(Radiology. 2001;220:13-30.)
© RSNA, 2001
MR Imaging of the Breast for the Detection, Diagnosis, and Staging of Breast Cancer1
Susan G. Orel, MD and
Mitchell D. Schnall, MD
1 From the Department of Radiology, Hospital of the University of Pennsylvania, 3400 Spruce St, Philadelphia, PA 19104. Received June 12, 2000; revision requested July 17; revision received September 14; accepted October 2. Address correspondence to S.G.O. (e-mail: orel@oasis.rad.upenn.edu).
 |
ABSTRACT
|
|---|
With the introduction of contrast agents, advances in surface coil technology, and development of new imaging protocols, contrast agent–enhanced magnetic resonance (MR) imaging has emerged as a promising modality for detection, diagnosis, and staging of breast cancer. The reported sensitivity of MR imaging for the visualization of invasive cancer has approached 100%. There are many examples in the literature of MR imaging–demonstrated mammographically, sonographically, and clinically occult breast cancer. Often, breast cancer detected on MR images has resulted in a change in patient care. Despite these results, there are many unresolved issues, including no defined standard technique for contrast-enhanced breast MR imaging, no standard interpretation criteria for evaluating such studies, no consensus on what constitutes clinically important enhancement, and no clearly defined clinical indications for the use of MR imaging. Furthermore, this technology remains costly, and issues of cost-effectiveness and cost competition from percutaneous biopsy have yet to be fully addressed. These factors along with the lack of commercially available MR imaging–guided localization and biopsy systems have slowed the transfer of this imaging technology from research centers to clinical breast imaging practices. Technical requirements, potential clinical applications, and potential pitfalls and limitations of contrast-enhanced MR imaging as a method to help detect, diagnose, and stage breast cancer will be described.
Index terms: Breast, MR, 00.121411, 00.121412, 00.121415, 00.12143 • Breast neoplasms, diagnosis, 00.31, 00.32, 00.81 • Breast neoplasms, MR, 00.121411, 00.121412, 00.121415, 00.12143 • Breast neoplasms, staging • State of the Art
|
INTRODUCTION
|
|---|
Over the past 2 decades, tremendous advances have been made in the field of magnetic resonance (MR) imaging of the breast—from early experience, prior to use of contrast agents, that suggested that MR imaging had little to offer as a breast imaging method, to more recent experience that has demonstrated not only that breast cancer can be visualized on MR images but also that mammographically and clinically occult breast cancer can be detected on MR images. Developments have been made in surface coil technology, resulting in improved spatial resolution, and in imaging protocols, resulting in improved temporal resolution. Development of MR-guided breast biopsy techniques continues.
Even with all of the advances that have been made, however, MR imaging of the breast as a method to detect and stage breast cancer is a technology in development. Several unresolved issues remain. There is, as yet, no defined "standard" or "optimal" technique for the performance of contrast agent–enhanced MR imaging of the breast. There are no standard interpretation criteria for evaluating breast MR imaging studies and no unifying definition of what constitutes potentially clinically important contrast enhancement. These factors, along with the lack of commercially available MR imaging–guided localization and biopsy systems, have slowed the transfer of this imaging technology from research centers to clinical breast imaging practices.
There are also other unresolved issues. There are no well-established indications for MR imaging of the breast. When should MR imaging be used? Should MR imaging be used only as an adjunctive diagnostic test or should it be used for breast cancer screening? Which patients would benefit most from MR imaging? Are breast cancers detected only with MR imaging of clinically important cancers? Even if MR imaging can be used to detect breast cancer, will this expensive imaging test be cost-effective? Clinical investigations addressing these and many other questions continue. The goal of this article is to describe the technical factors involved in performing contrast-enhanced breast MR imaging; to review the multitude of proposed methods for interpreting these MR studies; and to discuss potential clinical applications, pitfalls, and limitations of contrast-enhanced MR imaging as an imaging tool to help detect, diagnose, and stage breast cancer.
|
BREAST MR IMAGING TECHNIQUE
|
|---|
Hardware
To date, most of the breast MR imaging studies that have been reported in the literature have been performed with high-field-strength (1.0–1.5-T) MR imaging systems (1–11). The open architecture and low cost of lower field-strength open MR systems is potentially very appealing for breast MR imaging (12). There are, however, several factors that are encountered when attempting to perform contrast-enhanced breast MR imaging at lower field strength. One concern is the signal-to-noise ratio. Low-field-strength systems provide an inherently lower signal-to-noise ratio per unit time of sampling data. In most applications of low-field-strength imaging, a lower imaging bandwidth is used to provide adequate signal-to-noise ratio for imaging. The lower imaging bandwidth will result in longer acquisition times. Therefore, the need to compromise spatial or temporal resolution is exacerbated to a large extent at lower field strengths. A second issue that is encountered at lower field strengths is the issue of fat suppression. The separation between fat and water peaks, in absolute frequency, decreases as field strength decreases. This makes spectrally selected fat suppression extremely difficult to perform. A third and perhaps most important aspect of low-field-strength imagers with respect to breast MR imaging is the difference in contrast characteristics. The T1 times of tissues tend to decrease as field strength decreases. There is much less change in the relaxivity of gadolinium chelates as field strength decreases. Therefore, the difference in contrast between native tissue and tissue enhanced with a given concentration of gadolinium chelate for a given T1-weighted imaging sequence will be less at the lower field strengths. This has the potential to result in the lack of demonstration of enhancing structures.
Although these issues do not preclude performance of contrast-enhanced breast MR imaging at lower field strengths, they do raise the concern that breast MR imaging performed at a field strength as low as 0.2–0.3 T may not simulate the high-field-strength experience.
