Sarah J Vinnicombe
BSc (Hons) MRCP FRCR
Radiologist and Lead, Breast Radiology
St Bartholomew’s Hospital
Barts and The
London NHS Trust
E: s.j.vinnicombe @qmul.ac.uk
Though contrast-enhanced magnetic resonance imaging (MRI) of the breast was first described in the late 1980s, it has only really gained widespread acceptance in the last few years.(1,2) The reasons for this should be kept in mind when a breast MR service is being established in an institution. Until very recently, most radiologists in the UK reading breast MR studies were not breast radiologists. Today, training in breast radiology requires experience in interpreting breast MR studies. This reflects the increasing recognition of breast MR as a very powerful tool in breast imaging.
The development of faster, more robust gradient echo sequences, advances in coil design, the advent of MR-guided breast biopsy facilities and the standardisation of data acquisition and interpretation have all contributed to the incorporation of MR into the breast imaging paradigm.
Establishing a breast MR service
It is crucial to have the support of clinical colleagues before embarking on this daunting task. There is a very steep learning curve in evaluating breast MRI, and early on it is vital to obtain clinicopathological correlation with the MR findings. At St Bartholomew’s Hospital breast MR was performed on a cohort of patients scheduled for mastectomy, and the pathological specimens were subjected to 5mm subgross sectioning in the most appropriate plane for direct correlation with the MR images.
For premenstrual patients, examinations should be scheduled whenever possible for days seven to 12 of the menstrual cycle; for postmenopausal patients, consideration should be given to discontinuing hormone replacement therapy for six weeks prior to the examination. These measures can help reduce confusion from nonspecific progesterone-induced enhancement of the breast parenchyma.
Immediately prior to the scan, the breast should be examined and any scars or masses marked with an oil capsule, so that the relationship between the area of clinical concern and any abnormal MR findings can be ascertained. The technician must establish a good rapport with the patient and explain the necessity for absolute immobility during the dynamic scans. Careful positioning within the dedicated double breast coil is essential; any ridges or skin folds can cause local field inhomogeneities resulting in nonuniform fat suppression. Some authors advocate the use of gentle compression to minimise motion, but excessive compression may diminish contrast enhancement within the breast, resulting in a false-negative examination. Use of a powered injector ensures rapid delivery of a standardised bolus injection of contrast material. There is some evidence for the use of a larger dose of gadolinium-DTPA (0.15–0.2mmol/kg) with 3D gradient echo acquisitions but this is may be unnecessary with newer agents such as gadolinium–BOPTA, with its blood pool effect and relatively greater, more prolonged enhancement.(3)
The efficacy of breast MR lies in its ability to detect breast cancer, either in situ or invasive, which is not visible on conventional imaging. Though standard T1 and T2 weighted spin-echo sequences do yield a certain amount of diagnostic information, such as the presence of cysts or blood products, the key is the acquisition of a dynamic contrast-enhanced T1-weighted gradient echo scan, preferably 3D, of both breasts, with sufficiently high spatial and temporal resolution, so that the morphological and temporal enhancement characteristics of any abnormalities can be fully evaluated. In recent years, all the major manufacturers have developed sequences capable of providing in-plane resolution under 1mm, with a Z-plane resolution of under 2mm, while at the same time providing homogeneous fat suppression with no additional time penalty, such that both breasts can be imaged in their entirety within 60 seconds. This is facilitated with parallel channel acquisition techniques. Though coronal acquisitions minimise partial volume averaging of ductal structures as they course towards the nipple, most clinicians find axial and sagittal acquisitions intuitively easier to comprehend.
In the past, the great variety of sequences, methods and diagnostic criteria has made comparison of results from different centres extremely difficult. Standardisation of data acquisition and interpretation will facilitate rigorous scientific evaluation of the diagnostic performance of the technique. In an attempt to address some of these issues, the American College of Radiology (ACR) has produced a document on the performance of breast MRI.(4) The most recent edition of the ACR Breast Imaging Reporting and Data System has, for the first time, included a section on breast MRI, so that standardised descriptive terminology can be employed.(5) This will undoubtedly help those radiologists with little experience in breast MRI. Finally, every breast centre performing breast MRI should ideally have the facility for MRI-guided breast biopsy. The sensitivity of the technique is such that there will inevitably be unsuspected suspicious findings whose malignant nature must be proven before the treatment plan is altered. There is some evidence, for example from the MARIBS trial of MRI in screening of high-risk patients, that some biopsies are performed because of reader inexperience; the development of the BIRADS lexicon should go some way towards preventing this. Many manufacturers have now developed breast intervention coils and MRI-compatible needles, guidewires and marker clips.
