NIMG-33. [18F]-FLUOROMISONIDAZOLE RELATIVE PET INFLUX RATE IMPROVES DIAGNOSTIC ACCURACY FOR GLIOBLASTOMA PROGRESSION FROM THERAPY-INDUCED NEUROINFLAMMATORY PSEUDOPROGRESSION

Abstract

Abstract Glioblastoma is the most common and aggressive form of primary brain tumor in adults. Standard of care gadolinium-enhanced MRI (Gd-MRI) fails to capture hypoxia, a disease-defining feature of glioblastoma. Hypoxia potentiates resistance to chemo-radiation therapy (CRT) but is not extensively observed with therapy-induced pseudoprogression (PSP). This neuroinflammatory response to CRT occurs in approximately 40% of patients and is indistinguishable from disease progression by Gd-MRI, potentially causing unnecessary termination of effective therapy in patients with PSP. Additionally, delayed second line-therapy can occur in patients with resistance to CRT. This research evaluated whether [18F]-fluoromisonidazole (FMISO) positron emission tomography (PET), a noninvasive metric of tissue hypoxia, improves diagnostic accuracy in this context. Clinically, PET quantification employs the standardized uptake value (SUV), representing summed counts over an acquisition period. To add accuracy to this assessment, static 40-minute PET data acquired starting 90 minutes after FMISO injection was reconstructed into 20x2-minute frames and a relative Patlak model was applied. This technique forgoes blood sampling and extensive examination times required by traditional dynamic PET studies. The model produces two parameters that separately characterize the FMISO relative influx rate (Ki') and blood volume (VB'). In a cohort of 16 patients (3 IDH-mutated) imaged at a time of presumed disease progression, results showed that Ki' within the Gd-MRI enhancing lesion predicts future diagnosis of true progression (n = 11) or inflammation (n =5 ) with a sensitivity and specificity of 91% and 80% respectively. This outperforms diagnosis made with Gd-MRI alone, achieving ~70% for the same metrics. A t-test assuming unequal variance of cohort-wide mean Ki' tended toward significance (p = 0.07) for differentiating progressive disease (0.0036 ± 0.0016 min-1) from PSP (0.0011 ± 0.0017 min-1). RESULTS: from this study suggest relative Patlak analysis adds specificity to Gd-MRI and provides clinically relevant information regarding disease status.