ID #927 Comparing 3D Methodologies for DIPG Tumor Volume Quantification at Diagnosis
Candice Van Skike, Rachel Collins, Sibo ZhaoAbstract
Background
Two-dimensional, cross-sectional tumor measurements are traditional response assessment criteria for DIPG (e.g. DIPG-RAPNO). However, volumetric endpoints correlate better with prognosis. The purpose of this study is to compare ellipsoid volume calculations with volumetric quantification using 3D modeling software.
Methods
We conducted single-center retrospective review of patients with DIPG who had retrievable diagnostic imaging (n = 13). The ellipsoid formula ([π/6]*l*w*h) was used to estimate tumor volume using standard radiographic measurements. 3D volume was calculated using semi-automated techniques in 3D Slicer. Volumes for total tumor, enhancement, and necrosis were calculated using both methods.
Results
Median DIPG lesion volume at diagnosis was 29.0cc (IQR: 23.5-45.3cc) using the ellipsoid formula (3D-EF), compared to 29.9cc (IQR: 26.1-40.7cc) using 3D quantification (3D-Q). Three patients (23%) had enhancing regions with central necrosis. Semi-automated 3D-Q precisely delineated ring enhancement (mean 6.5cc) from necrosis (mean 1.3cc), whereas radiographic measurements quantified enhancement+necrosis together (mean 5.2cc). Two additional patients (15%) had small, focal areas of enhancement without necrosis (mean: 3D-EF=0.5cc; 3D-Q=0.3cc). Although total tumor volume measurements were highly correlated (r = 0.95, t(12)=10.08, p < 0.0001), quantification of enhancement and necrosis were discordant between methods (r = 0.44, t(4)=0.84, p = 0.46). Enhancement with or without necrosis was associated with reduced overall survival (7±5 vs 17±10 months).
Discussion
Current DIPG tumor surveillance methods may be improved with three-dimensional measurements instead of the standard two-dimensional cross-product. In our sample, the ellipsoid formula performed similarly to volumetric modeling when calculating total lesion volume at diagnostic presentation. However, three-dimensional modeling better delineates enhancement and necrosis. More accurate volumetric models are needed to assess treatment response and help differentiate pseudoprogression from progressive disease after radiation. Further, enhanced accuracy is required as tumor quantification for clinical trials becomes more complex, e.g. cystic/necrotic regions are excluded from tumor measurement.