It is unclear whether women with ductal carcinoma in situ (DCIS), like their counterparts with invasive breast cancer, warrant genetic risk assessment and testing on the basis of high-risk variables. The authors of this report identified predictive factors for mutations in the breast cancer-susceptibility genes BRCA1 and BRCA2 in women who were diagnosed with DCIS.
One hundred eighteen women with DCIS who were referred for genetic counseling and underwent genetic testing for BRCA1/BRCA2 mutations between 2003 and 2010 were included in the study. Logistic regression models were fit to determine the associations between potential predictive factors and BRCA status.
Of 118 high-risk women with DCIS, 27% (n = 32) tested positive for BRCA1/BRCA2 mutations. Of those, 10% (n = 12) and 17% (n = 20) had BRCA1 and BRCA2 mutations, respectively. Age, race, and tumor characteristics did not differ between BRCA noncarriers and carriers. In a multivariate logistic model, ≥2 relatives with ovarian cancer (OC) (odds ratio [OR], 8.81; 95% confidence interval [CI], 1.38-56.29; P = .034), and a score ≥10% according to the BRCAPRO mathematical model for calculating the probability that a particular family member carries a germline BRCA mutation (OR, 6.37; 95% CI, 2.23-18.22; P = .0005) remained as independent significant predictors for a BRCA mutation. Fifty-seven percent of mutation carriers but only 25% of noncarriers underwent prophylactic mastectomy(P = .0037). This difference remained significant for patients aged ≤40 years (P = .025).
Ductal carcinoma in situ (DCIS) is the fourth leading cause of cancer among women in the United States. Between 1980 and 2001, age-adjusted rates of DCIS increased 7.2-fold, likely because of the widespread use of screening mammography.1 DCIS is not immediately life-threatening, but it is associated with an increased risk of invasive breast cancer (IBC). It is estimated that 14% to 50% of DCIS lesions will progress to IBC if left untreated.2 The identification of deleterious mutations in the breast cancer-susceptibility genes BRCA1 and BRCA2 has important implications for carriers in general, because they are the principal cause of hereditary breast and ovarian cancer syndrome (HBOC).3, 4 Women who carry a germ line mutation in BRCA1 or BRCA2 have a 43% to 84% risk of developing breast cancer (BC) and a 22% to 39% risk of developing ovarian cancer (OC) by age 70 years.5-7
The association of BRCA mutations with IBC is well established; however, to date, there is no consensus on whether or not DCIS is part of the BRCA-associated hereditary breast-ovarian cancer spectrum. Although early studies revealed a lower prevalence of DCIS associated with IBC in BRCA1/BRCA2 mutation carriers,8-11 the subsequent identification of DCIS in prophylactic mastectomy specimens from carriers led investigators to reconsider DCIS as a component of HBOC.12-16 Although 2 studies have reported a similar incidence of DCIS in patients with BRCA1/BRCA2-positive tumors compared with BRCA-negative tumors,15, 17 Kauff et al18 observed that preinvasive lesions actually were more common in prophylactic mastectomy tissue specimens from carriers than in breast tissue obtained at autopsy from unaffected women who had no known hereditary predisposition to BC.
A few retrospective studies have examined the prevalence of BRCA1/BRCA2 mutations in women diagnosed with DCIS and have reported mutation rates ranging between 3.3%14 and 13%.11, 19 Those studies support the concept that women with DCIS, like their counterparts with IBC, warrant genetic risk assessment and testing on the basis of high-risk variables. Notably, the knowledge of a BRCA1/BRCA2 mutation is likely to significantly change the assessment of a DCIS patient's risks for future cancers and the cancer prevention/risk reduction recommendations that would be considered. Therefore, it is important to define the disease and clinical predisposing conditions for mutations in the BRCA1/BRCA2 genes as accurately as possible to facilitate genetic counseling and risk estimation in women with DCIS. Herein, we report the prevalence of deleterious BRCA1/BRCA2 mutations and further identify the predictive factors for BRCA1/BRCA2 mutations in a selected population of women with pure DCIS who had no previous history of BC.
MATERIALS AND METHODS
Patient Population and Data Collection
The prospectively maintained Breast Cancer Management System database at the University of Texas M. D. Anderson Cancer Center (MDACC) identified 118 women with DCIS who were referred for genetic counseling and underwent genetic testing for mutations in the BRCA1 and BRCA2 genes between 2003 and 2010. All women were either self-referred or physician-referred based on family history (FH) or young age (≤50 years). Women whose original pathologic specimens were not reviewed by dedicated breast pathologists at MDACC or who had a report that indicated definite invasion or microinvasion were excluded from the analysis. Also excluded were women who had a previous history of BC or whose BRCA mutations indicated a variant of uncertain significance.
