| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Breast Imaging |
1 From the Departments of Radiology (W.H., P.R.F., K.D., L.A.T., T.M.B.) and Surgery (P.R.F., B.O.), State University of New York, Stony Brook. Received April 1, 2003; revision requested June 20; final revision received November 17; accepted January 5, 2004. Supported by the Susan G. Komen Breast Cancer Foundation. Address correspondence to W.H., Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021 (e-mail: huangw2@mskcc.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The combined MR imaging–MR spectroscopy protocol was performed in 50 patients after positive findings at mammography but prior to biopsy. Single-voxel proton MR spectroscopy and perfusion MR imaging were conducted only if DCE MR images showed rapid contrast enhancement in the lesion. Biopsy results were used as the reference for comparison with MR results and for calculation of sensitivity and specificity in the detection of breast malignancy.
RESULTS: DCE MR imaging alone showed 100% sensitivity and 62.5% specificity. The specificity improved to 87.5% with the addition of 1H MR spectroscopy and to 100% with the further addition of perfusion MR imaging. Twenty-eight patients underwent both MR spectroscopy and perfusion MR imaging. Two patients underwent MR spectroscopy but declined to undergo perfusion MR imaging. The remaining 20 patients had negative results at DCE MR imaging and therefore did not undergo the additional examinations.
CONCLUSION: The combined MR protocol of DCE MR imaging, 1H MR spectroscopy, and perfusion MR imaging has high sensitivity and specificity in the diagnosis of breast cancer.
© RSNA, 2004
Index terms: Breast neoplasms, diagnosis, 00.31, 00.32 • Breast neoplasms, MR, 00.12143, 00.12144, 00.12145 • Magnetic resonance (MR), contrast enhancement, 00.12143 • Magnetic resonance (MR), perfusion study, 00.12144 • Magnetic resonance (MR), spectroscopy, 00.12145
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In recent years, results of many studies have shown that the noninvasive techniques of magnetic resonance (MR) imaging have strong potential to improve sensitivity and specificity in the diagnosis and evaluation of breast cancer. MR imaging techniques, particularly those involving the administration of contrast agents, have been performed in selected patients for the diagnosis and evaluation of breast tumors. Dynamic contrast agent–enhanced (DCE) MR imaging, in which the passage of a contrast agent through mammary tissue is monitored after a bolus injection, is now an integral part of a proposed standard diagnostic protocol for breast cancer (1) when MR imaging is being performed. The advantages of this approach stem from the observation that even the qualitative time courses of MR imaging signal intensity in a region of interest exhibit reproducible patterns that appear to be capable of enabling discrimination between benign and malignant lesions and even between different types of malignancy (2,3). Although the results of investigations have varied greatly, the sensitivity of T1-weighted DCE MR imaging for breast malignancy has been consistently reported to be excellent (88%–100%) (4–12). Qualitative analyses of the temporal changes in signal intensity after bolus contrast agent injection have shown that malignant tissues generally enhance early compared with benign tissue, with a large and rapid increase in signal intensity. This is presumably caused by the inherent leakiness of the tumor vasculature and/or by increased vascularization. However, the reported specificity of DCE MR imaging has been variable, ranging from 37% to 97% (4–13). Although there is good evidence that carcinomas tend to enhance faster and wash out earlier than do benign tissues, there are exceptions to this pattern, and there is considerable overlap in response between benign and malignant lesions. For example, fibroadenomas sometimes demonstrate an enhancement pattern similar to that of invasive cancer (14).
There exist a plethora of semiquantitative and quantitative analysis methods that have been applied to imaging data in an attempt to differentiate benign from malignant breast lesions (4–9,13,15,16). However, these methods assume an effectively infinitely fast rate of equilibrium transcytolemmal water exchange (equivalent to assuming that the linear relationship between R1 [1/T1] and contrast agent concentration holds) during contrast agent bolus passage through breast tissue; this assumption leads to substantial underestimation of pharmacokinetic parameters (17,18). Furthermore, quantitative methods require a lengthy process of data analysis and are not quite practical for breast cancer diagnosis in a clinical setting.
