1654
Delineating Benign from Malignant Breast Lesions Using Restriction Spectrum Imaging1Radiology, University of California San Diego, La Jolla, CA, United States, 2School of Medicine, University of California San Diego, La Jolla, CA, United States
Synopsis
Non-contrast diffusion MRI holds great potential to screen women for breast cancer. Restriction spectrum imaging (RSI) is an advanced diffusion-weighted imaging (DWI) technique that reflects the high nuclear to cytoplasm ratio observed in cancer cells. This abstract explores RSI as a technique to non-invasively identify malignant from benign masses on non-contrast MRI by measuring RSI cellularity index (RSI-CI). Biopsy-proven malignant masses demonstrate high cellularity index compared to benign lesions. In this pilot study, RSI differentiates malignant from benign masses without contrast imaging, and could prove useful as a screening tool.
Introduction
Women at high risk for breast cancer undergo annual dynamic contrast-enhanced MRI (DCE-MRI) as early as the age of 25. While DCE-MRI has higher sensitivity (90%) for detecting neoplastic lesions compared to conventional imaging (ultrasound, mammography), the concurrent low specificity (72%)[1] leads to a high false positive rate and unnecessary biopsies. The increased exposure to gadolinium with annual DCE-MRI[2] for screening purposes is also a concern. Developing a highly sensitive and specific protocol for breast cancer screening without the need for contrast would prove advantageous. Diffusion weighted imaging (DWI) has shown potential for cancer screening[3–6], however, utilization of DWI for breast imaging is limited by large geometric distortions. Restriction spectrum imaging (RSI) is an advanced DWI imaging technique that isolates the diffusion signal of water theoretically attributed to high nuclear-to-cytoplasmic ratio observed in cancer cells. In addition, RSI incorporates geometric distortion correction. The RSI cellularity index (RSI-CI) statistically measures, in standard deviations (Z-score), the discrepancy of signal from a spherically restricted diffusion pool in a tissue of interest with respect to that of average healthy tissue[7]. Since RSI is sensitive to highly cellular tissues with increased nuclear to cytoplasm ratio, it has proven useful in tumor identification and early treatment evaluation in breast[8,9] and prostate cancer[10]. We hypothesize that RSI can be utilized as a general breast cancer screening tool to provide high sensitivity and specificity in non-invasively differentiating malignant from benign masses.Methods
The MRIs of 21 patients with biopsy-proven malignant lesions and 15 patients with biopsy-proven benign lesions, or lesions read as benign by an expert radiologist, were retrospectively included in the study. A total of 21 malignant lesions and 25 benign lesions were evaluated, as some patients had multiple benign lesions. Imaging parameters: T2 fat suppressed FSE— TE/TR=107/4520ms, FA=111°, FOV=340×340mm2, voxel size=0.66×0.66×5mm3; Multi-shell DWI — TE/TR=82/9000ms, b-values (number of diffusion directions) = 0, 500(6), 1500(6), 4000(15) s/mm2, FOV=340×340mm2, voxel size=2.66×2.66×5mm3, PE direction L/R. Lesions were identified with the gold standard DCE-MRI and regions of interest (ROI) were manually drawn on these data in areas of pathology-proven malignant or benign lesion. Multi-shell DWI data were processed with RSI pipeline to correct for distortion artifacts using reverse polarity gradient (RPG)[11] and to estimate RSI-CI. Average RSI-CI (Z-score) were measured on each ROI (Fig. 1) for restriction spectrum imaging cellularity maps (CM). An ROI was drawn on post-contrast images in areas of either known malignancy or benign regions for each patient. The areas selected by these ROIs were then used to measure the corresponding RSI Z-scores. Unpaired two-tailed t-tests were performed to evaluate the significance of signal variation between malignant and benign Z-scores.Results
