Abstract
Metastatic prostate cancer may respond initially to hormone suppression, with involution of tumor sites, but ultimate tumor progression is inevitable. Our aim was to detect the proportion of bone and soft-tissue lesions that represent metabolically active tumor sites in patients with progressive metastatic prostate cancer. Methods: In a prospective study, we compared 18F-FDG and l-methyl-11C-methionine (11C-methionine) PET with conventional imaging modalities (CIM), which included the combination of 99mTc-methylene diphosphonate scintigraphy, CT, or MRI. Twelve patients with prostate cancer, increasing levels of prostate-specific antigen (PSA), and at least 1 site (index lesion) with new or increasing disease on CIM were studied. The total numbers of soft-tissue and bone-tissue lesions, in a site-by-site comparison, were calculated for all imaging modalities. Results: The sensitivities of 18F-FDG PET and 11C-methionine PET were 48% (167/348 lesions) and 72.1% (251/348 lesions), respectively, with CIM being used as the 100% reference (348/348). 11C-Methionine PET identified significantly more lesions than 18F-FDG PET (P < 0.01). All 12 patients with progressive metastatic prostate cancer had at least 1 lesion site of active metabolism for 18F-FDG or 11C-methionine, which could be used as an index lesion to monitor the metabolic response to therapy. A significant proportion of lesions (26%) had no detectable metabolism of 18F-FDG or 11C-methionine. Although technical factors cannot be totally excluded, we believe that metabolically inactive sites may be necrotic or dormant. More than 95% (251/258) of metabolically active sites (72% of the total number of lesions detected by CIM) metabolize 11C-methionine. 18F-FDG uptake is more variable, with 65% of metabolically active sites (48% of the total number of lesions detected by CIM). Conclusion: These findings reflect the different biologic characteristics of the lesions in a heterogeneous tumor such as prostate cancer and suggest that a time-dependent metabolic cascade may occur in advanced prostate cancer, with initial uptake of 11C-methionine in dormant sites followed by increased uptake of 18F-FDG during progression of disease.
Prostate carcinoma represents the most common malignancy in men in the United States. An estimated 184,500 men were affected in 1998, with fatal outcomes in 39,200 cases, making this the second leading cancer killer in men behind lung cancer (1). As a result of vascular spread, approximately 30% of these patients show bony metastasis at the time of diagnosis and >80% show bony metastasis at the time of death (2,3). Through lymphatic spread the metastasis can involve, in decreasing order, the obturator, internal iliac, common iliac, presacral, and paraaortic nodes. Hematogenous metastasis occurs in bone much more frequently than in lung, liver, and adrenal glands (4).
Radical prostatectomy and radiotherapy represent the 2 main curative modalities in cases of localized disease; however, hormonal manipulation is generally accepted as the treatment of choice in metastatic disease. Still, there is much controversy in some aspects of patient management. One of several reasons for these controversies is that the malignancy is heterogeneous, ranging from a relatively indolent tumor in some patients to an extremely aggressive disease in other patients.
Recent developments of high-resolution imaging modalities, such as MRI and spiral CT, have contributed to the early detection and more accurate staging of malignant tumors because of precise morphologic information on the lesion and surrounding normal tissue. However, these imaging modalities and bone scintigraphy are limited in their ability to distinguish benign from malignant masses, such as determining if a slightly enlarged lymph node is involved by cancer. Thus, an obvious need exists for better imaging modalities that could more accurately characterize individual lesions in vivo so that decisions regarding treatment options in a particular patient could also be based on knowledge of the metabolic activity and biologic characteristics of the tumor. Furthermore, better assessment of response to therapy is also of great importance.
PET with 18F-FDG, a tracer of glucose metabolism, has been highly successful for imaging a wide variety of tumors and for monitoring response to therapy (5). This agent is transported, phosphorylated, and metabolically trapped into tumor cells as a glucose substitute (6).
l-methyl-11C-Methionine (11C-methionine) is also used to image a variety of tumors, including lymphomas and tumors of the brain, lung, and head or neck. 11C-Methionine accumulation in tumor cells is attributed to increased amino acid transport and metabolism (7,8). However, only a few clinical studies have compared 18F-FDG PET and 11C-methionine PET in detecting tumor lesions in the same patient, and none of these studies was with prostate cancer patients (9–11).
