Article Text

PDF
379 PREDICTING EXTRACAPSULAR EXTENSION OF PROSTATE CANCER USING MAGNETIC RESONANCE SPECTROSCOPY.
  1. B. B. Woodbury,
  2. D. R. Broome,
  3. B. A. Holshouser,
  4. L. C. Sowers
  1. Loma Linda University, Loma Linda, CA

Abstract

Purpose To determine whether the percentage of the volume of the prostate showing abnormal proton magnetic resonance spectroscopy correlates with a higher likelihood of extracapsular extension (ECE) of prostate cancer on high-resolution MR imaging.

Methods Patients with a history of androgen blockade therapy, pelvic radiation therapy, more than 2 weeks of proton beam therapy or an incomplete data set were excluded from the study (n = 26 of 180 patients screened). Pelvic MRI was performed using a 1.5 T scanner with an abdominal phased array coil and an endorectal coil to obtain high-resolution T2-weighted images of the prostate. An experienced radiologist, blinded to MRS results, determined the presence of ECE. The endorectal coil was used for PRESS 3D spectroscopic imaging with frequency selective water/lipid signal suppression to obtain spectra. Nominal voxel resolution was 6 × 6 × 6 mm3. In consensus with an experienced radiologist, voxels were assigned to 2 regions (central gland and peripheral zone) in the prostate on the basis of the MRSI matrix overlaid on T2-weighted images. The spectra from these voxels were fitted in the time domain with model functions for the citrate (Ci), creatine (Cr), and choline (Cho) signals using proprietary software (PRISMA, v1.5, University Medical Center, Nijmegen, Netherlands). The fit results include integral values (+ SD) for the individual signals. Values for the (Cho + Cr)/Ci ratio (CC/C ratio) as a marker for tumor tissue were calculated and compared to published normal values for each region. For our analysis, spectra were considered abnormal (tumor) if the CC/C ratio was > 5 SD above normal. The percentage of abnormal spectra was calculated by dividing the number of voxels exhibiting abnormal spectra by the total number of voxels within the region of the prostate for each patient.

Results The percentage of abnormal spectra was not found to be correlated with extracapsular extension. Using a discriminant analysis, the percentage of abnormal spectra alone was able to predict extracapsular extension with 65% accuracy. When combined with PSA levels, the prediction accuracy increased to 75% (prediction accuracy of PSA alone was 73%).

Conclusion Percentage of abnormal spectra alone does not yield accurate prediction of extracapsular extension of tumor outside of the prostate. It will, however, improve prediction when combined with other methods of prostate cancer assessment, such as PSA levels.

Statistics from Altmetric.com

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.