| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Breast Imaging |
1 From the Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021. From the 2004 RSNA Annual Meeting. Received June 21, 2005; revision requested August 5; revision received November 2; final version accepted November 16. Supported by the Memorial Sloan-Kettering Sloan Kettering Research and Development Fund. Address correspondence to L.B. (e-mail: bartelll{at}mskcc.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: After institutional review board approval and informed consent were obtained for this HIPAA-compliant study, breast MR spectroscopy was performed in patients with suspicious or biopsy-proved malignant lesions measuring 1 cm or larger at MR imaging. Single-voxel MR spectroscopy data were collected from a single rectangular volume of interest that encompassed the lesion. MR spectroscopy findings were defined as positive if the signal-to-noise ratio of the choline resonance peak was greater than or equal to 2 and as negative in all other cases. MR spectroscopy findings were then compared with histologic findings.
Results: A total of 56 patients (age range, 20–77 years) with 57 lesions were imaged. The median lesion size at MR imaging was 2.3 cm (range, 1–15 cm). Histologically, 31 (54%) of 57 lesions were malignant, and 26 (46%) were benign. A choline peak was present in 34 of 57 lesions (including all cancers) and in three of 26 benign lesions, giving MR spectroscopy a sensitivity of 100% and a specificity of 88%. In 40 lesions of unknown histologic type, the use of MR spectroscopy as an adjunct to MR imaging would have significantly (P < .01) increased the positive predictive value of biopsy from 35% to 82%. If biopsy had been performed only on those lesions with a choline peak at MR spectroscopy, biopsy may have been spared in 23 (58%) of 40 lesions, and none of the cancers would have been missed.
Conclusion: Proton MR spectroscopy was successfully incorporated into breast MR imaging studies for lesions measuring 1 cm or larger. This technique may be useful in reducing the number of lesions detected at MR imaging that require biopsy.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The diagnostic value of 1H MR spectroscopy is typically based on the detection of elevated levels of choline compounds, which are a marker of active tumor (7). The results of several ex vivo MR spectroscopy studies have shown elevated levels of choline, such as phosphocholine and glycerophosphocholine (8,9), in cancerous human mammary cells. It has been suggested that choline levels may not be as greatly elevated in some breast cancers as they are in others, determined by the biologic aggressiveness; thus, the ability of MR spectroscopy to demonstrate abnormal choline levels in breast cancer has been variable (10). Multiple in vivo 1H MR spectroscopy studies aimed at improving discrimination between benign and malignant breast lesions have been performed at several centers (10–19). In addition to being used for breast cancer diagnosis, in vivo 1H MR spectroscopy has also been used to monitor breast cancer response to chemotherapy (20,21).
We hypothesized that MR spectroscopy may improve the positive predictive value of MR imaging of the breast. Thus, the purpose of our study was to prospectively evaluate the diagnostic performance of MR spectroscopy in patients with suspicious lesions or biopsy-proved cancers at MR imaging by using histologic findings as the reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The inclusion criteria for participation included patients who were 18 years of age or older, patients who were undergoing diagnostic MR imaging for a biopsy-proved breast lesion that measured 1 cm or larger, and patients who were undergoing MR imaging–guided biopsy or needle localization for a breast lesion that measured 1 cm or larger on MR images. Exclusion criteria were the inability to undergo or complete the MR imaging examination, neoadjuvant chemotherapy for breast cancer, and the presence of a breast hematoma (from either recent surgery or biopsy) adjacent to the suspicious lesion.
Patients were selected prospectively from a consecutive pool of patients who were undergoing diagnostic MR imaging or MR imaging–guided biopsy or needle localization. To identify patients with lesions measuring 1 cm or larger, a radiologist (L.B.) with 4 years of experience as a breast imaging specialist reviewed, on a weekly basis, the diagnostic MR images of patients who were scheduled to undergo MR imaging–guided procedures. This radiologist also reviewed, again on a weekly basis, a separate list of patients undergoing diagnostic MR imaging of the breast to identify patients with a biopsy-proved lesion measuring 1 cm or larger. The appropriate clinician was subsequently informed regarding eligible patients prior to the day of the study so that the clinician could inform the patient.
