In Fibroglandular, Dense Breasts, MRI Outperforms Mammography in Detecting Multiple, Malignant Foci

MILAN, Italy—In women with fibroglandular or dense breasts, magnetic resonance imaging (MRI) is more sensitive than mammography for detection of multiple malignant foci, suggesting that a dynamic MRI examination is warranted before treatment planning in this group of patients, a team of Italian radiologists and surgeons has concluded. Yet in breasts with an almost completely fatty pattern, both techniques had comparable sensitivity, their multicenter, prospective, nonrandomized study showed. Further, while MRI achieved a 17% gain in sensitivity over mammography in detection of invasive foci, the two techniques had similar sensitivity in detection of in situ foci, and neither had a strong positive predictive value (PPV), the researchers found.

In their analysis of 99 breasts in 90 women with planned mastectomies who underwent both diagnostic techniques, with pathologic examination of the whole excised breast used as the gold standard, mammography missed more larger and invasive cancer foci than MRI.

The team, led by Francesco Sardanelli, MD, of the Department of Diagnostic Imaging, Istituto Policlinico San Donato, Milan, Italy, noted that their results were expected and confirmed findings of several large series indicating that MRI has significantly greater sensitivity than mammography for correct diagnosis of multifocal, multicentric cancer (Am J Radiol 183:1149-1157, 2004).

The value of pathologic analysis of the whole excised breast in this study, they said, is that it provided a true picture of the proportion of false-negative findings with mammography and MRI, adding that, to their knowledge, theirs is the first large study to use a whole-breast pathologic exam as a gold standard for evaluating mammography vs MRI in detection of multifocal, multicentric disease.

The global trend toward a conservative surgical approach to breast cancer, they emphasized, should be based on exclusion of undetected malignant foci in the breast, which they noted are frequently responsible for relapses after conservative surgery.

The 18-center trial was performed between 1998 and 2000. All participating sites had proven breast imaging experience. Of 153 patients enrolled, 90 were evaluable. The women in the study had a mean age of 59 years (range, 43 to 75). They had proven breast cancer and planned mastectomies, 9 of which were bilateral, for a total of 99 breasts examined. Prior to surgery, the patients underwent both mammography and gadolinium-enhanced dynamic MRI. After imaging, pathologic exams of the entire excised breast were performed onsite.

Lesion-Matching Process

All examinations were evaluated for the presence of malignant foci by a team of two off-site radiologists who were aware that mastectomies were planned but were blinded to the results of the pathologic examination. A third radiologist compared the location of the breast foci in pathologic sections, on mammography, and on MRI, using a nine-region map of eight segments plus the nipple region.

A database created through this lesion-matching process enabled investigators to perform analyses on both a lesion-by-lesion and a breast-by breast basis. Based on the pathologic report, cancer was diagnosed as unifocal when a single malignant focus was found, as multifocal when more than one malignant focus was shown in the same segment or in adjacent segments of the map, and as multicentric when foci were identified in noncontiguous regions.

Pathology revealed a total of 188 malignant foci (158 invasive, 30 in situ) among the 99 breasts examined: 52 lesions were unifocal, 29 were multifocal, and 18 were multicentric.

Sensitivity

The overall sensitivity for detection was 66% for mammography and 81% for MRI (P < .001). MRI also had 17% greater sensitivity in detecting invasive foci (72% for mammography vs 89% for MRI, P < .001). However, the two techniques were not significantly different in their detection of in situ foci (mammography 37%, MRI 40%, P > .05).

Mammography missed a total of 64 malignant foci and MRI missed a total of 36; the median diameter of missed foci was 8 mm with mammography vs 5 mm with MRI (P = .033). Both mammography and MRI detected the nine cases of bilateral cancer.

Striking differences between the techniques were noted in dense or fibroglan-dular breasts (see Figures 1 and Figures 2), for which mammography and MRI had sensitivities of 60% and 81%, respectively (P < .001); the PPV of the techniques was similar, however, at 78% and 71%, respectively.

In contrast, results in breasts with an almost completely fatty pattern showed that sensitivity of mammography vs MRI was comparable, at 75% and 80%, respectively, as was their PPV, at 73% and 65%, respectively, with neither difference reaching statistical significance. The overall PPV was not significantly different for either diagnostic technique, at 76% for mammography and 68% for MRI
(P > .05).

In terms of staging of the lesions, both techniques correctly staged only 51% of the breasts, with mammography under-staging 30% and overstaging 19% of the remaining breasts; conversely, MRI understaged 19% and overstaged 30% of remaining breasts.

False-Positive Findings

In discussing the results, the investigators commented that, while a PPV of 76% for mammography might be considered noteworthy, the technique was associated with a total of 40 false-positive lesions in the study, and the high PPV seen "can be attributed in large part to selection of potential candidates for mastectomy with almost two (1.9) malignant foci per breast, of whom 47% had multifocal, multicentric cancer."

They speculated that the fact that 19% of malignant foci were undetected by MRI may have been a result of their small size (5 mm, vs 8 mm for foci detected by MRI), or perhaps a low level of angiogenesis.

The investigators concluded that MRI may be most practical following a second "conventional evaluation" such as a "second mammography to acquire tailored views or magnifications, or a targeted sonography after MRI," adding that MRI-guided breast biopsy should be performed when MRI-detected foci are undefined. Because their findings confirmed that, like mammography, the specificity of MRI is "relatively low," they concluded that "availability of MRI-guided biopsy [should be] mandatory for centers that intend to use MRI as a diagnostic tool."

The investigators suggested that the specificity of MRI could be improved "by integrating the best morphologic and dynamic data, by means of proton MR spectroscopy, and by the use of short-term antiestrogen (tamoxifen) medication before MRI."

Further gains in sensitivity of MRI, they wrote, might be achieved by increasing the in-plane and through-plane spatial resolution.

Going forward, the research team emphasized that only further research can determine the true clinical value of using only MRI to detect malignant foci in the breast, noting that "randomized studies comparing the outcome of patients undergoing pretreatment MRI are needed to define the effects of a more precise evaluation of the extent of disease on relapse rate, quality of life, and survival rate."