|Home | About | Journals | Submit | Contact Us | Français|
Radial scars (RS) are benign proliferative lesions associated with an increased risk of subsequent breast cancer. However, it remains unclear whether RS are an independent risk factor for breast cancer or whether their association with breast cancer is due to their common occurrence with other proliferative lesions known to increase breast cancer risk.
We performed an updated analysis of the association between RS and subsequent breast cancer risk in a nested case-control study among 460 cases and 1792 controls with benign breast disease (BBD) in the Nurses’ Health Studies. Logistic regression was used to estimate associations between RS in BBD biopsy specimens and breast cancer risk, adjusted for matching factors and breast cancer risk factors, including histologic category of concurrent BBD.
In multivariable models prior to adjustment for histologic category of BBD, RS were associated with a two-fold increased risk of breast cancer (odds ratio [OR] = 2.0; 95% confidence interval [95% CI]: 1.4, 2.8). This association was attenuated but still significant after adjustment for BBD histologic category (OR = 1.6; 95% CI: 1.1, 2.3). In models adjusted for BBD histologic category and other covariates, risk appeared greater among women with multiple RS (1 RS, OR = 1.5; 95% CI: 0.9, 2.3; ≥2 RS, OR = 2.7; 95% CI: 1.5, 5.0; p-heterogeneity = 0.12). There were also suggestions of a greater risk associated with RS among women age ≥50 years at biopsy (p-heterogeneity = 0.07) and for estrogen receptor-negative/progesterone receptor-negative (ER−/PR−) tumors compared with other hormone receptor subtypes (p-heterogeneity = 0.19).
RS appear to be an independent histologic risk factor for breast cancer. Larger studies are needed to further evaluate whether risk is increased when multiple RS are present and whether associations vary by age at biopsy or by hormone receptor status of the breast tumor.
Benign breast disease (BBD) comprises a diverse group of nonmalignant breast abnormalities commonly detected via screening mammography. Although some types of BBD may be premalignant, most are believed to be markers of a generalized increased risk of breast cancer . This risk varies by BBD histologic category; compared with women in the general population, women with nonproliferative disease do not appear to be at elevated risk, whereas those with proliferative disease without atypia (PDWA) are at 1.5–2.0-fold greater risk, and women with proliferative disease with atypia, or atypical hyperplasia (AH), have a 3.5–5.0–fold increased risk [2–4].
Radial scars (RS) are a type of proliferative benign breast lesion characterized by a central fibroelastic core from which ducts and lobules radiate circumferentially . Prior studies have suggested that women with RS are at a two-fold increased risk of breast cancer [6–8], but the nature of the relationship between RS and breast cancer risk remains unclear. Some speculate that RS may be premalignant [9–11] due to their mammographic and histologic resemblance to carcinoma, their frequent association with neoplastic and pre-neoplastic lesions [12–14], and the detection of carcinoma in some RS [14,15,13]. Even if RS are not themselves precancerous, RS may still be independent markers of breast cancer risk that could aid in the identification of women most likely to benefit from risk-reduction strategies. However, as RS are frequently detected in conjunction with other proliferative BBD lesions, the elevated breast cancer risk among women with RS may simply reflect the association of RS with other proliferative lesions, particularly the highest-risk AH lesions.
To investigate whether RS are associated with breast cancer while accounting for the presence of other high-risk BBD lesions, we and two other groups have examined whether the risk associated with RS remains within categories of proliferative BBD (PDWA and AH). In our previous nested case-control study in the Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII) with follow-up from 1976–1992 , RS were still associated with a nearly two-fold increased risk after stratifying by BBD category, providing evidence that RS are an independent histologic risk factor for breast cancer; however, the association reached statistical significance only among women with PDWA. In retrospective cohort studies in the Nashville Breast Cohort and Mayo Benign Breast Disease Cohort, no increased risk was observed among women with RS within proliferative BBD categories, suggesting instead that the association between RS and breast cancer is primarily, if not entirely, due to the presence of other proliferative lesions.