Average RSI-CI (Z-score) was found to be statistically different (p<0.0001) between malignant and benign lesions (Fig. 2). The average Z-score of biopsy-proven malignant lesions was 6.55, while the average Z-score of lesions classified as benign by an expert radiologist or biopsy was 0.322. Through this pilot study, initial evaluation of RSI for classification of benign versus malignant lesions suggests that a higher Z-score may be used to delineate likely malignant lesions, while a lower Z-score is suggestive of benign lesions. Sensitivity of using Z-score above 2 to determine malignancy is 96%, and specificity of using Z-score below 2 to determine benign lesions is 86% (Table. 1).Discussion
Initial results from the present study suggest that RSI-CI is highly sensitive (96%) and specific (86%) to breast cancer malignancies. Z-scores slightly above 2 in benign lesions may be attributed to incomplete geometric distortion correction and partial volume effect in small lesions. The ROIs were drawn in DCE-MR images and transferred to distortion corrected RSI-CI maps. A discrepancy between anatomical and DW images corrected for geometric distortions (with RPG) for breast applications is on average 1.8±1.0 pixels[12], and may contribute to artificially increased RSI-CI scores. This work suggests that RSI may be a reliable diffusion imaging technique in screening for breast malignancy with MRI as it is a non-contrast technique with high sensitivity and specificity. Future work will be required to determine a robust RSI-CI cutoff in a larger population to assist clinicians in determining probability of malignancy and potentially stratifying grade of disease.Acknowledgements
NIH EB-RO1000790, UCSD Clinician Scientist ProgramReferences
1. Menezes GL, Stehouwer BL. Gisela LG Menezes, Floor M Knuttel, Bertine L Stehouwer, Ruud M Pijnappel, Maurice AAJ van den Bosch. World J Clin Oncol. 2014;5(2):61-70. doi:10.5306/wjco.v5.i2.61
2. Kanal E, Tweedle MF. Residual or Retained Gadolinium: Practical Implications for Radiologists and Our Patients 1. Radiol n Radiol. 2015;275(3). doi:10.1148/radiol.2015150805
3. Zhang L, Tang M, Min Z, Lu J, Lei X, Zhang X. Accuracy of combined dynamic contrast-enhanced magnetic resonance imaging and diffusion-weighted imaging for breast cancer detection: a meta-analysis. doi:10.1177/0284185115597265
4. Bickelhaupt S, Laun F, Tesdorff J, Lederer W, Daniel H, Stieber A, Delorme S SH. Fast and noninvasive characterization of suspicious lesions Detected at Breast cancer X-ray screening: Capability of Diffusion-weighted MR Imaging with MIPs 1. Radiol Clin ATOS Clin Heidelb. 278:69120. doi:10.1148/radiol.2015150425
5. Partridge SC, DeMartini WB, Kurland BF, Eby PR, White SW, Lehman CD. Differential diagnosis of mammographically and clinically occult breast lesions on diffusion-weighted MRI. J Magn Reson Imaging. 2010;31(3):562-570. doi:10.1002/jmri.22078
6. Yabuuchi H, Matsuo Y, Sunami S, et al. Detection of non-palpable breast cancer in asymptomatic women by using unenhanced diffusion-weighted and T2-weighted MR imaging: comparison with mammography and dynamic contrast-enhanced MR imaging. Eur Radiol. 2011;21:11-17. doi:10.1007/s00330-010-1890-8
7. White NS, McDonald CR, Farid N, et al. Diffusion-weighted imaging in cancer: Physical foundations and applications of restriction spectrum imaging. Cancer Res. 2014. doi:10.1158/0008-5472.CAN-13-3534
8. Rakow-penner R, White N, Abudu B, et al. Restriction Spectrum Imaging in Breast Cancer. 2015:1000790.
9. Rakow-penner R, Albino-kroesing N, Abudu B, et al. Restriction Spectrum Imaging in Breast Cancer : Improved Evaluation of Response to Neoadjuvant Chemotherapy Introduction : Methods : Results : :2-3.
10. Rakow-Penner RA, White NS, Margolis DJA, et al. Prostate Diffusion Imaging with Distortion Correction HHS Public Access. Magn Reson Imaging. 2015;33(9):1178-1181. doi:10.1016/j.mri.2015.07.006
11. Holland D, Kuperman JM, Dale
AM. Efficient Correction of Inhomogeneous Static Magnetic Field-Induced
Distortion in Echo Planar Imaging. 2009. doi:10.1016/j.neuroimage.2009.11.044
Figures
