Diagnostic Value and Reproducibility of Positron Emission Tomography with 18f-Fluciclovine in High- and Low-Grade Glioma

Abstract

Positron emission tomography (PET) with amino acid analogs is a useful tool in the diagnostics and therapy planning of gliomas. PET-based delineation of the tumor volume enables delivery of a more targeted radiation dose and sparing of normal brain tissue, with potential consequences for patient survival. PET tracer, 18F-fluciclovine, has Orphan Drug designation in the US and Europe for imaging gliomas. Here, we explored its diagnostic performance and reproducibility in high- and low-grade gliomas when interpreted in combination with contrast-enhanced T1-weighted MRI (hereafter referred to as “18F-fluciclovine + CE-T1W”). In a previous prospective study, 18F-fluciclovine PET and MRI (CE-T1W and fluid-attenuated inversion recovery [FLAIR]) were performed in patients with suspected gliomas (Anaplastic astrocytoma (7), Anaplastic oligodendroglioma (1), Diffuse astrocytoma (7), Glioblastoma (14), Oligoastrocytoma (2), Oligodendroglioma (3), other: Anaplastic ependymoma (1)) Imaging data from thirty five individuals with a histopathologically confirmed diagnosis of glioma were analyzed. A neuroradiologist read the MRI images, while three blinded, 18F-fluciclovine-naïve, experienced nuclear medicine physicians interpreted 18F-fluciclovine + CE-T1W. Diagnostic performance was determined for CE-T1W, FLAIR and 18F-fluciclovine + CE-T1W. Inter- and intra-reader reproducibility (via a re-read of a random 20% of images) were assessed using Cohen’s kappa and Fleiss’ kappa statistics. Results were stratified by tumor grade, i.e. high and low. Across the three readers, the positive predictive value (PPV) for 18F-fluciclovine + CE-T1W was 96%, the negative predictive value (NPV) was 38-42%, the sensitivity was 66−71%, and specificity was 89%. For high-grade tumors (n = 15), the PPV, NPV, sensitivity and specificity for 18F-fluciclovine + CE-T1W for the three readers overall were 100.0%, 40.0%, 83.3% and 100.0%, respectively. In low-grade tumors (n = 11), these were 100.0%, 45.5%, 50.0% and 100.0%, respectively. The biopsy samples for the remainder (n=9) demonstrated malignancy but the neuro-pathologist could not confidently assign a grade. Inter-reader concordance was 89% (Fleiss’ Kappa = 0.86) overall, 100% (Fleiss’ Kappa=1.00) for high-grade, and 82.4% (Fleiss’ Kappa = 0.74) for low-grade gliomas. All pair-wise inter-reader comparisons showed 100% agreement (Cohen’s Kappa = 1.0) for high-grade gliomas and ranged from 82.4−94.1% (Cohen’s Kappa = 0.6−0.87) for low-grade. Intra-reader concordance was 83.3−100% (Cohen’s Kappa =0.66−1.00) with no discernible differences when stratifying by high- or low-grade gliomas. 18F-Fluciclovine PET + CE-T1W MRI show reproducible results and good diagnostic performance for imaging gliomas, a prerequisite for accurate PET-guided radiotherapy planning.