To clarify these conflicting results, we conducted an updated and more powerful analysis of the association between RS and breast cancer risk in the NHS/NHSII, with extended follow-up providing approximately 50% more cases than the previous assessment. We also expanded our previous analyses by evaluating whether the association between RS and breast cancer varies by time since or age at biopsy or by tumor hormone receptor subtype.
We conducted a nested case-control study among participants with biopsy-confirmed BBD in the NHS and NHS II. The NHS is an ongoing prospective cohort study that began in 1976, when 121,700 female registered nurses between the ages of 30 and 55 years completed a mailed questionnaire. The NHS II is a separate cohort study consisting of 116,609 female registered nurses who were between the ages of 25 and 42 years when the study began in 1989. Participants in both cohorts have been followed via biennial questionnaires that have collected information on lifestyle factors and incident disease. On the 1976, 1978, and 1980 questionnaires, participants were asked if they had ever been diagnosed with fibrocystic or other BBD and whether this diagnosis had required hospitalization; from 1982 onward, the questionnaires have inquired specifically about BBD confirmed by biopsy. The follow-up methods used in this cohort are very similar to those for the original NHS. On each biennial questionnaire, participants have been asked if they have ever been diagnosed with BBD and if so, whether the diagnosis was confirmed by biopsy or aspiration. The follow-up rate for each NHS/NHSII two-year cycle has been greater than 90% of the original cohorts.
Within the subcohort of women who reported a prior diagnosis of BBD confirmed by biopsy, eligible cases were women who reported a first diagnosis of breast cancer between 1976 and the return of the 1998 questionnaire in the NHS or between 1989 and return of the 1995 questionnaire in the NHS II. Self-reported breast cancers were confirmed by review of medical records, and both invasive breast cancer and carcinoma in situ were included in the study. Eligible controls were women who completed the questionnaire for the same year that the breast cancer case was reported and had a previous diagnosis of biopsy-confirmed BBD, but were free from breast cancer; within each NHS cohort, they were matched to up to four breast cancer cases on age at breast cancer diagnosis or index date, year of benign breast biopsy, and time since benign biopsy. Informed consent was obtained from all participants. The study was approved by the Human Subjects Research Committee of Brigham and Women’s Hospital, Boston, Massachusetts.
Eligible cases and controls were contacted for permission to obtain their BBD pathology records and biopsy specimens, and specimens were then obtained from hospital pathology departments when possible. These methods have been described in detail elsewhere [16,17]. Briefly, >70% of the 1310 cases and 6055 controls who were originally identified for the study confirmed the diagnosis and granted permission, and specimens were subsequently obtained for 48% of those who had granted permission (438 cases and 2096 controls). The primary reason given by hospital pathology departments for not sending specimens was that they had been destroyed or were no longer available (36%) [16,17]. Biopsy slides were independently reviewed by one of three study pathologists (SJS, JLC, LCC) who were blinded to the participants’ case or control status. The pathologists completed a detailed worksheet with information on the morphologic features of each specimen, including the presence and number of RS. Biopsies were then classified as nonproliferative, PDWA, or AH, according to the criteria of Page et al. [18,19], which have been used in previous investigations in this cohort [16,20,2]. All biopsies, including bilateral biopsies, were classified according to the most severe changes present, and specimens with possible or definite AH were jointly reviewed by two pathologists. As RS are considered to be a proliferative lesion, biopsies classified as nonproliferative excluded those with RS.
After excluding participants whose benign biopsy specimens were of poor quality or who had no breast tissue for evaluation, evidence of carcinoma in situ or invasive carcinoma, invalid dates of diagnosis, or insufficient information on laterality, 469 cases and 1,842 controls were available for our analyses. An additional 59 participants were excluded for missing data on covariates. Our final population for analysis consisted of 460 cases and 1792 controls.
