Three-dimensional magnetic resonance spectroscopic imaging of brain and prostate cancer

Abstract

Clinical applications of magnetic resonance spectroscopic imaging (MRSI) for the study of brain and prostate cancer have expanded significantly over the past 10 years. Proton MRSI studies of the brain and prostate have demonstrated the feasibility of noninvasively assessing human cancers based on metabolite levels before and after therapy in a clinically reasonable amount of time. MRSI provides a unique biochemical "window" to study cellular metabolism noninvasively. MRSI studies have demonstrated dramatic spectral differences between normal brain tissue (low choline and high N-acetyl aspartate, NAA) and prostate (low choline and high citrate) compared to brain (low NAA, high choline) and prostate (low citrate, high choline) tumors. The presence of edema and necrosis in both the prostate and brain was reflected by a reduction of the intensity of all resonances due to reduced cell density. MRSI was able to discriminate necrosis (absence of all metabolites, except lipids and lactate) from viable normal tissue and cancer following therapy. The results of current MRSI studies also provide evidence that the magnitude of metabolic changes in regions of cancer before therapy as well as the magnitude and time course of metabolic changes after therapy can improve our understanding of cancer aggressiveness and mechanisms of therapeutic response. Clinically, combined MRI/MRSI has already demonstrated the potential for improved diagnosis, staging and treatment planning of brain and prostate cancer. Additionally, studies are under way to determine the accuracy of anatomic and metabolic parameters in providing an objective quantitative basis for assessing disease progression and response to therapy.

Figure 1

Selected axial and sagittal spectral arrays from a 3D MRSI study of a poorly enhancing glioblastoma multiforme before brachytherapy. Regions of elevated choline and reduced (or absent) NAA resonances corresponded spatially to the tumor mass. The three-dimensional MRSI acquisition provided a total of 140 spectra from voxels located within the mass and in neighboring brain tissue. (From Ref. [94]; with permission.)

Figure 2

This graph summarizes the MRSI results from histology-proven viable brain tumors. The results demonstrated abnormal MR spectra in all tumors. Choline/NAA ratios were significantly (p<0.001) elevated, with no overlap with the normal values. (From: Proton Chemical Shift Imaging of Cancer. In Magnetic Resonance Imaging of the Body (3rd ed). CB Higggins, H Hricak, and A Helms (Eds). Lippincott-Raven Press, New York; with permission.)

Figure 3

MRI/MRSI and PET data from recurrent anaplastic astrocytoma before Gamma knife radiosurgery. The partially enhancing mass lesion seen on the post-gadolinium MR image (A, left) coincided spatially with the region of increased FDG uptake on the positron emission tomography image (A). The 1 cm³ spectral voxel centered on this lesion (B) demonstrated elevated choline and no NAA resonance. Following radiosurgery choline intensities in the spectra decreased significantly by 1 month and even further by 5 months. This example demonstrates the utility of MRSI for following radiosurgery treatment.

Figure 4

MRI/MRSI and PET data of a recurrent patient with anaplastic astrocytoma. The MR images demonstrated a small region of enhancment (top left) and a larger region of high signal intensity on the T2-weighted image (top middle). The PET scan demonstrated FDG uptake much lower than cortical gray matter (top right and was inconclusive for tumor). High-resolution 0.34 cm³ MRSI data were acquired using a figure-8 phased array coil placed at the top of the head. The choline resonance was highly elevated in the region of contrast enhancement but also extended posteriorly. Subsequent disease progression and surgical resection confirmed the presence of tumor. This study demonstrates the ability of high-resolution MRSI to not only detect viable cancer but also to assess the spatial extent of the non-enhancing regions of the tumor. (From Ref. [77]; with permission.)

Figure 5

Sequential MR spectra from three spatial locations are shown following external beam radiation therapy with a brachytherapy boost. The spectra from a tumor volume that was within the implant target (top row) showed a dramatic decrease in choline by 10 weeks. Tumor spectra from a region outside of the implant target (middle row) showed a slower reduction in choline levels. Spectra from adjacent brain tissue (bottom row) showed no significant change during the 25-week study. (From Ref. [94]; with permission).

Figure 6

Fifty-eight-year-old man with pathologic stage pT3a prostate cancer, Gleason score 5. (A) Reception profile corrected FSE T2-weighted (TR5000/TE 102 msec) axial MR image through the midprostate obtained using an endorectal coil. A tumor focus (arrows) is seen as an area of decreased signal intensity in the peripheral zone of the right gland. (B) Histopathologic section (hematoxylin-eosin stain; magnification 100x) confirmed tumor in the peripheral zone of the right midgland (1) which abuts the inked prostatic margin (a) and is interspersed between normal prostatic glands (b). (C) 0.24 cm³ spectra obtained from area of imaging abnormality (1) in the right peripheral zone demonstrates elevated choline and reduced citrate, a pattern consistent with definite cancer. (D) MR spectra obtained from normal left peripheral zone (2) demonstrates a normal spectral pattern with citrate dominant and no abnormal elevation in choline. (From Ref. [191]; with permission.)

