| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Departments of Radiology (U.G.M.L., D.B.V., H.H., M.G.S., A.B., J.S., A.S., R.G.M., J.K.), Urology (P.R.C.), and Pathology (I.C.), University of California–San Francisco, Magnetic Resonance Science Center, 1 Irving St, Suite AC-109, San Francisco, CA 94143-1290. Received September 26, 2000; revision requested November 12; revision received March 22, 2001; accepted April 11. Supported in part by grants NIH R01-59897 and NIH R29-64667 and the Association for the Cure of Cancer of the Prostate Foundation. Address correspondence to J.K. (e-mail: johnk@mrsc.ucsf.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Combined MR imaging/3D MR spectroscopic imaging examinations were performed in 16 hormone-treated patients and 48 nontreated matched control patients before radical prostatectomy and step-section histopathologic analysis. At MR imaging, cancer presence within the peripheral zone was assessed on a per sextant basis by two readers. At 3D MR spectroscopic imaging, cancer was identified by using (choline plus creatine)–to-citrate ratios at cutoff values of 2 and 3 SDs above mean normal peripheral zone values. Data were compared by using receiver operating characteristic analysis.
RESULTS: There was no significant difference in the ability of combined MR imaging/3D MR spectroscopic imaging to localize prostate cancer in treated versus control patients. For MR imaging alone, the sensitivity and specificity were 91% and 48% (reader 1) and 75% and 60% (reader 2) in treated patients versus 79% and 60% (reader 1) and 84% and 43% (reader 2) in control patients. For 3D MR spectroscopic imaging alone (>3 SDs cutoff), higher specificity (treated, 80%; controls, 73%) but lower sensitivity (treated, 56%; controls, 53%) was attained. In treated patients, high sensitivity or specificity (up to 92%) was achieved when either or both modalities indicated cancer.
CONCLUSION: When performed within 4 months after initiating hormone deprivation therapy, combined MR imaging/3D MR spectroscopic imaging had the same accuracy in localizing prostate cancer as in nontreated patients.
Index terms: Magnetic resonance (MR), spectroscopy, 844.12145 • Prostate, MR, 844.121411, 844.121415, 844.12145 • Prostate neoplasms, 844.32
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Currently, there is no consensus on the optimal duration of either neoadjuvant or intermittent hormone deprivation therapy (8,9). Therefore, to truly individualize treatment, a noninvasive means of assessing the effectiveness of therapy and accurately measuring the presence and local extent of residual cancer is required.
In nontreated patients with prostate cancer, the addition of metabolic data provided by using 3D proton magnetic resonance (MR) spectroscopic imaging, as compared with the data obtained by using MR imaging alone, has substantially improved both the localization of cancer to a sextant of the prostate (10) and the prediction of the extracapsular spread of cancer (11). To our knowledge, however, to date, there have been no published studies to assess the capability of combined MR imaging/3D MR spectroscopic imaging to depict the presence and extent of cancer during hormone deprivation therapy. In a previous study (12), a substantial time-dependent loss of the prostatic metabolites choline, creatine, and citrate was observed during hormone deprivation therapy, and it resulted in the complete loss of all observable metabolites (ie, total metabolic atrophy) in 25% of patients who had been receiving therapy for more than 16 weeks.
The time course of prostatic metabolite loss, like the time course of reduction in serum prostate-specific antigen levels after hormone deprivation therapy, may provide important prognostic information (7,13,14). However, the loss of prostatic metabolites with increasing duration of therapy may also reduce the accuracy of 3D MR spectroscopic imaging in depicting prostate cancer during hormone deprivation therapy. The goal of this matched case–controlled study was to determine the accuracy of combined MR imaging/3D MR spectroscopic imaging in localizing prostate cancer to a sextant of the prostate gland in patients receiving neoadjuvant hormone deprivation therapy prior to radical prostatectomy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patients Treated with Hormone Deprivation Therapy
Because the use of neoadjuvant hormone deprivation therapy prior to surgery was not a common clinical practice at our institution, only 16 patients (mean age, 60.4 years ± 6.9 [SD]) with localized cancer received neoadjuvant hormone deprivation therapy prior to radical prostatectomy during the study. All 16 patients underwent MR imaging/3D MR spectroscopic imaging after beginning hormone deprivation therapy and prior to surgery (mean time between imaging and surgery, 12 days ± 10). Five patients received standard doses of luteinizing hormone–releasing hormone agonist monotherapy with subcutaneous depot injection of either leuprolide acetate (Lupron; Tap Pharmaceuticals, Deerfield, Ill), 5–7 mg intramuscularly, or goserelin acetate (Zoladex; ICI Pharma, Wilmington, Del), 5–10 mg intramuscularly, every 28 days. Eleven patients received combined hormone deprivation therapy with standard doses of both luteinizing hormone– releasing hormone analogues (leuprolide acetate and goserelin acetate) and antiandrogens. The antiandrogen was flutamide (Eulexin; Schering, Kenilworth, NJ), 250 mg taken orally every 8 hours, or bicalutamide (Casodex; Zeneca, Wilmington, Del), 50 mg taken orally every 24 hours. The mean therapy duration was 9 weeks (range, 2–18 weeks). None of the patients had received any other treatment prior to hormone deprivation therapy or prior to the MR imaging/3D MR spectroscopic imaging examination.
