Futility of Fluorodeoxyglucose F 18 Positron Emission Tomography in Initial Evaluation of Patients With T2 to T4 Melanoma

Background  Evaluation of newly diagnosed patients with melanoma for metastasis is requisite to treatment planning. The reported diagnostic yield of whole-body conventional radiological imaging in initial staging of patients with melanoma is low. However, the diagnostic yield of positron emission tomography (PET) for distant metastases is unclear.

Hypothesis  There is no utility of PET as part of a routine metastatic survey in patients with T2 to T4 melanoma.

Design  Retrospective review of a cohort study between December 1998 and July 2004.

Setting  University hospital tertiary care center.

Patients and Methods  There were 64 patients with T2 to T4 melanomas who underwent PET for detection of occult metastases at our institution. All patients underwent surgical excision of the primary lesion and sentinel lymph node dissection. Data included were pathologic findings of the primary lesion and sentinel lymph nodes, laboratory data, and radiological reports. None of the patients had clinically suspected regional or distant metastases prior to PET. The diagnostic yield of PET was evaluated through retrospective analysis. Positive scans were then correlated for accuracy with follow-up imaging, biopsy, and clinical information when available.

Results  Positron emission tomography did not reveal occult distant metastases in any of the patients. Positron emission tomographic scans showed no abnormalities in 94% of these patients. In 2 patients (3%), false-positive findings were reported on PET (muscular activity and intranodal melanocytic nevocellular inclusion). Further, PET was not useful in predicting regional lymph node metastases. Nineteen of 64 patients had positive sentinel lymph nodes, and only 2 (11%) were identified on PET. Overall, PET did not change clinical management in any of the patients.

Conclusions  This study suggests no utility for PET in the detection of occult metastases in patients at initial diagnosis of melanoma. Omission of PET imaging from preoperative evaluations for patients with melanoma is recommended.

Positron emission tomography (PET) using fluorodeoxyglucose F 18 has proven useful for the initial staging of several malignancies, including colon, lung and esophageal carcinomas, and lymphoma. In patients with melanoma, the detection of regional lymph node and distant metastases substantially alters treatment and adversely affects prognosis. In patients at high risk for or with known distant metastases, PET has been shown to have a sensitivity of greater than 90%. Compared with conventional imaging for patients with melanoma with known or suspected distant metastases, it has been reported that PET has a sensitivity and specificity of 94% and 83%, respectively, compared with 55% and 84%, respectively, for computed tomography (CT).¹ For regional staging of primary melanoma, many studies have shown that PET cannot replace sentinel lymph node biopsy in T1 tumors. Since the current lower limit of PET resolution is 80 mm³, PET has no ability to reliably diagnose microscopic nodal disease.^2,3 Therefore, PET is often used to confirm clinical suspicion of metastases, evaluate patients for recurrent disease, and restage prior to changes in medical or surgical management for patients with local or distant metastases.¹

The value of PET for initial staging of melanoma in patients without clinical evidence of distant metastases is unclear. A recent report of 185 patients with primary melanoma and at least 1 positive sentinel lymph node showed that conventional radiologic imaging—chest radiography, CT, and magnetic resonance imaging—was not indicated in the initial workup of patients with melanoma being evaluated with sentinel lymph node mapping because the diagnostic yield was less than 1%.⁴ However, PET imaging was not included in that study. In another retrospective study, 101 patients in Switzerland with primary melanoma (Breslow thickness >1.0 mm) underwent baseline staging using PET. No distant metastases were detected. In 2 of these 101 patients, PET findings were suspicious for metastases, but we were unable to confirm it by CT and the patients did not show any progression on follow-up, therefore, we consider them to be false-positive findings.⁵

In this retrospective analysis, we explore the utility of whole-body PET imaging in 64 patients with T2 to T4 melanomas prior to sentinel lymph node dissection without clinically suspected metastases at our tertiary care institution.