This study’s aim was to assess the strengths and limitations of imaging patients with progressive metastatic prostate cancer with 18F-FDG PET and 11C-methionine PET, in comparison with conventional imaging modalities (CIM), which included the combination of bone scanning, CT, and MRI.
MATERIALS AND METHODS
Patients were recruited prospectively and entered into the PET protocol study after written informed consent was obtained, following the Institutional Review Board Guidelines of our institution. From this group of patients, we included in our study the first 12 who had whole-body 18F-FDG PET and 11C-methionine PET on the same day.
Patient Eligibility Criteria
Patients had histologically proven prostate adenocarcinoma and disease progression as shown by (a) a >50% increase in serum prostate-specific antigen (PSA) levels sustained for a minimum of 3 observations obtained at least 1 wk apart, and (b) development of new lesions or worsening of preexisting lesions on bone scintigraphy, CT, or MRI.
Pretreatment Evaluation
Patient evaluation included history and physical examination, complete blood count, PSA (within 2 wk of study entry), and urinalysis to exclude urinary tract infection (within 2 wk). Bone scintigraphy, modified MRI of the pelvis to image the prostate (including the retroperitoneum), CT of the abdomen and pelvis, and chest radiography (all of them done within 4 wk of the PET scan) were also included.
11C-Methionine PET
The cyclotron core facility staff of our institution provided the radiopharmaceuticals. 11C-Methionine was synthesized using a modification of the method described by Ishiwata et al. (7).
All patient scanning was performed on an Advance PET scanner (General Electric Medical Systems, Milwaukee, WI). This camera has a field of view of 55 cm in diameter and 15.2 cm in axial length. All scanning was performed in 2-dimensional (septa in) mode. The transaxial resolution is 3.8 mm full width at half maximum (FWHM) at the center of the field of view, increasing to 7.3 mm at a radial distance of 20 cm. The average axial resolution decreases from 4.0 mm FWHM at the center to 6.6 mm at 20 cm. Patients fasted for 6 h before PET; however, liberal water intake was encouraged.
A transmission scan was first obtained from the level of the upper neck to the pelvis, at 4 min per bed position. Immediately thereafter, the patient was injected with 370 MBq (10 mCi) 11C-methionine. Acquisition of the emission scan was started 10 min later, at 6 min per bed position. All scans were reconstructed using vendor-provided filtered backprojection algorithms.
18F-FDG PET
18F-FDG was prepared using the PETtrace FDG MicroLab (General Electric Medical Systems). The 18F-FDG PET scan was obtained a minimum of 2 h after completing the 11C-methionine PET scan. Forty-five minutes after the injection of 370 MBq (10 mCi) 18F-FDG, an emission scan was acquired from the neck to the pelvis, for 6 min per bed position.
Subsequently, a transmission scan was acquired for 4 min per bed position. This study was also reconstructed using vendor-provided filtered backprojection algorithms.
MRI
Modified MRI of the pelvis, including the retroperitoneum, was performed on a 1.5-T Signa MRI scanner (General Electric Medical Systems), using axial spin-echo T1- and T2-weighted images, with and without fat suppression. Selected patients also had MRI of the lower thoracic and lumbar spine, without gadolinium.
CT
Abdominal and pelvic CT scans were obtained on a high-speed CT/i helical scanner (General Electric Medical Systems). Contiguous 1-cm-thick tomographic slices were obtained after the intravenous bolus injection of nonionic contrast material.
Bone Scintigraphy and Bone Scan Index
Bone scans were done approximately 2 h after the intravenous injection of 925 MBq (25 mCi) 99mTc-methylene diphosphonate. Whole-body anterior and posterior images were acquired on a dual-head Genesys gamma camera (ADAC Laboratories, Milpitas, CA).
The bone scan index (BSI), a semiquantitative index that reflects the percentage of bony skeleton involved by the tumor, was also calculated. The BSI has been shown to have minimal interobserver and intraobserver variability and parallels the rise in the PSA level (12).
Data Analysis
The CT scans and MR images were read independently by a hospital radiologist. The bone scans were also read independently, with the BSI and the total number of lesions calculated for each patient.
The presence of bone metastasis was established on the basis of the bone scan findings of extensive characteristic lesions or, in cases with few abnormalities, on their subsequent progression on follow-up scans together with an increase in serum PSA level.