At step-section histopathologic analysis performed after surgery, 12 patients (three with unilateral malignancy and nine with bilateral multifocal malignancy) had pT2 cancer and four patients (three with unilateral extracapsular spreading, one with bilateral extracapsular spreading, and none with seminal vesicle invasion) had pT3 cancer. The median Gleason score of the surgical specimens was 6.5 (range, 5–9).
Nontreated (Control) Patients
The selected control patients who underwent combined MR imaging/3D MR spectroscopic imaging were the first 48 patients with prostate cancer who had not received any therapy prior to radical prostatectomy and who matched the treated patients in age, Gleason score, and clinical tumor stage. The mean age of the control patients (60.6 years ± 6.2 [SD]) was not significantly different from that of the treated patients (P = .92, two-tailed Student t test). For each treated patient, three control patients were matched in pathologic TNM stage and Gleason score (Tables 1 and 2).
|
|
2 test). The median Gleason score of the surgical specimens obtained from the control patients was 6.5 (range, 5–10). All examinations were performed with a 1.5-T whole-body MR unit (Signa; GE Medical Systems, Milwaukee, Wis) by using a body coil for radio-frequency transmission and an expandable endorectal coil (Medrad, Pittsburgh, Pa) as part of a pelvic phased-array coil system (GE Medical Systems) for MR signal reception (15,16). The pelvic phased array consists of four external coil elements (two above and two underneath the pelvis) that provide coverage of the entire pelvis. The expandable endorectal coil plugs into the phased array and becomes part of the coil array; this system provides the sensitivity necessary to obtain high–spatial-resolution MR imaging and MR spectroscopic data from the prostate. The total examination time, including coil placement, patient positioning, and combined MR imaging/3D MR spectroscopic imaging, was 50–60 minutes.
MR Imaging
Sagittal T2-weighted fast spin-echo localizer images (1,000/85 [repetition time msec/echo time msec], 5.0-mm section thickness, 256 x 192 matrix, 1.5-mm intersection gap, 24-cm field of view, one signal acquired) were obtained to check the endorectal coil position and to define the subsequent MR imaging sequences. Transverse T1-weighted spin-echo images (600/12, 4-mm section thickness, 1-mm intersection gap, 24-cm field of view, 256 x 192 matrix, two signals acquired) were obtained from the symphysis pubis to the aortic bifurcation. Then, T2-weighted fast spin-echo images (5,000–6,000/108 [effective], echo train length of 16, 3-mm contiguous interleaved sections, right-to-left phase encoding, 14-cm field of view, 256 x 192 matrix, two signals acquired, no phase wrap) were obtained in the transverse and coronal planes through the prostate and seminal vesicles. All images were transferred offline and analytically corrected for the reception profile of the endorectal and external pelvic phased-array coils (17,18). The total time for setup and acquisition of the MR imaging data was 25–30 minutes.
Three-dimensional MR Spectroscopic Imaging
The MR spectroscopy volume was selected by using the point-resolved spectroscopic (PRESS) technique (19). The selected PRESS volume was graphically prescribed by using consecutive high-spatial-resolution transverse T2-weighted fast spin-echo images of the prostate and was evaluated by means of imaging in the transverse plane. The echo delay of the PRESS sequence (echo time = 130 msec) was optimized for the quantitative detection of both citrate and choline (16,20). For water and lipid suppression, two band selective inversion with gradient dephasing, or BASING, pulses were used with outer-volume suppression during point-resolved spectroscopic excitation (21–23). The outer-volume suppression pulses, which abutted all six sides of the PRESS volume, were applied prior to PRESS excitation. Magnetic field homogeneity was optimized by using both automated and manual shimming until a 15-Hz or smaller water line width was attained. Finally, phase encoding was applied in three dimensions to produce 3D MR spectroscopic arrays of proton spectra with a nominal resolution of 0.24 cm3 from throughout the prostate.
Three-dimensional MR spectroscopic imaging data sets were acquired by using 1,000/130, a spectral width of 1,250 Hz, 512 complex points, and either 8 x 8 x 8 spectra with two signals acquired or 16 x 8 x 8 spectra with one signal acquired, depending on the dimensions of the prostate (data acquisition time = 17 minutes). Forty to 320 of the 512 or 1,024 1H MR spectra obtained were acquired from within the prostate. Typically, 75%–100% of the peripheral zone of the prostate was imaged at 3D MR spectroscopic imaging. The total time for setup and acquisition of the 3D MR spectroscopic imaging data was 25–30 minutes.
