Abstract
BACKGROUND: Primary central nervous system lymphoma (PCNSL) can be differentiated from glioblastoma multiforme (GBM) using positron emission tomography (PET) with [(18)F]fluoro-2-deoxy-D-glucose (FDG). However, differentiation is often difficult with magnetic resonance imaging (MRI) or FDG PET alone. We have used various PET tracers to aid glioma diagnosis; here, we assessed whether multiple PET tracers improve the distinction between GBM and PCNSL. METHODS: We studied 148 patients with newly diagnosed brain tumors: 96 with GBM and 52 with PCNSL (according to the 2016 World Health Organization classification). Tumor-to-normal tissue ratios (TNRs) were calculated for FDG, L[methyl[(11)C]]methionine, and 3′deoxy3′[(18)F]fluorothymidine (FLT). Tumor-to-blood ratio (TBR) was measured for [(18)F]fluoromisonidazole (FMISO). RESULTS: Median FDG TNR was 1.52 (interquartile range [IQR], 1.13–2.13) for GBM and 2.89 (IQR, 1.95–3.85) for PCNSL. MET TNR was 6.38 (IQR, 4.68–7.48) for GBM and 4.01 (IQR, 3.28–7.52) for PCNSL. FLT TNR was 17.35 (IQR, 11.10–22.28) for GBM and 25.61 (IQR, 15.91–55.30) for PCNSL. FMISO TBR was 2.72 (IQR, 2.22–3.62) for GBM and 1.58 (IQR, 1.04–1.93) for PCNSL. Receiver operating characteristic analysis showed areas under the curve of 0.78 (FDG), 0.62 (MET), 0.69 (FLT), and 0.85 (FMISO). FLT TNR had the highest sensitivity of 83.2% while FMISO TBR had the highest specificity of 84.2%. Among the four tracers, FLT showed the highest sensitivity and FMISO showed the highest specificity. Dual‑tracer combinations did not improve overall diagnostic accuracy beyond the best single tracers. CONCLUSIONS: Among four PET tracers, FMISO was most effective in distinguishing GBM from PCNSL. Diagnostic accuracy was highest when each tracer was used individually.