Pathologic characteristics of second breast cancers after breast conservation for ductal carcinoma in situ
The number of women diagnosed with ductal carcinoma in situ (DCIS) is increasing. Although many eventually develop a second breast cancer (SBC), little is known about the characteristics of SBCs. The authors described the characteristics of SBC and examined associations between the pathologic features of SBC and index DCIS cases.
Women were identified in the National Comprehensive Cancer Network Outcomes Database who were diagnosed with DCIS from 1997 to 2008 and underwent lumpectomy and who subsequently developed SBC (including DCIS or invasive disease that occurred in the ipsilateral or contralateral breast). The Fisher exact test and the Spearman test were used to examine associations between the pathologic characteristics of SBC and index DCIS cases.
Among 2636 women who underwent lumpectomy for DCIS, 150 (5.7%) experienced an SBC after a median of 55.5 months of follow-up. Of these 150 women, 105 (70%) received adjuvant radiotherapy, and 50 (33.3%) received tamoxifen for their index DCIS. SBCs were ipsilateral in 54.7% of women and invasive in 50.7% of women. Among the index DCIS cases, 60.6% were estrogen receptor (ER)-positive, and 54% were high grade, whereas 77.5% of SBCs were ER-positive, and 48.2% were high grade. Tumor grade (P = .003) and ER status (P = .02) were associated significantly between index DCIS and SBC, whereas tumor size was not (P = .87).
After breast conservation for DCIS, SBC in either breast exhibited pathologic characteristics similar to the index DCIS, suggesting that women with DCIS may be at risk for developing subsequent breast cancers of a similar phenotype. Cancer 2012. © 2012 American Cancer Society.
With the growing adoption of screening mammography, the incidence of ductal carcinoma in situ (DCIS) has increased markedly over the past 25 years.1-4 Currently, DCIS accounts for approximately 25% of all newly diagnosed breast cancers in the United States1-5 and 50% of the malignancies identified by mammography.6-9 It is estimated that greater than 1 million women will be living with a diagnosis of DCIS in the United States by the year 2020. The majority of women with DCIS in the United States today undergo breast-conserving surgery.1
Women with a history of DCIS are at risk for developing second breast cancer (SBC), including DCIS or invasive disease that occurs in the ipsilateral or contralateral breast. SBC may represent regrowth of the initial lesion (ie, recurrence) or development of a new primary tumor in either breast. Clinical trials and retrospective series have demonstrated that, after lumpectomy for DCIS, the risk of an ipsilateral breast tumor recurrence ranges from 2% to 20% at 5 years.10-14 This risk is higher than that observed in the general population. One study has suggested that a clonal relation may exist between index DCIS and second, ipsilateral DCIS events.15 Women with DCIS also have a 3% to 6% risk of developing a contralateral breast cancer 5 to 10 years after their initial DCIS diagnosis.10, 11, 16-18 Previous studies suggest that, among women with a history of DCIS, approximately 50% of ipsilateral breast cancers10-12, 19-22 and >50% of contralateral breast cancers are invasive.
Although several studies have examined risk factors for SBC after lumpectomy for DCIS, there are minimal data on the pathologic characteristics of SBC events, such as tumor size, tumor grade, and hormone receptor status. One prior study examined the pathologic characteristics of SBCs compared with the index DCIS,23 but that study included only ipsilateral SBC events and was in the context of a clinical trial, thereby limiting the ability to generalize results.24 Despite the large number of patients at risk, little is known about the pathologic characteristics of SBC events among women who undergo lumpectomy for DCIS as part of routine care.
Our objectives were to characterize the pathologic features of SBC events among women who underwent lumpectomy for DCIS and to compare these features with those of the index DCIS. Examination of SBC events in both ipsilateral and contralateral breasts may inform our understanding of the behavior or biology of DCIS.
MATERIALS AND METHODS
The National Comprehensive Cancer Network (NCCN) Breast Cancer Outcomes Database was the source of data for this analysis. The Outcomes Database Project is a prospective registry of all patients who are treated at participating NCCN member institutions and includes chart abstraction and patient survey data. The institutional review boards at each center approved the protocol as well as data collection, transmission, and repository procedures. Chart reviews were conducted to identify SBC events; charts were analyzed by trained abstractors at 4 months, 9 months, and 18 months after index DCIS presentation and annually thereafter. Rigorous quality-assurance processes were in place to ensure the accuracy of data in the registry, including initial and follow-up data-management training, online edit checking during web-based data entry, programmed logic checks against the pooled data repository, routine quality-assurance reports to the NCCN centers for rectification by data managers, and annual on-site audits.
