| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Departments of Radiology (F.V.C., H.S.T., A.Q., M.G.S., D.B.V., J.K.), Epidemiology and Biostatistics (Y.L.), Radiation Oncology (M.R., B.P.), and Urology (K.S.), University of California San Francisco, 505 Parnassus Ave, Box 0628, M-372, San Francisco, CA 94143-0628. Received December 22, 2003; revision requested February 26, 2004; revision received March 8; accepted April 1. Supported by NIH grants CA79980 and CA59897. Address correspondence to F.V.C. (e-mail: fergus.coakley@radiology.ucsf.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Endorectal MR imaging and MR spectroscopic imaging were performed in 21 patients with biochemical failure after external beam radiation therapy for prostate cancer. Two readers independently and retrospectively reviewed MR images and rated the likelihood of recurrent tumor on a five-point scale. Spectroscopic voxels were considered suspicious for malignancy if the choline level was elevated and citrate was absent. Receiver operating characteristic curve analysis was used to assess cancer detection in each side of the prostate with endorectal MR imaging and spectroscopic imaging at different thresholds based on the scores assigned by the two readers and on the number of suspicious voxels in each hemiprostate, respectively. The presence or absence of cancer at subsequent transrectal biopsy was used as the standard of reference.
RESULTS: Biopsy demonstrated locally recurrent prostate cancer in nine hemiprostates in six patients. The area under the receiver operating characteristic curve for the detection of locally recurrent cancer with MR imaging was 0.49 and 0.51 for readers 1 and 2, respectively. By using the number of suspicious voxels to define different diagnostic thresholds, the area under the receiver operating characteristic curve for MR spectroscopic imaging was significantly (P < .005) higher, at 0.81. In particular, the presence of three or more suspicious voxels in a hemiprostate showed a sensitivity and specificity of 89% and 82%, respectively, for the diagnosis of local recurrence. Seven hemiprostates demonstrated complete metabolic atrophy at spectroscopic imaging and only postirradiation atrophy at biopsy.
CONCLUSION: Preliminary data suggest that MR spectroscopic imaging, but not endorectal MR imaging, may be of value for the depiction of locally recurrent prostate cancer after radiation therapy.
© RSNA, 2004
Index terms: Magnetic resonance (MR), spectroscopy, 844.12145 • Prostate neoplasms, MR, 844.121411, 844.12145 • Radiation therapy, 844.1299
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
A total of 21 patients were identified and included in the study population. Mean patient age was 68 years (range, 45–80 years). Patients who underwent radiation therapy at our institution (n = 12) or in whom we were able to confirm radiation dose delivered at an outside institution (n = 4) received a mean dose of 74.7 Gy (range, 68.4–81.0 Gy); the dose administered to five patients treated at outside institutions was unknown. The pretreatment clinical stage of these patients was T1 (n = 2), T2 (n = 6), T3 (n = 5), or unknown (n = 8). Eleven patients underwent adjuvant hormonal therapy after external beam radiation therapy. The median interval from radiation therapy to MR imaging was 29 months (range, 14–48 months), and that from MR imaging to subsequent biopsy was 100 days (range, 0–363 days). All biopsies were performed by using a previously described technique (16). Biopsy was performed in a standard sextant fashion and was not affected or targeted by MR findings. None of the patients commenced hormonal therapy in the interval between imaging and biopsy.
These patients were referred to the department of radiation oncology because of biochemical failure after radiation therapy. Biochemical failure was defined by using the American Society for Therapeutic Radiology and Oncology definition of three consecutive increases in serum prostate-specific antigen level after a nadir has been reached (3). Such patients were then referred for MR imaging and MR spectroscopic imaging. The mean serum prostate-specific antigen level prior to MR imaging was 2.3 ng/mL (range, 0.4–4.8 ng/mL). None of the patients had a palpable cancer recurrence at digital rectal examination. One of the authors (B.P.) reviewed the medical records of all patients to determine the trend in prostate-specific antigen level after MR imaging and to determine what additional treatment, if any, was administered after biopsy.
MR Technique
MR imaging was performed on a 1.5-T whole-body MR imager (Signa; GE Medical Systems, Milwaukee, Wis) by using a previously described technique (17,18). In brief, patients were examined in the supine position. A body coil was used for excitation, and a pelvic phased-array coil (GE Medical Systems) was used in combination with a commercially available balloon-covered expandable endorectal coil (Medrad, Pittsburgh, Pa) for signal reception.