As the demand for breast MR imaging increases, dedicated breast MR systems, which would be more accessible and potentially less costly than breast imaging with a standard 1.5-T system, may be needed in the future. However, additional technical improvements in dedicated breast MR imaging systems along with continued clinical investigation to compare the results of imaging at high field strength with those at middle to low field strength are needed before widespread clinical implementation of a dedicated breast MR imaging system.
Breast Coils
Breast MR imaging is typically performed with the patient in the prone position, lying on a platform placed in the MR imager that allows the breast to extend dependently from the patient (11,13–15). Prone patient positioning minimizes the effects of respiratory motion and is preferred to supine imaging, which results in lower image quality.
A dedicated breast surface coil should be used when performing breast MR imaging. Many types of surface coils are available (11,13–15). These include both single-breast and bilateral-breast designs. These coils are also available commercially as single coils or phased arrays. In our experience, the performance of commercially available phased-array bilateral-breast coils is better than that of non–phased-array coils. There are some coils that have been designed in such a way as to provide access to the breast while the patient is lying on the coil (12,13). This provides the opportunity for follow-up MR-guided breast biopsy, if necessary.
Most coil designs rely on a solenoid or solenoid-like coil design that allows imaging of the breast in its natural dependent state. An alternative design consists of mild compression of the breast in the medial to lateral direction between two plates, each of which contains a two-coil phased array (13). This design offers substantial advantages over conventional breast coils. It reduces the dimension of the breast in the medial to lateral direction, borrowing approximately half of the number of sections to image the breast in the sagittal orientation. In addition, this design allows placement of the center of the breast closer to the radio-frequency coil elements, resulting in an increase in signal-to-noise ratio by a factor of approximately 3 over that with a conventional solenoid-like breast coil design. Another is that this design stabilizes the breast and is a natural configuration for MR imaging–guided biopsy.
Use of Paramagnetic Contrast Agents
The cornerstone of breast MR imaging is contrast-enhanced imaging. Most investigators report the use of a contrast agent dose of 0.1 mmol per kilogram of body weight. While the results of one study (16) suggested that the conspicuity of malignant lesions was improved at a gadolinium dose of 0.16 mmol/kg as compared with a dose of 0.1 mmol/kg, there are as yet no large-scale well-controlled studies in which a higher dose of contrast agent has been shown to lead to greater diagnostic accuracy.
The contrast agent is injected intravenously, usually as a bolus, and is followed by a saline flush. To ensure that contrast-enhanced images can be obtained immediately after the contrast agent injection, tuning and gain adjustments should be performed before the injection and should not be readjusted for the remainder of the contrast-enhanced sequence.
Imaging Protocols
Protocols for breast MR imaging vary dramatically (11,15,17). Although the hallmark of breast MR imaging is contrast-enhanced imaging, T2-weighted images can be valuable in establishing the diagnosis of a cyst or fibroadenoma (4,18). Therefore, it is recommended that a fast spin-echo T2-weighted sequence through the breast be acquired routinely prior to administration of contrast material. While in many cases, the T2-weighted image will not be valuable, this will not be known until the contrast-enhanced image is viewed. The T2-weighted sequence is best obtained with fat suppression and a section thickness of approximately 3 mm.
The common element in contrast-enhanced breast MR imaging is acquisition of T1-weighted images before and after intravenous administration of contrast material. With the use of paramagnetic contrast agents and gradient-echo sequences, both two-dimensional and three-dimensional (3D), have largely replaced spin-echo sequences (15,17). Gradient-echo sequences are more time efficient than spin-echo sequences, allowing for faster imaging times. In addition, gradient-echo sequences offer inherently better T1 contrast (with a given number of sections, T1-weighted imaging allows use of much shorter repetition and echo times) and are much more sensitive to the T1-shortening effects of gadolinium chelates than are spin-echo sequences. Although the use of both two-dimensional and 3D imaging sequences has been described, for reliable detection of small (<5-mm) lesions it is generally agreed that high-spatial-resolution 3D techniques should be used to provide large area coverage with thin (
3-mm) sections with no intersection gap (11,15,17).
Beyond these general guidelines, a wide variety of acquisition schemes have been proposed. Some investigators (1,2, 5–10,19,20) have developed contrast-enhancement acquisition schemes that emphasize temporal resolution. These dynamic techniques are designed to study enhancement kinetics after contrast agent administration. Dynamic techniques with time resolution varying from a few seconds to a minute have been described. Others (3,4) have stressed the importance of high spatial resolution to detect small lesions and characterize lesion architecture. High-spatial-resolution acquisitions with in-plane spatial resolution as high as 300 µm have been described (4).
Although for many years, the merging of these two schools of thought was difficult, advances in gradient coil technology have reduced the trade-offs between spatial and temporal resolution. These technical advances have led to a convergence in the temporal- and spatial-resolution schools of thought. It is possible to achieve relatively high spatial resolution with a temporal resolution of approximately 1–2 minutes. Today, typical contrast-enhanced breast MR imaging acquisitions have a temporal resolution on the order of 1 minute, with an in-plane pixel size of less than 1 mm and a section thickness of approximately 3 mm or less. With a modern imager with high-performance gradients, a fat-suppressed 256 x 192 x 60-pixel volume can be obtained in approximately 1 minute. This allows reasonably high-resolution images of the breast to be obtained in a time frame that allows one to perform at least a qualitative analysis of enhancement kinetics. From this compromise type of protocol, other protocols can be designed for higher spatial resolution at the expense of temporal resolution or higher temporal resolution at the expense of spatial resolution. This decision is based somewhat on individual preferences for image interpretation with regard to the relative performance of kinetic information versus architectural features.
