of the nonsynchronous
biopsies were statistically significantly associated with changes in
histological impression (Supplementary Table 7, available
online). In women with PDWA in their first sample, age (55þ years
more likely) and usual ductal hyperplasia (none more likely)
were statistically significantly associated with changes in
impression, but not interval to second biopsy or laterality. In AHs,
only age (55þ years less likely) and number of atypical foci (3þ
less likely) were marginally associated with change in
Changes in Overall Histological Classification and Breast
Cancer Risk in the MBC Group
Overall risk estimates by histological category were similar for
women with multiple biopsies compared with women with
only a single biopsy (Supplementary Table 8, available online).
Breast cancer risk was uniformly modified in the expected
direction as a result of changes in overall histological classification
between first and second biopsy (Table 3). When subset to NP
on initial biopsy, women with PDWA/AH at follow-up biopsy
were more likely to develop breast cancer than women with NP
in both initial and later biopsies (HR ¼ 1.77, 95% CI ¼ 1.06 to 2.94,
P ¼ .03) (Table 3 and Figure 2A).
Similarly, when subset to the category of PDWA on initial
biopsy (Table 3 and Figure 2B), breast cancer risk decreased in
women whose second biopsy showed NP (HR ¼ 0.49, 95%
CI ¼ 0.25 to 0.98) and increased in women whose second biopsy
showed AH (HR ¼ 1.49, 95% CI ¼ 0.73 to 3.05), compared with the
referent group of women whose second biopsy remained PDWA
(overall P ¼ .04).
Among the 65 women diagnosed with AH at initial biopsy in
the MBC, breast cancers developed in seven out of 23 (30.4%)
who retained atypia vs eight out of 42 (19.0%) of those without
atypia at subsequent biopsy. Although not statistically
significant (possibly because of small numbers), the hazard ratio was
decreased in women whose later biopsies did not show AH
(HR ¼ 0.57, 95% CI ¼ 0.10 to 3.32, vs 1.0 ref for persistent AH)
(Table 3 and Figure 2C). However, when comparing with a
reference group with NP in both samples, women with AH at first
biopsy only had hazard ratio of 3.36, which increased to 7.30
among women with AH in both initial and later biopsy. There
was no evidence supporting risk differences between the MBC
at follow-up biopsy and women without a follow-up biopsy for
any of the modeled characteristics (P ¼ .77 for PDWA histologic
impression vs NP, P ¼ .90 for AH vs NP, P ¼ .34 for age at biopsy
45–55 vs <45 years, P ¼ .85 for age 56þ vs <45 years, P ¼ .53 for
year of biopsy 1982–1991 vs 1967–1981, P ¼ .49 for 1992–2014 vs
1967–1981, P ¼ .46 for partial lobular involution vs none, P ¼ .47
for complete lobular involution vs none, P ¼ .20 for weak family
history vs none, P ¼ .82 for strong family history vs none)
0.97 (0.64 to 1.47)
1.30 (0.73 to 2.30)
1.06 (0.67 to 1.67)
1.59 (0.95 to 2.65)
1.79 (1.20 to 2.66)
4.60 (2.41 to 8.79)
1.03 (0.66 to 1.62)
0.69 (0.33 to 1.45)
1.00 (0.97 to 1.02)
1.25 (0.80 to 1.94)
1.35 (0.88 to 2.07)
1.77 (1.22 to 2.57)
3.40 (2.08 to 5.55)
0.81 (0.47 to 1.38)
0.60 (0.34 to 1.07)
*Follow-up variables were used as the time covariate in the regression modeling. AH ¼ atypical hyperplasia; CI ¼ confidence interval; HR ¼ hazard ratio; NP ¼
nonproliferative disease; PDWA ¼ proliferative disease without atypia.
†The P value was calculated using a two-sided type 3 Wald test.
‡Proportional subdistribution hazards model with death as a competing risk. Time was modeled as time from index biopsy to cancer for index biopsy characteristics
and time from second biopsy to cancer for secondary biopsy characteristics. Age at index biopsy was used as an adjustment term for the characteristics at index
biopsy, and age at second biopsy was used for characteristics at second biopsy.
§Involution status was not assessed in samples with fewer than four background lobules.
(Supplementary Table 9 and Supplementary Figure 1, available
To determine if our results were robust to the changes in the
minimum time interval required between biopsies, we ran
sensitivity analyses that re-examined associations of change in
histologic impression with breast cancer risk after requiring a
minimum of one year (resulting in a sample size of n ¼ 1288),
three years (n ¼ 983), and five years (n ¼ 754) between biopsies.