Processing of 3D MR Spectroscopic Imaging Data
All data were transferred offline and processed on a workstation (UltraSparc; SUN Microsystems, Mountain View, Calif) by using software developed for 3D MR spectroscopic imaging studies in our laboratory. Spectral data sets were zero filled to 1,024 points, multiplied by a 2-Hz Lorentzian function, and Fourier transformed in the time domain and in three spatial domains. The 3D MR spectroscopic imaging data were analytically corrected for the radio-frequency magnetic induction field B1 reception profile of the pelvic phased-array endorectal coil system (17,18). After phase, frequency, and baseline corrections were performed, the corrected spectra were analyzed to provide estimates of peak integral areas and random noise (and hence the SD in estimates) for choline, creatine, and citrate resonances for each 0.24 cm3 voxel (16,24,25).
MR Image Evaluation
All MR images were displayed offline on the monitor of the workstation by using image display language–based software (Research Systems, Boulder, Colo) developed in our laboratory. All patient identification information except the examination number was removed. MR images were randomized according to examination number and interpreted retrospectively by two independent readers (U.G.M.L. and J.S.). Both readers knew that the patients had biopsy-proven cancer, but they were unaware of the patients’ treatment status and all other clinical, 3D MR spectroscopic imaging, and histopathologic findings. Reader 1 (J.S.), having interpreted at least twice as many prostate MR imaging studies (more than 500 during the preceding 5 years) as reader 2 (U.G.M.L.), was more experienced. MR images were considered to be of diagnostic quality when the gland and pelvis were adequately depicted and the images did not have large or extensive motion artifacts or artifacts arising from a hip prosthesis.
For direct comparison of the MR imaging, 3D MR spectroscopic imaging, and histopathologic findings, the peripheral zone of the prostate was evaluated on a per-sextant basis (10). The right and left prostatic bases, middle gland, and apex were defined on transverse T2-weighted fast spin-echo images. The base of the prostate extended from the bladder floor and seminal vesicle–ejaculatory duct junctions to the level cranial to the transverse MR image section with the largest transverse diameter of the prostate. The middle gland extended from the level of the largest transverse diameter of the prostate to the caudal (inferior) level of the verumontanum, and the apex extended from the next caudal level to the urogenital diaphragm. The MR image positions of the base, middle gland, and apex were noted separately for subsequent reference during the interpretation of 3D MR spectroscopic imaging findings. The confines of the peripheral zone were defined by the low-signal-intensity bands of the prostatic capsule and pseudocapsule on transverse T2-weighted fast spin-echo images (26,27).
The presence of prostate cancer, identified as an area of decreased signal intensity within the peripheral zone on T2-weighted images, was assessed by each of the two readers independently. For each sextant, the likelihood of tumor presence was rated separately by using the following five-point scale: A rating of 1 meant definitely no cancer; 2, probably no cancer; 3, indeterminate findings; 4, probably cancer; and 5, definitely cancer. The level of confidence in detecting cancer was based on the clarity of the T2 signal intensity decrease and on the shape and conspicuity of the lesion at T2-weighted imaging. A more rounded smudgy-appearing lesion at T2-weighted imaging was associated with a higher probability of being cancer, whereas more triangular streaky lesions were associated with other processes, such as prostatitis or postbiopsy hemorrhage.
Although extracapsular spreading of cancer and seminal vesicle invasion were identified for tumor staging by using previously reported imaging criteria (28), these findings were not evaluated further in this study. The central gland was not evaluated for possible cancer, because earlier reports indicate the inaccuracy of MR imaging (29) and MR spectroscopy (16) in depicting cancer in this region of the prostate. For direct comparison between MR imaging and 3D MR spectroscopic imaging data, only those sextants that were covered by 3D MR spectroscopic imaging were evaluated on the MR images. All findings were entered onto a standardized spreadsheet that corresponded to the histopathologic data forms and that was developed for this study.
Three-dimensional MR Spectroscopic Image Evaluation
Using the gradients used for MR image evaluation and a constant patient position during combined MR imaging/3D MR spectroscopic imaging facilitated spatial alignment of the MR imaging and 3D MR spectroscopic imaging data. The outline of the PRESS volume and the 3D MR spectroscopic imaging phase-encoding grid were superimposed on the transverse T2-weighted fast spin-echo images, and the corresponding spectral arrays were plotted. The imaging levels of the prostatic base, middle gland, and apex were the same as those used for the MR image evaluations.