The use of more novel acquisition sequences such as nonrectilinear k-space sampling and echo-planar imaging may provide the ability to reduce the need for a compromise between spatial and temporal resolutions by improving total acquisition speed (19–22). Currently, these sequences are not commercially available for breast MR imaging. Over time, it is anticipated that these acquisition techniques will provide both high-spatial- and high-temporal-resolution images of the breast and may ultimately dominate MR imaging protocols.
Plane of Acquisition
When imaging a single breast, any acquisition plane can be used, although the sagittal plane appears to be the most popular. This is probably due to the relative ease of correlating a finding identified on a sagittal MR image with a finding on a mediolateral oblique or mediolateral mammographic view. Bilateral examinations are most often performed in the coronal or transverse plane, with a rectangular field of view oriented with the long axis in the medial-to-lateral direction. Imaging in the transverse plane is advantageous, relative to imaging in the coronal plane, because a finding on a transverse MR image can then be correlated with a similarly positioned craniocaudal mammographic view. Newer acquisition schemes that can be used to perform concurrent slab-interleaved 3D acquisitions of both breasts in the sagittal plane have been described (23) and offer potential efficiencies by not imaging the empty space between the breasts.
Fat Suppression
Unlike mammography, in which lesion conspicuity is enhanced against a fatty background, at MR imaging an enhancing lesion may not be discernible because it becomes isointense to fat. Suppression of the fat signal can be attained in two ways: (a) "active" fat suppression, where the fat signal is suppressed prior to the injection of contrast agent, and (b) "passive" fat suppression with postprocessing image subtraction.
Multiple active fat-suppression schemes have been described (11,15). Standard presaturation of fat during every repetition time can be used; however, this results in a dramatic lengthening of examination time over that of the standard non–fat-suppressed imaging protocol. Several more time-efficient alternatives have been described. The rotating delivery of excitation off resonance, or RODEO, sequence uses a spectrally selective excitation pulse to excite only the water in the breast at 3D gradient-echo breast MR imaging (3). Because this radio-frequency pulse is only spectrally selective, the imaging volume is selected by using a coil that has sensitivity restricted to the breast. This is best performed with a coil that both transmits and receives. A robust, time-efficient, fat-suppression technique is the spectrally selective inversion described by Foo et al (24). For this technique, a partial inversion pulse that is spectrally selective for fat is applied intermittently throughout the pulse sequence to suppress the signal from fat. With typical implementations of this technique, fat suppression can be robustly achieved with only a minor increase in acquisition time. This technique can easily be implemented with any coil and any 3D gradient-echo sequence. The major advantages of active fat suppression are insensitivity to patient motion and ease of lesion detection, with enhancing lesions appearing bright against a dark background. The major disadvantage of fat suppression is sensitivity to field inhomogeneity, which may preclude satisfactory fat suppression in some cases.
Fat suppression can also be achieved by means of postprocessing image subtraction (passive fat suppression). Identical imaging parameters before and after contrast agent administration are required. Subtracted images will yield the difference in signal intensity between the unenhanced and enhanced images, directly proportional to the degree of enhancement. Two advantages of image subtraction are minimized acquisition times, because inversion-recovery and saturation pulses are not used, and insensitivity to field inhomogeneity. The major disadvantages of image subtraction are decreased signal-to-noise ratio and sensitivity to patient motion, which can result in misregistration.
|
BREAST MR PROTOCOL OF THE HOSPITAL OF THE UNIVERSITY OF PENNSYLVANIA
|
|---|
As an example, we present the clinical breast MR imaging protocol ("high-resolution" single-breast examination) used at the Hospital of the University of Pennsylvania (Philadelphia).
Patients undergo imaging in the prone position with a four-coil compression breast array (13). Compression is applied gently so as not to cause patient discomfort. A transverse localizing image is acquired, followed by a sagittal T2-weighted fast spin-echo acquisition with fat suppression. The fast spin-echo acquisition is performed over an 18-cm field of view by using a 256 x 256 matrix and an echo train length of 16. The acquisition bandwidth is 32 kHz. Section thickness is 3 mm with a 1-mm intersection gap.
Nonenhanced and enhanced spoiled gradient-echo (SPGR) images are then obtained. The images are obtained by using a 3D SPGR sequence with intermittent fat-selective partial inversion for fat suppression. Images are obtained over a 512 x 256 x 32 matrix in the sagittal plane. The typical section thickness is 2.0–2.5 mm, depending on the size of the breast. The volume is adjusted so that the 32 sections completely fill the dimensions of the gently compressed breast. The field of view varies from 16 to 18 cm. A repetition time of 9.2 msec and an echo time of 2.1 msec (9.2/2.1) are used. The fat-selective inversion pulse is applied twice every pass through the z axis of k space, or every 16 repetition times. Images are acquired before administration of the contrast agent. Contrast-enhanced imaging is initiated simultaneously with completion of an injection of 20 mL of gadolinium chelate. The injection is then followed by a saline flush, which occurs during contrast-enhanced imaging. The total imaging time is 90 seconds. Two sequential contrast-enhanced studies are obtained.
Potential advantages of this imaging protocol include the compression multicoil array, which permits imaging of the entire breast with a decreased number of sections, and a fat-suppression technique that is time efficient such that images can be obtained with high spatial resolution and relatively high temporal resolution. The major disadvantage of this protocol is that only one breast can be imaged at a time. Ultimately, bilateral-breast imaging will be preferable in many clinical situations, including breast cancer staging and screening of patients at high risk. We are currently investigating an imaging protocol that permits high-resolution fat-suppressed imaging of both breasts in the sagittal plane (23).
|
MR IMAGING–GUIDED BIOPSY
|
|---|
Similar to the requirement for mammographically and ultrasonographically (US)-guided localization and biopsy systems for obtaining a histologic diagnosis of clinically occult lesions, an MR imaging–guided localization and biopsy system is needed to obtain a histologic diagnosis of lesions detected on MR images that are mammographically, sonographically, and clinically occult (Figs 1, 2). At a May 1, 1997, meeting of the Federal Multi-Agency Consortium on Imaging Technologies to Improve Women’s Health, the lack of MR imaging–guided localization and biopsy technology was cited as one of the major technical challenges for breast MR imaging (25).
