| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, New York 10021
2 Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, New York 10021
3 Department of Surgery, Columbia University Medical Center, 161 Fort Washington Avenue, New York, New York 10032
4 Department of Radiology, Hospital of the University of Pennsylvania, 1 Silverstein, 3400 Spruce Street, Philadelphia, PA 19104
5 Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, New York 10021
Correspondence: Address correspondence and reprint requests to: Mary S. Brady, MD, FACS; E-mail: bradym{at}mskcc.org.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: We performed a prospective trial to evaluate the clinical utility of whole-body fluorine 18–labeled deoxyglucose-PET scanning used in addition to standard imaging (contrast-enhanced computed tomographic [CT] imaging of the chest, abdomen, and pelvis) in preoperative stage IIC (T4N0M0), III (any T, N1 to N3, M0), and IV (any T, any N, M1) melanoma patients. Pathologic or clinical follow-up within 4 to 6 months of the imaging studies was used to determine the accuracy of preoperative PET and CT scan findings.
Results: Preoperative imaging findings led to a change in clinical management in 36 (35%) of 103 patients. In 32 (89%) of these patients, the information was accurate. Findings on PET scan alone (14 of 36; 39%) or in combination with CT (20 of 36; 56%) resulted in a treatment change in most patients (34 of 36; 94%). The most common decision was to cancel the operation (19 of 36; 53%). PET scanning was more sensitive than CT scanning in detecting occult disease (68% vs. 48%; P = .05), but both tests were highly specific (92% vs. 95%; P = .7, PET vs. CT).
Conclusions: PET scanning facilitates the appropriate management of high-risk melanoma patients being considered for operative intervention. PET imaging in addition to CT scanning should be strongly considered before operation in patients at high risk for occult metastatic disease.
Key Words: [(18)]F Fluorodeoxyglucose-positron emission tomography scanning • Melanoma • Preoperative • Management • High risk
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Most patients undergoing resection of metastatic melanoma are those with clinically palpable or visually apparent regional soft tissue or nodal disease. Patients with isolated distant sites of disease are also considered for operative resection. The standard approach to preoperative imaging in these patients consists of a contrast-enhanced computed tomographic (CT) scan of the chest, abdomen, and pelvis. A magnetic resonance imaging (MRI) or CT scan of the brain is often obtained as well, although its utility in asymptomatic patients is questionable. If these studies are negative for evidence of occult sites of disease, the surgeon will proceed with resection of clinically apparent regional or distant disease.
Positron emission tomography (PET) scanning may be a more sensitive diagnostic evaluation in patients at high risk for clinically and radiologically occult metastasis because it allows visualization of tumor based on uptake of fluorine 18–labeled deoxyglucose (FDG), thereby complimenting nonfunctional imaging (CT). Melanoma is a glucose-avid tumor,1 and previous studies suggested that PET scanning is more sensitive than standard imaging in patients with high-risk melanoma.2–5 Most of these studies were retrospective in design, however, and included patients with early as well as advanced disease.
There are limited data that address the utility and clinical yield of standard diagnostic imaging in pre-operative high-risk melanoma patients. Finkelstein et al.6 have recently reported that the combination of FDG-PET and conventional imaging (CT and/or MRI of the chest, abdomen, and pelvis) had a sensitivity of 88%, a specificity of 91%, and a positive and negative predictive value of 91% and 88%, respectively. They studied a small number of patients (n = 18) undergoing metastasectomy with stage IV melanoma.
We performed this study to determine whether FDG-PET scanning is useful as a preoperative imaging modality in stage IIC, III, and IV melanoma patients considered for surgical management. We designed the study to test the hypothesis that PET scanning in addition to conventional imaging (contrast-enhanced CT of the chest, abdomen, and pelvis) increases detection of occult sites of metastasis and thereby facilitates appropriate clinical management. We were interested in determining the additive value of PET imaging as opposed to determining whether PET was superior to standard imaging.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patients underwent a whole-body FDG-PET scan in addition to a CT scan of the chest, abdomen, and pelvis with oral and intravenous contrast. Outside CT scans were accepted if the quality of the study was sufficient as determined by the study radiologist (S.H.). Participating surgeons completed a staging assessment of the patient once the results of the CT and PET studies were available. The physician determined the stage of disease according to CT alone and then according to CT and PET imaging. No attempt was made to determine the stage of disease according to PET alone because this would have required the surgeon and nuclear medicine physician to be blinded to the CT results. It was our intent to determine the additive clinical utility of PET, because CT scanning was considered the standard preoperative imaging evaluation for these patients.
