High predictive value of F-18 FDG PET patterns of the spine for metastases or benign lesions with good agreement between readers.

Two nuclear medicine physicians retrospectively evaluated fluorine-18 fluorodeoxyglucose (FDG) positron emission tomography (PET) spine abnormalities in patients with cancer with the purpose of identifying straightforward criteria for benign versus malignant spine abnormalities. Four hundred seventy-five consecutive patients with colon, breast, and lung cancer were evaluated with FDG. Thirty-two patients (32) had spine abnormalities, 30 of 32 patients had adequate follow up for a final diagnosis, and 29 of 30 patients' studies were available to both PET readers for this retrospective review. The readers categorized the FDG PET abnormalities as benign, metastatic, or equivocal using a straightforward set of criteria. A final diagnosis was made using magnetic resonance imaging (MRI), computed tomography (CT), plain films, bone scans, previous studies, and clinical follow up. A single spinal focus of increased FDG activity had a relatively high probability of being a spinal metastasis (71%); and the more foci, the higher the probability. Segmental decreased activity of the spine after radiation therapy indicated benignity. The only discrepancies were with 3 abnormalities, each called metastasis by 1 reader and equivocal by the other, with a final diagnosis of metastasis in each case. Equivocal patterns required CT or MR correlation, because these could be either malignant or benign. However, abnormal patterns fulfilling either the benign or metastatic criteria described here resulted in the correct diagnoses of benign spinal changes or spinal metastases, respectively, in 100% of cases with low interobserver variation. No study was interpreted as benign by 1 reader and metastasis by the other. The 2 nuclear medicine readers agreed in their interpretations in 90% of cases.