Nuclear magnetic resonance in the diagnosis of breast cancer.

We believe that the preponderance of evidence suggests that MRI is less accurate than conventional mammography in the diagnosis of primary cancer of the breast. Thus, it currently has no established place in algorithms for the evaluation of patients suspected of having breast cancer. MRI could be used to evaluate masses with mammographically smooth, well-defined margins, since high signal intensity (greater than fat) in a T2-weighted image is a highly specific indicator of benignancy in such lesions. However, most of these masses are cysts and can be reliably and less expensively identified as such by sonography. Nonetheless, MRI might be used to re-evaluate a smooth, well-defined mass if sonography has failed to identify the lesion as a cyst. MRI might be particularly useful in this regard if a lesion is difficult to evaluate by other modalities because it is located adjacent to the chest wall, is deep within a very large breast, or is obscured by a breast prosthesis. MRI with Gd-DTPA may be useful in evaluating radiographically dense breasts or in differentiating breast malignancies from irregular dysplastic or scar tissue. However, further investigation of this technique is needed. It has been hoped that in vivo measurement of T1 and T2 or in vivo NMR spectroscopy might improve the accuracy of noninvasive diagnosis of cancer of the breast. However, there is currently no credible evidence that in vivo measurements of relaxation times provide useful indexes for the diagnosis of breast cancer. In vivo NMR spectroscopy of nuclei other than P may ultimately provide reliable criteria for noninvasive diagnosis of breast cancer in humans, but the technique is currently in its infancy.