HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Yu, K. K. Articles by Kurhanewicz, J. Search for Related Content PubMed PubMed Citation Articles by Yu, K. K. Articles by Kurhanewicz, J. Related Collections Related Article

Prostate Cancer: Prediction of Extracapsular Extension with Endorectal MR Imaging and Three-dimensional Proton MR Spectroscopic Imaging¹

¹ From the Departments of Radiology (K.K.Y., J.S., H.H., D.B.V., R.G.M., S.J.N., J.K.), Pathology (C.J.Z.), and Urology (P.R.C.), University of California at San Francisco, 505 Parnassus Ave, San Francisco, CA 94143-0628. Received June 23, 1998; revision requested August 8; final revision received February 3, 1999; accepted June 8. K.K.Y. supported by a GE-AUR Radiology Research Academic Fellowship. J.S. supported by the Deutsche Forschungsgemeinschaft. J.K. supported by grants RO1-59897 and R29-64667 from the National Institutes of Health. Address reprint requests to H.H. (e-mail: Hedvig.Hricak@radiology.ucsf.edu).

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
PURPOSE: To determine if the addition of three-dimensional (3D) proton magnetic resonance (MR) spectroscopic imaging to endorectal MR imaging helps diagnose extracapsular extension (ECE) of prostate cancer.

MATERIALS AND METHODS: Endorectal MR imaging and 3D MR spectroscopic imaging were performed in 53 patients with prostate cancer before radical prostatectomy. MR imaging studies were evaluated by two independent readers unaware of histopathologic findings. The presence of ECE was graded on a five-point scale. At 3D MR spectroscopic imaging, cancer was diagnosed if the ratio of choline plus creatine to citrate was 2 or more SDs above normal. The accuracy of MR imaging alone was compared with that of combined MR imaging and 3D MR spectroscopic imaging, with use of the step-section histopathologic results as the standard of reference.

RESULTS: For the less experienced reader, the addition of 3D MR spectroscopic imaging to MR imaging significantly improved accuracy (area under the receiver operating characteristic curve [A_z] = 0.75 vs A_z = 0.62, P < .05). For the more experienced reader, the addition improved accuracy but not significantly (A_z = 0.86 vs A_z = 0.78). The addition also reduced interobserver variability (A_z = 0.86 vs A_z = 0.75).

CONCLUSION: The addition of 3D MR spectroscopic imaging to MR imaging improves accuracy for less experienced readers and reduces interobserver variability in the diagnosis of ECE of prostate cancer.

Index terms: Diagnostic radiology, observer performance • Magnetic resonance (MR), spectroscopy, three-dimensional, 844.12145 • Prostate, MR, 844.12141, 844.12145 • Prostate, neoplasms, 844.32 • Receiver operating characteristic (ROC) curve

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
One of the most important factors affecting the prognosis and choice of treatment in patients with prostate cancer is the presence of extracapsular extension (ECE) (1–4). Diagnostic imaging may assist assessment of the extent and stage of disease; however, the role of diagnostic imaging in the pretreatment evaluation of patients with prostate cancer remains unclear. Transrectal ultrasonography (US) is the most widely used imaging modality for local staging of prostate cancer; however, transrectal US is not very accurate in the prediction of ECE (5,6). Endorectal magnetic resonance (MR) imaging shows the most promise for local staging but has not yet realized its potential owing to problems with interobserver variability and variable diagnostic accuracy (7–15). Differences in reader experience and lack of accepted diagnostic criteria have resulted in a wide range of results reported for MR imaging in the detection of ECE (8–10,16–19). Further improvements in MR imaging technology are needed to improve reproducibility and to simplify diagnosis. In particular, a means must be found to improve the performance of less experienced readers.

Development of endorectal surface coils has allowed application of localized three-dimensional (3D) proton MR spectroscopic imaging to the evaluation of the prostate gland with a spatial resolution of up to 0.24 cm³ (19). With use of 3D MR spectroscopic imaging, significantly higher choline and significantly lower citrate levels were observed in regions of cancer compared with those in areas of benign prostatic hypertrophy and normal prostatic tissue (19). The ratio of these metabolites (choline to citrate) in regions of cancer did not overlap with ratios in the normal peripheral zone, which suggests that 3D MR spectroscopic imaging may improve tumor detection and estimation of tumor volume compared to those with MR imaging alone. This is particularly helpful in patients with postbiopsy hemorrhage or prior hormonal treatment (19–25). The fact that histopathologic studies have demonstrated a strong correlation between the volume of prostate cancer and the presence of ECE (26–28) suggests the possibility that 3D MR spectroscopic imaging estimates of tumor volume may be used to assist the diagnosis of ECE, similar to the way in which serum prostate-specific antigen levels are used to assess the risk of ECE.