Transverse T1-weighted spin-echo images were obtained from the aortic bifurcation to the symphysis pubis by using the following parameters: repetition time msec/echo time msec, 700/8; section thickness, 5 mm; intersection gap, 1 mm; field of view, 24 cm; matrix, 256 x 192; frequency direction, transverse (to prevent obscuration of pelvic nodes by endorectal coil motion artifact); and number of signals acquired, one.
Thin-section high-spatial-resolution transverse and coronal T2-weighted fast spin-echo images of the prostate and seminal vesicles were obtained by using the following parameters: 5000/96 (effective); echo train length, 16; section thickness, 3 mm; no intersection gap; field of view, 14 cm; matrix, 256 x 192; frequency direction, anteroposterior (to prevent obscuration of the prostate by endorectal coil motion artifact); and number of signals acquired, three.
Three-dimensional proton MR spectroscopic imaging was performed by using a water- and lipid-suppressed double spin-echo point-resolved spectroscopic sequence (PRESS). The volume for MR spectroscopic imaging was selected to maximize coverage of the prostate while minimizing inclusion of periprostatic fat and rectal air. Three pairs of outer-volume saturation bands were placed symmetrically around the prescribed spectroscopic volume prior to excitation. Magnetic field homogeneity was optimized for the selected volume by using both automated and manual shimming until a water line width of 15 Hz or less was attained. Water and lipid suppression was achieved by using band-selective inversion with gradient dephasing (19). A spectroscopic volume (transverse x anteroposterior x craniocaudal measurements, 110 x 55 x 55 mm) was centered in the PRESS volume and phase encoded in three dimensions with 16 x 8 x 8 phase-encoding steps and with one signal acquired. This resulted in spectroscopic measurements of 1024 voxels per patient, with a nominal spectral resolution of 0.32 cm3. PRESS imaging parameters were 1000/130 with a 17-minute acquisition time (an echo time of 130 msec is optimal for detection of choline and citrate [17]). Spectroscopic data were processed by using a combination of in-house software and commercial software tools (IDL; RSI, Boulder, Colo). Spectral data were apodized with a 1-Hz Gaussian function and Fourier transformed in the time domain and three spatial domains. The resultant data were zero-filled once in the time domain (1024 data points), after which the central 50% of each spectrum was extracted to obtain 512 data points across a 625-Hz spectral width.
MR Image Interpretation
Two attending radiologists (F.V.C., A.Q., with 8 and 5 years of experience, respectively, in the interpretation of endorectal MR and MR spectroscopic images of the prostate) independently reviewed all of the MR images without knowledge of the MR spectroscopic findings. Studies were reviewed at a picture archiving and communication system workstation (Impax; Agfa, Mortsel, Belgium). Readers were aware that patients had prostate cancer treated with external beam radiation therapy but unaware of all other clinical and histopathologic findings. Readers rated the overall quality of each MR imaging study as excellent, intermediate, or nondiagnostic. Quality was considered excellent if the images showed high spatial and contrast resolution and absence of artifacts, or nondiagnostic if poor resolution and/or extensive artifacts precluded meaningful assessment. All imaging studies that were not rated as either excellent or nondiagnostic were considered to be of intermediate quality. Readers also rated both the degree of postirradiation zonal indistinctness and the degree of diffuse reduction of T2 signal intensity in the gland as absent, intermediate, or marked. Zonal indistinctness was considered absent if the prostate demonstrated zonal differentiation similar to that seen in nonirradiated prostate. Zonal indistinctness was considered marked if the peripheral zone and transitional zone were nearly indistinguishable. All other degrees of zonal indistinctness were considered intermediate. Diffusely reduced T2 signal intensity was considered absent if the T2 signal intensity in the prostate resembled that in a nonirradiated gland. Diffusely reduced T2 signal intensity was considered marked if the T2 signal intensity in the prostate was similar to or lower than that in nearby nonirradiated muscle. All other degrees of diffusely reduced T2 signal intensity were considered intermediate. Readers then subjectively rated the likelihood of malignancy in each side of the prostate on the following five-point scale: 1, definitely absent; 2, probably absent; 3, indeterminate; 4, probably present; and 5, definitely present. Malignancy was defined as focal or masslike T2 signal hypointensity in the peripheral zone of the prostate (12–14). Although the rating for likelihood of malignancy was subjective, images with a score of 1 were characterized by uniform T2 signal intensity throughout the peripheral zone, without focal abnormality; images with a score of 3 were characterized by an ill-defined area of questionably reduced T2 signal intensity; and images with a score of 5 were characterized by an unequivocal masslike region of reduced T2 signal intensity.
