[Usefulness and limitation of FDG-PET in the diagnosis of primary central nervous system lymphoma].

Primary central nervous system lymphoma(PCNSL)can be diagnosed with its typical neuroradiological findings. However, the diagnosis of atypical PCNSL is sometimes difficult because of its mimicking of other tumorous and non-tumorous brain diseases. Positron emission tomography(PET)with 2-[18F]-fluoro-2-deoxy-D-glucose(FDG)provides information on the glucose metabolism of brain tumors. This study was conducted to clarify the usefulness and limitation of FDG-PET in the diagnosis of PCNSL. We performed FDG-PET in 25 histologically-proven PCNSL cases and measured the maximum standardized uptake value(SUVmax)and the tumor-to-normal tissue count density(T/N)ratio of the tumors. In 25 histologically-proven cases of PCNSL, 19 showed typical neuroradiological findings and 6 showed atypical findings such as disseminated or no lesions. We also performed FDG-PET in 28 histologically-proven glioblastoma multiforme(GBM)and 7 clinically and neuroradiologically-suspected but histologically-unproven PCNSL cases and compared the uptake values with histologically-proven PCNSL. Typical PCNSL showed very high FDG uptake in the tumors with a mean SUVmax of 16.5±5.5 and a mean T/N ratio of 3.04±0.88 and these values were significantly higher than those in GBM(mean SUVmax of 10.0±4.7, p<0.001 and mean T/N ratio of 1.54±0.68, p<0.001). Receiver operating characteristic(ROC)curve analysis demonstrated that the T/N ratio had a higher accuracy in discrimination between PCNSL and GBM with a sensitivity of 0.95 and a specificity of 0.75 when a cutoff value was set on the T/N ratio of 1.8. On the other hand, atypical PCNSL showed mild FDG uptake in the tumor with a mean SUVmax of 7.6±2.4 and a mean T/N ratio of 1.33±0.47, which were not significantly different from the values in GBM. Patients with suspected PCNSL showed very high and similar FDG uptake values to those in histologically proven typical PCNSL, with a mean SUVmax of 18.3±9.4 and a mean T/N ratio of 2.70±0.67. These patients responded very well to chemotherapy and radiotherapy for PCNSL. In conclusion, FDG-PET is very useful in the diagnosis of typical PCNSL and can differentiate PCNSL from GBM with a high sensitivity and specificity. Moreover, FDG-PET has a supplementary role to the neuroradiological diagnosis of histologically unproven PCNSL. However, the role of FDG-PET is limited in the diagnosis of atypical PCNSL, but dynamic analysis of FDG-PET may overcome this issue.