Indications for breast MRI
Breast MRI is best used when there is either a known cancer, or a very high probability of finding one. The main indications are shown in Table 1. The ability of MRI to detect multifocal, multicentric disease that is otherwise occult renders it very accurate in preoperative staging.(6-9) It is the most accurate method of measuring tumour size, proximity of disease to the nipple/subareolar complex, skin involvement or chest wall involvement – all features that may make primary surgery difficult.(10) MRI can detect mammographically occult ductal carcinoma in situ (DCIS), provided that it is of sufficiently high grade to induce neoangiogenesis.(11,12) The ability to identify this confidently before operating facilitates appropriate surgical planning (Figure 1). MRI can result in a change of surgical treatment in up to 20% of patients, resulting in fewer surgical procedures to attain adequate local tumour control with fewer local recurrences.(6,7,13) It is particularly valuable in those whose disease is difficult to assess by conventional means, such as young women with dense breasts and women with lobular carcinoma.(14-16) MRI can also be used to screen the contralateral breast for synchronous occult carcinomas, which may be found in 4–9% of patients.(17,18)
What is unclear at present is whether the additional foci of disease that MRI detects, whether in situ or invasive, are in fact clinically relevant. A possible consequence of the great sensitivity of breast MRI is overtreatment of the disease with an excess of mastectomies. This question should, among others, be answered by the COMparative effectiveness of Magnetic resonance Imaging in breast CancEr trial (COMICE), a multicentre prospective trial underway in the UK looking primarily at the efficacy of preoperative breast MRI in women with breast cancer scheduled for breast conservation.
The very high negative predictive value of breast MRI can be used to exclude malignancy where there are ambiguous clinical or conventional imaging findings, and where biopsy is negative or where, for some reason, cannot be performed. It should be appreciated that a negative examination does not preclude the presence of low-grade DCIS or high-risk lesions, such as atypical hyperplasias, where enhancement may be minimal. In those women who present with adenocarcinoma metastatic to the axillary lymph nodes in whom an occult breast primary tumour is suspected, MRI may identify the primary tumour (Figure 2).(19) The importance of this lies in the fact that the patient may be then be treatable with breast conservation as opposed to the less desirable options of mastectomy or whole-breast irradiation.(20)
The ability of breast MRI to depict accurately the extent of the tumour within the breast makes it well suited to assessing the response to neoadjuvant chemotherapy. Numerous studies have demonstrated that it correlates most closely with final histopathological status after surgery.(21) Chemotherapy-induced fibrosis (which can appear very mass-like on mammography and ultrasound) generally enhances much less than residual tumour. Furthermore, MRI more accurately predicts response than conventional means after only a few cycles of chemotherapy – a typical pattern being a change in the shape of the enhancement curve from a typically malignant pattern to a more benign one (Figure 3), as well as a reduction in the size of the mass. Therefore, an early change of therapy can be offered to nonresponders before toxicity has ensued from an ineffective treatment.
MRI may also be of clinical use in the assessment of the amount of residual disease in the breast after wide local excision with positive pathological margins. Though surgery itself causes an inflammatory response resulting in enhancement around the scar and postoperative seroma, this can often be differentiated from residual foci of invasive disease or high-grade DCIS provided they are large enough, though not necessarily from low-grade DCIS.(22,23)
MRI is now the established method of choice for differentiating between florid post-treatment scarring and scar recurrence. Mammography will only detect up to 30% of scar recurrences, unless they are associated with microcalcification. By contrast, MRI will detect nearly all recurrences. Post-treatment changes can be marked, especially after radiotherapy, with persistent reactive enhancement around the scar for up to 12–18 months. Nonetheless, breast MRI can be extremely useful even as soon as nine months after treatment, since the very high negative predictive value of a negative scan effectively precludes scar recurrence.
Breast MRI is also the method of choice in the evaluation of breast implants. The accuracy of MRI for implant rupture, both intra- and extracapsular, far exceeds that of conventional imaging. Newer MRI units are all capable of producing high-resolution “silicon-only” sequences, where only silicon returns high signal. MRI is also extremely sensitive in the detection of malignancy in the augmented breast, which is notoriously difficult with conventional imaging. Therefore, it is likely to become the method of choice for surveillance and problem solving in the reconstructed breast after oncoplastic procedures (Figure 4).