Demographic and clinical information was collected from the prospectively maintained breast cancer research database under institutional review board-approved protocols and included age at the time of diagnosis, race, ethnicity (Ashkenazi Jewish [AJ] ancestry or non-AJ ancestry), FH of BC and/or OC in at least 1 first-degree and/or second-degree relative, number of relatives affected with BC and/or OC (first-degree and/or second-degree relatives only), status according to BRCAPRO (a mathematical model for calculating the probability that a particular family member carries a germline BRCA mutation), histopathologic features of tumors, type of surgery, and recurrence information. An FH of BC was considered to be early onset if the affected family member was diagnosed at age ≤50 years.
Pathologic Assessment and Mutation Analysis
All patients who were included in the analysis underwent definitive surgery and had their pathologic specimens reviewed by dedicated breast pathologists at MDACC. Tumor grade was defined according to the modified Black nuclear grading system. Immunohistochemical analysis was used to determine estrogen receptor (ER) and progesterone receptor (PR) status. Nuclear staining ≥10% for ER or PR was considered strongly positive. BRCA testing was performed using germline DNA (from blood) by Myriad Genetics Laboratories, Inc. (Salt Lake City, Utah), and the test results were categorized as either positive or negative for a deleterious mutation.
The University of Texas Southwestern Cancer Gene program, version 5.1, was used to calculate the BRCAPRO scores. This program is available online at http://www.utsouthwestern.edu/cancergene/ (Accessed July 27, 2011). Age, age at death, BC and/or OC diagnoses, and ages at diagnoses were entered for patients and their first-degree and second-degree relatives. BRCAPRO takes into account only invasive breast cancers; however, to account for DCIS, data often are entered into BRCAPRO slightly differently between users, both between and within institutions. Because there is no standard to date, we established criteria in our institution to eliminate user variability, and individuals were entered into the model as having had invasive BC 10 years after their DCIS diagnosis.
Statistical Analysis and Outcome Measures
Patient demographics and clinical characteristics have been tabulated and compared among the 3 groups, which were defined according to BRCA status (noncarrier, BRCA1 carrier, or BRCA2 carrier) and between noncarriers and carriers, using chi-square tests for categorical variables and t tests for continuous variables.20 Univariate and multivariate logistic regression analyses21 on BRCA mutation status (carrier vs noncarrier) were done to identify factors that were predictive of BRCA1/BRCA2 mutations in the patients with DCIS. A logistic regression model was obtained by first including an initial set of candidate predictor variables with P values ≤ .05 in the univariate analysis. A stepwise backward elimination (BE) was then performed using .05 for the significance level of the Wald chi-square statistic for an effect to stay in the model. Once the list of variables to be used in our final model was selected, the functional form of each variable and multicollinearity between the variables were examined. The distribution of recurrence-free survival (RFS), defined as the time from surgery until the first date of documented disease recurrence or the date of last follow-up, was estimated using the Kaplan-Meier product-limit method.22P values ≤ .05 were considered statistically significant, and all tests were 2-sided. Statistical analysis was carried out using the SAS software package (version 9.1.3; SAS Institute Inc., Cary, NC) and S-Plus 8.0 (Insightful Corporation, Seattle, Wash).
Of 118 women who were included in our analysis, 27% (n = 32) tested positive for BRCA1/BRCA2 mutations, 10% (n = 12) carried a BRCA1 mutation, and 17% (n = 20) carried a BRCA2 mutation. Table 1 lists the deleterious mutations in the BRCA1 and BRCA2 genes detected in the study cohort.
Table 1. Deleterious Breast Cancer-Susceptibility Gene Mutations Detected in the Study Cohort
Abbreviations: A, alanine; BRCA1/BRCA2, breast cancer-susceptibility genes 1 and 2, respectively; C, cysteine; del, deletion; E, glutamic acid; G, glycine; ins, insertion; K, lysine; Q, glutamine; R, arginine; S, serine; T, threonine; W, tryptophan, X, unspecified amino acid; Y, tyrosine.