T2*-weighted perfusion MR imaging (19–22) and hydrogen 1 (1H) MR spectroscopy (23–26) have also been examined as promising tools for improving specificity in the detection of breast malignancy. The former technique is based on measurement of the increased perfusion that is typical in malignant tumors; the latter is based on the detection of the 1H nuclear MR of choline-containing compounds (Cho), which, when enhanced, serves as the marker of active tumors (27).
The purpose of this study was to prospectively determine if a combined MR protocol that includes T1-weighted DCE MR imaging, 1H MR spectroscopy, and T2*-weighted perfusion MR imaging improves specificity in the diagnosis of breast cancer.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The MR examination protocol included T1-weighted DCE MR imaging, single-voxel 1H MR spectroscopy, and T2*-weighted perfusion MR imaging. In preparation for this study, a retrospective analysis was performed (results were not published). Clinical DCE MR imaging data (acquired with the same protocol used in our study) and pathologic results were compared in more than 30 patients who had mammographic scores of BIRADS 4 or BIRADS 5 at our institution. Results of this retrospective examination showed there were no false-negative findings when using the DCE MR imaging protocol alone. Therefore, to determine if our combined MR protocol would improve specificity in breast cancer diagnosis, 1H MR spectroscopy and perfusion MR imaging were performed only after positive findings were observed at DCE MR imaging. Prior to the MR examinations, informed consent was obtained from the patients once the nature of the procedures had been fully explained. This MR imaging–MR spectroscopy study was conducted with the approved institutional review board protocol.
MR Examinations
The MR imaging and MR spectroscopy data for all 50 patients were acquired by using a 1.5-T whole-body MR imager (Edge; Marconi Medical Systems, Cleveland, Ohio). The body coil was used as the transmitter, and a dedicated four-channel phased-array breast coil (USA Instruments, Aurora, Ohio) was used as the receiver.
After pilot imaging had been performed, T1-weighted DCE MR imaging was performed by using a three-dimensional spoiled gradient-recalled echo pulse sequence to acquire eight frames (data sets) of sagittal volumetric images continuously over time, spatially covering the whole breast. Parameters were as follows: 9.0/3.8 (repetition time msec/echo time msec), 30° flip angle, 5-mm section thickness, 24-cm field of view, and 64 x 256 matrix size. Each frame of images typically contained 18–26 sections, depending on the breast size; this resulted in an acquisition time of 10.4–15.0 seconds for each frame (the temporal resolution of the DCE MR imaging study). At the start of the second frame data acquisition, the contrast agent (gadodiamide, Omniscan; Nycomed, Princeton, NJ) was delivered intravenously as a dose of 0.1 mmol per kilogram of body weight, at a rate of 2 mL/sec, by using an MR-compatible programmable power injector (Spectris; Medrad, Indianola, Pa); this was followed with a 15-mL flush of isotonic saline solution. The total injection time was less than 15 seconds. The first frame of the DCE MR images was subtracted from each frame of images by using commercial image-processing software (Breast Uptake; Marconi Medical Systems).
From the subtracted MR images, if contrast enhancement was observed in later frames in the lesion with positive mammographic diagnosis, a region of interest was drawn encompassing the enhanced lesion, and a plot of signal intensity-versus-image frame was obtained. Typically in this study, signal intensity that reached a plateau by the fourth frame was defined as a positive finding for malignancy at DCE MR imaging. If there was continuously increasing signal intensity in the enhanced region through eight frames of data acquisition or if there was no enhancement at all, this was defined as a negative finding. In case of lesion enhancement, two authors (either W.H. and K.D. or W.H. and P.R.F.) drew the region of interest independently to confirm the shape of the signal intensity curve. The study protocol was discontinued in patients with negative findings at DCE MR imaging. Patients with positive findings continued the protocol, undergoing further examination with 1H MR spectroscopy and perfusion MR imaging.
A single-voxel proton spectrum was collected from the enhanced lesion with a point-resolved spectroscopic pulse sequence (2,000/135; 128 signals acquired). The subtracted DCE MR sections covering the enhanced lesion were used as pilot images for placement of the rectangular MR spectroscopy voxel, which encompassed the entire enhanced lesion area. The voxel size ranged from 1.6 to 9.0 cm3, depending on the size of the enhanced lesion that was being examined. With the commercial MR spectroscopy data-processing software (Marconi Medical Systems), the raw spectral data were processed by using 3-Hz line broadening, Fourier transformation, and phase and baseline corrections. The detection of an apparent Cho resonance peak at 3.23 ppm (signal-to-noise ratio, 
2) was defined as a positive finding at MR spectroscopy, and a negative finding was defined otherwise.