The distribution of breast cancer risk factors was examined according to RS presence among the controls, adjusting for age at benign biopsy when appropriate. Unconditional logistic regression with adjustment for matching factors was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for breast cancer risk according to RS presence and number. The following covariates, measured in the cycle prior to the breast cancer diagnosis or index date, were considered as potential confounding factors (see Table 2 footnote for variable classifications used): first-degree family history of breast cancer, age at menarche, menopausal status, jointly classified parity and age at first birth, body mass index (BMI) at age 18, and postmenopausal hormone (PMH) use. Because BMI at age 18 appeared to be more strongly related to breast cancer risk than current BMI, we chose to consider adjustment for BMI at age 18 and change in weight since age 18 rather than current BMI. We selected the four covariates producing the greatest change in the OR for inclusion in multivariable models; further adjustment for other covariates did not substantially alter estimates.
In our first set of models, we examined the overall association between RS and breast cancer risk. To assess whether RS were associated with breast cancer risk independent of concurrent proliferative disease, additional adjustment was made for histologic category of BBD in another set of models. We also jointly classified women according to proliferative category and RS presence and stratified by proliferative category to evaluate whether the association between RS and breast cancer risk differed by proliferative category; a likelihood ratio test was used to test for interaction. We examined associations between RS number and breast cancer risk in additional unconditional logistic regression models.
We examined relationships between RS and combined estrogen receptor (ER)/progesterone receptor (PR) status as well as laterality of the subsequent breast cancer (i.e., whether the breast tumor was in the same or opposite breast as BBD) in polytomous logistic regression models. Pathology reports were used as the primary source of information on ER status, PR status, and laterality. Information on receptor status from previously performed tissue microarray analyses (TMA)  was used when ER or PR status was not available from pathology reports. Receptor status abstracted from pathology reports has been shown in our cohort to have a high concordance with that obtained from TMA . Women with either bilateral BBD or breast cancer were excluded from the laterality analyses.
To examine whether associations between RS and breast cancer might depend on time since BBD biopsy and age at biopsy, we conducted stratified analyses and tested for interaction in logistic regression models. In stratified analyses, time since BBD biopsy was dichotomized as <10 years and >10 years and age at biopsy as <50 years and >50 years to allow comparison with results from the Nashville and Mayo studies; Wald tests were used to assess whether differences by these factors were statistically significant. We also examined relationships between RS and combined ER/PR status as well as laterality of the breast tumor in polytomous logistic regression models. SAS 9.3 (SAS Institute, Cary, NC) was used for all analyses except tumor subtype analyses, for which STATA 12.1 (StataCorp LP, College Station, TX) was used. All tests were two-sided, with p <0.05 indicating statistical significance.
Among all study participants, the mean age at BBD biopsy was 44.6 years (SD = 10.3 years), the mean time since biopsy was 9.7 years (SD = 7.2 years), and the mean number of slides reviewed was 4.5 (SD = 4.2). Women with RS were on average slightly older at BBD biopsy than women without RS. After adjustment for age at biopsy, the distribution of breast cancer risk factors was generally similar among women with and without RS, although women with RS had a lower BMI at both BBD biopsy and age 18 as well as slightly higher parity (Table 1).
In multivariable logistic regression models controlling for matching factors and covariates other than BBD histologic category (Table 2), women with RS were at a two-fold higher risk of breast cancer compared with those without RS (OR = 2.0; 95% CI: 1.4, 2.8). After additional adjustment for BBD histologic category, this association remained significant but was moderately attenuated (OR = 1.6; 95% CI: 1.1, 2.3). These associations were essentially unchanged when only invasive breast cancer cases were included (data not shown). The presence of RS was associated with an increased risk of breast cancer both among women with PDWA (OR = 1.8, 95% CI: 1.1, 2.8) and AH (OR = 1.5; 95% CI: 0.8, 2.9). While results were significant only among women with PDWA, there was no evidence that the association differed by BBD proliferative category (p-heterogeneity = 0.38). Women with both RS and AH were at the highest risk compared with women with nonproliferative disease and no RS (OR = 4.8; 95% CI: 2.6, 8.9), although this risk did not differ significantly from that among women with AH and no RS.