The purpose of our study was to determine if 3D proton MR spectroscopic imaging can be used to predict ECE in patients with prostate cancer and whether the addition of 3D MR spectroscopic imaging to endorectal MR imaging improves the diagnosis of ECE over that with MR imaging or 3D MR spectroscopic imaging alone.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Patients
This was a retrospective cross-sectional study. Between May 1992 and July 1997, 542 consecutive patients with a histopathologic diagnosis of prostate cancer were referred for combined endorectal and phased-array coil MR imaging and 3D MR spectroscopic imaging. We reviewed the medical records and MR imaging and 3D MR spectroscopic imaging studies to determine whether these patients met the following inclusion criteria: (a) MR imaging and 3D MR spectroscopic imaging and radical prostatectomy were performed at our institution (n = 89); (b) results of step-section pathologic examination were available for review (n = 87); (c) transverse plane, endorectal-coil T1- and T2-weighted MR images were retrievable for review (n = 79); (d) 3D MR spectroscopic imaging data were retrievable (n = 73); (e) 3D MR spectroscopic imaging studies had diagnostic signal-to-noise ratio characteristics (n = 68); (f) radical prostatectomy was performed within 3 months after MR imaging and 3D MR spectroscopic imaging (n = 62); and (g) no hormonal therapy was administered prior to surgery (n = 53).

Fifty-three patients met all the inclusion criteria. Mean patient age was 60 years ± 7 (SD). The preoperative prostate-specific antigen level was 8 ng/mL ± 5. Pathologic Gleason score was 6 ± 1 (range, 4–10). The interval between MR examination and radical retropubic prostatectomy was 27 days ± 23 (range, 1–86 days).

MR Imaging Technique
MR imaging and 3D MR spectroscopic imaging were performed on a 1.5-T MR imaging system (Signa; GE Medical Systems, Milwaukee, Wis). The endorectal coil (Medrad, Pittsburgh, Pa) was inserted into the rectum and connected to a pelvic phased-array coil, and combined images were obtained. After the coil position was verified with sagittal fast spin-echo T2-weighted localizer imaging, transverse fast spin-echo T2-weighted images were obtained from below the prostatic apex to above the seminal vesicles with use of the following parameters: repetition time msec/echo time msec (effective) of 4,000–5,000/102, 3-mm section thickness and no intersection gap, three signals acquired, 14-cm field of view, 256 x 192 matrix, and no phase wrap. Transverse T1-weighted images (500–700/12, 4-mm-thick sections and 1-mm gap, two signals acquired, 14-cm field of view, 256 x 192 matrix, no phase wrap) were then obtained from below the prostatic apex to the level of the aortic bifurcation. All images were analytically corrected for the reception profile of the endorectal and pelvic phased-array coils to reduce the near-field artifact associated with the endorectal coil (29). The total examination time, including patient positioning, coil placement, MR imaging, and 3D MR spectroscopic imaging was typically less than 60 minutes.

Three-dimensional MR Spectroscopic Imaging Technique
From the high-spatial-resolution transverse fast spin-echo T2-weighted images, a spectroscopic volume was selected with use of the point-resolved spectroscopic technique to encompass as much of the prostate as possible while excluding periprostatic fat (19,30,31). Magnetic field homogeneity was optimized for the selected volume by means of shimming of the water resonance until a line width of approximately 5 Hz was attained. The echo delay of the point-resolved spectroscopic sequence (130 msec) was optimized for the quantitative detection of both choline and citrate. After a whole-voxel water- and lipid-suppressed spectrum was stored, a 3D MR spectroscopic imaging data set was acquired with a nominal spatial resolution of 0.24–0.70 cm³. Studies were performed with 1,000/130, spectral width of 1,250 Hz, 512 points, and 8 x 8 x 8 with two signals acquired per phase-encoding step or 16 x 8 x 8 with one signal acquired, yielding 512 or 1,024 proton MR spectra, respectively, of which between 40 and 332 were from within the prostate, depending on gland size and spatial resolution. Technical improvements during the study resulted in improved craniocaudal coverage of the peripheral zone, which was increased from 20%–50% to 70%–100%. For this study, 3D MR spectroscopic imaging provided peripheral zone coverage of less than 25% in one patient, 25.0%–49.9% in 19 patients, 50.0%–74.9% in 21 patients, and 75%–100% in 12 patients.

MR Image Analysis
All images were interpreted retrospectively by two independent readers (K.K.Y., H.H.) who were aware that all patients had biopsy-proved prostate cancer. Both readers, however, were blinded to pertinent clinical data (ie, prostate-specific antigen, Gleason score, clinical stage), the results of 3D MR spectroscopic imaging, and surgical-pathologic findings. Reader 1 (more than 5 years experience) had more experience than reader 2 (2 years experience) in interpreting MR studies of the prostate.

The diagnosis of ECE was based on MR imaging findings previously reported to be the most specific for ECE on the basis of multivariate analysis (10,18). These findings were the presence of an irregular capsular bulge, obliteration of the rectoprostatic angle, and asymmetry or direct involvement of the neurovascular bundles. On the basis of these findings, the likelihood of ECE for the right and left prostatic lobes was estimated by the readers with a five-point rating scale: 1, no ECE; 2, probably no ECE; 3, possible ECE; 4, probable ECE; 5, definite ECE. For calculation of sensitivity and specificity, these results were dichotomized so that scores of 1–3 were rated as ECE absent and scores of 4 and 5 were rated as ECE present.

