Presented in part at the 45th annual meeting of the American Society for Therapeutic Radiology and Oncology, Salt Lake City, Utah, October 19–23, 2003.
Ductal carcinoma in situ (DCIS) is detected most commonly on routine screening mammography in the asymptomatic patient, and has a long natural history. The objective of the current study was to determine the long-term outcome after breast-conservation surgery followed by definitive breast irradiation for women with mammographically detected DCIS of the breast.
In total, 1003 women with unilateral, mammographically detected DCIS of the breast underwent breast-conserving surgery followed by definitive breast irradiation. These women were treated in 10 institutions in North America and Europe. The median follow-up was 8.5 years (mean, 9.0 years; range, 0.2–24.6 years).
The 15-year overall survival rate was 89%, and the 15-year cause-specific survival rate was 98%. The 15-year rate of freedom from distant metastases was 97%. In total, there were 100 local failures (10%) in the treated breast. The 15-year rate of any local failure was 19%, and the 15-year rate of local only first failure was 16%. Patient age ≥ 50 years at the time of treatment and negative final pathology margins from the primary tumor excision both were associated independently with a lower risk of local failure in univariate analysis (P = 0.00062 and P = 0.024, respectively) and in multivariate analysis (P = 0.00057 and P = 0.0026, respectively). For favorable subgroups of patients age ≥ 50 years or with negative resection margins, the 10-year risk of local failure was ≤ 8%.
Ductal carcinoma in situ (DCIS) (intraductal carcinoma) of the breast is detected increasingly in the asymptomatic patient on routine screening mammography. Surveillance, Epidemiology, and End Results (SEER) data document a dramatic increase in the detection of DCIS associated with mammographic screening.1 Such mammographically detected DCIS lesions generally are small and have a long natural history. Therefore, long-term outcome studies are required to evaluate the effects of treatment.
Breast-conservation using excision (or so-called lumpectomy) followed by radiation treatment is an accepted local management strategy for patients with small, mammographically detected DCIS of the breast. Three randomized studies have compared lumpectomy with lumpectomy plus radiation treatment.2–7 All three studies demonstrated that the risk of local recurrence in the ipsilateral, treated breast was reduced by approximately half with the addition of radiation after lumpectomy.
Little information has been published concerning long-term outcome after treatment of mammographically detected DCIS in routine clinical practice. Therefore, the current study was performed to evaluate the long-term outcome for women with mammographically detected DCIS of the breast who underwent breast-conserving surgery followed by definitive breast irradiation. To perform this study, patient data were combined from 10 institutions in 4 countries in North America and Europe.
MATERIALS AND METHODS
In this article, we report the outcome of treatment for 1003 women with unilateral mammographically detected DCIS of the breast who underwent breast-conserving surgery followed by definitive breast irradiation. These patients were identified from a collaborative, multi-institutional database and were treated during 1973 to 1995 at 10 institutions in 4 countries in North America and Europe (Table 1).8–39 The current study has expanded the patient population substantially from prior collaborative studies by including patients who were treated through 1995 to increase the number of patients in the database and to evaluate the results of more recently treated patients.8–17 Data were obtained from a retrospective review of the treatment records from the patients who were treated at each institution. The selection of patients for treatment, the method of treatment, including the technical delivery of the radiation treatment, and the method of follow-up were sufficiently uniform among the institutions to allow combining these patients into a single data set for analysis.
Table 1. Distribution of Patients According to the Treating Institution
No. of patients
The sum of the numbers in the column does not equal 100% because of rounding.
Institut Curie, Paris, France
Netherlands Cancer Institute, Amsterdam, The Netherlands
British Columbia Cancer Agency, Victoria, BC
Duke University, Durham, NC
Mallinckrodt Institute of Radiology, St. Louis, MO
University of Michigan, Ann Arbor, MI
University of Pennsylvania, Philadelphia, PA
University of Texas M. D. Anderson Cancer Center, Houston, TX
Criteria for the inclusion of patients in the current study were: 1) American Joint Committee on Cancer TisN0M0 DCIS of the breast at presentation40, 41; 2) clinically occult, mammographically detected DCIS; 3) no physical examination finding, such as a breast mass or bloody nipple discharge; 4) unilateral disease at presentation; 5) treatment with breast-conserving surgery followed by definitive breast irradiation; 6) a treatment date of 1995 or earlier; 7) definitive whole-breast irradiation to a dose ≥ 4000 centigrays (cGy); 8) no adjuvant systemic chemotherapy or hormonal treatment; 9) no Paget disease of the nipple; 10) no prior or concurrent invasive or microinvasive carcinoma of the ipsilateral or contralateral breast; and 11) no prior or concurrent malignancy other than DCIS, except for nonmelanoma skin cancer. Recognizing the wide variation in accepted radiation fractionation regimens, patients who were treated with whole-breast radiation ≥ 4000 cGy were included in the current study, whereas patients who were treated with < 4000 cGy were considered to have received an inadequate radiation dose and were excluded from the study.