Approval for this study was obtained from the institutional review board at Wake Forest University, Winston-Salem, NC. A total of 178 patients with melanoma (Breslow thickness >1 mm) underwent sentinel lymph node biopsy between December 1998 and July 2004 at our tertiary care institution. Of these, 66 patients had a PET scan for staging purposes based on attending preference, but predominantly for lesions thicker than 2 mm. Two of these patients had clinically suspected metastases (confirmed with PET), leaving 64 asymptomatic patients for evaluation. All patients underwent wide excision of the primary lesion and sentinel lymph node dissection. Data recorded for this review include primary lesion surgical pathologic findings, sentinel lymph node pathologic findings, laboratory data, and radiological reports. The diagnostic yield of PET was evaluated through retrospective report analysis. Positive scans were then correlated for accuracy with follow-up imaging, biopsy, and clinical information.

After fasting for 4 hours, patients were injected with 15 to 20 mCi (555-740 MBq) of fluorodeoxyglucose F 18. Forty-five minutes later, images were obtained on a GE Advance PET scanner (GE Medical Systems, Milwaukee, Wis). Attenuation-corrected images were obtained using transmission scans. For head, neck, and extremity lesions, nonattenuation-corrected images were also analyzed. Concurrent CT (when available) was evaluated side-by-side with the PET images. Images were read by 1 of 4 physicians trained in nuclear medicine.

Preoperatively, 0.5 to 1.0 mCi (18.5-37 MBq) of technetium Tc 99m–filtered sulfur colloid was injected intradermally in 4 quadrants around the primary tumor or biopsy site. In most cases, a lymphoscintigram was obtained to determine the draining lymph node basin. The area over the sentinel lymph node, as seen by the external gamma camera, was marked on the skin. Perilesional intradermal injections of isosulfan blue were performed at surgery, followed by 5-minute massages of the site. A gamma probe (Neoprobe 2000; Neoprobe Corp, Dublin, Ohio) was used intraoperatively to detect sentinel lymph nodes. Lymph nodes were considered sentinel if they were blue and/or concentrated radiotracer. All nodes with more than 10% of the hottest sentinel node recovered were also considered sentinel lymph nodes. After identification of sentinel lymph nodes with gamma counter and blue dye, intraoperative imprint cytology (touch preparation) was evaluated and reported as positive or negative.⁶ Permanent section pathologic evaluation was performed on at least 3 sections through each sentinel node. If the standard histologic finding was negative, an additional 3 levels were obtained for immunohistochemistry with S100 protein and HMB-45. Nodal micrometastasis was defined as less than 2 mm; macrometastasis was defined as greater than 2 mm.

Sixty-four patients (23 women, 41 men) aged 24 to 86 years (mean, 57 years) were included in the study. Tumor characteristics are summarized in Table 1. None of the patients had clinically suspected regional or distant metastases. Average length of follow-up was 19 months. There were 11 patients with nodal micrometastases and 6 patients with nodal macrometastases. Of the 4 patients with macrometastases that were negative by PET, the metastases ranged from 2 to 6 mm in size.

On initial diagnosis, PET scans were normal in 60 of 64 patients (94%). Neither PET nor concurrent CT (when available) revealed occult distant metastases (Table 2). In 2 (3%) cases of positive findings on PET, increased activity was noted in benign processes. One patient had increased fluorodeoxyglucose F 18 uptake in a muscular distribution (extensor digitorum brevis) with no corresponding abnormalities on skin examination (Figure 1). The second patient had increased uptake in a microscopic intranodal melanocytic nevocellular inclusion (nodal nevus) of a sentinel lymph node, confirmed on histopathologic findings (Figure 2). The patient was not rescanned postoperatively.

In the other 2 cases with positive findings (3%), regional metastatic lymphadenopathy was found on PET and confirmed at sentinel lymph node biopsy. The first patient had an ulcerating 3.5 × 3.5-cm scalp lesion, which was more than 4 mm thick with clinically nonpalpable lymph nodes. Positron emission tomography showed subcutaneous metastatic deposits as well as cervical lymphadenopathy (Figure 3). At surgery, 2 grossly positive sentinel lymph nodes were discovered. Subsequent cervical lymphadenectomy yielded 9 of 45 positive lymph nodes.