Lymph node metastasis was diagnosed on the basis of typical MRI or CT scan findings (or both) and was confirmed by interval progression over a 12-mo course. Lung metastasis and mediastinal metastasis were confirmed on the basis of persistent suspicious findings on follow-up chest x-rays, CT scans, and PET scans.
Transaxial, coronal, and sagittal PET images were evaluated visually by the same physician, who was unaware of the data from the other imaging modalities. The 11C-methionine PET and 18F-FDG PET scans were read independently. A PET scan was considered to be abnormal when a focus or an area of increased 18F-FDG or 11C-methionine uptake was above what is considered to be normal background. Regional 18F-FDG and 11C-methionine uptake in tumor tissue from the PET image data was quantified using standardized uptake values (SUVs). Each lesion was identified in the transaxial image, and SUV images were generated. Circular regions of interest were carefully drawn around each and every lesion on the 18F-FDG PET and the 11C-methionine PET images. The SUV values represent the maximal radioactivity concentration in the region of interest, normalized for injected dose and body weight.
Statistical Analysis
The sensitivity of each of the imaging modalities was calculated according to the number of lesions detected for each one of the studies compared with that of CIM (bone scan, CT scan, and MR image).
The SUV for each one of the lesions was determined on both PET scans. In addition, the mean, median, minimum, and maximum SUV for the bone and soft-tissue lesions was calculated for the 18F-FDG PET and 11C-methionine PET scans. The mean SUVs were compared by the paired t test. The difference between the proportion of lesions identified by 18F-FDG PET and 11C-methionine PET was compared using a McNemar test incorporating an adjustment method described by Eliasziw and Donner (13). Such an adjustment is necessary because the lesions on the same patient are likely to be correlated and one of the assumptions of the unadjusted McNemar test is violated. Statistical significance was determined from a P < 0.05.
Correlation between each pair of variables per patient, including the PSA, BSI, and number of lesions detected by each of the different imaging modalities, was examined by the Spearman rank correlation test.
Agreement between each pair of modalities was evaluated using the κ statistic, where κ = 0 when the agreement between the 2 modalities equals that expected by chance alone and κ = 1 when there is perfect agreement (14).
RESULTS
A total of 24 PET scans were obtained on 12 patients. Gleason scores of the primary prostate neoplasm ranged from 6 to 10. Table 1 presents details about the patients, including age, hormonal status, primary therapy given at the time of the scan, and PSA value upon entering the study.
The combination of CT, MRI, and bone scanning (CIM) detected a total of 348 of 348 lesions (100%), of which 325 of 348 (93%) were bone metastases and 23 of 348 (7%) were abdominal and pelvic soft-tissue lesions. CT and MRI of the chest were not part of the study, although a few patients had CT scans of the chest on follow-up visits. However, most of the lesions in metastatic prostate cancer will have been detected in our study by the combination of bone scanning plus CT and MRI of the abdomen and pelvis because the number of soft-tissue lesions above the diaphragm normally represents a small percentage, if any, of the total number of lesions.
PET scan findings were compared with the results of CIM. A total of 314 of 325 lesions (96.6%) were detected by bone scans on the 12 patients; however, 45 bone lesions were detected exclusively by PET, with 37 of them detected by 11C-methionine PET. In 3 patients, 8 cervical spine metastases were detected only by PET. Nine additional suspicious bone lesions were detected only by MRI (in patients 2, 3, and 10). The number of bone metastases varied greatly between patients; some patients had multiple bone metastases (86 lesions), whereas others had none, entering the study on the basis of soft-tissue involvement (Fig. 1).
The BSI ranged from 0% to 69.9%. The bone scans of 2 of the patients with the highest values were cases of very extensive osseous metastases (superscan). The BSI had a highly significant correlation with the total number of suspicious lesions on the bone scan (0.876), which was slightly less correlated with the serum PSA value (0.859) (Table 2).
18F-FDG PET detected a total 167 of 348 lesions (48.0%), including 157 of 325 bone lesions (48.3%), 7 of 23 abdominal or pelvic lesions (30.4%), and 3 chest or neck soft-tissue lesions. In general, when the patient had a prominent tumor burden, osseous metastases were detected readily by 18F-FDG PET. The SUV of bone lesions ranged from 1.2 to 7.7, with 3.4 as the mean and 3.1 as the median. For soft-tissue lesions, the SUV ranged from 2.2 to 7.9, with 4.1 as the mean and 3.6 as the median (Table 3).