View larger version (171K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1a. MR imaging-guided wire localization in a patient with a palpable mass shown to be carcinoma at fine-needle aspiration biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21/2.4) reveal not only the mammographically visible and clinically palpable carcinoma (arrow on left) in the upper outer quadrant but also an enhancing mass (arrow on right) in the subareolar portion of the breast. (b) Sagittal fat-suppressed contrast-enhanced gradient-echo MR image (21/2.4) shows imaging-guided wire localization. Arrow = signal void from MR-compatible titanium needle. (c) Mediolateral and (d) craniocaudal mammographic views reveal wire placed with MR imaging guidance. No lesion is seen along the wire. The known cancer (arrow) can be identified in the upper outer quadrant. Excisional biopsy revealed a 1.5-cm invasive ductal carcinoma along the wire and a 1.5-cm invasive ductal carcinoma in the upper outer quadrant.
|
|
View larger version (160K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1b. MR imaging-guided wire localization in a patient with a palpable mass shown to be carcinoma at fine-needle aspiration biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21/2.4) reveal not only the mammographically visible and clinically palpable carcinoma (arrow on left) in the upper outer quadrant but also an enhancing mass (arrow on right) in the subareolar portion of the breast. (b) Sagittal fat-suppressed contrast-enhanced gradient-echo MR image (21/2.4) shows imaging-guided wire localization. Arrow = signal void from MR-compatible titanium needle . (c) Mediolateral and (d) craniocaudal mammographic views reveal wire placed with MR imaging guidance. No lesion is seen along the wire. The known cancer (arrow) can be identified in the upper outer quadrant. Excisional biopsy revealed a 1.5-cm invasive ductal carcinoma along the wire and a 1.5-cm invasive ductal carcinoma in the upper outer quadrant.
|
|
View larger version (102K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1c. MR imaging-guided wire localization in a patient with a palpable mass shown to be carcinoma at fine-needle aspiration biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21/2.4) reveal not only the mammographically visible and clinically palpable carcinoma (arrow on left) in the upper outer quadrant but also an enhancing mass (arrow on right) in the subareolar portion of the breast. (b) Sagittal fat-suppressed contrast-enhanced gradient-echo MR image (21/2.4) shows imaging-guided wire localization. Arrow = signal void from MR-compatible titanium needle . (c) Mediolateral and (d) craniocaudal mammographic views reveal wire placed with MR imaging guidance. No lesion is seen along the wire. The known cancer (arrow) can be identified in the upper outer quadrant. Excisional biopsy revealed a 1.5-cm invasive ductal carcinoma along the wire and a 1.5-cm invasive ductal carcinoma in the upper outer quadrant.
|
|
View larger version (93K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1d. MR imaging-guided wire localization in a patient with a palpable mass shown to be carcinoma at fine-needle aspiration biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21/2.4) reveal not only the mammographically visible and clinically palpable carcinoma (arrow on left) in the upper outer quadrant but also an enhancing mass (arrow on right) in the subareolar portion of the breast. (b) Sagittal fat-suppressed contrast-enhanced gradient-echo MR image (21/2.4) shows imaging-guided wire localization. Arrow = signal void from MR-compatible titanium needle . (c) Mediolateral and (d) craniocaudal mammographic views reveal wire placed with MR imaging guidance. No lesion is seen along the wire. The known cancer (arrow) can be identified in the upper outer quadrant. Excisional biopsy revealed a 1.5-cm invasive ductal carcinoma along the wire and a 1.5-cm invasive ductal carcinoma in the upper outer quadrant.
|
|
View larger version (169K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. MR imaging-guided core biopsy in a patient with a palpable mass shown to be invasive ductal carcinoma at US-guided core biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21.2/2.1) reveal two enhancing lesions, one corresponding to the known cancer (solid arrow) and a second lesion (open arrow) that was occult at mammography and US. (b) MR imaging-guided core biopsy of the additional lesion identified only at MR imaging was performed by using a 14-gauge disposable biopsy gun (Monopty; Bard Urological, Covington, Va). These biopsy results demonstrated invasive ductal carcinoma. Arrow = introducer needle of biopsy gun.
|
|
View larger version (82K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. MR imaging-guided core biopsy in a patient with a palpable mass shown to be invasive ductal carcinoma at US-guided core biopsy. (a) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR images (21.2/2.1) reveal two enhancing lesions, one corresponding to the known cancer (solid arrow) and a second lesion (open arrow) that was occult at mammography and US. (b) MR imaging-guided core biopsy of the additional lesion identified only at MR imaging was performed by using a 14-gauge disposable biopsy gun (Monopty; Bard Urological, Covington, Va). These biopsy results demonstrated invasive ductal carcinoma. Arrow = introducer needle of biopsy gun.
|
|
Currently, MR imaging–guided breast localization and biopsy are performed by using dead-reckoning navigation similar to that used for stereotactic biopsy. Three-dimensional images of the breast are obtained with fiducial markers (reference markers containing copper sulfate) either on the breast or on a coil, to localize lesions within the breast (Fig 3). The needle position is then calculated relative to the fiducial marker for proper placement of the needle into the lesion. The patient is withdrawn from the bore of the magnet, and the needle (MR-compatible titanium) is placed. Several prototype breast biopsy coils have been designed that include either a large open window for placement of the needle or a grid of holes that can be used to guide needle placement (26–32). The patient can then be returned into the MR system for confirmation of the needle position. Once the needle position is confirmed, sampling or wire placement can be performed.