Single-Voxel 1H MR Spectroscopy
Data from single-voxel (ie, one single rectangular volume of interest) 1H MR spectroscopy were acquired by using a 1.5-T whole-body MR imager (LX or Excite; GE Medical Systems, Milwaukee, Wis). The body coil was used as the transmitter, and either a dedicated four-channel or an eight-channel phased-array breast coil (MR Imaging Devices, Waukesha, Wis) was used as the receiver. Pre- or postcontrast sagittal T1-weighted MR images with fat saturation (acquired with three-dimensional spoiled gradient-recalled acquisition in the steady state, 35° flip angle, 18.8/4.2 [repetition time msec/echo time msec], 3.0-mm section thickness, 20-cm field of view, and 256 x 192 matrix) were used as scout images for placement of the rectangular MR spectroscopy voxel. Voxel placement was performed by one of two physicists (W.H. or S.B.T., with 5 and 2 years of experience in MR spectroscopy of the breast, respectively) and encompassed the enhanced lesion. The range of the voxel size was 1.0–11.2 cm3, with a median of 3.8 cm3. The proton spectrum was collected with a point-resolved spectroscopy sequence (2000/135, 128 signals acquired). The automatic shimming on the unsuppressed water signal usually achieved less than 20 Hz of full width at half maximum. If the full width at half maximum was more than 20 Hz following automatic shimming, quick manual shimming (by W.H. or S.B.T.) was performed to adjust the full width at half maximum to less than 20 Hz.
The total imaging time, including the preimaging adjustment for shimming and water suppression, was initially about 20 minutes and was decreased to 10 minutes as experience was gained by the end of patient accrual.
Data Processing
The acquired raw MR spectroscopy data were transferred to a computer workstation and were processed by using a software program (SAGE/IDL; GE Medical Systems) with a 5-Hz line broadening Fourier transformation and phase and baseline corrections. After Fourier transformation and phase correction, the multichannel data sets were added together to generate the final spectrum. A choline resonance peak was searched for at a frequency of 3.23 ppm. The baseline was corrected by using a cubic spline method, which was performed by choosing data points in the frequency range of 3.5–4.0 and 2.5–3.0 ppm and setting these data points to zero. The signal-to-noise ratio of the choline peak was calculated as the ratio of choline peak amplitude to noise amplitude, which was measured in the flat noise baseline region (>6 or <0 ppm). Because the magnetic field strength, breast coil, range of lesion size, pulse sequence, and data acquisition parameters used in this study were almost identical to those used by Huang et al (18), we adopted the same signal-to-noise ratio threshold for choline peak to determine if the MR spectroscopy result was positive or negative. Results were deemed positive when the signal-to-noise ratio was greater than or equal to 2 and negative in all other cases. To eliminate bias, histologic findings were revealed to the physicists (W.H., S.B.T.) only after the signal-to-noise ratio of the choline peak was determined.
MR Imaging and Image Review
Eligible patients who were scheduled to undergo diagnostic breast MR imaging were first imaged according to the standard departmental protocol (22). The biopsy-proved lesion was identified by reviewing the postcontrast T1-weighted MR images. MR spectroscopy of the lesion was then performed.
For eligible patients who were scheduled to undergo MR imaging–guided biopsy or needle localization, the lesion was first identified by performing pre- or postcontrast T1-weighted MR imaging (the former was preferred). MR spectroscopy of the lesion was then performed, followed by the scheduled MR imaging–guided biopsy or needle localization and surgical excision (22,23).
MR images were reviewed by one radiologist with 4 years of experience as a breast imaging specialist (L.B.). T2-weighted MR images, unenhanced T1-weighted MR images, and T1-weighted MR images that were obtained within the first 2 minutes after intravenous contrast material injection were reviewed on a picture archiving and communication system workstation (GE Medical Systems) by a radiologist. The radiologist could page back and forth through sequential sections and adjust the window and level settings at the workstation. Recorded data included lesion size (largest measurement in one plane), morphologic features (mass vs nonmass), and final assessment categories according to the Breast Imaging Reporting and Data System (BI-RADS) lexicon for MR imaging (24).
After these data were recorded, histologic findings were reviewed and compared with MR spectroscopy results and MR imaging interpretations (L.B.). Data were entered into a computerized spreadsheet (Excel; Microsoft, Redmond, Wash).
Statistical Analysis
To evaluate the performance of MR spectroscopy, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of this technique in the population of women with MR imaging findings that warranted biopsy were calculated after correlation with histologic results. To assess the effect of MR spectroscopy on the positive predictive value of biopsy, the positive predictive value of MR imaging in patients with BI-RADS category 4 or 5 lesions was calculated and compared with the positive predictive value of MR imaging after MR spectroscopy was performed, with histologic findings used as the reference standard.