Three-dimensional MR Spectroscopic Imaging Data Analysis
All 3D MR spectroscopic imaging data were processed with an UltraSparc workstation (Sun Microsystems, Mountain View, Calif) by using software developed for 3D MR spectroscopic imaging studies. The spectral data sets were apodized with a 2-Hz Lorentzian function and Fourier transformed in the time domain and three spatial domains. To correlate metabolic data with anatomy and histopathologic findings within the same study, data were displayed by plotting proton MR spectral arrays over the corresponding transverse fast spin-echo T2-weighted images. Use of the same equipment and patient position within an examination allowed alignment of MR imaging and 3D MR spectroscopic imaging data. After frequency, phase, and baseline correction, integral areas for the choline, creatine, and citrate resonances were calculated (19,30). In addition to the peak parameters, the quantification algorithm was used to estimate random noise and, hence, the accuracy of the estimates of peak metabolite areas. To discriminate between cancer and normal prostatic tissue in the peripheral zone, we calculated the peak area ratio of choline plus creatine to citrate for each voxel (19). These ratios and the signal-to-noise ratios for choline and citrate were reported as the mean plus or minus SD. Findings were considered cancer when they (a) were in the peripheral zone, (b) had a signal-to-noise ratio more than 3 SDs of root mean square noise, (c) were not contaminated with periprostatic lipid, and (d) had a ratio of choline plus creatine to citrate of greater than 0.75, which corresponded to 2 SDs above the ratio in the normal peripheral zone (19).

Because technical improvements during the course of the study allowed more complete gland coverage in the more recent 3D MR spectroscopic imaging examinations, it was not possible to determine absolute tumor volume in our study. Three-dimensional MR spectroscopic imaging was used not to determine absolute tumor volume but to stratify the study group into patients with relatively more or relatively less extensive tumor. Relative tumor extent was estimated on 3D MR spectroscopic images by calculating the number of cancer voxels per section for each lobe (right and left) of the prostate. By dividing the number of cancer voxels by the number of sections obtained, we corrected for differences in gland coverage at 3D MR spectroscopic imaging and provided a more accurate determination of relative tumor extent.

Correlation of MR Imaging and 3D MR Spectroscopic Imaging Findings with Histopathologic Findings
One of the authors ( J.S.), who was not a reader of either the MR or 3D MR spectroscopic images, assembled all the images for interpretation, maintained the database of surgical-pathologic information, and correlated MR imaging and 3D MR spectroscopic imaging findings with histopathologic results.

For histopathologic examination after surgical resection, the prostate gland was coated with India ink and fixed in 10% buffered formaldehyde. The gland was transversely sectioned at 3–4-mm intervals in a plane perpendicular to the long axis (base to apex) of the gland. The presence and location of ECE was determined by a pathologist (C.J.Z.) and entered on standardized pathology forms with diagrams that corresponded to those on the MR imaging and 3D MR spectroscopic imaging data forms. These diagrams were used to depict and compare the sites of ECE. ECE was defined as full-thickness capsular penetration with extension of cancer into the periprostatic soft tissue (11). Invasion into but not through the capsule was classified as disease confined to the gland (32). Positive surgical margins at the base and apex of the prostate were also considered ECE because of the lack of a complete prostatic capsule at the base and apex.

Statistical Analysis
Receiver operating characteristic (ROC) analysis was used to compare the results of MR imaging (rating scale from 1 to 5), 3D MR spectroscopic imaging (number of cancer voxels per section), and combined MR imaging and 3D MR spectroscopic imaging. Results from MR imaging and 3D MR spectroscopic imaging were combined by giving each technique equal weighting. Since the range of values for MR imaging interpretation was 1–5 and the observed range of values for the number of cancer voxels for 3D MR spectroscopic imaging interpretation was 0.0–5.3, the MR imaging rating for the likelihood of ECE was added to the number of cancer voxels per section as determined at 3D MR spectroscopic imaging. Paired comparisons of ROC curves were performed by means of permutation tests, to evaluate the ability of each diagnostic technique to correctly identify ECE for each side of the prostate. The statistical significance (P < .05) of differences in tumor extent between patients with and those without ECE were determined by means of an unpaired Student t test. Standard descriptive statistics were calculated for each test, including sensitivity, specificity, positive predictive value, and negative predictive value. The significance of differences in descriptive statistics was assessed by means of the McNemar test. All statistical analyses were performed with commercially available software.