View larger version (97K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. MR imaging biopsy and localization device. The breast is compressed between a fixed medial plate and a mobile lateral plate. The lateral plate, made of polycarbonate plastic (Lexan; GE Plastics, Pittsfield, Mass), is removable and can be sterilized. Copper sulfate reference markers (arrows) are placed arbitrarily into holes in the plate.
|
|
In addition to the approach described in the preceding paragraph, the open architecture of MR imagers such as the Signa SP (GE Medical Systems, Milwaukee, Wis) allows interactive placement of needles directly with MR guidance. By using an open-architecture MR coil, a needle can be interactively placed toward the lesion with direct MR guidance. This approach has been reported by a Stanford University (Calif) group (33), who have successfully performed needle localizations and core-needle biopsies with MR imaging guidance.
MR imaging–compatible needles are available from many manufacturers. There are still multiple problems associated with MR needles, such as extensive artifact on gradient-echo images and poor performance of core biopsy guns. Continued development of higher quality needles and core biopsy guns are important to the further development of MR imaging–guided breast biopsy.
There are several limitations of MR imaging–guided localization. One limitation with many systems is access to only the lateral side of the breast. This often will not be the shortest path to the lesion and could result in surgical removal of additional breast tissue and a potentially suboptimal cosmetic result. A system that would permit easy access to at least the medial side of the breast and potentially to any part of the breast would be advantageous. A second limitation of MR imaging–guided localization is the inability to verify lesion removal in many cases (34). In mammography, imaging of the specimen is routinely performed to document lesion removal. In contrast, MR imaging of the specimen is not feasible, since most lesions are visualized only transiently after contrast agent administration, and the specimen is avascular. A tissue marker such as a titanium clip that could be placed into the lesion at the time of MR-guided localization is being investigated. This would allow documentation of lesion removal. Both of these limitations could be overcome with an MR imaging–guided core biopsy system. The results of a preliminary investigation (35) with such systems, including a vacuum-assisted device, have been promising.
Up to now, MR imaging–guided localization and biopsy systems have been available almost exclusively at research centers. It is only recently that such systems have become commercially available. Continued technical developments in both localization and biopsy systems will be critical to the ultimate success of MR imaging guidance for successful biopsy of cancers that can be identified only at MR imaging.
|
IMAGE INTERPRETATION: WHAT CONSTITUTES CLINICALLY IMPORTANT ENHANCEMENT?
|
|---|
In addition to the many technical factors involved in performing breast MR imaging, another potential barrier to the widespread adoption of MR as a breast imaging method is the absence of guidelines for interpreting MR studies. After intravenous administration of a contrast agent, it may appear that enhancing "lesions" are present throughout the breast. How are the potentially clinically important lesions identified?
At present, there are no standard interpretation criteria for evaluating breast MR studies. There are probably as many definitions of suggestive enhancement as there are investigators. Each investigator has chosen a specific definition of suggestive enhancement that, on the basis of their experience, yields the highest sensitivity and specificity for the detection of breast cancer. This, in turn, is reflective of multiple factors, including obtainable temporal and spatial resolution, patient population, and histologic features of lesions, as well as investigator preferences.
There have been two major approaches to image interpretation: (a) evaluation of enhancement kinetics following contrast agent administration and (b) evaluation of lesion morphology. The results of early studies (1,2) in Europe conducted after the introduction of intravenous gadolinium chelates generated considerable enthusiasm for a quantitative assessment of contrast enhancement. The authors of these studies demonstrated that malignant lesions consistently enhance after the administration of gadolinium chelate and tend to enhance earlier and to a greater degree than do benign lesions (1,2). On the basis of the promising results of these studies, it was suggested that enhancement kinetics could be used to identify suggestively enhancing lesions and differentiate these from benign enhancement. Several empiric measurements of enhancement have been used, including the maximum rate of enhancement (slope of enhancement uptake) and the increase in signal intensity after contrast agent administration (1,2,5–10,19,20,36–40).
Recommendations for the interval at which the measurements should be performed and the optimal threshold level above which enhancement should be considered suggestive have varied widely. Kaiser and Zeitler (2) classified a lesion as suspicious if the lesion demonstrated a 100% increase in signal intensity within the first 2 minutes after contrast agent injection (2). Heywang-Kobrunner (1) considered enhancement above 300 normalized units to be important. Gribbestad et al (6) defined important enhancement as a greater than 70% increase in signal intensity after 60 seconds. Gilles et al (7) defined enhancement concomitant with early vascular enhancement as suggestive. Boetes et al (5) defined enhancement within 11.5 seconds after aortic enhancement as suggestive. Kuhl et al (39,40) classified a lesion as suspicious if it demonstrated a greater than 60% increase in signal intensity that was apparent on the first contrast-enhanced image (40 seconds). Stomper et al (41) defined enhancement that is two or more times greater than the unenhanced signal intensity as suggestive. Quantitative methods have also included sophisticated physiologic modeling of uptake and washout of contrast agents (4,22,42,43).