Perhaps the area in which there has been most interest in breast MRI recently is in screening women at very high risk of breast cancer. A number of studies have now reported from Europe and the US, and it is clear that MRI is capable of detecting breast cancers that are otherwise occult.(24-26) The MARIBS study (MAgnetic Resonance Imaging for Breast Screening) looked at women under 50 with genetic mutations. The results were striking, especially for women with BRCA 1 mutations, where MRI detected many more cancers than mammography alone (sensitivity 77% and 40% respectively). The Dutch group have reported similar results.(26) Of critical importance, the MRI-detected cancers were generally high grade but small and lymph node negative, suggesting that there are likely to be survival benefits from their early detection. However, it is not known whether MRI has benefits in women at lower risk or in women with breasts that are difficult to evaluate by conventional means.(27)
Perspectives on breast MRI
The sensitivity of breast MRI for breast cancer detection compared with conventional techniques is beyond doubt, but its uncritical use must be questioned, since this will inevitably lead to increased numbers of unnecessary MRI-guided and open surgical biopsies. Attention to all the factors discussed above will go some way to prevent this. The development of blood pool contrast agents and, ultimately, tumour-specific contrast agents will increase the specificity of the examination and improve the positive predictive value of MRI biopsy recommendations, which at the moment is relatively low. Proton spectroscopy (1H MRIS) may also have a role in this context. Levels of choline compounds are elevated in breast cancers to a variable degree, and 1H MRS may be useful not only in differentiating between benign and malignant breast lesions (and hence decreasing the benign biopsy rate), but also potentially in monitoring response to neoadjuvant chemotherapy.(28,29) Computer-aided detection systems and software packages evaluating pharmacokinetics are appearing, though these are no substitute for a sound understanding of the principles behind breast MRI. It is clear that the role of breast MRI can only expand, but for the foreseeable future it will be used in conjunction with mammography, which remains the only diagnostic method proven to reduce mortality from breast cancer.
- Kaiser WA, Zeitler E. Radiology 1989;170:681-6.
- Heywang-Köbrunner SH, Hahn D, Schmid H, et al. J Comput Assist Tomogr 1986;10:199-204.
- Heywang-Köbrunner SH, Haustein J, Pohl C et al. Radiology 1994;191:639-46.
- Morris EA, Harms S. ACR practical guideline for the performance of magnetic resonance imaging (MRI) of the breast. ACR 2004, 269-74.
- ACR. Breast Imaging and Reporting Data System, Breast Imaging Atlas. Reston, VA: ACR, 2003.
- Fischer U, Kopka L, Grabbe E. Radiology 1999;213:881-8.
- Liberman L, Morris EA, Dershaw DD, et al. Am J Roentgenol 2003;180:901-10.
- Bedrosian I, Mick R, Orel SG et al. Cancer 2003;98:468-73.
- Schelfout K, Van Goethem M, Kersschott E, et al. Eur J Surg Oncol 2004;30:501-7.
- Morris EA, Schwartz LH, Drotman MB, et al. Radiology 2000;214:67-72.
- Neubauer H, Li M, Kuehne-Heid R, et al. Br J Radiol 2003;76:3-12.
- Ikeda O, Nishimura R, Miyayama H, et al. Acta Radiol 2004;45:721-5.
- Fischer U, Zachariae O, Baum F et al.Eur Radiol 2004;14:1725-31.
- Boetes C, Veltman J, van Die L, et al. Breast Cancer Res Treat 2004;86:31-7.
- Weinstein SP, Orel SG, Heller R, et al. Am J Roentgenol 2001;176:399-406.
- Kneeshaw PJ, Turnbull LW, Smith A, et al. Eur J Surg Oncol 2003;29:32-7.
- Lee SG, Orel SG, Woo IJ, et al. Radiology 2003;226:773-8.
- Wiener JI, Schilling KJ, Adami C, et al. Am J Roentgenol 2005;184:878-86.
- Orel SG, Weinstein SP, Schnall MD, et al. Radiology 1999;212:543-9.
- Olson JA, Morris EA, Van Zee KJ, et al. Ann Surg Oncol 2000;7:404-5.
- Partridge SC, Gibbs JE, Lu Y, et al. Am J Roentgenol 2002;179:1193-9.
- Lee JM, Orel SG, Czerniecki BJ, et al. Am J Roentgenol 2004;182:473-80.
- Orel SG, Reynolds C, Schnall MD, et al. Radiology 1997;205:429-36.
- Leach MO, Boggis CR, Dixon AK, et al.Lancet 2005;365:1769-78.
- Warner E, Plewes DB, Hill KA, et al. JAMA 2004;292:1317-25.
- Kriege M,Brekelmans CT, Boetes C, et al. N Engl J Med 2004;351:427-37.
- Morris EA,Liberman L, Ballon DJ, et al. Am J Roentgenol 2003;181:619-26.
- Bartella L, Morris EA, Dershaw DD, et al. Radiology 2006;239:686-92.
- Meisami S, Bolan PJ, Baker EH, et al.Radiology 2004;233:424-31.