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 
Histopathologic Data
At pathologic examination, 33 of the 53 patients (62%) had disease confined to the prostate gland (four with stage pT2a disease; four, stage pT2b; and 25, stage pT2c). Positive basilar or apical margins, which were considered ECE in our analysis, were found unilaterally in four patients and bilaterally in one patient. Among the 20 of 53 patients (38%) who had ECE, 16 had unilateral ECE (stage pT3a), four had bilateral ECE (stage pT3b), and two had seminal vesicle invasion (stage pT3c [both of these patients also had ECE]). Overall, ECE was considered present in 24 of 106 prostatic lobes (right and left side).

MR Imaging Data
Descriptive statistical data for readers 1 and 2 are summarized in Table 1. The more experienced reader (reader 1) demonstrated significantly better performance than did the less experienced reader (reader 2), with area under the ROC curve (A_z) of 0.78 and 0.62, respectively (Fig 1, P < .01). By relying on the high specificity of MR imaging findings for the diagnosis of ECE, both readers achieved nearly identical specificities (95% for reader 1 and 94% for reader 2) when ratings of 4 and 5 were used to diagnose ECE. Sensitivity, however, was significantly higher for reader 1 than for reader 2 (54% vs 17%, P < .01).

View larger version (20K): [in this window] [in a new window]   Figure 1. ROC curves compare the diagnostic performance of MR imaging alone (Reader 1, Reader 2) and 3D MR spectroscopic imaging (MRSI) alone for the diagnosis of ECE. The more experienced reader (reader 1) had better diagnostic performance than did the less experienced reader (reader 2). The diagnostic performance of 3D MR spectroscopic imaging (Az = 0.76) was comparable to that of reader 1 (Az = 0.78) and better than that of reader 2 (Az = 0.62).

View larger version (154K): [in this window] [in a new window]   Figure 2a. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (192K): [in this window] [in a new window]   Figure 2b. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (166K): [in this window] [in a new window]   Figure 2c. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (154K): [in this window] [in a new window]   Figure 2d. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (35K): [in this window] [in a new window]   Figure 2e. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (66K): [in this window] [in a new window]   Figure 2f. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (61K): [in this window] [in a new window]   Figure 2g. Histopathologic stage pT3a prostate cancer in a 57-year old man. (a) Fast spin-echo T2-weighted (5,000/102) transverse MR image through the middle of the prostate was obtained with an endorectal coil. A large tumor focus is seen as an area of decreased signal intensity in the left peripheral zone (*). There is slight irregular bulging of the adjacent prostatic margin (arrows) but no gross ECE. (b) Photomicrograph of histopathologic section indicates left ECE, with tumor (a) extending into the periprostatic fat. Tumor (b) can also be seen within the peripheral zone of the left middle gland. (Hematoxylin-eosin stain; original magnification, x40.) (c-e) The same transverse MR image as in a with areas of abnormal metabolism (ratio of choline plus creatine to citrate more than 2 SDs above normal), as demonstrated at 3D MR spectroscopic imaging, overlaid in red. The individual MR spectra corresponding to each of the voxels of the bottom two rows of the 3D MR spectroscopic imaging grid are detailed in e and demonstrate extensive metabolic abnormality in the left peripheral zone. (f, g) T2-weighted images through the base (left), middle gland (middle), and lower middle gland (right) (in f) with areas of 3D MR spectroscopic imaging abnormality overlaid in red (in g). Extensive areas of cancer can be seen predominantly in the left peripheral zone, with tumor also noted in the peripheral zone of the right lower middle gland.

View larger version (23K): [in this window] [in a new window]   Figure 3. ROC curves compare the performance of combined MR imaging and 3D MR spectroscopic imaging (Reader 1 + MRSI, Reader 2 + MRSI) to that with MR imaging alone (Reader 1, Reader 2). The addition of 3D MR spectroscopic imaging to MR imaging improved the diagnostic performance of both reader 1 (Az = 0.86) and reader 2 (Az = 0.75) over that with MR imaging alone (Az = 0.78 and 0.62, respectively). The addition of 3D MR spectroscopic imaging to MR imaging for reader 2 improved the performance (Az = 0.75) to a level comparable to that for reader 1 with MR imaging alone (Az = 0.78).

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

The strict application of specific MR imaging criteria for the diagnosis of ECE in our study allowed reproducible high specificity to be achieved by both readers of MR images (95% for reader 1, 94% for reader 2). Sensitivity (54% for reader 1, 17% for reader 2), however, was variable and, with one exception (37), lower than that in previous studies (8–10,12,13,36,38). In addition to the use of diagnostic criteria with high specificity in our study, the lower sensitivity may reflect a shift toward lower stage disease in our patient population, with a lower prevalence of ECE (38%) than was reported in previous studies (49%–62%) (8,9,12,13,38). None of the patients in our study had specimens that showed gross ECE at histopathologic examination, which would have been easier to detect at MR imaging than was early capsular penetration. Our findings are consistent with those of Cornud et al (37), who reported similarly high specificity and low sensitivity for the diagnosis of ECE with use of strictly defined criteria (irregular bulging of the prostatic contour and tumor visibility in the periprostatic fat). Because high diagnostic specificity minimizes false-positive results and helps ensure that few patients will be deprived of potentially curative surgery, it is a test characteristic desirable in treatment selection. It has also been found that the cost-effectiveness of endorectal MR imaging for the detection of ECE is increased at high levels of specificity (39).