Results are presented in Supplementary Table 8 (available
online). We see some attenuation in effect for women progressing
It has long been known that a substantial proportion of women
with BBD will have chronic, recurring, or worsening clinical
and/or radiographic abnormalities. Hence, many will undergo
more than one biopsy over their lifetime in order to rule out
malignancy. Although some studies have observed clinical breast
cancer risk association to this subset of BBD (
), the frequency
and pathological aspects of these cases have not been
adequately addressed. This study is the first to systematically
describe the histological features and outcome in a
wellcharacterized cohort of such women with long-term follow-up.
For women in the Mayo BBD cohort with multiple
nonsynchronous biopsies, the overall histological classification
changed from initial to second breast biopsy in nearly half
(43.9%). Changes in histological classification overall were about
twice as likely to involve progression to a higher-risk category
as regression to a lower-risk category. Importantly, altered
histological classification statistically significantly modified
risk for developing breast cancer in the expected or intuitive
direction, according to findings in the second biopsy in each of
the major histological categories of BBD.
A sizable proportion of women with NP in their initial biopsy
progressed to PDWA or AH in their second sample (41.0%). This
finding is clinically significant owing to increased cancer risk in
the group of women who “progressed” to PDWA and adds to our
understanding of NP. Most women with BBD—about 65% across
multiple studies—are diagnosed with NP. Our results imply that
breast cancer risk status can evolve in many women with NP
and that BC risk can increase for some of these women.
Despite a lack of systematic studies that have tracked
longitudinal histological changes over time among women with BBD,
it is presumed that some changes will occur. For example, there
are well-described alterations in mammographic appearances
of benign breast tissue, possibly related to physiological
). Our group has also observed progression of
lobular involution in patients who have undergone more than
one benign biopsy (
). Such findings are consistent with the
concept that breast cancer risk for individual women changes
over time and is reflected in tissue characteristics.
One potential limitation to our study is that a breast biopsy
provides a small sample of any patient’s total breast tissue, and
small proliferative or atypical lesions may not be present in the
biopsy. However, we have observed statistically significant
*All analyses adjusted for age at index biopsy, year of index biopsy, extent of lobular involution, family history of breast cancer, and time between index and second
biopsy. AH ¼ atypical hyperplasia; CI ¼ confidence interval; HR ¼ hazard ratio; NP ¼ nonproliferative disease; PDWA ¼ proliferative disease without atypia.
†Proportional subdistribution hazards model with death as a competing risk. Time was modeled as time from index biopsy to cancer for index biopsy characteristics
and time from second biopsy to cancer for secondary biopsy characteristics. The P value was calculated using a two-sided type 3 Wald test.
changes in breast cancer risk based upon such samples,
suggesting that diagnostic breast biopsy samples may reasonably
reflect the larger field of at-risk tissue. Moreover, the finding of
additional AH in 35% of second biopsies of patients who had AH
in their initial biopsy further supports the notion that evolution
of pathologic findings is not merely driven by random sampling
effects but favors a “field effect” hypothesis.
Another potential limitation of our study is that the findings
were necessarily made in a clinically distinct subset of BBD
women who underwent more than one benign biopsy for
clinical reasons. Although the MBC subset does not appear greatly
different from other subjects in the Mayo BBD cohort on the
basis of initial overall histology or cancer incidence, the two
differed statistically significantly by age and family history of
breast cancer and the fact that they had clinically actionable
recurrent ongoing clinical abnormalities. We cannot exclude the
possibility that women with successive benign biopsies are
more likely to develop histologically progressive breast lesions.
An additional limitation that should be noted is that these
breast biopsies were reviewed and characterized by a single
pathologist with particular expertise in breast pathology.
Assessment of breast cancer risk at the individual level should
consider the possibility of variability by individual pathologists
in characterization of breast lesions into nonproliferative,
PDWA, and atypia categories; this issue also emphasizes the
importance of clear parameters for histological classification.
In summary, we found that overall risk estimates by
histological category were similar for women with multiple biopsies at
their subsequent biopsy as compared with women with only a
single biopsy. However, for women with multiple breast
biopsies, the histologic classification in prior biopsies stratified their
long-term BC risk estimates. Women with progressive epithelial
proliferation have increased risk, and those with less
proliferation have decreased risk. These findings have important
implications for optimization of clinical management for the
approximately 100 000 women per year in the United States
who have multiple, metachronous benign biopsies and provide
insight into the natural history of benign breast disease and its
role in breast carcinogenesis.