All women underwent breast-conserving surgery followed by definitive breast irradiation. The surgical treatment included complete gross excision of the primary tumor site, in general, using mammographic needle localization. Reexcision of the primary tumor site was performed for 470 patients (47%). Reexcision was performed most commonly in an attempt to obtain negative resection margins or for residual microcalcifications seen on postbiopsy mammogram.
Pathologic axillary lymph node staging was performed for 210 patients (21%), and all of those patients had pathologically negative lymph nodes. The remaining 793 patients (79%) had clinically negative axillae and did not undergo pathologic axillary lymph node staging.
Radiation treatment was delivered with definitive intent according to the policy of the institution at which each patient was treated. The median whole breast dose was 5000 cGy (mean, 4914 cGy; range, 4125–6468 cGy). A boost to the primary tumor site was delivered after radiation treatment to the whole breast in 722 patients (72%). The boost dose was delivered using electrons in 573 patients, photons in 76 patients, and an implant in 73 patients. The total dose delivered to the primary tumor site was defined as the sum of the whole-breast radiation dose plus the boost dose. The median total dose was 6000 cGy (mean, 5842 cGy; range, 4125–8300 cGy). The total dose delivered was ≥ 5000 cGy in 928 of 1003 patients (93%) and ≥ 6000 cGy in 607 of 1003 patients (61%). Among the 75 patients who received a total dose < 5000 cGy, 62 patients received a total dose of 4125–4400 cGy, and the remaining 13 patients received a total dose of 4500–4990 cGy. All 62 patients who received a total dose of 4125–4400 cGy were treated to the whole breast using larger daily fractions (263–275 cGy) without a breast boost. Regional lymph node radiation was delivered to ≥ 1 regional site(s) in 45 patients (4%), predominantly early in the study period.
The Kaplan–Meier method was used to calculate actuarial curves for survival, freedom from distant metastases, local control, and contralateral breast carcinoma.42 The period measured was calculated from the start of definitive breast irradiation, not at the time of diagnosis of DCIS. The log-rank test was used for statistical comparisons between groups.43 A multivariate Cox proportional hazards regression model was used to evaluate the independent prognostic significance of the variables.44 Excluded from the model were potential prognostic variables for which a large fraction of the patients had incomplete information.
A local failure was scored for a failure that occurred within the treated breast. All local failures, including the first and subsequent events and including DCIS and invasive local failures, were included in the calculation of any local failure. Local only first failure was defined as a local failure that occurred in the breast as the first and only site of failure without any other prior or concurrent event (i.e., regional failure, distant failure, contralateral breast carcinoma, or second malignant neoplasm). The location of the local failure was scored according to the method of Recht et al.45 A regional failure was scored for a failure that occurred within the ipsilateral axillary, supraclavicular, infraclavicular, and/or internal mammary lymph node region(s). For analysis of the development of a contralateral breast carcinoma, both invasive carcinoma and DCIS were scored as events.
For overall survival, a patient who died from any cause was scored as a failure. For cause-specific survival, a patient was required to be dead from breast carcinoma to be scored as a failure. For analysis of freedom from distant metastases, a patient was scored as a failure at the time of first evidence of distant metastatic disease. For analysis of cause-specific survival and freedom from distant metastases, a patient who developed a nonbreast second malignant neoplasm (other than nonmelanoma skin cancer) was censored at the time of occurrence of the second malignant neoplasm.
The median follow-up for all patients was 8.5 years (mean, 9.0 years; range, 0.2–24.6 years). The median follow-up for surviving patients was 8.6 years (mean, 9.1 years; range, 0.2–24.6 years). The numbers of patients who were alive and available for analysis at 5 years, 10 years, and 15 years were 873 patients, 363 patients, and 68 patients, respectively.