The study cohort consisted of women with DCIS who received some or all of their care at 1 of 7 institutions participating in the NCCN Breast Cancer Outcomes Database Project: The Ohio State University Comprehensive Cancer Center, City of Hope Cancer Center, Dana-Farber/Brigham & Women's Cancer Center and Massachusetts General Hospital Cancer Center, Fox Chase Cancer Center, The University of Texas M. D. Anderson Cancer Center, Roswell Park Cancer Institute, and University of Michigan Comprehensive Cancer Center. Patients were eligible for this analysis if they were diagnosed with DCIS between July 1, 1997 and December 31, 2008; if they underwent breast-conserving surgery, and not mastectomy, for their index DCIS; and if they had no prior history of breast cancer. SBC was defined as subsequent DCIS or invasive cancer in the ipsilateral or contralateral breast. For patients who had more than 1 breast cancer episode after the index DCIS event (n = 4), only the first episode was considered in the analysis. No patients experienced simultaneous, bilateral SBCs. Data collected from patients' medical records for this analysis included sociodemographic characteristics, index DCIS characteristics, therapies administered, and SBC characteristics. Follow-up was based on the date the patient last visited an NCCN institution.
Comorbidities at the time of presentation to the NCCN institution were classified and scored using either the Charlson Comorbidity Index25 (based on chart review) or the modified version of this index developed by Katz et al26 (based on a patient survey). Patients were considered to have received adjuvant tamoxifen if the drug was started on or after the index DCIS diagnosis, before any SBC, and within 365 days after first presentation to an NCCN center. No patient received an aromatase inhibitor. The same rules were applied when considering receipt of adjuvant radiation therapy.
Estrogen receptor (ER) and progesterone receptor (PR) status were analyzed separately and were considered either positive or negative based on immunohistochemical staining. Grade 1 or well differentiated tumors were defined as low grade, grade 2 or moderately differentiated tumors were defined as intermediate grade, and grade 3 or poorly differentiated tumors were defined as high grade. If a patient with invasive cancer had both histologic and nuclear grade recorded and they were not the same, then the higher value was used. Tumor size was measured in centimeters and was based on the pathologic specimen. Surgical margins were considered negative if there was no tumor within 2 mm of ink and were defined as close if they measured between 0 and 2 mm; in patients who underwent more than 1 excision, margin status was based on the final excision. Type of SBC was categorized as DCIS or invasive; tumors with invasive disease plus DCIS in the same specimen were recorded as invasive. The time to SBC was dichotomized at the median follow-up in months since index DCIS diagnosis in the analytic cohort. Tumor size and margin status for patients with DCIS were not collected before July 1, 1999; and hormone receptor status data for DCIS was not collected before April 2003. Variables were coded as “unknown” when there was not sufficient information about the data element in the medical record.
For categorical variables, the Fisher exact test was used to examine associations between pathologic characteristics of the index DCIS compared with SBC as well as associations between adjuvant treatments received and SBC characteristics. For tumor size, which was the 1 continuous variable, Spearman rank correlation was used to assess associations. All P values were 2-sided. In addition, we performed analyses according to breast laterality to test whether associations between index DCIS and SBC characteristics would be affected.
Patients who had ER status information available for both the index DCIS and the SBC (n = 51) contributed to the analyses regarding ER status. Patients who had PR status information available for both the index DCIS and the SBC (n = 47) contributed to the analyses regarding PR status. Patients who had grade information available for both the index DCIS and the SBC (n = 42) contributed to the analyses regarding tumor grade.
