INTRODUCTION

Breast-conservation treatment (BCT) including definitive breast irradiation is well established as an alternative to mastectomy for the treatment of early-stage breast carcinoma. Randomized trials comparing BCT with mastectomy have demonstrated equivalent long-term survival.^1-4

For the patient undergoing evaluation for BCT, the standard primary imaging modality to evaluate the breast is mammography, with correlation ultrasound as indicated. However, magnetic resonance imaging (MRI) of the breast is increasingly used to evaluate potential candidates for BCT.^4-11 Multiple studies have demonstrated the ability of breast MRI to detect mammographically and clinically occult foci of cancer in the breast in approximately 25% to 30% of patients.^5,8-14 However, the clinical significance of additional disease detected using breast MRI remains uncertain.

The present study was performed to assess the potential value of integrating breast MRI into the clinical evaluation of women with newly diagnosed early-stage invasive breast carcinoma or ductal carcinoma in situ (DCIS; intraductal carcinoma).

PATIENTS AND METHODS

The study population consisted of a total of 756 women who underwent BCT including definitive breast irradiation at the Hospital of the University of Pennsylvania (Philadelphia, PA) during 1992 to 2001. The study population was limited to patients who met the following criteria: American Joint Commission on Cancer (AJCC) clinical and pathologic stage 0, I, or II disease (TisN0M0, T1N0M0, T2N0M0, T1N1M0, or T2N1M0) using the fifth edition of the AJCC staging manual¹⁵; unilateral disease at presentation; invasive breast carcinoma or DCIS of the breast; definitive locoregional management using BCT consisting of breast-conserving surgery followed by definitive breast irradiation; definitive breast irradiation at the Hospital of the University of Pennsylvania; definitive breast irradiation to a total dose of 60 Gy or more; if performed, breast MRI study performed before definitive breast irradiation; treatment dates of 1992 to 2001; no prior or concurrent malignancy (breast or other site), except for nonmelanoma skin cancer; and for the patients with invasive breast carcinoma, pathologic axillary lymph node staging performed.

Routine breast imaging included conventional mammography, with correlation ultrasound as indicated. Of the study population of 756 patients, 215 patients (28%) had also undergone a breast MRI study at the time of initial diagnosis and evaluation of breast cancer, and 541 patients (72%) had not undergone a breast MRI study. All of the breast MRI studies were performed in the Department of Radiology at the Hospital of the University of Pennsylvania.

For patients with early-stage breast cancer being evaluated for BCT at the Hospital of the University of Pennsylvania, breast MRI has been integrated into the work-up and evaluation for some of the patients with newly diagnosed invasive carcinoma or DCIS beginning in the early 1990s. Therefore, although recognizing the retrospective, nonrandomized nature of the present study, the dates for the present study were limited to patients treated beginning in 1992 so that the patients with and without a breast MRI study were treated during the same time period.

The technical method of performing breast MRI has been well described.^5,16-20 The clinical and radiologic findings were available to the radiologist at the time of breast MRI interpretation. For all patients, the breast MRI was considered a potential source of information to aid in their clinical management. For patients who required MRI-guided wire localization of a suspicious lesion identified on MRI, a proprietary MRI needle localization system was used.

Because of varying patient presentation, the breast MRI studies, when performed, were obtained at various points during the clinical course of patient management. Of the 215 patients who had undergone a breast MRI study, the sequences for obtaining the breast MRI studies were: (a) after a suspicious mammogram, but before tissue sampling for 58 patients (27%); (b) after a core biopsy or fine-needle aspiration cytology, but before excision for 49 patients (23%); (c) after an excision, but before re-excision for 80 patients (37%); (d) after an excision, with no re-excision performed for 14 patients (7%); (e) after re-excision for 12 patients (6%); or (f) after second re-excision for two patients (< 1%).

BCT at the University of Pennsylvania has been well described.^21-25 In brief, definitive locoregional treatment for all women consisted of breast-conserving surgery followed by definitive breast irradiation. The surgical treatment included complete gross excision of the primary tumor. Re-excision of the primary tumor site was performed for 437 patients (58%). Pathologic axillary lymph node staging was performed for all patients with invasive carcinoma. Early in the study period, pathologic axillary staging was generally performed using a lower axillary lymph node dissection, and later in the study period, sentinel lymph node biopsy became more commonly used.

