USING MAGNETIC RESONANCE IMAGING TO DETERMINE THE EXTENT OF A CANCER

MRI appears to be the most effective imaging technique for demonstrating additional occult foci of breast cancer in a breast in which a cancer has been detected. Some have begun to use MRI to establish the extent of the lesion for treatment planning. This becomes somewhat more difficult in that benign lesions may also enhance, and, although sensitivity is high, specificity of MRI is moderate to low using just enhancement. Separating true positive foci of cancer from falsely positive tissue is problematic. Efforts to measure the time course of tissue enhancement have had variable results, but breast cancer appears to enhance more rapidly than do benign lesions.

Magnetic Resonance Enhancement Behavior of Breast Lesions: Telling Benign from Malignant

There has been a great deal of interest in using MRI to differentiate benign lesions from malignant. Almost all invasive malignancies enhance with gadolinium (Gd)-DTPA and some research has shown that, with dynamic MR imaging techniques, malignancies may enhance at much more rapid initial rates than benign lesions. This rapid initial enhancement rate of malignancies is likely due to tumor angiogenesis. Malignant lesions are known to require the recruitment of a large concentration of tumor neovessels to permit their continued growth beyond a few millimeters. New vessels recruited by the tumors also have abnormal basement membranes. This results in vessel leakiness and an increase in surrounding interstitial fluid pressure.The increased concentration of vessels at the tumor site and their leakiness likely account for the rapid accumulation of Gd-DTPA in breast cancers. Some benign breast lesions also enhance with Gd-DTPA. This may also be due to higher vascularity, although whether the enhancement is due to increased concentrations of neovessels or large feeding vessels in some benign entities is not clear. The initial enhancement rate (within the first minute) is usually less rapid in benign diagnoses than in malignancies, offering the opportunity of distinguishing benign lesions from malignant by their enhancement pattern. It is important to note that if MR imaging is delayed and scanning begins even only minutes after the injection of Gd-DTPA, some benign lesions will show a higher signal intensity than malignant lesions. It is not the absolute enhancement that may permit separation of benign from malignant lesions but the dynamics of the enhancement. It should be remembered that if the goal is to use MRI to differentiate benign from malignant lesions, the analysis of the enhancement soon after injection of the contrast appears to be critical.

Some investigators have disputed the consistency of early dynamic MR in distinguishing benign from malignant lesions, as several benign lesions demonstrated a more rapid rate of enhancement than expected. Furthermore, several invasive cancers, including several infiltrating lobular carcinomas, two malignant phylloides tumors, one tubular carcinoma, and colloid and mucinous carcinomas, have been reported as having slow enhancement profiles. Although differences in vessel density may be the explanation for such variation, studies correlating low vessel density with slow enhancement of certain cancer cell types have not been performed. However, pathologic studies have shown a low expression of endothelial growth factor receptor messenger RNAs in infiltrating lobular carcinomas (as opposed to strong expression in invasive ductal carcinomas), which may account for decreased vascularity of this cell type.

Some authors report that the morphology of the lesion and its pattern of enhancement may permit the separation of benign from malignant processes. As has been shown by mammography and ultrasound, cancers demonstrate more irregularly shaped borders than do benign tumors. However, because neither mammography nor ultrasound has been able to differentiate many lesions morphologically with sufficient accuracy to avoid a tissue diagnosis, it is unlikely that MRI will be any more successful. Nevertheless, morphologic studies using MRI bear careful evaluation, because MRI provides cross-sectional morphology, which, combined with enhancement dynamics, may improve its accuracy. Numes. constructed a flow chart model to try to use these enhancement and architectural characteristics to differentiate benign lesions from malignant. They used the results from 98 cases to produce the model and then tested it on 94 different cases. They found that a lesion was likely to be cancer, primarily, if it enhanced. A high degree of enhancement, an irregular pattern of enhancement, irregular borders, and the lack of any internal septations were associated with malignancy. Cancer was very unlikely if the lesion was not visible after gadolinium infusion, if its borders were smooth or lobulated, or if the mass was regular in shape or it had nonenhancing internal septations.

The early results from trials are frequently the most successful. More work is necessary in applying patterns of enhancement shape, rate, or combination of MR data with mammography and sonography before benign and malignant lesions can be distinguished with sufficient accuracy to avoid a safe tissue diagnosis.