The second patient had axillary lymphadenopathy on PET, with grossly involved lymph nodes at surgery (Figure 4). Positron emission tomography was unable to predict the status of the sentinel lymph nodes. Of 64 patients, 19 had positive sentinel lymph nodes at surgery. As discussed above, only 2 (11%) of 19 were identified on PET (both macrometastases). Overall, PET did not change management in any of the patients.

Initial staging and evaluation of patients with melanoma is crucial to initiating appropriate therapy. The finding of distant metastases portends an ominous prognosis and surgical salvage is a realistic possibility for a small minority of such cases. In an effort to evaluate patients with intermediate-to-thick lesions for distant metastasis, we performed screening PET imaging during a period of 4 years. This report is a retrospective evaluation of that practice. This review clearly shows no utility for staging the primary lesions of patients with T2 to T4 melanomas in the absence of signs or symptoms suggestive of distant disease.

These results echo similar findings in other studies. The limited utility of conventional imaging (chest radiography, CT, magnetic resonance imaging) was reported in 185 patients undergoing sentinel lymph node mapping, which showed that less than 1% were positive for distant metastases.⁴ It is also consistent with a series of 101 Swiss cases in which PET identified no distant metastases in baseline staging.⁵ In this series of 64 patients with T2 to T4 melanomas, none had clinically occult metastases by PET or CT. In 19 patients with nodal metastasis identified via sentinel lymph node mapping, none had distant metastases by PET. Positron emission tomography did not change medical or surgical management of any patient.

It is interesting that the number of patients with true-positive findings on PET equaled the number of those with false-positive findings. One false-positive finding that showed malignant-range activity on PET, originally considered worrisome for a second melanoma, turned out to be muscular uptake in the extensor digitorum brevis. No associated abnormality was discovered on clinical inspection. This uptake was attributed to focal inflammation or injury. A second false-positive case involved malignant-range uptake in an intranodal melanocytic nevocellular inclusion (nodal nevus) that was confirmed histopathologically. Nodal nevi can mimic metastases in the sentinel lymph nodes of patients with melanoma. However, when cytologically compared with the primary melanoma, these nevi are bland nevocellular aggregates that do not resemble melanoma.⁷ A PubMed database search revealed no reported cases of increased fluorodeoxyglucose F 18 uptake in nodal nevi. Sentinel lymph node dissection confirmed the 2 true-positive PET findings, both in sentinel lymph nodes that were grossly positive at surgery.

As with any imaging modality, false-positive findings and indeterminate lesions occur. However, if PET adds no clinically useful staging information to this subset of patients, consideration of iatrogenic physical and psychosocial morbidity associated with evaluation of false-positive imaging results are substantial.

Some limitations to this study must be considered. Since this represents a retrospective analysis, the possibility of selection bias exists. However, there were no PET findings in the 64 patients to make this significant. Also, the small sample size cannot exclude a minor potential utility for PET in the initial evaluation of melanoma; however, the cost of PET makes such an insignificant rate of detection less cost-effective. The thicker the lesion, the more likely it is that metastatic disease exists. Because there were only 22 patients with thick (T4) lesions, there is a possibility of some limited utility of PET for thicker lesions and this should be considered. Further, in patients with clinically apparent metastasis (which we did not assess in this study), PET may be of greater utility. With the advent of PET and CT, the question also arises whether this modality will be more effective than PET alone in the detection of regional and distant melanoma metastases. Our group plans to study this question over the next few years.

Meanwhile, this study of 64 patients suggests little utility for PET in the detection of occult melanoma metastases at initial diagnosis. Although we continue to suggest PET evaluation for selected patients with signs and symptoms of metastatic melanoma, we see no role for routine PET imaging for asymptomatic patients who will be evaluated with sentinel node mapping. Further, it remains clear that negative PET imaging cannot be considered a viable substitute for sentinel node mapping and biopsy.

Correspondence: Edward A. Levine, MD, Surgical Oncology, Medical Center Blvd, Winston-Salem, NC 27157 (elevine@wfubmc.edu).

Accepted for Publication: June 14, 2005.