A prospective study of the accuracy of 18fluorodeoxyglucose positron emission tomography (18FDG PET) in identifying sites of metastasis prior to pelvic exenteration

Purpose. To determine the accuracy of (18)FDG PET in identifying sites of metastatic disease prior to pelvic exenteration or radical resection in patients (pts) with recurrent cervical or vaginal cancers. Methods. Pts with recurrent cervical or vaginal cancer being evaluated for surgical resection were enrolled in a prospective Study approved by the institutional human Subjects review board. All patients underwent (18)FDG PET scans as well as CT and/or MRI scans and were required to have pathologic confirmation of any sites suggestive of tumor recurrence. Results. Between 1998 and 2002 a total of 27 pts were enrolled on the study. Seven patients did not complete all study requirements and are excluded from further analysis. All pts had undergone prior pelvic radiation therapy and five patients had also received chemotherapy. CT/MRI scans identified three patients with possible metastatic disease in the following sites: (1) iliac nodes (2 pts) and (2) lungs (1 pt). After surgical and pathological evaluation, only one of these sites, the lungs, was confirmed to have metastatic disease. PET scans identified possible metastatic disease in nine patients and included the following sites: (1) pelvic nodes (4 pts), (2) para-aortic nodes (2 pts), (3) axillary node (1 pt), (4) bowel wall (1 pt) and (5) lungs (1 pt). After surgical and pathologic evaluation metastatic disease was identified in five of these pts at the following sites: iliac nodes, 2; para-aortic nodes, 1; bowel wall, 1; and lungs, 1. Of the sites identified by PET scan as areas of inetastasis CT scan only identified the pulmonary metastasis. Conclusion. (18)FDG PET was found to have a sensitivity of 100% and a specificity of 73% in detecting sites of extra-pelvic metastasis and may be the most accurate test to determine eligibility for pelvic exenteration. (c) 2007 Elsevier Inc. All rights reserved.