Mutation Carrier Rate by Patient Demographics and Clinical Characteristics
The prevalence of BRCA1 and BRCA2 mutations with regard to patient demographics and clinical characteristics is displayed in Table 2. Seven patients had a previous history of OC, including 4 patients with BRCA1 mutations and 3 patients with BRCA2 mutations. Overall, 45% of patients had an FH of BC, 7% had an FH of OC, and 2.5% had both in at least 1 first-degree or second-degree relative. Among 51% of patients who did not report any FH of BC or OC, the BRCA1 and BRCA2 mutation prevalence was 10% (6 of 60 patients) and 8% (5 of 60 patients), respectively. Patients with an FH of BC had a higher proportion of BRCA2 mutations compared with patients without any FH of BC (26% vs 9%). Similarly, compared with patients who had no FH of OC, patients with an FH of OC had higher proportions of BRCA1 and BRCA2 mutations (8% vs 38% and 16% vs 25%, respectively). BRCA2 mutations were identified in 25 of 74 patients (24%) who had ≥2 family members diagnosed with BC compared with 1 of 27 patients (4%) who had a single relative affected by BC. The BRCA1/BRCA2 mutation rates were similar in patients who had an FH of early onset BC versus late-onset BC.
Table 2. The Prevalence of Breast Cancer-Susceptibility Gene 1 (BRCA1) and BRCA2 Mutations by Patient Demographics and Baseline Disease Characteristics
No. of Patients (%)
BRCA1, N = 12
BRCA2, N = 20
BRCA Noncarriers, N = 86
Abbreviations: AJ, Ashkenazi Jewish; BC, breast cancer; BRCAPRO, mathematical model for calculating the probability that a particular family member carries a germline mutation of the BRCA1 and BRCA2 genes; DCIS, ductal carcinoma in situ; DX, diagnosis; EBC, early onset breast cancer; ER, estrogen receptor; FH, family history; LBC, late-onset breast cancer; OC, ovarian cancer; PR, progesterone receptor.
FH of BC
FH of OC
FH of onset of BC
No. of relatives with BC DX
No. of relatives with OC DX
Of 98 patients with known BRCAPRO status, 25 had an FH of BC and/or OC that exceeded a 10% prior probability of carrying a BRCA mutation using the BRCAPRO model.23 Among patients who had a BRCAPRO probability of <10%, BRCA1 and BRCA2 mutations were detected in 7% and 11%, respectively. Among patients who had a BRCAPRO mutation probability ≥10%, 50% of the patients aged ≤40 years at first diagnosis tested positive for BRCA mutations compared with 65% of patients aged >40 years at first diagnosis (data not shown).
In univariate analyses (Table 3), age at diagnosis and race, as a variable with 4 groups (AJ white, non-AJ white, black, and others), were not predictive of mutation status (P > .47). No significant differences were noted in ER/PR status or nuclear grade with respect to BRCA mutation status. The BRCA1/BRCA2 mutation rate was significantly higher in patients who had bilateral synchronous DCIS compared with patients who had unilateral DCIS (80% vs 25%; P = .018). Patients who had an FH of OC had a higher risk of having BRCA mutations compared with patients who had no FH of OC (63% vs 25%; P = .033). This same trend was observed among patients who had ≥2 family members with OC compared with those who had fewer relatives with OC (P = .004). Conversely, BRCA mutation status was not associated significantly with an FH of BC, early onset BC, or the number of relatives affected by BC (P > .06). BRCA1/BRCA2 mutations were detected more frequently among patients who had a BRCAPRO calculated probability ≥10% compared with patients who had a BRCAPRO mutation probability <10% (60% vs 18%; P = .0001).
Table 3. Association Between Breast Cancer-Susceptibility Gene Positivity and Patients' Demographics and Baseline Disease Characteristics
No. of Patients (%)
BRCA Noncarriers, N = 86
BRCA Carriers, N = 32
Abbreviations: AJ, Ashkenazi Jewish; BC, breast cancer; BRCA, breast cancer-susceptibility gene; BRCAPRO, mathematical model for calculating the probability that a particular family member carries a germline mutation of the BRCA1 and BRCA2 genes; DCIS, ductal carcinoma in situ; DX, diagnosis; EBC, early onset breast cancer; ER, estrogen receptor; FH, family history; LBC, late-onset breast cancer; OC, ovarian cancer; PR, progesterone receptor.
FH of BC
FH of OC
FH of onset of BC
No. of relatives with BC DX
No. of relatives with OC DX
Table 4 provides the multivariate logistic regression model for BRCA incorporating patient and disease characteristics. The presence of ≥2 family members diagnosed with OC and a BRCAPRO mutation probability ≥10% remained as independent, significant predictors for a BRCA mutation. Patients who had ≥2 relatives with OC had a greater odds of being BRCA1/BRCA2 positive than those who had fewer relatives with OC (P = .0345). Specifically, patients who had ≥2 relatives with OC were more likely to have BRCA mutations compared with patients who had no relatives with OC (odds ratios [OR], 8.81; 95% confidence interval [CI], 1.38-56.29; P = .0214). However, the BRCA mutation rate did not differ significantly between patients who had ≥2 relatives with OC versus patients who had a single relative with OC (P = .127). A BRCAPRO mutation probability ≥10% was associated with a 6-fold increase in the likelihood of identifying a mutation (OR, 6.37; 95% CI, 2.23-18.22; P = .0005).