Figure 7

Comparison of axial endorectal coil/pelvic phased array FSE prostate images (A) beforeand (B) after performing an analytic correction for the reception profiles of the endorectal and pelvic phased array coils. In the corrected image, the high signal intensity close to the endorectal coil has been removed allowing for improved visualization of prostate cancer (low T2 signal intensity white arrows) and the prostatic capsule (thin dark line encompassing the prostate, black arrow).

Figure 8

(A) A reception profile corrected axial FSE prostate image taken from the midgland of a patient with prostate cancer. Prostate cancer is anatomically identified as a region of low signal intensity in the right peripheral zone (white arrows) compared to the high signal intensity of the left peripheral zone (blackarrow). (B) Same image as in (A) with corresponding PRESS selected volume (thick solid white line) and a small section (6x6 voxels) of the phase encoded spectral array. (C) 0.24 cm³ proton spectra associated with the region of the prostate indicated by the phase encoded grid in (B). Prostate cancer can be metabolically identified as region of elevated choline and reduced citrate in the right peripheral zone compared to the healthy tissue (low choline and high citrate) in the left peripheral zone. (D) A (choline+creatine)/citrate peak area ratio image (red) overlaid on a the corresponding T2-weighted image. The (choline+creatine)/citrate peak area ratio has been thresholded such that all intensities below three standard deviations of the mean healthy peripheral zone value are translucent whereas those above the threshold demonstrate increasing intensities of red.

Figure 9

Plot of individual (x's) and mean (circles) choline/normal peripheral zone choline (A) citrate/normal peripheral zone citrate (B) area ratios for the central gland, periurethral tissues, BPH and prostate cancer. (C) Plot of individual (x's) and mean (circles) (creatine+choline)/citrate area ratios for regions of prostate cancer, BPH, normal peripheral zone in young volunteers and patients. Note the absence of overlap of individual (choline+creatine)/citrate ratios between regions of normal peripheral zone (Pz) and prostate cancer. Also note that older individuals demonstrated peripheral zone ratios that were not different from the younger volunteer values. (From Ref. [36]; with permission.)

Figure 10

Representative 0.24 cm³ PRESS ¹H spectra (TR-1 sec, TE-130 ms) “voxel shifted” to be within regions of cancer based on T2-weighted MRI and subsequent step-section histopathology of the surgically resected prostate. A region of Gleason Grade; (A) 5 (2+3), (B) 6 (3+3), (C) 7 (3+4), (D) 8 (4+4), and (E) metastatic prostate cancer in brain.

Figure 11

ROC curve comparing the diagnostic performance of MRI alone and combined with MRSI. The addition of MRSI improved the diagnostic performance of MRI in the localization of prostate cancer to a sextant of the prostate. (Adapted from Ref. [191] with permission.)

Figure 12

An axial T2-weighted images through the prostate of a patient who had significantly enlarged central gland due to benign prostatic hyperplasia (BPH, dashed arrow) and a compressed peripheral zone. Due to an elevated PSA (12 ng/ml), he underwent a TRUS guided biopsy that was negative. The sampling error of TRUS guided biopsy which is already large in the average-size gland is further increased in very large prostates. Prostate cancer was identified in a subsequent MRI/MRSI exam based on low signal intensity on the T2-weighted image (A, solid arrow), and by elevated choline and reduced citrate in the corresponding 0.24 cm³ array of proton spectra (C). A subsequent MRI/MRSI MR targeted, TRUS guided biopsy confirmed the presence of cancer in the right peripheral zone.

Figure 13

ROC curve comparing the diagnostic performance of MRI alone and combined with MRSI. The addition of MRSI improved the diagnostic performance of MRI in the prediction of spread of cancer outside the prostate. (Adapted from Ref. [112] with permission.)

Figure 14

(A) Representative 0.24 cm³ PRESS ¹H spectrum of necrosis and corresponding histologic slide of necrotic biopsy tissue. (B) Representative ¹H spectrum of BPH and corresponding histologic slide of BPH biopsy tissue. (C) Representative ¹H spectrum of prostate cancer and corresponding histologic slide of malignant biopsy tissue. (Adapted from Ref. [169].)