A clinical decision was scored by the treating physician, and the imaging study or studies that led to the decision were identified and scored on the study form. Additional imaging studies were obtained before surgery at the discretion of the treating physician to clarify PET and/or CT findings or when otherwise deemed appropriate. Clinical interpretations of the CT scan and PET scan available to the physician before surgery were used to complete the staging forms and score the management decision. A management decision was made after both imaging studies were obtained. The surgeon identified the study or studies that facilitated the decision, but no attempt was made to elicit a decision on the basis of one study without knowledge of the results of the other. The purpose of the study was to determine the additive utility of PET scanning and not to compare it directly with CT imaging.
Imaging findings that were "indeterminate" by the reviewing radiologist or nuclear medicine physician were not considered positive studies. Imaging findings interpreted as "suspicious for" or "consistent with" metastatic melanoma were considered positive preoperative imaging studies for evidence of occult metastatic disease. Nuclear medicine physicians and radiologists were not blinded to the results of the other preoperative imaging studies or to the patient’s clinical history.
True-negative (TN) PET or CT scans were considered those that were negative for evidence of occult metastatic disease at study entry in patients who remained without recurrence detected by any method for 4 months after the initial scans. False-negative (FN) scans were read as normal, but patients developed evidence of metastatic melanoma within 4 months of the initial scans or the other imaging study correctly identified occult disease. In addition, FN scans occurred when findings that were not originally considered consistent with metastatic disease were identified as such with additional clinical or radiologic follow-up. True-positive (TP) scans demonstrated occult metastatic disease that was confirmed by pathologic, radiological, or clinical follow-up at any point up to 6 months after study entry. False-positive (FP) scans occurred when the preoperative imaging study was considered positive for evidence of occult metastatic disease but this was subsequently (within 4–6 months) determined to be incorrect. In this way, we could determine the real utility of the imaging studies, keeping in mind that early recurrence of metastasis (i.e., within 4 months) after operative resection of known disease could be considered a failure of preoperative imaging to facilitate selection of appropriate patients for surgical resection.
Imaging studies were considered TP when any site of occult disease identified before surgery was subsequently confirmed. We did not require that the imaging study identify all sites of occult disease to be considered a TP study. The analysis was performed on a patient basis and not a disease site basis. If an imaging study correctly identified occult metastasis in the lung, for example, but also identified suspicious lesions in the liver that were subsequently not confirmed, the imaging study was still considered a TP study for detection of occult metastasis.
The study was designed so that patients would undergo repeat PET and CT scans 4 to 6 months after completion of their initial studies to facilitate the scoring of the initial scans. Although compliance with this was problematic, sufficient radiological, clinical, and pathologic information was available to allow for accurate scoring of the initial imaging studies in all patients. All initial clinical decision scores were provided by the enrolling physician, and then all clinical and radiologic data were reviewed by the investigating surgeons (M.S.B. and K.S.), a nuclear medicine physician (T.A.), and a research study assistant (A.P.) to determine whether the studies provided accurate information.
PET Imaging
Whole-body FDG-PET scans were obtained by using the ADVANCE (General Electric Medical Systems, Waukesha, WI) whole-body PET scanner. All patients were instructed to fast for 6 hours before the PET study. A fasting blood sugar sample was obtained just before intravenous injection of FDG. Imaging began 50 to 60 minutes after injection of FDG. Patients were positioned with their arms at their sides on the scanner bed. Foley catheters were not used. Positioning allowed acquisition of sequential two-dimensional static emission scans from the vertex to the feet. Two-dimensional static emission scans were acquired at 4 minutes per view. Corresponding transmission images were acquired for 2 minutes per view by using a germanium 68 rotating source for each corresponding emission bed position. The total scan time was approximately 70 to 90 minutes.
CT Scanning
CT scans of the chest, abdomen, and pelvis were performed within 4 weeks of the PET scan. Oral and intravenous contrast were used. CT scans were performed with contiguous sections and bolus injection of intravenous contrast. Outside CT scans were accepted if performed within 4 weeks of PET and if judged to be of sufficient quality by the reviewing radiologist (S.H.).
Statistical Analysis
The gold standard was obtained through pathology for resected patients and through clinical status and imaging at 6 months, as explained in detail previously. The sensitivity and specificity of PET and CT were estimated by using TP and TN proportions, respectively, and were compared by using a McNemar test.8 We used the exact permutation distribution to obtain the P values.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
There were 55 women and 48 men, with a median age of 60 years (range, 21–88 years). Most of the patients had American Joint Committee on Cancer stage III melanoma (n = 74; 72%). There were 17 (17%) patients with stage IV disease and 12 patients with stage IIC (12%). Most patients in the study had primary cutaneous melanoma of the extremity (n = 54) or trunk (n = 30). Four patients had a primary cutaneous melanoma of the head and neck. The remaining patients had melanoma of unknown primary origin (n = 12) or mucosal melanoma (n = 3).