For each sextant, the likelihood of tumor presence was determined by using the (choline plus creatine)–to-citrate ratio (CC/C) as follows: As previously described, 3D MR spectroscopic imaging voxels were considered to be usable when at least 75% of a voxel was within the peripheral zone, when they were not contaminated by insufficiently suppressed water or lipid, and when they did not include tissue surrounding the urethra and ejaculatory ducts (12). The determination of whether a voxel was useable was made by consensus between two readers (J.K. and D.V.), each of whom had more than 6 years experience in interpreting 3D MR spectroscopic imaging data of the prostate. For all useable spectral voxels, choline and citrate peak area–to-noise, choline-to-creatine, and (choline plus creatine)–to-citrate peak area ratios were calculated. Spectral voxels in which the choline and citrate peak area–to-noise ratios were below 5 were considered to be absent of metabolites, and this phenomenon, owing to its similarity to morphologic atrophy, was called metabolic atrophy (12). Spectra that demonstrated metabolic atrophy were interpreted as normal for purposes of comparison with the histopathologic findings.
For voxels that demonstrated detectable prostatic metabolites, the CC/C ratio was calculated. Two cutoffs—that is, a CC/C of more than 2 SDs and that of more than 3 SDs above the mean value for normal peripheral zone tissue—were used to identify prostate cancer on the basis of metabolic findings (16,30). The mean CC/C ratio for normal tissue prior to therapy and with increasing therapy duration was determined in a previous MR imaging/3D MR spectroscopic imaging study (12). The mean normal-tissue CC/C ratio used to identify cancer was adjusted for duration of treatment, because it was previously observed that citrate levels decreased faster than did choline and creatine levels during hormone deprivation therapy (12). The faster decrease in citrate compared with choline and creatine resulted in an increase in the mean CC/C ratio for normal tissue with increasing therapy duration: 0.24 ± 0.13 before hormone deprivation therapy, 0.33 ± 0.17 with 1–6 weeks of therapy, 0.41 ± 0.32 with 7–16 weeks of therapy, and 0.40 ± 0.21 with more than 16 weeks of therapy. In voxels without detectable citrate (ie, citrate peak area–to-noise ratio, <5) (12), elevated choline (choline signal-to-noise ratio, >5 and choline-to-creatine ratio, >1) was used as a marker of cancer presence.
For descriptive statistical analysis and receiver operating characteristic (ROC) analysis, the 3D MR spectroscopic imaging results for each sextant of the peripheral zone of the prostate were rated in a manner similar to that used to rate the MR imaging findings: A rating of 5 was given when the CC/C ratio was more than 3 SDs above the normal population average; a rating of 3, when the CC/C was more than 2 SDs above the normal population average or the choline-to-creatine ratio was elevated; and a rating of 1 when the CC/C ratio was less than 2 SDs or above the normal population average.
Pathologic Review
After radical prostatectomy, each prostate was coated with India ink and fixed in 10% buffered formaldehyde. Transverse step sections were obtained at 3–4-mm intervals in a plane perpendicular to the long axis (base-to-apex) of the prostate and its posterior surface. Each section was subdivided into quarter slices by cuts in the frontal and sagittal planes to fit specimen cassettes. All prostate blocks were labeled according to location to allow whole-mount reconstruction, if necessary. The presence and extent of cancer for each block were determined by the same pathologist (I.C.) and entered onto standardized pathology forms with labeling such that the tumor locations could be reconstructed.
Correlation of MR imaging and 3D MR Spectroscopic Imaging Results with Histopathologic Findings
One of the authors (A.B.), who did not participate in the MR imaging and 3D MR spectroscopic imaging interpretations, maintained the database of histopathologic information. With guidance from the schematic representation of the prostate specimen, histopathologic findings were summarized on a per-sextant basis. Tumor sites on the MR or 3D MR spectroscopic images were considered to match the histopathologic findings when the tumor was present in the peripheral zone of the same sextant of the prostate (ie, right or left base, right or left middle gland, or right or left apex) within the range of one MR imaging section (ie, ±3–4-mm craniocaudal distance). In addition, small tumors—that is, those occupying less than 25% of a sextant—were required to be in the same anterior or posterior location of the sextant.
Statistical Analyses
On the basis of histopathologic findings, the MR imaging and 3D MR spectroscopic imaging findings for each sextant were encoded individually as true-positive, true-negative, false-positive, or false-negative. For descriptive statistical data, including sensitivity and specificity with corresponding 95% CIs, MR imaging findings were dichotomized. Scores of 1–3 represented a finding of no cancer in the sextant, whereas scores of 4 and 5 represented a finding of cancer in the sextant. Three-dimensional MR spectroscopic imaging data were evaluated twice: once with scores 3 and 5 representing cancer and once with only a score of 5 representing cancer.