11C-Methionine PET scanning detected a total of 251 of 348 lesions (72.1%), with 227 of 325 bone metastases (69.8%), 16 of 23 abdominal and pelvic soft-tissue lesions (69.6%), and 8 neck and chest soft-tissue lesions. In the cases of the superscan, the number of bone metastases detected by 11C-methionine PET correlated much better than 18F-FDG PET with the number of lesions seen on the bone scan. In general, lesions detected by 18F-FDG PET were also detected by 11C-methionine PET; conversely, 11C-methionine PET detected many more additional lesions than 18F-FDG PET. However, in 2 patients, 18F-FDG PET detected 5 bone and 2 soft-tissue lesions not detected by 11C-methionine PET. The SUV of bone lesions ranged from 13.1 to 1.5, with 4.8 as the mean and 4.7 as the median. For soft-tissue lesions, the SUV ranged from 8.9 to 2.8, with 5.1 as the mean and 4.9 as the median. The difference in means was found to be statistically significant (P < 0.01) with a paired t test. Hence, we conclude that the mean uptake of 11C-methionine was higher than the mean uptake of 18F-FDG.
The difference in the proportion of lesions detected was also found to be statistically significant (P < 0.01) using an adjusted McNemar test (13). On the basis of this study, the sensitivity for the detection of bone and soft-tissue metastases with 18F-FDG PET was 48.3% and 30.4%, respectively; whereas for 11C-methionine PET it was considerably better, 69.8% and 69.6%, respectively. Table 4 summarizes the results of the study.
DISCUSSION
Over the years, 18F-FDG PET has proven to be a formidable imaging technique for a great variety of malignant tumors (15). Since the initial descriptions by Di Chiro et al. (16) of increased 18F-FDG uptake in brain neoplasms, there has been continuous interest and research in the applications of 18F-FDG PET in oncology. Nevertheless, its application in such an important neoplasm as prostate cancer has not met the expectations. One of the main reasons has been the limited sensitivities of 18F-FDG PET in metastatic prostate cancer, ranging from as low as 18% in a study done by Yeh et al. (17) to 65% in a study done by Shreve et al. (18). Using 11C-methionine PET, Nilsson et al. (19) found methionine uptake in a considerable number of lesions in patients with androgen-resistant prostate cancer. Inoue et al. (11) compared 18F-FDG PET with 11C-methionine PET in a variety of tumors, which did not include prostate cancer, and imaged only a limited part of the patient’s body; however, the sensitivities were 64.5% and 61.3% for 18F-FDG PET and 11C-methionine PET, respectively. More recently, we have shown that a high proportion of clinically active index lesions are metabolically active, with sensitivities of 85% and 95% for 18F-FDG PET and 11C-methionine PET, respectively (20). However, this study analyzed only index lesions (index lesions being the clinically active lesions that were detected by conventional imaging studies, which could be positive on a PET scan, representing a finding for future follow-up studies).
Evaluation of the extent of disease in prostate cancer, especially when there is biochemical or clinical evidence of disease progression, is of vital importance. Traditionally, this evaluation has been done with serial bone and CT scans and more recently with MR images. Nevertheless, these imaging techniques, though very precise anatomically (especially MRI), are limited in the extent of the body imaged and in their ability to distinguish benign from malignant tissue. Even with serial bone scans, it is frequently difficult to evaluate treatment response because the scan appearances may falsely indicate progression attributed to a flare-up response of the bone in the process of repairing or healing from the injury of the tumor. Moreover, sclerotic non–tumor-bearing lesions tend to appear on bone scans as suspicious findings for long periods of time.
The serum tumor marker PSA has represented a great advance in the management of the patient with prostate cancer. However, it is limited in its ability to evaluate and quantify or semiquantify the tumor burden. In addition, sometimes tumor progression may not be accompanied by a parallel rise in PSA because of, for example, tumor dedifferentiation (21). Thus, there is an obvious need for a metabolic imaging modality such as PET, which reflects the differences in the metabolism of tumor cells in comparison with normal tissue.
Our study shows, as reported by Macapinlac et al. (22), that whole-body 11C-methionine PET is a feasible technique that can produce high-quality images, despite the short physical half-life of 11C-methionine (Fig. 2). We have been able to routinely perform 11C-methionine PET and 18F-FDG PET scanning on the same day separated by a 3- or 4-h period.
