LAUSR

Abstract OBJECTIVE: The prognosis and treatment of pediatric low-grade gliomas (pLGGs) is influenced by their molecular subtype. MRI remains the mainstay for initial work-up and surgical planning. We aimed to determine the relationship between imaging patterns and molecular subtypes of pLGGs. METHODS: This is a bi-institutional retrospective study for patients diagnosed from 2004 to 2021 with pathologically confirmed pLGG, molecularly defined as BRAF fusion (KIAA1549-BRAF), BRAF V600E mutation, or wild-type (negative for both BRAF V600E mutation and BRAF fusion). Two neuroradiologists, blinded, independently reviewed imaging parameters on the initial MRI and discrepancies were solved by consensus. Bivariate analysis was used followed by pairwise comparison of Dwass, Steel, and Critchlow-Fligner methods to compare the 3 molecular subtypes. Agreement between reviewers was assessed using Kappa (k). RESULTS: 70 patients were included: 30 with BRAF fusion, 19 with BRAF V600E mutation, and 21 wild-type. There was substantial agreement between the two readers for overall imaging variables (k=0.75). BRAF fusion tumors compared to V600E and wild-type had larger size (p=0.0022), greater mass effect (p=0.0053), and increased rate of hydrocephalus (p=0.0002). BRAF fusion tumors had increased frequency of diffuse enhancement compared with BRAF V600E and wild-type (p <0.0001). BRAF V600E mutant tumors were more often located in a cerebral hemisphere (p <0.0001). Diffusion restriction (qualitatively) was uncommon but only seen in BRAF V600E (p=0.0042) with lower ADC ratio (quantitatively) (p=0.003). Additionally, BRAF V600E mutant tumors appeared more infiltrative than BRAF fusion and wild-type (p=0.0002). CONCLUSION:BRAF fusion and BRAF V600E mutant pLGG have unique imaging features that can be used to differentiate from each other and wild-type pLGG using standard radiology review with high inter-reader agreement. In the era of targeted therapy, these features can be useful for therapeutic planning prior to surgery.