Although the use of strict diagnostic criteria produced high and reproducible specificity in our study, low and variable sensitivity remained a problem. Findings in histopathologic studies have demonstrated that prostate cancer volume is a significant predictor of several tumor prognostic factors, including ECE (26–28,40). Stamey et al (26) noted that the prevalence of ECE was only 18% in tumors with volume less than 3 cm³ as compared with 79% in tumors with volume more than 3 cm³. Previous attempts to estimate cancer volume preoperatively with use of sextant biopsy, transrectal US, or MR imaging have been disappointing (23–25,41,42). Three-dimensional MR spectroscopic imaging may provide more accurate tumor volume estimation than has previously been possible by allowing detection of significant alterations in metabolite levels of choline and citrate in the peripheral zone of the prostate (19,43). In our study, we evaluated 3D MR spectroscopic imaging as a potential predictor of ECE but did not determine a specific tumor volume threshold because variations in 3D MR spectroscopic imaging technique during the course of the study prevented determination of absolute tumor volume. A specific volume threshold would be determined best in a prospective study with use of current state-of-the-art 3D MR spectroscopic imaging technique to provide complete gland coverage. The results can then be compared with thresholds previously reported in the histopathologic studies by Stamey et al (26) (3.0 cm³) and McNeal et al (40) (5.4 cm³). In our study, 3D MR spectroscopic imaging was used to stratify the study group on the basis of relative tumor extent, and the risk of ECE in each group was determined. Patients with the least extensive tumor at 3D MR spectroscopic imaging (<1 cancer voxel per section) were found to have only a 6% risk of ECE, whereas patients with the most extensive tumor (>4 cancer voxels per section) had an 80% risk of ECE (Table 2). These results support the use of 3D MR spectroscopic imaging as a predictor of ECE. Compared with MR imaging, 3D MR spectroscopic imaging demonstrated diagnostic accuracy comparable to that for the more experienced reader (reader 1, A_z = 0.78) and significantly better than that for the less experienced reader (reader 2, A_z = 0.62).

After evaluating 3D MR spectroscopic imaging as a potential predictor of ECE, we investigated whether the addition of 3D MR spectroscopic imaging to MR imaging can improve the diagnosis of ECE. For the more experienced reader (reader 1), ROC analysis demonstrated that the addition of 3D MR spectroscopic imaging to MR imaging (A_z = 0.86) resulted in significant improvement in diagnostic performance over that with MR imaging alone (A_z = 0.78) or with 3D MR spectroscopic imaging alone (A_z = 0.76). For the less experienced reader (reader 2), the addition of 3D MR spectroscopic imaging to MR imaging (A_z = 0.75; sensitivity, 46%) resulted in significant improvement in diagnostic performance over that with MR imaging alone (A_z = 0.62; sensitivity, 17%). These preliminary results suggest that the addition of 3D MR spectroscopic imaging to MR imaging may be of potential benefit to readers with different experience levels in the interpretation of prostate MR images.

In addition to the inability to determine absolute tumor volume discussed earlier, other limitations of our study include the lack of evaluation of the role of tumor location, particularly tumor proximity to the neurovascular bundle, and lack of a truly inexperienced reader of MR images. Although our early results show promise for 3D MR spectroscopic imaging as a predictor of ECE that may help improve the performance of readers of MR images with different experience levels, these results should be considered preliminary and need to be validated in further prospective studies. For our results to be generalizable to the radiology community at large, it will be necessary to show that the addition of 3D MR spectroscopic imaging to MR imaging will improve the performance of truly inexperienced readers in different practice settings. The results of the current study provide guidelines about the way 3D MR spectroscopic imaging can be combined with MR imaging in the planning of further prospective studies.

In summary, findings in our study indicate that the strict application of specific MR imaging criteria for the diagnosis of ECE allows reproducible high specificity to be achieved by readers with varying levels of experience in the interpretation of endorectal MR images. Our results suggest that 3D proton MR spectroscopic imaging has potential as a predictor of ECE; the addition of 3D MR spectroscopic imaging to MR imaging decreased the interobserver variability and improved the diagnostic accuracy of MR imaging in the diagnosis of ECE in patients with clinically localized prostate cancer.

	Footnotes

 
See also the article by Scheidler et al (pp 473–480 ) in this issue.

Abbreviations: A_z = area under the receiver operating characteristic curve ECE = extracapsular extension ROC = receiver operating characteristic 3D = three-dimensional

Author contributions: Guarantor of integrity of entire study, H.H.; study concepts, H.H.; study design, K.K.Y., J.S., H.H., J.K.; definition of intellectual content, H.H., S.J.N., P.R.C., J.K.; literature research, J.S.; clinical studies, R.G.M., H.H., K.K.Y., C.J.Z.; data acquisition, R.G.M., J.K., D.B.V.; data analysis, R.G.M., J.S., J.K.; statistical analysis, K.K.Y.; manuscript preparation, K.K.Y., J.S., H.H., J.K.; manuscript editing, H.H., J.K.; manuscript review, H.H., P.R.C., C.J.Z., S.J.N.