We identified 150 women who were diagnosed with DCIS between July 1, 1997 and December 31, 2008; who underwent breast-conserving surgery as definitive local therapy for their DCIS; and eventually developed an SBC. This represented 5.7% of the cohort who underwent breast-conserving surgery for DCIS during the study period. Among those who experienced an SBC event, the median follow-up after the index DCIS diagnosis was 55.5 months (range, 11.2-139.1 months), and the median age at diagnosis of the index DCIS was 53.5 years (range, 36.0-83.5 years). Eighty-two percent of patients were white, 57% were postmenopausal, and 72% had a comorbidity score of 0 (Table 1).
Table 1. Baseline Clinical Characteristics of Women who Experienced an Eventual Second Breast Cancer (N = 150)
|Age at diagnosis, y|| |
| Median [range]||53.5 [36.0-83.5]|
| <40||7 (4.7)|
| From 40 to <50||50 (33.3)|
| From 50 to <60||45 (30)|
| From 60 to <70||28 (18.7)|
| ≥70||20 (13.3)|
| White/non-Hispanic||123 (82)|
| African American/non-Hispanic||13 (8.7)|
| Asian/non-Hispanic||7 (4.7)|
| Hispanic||6 (4)|
| Other||1 (0.7)|
|Menopausal status|| |
| Premenopausal||65 (43.3)|
| Postmenopausal||85 (56.7)|
|Body mass index, kg/m2|| |
| <20||9 (6)|
| From 20 to <25||44 (29.3)|
| From 25 to <30||34 (22.7)|
| ≥30||38 (25.3)|
| Unknown||25 (16.7)|
| Screening mammography||127 (84.7)|
| Symptomatic||20 (13.3)|
| Unknown||3 (2)|
|Year of diagnosis|| |
| 1997-2000||45 (30)|
| 2001-2005||77 (51.3)|
| 2005-2008||28 (18.7)|
|Comorbidity score|| |
| 0||108 (72)|
| ≥1||42 (28)|
Histopathologic characteristics of the index DCIS are summarized in Table 2. The median tumor size was 1.0 cm (range, 0.1-6.0 cm), and 95% of patients had negative surgical margins. Fifty-four percent of tumors were high grade, 61% were ER-positive, and 58% were PR-positive. For the index DCIS, 24% of patients received no adjuvant radiotherapy or tamoxifen treatment, 43% had received radiotherapy without endocrine therapy, 27% had received radiotherapy and tamoxifen, and 6% had received tamoxifen without radiotherapy. The median duration of tamoxifen use was 26.9 months (range, 0.2-66.8 months). At the time of the current analysis, 11 patients still were receiving tamoxifen, and the reason for stopping tamoxifen among the 39 other patients was not available.
Table 2. Histopathologic Characteristics of Index Ductal Carcinoma In Situ Among Women who Experienced an Eventual Second Breast Cancer (N = 150)
|Tumor size, cma|| |
| Median [range]||1.0 [0.1-6.0]|
| <1||42 (28)|
| From 1 to <2||24 (16)|
| From 2 to <3||13 (8.7)|
| ≥3||12 (8)|
| Unknown||59 (39.3)|
|Margin distance from tumora|| |
| Involved||6 (4)|
| Close, <2 mm||1 (0.7)|
| Negative, ≥ 2 mm||132 (88)|
| Unknown||11 (7.3)|
| High||27 (18)|
| Intermediate||16 (10.7)|
| Low||7 (4.7)|
| Unknown||100 (66.7)|
|ER statusb|| |
| Positive||40 (26.7)|
| Negative||26 (17.3)|
| Unknown||84 (56)|
|PR statusb|| |
| Positive||36 (24)|
| Negative||26 (17.3)|
| Unknown||88 (58.7)|
| Comedo||56 (37.3)|
| Noncomedo||93 (62)|
| Unknown||1 (0.7)|
| Yes||4 (2.7)|
| No||144 (96)|
| Unknown||2 (1.3)|
The median time to SBC was 33.2 months (range, 3.8-120.3 months) (Table 3). SBCs were similarly likely to be invasive (51%; 95% confidence interval, 42%-59%) versus noninvasive (49%; 95% confidence interval, 41%-58%) and were similarly likely to occur in the ipsilateral breast (55%; 95% confidence interval, 46%-63%) versus the contralateral breast (45%; 95% confidence interval, 37%-54%). Of the invasive SBCs, tumor size was <2 cm in 82% of patients, and 78% were lymph node-negative.