In contrast to these quantitative methods, in which one or more kinetic measurements are used, a qualitative method has been described for evaluation of the overall shape of the enhancement curve (39,40). This type of approach permits visual inspection of the enhancement curve as an alternative to complex mathematical models or the need to quantify the amount or rate of absolute enhancement. Three types of time-intensity curves have been described: type I (steady enhancement), where a persistent increase in signal intensity is present beyond 2 minutes after contrast agent injection; type II (plateau), where the maximum signal intensity is achieved in the first 2 minutes and then remains fairly constant; and type III (washout), where the maximum signal intensity is achieved in the first 2 minutes and then decreases over time. It has been reported that benign lesions tend to exhibit a type I curve, and malignant lesions tend to exhibit a type III curve. It has also been reported (39,40) that evaluation of these curves results in improved specificity and accuracy, as compared with the evaluation of enhancement rates.
For all of these approaches in which enhancement kinetics are evaluated—both quantitative and qualitative methods—accurate placement of a region of interest over the area(s) of most rapid and intense enhancement is critical. Statistically significant interobserver variability and bias in placement of the region of interest have been reported, although this variability can be improved when a semiautomated method with parametric images is used (44). In addition, placement of a region of interest and generation of a time-intensity curve for both ductal enhancement (ie, seen in ductal carcinoma in situ [DCIS]) and regional enhancement (ie, seen in both benign and malignant lesions) may not be as accurate as those for solid masses.
The second major approach to image interpretation has been the evaluation of lesion morphology. Although the authors of early reports were optimistic that the evaluation of enhancement kinetic measurements could be used to differentiate malignant from benign lesions, results from later studies demonstrated considerable overlap in the enhancement characteristics of malignant and benign lesions, which is likely due to histologic variability (4,38,41,45,46). On the basis of this overlap, it was suggested (4) that architectural features identified on high-spatial-resolution images could be used to characterize lesions as likely malignant or likely benign (Fig 4). Architectural features that have been reported as suggestive of malignancy include a mass with irregular or spiculated borders, a mass with peripheral enhancement, and ductal enhancement (4,7,45–52). Architectural features that have been reported suggestive of a benign process include a mass with smooth or lobulated borders, a mass demonstrating no contrast enhancement, a mass with nonenhancing internal septa, and patchy parenchymal enhancement (3,4,45,53). A decision tree incorporating several architectural features has been described (54). Because reader variability remains a concern, an imaging lexicon, similar to the Breast Imaging Reporting and Data System, or BI-RADS, lexicon used in mammography in which the architectural features are defined and illustrated is needed. Development of such a lexicon is in progress (55).
View larger version (168K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. Lesion morphology. (a) Linear enhancement of DCIS. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an area of linear enhancement (arrows) in a patient with bloody nipple discharge. MR imaging-guided wire localization revealed DCIS. (b) Peripheral enhancement in invasive carcinoma. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with a palpable mass reveals a lesion (arrows) that is enhancing around the periphery. Invasive ductal carcinoma was found at excisional biopsy. (c) Internal septa in fibroadenoma. Sagittal fat-suppressed T2-weighted fast spin-echo MR image (4,000/120) in a patient with a developing mass identified at mammography reveals an enhancing lesion (arrow) containing nonenhancing septa. Fibroadenoma was found at excisional biopsy. (d) Regional enhancement. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with newly diagnosed invasive breast cancer identified at mammography and physical examination reveals an area of regional enhancement (arrows) that was adjacent to the known breast cancer. At excisional biopsy, extensive DCIS was found adjacent to the known invasive breast cancer.
|
|
View larger version (152K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. Lesion morphology. (a) Linear enhancement of DCIS. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an area of linear enhancement (arrows) in a patient with bloody nipple discharge. MR imaging-guided wire localization revealed DCIS. (b) Peripheral enhancement in invasive carcinoma. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with a palpable mass reveals a lesion (arrows) that is enhancing around the periphery. Invasive ductal carcinoma was found at excisional biopsy. (c) Internal septa in fibroadenoma. Sagittal fat-suppressed T2-weighted fast spin-echo MR image (4,000/120) in a patient with a developing mass identified at mammography reveals an enhancing lesion (arrow) containing nonenhancing septa. Fibroadenoma was found at excisional biopsy. (d) Regional enhancement. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with newly diagnosed invasive breast cancer identified at mammography and physical examination reveals an area of regional enhancement (arrows) that was adjacent to the known breast cancer. At excisional biopsy, extensive DCIS was found adjacent to the known invasive breast cancer.
|
|
View larger version (120K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c. Lesion morphology. (a) Linear enhancement of DCIS. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an area of linear enhancement (arrows) in a patient with bloody nipple discharge. MR imaging-guided wire localization revealed DCIS. (b) Peripheral enhancement in invasive carcinoma. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with a palpable mass reveals a lesion (arrows) that is enhancing around the periphery. Invasive ductal carcinoma was found at excisional biopsy. (c) Internal septa in fibroadenoma. Sagittal fat-suppressed T2-weighted fast spin-echo MR image (4,000/120) in a patient with a developing mass identified at mammography reveals an enhancing lesion (arrow) containing nonenhancing septa. Fibroadenoma was found at excisional biopsy. (d) Regional enhancement. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with newly diagnosed invasive breast cancer identified at mammography and physical examination reveals an area of regional enhancement (arrows) that was adjacent to the known breast cancer. At excisional biopsy, extensive DCIS was found adjacent to the known invasive breast cancer.
|
|
View larger version (128K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4d. Lesion morphology. (a) Linear enhancement of DCIS. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an area of linear enhancement (arrows) in a patient with bloody nipple discharge. MR imaging-guided wire localization revealed DCIS. (b) Peripheral enhancement in invasive carcinoma. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with a palpable mass reveals a lesion (arrows) that is enhancing around the periphery. Invasive ductal carcinoma was found at excisional biopsy. (c) Internal septa in fibroadenoma. Sagittal fat-suppressed T2-weighted fast spin-echo MR image (4,000/120) in a patient with a developing mass identified at mammography reveals an enhancing lesion (arrow) containing nonenhancing septa. Fibroadenoma was found at excisional biopsy. (d) Regional enhancement. Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.2) in a patient with newly diagnosed invasive breast cancer identified at mammography and physical examination reveals an area of regional enhancement (arrows) that was adjacent to the known breast cancer. At excisional biopsy, extensive DCIS was found adjacent to the known invasive breast cancer.
|
|
It is becoming increasingly clear that while most investigators have used either enhancement kinetics or lesion morphology in the attempt to differentiate benign enhancing lesions from enhancing breast cancer, an integrated interpretation strategy where enhancement kinetics data and morphologic feature analysis used together for image interpretation may be superior to the use of either method alone (56).
