Qualitative analysis of spinal intramedullary lesions using PET/CT.

OBJECT Although the usefulness of PET for brain lesions has been established, few reports have examined the use of PET for spinal intramedullary lesions. This study investigated the diagnostic utility of PET/CT for spinal intramedullary lesions. METHODS l-[methyl-C]-methionine (MET)- or [F]-fluorodeoxyglucose (FDG)-PET/CT was performed in 26 patients with spinal intramedullary lesions. The region of interest (ROI) within the spinal cord parenchyma was placed manually in the axial plane. Maximum pixel counts in the ROIs were normalized to the maximum standardized uptake value (SUV) using subject body weight. For FDG-PET the SUV was corrected for lean body mass (SUL) to exclude any influence of the patient's body shape. Each SUV was analyzed based on histopathological results after surgery. The diagnostic validity of the SUV was further compared with the tumor proliferation index using the MIB-1 monoclonal antibody (MIB-1 index). RESULTS A total of 16 patients underwent both FDG-PET and MET-PET, and the remaining 10 patients underwent either FDG-PET or MET-PET. Pathological diagnoses included high-grade malignancy such as glioblastoma multiforme, anaplastic astrocytoma, or anaplastic ependymoma in 5 patients; low-grade malignancy such as hemangioblastoma, diffuse astrocytoma, or ependymoma in 12 patients; and nonneoplastic lesion including cavernous malformation in 9 patients. Both FDG and MET accumulated significantly in high-grade malignancy, and the SUL and SUV correlated with the tumor proliferation index. Therapeutic response after chemotherapy or radiation in high-grade malignancy was well monitored. However, a significant difference in SUL and SUV for FDG-PET and MET-PET was not evident between low-grade malignancy and nonneoplastic lesions. CONCLUSIONS Spinal PET/CT using FDG or MET for spinal intramedullary lesions appears useful and practical, particularly for tumors with high-grade malignancy. Differentiation of tumors with low-grade malignancy from nonneoplastic lesions may still prove difficult. Further technological refinement, including the selection of radiotracer or analysis evaluation methods, is needed.