All preoperative PET scans were performed and interpreted at Memorial Sloan-Kettering Cancer Center. Of the 103 patients, 36 had good-quality CT scans available at the time of accrual to the study. The study was designed so that all patients would undergo a repeat PET and CT scan 4 to 6 months from study entry. At this point, the initial imaging studies were scored for accuracy. Unfortunately, only 25 patients had both follow-up studies within the required time interval, although all of the patients had subsequent CT, MRI, or PET scans or pathologic evaluation that allowed determination of the accuracy of the original imaging studies. The most common reason for failure to obtain the follow-up PET and CT scan within the time interval designated by the study was patient noncompliance. There were 21 patients who underwent only follow-up CT scans within 4 to 6 months of the initial scans, and 15 patients had follow-up PET scans only. Despite this, all patients had sufficient radiological, pathologic, or clinical information available during the follow-up interval to allow for accurate assessment of the preoperative imaging studies. "Sufficient information" was unequivocal evidence of metastatic disease as determined by pathology assessment and/or unequivocal radiological and/or clinical progression of disease.
The preoperative plan for most patients accrued to the study (57 of 103; 55%) was regional lymphadenectomy (47 therapeutic and 10 sentinel node directed). The remaining patients were being considered for resection of metastatic disease (n = 15), resection of regional nodal and in-transit disease (n = 6), resection or regional chemotherapy for in-transit disease (n = 13), or resection of deep primary melanoma (n = 12). These data are depicted in Table 1
.
|
With clinical and radiological follow-up, it was determined that in 17 of 18 patients with positive PET and CT, both studies were TP (94%). In one patient, both the PET and CT were FP (6%). Of the 17 patients with positive PET scans and negative CT scans, in 13 patients the PET was TP (76%), and in 4 patients the PET was FP. In the six patients with positive CT scans and negative PET scans, four (67%) of six results were TP, and two (33%) of six were FP.
Treatment Changes Due to Preoperative Imaging Studies (n = 36)
Information detected by the preoperative imaging studies led to a treatment change in 36 (35%) of the 103 study patients. In 34 (94%) of the 36 patients, PET scan findings alone (n = 14) or in combination with CT scan findings (n = 20) resulted in the management change. In two patients, CT scanning resulted in a change in management (Table 2). The most common decision was to cancel the operation because of the detection of occult metastatic disease (19 of 36; 53%). In most of these patients (14 of 19; 74 %), this was due to findings on both CT and PET. In 5 (24%) of 19 patients, this was due to findings on PET only. In 32 (89%) of the 36 patients, the information was determined to be accurate once all clinical, pathologic, and radiological information was available. In 4 of 36 patients, the scan or scans were incorrect. This occurred in 2 of 20 patients in whom the combination of tests was interpreted as positive for occult disease (FP), in 1 patient with an FP PET, and in 1 patient in whom treatment was changed because of an FN CT scan.
|
Accuracy of Preoperative Imaging Studies
The accuracy of preoperative imaging was determined by reviewing all of the clinical and pathologic information available within 6 months of study entry and then by determining whether the disease was present on the imaging studies. There were 44 patients determined to have had occult metastatic disease at study entry on the basis of pathologic, clinical, and radiological follow-up within 4 to 6 months of accrual. Of the 44 patients, 34 patients had occult metastatic disease accurately identified on either PET or CT, for a sensitivity of the combination of studies of 77%. PET scanning was more sensitive than CT scanning (68% vs. 48%; P = .05), but both tests were highly specific (92% vs. 95%, PET vs. CT; P =.7; Table 3). The method used to determine the accuracy of the initial imaging findings in the 44 patients determined to have had occult disease at study entry was most commonly imaging studies (59%), followed by pathologic confirmation (36%) and, less commonly, by clinical assessment (5%). These data are summarized in Table 4.
|
|
There were 89 sites of confirmed occult disease detected by initial imaging studies in 34 patients (Table 5). Most occult disease was distant (71 of 89 sites; 80%) as opposed to regional (18 of 89; 20%) disease. The most common site of occult disease detected by preoperative imaging studies was distant soft tissue and/or distant nodes (38 of 89; 43%), followed by regional soft tissue and/or regional nodal disease (18 of 89; 20%). Less common sites for detection of occult disease were lung (10 of 89; 11%), bone (6 of 89; 7%), liver (5 of 89; 6%), and retroperitoneum (5 of 89; 6%).
|
).
|
.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Several more recent studies have supported the observations made by early reports and further defined the advantages and limitations of PET scanning in melanoma patients. Swetter et al.11 compared the sensitivity and specificity of PET scanning in a primarily retrospective study of 104 melanoma patients with both primary and recurrent disease. They compared PET with CT scanning for the detection of any site of melanoma and reported that PET was more sensitive than CT in detecting primary or recurrent melanoma (84% vs. 58% sensitivity, respectively). In addition, PET scanning had high specificity (97%) compared with CT (69%).