Although individual investigators have set criteria for what constitutes a potentially malignant enhancing lesion, one must remember that these investigators used a wide variety of imaging sequences. The reported results with one technique will not necessarily be transferable to other MR techniques, and no single set of interpretation criteria are uniformly accepted at the present time. There clearly are many possible methods for interpreting breast MR studies. Which method(s) is optimal remains to be defined. As long as there continues to be wide variability in breast MR imaging systems, including variable magnetic field strengths, available pulse sequences, and variable surface coils, it may be that the optimal interpretation method will differ on the basis of available hardware and software. It may also be that the interpretation method will vary on the basis of the indication for the MR examination. With continued technical improvements, the potential exists for imaging with both high temporal and spatial resolution, without substantial compromise on either side, such that optimized kinetic and morphologic information can be obtained in one study.
Despite the present lack of consensus on how best to obtain breast MR images and in the absence of a unifying scheme for image interpretation, there is greater than ever agreement among investigators in terms of what is regarded as crucial for breast MR technique and also what criteria should be applied for lesion diagnosis. MR imaging is a robust technique that offers many ways to detect and classify breast lesions. Despite the fact that so many fundamentally different approaches have been taken and although so many different concepts of what constitutes clinically important enhancement have been presented, there is still almost universal agreement that breast MR imaging is an extremely sensitive imaging technique for the detection of breast cancer. Owing to the wide variety of imaging parameters and diagnostic criteria provided by breast MR imaging, standardization of technique and interpretation may take longer than with other breast imaging techniques.
|
POTENTIAL CLINICAL APPLICATIONS
|
|---|
There is a rapidly growing clinical demand for the use of breast MR imaging to evaluate patients with a breast-related problem that may not be adequately addressed with conventional imaging methods. While, at the present time, the clinical indications for MR imaging of breast as a method to help detect and diagnose breast cancer remain to be defined, several potential clinical applications have emerged on the basis of the results of clinical investigation thus far.
|
LESION DETECTION
|
|---|
Can MR imaging help decrease the number of breast biopsies performed for mammographically or clinically demonstrated abnormalities? One of the major limitations of mammography is overlap in the appearances of benign and malignant lesions. Approximately 75% of mammographically detected suspected or indeterminate lesions will prove to be benign at biopsy (57). A similar limitation of physical examination also results in the biopsy of many palpable benign lesions. An imaging method that could help reduce the number of breast biopsies performed for benign lesions is desirable, assuming that the improved specificity will not be at the cost of reduced sensitivity for the detection of cancer. Because investigators of breast MR imaging have reported higher sensitivities and specificities for MR imaging compared with those for mammography, it has been suggested that MR imaging could be used to further characterize indeterminate lesions detected at mammography, US, or physical examination. A lesion that appears unequivocally benign at MR imaging could then be followed rather than excised. The addition of MR imaging could, therefore, result in a decrease the number of biopsies performed for benign causes.
How has MR imaging performed in this role? It is not surprising, based on the plethora of potential interpretation methods, that comparison of specificity data reported in various series is difficult. Review of the literature yields a wide range of results, with reported specificities ranging from 37% to 97% (1–3,5,7, 8,19,36,39,51,58–60). This wide range is probably the result of multiple factors, including patient selection, patient age, lesion histologic characteristics, and variability in interpretation criteria.
The potentially limited specificity of MR imaging has been attributed to the fact that, in addition to cancer, many benign lesions as well as presumably normal breast tissue may enhance after administration of contrast material. Enhancement has been seen in many benign lesions, including fibroadenoma, proliferative and nonproliferative fibrocystic change, and mastitis, as well as increased-risk lesions such as radial scars, atypical ductal hyperplasia, and lobular carcinoma in situ (3,4,7,8,18,37,41,58,59). In addition, presumably normal breast tissue may enhance, and this has been shown to vary in degree at different times of the menstrual cycle (39,61). Enhancement of benign lesions along with presumably normal breast tissue has demonstrated overlap with enhancement of malignant lesions, both in terms of kinetic measurements and morphologic appearances, which precludes complete differentiation of benign from malignant lesions.
The primary benefit of a noninvasive test such as MR imaging undertaken prior to tissue diagnosis (ie, with core biopsy or excisional biopsy) is that it can be used to determine which lesions are likely to be benign, so that mammographic or clinical follow-up could be used in place of tissue diagnosis. When a test is used in this capacity, it must be sufficiently sensitive that the negative predictive value is very high (approaching 100%) without sacrificing specificity. In addition, if these goals are met, it will also have to be determined whether the use of the test is cost-effective. Does MR imaging meet these requirements? Can MR imaging be used to determine which lesions are likely benign and do not require tissue diagnosis, and will MR imaging be a cost-effective technology? The overlap in enhancement characteristics and morphologic appearances of benign and malignant lesions remains a problem and may limit the use of MR imaging in this capacity. If, however, after a large number of patients are examined, certain MR features are shown to be close to 100% reliable for diagnosis of lesions as benign and mammographic and clinical findings do not suggest that the lesion is highly likely to be malignant, then MR imaging findings could be used to support the use of mammographic or clinical follow-up rather than tissue diagnosis.