The final pathology margin status from the primary tumor excision was determined from a retrospective review of the patient records. Pathology margins were scored as negative, positive, or close according to the policy of each participating institution. The final pathology margin was defined as the pathology margin from the reexcision, if performed, or the pathology margin from the primary surgical excision, if a reexcision had not been performed. The final pathology margin was considered negative when no tumor was identified in the reexcision specimen. The final pathology margin was scored by all participating institutions as positive when tumor was identified at the inked margin. Eight of the 10 participating institutions used 2 mm to differentiate between negative margins (> 2 mm or ≥ 2 mm) and close margins (≤ 2 mm or < 2 mm). One institution used 2–3 mm for this differentiation, and 1 institution used 3 mm. Routine inking of breast biopsy specimens was not performed commonly during the earlier years of this study.
Table 2 shows the patient and tumor characteristics. The median patient age at presentation was 53 years (mean, 54 years; range, 26–86 years). The predominant mammographic finding at presentation was microcalcifications alone (84%). Pathologically confirmed negative margins of excision were found in 599 patients (60%).
Table 2. Patient and Tumor Characteristics
No. of patients
Mammogram findings at presentation
Microcalcifications and mass
Clinical tumor size
≤ 2.0 cm
> 5.0 cm
Final pathology margin
Pathologic lymph node status
Table 3 shows 5-year, 10-year, and 15-year outcome data for survival, freedom from distant metastases, local failure, and contralateral breast carcinoma. The curves for overall survival and cause-specific survival are shown in Figure 1, and the curves for local failure are shown in Figure 2.
Table 3. Outcome Data
Percent (95% CI)
At 5 yrs
At 10 yrs
At 15 yrs
95% CI: 95% confidence interval.
Includes the one patient who had angiosarcoma as local failure.
Excludes two patients with unknown information.
Includes one patient with Paget disease plus associated ductal carcinoma in situ.
Excludes 11 patients with unknown information.
Includes any contralateral breast carcinoma (first event or other event).
Local failure (first event or other event) developed in 100 patients (10%). The median time to local failure was 5.3 years (mean, 6.1 years; range, 0.6–17.7 years). The median time to invasive local failure was 5.9 years (mean, 6.7 years; range, 1.0–17.7 years). The median time to DCIS local failure was 4.5 years (mean, 5.2 years; range, 0.6–11.7 years).
Of the 100 patients who had local failures, isolated local failure in the treated breast was the first event in 85 patients, and local-regional failure was the first event in 5 patients (Table 4). The most common types of first events after treatment were local only failure, contralateral breast carcinoma, and second malignant neoplasm. Of the 1003 patients, 15 patients developed metastatic disease as follows: 1) after an invasive local recurrence in 3 patients; 2) after local-regional recurrence in 4 patients; 3) after a contralateral invasive breast carcinoma in 2 patients; 4) concurrent with a contralateral invasive breast carcinoma in 2 patients; 5) concurrent with a second malignant neoplasm in 2 patients; 6) after a second malignant neoplasm in 1 patient; and 7) as the first event in 1 patient.
Table 4. Type of First Event
Type of first event
No. of patients
DCIS: ductal carcinoma in situ; NOS: not otherwise specified.
All five patients with concurrent local (invasive carcinoma) and axillary lymph node failure.
Includes two patients with concurrent contralateral infiltrating ductal carcinoma of the breast and distant metastases and one patient with concurrent contralateral infiltrating ductal carcinoma of the breast and (ipsilateral) DCIS local recurrence.
Includes two patients with concurrent second malignant neoplasm and distant metastases.
Univariate analysis was performed to determine which potential clinical and pathologic parameters were associated with local control (Table 5). The major findings from this analysis were that patient age at the time of treatment and final pathology margin status from the primary tumor excision both were associated with local recurrence (P = 0.00062 and P = 0.024, respectively). Figure 3 shows local failure as a function of patient age at the time of treatment, and Figure 4 shows local failure as a function of final pathology margin status. The risk of developing contralateral breast carcinoma was not different when analyzed by age group (P = 0.63).