The sensitivity of 72.1% (251/348 lesions) obtained in our study for the detection of bone and soft-tissue metastases with 11C-methionine PET is fairly high, at least in comparison with previously reported values with 18F-FDG PET of 18%–65% (17,18). However, the sensitivity of 48% (167/348 lesions) for 18F-FDG PET in our study (for bone and soft-tissue metastasis) is more in concordance with these earlier results. The ability of 11C-methionine PET to detect more bone and soft-tissue metastases than 18F-FDG PET is not surprising (Fig. 3) if we consider that prostate cancer can be a heterogeneous tumor with a wide spectrum in biologic behavior, ranging from indolent tumors to more aggressive ones, often in younger men, which can kill the patient in a few years. Therefore, the difference in uptake of the 2 tracers probably reflects the diversity and differences in tumor metabolism, not only from patient to patient but also frequently in the same patient between metastases. Still, a significant proportion of lesions (26%) had no detectable metabolism of 18F-FDG or 11C-methionine. Although technical factors cannot be totally excluded, we believe that metabolically inactive sites may be necrotic or dormant. Thus, bone scans can overestimate the extent of disease. Conversely, a few bone lesions were detected earlier by PET rather than by bone scanning, which reflects the ability of PET to detect the initial tumor cluster located in the bone marrow before it can be seen on the bone scan (Fig. 3).
The difference in lesion detection between 11C-methionine and 18F-FDG (69.6% and 30.4%, respectively) is even more pronounced if we consider soft-tissue metastases in the abdomen and pelvis. This difference may be attributed to several causes: (a) Minimal activity of 11C-methionine was found in the bladder in comparison with 18F-FDG, with the latter interfering with the evaluation of pelvic structures. Even with new reconstruction algorithms, 11C-methionine is clearly superior for the evaluation of metastases in the pelvis. (b) In comparison with 11C-methionine, which has minimal excretion through the kidneys, the higher concentration of 18F-FDG in the collecting system, pelvis, and ureter of both kidneys diminishes the ability to detect abdominal retroperitoneal metastases. (c) Inherent metabolic differences of tumor cells favoring preferential uptake and metabolism of 11C-methionine over 18F-FDG may in part play a role. This might explain the differences observed in the SUVs of bone and soft-tissue metastases, where 11C-methionine PET has consistently higher values in both types of tissues.
We found that cervical spine metastases were more easily detected by PET than by bone scintigraphy. In 3 patients, a total of 8 cervical spine metastases not detected by bone scintigraphy were clearly evident on the 18F-FDG PET and 11C-methionine PET scans (Fig. 4). The study by Schirrmeister et al. (23) confirms this finding; they found that the sensitivity of bone scintigraphy in comparison with 18F PET was only 20% in the cervical spine. Conversely, rib metastases were the most difficult metastases to detect with PET, not only because the SUV in 18F-FDG PET was as low as 1.2 for some metastases but also because the small cross-sectional area of the ribs limited the ability to detect the metastasis, even with a dedicated PET scanner—especially if the lesions were not very prominent. The use of a rotating 3-dimensional display helps in the detection of metastases in the rib cage. In general, the SUVs were higher than 1.2. 11C-Methionine PET had consistently higher SUVs than 18F-FDG PET in both bone and soft-tissue lesions (Table 3), and the difference in means was statistically significant between these 2 modalities (P < 0.01). Similarly, a statistically significant difference was found between the proportions of lesions found by 11C-methionine PET and 18F-FDG PET (P < 0.01). Conversely, the mean SUVs between bone and soft-tissue lesions did not differ significantly for 18F-FDG PET (P = 0.25) and for 11C-methionine PET (P = 0.36).
Another question of interest for us was the correlation of PSA and BSI with the total number of lesions detected by each modality. The Spearman rank correlation was calculated between each pair of modalities or variables (Table 2). CT and MRI had the lowest correlation with the other modalities. Of all 21 possible pairwise correlations, the 6 that involved CT and MRI were the lowest, the highest of these being the one with PSA (0.785). 11C-Methionine PET and BSI had a higher correlation (0.916) than that for bone scan and BSI (0.876), whereas 18F-FDG PET and BSI had a lower correlation of 0.790. Both PET modalities correlated well with bone scanning (0.937 for 18F-FDG and 0.950 for 11C-methionine) and CIM (0.934 for 18F-FDG and 0.955 for 11C-methionine). Evaluation of the agreement between each pair of variables (κ statistic) shows that bone scan and CIM agree very well (0.930), which is not surprising because the bone scan is part of CIM and identifies most of the lesions. The agreement between 11C-methionine PET and CIM (0.539) is higher than that between 18F-FDG PET and CIM (0.431). The differences between 18F-FDG PET and 11C-methionine PET, in comparison with CIM, are more pronounced than the rank correlations reported above (Table 5).