In terms of cost-effectiveness, MR imaging must compete with percutaneous breast biopsy. Stereotactic and US-guided biopsy have demonstrated high sensitivity and specificity, comparable to those of excisional biopsy, at approximately one-fourth the cost excisional biopsy (62–67). With the increasing availability of percutaneous biopsy, MR imaging with a less than perfect specificity and sensitivity may not prove to be a cost-effective method to help differentiate benign from malignant lesions. However, cost-effectiveness studies that take into account not only objective costs but also subjective costs (eg, patient preferences) are needed to determine whether follow-up based on benign findings at MR imaging is optimal in comparison with proceeding either to core biopsy or excisional biopsy. For a woman who does not want to undergo breast biopsy (either excisional or percutaneous), who has a history of multiple biopsy-proved fibroadenomas, and who develops a new palpable mass, MR imaging might prove to be clinically useful if the lesion can reliably be shown to be benign. In addition, in patients found to have breast cancer, while this diagnosis can be established with percutaneous core biopsy, MR imaging may be still be advantageous because it can be used not only to characterize the lesion in question but also to evaluate the remainder of the breast, potentially leading to identification of unsuspected multifocal disease. Additional clinical studies, including cost-effective analyses, to directly compare MR imaging and imaging-guided biopsy are needed to address these issues.
|
BREAST CANCER DETECTION, DIAGNOSIS, AND STAGING
|
|---|
Although mammography remains the primary imaging modality used in the detection of early breast cancer, the frequency of false-negative results is estimated to be 5%–15% (68). The inability to detect breast cancer at mammography is often due to obscuration of the tumor by superimposed fibroglandular tissue. This imperfect sensitivity of mammography has led to the use of adjunctive imaging methods, including MR imaging. Although MR imaging has demonstrated variable specificity, the reported (2–7) sensitivity of this modality for the demonstration of invasive breast cancer has approached 100% in several series. The invasive cancers in these studies have predominantly been invasive ductal carcinoma. In terms of the sensitivity of MR imaging for detection of invasive lobular carcinoma, relatively little has been published (69). There have been reports of false-negative MR results in cases of invasive lobular cancer (7,69,70), although authors of recent reports suggest that MR imaging may be sensitive for the demonstration of this cancer (69,71). In contrast to high sensitivity of MR imaging for the demonstration of invasive cancer reported in multiple series, the reported (51,52,72,73) sensitivity of MR imaging for the demonstration of DCIS has been variable, ranging from 40% to 100%. This wide range probably reflects small numbers of cases, variable imaging techniques, and the histologic variability (including differences in degree of angiogenesis) of DCIS (74–76). There are, however, reports (31,47,51) of mammographically and clinically occult DCIS that was demonstrated at MR imaging.
On the basis of the promising results of these studies, it has been proposed that MR imaging, with its high sensitivity for the demonstration of invasive breast cancer, could be used as an adjunctive imaging modality when the results of conventional imaging and clinical examinations are equivocal and for breast cancer staging in patients with newly diagnosed disease.
|
PROBLEM SOLVING: IS CANCER PRESENT?
|
|---|
There are reports (77,78) that MR imaging can be used as a problem-solving tool in cases where a suggestive lesion is identified on only one mammographic view and in cases of equivocal mammographic or physical examination findings (Fig 5). MR imaging may also play a problem-solving role in the evaluation of patients who have undergone breast conservation therapy for breast cancer in whom equivocal changes are identified at mammography or physical examination. For these patients, posttreatment follow-up can sometimes be difficult, because posttreatment changes can mimic and obscure recurrent disease.
View larger version (186K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. MR imaging for problem solving. Images in a woman with two abnormalities identified at mammography, each seen on only one mammographic view. (a) Spot mediolateral oblique magnification mammogram demonstrates a cluster of calcifications (arrow). (b) Mediolateral mammographic view obtained during diagnostic work-up shows an area of possible architectural distortion (arrows), which was visible only on this view. No definite lesion was identified at US (not shown). (c) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an enhancing spiculated mass (bottom arrow) in the central lateral portion of the breast and a second enhancing lesion (top arrow) in the superior lateral portion. MR imaging-guided wire localization was performed for both lesions. Mammographic specimen radiography (not shown) demonstrated the calcifications in the superior lateral specimen and architectural distortion in the central lateral specimen. Two infiltrating ductal carcinomas were identified at histopathologic examination. The patient subsequently underwent mastectomy.
|
|
View larger version (149K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. MR imaging for problem solving. Images in a woman with two abnormalities identified at mammography, each seen on only one mammographic view. (a) Spot mediolateral oblique magnification mammogram demonstrates a cluster of calcifications (arrow). (b) Mediolateral mammographic view obtained during diagnostic work-up shows an area of possible architectural distortion (arrows), which was visible only on this view. No definite lesion was identified at US (not shown). (c) Sagittal fat-suppressed contrast-enhanced 3D fast SPGR MR image (9.2/2.1) reveals an enhancing spiculated mass (bottom arrow) in the central lateral portion of the breast and a second enhancing lesion (top arrow) in the superior lateral portion. MR imaging-guided wire localization was performed for both lesions. Mammographic specimen radiography (not shown) demonstrated the calcifications in the superior lateral specimen and architectural distortion in the central lateral specimen. Two infiltrating ductal carcinomas were identified at histopathologic examination. The patient subsequently underwent mastectomy.
|
|
View larger version (76K | |
TOP
ABSTRACT
INTRODUCTION