Radiation treatment was delivered to all patients with definitive intent. The median dose was 46 Gy for the radiation treatment to the whole breast (mean = 46.54 Gy; range, 44.75 to 50.4 Gy). A boost to the primary tumor site was delivered after radiation treatment to the whole breast for all patients, and the large majority of the boosts were delivered using electrons. The total dose delivered to the primary tumor site was defined as the sum of the dose from the whole-breast treatment plus the dose from the breast boost. The median total dose was 63 Gy (mean, 62.58 Gy; range, 60 to 68 Gy). Regional nodal irradiation was added as clinically indicated, generally for patients with pathologically positive lymph nodes, and was delivered to the supraclavicular fossa alone in 39 patients or to the supraclavicular fossa and full axilla (including a posterior axillary boost field) in 32 patients. The internal mammary lymph nodes were irradiated in three patients.

Adjuvant systemic therapy was added as clinically indicated on the basis of patient and tumor characteristics. For the patients who received adjuvant hormonal treatment, the large majority received adjuvant tamoxifen. The study period was before the routine use of adjuvant aromatase inhibitors for postmenopausal patients.

The Kaplan-Meier method was used to calculate outcomes for overall survival, cause-specific survival, freedom from distant metastases, local control, and contralateral breast cancer.²⁶ The time period was calculated as beginning at the start of definitive breast irradiation, not at the time of diagnosis of breast carcinoma. The log-rank test was used for statistical comparisons between groups.²⁷ When two groups of patients differed with respect to baseline characteristics, an adjusted analysis was performed using the Cox proportional hazards model.^28,29 The minimal number of covariates necessary to avoid overfitting the model were included, particularly when there were limited numbers of events of interest. CIs using the Cox proportional hazards model with covariates were calculated using the method suggested by Link.³⁰

A local failure was scored for recurrence of disease within the treated breast. For the calculation of any local failure, all local failures, including first and subsequent events and including DCIS and invasive local failures, were considered events. Local-only first failure was defined as a local failure that occurred in the breast as the first and only site of failure. A regional failure was scored for a failure that occurred within the ipsilateral axillary, supraclavicular, infraclavicular, and/or internal mammary nodal region(s). For the determination of a contralateral breast cancer, both invasive cancer and DCIS and both first events and any events were scored as events.

For overall survival, a patient who was dead as a result of any cause was scored as a treatment failure. For cause-specific survival, a patient was required to be dead as a result of breast cancer to be scored as a treatment failure. For analysis of freedom from distant metastases, a patient was scored as a treatment 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 time for all patients was 4.6 years (mean, 5.0 years; range, 0.1 to 13.5 years). The median follow-up time for the 215 patients with a breast MRI study was 4.8 years (mean, 5.1 year; range, 0.1 to 12.9 years). The median follow-up time for the 541 patients without a breast MRI study was 4.5 years (mean, 4.9 years; range, 0.1 to 13.5 years). The numbers of patients alive at 5 and 8 years were 340 and 158, respectively.

RESULTS

Patient, tumor, and treatment characteristics according to the use of breast MRI study are detailed in Table 1. The patients with a breast MRI study were slightly younger than the patients without a breast MRI study. The median age for the patients with a breast MRI study was 53 years (mean, 53 years; range, 25 to 85 years), and the median age for the patients without a breast MRI study was 56 years (mean, 57 years; range, 27 to 89 years). The patients with a breast MRI study also had slightly more favorable tumor characteristics in terms of clinical tumor size, when known, and pathologic axillary lymph node staging.

Outcomes according to the use of breast MRI study are shown in Table 2. There were no differences between the two groups for overall survival, cause-specific survival, freedom from distant metastases, any local failure, local-only first failure, and contralateral breast cancer (all P ≥ .16). The 8-year rate of any local failure was 3% for the patients with a breast MRI study and 4% for the patients without a breast MRI study (Fig 1). After adjusting for patient age and date of treatment, there continued to be no difference between the two groups for overall survival, cause-specific survival, freedom from distant metastases, any local failure, local-only first failure, and contralateral breast cancer (all P ≥ .19; data not shown). The site(s) of first failure after treatment are detailed in Table 3.

For the subset of 136 patients with DCIS, the 8-year rate of any local failure was 6% versus 6%, respectively (P = .58). For the subset of 620 patients with invasive carcinoma, the 8-year rate of any local failure was 3% versus 3%, respectively, (P = .62).