After MR spectroscopy, T2*-weighted perfusion MR imaging was performed in a single 5-mm sagittal section containing the enhanced lesion. The location of this section was chosen to be approximately through the center of the enhanced lesion and was based on the analysis of the DCE MR imaging data. A fast low-angle shot sequence was employed for perfusion imaging, and parameters were as follows: 54/35, 10° flip angle, 24-cm field of view, 92 x 256 matrix size, and 40 frames. The temporal resolution of the data acquisition was about 5 seconds. Intravenous bolus injection of gadodiamide (0.1 mmol per kilogram of body weight) was carried out at 4 mL/sec at the beginning of the sixth frame data acquisition. The perfusion MR data were processed (PROPAK software; Marconi Medical Systems) by using the standard area-under-the-curve algorithm (28–32) to construct the relative blood volume map that corresponded to the breast image section. The MR signal intensity-versus-time curve was converted to the 
R2*-versus-time curve based on the following relationship: R2* = (1/T2*) = –(1/TE) ln(SIt/SI0), where TE is echo time, SIt is the signal intensity at time t, and SI0 is the baseline signal intensity prior to contrast material injection. The R2*-versus-time curve was analyzed by using gamma-variate fit, and the area under the curve was computed that was proportional to blood volume. When compared with enhancement in normal breast tissue on the same image section, the observation of striking enhancement in the lesion area (at least a fivefold increase in signal intensity when compared with normal breast tissue) on the relative breast blood volume map was defined as a positive finding at perfusion MR imaging. No apparent enhancement (no enhancement at all or less than a fivefold increase in signal intensity when compared with normal breast tissue) was defined as a negative finding.
Statistical Analysis
The sensitivity and specificity in the detection of breast malignancy were calculated for each MR examination method and for the combination of methods; these calculations were based on the correlation of the MR data with the biopsy results used as the reference standard. The positive or negative findings at MR examinations were classified as true or false in comparison with pathologic findings. Sensitivity is the probability that results at imaging are positive in those patients who have the disease. Sensitivity is defined as [TP/(TP + FN)] · 100, where TP is the total number of true-positive results and FN is the total number of false-negative results. Specificity is the probability that results at imaging are negative in patients who do not have the disease. Specificity is defined as [TN/ (TN + FP)] · 100, where TN is the total number of true-negative results and FP is the total number of false-positive results.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In 39 of 50 patients, there was contrast enhancement in the suspicious lesions at DCE MR imaging. As an example, Figure 1a shows a sagittal DCE MR image of the breast obtained in a patient with a suspicious lesion at mammography that was later pathologically proved to be malignant. This image was the result of subtraction of a first frame image from the fourth frame image at the same location. The contrast-enhanced lesion is clearly visible on this image, and the placement of the MR spectroscopy voxel, encompassing the enhanced area, is also demonstrated. Figure 1b shows the graph of signal intensity-versus-image frame from the enhanced lesion area in this patient. The curve rises rapidly and reaches a plateau by the fourth frame, which implies positive findings at DCE MR imaging. Figure 1c shows a DCE MR image similar to the one in Figure 1a, although this image was obtained in another patient. The lesion was clearly enhanced following contrast administration; however, the curve of signal intensity-versus-image frame (Fig 1d) of this lesion displays continuous signal intensity increase, which implies negative findings at DCE MR imaging in this patient. The lesion was later confirmed to be benign at biopsy. Of 20 patients with negative DCE MR imaging findings, 11 had images with no contrast enhancement at all. Typically, when contrast enhancement was observed in the lesion, curve shapes like those shown in Figure 1b and 1d were used to distinguish positive from negative findings (in nine patients) at DCE MR imaging.