Magnetic Resonance Imaging and Ductal Carcinoma In Situ

The MR behavior of DCIS is less clear, as enhancement varies from a rapid rate to no enhancement. This variation in enhancement behavior may be due to the variation of neovessel recruitment in DCIS. Despite the fact that these are intraductal lesions, some DCIS can stimulate neovascularity. Soderstrom et al. suggest that virtually all DCIS can be detected. Others have not been as successful, because some cases of DCIS do not appear to enhance. In a review of 13 patients with pure DCIS, Orel  were able to identify ten of the lesions on MRI scans. In six cases, curvilinear enhancement was termed ductal. In three of the cases there was segmental or “regional” enhancement, and in one case the enhancement was termed peripheral. Three lesions were not evident on their scans, and two other lesions were not identified in six additional patients who had both invasive and intraductal cancer. Giles evaluated 36 women with DCIS and contrast enhancement evaluation using the subtraction technique. They were able to identify 34 of the lesions but failed to demonstrate two cases of comedocarcinoma. One problem that complicates some of the studies is the fact that there is a difference between knowing where DCIS was found, histologically, and “finding” it on an imaging study versus identifying the lesion prospectively and having it confirmed by the imaging.

The ability of MRI to detect DCIS, particularly poorly differentiated DCIS, is important. Not only do these lesions tend to progress to invasion sooner, but it is also critical for assessing the extent of the cancer. If DCIS is not recognized extending away from a primary invasive tumor, then excision, or even destroying the primary tumor in vivo (e.g., laser, cryotherapy, high-frequency ultrasound), will be likely to fail to achieve local control because residual, undetected (and hence untreated) tumor will lead to high recurrence rates.

Specific Applications of Breast Magnetic Resonance Lesion Imaging

The roles for MR in evaluating breast lesions are being defined. The ability to distinguish benign from malignant lesions on the basis of their dynamic enhancement pattern or their morphologic characteristics, or both, using the variety of techniques discussed previously remains under investigation in the United States.

MR imaging is already being used by some in the following specific areas:

1. Tumor staging to determine the extent of disease within the breast, which is important in permitting complete excision or tumor destruction if conservation therapy is chosen.
2. The differentiation of scar from cancer recurrence. Sometimes, a patient who has undergone lumpectomy or radiation therapy, or both, for cancer may later present with a suspicious mammogram or physical examination in the region of prior surgery. Some of these patients undergo reexcision of the scar due to the concern for local recurrence, although local recurrence only occurs in 1% of such patients each year. MRI may be useful in distinguishing scar tissue from local recurrence at the specified postoperative time interval. Although immediately after surgery and during the course of radiation therapy, a high density of new vessels from wound healing can result in enhancement on MR imaging, recent small series have suggested that scar tissue does not enhance with Gd-DTPA on MR. Dao et al. have suggested that MR imaging can be used to follow up women treated conservatively with excision and irradiation. Areas of surgery frequently enhance if they are studied within 6 to 12 months after treatment. After that period of healing, fixed fibrosis does not enhance, but recurrent breast cancer frequently does. If any enhancement is visualized in a scar older than 12 months, the possibility of recurrence increases. Although the early detection of recurrence in the breast may have useful therapeutic value, it has yet to be shown that detecting recurrences earlier alters mortality. Nevertheless, in the absence of data, early detection of recurrent breast cancer would seem to be a good idea.

Chest wall imaging, searching for a primary malignancy, and following the response of breast cancers to chemotherapy are other areas in which some find MR imaging of the breast to be useful. As noted earlier, we and others have found MR imaging to also be useful in searching for a primary breast cancer in patients with metastatic disease of unknown origin or in those with axillary adenopathy that is highly suspicious for a breast malignancy.

The evaluation of local tumor response to chemotherapy with MR breast imaging may also provide useful information to oncologists or surgeons who are employing neoadjuvant chemotherapy and wish to monitor tumor response.

Magnetic Resonance Imaging as a Possible Second-Level Screen

Another possibility, which has yet to be proved, is the use of MRI to detect breast cancer in asymptomatic, healthy-appearing women with negative mammograms. We have found a surprising number of mammographically and clinically occult synchronous breast cancers in the contralateral breasts of women with breast cancer detected only by MRI (4 of 17).

In addition, among eight women who presented with metastatic disease, six had their clinically and mammographically occult breast cancers detected by MRI. These anecdotal cases suggest the possibility of MRI as a second-level screening technology that may be able to detect early cancers that, at present, are mammographically as well as clinically occult.

Despite the fact that the experience with contrast-enhanced MRI of the breast has greatly increased, many of the data are anecdotal. The lack of large prospective studies to define scientifically the efficacy of MRI in breast evaluation has kept it largely as a research tool. The cost of the study and limited access to the magnets have greatly slowed the research. Before MRI can be applied to screening, much work is needed to validate its efficacy.