Table 3. Clinical Characteristics of Second Breast Cancers (N = 150)
|Patient follow-up since index DCIS diagnosis: Median [range], mo||55.5 [11.2-139.1]|
|Time to SBC: Median [range], mo||33.2 [3.8-120.3]|
| Ipsilateral||36.3 [7.9-120.3]|
| Contralateral||28.5 [3.8-101.5]|
|Presentation of SBC|| |
| Screening mammography||93 (62)|
| Symptomatic||22 (14.7)|
| Unknown||35 (23.3)|
|Laterality of SBC|| |
| Ipsilateral||82 (54.7)|
| Contralateral||68 (45.3)|
|Type of SBC|| |
| Ipsilateral|| |
| DCIS||48 (58.5)|
| Invasive||34 (41.5)|
| Contralateral|| |
| DCIS||26 (38.2)|
| Invasive||42 (61.8)|
|SBC according to index DCIS management [no. ipsilateral/contralateral]|| |
| Lumpectomy alone||36 (24) [26/10]|
| Lumpectomy + RT||64 (42.7) [36/28]|
| Lumpectomy + RT + tamoxifen||41 (27.3) [17/24]|
| Lumpectomy + tamoxifen||9 (6) [3/6]|
The median time to SBC was 36.3 months for ipsilateral tumors versus 28.5 months for contralateral tumors. A majority of ipsilateral SBC events were noninvasive (59%), whereas a majority of contralateral SBC events were invasive (62%). SBCs were high grade in 48% of patients, ER-positive in 77% of patients, and PR-positive in 59% of patients (Table 4).
Table 4. Histopathologic Characteristics of Second Breast Cancers (N = 150)
|Tumor size, cma|| || |
| Median [range]||0.7 [0.1-6.5]||1.1 [0.1-14]|
| <1||25 (33.8)||25 (32.9)|
| From 1 to <2||6 (8.1)||31 (40.8)|
| From 2 to <3||2 (2.7)||8 (10.5)|
| ≥3||4 (5.4)||4 (5.3)|
| Unknown||37 (50)||8 (10.5)|
|Grade|| || |
| High||15 (20.3)||26 (34.2)|
| Intermediate||3 (4.1)||29 (38.1)|
| Low||2 (2.7)||10 (13.2)|
| Unknown||54 (73)||11 (14.5)|
|ER statusb|| || |
| Positive||33 (44.6)||53 (69.7)|
| Negative||7 (9.5)||18 (23.7)|
| Unknown||34 (45.9)||5 (6.6)|
|PR statusb|| || |
| Positive||26 (35.1)||39 (51.3)|
| Negative||14 (18.9)||32 (42.1)|
| Unknown||34 (45.9)||5 (6.6)|
|HER2/neuc|| || |
| Overexpressed||—||15 (19.7)|
| Not overexpressed||—||55 (72.4)|
| Unknown||—||6 (7.9)|
|Subtype|| || |
| Comedo||25 (33.8)||—|
| Noncomedo||38 (51.4)||—|
| Unknown||11 (14.9)||—|
|Histologic type|| || |
| Ductal||—||61 (80.3)|
| Lobular||—||9 (11.8)|
| Ductal and lobular||—||3 (3.9)|
| Tubular/colloid/papillary||—||3 (3.9)|
|EIC|| || |
| Positive||—||15 (19.7)|
| Negative|| ||61 (80.3)|
|LVI|| || |
| Positive||—||15 (19.7)|
| Negative||—||53 (69.7)|
| Unknown||—||8 (10.5)|
|No. of lymph nodes involved|| || |
| cN0||—||7 (9.2)|
| 0||—||49 (64.5)|
| 1-3||—||9 (11.8)|
| ≥4||—||7 (9.2)|
| Unknown||—||4 (5.3)|
Comparisons between index DCIS and SBC pathologic features are provided in Table 5. Among patients who had an ER-positive index DCIS, 82% had an ER-positive SBC; and, among patients who had an ER-negative index DCIS, 53% had an ER-negative SBC. Seventy-one percent of patients who had a high-grade index DCIS had a high-grade SBC, 50% of patients with an intermediate-grade index DCIS had an intermediate-grade SBC, and no patients with a low-grade index DCIS had a high-grade SBC.
