Perfusion Showdown: Comparison of Multiple MRI Perfusion Techniques in the Grading of Pediatric Brain Tumors

• Published in: American journal of neuroradiology : AJNR Vol. 46; no. 7; p. 1464
• Main Authors: Ho, Chang Y., Supakul, Nucharin, Anthony, Greg, Wen, Qiuting, Chen, Paula, Allen, Maryann, Parker, Jason G., Gray, Benjamin R., Chen, Eric, Territo, Paul R., Persohn, Scott A., Kralik, Stephen F., Hutchins, Gary D.
• Format: Journal Article
• Language: English
• Published: United States 01.07.2025
• Subjects: Adolescent; Brain Neoplasms - diagnostic imaging; Brain Neoplasms - pathology; Child; Child, Preschool; Diffusion Magnetic Resonance Imaging - methods; Female; Humans; Infant; Magnetic Resonance Angiography - methods; Male; Neoplasm Grading; Reproducibility of Results; Sensitivity and Specificity
• ISSN: 0195-6108; 1936-959X; 1936-959X
• DOI: 10.3174/ajnr.A8635

There are multiple MRI perfusion techniques, with limited available literature comparing these techniques in the grading of pediatric brain tumors. For efficiency and limiting scan time, ideally only one MRI perfusion technique can be used in initial imaging. We compared DSC, dynamic contrast enhancement (DCE), and intravoxel incoherent motion (IVIM) along with ADC from DWI for differentiating high- versus low-grade pediatric brain tumors. Presurgical MRI at a single pediatric hospital for primary brain tumor of glial, neuronal, or embryonal origin including DWI, DSC, DCE, and IVIM with postsurgical histopathology results with World Health Organization (WHO) tumor grading were included. Tumor VOI was drawn on T2-weighted images registered to selected parametric maps from DWI, DSC, DCE, and IVIM. Multiple Bonferroni-corrected tests were performed on the mean tumor values for each selected parameter, including histogram analysis of 95th percentile relative CBV (rCBV), 5th percentile ADC, and 5th percentile D, to evaluate for significance between high- and low-grade tumors. Receiver operating characteristic (ROC) analysis was performed for significant test results. Seventy-two subjects were included with 36 low-grade and 36 high-grade tumors. test after Bonferroni correction yielded significant results for 5th percentile ADC ( = .003), mean ADC ( = .006), mean D ( = .009), and 5th percentile D ( = .02). Specifically, 95th percentile rCBV, mean rCBV, D*, f, and all DCE parameters were not significant for high- versus low-grade pediatric brain tumors after correction. Only mean rCBV was significant before but not after Bonferroni correction ( = .04 → = .35). ROC analysis for the parameters with test significance before and after Bonferroni correction had the following AUC, in descending order: 5th percentile ADC (0.791, 0.68-0.88, < .001), 5th percentile D (0.789, 0.68-0.88, < .001), mean D (0.76, 0.64-0.85, < .001), mean ADC (0.754,0.64-0.85, < .001), and mean rCBV (0.683, 0.56-0.79, = .004). Perfusion parameters from IVIM, DCE, and DSC were not able to significantly differentiate between high- versus low-grade pediatric brain tumors compared with ADC in our study. Fifth percentile ADC performed the best and can be the primary technique for grading pediatric brain tumors.