This work was supported by the National Cancer Institute
(CA187112 to ACD); Bankhead-Coley Foundation (5BC02 to
DCR and AN); Mayo Clinic Breast Cancer–Specialized
Program of Research Excellence (SPORE CA116201 to DWV,
DCR, AN, and LCH).
The study sponsors had no role in design of the study; the
collection, analysis, or interpretation of the data; the writing of the
manuscript; or the decision to submit the manuscript for
publication. The authors disclose no potential conflicts of interest
related to this study.
1. Gutwein LG , Ang DN , Liu H , et al. Utilization of minimally invasive breast biopsy for the evaluation of suspicious breast lesions . Am J Surg . 2011 ; 202 : 127 - 132 .
2. Silverstein M. Where's the outrage ? J Am Coll Surg . 2009 ; 208 : 78 - 79 .
3. Dupont WD , Parl FF , Hartmann WH , et al. Breast cancer risk associated with proliferative breast disease and atypical hyperplasia . Cancer . 1993 ; 71 : 1258 - 1265 .
4. Hartmann LC , Sellers TA , Frost MH , et al. Benign breast disease and the risk of breast cancer . N Engl J Med . 2005 ; 353 : 229 - 237 .
5. McDivitt RW , Stevens JA , Lee NC , Wingo PA , Rubin GL , Gersell D. Histologic types of benign breast disease and the risk for breast cancer . The Cancer and Steroid Hormone Study Group. Cancer . 1992 ; 69 : 1408 - 1414 .
6. Dupont WD , Page DL . Risk factors for breast cancer in women with proliferative breast disease . N Engl J Med 1985 ; 312 : 146 - 151 .
7. Dupont WD , Page DL , Parl FF , et al. Long-term risk of breast cancer in women with fibroadenoma . N Engl J Med . 1994 ; 331 : 10 - 15 .
8. Dyrstad SW , Yan Y , Fowler AM , Colditz GA . Breast cancer risk associated with benign breast disease: Systematic review and meta-analysis . Breast Cancer Res Treat . 2015 ; 149 : 569 - 575 .
9. Page DL , Dupont WD , Rogers LW , Rados MS . Atypical hyperplastic lesions of the female breast. A long-term follow-up study . Cancer . 1985 ; 55 : 2698 - 2708 .
10. Hartmann LC , Degnim AC , Santen RJ , Dupont WD , Ghosh K. Atypical hyperplasia of the breast-risk assessment and management options . N Engl J Med . 2015 ; 372 : 78 - 89 .
11. Hartmann LC , Radisky DC , Frost MH , et al. Understanding the premalignant potential of atypical hyperplasia through its natural history: A longitudinal cohort study . Cancer Prev Res (Phila) . 2014 ; 7 : 211 - 217 .
12. Colditz GA , Rosner B . Cumulative risk of breast cancer to age 70 years according to risk factor status: Data from the Nurses' Health Study . Am J Epidemiol . 2000 ; 152 : 950 - 964 .
13. Dupont WD , Page DL . Breast cancer risk associated with proliferative disease, age at first birth, and a family history of breast cancer . Am J Epidemiol . 1987 ; 125 : 769 - 779 .
14. Milanese TR , Hartmann LC , Sellers TA , et al. Age-related lobular involution and risk of breast cancer . J Natl Cancer Inst . 2006 ; 98 : 1600 - 1607 .
15. 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 . 1989 ; 81 : 1879 - 1886 .
16. Tyrer J , Duffy SW , Cuzick J. 2004 . A breast cancer prediction model incorporating familial and personal risk factors . Stat Med 23 : 1111 - 1130 .
17. Pankratz VS , Degnim AC , Frank RD , et al. Model for individualized prediction of breast cancer risk after a benign breast biopsy . J Clin Oncol . 2015 ; 33 : 923 - 929 .
18. 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 : 3137 - 3143 .
19. Quante AS , Whittemore AS , Shriver T , Strauch K , Terry MB . Breast cancer risk assessment across the risk continuum: Genetic and nongenetic risk factors contributing to differential model performance . Breast Cancer Res . 2012 ; 14 : R144 .
20. Fine JP , Gray RJ . A proportional hazards model for the subdistribution of a competing risk . JAMA . 1999 ; 94 : 496 - 509 .
21. Work ME , Reimers LL , Quante AS , Crew KD , Whiffen A , Terry MB . Changes in mammographic density over time in breast cancer cases and women at high risk for breast cancer . Int J Cancer . 2014 ; 135 : 1740 - 1744 .
22. Radisky DC , Visscher DW , Frank RD , et al. Natural history of age-related lobular involution and impact on breast cancer risk . Breast Cancer Res Treat . 2016 ; 155 : 423 - 430 .