[18F]Fluciclovine PET discrimination between high- and low-grade gliomas

Background: The ability to accurately and non-invasively distinguish high-grade glioma from low-grade glioma remains a challenge despite advances in molecular and magnetic resonance imaging. We investigated the ability of fluciclovine (F-18) PET as a means to identify and distinguish these lesions in patients with known gliomas and to correlate uptake with Ki-67. Results: Sixteen patients with a total of 18 newly diagnosed low-grade gliomas (n = 6) and high grade gliomas (n = 12) underwent fluciclovine PET imaging after histopathologic assessment. Fluciclovine PET analysis comprised tumor SUVmax and SUVmean, as well as metabolic tumor thresholds (1.3*, 1.6*, 1.9*) to normal brain background (TBmax, and TBmean). Comparison was additionally made to the proliferative status of the tumor as indicated by Ki-67 values. Fluciclovine uptake greater than normal brain parenchyma was found in all lesions studied. Time activity curves demonstrated statistically apparent flattening of the curves for both high-grade gliomas and low-grade gliomas starting 30 min after injection, suggesting an influx/efflux equilibrium. The best semiquantitative metric in discriminating HGG from LGG was obtained utilizing a metabolic 1 tumor threshold of 13* contralateral normal brain parenchyma uptake to create a tumor background (TBmean1.3) cutoff of 2.15 with an overall sensitivity of 975% and specificity of 95.5%. Additionally, using a SUVmax > 43 cutoff gave a sensitivity of 90.9% and specificity of 97.5%. Tumor SUV mean and tumor SUVmax as a ratio to mean normal contralateral brain were both found to be less relevant predictors of tumor grade. Both SUVmax (R = 0.71, p = 0.0227) and TBmean (TBmean1.3: R = 0.81, p = 0.00081) had a high correlation with the tumor proliferative index Ki-67. Conclusions: Fluciclovine PET produces high-contrast images between both low-grade and high grade gliomas and normal brain by visual and semiquantitative analysis. Fluciclovine PET appears to discriminate between lowgrade glioma and high-grade glioma, but must be validated with a larger sample size.