Table 4. Multivariate Logistic Regression Model for Breast Cancer-Susceptibility Gene Mutation Status
Abbreviations: BC, breast cancer; BRCAPRO, mathematical model for calculating the probability that a particular family member carries a germline breast cancer-susceptibility gene (BRCA) mutation; CI, confidence interval; DX, diagnosis; OC, ovarian cancer; OR, odds ratio.
No. of relatives with OC DX
≥2 (N = 7) vs 1 (N = 27)
≥2 (N = 7) vs 0 (N = 93)
≥10% (N = 25) vs <10% (N = 73)
Patients underwent definitive surgery either before (n = 30) or after (n = 96) genetic testing. The surgical intervention was breast-conserving surgery (BCS) for 31% of patients (n = 36), unilateral mastectomy for 34% (n = 40), and bilateral mastectomy for 35% (n = 41) (Table 5). Of 41 patients who underwent bilateral mastectomy, 2 patients underwent the surgery for bilateral DCIS, and 39 patients underwent the surgery for prophylactic reasons. Sixteen of 28 carriers (57%) underwent contralateral prophylactic mastectomy, whereas only 23 of 85 noncarriers (27%) underwent such surgery (P = .0037). This difference remained significant for patients aged ≤40 years. Among those patients, 7 of 9 carriers (78%) and 11 of 33 noncarriers (33%) underwent prophylactic mastectomy (P = .025). In addition, patients who had genetic testing before surgery underwent prophylactic mastectomy more frequently that those who had genetic testing after surgery (53% vs 31%).
Table 5. Association Between Breast Cancer-Susceptibility Gene Positivity and Surgery
No. of Patients (%)
Aged ≤40 Years
Aged >40 Years
Abbreviations: BCS, breast-conserving surgery; BRCA, breast cancer-susceptibility gene.
Three patients had bilateral ductal carcinoma in situ but underwent only unilateral mastectomy. They were not included in the table for cross-tabulation with BRCA mutation status.
Two patients underwent bilateral mastectomy for bilateral ductal carcinoma in situ. They were not included in the table for cross-tabulation with BRCA mutation status.
At a median follow-up of 3.31 years (range, 0.07-30 years) for survivors, 5% (n = 6) experienced an ipsilateral disease recurrence, 2 of which occurred in carriers. Of those 6 patients, 2 patients developed a chest wall recurrence after mastectomy, 2 reported an ipsilateral DCIS recurrence after BCS, and another 2 developed an ipsilateral IBC recurrence after BCS. BRCA status was not associated with an increased risk of recurrence (P = .77). The five-year RFS estimate for all patients was 93% (95% CI, 86%-100%) (Fig. 1). None of the patients in the series who had pure DCIS developed distant metastases or died of disease.
Our data indicate an overall 27% prevalence of deleterious BRCA1/BRCA2 mutations in high-risk women diagnosed with DCIS, supporting the presence of an in situ phase of carcinogenesis in the development of at least some BRCA-associated breast cancers.17 Our findings also suggest that the presence of at least 2 family members diagnosed with OC and a BRCAPRO mutation probability ≥10% are associated independently with higher BRCA1/BRCA2 mutation rates.
Previous studies have assessed the prevalence rates of BRCA1/BRCA2 mutations in patients with in situ carcinomas, with reported rates of up to 13%.11, 14, 19, 24 One of those studies evaluated the mutation rate in 10,000 consecutive patients who were referred for genetic testing to Myriad Genetic Laboratories. DCIS versus invasive status information was obtained from accompanying request forms.11 The prevalence of BRCA mutations was 13% in women who had DCIS diagnosed before age 50 years versus 24% in women who had IBC. Subsequently, Claus et al14 reported a BRCA1/BRCA2 mutation prevalence of 3.3% (12 of 369 patients) in a population-based sample of women with DCIS from the Connecticut Tumor Registry. In another study,24 deleterious mutations were identified in 10 of 79 women (12.7%) with DCIS who were referred for hereditary cancer risk assessment, similar to the frequency (14%) in IBC probands and slightly less than the 17% prevalence observed among women with IBC diagnosed before age 50 years. Recently, Hall et al19 conducted a cross-sectional analysis of the Myriad Genetics BRCA1/BRCA2 database and reported an overall 5.9% prevalence of BRCA1/BRCA2 mutations in non-AJ patients with carcinoma in situ (CIS) (ductal or lobular). In the absence of any FH of BC or OC at any age, the mutation prevalence in individuals with CIS and no personal history of IBC was 4.6% compared with 10.3% in patients who reported a personal history of IBC. Similar to our findings, patients with CIS had BRCA2 mutations (72.2%) more frequently than BRCA1 mutations (27.8%).
