Abstract
BACKGROUND: Although frequently used for tumor surveillance, the sensitivity of carcinoembryonic antigen (CEA) to detect recurrent colorectal cancer (CRC) is not optimal. Fluorine 18-fluoro-2-deoxy-glucose-positron emission tomography (18F FDG-PET) scans promise to improve recurrent CRC detection. We aimed to review PET scans of patients with clinically and/or radiologically suspicious tumor recurrence but normal CEA. METHODS: A retrospective review of an electronic database of 308 patients with CRC who had PET scans was performed. Only PET studies of patients with normal CEAs and suspected tumor recurrence who had pathological verification were selected for further analysis. Thirty-nine patients met the inclusion criteria. RESULTS: PET was positive in 26 patients (67%) and normal in 13 (33%). Histopathologic evidence of tumor recurrence was seen in 27 of the 39 patients (69%). When correlated with histopathology, PET was true positive in 22 patients, false positive in 4, true negative in 8 and false negative in 5. Overall, the accuracy of PET was 76.9%, negative predictive value (NPV) was 61.5%, and positive predictive value (PPV) was 84.6%. PPV value of PET for liver metastases was 88.8% compared to 73.3% for local recurrence. In two patients with confirmed recurrence, CEA became positive 2 months after PET scan indicating earlier detection of disease with PET. The false positive PET findings were mainly in the bowel and were secondary to acute/chronic inflammation and granulation tissue. In 3 patients with false negative PET, histopathology was consistent with mucinous adenocarcinoma. CONCLUSION: PET yields high PPV for recurrent CRC, particularly for liver metastases, in spite of normal CEA levels and should be considered early in the evaluation of patients with suspected tumor recurrence.