	References

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION References

 

1. Gibbons RP, Correa RJ, Jr, Brannen GE, Weissman RM. Total prostatectomy for clinically localized prostatic cancer: long-term results. J Urol 1989; 141:564-566.[Medline]
2. Chodak GW, Thisted RA, Gerber GS, et al. Results of conservative management of clinically localized prostate cancer. N Engl J Med 1994; 330:242-248.[Abstract/Free Full Text]
3. Fair WR, Aprikian A, Sogani P, Reuter V, Whitmore WF, Jr. The role of neoadjuvant hormonal manipulation in localized prostatic cancer. Cancer 1993; 71:1031-1038.[Medline]
4. Gibbons RP, Cole BS, Richardson RG, et al. Adjuvant radiotherapy following radical prostatectomy: results and complications. J Urol 1986; 135:65-68.[Medline]
5. Rifkin MD, Zerhouni EA, Gatsonis CA, et al. Comparison of magnetic resonance imaging and ultrasonography in staging early prostate cancer: results of a multi-institutional cooperative trial. N Engl J Med 1990; 323:621-626.[Abstract]
6. Salo JO, Kivisaari L, Rannikko S, Lehtonen T. Computerized tomography and transrectal ultrasound in the assessment of local extension of prostatic cancer before radical retropubic prostatectomy. J Urol 1987; 137:435-438.[Medline]
7. Quinn SF, Franzini DA, Demlow TA, et al. MR imaging of prostate cancer with an endorectal surface coil technique: correlation with whole-mount specimens. Radiology 1994; 190:323-327.[Abstract/Free Full Text]
8. Hricak H, White S, Vigneron D, et al. Carcinoma of the prostate gland: MR imaging with pelvic phased-array coils versus integrated endorectal—pelvic phased-array coils. Radiology 1994; 193:703-709.[Abstract/Free Full Text]
9. Tempany CM, Zhou X, Zerhouni EA, et al. Staging of prostate cancer: results of Radiology Diagnostic Oncology Group project comparison of three MR imaging techniques. Radiology 1994; 192:47-54.[Abstract/Free Full Text]
10. Outwater EK, Petersen RO, Siegelman ES, Gomella LG, Chernesky CE, Mitchell DG. Prostate carcinoma: assessment of diagnostic criteria for capsular penetration on endorectal coil MR images. Radiology 1994; 193:333-339.[Abstract/Free Full Text]
11. Schiebler ML, Schnall MD, Pollack HM, et al. Current role of MR imaging in the staging of adenocarcinoma of the prostate. Radiology 1993; 189:339-352.[Abstract/Free Full Text]
12. Harris RD, Schned AR, Heaney JA. Staging of prostate cancer with endorectal MR imaging: lessons from a learning curve. RadioGraphics 1995; 15:813-829.[Abstract]
13. Bartolozzi C, Menchi I, Lencioni R, et al. Local staging of prostate carcinoma with endorectal coil MRI: correlation with whole-mount radical prostatectomy specimens. Eur Radiol 1996; 6:339-345.[Medline]
14. Biondetti PR, Lee JK, Ling D, Catalona WJ. Clinical stage B prostate carcinoma: staging with MR imaging. Radiology 1987; 162:325-329.[Abstract/Free Full Text]
15. Schnall MD, Imai Y, Tomaszewski J, Pollack HM, Lenkinski RE, Kressel HY. Prostate cancer: local staging with endorectal surface coil MR imaging. Radiology 1991; 178:797-802.[Abstract/Free Full Text]
16. Langlotz C, Schnall M, Pollack H. Staging of prostatic cancer: accuracy of MR imaging. Radiology 1995; 194:645-646.[Free Full Text]
17. Schiebler ML, Yankaskas BC, Tempany C, et al. MR imaging in adenocarcinoma of the prostate: interobserver variation and efficacy for determining stage C disease. AJR 1992; 158:559-562.[Abstract/Free Full Text]
18. Yu KK, Hricak H, Alagappan R, Chernoff DM, Bacchetti P, Zaloudek CJ. Detection of extracapsular extension of prostate carcinoma with endorectal and phased-array coil MR imaging: multivariate feature analysis. Radiology 1997; 202:697-702.[Abstract/Free Full Text]
19. Kurhanewicz J, Vigneron DB, Hricak H, Narayan P, Carroll P, Nelson SJ. Three-dimensional H-1 MR spectroscopy of the in situ human prostate with high (0.24–0.7-cm³) spatial resolution. Radiology 1996; 198/VOLUME-NR>:795-805.[Abstract/Free Full Text]
20. White S, Hricak H, Forstner R, et al. Prostate cancer: effect of postbiopsy hemorrhage on interpretation of MR images. Radiology 1995; 195:385-390.[Abstract/Free Full Text]
21. Kaji Y, Kurhanewicz J, Hricak H, et al. Localizing prostate cancer in the presence of postbiopsy changes on MR images: role of proton MR spectroscopic imaging. Radiology 1998; 206:785-790.[Abstract/Free Full Text]
