Magnetic resonance imaging in the breast cancer patient: curb your enthusiasm.

There is no doubt that breast magnetic resonance imaging (MRI) identifies foci of cancer that are not evident on physical examination, mammogram, or ultrasound. The documentation in multiple studies that MRI identifies additional cancer in the ipsilateral breast of 11% to 31% of women with unicentric breast cancer has led to increasing use of this technology in the newly diagnosed breast cancer patient. Advocates of MRI cite as potential benefits improvement in the selection process for breast conserving surgery, a decrease in the number of surgical procedures required to obtain negative lumpectomy margins, and the synchronous detection of contralateral cancers. These benefits would provide a compelling rationale for the routine use of MRI, but, unfortunately, there are no data from prospective randomized trials to support these assertions. To date, the only information on the impact of MRI on local recurrence comes from a single retrospective study of 224 women treated with breast-conserving therapy (BCT) that reported a 40-month local recurrence rate of 1.2% for those who had preoperative MRI compared with 6.5% for those who did not (P .001). Unfortunately, the failure to adjust for differences in tumor size, nodal status, and, most importantly, the use of systemic therapy between groups render the results of this study uninterpretable. In this issue of the Journal of Clinical Oncology, Solin et al provide much-needed outcome data on the impact of MRI on local recurrence after BCT. Although their study is retrospective, factors known to influence local recurrence such as age, margin status, date of treatment, and the use of systemic therapy are carefully documented, and appropriate statistical adjustments for differences between groups were made. Of 756 patients reported, 215 had a breast MRI as part of their initial evaluation and 541 did not. The 8-year rates of any local failure or local-only first failure were 3% and 4% in the women with and without MRI, respectively. When patients with invasive carcinoma and intraductal carcinoma were analyzed separately, there was no trend toward a decrease in local recurrence with MRI in either histologic group. These results are noteworthy for several reasons. First, the radiologists performing breast MRI at the University of Pennsylvania (Philadelphia, PA) are leaders in the field, with extensive experience with a standardized technique of breast MRI and the ability to perform MRI-guided biopsies. Yet, despite this, no benefit of MRI in reducing the rate of local recurrence was observed. Second, patients with unfavorable findings on MRI were excluded from the BCT group, rather than simply observed to determine the significance of the mammographically occult cancer. Three studies examining changes in surgical therapy resulting from MRI findings found conversion from planned BCT to mastectomy to be the most common outcome, with mastectomy performed in 15.5% to 25% of cases because of the MRI demonstration of disease thought to be too extensive to be encompassed by lumpectomy. This suggests that between 81 and 136 women in the study by Solin et al who were treated with BCT without the benefit of an MRI were poor candidates for the procedure, yet only 13 experienced any type of local failure. In addition to failing to show a benefit for MRI, the study by Solin et al reaffirms that local failure after BCT is an extremely uncommon event, with 96% of women undergoing BCT without an MRI free of local recurrence at 8 years. Low rates of local failure after BCT are not limited to tertiary-care hospitals staffed with breast specialists. Wapnir et al observed 10-year cumulative local recurrence rates ranging from 4.8% to 10.1% across five National Surgical Adjuvant Breast and Bowel Project (NSABP) trials involving 2,669 node-positive patients treated between 1984 and 1994, and 10-year local recurrence rates of 3.5% to 6.5% were observed in node-negative patients receiving systemic treatment in NSABP trials during the same time period. In considering the potential benefit of MRI in reducing local failures after BCT, it is worth remembering that isolated local failure is observed in 8% of node-negative women treated with mastectomy without chestwall irradiation and 27.6% of their node-positive counterparts treated in the same fashion at 10 years of follow-up. The expectation that MRI will eliminate, or even significantly reduce, currently observed local failure rates fails to consider the dual nature of local recurrence. Some local recurrences are a result of tumor burden too large to control with radiotherapy, or inadequate treatment as evidenced by the increased rates of local failure observed in women with positive lumpectomy margins or those treated with lumpectomy without radiotherapy. However, because MRI detects some but not all of the pathologically evident, clinically occult cancer in the breast, it cannot be expected to reduce even this type of local recurrence to a zero incidence. A direct comparison of serial subgross sectioning and MRI found that the sensitivity of MRI for the detection of invasive cancer ranged from 81% to 89%, and was 40% for in situ disease. Other local failures are really a first site of metastases, with characteristics similar to those of chest-wall recurrences postmastectomy with a short interval to development and an increased risk of subsequent distant relapse. It is highly unlikely that the use of MRI will have any impact on this type of local recurrence. It is also interesting that Solin et al observed no difference in the incidence of contralateral breast cancer in the MRI and non-MRI groups, with 6% of women in each group developing a contralateral cancer at 8 years. This finding is consistent with data from 134,501 JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 26 NUMBER 3 JANUARY 2