Among women with RS, one RS lesion was present in 67.3% of specimens, two lesions were present in 16.7% of specimens, and 3 or more lesions were present in 16.0% of specimens. After adjustment for BBD histologic category and other covariates, women with multiple RS appeared to be at higher risk of breast cancer (RR for ≥2 RS vs. 0 RS = 2.7; 95% CI: 1.5, 5.0) than women with a single RS (RR for 1 vs. 0 RS = 1.5; 95% CI: 0.9, 2.3). However, this difference in risk did not reach significance (p = 0.12).
Multivariable associations between RS and breast cancer were slightly stronger and only significant in the first decade after BBD biopsy (Table 3). After adjustment for BBD histologic category, RS were associated with a 70% increased risk of breast cancer in the first 10 years since BBD biopsy (95% CI: 1.1, 2.7) compared with a 40% increased risk with subsequent follow-up (95% CI: 0.7, 2.6), although this difference by time since biopsy was not significant (p-heterogeneity = 0.57). RS appeared to be associated with breast cancer risk only among women over age 50 at BBD biopsy, with the difference by age approaching significance (p-heterogeneity= 0.07). After adjustment for BBD histologic category, the presence of RS was associated with a three-fold increased risk of breast cancer among women over age 50 years (OR = 3.3; 95% CI: 1.7, 6.1), whereas no increased risk was observed among younger women (OR = 1.1; 95% CI: 0.7, 1.8). These findings were very similar in analyses stratified by menopausal status (data not shown).
The association between RS and breast cancer did not vary significantly by breast cancer hormone receptor subtype in models adjusted for BBD histologic category and other covariates (p-heterogeneity = 0.19), although there was some suggestion of a stronger association for ER−/PR− than either ER+/PR+ or ER+/PR− breast cancer (Table 4). In fact, RS were only significantly associated with risk of ER−/PR− breast cancer, with risk of this subtype nearly three-fold higher for women with RS compared with those without RS (OR=2.8; 95% CI: 1.4, 5.8). RS appeared to be equally associated with ipsilateral and contralateral breast cancer (p-heterogeneity = 0.80).
This re-analysis of the association between RS and breast cancer in the NHS/NHSII confirms results from our previous analysis suggesting that RS are an independent histologic risk factor for breast cancer. As in the original analysis, RS were associated with a two-fold increased risk of breast cancer. This association was slightly attenuated after adjustment for BBD histologic category, suggesting that some of the risk associated with RS reflects their frequent co-occurrence with other proliferative lesions. However, RS remained associated with a nearly two-fold increased risk of breast cancer even after accounting for concurrent proliferative disease. Associations were suggestively stronger for ER−/PR− tumors, among women with multiple RS, and among women older than age 50.
As in our study, the Nashville Breast Cohort and Mayo Benign Breast Disease Cohort studies also observed RS to be associated with a two-fold increased risk of breast cancer, at least in the first 10 years since BBD biopsy. However, the substantial attenuation in risk within BBD proliferative categories in both studies suggests that most, if not all, of this increased risk can be accounted for by concurrent proliferative disease (Table 5). There was some suggestion in both the Nashville and Mayo cohorts as well as in the NHS/NHSII that any independent association between RS and breast cancer may be limited to women with PDWA, but this requires further examination in analyses with greater power to assess associations by BBD category.
We did not find evidence that the risk associated with RS is greater in the first decade after BBD biopsy, as suggested in the Nashville Breast Cohort. However, results from both our study and the Nashville Cohort suggest that the risk associated with RS may be stronger among women older than age 50. In fact, the association between RS and breast cancer, both overall and independent of other BBD lesions, appeared to be limited to women over age 50 in the NHS/NHSII. Although associations by age were based on relatively small numbers in both the NHS/NHSII and Nashville Cohort, an effect of age or menopausal status on the association between RS and breast cancer is supported by the more frequent co-occurrence of RS and breast cancer among women over age 50 reported in several studies [9,14,13].
