RADT-70. WHITE AND GRAY MATTER DIFFERENTIATION IMPACTS DISTRIBUTION OF TUMOR TREATING FIELDS

Abstract BACKGROUND Tumor Treating Fields (TTFields) are an established adjuvant therapy for glioblastoma patients in combination with conventional surgery, radiation and chemotherapy. To model TTFields inside the brain, a key point of debate centers on whether distinguishing between gray and white matters significantly impacts field distribution, compared to treating both as a single entity. This study attempts to resolve this issue. METHODS T1-weighted MPR MR datasets from 56 models from 28 patients were segmented using established methods (PMID: 29023236 and 28544575). Standard tissue properties were applied to the baseline models, while 2 variations for each were created by assigning properties of gray matter as white matter and vice versa. Plan quality metrics (PQMs) for electric field intensity (E), specific absorption rate (SAR), and current density (CD) were evaluated across coverage levels at 95%, 50% and 5% of tissue for volumetric comparisons between models. Coefficient of variance and percent differences relative to baseline models were used to compare model variants. RESULTS Percentage differences in PQM between model variants revealed differences in E, SAR, and CD when compared with the baseline model, for gross tumor volume (GTV), gray matter, white matter, dura, and scalp. The respective coefficient of variance for GTV coverage was -4.7, -7.0 and -11.7 for E95%, E50% and E5%; -7.5, -7.8 and +148.7 for SAR95%, SAR50% and SAR5%; and +9.6, +48.4 and +23.3 for CD95%, CD50% and CD5%. The average percentage difference for GTV coverage was -2.8%, -2.1% and -1.3% for E95%, E50% and E5%; -2.8%, -2.7% and +0.2% for SAR95%, SAR50% and SAR5%; and +3.0%, +0.4% and +0.9% for CD95%, CD50% and CD5%. CONCLUSION Percent difference in GTV metrics showed reduced coverage when gray and white matter were modeled as one entity. Variability metrics revealed significant disparities, underscoring the importance of differentiating tissue properties in computational modeling of TTFields.