ISRA May 2022

Role of MRI for Differentiating Pediatric Low-Grade Cerebral Tumors according to BRAF and NF1 Status

Jennie Belenky 1 Ariel Kerpel 1 Michal Yalon 2,3 Michal Lurye 2 Chen Hoffmann 1,3 Shai Shrot 1,3
1Section of Neuroradiology, Division of Diagnostic Imaging, Sheba Medical Center, Israel
2Department of Pediatric Hemato-Oncology, Sheba Medical Center, Israel
3Sackler School of Medicine, Tel Aviv University, Israel

Purpose: BRAF and NF1 status are distinctive features in pediatric low-grade brain tumors with prognostic and therapeutic implications. This retrospective study aimed to analyze the imaging features of BRAF V600E-mutant, BRAF-KIAA fusion, and wild-type BRAF versus NF1-associated pediatric low-grade brain tumors.

Methods: We retrospectively evaluated 40 pediatric patients with histologically proven pilocytic astrocytoma (PA, n=32), ganglioglioma (GG, n=4), pleomorphic xanthoastrocytoma (PXA, n=3), and diffuse astrocytoma grade 2 (n=1). Ten NF1 patients underwent conventional MRI with a diagnosis of a low-grade tumor without a biopsy. BRAF molecular analysis was performed for non-NF1 patients. Eleven patients presented BRAF V600E-mutant, twenty patients had BRAF-KIAA fusion, and nine BRAF wild-type tumors. Imaging studies were reviewed for dominant site, margin definition, hemorrhage, calcifications, cystic components, and contrast enhancement. Histogram analysis of tumoral diffusivity was performed.

Results: There were significant differences between evaluated low-grade tumors in their margins and cystic component (p=0.04 and p<0.001). Well-defined margins were characteristic for BRAF-KIAA or wild-type BRAF rather than BRAF V600E-mutant or NF1 tumors. None of the NF1 tumors showed a cystic component. Diffusion histogram metrics (mean, median, 10th and 90th percentiles), but not kurtosis or skewness, were different between tumors (p<0.005). Diffusivity was lowest in BRAF V600E-mutant tumors (10th percentile reached AUC of 0.9 on ROC analysis).

Conclusion: Imaging features of pediatric low-grade brain tumors, including quantitative diffusion metrics, may assist in predicting BRAF and NF1 status, suggesting radio-genomic correlation and might enable rapid molecular characterization.