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Despite the exciting progress that has been made in DCE MR imaging methods for breast cancer diagnosis, based on the characteristics of the time course of signal intensity change following bolus contrast media injection, there have been considerable difficulties with reproducibility from one MR imaging acquisition or pulse sequence to another and from site to site. Because of the variable results, no single standardized and generally accepted technique has emerged for DCE MR imaging. This has caused difficulties in making meaningful comparisons between different cancer types and between data from different imaging sites (33,34). There are discrepancies with regard to sequence parameter choices, numbers of sections, speed of acquisition, and dose of contrast medium; these discrepancies are based on available hardware and software, clinical indications, desired results, and personal experience. Therefore, it is desirable to perform quantitative analysis of the DCE MR imaging data to extract fundamental pathophysiologic quantities, such as microvascular perfusion and permeability of the breast lesion. These values are independent of MR data acquisition methods and parameters and may be used to differentiate malignant from benign breast lesions. Most semiquantitative and quantitative DCE MR data analysis methods employ the two-compartment Kety-Schmidt model (17,35) to compute pharmacokinetic parameters, which assumes an effectively infinitely fast rate of equilibrium transcytolemmal water exchange and, thus, a linear relationship between R1 and tissue contrast concentration during bolus contrast agent passage through breast lesions. However, these assumptions do not always reflect the actual physiologic environment, given the inhomogeneous nature of tumors, and lead to substantial underestimation of the pathophysiologic quantities (17,18). As a result, there have been no clearly defined thresholds for pathophysiologic quantities that can be used to differentiate benign from malignant breast tumors. Another drawback of quantitative analysis is that the data processing is complicated and lengthy, which may delay the diagnostic process. Also, the results of quantitative analysis may be more difficult to understand and interpret. At this time and in the near future, qualitative MR protocols with high sensitivity and specificity, such as the one we used for this study, may be the tools of choice for breast cancer diagnosis in clinical practice.
The target of MR data collection was well defined in this study; it was the breast lesion with a positive diagnosis based on conventional mammography. The lesion could be easily located on MR images according to geometric information on mammograms. In clinical practice, however, there are other complications to take into account when using this combined protocol of DCE MR imaging, 1H MR spectroscopy, and perfusion MR imaging, such as a patient’s menstrual cycle or the presence of multiple enhancing lesions. It has been shown (36,37) that both diffuse and nodular contrast enhancement of breast parenchyma can occur at DCE MR imaging during all phases of the menstrual cycle, especially during week 1 and week 4. There is less enhancement during weeks 2 and 3, especially during week 2. Since the breast lesion of interest was predefined in our study, menstrual cycle was not an issue of concern. If this breast MR protocol is to be employed for clinical purposes, we believe examination should be scheduled for the 2nd week of the patient’s cycle, whenever possible, to avoid potential complications. Although the DCE MR imaging technique involves the use of a three-dimensional sequence to collect data in the whole breast, the MR spectroscopy and perfusion MR imaging techniques used in this study are limited to data collection from one contrast-enhanced lesion and one image section containing the enhanced lesion, respectively. In reality, multiple contrast-enhanced lesions are often observed at different locations in one breast. In such cases, multisection or three-dimensional MR spectroscopic imaging techniques and multisection echo-planar MR imaging techniques are desirable to measure Cho level and relative blood volume in all the enhanced lesions. The echo-planar MR imaging method (21) enables data collection in the whole breast while maintaining or even shortening the acquisition time in comparison with that of the perfusion MR imaging method used in this study. There may be an extra 10–15 minutes required for MR spectroscopic imaging data acquisition, compared with the acquisition time of single-voxel MR spectroscopy. The possibility of using the MR spectroscopic and echo-planar MR imaging sequences in our methods for whole breast study is being investigated.
A limitation of this study was the patient population chosen. All patients had positive findings at mammography with either possible malignancy or a high probability of malignancy. To determine if this MR imaging–MR spectroscopy protocol may potentially be used as a standard screening tool for the general patient population following positive mammographic diagnosis but prior to biopsy, we hope to conduct a study among patients whose positive ratings at mammography are lower than BIRADS 4 and BIRADS 5.