ROC analysis was performed to compare the results of MR imaging alone and combined MR imaging and 3D MR spectroscopic imaging between the treated and nontreated (control) patients. A P value of less than .05 indicated a significant difference (two-tailed test based on the z statistic) (31,32). For combined MR imaging and 3D MR spectroscopic imaging, half of the MR imaging score (1 to 5) was added to half of the MR spectroscopic imaging score (1, 3, or 5). The results were rounded to the next whole number.
The 2 test with the Yates correction for continuity was performed to determine significant differences (P < .025) between the frequency of MR imaging findings and the frequency of 3D MR spectroscopic imaging findings, and their combinations, between the treated and control patients. In addition, whether the number of false or equivocal interpretations correlated with the therapy duration, type of therapy, Gleason score, or tumor location was determined. When the expected number of findings was fewer than five in any cell of the 2 x 2 table, the two-tailed Fisher exact test was performed instead of the 2 test (P < .025). Interreader agreement on MR image interpretation was quantified by using 
statistics, with a value of less than 0.20 considered to indicate poor agreement; a value of 0.20–0.39, fair agreement; a value of 0.40–0.59, moderate agreement, a value of 0.60–0.79, substantial agreement; and a value of 0.80 or greater, excellent agreement.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2 = 0.6834; P = .42). The data obtained enabled the analysis of differences between the treated and control patients by using one sextant per patient with a power of 80%, based on the z statistic. Individual patients had diverse morphologic and metabolic responses to hormone deprivation therapy, with the magnitude of responses depending on the type (single [monotherapy] or combined agents) and therapy duration. Figures 1–3 show representative transverse T2-weighted images and the corresponding proton spectral arrays obtained in a patient prior to hormone deprivation therapy, a patient after 9 weeks of monotherapy with leuprolide acetate, and a patient after 12 weeks of combined therapy (goserelin acetate plus bicalutamide). In the patient who received shorter-duration monotherapy, there were no visible prostate metabolic or morphologic effects of therapy. In contrast, long-term hormone deprivation therapy had a dramatic effect on both prostate morphologic features and metabolite levels.
|
|
|
|
|
|
|
|
|
In addition, in many voxels, there was a complete loss of all prostatic metabolites (ie, metabolic atrophy), which resulted in complete metabolic atrophy of an entire sextant in 16 (20%) of 80 cases in the treated patients—a significantly (2 = 26.3076; P < .001) greater number of cases of complete metabolic atrophy than that in the control patients: six (2.5%) of 242. In addition, metabolic atrophy was significantly more frequent in the histopathologically negative sextants (10 of 25 [40%]) than in the histopathologically positive sextants (six of 55 [11%]; 2 = 7.3636; P = .008).
Localization of Cancer at MR Imaging
The sensitivities and specificities achieved by the two independent readers in the interpretation of MR images obtained in the treated and control patients are summarized in Table 3. Neither the mean area under the ROC curve (Az) for MR imaging findings detected by reader 1 nor that for findings detected by reader 2 was significantly different between the treated (Fig 4) and control (Fig 5) patients: The mean Az obtained by reader 1 was 0.731 ± 0.066 for the treated group and 0.711 ± 0.035 for the control group (z = 0.2535; P > .5). The mean Az obtained by reader 2 was 0.697 ± 0.067 for the treated group and 0.639 ± 0.037 for the control group (z = 0.7674; P = .4). Readers 1 and 2 had fair agreement ( = 0.32) regarding findings in 57 (71%) of 80 sextants in the treated patients and moderate agreement ( = 0.43) regarding findings in 183 (76%) of 242 sextants in the control patients.
|
|
|
2 = 0.0548; P > .5) (Table 3); the specificity was 80% in the treated patients and 73% in the control patients (2 = 0.1753; P > .5) (Table 3). Accuracy was not significantly different between the treated (51 of 80 sextants [64%]) and control (147 of 242 sextants [61%]) patients (2 = 0.1200; P > .5). When the cutoff level was lowered to 2 SDs above the normal values, sensitivity increased to 80% in the treated patients and 80% in the control patients (2 = 0.0242; P > .5). However, specificity decreased to 60% in the treated patients and to 61% in the control patients (2 = 0.0171; P > .5). Accuracy increased to 74% (59 of 80 sextants) in the treated patients and to 73% (177 of 242 sextants) in the control patients (2 = 0.0015; P > .5 [not significant]) (Table 3).
Localization of Cancer by Using Combined MR Imaging/3D MR Spectroscopic Imaging
The addition of 3D MR spectroscopic imaging data improved the overall accuracy of MR imaging both in the treated and the control patients. For reader 1, the Az values improved markedly: The mean Az was 0.808 ± 0.052 for treated patients (Fig 4) and 0.759 ± 0.033 for control patients (Fig 5). For reader 2, the Az values improved to 0.795 ± 0.058 for treated patients (z = 2.0054; P = .048) and to 0.746 ± 0.028 for control patients (z = 4.0292; P < .001). The Az values for combined MR imaging/3D MR spectroscopic imaging were not significantly different between the treated and control patients (reader 1: z = 0.7990, P = .43; reader 2: z = 0.7763, P = .45).