Analyses were conducted by using a statistical software program (Stata 8.0 for Windows; Stata, College Station, Tex). Confidence intervals were calculated by using the SVYTAB procedure (Stata) to adjust for the multiple observations per patient. The positive predictive values of MR imaging with and without MR spectroscopy were compared by using methods described by Leisenring et al (25).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Histologic Findings and MR Imaging Features
At histologic analysis, 31 (54%) of 57 lesions were malignant and 26 (46%) were benign. Among the 31 cancers, 19 (61%) were invasive ductal carcinoma, eight (26%) were invasive lobular carcinoma, three (10%) had mixed invasive lobular and ductal features, and one (3%) was a ductal carcinoma in situ (DCIS) of high nuclear grade with extensive necrosis. Eighteen (69%) consisted of fibrocystic changes, ductal hyperplasia, benign breast parenchyma, stromal fibrosis, fat necrosis, fibroadenomas, sclerosing adenosis, and duct ectasia. The remaining eight (31%) of 26 benign lesions were high-risk lesions and included lobular carcinoma in situ, atypical lobular hyperplasia, atypical ductal hyperplasia, columnar cell alteration with atypia, chronic inflammation with atypia, and an atypical vascular lesion with lobular carcinoma in situ.
The median lesion size at MR imaging was 2.3 cm (range, 1–15 cm). Among the 57 lesions, 33 (58%) were mass lesions and 24 (42%) were nonmass lesions. Of these 57 lesions, 28 (49%) were BI-RADS category 4 lesions, 12 (21%) were BI-RADS category 5 lesions, and 17 (30%) were BI-RADS category 6 lesions. Carcinoma was found in five (18%) of 28 BI-RADS category 4 lesions and in nine (75%) of 12 BI-RADS category 5 lesions. All BI-RADS category 6 lesions were malignant.
MR Spectroscopy Results
A positive choline peak was present in 34 (60%) of 57 lesions. All 31 malignant lesions demonstrated a positive choline peak (Fig 1). Three (12%) of 26 benign lesions also demonstrated a positive choline peak. The three lesions with false-positive results included a fibroadenoma, a chronic inflammatory lesion with atypia (Fig 2), and a lesion with atypical ductal hyperplasia and columnar cell alteration. All lesions with a negative choline peak were benign (Fig 3).
|
|
|
|
|
|
To evaluate the effect that MR spectroscopy had on the original MR imaging recommendation for biopsy, the 17 BI-RADS category 6 lesions and all biopsy-proved cancers at the time of initial MR imaging were excluded from the calculations. Of the remaining 40 lesions, 14 (35%) were malignant and 26 (65%) were benign (Table 1). According to the interpretation of diagnostic MR imaging results, all 40 lesions had suspicious characteristics and biopsy was recommended; the positive predictive value of MR imaging in this group was 35%. MR spectroscopy demonstrated a choline peak in only 17 lesions, 14 of which were malignant. The positive predictive value in selecting women for biopsy significantly (P < .01) increased to 82%, with no false-negative recommendations when MR imaging and MR spectroscopy were combined. Biopsy could have been spared in 23 (58%) of 40 lesions, and none of the cancers would have been missed.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
Three false-positive cases were included in this study: a fibroadenoma, a chronic inflammatory lesion with atypia, and atypical ductal hyperplasia with columnar cell alteration. A false-positive choline peak has been previously reported with fibroadenoma (11,12). To our knowledge, the other two lesions have not been previously reported, but we are limited by the small number of published series on single-voxel MR spectroscopy of the breast. In view of the presence of atypia in these two lesions, excision would have been the standard of care. Further work is necessary to evaluate the frequency and characteristics of false-positive findings at MR spectroscopy.
Though absolute quantitation of choline concentration (17) may be more desirable in assessing suspicious breast lesions, this method requires additional time to collect data from an internal or external reference, to correct partial volume effects that result from the inclusion of adipose tissue in the MR spectroscopy voxel, and to carefully calibrate differences in relaxation times between the tissue choline signal and references. Therefore, MR spectroscopy may not be practical in the clinical setting. On the other hand, measurement of choline signal-to-noise ratio can usually provide a quick answer, which is important if MR spectroscopy is to be included in a clinical diagnostic protocol for breast cancer.
During hardware and software setup, the signal-to-noise ratio can be affected by many factors, including field strength, coil sensitivity profile, voxel size, pulse sequence, and data acquisition parameters. In this particular study, because our hardware and software setup and voxel size range were similar to those used by Huang et al (18), a threshold of choline signal-to-noise ratio of 2 was used to discriminate between benign and malignant lesions. Encouragingly, we achieved the same 100% sensitivity, as well as improved specificity. We did not find any correlation between choline signal-to-noise ratio and MR spectroscopy voxel size. This is probably because the choline level is related more to the malignant nature of the tumor than to the size of the tumor.
These two studies seem to suggest that, with this specific experimental setup, a threshold of choline signal-to-noise ratio of 2 is adequate for breast MR spectroscopy to discriminate between benign and malignant lesions with 100% sensitivity. By no means should the same threshold be blindly adopted for other MR spectroscopy studies of the breast if any element of the hardware and software setup is different. Successful shimming is essential for sufficient water suppression and good quality of the spectrum. Although proximity to skin and biopsy clips could be a problem and may produce susceptibility effects, we did not encounter such conditions in our small series.