MR Spectroscopic Image Evaluation and Interpretation
A spectroscopist (J.K.) with 12 years of experience in the performance and interpretation of MR spectroscopic imaging evaluated the spectra after baseline correction and automated frequency shift to optimize the alignment of statistically significant peaks (signal-to-noise ratio, >5:1) in the magnitude spectrum with the expected locations of choline, creatine, citrate, and residual water resonances. Subsequently, the spectra were phased by using the phase of the residual water and metabolite resonances and were baseline corrected by using local nonlinear fitting to the nonpeak regions of the spectra. The quality of each MR spectroscopic study was rated as excellent, good, fair, or poor on the basis of signal-to-noise ratio, overall shim, and the presence or absence of water- and/or lipid-induced baseline distortions. Specifically, a study was considered of excellent quality if the signal-to-noise ratios of all metabolites were greater than 10, all metabolic resonances were well resolved, and there were no baseline spectral distortions due to residual signal components in water or lipids. A study was considered of good quality if the signal-to-noise ratios of all metabolites were between 8 and 10, all metabolic resonances were reasonably well resolved, and there were minimal baseline spectral distortions due to residual signal components in water or lipids. Studies with lower signal-to-noise ratios were considered of fair quality if there was no spectral contamination from the lipid resonance. Studies with substantial spectral contamination were considered to be of poor quality.
After evaluation of quality, spectra were analyzed with a quantification algorithm that provided estimates of metabolite peak areas and random noise. Metabolite peak areas were obtained by numeric integration over regions corresponding to choline, creatine, and citrate. Spectra were inspected and compared with the corresponding transverse T2-weighted image and were considered useable if at least 75% of each voxel consisted of peripheral-zone tissue, the voxel did not contain urethral or ejaculatory ductal tissue, and the spectra were not contaminated by insufficiently suppressed water or lipid signal. No other spectral criteria were used to define a useable voxel. Choline, creatine, and citrate peak area-to-noise ratios and choline-to-creatine peak area ratios were calculated for all useable spectral voxels. Spectroscopic voxels were considered suspicious for malignancy if the choline level was elevated and citrate was absent. In the presence of detectable creatine (peak creatine area-to-noise ratio, >5:1), an elevated choline level was defined as a choline-to-creatine ratio greater than 1.5:1. In the absence of detectable creatine (peak creatine area-to-noise ratio, 
5:1), an elevated choline level was defined solely as a choline peak area-to-noise ratio greater than 5:1. Absence of citrate was defined as a citrate peak area-to-noise ratio less than 5:1. These criteria for the characterization of a voxel as suspicious were based on the described metabolic changes after radiation therapy for prostate cancer. It is known that the citrate and polyamine spectral peaks decrease rapidly and progressively with time after radiation therapy, in contrast to their levels in the untreated gland; these metabolites therefore are of limited use in the identification of cancer after treatment (20). Recurrent prostate cancer after other organ-preserving therapies, such as cryosurgery and androgen deprivation, is characterized by elevated choline levels (21–25). We could not base our criteria directly on reports of in vivo MR spectroscopic imaging in patients with biochemical failure after definitive radiation therapy, because no such studies had been published. A single ex vivo investigation of MR spectroscopy in prostate biopsy specimens after external beam radiation therapy confirmed that an elevated choline peak was one of the spectral characteristics in recurrent prostate cancer (15). The contiguity of suspicious voxels in each hemiprostate was also noted. Metabolic atrophy in a voxel was defined as the absence of choline and citrate peaks (peak area-to-noise ratio, <5:1).
