Maximum Uptake and Hypermetabolic Volume of 18F-FDOPA PET Estimates Overall Survival in Low-Grade Gliomas

Introduction: Magnetic resonance imaging (MRI), including fluid-attenuated inversion recovery (FLAIR), perfusion, and diffusion weighted imaging (DWI), plays a significant clinical role in the evaluation of morphology, cellularity, and vascularity of low-grade gliomas (LGGs). Prior studies have suggested the use of amino acid PET, including 18F-FDOPA PET, can provide critical metabolic information that can complement MRI; however, evaluation of LGGs using both 18F-FDOPA and MRI have been limited to small pilot studies and long-time prognosis was previously not reported. Recently, molecular subtypes like isocitrate dehydrogenase (IDH) gene mutation or a chromosomal 1p/19q co-deletion have been used for histopathological diagnosis in 2016 WHO classification, and have become essential for brain tumor classification, treatment decisions, and predicting prognosis. Thus, the purpose of the current study was to evaluate imaging characteristics of LGGs using 18F-FDOPA PET and multi-parametric MRI in a large population and determine the association with OS or molecular status.

Methods: Eighty-six patients who underwent 18F-FDOPA PET and were diagnosed as initial or recurrent LGGs according to the WHO 2007/2016 classification between 2007 and 2019 were retrospectively selected (IDH status, wild type 10, IDH mutant 58, unknown, 18; 1p19q status in IDH mutant, non-codeleted 20, codeleted 37, unknown 1; MGMT methylation status, unmethylated 17, methylated 25, unknown 44). FLAIRROI (hyperintensity area on FLAIR image) was segmented, and the normalized SUV max (nSUVmax) of 18F-FDOPA relative to striatum, FLAIRROI volume, 18F-FDOPA hypermetabolic volume (tumor-to-striatum ratios > 1), cerebral blood volume (CBV), and apparent diffusion coefficient low (ADClow), which was defined as the mean of lower Gaussian curve (a mixed model with a double Gaussian distribution fitting to the histogram of ADC), within FLAIRROI were calculated. Receiver-operator characteristic (ROC) analysis and Cox univariate regression were used for statistical analyses.

Results: ROC analysis revealed that combining PET and MRI parameters with sex and age could predict IDH and 1p19q status with 91.9% and 87.5% sensitivities and 71.4% and 85.7% specificities, respectively. Cox univariate regression analysis showed significant association between residual OS and IDH status (hazard ratio (HR)= 4.939, p=0.026), nSUVmax (HR=2.827, p=0.017), and 18F-FDOPA hypermetabolic volume (HR=1.048, p=0.001). Stratifying the LGGs into 3 groups by the nSUVmax (<1, 1-2, >2), the log-rank test and trend test showed significant differences (p<0.001, and 0.001, respectively). For initial LGGs (n=29), nSUVmax (HR=151.6, p=0.038) showed significant association with OS, while for residual or recurrent LGGs (n=57), volume of FLAIRROI (HR=1.006, p=0.049) and 18F-FDOPA hypermetabolic volume (HR 1.038, p=0.009) showed significant association with OS. CBV and ADClow did not associate with OS in all analyses (all, p>0.17).

Conclusions: In the largest study evaluating LGGs using 18F-FDOPA PET and MRI, nSUVmax and 18F-FDOPA hypermetabolic volume may be useful for estimating OS, and combining 18F-FDOPA PET and MRI may be useful for predicting molecular status in patients with LGGs.