With the greater than 2 SD cutoff, the addition of 3D MR spectroscopic imaging data to the MR imaging data reduced sensitivity in both the treated and control patients by 6%–10%. However, the resulting gain in specificity (16%–28%) was more pronounced (Table 3). The highest specificity was achieved with the combination of an MR imaging interpretation of probable or definite cancer with a 3D MR spectroscopic imaging cutoff of more than 3 SDs. For treated patients, reader 1 achieved a specificity of 92%, sensitivity of 51%, and positive predictive value of 93%, whereas reader 2 achieved a specificity of 88%, sensitivity of 42%, and positive predictive value of 89%. For the control patients, reader 1 achieved a specificity of 86%, sensitivity of 43%, and positive predictive value of 83%, whereas reader 2 achieved a specificity of 81%, sensitivity of 46%, and positive predictive value of 81% (Figs 4, 5).
The highest sensitivity was obtained with the combination of an MR imaging interpretation of probable or definite cancer or by using a 3D MR spectroscopic imaging cutoff of more than 2 SDs. For the treated patients, reader 1 achieved a sensitivity of 98%, specificity of 24%, and negative predictive value of 86%, whereas reader 2 achieved values of 98%, 32%, and 89%, respectively. For the control patients, reader 1 achieved a sensitivity of 93%, specificity of 40%, and negative predictive value of 78%, whereas reader 2 achieved values of 95%, 30%, and 77%, respectively.
False Interpretations at Combined MR Imaging/3D MR Spectroscopic Imaging
In the treated group, five (6%) of 80 sextants in five different patients showed findings that were false for cancer by combined MR imaging/3D MR spectroscopic imaging, as compared with the results of histopathologic analysis. There was no obvious reason (eg, therapy duration, type of therapy, Gleason score, tumor location, or metabolic atrophy) that explained the false calls with either modality. In the control group, 33 (14%) of 242 sextants in 21 different patients had false findings at both MR imaging and 3D MR spectroscopic imaging; this number was not significantly different from that for the treated group (2 = 2.4818; P = .115). In seven patients, more than one sextant was involved. As in the treated patients, there was no obvious reason for the false calls with either modality.
Equivocal Interpretations at Combined MR Imaging/3D MR Spectroscopic Imaging
For the treated patients, reader 1 had equivocal interpretations of findings—that is, those with a combined MR imaging and 3D MR spectroscopic imaging score of 2.5 or 3.0—in 21 (26%) of 80 sextants (nine histopathologically positive, 12 histopathologically negative) and reader 2 had equivocal calls in 29 (36%) of 80 sextants (17 histopathologically positive, 12 histopathologically negative). For the control patients, reader 1 had equivocal interpretations in 61 (25%) of 242 sextants (27 histopathologically positive, 34 histopathologically negative) and reader 2 had equivocal calls in 64 (26%) of 242 sextants (27 histopathologically positive, 37 histopathologically negative). For both readers, MR imaging was correct as often as 3D MR spectroscopic imaging was. The frequency of equivocal calls did not differ significantly between the treated and control patients (reader 1: 2 = 0.0014, P > .5; reader 2: 2 = 2.3563, P = .15). There was no obvious reason (eg, therapy duration, type of therapy, Gleason score, tumor location, or metabolic atrophy) to explain the equivocal calls with either modality for either the treated or control patients.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In nontreated patients, the combination of MR imaging and 3D MR spectroscopic imaging is more accurate than MR imaging alone in localizing cancer to a sextant of the gland (10). However, to our knowledge, the accuracy of combined MR imaging/3D MR spectroscopic imaging in depicting cancer after hormone deprivation therapy has not been established. Results of preliminary MR imaging studies performed in patients undergoing hormone deprivation therapy have indicated that the accuracy of cancer detection may be reduced. Chen et al (33) reported an overestimation of tumor presence with MR imaging alone performed in patients who had undergone 1–7 months of therapy. Recent study results also suggest that the accuracy of cancer detection achieved by using the CC/C ratio during hormone deprivation therapy may be reduced (12).
Mueller-Lisse et al (12) reported a significant (P < .05) time-dependent loss of prostatic choline, creatine, and citrate during hormone deprivation therapy. This loss resulted in the complete loss of all MR spectroscopically observable metabolites (ie, total metabolic atrophy) in 25% of patients who had received more than 16 weeks of therapy. In addition, it was observed that citrate levels decreased faster than did choline and creatine levels during therapy. This faster citrate level decrease resulted in an increase in the mean CC/C ratio in both normal and malignant tissue with increasing therapy duration and in a loss of all MR spectroscopically detectable citrate in 69% of the patients who had received more than 16 weeks of therapy. In the present work, we report—to our knowledge, for the first time—the accuracy of combined MR imaging/3D MR spectroscopic imaging in the localization of prostate cancer to a sextant of the prostate after short-to-intermediate–term hormone deprivation therapy.