Table 5. Local Failure Correlated with Various Potential Prognostic Parameters on Univariate Analysis
A multivariate Cox proportional hazards regression model evaluated patient age at the time of treatment, final pathology margin status, mammographic findings, institution at which the patient was treated, date of treatment, location of the primary tumor, and total radiation dose as potentially significant prognostic factors. After adjusting for other prognostic variables in the model, both patient age and final pathology margin status were associated independently with the risk of local recurrence (P = 0.00057 and P = 0.0026, respectively). Compared with patients who had negative final pathology margins (hazard ratio defined as 1.00), a statistically significant increase in local recurrence was found for patients who had positive margins (hazard ratio, 3.35; P = 0.00035) and close margins (hazard ratio, 1.90; P = 0.027). Compared with patient age ≤ 39 years (hazard ratio defined as 1.00), a statistically significant decrease in local recurrence was found for patients age 50–59 years (hazard ratio, 0.36; P = 0.0056) and patients age ≥ 60 years (hazard ratio, 0.23; P = 0.00026). All other variables were not associated significantly with local recurrence (all P ≥ 0.13), including the institution at which the patient was treated (P = 0.66).
The risk of local failure was evaluated according to the combination of patient age and final pathology margin status (Table 6). These data were calculated as crude incidences, because there were too few patients at risk in many of the subsets to determine actuarial outcomes.
Table 6. Crude Incidence of Local Failure According to Patient Age and Final Pathology Margin Status
The current study has generated a number of important observations related to the long-term outcome for patients with DCIS of the breast who undergo breast-conservation surgery followed by definitive breast irradiation. First, the 5-year, 10-year, and 15-year outcomes were reported after breast conservation with radiation treatment (Table 3) (Figs. 1 and 2). Second, in this study, we found that patient age at initial treatment and final pathology margin status were associated strongly with local failure (Table 5) (Figs. 3 and 4). Third, a detailed analysis was presented for the types of outcome events after breast-conservation treatment (Table 4).
To our knowledge, the results from the current study (Table 3) (Figs. 1 and 2) represent the largest study with 15-year outcome data for patients with DCIS who underwent breast-conservation surgery with radiation. The 15-year overall survival rate of 89% and the cause-specific survival rate of 98% demonstrate that very few patients die from breast carcinoma-related causes and that patients are more likely to die from causes other than breast carcinoma. Therefore, follow-up for patients with DCIS should emphasize long-term health maintenance strategies, as would be done for the general population, in addition to oncologic strategies to reduce the risk of disease recurrence. The high overall and cause-specific survival rates observed in the current study were similar to the rates reported in other studies, although no more than 10-year outcomes generally were reported.2, 3, 5, 26, 28, 30, 34, 38, 39
Patient age ≥ 50 years at the time of treatment and negative final pathology margins were associated with lower rates of local recurrence (Table 5) (Figs. 3 and 4). Similar results have been reported by other investigators.4, 6, 20, 23, 27, 28, 35, 37, 46 Younger patients (age ≤ 49 years) with negative resection margins appear to have an acceptable risk of local failure, although the data were based on small numbers of patients (Table 6). Therefore, obtaining negative resection margins appears to be particularly important for the younger patient with DCIS of the breast to minimize the risk of local failure.
Three randomized trials have demonstrated that the addition of radiation after breast-conservation surgery reduces the risk of ipsilateral local recurrence by about half.2–7 In the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-17 trial, the 12-year risk of local recurrence was reduced from 31.7% without radiation to 15.7% with radiation (P < 0.000005).3 The 4-year results from the European Organization for Research and Treatment of Cancer demonstrated a 16% rate of local recurrence without radiation, but the local recurrence rate was 9% with radiation (P = 0.005).5 In the United Kingdom, Australia, and New Zealand (UK/ANZ) DCIS trial, the crude incidence of local recurrence was reduced from 14% without radiation to 6% with radiation (P < 0.0001).7 However, the addition of radiation has not been associated with an improvement in overall survival, although longer follow-up may be required to evaluate this endpoint.2, 3, 5
A similar risk reduction associated with the addition of radiation treatment after lumpectomy was found in retrospective studies.46, 47 Although retrospective studies have suggested favorable subgroups of patients with a low risk of local recurrence after lumpectomy without radiation, prospective studies have not reproducibly identified those patients for whom radiation can be omitted safely.48
The most common first event after treatment in the current study was a local failure, a contralateral breast carcinoma, or a second malignant neoplasm (Table 4). This pattern of first events is similar to other reported data.2 Contralateral breast carcinoma was the first event in 71 patients (7%) (Table 4), and the development of distant metastatic disease was associated with contralateral invasive breast carcinoma in 4 patients. These results have important practical implications. Long-term bilateral mammographic screening after breast-conservation treatment is indicated, and adjuvant tamoxifen should be considered at the time of initial presentation with DCIS.