Data and Statistical Analysis
Because of the known limitations of tumor localization and registration based on sextant biopsy results (26,27), we used the hemiprostate (ie, the left and right sides of the gland) as the unit of analysis. The limitation of the prostatic sextant as a unit of analysis is illustrated by the results of a previous investigation of tumor localization with MR imaging and MR spectroscopic imaging: The accuracy of imaging for sextant localization was only 67% (157 of 234) to 74% (173 of 234), but that of imaging for tumor lateralization was 75% (80 of 106) to 88% (93 of 106) (12). The difference was, presumably, at least partially due to errors in registration between imaged sections and biopsy specimens. Such errors are likely to be magnified in the irradiated gland because of radiation-induced shrinkage and distortion of tissue. Recurrent cancer was determined to be present or absent in each hemiprostate on the basis of the presence or absence of an ipsilateral positive biopsy result. The ability of MR imaging and MR spectroscopic imaging to identify a hemiprostate as cancer-containing was analyzed by using logistic regression analysis with generalized estimation equations to account for clustering effects from multiple measurements in the same patient (28). For MR imaging, interobserver agreement was evaluated with weighted 
values. The degree of observer agreement indicated by the value was interpreted as follows: 0–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 
0.81, almost perfect agreement (29). We generated receiver operating characteristic curves for cancer detection with MR imaging and with MR spectroscopic imaging at different thresholds based on the scores assigned by the two readers and on the number of suspicious voxels in each hemiprostate, respectively. Differences in area under the receiver operating characteristic curve between MR imaging and MR spectroscopic imaging were tested for statistical significance by using the bootstrap method, and 95% confidence intervals were calculated to characterize these differences (30). To determine the ability of MR spectroscopic imaging to help identify a hemiprostate as cancer free, we also analyzed the relationship between biopsy findings and spectroscopic findings of metabolic atrophy. In all statistical analyses, a P value of less than .05 was considered to indicate a significant difference. Statistical calculations were performed by using software (S-Plus 2000; Insightful, Seattle, Wash).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Both readers rated 20 of the 21 MR imaging studies as excellent and rated one study as intermediate in quality. Reader 1 rated zonal indistinctness as intermediate in 19 and as marked in two cases. Reader 1 rated the degree of diffuse T2 signal hypointensity as intermediate in all 21 cases. Reader 2 rated the degree of diffuse T2 signal hypointensity as intermediate in 13 and as marked in eight cases. The area under the receiver operating characteristic curve for the detection of locally recurrent cancer with MR imaging was 0.49 for reader 1 (95% confidence interval: 0.33, 0.66) and 0.51 for reader 2 (95% confidence interval: 0.32, 0.67), indicating only slight interobserver agreement (weighted , 0.20; 95% confidence interval: 0.02, 0.36).
The quality of MR spectroscopic studies was rated as excellent in seven cases, good in six cases, fair in five cases, and poor in three cases. The mean number of useable peripheral-zone voxels in each hemiprostate was 27 (range, 14–64). The relation between the number of suspicious voxels in a hemiprostate and the likelihood of a positive finding at biopsy in that hemiprostate is detailed in Table 1. The number of suspicious voxels in the nine cancer-containing hemiprostates was zero (n = 1), three (n = 1), four (n = 2), or six or more (n = 5). The number of suspicious voxels in the 33 cancer-free hemiprostates was zero (n = 17), one (n = 5), two (n = 5), three (n = 2), or four or more (n = 4) (Table 2). When the number of suspicious voxels was used to define different diagnostic thresholds, the area under the receiver operating characteristic curve for the detection of locally recurrent cancer with MR spectroscopic imaging was 0.81 (95% confidence interval: 0.60, 0.95; Fig 1), significantly greater than the areas under the receiver operating characteristic curves for the two readers at MR imaging (P < .005). The finding of three or more suspicious voxels in a hemiprostate had a sensitivity and specificity of 89% and 82%, respectively, for the diagnosis of locally recurrent prostate cancer (Fig 2). Sensitivity and specificity were not significantly different according to the contiguity or noncontiguity of suspicious voxels. Of the six hemiprostates with three or more suspicious voxels but without cancer found at biopsy, four were in three patients who subsequently underwent hormonal therapy for continued increase in prostate-specific antigen level, and two were in two patients who had a positive result at biopsy in the contralateral hemiprostate (Table 2). Although all of the useable voxels in seven hemiprostates demonstrated metabolic atrophy (ie, complete metabolic atrophy was present in the hemiprostate), these hemiprostates were cancer free and demonstrated only histologic postirradiation atrophy at biopsy (Table 2, Fig 3).