Table 5. Comparison Between the Index Ductal Carcinoma In Situ and Second Breast Cancer Marker Expression or Tumor Grade
|ER status|| || || ||37/51 (72.5)||.02|
|Positive||28||6|| || || |
|Negative||8||9|| || || |
|PR status|| || || ||29/47 (61.7)||.23|
|Positive||20||10|| || || |
|Negative||8||9|| || || |
|Grade|| || || ||27/42 (64.3)|| |
|High||17||6||1|| || |
|Intermediate||6||7||1|| || |
ER status of the index DCIS was associated with ER status of the SBC (P = .02) (Table 5). This association for ER status was present even after stratification at the median age of diagnosis (53.5 years). Prior studies have demonstrated that approximately 70% of all breast cancers are ER-positive. Despite the baseline expectation that most SBCs would be ER-positive, we still observed a significantly greater number of hormone receptor-concordant tumors than would have been expected if the index DCIS and SBC events were independent (P = .001). In addition, tumor grade of the index DCIS was associated with grade of SBC (P = .003). There was no significant association between the index DCIS and the SBC with regard to PR status (P = .23) or tumor size (Spearman correlation = 0.02; P = .87); and, among the noninvasive SBCs, no significant association was observed with regard to DCIS subtype (comedo vs noncomedo; P = .07).
We performed analyses according to breast laterality to assess whether the pathologic associations varied according to side of the SBC (ipsilateral vs contralateral). Comparing ER status between the index DCIS and the SBC, there was a significant association among the 22 ipsilateral events (P = .002), but not the 29 contralateral events (P = .99). Comparing grade, there was a significant association among the 19 contralateral events (P = .01), but not the 23 ipsilateral events (P = .17). We considered analyzing the effects of radiotherapy and tamoxifen on pathologic associations but elected not to do so because of the small sample sizes in these subgroups.
Among the women in this study who underwent lumpectomy for DCIS and subsequently developed SBC, we observed significant associations for tumor grade and ER status between the index DCIS and the SBC event. Furthermore, we observed that developing an ipsilateral SBC was not significantly more likely than developing a contralateral SBC (55% vs 45%) and that developing invasive SBC was not significantly more likely than developing noninvasive SBC (49% vs 51%).
In a prior report that examined associations between the pathologic characteristics of index DCIS and SBC, concordant grade was observed in 62% of tumors, and concordant hormone receptor expression was observed in 63% of tumors,23 but that investigation focused only on the correlation between index DCIS and ipsilateral SBC events. In addition, a previous study using comparative genomic hybridization data suggested a clonal relation between index DCIS and second ipsilateral DCIS events.15 Our data, which included events in both breasts, reflected similar overall concordance rates for grade and ER status. The findings that pathologic associations between index DCIS and SBC events exist, and that ipsilateral SBC events are more alike than different from contralateral SBC events, raise the intriguing possibility that both DCIS characteristics and patient traits influence the type of breast cancer patients develop. Specifically, women with DCIS may be more likely to experience SBC with similar pathologic features in either breast.
Because the features of SBC are clinically relevant, our data may provide useful information about the potential type of SBC that may develop if a woman were to experience recurrence after undergoing lumpectomy for DCIS. Moreover, this information may influence her/her physician's decision-making with regard to adjuvant therapy for the index DCIS. For example, if a woman who is treated for an ER-positive DCIS were to develop an SBC, then it is most likely to be ER-positive again. The knowledge that an SBC event would be likely to have favorable prognostic features may make some clinicians and patients more comfortable with a decision to forgo adjuvant tamoxifen or radiotherapy for an index DCIS. Because patterns-of-care data indicate substantial heterogeneity in the management of DCIS, with many patients not receiving adjuvant tamoxifen or radiotherapy after lumpectomy,27-33 our data may help clinicians and patients predict the features of a potential SBC and thus better inform their decisions regarding index DCIS as they weigh the risks of possible “over treatment” of an index DCIS with the risks of SBC.