It is of considerable interest that we observed the highest incidence of BRCA mutations reported to date in a population of women with pure DCIS, most of whom (62%) had a BRCAPRO mutation probability <10%. These rates are comparable to the 28% prevalence observed in male BC11 and with the 24% associated with the BRCA2 999del5 mutation in Icelandic women diagnosed with IBC before age 40 years.25 The relatively higher prevalence estimates in the current study can be explained by the highly select nature of our patient population or because ours is a tertiary cancer care center, which automatically will have a selection bias. BRCA mutations may have been more prevalent in the clinic-based ascertainment because of increased BC screening in women. Alternatively, a heterogeneous clinical referral population may reflect the magnitude of the BRCA1/BRCA2 prevalence estimates reported in each study.
In our study cohort of women with pure DCIS who were referred for genetic risk assessment, we identified the predictive factors for BRCA1/BRCA2 mutations. Multivariate analysis revealed that ≥2 family members with OC (OR, 8.81) and a BRCAPRO mutation probability ≥10% (OR, 6.37) were strongly predictive of mutation status. Several studies11, 14, 19, 24 have identified an FH of OC and early onset BC as risk factors for BRCA mutations among DCIS probands. In our study, the association between FH of BC and BRCA mutation failed to reach statistical significance. This discrepancy may be explained in part by the misclassification bias resulting from errors in patient recall, because FH was self-reported and was not confirmed in other studies.11, 14, 19, 24
Hall et al19 determined that women who had early onset DCIS had a significantly increased risk of a BRCA1/BRCA2 mutation compared with women who had late-onset disease (aged ≥50 years; OR, 1.5; 95% CI, 1.1-2.1). This association was higher in women with very early onset disease (age <40 years vs ≥40 years; OR, 1.8; 95% CI, 1.3-2.3). Conversely, Smith et al24 observed similar mutation rates between patients with DCIS who were diagnosed at age ≤50 years and with those who were diagnosed at a later age. In our study, proband age at diagnosis was not predictive of mutation status, regardless of the calculated BRCAPRO estimate. Also, confirming the previous studies,24, 26 AJ ethnicity was not associated with an increased rate of mutations; however the small sample size and lack of age-matched controls limit the ability to reach a definitive conclusion.
The likelihood of identifying a BRCA mutation often is calculated using the BRCAPRO model. Yet the current BRCAPRO program does not account for DCIS as a risk factor in the likelihood of detecting a BRCA1/BRCA2 mutation. Although it was not intended for or validated in the DCIS population, our study supports the finding that further refinements in BRCAPRO may allow it to be used in this setting. Conversely, our findings also suggest that the BRCAPRO model may underestimate the relative contribution DCIS had on the likelihood of detecting a BRCA1/BRCA2 mutation, because the model predicted only 54% of the mutations that were actually identified in the study cohort. It is noteworthy that, in patients whose calculated BRCAPRO was <10%, the rate of positive mutations was as high as 18%; therefore, prospective studies are needed to determine the accuracy of the current BRCAPRO model for women with DCIS.
Despite the similarities between our study and the 2 clinic-based studies, our data are strengthened by the larger size of our population of 118 pure DCIS patients from a single institution, compared with smaller sample sizes in the Hwang et al15 (n = 20) and Smith et al24 (n = 79) studies. Nonetheless, our results must be interpreted in light of the limitations of the study. First, the small sample size in our study may have prevented statistically significant associations from emerging. Second, confidence in the conclusions is weakened by the possibility of selection bias, because BRCA genetic testing among patients with BRCAPRO mutation probability <10% can only be made through self-referral, and women who self-referred are likely to do so based on FH. Thus, the study group may not be a fair representation of the general population.
In conclusion, referral for BRCA genetic testing should be considered not only in women with IBC or OC but also in women with DCIS regardless of age at diagnosis, particularly when other personal or familial risk factors for BRCA mutations are present. Future studies are needed to establish criteria for BRCA mutation testing and to validate the proper incorporation of DCIS into risk models that calculate the probability of carrying a mutation. How this genetic knowledge, if known concurrent with a DCIS, would impact on optimal patient management, remains to be defined.
This work was funded in party by the Lynn Cohen Breast and Ovarian Cancer Project, Texas Business Women Funding, and the Nelly B. Connally Breast Cancer Research Fund.