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Although these results are promising, our study had several limitations, particularly the imperfect standard of reference used. Sextant biopsy, even with the extended method, is subject to sampling error, and the histopathologic interpretation of tissue from the irradiated prostate is difficult (33). Results of post–radiation therapy prostatic biopsies may be false-negative because of sampling error, false-positive because of delayed tumor regression, or indeterminate because of uncertainty about tumor viability (34). Despite these limitations, positive results of post–radiation therapy biopsies are independently predictive of ultimate treatment failure in multivariate analysis (34). We therefore have used sextant biopsy results as a surrogate marker for local recurrence, although it is likely a suboptimal marker. For example, it is possible that patients with positive spectroscopic findings and negative biopsy results may actually have small locally recurrent prostate cancers that were missed at biopsy. This possibility is partially supported by our finding that all five patients in whom three or more suspicious voxels were found at MR spectroscopic imaging in a biopsy-negative hemiprostate also had other evidence of disease recurrence: Three required hormonal therapy for continued increase in prostate-specific antigen level, and two had positive findings at biopsy in the contralateral hemiprostate (Table 2). In addition, the spatial correlation of biopsy findings and spectroscopic findings is inherently limited and does not allow a voxel-by-voxel analysis. Such an analysis would enable investigation into the cause of false-negative findings at MR spectroscopic imaging, which might be small tumor size, low cell density, or other unrecognized factors. Histologic analysis of salvage prostatectomy specimens might have been preferable as the standard of reference, but such surgery is rarely performed, and even the larger studies in which this method was used included small numbers of patients recruited over many years (35). Long-term patient surveillance would allow the use of clinical and biochemical outcomes as an alternative standard of reference but might not be particularly helpful in the differentiation of local from distant recurrence. That is, demonstration of a continued increase in prostate-specific antigen level in a patient with positive MR spectroscopic findings would not necessarily indicate that the spectroscopic findings were true-positive, since the prostate-specific antigen level might actually reflect distant recurrence. Conversely, demonstration of a continued increase in prostate-specific antigen level in a patient with negative MR spectroscopic imaging findings would not necessarily indicate that the spectroscopic findings were false-negative, for the same reason. Similarly, a positive or negative clinical outcome would not necessarily support a positive or negative biopsy result, respectively. The variable interval between MR spectroscopic imaging and biopsy, the small number of patients, and the fact that MR spectral analysis was confined to peripheral-zone voxels only, are additional important limitations. Given these limitations, we regard our results as preliminary. For example, our finding that the contiguity of abnormal voxels was not a significant factor appears surprising, since intuition suggests that a cluster of abnormal voxels would prompt greater concern than would the same number of voxels scattered noncontiguously throughout the peripheral zone. It is possible that a larger study would have shown such clustering to be a significant factor.
In conclusion, our preliminary data suggest that MR spectroscopic imaging, but not MR imaging, may be of value for the depiction of locally recurrent prostate cancer after radiation therapy.
|
FOOTNOTES |
|---|
Abbreviation: PRESS = point-resolved spectroscopy
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, H.S.T., F.V.C., J.K.; study concepts, all authors; study design, H.S.T., F.V.C., J.K.; literature research, H.S.T., F.V.C., J.K.; clinical and experimental studies, all authors; data acquisition, M.G.S., K.S., D.B.V., J.K., B.P.; data analysis/interpretation, H.S.T., J.K., F.V.C, Y.L., B.P.; statistical analysis, H.S.T., Y.L.; manuscript preparation, editing, and revision/review, all authors; manuscript definition of intellectual content and final version approval, H.S.T., F.V.C., J.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Plathow and W. A. Weber Tumor Cell Metabolism Imaging J. Nucl. Med., June 1, 2008; 49(Suppl_2): 43S - 63S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 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. R. Cabrera, F. V. Coakley, A. C. Westphalen, Y. Lu, S. Zhao, K. Shinohara, P. R. Carroll, and J. Kurhanewicz Prostate Cancer: Is Inapparent Tumor at Endorectal MR and MR Spectroscopic Imaging a Favorable Prognostic Finding in Patients Who Select Active Surveillance? Radiology, May 1, 2008; 247(2): 444 - 450. [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]
|
|
|
|
|
|
O. Rouviere, E. Sala, and H. Hricak MR Assessment of Recurrent Prostate Cancer after Radiation Therapy Radiology, February 1, 2007; 242(2): 635 - 637. [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]
|
|
|
|
 |
|
 G S Payne and M O Leach Applications of magnetic resonance spectroscopy in radiotherapy treatment planning Br. J. Radiol., September 1, 2006; 79(Special_Issue_1): S16 - S26. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) was used to define different diagnostic thresholds (area under the curve, 0.81).