Several earlier studies evaluated risk factors for the development of SBC. Some have suggested that selected women with DCIS who undergo lumpectomy without radiotherapy have low rates of ipsilateral SBC,34-37 particularly those who are older and have small, low-grade primary cancers and negative surgical margins.19, 38-41 Others have revealed high rates of SBC even among highly selected patients who underwent lumpectomy without receiving radiotherapy.42 Regardless, these and other studies have raised questions about who should versus should not receive adjuvant therapy (radiotherapy or tamoxifen) after undergoing lumpectomy for DCIS. Additional research is warranted, and our findings suggest that future clinical trials investigating when to omit adjuvant therapy after lumpectomy for DCIS should include ER status as a stratification factor in addition to tumor grade; this has not been the case in prior trials.37, 42
Among women from the NCCN Breast Cancer Outcomes Database who underwent lumpectomy for DCIS, 2.6% developed a contralateral SBC and 3.1% developed an ipsilateral SBC at a median of 4.6 years. Previous studies evaluating recurrence after DCIS have yielded conflicting results. After 8 years of follow-up in the National Surgical Adjuvant Breast and Bowel Project trial,16 there was a substantially higher proportion of ipsilateral SBC versus contralateral SBC as a first event after initial treatment for DCIS (19% ipsilateral vs 4% contralateral). Another study revealed a similar risk of ipsilateral SBC versus contralateral SBC after initial DCIS but only assessed invasive SBC events.18 Lower rates of ipsilateral SBC in our data may reflect improvements in breast imaging that have made it possible to detect smaller foci of DCIS upfront and to better delineate the lumpectomy cavity for radiotherapy planning; use of a radiotherapy boost, which was not used in the randomized trials from the 1980s and 1990s that evaluated breast-conserving treatment for DCIS10-12, 16; or possibly higher rates of guideline-concordant care delivered at NCCN centers. Other modern series also have demonstrated very low rates of ipsilateral SBC after breast-conserving treatment for DCIS with similar follow-up.43 Clearly, the ratio of ipsilateral to contralateral SBC will be influenced by the use of lumpectomy, radiotherapy, tamoxifen, and contralateral prophylactic mastectomy at the time of the index DCIS diagnosis.
A proportion of patients in our current series had missing pathologic characteristics. Causes of missing data include the following: Tumor size and margin status were not collected before July 1999, and hormone receptor status was not collected before April 2003. If the patterns of missing data are not random, then this could lead to overestimates or underestimates of the strength of associations between index DCIS and SBC pathologic features. However we do not have evidence that missing pathologic data were related to characteristics of the patients; thus, the observed associations should be unaffected. In addition, the overall associations we observed for tumor grade and ER status between index DCIS and SBCs were modified when analyzed by breast laterality; however, given the small sample sizes of the subgroups, we have refrained from drawing general conclusions from those results. Also, with our median follow-up of 55.5 months, it is possible that the SBC events we observed may have had different characteristics compared with potential, later SBC events; thus, longer follow-up is required to confirm our findings. Finally, patients who underwent management at an NCCN institution for an index DCIS may not have returned to the same NCCN institution for diagnosis or management of their SBC, which could lead to the under-reporting of SBC event rates. Nonetheless, we have no reason to believe that DCIS characteristics would lead to differential patient follow-up at these institutions; therefore, the pathologic associations we observed should remain valid even if there was under-reporting of the true incidence of SBC.
In conclusion, we have described the clinical and pathologic characteristics of SBC after breast conservation for DCIS. SBC occurred with almost equal frequency in the ipsilateral and contralateral breast, and the grade and ER status of the index DCIS and SBC events were significantly associated. These results suggesting that women with DCIS may be at risk of developing SBCs of a similar phenotype may have the potential to inform clinical decision-making at the time of the index DCIS diagnosis. Further work will be required to confirm our findings and define additional characteristics that may help guide optimal treatment recommendations for patients with DCIS.
This analysis was supported by contract HHSA290200500161 from the Agency for Healthcare Research and Quality. Collection of the data was supported in part by grant P50 CA89393 from the National Cancer Institute to Dana-Farber Cancer Institute. Dr. Hassett received salary support from an American Society of Clinical Oncology Career Development Award and a Susan G. Komen for the Cure Career Catalyst Award.
Note Added in Proof
CONFLICT OF INTEREST DISCLOSURES
Dr. Laronga has received honoraria from Genomic Health (Speaker's Bureau).