Stas et al.12 reviewed a consecutive series of 100 PET scans performed in 84 melanoma patients who presented with confirmed or suspected regional or distant recurrence. They reported that PET scans were more accurate at all sites when compared with conventional screening procedures (usually a chest x-ray, liver ultrasonography, and blood analysis), with the exception of the brain, for which conventional imaging was more accurate than PET scanning (93% vs. 82%, respectively). FN PET scans occurred in patients with small skin metastases and brain metastases. Wagner et al.13 reported the failure of PET scanning to detect microscopic nodal disease in patients undergoing sentinel lymph node biopsy. A recent update of this study confirmed that in patients with early-stage melanoma, PET scanning is unlikely to detect occult regional or distant disease.14
Two recent studies addressed the question of the utility of surgical staging by PET scanning. Tyler et al.15 reported their experience with 95 patients with clinically evident stage III melanoma in an attempt to determine the sensitivity and specificity of PET imaging in a homogeneous population of patients. They reported that findings on PET scan altered clinical management in 15% of these patients. Unfortunately, they reported an FP rate of 57%, primarily due to recent surgery. The sensitivity of PET scanning in these patients was 87%.
Finkelstein et al.6 reported a recent experience with preoperative PET scanning in 18 patients enrolled in immunotherapy protocols at the National Cancer Institute. They compared conventional imaging with PET scanning on a lesion-by-lesion basis in stage IV patients undergoing metastasectomy and found that the combination of PET and CT scanning increased the positive predictive value of imaging from 86% with CT or PET to 91% when CT and PET were both used. This study confirms the additive value of PET scanning in detecting metastatic disease in these high-risk patients.
Our study is unique because it demonstrates the utility of PET scanning in a prospective trial in a clinically relevant, well-defined group of high-risk melanoma patients. Our focus was to determine the additive clinical utility of PET for the individual high-risk patient. We required that patients remain without recurrence for 4 months from study entry for the initial imaging studies to be deemed TN. This requirement was made because in most patients who undergo resection of advanced disease, relapse at occult sites within 4 months of resection of known disease could be considered a failure to properly identify patients for operative management. No attempt was made to address the importance of local or regional control of disease despite recurrence elsewhere, but merely to determine, in high-risk melanoma patients undergoing resection with curative intent, whether the surgical intervention was appropriate.
Not surprisingly, we found that preoperative imaging was most likely to be accurate when findings were noted on PET and CT, as opposed to PET or CT alone. Like many new diagnostic tests, PET scanning is unlikely to replace the previous standard. FDG-PET scanning remains more expensive than CT imaging and is not yet widely available. FDG-PET scanning often provides additional imaging information, however, to allow more accurate staging when interpreted in conjunction with findings noted on CT scan. Surgeons reluctant to act on CT scan findings alone, which are often considered equivocal, were much more likely to base important management decisions on CT findings confirmed by PET or PET findings also noted on CT imaging.
One of the pitfalls of this study is the inherent bias present in the study population due to the fact that approximately one third of patients referred for resection of regional or distant disease had undergone CT scans before referral. This may have had the effect of making PET scanning seem relatively more useful than CT scanning, because those with unresectable occult metastatic disease by CT would not be referred for operative therapy. Despite this, the design of this study mimics quite closely the clinical situation that faces the oncologist who cares for patients with potentially resectable regional or distant disease.
Should PET scanning be considered the standard of care in the preoperative evaluation of patients undergoing resection of regional, distant, or advanced primary disease? Our data demonstrate that imaging findings identified by a combination of pre-operative PET and CT scanning resulted in an appropriate change in management in approximately one third of patients (32 of 103; 31%). It is important to remember, however, that this study was conducted in a large, tertiary referral center with significant PET expertise and a very experienced nuclear medicine staff. Nonetheless, as PET scanning is increasingly used in the management of patients with melanoma, enhanced expertise and experience will become more widely available. In this setting, preoperative CT and PET imaging should be strongly considered in patients with regional or distant melanoma considered for operative resection.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication February 8, 2005. Accepted for publication October 13, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. Krug, R. Crott, M. Lonneux, J.-F. Baurain, A.-S. Pirson, and T. Vander Borght Role of PET in the Initial Staging of Cutaneous Malignant Melanoma: Systematic Review Radiology, December 1, 2008; 249(3): 836 - 844. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Farma, A. A. Santillan, M. Melis, J. Walters, D. Belinc, D.-T. Chen, E. A. Eikman, and M. Malafa PET/CT Fusion Scan Enhances CT Staging in Patients with Pancreatic Neoplasms Ann. Surg. Oncol., September 1, 2008; 15(9): 2465 - 2471. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