DISCUSSION

The major finding from the present study is that the use of breast MRI was not associated with an improvement in outcomes after BCT with radiation. Although recognizing the retrospective, nonrandomized nature of the present study, these findings were observed in a large cohort of patients treated in a single institution, with the two groups treated over the same time period.

Maximizing local control is one of the goals of BCT. Patient selection for BCT strives to identify those tumors that are clinically unicentric on physical examination and mammography. However, the assumption that identifying otherwise clinically occult multifocal or multicentric disease with improved breast imaging will lead to improved local control warrants careful consideration. In pathologic studies of mastectomy specimens, foci of microscopic disease are commonly identified outside of the presumed surgical volume for excision of the primary tumor.^31,32 In randomized trials, the addition of radiation after breast-conserving surgery substantially reduces the risk of local recurrence, and for patients with invasive breast carcinoma, is associated with a small but real improvement in survival.^1,3,4 These results emphasize that radiation successfully controls the large majority of patients with microscopic foci of residual disease in the breast that are not resected at the time of breast-conserving surgery, and demonstrate that these microscopic foci will lead directly to local recurrence if left untreated by either radiation or mastectomy. However, even with the addition of radiation treatment, a small subset of patients develop local recurrence. The origin of such local failures may stem in part from one or more foci of disease at initial presentation that were too large for conventional doses of radiation or from tumor cells resistant to radiation treatment.

Notwithstanding attempts to improve local control, local failure after BCT will never be completely eliminated, nor should this be the goal of BCT. Retention of the breast will always leave the patient with some risk of local failure through the development of a new primary tumor of the breast or failure of primary treatment. For the patient with early-stage invasive breast carcinoma, mastectomy does not guarantee local control.

MRI has a high sensitivity for detecting disease in the breast, and for the patient with a known breast carcinoma, MRI is superior to mammography in delineating the extent of tumor in the breast.^5,8-14 The value of MRI may be particularly high in younger patients or patients with dense breast tissue.^8-10,13 A number of studies have demonstrated an improved yield for screening with breast MRI compared with mammography alone in high-risk patient populations.^33-37

For patients undergoing BCT, identifying clinically unsuspected foci of disease with improved breast imaging using breast MRI could potentially lead to better local control if such foci of disease were treated, for example, with wider surgical excision or mastectomy. Further, if a small subset of patients at high risk for local recurrence was identified and treated with a mastectomy, the remaining cohort of patients undergoing BCT would thereby have an improved rate of local recurrence. Alternatively, wider local excision might be sufficient for some high-risk patients to reduce their risk of local failure to a more acceptable level with BCT.

Even absent a lower rate of local failure after BCT for patients undergoing breast MRI, a number of potential advantages of adding breast MRI to conventional mammography have been described. Improved preoperative evaluation of the extent of disease should lead to improved tailoring of the surgical procedure.^6,7,9,10 For the patient requiring a mastectomy because of the extent of disease in the breast, one potential benefit of MRI might be to limit the number of surgical procedures before mastectomy. In practice, such patients often undergo a wide excision or re-excision with positive or close margins before subsequent mastectomy. However, one of the caveats of using breast MRI is that mastectomy should not be performed on the basis of MRI findings alone in the absence of pathologic confirmation of extensive disease in the breast.^4,6,7 Other potential advantages of adding breast MRI to mammography include the improved detection of contralateral breast cancers, improved imaging for invasive lobular carcinoma, and improved imaging for patients with dense breast tissue.^{5,8-10,13,17,18,38-42} Breast MRI might also be valuable for selecting patients for treatment with accelerated partial breast irradiation or breast-conserving surgery alone without radiation.^43,44

The use of breast MRI could potentially refine the criteria for BCT. The rate of breast conservation could be increased for certain subsets of patients, such as younger patients, patients with dense breast tissue on mammography, and patients with mammograms that are difficult to interpret. For patients with larger tumors, neoadjuvant chemotherapy has been shown to increase the rate of breast conservation, and MRI imaging may be the preferred imaging modality for monitoring the response of the primary tumor in this setting.^5,45-47 Breast MRI is indicated for the evaluation of the patient with breast implants or an axillary lymph node presentation, and might increase the use of BCT in these settings.^5,48-50

A number of arguments have been advanced against the routine use of breast MRI. Specificity for breast MRI is well known to be suboptimal. Breast MRI often identifies findings that prove benign on biopsy, and might increase the number of biopsy procedures. Breast MRI adds significantly to cost when used routinely, especially when additional procedures are required to evaluate MRI findings.