The most important finding of our study was that the sensitivity, specificity, and overall accuracy of cancer localization to a sextant of the prostate by combined MR imaging/3D MR spectroscopic imaging in patients without hormone deprivation therapy was equivalent to those in patients who had received up to 4 months of therapy. Unlike Chen et al (33), who reported an overestimation of tumor presence with MR imaging in 22 patients who received therapy, both MR image readers in our study achieved an equivalent overall accuracy in the interpretation of images obtained in the treated and control patients. This discrepancy between studies may be explained by differences in therapy duration between the patient populations. The therapy duration in our study was 2–18 weeks, whereas that in the Chen et al (33) study was 4–30 weeks. The discrepancy could be due in part also to differences in reader experience and image quality. Our study results suggest that follow-up MR imaging performed within the first 4 months after therapy is as accurate in the localization of prostate cancer as the baseline MR examination, if the pre- and posttherapy images are interpreted by the same reader.
Another important finding of our study was that in treated patients, the presence of metabolic atrophy was significantly more frequent (P = .008) in normal peripheral zone tissue than in cancerous tissue. Therefore, at early times after therapy, metabolic atrophy may be useful as a substitute marker for normal tissue when the CC/C ratio cannot be determined due to low metabolite levels. However, with increasing therapy duration, metabolic atrophy is observed prominently in malignant tissue also (12). It has been suggested previously that in the absence of detectable citrate, elevated choline levels (ie, choline signal-to-noise ratio >5 and choline-to-creatine ratio >1) may be useful in detecting residual prostate cancer after therapy (12). The findings of our study did not support this hypothesis: Elevated choline levels alone were not indicative of residual prostate cancer at this early time after therapy. Due to the small number of sextants with undetectable citrate levels but elevated choline levels in our study, further studies are required before a definite conclusion about the accuracy of elevated choline as a marker for residual cancer can be drawn.
As in a previous study involving nontreated patients with localized cancer (10), the combination of 3D MR spectroscopic imaging and MR imaging markedly improved ROC analysis results. For the less experienced reader, the approximately 10% gain in accuracy achieved with this combination, as compared with the accuracy of MR imaging alone, determined by using ROC analysis was statistically significant in both the treated and nontreated patients. For the more experienced reader, the results improved by about 5%–8%, which was almost but not significant. These findings demonstrate that MR imaging combined with 3D MR spectroscopy is more accurate than MR imaging alone in the detection and localization of cancer after up to 4 months of hormone deprivation therapy.
In addition, it has been shown that for nontreated patients with localized cancer, various combinations of MR imaging and 3D MR spectroscopy results can help in choosing the points on the ROC curve that provide high sensitivity or high specificity, depending on clinical demands (10). In the present study, this capability was extended to patients who had received up to 4 months of therapy for localized cancer. In both the nontreated and treated groups, high specificity (81%–92%) was achieved when both MR imaging and 3D MR spectroscopic imaging depicted cancer. A positive result of combined MR imaging/3D MR spectroscopic imaging at greater than 3 SDs above the mean value for normal tissue in the peripheral zone indicated the presence of cancer with high probability (positive predictive value, 81%–93%). A positive result of either examination provided high sensitivity (93%–98%). Conversely, a negative result of combined MR imaging/3D MR spectroscopic imaging at less than 2 SDs above the mean value excluded the presence of cancer with high probability (negative predictive value, 77%–89%). However, as indicated in previous studies by Chen et al (33) and Mueller-Lisse et al (12), the most dramatic morphologic and metabolic consequences of hormone deprivation therapy occur at therapy durations of longer than 9 weeks, which was the mean therapy duration in this study. Therefore, we believe that the sensitivity and specificity of combined MR imaging and 3D MR spectroscopy for the localization of cancer to a sextant of the prostate decreases with increasing therapy duration.
The analysis of false calls at both MR imaging and 3D MR spectroscopy revealed that no systematic method or interpretation errors occurred in either the treated or control cases. The frequency of false calls did not differ significantly between the two groups; this indicates that the reliability of combined MR imaging/3D MR spectroscopic imaging remains unchanged, at least during the first 4 months after treatment. The frequency of equivocal interpretations was similar between the treated and control cases; this indicates that short-term therapy does not interfere with the capability of combined MR imaging and 3D MR spectroscopy in enabling one to reach a diagnostic decision.