The patients reported in the current study predate the era of adjuvant tamoxifen for DCIS; therefore, current patients who are treated with tamoxifen for DCIS of the breast may have a lower local failure rate and a lower rate of developing a contralateral breast carcinoma than reported in the current study. Two randomized trials have studied the use of tamoxifen with differing results.3, 7, 49 In the NSABP B-24 trial, the 7-year rate of ipsilateral local recurrence was reduced from 11.1% without tamoxifen to 7.7% with tamoxifen (P = 0.02), and the 7-year rate of contralateral breast carcinoma was reduced from 4.9% without tamoxifen to 2.3% with tamoxifen (P = 0.01).3 The rate of all breast carcinoma events was reduced from 16.9% to 10.0% with tamoxifen (P = 0.0003). Positive estrogen receptor status was predictive for the benefit of adjuvant tamoxifen.50 Houghton et al. reported the results from the UK/ANZ DCIS trial.7 Although the rates of ipsilateral local recurrence (15% vs. 13%, respectively; P = 0.42), contralateral breast carcinoma (3% vs. 1%, respectively; P = 0.07), and all breast carcinoma events (18% vs. 14%, respectively; P = 0.13) were reduced with the addition of tamoxifen, these reductions were not statistically significant. It has not been shown that the addition of tamoxifen improves overall survival.3, 7, 49
The findings in the current study have practical implications for the management of surgical margins of resection. Margins of resection are one of the few variables that can be controlled by the treating physician, for example, by the extent of surgical resection or by the use of a reexcision. The results of the current study showed that negative resection margins were associated strongly with a decreased risk of local recurrence (Table 5) (Fig. 4). Adequate surgical excision with reexcision, if indicated, is important to maximize local control, and obtaining clear margins of resection remains the objective of surgical excision (or reexcision) for patients who undergo breast-conservation treatment. For the patient who presents with microcalcifications on mammography, a postbiopsy mammogram is indicated to confirm the removal of all microcalcifications before beginning definitive breast irradiation. Although mammography remains the standard imaging modality for evaluating the extent of disease in the breast, alternative methods of breast imaging, such as breast magnetic resonance imaging, are under investigation.51
The minimum negative margin width for breast-conservation treatment has not been established.4, 52 When radiation is added after lumpectomy, 2 mm is used commonly to differentiate between negative margins and close margins, although some institutions have used other definitions, such as 1 mm or 3 mm. Most of the institutions that participated in the current study used 2 mm to make this differentiation. In contrast, a wider negative margin width, generally ≥ 10 mm, may be required when radiation is omitted after lumpectomy.53, 54 In a prospective, single-arm study of wide excision alone with negative excision margins ≥ 1 cm, Wong et al. reported that the 5-year rate of local failure was 12.5%, and that study was closed to accrual early because of this high rate of local failure.55
The institution at which the patient was treated was not associated with local failure (P = 0.73) (Table 5). Although there were differences in the treatment algorithms used among the institutions that participated in the current study, the lack of any difference in the local recurrence rates among the participating institutions suggests that breast-conservation surgery with radiation can be applied widely in routine clinical practice. A multidisciplinary management approach, including physicians from multiple specialties, was emphasized at all of the participating institutions.
Limitations of the current study include the lack of a standard definition for margin evaluation and the lack of a central pathology review. However, the definition of negative, positive, and close margins was similar, but not identical, among the participating institutions. Thus, the finding of an increased risk of local failure for patients without negative resection margins (Table 5) (Fig. 4) should be reliable, even in the absence of a central pathology review.
The current report presents the 5-year, 10-year, and 15-year outcomes for patients with mammographically detected DCIS of the breast who underwent breast-conserving surgery followed by definitive breast irradiation. A decreased risk of local failure was associated with patient age ≥ 50 years at the time of treatment and negative margins of resection. These results support the use of breast-conserving surgery and definitive breast irradiation for the treatment of appropriately selected patients with mammographically detected DCIS of the breast.