In a study of breast magnetic resonance imaging (MRI) in women at high risk of developing breast cancer, 24% of women had lesions interpreted as "probably benign"; among these lesions, 7-10% were subsequently found to be cancer, according to a study published in the journal Cancer.

In this study, the first to evaluate the outcome of "probably benign" lesions of the breast detected by MRI, Liberman et al. at Memorial Sloan-Kettering Cancer Center found that more than half of these malignant lesions were of such size and quality that they were detectable only by MRI and not by physical exam or mammography. Furthermore, more than half of the lesions that proved to be cancer were ductal carcinoma in situ (DCIS), an early, noninvasive form of breast cancer with an excellent prognosis.

The authors conclude, "these data indicate that follow-up breast MR imaging in high-risk women can detect early breast carcinoma before it can be diagnosed by mammography or physical examination."

Radiologists characterize lesions of the breast as "negative," "benign," "probably benign," "suspicious" or "highly suggestive of malignancy." While it is perhaps obvious what to do with the benign, suspicious and highly suggestive lesions, probably benign lesions have been the source of concern for clinicians. Biopsies are expensive and have associated risks. Observation by follow-up imaging is noninvasive, but requires the existence of criteria to determine which lesions are appropriate for follow-up.

In mammography, short-term follow-up with repeat mammogram is the accepted practice to track changes - one supported by evidence gathered from previous investigations. Prior studies report "probably benign" breast lesions in 3% to 11% of mammograms, of which 0.2% to 2% are actually malignant. However, no previous outcome data exist for "probably benign" lesions detected by breast MRI, a technique which is now emerging as a highly sensitive adjunct to mammography for breast cancer detection. Distinguishing benign from malignant lesions at breast MRI may be challenging, and there are limited evidence-based data to guide physicians in cases of equivocal interpretations.

Liberman et al. investigated 367 women at high risk for breast cancer (due to prior breast cancer, family history of breast cancer, or previous biopsy showing atypia or lobular carcinoma in situ) who had normal mammograms and a subsequent screening breast MRI. The authors analyzed data for the frequency of "probably benign" interpretations of breast MRI screening and how often these lesions became malignant.

"Probably benign" interpretations were reported for 89 of 367 women (24%) and included single (38%), multiple and unilateral (25%) and multiple and bilateral (37%) lesions. Short-term follow-up MRI was recommended for 89 women; follow-up MRI was performed for 70 women (79%) at a median of 11 months after screening MRI. Twenty of the 89 women underwent subsequent biopsy; malignancy was found in nine women, constituting 10% (9/89) women with "probably benign" lesions in the study. Six of the 9 (67%) cancers were identified based on interval progression of lesions at follow-up MRI.

The authors comment that the 7-10% frequency of cancer in "probably benign" lesions detected by MRI is "...higher than the 0.2-2% frequencies of malignancy reported in studies of nonpalpable, mammographically detected, probably benign lesions." The higher frequency of cancer in these lesions may reflect the nature of breast MRI, which detects lesions that have increased blood flow, as well as the high-risk population having the breast MRI examination.

Of the 9 malignant lesions identified by MRI, 6 (67%) were nonpalpable on clinical exam and not detectable by mammography. Biopsy identified the malignancy as early, noninvasive DCIS in five (56%) of the nine malignancies; the remaining four (44%) malignancies were small infiltrating carcinomas (median size, 0.6 cm). Hence, for women at high risk of developing breast cancer, follow-up MR imaging was useful in identifying cancer at an early stage, with excellent prognosis.

"The findings of the current study indicate the importance of establishing evidence-based criteria for probably benign lesions on breast MR imaging studies," conclude the authors.

SOURCE:
Cancer, July 15, 2003