22. Chen M, Hricak H, Kalbhen CL, et al. Hormonal ablation of prostatic cancer: effects on prostate morphology, tumor detection, and staging by endorectal coil MR imaging. AJR 1996; 166:1157-1163.[Abstract/Free Full Text]
23. Sommer FG, Nghiem HV, Herfkens R, McNeal J, Low RN. Determining the volume of prostatic carcinoma: value of MR imaging with an external-array coil. AJR 1993; 161:81-86.[Abstract/Free Full Text]
24. Quint LE, Van Erp JS, Bland PH, et al. Carcinoma of the prostate: MR images obtained with body coils do not accurately reflect tumor volume. AJR 1991; 156:511-516.[Abstract/Free Full Text]
25. Jager GJ, Ruijter ET, van de Kaa CA, et al. Local staging of prostate cancer with endorectal MR imaging: correlation with histopathology. AJR 1996; 166:845-852.[Abstract/Free Full Text]
26. Stamey TA, McNeal JE, Freiha FS, Redwine EA. Morphometric and clinical studies on 68 consecutive radical prostatectomies. J Urol 1988; 139:1253-1261.[Medline]
27. Bostwick DG, Graham SD, Jr, Napalkov P, et al. Staging of early prostate cancer: a proposed tumor volume-based prognostic index. Urology 1993; 41:403-411.[Medline]
28. McNeal JE, Villers AA, Redwine EA, Freiha FS, Stamey TA. Capsular penetration in prostate cancer: significance for natural history and treatment. Am J Surg Pathol 1990; 14:240-247.[Medline]
29. Vigneron DB, Nelson SJ, Moyher S, Kelley DAC, Kurhanewicz J, Hricak H. An analytical correction of MR images obtained with endorectal or surface coils (abstr). JMRI 1993; 3:142.
30. Kurhanewicz J, Vigneron DB, Hricak H, et al. Prostate cancer: metabolic response to cryosurgery as detected with 3D H-1 MR spectroscopic imaging. Radiology 1996; 200:489-496.[Abstract/Free Full Text]
31. Star-Lack J, Nelson SJ, Kurhanewicz J, Huang LR, Vigneron DB. Improved water and lipid suppression for 3D PRESS CSI using RF band selective inversion with gradient dephasing (BASING). Magn Reson Med 1997; 38:311-321.[Medline]
32. Epstein JI, Carmichael MJ, Pizov G, Walsh PC. Influence of capsular penetration on progression following radical prostatectomy: a study of 196 cases with long-term followup. J Urol 1993; 150:135-141.[Medline]
33. Schiebler ML, Schnall MD, Outwater E. MR imaging of mucinous adenocarcinoma of the prostate. J Comput Assist Tomogr 1992; 16:493-494.[Medline]
34. Seltzer SE, Getty DJ, Tempany CM, et al. Staging prostate cancer with MR imaging: a combined radiologist-computer system. Radiology 1997; 202:219-226.[Abstract/Free Full Text]
35. Tempany CM, Rahmouni AD, Epstein JI, Walsh PC, Zerhouni EA. Invasion of the neurovascular bundle by prostate cancer: evaluation with MR imaging. Radiology 1991; 181:107-112.[Abstract/Free Full Text]
36. Chelsky MJ, Schnall MD, Seidmon EJ, Pollack HM. Use of endorectal surface coil magnetic resonance imaging for local staging of prostate cancer. J Urol 1993; 150:391-395.[Medline]
37. Cornud F, Belin X, Flam T, et al. Local staging of prostate cancer by endorectal MRI using fast spin-echo sequences: prospective correlation with pathological findings after radical prostatectomy. Br J Urol 1996; 77:843-850.[Medline]
38. Presti JC, Hricak H, Nayaran PA, Shinohara K, White S, Caroll PR. Local staging of prostatic carcinoma: comparison of transrectal sonography and endorectal MR imaging. AJR 1996; 166:103-108.[Abstract/Free Full Text]
39. Langlotz CP, Schnall MD, Malkowicz SB, Schwartz JS. Cost-effectiveness of endorectal magnetic resonance imaging for the staging of prostate cancer. Acad Radiol 1996; 3(suppl):S24-S27.
40. McNeal JE, Bostwick DG, Kindrachuk RA, Redwine EA, Freiha FS, Stamey TA. Patterns of progression in prostate cancer. Lancet 1986; 1:60-63.[Medline]
41. Cupp MR, Bostwick DG, Myers RP, Oesterling JE. The volume of prostate cancer in the biopsy specimen cannot reliably predict the quantity of cancer in the radical prostatectomy specimen on an individual basis. J Urol 1995; 153:1543-1548.[Medline]
42. Terris MK, McNeal JE, Stamey TA. Estimation of prostate cancer volume by transrectal ultrasound imaging. J Urol 1992; 147:855-857.[Medline]
43. Kurhanewicz J, Vigneron DB, Nelson SJ, et al. Citrate as an in vivo marker to discriminate prostate cancer from benign prostatic hyperplasia and normal prostate peripheral zone: detection via localized proton spectroscopy. Urology 1995; 45:459-466.[Medline]