The current study has several limitations. First, the present study is a nonrandomized, retrospective analysis of a cohort of patients from a single institution. A subtle source of bias in the selection of patients for breast MRI cannot be excluded. For example, a breast MRI was more commonly obtained for younger patients (Table 1), and could potentially have been more commonly obtained for patients with mammograms that were difficult to interpret (for example, with dense breast tissue).

Second, the value of MRI could be underestimated in that patients identified as having extensive disease on MRI and treated with mastectomy were not considered in this study. Further, the local recurrence rate after BCT in the present study could have been subtly improved for the patients with an MRI, for example, through the use of mastectomy or wider resection for patients at higher risk or with more extensive disease on MRI study.

Third, the 8-year rate of local recurrence of 4% for the patients without a breast MRI study (Table 2; Fig 1) is sufficiently low that demonstrating an improvement would be extremely difficult in a retrospective cohort study. The treated patients in the present study were largely at low risk for local recurrence with predominantly negative margins, predominantly small tumors, and, for the invasive carcinomas, a high rate of adjuvant systemic therapy (Table 1). Given the low baseline rate of local recurrence in those patients without a breast MRI study, a randomized study would require a substantially larger sample size than the 756 patients evaluated in the present study.

Finally, the results of the present study suggest that MRI is not indicated globally for all patients undergoing BCT. However, the present study could not address whether there are one or more subsets of patients who might benefit from breast MRI (eg, patients at higher risk, younger patients, patients with larger primary tumors, or patients with dense breast tissue).

There is little information on the influence of using breast MRI relative to the outcome after BCT. Fischer et al⁵¹ reported the outcome for 346 patients, of which only a subset (n = 224; 65%) had undergone BCT. The rate of local recurrence after BCT was lower for patients with a breast MRI compared with patients without a breast MRI (1.2% [one of 86] v 6.5% [nine of 138], respectively; P < .001).

For patients undergoing BCT with radiation, one question that may not be answered in a prospective, randomized trial is whether better outcomes will be achieved for patients who undergo breast MRI staging compared with patients who do not undergo breast MRI staging. In contemporary practice, the baseline risk of local recurrence is low (Table 2; Fig 1), and, therefore, detecting an improvement in local recurrence with the addition of breast MRI would require a large number of patients. Technical differences in breast MRI imaging among various institutions could also pose a significant problem to mounting a randomized trial. The Comparative Effectiveness of MRI in Breast Cancer (COMICE) trial in the United Kingdom is a randomized trial of 1,850 patients that will evaluate the impact of MRI on the selection of patients for BCT and the adequacy of breast-conserving surgery.^52,53

In summary, the present study has demonstrated that the use of a breast MRI study at the time of initial diagnosis and evaluation for patients with early-stage breast carcinoma was not associated with an improvement in outcomes after BCT. These findings suggest that MRI is not globally indicated for all patients with early-stage breast cancer at the time of initial diagnosis and evaluation. However, the results of the present study do not exclude either a small benefit for the use of breast MRI or a benefit for one or more subsets of patients.

Authors’ Disclosures of Potential Conflicts of Interest

Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment: N/A Leadership: N/A Consultant: Mitchell D. Schnall, Berlix Stock: N/A Honoraria: Mitchell D. Schnall, Siemens, Berlix Research Funds: Mitchell D. Schnall, Siemens Testimony: N/A Other: N/A

Author Contributions

Conception and design: Lawrence J. Solin, Susan G. Orel, Eleanor E. Harris, Mitchell D. Schnall

Financial support: Lawrence J. Solin

Administrative support: Lawrence J. Solin

Provision of study materials or patients: Lawrence J. Solin, Susan G. Orel, Eleanor E. Harris, Mitchell D. Schnall

Collection and assembly of data: Lawrence J. Solin

Data analysis and interpretation: Lawrence J. Solin, Wei-Ting Hwang, Eleanor E. Harris

Manuscript writing: Lawrence J. Solin, Susan G. Orel, Wei-Ting Hwang, Eleanor E. Harris, Mitchell D. Schnall

Final approval of manuscript: Lawrence J. Solin, Susan G. Orel, Wei-Ting Hwang, Eleanor E. Harris, Mitchell D. Schnall