One of the advantages of the PET scan is that it can provide an indication of the metabolic activity of the disease in a single study and anatomically localize metastatic sites in bone and soft tissues. As such, PET could be an extremely useful adjunct to refine increasingly accurate nomograms used to estimate the pre- and postsurgical likelihood of nodal and distant metastases in prostate cancer (24–26).
In ongoing clinical studies using 11C-methionine PET and 18F-FDG PET in prostate cancer patients, we have found that these PET studies can assess the response to treatment as early as 4 wk after starting or changing treatment, with improvement or worsening of the PET scan findings, in many cases with no significant changes on the bone scans. Similar to what was seen in a study by Agus et al. (27) using male mice with the CWR22 human prostate cancer xenograft model, the changes observed on the 18F-FDG PET scan have paralleled the changes seen in serum PSA (Fig. 5). Therefore, the main value of PET for the prostate cancer patient may be to evaluate changes in tumor burden and location of disease, using this information to monitor the effect of treatment and determine the prognosis of the patient. We have also found that 18F-FDG PET can detect native, residual, or recurrent disease in the prostate bed,largely because of iterative reconstruction algorithms that allow a better visualization of the anatomic structures adjacent to the bladder, including the prostate bed (Fig. 6).
This study had certain limitations. Our study, like others, was limited by the inability to prove with certainty the metastatic nature of all lesions detected with PET, because pathologic confirmation of all PET findings is impractical and unethical. In addition, there was a lack of CT and MRI studies of the chest to correlate the PET scan findings in the chest. The PET scans were obtained from the neck to the pelvis; therefore, bone scan findings in the lower extremities and skull (17 lesions in our study) could not be correlated.
The PET scanner itself has the known limitation of not being able to detect microscopic disease. Artifacts from the attenuation correction with filtered backprojection reconstruction techniques were present around areas of high activity concentration, such as the liver with 11C-methionine PET and the bladder and kidneys with 18F-FDG PET. In addition, skull metastases are not detected with 18F-FDG PET because of the high activity concentration in the brain. We have found that the intrinsically higher activities seen in the bone marrow with 11C-methionine PET in comparison with 18F-FDG PET does not limit the ability to differentiate a bone metastasis from normal bone tissue (Fig. 4).
CONCLUSION
Whole-body 11C-methionine PET is a feasible imaging technique that detects significantly more bone and soft-tissue lesions than 18F-FDG PET, allowing the assessment of tumor metabolism in the frequently biologically heterogeneous metastases of prostate cancer patients. Two possible explanations for the differences are as follows: (a) differences in tumor metabolism between patients and in the same patient between metastases, and (b) a time-dependent metabolic cascade in advanced prostate cancer, with initial uptake of 11C-methionine in dormant sites followed by increased uptake of 18F-FDG during progression of the disease. Therefore, 11C-methionine PET is a potentially valuable imaging technique for the assessment and monitoring of tumor response to treatment, an area in which CIM, including bone scans (despite having increased sensitivity), have well-known limitations.
Acknowledgments
The authors thank Ronald Finn, PhD, for preparation and quality assurance of the radiotracers; Richard Petrisco for data management; Dr. Samuel Yeh for helpful commentaries and reviewing the manuscript; and Dr. Roland Chisin for many helpful insights, suggestions, and reviewing the manuscript. This study was supported by grants from the Hascoe Foundation and National Institutes of Health/National Cancer Institute grant 1P50 CA86438 01.
Footnotes
Received Feb. 20, 2001; revision accepted Sep. 25, 2001.
For correspondence or reprints contact: Rodolfo Nuñez, MD, Division of Nuclear Medicine, Long Island Jewish Medical Center, 270-05 76th Ave., New Hyde Park, NY 11040.
E-mail: rnunez{at}lij.edu