Accuracy of 18F-FDG PET-CT in triaging lung cancer patients with suspected brain metastases for MRI

Objective Brain metastases are common in lung cancer. Whole-body 2-deoxy-2-[fluorine-18] fluoro-D-glucose ([F-18] FDG) PET/computed tomography (CT) is used for general staging, but MRI is the best modality for characterizing brain abnormalities. We aimed to determine whether PET/CT is suitable for selecting patients for MRI on the suspicion of brain metastases. Materials and methods F-18-FDG PET/CT (from the vertex to mid-thigh) was performed in 1108 consecutive patients suspected of lung cancer. The final diagnoses were extracted from medical records as lung cancer, with or without brain metastases, other kinds of cancers, or no cancer. The sensitivity, specificity, and positive predictive value for detecting brain metastases were calculated. Interobserver variation was tested in a subset of 88 PET/CT scans. Results Of the 1108 referred patients, 596 had lung cancer. Sixty-six had brain metastases. One PET/CT was false positive. Thirty-one scans were true positive among the 43 patients who were diagnosed with brain metastases 1 month before to 3 months after PET/CT (metastasis prevalence, 7.3%). Twelve PET/CT scans were false negative. Sensitivity, specificity, and positive predictive values were 72, 100, and 97%, respectively. Interobserver agreement between two experienced observers was high (kappa = 0.83), whereas agreement between the experienced and the inexperienced observer was poor. Conclusion The sensitivity of brain PET/CT for detecting brain metastases in lung cancer was above 70%, and the specificity was very high. Thus, PET/CT may be suitable for selecting patients for MRI in diagnostic centers that do not perform routine MRI in the pretherapeutic staging workup. The agreement among experienced readers was very high. Copyright (C) 2015 Wolters Kluwer Health, Inc. All rights reserved.