This article has been cited by other articles:

				D. A. McKenna, F. V. Coakley, A. C. Westphalen, S. Zhao, Y. Lu, E. M. Webb, B. Pickett, M. Roach III, and J. Kurhanewicz Prostate Cancer: Role of Pretreatment MR in Predicting Outcome after External-Beam Radiation Therapy--Initial Experience Radiology, April 1, 2008; 247(1): 141 - 146. [Abstract] [Full Text] [PDF]

				A. C. Westphalen, F. V. Coakley, A. Qayyum, M. Swanson, J. P. Simko, Y. Lu, S. Zhao, P. R. Carroll, B. M. Yeh, and J. Kurhanewicz Peripheral Zone Prostate Cancer: Accuracy of Different Interpretative Approaches with MR and MR Spectroscopic Imaging Radiology, December 1, 2007; 246(1): 177 - 184. [Abstract] [Full Text] [PDF]

				A. Shukla-Dave, H. Hricak, C. Moskowitz, N. Ishill, O. Akin, K. Kuroiwa, J. Spector, M. Kumar, V. E. Reuter, J. A. Koutcher, et al. Detection of Prostate Cancer with MR Spectroscopic Imaging: An Expanded Paradigm Incorporating Polyamines Radiology, November 1, 2007; 245(2): 499 - 506. [Abstract] [Full Text] [PDF]

				B. N. Bloch, E. Furman-Haran, T. H. Helbich, R. E. Lenkinski, H. Degani, C. Kratzik, M. Susani, A. Haitel, S. Jaromi, L. Ngo, et al. Prostate Cancer: Accurate Determination of Extracapsular Extension with High-Spatial-Resolution Dynamic Contrast-enhanced and T2-weighted MR Imaging Initial Results Radiology, October 1, 2007; 245(1): 176 - 185. [Abstract] [Full Text] [PDF]

				S. W. T. P. J. Heijmink, J. J. Futterer, T. Hambrock, S. Takahashi, T. W. J. Scheenen, H. J. Huisman, C. A. Hulsbergen-Van de Kaa, B. C. Knipscheer, L. A. L. M. Kiemeney, J. A. Witjes, et al. Prostate Cancer: Body-Array versus Endorectal Coil MR Imaging at 3 T--Comparison of Image Quality, Localization, and Staging Performance Radiology, July 1, 2007; 244(1): 184 - 195. [Abstract] [Full Text] [PDF]

				H. Hricak, P. L. Choyke, S. C. Eberhardt, S. A. Leibel, and P. T. Scardino Imaging Prostate Cancer: A Multidisciplinary Perspective Radiology, April 1, 2007; 243(1): 28 - 53. [Abstract] [Full Text] [PDF]

				F. G. Sommer, P. T. Scardino, H. Hricak, M. W. Kattan, and L. Wang Radiology Research and Rock Soup Radiology, February 1, 2007; 242(2): 637 - 639. [Full Text] [PDF]

				A. Graser, A. Heuck, B. Sommer, J. Massmann, J. Scheidler, M. Reiser, and U. Mueller-Lisse Per-Sextant Localization and Staging of Prostate Cancer: Correlation of Imaging Findings with Whole-Mount Step Section Histopathology Am. J. Roentgenol., January 1, 2007; 188(1): 84 - 90. [Abstract] [Full Text] [PDF]

				Y. J. Choi, J. K. Kim, N. Kim, K. W. Kim, E. K. Choi, and K.-S. Cho Functional MR Imaging of Prostate Cancer RadioGraphics, January 1, 2007; 27(1): 63 - 75. [Abstract] [Full Text] [PDF]

				A. Wetter, T. A. Engl, D. Nadjmabadi, K. Fliessbach, T. Lehnert, J. Gurung, W.-D. Beecken, and T. J. Vogl Combined MRI and MR spectroscopy of the prostate before radical prostatectomy. Am. J. Roentgenol., September 1, 2006; 187(3): 724 - 730. [Abstract] [Full Text] [PDF]

				S. N. Reske, N. M. Blumstein, B. Neumaier, H.-W. Gottfried, F. Finsterbusch, D. Kocot, P. Moller, G. Glatting, and S. Perner Imaging Prostate Cancer with 11C-Choline PET/CT J. Nucl. Med., August 1, 2006; 47(8): 1249 - 1254. [Abstract] [Full Text] [PDF]