Differences in methodology and study populations may explain why, unlike in the NHS/NHSII, RS were not found to be a significant independent risk factor for breast cancer in the Nashville and Mayo cohorts. As only our study adjusted for BBD histologic category to assess RS as a potential independent risk factor for breast cancer, it is possible that a modest independent association could have been missed due to small numbers in analyses stratified by BBD category in the Nashville and Mayo studies. In the Mayo study, the use of an external reference group (the Iowa SEER population) that included both women with and without RS may also have led to an underestimate of associations. The weaker associations in the Nashville study may in part reflect the slightly younger population in that cohort compared with the NHS/NHSII, as the association between RS and breast cancer appears to be stronger at older ages. The predominance of biopsies performed in the pre-mammographic era in the Nashville study may also have contributed to the weaker associations in that study, as evidence suggests that the larger RS lesions detected via mammography are more strongly associated with breast cancer risk than those detected primarily as incidental findings in the pre-mammographic era [14,6].
Our results suggest that RS elevate breast cancer risk beyond that of associated proliferative disease, but whether RS might themselves be premalignant or simply a marker for precancerous changes in the breast is uncertain. Some have hypothesized that RS represent an early breast cancer precursor lesion largely based on the finding of carcinoma and precancerous lesions within RS [9,23] and the histologic resemblance of RS to carcinoma, particularly tubular carcinoma [10,11]. This hypothesis is also supported by the detection of similar levels of hyaluronic acid, a component of the extracellular matrix, in RS and tubular carcinoma  as well as the observation of abnormal ER expression and chromosomal abnormalities in RS and carcinoma . However, these similarities between RS and carcinoma do not definitively demonstrate that RS are breast cancer precursors. The similar risk associated with ipsilateral and contralateral breast cancer in our study suggests that instead of being breast cancer precursors, RS are more likely markers for other precancerous changes in the breast . Further work is needed to understand how RS may be related to breast carcinogenesis, although it has been hypothesized that the presence of RS may signal a disruption in normal stromal-epithelial interactions , which are believed to play a role in the development of breast and other cancers [27–30]. The similarly increased mRNA expression of stromal vascular factors observed in RS and invasive carcinoma  and the loss of fibrocytic stromal cells in both RS and tubular carcinoma  support the notion that RS are related to precancerous stromal changes.
Our study is, to our knowledge, the first to examine the association between RS and breast cancer by hormone receptor subtype. The apparent stronger association between RS and ER−/PR− compared with ER+/PR+ or ER+/PR− breast cancer suggests that RS may signal the presence of non-hormonal premalignant changes in the breast. However, this stronger association with receptor-negative subtypes is somewhat puzzling given that previous clinical studies have reported RS to be most frequently found in conjunction with tubular carcinoma [15,32], a breast cancer subtype that is most commonly ER+/PR+ [33,34]. Given this unexpected association with hormone receptor-negative cancers and the small numbers in analyses by receptor subtype, our findings require confirmation in future studies.
Our study has several limitations and strengths. Even with approximately 50% more cases than the previous NHS analysis, power was limited in analyses examining potential effect modifiers and even more so in analyses by ER/PR subtype, as not all participants had information available on hormone receptor subtypes. Additionally, we were only able to obtain BBD biopsy slides for approximately one-third of study participants. However, the success rate did not differ by case-control status and women appeared to be missing completely at random, as those with and without specimens were similar with respect to their distribution of breast cancer risk factors, including BMI, parity, and PMH. Strengths of our study are its prospective design, the confirmation of breast cancer cases via medical record review, the centralized pathology review of BBD specimens, and the detailed collection of information on potential confounders.
In conclusion, our results suggest that RS are associated with risk of subsequent breast cancer independent of simultaneous proliferative disease. As RS may further elevate risk among women with other high-risk lesions, our findings have potentially important clinical implications. However, future studies with larger numbers of cases are needed to confirm our main results and to further evaluate whether risk is influenced by the presence of multiple RS, age at biopsy, and hormone receptor status of the breast tumor.
Funding/Support: Public Health Service Grants CA046475, CA087969, CA050385, T32 CA 09001, and SPORE in Breast Cancer CA089393 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services and the Breast Cancer Research Foundation, and the American Cancer Society (to G. A. Colditz).
In addition, we would like to thank the participants and staff of the Nurses' Health Study and the Nurses' Health Study II, for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY.
Conflicts of Interest: The authors declare that they have no conflicts of interest.