Results of one study (19) showed that the addition of T2*-weighted perfusion MR imaging to a DCE MR imaging study substantially improved specificity in the diagnosis of breast cancer, which is consistent with our findings. In that study, there was an interval of about 15 minutes between T1-weighted DCE MR imaging and T2*-weighted perfusion MR imaging to allow sufficient time for washout of the contrast agent injected during DCE MR imaging. Since the MR signal changes in opposite directions for these two techniques, excessive residual contrast material from the DCE MR imaging procedure will severely compromise the robustness of the signal change during bolus contrast material passage at perfusion MR imaging. In our experience, patients usually feel anxious and restless in the MR imager during idling time. This may lead to body movement and cause misplacement of the image section at single-section perfusion MR imaging. In our study, the addition of MR spectroscopy (about a 15-minute examination) between DCE MR imaging and perfusion MR imaging had the following advantages: (a) It allowed us to obtain more information to improve specificity in the detection of breast malignancy, (b) it allowed time for the washout of contrast agent, and (c) it made patients more at ease and thereby reduced the possibility of body movement. However, there is a limitation to MR spectroscopy. Because of the low signal-to-noise ratio and the impracticality of increasing the number of signals acquired and thus lengthening the imaging time, the MR spectroscopy data for lesions smaller than 1 cm3 are usually not reliable. Therefore, the diagnosis of small lesions needs to be based on data obtained at DCE MR imaging and perfusion MR imaging. In our study, the patients with false-positive findings at MR spectroscopy all had pathologically proved fibroadenomas. It appears, on the basis of our results, that the false-positive findings at MR spectroscopy can be corrected by taking perfusion MR imaging data into account.
In conclusion, results of this study in 50 patients with positive mammographic findings showed that a combined protocol of T1-weighted DCE MR imaging, 1H MR spectroscopy, and T2*-weighted perfusion MR imaging had high sensitivity and specificity in the diagnosis of breast cancer. We believe this protocol is easy to implement at clinical MR imaging sites and has the potential to be used as a tool to help prevent unnecessary biopsies following positive mammographic diagnosis.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, W.H., T.M.B.; study concepts and design, W.H., T.M.B.; literature research, W.H., L.A.T.; clinical studies, W.H., K.D., L.A.T.; data acquisition, W.H., K.D., L.A.T.; data analysis/interpretation, W.H., P.R.F., B.O., L.A.T.; manuscript preparation, W.H.; manuscript definition of intellectual content, W.H., P.R.F., T.M.B.; manuscript editing, W.H., P.R.F., B.O., T.M.B., L.A.T.; manuscript revision/review, W.H., L.A.T.; manuscript final version approval, W.H.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. H. G. M. Peters, I. H. M. Borel Rinkes, N. P. A. Zuithoff, W. P. T. M. Mali, K. G. M. Moons, and P. H. M. Peeters Meta-Analysis of MR Imaging in the Diagnosis of Breast Lesions Radiology, December 1, 2007; 246(1): 116 - 124. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Bartella, S. B. Thakur, E. A. Morris, D. D. Dershaw, W. Huang, E. Chough, M. C. Cruz, and L. Liberman Enhancing Nonmass Lesions in the Breast: Evaluation with Proton (1H) MR Spectroscopy Radiology, October 1, 2007; 245(1): 80 - 87. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Bartella and W. Huang Proton (1H) MR Spectroscopy of the Breast RadioGraphics, October 1, 2007; 27(suppl_1): S241 - S252. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. G. ODLE Breast MR Radiol. Technol., September 1, 2006; 78(1): 45M - 66M. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. K. Kuhl, P. Jost, N. Morakkabati, O. Zivanovic, H. H. Schild, and J. Gieseke Contrast-enhanced MR Imaging of the Breast at 3.0 and 1.5 T in the Same Patients: Initial Experience Radiology, June 1, 2006; 239(3): 666 - 676. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Bartella, E. A. Morris, D. D. Dershaw, L. Liberman, S. B. Thakur, C. Moskowitz, J. Guido, and W. Huang Proton MR Spectroscopy with Choline Peak as Malignancy Marker Improves Positive Predictive Value for Breast Cancer Diagnosis: Preliminary Study Radiology, June 1, 2006; 239(3): 686 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Meisamy, P. J. Bolan, E. H. Baker, M. G. Pollema, C. T. Le, F. Kelcz, M. C. Lechner, B. A. Luikens, R. A. Carlson, K. R. Brandt, et al. Adding in Vivo Quantitative 1H MR Spectroscopy to Improve Diagnostic Accuracy of Breast MR Imaging: Preliminary Results of Observer Performance Study at 4.0 T Radiology, August 1, 2005; 236(2): 465 - 475. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