This is an ongoing project and the size of the population is small, which is an obvious limitation of this study. This population, however, allowed us to evaluate MR spectroscopy in the breast at 1.5 T without the technical failures owing to lesion size. It also enabled us to compare results in all biopsy cases without altering patient care. As a consequence of our small population, we had only one case of DCIS. Further analysis in a larger series with a wider variety of histologic types (including more DCIS cases) is necessary. Our data do not enable calculation of the sensitivity, specificity, negative predictive value, or accuracy of breast MR imaging alone because only lesions classified as BI-RADS category 4, 5, or 6 by using MR imaging criteria were included. Additional study of MR spectroscopy applied to normal MR imaging examinations with long-term follow-up would be useful. Because smaller lesions result in a longer imaging time to obtain meaningful data with a 1.5-T magnet, this study included only those MR imaging lesions that measured 1 cm or larger. Single-voxel MR spectroscopy was used so that only one lesion could be evaluated at a time. Techniques that enable assessment of multiple voxels require investigation.
In conclusion, MR spectroscopy is a promising technique that may decrease the number of benign biopsy results generated with MR imaging in the clinical setting. MR spectroscopy is fast and well tolerated and could be readily incorporated into a breast MR imaging examination. In our population, adjunct use of MR spectroscopy would have significantly (P < .01) increased the positive predictive value of biopsy from 35% to 82% for lesions detected at MR imaging and would have spared biopsy in 58% of lesions, without missing any cancers. Our findings, if confirmed in a larger series, suggest that single-voxel MR spectroscopy may be used to evaluate suspicious lesions detected at MR imaging as part of the diagnostic work-up to assist in determining the need for biopsy. Additional studies with a larger population and wider variety of histologic types are necessary for further assessment of this promising method. Chemical shift MR imaging will enable us to examine the whole breast in the future, and higher-field-strength magnets will enable us to study smaller lesions. The door to breast MR spectroscopy in the clinical setting is now beginning to open.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: BI-RADS = Breast Imaging Reporting and Data System • DCIS = ductal carcinoma in situ
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, L.B.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, L.B., E.A.M., L.L., S.B.T., W.H.; clinical studies, L.B., E.A.M., D.D.D., L.L., S.B.T., J.G., W.H.; statistical analysis, C.M.; and manuscript editing, L.B., E.A.M., D.D.D., L.L., W.H.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Sardanelli, A. Fausto, G. Di Leo, R. de Nijs, M. Vorbuchner, and F. Podo In Vivo Proton MR Spectroscopy of the Breast Using the Total Choline Peak Integral as a Marker of Malignancy Am. J. Roentgenol., June 1, 2009; 192(6): 1608 - 1617. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-M. Lee, J. B. Kaplan, M. P. Murray, and L. Liberman Complete Excision of the MRI Target Lesion at MRI-Guided Vacuum-Assisted Biopsy of Breast Cancer Am. J. Roentgenol., October 1, 2008; 191(4): 1198 - 1202. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Q. Yu, J. Yang, and P. Wang Malignant Tumors and Chronic Infections in the Masticator Space: Preliminary Assessment with In Vivo Single-Voxel 1H-MR Spectroscopy AJNR Am. J. Neuroradiol., April 1, 2008; 29(4): 716 - 719. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. K. Kuhl, F. Traber, J. Gieseke, W. Drahanowsky, N. Morakkabati-Spitz, W. Willinek, M. von Falkenhausen, C. Manka, and H. H. Schild Whole-Body High-Field-Strength (3.0-T) MR Imaging in Clinical Practice * Part II. Technical Considerations and Clinical Applications Radiology, April 1, 2008; 247(1): 16 - 35. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Wang, P.-L. Torng, T.-P. Liu, K.-L. Chen, and T. T.-F. Shih Proton MR Spectroscopy in Normal Breasts Between Pre- and Postmenopausal Women: A Preliminary Study Am. J. Roentgenol., February 1, 2008; 190(2): 505 - 510. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. L. Sah, R. Sharma, H. Kandpal, A. Seith, S. Rastogi, S. Bandhu, and N. R. Jagannathan In Vivo Proton Spectroscopy of Giant Cell Tumor of the Bone Am. J. Roentgenol., February 1, 2008; 190(2): W133 - W139. [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]
|
|
|
|
|
|
P. Stanwell and C. Mountford In Vivo Proton MR Spectroscopy of the Breast RadioGraphics, October 1, 2007; 27(suppl_1): S253 - S266. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. G. ODLE Breast MR Radiol. Technol., September 1, 2006; 78(1): 45M - 66M. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
