Intraoperative MR Imaging: Can Image Guidance Improve Therapy?

Abstract

Current estimates are that approximately 189,000 new cases of prostate cancer and 205,000 new cases of breast cancer will be diagnosed this year in the United States alone (1). Furthermore, both diseases most commonly manifest as organ-confined disease that is treated with local therapy alone. Two articles in this issue of Academic Radiology present groundbreaking work on the use of intraprocedural magnetic resonance (MR) imaging to guide local therapy for breast (2) and prostate cancer (3). Survival in patients with stage I breast cancer after mastectomy is similar to that after breast-conserving therapy (ie, lumpectomy combined with radiation therapy). While lumpectomy is straightforward for well-defined palpable masses, it is often difficult for patients with nonpalpable lesions, despite the use of preoperative imagingguided localization and lesion marking with a hookwire(s). Indeed, initial lumpectomy is incomplete in a substantial number of patients, as evidenced by the presence of transected tumor at the inked margins of the surgical specimen. These patients thus require at least one reexcision to achieve complete tumor removal. The goal of MR imaging–guided excision, as first proposed by Gould et al (4) and substantially expanded on in this issue by Hirose et al (2), is to guide resection with intraoperative MR imaging, with the goal of attaining tumor-free resection margins and thereby obviating reexcision. The specific technique employed by Hirose et al (2) builds on the success of intraoperative MR-guided resection of brain tumors. As with brain tumors, contrast material–enhanced MR imaging clearly reveals the size and extent of invasive breast tumors better than palpation or other imaging modalities, and it is thus a natural choice for guiding tumor removal. The procedure also shares the complexities of brain tumor resection in that a “bloodless” surgical field must be achieved before any imaging, to prevent confusion between tumor enhancement and the accumulation of contrast material within hemorrhage. There are additional challenges to breast surgery with MR imaging guidance, the principal one being the lack of a “blood-brain barrier” in the breast, which means that the preferential enhancement of breast tumors is evanescent compared with the continual enhancement of surrounding breast tissue. Although this window of tumor visualization can be prolonged by imaging patients during the time in the menstrual cycle when normal glandular enhancement is slowest, the period of adequate visualization is not as long as a lumpectomy procedure. To counter this problem, Hirose et al (2) perform a second injection, but residual enhancement from the initial injection complicates this approach. Clearly, the technique would be much more practical with longer-acting tumorspecific agents, such as intravascular agents or angiogenesis-specific liposomal contrast agents, which are under development. Motion is another challenge not encountered with brain tumor resection. Breath-hold imaging techniques are required for adequate image quality, and the registration between preand postcontrast images (required for subtraction-based fat suppression) is extremely difficult in supine patients, especially when even small misregistration artifacts can overwhelm potentially small foci of residual tumor. Furthermore, once any additional resection is performed, another injection and subtraction must be performed to assess for residual tumor. Ultimately, the Acad Radiol 2002; 9:875–877