Recent technical improvements in MR spectroscopic imaging techniques may reduce the number of false and equivocal calls. For example, the current use of very selective outer-voxel suppression pulses to better conform the point-resolved spectroscopy–selected volume to the prostate has allowed complete coverage of the prostate at imaging, with reduced contamination from lipid surrounding the prostate (34). In addition, the sensitivity and specificity of MR spectroscopic identification of cancer could be increased by increasing the number of metabolic markers for cancer and by determining the optimal combination of MR imaging and MR spectroscopy criteria for cancer detection. In two recent studies, the polyamine spermine was identified as a new metabolic marker for prostate cancer (35,36).
In conclusion, results of this matched case–controlled study demonstrate that MR imaging combined with 3D proton MR spectroscopy of the prostate performed within the first 4 months after hormone deprivation therapy for localized prostate cancer is as accurate, reliable, and likely to help reach a diagnostic decision as that performed prior to therapy. The addition of 3D MR spectroscopic imaging to MR imaging improved the detection and localization of prostate cancer to a sextant of the prostate in both treated and nontreated patients. Our study results demonstrate the potential usefulness of combined MR imaging/3D MR spectroscopic imaging for monitoring the efficacy of hormone deprivation therapy in patients with localized prostate cancer.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, J.K.; study concepts and design, U.G.M.L., J.K., D.B.V., H.H.; literature research, A.S., U.G.M.L.; clinical studies, H.H., P.R.C., I.C.; data acquisition, M.G.S., R.G.M., A.B.; data analysis/interpretation, U.G.M.L., J.K., J.S.; statistical analysis, U.G.M.L., A.S.; manuscript preparation, U.G.M.L., J.K., M.G.S., A.B.; manuscript definition of intellectual content, U.G.M.L., J.K., D.B.V.; manuscript editing, J.K., D.B.V., M.G.S.; manuscript revision/review, U.G.M.L., D.B.V., M.G.S., J.K.; manuscript final version approval, U.G.M.L., J.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. Casciani, E. Polettini, E. Carmenini, I. Floriani, G. Masselli, L. Bertini, and G. F. Gualdi Endorectal and Dynamic Contrast-Enhanced MRI for Detection of Local Recurrence After Radical Prostatectomy Am. J. Roentgenol., May 1, 2008; 190(5): 1187 - 1192. [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]
|
|
|
|
|
|
D. J. A. Margolis, J. M. Hoffman, R. J. Herfkens, R. B. Jeffrey, A. Quon, and S. S. Gambhir Molecular Imaging Techniques in Body Imaging Radiology, November 1, 2007; 245(2): 333 - 356. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Kundra, P. M. Silverman, S. F. Matin, and H. Choi Imaging in Oncology from The University of Texas M. D. Anderson Cancer Center: Diagnosis, Staging, and Surveillance of Prostate Cancer Am. J. Roentgenol., October 1, 2007; 189(4): 830 - 844. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. Bauer, T. D. Henning, D. Mueller, Y. Lu, S. Majumdar, and T. M. Link Volumetric Quantitative CT of the Spine and Hip Derived from Contrast-Enhanced MDCT: Conversion Factors Am. J. Roentgenol., May 1, 2007; 188(5): 1294 - 1301. [Abstract] [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]
|
|
|
|
|
|
A. Miguel-Dasit, L. Marti-Bonmati, R. Aleixandre, P. Sanfeliu, and D. Bautista Publication of Material Presented at Radiologic Meetings: Authors' Country and International Collaboration Radiology, May 1, 2006; 239(2): 521 - 528. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Pucar, A. Shukla-Dave, H. Hricak, C. S. Moskowitz, K. Kuroiwa, S. Olgac, L. E. Ebora, P. T. Scardino, J. A. Koutcher, and K. L. Zakian Prostate Cancer: Correlation of MR Imaging and MR Spectroscopy with Pathologic Findings after Radiation Therapy-Initial Experience Radiology, August 1, 2005; 236(2): 545 - 553. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. V. Coakley, H. S. Teh, A. Qayyum, M. G. Swanson, Y. Lu, M. Roach III, B. Pickett, K. Shinohara, D. B. Vigneron, and J. Kurhanewicz Endorectal MR Imaging and MR Spectroscopic Imaging for Locally Recurrent Prostate Cancer after External Beam Radiation Therapy: Preliminary Experience Radiology, November 1, 2004; 233(2): 441 - 448. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. M. Link, B. B. Koppers, T. Licht, J. Bauer, Y. Lu, and E. J. Rummeny In Vitro and in Vivo Spiral CT to Determine Bone Mineral Density: Initial Experience in Patients at Risk for Osteoporosis Radiology, June 1, 2004; 231(3): 805 - 811. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) and for combined MR imaging/3D MR spectroscopic imaging (
) in patients receiving hormone deprivation therapy. The Az values for combined MR imaging/3D MR spectroscopic imaging were larger than those for MR imaging alone (P = .100 for reader 1, P = .048 for reader 2).